Circulating levels of insulin-like growth factor-I and risk of ovarian cancer†
The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute.
Insulin-like growth factor (IGF)-I, a mitogenic and anti-apoptotic peptide, has been implicated in the development of several cancers. We hypothesized that high circulating IGF-I concentrations may be associated with an increased risk of ovarian cancer. A case–control study was nested within 3 prospective cohorts in New York (USA), Umeå (Sweden) and Milan (Italy). One hundred thirty-two women with primary invasive epithelial ovarian cancer diagnosed at least 1 year after blood donation were case subjects. For each case, 2 control subjects were selected, matching the case subject on cohort, menopausal status, age and date of recruitment (n = 263). Only women who did not use exogenous hormones at blood donation were included in the study. There was no association between IGF-I concentrations and ovarian cancer risk in the study group as a whole. In analyses restricted to subjects who had developed ovarian cancer at a young age (<55), circulating IGF-I was directly and strongly associated with ovarian cancer risk (OR = 4.97; 95% CI = 1.22–20.2 for the top vs. the bottom IGF-I tertile after adjustment for parity, BMI categories and smoking). There was no significant association of IGF binding protein-3 with ovarian cancer risk. We found a strong direct relationship between circulating IGF-I levels and risk of developing ovarian cancer before age 55. Additional, larger studies of this association are needed to provide more precise estimates of effect. © 2002 Wiley-Liss, Inc.
Several hypotheses about ovarian carcinogenesis have been proposed, implicating incessant ovulation,1 gonadotropins,2, 3 steroid hormones,3, 4 inflammatory processes5 or retrograde carcinogen transportation6 as key etiological factors. The existing epidemiologic evidence gives support to the hypothesis postulating that frequent ovulation is related to increased ovarian cancer risk, but additional hormonal factors such as steroid hormones, gonadotropins, insulin or insulin-like growth factor (IGF)-I are also believed to play a role.7
Recent epidemiological studies have related elevated circulating levels of IGF-I, measured as absolute concentrations, or relative to levels of IGF-binding protein-3 (IGFBP-3), to increased risk of cancers of the breast, prostate and colon.8, 9, 10, 11, 12, 13 The principal mechanisms by which IGF-I is believed to influence cancer risk involve increased cell proliferation and inhibition of apoptosis,14 effects that have been demonstrated in many cell types, including normal and neoplastic epithelial ovarian cells.15, 16, 17 Increased proliferation rates and the impairment of apoptosis may allow cells that have harbored mutations in proto-oncogenes and tumor suppressor genes to survive and expand clonally. Initial data linking the IGF-I system to ovarian cancer come from observations that IGF-I levels are higher in cystic fluid from invasive malignant ovarian neoplasms than in cystic fluid from benign neoplasms.18 IGF-I receptors are present in surgical specimens from primary or metastatic ovarian tumors17 and carcinoma cells derived from fresh, untreated ovarian cancers express all major components of the IGF-I system, IGF-peptides, type-I IGF-I receptor, IGFBPs, and demonstrate functional responses to exogenous IGF.19
IGFBP-3, the major carrier protein of IGF-I in the circulation, is known to reduce IGF-I tissue availability and to decrease IGF-I biological activity, and has also been shown to have an IGF-I-independent inhibitory effect on cell growth.14 In 1 small study, IGFBP-3 levels were found to be decreased in patients with epithelial ovarian cancer compared to women with benign lesions or healthy controls20 and lower levels of IGFBP-3 in surgically removed ovarian tumor tissues, were associated with unfavorable prognosis, such as large size and advanced stage of the residual tumor.21
We describe findings from a pooled case–control study nested within 3 prospective studies in New York (USA), Umeå (Sweden) and Milan (Italy). The study's leading hypothesis was that the risk of ovarian cancer increases with increasing circulating levels of IGF-I or decreased circulating levels of IGFBP-3 in the years preceding clinical diagnosis.
MATERIAL AND METHODS
The collaborating cohorts have been described in detail previously9, 13, 22 and included the New York University Women's Health Study (NYUWHS), the Northern Sweden Health and Disease Study (NSHDS), and the Study of Hormones and Diet in the Etiology of Breast Cancer (ORDET). Main characteristics of these cohorts are presented in Table I.
Table I. Characteristics of The Cohort Studies Included in The Pooled Analysis of IGF-I, IGFBP-3 and Ovarian Cancer
|NYUWHS (New York, USA)||Mammographic screening clinic||1985–1991||14,275||32–70||Dec. 1998||74||148||55.8 (35.5–65.6)|
|NSHDS (Umeå, Sweden)||General population||1986–present||43,268||30–70||Sep. 2000||42||83||59.4 (30.1–69.5)|
|ORDET (Milan, Italy)||Healthy volunteers and women attending breast cancer prevention unit||1987–1992||10,788||35–70||Jan. 1997||16||32||49.8 (36.1–61.8)|
At recruitment, subjects in the NYUWHS and the NSHDS were asked to complete a self-administered questionnaire to collect demographic, lifestyle and medical information and to donate a venous blood sample. The baseline questionnaires in NYUWHS and ORDET cohorts included a detailed section on reproductive history. Information about exogenous hormone use was collected at baseline from ORDET subjects and from follow-up questionnaires for NYUWHS. In the NSHDS, a reproductive history and exogenous hormone use questionnaire was administered prospectively to 47% of the subjects and a similar questionnaire was sent out retrospectively to all women, selected to participate in the study (response rate of 95%). Data were also collected from medical records of the NSHDS case subjects; for a few deceased women (n = 13) these records were the only source of information about reproductive history and hormone use available. The NSHDS and ORDET components of the study included only Caucasian subjects, while in NYUWHS information about race was available for 78% of the subjects included in the present study. Among them 88% of the women indicated that they were non-Hispanic whites, 6% as black, 3% as Hispanic, 3% as other ethnicity.
Identification of ovarian cancer cases and selection of control subjects
Case subjects were cohort members with primary, invasive epithelial ovarian cancer that was diagnosed at least 1 year after the initial blood donation, without preceding cancer diagnosis and identified within the parent cohort by the date of the last complete follow-up. Follow-up in the NYUWHS consists of periodic contact by mail and telephone, as well as record linkages with state-wide tumor registries (New York, New Jersey, Connecticut and Florida) and the U.S. National Death Index. It was estimated that follow-up is approximately 95% complete for breast cancer cases diagnosed during the study period.23 In the NSHDS, ovarian cancer cases were identified through linkage with regional and national cancer registries and the vital status of the study participants was ascertained by linkage with the regional and national registries for all-cause mortality. The ORDET database was linked to the local cancer registry (Lombardy Cancer Registry) to identify ovarian cancer cases and to the regional residents' files to check the vital status of cohort members. A total of 132 ovarian cancer cases were included for our study from the 3 parent cohorts (Table I).
Among tumors with histological verification (n = 115), 48% were of serous (n = 55), 12% of endometrioid (n = 14), 11% of mucinous (n = 13) and 7% of clear cell types (n = 8), whereas the remaining 22% were classified as carcinoma not-otherwise specified (n = 17), or as mixed (n = 1) or undifferentiated (n = 4) carcinomas.
For each case subject, 2 control subjects were selected at random among appropriate risk sets. The risk set for a given case included all cohort subjects alive, free of cancer, and who have not had a bilateral ovariectomy, and matched the case on cohort, menopausal status at enrollment, age (±6 months), date at recruitment (±3 months) and, for premenopausal subjects, day of the menstrual cycle at blood donation (for NYUWHS and ORDET subjects only). Potential case and control subjects from the NSHDS who reported use of exogenous hormones at the time of blood donation were not considered eligible, whereas in the NYUWHS and ORDET cohorts, subjects reporting hormone use at baseline were not recruited. The matching for menopausal status was confirmed by follicle-stimulating hormone (FSH) measurement. A total of 263 control subjects were identified and included in our study (Table I).
The Ethical Review Boards of New York University School of Medicine, the University of Umeå, Istituto Nazionale Tumori in Milan and the International Agency for Research on Cancer, in Lyon, France, reviewed and approved the present study.
The hormone analyses were carried out on serum samples obtained from the NYUWHS subjects and plasma samples, in which heparin was used as an anticoagulant, obtained from the NSHDS and ORDET subjects.
The laboratory analyses were carried out at the Hormone Laboratory at the International Agency for Cancer Research, Lyon, France. Samples from case subjects and their matched control subjects were always analyzed in the same assay kit and on the same day. Laboratory personnel were unable to distinguish among case and control samples. To control the quality of the peptide measurements, aliquots from a pool of quality control plasma and 3 standard sera were inserted randomly in each batch. Peptide concentrations were measured by double-antibody, immunoradiometric assays with reagents from Diagnostic System Laboratories, (Webster, Texas). Total IGF-I was measured after acid-ethanol precipitation of IGFBPs. The mean intra-batch coefficients of variation were 1.5% for an IGF-I concentration of 150 ng/ml and 4.8% for an IGFBP-3 concentration of 3,800 ng/ml. The mean inter-batch coefficients of variation were 3.4% for an IGF-I concentration of 150 ng/ml, and 7.2% for an IGFBP-3 concentration of 3800 ng/ml. Among 86 subjects from the NYUWHS cohort who had a second blood sample taken 11–60 months after the first blood donation (mean duration between visits 18 months), intra-class correlations (calculated from variance components estimated by the ‘MIXED’ procedure in the Statistical Analysis System [SAS Institute, Cary, NC])24 between repeated peptide measurements were 0.86 for IGF-I, and 0.93 for IGFBP-3. These results confirm previous findings that single serum measurements of IGF-I or IGFBP-3 are representative of the peptide levels for a period of at least 1 year.11, 25, 26 FSH levels were measured by immunoradiometric assay with reagents from Diagnostic System Laboratories, (Webster, Texas). For a FSH concentration of 15 μ IU/ml, the mean intra-batch and the inter-batch coefficients of variation were 4.2% and 12.6%, respectively.
An analysis of covariance was used to investigate subgroup differences in mean IGF-I and IGFBP-3 levels (by case/control status, study cohort, oral contraceptives [OC] use, parity, diabetes diagnosis, family history of breast cancer, BMI and smoking categories), adjusting for potential confounders. This analysis was carried out using the Generalized Linear Models (GLM) SAS procedure. Pearson's partial correlations, adjusted for age and study cohort, were calculated on the basis of sums of squares error estimates, also obtained by the SAS “GLM” procedure.
Odds ratios for disease by quartile levels of the hormone variables were estimated by conditional logistic regression models, using the SAS “PHREG” procedure. Cohort-specific quartile cut-off points were determined according to variable distributions in the cases and controls combined. Likelihood ratio tests were used to assess linear trends in ORs over the quartiles, giving quantitative scores of 1–4 to the 4 levels. All statistical tests and corresponding p-values were 2-sided, and p-values < 0.05 were considered statistically significant. The potential confounding effects of ages at menarche and menopause, parity, BMI (in categories ≤23, 23–25, 25–30, ≥30), use of OC and hormone replacement therapy, smoking, physical activity, education and selected medical conditions (diabetes, family history of breast cancer and hypertension) were examined by including these factors in the conditional logistic regression models.
At recruitment, 33% of the case subjects (44 women) were premenopausal. Mean age at cancer diagnosis was 59.5 ± 8.8 years (median 61.1 years). The time between recruitment and cancer diagnosis ranged from 12 months to 13.3 years, with an average of 5.7 years (median 5.4 years). Eighty-six percent of the case subjects were diagnosed at least 2 years after cohort recruitment, whereas 73% were diagnosed after more than 3 years.
Case subjects tended to report less frequently a history of full-term pregnancy (68% vs. 82%, p < 0.01), to be leaner (BMI 25.0 vs. 26.0, p = 0.03), and to have reached their menopause at an older age (50.2 vs. 48.8 years, p = 0.05) than controls (Table II). There were no significant differences between case and control subjects in their mean age at menarche, frequency of OC use, smoking habits, diagnosis of diabetes or family history of breast cancer.
Table II. Factors For Cases And Control Subjects1
|BMI kg/m2 (range)||25.0 (24.3–25.7)||26.0 (25.5–26.4)||0.032|
|Age at menarche (range)||12.9 (12.6–13.2)||12.9 (12.7–13.1)||0.732|
|Full-term pregnancy, % (n)|| || ||0.0043|
| Never||30 (40)||18 (47)|| |
| Ever||68 (90)||82 (215)|| |
| Missing||2 (2)||0 (1)|| |
|OC use|| || ||0.143|
| Never||62 (82)||55 (144)|| |
| Ever||23 (31)||30 (79)|| |
| Missing||14 (19)||15 (40)|| |
|Age at menopause (range)||50.2 (49.0–51.3)||48.8 (48.0–49.6)||0.052|
|Smoking, % (n)|| || ||0.133|
| Never||51 (67)||41 (107)|| |
| Ex-smoker||23 (31)||25 (67)|| |
| Current smoker||10 (13)||13 (34)|| |
| Missing||16 (21)||21 (55)|| |
Mean IGF-I and IGFBP-3 levels differed between cohorts. In matched sets where the case was diagnosed before age 55, IGF-I concentrations were highest in ORDET subjects, intermediate in NYUWHS subjects and lowest in the NSHDS (Table III).
Table III. Age at Cancer Diagnosis of The Case Subjects1
|Diagnosed before age 55|
| NYUWHS||19/38||228.4 (203.6–253.3)||194.9 (177.4–212.5)||3,962 (3,709–4,215)||3,770 (3,592–3,949)|
| NSHDS||13/25||185.7 (153.2–218.2)||182.0 (158.6–205.4)||3,208 (2,472–3,943)||3,228 (2,741–3,716)|
| ORDET||9/18||263.6 (204.5–322.8)||216.0 (174.1–257.8)||3,971 (3,575–4,368)||3,719 (3,438–3,999)|
| All subjects||41/81||226.8 (206.6–247.0)||200.0 (185.4–214.6)||3,725 (3,467–3,984)||3,579 (3,397–3,762)|
|Diagnosed at or after age 55|
| NYUWHS||55/110||162.7 (144.9–180.6)||165.0 (152.4–177.6)||3,917 (3,721–4,113)||3,881 (3,742–4,019)|
| NSHDS||29/57||137.6 (121.0–154.1)||132.7 (121.0–144.4)||3,795 (3,383–4,207)||3,822 (3,543–4,102)|
| ORDET||7/14||141.0 (86.4–195.5)||153.8 (115.2–192.4)||3,651 (3,030–4,272)||3,843 (3,404–4,282)|
| All subjects||91/181||147.7 (133.3–162.1)||148.5 (137.1–159.9)||3,832 (3,633–4,031)||3,831 (3,676–3,987)|
There was a moderate direct correlation between levels of IGF-I and IGFBP-3 (r = 0.46, p < 0.0001) and there was an inverse correlation of IGF-I with age (r = −0.42, p < 0.0001). Only very weak correlations were observed between levels of IGF-I and IGFBP-3 with height and BMI (data not shown) and the mean levels of both peptides were similar across BMI categories. Mean IGF-I levels were the highest in overweight women with BMI between 25 and 30, compared to women with BMI ≤23 and BMI ≥30, confirming previous observations of a non-linear relationship between BMI and IGF-I.27 After adjustment for age at sampling, study cohort and case–control status, mean IGF-I and IGFBP-3 levels showed no significant differences according to menopausal status at baseline, ever-use of OC, pre-existing diabetes diagnosis or family history of breast cancer. Women reporting a history of full-term pregnancy tended to have lower mean IGF-I (170.4 vs. 184.6 ng/ml, p < 0.08) and IGFBP-3 (3,745 vs. 3,933 ng/ml, p < 0.08) levels than nulligravid women. Smokers tended to have a slightly higher mean IGF-I level compared to non-smokers but the difference did not reach statistical significance (182.0 vs. 173.1, p < 0.28).
In the 3 cohorts combined, mean IGF-I concentrations were only slightly higher (4.6%) in the case than in the control subjects (176.1 vs. 168.3 ng/ml IGF-I, p = 0.24). For women who had developed cancer before age 55 or in those who were age 50 or less at recruitment this difference increased to 13% (226.8 vs. 200.0 ng/ml, p = 0.03) and 18% (240.0 vs. 203.5 ng/ml, p < 0.004), respectively. The difference in mean IGF-I levels between case and control subjects was evident for the NYUWHS and ORDET subjects, whereas in NSHDS no difference in mean IGF-I in case and control subjects were observed (Table III). There were no significant differences in mean IGFBP-3 levels in case and control subjects (Table III).
Overall, for all age groups combined, there were no significant associations between IGF-I and IGFBP-3 concentrations and ovarian cancer risk, either before or after adjustment for potential confounders (OR [95% CI] 1.10 [0.56–2.19], 0.80 [0.39–1.62], 1.38 [0.68–2.84], p = 0.57 for quartiles of IGF-I and 1.23 [0.62–2.45], 1.14 [0.57–2.27], 0.84 [0.41–1.72], p = 0.60 for quartiles of IGFBP-3). Adjustment of the IGF-I models for IGFBP-3, and of the IGFBP-3 models for IGF-I, did not alter these results. A similar lack of association between peptide hormone levels and ovarian cancer risk was observed in the women who were 55 or older at ovarian cancer diagnosis (Table IV).
Table IV. Risk of Ovarian Cancer by Tertiles of IGF-I and IGFBP-31
|Index case subject diagnosed before age 55|
| IGF-I (cases/controls)||9/29||12/29||20/23|| |
| Unadjusted2||1.00||1.58 (0.48–5.18)||4.73 (1.31–17.1)||< 0.02/0.02|
| Adjusted3||1.00||1.79 (0.47–6.78)||4.97 (1.22–20.2)||< 0.03/< 0.02|
| Adjusted4||1.00||1.72 (0.45–6.51)||4.98 (1.21–20.6)||< 0.03/0.02|
| IGFBP-3 (cases/controls)||6/31||17/23||16/27|| |
| Unadjusted2||1.00||3.48 (1.22–9.94)||2.59 (0.94–7.13)||< 0.08/0.33|
| Adjusted3||1.00||3.14 (0.96–10.2)||1.90 (0.63–5.75)||< 0.31/< 0.54|
| Adjusted4||1.00||2.79 (0.76–10.2)||0.87 (0.23–3.25)||0.79/0.77|
|Index case subject diagnosed at or after age 55|
| IGF-I (cases/controls)||28/60||32/57||31/64|| |
| Unadjusted2||1.00||1.21 (0.65–2.26)||1.04 (0.55–1.95)||< 0.92/< 0.90|
| Adjusted3||1.00||1.15 (0.57–2.32)||1.04 (0.51–2.10)||< 0.92/0.83|
| Adjusted4||1.00||1.02 (0.47–2.21)||0.94 (0.41–2.18)||< 0.88/0.78|
| IGFBP-3 (cases/controls)||26/58||30/55||29/61|| |
| Unadjusted2||1.00||1.16 (0.63–2.14)||1.02 (0.52–2.00)||0.93/< 0.95|
| Adjusted3||1.00||1.13 (0.57–2.22)||0.98 (0.46–2.08)||0.98/0.76|
| Adjusted4||1.00||1.19 (0.58–2.45)||1.09 (0.46–2.62)||0.83/< 0.93|
Among women, who were younger than 55 when diagnosed with ovarian cancer (of whom 35 case and 62 control subjects where premenopausal at blood donation), there was a direct association between IGF-I concentration and ovarian cancer risk (Table IV). This increase in risk, however, was confined to the highest tertile of IGF-I concentrations. Further adjustments for levels of IGFBP-3 only slightly influenced these results (Table IV). A similar direct association was observed in the group of women who were age 50 or younger at recruitment [OR (95% CI) 1.03 (0.28–3.83) and 4.82 (1.26–18.5), p = 0.02 for the second and top IGF-I tertile]. There was, however, a significant (about 80%) overlap between these subgroups of young women. Restriction of these analyses to subjects, diagnosed 2 or more years after blood donation, did not influence the strength and the direction of the association between IGF-I levels and ovarian cancer diagnosed before age 55 (data not shown). Due to small number of observations per cohort, the study lacked the statistical power to calculate meaningful tests for interaction between IGF-I concentrations and cohort sub-populations.
Among women, who were younger than 55 when diagnosed with ovarian cancer, IGFBP-3 levels appeared to be directly related to an increase in ovarian cancer risk, but there was no evident trend of increase and all confidence intervals included unity. Adjustment of the IGFBP-3 models for levels of IGF-I reduced considerably all point estimates (Table IV).
To our knowledge this is the first epidemiological study to investigate the relationship between prediagnostic circulating IGF-I and IGFBP-3 levels and risk of ovarian cancer. We observed a strong direct association of IGF-I levels with ovarian cancer risk among women who were younger than age 55 at cancer diagnosis. No association of IGF-I with ovarian cancer risk was observed in the study population as a whole, or in women with cancer diagnoses after age 55. IGFBP-3 did not seem to be related to ovarian cancer risk in our study population.
Our study is part of an on-going collaborative project between 3 prospective cohorts in New York (USA), Umeå (Sweden) and Milan (Italy) on endogenous hormones and ovarian cancer risk. Pooling the data gave us the possibility to investigate prospectively this comparatively less frequent cancer with reasonable statistical power. One advantage of the nested case–control study within a prospective cohort is that cancer case and control subjects originate from the same, well-defined source population, thereby minimizing the risk of control selection biases. Furthermore, prospective studies have the advantage that blood samples are obtained before the clinical manifestation of the disease, so that the observed case–control difference in blood hormone levels is unlikely to be an effect of the disease. The direct association of IGF-I with ovarian cancer risk in women younger than age 55 at ovarian cancer diagnosis did not disappear after exclusion of case subjects diagnosed within 2 years after blood donation, suggesting that the observed case–control differences were not the result of the presence of latent tumors, sufficiently advanced in stage to influence circulating hormone levels.
The mean IGF-I levels differed between the cohorts. We believe that these differences were most likely due to differences in the blood collection, processing or storage conditions among the 3 cohort studies, and used cohort-specific cut-off points to estimate relative risks of ovarian cancer by tertiles of IGF-I. When applying study-wide cut-off points, however, the relative risks for ovarian cancer for IGF-I tertiles in the younger women were of similar magnitude.
IGF-I may be involved in ovarian carcinogenesis through several mechanisms. It may exert a direct effect by increasing cell proliferation and inhibition of apoptosis,17 and experimental studies have indeed shown that malignant transformation of ovarian epithelial cells (the cells from which ovarian cancer is believed to originate) can be induced by overexpression of the IGF-1 receptor.28 These mitogenic and anti-apoptotic effects of IGF-I might be particularly relevant during ovulation related tissue remodeling of the surface epithelium.29 In addition, IGF-I may also influence ovarian cancer risk through modulation of the synthesis and bioavailability of sex steroid hormones, which have been implicated in ovarian cancer etiology.4, 7 IGF-I has been shown to enhance the activity and expression of enzymes involved in the synthesis of androgens in ovarian stromal tissue in vitro7, 30, 31, 32 and to downregulate the hepatic production of SHBG, which determines the free, bioavailable fraction of steroid hormones.33
The increase in relative risk exclusively among women with diagnosis of ovarian cancer before age 55 may reflect an interaction between elevated IGF-I and the ovarian steroidogenic or ovulatory activity before menopause. Alternatively, it is possible that mean IGF-I levels decline more rapidly with age among women whose levels were initially high, and who were at increased ovarian cancer risk due to the direct effects of IGF-I irrespective of any interactions with ovarian activity. Similar to our observation of an effect of IGF-I on ovarian cancer risk only at a comparatively young age, other studies have shown an association of IGF-I with breast density and breast cancer risk exclusively among premenopausal women and women with relatively early cancer diagnosis.8, 9, 34
Unfortunately, our data did not allow us to account for the effect of family history of ovarian cancer in our analyses. Genetic factors are believed to contribute about 40–60% of the variation in endogenous IGF-I levels, but at present, there is no data relating genes conferring increased risk of ovarian cancer (such as BRCA1 or BRCA2) to genes involved in the synthesis and biological activity of IGF-I. Future studies will be necessary to investigate the possible interaction of high-risk genetic background with IGF-I levels.
The observed increase in risk with elevated levels of IGFBP-3 in women with ovarian cancer diagnosis before age 55 was not entirely anticipated but coincides with other studies relating IGF-I with cancer risk.10, 25 The concomitant increase in ovarian cancer with levels of both IGF-I and IGFBP-3 might be a reflection of their common regulation by growth hormone,35 and thus suggests that increased pituitary growth hormone secretion might be at the origin of the elevated IGF-I levels among women at increased ovarian cancer risk. The association of IGF-I with ovarian cancer remained significant after adjustments for levels of IGFBP-3, whereas the confidence intervals of IGFBP-3 models always included unity, and the point estimates for IGFBP-3 were significantly reduced after adjustment for levels of IGF-I.
The findings of our study suggest that elevated IGF-I levels may be implicated in the development of ovarian cancer, diagnosed before age 55. These results adjoin ovarian cancer to the group of common tumors in the economically developed countries for which the IGF-I system is believed to play an important etiological role. IGF-I levels are intricately related to nutritional status and energy balance, and might provide a link between nutritional life-style factors such as energy- and protein-dense diets and lack of physical activity and cancer development.33
Ms. Y. Afanasyeva, Ms. L. Quinones and Ms. D. Masciangelo provided technical assistance in the NYU Women's Health Study; Ms. Å. Ågren and Mr. H. Sjodin helped with the management of the Swedish Biobank database. Mr. D. Achaintre, Ms. J. Bouzac and Ms. B. Vozar contributed to the laboratory analyses and Ms. J. Dehedin in the manuscript preparation.