Discovery of Novel Lipid Profiles in PCOS: Do Insulin and

C L I N I C A L R E S E A R C H A R T I C L E

Discovery of Novel Lipid Profiles in PCOS: Do Insulinand Androgen Oppositely Regulate BioactiveLipid Production?

Shengxian Li,1* Qianqian Chu,4* Jing Ma,1* Yun Sun,2 Tao Tao,1 Rong Huang,1

Yu Liao,1 Jiang Yue,1 Jun Zheng,1 Lihua Wang,1 Xinli Xue,4,7 Mingjiang Zhu,4

Xiaonan Kang,3 Huiyong Yin,4,8† and Wei Liu1†

1Department of Endocrinology and 2Shanghai Key Laboratory for Assisted Reproduction and ReproductiveGenetics, Center for Reproductive Medicine and 3Department of Biobank, Renji Hospital, School ofMedicine, Shanghai Jiao Tong University, Shanghai 200127, China; 4Key Laboratory of Food Safety Researchand 5Mass Spectrometry Research Center, Institute for Nutritional Sciences, Shanghai Institutes for BiologicalSciences, Chinese Academy of Sciences, Shanghai 200031, China; 6Key Laboratory of Food Safety RiskAssessment, Ministry of Health, Beijing 100021, China; 7University of the Chinese Academy of Sciences,Shanghai 200031, China; and 8School of Life Science and Technology, ShanghaiTech University, Shanghai200031, China

Context: Polycystic ovary syndrome (PCOS) is a complex syndrome showing clinical features of anendocrine/metabolic disorder, including hyperinsulinemia and hyperandrogenism. Polyunsaturatedfatty acids (PUFAs) and their derivatives, both tightly linked to PCOS and obesity, play importantroles in inflammation and reproduction.

Objective: This study aimed to investigate serum lipid profiles in newly diagnosed patients withPCOS using lipidomics and correlate these features with the hyperinsulinemia and hyper-androgenism associated with PCOS and obesity.

Design and Setting: Thirty-two newly diagnosed women with PCOS and 34 controls were dividedinto obese and lean subgroups. A PCOS rat model was used to validate results of the human studies.

Main Outcome Measures: Serum lipid profiles, including phospholipids, free fatty acids (FFAs), andbioactive lipids, were analyzed using gas chromatography–mass spectrometry (MS) and liquidchromatography–MS.

Results: Elevation in phosphatidylcholine and a concomitant decrease in lysophospholipid werefound in obese patients with PCOS vs lean controls. Obese patients with PCOS had decreased PUFAlevels and increased levels of long-chain saturated fatty acids vs lean controls. Serum bioactive lipidsdownstream of arachidonic acid were increased in obese controls, but reduced in both obeseand lean patients with PCOS vs their respective controls.

ISSN Print 0021-972X ISSN Online 1945-7197Printed in USACopyright © 2017 by the Endocrine SocietyThis article has been published under the terms of the Creative Commons AttributionLicense (CC BY; https://creativecommons.org/licenses/by/4.0/).Received 14 July 2016. Accepted 23 November 2016.First Published Online 28 November 2016

*These authors contributed equally to this study.†These authors are joint senior authors.Abbreviations: AA, arachidonic acid; AR, androgen receptor; BMI, bodymass index; CON,control group; COX, cyclooxygenase; DHA, docosahexaenoic acid; DHEAS, dehy-droepiandrosterone sulfate; DHT, dihydrotestosterone; EET, epoxyeicosatrienoic acid; FAI,free androgen index; FFA, free fatty acid; GC, gas chromatography; HDL, high-densitylipoprotein; HETE, hydroxyeicosatetraenoic acid; HF, high-fat-diet–fed group; HF + DHEA,high-fat-diet–fed and hyperandrogenism group; HODE, hydroxyoctadecadienoic acid;HOMA-IR, homeostatic model assessment–insulin resistance; LA, linoleic acid; LC, leancontrol; LDL, low-density lipoprotein; LOX, lipoxygenase; LP, lean with polycystic ovarysyndrome; LPC, lysophospholipid of phosphatidylcholine; MS, mass spectrometry; OB,obese; OP, obese with polycystic ovary syndrome; PC, phosphatidylcholine; PCOS,polycystic ovary syndrome; PG, prostaglandin; PI, phosphatidylinositol; PS, phosphati-dylserine; PUFA, polyunsaturated fatty acid; SHBG, sex hormone–binding globulin; TG,triglyceride; TT, total testosterone; TXB2, thromboxane B2; T2DM, type 2 diabetesmellitus.

810 press.endocrine.org/journal/jcem J Clin Endocrinol Metab, March 2017, 102(3):810–821 doi: 10.1210/jc.2016-2692

Downloaded from https://academic.oup.com/jcem/article-abstract/102/3/810/3061899by ShanghaiTech University useron 13 December 2017

https://creativecommons.org/licenses/by/4.0/

http://press.endocrine.org/journal/jcem

http://dx.doi.org/10.1210/jc.2016-2692

Conclusions: Patients with PCOS showed abnormal levels of phosphatidylcholine, FFAs, and PUFAmetabolites. Circulating insulin and androgens may have opposing effects on lipid profiles inpatientswith PCOS, particularly on the bioactive lipidmetabolites derived fromPUFAs. These clinicalobservations warrant further studies of the molecular mechanisms and clinical implications of PCOSand obesity. (J Clin Endocrinol Metab 102: 810–821, 2017)

Polycystic ovary syndrome (PCOS) is a complexmultisystem syndrome that occurs in postpubertal

women. According to diagnostic criteria of the NationalInstitute of Child Health and Human Development/National Institutes of Health, 4% to 10% of women ofreproductive age have PCOS (1–3). The major manifesta-tions of PCOS are dilute ovulation/anovulation, polycysticovaries, and hyperandrogenism. PCOS is often associatedwith obesity, insulin resistance, abnormal glucose tolerance,lipid metabolic disorders, and other metabolic abnormali-ties (4, 5). The prevalence of obesity inwomenwith PCOS isapproximately 50% to 80% (4). Obesity is associated withabnormal adipokine secretion, elevated serum free fattyacids (FFAs), and metabolic disorders associated with ab-normal steroid hormone function in adipose tissue, resultingin low-grade chronic inflammation, insulin resistance, andabnormal glucose tolerance (4, 6). Approximately 20% ofobese patients with PCOS have impaired glucose toleranceor type 2 diabetes mellitus (T2DM) (7); in addition, leanpatients with PCOS show increased prevalence of impairedglucose tolerance and T2DM compared with womenwithout PCOS (5). The prevalence of insulin resistance isbetween 44% and 70% (4, 8, 9), and it manifests as im-paired insulin-mediated suppression of lipolysis and lipidoxidation (10, 11), resulting in increased serum FFAs inobese women with PCOS that are matched for body massindex (BMI) (12).

Another core pathophysiologic feature of PCOS ishyperandrogenism. Androgen promotes lipolysis invisceral fat cells, which is an early and possibly primarymetabolic defect in PCOS (13, 14). However, how in-sulin and hyperandrogenism contribute to the adverselipid profile present in PCOS, which includes alteredconcentrations of phospholipids, FFAs, and bioactivelipid metabolites derived from polyunsaturated fattyacids (PUFAs), remains poorly defined. Notably, pros-taglandins (PGs), which are cyclooxygenase (COX)–generated metabolites of arachidonic acid (AA), influ-ence both the development of reproductive defects andchronic inflammation, which are hallmarks of PCOS.As a result of defects in follicular maturation, COXknockout mice show abnormal ovulation and fertil-ization of their oocytes (15–17). COX inhibition canalso block blastocyst development, which suggests thatCOX and its products are involved in early embryonicdevelopment (18). In addition, PGE2, PGI2, PGF2a, PGD2,

and 15-deoxy-D12,14-PG J2 also play major roles infollicular/oocyte development, ovulation, and fertilization(19–23). Furthermore, PGs are involved in the chronicinflammation that develops alongside the metabolic dis-orders associated with PCOS and obesity. However, thecirculating levels of these important lipid mediators havenot been systematically defined in patientswith PCOS, andit is unknown how obesity and hyperandrogenism affectthe production of these bioactive lipids in these patients.

As a branch of metabolomics, lipidomics is used tosystematically investigate a large range of lipids in a givenbiological system. It has become an indispensable tool tostudy lipidmetabolism in human disease. In this study, wecarried out systematic lipidomic profiling of the serumlipids that comprise the major lipid metabolic pathways(Fig. 1) in lean and obese patients with PCOS andmatched control subjects using mass spectrometry(MS)–based lipidomic methods. We observed signifi-cantly higher levels of phosphatidylcholine (PC) andconcomitantly lower lysophospholipid (LPC) in obesePCOS groups. Concentrations of PUFAs, such as AA,linoleic acid (LA), and docosahexaenoic acid (DHA),were also lower, whereas saturated long-chain fatty acids(FAs) were higher. Interestingly, however, major down-stream bioactive lipids generated from PUFAs were sig-nificantly higher in sera from obese controls and lower inobese and lean patients with PCOS when compared withtheir respective controls. These data suggest that obesity,which is typically accompanied by hyperinsulinemia,stimulates the production of bioactive lipids, whereasandrogen has the opposite effect. Similar results wereobtained using a rat model of PCOS that is obese anddemonstrates hyperandrogenism. Our clinical observa-tions imply that further mechanistic studies should beundertaken with regard to the antagonistic effects ofinsulin and androgen on circulating lipids in patients withPCOS. A deeper understanding of the molecular mech-anisms that are involved in the hormonal and metabolicdisorders present in PCOS may be able to guide newclinical interventions in future.

Materials and Methods

SubjectsThis clinical study was approved by the Internal Review and

Ethics Boards of Renji Hospital, which is affiliated withShanghai Jiao Tong University. All of the participants signed

doi: 10.1210/jc.2016-2692 press.endocrine.org/journal/jcem 811


http://dx.doi.org/10.1210/jc.2016-2692


informed consent to be included in the study. Both newly di-agnosed patients with PCOS and healthy controls were enrolledin the clinics of Renji Hospital. The study subjects were dividedinto 4 groups: lean control (LC; BMI,24 kg/m2; n = 18), obese(OB; BMI$28 kg/m2; n = 16), obese PCOS (OP; BMI$28 kg/m2;n = 15), and lean PCOS (LP; BMI ,24 kg/m2; n = 17), on thebasis of the Chinese criteria for BMI categories (24). The di-agnostic criteria for PCOSwere based on the unified standardsformulated by the Rotterdam International Conference in2003. Patients with any 2 of the following 3 conditions werediagnosed with PCOS: (1) infrequent ovulation or anov-ulation; (2) hyperandrogenism or clinical manifestations ofhigh blood androgen; (3) ultrasound findings of polycysticovaries in 1 or 2 ovaries, or$12 follicles measuring 2 to 9 mmin diameter, and/or ovarian volume $10 mL. Exclusion cri-teria included congenital adrenal hyperplasia, androgen-secreting tumors, Cushing syndrome, thyroid dysfunction,hyperprolactinemia, and other diseases (25). Patients in thecontrol groups exhibited normal menstruation, no clinical orbiochemical signs of hyperandrogenism, normal ovaries as de-fined by B-mode ultrasonic examination, no family history ofendocrine andmetabolic diseases, andno family history of PCOS.

Clinical measurementsFasting plasma glucose concentration was quantified using

the glucose oxidase method. Serum lipids [total cholesterol,triglyceride (TG), high-density lipoprotein (HDL), and low-density lipoprotein (LDL)] were measured by enzymaticassays (Cobas auto analyzer; Roche Diagnostics, Basel, Swit-zerland). Serum insulin was measured by radioimmunoassay.The concentrations of serum hormones [follicle stimulatinghormone, luteinizing hormone, sex hormone–binding globulin(SHBG), and dehydroepiandrosterone sulfate (DHEAS)] weredetermined using chemiluminescence (Elecsys Auto analyzer;Roche Diagnostics). Total testosterone (TT) was measured byliquid chromatography–MS according to protocols previously

reported (26). The intra-assay and inter-assay coefficients of variation were ,6%and 10%, respectively, for all analyses. Theinsulin resistance index was calculated ashomeostatic model assessment–insulin re-sistance (HOMA-IR). HOMA-IR = (fastingplasma glucose 3 fasting serum insulin) /22.5. The free androgen index (FAI) wascalculated as (TT 3 100) / SHBG. Thesubjects also received an ultrasound ex-amination in the Department of Obstetricsand Gynecology on a GE Voluson E8System (GE Healthcare, Madison, WI)between the second and fifth days aftermenstruation.

Lipidomics/metabolomicsSolvents for sample preparation andMS

analysis, such as methanol, chloroform,and water, were purchased from Burdickand Jackson (Muskegon, MI). Other high-performance liquid chromatography–quality solvents, including methanol, wa-ter, 2-propanol, hexane, and acetonitrile,were purchased from either Fisher Chem-ical (Phillipsburg, NJ) or EM Science

(Gibbstown, NJ). Standards and deuterated standards for AAmetabolites, including PGD2, PGE2, PGF2a, PGI2, throm-boxane B2 (TXB2), hydroxyeicosatetraenoic acids (HETEs),hydroxyoctadecadienoic acids (HODEs), and epoxyeicosa-trienoic acids (EETs), were purchased from Cayman Chemical(Ann Arbor, MI) and used without further purification. Fattyacids of the highest purity (.99%) were purchased from Nu-Chek Prep (Elysian, MN). All other chemical reagents werefrom Sigma-Aldrich (St. Louis, MO).

All lipid extracts from human or rat plasmawere prepared asdescribed previously, with slight modifications (27, 28). Forplasma lipidomics, 200 mL of plasma was extracted after ad-dition of the internal standard mixture. Shotgun lipidomics wasperformed on a TSQ Vantage triple-quadrupole mass spec-trometer (Thermo Scientific, San Jose, CA) according topublished protocols (28). Tandem MS scan fragment andcollision energy for each lipid class were optimized as pre-viously reported (27, 28). Fatty acids were analyzed as fattyacid methyl esters using gas chromatography (GC)–MS(Agilent 6890N/5975B; Agilent, Santa Clara, CA) in thepositive-ion mode of electron impact–MS according topublished protocols (29, 30). Fatty acid methyl esters wereanalyzed by GC (Agilent 6890 GC with SP-2560 capillarycolumn; Agilent; and 100 m 3 0.25 mm 3 0.2 mm film;Supelco, Bellefonte, PA).

Fatty acidmetabolites of AA and other PUFAswere analyzedusing a previously reported targeted metabolomic method(31–33). Briefly, after addition of a mixture of deuterated in-ternal standards to 400 mL of serum, the pH of the solution wasadjusted to 3.0 using 1 N HCl. Liquid-liquid extraction of themixture was carried out twice using hexane:methyl t-butyl ether(50:50, v/v). The samples were separated on a PhenomenexKinetix C18 column (3 mm, 100 3 2.1 mm; Phenomenex,Torrance, CA) using a Thermo Accela UPLC system (ThermoScientific) at a rate of 0.4 mL/min, using a gradient of mobilephase A (water:acetonitrile:formic acid 63:37:0.02, v/v/v) and

Figure 1. Pathways whereby bioactive lipid mediators are generated from phospholipids andFAs. AA is C20:4; DHA is C22:6; LA is C18:2. ALA, alpha linolenic acid C18:3a; DGLA,dihomo-g-linolenic acid C18:3g; DPA, docosapentaenoic acid C22:5; EPA, eicosapentaenoicacid C20:5; Lpcat, lysophosphatidylcholine acyltransferase; LT, leukotriene; LX, lipoxin; PLA2,phospholipase A2; P450epo, cytochrome P450 epoxygenase; RVD, resolvin D; RVE, resolvin E;TXA2, thromboxane A2.

812 Li et al Serum Lipidomic Profiles in Patients With PCOS J Clin Endocrinol Metab, March 2017, 102(3):810–821


mobile phase B (acetonitrile:isopropanol 50:50, v/v). MSanalysis was carried out on a TSQ Vantage triple-quadrupolemass spectrometer (Thermo Scientific). The mass spectrometerwas operated in the negative ion mode using multiple reactionmonitoring. Data acquisition and analysis were performedusing Xcalibur software, version 2.0 (Thermo Scientific).

Animal experimentsAnimal experiments were performed according to pro-

tocols approved by the Animal Care Committee of the Schoolof Medicine affiliated with Shanghai Jiaotong University.Thirty 21-day-old female Sprague-Dawley rats (50 to 60 g)from the Shanghai Laboratory Animal Center of the ChineseAcademy of Science, Shanghai, China, were individuallyhoused in ventilated cages in a temperature- and humidity-controlled environment. The animalswere randomly divided into3 groups: normal controls (CON; n = 10), a high-fat-diet–fedgroup (HF; n = 10), and a high-fat-diet–fed and hyper-androgenism group (HF + DHEA; n = 10), the last beingcomposed of obese rats with PCOS and high levels of an-drogen. Rats in the HF +DHEA group were fed a high-fat dietand subcutaneously injected with DHEA (Sigma-Aldrich)daily at a dose of 6 mg/100 g of body weight in 0.2 mL ofoil as a vehicle for 8 weeks, whereas the control group wasfed a normal diet and injected with vehicle only. The high-fatdiet comprised lard (20%), sugar (4%), whole milk powder(2%), cholesterol (1.5%), cholate (0.75%), plus a basal dietcomponent (71.75%, containing 20% protein, 4% fat, 5%crude fiber, 8% crude ash, and 52.5% nitrogen-free extract).When the body mass of the HF group became 30% greaterthan that of the control group and assessment of vaginalsmears showed a disappearance of estrous cyclicity in the HF +DHEAgroup, fasting bloodwas collected for themeasurement ofbiochemical parameters including glucose, insulin, lipids, and sexhormones, as well as for MS analysis for FFAs and lipid me-tabolites. Rat ovaries were also collected for paraffin embeddingand hematoxylin and eosin staining to evaluate follicularchanges. Vaginal smears were evaluated microscopically afterWright-Giemsa staining.

Statistical analysisAll data were analyzed using SPSS, version 16.0 (SPSS,

Chicago, IL). Continuous data are presented as means 6standard errors of the mean. The nonnormal distribution datawere analyzed after logarithmic transformation. Intergroupcomparisons were performed using 1-way analysis of variance,and categorical variables were analyzed using the x2 test. TheMann-WhitneyU test was used to analyze nonparametric data.P , 0.05 was considered statistically significant.

Results

Characteristics of the study subjectsThe characteristics of the 4 groups of study subjects are

summarized in Table 1. The average age of the LP groupwas lower than that of groups LC, OB, or OP. The OPgroup showed more marked metabolic disturbances thanthe other 3 groups, including higher levels of fastingblood glucose, fasting serum insulin, HOMA-IR, TG, andLDL, as well as lower levels of HDL. Higher levels of TT

and DHEAS were observed in subjects with PCOS (OPand LP groups) vs those with no PCOS (LC and OBgroups), whereas there were no significant differenceswithin PCOS groups. The level of SHBG in the OP groupwas markedly lower than that in the LP group, whereasthe FAI was significantly higher in the OP group than inthe LP group. Interestingly, the incidence of nonalcoholicfatty liver disease was much higher in the PCOS groups(OP and LP) than in their respective non-PCOS controlgroups (OB and LC).

Serum profile of glycerophospholipidSerum lipidomic changes associated with PCOS were

analyzed by a shotgun lipidomic approach, includingevaluation of phosphatidic acid, phosphatidylglycerol,phosphatidylserine (PS), phosphatidylinositol (PI), PC,and lysophospholipid of PC (LPC). This method rou-tinely detects and quantifies ;400 individual lipid ions,including PC, phosphatidylglycerol, PI, PS, phospha-tidic acid, LPC, and ceramide. Among these lipids, weobserved significantly higher levels of PC in the OPgroup than in the LC group, whereas, surprisingly, theconcentrations of LPC were lower in the OP group thanin the LC group (Table 2). However, there were nodifferences among the LP, LC, and OB groups. Furtheranalyses of PC and LPC demonstrated that the PCspecies that were present at higher concentrations wereC18:2, C20:2, C20:3, and C20:4, whereas the LPCspecies that were present at lower concentrations wereC16:0, C18:0, and C18:1. Interestingly, we also foundthat the concentrations of ceramide in the OB, OP, andLP groups were significantly higher than those in theLC group.

Serum profile of FFAsSerum FFAs were analyzed using GC-MS. The con-

centrations of PUFAs, including LA (C18:2) and AA(C20:4), were significantly lower in the OP group than inthe other 3 groups, whereas the concentration of DHA inthe LP group was higher than in the LC and OP groups.By contrast, levels of the saturated long-chain fatty acidsC20:0 and C24:0 were highest in the OP group, whereasthe concentration of C22:0 was higher in the OP and LPgroups than in the OB and LC groups (Fig. 2; Supple-mental Table 1).

Downstream metabolites of PUFAsAccumulating evidence suggests that these lipid media-

tors play an important role in inflammation and obesity;therefore, the unique patterns of PUFA-containing PC andLPC that we observed in the OP group prompted us tofurther analyze the downstreambioactive lipids. There are 3major enzymatic pathways that metabolize PUFAs. COXsconvert AA into PGs, such as PGE2, PGD2, PGF2a, PGI2,



http://dx.doi.org/10.1210/jc.2016-2692


and TXB2. Specific lipoxygenases (LOXs) metabolize AAinto 5-, 12-, and 15-HETEs, and 5-LOX can also generateleukotrienes. Cytochrome P450 enzymes represent the thirdmajor metabolic pathway for AA, generating EETs,diHETEs, and 19-, or 20-HETEs. In addition to these 3enzymatic pathways, free radical–induced lipid perox-idation generates structurally similar metabolites to theenzymatic products (34, 35). Some of these compounds,such as isoprostanes, can be used as markers for oxi-dative stress.

Because of the low abundance of these downstreammetabolites, we developed a targeted metabolomic ap-proach to systematically quantify all of the major AAmetabolites using liquid chromatography–MSandmultiplereaction monitoring. As shown in Fig. 3 and SupplementalTable 2, the levels of COX metabolites, including PGI2,PGE2, PGD2, PGFmetabolite, and TXB2, in theOBgroup,were higher than those in the LC group, whereas the levelsof the same metabolites were decreased in the OP and LPgroups vs control subjects. Similar patterns were also ob-served for the metabolites generated by the LOX and P450pathways.Notably, the concentrations of leukotrieneB4, 5-HETE, 12-HETE, 5(S),6(R),15(R)-LXA4, and 5(S),14(R)-LXB4 in the LP group were markedly lower than those intheOB group. Interestingly, metabolites of other PUFAs,such as LA, eicosapentaenoic acid, and DHA, showeda similar trend, i.e., OB . LC . OP . LP (Fig. 3;Supplemental Table 2). Androgen attenuated the ele-vated production of bioactive lipid mediators thatwas associated with obesity. The differences in the

metabolic profiles (LC compared with LP and OBcompared with OP) remained after adjusting for ageand BMI.

Lipid profiles in obese rats and a rat model of PCOSTo further investigate the effects of obesity, hyper-

insulinemia, and hyperandrogenism on themetabolism ofAA and other PUFAs, we used awell-established high-fat-diet– and DHEA-induced rat model of obese PCOS (36).The model has a number of clinical and pathologicalfeatures in common with obese patients with PCOS, suchas obesity, hyperinsulinemia, hyperandrogenism, ovarianfollicle immaturity, and ovulation disorders (37, 38).Compared with the control group (CON; normal diet),the HF group had significantly higher body mass andprominent hyperinsulinemia. Rats in the HF + DHEAgroup showed hyperandrogenism (Supplemental Ta-ble 3). Vaginal smears stained with Wright-Giemsarevealed that rats in the CON and HF groups had reg-ular estrous cycles, whereas estrous cyclicity disappearedin the HF + DHEA group (Supplemental Fig. 1). He-matoxylin and eosin staining of ovarian specimensrevealed that, unlike in the CON and HF groups, in HF +DHEA rats the number of corpora lutea was reduceddramatically, whereas the numbers of small and atreticfollicles were increased. In addition, no mature ordominant follicles were observed (Supplemental Fig. 2).In summary, therefore, the HF + DHEA rat modelshowed insulin resistance, a lack of estrous cyclicity,hyperandrogenism, and follicular/ovulatory disorders,

Table 1. Characteristics of the Study Participants

LC (n = 18) OB (n = 16) LP (n = 17) OP (n = 15)

Age, y 27.11 6 1.07 30.75 6 0.82 24.88 6 1.12a,d 29.33 6 1.02BMI, kg/m2 19.68 6 0.34 29.00 6 0.92b 19.63 6 0.42d 31.56 6 1.50b,d,f

FBG, mmol/L 4.65 6 0.0.08 5.16 6 0.24 4.47 6 0.11 5.21 6 0.29a,c,f

Fins, mIU/L 5.85 6 0.33 14.04 6 1.83a 10.83 6 1.04a 36.98 6 4.06b,d,f

HOMA-IR 1.21 6 0.07 3.38 6 0.54a 2.17 6 0.24 9.15 6 1.52b,d,f

TT, nmol/L 1.10 6 0.09 1.23 6 0.09 2.22 6 0.20b,d 2.00 6 0.19b,d

SHBG, nmol/L 85.17 6 4.68 51.24 6 6.10a 59.72 6 10.48 16.66 6 2.26b,d,f

DHEAS, mg/mL 177.73 6 12.43 177.30 6 17.84 288.96 6 33.85b,c 295.95 6 23.64a,c

FAI 1.38 6 0.14 2.53 6 0.21 3.93 6 0.42 15.39 6 2.21b,d,f

TG, mmol/L 1.00 6 0.21 2.11 6 0.34b 1.98 6 0.04a 2.67 6 0.10b,e

TC, mmol/L 4.16 6 0.25 4.51 6 0.36 4.86 6 0.22b,c 4.31 6 0.20a,c,f

HDL, mmol/L 1.51 6 0.07 1.22 6 0.07a 1.56 6 0.09c 1.04 6 0. 06b,c,e

LDL, mmol/L 2.08 6 0.10 2.93 6 0.15a 2.47 6 0.15 3.08 6 0.19b,f

Data are mean 6 standard error of the mean.

Abbreviations: FBG, fasting blood glucose; Fins, fasting blood insulin; TC, total cholesterol.aP , 0.05.bP , 0.01 compared with group LC.cP , 0.05.dP , 0.01 compared with group OB.eP , 0.05.fP , 0.01 compared with group LP.



consistent with the main pathophysiological features ofobese patients with PCOS.

Lipidomic studies were performed on rat serum andshowed that total FFAs in the HF + DHEA group weresignificantly higher than in the CON and HF groups(Supplemental Table 4). In particular, AA was significantlyhigher in this group than in the other 2 groups (HF +DHEA, 22.2462.12mg/mL, vsCON, 6.0660.57mg/mL,and HF, 6.47 6 0.35 mg/mL).

Further analyses of AA metabolites generated by the 3major enzymatic pathways (COX, LOX, and P450)revealed remarkably similar changes to those observed inthe human clinical samples, i.e., HF . CON . HF +DHEA (equivalent to OB. CON.OP). The changes inCOX and LOX metabolites were significant. The dif-ferences in LA and DHA also resembled the pattern

observed with the human clinical samples (Fig. 4; Sup-plemental Table 5).

Discussion

PCOS is a complex and heterogeneous clinical syndrome.Women with this syndrome have an increased risk ofdeveloping T2DM and cardiovascular diseases. Emerg-ing evidence suggests that chronic low-grade inflamma-tion is involved in the pathogenesis of PCOS and obesityand is linked to insulin resistance (39). Besides the insulinresistance and associated compensatory hyperinsulinemia,hyperandrogenism is another core pathophysiologicalchange of PCOS. However, the association between thesepathological features and the characteristic changes inthe lipid profiles of patients with PCOS has not been

Figure 2. Serum saturated long-chain fatty acids in lean and obese patients and patients with PCOS. *P , 0.05 compared with LC; #P , 0.05compared with OB; &P , 0.05 compared with LP.

Table 2. Serum Phospholipids in Lean and Obese Patients and Patients With PCOS

LC OB LP OP

PA, pmol/mL 2016.9 6 345.6 3864.2 6 1558.5 1938.3 6 360.2 3314.9 6 829.1Phosphatidylglycerol, pmol/mL 1813.3 6 445.9 3252.3 6 856.5 390.7 6 72.8d 2038.8 6 615.1PI, pmol/mL 3068.3 6 492.8 4099.4 6 780.1 1510.3 6 276.9c 3399.3 6 803.5PS, pmol/mL 3211.2 6 1519.9 1,3573.7 6 6010.8 1183.3 6 327.3 1,2974.2 6 8466.0Cer, pmol/mL 2375.8 6 172.5 3154.1 6 116.9a 3860.7 6 297.6b,c 3105.2 6 219.5a,e

LPC, nmol/mL 497.2 6 66.0 355.9 6 32.1 382.3 6 80.7 314.2 6 51.6a

PC, nmol/mL 619.3 6 54.8 761.7 6 44.9 792.0 6 97.6 856.9 6 59.2a

PC:D16:0–18:2 176.1 6 17.3 203.9 6 10.2 236.0 6 39.4 249.9 6 18.3a

PC:D18:2–18:2/D16:0–20:4 78.7 6 8.6 107.1 6 9.3 95.2 6 23.3 107.2 6 9.6PC:D18:2–20:2/D18:0–20:4 50.4 6 4.0 71.5 6 6.6a 58.5 6 8.0 66.6 6 5.4a

PC:D18:1–18:2/D16:0–20:3 39.1 6 5.2 44.7 6 8.6 56.3 6 6.9 58.2 6 7.0a

PC:D18:0–18:2/D18:1–18:1 88.1 6 9.2 98.2 6 11.3 117.8 6 11.6 127.7 6 13.3a

LPC:16:0 133.2 6 21.8 124.9 6 21.7 76.3 6 17.2a 85.7 6 11.2a

LPC:P18:1 15.1 6 4.8 4.8 6 1.5a 10.8 6 3.2 4.7 6 0.98a

LPC: P18:0 19.2 6 7.0 9.2 6 2.6 12.5 6 2.9 7.4 6 2.4a

LPC:18:1 34.4 6 5.7 24.6 6 2.0 40.7 6 11.2 20.2 6 2.0a,e

LPC:18:0 53.8 6 7.5 51.2 6 7.1 48.8 6 14.5 42.8 6 8.1

Data are mean 6 standard error of the mean.

Abbreviations: Cer, ceramide; PA, phosphatidic acid.aP , 0.05.bP , 0.01 compared with group LC.cP , 0.05.dP , 0.01 compared with group OB.eP , 0.05.



http://dx.doi.org/10.1210/jc.2016-2692


Figure 3. Serum bioactive lipids derived from AA, eicosapentaenoic acid (EPA), DHA, and LA. AA is C20:4; DHA is C22:6; EPA is C20:5; LA isC18:2. *P , 0.05 compared with LC; #P , 0.05 compared with OB; &P , 0.05 compared with LP. LT, leukotriene; LTB4, leukotriene B4; LX,lipoxin; LXA4, lipoxin A4; PGFM, prostaglandin F metabolite; PGI2, prostacycline I2; P450epo, cytochrome P450 epoxygenase; 15d-PGJ2,15-deoxy-D12,14-prostaglandin J2.



Figure 4. Lipid metabolites derived from AA, eicosapentaenoic acid (EPA), DHA, and LA in rat serum. AA is C20:4; DHA is C22:6; EPA is C20:5; LAis C18:2. *P , 0.05 compared with CON; #P , 0.05 compared with HF. LTB4 leukotriene B4; LXA4, lipoxin A4; PGFM, prostaglandin F metabolite;PGI2, prostacycline I2; P450epo, cytochrome P450 epoxygenase; RVD, resolvin D; RVE, resolvin E; 15d-PGJ2, 15-deoxy-D

12,14-prostaglandin J2.



http://dx.doi.org/10.1210/jc.2016-2692


systematically investigated. This study provides anextensive profile of serum lipid metabolism in bothhumans and a rat model of PCOS and associatedmetabolic disorders, and clearly demonstrates thatthere are significant changes in the production ofphospholipids, FFAs, and bioactive lipids in lean andobese patients with PCOS compared with womenwithout PCOS.

Phospholipids are the main constituents of the cellmembrane, and an altered profile may have a profoundimpact on cell function. Because of the complexity ofphospholipid metabolism, lipidomics is the tool of choiceto study systematic changes in phospholipids and tocorrelate these changes with disease. PC is 1 of the majorphospholipids, and can be hydrolyzed to LPC and FFAsby phospholipases. The ratio of PC/LPC in serum hasbeen studied in hepatic diseases associated with in-flammation or infection. For example, LPC is decreasedin a model of drug-induced liver injury, and in patientswith viral hepatitis, alcoholic liver cirrhosis, and acutedeterioration of liver function (40, 41). Furthermore,lower levels of LPC have been reported in patients withnonalcoholic steatohepatitis (42, 43). Thus, our ob-servations of the features of phospholipid profiles inPCOS serum may reflect the inflammatory status of theliver, associated with the higher incidence of fatty liverin PCOS (Supplemental Fig. 3). The current study isconsistent with a previous study in which serum LPCs(C16:0, C18:0, C18:1) were decreased in an animal modelof fatty liver disease (44). AA in serum is primarily derivedfrom the hydrolysis of phospholipids at sn-2 (45). Thus,the abnormal PC/LPC in the OP group may partiallycontribute to the decreased serum AA level.

Bioactive derivatives of AA, especially PGs, play amajor role in inflammation and reproduction. Becauseserum AA was lower in the OP group, we also quantifiedthe downstream metabolites of AA and other PUFAsusing a targeted metabolomic approach. We found thatsimilar patterns of metabolites were derived from theaction of COX, LOX, and P450: their concentrationsfollowed the pattern OB . LC . OP . LP. The mostsignificant difference was between the OB and LP groups(Fig. 3; Supplemental Table 2). Interestingly, COX andLOXmetabolites generated from other PUFAs, includingeicosapentaenoic acid, DHA, and LA, showed patternsresembling those derived from AA (Fig. 3; SupplementalTable 2).

Whereas the OB group presented with obvious insulinresistance, but without marked hyperandrogenemia, theLP group had hyperandrogenemia and mild insulin re-sistance. Patients in the OP group were insulin resistantand hyperandrogenemic. On the basis of the differencesin AA metabolites among the 4 groups, we hypothesize

that insulin upregulates the expression and activity ofCOX, LOX, and other metabolic enzymes, whereasandrogen downregulates their expression and/or activity.Because patients in the OP group had both insulin re-sistance and hyperandrogenemia, the levels of the me-tabolites in this group were in the midrange.

As shown in Table 1, fasting insulin was much higherin the OP group than in the other groups. Even in thepresence of such a high concentration of insulin, however,the high androgen concentration still appeared to becapable of significantly lowering the production of bio-active lipids. These data, together with the fact that thelevels of other metabolites in the OP group were close to,or even lower than, those in the control groups, stronglysuggest that androgen can reverse or override the stim-ulatory effects of insulin. Alternatively, hyperinsulinemiamay reduce the level of SHBG, leading to a high FAIvalue, and thereby amplification of the effect of andro-gens. The underlying molecular mechanisms of the an-tagonistic actions of insulin and androgen on bioactivelipids require further investigation.

Importantly, we successfully recapitulated dataobtained from the human patients with PCOS in a well-established HF/DHEA-induced rat model of PCOS. Se-rum FFAs, including AA, were increased after injection ofDHEA. The increase in free AAwasmost likely a result ofthe upregulation of lipolysis. In female animals, supra-physiological doses of androgens promote fat mobiliza-tion and decomposition (46, 47). Our polymerase chainreaction results demonstrated that the expression ofhormone-sensitive lipase messenger RNA was signifi-cantly upregulated in the adipose tissue of the HF +DHEA group vs the CON and HF groups (SupplementalFig. 4).

In the HF group, although there was no significantchange in AA compared with controls, AA metaboliteswere markedly increased, which is consistent with theclinical findings (Fig. 4). Evenwith the significant increaseof AA in the HF + DHEA group, we observed a profounddecrease in the main AA metabolites, which suggests astrong inhibitory effect of androgen on bioactive lipidproduction. Similar results were also obtained withregard to LA, DHA, and other PUFA metabolites (Fig. 4;Supplemental Table 5). The data obtained from the an-imal studies were consistent with those from the humanstudies: obesity, which is commonly accompanied bycompensatory hyperinsulinemia, promotes PUFA meta-bolism, whereas androgen has an inhibitory effect onPUFA metabolism.

It is well established that obesity is associated withchronic low-grade inflammation. Previously, increasedlevels of circulating androgens in obese women withPCOS had been thought to lead to enhanced



inflammation, which would aggravate the metabolic dis-order. However, recent studies show that the elevatedcirculating androgens may exert anti-inflammatory effectswhen obesity is present (48, 49). Nonetheless, the un-derlying molecular mechanism remains to be elucidated.Our data show that the COX, LOX, and P450 metabolicpathways were upregulated in obese women, whichresulted in higher levels of bioactive lipids. Conversely,concentrations of these bioactive lipids were lower whenhyperandrogenism was present, as in both the obese pa-tients with PCOS and the obese PCOS rat model. Thus,our data provide insights into an anti-inflammatory roleof androgens, which is at least partially mediated by areduction in AA-derived inflammatory lipids.

Androgens also play an important part in lipidmetabolism. Previous studies have focused mainly oncholesterol (total, HDL, and LDL), oxidized LDL, andTGs and have shown that androgens also increase li-polysis in visceral fat cells in women with PCOS (15, 16).To our knowledge, however, the effects of androgen onthe metabolic profiles of PUFA, especially in obesewomen and women with PCOS, have not been reported.Our observations that androgens antagonize the pro-duction of PGs may have clinical implications for thereproductive disorders present in PCOS.

COX2 is involved in various reproductive functions inthe ovary, including oocyte maturation, ovulation, earlyembryonic development, implantation, and parturition,as previously mentioned (16–18). Nevertheless, the ef-fects of androgen on COX2 expression and activity re-main controversial. Yazawa et al. (50) found thatdihydrotestosterone (DHT) treatment induced expres-sion of the COX2 gene in granulosa cells from normalrats and human ovarian granulosa-like tumor cell line(KGN) in vitro, whereas other studies suggest a morecomplicated method of COX2 regulation by DHT. Forexample, although DHT upregulated COX2 expressionin the absence of induced inflammation via an androgenreceptor (AR)–dependent mechanism, it downregulatedCOX2 expression in cytokine- or LPS-induced inflam-mation via an AR-independent mechanism (51, 52).Thus, DHTdifferentially influences COX2 levels throughAR-dependent and independent mechanisms, dependingon the physiological and pathophysiological state of thecells (51). If the follicle is in an inflammatory microen-vironment, COX2 expression may be inhibited by theelevated androgen, and the luteinizing hormone–inducedPG levels will be decreased, eventually resulting in im-paired spontaneous ovulation in women with PCOS. Weare currently investigating the molecular mechanismswhereby hyperandrogenism contributes to ovulatorydysfunction through modulation of bioactive lipids, in-cluding PGs.

In summary, we observed an abnormal PC/LPC ratioin obese patients with PCOS, whichmay result in changesin serum AA levels. Obesity, which is usually accompa-nied by compensatory hyperinsulinemia, promoted themetabolism of AA and other PUFAs, whereas androgenshad an inhibitory effect. Further understanding of themolecular mechanisms that lead to the altered lipidprofiles we identified, together with genomic and pro-teomic studies, may provide new insights into the path-ogenic mechanisms of PCOS and suggest novel potentialtherapeutic strategies.

Acknowledgments

The authors thank the research volunteers for their participationin this study.

Address all correspondence and requests for reprints to: WeiLiu, PhD,Department of Endocrinology, RenjiHospital, Schoolof Medicine, Shanghai Jiao Tong University. 160 Pujian Road,Shanghai 200127, China. E-mail: [email protected].

This study was supported by the National Scientific Foun-dation of China (grant nos. 81270875, 81471424, and81471029). H.Y. acknowledges financial support from theMinistry of Science and Technology of China (grant nos.2016YFC0903403, 2016YFD0400205) and the NationalNatural Science Foundation of China (grant nos. 31470831,91439103, 91539127). H.Y. is an Associate Fellow at theCollaborative Innovation Center for Cardiovascular DiseaseTranslational Medicine at Nanjing Medical University.

Disclosure Summary: The authors have nothing to disclose.

References

1. Knochenhauer ES, Key TJ, Kahsar-Miller M, Waggoner W, BootsLR, Azziz R. Prevalence of the polycystic ovary syndrome in un-selected black and white women of the southeastern United States:a prospective study. J Clin Endocrinol Metab. 1998;83(9):3078–3082.

2. Diamanti-Kandarakis E, Kouli CR, Bergiele AT, Filandra FA,Tsianateli TC, Spina GG, Zapanti ED, Bartzis MI. A survey of thepolycystic ovary syndrome in the Greek island of Lesbos: hormonaland metabolic profile. J Clin Endocrinol Metab. 1999;84(11):4006–4011.

3. Azziz R, Woods KS, Reyna R, Key TJ, Knochenhauer ES, YildizBO. The prevalence and features of the polycystic ovary syndromein an unselected population. J Clin Endocrinol Metab. 2004;89(6):2745–2749.

4. Diamanti-Kandarakis E, Dunaif A. Insulin resistance and thepolycystic ovary syndrome revisited: an update onmechanisms andimplications. Endocr Rev. 2012;33(6):981–1030.

5. Legro RS, Kunselman AR, Dodson WC, Dunaif A. Prevalence andpredictors of risk for type 2 diabetes mellitus and impaired glucosetolerance in polycystic ovary syndrome: a prospective, controlledstudy in 254 affectedwomen. J Clin EndocrinolMetab. 1999;84(1):165–169.

6. Salehi M, Bravo-Vera R, Sheikh A, Gouller A, Poretsky L. Path-ogenesis of polycystic ovary syndrome: what is the role of obesity?Metabolism. 2004;53(3):358–376.

7. Dunaif A, Graf M, Mandeli J, Laumas V, Dobrjansky A. Char-acterization of groups of hyperandrogenic women with acanthosis



mailto:[email protected]

http://dx.doi.org/10.1210/jc.2016-2692


nigricans, impaired glucose tolerance, and/or hyperinsulinemia.J Clin Endocrinol Metab. 1987;65(3):499–507.

8. Vigil P, Contreras P, Alvarado JL, Godoy A, Salgado AM, CortesME. Evidence of subpopulations with different levels of insulinresistance inwomenwith polycystic ovary syndrome.HumReprod.2007;22(11):2974–2980.

9. Ciampelli M, Leoni F, Cucinelli F, Mancuso S, Panunzi S, DeGaetano A, Lanzone A. Assessment of insulin sensitivity frommeasurements in the fasting state and during an oral glucose tol-erance test in polycystic ovary syndrome and menopausal patients.J Clin Endocrinol Metab. 2005;90(3):1398–1406.

10. Groop LC, Bonadonna RC, Simonson DC, Petrides AS, Shank M,DeFronzo RA. Effect of insulin on oxidative and nonoxidativepathways of free fatty acid metabolism in human obesity. Am JPhysiol. 1992;263(1 Pt 1):E79–E84.

11. Højlund K, Glintborg D, AndersenNR, Birk JB, Treebak JT, FrøsigC, Beck-Nielsen H, Wojtaszewski JF. Impaired insulin-stimulatedphosphorylation of Akt and AS160 in skeletal muscle of womenwith polycystic ovary syndrome is reversed by pioglitazone treat-ment. Diabetes. 2008;57(2):357–366.

12. Holte J, Bergh T, Berne C, Wide L, Lithell H. Restored insulinsensitivity but persistently increased early insulin secretion afterweight loss in obese women with polycystic ovary syndrome. J ClinEndocrinol Metab. 1995;80(9):2586–2593.

13. Ek I, Arner P, Ryden M, Holm C, Thorne A, Hoffstedt J, Wah-renberg H. A unique defect in the regulation of visceral fat celllipolysis in the polycystic ovary syndrome as an early link to insulinresistance. Diabetes. 2002;51(2):484–492.

14. XuX, De Pergola G, Eriksson PS, Fu L, Carlsson B, Yang S, Eden S,Bjorntorp P. Postreceptor events involved in the up-regulation ofbeta-adrenergic receptor mediated lipolysis by testosterone in ratwhite adipocytes. Endocrinology. 1993;132(4):1651–1657.

15. O’Connor A, Gibney J, Roche HM. Metabolic and hormonal as-pects of polycystic ovary syndrome: the impact of diet. Proc NutrSoc. 2010;69(4):628–635.

16. Sirois J, Dore M. The late induction of prostaglandin G/Hsynthase-2 in equine preovulatory follicles supports its role as adeterminant of the ovulatory process. Endocrinology. 1997;138(10):4427–4434.

17. Richards JS. Sounding the alarm—does induction of prostaglandinendoperoxide synthase-2 control the mammalian ovulatory clock?Endocrinology. 1997;138(10):4047–4048.

18. Lim H, Paria BC, Das SK, Dinchuk JE, Langenbach R, TrzaskosJM, Dey SK. Multiple female reproductive failures in cyclo-oxygenase 2-deficient mice. Cell. 1997;91(2):197–208.

19. Nunes C, Silva JV, Silva V, Torgal I, Fardilha M. Signallingpathways involved in oocyte growth, acquisition of competenceand activation. Hum Fertil (Camb). 2015;18(2):149–155.

20. Jeremy JY, Okonofua FE, Thomas M, Wojdyla J, Smith W, CraftIL, Dandona P. Oocyte maturity and human follicular fluidprostanoids, gonadotropins, and prolactin after administration ofclomiphene and pergonal. J Clin Endocrinol Metab. 1987;65(3):402–406.

21. Kohda H, Mori T, Nishimura T, Kambegawa A. Co-operation ofprogesterone and prostaglandins in ovulation induced by humanchorionic gonadotrophin in immature rats primed with pregnantmare serum gonadotrophin. J Endocrinol. 1983;96(3):387–393.

22. Tokugawa Y, Kunishige I, Kubota Y, Shimoya K, Nobunaga T,Kimura T, Saji F, Murata Y, Eguchi N, Oda H, Urade Y, HayaishiO. Lipocalin-type prostaglandin D synthase in human male re-productive organs and seminal plasma. Biol Reprod. 1998;58(2):600–607.

23. SchaiffWT,CarlsonMG, Smith SD, LevyR,NelsonDM, SadovskyY. Peroxisome proliferator-activated receptor-gamma modulatesdifferentiation of human trophoblast in a ligand-specific manner.J Clin Endocrinol Metab. 2000;85(10):3874–3881.

24. Zhou BF; Cooperative Meta-Analysis Group of the WorkingGroup on Obesity in China. Predictive values of body mass index

andwaist circumference for risk factors of certain related diseases inChinese adults—study on optimal cut-off points of bodymass indexand waist circumference in Chinese adults. Biomed Environ Sci.2002;15(1):83–96.

25. Rotterdam ESHRE/ASRM-Sponsored PCOS consensus workshopgroup.Revised 2003 consensus on diagnostic criteria and long-termhealth risks related to polycystic ovary syndrome (PCOS). HumReprod. 2004;19(1):41–47 (Review).

26. Wang C, Catlin DH, Demers LM, Starcevic B, Swerdloff RS.Measurement of total serum testosterone in adult men: comparisonof current laboratory methods versus liquid chromatography-tandem mass spectrometry. J Clin Endocrinol Metab. 2004;89(2):534–543.

27. Han X, Yang J, Cheng H, Ye H, Gross RW. Toward fingerprintingcellular lipidomes directly from biological samples by two-dimensional electrospray ionization mass spectrometry. AnalBiochem. 2004;330(2):317–331.

28. Han X, Yang K, Cheng H, Fikes KN, Gross RW. Shotgun lip-idomics of phosphoethanolamine-containing lipids in biologicalsamples after one-step in situ derivatization. J Lipid Res. 2005;46(7):1548–1560.

29. Yu DX, Sun Q, Ye XW, Pan A, Zong G, Zhou YH, Li HX, Hu FB,Lin X. Erythrocyte trans-fatty acids, type 2 diabetes and cardio-vascular risk factors in middle-aged and older Chinese individuals.Diabetologia. 2012;55(11):2954–2962.

30. Zhang G, Sun Q, Hu FB, Ye X, Yu Z, Zong G, Li H, Zhou Y, Lin X.Erythrocyte n-3 fatty acids and metabolic syndrome in middle-agedand older Chinese. J Clin Endocrinol Metab. 2012;97(6):E973–E977.

31. Yang J, Schmelzer K, Georgi K, Hammock BD. Quantitativeprofiling method for oxylipin metabolome by liquid chromatog-raphy electrospray ionization tandem mass spectrometry. AnalChem. 2009;81(19):8085–8093.

32. Buczynski MW, Dumlao DS, Dennis EA. Thematic Review Series:proteomics. An integrated omics analysis of eicosanoid biology.J Lipid Res. 2009;50(6):1015–1038.

33. Deems R, Buczynski MW, Bowers-Gentry R, Harkewicz R, DennisEA. Detection and quantitation of eicosanoids via high perfor-mance liquid chromatography-electrospray ionization-mass spec-trometry. Methods Enzymol. 2007;432:59–82.

34. Yin H, Xu L, Porter NA. Free radical lipid peroxidation: mecha-nisms and analysis. Chem Rev. 2011;111(10):5944–5972.

35. Milne GL, Yin H, Hardy KD, Davies SS, Roberts LJ II. Isoprostanegeneration and function. Chem Rev. 2011;111(10):5973–5996.

36. Zhang H, Yi M, Zhang Y, Jin H, Zhang W, Yang J, Yan L, Li R,Zhao Y, Qiao J. High-fat diets exaggerate endocrine and metabolicphenotypes in a rat model of DHEA-induced PCOS.Reproduction.2016;151(4):431–441.

37. Ressler IB, Grayson BE, Ulrich-Lai YM, Seeley RJ. Diet-inducedobesity exacerbates metabolic and behavioral effects of polycysticovary syndrome in a rodent model. Am J Physiol EndocrinolMetab. 2015;308(12):E1076–E1084.

38. Lai H, Jia X, YuQ, Zhang C, Qiao J, Guan Y, Kang J. High-fat dietinduces significant metabolic disorders in a mouse model ofpolycystic ovary syndrome. Biol Reprod. 2014;91(5):127.

39. Repaci A, Gambineri A, Pasquali R. The role of low-grade in-flammation in the polycystic ovary syndrome.MolCell Endocrinol.2011;335(1):30–41.

40. Lian JS, LiuW, Hao SR, Guo YZ,HuangHJ, Chen DY, Xie Q, PanXP, Xu W, Yuan WX, Li LJ, Huang JR. A serum metabonomicstudy on the difference between alcohol- and HBV-induced livercirrhosis by ultraperformance liquid chromatography coupled tomass spectrometry plus quadrupole time-of-flight mass spectrom-etry. Chin Med J (Engl). 2011;124(9):1367–1373.

41. Yang J, Zhao X, Liu X, Wang C, Gao P, Wang J, Li L, Gu J,Yang S, Xu G. High performance liquid chromatography-massspectrometry for metabonomics: potential biomarkers for acutedeterioration of liver function in chronic hepatitis B. J ProteomeRes. 2006;5(3):554–561.



42. Puri P, Wiest MM, Cheung O, Mirshahi F, Sargeant C, Min HK,Contos MJ, Sterling RK, Fuchs M, Zhou H, Watkins SM, SanyalAJ. The plasma lipidomic signature of nonalcoholic steatohepatitis.Hepatology. 2009;50(6):1827–1838.

43. Kalhan SC, Guo L, Edmison J, Dasarathy S, McCullough AJ,Hanson RW, Milburn M. Plasma metabolomic profile in non-alcoholic fatty liver disease. Metabolism. 2011;60(3):404–413.

44. Tanaka N, Matsubara T, Krausz KW, Patterson AD, Gonzalez FJ.Disruption of phospholipid and bile acid homeostasis in mice withnonalcoholic steatohepatitis. Hepatology. 2012;56(1):118–129.

45. Zhou L, Nilsson A. Sources of eicosanoid precursor fatty acid poolsin tissues. J Lipid Res. 2001;42(10):1521–1542.

46. De Pergola G. The adipose tissue metabolism: role of testosteroneand dehydroepiandrosterone. Int J Obes Relat Metab Disord.2000;24(Suppl 2):S59–S63 (Review).

47. Arner P. Effects of testosterone on fat cell lipolysis. Species dif-ferences and possible role in polycystic ovarian syndrome. Bio-chimie. 2005;87(1):39–43.

48. WahrenbergH, Ek I, Reynisdottir S, CarlstromK, Bergqvist A, ArnerP. Divergent effects of weight reduction and oral anticonception

treatment on adrenergic lipolysis regulation in obese womenwith thepolycystic ovary syndrome. J Clin Endocrinol Metab. 1999;84(6):2182–2187.

49. Gonzalez F, Sia CL, Stanczyk FZ, Blair HE, Krupa ME. Hyper-androgenism exerts an anti-inflammatory effect in obese womenwith polycystic ovary syndrome. Endocrine. 2012;42(3):726–735.

50. Yazawa T, Kawabe S, Kanno M, Mizutani T, Imamichi Y, Ju Y,Matsumura T, Yamazaki Y, Usami Y, KuribayashiM, ShimadaM,Kitano T, Umezawa A, Miyamoto K. Androgen/androgen receptorpathway regulates expression of the genes for cyclooxygenase-2and amphiregulin in periovulatory granulosa cells. Mol CellEndocrinol. 2013;369(1-2):42–51.

51. Osterlund KL, Handa RJ, Gonzales RJ. Dihydrotestosterone al-ters cyclooxygenase-2 levels in human coronary artery smoothmuscle cells. Am J Physiol Endocrinol Metab. 2010;298(4):E838–E845.

52. Zuloaga KL, Gonzales RJ. Dihydrotestosterone attenuates hypoxiainducible factor-1a and cyclooxygenase-2 in cerebral arteriesduring hypoxia or hypoxia with glucose deprivation. Am J PhysiolHeart Circ Physiol. 2011;301(5):H1882–H1890.



http://dx.doi.org/10.1210/jc.2016-2692


Documents

Discovery of Novel Lipid Profiles in PCOS: Do Insulin and