Insulin-like growth factor 1 (IGF1), IGF binding protein 3(IGFBP3), and breast cancer risk: pooled individual dataanalysis of 17 prospective studies

7/30/2019 Insulin-like growth factor 1 (IGF1), IGF binding protein 3(IGFBP3), and breast cancer risk: pooled individual dataan

1/13

Articles

530 www.thelancet.com/oncology Vol 11 June 2010

Lancet Oncol 2010; 11: 53042

Published Online

May 17, 2010

DOI:10.1016/S1470-

2045(10)70095-4

See Reflection and Reaction

page 501

*See end of paper for members

and affi liatio ns.Correspondence to:

Prof Tim Key, Endogenous

Hormones and Breast Cancer

Collaborative Group, Cancer

Epidemi ology Unit, Nuffi eld

Department of Clinical Medicine,

University of Oxford, Richard

Doll Building, Roosevelt Drive,

Oxford, OX3 7LF, UK

[email protected]

Insulin-like growth factor 1 (IGF1), IGF binding protein 3

(IGFBP3), and breast cancer risk: pooled individual dataanalysis of 17 prospective studies

The Endogenous Hormones and Breast Cancer Collaborative Group*

SummaryBackground Insulin-like growth factor 1 (IGF1) stimulates mitosis and inhibits apoptosis. Some published resultshave shown an association between circulating IGF1 and breast-cancer risk, but it has been unclear whether thisrelationship is consistent or whether it is modified by IGF binding protein 3 (IGFBP3), menopausal status, oestrogenreceptor status or other factors. The relationship of IGF1 (and IGFBP3) with breast-cancer risk factors is also unclear.The Endogenous Hormones and Breast Cancer Collaborative Group was established to analyse pooled individual datafrom prospective studies to increase the precision of the estimated associations of endogenous hormones with breast-cancer risk.

Methods Individual data on prediagnostic IGF1 and IGFBP3 concentrations were obtained from 17 prospectivestudies in 12 countries. The associations of IGF1 with risk factors for breast cancer in controls were examined bycalculating geometric mean concentrations in categories of these factors. The odds ratios (ORs) with 95% CIs ofbreast cancer associated with increasing IGF1 concentrations were estimated by conditional logistic regression in4790 cases and 9428 matched controls, with stratification by study, age at baseline, and date of baseline. All statisticaltests were two-sided, and a p value of less than 005 was considered significant.

Findings IGF1 concentrations, adjusted for age, were positively associated with height and age at first pregnancy,inversely associated with age at menarche and years since menopause, and were higher in moderately overweightwomen and moderate alcohol consumers than in other women. The OR for breast cancer for women in the highestversus the lowest fifth of IGF1 concentration was 128 (95% CI 114144; p


2/13

Articles

www.thelancet.com/oncology Vol 11 June 2010 531

hormones with breast-cancer risk.25 In this study weundertook a collaborative analysis of data from 17 studiesto investigate the associations of IGF1 and IGFBP3 withbreast-cancer risk. We also examined consistencybetween studies, associations in subgroups includingmenopausal status at blood collection and oestrogenreceptor status, the effects of adjustment of IGF1 andIGFBP3 for each other and for other risk factors, and thejoint associations of IGF1, oestradiol, and testosteronewith breast cancer risk in postmenopausal women.

MethodsData collectionStudies were eligible for the collaborative analysis if theyhad prospectively collected blood samples and data oncirculating IGF1, IGFBP3, and breast-cancer risk.

Potentially eligible studies were identified throughPubMed using the terms IGF1, IGFBP3, and breast

cancer, by searching the reference lists of identifiedstudies, and by correspondence with study investigators.17 eligible studies were identified: CLUE I and CLUE IIfrom the USA;17 European Prospective Investigation intoCancer and Nutrition (EPIC), from Europe;16 Guernseystudy, from the UK;13 Janus biobank study, from Norway;20Danish Diet, Cancer, and Health study (KKH), fromDenmark;12 Kaiser Permanente-Orentreich FoundationStudy (KP-OFAS) study, from the USA;9 Malm andNorthern Sweden studies, from Sweden;8 MelbourneCollaborative Cohort Study (MCCS), from Australia;19Nurses Health Study, from the USA;6,15 Nurses HealthStudy II, from the USA;18 New York University WomensHealth Study (NYU WHS), from the USA;7 Study ofHormones and Diet in the Etiology of Breast Tumours(ORDET), from Italy;10 Prostate, Lung, Colorectal, and

Ovarian Cancer Screening Trial (PLCO), from the USA;22

Monitoring Project on Cardiovascular Disease Risk Factors

Recruitment

period

Fasting status Storage

temperature

Matching criteria

Age at bloodcollection

Date of blood sample Other matching criteria and comments

CLUE I and CLUE II, USA17 1974 and 1989 Non-fasting 70C 1 year 14 days Participation in one or both cohorts, menopausal status,

ethnic group, freeze and thaw history of serum sample

EPIC, Europe16 199298 Matched Mostly 196C* 6 months No (incidence density

sampling)

Time of day at blood collection, menopausal status, phase of

cycle in premenopausal, subcohort

Guernsey, UK13 197791 Non-fasting 20C 2 years 1 year Menopausal status, phase of cycle in premenopausal

Janus Biobank, Norway20 198697 Non-fasting 25C All 4042 years 6 months Originally unmatched; matched sets created for this analysis

KKH, Denmark12 199397 Non-fasting 150C Same half-year No Known or probable postmenopausal

KP-OFAS, USA9 196471 Non-fasting 23C until1980 then 40C

Age 1 year Menopausal status

Malm/Ume, Sweden8 198598 Some matched,

some non-fasting

80 C 1 year 1 year Menopausal status

MCCS, Australia19 199094 Non-fasting


3/13

Articles


(PPHV) and Prospect-EPIC, from the Netherlands;11 Studyof Osteoporotic Fractures (SOF), from the USA;26 and theWomens Health Initiative, Observational Study (WHI-OS), from the USA.21

Table 1 summarises the study designs. Details of therecruitment of participants, informed consent, ethicsapprovals, and definitions of reproductive variables are inthe original publications. Collaborators were asked toprovide data on concentrations of IGF1 and IGFBP3, andalso on the sex hormones oestradiol and testosterone.Details of the assay methods for IGF1 and IGFBP3 areshown in table 2; 10 studies used serum, six used plasma,and one used both, but for convenience we refer to plasmaconcentrations throughout this paper. Details of the assaymethods for the sex hormones are in the originalpublications. Collaborators also provided data on

reproductive, anthropometric, and other characteristicsfor each woman in their study. Menopausal status at thetime of blood collection was defined on the basis ofquestions about the number of menstrual periods in theprevious year and details of any hysterectomy andovariectomy; the details varied slightly between studiesand are in the original publications. Women were excludedfrom the analyses if they were perimenopausal or ofunknown menopausal status, if they were using hormone-replacement therapy or other exogenous sex hormones atthe time of blood collection, or if data were missing fordates of birth, blood collection, or diagnosis (for cases).

Statistical analysisOf the 17 studies that contributed data, 11 provided data forwomen who were premenopausal at blood collection, and15 provided data for women who were postmenopausal atblood collection. Data from premenopausal women andpostmenopausal women in the same cohort were treatedas separate sub-cohorts. For the cross-sectional analyses,data were included from all women in the original studieswho had not been diagnosed with breast cancer (n=10 022);in the analyses of breast-cancer risk the data were arrangedin matched sets, and some potential controls were notmatched; therefore, the number of controls with data onIGF1 is less in the risk analyses (n=9428).

Concentrations of IGF1 and IGFBP3 were positivelyskewed; therefore, log-transformed concentrations wereused for all parametric analyses. Correlations between

IGF1 and IGFBP3 among premenopausal and post-menopausal controls were calculated using standardisedlog-transformed concentrations within each study, thestandardised values being calculated by subtracting themean log concentration and dividing by the standarddeviation of the log concentration. The associations ofIGF1 with risk factors for breast cancer were examinedin the controls using linear regression, calculatinggeometric mean concentrations and 95% CIs accordingto categories of these factors. Geometric means wereadjusted for study and age (age categories as in figure 1),as appropriate. F tests were used to test for heterogeneity

Sample IGF1 assay Intra-assay CV Inter-assay CV IGFBP3 assay Intra-assay CV Inter-assay CV

CLUE I and CLUE II, USA17 Phase 1 serum;

phase 2 plasma

ELISA (DSL) Serum 40%;

plasma 32%

Serum 59%;

plasma 59%

ELISA (DSL) Serum 59%

Plasma 37%

Serum 60%

Plasma 65%

EPIC, Europe16 Serum ELISA (DSL) 62% 162% ELISA (DSL) 72% 97%

Guernsey, UK13 Serum ELISA (DSL) Overall CV 66% RIA (in-house) Overall CV 39%

Janus Biobank, Norway20 Serum RIA (in-house) 8% 10% RIA (in-house) 5% 8%

KKH, Denmark12 Serum TRIFMA (DELFIA)


4/13

Articles


in the geometric mean hormone concentrationsbetween the categories of risk factors, and whereappropriate to test for trends across the categories, withthe ordered categories scored from 1 to the maximumnumber of categories. The heterogeneity betweenstudies in the associations of IGF1 with risk factors wasassessed by adding a study factor interaction term tothe model and using the F test to calculate itssignificance. A similar approach was used to assessheterogeneity according to menopausal status.

11 of the original studies contributing to the collaborativeanalysis had used matched nested case-control designs,and the remaining six had used unmatched controls or acase-cohort design (Janus,20 KKH Denmark,12 MCCS,19PLCO,22 SOF,26 and WHI-OS21). Some used densitysampling, meaning that an individual participant could

appear more than once in a data file; in order to avoiddouble-counting women in the cross-sectional analyses,we created a pooled dataset in which duplicateobservations were deleted, with the case observationretained where a participant appeared as both a casepatient and a control. We retained the original matchedsets where available, otherwise for the case-cohort studieswe created new matched sets in which each case wasmatched with up to four controls, matching by study,date of blood collection (plus or minus 24 months), age atblood collection (plus or minus 24 months) and, for Janusonly,20 county of residence.

Conditional logistic regression was used to calculate theodds ratio (OR) for breast cancer in relation to the plasma

concentrations of IGF1 and IGFBP3, categorising womenin each study according to the quintiles of hormoneconcentration for the controls in that study. Study-specificcut-points were used because the absolute concentrationsof IGF1 and IGFBP3 vary between studies because oflaboratory variation; further explanation of this approachis provided in previous publications.25,27 To provide asummary measure of risk, we calculated a linear trend byscoring the fifths of the plasma IGF1 or IGFBP3concentrations as 0, 025, 05, 075, and 1; under theassumption of linearity, a unit change in this trendvariable is equivalent to the OR comparing the highestwith the lowest fifth of hormone concentration.27Heterogeneity in linear trends among studies was

assessed using a test, calculating the statistic as thedifference between the sum of the model values foreach study and the model value from the all-studiesanalysis. We also used tests to examine whether therewas evidence of heterogeneity in the associations of IGF1with breast-cancer risk according to subgroups defined bymenopausal status at blood collection and other factors.

We examined the effect on the association with breast-cancer risk of adjusting IGF1 and IGFBP3 for each other.We also investigated the associations of IGF1 with breast-cancer risk after adjusting, one factor at a time, for variousestablished reproductive and hormonal risk factors forbreast cancer: age at menarche (


5/13

Articles


menopause (natural or surgical); time since menopause(04, 514, 15 years; natural postmenopausal womenonly); previous use of hormone-replacement therapy(never or ever). For postmenopausal women, we alsoinvestigated the associations of IGF1 with breast-cancerrisk after adjustment for plasma concentrations ofoestradiol and testosterone, and the associations of IGF1with breast-cancer risk with joint classification accordingto plasma oestradiol and testosterone concentrations.

All statistical tests were two-sided, and statisticalsignificance was set at the 5% level. All analyses weredone using Stata version 9.0.

Role of the funding sourceThe funding source had no role in study design, datacollection, data analysis, data interpretation, or the

writing of the report. The members of the writing team

had full access to all data in the study. The correspondingauthor had the final responsibility for the decision tosubmit for publication.

ResultsTable 3 shows the characteristics of the cases and controlsin each study. There were 4790 cases and 9428 matchedcontrols. Mean age at baseline ranged from 355 (SD 78)to 477 (SD 31) for premenopausal women, and from543 (SD 61) to 718 (SD 49) for postmenopausalwomen. Most women had had a full-term pregnancy, andmost postmenopausal women had reported a naturalmenopause. Across the studies, mean BMI ranged from231 (SD 35) to 284 (SD 63) kg/m, and the mediantime between blood collection and diagnosis ranged from1 (IQR 03) to 17 (IQR 818) years. Geometric mean

concentrations of IGF1 ranged from 193 (95% CI

Number Age (years) Nulliparous,

n (%)

Natural menopause,

n (%)

BMI (kg/m2) Years to

diagnosis*

Geometric mean (95% CI)

IGF1, nmol/L

Geometric mean (95% CI)

IGFBP3, nmol/L

Premenopausal at blood collection

CLUE I & CLUE II, USA17

Cases 87 431 (61) 5 (19%) 267 (59) 11 (822) 253 (233275) 680 (657704)

Controls 87 430 (60) 2 (8%) 259 (44) 247 (227268) 687 (663711)

EPIC, Europe16

Cases 409 455 (48) 66 (17%) 249 (43) 2 (14) 339 (329349) 1196 (11561239)

Controls 793 454 (48) 110 (15%) 253 (43) 337 (329344) 1166 (11371196)

Guernsey, UK13

Cases 69 406 (48) 6 (9%) 247 (37) 14 (1016) 215 (198234) 1593 (15111680)

Controls 200 405 (44) 24 (12%) 242 (39) 221 (213229) 1687 (16451730)

Janus Biobank, Norway20

Cases 323 405 (05) 4 (26) 270 (262278) 1768 (17231815)

Controls 639 406 (08) 262 (256268) 1794 (17591830)

KP-OFAS, USA9

Cases 89 357 (79) 18 (21%) 239 (53) 13 (818) 317 (295342) 813 (765864)

Controls 89 355 (78) 19 (22%) 236 (44) 305 (283328) 797 (749848)

Malm-Ume, Sweden8

Cases 141 471 (50) 9 (7%) 245 (40) 2 (14) 241 (226257) 1263 (12181309)

Controls 256 471 (49) 11 (5%) 248 (41) 236 (225247) 1248 (12111286)

MCCS, Australia19

Cases 160 468 (42) 38 (24%) 260 (49) 4 (27) 232 (220246) 1076 (10411112)

Controls 594 461 (40) 94 (16%) 262 (51) 241 (234247) 1109 (10911127)

Nurses Health Study, USA6,15

Cases 194 477 (31) 13 (7%) 245 (42) 7 (38) 274 (261288) 1276 (12311322)Controls 262 476 (31) 13 (5%) 254 (50) 270 (259281) 1292 (12541331)

Nurses Health Study II, USA18

Cases 231 436 (40) 53 (23%) 249 (51) 2 (14) 310 (298322) 1729 (16971763)

Controls 454 433 (38) 84 (19%) 253 (61) 308 (299318) 1726 (17011751)

NYU WHS, USA7

Cases 172 444 (48) 78 (51%) 239 (40) 5 (36) 268 (255281) 1156 (11101205)

Controls 483 442 (47) 175 (41%) 245 (45) 264 (257272) 1127 (11031152)

ORDET, Italy10

Cases 62 443 (50) 8 (13%) 241 (37) 1 (03) 209 (190230) 1276 (11921367)

Controls 239 438 (45) 25 (11%) 244 (41) 193 (184203) 1201 (11501254)

(Continues on next page)


6/13

Articles


Number Age (years) Nulliparous,n (%)

Natural menopause,n (%)

BMI (kg/m2) Years todiagnosis*

Geometric mean (95% CI)IGF1, nmol/L

Geometric mean (95% CI)IGFBP3, nmol/L

(Continued from previous page)

Postmenopausal at blood collection

CLUE I & CLUE II, USA17

Cases 73 606 (52) 5 (12%) 38 (86%) 265 (55) 7 (310) 220 (198244) 726 (698754)

Controls 73 603 (52) 6 (13%) 36 (77%) 253 (55) 203 (184224) 697 (668727)

EPIC, Europe16

Cases 677 601 (57) 84 (13%) 558 (82%) 272 (45) 2 (14) 281 (274289) 1199 (11641234)

Controls 1302 601 (57) 174 (14%) 1074 (82%) 268 (47) 272 (267277) 1153 (11281177)

Guernsey, UK13

Cases 47 589 (58) 12 (26%) 43 (91%) 256 (35) 13 (1015) 168 (151187) 1596 (14901709)

Controls 139 590 (58) 20 (14%) 132 (95%) 252 (35) 170 (161179) 1596 (15361657)

KKH, Denmark12

Cases 195 575 (40) 30 (15%) 164 (85%) 263 (48) 2 (13) 172 (166179) 1496 (14571537)

Controls 195 575 (40) 29 (15%) 160 (84%) 263 (45) 168 (162174) 1450 (14141487)

KP-OFAS, USA9

Cases 27 586 (59) 5 (21%) 27 (100%) 243 (22) 17 (818) 250 (222282) 779 (718846)

Controls 27 586 (59) 7 (29%) 27 (100%) 231 (35) 282 (249318) 774 (692866)

Malm-Ume, Sweden8

Cases 222 606 (51) 29 (14%) 199 (90%) 265 (42) 2 (03) 181 (172191) 1221 (11561290)

Controls 401 606 (51) 32 (9%) 377 (94%) 258 (44) 179 (170188) 1137 (10791199)

MCCS, Australia19

Cases 257 615 (52) 36 (14%) 205 (82%) 278 (47) 4 (26) 200 (191209) 1143 (11111175)

Controls 993 613 (51) 118 (12%) 760 (79%) 275 (50) 186 (182191) 1094 (10771111)

Nurses Health Study, USA6,15

Cases 239 615 (47) 13 (6%) 164 (73%) 270 (54) 3 (14) 203 (194213) 1288 (12411337)

Controls 470 616 (47) 34 (7%) 333 (74%) 264 (47) 201 (195208) 1306 (12721341)

NYU WHS, USA7

Cases 98 591 (35) 24 (30%) 78 (80%) 264 (42) 4 (35) 207 (194222) 1094 (10381154)

Controls 171 590 (35) 34 (23%) 138 (81%) 256 (48) 205 (194218) 1073 (10311118)

ORDET, Italy10

Cases 60 587 (49) 7 (12%) 50 (83%) 263 (39) 2 (13) 152 (138167) 1286 (12091368)

Controls 220 582 (49) 28 (13%) 169 (77%) 267 (42) 155 (147165) 1279 (12391320)

PLCO, USA22

Cases 386 638 (52) 34 (9%) 295 (77%) 281 (52) 3 (15) 274 (265284) 1603 (15731635)

Controls 468 636 (52) 35 (7%) 361 (77%) 275 (54) 268 (260277) 1606 (15791633)

PPHV, Netherlands11

Cases 77 545 (33) 77 (100%) 264 (42) 5 (38) 239 (219261) 1351 (12971407)

Controls 167 544 (38) 167 (100%) 264 (43) 220 (208233) 1308 (12761340)

Prospect-EPIC, Netherlands11

Cases 15 543 (61) 1 (7%) 15 (100%) 254 (41) 2 (23) 185 (153223) 1090 (9921198)

Controls 35 544 (62) 4 (11%) 35 (100%) 264 (47) 189 (167215) 1078 (10271133)

SOF, USA26

Cases 101 708 (47) 18 (18%) 87 (86%) 277 (53) 2 (14) 148 (140157) 1395 (13361456)

Controls 235 718 (49) 53 (23%) 203 (86%) 265 (43) 146 (139152) 1361 (13201404)

WHI-OS, USA21

Cases 379 658 (72) 55 (15%) 268 (71%) 284 (63) 3 (24) 173 (167179) 1444 (14151474)

Controls 436 645 (74) 65 (15%) 282 (65%) 277 (66) 171 (165177) 1450 (14251476)

Values are mean (SD) unless otherwise indicated, percentages exclude women with missing values. *Median (inter-quartile range) time between blood collection and diagnosis for cases. Numbers are for women

with an IGF1 measurement. EPIC=European Prospective Investigation into Cancer and Nutrition. KKH=Danish Diet, Cancer, and Health study. KP-OFAS=Kaiser Permanente-Orentreich Foundation Study.

MCCS=Melbourne Collaborative Cohort Study. NYU WHS=New York University Womens Health Study. ORDET=Study of H ormones and Diet in the Etiology of Breast Tumours. PLCO=Prostate, Lung, Colorectal,

and Ovarian Cancer Screening Trial. PPHV=Monitoring Project on Cardiovascular Disease Risk Factors; SOF=Study of Osteoporotic Fractures. WHI-OS=Womens Health Initiative, Observational Study.

Table 3: Participant characteristics by study and case-control status


7/13

Articles


184203) to 339 (329349) nmol/L for premenopausal

women, and from 146 (139152) to 282 (249318)nmol/L for postmenopausal women. Geometric meanconcentrations of IGFBP3 ranged from 680 (95% CI657704) to 1794 (17591830) nmol/L forpremenopausal women and from 697 (668727) to1606 (15791633) nmol/L for postmenopausal women.

Data on IGF1 and IGFBP3 were available for 10 022 and9889 controls, respectively (these numbers are largerthan those for the matched-set analyses because data forunmatched controls were included in the cross-sectionalanalyses). IGF1 and IGFBP3 were associated with eachother, with correlations of 038 and 050 (data not shown)in premenopausal and postmenopausal women,

respectively (both p


8/13

Articles


those in the highest fifth of IGFBP3 had an OR of 113(95% CI 099128) compared with women in the lowestfifth (test for trend p=0062).

Measurements of both IGF1 and IGFBP3 in completematched sets were available for 4727 cases and9196 controls. Adjustment of the association betweenIGF1 and breast-cancer risk for IGFBP3 had no significanteffect on the OR; the OR for linear trend before adjustmentwas 124 (95% CI 111138) and after adjustment was124 (110141). By contrast, adjustment of the associationbetween IGFBP3 and breast-cancer risk for IGF1 reducedthe OR for linear trend from 112 (95% CI 100126) to099 (087114). Analyses of breast-cancer risk in relationto the molar ratio of IGF1 to IGFBP3 showed a significantpositive association, but the magnitude was less than forthe analyses of IGF1; ORs in increasing fifths of the ratio

were 117 (95% CI 104132), 112 (099126),114 (101129) and 123 (108140) (test for trendp=0009). Further stratified analyses showed that IGFBP3was not associated with breast-cancer risk within thirds ofIGF1 (webappendix p 6).

Figure 4 shows the associations with breast cancer ofan 80 percentile difference in IGF1 according tosubgroups of various factors. The ORs varied accordingto oestrogen-receptor status; the OR for a linear trend inIGF1 was significant among oestrogen-receptor positivecases (OR 138, 95% CI 114168), but not for oestrogen-receptor negative tumours (OR 080, 057113) and thetest for heterogeneity was significant (p=0007). For theother factors there was no significant heterogeneity in

the association of IGF1 with breast-cancer risk.We examined the effect on the association of IGF1 withbreast-cancer risk of adjustment, one factor at a time, forheight, age at menarche, number of full-term pregnancies,age at first full-term pregnancy, use of hormonalcontraceptives, type of menopause (postmenopausal only),time since menopause (postmenopausal only), previoususe of hormonal therapy for menopause (postmenopausalonly), BMI (premenopausal and postmenopausal analysedseparately), plasma oestradiol concentration (post-menopausal only), plasma testosterone concentration(postmenopausal only), time of day of blood collection, andthe phase of the menstrual cycle at blood collection(premenopausal only). None of these adjustments altered

the OR for IGF1 and breast cancer by more than 2% (datanot shown) with the exception of adjustment fortestosterone in postmenopausal women, which reducedthe OR for an 80 percentile difference in IGF1 from 130(95% CI 108155) to 124 (104149).

The relationship of IGF1 with breast-cancer risk forpostmenopausal women was examined together with theassociations with oestradiol and testosterone (table 4). Inboth these joint analyses, the OR increased fairlyconsistently across thirds of concentrations of both IGF1and the sex hormone, with no significant interaction.

IGFBP3 was not significantly associated with breast-cancer risk in any study of premenopausal women, and

was significantly positively associated with risk in three

out of 15 studies of postmenopausal women (webappendixpp 3,4). There was significant heterogeneity in theassociation of IGFBP3 with breast-cancer risk accordingto oestrogen receptor status; IGFBP3 was non-significantlypositively associated with risk for oestrogen-receptor-positive breast cancer and non-significantly inverselyassociated with risk for oestrogen-receptor-negative breastcancer (test for heterogeneity p=0039; webappendix p 5).

DiscussionThe results of this collaborative analysis show that plasmaconcentrations of IGF1 are positively associated withbreast-cancer risk. The association is not substantially

Figure 3: Odds ratios (OR) for breast cancer associated with IGF1 concentrations in women who were

premenopausal at blood collection (A), and postmenopausal at blood collection (B)

The OR is the estimate of the linear trend for IGF1 obtained by replacing the categorical variables representing the

fifths of concentration in controls by a continuous variable scored as 0, 025, 05, 075, and 1. The black squares

indicate the ORs and the horizontal lines show the 95% CIs. The area of each square is proportional to the amount

of statistical information (inverse of the variance of the logarithm of the OR). The diamonds indicate the OR and

95% CI for all studies combined. Estimates are from conditional logistic regression on case-control sets matched

within each study. EPIC=European Prospective Investigation into Cancer and Nutrition. KKH=Danish Diet, Cancer, and

Health study. KP-OFAS=Kaiser Permanente-Orentreich Foundation Study. MCCS=Melbourne Collaborative Cohort Study.

NYU WHS=New York University Womens Health Study. ORDET=Study of Hormones and D iet in the Etiology of Breast

Tumours. PLCO=Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. PPHV=Monitoring Project on Cardiovascular

Disease Risk Factors. SOF=Study of Osteoporotic Fractures. WHI-OS=Womens Health Initiative, Observational Study.

Study OR (95% CI)Cases/controls

Ratio of median

concentrations:topbottom fifth

OR and 95% CI

CLUE I and CLUE II, USA 17 87/87 26 141 (055363)

EPIC, Europe16 409/793 22 113 (078163)

Guernsey, UK13 69/200 20 124 (056277)

Janus Biobank, Norway20 323/639 20 121 (081180)

KP-OFAS, USA9 89/89 24 269 (093778)

Malmo-Umea, Sweden8 141/256 26 115 (063208)

MCCS, Australia19 160/594 23 072 (044120)

Nurses Health Study, USA6,15 194/262 24 201 (106381)

Nurses Health Study II, USA18 231/454 22 081 (049133)

NYU WHS, USA7 172/483 23 134 (079228)

ORDET, Italy10 62/239 26 264 (103680)

All studies 1937/4096 118 (100140)

CLUE I and CLUE II, USA 17 73/73 26 220 (078619)

EPIC, Europe16 677/1302 23 148 (111197)

Guernsey, UK13 47/139 23 100 (038260)

KKH, Denmark12 195/195 19 146 (080266)

KP-OFAS, USA9 27/27 21 043 (010186)

Malmo-Umea, Sweden8 222/401 32 112 (068184)

MCCS, Australia19 257/993 26 164 (110246)

Nurses Health Study, USA6,15 239/470 26 111 (070176)

NYU WHS, USA7 98/171 24 132 (063277)

ORDET, Italy10 60/220 29 075 (034164)

PLCO, USA22 386/468 23 128 (087188)

PPHV, Netherlands11 77/167 24 273 (124599)

Prospect-EPIC, Netherlands11 15/35 27 113 (016783)

SOF, USA26 101/235 23 111 (056219)

WHI-OS, USA21 379/436 24 102 (068152)

All studies 2853/5332 130 (113149)

025 05 1 2 4

B

A


9/13

Articles


modified by menopausal status at blood collection or byIGFBP3 concentrations, but seems to be confined tooestrogen-receptor-positive tumours.

The strengths of our study are that the data and plasmasamples were all collected prospectively, that it includesalmost all the available data from published studiesworldwide, and that we were able to adjust for otherpotential risk factors, including endogenous sexhormones. A potential weakness is that the study designsand methods for measuring IGF1 and IGFBP3 and otherrisk factors were not standardised. IGF1 and IGFBP3concentrations varied substantially between studies, andthis is likely to reflect differences in assay methods. Ouranalysis allowed for this by defining study-specificquintiles of IGF1 and IGFBP3 concentrations. Thismethod assumes the true concentrations across the

quintiles are similar in all the studies, and if thisassumption is not correct then the estimates of ORsmight be biased.27 However, because heterogeneitybetween studies in risk estimates was not evident, thisassumption does seem reasonable. There was someevidence of heterogeneity between studies in some of thecross-sectional analyses, suggesting that caution shouldbe maintained in the interpretation of these analyses.

Previous studies have examined the associations ofIGF1 with other factors. Relevant publications are citedbelow, but it should be noted that some of these are fromthe studies contributing to this collaborative analysis.

IGF1 was inversely associated with age, with no obviousadditional decline in concentrations around age 50 years,

suggesting that menopause itself does not have a markedeffect on IGF1. This is consistent with previousobservations.5,2832

IGF1 was 7% higher in the tallest women than in theshortest women. No significant association of IGF1 withheight was noted in two previous analyses in adults, 29,33but these results might be compatible with the smallassociation noted in the current analysis. IGF1 washigher in women with a BMI of 250 to 274 kg/m thanin thinner or more overweight women, as describedpreviously.34 Most circulating IGF1 is produced by theliver, and it is possible that a low BMI is associated withlow IGF1 synthesis due to a relatively low supply ofnutrients to the liver, whereas obesity is associated with

low IGF1 synthesis in the liver due to compromised liverfunction.35

IGF1 was not associated with smoking, consistent withprevious observations.29,30,36 In relation to alcohol, IGF1was higher for women who drank a small amount thanfor those who drank no alcohol or those who drank 20 g ormore per day. Other observational studies had similarresults.29,3740 In randomised trials, 15 g/d of alcohol had noeffect on IGF1 in postmenopausal women, whereas 30 g/dcaused a decrease in IGF1 by 95% for premenopausalwomen and by 49% for postmenopausal women.41,42

IGF1 did not differ between women with or without afirst-degree family history of breast cancer. The inverse

Figure 4: Odds ratios (OR) for breast cancer associated with IGF1 concentration, according to menopausal

status at blood collection and other factors

The OR is the estimate of the linear trend obtained by replacing the categorical variables representing the fifths of

IGF1 concentration in controls by a continuous variable scored as 0, 025, 05, 075 and 1. Black squares indicate

the OR and the horizontal lines show the 95% CIs. The area of each square is proportional to the amount of

statistical information (inverse of the variance of the logarithm of the OR). The vertical dotted line indicates the OR

for all studies. Tests for heterogeneity are for the difference in the association of IGF1 with breast-cancer risk

between subgroups. Estimates are from conditional logistic regression on case-control sets matched within each

study. HRT=Hormone replacement therapy.

Factor and subset Cases/Controls OR (95% CI) OR and 95% CI

All studies

Menopausal status

Pre-menopausal

Postmenopausal

Age at diagnosis (years)


10/13

Articles


association we observed between IGF1 and age at menarchehas been noted previously.43,44 We observed a non-linearassociation between IGF1 and parity, with the lowestconcentrations for women who had four or more full-termpregnancies; previous studies have not reported anyassociations between IGF1 and parity.30,4345 IGF1 was alsopositively associated with age at first full-term pregnancy.IGF1 was not associated with type of menopause, but inpostmenopausal women was higher in those who hadmenopause most recently. IGF1 was marginally higher forwomen who had previously used hormonal contraceptivesthan for women who had not, but did not vary according toprevious use of hormonal therapy for menopause. Infuture analyses we will examine the relationships of IGF1with endogenous sex hormones.

The associations of IGF1 with height, age at menarche,

age at first full-term pregnancy, and time sincemenopause are compatible with the possibility that thesefactors affect breast-cancer risk partly through theirrelationships with IGF1.

IGF1 concentrations were positively associated withbreast-cancer risk, with a highly significant trend and noevidence of heterogeneity between studies. Women in thehighest fifth of IGF1 had a 28% higher risk of breastcancer than women in the lowest fifth. This associationdid not vary significantly according to menopausal statusat blood collection or according to the risk factors forbreast cancer examined, and was not attenuated byadjustment for other risk factors including IGFBP3,reproductive factors, and, for postmenopausal women,

BMI, oestradiol, and testosterone. If the association wasdue to an effect of preclinical tumours on IGF1 (reversecausality),46 then it would be expected to be weaker in thosewith a greater time interval between blood collection anddiagnosis. This was not the case, and the association washighly significant in patients from whom blood had beencollected at least 4 years before diagnosis.

A previous meta-analysis based on studies published upto 2006 concluded that the association of IGF1 with breast-cancer risk is limited to premenopausal women,47 but ouranalysis includes four large more recent studies with over1500 additional patients and shows a clear association ofIGF1 with breast-cancer risk in postmenopausal women.

Our analyses were all based on a single hormone

measure for each woman. Measurements of hormoneconcentrations are subject to largely random errorassociated with assay variation and fluctuations inplasma concentrations within individual women. Fivestudies have reported the reproducibility of IGF1 overperiods of between 1 and 15 years in samples of between13 and 138 women. The correlations (intra-class orSpearman) between baseline and repeat measuresranged from approximately 04 to 09 over 1 to15 years.10,17,19,4850 It is therefore likely that the observedassociation between IGF1 concentrations and breast-cancer risk is an underestimate of the true association,but more reproducibility data are required.

The association of IGF1 with breast-cancer risk wasconfined to oestrogen-receptor-positive tumours. Furtherwork is needed to examine the potential biological basisfor this observation. Laboratory studies have shown thatoestrogen increases IGF receptor levels in breast-cancercells,51 whereas in oestrogen-receptor-negative breast-cancer cells the levels of IGF1 receptor are decreased, andIGF1 is non-mitogenic.52

IGFBP3 was positively associated with breast-cancerrisk, but this association was weak, and was eliminatedby adjustment for IGF1, suggesting that the associationof IGFBP3 with risk is due to its positive correlationwith IGF1. It seems that, at least in the current dataset,the IGFBP3 measures do not add substantial

information in assessing the relationship of IGF1 withbreast-cancer risk. In addition to its role in transportingIGF1, laboratory studies have shown that IGFBP3 canhave direct effects on cell behaviour which can promoteapoptosis, but under other circumstances can actagainst apoptosis.53 Data on IGFBP-1 and IGFBP-2 havealso been contributed for our collaborative analyses, butcurrently there are too few data to provide robustanalyses. Better understanding of the roles of IGF-binding proteins as potential modulators of theassociation between IGF1 and breast-cancer risk mightcome from further data on IGFBP1 and IGFBP2, frommeasures of intact IGFBP3,54 or from measures ofbioavailable IGF1.55

The OR for IGF1 is smaller than the ORs for bothoestrogens and androgens and breast-cancer risk inpostmenopausal women, which have been shown in datamostly from the same epidemiological studies; highconcentrations of oestradiol and testosterone areassociated with around a doubling in breast-cancerrisk.25,5658 In our analyses, adjustment for oestradiol andtestosterone had little effect on the association of IGF1with breast-cancer risk for postmenopausal women, andthere was no evidence of an interaction between IGF1and oestradiol or testosterone in relation to breast-cancerrisk. Nevertheless, a better understanding of the jointeffects of hormones on breast-cancer risk is needed.59

Low Medium High

Cases/

controls

OR (95% CI) Cases/

controls

OR (95% CI) Cases/

controls

OR (95% CI)

Thirds of oestradiol (test of interaction: 24=231, p=0679)

Low 176/474 100 (ref) 205/471 120 (094153) 194/475 121 (094155)

Medium 16 0/396 1 25 ( 096163) 225/415 16 1 (126206) 196/424 1 38 ( 1071 77)

High 212/405 170 (132219) 240/404 189 (147243) 267/403 208 (162267)

Thirds of testosterone (test of interaction: 24=379, p=0436)

Low 125/438 100 (ref) 156/424 127 (096166) 157/378 147 (111195)

Medium 152/ 404 1 35 (1031 79) 181/389 16 5 (1252 17) 181/392 1 62 (1232 13)

High 175/345 188 (141249) 214/385 201 (154264) 227/426 191 (147249)

Table 4: Relationships of IGF1 with breast-cancer risk among postmenopausal women, according to

plasma concentrations of oestradiol and testosterone


11/13

Articles


The association of IGF1 with breast-cancer risk was notaltered by adjusting for age at menarche, parity, age at firstfull-term pregnancy, use of exogenous hormones, andBMI, suggesting that the relationship of IGF1 with breast-cancer risk is not confounded by these other risk factors.

This collaborative analysis has confirmed a positiveassociation between IGF1 and breast cancer risk. It isnot known whether this association is causal, but thereare plausible biological mechanisms that could explainsuch an effect.1,2 The magnitude of the observedassociation is modest, but the true association could besubstantially larger because of measurement error, andfurther work is needed to reliably quantify therelationship. If the association is causal then it mighthave important implications for prevention. Plasmaconcentrations of IGF1 are influenced by nutritional

factors such as energy and protein intake,60

and thepossibility of lowering breast-cancer risk by reducingIGF1 should be explored.

Contributors

Writing committee: Timothy J Key, Paul N Appleby, Gillian K Reeves(Cancer Epidemiology Unit, Nuffi eld Department of Clinical Medicine,University of Oxford, Oxford, UK); Andrew W Roddam (CancerEpidemiology Unit, Nuffi eld Department of Clinical Medicine,University of Oxford, Oxford, UK, and Global Biostatistics andEpidemiology, Amgen, Uxbridge, UK).Co-authors from CLUE I and CLUE II: KJ Helzlsouer (Department ofEpidemiology, The Johns Hopkins University Bloomberg School ofPublic Health, Baltimore, MD, USA and Prevention and ResearchCenter, Mercy Medical Center, Baltimore, MD, USA); AJ Alberg(Department of Epidemiology, The Johns Hopkins UniversityBloomberg School of Public Health, Baltimore, MD, USA, and HollingsCancer Center, Medical University of South Carolina, Charleston, SC,

USA); DE Rollison (Department of Interdisciplinary Oncology,H Lee Moffi tt Cancer Center, Tampa, FL, USA).Co-authors from EPIC: K Overvad (Department of Clinical Epidemiology,Aarhus University Hospital, Aarhus, Denmark); R Kaaks (DKFZ,Heidelberg, Germany); D Trichopoulos (Department of Epidemiology,Harvard School of Public Health, Boston, MA, USA, and Bureau ofEpidemiologic Research, Academy of Athens, Athens, Greece);F Clavel-Chapelon (Inserm ERI20 and Paris South University EA4045,Paris, France); P Vineis (ISI Foundation, Torino, Italy, and ImperialCollege, London, UK); M-D Chirlaque (Department of Epidemiology,Regional Health Authority, Murcia and CIBER Epidemiologa y SaludPblica [CIBERESP], Spain); PHM Peeters (Julius Center, UniversityMedical Center Utrecht, the Netherlands, and Imperial College, London,UK); S Rinaldi (International Agency for Research on Cancer, Lyon,France); E Riboli (Imperial College, London, UK).Co-authors from Guernsey: NE Allen (Oxford University, Oxford, UK);DS Allen, IS Fentiman (Guys Hospital, London, UK); JM Holly (Bristol

University, Bristol, UK).Co-authors from Janus Biobank: LJ Vatten (Trondheim University,Trondheim, Norway); JM Holly, D Gunnell (Bristol University, Bristol,UK); S Tretli (Norwegian Cancer Registry, Oslo, Norway).Co-authors from KKH: H Grnbk (Medical Department V, AarhusUniversity Hospital, Aarhus, Denmark); A Tjnneland (Danish CancerSociety, Copenhagen, Denmark); K Overvad (Department of ClinicalEpidemiology, Aarhus University Hospital, Aarhus, Denmark).Co-authors from KP-OFAS: R Krajcik (Burke-Cornell Medical ResearchInstitute, White Plains, NY, USA).Co-authors from Malm/Ume: J Manjer (Department of Surgery, MalmUniversity Hospital, Malm, Sweden); P Lenner (Department ofRadiation Sciences, Oncology, Ume University, Ume, Sweden); RKaaks (DKFZ, Heidelberg, Germany); G Hallmans (Department ofClinical Medicine and Public Health, Ume University Hospital, 901 85Ume, Sweden).

Co-authors from MCCS: L Baglietto, DR English, GG Giles, G Severi(Cancer Epidemiology Centre, Cancer Council of Victoria, and School ofPopulation Health, University of Melbourne, Melbourne, VA, Australia);

HA Morris (Hanson Institute, Adelaide, SA, Australia).Co-authors from Nurses Health Study and Nurses Health Study II:

SE Hankinson, ES Schernhammer, for the Nurses Health Study ResearchGroup (Channing Laboratory, Department of Medicine, Brigham andWomens Hospital and Harvard Medical School, and Department ofEpidemiology, Harvard School of Public Health, Boston, MA, USA).Co-authors from NYU WHS: K Koenig, A Zeleniuch-Jacquotte(Department of Environmental Medicine, New York University School ofMedicine, New York, NY, USA); AA Arslan, P Toniolo (Department ofObstetrics and Gynecology, New York University School of Medicine,New York, NY, USA); RE Shore (Radiation Effects Research Foundation,Hiroshima, Japan).Co-authors from ORDET: V Krogh, A Micheli, F Berrino (FondazioneIstituto Nazionale Tumori, Milan, Italy; P Muti, Istituto NazionaleTumori Regina Elena, Rome, Italy).Co-authors from PLCO: C Schairer, RG Ziegler (Division of CancerEpidemiology and Genetics, National Cancer Institute, Bethesda, MD,USA); CD Berg (Division of Cancer Prevention, National Cancer Institute,Bethesda, MD, USA); CA McCarty for the Prostate, Lung, Colorectal, andOvarian Cancer Screening Trial investigators (Center for HumanGenetics, Marshfield Clinic Research Foundation, Marshfield, WI, USA).Co-authors from PPHV and Prospect-EPIC: PHM Peeters (Julius Center,University Medical Center Utrecht, the Netherlands, and ImperialCollege, London, UK); HB Bueno-de-Mesquita (Center for Nutrition andHealth, National Institute of Public Health and the Environment,Bilthoven, the Netherlands).Co-authors from SOF: JA Cauley (Department of Epidemiology, Universityof Pittsburgh, PA, USA); Li Yung Lui, SR Cummings (San FranciscoCoordinating Center, California Pacific Medical Center, San Francisco,CA, USA); and the Study of Osteoporotic Fractures Research Group.Co-authors from Womens Health Initiative Observational Study: MJ Gunter,TE Rohan, HD Strickler (Department of Epidemiology and PopulationHealth, Albert Einstein College of Medicine, New York, NY, USA).

Conflict of interest statement

For the writing committee, TJK, PNA, and GKR declared no conflicts ofinterest. AWR declares current employment by Amgen.

Acknowledgments

Centralised pooling, checking, and analysis of data were supported byCancer Research UK. We thank the women who participated in thecollaborating studies, the research staff, the collaborating laboratoriesand the funding agencies in each of the studies.

References1 Pollak M. Insulin and insulin-like growth factor signalling in

neoplasia. Nat Rev Cancer2008; 8: 91528.

2 Samani AA, Yakar S, LeRoith D, Brodt P. The role of the IGFsystem in cancer growth and metastasis: overview and recentinsights. Endocr Rev2007; 28: 2047.

3 Furlanetto RW, DiCarlo JN. Somatomedin-C receptors and growtheffects in human breast cells maintained in long-term tissueculture. Cancer Res 1984; 44: 212228.

4 Pollak MN, Perdue JF, Margolese RG, Baer K, Richard M. Presence

of somatomedin receptors on primary human breast and coloncarcinomas. Cancer Lett1987; 38: 22330.

5 Peyrat JP, Bonneterre J, Hecquet B, et al. Plasma insulin-likegrowth factor-1 (IGF-1) concentrations in human breast cancer.Eur J Cancer1993; 29A: 49297.

6 Hankinson SE, Willett WC, Colditz GA, et al. Circulatingconcentrations of insulin-like growth factor-I and risk of breastcancer. Lancet1998; 351: 139396.

7 Toniolo P, Bruning PF, Akhmedkhanov A, et al. Serum insulin-likegrowth factor-I and breast cancer. Int J Cancer2000; 88: 82832.

8 Kaaks R, Lundin E, Rinaldi S, et al. Prospective study of IGF-I,IGF-binding proteins, and breast cancer risk, in northern andsouthern Sweden. Cancer Causes Control2002; 13: 30716.

9 Krajcik RA, Borofsky ND, Massardo S, Orentreich N. Insulin-likegrowth factor I (IGF-I), IGF-binding proteins, and breast cancer.Cancer Epidemiol Biomarkers Prev2002; 11: 156673.


12/13

Articles


10 Muti P, Quattrin T, Grant BJ, et al. Fasting glucose is a risk factorfor breast cancer: a prospective study. Cancer Epidemiol BiomarkersPrev2002; 11: 136168.

11 Keinan-Boker L, Bueno De Mesquita HB, Kaaks R, et al. Circulatinglevels of insulin-like growth factor I, its binding proteins -1,-2, -3,C-peptide and risk of postmenopausal breast cancer. Int J Cancer2003; 106: 9095.

12 Grnbk H, Flyvbjerg A, Mellemkjr L, et al. Serum insulin-likegrowth factors, insulin-like growth factor binding proteins, andbreast cancer risk in postmenopausal women. Cancer EpidemiolBiomarkers Prev2004; 13: 175964.

13 Allen NE, Roddam AW, Allen DS, et al. A prospective study ofserum insulin-like growth factor-I (IGF-I), IGF-II, IGF-bindingprotein-3 and breast cancer risk. Br J Cancer2005; 92: 128387.

14 Rinaldi S, Toniolo P, Muti P, et al. IGF-I, IGFBP-3 and breastcancer in young women: a pooled re-analysis of three prospectivestudies. Eur J Cancer Prev2005; 14: 49396.

15 Schernhammer ES, Holly JM, Pollak MN, Hankinson SE.Circulating levels of insulin-like growth factors, their bindingproteins, and breast cancer risk. Cancer Epidemiol Biomarkers Prev2005; 14: 699704.

16 Rinaldi S, Peeters PH, Berrino F, et al. IGF-I, IGFBP-3 and breastcancer risk in women: The European Prospective Investigationinto Cancer and Nutrition (EPIC). Endocr Relat Cancer2006;13: 593605.

17 Rollison DE, Newschaffer CJ, Tao Y, Pollak M, Helzlsouer KJ.Premenopausal levels of circulating insulin-like growth factor I andthe risk of postmenopausal breast cancer. Int J Cancer2006;118: 127984.

18 Schernhammer ES, Holly JM, Hunter DJ, Pollak MN,Hankinson SE. Insulin-like growth factor-I, its binding proteins(IGFBP-1 and IGFBP-3), and growth hormone and breast cancerrisk in The Nurses Health Study II. Endocr Relat Cancer2006;13: 58392.

19 Baglietto L, English DR, Hopper JL, Morris HA, Tilley WD,Giles GG. Circulating insulin-like growth factor-I and bindingprotein-3 and the risk of breast cancer. Cancer Epidemiol BiomarkersPrev2007; 16: 76368.

20 Vatten LJ, Holly JM, Gunnell D, Tretli S. Nested case-control studyof the association of circulating levels of serum insulin-like growth

factor I and insulin-like growth factor binding protein 3 with breastcancer in young women in Norway. Cancer Epidemiol BiomarkersPrev2008; 17: 2097100.

21 Gunter MJ, Hoover DR, Yu H, et al. Insulin, insulin-like growthfactor-I, and risk of breast cancer in postmenopausal women.

J Natl Cancer Inst2009; 101: 4860.

22 Schairer C, McCartyCA, IsaacsC, et al. Circulating insulin-likegrowth factor (IGF)-I and IGF binding protein (IGFBP)-3 levels andpostmenopausal breast cancer risk in the Prostate, Lung, Colorectal,and Ovarian Cancer Screening Trial (PLCO) Cohort. HormonesCancer2010; 1: 10011.

23 Firth SM, Baxter RC. Cellular actions of the insulin-like growthfactor binding proteins. Endocr Rev2002; 23: 82454.

24 Dupont J, Le Roith D. Insulin-like growth factor 1 and oestradiolpromote cell proliferation of MCF-7 breast cancer cells: newinsights into their synergistic effects. Mol Pathol2001; 54: 14954.

25 Endogenous Hormones and Breast Cancer Collaborative Group.Endogenous sex hormones and breast cancer in postmenopausal

women: reanalysis of nine prospectivestudies.J Natl Cancer Inst2002; 94: 60616.

26 Cauley JA, Lucas FL, Kuller LH, Stone K, Browner W,Cummings SR. Elevated serum estradiol and testosteroneconcentrations are associated with a high risk for breast cancer.Study of Osteoporotic Fractures Research Group. Ann Intern Med1999; 130: 27077.

27 Key TJ, Appleby PN, Allen NE, Reeves GK. Pooling biomarker datafrom different studies of disease risk, with a focus on endogenoushormones. Cancer Epidemiol Biomark Prev2010; 19: 96065.

28 Clemmons DR, Van Wyk JJ. Factors controlling bloodconcentration of somatomedin C. Clin Endocrinol Metab 1984;13: 11343.

29 Goodman-Gruen D, Barrett-Connor E. Epidemiology of insulin-likegrowth factor-I in elderly men and women. The Rancho BernardoStudy. Am J Epidemiol1997; 145: 97076.

30 Lukanova A, Toniolo P, Akhmedkhanov A, et al. A cross-sect ionalstudy of IGF-I determinants in women. Eur J Cancer Prev2001;10: 44352.

31 Brabant G, von zur Mhlen A, Wster C, et al. Serum insulin-likegrowth factor I reference values for an automatedchemiluminescence immunoassay system: results froma multicenter study. Horm Res 2003; 60: 5360.

32 Landin-Wilhelmsen K, Lundberg PA, Lappas G, Wilhelmsen L.Insulin-like growth factor I levels in healthy adults. Horm Res 2004;62 (suppl 1): 816.

33 Schernhammer ES, Tworoger SS, Eliassen AH, et al. Body shapethroughout life and correlations with IGFs and GH. Endocr RelatCancer2007; 14: 72132.

34 Lukanova A, Sderberg S, Stattin P, et al. Nonlinear relationship ofinsulin-like growth factor (IGF)-I and IGF-I/IGF-binding protein-3ratio with indices of adiposity and plasma insulin concentrations(Sweden). Cancer Causes Control2002; 13: 50916.

35 Vlzke H, Nauck M, Rettig R, et al. Association between hepaticsteatosis and serum IGF1 and IGFBP-3 levels in a population-basedsample. Eur J Endocrinol2009; 161: 70513.

36 Renehan AG, Atkin WS, ODwyer ST, Shalet SM. The effect of

cigarette smoking use and cessation on serum insulin-like growthfactors. Br J Cancer2004; 91: 152531.

37 Kaklamani VG, Linos A, Kaklamani E, Markaki I, Mantzoros C.Age, sex, and smoking are predictors of circulating insulin-likegrowth factor 1 and insulin-like growth factor-binding protein 3.

J Clin Oncol1999; 17: 81317.

38 Jernstrm H, Deal C, Wilkin F, et al. Genetic and nongeneticfactors associated with variation of plasma levels of insulin-likegrowth factor-I and insulin-like growth factor-binding protein-3 inhealthy premenopausal women. Cancer Epidemiol Biomarkers Prev2001; 10: 37784.

39 Holmes MD, Pollak MN, Willett WC, Hankinson SE. Dietarycorrelates of plasma insulin-like growth factor I and insulin-likegrowth factor binding protein 3 concentrations. Cancer EpidemiolBiomarkers Prev2002; 11: 85261.

40 DeLellis K, Rinaldi S, Kaaks RJ, Kolonel LN, Henderson B,Le Marchand L. Dietary and lifestyle correlates of plasma insulin-likegrowth factor-I (IGF-I) and IGF binding protein-3 (IGFBP-3): themultiethnic cohort. Cancer Epidemiol Biomarkers Prev2004; 13: 144451.

41 Lavigne JA, Wimbrow HH, Clevidence BA, et al. Effects of alcoholand menstrual cycle on insulin-like growth factor-I and insulin-likegrowth factor binding protein-3. Cancer Epidemiol Biomarkers Prev2004; 13: 226467.

42 Lavigne JA, Baer DJ, Wimbrow HH, et al. Effects of alcohol oninsulin-like growth factor I and insulin-like growth factor bindingprotein 3 in postmenopausal women. Am J Clin Nutr2005; 81: 50307.

43 Johansson H, Baglietto L, Guerrieri-Gonzaga A, et al. Factorsassociated with circulating levels of insulin-like growth factor-I andinsulin-like growth factor binding protein-3 in 740 women at riskfor breast cancer. Breast Cancer Res Treat2004; 88: 6373.

44 Probst-Hensch NM, Wang H, Goh VH, Seow A, Lee HP, Yu MC.Determinants of circulating insulin-like growth factor I andinsulin-like growth factor binding protein 3 concentrations in acohort of Singapore men and women. Cancer Epidemiol BiomarkersPrev2003; 12: 73946.

45 Morimoto LM, Newcomb PA, White E, Bigler J, Potter JD. Variationin plasma insulin-like growth factor-1 and insulin-like growth factor

binding protein-3: personal and lifestyle factors (United States).Cancer Causes Control2005; 16: 91727.

46 Brabant G, Wallaschofski H. Normal levels of serum IGF-I:determinants and validity of current reference ranges.Pituitary2007; 10: 12933.

47 Renehan AG, Harvie M, Howell A. Insulin-like growth factor(IGF)-I, IGF binding protein-3, and breast cancer risk: eight yearson. Endocr Relat Cancer2006; 13: 27378.

48 Kaaks R, Toniolo P, Akhmedkhanov A, et al. Serum C-peptide, insulin-like growth factor (IGF)-I, IGF-binding proteins, and colorectal cancerrisk in women.J Natl Cancer Inst2000; 92: 1592600.

49 Missmer SA, Spiegelman D, Bertone-Johnson ER, Barbieri RL,Pollak MN, Hankinson SE. Reproducibility of plasma steroidhormones, prolactin, and insulin-like growth factor levels amongpremenopausal women over a 2- to 3-year period.Cancer Epidemiol Biomarkers Prev2006; 15: 97278.


13/13

Articles

50 Borofsky ND, Vogelman JH, Krajcik RA, Orentreich N. Utility ofinsulin-like growth factor-1 as a biomarker in epidemiologic studies.Clin Chem2002; 48: 224851.

51 Maor S, Mayer D, Yarden RI, et al. Estrogen receptor regulates insulin-like growth factor-I receptor gene expression in breast tumor cells:involvement of transcription factor Sp1. Endocrinol2006; 191: 60512.

52 Surmacz E, Bartucci M. Role of estrogen receptor alpha inmodulating IGF-I receptor signaling and function in breast cancer.

J Exp Clin Cancer Res 2004; 23: 38594.

53 Perks CM, Holly JM. IGF binding proteins (IGFBPs) and regulationof breast cancer biology.J Mammary Gland Biol Neoplasia 2008;13: 45569.

54 Rinaldi S, Kaaks R, Zeleniuch-Jacquotte A, et al. Insulin-like growthfactor-I, IGF binding protein-3, and breast cancer in young women:a comparison of risk estimates using different peptide assays.Cancer Epidemiol Biomarkers Prev2005; 14: 4852.

55 Brugts MP, Ranke MB, Hofland LJ, et al. Normal values ofcirculating insulin-like growth factor-I bioactivity in the healthypopulation: comparison with five widely used IGF-I immunoassays.

J Clin Endocrinol Metab 2008; 93: 253945.

56 Missmer SA, Eliassen AH, Barbieri RL, Hankinson SE.Endogenous estrogen, androgen, and progesterone concentrationsand breast cancer risk among postmenopausal women.

J Natl Cancer Inst2004; 96: 185665.57 Kaaks R, Rinaldi S, Key TJ, et al. Postmenopausal serum androgens,oestrogens and breast cancer risk: the European prospectiveinvestigation into cancer and nutrition. Endocr Relat Cancer2005;12: 107182.

58 Eliassen AH, Missmer SA, Tworoger SS, et al. Endogenous steroidhormone concentrations and risk of breast cancer amongpremenopausal women.J Natl Cancer Inst2006; 98: 140615.

59 Trichopoulos D, Bamia C, Lagiou P, Trichopoulou A.Are mammotropic hormones mainly permissive for thedevelopment of breast cancer? Int J Cancer2006; 118: 286365.

60 Crowe FL, Key TJ, Allen NE, et al. The association between diet andserum concentrations of IGF-I, IGFBP-1, IGFBP-2, and IGFBP-3 inthe European Prospective Investigation into Cancer and Nutrition.Cancer Epidemiol Biomarkers Prev2009; 18: 133340.

Documents

Insulin-like growth factor 1 (IGF1), IGF binding protein 3(IGFBP3), and breast cancer risk: pooled individual dataanalysis of 17 prospective studies