Dissertations in Health Sciences - UEFepublications.uef.fi/pub/urn_isbn_978-952-61-2107-9/urn...examined in another subpopulation (n=1373) of METSIM. n-6 fatty acids, except for All

DIS

SE

RT

AT

ION

S | M

AR

KU

S T

AK

KU

NE

N | C

IRC

UL

AT

ING

FA

TT

Y A

CID

S – A

SS

OC

IAT

ION

S W

ITH

DIE

T... | N

o 350

uef.fi

PUBLICATIONS OF THE UNIVERSITY OF EASTERN FINLAND

Dissertations in Health Sciences

ISBN 978-952-61-2106-2ISSN 1798-5706



MARKUS TAKKUNEN

CIRCULATING FATTY ACIDS – ASSOCIATIONS WITH DIET, GENETIC VARIATIONS, LOW-GRADE

INFLAMMATION AND TYPE 2 DIABETES

Fatty acids in erythrocyte membranes and plasma are used as objective biomarkers

of dietary fat intake. These circulating fatty acids are particularly good in reflecting fish oil intake. In this thesis, circulating marine n-3 fatty acids were associated with lower

incidence of type 2 diabetes. N-6 fatty acids, except for linoleic acid, were associated

with higher low-grade inflammation. Novel evidence was found suggesting that gene-diet interactions modulate circulating fatty acids.

MARKUS TAKKUNEN

Circulating fatty acids – associations with diet, genetic variations, low-grade inflammation and type 2 diabetes

MARKUS TAKKUNEN

Circulating fatty acids – associations with diet, genetic variations, low-grade inflammation and type 2 diabetes

To be presented by permission of the Faculty of Health Sciences, University of Eastern Finland for public examination in CA102, Kuopio, on Friday, 10th of June, at 12 o’clock noon

Publications of the University of Eastern Finland Dissertations in Health Sciences

Number 350

Institute of Public Health and Clinical Nutrition, School of Medicine, Faculty of Health Sciences, University of Eastern Finland

Kuopio 2016

Printing Office: Grano Oy Jyväskylä, 2016

Series Editors:

Professor Tomi Laitinen, M.D., Ph.D. Institute of Clinical Medicine, Clinical Physiology and Nuclear Medicine

Faculty of Health Sciences

Professor Hannele Turunen, Ph.D. Department of Nursing Science


Professor Kai Kaarniranta, M.D., Ph.D. Institute of Clinical Medicine, Ophthalmology


Associate Professor (Tenure Track) Tarja Malm, Ph.D. A.I. Virtanen Institute for Molecular Sciences


Lecturer Veli-Pekka Ranta, Ph.D. (pharmacy) School of Pharmacy


Distributor: University of Eastern Finland

Kuopio Campus Library P.O.Box 1627

FI-70211 Kuopio, Finland http://www.uef.fi/kirjasto

ISBN (print): 978-952-61-2106-2 ISBN (pdf): 978-952-61-2107-9

ISSN (print): 1798-5706 ISSN (pdf): 1798-5714

ISSN-L: 1798-5706

III

Author’s address: Institute of Public Health and Clinical Nutrition University of Eastern Finland Kuopio Finland Email: [email protected]

Supervisors: Associate professor Ursula Schwab, Ph.D.

Institute of Public Health and Clinical Nutrition University of Eastern Finland Kuopio Finland Email: [email protected] Professor Matti Uusitupa, M.D., Ph.D. Institute of Public Health and Clinical Nutrition University of Eastern Finland Kuopio Finland Email: [email protected] Docent Vanessa de Mello Laaksonen, Ph.D. Institute of Public Health and Clinical Nutrition University of Eastern Finland Kuopio Finland Email: [email protected]

Reviewers: Professor Suvi Virtanen, M.sc., M.D., Ph.D.

Nutrition Unit National Institute for Health and Welfare Helsinki Finland & School of Health Sciences University of Tampere Tampere Finland

Docent Kirsi Laitinen, Ph.D. Institute of Biomedicine University of Turku Turku Finland

Opponent: Professor Harri Niinikoski, M.D., Ph.D.

Institute of Biomedicine University of Turku Turku Finland

IV

V

Takkunen, Markus Circulating fatty acids – associations with diet, genetic variations, low-grade inflammation and type 2 diabetes University of Eastern Finland, Faculty of Health Sciences Publications of the University of Eastern Finland. Dissertations in Health Sciences 350. 2016. 78 p. ISBN (print): 978-952-61-2106-2 ISBN (pdf): 978-952-61-2107-9 ISSN (print): 1798-5706 ISSN (pdf): 1798-5714 ISSN-L: 1798-5706 ABSTRACT: Proportions of fatty acids measured in erythrocyte membranes (EM) and plasma are commonly used as objective biomarkers of dietary fat quality. These circulating fatty acids also provide insight into fatty acid metabolism in the human body, and activities of certain desaturase enzymes are often estimated using circulating fatty acids. In this thesis the use of circulating fatty acids as biomarkers of dietary fat intake and their associations with genetic variation, low-grade inflammation and type 2 diabetes (T2D) were examined.

The associations of major sources of dietary fat, estimated by a qualitative food frequency questionnaire, with EM fatty acid composition were investigated in a subpopulation (n=1033) of Metabolic Syndrome in Men study (METSIM). All major sources of dietary fat studied (fish, meat, dairy fat, spreads and cooking fat) were associated with EM fatty acid composition. As expected, the strongest associations were found between fish products and marine n-3 fatty acids in EM.

The associations between EM fatty acids and markers of low-grade inflammation were examined in another subpopulation (n=1373) of METSIM. All n-6 fatty acids, except for linoleic acid (18:2n-6), were associated with higher low-grade inflammation, whereas the anti-inflammatory associations of n-3 fatty acids were modest. Palmitoleic acid (16:1n-7) was associated with a higher concentration of C-reactive protein, whereas its elongation product, vaccenic acid (18:1n-7), was associated with higher adiponectin concentration.

The associations of serum fatty acid composition with T2D incidence, insulin secretion and insulin sensitivity were analyzed in a prospective cohort study (n=407) using repeated measurements based on the randomized Finnish Diabetes Prevention Study. Higher marine n-3 fatty acids and Δ5 desaturase (D5D) activity predicted lower T2D incidence during the long follow-up (median 11 y). These same n-3 fatty acids and D5D also tended to be associated with higher insulin sensitivity.

We hypothesized that a polymorphism in the FADS1 gene (rs174550) could modulate the observed association between intake of marine fatty acids and circulating polyunsaturated fatty acids in EM and plasma. In the participants of the METSIM study, who were homozygous for minor alleles (C/C) of rs174550 (n=168), the association between diet and circulating long-chain n-3 fatty acids was stronger than in those men who were carries of major alleles (T, n=794). Polymorphisms in the same locus were associated with hepatic expression of FADS1 mRNA in the separate Kuopio Obesity Surgery study.

This thesis provides more evidence that endogenous and dietary fatty acids play a role in the development of chronic diseases. Novel evidence was found indicating that gene-diet interactions modulate circulating fatty acid concentrations, which should be considered in future studies that examine the role of dietary and circulating fatty acids in the development of chronic diseases. National Library of Medicine Classification: QU 90, QU 145, QZ 150, QU 500, WK 810 Medical Subject Headings: Fatty Acids; Biomarkers; Humans; Diet; Diabetes Mellitus, Type 2; Inflammation; Fatty Acid Desaturases; Polymorphism, Genetic; Dietary Fats; Erythrocyte Membrane;

VI

VII

Takkunen, Markus Veren rasvahappokoostumus – yhteydet ruokavalioon, geneettisiin variaatioihin, lievään tulehdukseen ja tyypin 2 diabetekseen Itä-Suomen yliopisto, terveystieteiden tiedekunta Publications of the University of Eastern Finland. Dissertations in Health Sciences 350. 2016. 78 p. ISBN (print): 978-952-61-2106-2 ISBN (pdf): 978-952-61-2107-9 ISSN (print): 1798-5706 ISSN (pdf): 1798-5714 ISSN-L: 1798-5706 TIIVISTELMÄ: Rasvahappojen suhteellisia osuuksia punasoluissa ja plasmassa käytetään usein objektiivisina biomarkkereina ruokavalion rasvan laadulle. Verestä mitatut rasvahapot kertovat samalla elimistön rasvahappometaboliasta ja niiden perusteella voidaan esimerkiksi arvioida desaturaasientsyymien aktiivisuutta. Väitöskirjatyössä tutkittiin veren rasvahappokoostumuksen yhteyksiä ruokavalioon, geneettisiin variaatioihin sekä yhteyksiä lievään tulehdukseen ja tyypin 2 diabeteksen (T2D) ilmaantuvuuteen.

Metabolinen oireyhtymä miehillä (MOM) –tutkimuksen osajoukossa (n=1033) tarkasteltiin rasvaa sisältävien ruokien, joiden määrää arvioitiin laadullisella frekvenssikyselyllä, yhteyksiä punasolujen rasvahappokoostumukseen. Kaikki merkittävät rasvanlähteet (kala, liha, maitotuotteet, levitteet ja ruokaöljyt) olivat yhteydessä punasolujen rasvahappokoostumukseen. Yhteydet olivat voimakkaimmat kalarasvojen ja punasoluista mitattujen pitkäketjuisten n-3 rasvahappojen välillä.

Toisessa MOM-tutkimuksen osajoukossa (n=1373) tutkittiin punasolujen rasvahappokoostumuksen yhteyttä lievään tulehdukseen. N-6 rasvahapot, paitsi linolihappo (18:2n-6), olivat yhteydessä korkeampiin tulehdusmerkkiainetasoihin. N-3 rasvahappojen anti-inflammatoriset yhteydet olivat melko vaatimattomat. Endogeeninen palmitoleiinihappo (16:1n-7) oli yhteydessä suurempaan CRP:hen ja tämän rasvahapon elongaatiotuote, vakseenihappo (18:1n-7), oli puolestaan yhteydessä anti-inflammatorisen adiponektiinin pitoisuuteen.

Kohorttitukimuksessa (n=407), joka perustui randomoituun suomalaiseen diabeteksen ehkäisytutkimukseen (DPS), analysoitiin toistetusti mitattujen seerumin rasvahappopitoisuuksien yhteyksiä T2D:een, insuliiniherkkyyteen ja insuliinin eritykseen. Pitkäketjuiset n-3 rasvahapot ja Δ5-desaturaasientsyymin aktiivisuus olivat yhteydessä matalampaan T2D:n ilmaantuvuuteen. Nämä kyseiset löydökset vaikuttivat selittyvän paremmalla insuliiniherkkyydellä.

Hypoteesina oli myös, että FADS1-geenin pistevaihtelu (rs174550) muuntaisi havaittua yhteyttä ruokavalion kalarasvojen ja verestä mitattujen monityydyttymättömien rasvahappojen välillä. MOM-tutkimuksessa (n=962) heillä, joilla oli molemmat harvinaisemmat alleelit (C/C) rs174550:sta, oli voimakkaampi yhteys kalarasvojen ja punasoluista ja plasmasta mitattujen pitkäketjuisten n-3 rasvahappojen välillä kuin heillä, joilla oli yleisempiä alleeleja (T). FADS1:n pistevaihtelut olivat yhteydessä maksan FADS1:n mRNA-ilmentymiseen erillisessä kuopiolaisten lihavuusleikattujen aineistossa.

Tämä väitöskirjatyö esittää lisää uuttaa näyttöä siitä, että rasvahapoilla on merkitystä eri kroonisten sairauksien kehityksessä. Uutta tietoa saatiin siitä, että geeni-ruokavalio yhdysvaikutukset voivat muuntaa veren rasvahappokoostumusta, mikä on huomioitava tulevissa tutkimuksissa, jotka käsittelevät rasvahappojen yhteyksiä sairauksiin. Luokitus: QU 90, QU 145, QZ 150, QU 500, WK 810 Yleinen suomalainen asiasanasto: rasvahapot; markkerit; ihminen; ravitsemus; aikuistyypin diabetes; tulehdus; entsyymit; geneettinen monimuotoisuus; ravintorasvat; punasolut;

VIII

IX

Acknowledgements

I am grateful for the instructions and support by the supervisors of my thesis, associate professor Ursula Schwab, professor Matti Uusitupa and docent Vanessa de Mello Laaksonen. This dissertation would have not been possible without their guidance. I am especially indebted to professor Matti Uusitupa because of his enthusiasm and planning related to my dissertation. I give my thanks to Jyrki Ågren, whose special insight in biochemistry was of high value. I am grateful for the support of Jaana Lindström and professor Jaakko Tuomilehto from the DPS study. I also want to thank all other co-authors who participated to the substudies of this dissertation. I am also thankful for all the other members in the study groups of the METSIM, DPS and KOBS studies.

I want to express my gratitude to professor Markku Laakso and professor Johanna Kuusisto for allowing me to use data from the METSIM study in my thesis. I am thankful to professor Jussi Pihlajamäki for allowing me to use data from the KOBS study for one of the substudies. I thank laboratory technician Sirkku Karhunen for her excellent technical assistance. I thank Teemu Kuulasmaa for the outstanding data management in the METSIM study.

I thank docent Kirsi Laitinen and professor Suvi Virtanen for reviewing the thesis and providing comments that helped me significantly to improve the clarity of the thesis. I express my thanks to docent David Laaksonen for proofreading my thesis.

I thank my family and friends for their support during the past years. I am eternally grateful to Maria for her love and support. Without her, this thesis would have not been possible.

X

XI

List of the original publications

This dissertation is based on the following original publications:

I Takkunen M, Ågren J, Kuusisto J, Laakso M, Uusitupa M, Schwab U. Dietary fat in relation to erythrocyte fatty acid composition in men. Lipids 48: 1093-102, 2013.

II Takkunen MJ, de Mello VD, Schwab US, Ågren JJ, Kuusisto J, Uusitupa MI. Associations of erythrocyte membrane fatty acids with the concentrations of C-reactive protein, interleukin 1 receptor antagonist and adiponectin in 1373 men. Prostaglandins Leukot Essent Fatty Acids 91: 169-74, 2014.

III Takkunen MJ, Schwab US, de Mello VD, Eriksson JG, Lindström J, Tuomilehto J,

Uusitupa MI, DPS Study Group. Longitudinal associations of serum fatty acid composition with type 2 diabetes risk and markers of insulin secretion and sensitivity in the Finnish Diabetes Prevention Study. Eur J Nutr, 2015. (In press, published online)

IV Takkunen MJ, de Mello VD, Schwab US, Kuusisto J, Vaittinen M, Ågren JJ,

Laakso M, Pihlajamäki J, Uusitupa MI. Gene-diet interaction of a common FADS1 variant with marine polyunsaturated fatty acids for fatty acid composition in plasma and erythrocytes among men. Mol Nutr Food Res, 60: 381-9, 2016.

The publications were adapted with the permission of the copyright owners.

XII

XIII

Contents

1 INTRODUCTION ........................................................................................................................... 1 2 REVIEW OF THE LITERATURE ................................................................................................. 3

2.1 Fatty acids ................................................................................................................................... 3 2.1.1 Classification ....................................................................................................................... 3 2.1.2 Dietary sources and absorption ........................................................................................ 4 2.1.3 Fatty acid metabolism ........................................................................................................ 5 2.1.4 Impact of genetic variations on desaturation ................................................................. 8

2.2 Biomarkers of dietary fat .......................................................................................................... 8 2.2.1 Measurement and levels of fatty acids in different tissues .......................................... 8 2.2.2 Circulating biomarkers of fat intake ................................................................................ 9 2.2.3 Estimating enzyme activities by biomarkers of dietary fat ........................................ 10 2.2.4 Gene-diet interactions when studying biomarkers of dietary fat .............................. 11

2.3 Circulating biomarkers of dietary fat and low-grade inflammation ............................... 11 2.4 Circulating biomarkers of dietary fat and type 2 diabetes ................................................ 19

3 AIMS OF THE STUDY ................................................................................................................ 21 4 METHODS ..................................................................................................................................... 23

4.1 Metabolic Syndrome in Men Study (METSIM) (Studies I, II, IV) ..................................... 23 4.1.1 Study population and design .......................................................................................... 23 4.1.2 Biochemical and clinical measurements ........................................................................ 24 4.1.3 Fatty acid composition in erythrocytes and plasma .................................................... 25 4.1.4 Food frequency questionnaire ........................................................................................ 25

4.2 Finnish Diabetes Prevention Study (DPS) (Study III) ........................................................ 27 4.2.1 Study population and design .......................................................................................... 27 4.2.2 Biochemical and clinical measurements ........................................................................ 28 4.2.3 Fatty acid composition in serum .................................................................................... 28 4.2.4 Calculations of insulin secretion, insulin sensitivity and desaturase indices .......... 29

4.3 Kuopio Obesity Surgery Study (KOBS) (Study IV) ............................................................ 29 4.3.1 Study population and methods ...................................................................................... 29 4.3.2 Measurement of hepatic FADS1 mRNA expression ................................................... 29

4.4 Statistical methods ................................................................................................................... 30 4.4.1 Study I ................................................................................................................................ 30 4.4.2 Study II ............................................................................................................................... 30 4.4.3 Study III .............................................................................................................................. 30 4.4.4 Study IV ............................................................................................................................. 31

5 RESULTS ........................................................................................................................................ 33

5.1 Characteristics of the study populations .............................................................................. 33 5.2 Dietary fat and erythrocyte membrane fatty acids (Study I) ............................................ 37

XIV

5.3 Interactions between marine fatty acid intake and a FADS1 variant for circulating fatty acids (Study IV) .................................................................................................................... 40 5.4 Erythrocyte membrane fatty acids and low-grade inflammation (Study II) .................. 43 5.5 Prospective associations of serum fatty acids with type 2 diabetes, insulin secretion and insulin sensitivity (Study III) ............................................................................................... 45

6 DISCUSSION ................................................................................................................................ 51

6.1 Main sources of fat estimated by a FFQ are related to erythrocyte membrane fatty acid composition .................................................................................................................................... 51

6.1.1 Principal findings ............................................................................................................. 51 6.1.2 Meat and fish intake in relation to n-6 fatty acids in erythrocyte membranes ........ 51 6.1.3 Dairy fat and erythrocyte membrane fatty acids ......................................................... 51 6.1.4 Vegetable oil based fats and erythrocyte membrane fatty acids ............................... 52

6.2 A common FADS1 variant may modify the relationship between marine fatty acids and circulating fatty acids ............................................................................................................ 53

6.2.1 Principal findings ............................................................................................................. 53 6.2.2 Gene-diet interactions for circulating fatty acids ......................................................... 53 6.2.3 Mechanism for the gene-diet interactions? ................................................................... 53

6.3 N-6 fatty acids but not linoleic acid in erythrocyte membranes associate with low-grade inflammation ....................................................................................................................... 54

6.3.1 Principal findings ............................................................................................................. 54 6.3.2 N-6 fatty acids and low-grade inflammation ............................................................... 54 6.3.3 N-3 fatty acids and low-grade inflammation ............................................................... 55 6.3.4 Estimated enzyme activities and low-grade inflammation ........................................ 55

6.4 Marine n-3 fatty acids and Δ5 desaturase activity predict lower type 2 diabetes incidence and higher insulin sensitivity .................................................................................... 56

6.4.1 Principal findings ............................................................................................................. 56 6.4.2 Marine n-3 fatty acids and the risk of type 2 diabetes ................................................ 56 6.4.3 Other serum fatty acids and the risk of type 2 diabetes ............................................. 57 6.4.4 Desaturase enzymes and the risk of type 2 diabetes ................................................... 57

6.5 Strengths and limitations ....................................................................................................... 58 7 CONCLUSIONS AND FUTURE IMPLICATIONS ............................................................... 61 8 REFERENCES ................................................................................................................................ 63 ORIGINAL PUBLICATIONS (I-IV)

XV

Abbreviations

ALA Alpha-linolenic acid (18:3n-3)

ARA Arachidonic acid (20:4n-6)

CE Cholesteryl esters

CoA Coenzyme A

CRP C-reactive protein

CVD Cardiovascular disease

D5D Δ5 desaturase

D6D Δ6 desaturase

DGLA Dihomo-gamma-linolenic

acid (20:3n-6)

DHA Docosahexaenoic acid

(22:6n-3)

DPA Docosapentaenoic acid

(22:5n-3)

DPS Finnish Diabetes Prevention

Study

EM Erythrocyte membranes

EPA Eicosapentaenoic acid

(20:5n-3)

FFQ Food frequency questionnaire

IL-1Ra Interleukin-1 receptor

antagonist

IL-6 Interleukin-6

KOBS Kuopio Obesity Surgery

Study

LA Linoleic acid (18:2n-6)

Marine EPA (20:5n-3), DPA (22:5n-3)

fatty acids and DHA (22:6n-3)

METSIM Metabolic syndrome in men

study

MUFA Monounsaturated fatty acids

OGTT Oral glucose tolerance test

PC Phosphatidylcholine

PL Phospholipids

PUFA Polyunsaturated fatty acids

SCD1 Stearoyl-CoA desaturase-1

SFA Saturated fatty acids

SNP Single nucleotide

polymorphism

T2D Type 2 diabetes

TG Triacylglycerides

WHO World Health Organization

XVI

1 Introduction Fatty acids are a major source of energy in human diet due to their high energy content (9 kcal/g). However, their biological impact is much wider. Fatty acids are essential parts of cellular membranes, bind to various types of receptors and transcription factors (1,2), act as precursors to paracrine mediators (e.g., prostaglandins) and may even exhibit lipotoxic effects (2). Fatty acids are also thought to have an important contribution to human disease and health, and it is considered that certain types of fatty acids are more beneficial in the prevention of common chronic diseases, such as cardiovascular diseases (CVD).

Replacing unsaturated fatty acids with saturated fatty acids (SFA) in the diet increases plasma LDL cholesterol concentration, which in turns predisposes to the development of CVD (3,4). Similarly, trans-fatty acids increase LDL cholesterol concentrations and the risk of CVD, probably even more than SFA (5,6). N-3 fatty acids are generally thought to protect from CVD and lower plasma triglyceride concentrations (3,7), but the latest intervention studies supplementing n-3 fatty acids in the form of fish oil have not found any effect on CVD (8,9). Fatty acids may also play a part in the development of type 2 diabetes (T2D). In T2D and its prevention preference of unsaturated over saturated fatty acids is promoted (10), because this tends to improve insulin sensitivity (3,11). The role of n-3 fatty acids in the etiology of T2D, however, is unclear because in cohort studies n-3 fatty acids have been associated with both increased and decreased T2D risk, and supplementation has not improved insulin sensitivity in human intervention studies (3,12,13).

Furthermore, n-3 fatty acids are often referred as anti-inflammatory and n-6 fatty acids as proinflammatory fatty acids due to their eicosanoids products. This might be of high importance when considering the effects of these fatty acids on chronic diseases, as low-grade inflammation is one main pathogenic mechanism behind chronic diseases such as CVD, cancer and T2D (14-16). Yet, the evidence from intervention studies that n-3 and n-6 fatty acids modulate low-grade inflammation in the general population is insufficient (17-19).

Fatty acid composition in plasma and various tissues have been successfully used as biomarkers of certain dietary fatty acids in a number of studies (20). These biomarkers are more objective estimates of dietary fat intake than dietary questionnaires, and they are more feasible to measure in large scale studies. They also provide insight in the endogenous metabolism of fatty acids. Thus, biomarkers of dietary fat intake offer a valuable tool to study the health effects of different types of fatty acids.

The aim of this thesis was to investigate the use of circulating fatty acids measured in erythrocytes and plasma as measures of fat intake and their relationship with genetic variation, low-grade inflammation and the risk of T2D in observational settings.

2

3

2 Review of the literature

2.1 FATTY ACIDS

2.1.1 Classification Fatty acids consist of a functional group, carboxylic acid (-COOH), and of a hydrocarbon chain with a variable length. Fatty acids are classified as saturated if the hydrocarbon chain does not contain double bonds. Unsaturated fatty acids contain one (monounsaturated) or more (polyunsaturated) double bonds in the hydrocarbon chain. The unsaturated fatty acids are further classified into different groups depending on which carbon atom, counting from the non-functional end (i.e. the methyl terminal end), has the first double bond. These fatty acid groups are referred e.g. as n-3, n-6 or n-9 fatty acids, or respectively, as omega-3, omega-6 and omega-9 fatty acids. This grouping is important because these fatty acid groups have different biological properties, and n-3 and n-6 fatty acids are not de novo synthesized in the human body (21). Consequently, two important n-3 and n-6 fatty acids, alpha-linolenic acid (ALA, 18:3n-3) and linoleic acid (LA, 18:2n-6), are referred as essential fatty acids. N-3 fatty acids with longer chain length than ALA, i.e. eicosapentaenoic (EPA, 20:5n-3), docosapentaenoic (DPA, 22:5n-3) and docosahexaenoic (DHA, 22:6n-3) acids are often referred as marine fatty acids, indicating that their source in diet is largely of marine orgin, e.g. fish. Furthermore, the unsaturated fatty acids can be divided into trans- or cis-fatty acids depending on the isomer structure of the double bonds. Cis-fatty acids are the more commonly occurring fatty acid type in the nature.

Fatty acids can be divided according to their chain length. Fatty acids with an odd-numbered chain length are thought not to be de novo synthesized in the human body, but instead are produced by bacteria in the gut of ruminant animals (22). Fatty acids are also sometimes referred as short-chain fatty acids (< 7 carbon atoms), medium-chain fatty acids (>6 carbon atoms), long-chain fatty acids (>12 carbon atoms) and very-long-chain fatty acids (>20 carbons). Most fatty acids are straight chain fatty acids, but fatty acids with branched chains also exist in humans (23).

Fatty acids occur as free fatty acids in the body, but they are more often incorporated into other molecules. The most common of these are glycerol, making up triacylglycerides (TG); phosphate, making up very diverse types of phospholipids (PL); and cholesterol, forming cholesteryl esters (CE). Fatty acids bind also to other organic molecules, such as carbohydrates.

4

Figure 1. An example of a fatty acid. The trivial name is oleic acid (18:1n-9). Systematic name is (cis-9-)octadecenoic acid. Sometimes the abbreviation OA is used.

The nomenclature of fatty acids is diverse. Both trivial names and systematic names are used. Besides, some fatty acids are often referred to by their abbreviations, such as EPA. The structure of fatty acid is expressed as number of carbon atoms (e.g. 22), number of double bonds in the hydrocarbon chain (:5) and the first double bond from the methyl end (n-3). Figure 1 shows another example of a common unsaturated cis-fatty acid, oleic acid (18:1n-9) that has a chain length of 18 carbons and one double bond at the n-9 position in the cis conformation.

2.1.2 Dietary sources and absorption Men consume on average roughly 90 g and women 70 g of fat daily in Finland, totaling slightly over a third of daily energy intake (24). Some common sources of dietary fat are listed in Table 1. Dairy products and red meat are examples of food products that contain mostly SFA and monounsaturated fatty acids (MUFA) and only small amounts of polyunsaturated fatty acids (PUFA). Most of the trans-fatty acids are derived from dairy and meat products in Finland (25). Fish and chicken products have somewhat different fatty acid composition compared with beef and pork (Table 1). Eggs and fish contain mainly unsaturated fatty acids. The marine products, such as fish and fish oil, are the main sources of EPA and DHA. Nuts and seeds are also good sources of unsaturated fatty acids, and for example peanuts and sunflower seeds are high in essential fatty acid, LA. Rapeseed oil (Canola oil) is a commonly used cooking oil in Finland and in addition to LA, it has an especially high content of the other essential fatty acid, ALA, unlike other commonly used vegetable oils. Olive oil is a good source of LA and MUFA, especially oleic acid (18:1n-9).

Fat absorption takes place in the small intestine as reviewed in (26) and summarized here. First, fat is emulsified by bile acids and lecithin secreted by the liver. As the surface area of the fat increases, triglycerides and other fat molecules are hydrolyzed by enzymes called lipases, which are mainly secreted by the pancreas. Thus, free fatty acids and monoglycerides are produced. Then the free fatty acids and monoglycerides in combination with bile acids form very small micelles, which transport the fatty acids to epithelial cells. Fatty acids with short chain lengths (≤12 carbon atoms), which are more water soluble, are then absorbed directly into the portal blood and transported to the liver. Monoglycerides and long-chain fatty acids are instead taken up by endoplasmic reticulum of enterocytes and recombined into triglycerides. Triglycerides are then packed into lipoproteins called chylomicrons and secreted into lymphatic vessels. These vessels combine into the thoracic

5

duct that ends at the subclavian vein and finally releases the chylomicrons to circulating blood.

2.1.3 Fatty acid metabolism The main events in fatty acid metabolism after absorption are summarized in this paragraph as reviewed in (28). After the chylomicrons have entered the circulating blood, an enzyme called lipoprotein lipase, located in the endothelial cells of capillaries, hydrolyzes triglycerides into free fatty acids. Fatty acids are also transported from the liver by another type of lipoprotein, very low-density lipoprotein (VLDL). In adipocytes, the free fatty acids are again combined into triglycerides and stored as lipid droplets. In muscle cells, however, the fatty acids are primarily oxidized for energy. Fatty acids are activated by coenzyme A (CoA) and are then β-oxidized inside mitochondria. Each round of β-oxidation shortens the fatty acids by two carbons, and acetyl-CoA, NADH and FADH2 are produced. These are used in the citric acid cycle and finally in the respiratory chain to produce energy, i.e. ATP and heat. Unlike other fatty acids, very-long chain fatty acids are β-oxidized and branched fatty acids α-oxidized inside peroxisomes. While a person is fasting and fatty

Table 1. Examples of dietary sources of fatty acids.

Minced meat (pork and

beef) Butter Egg Salmon Fish oil

capsule* Peanuts

Rape-seed oil

Olive oil

Typical portion size (g)

80-230 3-10 60 100-215 1-3 20-100 10-40 10-40

/100 g of product

Energy (kcal) 221 727 143 195 760 568 884 884

Fatty acids (g) 13.5 76.7 6.6 11.9 72.6 35.1 98.4 93.3

Saturated fatty acids (g)

5.6 52.8 2.1 2.5 5.1 6.3 5.7 14.0

Monounsaturated fatty acids (g)

6.1 19.4 3.3 4.5 16.1 17.2 59.6 68.4

N-3 fatty acids (g) 0.2 0.4 0.2 4.2 50.3 <0.1 10.9 0.5

N-6 fatty acids (g) 0.9 1.3 1.0 0.7 1.1 11.6 22.1 10.4

Trans fatty acids (g)

0.2 2.0 0 0 - 0 0 0

Linoleic acid, 18:2n-6 (g)

0.9 1.0 0.9 0.5 - 11.6 22.1 10.4

Alpha-linolenic acid, 18:3n-3 (g)

0.1 0.4 0.1 0.3 - <0.1 10.9 0.5

EPA, 20:5n-3 (g) <0.1 0 0 0.8 26.3 0 0 0

DHA, 22:6n-3 (g) <0.1 0 0.1 2.3 18.2 0 0 0

Portion sizes and amounts of fatty acids and energy (/100 g) are based on Fineli® food composition database (27). *Based on a label of a common commercial supplement available in Finnish grocery stores. EPA, Eicosapentaenoic acid; DHA, docosahexaenoic acid.

6

acids are the primary source of energy, the liver converts the fatty acids into ketone bodies that are used as energy in extrahepatic tissues.

Fatty acids are also de novo synthesized in the human body, especially by the liver, as reviewed in (21) and summarized in the following. Acetyl-CoA is used to produce malonyl-CoA that is the substrate for fatty acid synthesis. The synthesis is catalyzed by an enzyme system called fatty acid synthase. This system is located in cytosol and it lengthens the fatty acid chain by two carbons in each cycle. The end product of fatty acid synthase is mainly palmitic acid (16:0). Palmitic acid in turn may be elongated or desaturated and used to produce phospholipids and triglycerides in the endoplasmic reticulum.

The processes related to modifying the fatty acids by elongation and desaturation are pivotal concerning this thesis. Figure 2 presents the outline of the main fatty acid modifications in human body adapted from several sources (29-34). Saturated and monounsaturated fatty acids can be de novo synthesized and modified in the human body (Figure 2A). N-3 and n-6 fatty acids can only be synthesized from the essential fatty acids ALA and LA, or they have to be acquired from the diet (Figure 2B).

After synthesis or absorption, fatty acids are elongated by a group of enzymes, of which the rate-limiting enzymes are called elongases (30). In humans seven different elongases are known with varying specificity for different fatty acid substrates (30). Addition of double bonds to fatty acids is catalyzed by three different types of desaturases: stearoyl-CoA desaturase-1 (SCD1), Δ5 desaturase (D5D) and Δ6 desaturase (D6D). SCD1 acts on saturated fatty acids and D5D and D6D on PUFA as shown in the Figure 2. Production of EPA, DPA and DHA from ALA is lower in men than in women and is also down-regulated by intake of EPA and DHA (35-37). The activity of SCD1 is also upregulated by a diet high in SFA in comparison with a diet high in MUFA (38). In addition to what is shown in the Figure 2, DHA can be retroconverted into EPA and DPA in the human body, even though this seems to happen in only minor quantities (39).

Besides being added to other organic molecules to produce e.g. triacylglycerides or phospholipids, certain PUFA are used as substrates by enzymes called cyclooxygenases and lipoxygenases to produce eicosanoids and docosanoids (Figure 2B). These products, such as prostaglandins and leukotrienes, act as paracrine mediators in processes such as inflammation and thrombosis (33).

7

Figure 2. An outline of the major modifications of fatty acids in the human body as adapted from several sources (29-34). Panel A presents the endogenous pathway, and panel B the pathways starting from the essential fatty acids, ALA and LA, i.e. the modification of n-3 and n-6 fatty acids. FADS1 and FADS2 are the genes encoding D5D and D6D.

8

2.1.4 Impact of genetic variations on desaturation D5D and D6D enzymes are encoded by the FADS1 and FADS2 genes, respectively, located in the FADS gene cluster in chromosome 11. In 2006, Schaeffer et al. showed that polymorphisms in FADS1 and FADS2 genes are strongly related to fatty acid composition in serum (34). There are several common (minor allele frequency > 5%) single nucleotide polymorphisms (SNP) in FADS gene cluster that associate with levels of circulating PUFA, and many of these SNP, located relatively close together, are in linkage disequilibrium, i.e. they represent similar genetic variation of this genetic region. Most of the minor alleles of these SNP associate with increased levels of substrates of the desaturase enzymes and decreased levels of products, especially arachidonic acid (ARA, 20:4n-6) (34,40,41). These associations are indeed in accordance with the fatty acid metabolism pathway presented in Figure 2B and suggest especially decreased activity of D5D and also D6D caused by these variations. Although the exact mechanisms and functional SNP are still unknown, SNP in the FADS region associate with decreased hepatic expression of FADS1 and its protein product in human liver, which is expected to lead to decreased D5D activity (42). One example of a common SNP in FADS region is the intron variant of FADS1 gene, rs174550, which is strongly associated with fatty acid composition in erythrocytes, and is also, as explained above, associated with decreased activities of D5D and D6D (43).

SCD1 is encoded by the SCD gene on chromosome 10. There are also several SNP in the SCD gene, but their effect on SCD1 activity is unclear and probably low (44-46). 2.2 BIOMARKERS OF DIETARY FAT

2.2.1 Measurement and levels of fatty acids in different tissues Gas chromatography is the prevailing method used to quantify fatty acid levels in tissues. Besides gas chromatography, other methods such as nuclear magnetic resonance or mass spectroscopy are sometimes used (47,48). Fatty acids can and have been measured in a myriad of tissues, e.g. leukocytes, thrombocytes, erythrocytes, plasma, adipocytes, sperm, milk, skin, muscle and liver (20).

Lipid fractions are often, but not always, separated before the fatty acid quantification, for example in plasma, to separate PL, TG, CE and free fatty acids, chromatographic methods such as solid phase or thin layer chromatography are used (49). Before gas chromatography, the fatty acids need to be separated from the lipid molecules. This is done by a process called trans-esterification that produces fatty acid methyl esters. In gas chromatography, the fatty acids are separated by their retention times and can be quantified by their respective peak areas by flame ionization detectors. Although gas chromatography produces quantifiable results (mol/l) of each detected fatty acid when an internal standard is added, the common practice is to report fatty acids relative to all fatty acids, i.e. as weight or molar percentage.

The fatty acid composition of different tissues and lipid types greatly vary, and fatty acid composition of different tissues is extensively presented in (20). As an example of an important difference, CE contain abundantly LA (~50 mol%), due to the specificity of lecithin-cholesterol acyl transferase for this fatty acid, but contains very little of the n-3 fatty acids EPA, DPA and DHA (20). Other good examples are the very long-chain fatty acids, such as nervonic (24:1n-9) and lignoceric (24:0) acids, which are almost solely found in a type of PL called sphingolipids due to the action of enzymes called ceramide synthases (30).

9

In TG oleic acid (18:1n-9) is the predominant fatty acid and usually totals over one third of all fatty acids in TG (20). Despite these differences, most individual circulating fatty acids correlate with between different sites of measurement moderately or strongly (20,50). Of circulating fatty acids, erythrocyte membrane (EM) fatty acids are especially interesting to study, since erythrocytes, unlike thrombocytes, white blood cells and lipoproteins, have a long three- to four-month lifespan, and mature erythrocytes also lack most of the capability for fatty acid metabolism (51,52). Some examples of fatty acid distributions for circulating fatty acids in total serum, EM and plasma CE, PL and TG can also be found in chapter 5.1.

2.2.2 Circulating biomarkers of fat intake It was shown already over 50 years ago that circulating levels of fatty acids respond to dietary changes (53,54). Nowadays, circulating fatty acids in EM and plasma are commonly used in the validation of dietary questionnaires, as measures of adherence in intervention trials and in the assessment of dietary fat type, whereas platelets and white blood cells are less often used (20). Certain fatty acid concentrations in blood can be considered as objective estimates of dietary intake of fat. Dietary questionnaires can be biased by the behavior of the responder and investigator. Furthermore, dietary questionnaires are somewhat laborious to fill in, performing them in large scale studies is time consuming, and the accuracy of databases for calculating fatty acid composition in different foods may be limited. As peripheral blood is commonly drawn in studies, measurement of fatty acids in plasma or erythrocytes is a good alternative to gain objective information on fat intake. However, the individual fatty acid levels in blood need to be carefully interpreted, because, in addition to the dietary intake, endogenous fatty acid metabolism contributes to circulating fatty acid composition (see 2.1).

Circulating biomarkers of fat intake are most appropriate when estimating intake of EPA and DHA, and the correlations between dietary questionnaires and circulating fatty acids are usually in the range r=~0.4-0.6 (55). The sum of EPA+DHA, also known as the omega-3 index, increases in response to n-3 supplementation, and a low omega-3 index has been proposed to act as a risk predictor for coronary heart disease (56). This index also correlates strongly (r=0.9) with levels of EPA, DPA and DHA in whole blood and plasma PL (56). A randomized controlled trial supplementing varying amounts of EPA+DHA in healthy men and women showed that the dose explained almost 70% of total variation in the change of erythrocyte EPA+DHA, and the second strongest predictor was the baseline omega-3 index (5% of variation) (57). Both fish oil and DPA supplementations increase levels of circulating DPA notably (58,59). At least concerning EPA and DHA, erythrocytes seem to be more stable in their fatty acid composition and reflect the intake of EPA and DHA over longer period of time (months) than plasma (days-weeks) (60-63). This difference is supposed to be related to the three- to four-month life cycle of erythrocytes and their lack of capability for fatty acid synthesis. Although this makes erythrocytes more useful in long-term studies, on the other hand, EM fatty acid composition cannot be used in very short-term studies. The half-life of EPA in the different plasma fractions PL, TG and CE is in the order of days (60,64).

As the only source of the essential fatty acids ALA and LA is diet, it is not surprising that their circulating levels can be used as biomarker. For LA, the correlations between dietary questionnaires range usually between r=~0.2-0.7 and for ALA r=~0.1-0.4, and their circulating levels increase after their intake is increased (20). Both plasma (free fatty acids,

10

PL, CE and TG) and erythrocytes (total PL and phosphatidyl choline (PC)) seem to reflect increased intake of LA over a similar time period (days-weeks) (65). Furthermore, it is known that circulating levels of ARA change significantly even after low-dose ARA supplementation (66).

Some of the circulating saturated fatty acids can also be used to reflect SFA intake. Myristic (14:0), pentadecanoic (15:0) and heptadecanoic (17:0) acid seem to reflect saturated fat intake especially from dairy products (67,68). However, pentadecanoic and heptadecanoic acids are also found in many other food products such as fish and meat and concerns have been raised about their use as biomarkers of dairy fat (69). Palmitic acid also seems to correlate with its estimated intake to small extent (20), but it has to be remembered that this fatty acid is the main product of fatty acid synthase. In one study pentadecanoic acid in plasma and erythrocytes seemed to reflect changes in SFA intake over similar period of time (days-weeks) (65).

Circulating MUFA are not generally used as biomarker of dietary fat. However, in populations with high intake of olive oil, oleic acid (18:1n-9) acts as a biomarker for oleic acid intake (70).

Circulating trans-fatty acids, which are not produced in the human body, are sometimes used to reflect their intake. Trans fatty acids, such as trans-elaidic (trans-18:1n-9) and trans-palmitoleic (trans-16:1n-7), reflect their intake from partly hydrogenated fats, dairy products and beef that contain some (Table 1) trans-fatty acids (70,71).

2.2.3 Estimating enzyme activities by biomarkers of dietary fat Product-to-precursor ratios are commonly used in studies to estimate enzyme activities related to endogenous synthesis and modification of fatty acids presented in Figure 2. Particularly, estimated enzyme activities are used for the three desaturase enzymes, SCD1, D5D and D6D, e.g. (72,73). The validation evidence for desaturase indices, however, is still somewhat limited. SCD1 is estimated by the ratio of 16:1n-7/16:0 or 18:1n-9/18:0. In a study in which SCD mRNA expression and estimated activity were measured in adipose tissue, both ratios 16:1n-7/16:0 and 18:1n-9/18:0 correlated with the mRNA expression (74). However, in another study in which SCD expression was measured in liver, plasma and liver, the ratio of 16:1n-7/16:0 reflected mRNA expression better than the ratio of 18:1n-9/18:0. The 16:1n-7/16:0 ratio may therefore be a more suitable index, possibly because it is less affected by diet (75). D5D and D6D are usually estimated by ratios of 20:4n-6/20:3n-6 and 18:3n-6/18:2n-6, respectively. Sometimes in phospholipid fractions also the ratio of 20:3n-6/18:2n-6 is used for D6D, because the amount of 18:3n-6 in PL is low. However, there is currently little evidence that these indices are associated with liver or adipose tissue mRNA expression of FADS1 or FADS2 (74,75). Still one study found that in children D6D activity is associated with mRNA expression of both FADS1 and FADS2 in peripheral blood (76). In another study a common FADS1 SNP was clearly associated with both D5D and D6D indices measured in erythrocytes (77), providing some validation for the use of these indices.

Besides estimated desaturase activities, indices for de novo lipogenesis and elongase activity are used. Because LA is not synthetized in the human body and palmitic acid (16:0) is the main product of de novo lipogenesis, the ratio of these two, 16:0/18:2n-6 is used to reflect de novo lipogenesis. In plasma TG this index is increased when hepatic de novo lipogenesis is induced by a high-carbohydrate diet, and this ratio correlates with hepatic

11

mRNA expression of genes related to de novo lipogenesis (75,78,79). Still, this index does not distinguish between de novo lipogenesis and diets high in palmitic acid and low in LA. Elongase activities can be estimated for example by ratios of 18:0/16:0, 18:1n-7/16:1n-7 and 22:4n-6/20:4n-6, but there is little evidence for validity of these indices (75).

2.2.4 Gene-diet interactions when studying biomarkers of dietary fat It is possible that besides the direct effects of individual genes and diet on circulating fatty acid composition, their combined effect may modulate fatty acid composition differently, i.e. a genotype may increase or decrease the effect of diet on circulating fatty acids. A number of recent studies have investigated gene-diet interactions of fat intake and FADS1 and FADS2 polymorphisms for circulating biomarkers of dietary fat both in intervention and observational settings (77,80-87). The data, however, is still quite limited to draw firm conclusions about the gene-diet interactions. The studies tend to suggest that intake of ALA interacts with variants in FADS cluster for circulating EPA (82,84,86). Secondly, it is possible that intake of PUFA from marine sources, may interact with FADS SNP for D5D and D6D activities and related fatty acid products (80,81,83,85,87). Interestingly, the direction of interaction may differ depending in which tissue fatty acids are measured (80,83,84), which may suggest that the nature of these interactions is very complex or that some of the observed interactions have arisen by chance.

Two studies have also investigated if the relationship between diet and circulating SCD1 indices is modified by SNP in SCD. One study found that the type of dietary oil (olive oil vs. sunflower oil) and SCD variants may have an interaction with the SCD1 index based on 18:1n-9/18:0, but no interaction with 16:1n-7/16:0 was found (46). Another study, in which fish oil was supplemented, found no gene-diet interaction for either of SCD1 indices (88).

In light of the aforementioned, it is clear that more research is needed to understand the nature of gene-diet interactions, since these interactions may play a part in diet-disease and gene-disease relationships and may also affect how the biomarkers of dietary fat should be applied in further studies. 2.3 CIRCULATING BIOMARKERS OF DIETARY FAT AND LOW-GRADE INFLAMMATION

Low-grade inflammation is a phenomenon in which concentrations of inflammatory markers, such as C-reactive protein (CRP), are mildly increased without existence of actual infectious or inflammatory disease. Low-grade inflammation is also linked to progression of atherosclerosis and T2D (14,16). Higher low-grade inflammation is related to other common risk factors of chronic diseases, such as aging, smoking, obesity, high waist circumference and low physical activity (89).

An extensive literature review has indicated that many dietary components, including fatty acids, could play a role in low-grade inflammation and suggested that trans- and saturated fatty acids are proinflammatory and especially marine n-3 fatty acids are anti-inflammatory (90). Furthermore, n-6 fatty acids are often referred to as pro-inflammatory fatty acids because the eicosanoid products of ARA have often stronger effects on inflammation than the eicosanoid products of EPA (33). In contrast, n-3 fatty acids are commonly considered to be anti-inflammatory fatty acids because they replace and

12

compete for enzymatic pathways with ARA in cells and give rise to anti-inflammatory products such as resolvins (33).

There is no clear view which of the inflammatory markers should be used to describe low-grade inflammation, and a myriad number of markers of various types have been investigated (90). CRP is one of the most commonly reported markers of inflammation. It is an acute-phase protein secreted from liver and acts in the complement system. It seems to have no causal role, however, and thus serves solely a marker of low-grade inflammation (91). Interleukin-6 (IL-6) and adiponectin concentrations are also very commonly reported (Table 2). IL-6 is a pro-inflammatory cytokine secreted by leukocytes and adipocytes and probably has a causal role at least in CVD (92,93). Adiponectin on the other hand, is an anti-inflammatory adipokine secreted by adipocytes and also, for example, decreases glucose production in liver and is associated with decreased risk of T2D (94,95). Interleukin-1 receptor antagonist (IL-1Ra) has so far been used less often (Table 2), but was included for this thesis. IL-1Ra is an anti-inflammatory cytokine that blocks interleukin-1 signaling. However, it has proinflammatory associations and is increased for example in T2D and non-alcoholic steatohepatitis probably since it is secreted by liver and adipose tissue in pro-inflammatory situations to impede the effect of interleukin-1β (90,96,97).

Many observational studies in humans have investigated the associations of biomarkers of dietary fat with a number of circulating markers of inflammation. Due to high number of studies and inflammatory markers, the overview of studies in Table 2 is focused only on circulating biomarkers of dietary fat and their associations with selected important markers of low-grade inflammation, i.e. with concentrations of pro-inflammatory CRP, IL1-Ra and IL-6 and with concentrations of anti-inflammatory adiponectin. The associations of circulating fatty acids with markers of low-grade inflammation were generally weak in these studies, and the studies often reported only some of the measured fatty acids (Table 2).

Saturated fatty acids were generally associated with higher concentrations of inflammatory markers (Table 2). Of individual saturated fatty acids, myristic acid (14:0) tends to be associated with higher low-grade inflammation, whereas for longer individual SFA the associations are not clear and associations to both directions have been observed.

For total circulating MUFA the associations were equivocal (Table 2). Palmitoleic acid (16:1n-7) associates somewhat consistently with high low-grade inflammation. This may be explained by the fact that SCD1 activity also tends to associate with higher inflammation. Accordingly, also oleic acid (18:1n-9) tended to associate with increased low-grade inflammation. For longer MUFA the data are insufficient. In contrast, one large-scale study found that trans-16:1n-7 was associated with lower CRP and IL-6 (71).

Total PUFA, n-3 and n-6 fatty acids were mainly associated with lower low-grade inflammation, even though the n-6/n-3-ratio tended to be associated with higher inflammation (Table 2). Of n-3 fatty acids, ALA does not show consistent association with low-grade inflammation, whereas EPA, DPA and DHA tend to be associated with lower inflammation. Of the n-6 fatty acids, LA seems to be associated with lower inflammation, and 18:3n-6 and dihomo-gamma-linolenic acid (20:3n-6, DGLA) seem to be pro-inflammatory. Neither ARA nor 22:4n-6 is consistently associated with low-grade inflammation. Only a few studies have investigated associations with estimated D5D and D6D. In those studies, it seems that associations with D5D are anti-inflammatory and with

13

D6D pro-inflammatory, which is in accordance with the associations of the individual n-6 fatty acids, i.e. LA, 18:3n-6, DGLA and ARA.

It is noteworthy that most of the reviewed studies are cross-sectional, and therefore provide little insight into whether the fatty acids affect inflammation or the other way around. Only study by Peterson et al. has investigated longitudinal associations (over 20 years) of circulating fatty acids with CRP in a population setting (98). This study found pro-inflammatory associations with palmitoleic and oleic acid and with SCD1 and D6D activities, but interestingly, also with EPA. Moreover, LA was associated with lower CRP. Furthermore, three dietary intervention studies have investigated correlations between change in circulating n-6 and n-3 fatty acids and inflammatory markers, but found no associations (99-101).

Table 2. Overview of studies on the association of circulating fatty acids with markers of low-grade inflammation in adults in the order of publication year.

Ref Subjects and setting*

n (male%) Mean age, years

Fatty acid CRP IL-6 IL-1Ra Adipo-nectin

(102) Coronary angiography patients

269 (63.6%) 60 Granulocyte EPA 0

Granulocyte DPA 0

Granulocyte DHA ↓

Granulocyte LA 0

Granulocyte 18:3n-6 0

Granulocyte ARA 0

(103) Spaniards 232 (58%) 40 Serum 14:0 0 0/↑ Serum 16:0 0 0/↑ Serum 18:0 0 0/↑ Serum 24:0 0 0/↑ Serum LA 0/↓ 0/↓ Serum EPA 0/↓ 0 Serum DHA 0/↓ 0 Serum SFA 0 0/↑ Serum Σn-6 0 0/↓ Serum Σn-3 0/↓ 0 (104) Healthy

Caucasians 116 (66%) 39 Plasma 14:0 0/↓

Plasma 16:0 ↓

Plasma 20:1n-9 ↓

Plasma SFA 0/↓

Plasma EPA 0

Plasma DHA 0

Plasma ALA 0

Plasma LA 0

Plasma Others 0

(105) Italians 1,123 (45%) 68 Plasma LA 0 0 0/↓ Plasma ARA 0 ↓ ↓ Plasma ALA ↓ 0 ↓ Plasma EPA 0 ↓ 0/↓ Plasma DHA 0 ↓ ↓ Plasma Σn-3 0 ↓ ↓ Plasma Σn-6 0 0 ↓

Plasma Σn-6/Σn-3 0 ↑ ↑

Continued on the next page.

14

Table 2. Continued.




(98) Swedish 50y men, 20y longitudinal study

767 (100%) 50 until 70

CE 16:1n-7 ↑ CE 18:1n-9 ↑ CE 18:3n-6 0/↑ CE LA ↓

CE SCD1(-16) ↑

CE D6D ↑ CE D5D 0 CE EPA 0/↑ CE Others 0

Cross-sectional 320 (100%) 70

CE LA 0/↑ CE DGLA ↑ CE D5D 0/↓ CE D6D 0/↑ CE Others 0 (106) Female

nurses ~270 (0%) 60 EM EPA ↑ 0

EM DPA 0 0 EM DHA 0 0 Plasma EPA 0 0 Plasma DPA 0 ↓ Plasma DHA 0 0 (101) 12-wk

intervention (low carbohydrate vs. low fat diet)

40 (50%) 18-55 PL ΔARA 0 0


876 (77%) 60 EM ARA/LA ↑ (108) Patients with

stable coronary heart disease

992 (82%) 67 EM EPA+DHA ↓ ↓

(109) 70 y old Swedes

264 (56%) 70 CE 14:0 0 0 CE 16:0 0 0 CE 16:1n-7 ↑ 0 CE 18:0 0/↓ 0 CE 18:1n-9 0/↑ 0 CE LA ↓ 0 CE 18:3n-6 ↑ 0 CE ALA 0 0 CE DGLA 0/↑ 0 CE ARA 0 0 CE EPA 0 0 CE DHA 0 0

CE SCD1(-16) ↑ 0

(110) English civil servants

348 (100%) 79 PL SFA ↑ PL MUFA 0 PL PUFA ↓ Continued on the next page.

15

Table 2. Continued.





291 (63.6%) 60 Thrombocyte EPA 0/↑

Thrombocyte DHA 0

Granulocyte EPA 0/↑

Granulocyte DHA 0/↑

(112) Healthy Australians

124 (37.1%) 48 Plasma Σn-3 ↓ Plasma EPA ↓ Plasma DPA ↓ (113) With stable

coronary heart disease

956 (82%) 67 Whole blood EPA+DHA

↓ ↓ (114) Healthy

Greeks 374 (51%) 42 Plasma SFA ↑ ↑

Plasma MUFA ↓ 0/↓ Plasma PUFA ↓ ↓ Plasma Σn-3 ↓ ↓ Plasma Σn-6 ↓ ↓

Plasma Σn-6/Σn-3 0/↑ 0/↑

Plasma LA ↓ 0/↓ Plasma ARA 0/↓ 0/↓ Plasma ALA ↓ 0/↓ Plasma EPA ↓ 0 Plasma DPA ↓ ↓ Plasma DHA ↓ 0/↓

Plasma EPA+DHA 0/↓ 0

(71,115) US citizens 3,630 (45%) 75 PL 16:1n-7 0/↑ 3,736 (45%) 75 PL trans16:1n-7 0/↓

Female nurses 327 (0%) 60 EM

trans16:1n-7 ↓ ↓ (116) Yup'ik

Eskimos 357 (41%) 45 EM EPA 0/↓ 0 0

EM DHA 0/↓ 0 0

(117) Middle-aged Germans

1,980 (38%) 50 EM LA 0/↓ 0/↑

EM 18:3n-6 0/↑ 0/↓

EM DGLA ↑ 0/↓

EM ARA 0 0

EM 22:4n-6 0 0/↓

EM ALA 0 0/↑

EM EPA 0 0

EM DPA 0/↓ 0

EM DHA 0 0 Continued on the next page.

16

Table 2. Continued.




(118) Patients referred to an oral glucose tolerance test

70 (40%) 51 EM-PC 16:0 0

EM-PC LA 0

EM-PC EPA ↑

EM-PC DHA ↑

EM-PE 16:0 0

EM-PE LA 0

EM-PE EPA 0/↑

EM-PE DHA 0/↑

Plasma-PC 18:0 ↓

(119) Patients with stable heart failure

183 (80.9%) 56 Plasma MUFA ↑

Plasma 18:1n-9 ↑

Plasma 18:1n-7 ↑

Plasma EPA ↓

Plasma 20:4n-3 ↓

Plasma 22:0 ↓

Plasma 22:4n-6 0

Plasma 23:0 ↓ Plasma 24:0 ↓ (120) Finns 1,395

(100%) 52 Serum Σn-3 ↓

Serum EPA+DPA +DHA

0/↓

Serum EPA 0 Serum DPA ↓ Serum DHA 0/↓ Serum ALA 0 (121) Elderly

Chinese 3,107 (43%) 58 EM 16:1n-7 ↑ ↓

(122) Japanese

municipal employees

489 (58%) 42 CE 16:0 0/↑ CE ALA 0/↓ CE DGLA ↑ CE LA 0 CE EPA 0 CE DHA 0 CE 16:1n-7 0 CE ARA 0 CE Others 0 CE SCD1 0 CE D5D ↓ (123) US citizens 2,848 (47%) 62 PL LA ↓ 0 PL 18:3n-6 0 ↑ PL DGLA ↑ ↑ PL ARA 0 0 PL ALA 0 0 PL EPA 0 ↓ PL DPA ↓ ↓ PL DHA 0 ↓ Continued on the next page.

17

Table 2. Continued.




(99) Heart failure patients, baseline

1,203 (80.7%)

67 PL Σn-3 0 ↓

PL EPA ↓ ↓

PL DHA 0 ↓

3 mo intervention with EPA+DHA (1g/d) /placebo

PL EPAΔ 0

0

PL DHAΔ 0 0

(124) Healthy Koreans

593 (100%) 49 PL 18:1n-9 0

PL LA ↑

PL DGLA 0

PL ARA 0

PL ALA 0

PL EPA 0

PL DPA ↑

PL DHA 0

PL D6D 0

PL D5D 0

PL SCD1(-16) 0

PL SCD1(-18) 0

(125) Patients with peripheral artery disease

64 (100%) 67 EM EPA+DHA ↓ 0/↓ (126) Healthy

Japanese 195 (100%) 50 EM 24:0 0/↑

(127) 20-29y old

Canadians 965 (29%) 23 Plasma 14:0 ↑ ↑

Plasma 16:0 ↑ 0/↑ Plasma 18:0 0/↓ 0/↓

Plasma 16:1n-7 ↑ ↑

Plasma 18:1n-9 0/↑ 0

Plasma 18:1n-7 0/↑ 0

Plasma LA 0/↓ 0/↓

Plasma 18:3n-6 0 0

Plasma DGLA ↑ ↑ Plasma ARA 0/↑ 0

Plasma 22:4n-6 0 0

Plasma ALA 0/↑ 0 Plasma EPA 0/↓ 0 Plasma DPA 0 0 Plasma DHA 0 0


18

Table 2. Continued.




(128) Morbidly obese women referred to bariatric surgery

16 (0%) 43 Serum PUFA ↓ 0 Serum Σn-3 ↓ 0 Serum Σn-6 ↓ 0 Serum SFA ↑ 0 Serum MUFA ↑ 0 Serum 12:0 ↑ ↑ Serum 14:0 ↑ ↑ Serum 16:0 0 0 Serum 18:0 0 0 Serum 20:0 0 0 Serum 22:0 0 ↓ Serum 24:0 0 0

Serum 14:1n-5 ↑ 0

Serum 16:1n-7 ↑ ↑

Serum 18:1n-9 ↑ 0

Serum 20:1n-9 0 ↑

Serum 24:1n-9 0 0

Serum LA ↓ 0 Serum DGLA 0 0 Serum ARA ↓ ↓

Serum 22:4n-6 0 0

Serum EPA ↓ 0 Serum DPA 0 0 Serum DHA ↓ ↓

Serum Σn-6/Σn-3 ↑ 0

(129) Cree (Canadians)

744 (45%) 40 EM EPA 0 0 EM DPA ↓ 0 EM DHA 0 0 (130) Health

examination participants in Korea

1,022 (52%) 47 PL MUFA ↑ PL 18:1n-9 ↑

PL SCD1(-18) ↑

PL 16:1n-7 ↑ DGLA ↑ (100) Healthy

adults 112 (52%) 26 EM ALA 0/↑ 0

EM EPA 0 0 EM DPA ↓ 0 EM DHA 0 0 EM EPA+DHA 0 0 EM LA 0/↓ ↓ EM ARA 0 0

5 mo intervention with fish oil (0, 0.3, 0.6, 0.9 or 1.8 g/d)

EM AllΔ 0 0


19

Table 2. Continued.




(131) Middle-aged Germans

1,533 (37%) 49 EM SCD1(-16) 0/↑ 0

EM D5D 0/↓ 0

EM D6D 0 0

(132) US nurses 762 (0%) 60 Plasma 20:0+22:0+24:0

↓ ↑

EM 20:0+22:0+24:0

0 0

US health professionals

1,265 (100%)

64 Plasma 20:0+22:0+24:0

0 0 ↑

EM 20:0+22:0+24:0

0 0 0

(59) Healthy subjects with high triglycerides

26 (88%) 44 EM DPA 0 0

Inverse associations marked with '↓', no association with '0', positive association with '↑' and '↓/0' or '0/↑' means that there were associations in some subgroups or models. PL and CE refer to plasma fractions. *All reported associations were cross-sectional if not otherwise stated. ALA, Alpha-linolenic acid (18:3n-3); ARA, Arachidonic acid (20:4n-6); CE, Cholesteryl esters; CRP, C-reactive protein; DGLA, Dihomo-gamma-linolenic acid (20:3n-6); DHA, Docosahexaenoic acid (22:6n-3); DPA, Docosapentaenoic acid (22:5n-3); D5D, Δ5 desaturase; D6D, Δ6 desaturase; EM, Erythrocyte membranes; EPA, Eicosapentaenoic acid (20:5n-3); IL-1Ra, Interleukin-1 receptor antagonist; IL-6, Interleukin-6; LA, Linoleic acid (18:2n-6); MUFA, Monounsaturated fatty acids; PC, Phosphatidylcholine; PE, Phosphatidylethanolamine; PL, Phospholipids; PUFA, Polyunsaturated fatty acids; SCD1(-16), Stearoyl-CoA desaturase-1 (16:1n-7/16:0); SCD1(-18), Stearoyl-CoA desaturase-1 (18:1n-9/18:0); SFA, Saturated fatty acids.

2.4 CIRCULATING BIOMARKERS OF DIETARY FAT AND TYPE 2 DIABETES

In the management and prevention of T2D an increase in dietary unsaturated fatty acids and decrease in SFA are generally recommended (10). Unsaturated fatty acids, when compared with SFA may improve insulin sensitivity (3,11). Especially intake of LA seems to be protective from T2D (3). Dietary n-3 fatty acids, however, do not seem to play a role in insulin sensitivity according to controlled intervention studies (13). Furthermore, there has been inconsistencies between the prospective associations of dietary n-3 with T2D incidence, as the Asian studies have found inverse associations and the Western studies positive associations (133).

The mechanisms on how dietary fat modulates insulin sensitivity in humans are poorly understood (134). Experimental studies have revealed, for example, that SFA induces innate inflammation and synthesis of diacylglycerides and ceramides, all of which impair insulin signaling (134).

To better understand the effect of dietary fatty acids on T2D some studies have investigated the prospective associations between circulating biomarkers of dietary fat and T2D incidence measuring either plasma or EM fatty acids (71,115,135-145). These studies suggest that many individual fatty acids are associated with T2D risk, but the associations

20

are not entirely consistent. Circulating total SFA seem to predict higher risk of T2D (137,142), but associations with individual SFA vary (143). Shorter SFA (14:0, 16:0 and 18:0) are associated with increased risk and odd-chain (15:0 and 17:0) and very-long-chain fatty acids with decreased risk of T2D (135,138,143,145). For total MUFA there seems to be no clear association, but proportion of palmitoleic acid (16:1n-7), the product of SCD1, tends to associate with an increased risk of T2D (136,140,142,146). Circulating total PUFA show trends towards a lower risk of T2D in some studies (135,137,138). Of the n-6 fatty acids, LA has been associated with lower T2D risk quite consistently (136-138,140,142), whereas 18:3n-6 and DGLA predict higher risk (136-138,142). Circulating n-3 fatty acids have been associated with T2D only in some studies, in which the associations have been inverse for ALA, EPA, DPA and DHA (135,137,139,141). Trans-16:1n-7, which is partly derived from dairy products, has been associated with decreased risk of T2D in only one study (71), whereas trans-16:1n-9 and trans-18:1 were associated with increased risk in another study (144). Interestingly, a recent study using EM showed that a high lipophilic index, i.e. decreased cellular membrane fluidity based on fatty acid composition, was associated with higher T2D risk (146). This suggests one mechanism by which individual fatty acids may be related to T2D risk, because for instance saturated and trans-fatty acids decrease fluidity due to their high melting points, whereas PUFA increase fluidity.

Activities of SCD1 and D6D have been associated with higher risk of T2D, whereas D5D associated with a lower risk of T2D (138,142). A recent study investigated the possible mechanisms behind these associations with desaturase indices and suggested that accumulation of liver fat and plasma triglycerides could mediate the associations with T2D (131). There is some evidence that SCD1, D5D and D6D enzymes are causally involved in the development of T2D (44,147,148).

Few prospective studies have investigated the longitudinal associations between circulating fatty acids and the two main pathogenic mechanisms of T2D, insulin secretion and insulin resistance. In two studies within the Metabolic syndrome in men study (METSIM) using insulin sensitivity and secretion indices based on oral glucose tolerance tests (OGTT) over 5 years’ time, high estimated activity of SCD1 and D6D and low activity of D5D were associated with lower insulin sensitivity and SCD1 and D6D were also associated with lower insulin secretion (140,142). High SFA in plasma was associated with decreased insulin sensitivity, and high SFA and decreased LA were associated with decreased insulin secretion. Furthermore, in one cross-sectional study, which used intravenous measures of insulin sensitivity and secretion, 15:0 was associated with both higher insulin sensitivity and secretion (145).

21

3 Aims of the study

The general aims of the thesis were to use circulating fatty acids as biomarkers of dietary fat and investigate their associations with genetic variation, low-grade inflammation and the risk of T2D. The main hypotheses are summarized in Figure 3.

Specific aims in the substudies: 1. To establish the associations of major sources of dietary fat estimated by a previously

unvalidated qualitative food frequency questionnaire (FFQ) with EM fatty acid composition in a population-based sample of men. Study I.

2. To examine how the EM fatty acids and estimated desaturase and elongase enzyme

activities associate with markers of low-grade inflammation in another population-based sample of men. Study II.

3. To study prospective associations of serum fatty acids and estimated desaturase

activities with T2D incidence, insulin sensitivity and insulin secretion in overweight men and women with impaired glucose tolerance participating in a lifestyle intervention study. Study III.

4. To examine if a common, known FADS1 variant (rs174550), representing genetic

variation in the FADS gene cluster, could modify the relationship between dietary PUFA intake from fish and fish oil and circulating PUFA in a population-based sample of men. To confirm the association between genetic variation in FADS gene cluster and hepatic FADS1 expression in a separate sample of obesity surgery patients. Study IV.

22

Figure 3. A diagram of the hypothesized associations in the four substudies.

23

4 Methods

4.1 METABOLIC SYNDROME IN MEN STUDY (METSIM) (STUDIES I, II, IV)

4.1.1 Study population and design A total of 10,197 men, aged from 45 to 70 years, participated in the baseline study of METSIM in 2005–2010. In this population based study, all men in this age group listed in the population register of the city of Kuopio in Eastern Finland (97,000 inhabitants) were contacted by an invitation letter. The participation rate was 70%. The original study protocol has been reported elsewhere (149). EM fatty acid composition was measured in a random sample of 1,395 men who participated in the study in 2006–2010 and did not have a diabetes diagnosis prior the METSIM study (Figure 4A). In study II, subjects who were considered to have acute inflammation (n=20), i.e. a cut-off value of CRP over 10 mg/l was chosen, or who had missing data (n=2) were excluded. Thus, in the study II, cross-sectional associations between EM fatty acids and markers of low-grade inflammation were examined in 1,373 men.

The METSIM participants who were free of diabetes at baseline were invited to a phase 2 study in average 5 years later. Altogether 1,033 men, aged from 47 to 75 years and who participated consecutively in the phase 2 examination in 2010–2011, were asked to complete a FFQ and their EM fatty acid composition was measured (Figure 4B). In the cross-sectional study I, the final number of subjects in each statistical model ranged between 904 and 1012 mainly depending on the availability of the FFQ data. In study IV, marine PUFA intake could be estimated using a FFQ in 983 men who had also EM fatty acid composition measured. Plasma fatty acid composition was also measured later but in fewer men (n=874). The men who reported a very high marine PUFA intake (> 3.5 SD) were excluded (n=10). Ten participants had missing genotype data and nine participants had other missing data and they were also excluded. Thus, the cross-sectional study IV included 962 men who had their EM (n=954) or plasma (n=845) fatty acid composition measured.

The METSIM study was approved by the Ethics Committee of the University of Kuopio and Kuopio University Hospital and it was conducted in accordance with the Helsinki Declaration. All participants signed a written informed consent.

24

Figure 4. Design of the substudies. CE, Cholesteryl esters; CRP, C-reactive protein; DHA, Docosahexaenoic acid (22:6n-3); EM, Erythrocyte membranes; EPA, Eicosapentaenoic acid (20:5n-3); FADS1, Gene encoding Δ5 desaturase; FFQ, Food frequency questionnaire; IL-1Ra, Interleukin-1 receptor antagonist; PL, Phospholipids; TG, Triacylglycerides; T2D, Type 2 diabetes; WHO, World Health Organization.

4.1.2 Clinical and biochemical measurements The subjects had a 1-day outpatient visit to the Clinical Research Center at the University of Eastern Finland where clinical measurements were completed and blood samples were drawn after an overnight fast. Height was measured (without shoes) in the Frankfurt position to the nearest 0.5 cm. Weight was measured in light clothing with a digital scale (Seca 877, Seca, Hamburg, Germany), and rounded to the nearest 0.5 kg. Waist circumference was measured at the midpoint between the lowest rib and iliac crest to the nearest 0.5 cm. The participants were also asked about their smoking habits, alcohol consumption, medications and leisure-time physical activity.

Two different immunoturbidimetric assays by Roche (Roche Diagnostics GmbH, Mannheim, Germany) with a Roche Cobas C501 analyzer (Hitachi High Technology Co., Tokyo, Japan) and by Beckman Coulter (Fullerton, CA, USA) were used for the plasma high-sensitivity CRP measurements (Study II). The results of the Beckman Coulter method were converted to the Roche method level. Altogether 51 subjects had CRP concentrations below the detection limits (0.15 mg/l and 0.20 mg/l) and the value of 0.14 mg/l was used for them. For the measurement of plasma IL-1Ra the Quantikine immunoassay was used (R&D Systems Inc., Minneapolis, MN, USA). For the total plasma adiponectin immunoassay the

25

Human Adiponectin ELISA Kit (Linco Research, St Charles, MI, USA) was used. Intra-assay CV% was 12.8% and inter-assay CV% varied between 13.9% and 28% for the adiponectin measurements. For the CRP and IL-1Ra measurements intra-assay and inter-assay CV% were ≤6.7% and ≤14.0%, respectively.

Genotyping (Study IV) was performed from blood using the Illumina HumanOmniExpress BeadChip manufactured by Illumina (San Diego, CA) (150). For studies III and IV we selected an intron-variant of FADS1 gene, rs174550, which is a known glucose-raising variant and is strongly associated with EM fatty acid composition (43,151). The SNP was in Hardy-Weinberg equilibrium (P=0.45) with a minor allele frequency of 42%.

4.1.3 Fatty acid composition in erythrocytes and plasma Erythrocytes were separated from EDTA-blood by centrifugation at 1000g for 10 min (4 °C) and hemolyzed in the tris–HCl buffer (pH 7.6, 10 mmol/l). Erythrocyte membranes were prepared by centrifugation of hemolysate at 30,000g for 30 min at 4 °C. Membrane sediment was resuspended in 0.5 ml of distilled water. Fatty acid methyl esters were prepared by direct transesterification, which gives more complete recovery especially of sphingomyelin-derived fatty acids compared to separate extraction and transesterification (152). After mixing 0.1 ml of membrane suspension and 2 ml of methanol–toluene (4:1, v:v) in a glass tube, 0.2 ml of acetyl chloride was slowly added and this mixture was incubated at 100 °C for 1 h. After cooling in cold water 5 ml of 6 % K2CO3 was carefully added and then vigorously shaken. Toluene was separated into an upper phase by centrifugation at 2,000g for 5 min and injected into the gas chromatograph (Agilent Technologies 7890A) with a 25-m FFAP column (Agilent Technologies, Wilmington, DE). Pure standards (NU Chek Prep, Inc., Elysian, MA) were used for the identification of fatty acid methyl esters and for the preparation of calibration curves. Heptadecanoic acid (17:0) methyl ester served as an internal standard.

In plasma, the CE, TG and PL lipid fractions were extracted using chloroform-methanol (2:1), separated using a solid-phase aminopropryl column (49) and then measured using gas chromatography (153) as explained before. Briefly, separated lipid fractions were transmethylated with boron trifluoride in methanol, and fatty acid methyl esters were separated using the same equipment as explained above. Nonadecanoic acid (19:0) served as the internal standard.

Fatty acid quantities were converted into molar percentages (mol%). Estimated enzyme activities were measured similarly as in other studies, e.g. (75). D5D activity was estimated by the product-to-precursor ratio 20:4n-6/20:3n-6; D6D activity was estimated by the ratio 20:3n-6/18:2n-6 in EM (Study II) and 18:3n-6/18:2n-6 in CE and TG (Study IV). Elongase activities were estimated by the ratios 18:0/16:0 and 18:1n-7/16:1n-7 and SCD1 activity by the ratios 16:1n-7/16:0 and 18:1n-9/18:0 in Study II.

4.1.4 Food frequency questionnaire We used a previously unvalidated qualitative FFQ similar to the one previously used in the FINRISK study (154) with minor modifications. In our study the information about fish species and dairy products consumed was inquired about in more detail, bread type classifications were more specific, and the brand names for spreads and cooking fats were updated.

26

During their visit to the research center, a total of 1033 participants filled in the FFQ. The participants were asked to choose from eight frequency categories (ranging from never/less than once a month to four or more frequently per day) how often they had used specified food products (70 items) during the past 12 months. Bread consumption was inquired by questions in which the subjects chose how many slices they eat from eight categories (never/slice a week to six or more slices a day). The subjects were also allowed to freely specify how many glasses (2 dl) of milk, hot chocolate and sour milk they drink. In subsequent questions milk, spread and cooking fat types usually consumed were inquired. Lastly, the participants were asked which nutrient supplements they had used in previous 6 months.

The FFQ results were used to create continuous variables which were presumed to relate to different levels of fatty acid intakes. Because the qualitative FFQ did not include portion sizes, we used typical portion sizes as approximations. Four questions concerning fish types consumed were combined by multiplying frequencies by PUFA content of different fish types (e.g. twice a day multiplied by 2.6 g of PUFA per 100 g of rainbow trout or salmon). The use of fish oil supplement was included in this sum of fish-derived PUFA intake (approximation of 0.65 g/day of PUFA was used). For milk products SFA and MUFA content was calculated similarly. For different milk products, estimations of portion sizes were used (e.g. cheese 10 g per portion, yoghurt 150 g per portion). SFA, MUFA and PUFA intakes from meat were correspondingly calculated summing relevant variables weighted by portion size (e.g. cold cuts 20 g, meat 100 g) and fatty acid content (e.g. meat PUFA 1.4 g/100 g). Additionally, SFA, MUFA and PUFA intakes from spreads and cooking fats were estimated by multiplying bread consumption by fatty acid content of spread typically consumed (approximation of 5 g spread per slice was used) and summing this with cooking fat type typically used (an approximation of 10 g/day was used for all types of cooking fats) weighted by its fatty acid content. All SFA, MUFA and PUFA contents of different food products were derived from the Finnish Food Composition Database (Fineli®, National Institute for Health and Welfare, Helsinki, Finland). Additionally, total fast food consumption was estimated by summing unweighted frequencies per day of different types of fast food.

Subjects with missing or unclear answers were excluded when the intake variables were formed. Since there were high intercorrelations (rs > 0.90) within the food groups weighted by either SFA, MUFA or PUFA contents, certain fat intake variables were not used in the final statistical analyses (Table 3). In other words, the variable of SFA from milk and milk products closely reflects total fat intake from milk, meat intake variables reflect total fat intake from meat, and the variable of PUFA from spreads and cooking fat reflects both MUFA and PUFA intake from spreads and cooking fat. Final numbers of subjects and specific information of each variable are presented in Table 3.

27

Table 3. The calculation of fat intake variables for specific food groups.

Food group Products in the FFQ

Number of food items included

Created fat intake variables

Weighted bya nb

Fish and fish oil

Oily fishes (e.g. rainbow trout), Baltic herring, freshwater fishes with medium fat content (e.g. vendace), low fat fishes (e.g. perch), fish oil supplement

5 PUFA Fatty acids, portion size

983

Milk and milk products

Milk, hot chocolate, sour milk (fat > 1%), unflavored yoghurt (fat > 1%), flavored yoghurt (fat > 1%), cheese (fat ≤ 17%), cheese (fat > 17%), ice cream

8 SFA, MUFAc Fatty acids, portion size, milk type mainly used

951

Meat Meat (minced meat, beef etc.), sausages, cold cuts, cold cuts (full meat)

4 SFAd, MUFAd, PUFA

Fatty acids, portion size

1012

Spread and cooking fat

Spread and cooking fat type typically used

2 SFA, MUFAd, PUFA

Fatty acids, portion size, bread slices

967

Butter use

Spread and cooking fat type typically used

2 Yes/ No

‒ 122/871

Fast food Pizza, hamburgers, French fries, ready meals, nibbles (potato chips, popcorn)

5 Overall frequency

‒ 977

aBefore summing individual eating frequencis, they were multiplied by approximated portion sizes, fatty acid (SFA, MUFA or PUFA as g/100g) compositions of the food products (Fineli®) and using data from the FFQ (milk type and bread slices). bFor whom FFQ and erythrocyte fatty acid composition data available. cThe MUFA variable was omitted from the statistical analyses due to the high intercorrelation (rs > 0.90) with the SFA variable in the same food group. dThe variables were omitted from the statistical analyses due to the high intercorrelation (rs > 0.90) with the PUFA variable in the same food group. FFQ, Food frequency questionnaire; MUFA, Monounsaturated fatty acids; PUFA, Polyunsaturated fatty acids; SFA, Saturated fatty acids.

4.2 FINNISH DIABETES PREVETION STUDY (DPS) (STUDY III)

4.2.1 Study population and design Originally, 522 people at a high risk of T2D participated in the randomized, controlled DPS study (NCT00518167, ClinicalTrials.gov) (155,156). Eligibility criteria were age 40–65 years, overweight (BMI ≥ 25 kg/m2) and impaired glucose tolerance at repeated OGTT (mean value of 2 h plasma glucose concentrations 7.8–11.0 mmol/l in two OGTT). People with previous diagnosis of diabetes, severe chronic disease or unstable clinical conditions related to glucose metabolism were excluded from the study. The study baseline measurements took place in 1993–1998 (recruitment period), and the participants were randomized into a control (n = 257) or intensive lifestyle intervention group (n = 265) in five study centers in Finland. In the DPS, the main endpoint was the diagnosis of diabetes defined by the World Health Organization (WHO) 1985 criteria (plasma fasting glucose ≥7.8 or 2 h glucose ≥11.1). All diabetes diagnoses were ascertained by a second positive OGTT and confirmed by a physician. The active period of the lifestyle intervention study lasted 1–6 years (median 4 years), and is referred hereafter as the intervention period. Ten participants dropped out before the first annual visit. After the end of the intervention period in 2001, the annual follow-up visits were continued and 89 % of all 522 participants attended at least one follow-up visit in 2001–2009 (157).

28

The goals of the intensive, individualized lifestyle intervention included a weight loss of at least 5%, moderate daily physical activity (4 h/week), fat intake below 30% and saturated fat intake below 10% of total energy intake and the intake of dietary fiber 15 g/1000 kcal or more. The control group received only general information about healthy diet and benefits of physical exercise in the beginning of the study. According to the original study design, participants in both groups discontinued the study after the diagnosis of T2D.

Because the original eligibility criteria were based on the WHO 1985 diabetes criteria, which were revised in 1999, study III included 407 participants who were non-diabetic by the revised WHO 1999 criteria at baseline, and had one or more stored serum samples available from the intervention period (1-6 years) (Figure 4C). In total 96 participants were excluded because of high glucose levels. Thus, total serum fatty acid composition was not measured from participants who had fasting plasma glucose ≥7.0 mmol/l (n = 62) or 2 h glucose ≥11.1 mmol/l (n = 43) at the baseline OGTT, which was the latest OGTT of the two performed before the study baseline examination. Due to other missing data, the number of participants varied in final statistical models.

DPS has been conducted according to the Declaration of Helsinki, and all participants have provided a written informed consent. A separate consent was asked for genotyping. The study design has been approved by the Ethics Committee of the National Public Health Institute of Finland.

4.2.2 Biochemical and clinical measurements At baseline and each annual visit, including the extended follow-up (2001–2009), an OGTT was performed, serum lipids were measured after overnight fast, several anthropometric and other clinical measurements were taken, and participants filled in a physical activity questionnaire, a 3-day food record and a medical health questionnaire (156,158). Participants smoking at least weekly at any time of the intervention period were considered as smokers. Alcohol, dietary fiber, carbohydrate and energy intakes calculated from 3-day food record data and were available only until the third year of the study. Thus, for the fourth and fifth year, the intakes of the third year were used as covariates in the statistical models. Genotyping was performed from blood using MetaboChip, a custom Illumina iSelect array manufactured by Illumina Inc. (San Diego, CA, USA) (150). 4.2.3 Fatty acid composition in serum The total serum fatty acid composition was measured by TETHYS Bioscience Inc. (Emeryville, CA, USA) in 2010 using stored (−80 °C) serum samples taken during the active intervention period. The lipids from serum were extracted in the presence of authentic internal standards by the method of Folch et al. (159) using chloroform and methanol (2:1). The total lipid extract was trans-esterified in 1 % sulfuric acid in methanol in a sealed vial under a nitrogen atmosphere at 100 °C for 45 min. The resulting extract was neutralized with 6 % potassium carbonate, and the fatty acid methyl esters were extracted with hexane. The fatty acid methyl esters were separated and quantified by capillary gas chromatography (Agilent Technologies model 6890) equipped with a 30-m DB-88MS capillary column (Agilent Technologies) and a flame ionization detector. The proportions of fatty acids were converted into molar percentages (mol%).

29

4.2.4 Calculations of insulin secretion, insulin sensitivity and desaturase indices The OGTT results at 0, 30 and 120 min were used to calculate surrogate indices of insulin secretion and sensitivity as previously described (160). As an index of insulin sensitivity, the Matsuda insulin sensitivity index was calculated (161). Insulin secretion was estimated by the ratio of insulin area under the curve to glucose area under the curve based on glucose and insulin concentrations at 0 and 30 min (149). Disposition index was calculated as insulin sensitivity index times insulin secretion index (149). Concentrations of plasma glucose and insulin at 30 min were not measured at baseline. Thus, the indices could be calculated only for five participants at baseline.

Desaturase activities were measured similarly as in other studies, e.g. (75). To estimate SCD1 activity, the ratio 16:1n-7/16:0 was used. D5D and D6D activities were estimated by product-to-precursor ratios 20:4n-6/20:3n-6 and 18:3n-6/18:2n-6, respectively. For validation of estimated D5D and D6D activities measured in total serum, we selected an intron variant of FADS1 gene (rs174550).

4.3 KUOPIO OBESITY SURGERY STUDY (KOBS) (STUDY IV)

4.3.1 Study population and methods All patients undergoing obesity surgery in Kuopio University Hospital are recruited into an ongoing study investigating metabolic consequences of obesity surgery (162). The participants were instructed to follow a preoperative very low calorie diet (VLCD) (600-800 kcal/d) and consume special products designed for VLCD for four weeks prior to surgery. Liver biopsies were obtained during the surgery.

The variant rs174547 (FADS1) was genotyped using the TaqMan SNP Genotyping Assay (Applied Biosystems, Foster City, CA). This SNP is in complete linkage disequilibrium (r2=1) with rs174550 according to the HapMap CEU population (http://hapmap.ncbi.nlm.nih.gov). In total, 240 patients were included in the cross-sectional study investigating the effect of the genotype on mRNA expression (Figure 4D). Because dietary data was not available in the KOBS, we considered PL-EPA and PL-DHA as estimates of habitual dietary intakes of EPA and DHA in the 91 participants who had these measurements available.

The study protocol has been approved by the Ethics Committee of the Northern Savo Hospital District, and it was performed in accordance with the Helsinki Declaration. Written informed consent was obtained from the subjects.

4.3.2 Measurement of hepatic FADS1 mRNA expression TruSeq targeted RNA expression, a custom gene panel by Illumina was used to measure FADS1 gene expression level in liver. Total RNA (150ng) from liver was reverse-transcribed using the ProtoScript II Reverse Transcriptase (New England BioLabs, Ipswich, MA). Oligo pool targeted regions of interest were hybridized to cDNA. Next, hybridized cDNA was extended by DNA polymerase followed by ligation using DNA ligase. The extension-ligation products were amplified with PCR and AMPure XP beads (Beckman Coulter, Brea, CA) were used to clean up the PCR products. Equal volumes of the products were pooled together and quantitated with the DNA 1000 chip (Agilent Technologies, Santa Clara, CA). Finally, the pooled sample was diluted, denatured and sequenced with MiSeq (Illumina). The reads for gene expression of each gene in the custom gene panel per sample were

30

normalized based on the total number of all gene expression reads of the corresponding sample. 4.4 STATISTICAL METHODS

4.4.1 Study I Spearman’s rank correlation test was used to analyze the association between dietary fat and EM fatty acids. Log-transformation was used to correct for skewness of variables, when appropriate. Differences between the means of continuous variables were tested by t-tests. Linear regression and ANCOVA were used to adjust the models for age, BMI, smoking, statin medication and alcohol consumption. All statistical analyses were performed using SPSS Statistics 19 (IBM, Armonk, NY).

4.4.2 Study II Distributions of CRP, IL-Ra and adiponectin were skewed, and they were log-transformed for the statistical analyses. Pearson correlations were calculated for unadjusted correlations between the markers of inflammation and EM fatty acids. Confounder-adjusted partial correlations (rpartial) were calculated using linear regression adjusted for age, smoking, alcohol consumption (g/wk), waist circumference, statin medication and exercise at leisure time (little/none vs. regularly at least 30 min at a time). Williams׳ test was used to compare correlation coefficients. The significance level (α=0.05) was corrected for multiple testing (between three markers of inflammation and 20 fatty acids) with the Bonferroni method and P-value below 0.05/[3×20]=0.00083 was considered statistically significant. This corresponds to r and rpartial≈|0.09| when n=1373. Analyses were performed using SPSS Statistics v. 19 (IBM, Armonk, NY).

4.4.3 Study III The intervention and control groups were pooled for the analyses. Skewed variables were log-transformed when deemed necessary. Partial correlation coefficients were adjusted for age, sex, study group, study center, smoking, alcohol intake, waist circumference and physical activity at leisure. ANCOVA was used to analyze association of rs174550 with estimated D5D and D6D activities, and effect sizes were estimated using η2 and partial η2.

We analyzed 23 fatty acids, calculated ratios and sums in relation to four outcome variables (T2D incidence and insulin sensitivity, insulin secretion and disposition indices). Cox regression was used to examine longitudinal associations of serum fatty acids with T2D incidence (WHO 1985 criteria) until 2009. Participants who did not develop diabetes were censored at their last visit (1–16 years after randomization, median 11 y). Schoenfeld residuals were plotted against time to test the proportional hazards assumption.

Both Cox regression and mixed models were adjusted for putative confounding factors including age, smoking (yes/no), sex, study group, alcohol consumption (g/day), total leisure time physical activity (h/week), waist circumference, serum concentration of triglycerides, dietary fiber intake (g/day), carbohydrate intake (g/day) and total energy intake. Serum triglycerides were considered as a confounding variable because the fatty acid composition of TG differs from the fatty acid composition of PL and CE. Thus, changes in serum concentration of total triglycerides alter the proportions of total serum fatty acids (20). For the repeated measurements analysis, linear mixed models with a random individual effect were used to examine longitudinal associations of (lagged) serum fatty

31

acids with each subsequent year’s (non-lagged) insulin sensitivity, insulin secretion and disposition indices during the intervention period. In the linear mixed models, besides the confounding variables of each respective year, the visit year and the interaction term visit × group were also included in the models. In the Cox regression models, stratification was done by the study center, and in the linear mixed models, the study center was included as a random factor. We also considered lagged outcome variables, i.e. concentrations of fasting and 2 h glucose in the Cox regression models and insulin sensitivity, insulin secretion and disposition indices in linear mixed models, as possible confounders. After adding these lagged outcome variables into the models, the random individual and study center effects became redundant in the mixed models and, hence, a first-order autoregressive covariance matrix by annual visits was applied.

Insulin sensitivity, insulin secretion and disposition indices were log-transformed for the analyses, and the coefficients were potentiated (10β) to express the results as proportional difference per 1 SD of fatty acids. P-value <0.05 was considered as statistically significant. Statistical analyses were performed using SPSS, version 21 (IBM, Armonk, NY).

4.4.4 Study IV Skewed variables were log-transformed when necessary. ANOVA, Kruskal-Wallis and chi-square tests were used to compare variables across categorical groups. Linear regression was used to examine associations between the FADS1 genotype (rs174550) and standardized fatty acid proportions. To construct additive linear regression models, the genotype was included as a continuous variable, which allows presentation of the regression coefficients in relation to the SDs of a fatty acid proportion per minor allele. Unlike in the main analysis, the regression lines were plotted for all three allele pairs (T/T, T/C and C/C) for illustrative purposes. Linear regression models also included marine PUFA intake (continuous) and a continuous interaction term marine PUFA intake*genotype. To account for possible confounding factors between intake and circulating fatty acids, the regression models were adjusted for age, statin medication (yes/no), smoking (yes/no), BMI (kg/m2) and alcohol consumption (g/wk). As a sensitivity analysis, we analyzed the interactions after excluding men who used fish oil supplements. Additionally, we did a further adjustment for PUFA intake from cooking fats and spreads. We considered P-value <0.05 as nominally significant. The significance level was corrected for multiple testing by the Bonferroni method. P-value <0.003 (=0.05/15) was considered as statistically significant after correction only for the number of different fatty acids (n=10) and their ratios and sums (n=5), because the proportions of fatty acids correlate between EM and plasma CE, TG and PL. SPSS Statistics v. 21 (IBM, Armonk, NY) was used for the statistical analyses. Post-hoc power calculations were performed using R 3.2.4 for Windows (http://www.statmethods.net/stats/power.html).

32

33

5 Results

5.1 CHARACTERISTICS OF THE STUDY POPULATIONS

The characteristics of the METSIM and KOBS study populations are shown in Table 4. The plasma concentrations of CRP and IL-1Ra were positively correlated (r=0.325), whereas their correlations with concentrations of plasma adiponectin were inverse (r=−0.108 and r=−0.129, respectively) in the METSIM baseline study, indicating that they only partly reflect the same variation in low-grade inflammation. The FADS1 variant rs174550 was not associated with any of the characteristics in the METSIM phase 2 study, which was expected. Proportions of fatty acids in EM in the METSIM are presented in Table 5. Table 6 shows PUFA composition in plasma CE, TG and PL in the METSIM phase 2 study and serum PL in the KOBS study.

The characteristics of the DPS study participants are presented in Table 4. Almost all characteristics and the proportions of fatty acids changed during the intervention period (1-6 years), justifying the use of repeated statistical models (Table 4 and Table 7). At the DPS baseline, most dietary fatty acids were associated with their respective proportions in serum, as expected (Table 8), showing that the serum samples, despite the long storage, reflect dietary fat type. Because the desaturase activities are usually calculated in defined lipid fractions instead of whole serum used in DPS, we validated our estimated D5D and D6D activities against a previously known intron variant rs174550 of FADS1 gene. Rs174550 explained 12 % (P = 6.6 × 10−11) of variance in baseline D5D and 20 % (P = 2.5 × 10−18) of variance in baseline D6D. The adjustment for estimated D5D or D6D in the ANCOVA did not blunt the explained variance as rs174550 explained 20 % (P = 2.9 × 10−17) of variance in D5D and 26 % (P = 1.2 × 10−24) of variance in D6D. These results indicate that the estimated activities indeed reflect real desaturase enzymes activities.

34

Table 4. Characteristics of the METSIM, KOBS and DPS studies.

METSIM baseline

(n=1373)

METSIM phase 2

(n=1031) KOBS

(n=240) DPS baseline

(n=387)

Male (%) 100 100 30 33 Intervention group (%) - - - 51 Age (years) 55 ±5.6 64 ±5.9 48 ±8.9 55 ±7.1

Smokers (%) 18 12 - 12 No/little exercise during leisure (%) 31 34 - - Leisure-time physical activity (h/wk) - - - 7.3 ±6.1*

Statin treatment (%) 14 36 33a - Use of fish oil supplements (%) - 23 - - Consumption of fish at least once a week (%) - 85 - -

FADS1 variant rs174550/rs174547 (%) T/T - 32 30 32 T/C - 50 49 49 C/C - 18 21 19

Alcohol consumption (g/wk) 102 ±123 81 ±108 - 43 ±98*

Fiber intake (g/d) - - - 20 ±7.7

Carbohydrate intake (g/d) - - - 191 ±57*

Total fat intake (g/d) - - - 73 ±29*

Energy intake (kcal/d) - - - 1777 ±528*

Waist circumference (cm) 96 ±9.8 98 ±10 - 101 ±11*

Body mass index (kg/m2) 26 ±3.5 27 ±3.7 43 ±5.4 31 ±4.7*

HbA1c (%) 5.6 ±0.40 5.8 ±0.37 - - Fasting plasma glucose (mmol/l) 5.8 ±0.60 5.8 ±0.60 6.5 ±1.8 5.9 ±0.59*

2-hour glucose (mmol/l) - - - 8.6 ±1.3*

Serum triglycerides (mmol/l) - - - 1.7 ±0.71*

Matsuda insulin sensitivity index0,30,120 - - - 4.1b ±1.9

Insulin secretion index0,30 - - - 34c ±21

Disposition index - - - 116.4b ±45*

C-reactive protein (mg/l) 1.5 ±1.6 - - - Interleukin-1 receptor antagonist (pg/ml) 202 ±128 - - -

Adiponectin (µg/ml) 7.8 ±4.3 - - -

Values are means ±SD or proportions. The data of two separate subpopulations of METSIM are shown (random sample at baseline and participants who participated in the 5-y follow-up study consecutively). an=58, bn=171 at year 1, cn=175 at year 1 *P<0.05 for change during the intervention period of DPS. P-values were calculated in the whole cohort (n=407) using repeated linear mixed model. METSIM, Metabolic syndrome in men study; KOBS, Kuopio obesity surgery study; DPS, Finnish diabetes prevention study.

35

Table 5. Proportions of all measured erythrocyte membrane fatty acids (mol% of total fatty acids).

METSIM baseline

(n=1373) METSIM phase 2 (n=1031)

14:0 0.43 ±0.08 0.46 ±0.08 16:0 22 ±0.93 23 ±1.0 18:0 16 ±0.55 15 ±0.55 20:0 0.43 ±0.05 0.43 ±0.05 22:0 1.8 ±0.21 1.8 ±0.21 24:0 5.4 ±0.47 5.5 ±0.44 16:1n-7 0.41 ±0.15 0.45 ±0.14 18:1n-7 1.1 ±0.11 1.2 ±0.12 18:1n-9 12 ±0.75 13 ±0.89 20:1n-9+11 0.33 ±0.05 0.34 ±0.05 24:1n-9 5.7 ±0.56 5.9 ±0.54 18:3n-3 0.19 ±0.05 0.19 ±0.05 20:5n-3 1.5 ±0.59 1.5 ±0.55 22:5n-3 2.6 ±0.38 2.5 ±0.33 22:6n-3 6.2 ±1.1 6.0 ±1.1 18:2n-6 8.3 ±1.1 8.1 ±1.1 20:3n-6 1.5 ±0.28 1.4 ±0.26 20:4n-6 12 ±1.1 11 ±1.1 22:4n-6 1.9 ±0.44 1.8 ±0.43 22:5n-6 0.33 ±0.09 0.31 ±0.08 Values are mean ±SD in mol%. The data of two separate subpopulations of METSIM are shown (random sample at baseline and participants who participated in the 5-y follow-up study consecutively). METSIM, Metabolic syndrome in men study.

Table 6. Proportions of plasma PUFA (mol % of total fatty acids in each fraction) in the METSIM phase 2 study and serum phospholipids in KOBS.

Cholesteryl esters (n=845)

Triglycerides (n=845)

Phospholipids (n=845)

KOBS phospholipids (n=91)

18:3n-3 1.0 ±0.27 2.0 ±0.71 0.35 ±0.12 0.20 ±0.07

20:5n-3 2.5 ±1.2 0.83 ±0.63 2.4 ±1.1 1.6 ±0.64

22:5n-3 NA 0.83 ±0.29 1.4 ±0.22 1.2 ±0.16

22:6n-3 1.1 ±0.3 2.2 ±1.5 5.9 ±1.4 6.4 ±1.3

18:2n-6 48 ±5.2 14 ±3.4 19 ±2.8 17 ±2.4

18:3n-6 0.89 ±0.38 0.36 ±0.19 NA NA

20:3n-6 0.73 ±0.17 0.30 ±0.08 2.7 ±0.63 3.1 ±0.70

20:4n-6 6.7 ±1.8 1.3 ±0.45 9.2 ±1.9 11 ±2.3

22:4n-6 NA NA 0.31 ±0.08 0.25 ±0.06

22:5n-6 NA NA 0.18 ±0.07 0.19 ±0.06

n-3 4.7 ±1.4 5.8 ±2.5 10 ±2.4 9.3 ±1.8

n-6 57 ±4.5 16 ±3.6 31 ±2.5 32 ±1.6 Values are means (mol%) ±SD of fatty acids measured in plasma and serum. KOBS, Kuopio obesity surgery study; METSIM phase 2, 5-y follow-up study of the Metabolic syndrome in men study; NA, not available.

36

Table 7. Proportions of serum fatty acids (mol% of total fatty acids) in the DPS study population at baseline (n=387).

14:0 1.5 ±0.49

15:0 0.28 ±0.08*

16:0 24 ±2.0

18:0 7.1 ±0.77*

16:1n-7 3.1 ±0.99*

18:1n-9 21 ±2.3*

18:2n-6 25 ±3.9*

20:3n-6 1.4 ±0.29*

20:4n-6 5.0 ±1.1*

18:3n-3 1.0 ±0.27*

20:5n-3 1.6 ±1.2*

22:5n-3 0.59 ±0.16*

22.6n-3 2.9 ±1.1*

Trans 16:1n-7 0.19 ±0.07*

Trans 18:1n-9 0.59 ±0.27*

Values are proportions or means ±SD. * P<0.05 for change during the intervention period of DPS. P-values were calculated in the whole cohort (n=407) using repeated linear mixed model. DPS, Finnish diabetes prevention study.

37

Table 8. Partial correlations of fatty acid intakes with proportions of total serum fatty acids at baseline of DPS (n=387).

Fatty acid intake (g/100g of total fat)

Serum fatty acids (mol%)

SFA MUFA 18:3n-3 n-3

without 18:3n-3

18:2n-6 20:4n-6 PUFA Trans

14:0 0.224*** 15:0 0.287*** 16:0 0.144** 18:0 0.051 SFA 0.198*** 16:1n-7 0.219*** -0.189*** 18:1n-9 0.073 MUFA -0.008 18:3n-3 0.203*** 0.199***

20:5n-3 0.143* 0.006

22:5n-3 0.132* -0.034

22:6n-3 0.193*** 0.024

n-3 0.072 0.196*** 0.048

18:2n-6 0.201*** 0.203***

20:3n-6 -0.096

20:4n-6 0.115* 0.068

n-6 0.200*** -0.113* 0.205***

PUFA 0.163** 0.005 0.198*** 0.002 0.210*** Trans 16:1n-7 0.061

Trans 18:1n-9 0.124*

Values are partial correlation coefficients adjusted for age, sex, intervention group, study center, smoking, alcohol intake, waist circumference and physical activity at leisure time. Fatty acid intakes (g/100g of total fat) have been calculated based on a 3-day food diary at baseline. *P<0.05, **P<0.01 and ***P<0.001. DPS, Finnish diabetes prevention study. MUFA, Monounsaturated fatty acids; PUFA, Polyunsaturated fatty acids; SFA, Saturated fatty acids.

5.2 DIETARY FAT AND ERYTHROCYTE MEMBRANE FATTY ACIDS (STUDY I)

Correlations between EM fatty acids and food groups assessed by the FFQ are reported in Table 9. PUFA from weighted fish and fish oil supplement intake was associated with PUFA in EM. The highest positive correlation was observed with EPA. All EM n-6 PUFA were inversely associated with fish and fish oil supplement intake. PUFA from meat products were associated with higher ARA, adrenic acid (22:4n-6) and osbond acid (22:5n-

38

6) proportions, whereas the correlations with EPA and DHA proportions were inverse. Intake of PUFA from spreads and cooking fat was associated with lower proportions of ARA and 22:5n-6 but with higher proportions of ALA and LA. In addition, use of butter in cooking or on bread was related to lower proportions (mol%) of ALA and LA (butter users 0.16 ± 0.04 vs. non-users 0.19 ± 0.05, P < 0.001 and 7.77 ± 1.02 vs. 8.12 ± 1.11, P = 0.001, respectively).

Table 9. Spearman correlations between erythrocyte membrane fatty acids and diet weighteda by fatty acid content in the METSIM phase 2 study.

SFA PUFA

Milk and

milk products

Spread and

cooking fat

Fish and fish oil supplements

Spread and cooking fat

Meat products

14:0 0.186*** 0.188*** - − 0.128*** - 16:0 - - - - - 18:0 - - − 0.077 - - 20:0 - - − 0.070 - - 22:0 0.132*** 0.086* − 0.182*** − 0.089* - 24:0 - - - - - 16:1n-7 - - - − 0.182*** - 18:1n-7 - - - - - 18:1n-9 - - - - - 20:1n-9+11 - - - 0.065 - 24:1n-9 − 0.159*** − 0.142*** - 0.143*** - 18:3n-3 - - - 0.229*** - 20:5n-3 − 0.091* - 0.415*** - − 0.136*** 22:5n-3 - - 0.141*** - - 22:6n-3 - - 0.340*** - − 0.127*** 18:2n-6 0.104* 0.081 − 0.121** 0.160*** - 20:3n-6 - - − 0.149*** - - 20:4n-6 - - − 0.272*** − 0.116** 0.160*** 22:4n-6 - - − 0.296*** - 0.128***

22:5n-6 0.099* 0.093* − 0.269*** − 0.180*** 0.132*** Spearman's correlation coefficients were calculated between each erythrocyte fatty acid proportion and saturated or polyunsaturated fatty acid intake by food group taking into account approximated portion sizes using the FFQ. Only significant associations are shown (P < 0.05). aThe eating frequencies in the FFQ were multiplied by either SFA or PUFA content of different food items. *P < 0.01, **P < 0.001 and ***P < 0.0001. FFQ, Food frequency questionnaire; METSIM, Metabolic syndrome in men study; PUFA, Polyunsaturated fatty acids; SFA, Saturated fatty acids.

The proportions of particular SFA and MUFA in EM were only weakly associated with diet (Table 9). Intake of dairy SFA was positively related to proportions of myristic acid (14:0) and behenic acid (22:0). The proportion of nervonic acid (24:1n-9) was associated inversely with SFA intake from milk and spreads and cooking fat, whereas there was a positive correlation between the proportion of nervonic acid and PUFA from spreads and cooking fat. The proportion of palmitoleic acid (16:1n-7) was also inversely related to PUFA from spreads and cooking fat. In agreement, the use of butter in cooking or on bread was associated with higher proportions (mol%) of myristic and behenic acids in erythrocytes

39

(0.50 ± 0.08 vs. 0.45 ± 0.08, P < 0.001 and 1.82 ± 0.21 vs. 1.78 ± 0.21, P = 0.021) as well as with higher palmitoleic acid proportion (0.50 ± 0.15 vs. 0.44 ± 0.14, P < 0.001). Consistently, the proportion of nervonic acid was lower in butter users than non-users (5.71 ± 0.61 vs. 5.96 ± 0.53, P < 0.001). Consumption of fast food was infrequent (0.14 times a day) and associations remained very low.

Next we analyzed the associations using linear regression models adjusted for age, BMI, smoking, statin medication and alcohol. The adjusted results were virtually similar to the unadjusted statistics. Furthermore, adding both variables of spreads and cooking fat intake weighted by SFA and PUFA in the same regression model they were both still associated, in opposite directions, with the proportions of myristic (B = 0.018 and B = −0.011, both P < 0.001) and behenic acids (B = 0.021, P = 0.011 and B = −0.024, P = 0.003). Adding the SFA and PUFA weighted spreads and cooking fat intake variables to the same regression model, both the SFA and PUFA variables were associated with the proportion of nervonic acid in opposite directions (B = −0.093 and B = 0.098, both P < 0.001). The opposite effect of dairy SFA and PUFA from spreads and cooking fat on the proportion of nervonic acid is further illustrated in Figure 5.

Figure 5. Values are mean (mol%) and standard errors of the proportion of nervonic acid (24:1n-9) in erythrocyte membranes in relation to diet (n = 904) in the Metabolic syndrome in men (METSIM) phase 2 study. Please note the differing order of categories on the horizontal axes. *Dairy fat consumption is the sum of saturated fatty acid weighted consumption of spread and cooking fat and milk products divided into tertiles. PUFA, Polyunsaturated fatty acids.

40

5.3 INTERACTIONS BETWEEN MARINE FATTY ACID INTAKE AND A FADS1 VARIANT FOR CIRCULATING FATTY ACIDS (STUDY IV)

The minor alleles (C) of FADS1 intron-variant (rs174550) were strongly and additively associated with lower estimated activities of D5D and D6D in plasma and EM (Table 10). Substrates of D6D, ALA and LA, were correspondingly higher in EM, CE and PL but not in TG, and proportion of 18:3n-6, product of D6D, was lower among carriers of the minor allele. Substrate of D5D, 20:3n-6 was higher only in EM. Other PUFA with longer chains, i.e. products of the D5D enzyme, ARA and EPA and their elongation products, DPA and 22:4n-6, were lower in the carriers of minor allele. Levels of fatty acids that are further down in the metabolic pathway, i.e. DHA and 22:5n-6, were not associated with the genotype.

We examined continuous interactions between the genotype and marine PUFA intake for the outcomes of individual circulating fatty acids (Table 10). The same estimate of marine PUFA intake was used as in Study I. We found nominally significant interactions between the genotype and dietary marine fatty acids for EPA in CE (P-interaction=0.035) and EM (P-interaction=0.032) and for DPA in PL (P-interaction=0.007). There was also a non-significant trend of the interaction for EPA in PL (P-interaction=0.062). The interaction coefficients in linear regression models were positive in EM and all fractions of plasma for proportions of EPA and DPA. None of the potential interactions reached statistical significance level after correction for multiple comparisons. We plotted the associations between marine PUFA intake and proportions of EPA and DPA according to each of the allele pairs (Figure 6). This illustration revealed that the associations between the intake and proportions of EPA and DPA seemed to be less pronounced in those men who were homozygous for the major allele (T/T) compared with carriers of the minor allele.

As a sensitivity analysis, we excluded the participants who used fish oil supplements. There was a statistically significant interaction between marine PUFA intake and rs174550 for EM-EPA (P-interaction = 0.0026, n = 732). There was also a nominally significant interaction for D6D in TG (P-interaction=0.022, n=649) and a non-significant trend for interaction for DPA in PL (P-interaction=0.054, n=649). Otherwise the gene-diet interactions remained non-significant.

We further adjusted the models for PUFA intake from cooking oil and spreads, which are major sources of ALA and LA. However, this adjustment had little effect on the observed interactions (data not shown). Nonetheless, the gene-diet interactions between marine PUFA intake (including supplements) and rs174550 for EPA in TG (P-interaction=0.035) and for D5D in CE (P-interaction=0.048) and for D6D in CE and TG (P-interaction=0.047 and P-interaction=0.043, respectively) became nominally significant.

To investigate potential mechanisms underlying the observed interactions between FADS1 genotype and dietary fatty acids for circulating fatty acids we analyzed gene expression levels of FADS1 in the liver samples in KOBS. This analysis showed that the minor allele (C) of rs174547 is strongly associated with reduced hepatic mRNA expression of FADS1 (Figure 7). We hypothesized that high intake of EPA and DHA could decrease mRNA expression of FADS1 gene in the liver and consequently decrease D5D activity, resulting in decreased endogenous production of EPA and DPA. However, we could not observe any association between PL-EPA or PL-DHA and hepatic mRNA expression of FADS1, nor did we find any evidence that rs174547 would modify these associations (data not shown).

41

Ta

ble

10.

The

effe

ct o

f a

com

mon

FAD

S1

varian

t (r

s174

550)

on

fatt

y ac

id c

ompo

sitio

ns a

nd t

he int

erac

tion

betw

een

mar

ine

PUFA

int

ake

and

rs17

4550

fo

r ci

rcul

atin

g fa

tty

acid

com

posi

tions

in t

he M

ETSIM

pha

se 2

stu

dy.

Er

yth

rocy

tes

(n=

95

4)

Ch

ole

ster

yl e

ster

s (n

=8

45

) Tr

igly

ceri

des

(n

=8

45

) P

ho

sph

olip

ids

(n=

84

5)

SD

per

al

lele

P

-val

ue

P-

inte

ract

io

n

SD

per

al

lele

P

-val

ue

P-

inte

ract

ion

SD

per

al

lele

P

-val

ue

P-

inte

ract

io

n

SD

per

al

lele

P

-val

ue

P-

inte

ract

ion

18

:3n-

3 0.

33

4.4x

10-6

N

S

0.32

2.

0x10

-5

NS

0.12

a N

S

NS

0.32

a 1.

8x10

-5

NS

20:5

n-3

-0.2

8a

9.0x

10-5

0.

032

-0.5

5a

1.2x

10-1

3 0.

035

-0.5

8a

8.0x

10-1

5 N

S

-0.4

8a

2.7x

10-1

0 N

S

22:5

n-3

-0.3

1 7.

0x10

-5

NS

NA

NA

NA

-0.3

7 1.

2x10

-6

NS

-0.5

4 2.

9x10

-12

0.00

7

22:6

n-3

0.11

N

S

NS

-0.1

2a

NS

NS

-0.1

4a

NS

NS

-0.0

3 N

S

NS

18:2

n-6

0.50

4.

8x10

-13

NS

0.40

7.

0x10

-8

NS

0.03

N

S

NS

0.52

3.

0x10

-13

NS

18:3

n-6

NA

NA

NA

-0.6

9a

5.0x

10-2

2 N

S

-0.8

7a

1.2x

10-3

7 N

S

NA

NA

NA

20:3

n-6

0.69

4.

0x10

-23

NS

-0.1

0 N

S

NS

-0.2

3 4.

2x10

-3

NS

0.09

N

S

NS

20:4

n-6

-0.5

0 1.

0x10

-12

NS

-0.7

6a

8.6x

10-3

5 N

S

-0.6

7a

1.9x

10-2

0 N

S

-0.6

7 1.

4x10

-24

NS

22:4

n-6

-0.2

8 1.

8x10

-4

NS

NA

NA

NA

NA

NA

NA

-0.2

3 3.

0x10

-3

NS

22:5

n-6

0.03

a N

S

NS

NA

NA

NA

NA

NA

NA

-0.0

9 N

S

NS

n-3

-0.0

6 N

S

NS

-0.3

4a

6.6x

10-6

N

S

-0.1

7a

3.1x

10-2

N

S

-0.2

1 6.

9x10

-3

NS

n-6

0.04

N

S

NS

0.12

N

S

NS

-0.1

0 N

S

NS

0.09

N

S

NS

D6D

(18

:3n-

6/18

:2n-

6)

NA

NA

NA

-0.6

9a

6.0x

10-2

2 N

S

-0.8

8a

9.5x

10-3

7 N

S

NA

NA

NA

D5D

(20

:4n-

6/20

:3n-

6)

-0.7

9 3.

8x10

-32

NS

-0.5

6a

6.0x

10-1

5 N

S

-0.4

6a

7.4x

10-9

N

S

-0.5

2a

2.2x

10-1

2 N

S

n-6/

n-3

0.04

N

S

NS

0.39

2.

8x10

-7

NS

0.16

5.

1x10

-2

NS

0.27

4.

5x10

-4

NS

The

asso

ciat

ions

wer

e ca

lcul

ated

usi

ng l

inea

r re

gres

sion

adj

uste

d fo

r ag

e, s

mok

ing,

BM

I, s

tatin

med

icat

ion,

alc

ohol

con

sum

ptio

n an

d in

clud

ed g

enot

ype

(rs1

7455

0),

mar

ine

PUFA

inta

ke f

rom

fis

h an

d fis

h oi

l sup

plem

ents

and

a c

ontin

uous

inte

ract

ion

term

gen

otyp

e*m

arin

e PU

FA in

take

.

a The

pro

port

ion

of f

atty

aci

d or

inde

x w

as lo

g-tr

ansf

orm

ed b

efor

e th

e an

alys

is.

MET

SIM

, M

etab

olic

syn

drom

e in

men

stu

dy;

NS,

Non

-sig

nific

ant

(P >

0.05

); N

A,

Not

ava

ilabl

e; P

UFA

, Po

lyun

satu

rate

d fa

tty

acid

s.

42

Figure 6. Association of PUFA intake from fish and fish oil supplements with EPA in erythrocyte membranes (A), plasma cholesteryl esters (B) and plasma phospholipids (C) and DPA in plasma phospholipids (D) stratified by different genotypes of the FADS1 variant rs174550 in the Metabolic syndrome in men (METSIM) phase 2 study. Solid lines represent men who are homozygous for the major allele (T/T), dotted lines represent heterozygous (T/C) and sparsely dotted lines represent homozygous for the minor allele (C/C). Please note the logarithmic y-axis in panels A, B and C. P-values and unstandardized regression coefficients (β) for the illustrated gene-diet interactions were calculated using linear regression adjusted for age, smoking, BMI, statin medication and alcohol consumption. DPA, Docosapentaenoic acid; EPA, Eicosapentaenoic acid; PUFA, Polyunsaturated fatty acids.

A

B

C D

P-interaction=0.032 P-interaction=0.035

P-interaction=0.062 P-interaction=0.007

βT/T=0.09

βT/C=0.10

βC/C=0.15

βT/T=0.08

βT/C=0.12

βC/C=0.13

βT/T=0.08

βT/C=0.12

βC/C=0.13

βT/T=-0.01

βT/C=0.06

βC/C=0.09

43

Figure 7. Boxplots illustrating associations of the FADS1 variant rs174547 with hepatic FADS1 mRNA expression in the Kuopio obesity surgery study (KOBS) population (n=240). The P-value was calculated using the Kruskal-Wallis test. The FADS1 variant rs174547 is in complete linkage disequilibrium with rs174550.

5.4 ERYTHROCYTE MEMBRANE FATTY ACIDS AND LOW-GRADE INFLAMMATION (STUDY II)

In study II we examined the cross-sectional associations of EM fatty acids and circulating markers of low-grade inflammation. Unadjusted Pearson correlations between EM fatty acids and CRP, IL-1Ra and adiponectin were quite weak, but statistically significant even after correction for multiple testing (corresponding to |r|≥0.09) as shown in Table 11. The strongest positive correlations were observed between the proportions of palmitoleic acid (16:1n−7) and ARA and CRP (r=0.224 and r=0.172, respectively), and these two EM fatty acids were related to IL-1Ra as well (r=0.132 and r=0.160, respectively). Contrary to the other n−6 EM fatty acids, the proportion of LA was inversely correlated with both CRP and IL-1Ra (r=−0.134 and r=−0.172, respectively). The correlations between four individual n−3 fatty acids measured (ALA, EPA, DPA and DHA) and CRP were inverse, but only DPA and DHA were significantly correlated with CRP (r=−0.091 for both). Proportions of palmitic (16:0) and cis-vaccenic (18:1n−7) acids correlated positively with concentrations of adiponectin (r=0.156 and r=0.169, respectively).

Table 12 shows the confounder-adjusted partial correlations of CRP, IL-1Ra and adiponectin with those EM fatty acids, which were significantly associated with any of the three markers of inflammation after the adjustment for age, smoking, alcohol consumption, waist circumference, statin medication and exercise. The adjustment for these factors attenuated the associations and most of the weak correlations lost their statistical significance. None of the individual n−3 fatty

P=1.5x10-10

44

acids nor LA were significantly associated with inflammatory markers (|rpartial|<0.09) in the adjusted model.

The ratios and sums of multiple EM fatty acids in relation to markers of inflammation are shown in Table 12. The sum of n−3 fatty acids correlated inversely with concentrations of CRP in the adjusted model, but this correlation disappeared after further adjustment for the proportion of ARA (rpartial=−0.039, P=0.154). Because the unadjusted correlations showed anti-inflammatory associations of LA, we also calculated the sum of n−6 without LA. The correlations of the sum of n−6 PUFA without LA with CRP and IL1-Ra were slightly stronger than those of the sum of n−6 PUFA including LA (rpartial=0.139 vs. rpartial=0.099, P=0.014 and rpartial=0.115 vs. rpartial= 0.060, P<0.00083, respectively). Other calculated ratios and sums, including estimated SCD1 activity based on 18:1n−9/18:0-ratio and estimated D5D activity (20:4n−6/20:3n−6) were not significantly associated with inflammatory markers, nor were the sums of SFA, MUFA or PUFA.

Table 11. Pearson correlations between markers of inflammation and proportions of erythrocyte membrane fatty acids in the METSIM baseline study.

CRP IL-1Ra Adiponectin

14:0 -0.067 0.113 16:0 -0.056 0.156 18:0 0.090 0.081 -0.123 20:0 -0.102 -0.068 22:0 -0.062 24:0 0.061 -0.126 16:1n-7 0.224 0.132 18:1n-7 -0.075 0.169 18:1n-9 20:1n-9+11 -0.096 -0.079 24:1n-9 18:3n-3 -0.075 -0.068 20:5n-3 -0.083 0.073 22:5n-3 -0.091 0.092 22:6n-3 -0.091 -0.077 18:2n-6 -0.134 -0.172 20:3n-6 0.089 -0.114 20:4n-6 0.172 0.160 -0.066 22:4n-6 0.136 0.152 -0.063 22:5n-6 0.155 0.102 Nominally significant correlations are shown (P < 0.05). Correlations in bold were significant after correction for multiple testing (P < 0.00083 = 0.05/60). CRP, C-reactive protein; IL-1Ra, Interleukin-1 receptor antagonist; METSIM, Metabolic syndrome in men study.

45

5.5 PROSPECTIVE ASSOCIATIONS OF SERUM FATTY ACIDS WITH TYPE 2 DIABETES, INSULIN SECRETION AND INSULIN SENSITIVITY (STUDY III)

During the extended follow-up of DPS, the incidence of diabetes was 40.5 % (n=67 in the intervention and n=88 in the control group) among participants whose serum fatty acids were measured at baseline. Higher baseline serum proportions of ARA, EPA, DPA and DHA and higher estimated D5D predicted lower incidence of T2D in confounder-adjusted Cox regression models (Table 13). Comparing the lowest with the highest tertile of each fatty acid proportion at baseline, the unadjusted incidence rates per 100 person years at risk were 5.9 versus 3.0 for EPA, 6.5 versus 3.0 for DPA, 5.7 versus 3.0 for DHA and 6.3 versus 3.4 for D5D (Figure 8). Although there was no significant interaction between fatty acids and study groups, the associations with T2D tended to be stronger in the control than in the intervention group (data not shown).

The confounder-adjusted Cox regression analysis was repeated using the serum fatty acids measured at the first annual visit among participants (n = 322) who were non-diabetic at that time, and the associations were comparable with the results based on the baseline proportions of fatty acids (data not shown). However, ALA and LA predicted a higher risk of T2D in the adjusted model (HR per 1 SD 1.29, 95 % CI 1.07–1.56, P = 0.009 and HR = 1.45, 95 % CI 1.12–1.87, P = 0.005, respectively). The analysis was also repeated using the fatty acids measured at the

Table 12. Confounder-adjusted partial correlationsa between markers of inflammation and erythrocyte membrane fatty acids and estimated enzyme activities in the METSIM baseline study.

CRP IL-1Ra Adiponectin

16:1n-7 0.096

18:1n-7 -0.086 0.139

20:3n-6 0.067 -0.091

20:4n-6 0.126 0.103

22:4n-6 0.093 0.104

22:5n-6 0.104 0.055

SCD1 (16:1n-7/16:0) 0.102

D6D (20:3n-6/18:2n-6) 0.080 -0.103

Elongase (18:0/16:0) -0.107

Elongase (18:1n-7/16:1n-7) -0.124

Σn-3 -0.098 -0.072

Σn-6 0.099 0.060

Σn-6 (-LA) 0.139 0.115

Σn-6/Σn-3 0.103 0.069

Nominally significant partial correlations calculated using linear regression are shown. Coefficients in bold were significant after the adjustment for multiple testing (P < 0.00083 = 0.05/60). aAdjusted for age, smoking, alcohol consumption (g/wk), waist circumference, statin medication and exercise. -LA, without linoleic acid; CRP, C-reactive protein; IL-1Ra, Interleukin-1 receptor antagonist; METSIM, Metabolic syndrome in men study.

46

second annual visit, and the results were similar to the results at baseline and first year (data not shown).

During the intervention period, several fatty acids were associated with insulin sensitivity at the subsequent annual examination in the unadjusted model, but after the adjustments, most of the associations became nonsignificant (Table 14). Marine n-3 fatty acids and D5D were longitudinally associated with higher insulin sensitivity. D6D and SCD1 were also associated with higher insulin sensitivity in the fully adjusted model. Saturated fatty acids were associated with lower insulin sensitivity, though not in the fully adjusted model. There were significant inverse associations of D5D and EPA and a direct association of DGLA with insulin secretion measured at the succeeding annual visit, but otherwise the proportions of fatty acids were mainly unrelated to insulin secretion index (Table 15). In order to take better into account how insulin secretion is affected by insulin sensitivity, we analyzed the longitudinal associations of fatty acids with disposition index. A low proportion of palmitic acid (16:0) and a high proportion of LA was associated with disposition index of the subsequent year in the unadjusted repeated linear mixed model (proportional difference per 1 SD of fatty acid: β = 0.96, P = 0.001 and β = 1.03, P = 0.017, respectively), but otherwise individual fatty acids and desaturases were unassociated with disposition index. In models 1 and 2 (see the legend of Table 14), there were no significant associations with disposition index. In models further adjusted for disposition index of the previous year, MUFA and SCD1 activity were directly associated with DI30 (β = 1.05, P < 0.001 and β = 1.03, P = 0.019). In the fully adjusted model, however, LA was associated with lower disposition index (β = 0.97, P = 0.027).

47

Table 13. Type 2 diabetes incidence during an extended follow-up in association with baseline proportions of serum fatty acids and estimated desaturase activities in Cox regression model in DPS (n=383).

Unadjusted Adjusteda

HR CI 95% P-value

HR CI 95% P-value

(per 1SD) (per 1SD)

14:0 1.15 0.99 - 1.35 0.076 1.03 0.86 - 1.24 0.748 15:0 0.92 0.78 - 1.08 0.315 1.04 0.86 - 1.25 0.719 16:0 1.29 1.10 - 1.51 0.002 1.05 0.86 - 1.28 0.668 18:0 1.02 0.87 - 1.20 0.785 1.01 0.85 - 1.20 0.934 16:1n-7 1.12 0.96 - 1.32 0.157 0.95 0.78 - 1.15 0.575 18:1n-9 1.30 1.11 - 1.52 0.001 1.14 0.91 - 1.43 0.258 18:2n-6 0.91 0.77 - 1.07 0.235 1.20 0.97 - 1.47 0.087 20:3n-6 1.07 0.92 - 1.24 0.410 1.07 0.90 - 1.29 0.449 20:4n-6 0.78 0.66 - 0.92 0.003 0.80 0.67 - 0.95 0.013 18:3n-3 1.10 0.94 - 1.29 0.250 1.03 0.86 - 1.25 0.745 20:5n-3 0.74 0.62 - 0.88 4.67E-04 0.72 0.58 - 0.88 0.002 22:5n-3 0.72 0.61 - 0.86 2.00E-04 0.74 0.61 - 0.90 0.003 22.6n-3 0.73 0.61 - 0.88 9.38E-04 0.73 0.59 - 0.90 0.004 Trans 16:1n-7 0.90 0.77 - 1.05 0.190 1.00 0.84 - 1.19 0.975 Trans 18:1n-9 0.99 0.86 - 1.15 0.912 1.06 0.90 - 1.25 0.488

ΣSFA 1.26 1.08 - 1.48 0.004 1.04 0.86 - 1.26 0.667 ΣMUFA 1.28 1.09 - 1.50 0.002 1.07 0.86 - 1.33 0.553 ΣPUFA 0.76 0.65 - 0.89 6.53E-04 0.93 0.74 - 1.16 0.495 Σn-3 0.75 0.63 - 0.88 0.001 0.70 0.57 - 0.86 0.001 Σn-6 0.85 0.73 - 1.00 0.051 1.10 0.90 - 1.36 0.355 SCD1 activity 1.05 0.90 - 1.24 0.525 0.93 0.76 - 1.12 0.430 D6D activity 1.04 0.89 - 1.22 0.613 0.95 0.79 - 1.14 0.568

D5D activity 0.75 0.62 - 0.90 0.002 0.78 0.64 - 0.94 0.011 Cumulative type 2 diabetes incidence was 40.5% during the extended follow-up (median 11 years). Significant associations are bolded. aAdjustments were made for age, sex, study group, smoking, alcohol intake, waist circumference, physical activity at leisure time, fiber intake, carbohydrate intake, energy intake, serum triglyceride concentration and concentrations of plasma fasting and 2h glucose. The model was stratified by the five study centers. D5D activity, 20:4n-6/20:3n-6; D6D activity, 18:3n-6/18:2n-6; DPS, Finnish diabetes prevention study; MUFA, Monounsaturated fatty acids; PUFA, Polyunsaturated fatty acids; SCD1 activity (16:1n-7/16:0); SFA, Saturated fatty acids.

48

Figure 8. Kaplan-Meier curves demonstrating the probability of remaining free of diabetes according to tertile of serum EPA, DPA, DHA and Δ5 desaturase activity (20:4n-6/20:3n-6) in the Finnish diabetes prevention study (DPS). DHA, Docosahexaenoic acid; DPA, Docosapentaenoic acid; EPA, Eicosapentaenoic acid.

49

Table 14. Longitudinal associations of proportions of total serum fatty acids and insulin sensitivity index during the intervention period.

Unadjusted Model 1a Model 2b Model 3c

(n=376, observations = 1084) (n=329,

observations=696)

βd P-value

β P-value

β P-value

β P-value (per

1SD) (per 1SD)

(per 1SD)

(per 1SD)

14:0 0.96 0.003 0.96 0.003 0.98 0.232 1.00 0.827 15:0 0.99 0.670 0.99 0.245 0.99 0.235 1.00 0.699 16:0 0.95 3.11E-04 0.96 0.007 0.98 0.310 1.02 0.203 18:0 0.98 0.076 0.97 0.044 0.97 0.012 0.99 0.249 16:1n-7 0.96 0.018 0.99 0.645 1.02 0.252 1.02 0.054 18:1n-9 0.96 0.003 0.97 0.013 0.99 0.615 1.01 0.735 18:2n-6 1.05 0.002 1.03 0.040 1.01 0.750 0.97 0.018 20:3n-6 0.96 0.003 0.97 0.019 0.96 0.011 0.98 0.115 20:4n-6 1.02 0.122 1.03 0.079 1.00 0.814 1.01 0.445 18:3n-3 1.00 0.998 0.99 0.367 1.00 0.870 0.99 0.162 20:5n-3e 1.05 0.001 1.04 0.002 1.03 0.038 1.03 0.016 22:5n-3 1.03 0.024 1.02 0.176 1.02 0.218 1.03 0.024 22.6n-3 1.04 0.003 1.04 0.014 1.02 0.166 1.02 0.098 Trans 16:1n-7 1.02 0.177 1.01 0.430 1.01 0.39 0.99 0.257 Trans 18:1n-9 1.00 0.775 1.00 0.813 1.00 0.794 0.98 0.100 ΣSFA 0.94 2.53E-05 0.95 3.78E-04 0.97 0.040 1.01 0.479 ΣMUFA 0.95 0.001 0.97 0.033 1.00 0.991 1.02 0.219 ΣPUFA 1.06 1.33E-05 1.05 0.001 1.02 0.218 0.98 0.220 Σn-3e 1.05 0.001 1.04 0.006 1.03 0.063 1.02 0.051 Σn-6 1.04 0.002 1.03 0.023 1.01 0.770 0.97 0.023 SCD1 activity 0.97 0.069 1.00 0.841 1.03 0.145 1.02 0.05 D6D activity 0.99 0.367 0.99 0.663 1.02 0.28 1.02 0.036 D5D activity 1.05 3.45E-04 1.05 0.001 1.03 0.021 1.02 0.185 Longitudinal associations were analyzed using repeated linear mixed models with individual effect as a random factor (except for the Model 3). The proportions of fatty acids measured at the baseline and annual visits during the intervention period (1-6 years) were regressed on each subsequent years' insulin sensitivity index. Significant associations are bolded. aModel 1: Adjustments were made for age, sex, study group, visit, visit*group, smoking, alcohol intake, waist circumference and physical activity at leisure time. The five study centers were considered as a random factor. bModel 2: Model 1 + fiber intake, carbohydrate intake, energy intake and concentration of serum triglycerides. cModel 3: Model 2 + outcome variable (insulin sensitivity index) at the annual visits with a lag of one year. The random factors (study center and individuals) were redundant in Model 3 and thus autoregressive covariance structure by visits was applied. dβ was potentiated because the outcome variable (insulin sensitivity index) was logarithmized for the analyses and is expressed as the proportional difference per 1 SD of fatty acids. eLog-transformed. D5D activity, 20:4n-6/20:3n-6; D6D activity, 18:3n-6/18:2n-6; MUFA, Monounsaturated fatty acids; PUFA, Polyunsaturated fatty acids; SCD1 activity (16:1n-7/16:0); SFA, Saturated fatty acids.

50

Table 15. Longitudinal associations of proportions of total serum fatty acids and insulin secretion index during the intervention period (1-6 years).

Unadjusted Model 1a Model 2b Model 3c

(n=378 , observations=1106) (n=336, observations=721)

βd P-value

β P-value

β P-value

β P-value

(per 1SD) (per 1SD)

(per 1SD)

(per 1SD)

14:0 1.01 0.237 1.01 0.322 0.99 0.668 0.98 0.062 15:0 1.02 0.047 1.02 0.068 1.02 0.066 1.00 0.881 16:0 1.01 0.611 1.01 0.615 0.99 0.353 0.99 0.382 18:0 1.01 0.300 1.01 0.333 1.02 0.189 1.00 0.734 16:1n-7 1.01 0.384 1.00 0.874 0.98 0.232 1.00 0.708 18:1n-9 1.02 0.064 1.02 0.057 1.01 0.545 1.03 0.012 18:2n-6 0.99 0.415 0.99 0.507 1.01 0.493 1.01 0.436 20:3n-6 1.05 0.001 1.04 0.002 1.05 0.002 1.02 0.041 20:4n-6 0.98 0.083 0.98 0.158 0.99 0.621 0.99 0.209 18:3n-3 1.01 0.639 1.01 0.287 1.01 0.498 1.01 0.422 20:5n-3e 0.98 0.064 0.97 0.041 0.98 0.193 0.98 0.016 22:5n-3 0.99 0.207 0.99 0.358 0.99 0.487 0.98 0.056 22.6n-3 0.98 0.126 0.98 0.111 0.99 0.421 0.99 0.149 Trans 16:1n-7 1.00 0.955 0.99 0.395 0.99 0.359 1.00 0.925 Trans 18:1n-9 1.01 0.581 1.00 0.963 0.99 0.640 1.01 0.586 ΣSFA 1.01 0.244 1.01 0.277 1.00 0.926 0.99 0.339 ΣMUFA 1.03 0.051 1.02 0.070 1.01 0.707 1.03 0.037 ΣPUFA 0.98 0.056 0.98 0.081 1.00 0.892 0.99 0.428 Σn-3e 0.98 0.074 0.98 0.081 0.99 0.295 0.98 0.064 Σn-6 0.99 0.310 0.99 0.418 1.01 0.487 1.01 0.637 SCD1 activity 1.02 0.260 1.00 0.793 0.99 0.416 1.00 0.857 D6D activity 1.00 0.843 0.99 0.638 0.98 0.084 0.98 0.018 D5D activity 0.95 7.93E-05 0.95 3.78E-04 0.96 0.002 0.98 0.018 See the legend of Table 14.

51

6 Discussion

6.1 MAIN SOURCES OF FAT ESTIMATED BY A FFQ ARE RELATED TO ERYTHROCYTE MEMBRANE FATTY ACID COMPOSITION

6.1.1 Principal findings In agreement with many previous observational studies (50,72,163-166), we observed the strongest correlations between marine PUFA intake, estimated using the FFQ, and the proportions of marine n-3 in EM. This indicates that the FFQ and the calculated marine fatty acid intake estimate can be used to estimate total EPA, DPA and DHA intake. All other intake variables of describing major sources of fat were related to EM fatty acid composition, even though less strongly. As a novel finding in an observational setting, we reported that dairy fat associated with a higher proportion of behenic acid and a lower proportion of nervonic acid, whereas, conversely, vegetable oil based fats associated with a lower proportion of behenic acid and a higher proportion of nervonic acid.

6.1.2 Meat and fish intake in relation to n-6 fatty acids in erythrocyte membranes All erythrocyte n-6 fatty acids were inversely associated with PUFA intake from marine sources, which is probably due to well-known competition between n-3 and n-6 for metabolic pathways (63). This is also in line with an observational Japanese study that found inverse associations between EPA and DHA intake and the EM-ARA proportion (166). Furthermore, among pregnant Canadian women, there were decreased proportions of adrenic and osbond acids in erythrocyte ethanolamine phosphoglycerides related to higher dietary EPA and DHA intakes (167). We clearly demonstrated that intake of fish and fish oil supplements is stronger determinant of EM n-6 fatty acids than the intake of meat, which was positively, but weakly, associated with proportions of n-6 fatty acids. A significant positive association of ARA intake with membrane phospholipids is rarely reported in observational studies. However, evidence from intervention studies is similar to our results and shows that ARA intake increases proportions of ARA and adrenic acid (22:4n-6) in plasma PL and EM (66,168). Our results also showed a positive association of meat PUFA with osbond acid (22:5n-6).

6.1.3 Dairy fat and erythrocyte membrane fatty acids Positive correlations between diet and erythrocyte proportions of palmitoleic and myristic acids have been reported in previous studies (50,163). This is explained by the fact that both myristic acid (after elongation) and palmitic acid can be desaturated into palmitoleic acid by SCD1, and these fatty acids originate from the same sources, largely dairy fat. In the present study, positive correlations were found between SFA weighted spreads and cooking fat and proportions of myristic and behenic acids, and butter users had a higher proportion of palmitoleic acid. Similarly, SFA weighted milk and milk products intake was associated with higher myristic and behenic acid proportions. Both of these SFA intake variables correlated also negatively with nervonic acid proportion. Although these two different variables of SFA intake were composed of entirely different sets of questions in the FFQ, their correlations with EM fatty acids were strikingly similar and thus seem to represent the effect of dairy fats on EM fatty acid composition.

52

The association between dairy fat and behenic acid is unlikely to be explained by its intake as there is only little behenic acid in dairy fat and it is poorly absorbed (171). The more probable explanation is the increased production of behenic acid from shorter chain SFA present in dairy products. However, there was no association with lignoceric acid (24:0).

6.1.4 Vegetable oil based fats and erythrocyte membrane fatty acids We observed moderate to weak correlations between PUFA from spreads and cooking fat and EM fatty acid composition, providing some validation for this particular intake variable. Among Finns spreads and cooking oils account over half of the total PUFA intake, and over two thirds of ALA intake because in Finland low-erucic acid rapeseed oil is commonly used (24). Expectedly, ALA and LA, correlated positively with PUFA weighted spreads and cooking fat intake, whereas use of butter was associated with lower proportions of ALA and LA. Similar results have been reported in previous studies with LA (50,72,163) and ALA (163,165) in EM, even though in some less recent studies correlations with LA have been stronger (169,170). Spreads and cooking fat PUFA were also associated with lower ARA and osbond acid (22:5n-6), but importantly, not with lower EPA or DHA.

The influence of PUFA from cooking fat and spreads on EM SFA and MUFA was almost reciprocal to the effect of dairy fat, and there was actually a negative correlation with the proportion of palmitoleic acid. Even after both SFA and PUFA intakes from spreads and cooking fats were inputted to the same model, the opposite associations with myristic, behenic and nervonic acid persisted. It has to be noted that behenic and nervonic acids are specific for the sphingomyelin fraction of phospholipids and that this fraction has very little PUFA (172,173) and thus, should not be directly affected by PUFA intake. This and the fact that behenic acid was also negatively related to dietary marine PUFA, supports an indirect mechanism by which PUFA could suppress lipogenesis, elongases and SCD1, possibly by down regulating genes related to lipid metabolism (1).

Proportions of nervonic and lignoceric acids in rat liver sphingomyelin have been shown to be affected by their direct intake and to a lesser extent by their precursors (174). In human platelets, borage oil, which contains some nervonic acid, was shown to decrease the proportion of behenic and increase the proportion of nervonic acid (173). Thus, the association of a higher proportion of nervonic acid with PUFA from cooking fat and spreads in the present study is most likely explained by minute concentrations of nervonic acid and its immediate precursors 20:1n-9 and erucic acid (22:1n-9) in vegetable oils and partly hydrogenated fats used by the Finns. This is also supported by a new large study, which related higher proportions of long-chain MUFA, including nervonic acid, in plasma PL to the incidence of congestive heart failure and found several food products, such as poultry, that were associated with the proportion of nervonic acid (175). However, proportions of nervonic acid in plasma phospholipids and EM do not seem to be related (50). Interestingly, a lower proportion of nervonic acid has been weakly associated, in case-control setting, with a higher risk of neurological diseases (176,177).

We did not find any associations with oleic acid proportions, which is contrary to the high correlations reported among Italians (72). This is likely to be explained by lower olive oil consumption in Finland and demonstrates regional differences in EM and other lipid fractions (70).

53

6.2 A COMMON FADS1 VARIANT MAY MODIFY THE RELATIONSHIP BETWEEN MARINE FATTY ACIDS AND CIRCULATING FATTY ACIDS

6.2.1 Principal findings Our results suggested that the effect of marine PUFA intake on circulating marine n-3 fatty acids, EPA and its elongation product DPA, could be modulated by genetic variations in FADS1 gene. In men homozygous for the major allele of FADS1 intron-variant, rs174550, the relationship between marine PUFA intake and EM and plasma EPA and DPA was weaker than in the carriers of the minor allele. Although the interactions did not reach statistical significance after correction for multiple comparisons in the main analyses, we found a significant gene-diet interaction for EPA in EM after excluding the users of fish oil supplements, which strengthens our results. In a separate cohort of obesity surgery patients, the FADS1 variant rs174547 was associated with markedly reduced hepatic mRNA expression of FADS1. Rs174547 is in complete linkage disequilibrium with rs174550, offering suggestive evidence that the observed nominal interactions may be explained by differential effect of the genotypes on endogenous D5D activity in the liver.

6.2.2 Gene-diet interactions for circulating fatty acids There is emerging evidence for a gene-diet interaction between the intake of n-3 fatty acids and both FADS1 and FADS2 genotypes (77,80-87). A recent intervention study by Gillingham et al. found that intake of ALA interacts with FADS1 and FADS2 SNP for proportions of EPA and DPA in plasma (86). Large observational studies from the CHARGE consortium have also found similar interactions of ALA intake for plasma EPA, DPA and DHA (82,84).

Studies have also suggested gene-diet interactions for estimated D5D and D6D activities. Supplementation of EPA+DHA led to an increase in estimated D5D activity only in carriers of minor allele of upstream variant of FADS1 in a recent study by Al-Hilal et al. (81), whereas another EPA+DHA supplementation study by Cormier et al. found interactions only for D6D activity (87). In a large observational study by Zietemann et al., the effect of dietary n-6/n-3 ratio on D5D activity was modulated by a FADS1 variant (77).

In contrast with our novel finding suggesting interactions between marine PUFA intake and FADS1 variants for circulating levels of EPA and DPA, previous studies have found little evidence for such interactions between intake of fish or fish oil and gene variants in the FADS cluster. Yet in line with our study, a small EPA+DHA supplementation study in young men (n=12) found that there was a greater response in plasma and EM EPA in the carriers of minor allele of a FADS1 variant. There was no significant interactions, however, but statistical power was very limited (85). Furthermore, we showed that the interactions for circulating EPA and DPA were to the same direction in EM and plasma lipid fractions. This disagrees with a recent study suggesting that gene-diet interactions could differ by fraction (plasma vs. erythrocytes) (84).

6.2.3 Mechanism for the gene-diet interactions? The potential gene-diet interactions for proportions of EPA and DPA could be explained by changes in desaturation and elongation of their precursor ALA, which is found in substantial amounts in spreads and rapeseed oils commonly used in Finland. Previous studies suggest that the activities of D5D and D6D enzymes are modulated by the intake of n-3. Based on estimated product-to-precursor ratios, supplementation of EPA+DHA seems to decrease D6D activity and increase D5D activity (38,81,87). However, these results do not necessarily indicate that

54

expression of FADS1 or FADS2 are changed, but could, for instance, be related to decreased proportions of certain n-6 fatty acids due to substrate competition with n-3 fatty acids. Indeed, we found no significant associations between habitual intake of marine PUFA and D5D and D6D activities. It is known that human liver is a major site for expression of FADS1 and FADS2 (178). In a recent study, genetic variations in FADS1 were associated with decreased expression of FADS1 mRNA and protein in the human liver (42). We were able to confirm this finding by showing that minor alleles of a FADS1 variant associate strongly with reduced hepatic mRNA expression in the separate KOBS study cohort. This result is also in line with our finding that minor alleles of rs174550 associates with lower estimated D5D activity.

Unfortunately, we were unable to find a definite mechanism for the possible interactions observed in the METSIM study, as we found no indication in the morbidly obese KOBS patients that serum EPA or DHA, as proxies of EPA and DHA intake, were related to mRNA expression of FADS1. This is not a surprise due to the small number of participants (n=91), all of whom were on standardized VLCD before the bariatric surgery when the liver samples were obtained.

6.3 N-6 FATTY ACIDS BUT NOT LINOLEIC ACID IN ERYTHROCYTE MEMBRANES ASSOCIATE WITH LOW-GRADE INFLAMMATION

6.3.1 Principal findings We found positive associations between n−6 EM fatty acids, except for LA, and higher low-grade inflammation estimated by concentrations of CRP, IL-1Ra and adiponectin in a cross-sectional setting. The proportion of palmitoleic acid in EM was also associated with slightly elevated circulating concentrations of these inflammatory markers. Interestingly, the proportion of cis-vaccenic acid (18:1n-7) was positively and relatively strongly associated with adiponectin concentration. The sum of n−3 in EM was inversely associated with CRP, but this association disappeared after further adjustment for the proportion of ARA in EM. In general, the associations were rather weak, which is in line with earlier studies using biomarkers of dietary fat quality (98,106,108,109,117,120,121,126).

6.3.2 N-6 fatty acids and low-grade inflammation Certain n−6 PUFA could in theory aggravate low-grade inflammation because of their pro-inflammatory eicosanoid end-products mainly synthesized from ARA. Indeed, we found that the proportions of ARA, 22:4n−6 and 22:5n−6 correlated positively with concentrations of CRP and IL1-Ra, and the proportion of DGLA was associated with lower concentrations of adiponectin. Previous studies have only reported pro-inflammatory associations with DGLA and ARA/LA-ratio in EM (107,117), and in some studies using plasma fatty acids even anti-inflammatory associations with ARA have been detected (105,114,128). Discrepant findings across studies could be explained by several factors. Differences in diets between study populations, different adjustments and sample sizes in statistical models as well as differences in genetic background, average body weight and age could explain the inconsistent results. Furthermore, the associations could differ if fatty acids of plasma were used instead of EM, as is suggested by one study (106).

Even a low-dose supplementation of ARA leads to higher proportion of ARA in EM (66). Since a small placebo-controlled supplementation study by Kakutani et al. did not find any inflammatory effects of ARA (17), it is possible that the inflammatory associations are not explained by ARA itself. As stated earlier, we observed that meat consumption is associated

55

with the proportion of ARA in EM. Therefore, concentrations of CRP and IL-1Ra positively associated with ARA and other long-chain n−6 fatty acids could be due to high meat consumption, which is in line with some other studies reporting an association between meat consumption and elevated concentrations of CRP (179,180). Thus, the association may not be explained by fatty acid themselves, but rather higher animal product consumption leading to higher proportion of ARA or other factors unaccounted in this study, such as poorer diet (e.g. lower consumption of vegetables and whole grain products) or lower socioeconomic status.

Unlike the other n−6 fatty acids, the proportion of LA showed a trend towards lower inflammation in the unadjusted correlations with concentrations of CRP and IL-1Ra as also reported by earlier studies using EM (117) and serum fatty acids (98). These findings are important because LA is an essential fatty acid and the most common PUFA in the western diet. Dietary intervention studies also show, in line with our study, that LA does not increase inflammation (18) and could even have anti-inflammatory effects (181).

6.3.3 N-3 fatty acids and low-grade inflammation N−3 PUFA are often referred to as anti-inflammatory fatty acids. In this study we found a weak inverse association of total n−3 fatty acids with CRP, but no significant associations with IL-1Ra or adiponectin. Moreover, this inverse association seemed to be explained by lower proportion of ARA rather than by n−3 PUFA themselves, because the association disappeared after the adjustment for ARA. In some of the earlier studies using n-3 fatty acids measured in EM a trend to an anti-inflammatory direction is also observed as n-3 fatty acids have been associated with low CRP and IL-6 (108,116). However, one study reported that proportion of EPA in EM was associated with high CRP (106). The proportion of n−3 PUFA measured in serum of 1395 middle-aged men from Eastern Finland was associated with slightly lower concentrations of CRP (120). Other studies using proportions of plasma fatty acids have shown similar mixed results, some have shown similar inverse associations (105), some lack of associations (109) and one study even a positive association of EPA with CRP (98). The reasons for the mixed results is not entirely clear since the previous study have all measured at least CRP and some also various other markers of inflammation. The partly discrepant findings could be related to same factors as discussed above in relation to n-6 fatty acids. Thus, to summarize, anti-inflammatory associations of n−3 PUFA seem to be modest and possibly mediated by decreased proportions of ARA.

6.3.4 Estimated enzyme activities and low-grade inflammation Palmitoleic acid and SCD1 have recently received increasing attention. In the present study, both proportion of palmitoleic acid and SCD1 were positively associated with CRP. How palmitoleic acid or SCD1 activity would be involved in inflammation is not yet known. The strongest evidence that the proportion of palmitoleic acid and SCD1 is associated with low-grade inflammation comes from a longitudinal study by Petersson et al. (98), in which the proportions of palmitoleic and oleic acids as well as SCD1 in serum CE at the age of 50 were related to CRP measured 20 years later in a population based cohort of 767 Swedish men. More support for an inflammatory nature of palmitoleic acid comes from two later cross-sectional studies in which this fatty acid was measured in CE (109) and EM (121). There is no general agreement between the studies, because reports with lack of association (122) or even opposite, i.e. anti-inflammatory, association with fibrinogen (115) exist. It has even been proposed that palmitoleic acid acts as a beneficial “lipokine” with possible anti-inflammatory properties (182).

56

Another commonly used SCD1 index, 18:1n−9/18:0-ratio, was not associated with inflammation in our study, making it less likely that SCD1 or its activity by itself would be associated with inflammation.

We observed that the proportion of cis-vaccenic acid (18:1n-7), an elongation product of palmitoleic acid, was positively associated with concentrations of adiponectin, and this association was actually similar in strength (but opposite in direction) as that of waist circumference with adiponectin (data not shown). Associations of cis-vaccenic acid measured in EM with adipokines or inflammatory markers have not been reported earlier. Matsumori et al. (126) reported that cis-vaccenic acid in EM was decreased in metabolic syndrome, and Djousse et al. (183) reported that cis-vaccenic acid in EM was associated with decreased risk of coronary heart disease, whereas the association with palmitoleic acid was just the opposite. We found also that elongase activity (18:1n−7/16:1n−7) was associated with low CRP and another elongase activity (18:0/16:0) with low adiponectin. These elongase indices may reflect different elongase enzymes (184), and this probably explains the contradictory findings between the two indices. Furthermore, elongase activity could partly explain the opposing associations of palmitoleic and cis-vaccenic acids with low-grade inflammation, even though 18:1n−7/16:1n−7-ratio was not associated with adiponectin.

6.4 MARINE N-3 FATTY ACIDS AND Δ5 DESATURASE ACTIVITY PREDICT LOWER TYPE 2 DIABETES INCIDENCE AND HIGHER INSULIN SENSITIVITY

6.4.1 Principal findings In this prospective cohort study, based on the randomized DPS, we used objective biomarkers of dietary fat quality and showed that the proportions of marine n-3 fatty acids and D5D activity were associated with lower diabetes incidence during the extended follow-up. Furthermore, marine n-3 fatty acids and D5D were associated, though less consistently, with higher insulin sensitivity in longitudinal models. SFA, MUFA, trans-fatty acids, ALA or LA were not consistently associated with T2D risk or related traits. The novelty of our study was that we analyzed serum fatty acid composition and insulin sensitivity and secretion indices repeatedly at several annual visits, and not only once as in earlier studies.

6.4.2 Marine n-3 fatty acids and the risk of type 2 diabetes Earlier studies have revealed inconsistent results considering the circulating proportions of n-3 fatty acids and incidence of T2D (135-137,139). In a recent Finnish cohort, no association between n-3 fatty acids in EM and incident T2D was seen in men (140), whereas in another longer-term prospective Finnish study involving both men and women, the total serum proportion of marine n-3 fatty acids predicted lower T2D incidence (141). In a meta-analysis of 16 prospective cohort studies by Wu et al. (185), no significant association between EPA + DHA intake, including both fish and fish oil supplement intake, and T2D incidence was found. To our knowledge, we showed for the first time that the proportions of individual marine n-3 fatty acids, EPA, DPA and DHA, and total n-3 fatty acids predict lower T2D incidence in people who had a high risk of T2D, i.e., among participants who were middle-aged, overweight and had impaired glucose tolerance at baseline.

We found that the high proportions of EPA, DPA and DHA were longitudinally associated, though rather weakly, with higher insulin sensitivity, which suggests a mechanism which could explain why these same fatty acids were associated with lower T2D risk. However, n-3 fatty

57

acids in EM were not associated with insulin secretion or sensitivity in a recent population-based cohort study of Finnish men (140).

The proportions of serum n-3 fatty acids are largely modified by the diet (20). In the present study, the associations of dietary n-3 fatty acids with their proportions in serum were similar or slightly weaker than those reported in earlier studies (20). In the 1990s, when the DPS study was conducted, the use of n-3 supplements was rare in Finland, and the main source of marine n-3 fatty acids was fatty fish. In two large randomized placebo-controlled studies examining the effects of moderate EPA and DHA supplementation on cardiovascular morbidity in mainly diabetic participants, the supplementation did not have any effects on glycated hemoglobin or plasma glucose concentrations (8,9). Likewise, there is no evidence that supplementation of n-3 fatty acids improves insulin sensitivity based on several intervention studies (13). Thus, a more plausible explanation for our results is that fatty fish intake, rather than n-3 fatty acids themselves, may protect against T2D. This is in agreement with a recent meta-analysis of cohort studies examining fatty fish consumption in relation to T2D incidence (12). One explanation for the protective effect of fish or fatty fish might be that the frequent consumption of fish may result in lower consumption of meat, which is proposed to increase the risk of T2D probably by various mechanisms (10). Unfortunately, we had no detailed information about the meat intake for this assessment. Fatty fish also contains D-vitamin, which has been linked to a decreased risk of T2D in observational studies (186). Furthermore, very low intakes of DHA and EPA could be detrimental. Unlike cohort studies, intervention trials comparing population mean intake (placebo) to high intake (supplement) may not find the negative effect of very low intakes on the development of T2D. On the other hand, it has to be remembered that humans can synthesize EPA and DHA from ALA. 6.4.3 Other serum fatty acids and the risk of type 2 diabetes Our results related to the associations of ALA and LA with incident diabetes and insulin sensitivity and secretion were conflicting, but overall suggested no association. ALA and LA measured at the first follow-up visit were both associated with higher T2D incidence, contrary to previous studies (136,137,139). This could be at least in part explained by the dietary changes made by the participants during the first year of the study according to the dietary goals to reduce the intake of SFA and increase the intake of unsaturated fats, or by the adjustment set, since in some models the association of LA with insulin sensitivity and disposition index was inverse and in some others it was direct. Our results show no association between serum trans-fatty acids and T2D or related traits, but the intake of trans-fatty acids is and has been low in Finland. Previous studies have suggested that the proportions of SFA may be related to higher incidence of T2D, with the exception of pentadecanoic acid (15:0) (135,137,143). We could not confirm this, even though SFA tended to be associated with lower insulin sensitivity.

6.4.4 Desaturase enzymes and the risk of type 2 diabetes Fatty acid desaturase enzymes have lately received increased interest concerning their role on predicting the development of T2D. Both D5D and D6D enzymes are suggested to be involved in the development of T2D. D5D has been associated inversely and D6D directly with T2D incidence (147), and one Mendelian randomization study indicated that these associations could be causal (138). Polymorphisms in the SCD gene have also been associated with obesity and insulin sensitivity (44).

58

Our results showed that D5D was associated with lower T2D incidence and higher insulin sensitivity, and the fatty acid contributing to its estimated activity, DGLA, was longitudinally associated with lower insulin sensitivity. We also found that D5D activity was associated inversely with insulin secretion, whereas DGLA was associated positively with insulin secretion. We interpreted these results to be due to higher insulin sensitivity, i.e., higher insulin sensitivity associates with lower insulin secretion, in participants with high D5D activity. This was supported by null associations found with the disposition index, which takes insulin sensitivity into account. Likewise, previous prospective studies have shown that the proportion of DGLA associates with higher T2D risk (135-137), and D5D activity with lower T2D risk (135,138). D5D in erythrocytes was also found to be longitudinally associated with higher insulin sensitivity (140). The favorable associations of D5D could be explained by its role as a regulating enzyme in endogenous production of ARA, which is the main precursor of eicosanoids.

Our results do not agree with the findings from studies that have found estimated D6D or SCD1 activities to be positively associated with a risk of T2D (135,138,140,143), low insulin sensitivity and low disposition index (140). We also calculated SCD1 activity as the ratio of 18:1n-9 to 18:0, but this did not change our results (data not shown). The lack of associations of D6D and SCD1 with T2D risk, their weak positive associations with high insulin sensitivity and direct association of SCD1 with disposition index in our study might be explained by the fact that all participants in our study were overweight and had impaired glucose metabolism at the baseline, making reverse causation less likely. Moreover, in PL fractions, which have been used in previous studies (135,140), the proportion of 18:3n-6 is very small and D6D is usually calculated as a ratio of 20:3n-6 to 18:2n-6, which also describes elongase activity. It is also important to note that we calculated the desaturase activities in whole serum unlike in most of the previous studies. However, we also showed that D5D and D6D activities are associated strongly with a known intron variant of FADS1 gene, providing indirect validation for using these ratios. SCD1 activity should be primarily estimated in defined lipid fractions instead of mixed lipid fractions because variations in the amount of TG may lead to false associations (187). We took this into account by adjusting our models for the concentration of serum TG. Still, our results concerning SCD1 should be interpreted with caution, and no firm conclusions can be drawn regarding SCD1.

6.5 STRENGTHS AND LIMITATIONS

There are some limitations to this thesis that need to be considered. A major limitation of studies based on METSIM and KOBS is the cross-sectional setting. Especially considering the associations of EM fatty acids with low-grade inflammation, it is possible that the fatty acids levels and endogenous fatty acid metabolism are modified by inflammation and not the other way. Another limitation relating to this cross-sectional setting and low-grade inflammation, is the availability of only single measurements of inflammatory markers, whereas inflammation in actuality is a very dynamic condition. The strength of the METSIM study was the relatively large, regionally and ethnically homogenous and population based study population of middle-aged and elderly men. The lack of women in METSIM, even though limiting generalizability, makes the statistical analysis easier, because women have some differences in fatty acid metabolism compared with men (37).

59

There is also the obvious limitation of circulating biomarkers of dietary fat in this thesis, due to the fact that it is impossible to entirely separate the effects of diet and endogenous metabolism, therefore leaving room for various possible mechanisms behind the observed associations. On the other hand, the use of circulating biomarkers of dietary fat is more objective way to gather insight of dietary fat quality than the use of dietary questionnaires.

Due to the qualitative FFQ used in the METSIM study, the determination of exact fatty acids intakes was impossible. Instead we used a more practical approach, providing direct information on how specific food groups, which are major sources of fat, affect EM fatty acid composition. The FFQ was previously unvalidated, and we calculated the fatty acid intakes using approximated portion sizes, which are clear weaknesses of this study. Still, we showed correlations between the intake estimates and EM fatty acid composition, and especially fish and fish oil intake as well as vegetable oil intake were reflected in EM fatty acid composition as expected, providing some validation evidence. We also lacked information on the total energy intake and total fat intake. Therefore, the variables of fat intake could not be adjusted for them. Instead, we adjusted for common confounders, and the results were similar to the unadjusted statistics. Moreover, the use of EM and the FFQ together is suitable, because both aim to reflect diet within past months.

Unlike previous gene-diet interaction studies focusing only on one or two lipid fractions, we analyzed the gene-diet interactions applying altogether four different fractions of blood, i.e. EM and plasma CE, TG and PL, and our finding of rather consistent gene-diet interactions among all four fractions of circulating fatty acids strengthens the results. Selecting only one SNP as a marker of genetic variation suffices because the SNP in the FADS cluster strongly affecting the variation of tissue PUFA are in linkage disequilibrium (34). The SNP used in the gene-diet interaction study (rs174550) is also in linkage disequilibrium (r2≥0.80) in HapMap CEU population with several SNP used in the previous gene-diet interaction studies (e.g. rs1535, rs174546, rs174548, rs174556). The accuracy of our qualitative FFQ to capture total habitual intake of n-3 fatty acids may be limited, but on the other hand the main sources of EPA+DHA, fish and fish oil, are straightforward to enquire. Despite the relatively large study population (n=962), the power of our study to detect weak gene-diet interactions was probably limited, and for instance, the post-hoc power calculated for EM-EPA was only 21%. Furthermore, in the main analysis, none of the observed interactions reached statistical significance level adjusted for multiple comparisons. On the other hand, a significant interaction was found after exclusion of fish oil supplement users. Nevertheless, our results suggesting novel gene-diet interactions need to be confirmed in larger cohorts. Assuming similar effect size (0.5%) that was found for EM-EPA in the main analysis and significance level of α=0.05, a sample size of n=1600 is needed to detect the gene-diet interaction for EM-EPA with statistical power of 80%.

The strength of the longitudinal study based on the DPS was a highly compliant cohort of participants who were at a high risk of T2D at baseline, minimizing the chance of bias and reverse causation of the association between development of T2D and the proportions of fatty acids. To decrease the chance of reverse causation further, we adjusted the statistical models for baseline concentrations of plasma glucose, lagged insulin secretion, insulin sensitivity and disposition index. Despite the relatively small size of the study population in DPS, a considerable number of incident T2D cases occurred during the extended follow-up. The diagnosis of T2D was affirmed to the highest standard, i.e., based on the annually performed OGTT, and the final diagnosis was confirmed by a repeat OGTT. Importantly, data on serum

60

fatty acids and indices of insulin secretion and sensitivity were available at multiple time points, allowing us to construct longitudinal repeated measures models.

The data from lifestyle intervention and control groups were pooled for the study based on the randomized DPS, but we adjusted for the study group in our analyses. Still, the intervention setting and high-risk group in the DPS study may limit the generalizability of our results relating to the risk of T2D to the general population, especially because the intervention group made multiple lifestyle changes, which resulted in weight loss and changes in diet. For measuring insulin secretion and insulin sensitivity, we used OGTT-derived surrogate indices which are inferior to, but less invasive than the intravenous glucose tolerance test or hyperglycemic and euglycemic insulin clamp methods. It is of note that only a few studies have used total serum fatty acids, which could explain the different results in studies that used defined lipid fractions. Serum fatty acids consist largely of TG, PL, CE and free fatty acid fractions, which may have differing properties. Despite this, total serum fatty acids are related to dietary fat in similar manner as defined fractions (20). Although we used objective biomarkers adjusted for main lifestyle factors known to be involved in the development of T2D in participants who were already at a high risk of T2D, residual confounding cannot be ruled out. For instance, fish consumption is probably related to many lifestyle factors that cannot be fully accounted for.

61

7 Conclusions and future implications

Four distinct patterns between the main groups of dietary fat estimated by a qualitative FFQ and EM fatty acids were identified. The strongest and best known was the association of marine PUFA intake with higher proportions of marine n-3 PUFA and lower proportions of all n-6 PUFA. PUFA intake from meat was associated with higher proportions of long-chain n-6 and slightly lower n-3 PUFA in EM. Intake of PUFA from vegetable oils and spreads was associated, in a more complex manner, with higher proportions of ALA, LA and nervonic acid (24:1n-9), but also with lower proportions of some n-6 PUFA (e.g. ARA) and endogenously synthesized fatty acids (e.g. behenic acid). Fourth, dairy fat associated with higher proportions of myristic and behenic acids and lower levels of nervonic acid, which was reciprocal compared with PUFA from spreads and cooking fats. More research is needed on how and what dietary components affect very long-chain fatty acids typical for sphingomyelin and what is the impact of these fatty acids on health.

As a novel finding, we showed that FADS1 variants may modulate the relationship between marine PUFA intake and circulating marine n-3 fatty acids in men. In the carriers of minor alleles (C/C and C/T) of rs174550 the relationship between diet and circulating fatty acids seemed to be stronger than in the men homozygous for the major allele (T/T). Thus, the possibility of gene-diet interactions should be considered in future studies investigating the effects of dietary marine n-3 fatty acids on disease risk and when using these fatty acids as circulating biomarkers of dietary intake of these particular fatty acids. However, no evidence was found that EPA or DHA would affect or modulate hepatic mRNA expression of FADS1. Thus, there is a need for large-scale studies to observe an interaction between EPA+DHA intake and FADS1 genotype for hepatic FADS1 mRNA expression.

N-6 PUFA in EM, with the important exception of LA, which is the main dietary PUFA, were positively associated with concentrations of markers of low-grade inflammation. Another interesting finding was the novel, positive association of circulating vaccenic acid with adiponectin. The anti-inflammatory associations of n−3 fatty acids remained modest in this study. In serum, marine n-3 fatty acids and D5D activity predicted lower T2D incidence. These associations seemed to be explained by better insulin sensitivity related to marine n-3 fatty acids and D5D. One plausible interpretation is that the beneficial association of n-3 fatty acids with lower T2D risk could be due to higher fish consumption. In the light of the novel gene-diet interactions found in this thesis, in the future it would be interesting to study if genetic variations in FADS gene cluster modify the associations of dietary or circulating PUFA with low-grade inflammation and T2D.

In conclusion, dietary, endogenous and genetic factors as well as their complex interactions modulate circulating fatty acid composition, which has to be taken into account in the interpretation of studies that examine the role of dietary fatty acids in the development of chronic diseases.

62

63

8 References

(1) Sampath H, Ntambi JM. Polyunsaturated fatty acid regulation of genes of lipid metabolism. Annu Rev Nutr 2005;25:317-340.

(2) Gehrmann W, Elsner M, Lenzen S. Role of metabolically generated reactive oxygen species for lipotoxicity in pancreatic beta-cells. Diabetes Obes Metab 2010 Oct;12 Suppl 2:149-158.

(3) Schwab U, Lauritzen L, Tholstrup T, Haldorssoni T, Riserus U, Uusitupa M, et al. Effect of the amount and type of dietary fat on cardiometabolic risk factors and risk of developing type 2 diabetes, cardiovascular diseases, and cancer: a systematic review. Food Nutr Res 2014 Jul 10;58:10.3402/fnr.v58.25145. eCollection 2014.

(4) Hooper L, Martin N, Abdelhamid A, Davey Smith G. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev 2015 Jun 10;6:CD011737.

(5) Ascherio A. Trans fatty acids and blood lipids. Atheroscler Suppl 2006 May;7(2):25-27.

(6) Erkkila A, de Mello VD, Riserus U, Laaksonen DE. Dietary fatty acids and cardiovascular disease: an epidemiological approach. Prog Lipid Res 2008 May;47(3):172-187.

(7) Siscovick DS, Raghunathan TE, King I, Weinmann S, Wicklund KG, Albright J, et al. Dietary intake and cell membrane levels of long-chain n-3 polyunsaturated fatty acids and the risk of primary cardiac arrest. JAMA 1995 Nov 1;274(17):1363-1367.

(8) ORIGIN Trial Investigators, Bosch J, Gerstein HC, Dagenais GR, Diaz R, Dyal L, et al. N-3 Fatty Acids and Cardiovascular Outcomes in Patients with Dysglycemia. N Engl J Med 2012 Jul 26;367(4):309-318.

(9) Risk and Prevention Study Collaborative Group, Roncaglioni MC, Tombesi M, Avanzini F, Barlera S, Caimi V, et al. N-3 Fatty Acids in Patients with Multiple Cardiovascular Risk Factors. N Engl J Med 2013 May 9;368(19):1800-1808.

(10) Ley SH, Hamdy O, Mohan V, Hu FB. Prevention and management of type 2 diabetes: dietary components and nutritional strategies. Lancet 2014 Jun 7;383(9933):1999-2007.

(11) Vessby B, Uusitupa M, Hermansen K, Riccardi G, Rivellese AA, Tapsell LC, et al. Substituting dietary saturated for monounsaturated fat impairs insulin sensitivity in healthy men and women: The KANWU Study. Diabetologia 2001 Mar;44(3):312-319.

(12) Zhang M, Picard-Deland E, Marette A. Fish and Marine Omega-3 Polyunsaturated Fatty Acid Consumption and Incidence of Type 2 Diabetes: A Systematic Review and Meta-Analysis. Int J Endocrinol 2013;2013:501015.

64

(13) Akinkuolie AO, Ngwa JS, Meigs JB, Djousse L. Omega-3 polyunsaturated fatty acid and insulin sensitivity: a meta-analysis of randomized controlled trials. Clin Nutr 2011 Dec;30(6):702-707.

(14) Donath MY, Shoelson SE. Type 2 diabetes as an inflammatory disease. Nat Rev Immunol 2011 Feb;11(2):98-107.

(15) Vasto S, Carruba G, Lio D, Colonna-Romano G, Di Bona D, Candore G, et al. Inflammation, ageing and cancer. Mech Ageing Dev 2009 Jan-Feb;130(1-2):40-45.

(16) Libby P. Inflammation and cardiovascular disease mechanisms. Am J Clin Nutr 2006 Feb;83(2):456S-460S.

(17) Kakutani S, Ishikura Y, Tateishi N, Horikawa C, Tokuda H, Kontani M, et al. Supplementation of arachidonic acid-enriched oil increases arachidonic acid contents in plasma phospholipids, but does not increase their metabolites and clinical parameters in Japanese healthy elderly individuals: a randomized controlled study. Lipids Health Dis 2011 Dec 22;10:241-511X-10-241.

(18) Johnson GH, Fritsche K. Effect of dietary linoleic acid on markers of inflammation in healthy persons: a systematic review of randomized controlled trials. J Acad Nutr Diet 2012 Jul;112(7):1029-41, 1041.e1-15.

(19) Rangel-Huerta OD, Aguilera CM, Mesa MD, Gil A. Omega-3 long-chain polyunsaturated fatty acids supplementation on inflammatory biomakers: a systematic review of randomised clinical trials. Br J Nutr 2012 Jun;107 Suppl 2:S159-70.

(20) Hodson L, Skeaff CM, Fielding BA. Fatty acid composition of adipose tissue and blood in humans and its use as a biomarker of dietary intake. Prog Lipid Res 2008 Sep;47(5):348-380.

(21) Nelson D, Cox M. Chapter 21: Lipid Biosynthesis. Lehninger: Principles of Biochemistry. 5th ed. New York, NY: W.H. Freeman and Company; 2008. p. 805.

(22) Wu Z, Palmquist DL. Synthesis and biohydrogenation of fatty acids by ruminal microorganisms in vitro. J Dairy Sci 1991 Sep;74(9):3035-3046.

(23) Ran-Ressler RR, Devapatla S, Lawrence P, Brenna JT. Branched chain fatty acids are constituents of the normal healthy newborn gastrointestinal tract. Pediatr Res 2008 Dec;64(6):605-609.

(24) Helldán A, Raulio S, Kosola M, Tapanainen H, Ovaskainen M, Virtanen S. The national FINDIET 2012 survey. Helsinki: National institute for health and welfare; 2013.

(25) Pietinen P, Paturi M, Reinivuo H, Tapanainen H, Valsta LM. FINDIET 2007 Survey: energy and nutrient intakes. Public Health Nutr 2010 Jun;13(6A):920-924.

(26) Guyton AC, Hall JE. Chapter 65: Digestion and Absorption in the Gastrointestinal Tract. Textbook of Medical Physiology. 12th ed.: Saunders Elsevier; 2011. p. 789.

65

(27) National Institute for Health and Welfare, Nutrition Unit. Fineli. Finnish food composition database. Release 16. 2013; Available at: http://www.fineli.fi. Accessed 10/10, 2015.

(28) Nelson D, Cox M. Chapter 17: Fatty Acid Catabolism. Lehninger: Principles of Biochemistry. 5th ed. New York, NY: W.H. Freeman and Company; 2008. p. 647.

(29) Kunesova M, Phinney S, Hainer V, Tvrzicka E, Stich V, Parizkova J, et al. The responses of serum and adipose Fatty acids to a one-year weight reduction regimen in female obese monozygotic twins. Ann N Y Acad Sci 2002 Jun;967:311-323.

(30) Sassa T, Kihara A. Metabolism of very long-chain Fatty acids: genes and pathophysiology. Biomol Ther (Seoul) 2014 Feb;22(2):83-92.

(31) Wu JH, Lemaitre RN, Imamura F, King IB, Song X, Spiegelman D, et al. Fatty acids in the de novo lipogenesis pathway and risk of coronary heart disease: the Cardiovascular Health Study. Am J Clin Nutr 2011 Aug;94(2):431-438.

(32) Levin G, Duffin KL, Obukowicz MG, Hummert SL, Fujiwara H, Needleman P, et al. Differential metabolism of dihomo-gamma-linolenic acid and arachidonic acid by cyclo-oxygenase-1 and cyclo-oxygenase-2: implications for cellular synthesis of prostaglandin E1 and prostaglandin E2. Biochem J 2002 Jul 15;365(Pt 2):489-496.

(33) Calder PC. Polyunsaturated fatty acids and inflammatory processes: New twists in an old tale. Biochimie 2009 Jun;91(6):791-795.

(34) Schaeffer L, Gohlke H, Muller M, Heid IM, Palmer LJ, Kompauer I, et al. Common genetic variants of the FADS1 FADS2 gene cluster and their reconstructed haplotypes are associated with the fatty acid composition in phospholipids. Hum Mol Genet 2006 Jun 1;15(11):1745-1756.

(35) Burdge GC, Wootton SA. Conversion of alpha-linolenic acid to eicosapentaenoic, docosapentaenoic and docosahexaenoic acids in young women. Br J Nutr 2002 Oct;88(4):411-420.

(36) Burdge GC, Finnegan YE, Minihane AM, Williams CM, Wootton SA. Effect of altered dietary n-3 fatty acid intake upon plasma lipid fatty acid composition, conversion of [13C]alpha-linolenic acid to longer-chain fatty acids and partitioning towards beta-oxidation in older men. Br J Nutr 2003 Aug;90(2):311-321.

(37) Childs CE, Kew S, Finnegan YE, Minihane AM, Leigh-Firbank EC, Williams CM, et al. Increased dietary alpha-linolenic acid has sex-specific effects upon eicosapentaenoic acid status in humans: re-examination of data from a randomised, placebo-controlled, parallel study. Nutr J 2014 Dec 11;13(1):113-2891-13-113.

(38) Vessby B, Gustafsson IB, Tengblad S, Berglund L. Indices of fatty acid desaturase activity in healthy human subjects: effects of different types of dietary fat. Br J Nutr 2013 Sep 14;110(5):871-879.

66

(39) Brossard N, Croset M, Pachiaudi C, Riou JP, Tayot JL, Lagarde M. Retroconversion and metabolism of [13C]22:6n-3 in humans and rats after intake of a single dose of [13C]22:6n-3-triacylglycerols. Am J Clin Nutr 1996 Oct;64(4):577-586.

(40) Malerba G, Schaeffer L, Xumerle L, Klopp N, Trabetti E, Biscuola M, et al. SNPs of the FADS gene cluster are associated with polyunsaturated fatty acids in a cohort of patients with cardiovascular disease. Lipids 2008 Apr;43(4):289-299.

(41) Rzehak P, Heinrich J, Klopp N, Schaeffer L, Hoff S, Wolfram G, et al. Evidence for an association between genetic variants of the fatty acid desaturase 1 fatty acid desaturase 2 ( FADS1 FADS2) gene cluster and the fatty acid composition of erythrocyte membranes. Br J Nutr 2009 Jan;101(1):20-26.

(42) Wang L, Athinarayanan S, Jiang G, Chalasani N, Zhang M, Liu W. Fatty acid desaturase 1 gene polymorphisms control human hepatic lipid composition. Hepatology 2015 Jan;61(1):119-128.

(43) Zhu J, Sun Q, Zong G, Si Y, Liu C, Qi Q, et al. Interaction between a common variant in FADS1 and erythrocyte polyunsaturated fatty acids on lipid profile in Chinese Hans. J Lipid Res 2013 May;54(5):1477-1483.

(44) Warensjo E, Ingelsson E, Lundmark P, Lannfelt L, Syvanen AC, Vessby B, et al. Polymorphisms in the SCD1 gene: associations with body fat distribution and insulin sensitivity. Obesity (Silver Spring) 2007 Jul;15(7):1732-1740.

(45) Gong J, Campos H, McGarvey S, Wu Z, Goldberg R, Baylin A. Genetic variation in stearoyl-CoA desaturase 1 is associated with metabolic syndrome prevalence in Costa Rican adults. J Nutr 2011 Dec;141(12):2211-2218.

(46) Martin-Nunez GM, Cabrera-Mulero R, Rojo-Martinez G, Gomez-Zumaquero JM, Chaves FJ, de Marco G, et al. Polymorphisms in the SCD1 gene are associated with indices of stearoyl CoA desaturase activity and obesity: a prospective study. Mol Nutr Food Res 2013 Dec;57(12):2177-2184.

(47) Tukiainen T, Tynkkynen T, Makinen VP, Jylanki P, Kangas A, Hokkanen J, et al. A multi-metabolite analysis of serum by 1H NMR spectroscopy: early systemic signs of Alzheimer's disease. Biochem Biophys Res Commun 2008 Oct 24;375(3):356-361.

(48) Leidl K, Liebisch G, Richter D, Schmitz G. Mass spectrometric analysis of lipid species of human circulating blood cells. Biochim Biophys Acta 2008 Oct;1781(10):655-664.

(49) Agren JJ, Julkunen A, Penttila I. Rapid separation of serum lipids for fatty acid analysis by a single aminopropyl column. J Lipid Res 1992 Dec;33(12):1871-1876.

(50) Patel PS, Sharp SJ, Jansen E, Luben RN, Khaw KT, Wareham NJ, et al. Fatty acids measured in plasma and erythrocyte-membrane phospholipids and derived by food-frequency questionnaire and the risk of new-onset type 2 diabetes: a pilot study in the European

67

Prospective Investigation into Cancer and Nutrition (EPIC)-Norfolk cohort. Am J Clin Nutr 2010 Nov;92(5):1214-1222.

(51) Pittman JG, Martin DB. Fatty acid biosynthesis in human erythrocytes: evidence in mature erythrocytes for an incomplete long chain fatty acid synthesizing system. J Clin Invest 1966 Feb;45(2):165-172.

(52) Franco RS. The measurement and importance of red cell survival. Am J Hematol 2009 Feb;84(2):109-114.

(53) Farquhar JW, Ahrens EH,Jr. Effects of dietary fats on human erythrocyte fatty acid patterns. J Clin Invest 1963 May;42:675-685.

(54) Ahrens EH. The effect on human serum-lipids of a dietary fat, highly unsaturated, but poor in essential fatty acids. Lancet 1959;1(7064):115.

(55) Serra-Majem L, Nissensohn M, Overby NC, Fekete K. Dietary methods and biomarkers of omega 3 fatty acids: a systematic review. Br J Nutr 2012 Jun;107 Suppl 2:S64-76.

(56) Harris WS, Von Schacky C. The Omega-3 Index: a new risk factor for death from coronary heart disease? Prev Med 2004 Jul;39(1):212-220.

(57) Flock MR, Skulas-Ray AC, Harris WS, Etherton TD, Fleming JA, Kris-Etherton PM. Determinants of erythrocyte omega-3 fatty acid content in response to fish oil supplementation: a dose-response randomized controlled trial. J Am Heart Assoc 2013 Nov 19;2(6):e000513.

(58) Miller E, Kaur G, Larsen A, Loh SP, Linderborg K, Weisinger HS, et al. A short-term n-3 DPA supplementation study in humans. Eur J Nutr 2013 Apr;52(3):895-904.

(59) Skulas-Ray AC, Flock MR, Richter CK, Harris WS, West SG, Kris-Etherton PM. Red Blood Cell Docosapentaenoic Acid (DPA n-3) is Inversely Associated with Triglycerides and C-reactive Protein (CRP) in Healthy Adults and Dose-Dependently Increases Following n-3 Fatty Acid Supplementation. Nutrients 2015 Aug 4;7(8):6390-6404.

(60) Katan MB, Deslypere JP, van Birgelen AP, Penders M, Zegwaard M. Kinetics of the incorporation of dietary fatty acids into serum cholesteryl esters, erythrocyte membranes, and adipose tissue: an 18-month controlled study. J Lipid Res 1997 Oct;38(10):2012-2022.

(61) Harris WS, Pottala JV, Sands SA, Jones PG. Comparison of the effects of fish and fish-oil capsules on the n 3 fatty acid content of blood cells and plasma phospholipids. Am J Clin Nutr 2007 Dec;86(6):1621-1625.

(62) Harris WS, Varvel SA, Pottala JV, Warnick GR, McConnell JP. Comparative effects of an acute dose of fish oil on omega-3 fatty acid levels in red blood cells versus plasma: implications for clinical utility. J Clin Lipidol 2013 Sep-Oct;7(5):433-440.

68

(63) von Schacky C, Fischer S, Weber PC. Long-term effects of dietary marine omega-3 fatty acids upon plasma and cellular lipids, platelet function, and eicosanoid formation in humans. J Clin Invest 1985 Oct;76(4):1626-1631.

(64) Zuijdgeest-van Leeuwen SD, Dagnelie PC, Rietveld T, van den Berg JW, Wilson JH. Incorporation and washout of orally administered n-3 fatty acid ethyl esters in different plasma lipid fractions. Br J Nutr 1999 Dec;82(6):481-488.

(65) Hodson L, Eyles HC, McLachlan KJ, Bell ML, Green TJ, Skeaff CM. Plasma and erythrocyte fatty acids reflect intakes of saturated and n-6 PUFA within a similar time frame. J Nutr 2014 Jan;144(1):33-41.

(66) Hirota S, Adachi N, Gomyo T, Kawashima H, Kiso Y, Kawabata T. Low-dose arachidonic acid intake increases erythrocytes and plasma arachidonic acid in young women. Prostaglandins Leukot Essent Fatty Acids 2010 Aug;83(2):83-88.

(67) Smedman AE, Gustafsson IB, Berglund LG, Vessby BO. Pentadecanoic acid in serum as a marker for intake of milk fat: relations between intake of milk fat and metabolic risk factors. Am J Clin Nutr 1999 Jan;69(1):22-29.

(68) Wolk A, Furuheim M, Vessby B. Fatty acid composition of adipose tissue and serum lipids are valid biological markers of dairy fat intake in men. J Nutr 2001 Mar;131(3):828-833.

(69) Ratnayake WM. Concerns about the use of 15:0, 17:0, and trans-16:1n-7 as biomarkers of dairy fat intake in recent observational studies that suggest beneficial effects of dairy food on incidence of diabetes and stroke. Am J Clin Nutr 2015 May;101(5):1102-1103.

(70) Saadatian-Elahi M, Slimani N, Chajes V, Jenab M, Goudable J, Biessy C, et al. Plasma phospholipid fatty acid profiles and their association with food intakes: results from a cross-sectional study within the European Prospective Investigation into Cancer and Nutrition. Am J Clin Nutr 2009 Jan;89(1):331-346.

(71) Mozaffarian D, Cao H, King IB, Lemaitre RN, Song X, Siscovick DS, et al. Trans-palmitoleic acid, metabolic risk factors, and new-onset diabetes in U.S. adults: a cohort study. Ann Intern Med 2010 Dec 21;153(12):790-799.

(72) Fuhrman BJ, Barba M, Krogh V, Micheli A, Pala V, Lauria R, et al. Erythrocyte membrane phospholipid composition as a biomarker of dietary fat. Ann Nutr Metab 2006;50(2):95-102.

(73) Warensjo E, Riserus U, Vessby B. Fatty acid composition of serum lipids predicts the development of the metabolic syndrome in men. Diabetologia 2005 Oct;48(10):1999-2005.

(74) Sjogren P, Sierra-Johnson J, Gertow K, Rosell M, Vessby B, de Faire U, et al. Fatty acid desaturases in human adipose tissue: relationships between gene expression, desaturation indexes and insulin resistance. Diabetologia 2008 Feb;51(2):328-335.

69

(75) Peter A, Cegan A, Wagner S, Lehmann R, Stefan N, Konigsrainer A, et al. Hepatic lipid composition and stearoyl-coenzyme A desaturase 1 mRNA expression can be estimated from plasma VLDL fatty acid ratios. Clin Chem 2009 Dec;55(12):2113-2120.

(76) Chisaguano AM, Montes R, Perez-Berezo T, Castellote AI, Guerendiain M, Bustamante M, et al. Gene expression of desaturase (FADS1 and FADS2) and Elongase (ELOVL5) enzymes in peripheral blood: association with polyunsaturated fatty acid levels and atopic eczema in 4-year-old children. PLoS One 2013 Oct 22;8(10):e78245.

(77) Zietemann V, Kroger J, Enzenbach C, Jansen E, Fritsche A, Weikert C, et al. Genetic variation of the FADS1 FADS2 gene cluster and n-6 PUFA composition in erythrocyte membranes in the European Prospective Investigation into Cancer and Nutrition-Potsdam study. Br J Nutr 2010 Dec;104(12):1748-1759.

(78) Chong MF, Hodson L, Bickerton AS, Roberts R, Neville M, Karpe F, et al. Parallel activation of de novo lipogenesis and stearoyl-CoA desaturase activity after 3 d of high-carbohydrate feeding. Am J Clin Nutr 2008 Apr;87(4):817-823.

(79) Hudgins LC, Hellerstein M, Seidman C, Neese R, Diakun J, Hirsch J. Human fatty acid synthesis is stimulated by a eucaloric low fat, high carbohydrate diet. J Clin Invest 1996 May 1;97(9):2081-2091.

(80) Molto-Puigmarti C, Plat J, Mensink RP, Muller A, Jansen E, Zeegers MP, et al. FADS1 FADS2 gene variants modify the association between fish intake and the docosahexaenoic acid proportions in human milk. Am J Clin Nutr 2010 May;91(5):1368-1376.

(81) Al-Hilal M, Alsaleh A, Maniou Z, Lewis FJ, Hall WL, Sanders TA, et al. Genetic variation at the FADS1-FADS2 gene locus influences delta-5 desaturase activity and LC-PUFA proportions after fish oil supplement. J Lipid Res 2013 Feb;54(2):542-551.

(82) Lemaitre RN, Tanaka T, Tang W, Manichaikul A, Foy M, Kabagambe EK, et al. Genetic loci associated with plasma phospholipid n-3 fatty acids: a meta-analysis of genome-wide association studies from the CHARGE Consortium. PLoS Genet 2011 Jul;7(7):e1002193.

(83) Porenta SR, Ko YA, Gruber SB, Mukherjee B, Baylin A, Ren J, et al. Interaction of fatty acid genotype and diet on changes in colonic fatty acids in a Mediterranean diet intervention study. Cancer Prev Res (Phila) 2013 Nov;6(11):1212-1221.

(84) Smith CE, Follis JL, Nettleton JA, Foy M, Wu JH, Ma Y, et al. Dietary fatty acids modulate associations between genetic variants and circulating fatty acids in plasma and erythrocyte membranes: meta-analysis of 9 studies in the CHARGE consortium. Mol Nutr Food Res 2015 Jan 27.

(85) Roke K, Mutch DM. The role of FADS1/2 polymorphisms on cardiometabolic markers and fatty acid profiles in young adults consuming fish oil supplements. Nutrients 2014 Jun 16;6(6):2290-2304.

70

(86) Gillingham LG, Harding SV, Rideout TC, Yurkova N, Cunnane SC, Eck PK, et al. Dietary oils and FADS1-FADS2 genetic variants modulate [13C]alpha-linolenic acid metabolism and plasma fatty acid composition. Am J Clin Nutr 2013 Jan;97(1):195-207.

(87) Cormier H, Rudkowska I, Lemieux S, Couture P, Julien P, Vohl MC. Effects of FADS and ELOVL polymorphisms on indexes of desaturase and elongase activities: results from a pre-post fish oil supplementation. Genes Nutr 2014 Nov;9(6):437-014-0437-z. Epub 2014 Nov 4.

(88) Rudkowska I, Julien P, Couture P, Lemieux S, Tchernof A, Barbier O, et al. Cardiometabolic risk factors are influenced by Stearoyl-CoA Desaturase (SCD) -1 gene polymorphisms and n-3 polyunsaturated fatty acid supplementation. Mol Nutr Food Res 2014 May;58(5):1079-1086.

(89) Aronson D, Avizohar O, Levy Y, Bartha P, Jacob G, Markiewicz W. Factor analysis of risk variables associated with low-grade inflammation. Atherosclerosis 2008 Sep;200(1):206-212.

(90) Calder PC, Ahluwalia N, Brouns F, Buetler T, Clement K, Cunningham K, et al. Dietary factors and low-grade inflammation in relation to overweight and obesity. Br J Nutr 2011 Dec;106 Suppl 3:S5-78.

(91) C Reactive Protein Coronary Heart Disease Genetics Collaboration (CCGC), Wensley F, Gao P, Burgess S, Kaptoge S, Di Angelantonio E, et al. Association between C reactive protein and coronary heart disease: mendelian randomisation analysis based on individual participant data. BMJ 2011 Feb 15;342:d548.

(92) Interleukin-6 Receptor Mendelian Randomisation Analysis (IL6R MR) Consortium. The interleukin-6 receptor as a target for prevention of coronary heart disease: a mendelian randomisation analysis. Lancet 2012 Mar 31;379(9822):1214-1224.

(93) Niu W, Liu Y, Qi Y, Wu Z, Zhu D, Jin W. Association of interleukin-6 circulating levels with coronary artery disease: a meta-analysis implementing mendelian randomization approach. Int J Cardiol 2012 May 31;157(2):243-252.

(94) Turer AT, Scherer PE. Adiponectin: mechanistic insights and clinical implications. Diabetologia 2012 Sep;55(9):2319-2326.

(95) Li S, Shin HJ, Ding EL, van Dam RM. Adiponectin levels and risk of type 2 diabetes: a systematic review and meta-analysis. JAMA 2009 Jul 8;302(2):179-188.

(96) Pihlajamaki J, Kuulasmaa T, Kaminska D, Simonen M, Karja V, Gronlund S, et al. Serum interleukin 1 receptor antagonist as an independent marker of non-alcoholic steatohepatitis in humans. J Hepatol 2012 Mar;56(3):663-670.

(97) Herder C, Brunner EJ, Rathmann W, Strassburger K, Tabak AG, Schloot NC, et al. Elevated levels of the anti-inflammatory interleukin-1 receptor antagonist precede the onset of type 2 diabetes: the Whitehall II study. Diabetes Care 2009 Mar;32(3):421-423.

71

(98) Petersson H, Basu S, Cederholm T, Riserus U. Serum fatty acid composition and indices of stearoyl-CoA desaturase activity are associated with systemic inflammation: longitudinal analyses in middle-aged men. Br J Nutr 2008 Jun;99(6):1186-1189.

(99) Masson S, Marchioli R, Mozaffarian D, Bernasconi R, Milani V, Dragani L, et al. Plasma n-3 polyunsaturated fatty acids in chronic heart failure in the GISSI-Heart Failure Trial: relation with fish intake, circulating biomarkers, and mortality. Am Heart J 2013 Feb;165(2):208-15.e4.

(100) Flock MR, Skulas-Ray AC, Harris WS, Gaugler TL, Fleming JA, Kris-Etherton PM. Effects of supplemental long-chain omega-3 fatty acids and erythrocyte membrane fatty acid content on circulating inflammatory markers in a randomized controlled trial of healthy adults. Prostaglandins Leukot Essent Fatty Acids 2014 Oct;91(4):161-168.

(101) Forsythe CE, Phinney SD, Fernandez ML, Quann EE, Wood RJ, Bibus DM, et al. Comparison of low fat and low carbohydrate diets on circulating fatty acid composition and markers of inflammation. Lipids 2008 Jan;43(1):65-77.

(102) Madsen T, Skou HA, Hansen VE, Fog L, Christensen JH, Toft E, et al. C-reactive protein, dietary n-3 fatty acids, and the extent of coronary artery disease. Am J Cardiol 2001 Nov 15;88(10):1139-1142.

(103) Fernandez-Real JM, Broch M, Vendrell J, Ricart W. Insulin resistance, inflammation, and serum fatty acid composition. Diabetes Care 2003 May;26(5):1362-1368.

(104) Fernandez-Real JM, Vendrell J, Ricart W. Circulating adiponectin and plasma fatty acid profile. Clin Chem 2005 Mar;51(3):603-609.

(105) Ferrucci L, Cherubini A, Bandinelli S, Bartali B, Corsi A, Lauretani F, et al. Relationship of plasma polyunsaturated fatty acids to circulating inflammatory markers. J Clin Endocrinol Metab 2006 Feb;91(2):439-446.

(106) Sun Q, Ma J, Campos H, Rexrode KM, Albert CM, Mozaffarian D, et al. Blood concentrations of individual long-chain n-3 fatty acids and risk of nonfatal myocardial infarction. Am J Clin Nutr 2008 Jul;88(1):216-223.

(107) Martinelli N, Girelli D, Malerba G, Guarini P, Illig T, Trabetti E, et al. FADS genotypes and desaturase activity estimated by the ratio of arachidonic acid to linoleic acid are associated with inflammation and coronary artery disease. Am J Clin Nutr 2008 Oct;88(4):941-949.

(108) Farzaneh-Far R, Harris WS, Garg S, Na B, Whooley MA. Inverse association of erythrocyte n-3 fatty acid levels with inflammatory biomarkers in patients with stable coronary artery disease: The Heart and Soul Study. Atherosclerosis 2009 Aug;205(2):538-543.

(109) Petersson H, Lind L, Hulthe J, Elmgren A, Cederholm T, Riserus U. Relationships between serum fatty acid composition and multiple markers of inflammation and endothelial function in an elderly population. Atherosclerosis 2009 Mar;203(1):298-303.

72

(110) Clarke R, Shipley M, Armitage J, Collins R, Harris W. Plasma phospholipid fatty acids and CHD in older men: Whitehall study of London civil servants. Br J Nutr 2009 Jul;102(2):279-284.

(111) Rasmussen JG, Christensen JH, Schmidt EB. Adiponectin and marine n-3 fatty acids in patients referred for coronary angiography. Int J Cardiol 2009 Jun 26;135(2):248-250.

(112) Micallef MA, Munro IA, Garg ML. An inverse relationship between plasma n-3 fatty acids and C-reactive protein in healthy individuals. Eur J Clin Nutr 2009 Sep;63(9):1154-1156.

(113) Pottala JV, Garg S, Cohen BE, Whooley MA, Harris WS. Blood eicosapentaenoic and docosahexaenoic acids predict all-cause mortality in patients with stable coronary heart disease: the Heart and Soul study. Circ Cardiovasc Qual Outcomes 2010 Jul;3(4):406-412.

(114) Kalogeropoulos N, Panagiotakos DB, Pitsavos C, Chrysohoou C, Rousinou G, Toutouza M, et al. Unsaturated fatty acids are inversely associated and n-6/n-3 ratios are positively related to inflammation and coagulation markers in plasma of apparently healthy adults. Clin Chim Acta 2010 Apr 2;411(7-8):584-591.

(115) Mozaffarian D, Cao H, King IB, Lemaitre RN, Song X, Siscovick DS, et al. Circulating palmitoleic acid and risk of metabolic abnormalities and new-onset diabetes. Am J Clin Nutr 2010 Dec;92(6):1350-1358.

(116) Makhoul Z, Kristal AR, Gulati R, Luick B, Bersamin A, Boyer B, et al. Associations of very high intakes of eicosapentaenoic and docosahexaenoic acids with biomarkers of chronic disease risk among Yup'ik Eskimos. Am J Clin Nutr 2010 Mar;91(3):777-785.

(117) Enzenbach C, Kroger J, Zietemann V, Jansen EH, Fritsche A, Doring F, et al. Erythrocyte membrane phospholipid polyunsaturated fatty acids are related to plasma C-reactive protein and adiponectin in middle-aged German women and men. Eur J Nutr 2011 Dec;50(8):625-636.

(118) Min Y, Lowy C, Islam S, Khan FS, Swaminathan R. Relationship between red cell membrane fatty acids and adipokines in individuals with varying insulin sensitivity. Eur J Clin Nutr 2011 Jun;65(6):690-695.

(119) Oie E, Ueland T, Dahl CP, Bohov P, Berge C, Yndestad A, et al. Fatty acid composition in chronic heart failure: low circulating levels of eicosatetraenoic acid and high levels of vaccenic acid are associated with disease severity and mortality. J Intern Med 2011 Sep;270(3):263-272.

(120) Reinders I, Virtanen JK, Brouwer IA, Tuomainen TP. Association of serum n-3 polyunsaturated fatty acids with C-reactive protein in men. Eur J Clin Nutr 2012 Jun;66(6):736-741.

(121) Zong G, Ye X, Sun L, Li H, Yu Z, Hu FB, et al. Associations of erythrocyte palmitoleic acid with adipokines, inflammatory markers, and the metabolic syndrome in middle-aged and older Chinese. Am J Clin Nutr 2012 Nov;96(5):970-976.

73

(122) Poudel-Tandukar K, Sato M, Ejima Y, Nanri A, Matsushita Y, Imaizumi K, et al. Relationship of serum fatty acid composition and desaturase activity to C-reactive protein in Japanese men and women. Atherosclerosis 2012 Feb;220(2):520-524.

(123) Steffen BT, Steffen LM, Tracy R, Siscovick D, Hanson NQ, Nettleton J, et al. Obesity modifies the association between plasma phospholipid polyunsaturated fatty acids and markers of inflammation: the Multi-Ethnic Study of Atherosclerosis. Int J Obes (Lond) 2012 Jun;36(6):797-804.

(124) Kim OY, Lim HH, Lee MJ, Kim JY, Lee JH. Association of fatty acid composition in serum phospholipids with metabolic syndrome and arterial stiffness. Nutr Metab Cardiovasc Dis 2013 Apr;23(4):366-374.

(125) Grenon SM, Conte MS, Nosova E, Alley H, Chong K, Harris WS, et al. Association between n-3 polyunsaturated fatty acid content of red blood cells and inflammatory biomarkers in patients with peripheral artery disease. J Vasc Surg 2013 Nov;58(5):1283-1290.

(126) Matsumori R, Miyazaki T, Shimada K, Kume A, Kitamura Y, Oshida K, et al. High levels of very long-chain saturated fatty acid in erythrocytes correlates with atherogenic lipoprotein profiles in subjects with metabolic syndrome. Diabetes Res Clin Pract 2013 Jan;99(1):12-18.

(127) Perreault M, Roke K, Badawi A, Nielsen DE, Abdelmagid SA, El-Sohemy A, et al. Plasma levels of 14:0, 16:0, 16:1n-7, and 20:3n-6 are positively associated, but 18:0 and 18:2n-6 are inversely associated with markers of inflammation in young healthy adults. Lipids 2014 Mar;49(3):255-263.

(128) Kaska L, Mika A, Stepnowski P, Proczko M, Ratnicki-Sklucki K, Sledzinski T, et al. The relationship between specific Fatty acids of serum lipids and serum high sensitivity C- reactive protein levels in morbidly obese women. Cell Physiol Biochem 2014;34(4):1101-1108.

(129) Labonte ME, Dewailly E, Lucas M, Couture P, Lamarche B. Association of red blood cell n-3 polyunsaturated fatty acids with plasma inflammatory biomarkers among the Quebec Cree population. Eur J Clin Nutr 2014 Sep;68(9):1042-1047.

(130) Cho JS, Baek SH, Kim JY, Lee JH, Kim OY. Serum phospholipid monounsaturated fatty acid composition and Delta-9-desaturase activity are associated with early alteration of fasting glycemic status. Nutr Res 2014 Sep;34(9):733-741.

(131) Jacobs S, Schiller K, Jansen EH, Boeing H, Schulze MB, Kroger J. Evaluation of various biomarkers as potential mediators of the association between Delta5 desaturase, Delta6 desaturase, and stearoyl-CoA desaturase activity and incident type 2 diabetes in the European Prospective Investigation into Cancer and Nutrition-Potsdam Study. Am J Clin Nutr 2015 Jul;102(1):155-164.

(132) Malik VS, Chiuve SE, Campos H, Rimm EB, Mozaffarian D, Hu FB, et al. Circulating Very-Long-Chain Saturated Fatty Acids and Incident Coronary Heart Disease in US Men and Women. Circulation 2015 Jul 28;132(4):260-268.

74

(133) Zheng JS, Huang T, Yang J, Fu YQ, Li D. Marine N-3 polyunsaturated fatty acids are inversely associated with risk of type 2 diabetes in Asians: a systematic review and meta-analysis. PLoS One 2012;7(9):e44525.

(134) Deer J, Koska J, Ozias M, Reaven P. Dietary models of insulin resistance. Metabolism 2015 Feb;64(2):163-171.

(135) Krachler B, Norberg M, Eriksson JW, Hallmans G, Johansson I, Vessby B, et al. Fatty acid profile of the erythrocyte membrane preceding development of Type 2 diabetes mellitus. Nutr Metab Cardiovasc Dis 2008 Sep;18(7):503-510.

(136) Vessby B, Aro A, Skarfors E, Berglund L, Salminen I, Lithell H. The risk to develop NIDDM is related to the fatty acid composition of the serum cholesterol esters. Diabetes 1994 Nov;43(11):1353-1357.

(137) Wang L, Folsom AR, Zheng ZJ, Pankow JS, Eckfeldt JH, ARIC Study Investigators. Plasma fatty acid composition and incidence of diabetes in middle-aged adults: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Clin Nutr 2003 Jul;78(1):91-98.

(138) Kroger J, Zietemann V, Enzenbach C, Weikert C, Jansen EH, Doring F, et al. Erythrocyte membrane phospholipid fatty acids, desaturase activity, and dietary fatty acids in relation to risk of type 2 diabetes in the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study. Am J Clin Nutr 2011 Jan;93(1):127-142.

(139) Djousse L, Biggs ML, Lemaitre RN, King IB, Song X, Ix JH, et al. Plasma omega-3 fatty acids and incident diabetes in older adults. Am J Clin Nutr 2011 Aug;94(2):527-533.

(140) Mahendran Y, Agren J, Uusitupa M, Cederberg H, Vangipurapu J, Stancakova A, et al. Association of erythrocyte membrane fatty acids with changes in glycemia and risk of type 2 diabetes. Am J Clin Nutr 2014 Jan;99(1):79-85.

(141) Virtanen JK, Mursu J, Voutilainen S, Uusitupa M, Tuomainen TP. Serum omega-3 polyunsaturated fatty acids and risk of incident type 2 diabetes in men: the Kuopio Ischemic Heart Disease Risk Factor study. Diabetes Care 2014 Jan;37(1):189-196.

(142) Lankinen MA, Stancakova A, Uusitupa M, Agren J, Pihlajamaki J, Kuusisto J, et al. Plasma fatty acids as predictors of glycaemia and type 2 diabetes. Diabetologia 2015 Aug 16.

(143) Forouhi NG, Koulman A, Sharp SJ, Imamura F, Kroger J, Schulze MB, et al. Differences in the prospective association between individual plasma phospholipid saturated fatty acids and incident type 2 diabetes: the EPIC-InterAct case-cohort study. Lancet Diabetes Endocrinol 2014 Oct;2(10):810-818.

(144) Wang Q, Imamura F, Ma W, Wang M, Lemaitre RN, King IB, et al. Circulating and dietary trans fatty acids and incident type 2 diabetes in older adults: the Cardiovascular Health Study. Diabetes Care 2015 Jun;38(6):1099-1107.

75

(145) Santaren ID, Watkins SM, Liese AD, Wagenknecht LE, Rewers MJ, Haffner SM, et al. Serum pentadecanoic acid (15:0), a short-term marker of dairy food intake, is inversely associated with incident type 2 diabetes and its underlying disorders. Am J Clin Nutr 2014 Dec;100(6):1532-1540.

(146) Kroger J, Jacobs S, Jansen EH, Fritsche A, Boeing H, Schulze MB. Erythrocyte membrane fatty acid fluidity and risk of type 2 diabetes in the EPIC-Potsdam study. Diabetologia 2015 Feb;58(2):282-289.

(147) Kroger J, Schulze MB. Recent insights into the relation of Delta5 desaturase and Delta6 desaturase activity to the development of type 2 diabetes. Curr Opin Lipidol 2012 Feb;23(1):4-10.

(148) Sampath H, Ntambi JM. The role of stearoyl-CoA desaturase in obesity, insulin resistance, and inflammation. Ann N Y Acad Sci 2011 Dec;1243:47-53.

(149) Stancakova A, Javorsky M, Kuulasmaa T, Haffner SM, Kuusisto J, Laakso M. Changes in insulin sensitivity and insulin release in relation to glycemia and glucose tolerance in 6,414 Finnish men. Diabetes 2009 May;58(5):1212-1221.

(150) Voight BF, Kang HM, Ding J, Palmer CD, Sidore C, Chines PS, et al. The metabochip, a custom genotyping array for genetic studies of metabolic, cardiovascular, and anthropometric traits. PLoS Genet 2012;8(8):e1002793.

(151) Dupuis J, Langenberg C, Prokopenko I, Saxena R, Soranzo N, Jackson AU, et al. New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat Genet 2010 Feb;42(2):105-116.

(152) Lepage G, Roy CC. Direct transesterification of all classes of lipids in a one-step reaction. J Lipid Res 1986 Jan;27(1):114-120.

(153) Venalainen T, Schwab U, Agren J, de Mello V, Lindi V, Eloranta AM, et al. Cross-sectional associations of food consumption with plasma fatty acid composition and estimated desaturase activities in Finnish children. Lipids 2014 May;49(5):467-479.

(154) National Institute for Health and Welfare (THL), Helsinki, Finland. The FINRISKI 2007 study. 2008; Available at: http://urn.fi/URN:NBN:fi-fe201204193298. Accessed 02/20, 2013.

(155) Tuomilehto J, Lindstrom J, Eriksson JG, Valle TT, Hamalainen H, Ilanne-Parikka P, et al. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med 2001 May 3;344(18):1343-1350.

(156) Eriksson J, Lindstrom J, Valle T, Aunola S, Hamalainen H, Ilanne-Parikka P, et al. Prevention of Type II diabetes in subjects with impaired glucose tolerance: the Diabetes Prevention Study (DPS) in Finland. Study design and 1-year interim report on the feasibility of the lifestyle intervention programme. Diabetologia 1999 Jul;42(7):793-801.

76

(157) Lindstrom J, Peltonen M, Eriksson JG, Ilanne-Parikka P, Aunola S, Keinanen-Kiukaanniemi S, et al. Improved lifestyle and decreased diabetes risk over 13 years: long-term follow-up of the randomised Finnish Diabetes Prevention Study (DPS). Diabetologia 2013 Feb;56(2):284-293.

(158) Lindstrom J, Louheranta A, Mannelin M, Rastas M, Salminen V, Eriksson J, et al. The Finnish Diabetes Prevention Study (DPS): Lifestyle intervention and 3-year results on diet and physical activity. Diabetes Care 2003 Dec;26(12):3230-3236.

(159) Folch J, Lees M, Sloane Stanley GH. A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem 1957 May;226(1):497-509.

(160) de Mello VD, Lindstrom J, Eriksson J, Ilanne-Parikka P, Keinanen-Kiukaanniemi S, Sundvall J, et al. Insulin secretion and its determinants in the progression of impaired glucose tolerance to type 2 diabetes in impaired glucose-tolerant individuals: the Finnish Diabetes Prevention Study. Diabetes Care 2012 Feb;35(2):211-217.

(161) Matsuda M, DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care 1999 Sep;22(9):1462-1470.

(162) Simonen M, Mannisto V, Leppanen J, Kaminska D, Karja V, Venesmaa S, et al. Desmosterol in human nonalcoholic steatohepatitis. Hepatology 2013 Sep;58(3):976-982.

(163) Sun Q, Ma J, Campos H, Hankinson SE, Hu FB. Comparison between plasma and erythrocyte fatty acid content as biomarkers of fatty acid intake in US women. Am J Clin Nutr 2007 Jul;86(1):74-81.

(164) Harris WS, Pottala JV, Lacey SM, Vasan RS, Larson MG, Robins SJ. Clinical correlates and heritability of erythrocyte eicosapentaenoic and docosahexaenoic acid content in the Framingham Heart Study. Atherosclerosis 2012 Jun 7;225(2):425-431.

(165) Zhang B, Wang P, Chen CG, He QQ, Zhuo SY, Chen YM, et al. Validation of an FFQ to estimate the intake of fatty acids using erythrocyte membrane fatty acids and multiple 3d dietary records. Public Health Nutr 2010 Oct;13(10):1546-1552.

(166) Kawabata T, Hirota S, Hirayama T, Adachi N, Kaneko Y, Iwama N, et al. Associations between dietary n-6 and n-3 fatty acids and arachidonic acid compositions in plasma and erythrocytes in young and elderly Japanese volunteers. Lipids Health Dis 2011 Aug 13;10:138.

(167) Friesen RW, Innis SM. Linoleic acid is associated with lower long-chain n-6 and n-3 fatty acids in red blood cell lipids of Canadian pregnant women. Am J Clin Nutr 2010 Jan;91(1):23-31.

(168) Nelson GJ, Schmidt PC, Bartolini G, Kelley DS, Phinney SD, Kyle D, et al. The effect of dietary arachidonic acid on plasma lipoprotein distributions, apoproteins, blood lipid levels, and tissue fatty acid composition in humans. Lipids 1997 Apr;32(4):427-433.

77

(169) Sarkkinen ES, Agren JJ, Ahola I, Ovaskainen ML, Uusitupa MI. Fatty acid composition of serum cholesterol esters, and erythrocyte and platelet membranes as indicators of long-term adherence to fat-modified diets. Am J Clin Nutr 1994 Feb;59(2):364-370.

(170) Feunekes GI, Van Staveren WA, De Vries JH, Burema J, Hautvast JG. Relative and biomarker-based validity of a food-frequency questionnaire estimating intake of fats and cholesterol. Am J Clin Nutr 1993 Oct;58(4):489-496.

(171) Cater NB, Denke MA. Behenic acid is a cholesterol-raising saturated fatty acid in humans. Am J Clin Nutr 2001 Jan;73(1):41-44.

(172) Dougherty RM, Galli C, Ferro-Luzzi A, Iacono JM. Lipid and phospholipid fatty acid composition of plasma, red blood cells, and platelets and how they are affected by dietary lipids: a study of normal subjects from Italy, Finland, and the USA. Am J Clin Nutr 1987 Feb;45(2):443-455.

(173) Barre DE, Holub BJ. The effect of borage oil consumption on the composition of individual phospholipids in human platelets. Lipids 1992 May;27(5):315-320.

(174) Bettger WJ, Blackadar CB. Dietary very long chain fatty acids directly influence the ratio of tetracosenoic (24:1) to tetracosanoic (24:0) acids of sphingomyelin in rat liver. Lipids 1997 Jan;32(1):51-55.

(175) Imamura F, Lemaitre RN, King IB, Song X, Steffen LM, Folsom AR, et al. Long-chain monounsaturated Fatty acids and incidence of congestive heart failure in 2 prospective cohorts. Circulation 2013 Apr 9;127(14):1512-1521.

(176) Chen JR, Hsu SF, Hsu CD, Hwang LH, Yang SC. Dietary patterns and blood fatty acid composition in children with attention-deficit hyperactivity disorder in Taiwan. J Nutr Biochem 2004 Aug;15(8):467-472.

(177) Assies J, Lieverse R, Vreken P, Wanders RJ, Dingemans PM, Linszen DH. Significantly reduced docosahexaenoic and docosapentaenoic acid concentrations in erythrocyte membranes from schizophrenic patients compared with a carefully matched control group. Biol Psychiatry 2001 Mar 15;49(6):510-522.

(178) Cho HP, Nakamura M, Clarke SD. Cloning, expression, and fatty acid regulation of the human delta-5 desaturase. J Biol Chem 1999 Dec 24;274(52):37335-37339.

(179) Hickling S, Hung J, Knuiman M, Divitini M, Beilby J. Are the associations between diet and C-reactive protein independent of obesity? Prev Med 2008 Jul;47(1):71-76.

(180) van Woudenbergh GJ, Kuijsten A, Tigcheler B, Sijbrands EJ, van Rooij FJ, Hofman A, et al. Meat consumption and its association with C-reactive protein and incident type 2 diabetes: the Rotterdam Study. Diabetes Care 2012 Jul;35(7):1499-1505.

78

(181) Bjermo H, Iggman D, Kullberg J, Dahlman I, Johansson L, Persson L, et al. Effects of n-6 PUFAs compared with SFAs on liver fat, lipoproteins, and inflammation in abdominal obesity: a randomized controlled trial. Am J Clin Nutr 2012 May;95(5):1003-1012.

(182) Cao H, Gerhold K, Mayers JR, Wiest MM, Watkins SM, Hotamisligil GS. Identification of a lipokine, a lipid hormone linking adipose tissue to systemic metabolism. Cell 2008 Sep 19;134(6):933-944.

(183) Djousse L, Matthan NR, Lichtenstein AH, Gaziano JM. Red Blood Cell Membrane Concentration of cis-Palmitoleic and cis-Vaccenic Acids and Risk of Coronary Heart Disease. Am J Cardiol 2012 May 12;110:539-544.

(184) Green CD, Ozguden-Akkoc CG, Wang Y, Jump DB, Olson LK. Role of fatty acid elongases in determination of de novo synthesized monounsaturated fatty acid species. J Lipid Res 2010 Jul;51(7):1871-1877.

(185) Wu JH, Micha R, Imamura F, Pan A, Biggs ML, Ajaz O, et al. Omega-3 fatty acids and incident type 2 diabetes: a systematic review and meta-analysis. Br J Nutr 2012 Jun;107 Suppl 2:S214-27.

(186) Mitri J, Pittas AG. Vitamin D and diabetes. Endocrinol Metab Clin North Am 2014 Mar;43(1):205-232.

(187) Peter A, Cegan A, Wagner S, Elcnerova M, Konigsrainer A, Konigsrainer I, et al. Relationships between hepatic stearoyl-CoA desaturase-1 activity and mRNA expression with liver fat content in humans. Am J Physiol Endocrinol Metab 2011 Feb;300(2):E321-6.

DIS

SE

RT

AT

ION

S | M

AR

KU

S T

AK

KU

NE

N | C

IRC

UL

AT

ING

FA

TT

Y A

CID

S – A

SS

OC

IAT

ION

S W

ITH

DIE

T... | N

o 350

uef.fi



ISBN 978-952-61-2106-2ISSN 1798-5706



MARKUS TAKKUNEN

CIRCULATING FATTY ACIDS – ASSOCIATIONS WITH DIET, GENETIC VARIATIONS, LOW-GRADE

INFLAMMATION AND TYPE 2 DIABETES

Fatty acids in erythrocyte membranes and plasma are used as objective biomarkers

of dietary fat intake. These circulating fatty acids are particularly good in reflecting fish oil intake. In this thesis, circulating marine n-3 fatty acids were associated with lower

incidence of type 2 diabetes. N-6 fatty acids, except for linoleic acid, were associated

with higher low-grade inflammation. Novel evidence was found suggesting that gene-diet interactions modulate circulating fatty acids.

MARKUS TAKKUNEN

Documents

Dissertations in Health Sciences - UEFepublications.uef.fi/pub/urn_isbn_978-952-61-2107-9/urn...examined in another subpopulation (n=1373) of METSIM. n-6 fatty acids, except for All