IGF-1 and IGFBP-3: Risk of prostate cancer among men in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial
Version of Record online: 27 JUN 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 121, Issue 10, pages 2267–2273, 15 November 2007
How to Cite
Weiss, J. M., Huang, W.-Y., Rinaldi, S., Fears, T. R., Chatterjee, N., Chia, D., Crawford, E. D., Kaaks, R. and Hayes, R. B. (2007), IGF-1 and IGFBP-3: Risk of prostate cancer among men in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial. Int. J. Cancer, 121: 2267–2273. doi: 10.1002/ijc.22921
- Issue online: 25 SEP 2007
- Version of Record online: 27 JUN 2007
- Manuscript Accepted: 9 MAY 2007
- Manuscript Received: 16 MAR 2007
- insulin-like growth factors
IGF-1 and IGFBP-3 may influence risk of prostate cancer through their roles in cellular growth, metabolism and apoptosis, however, epidemiologic results have been inconsistent. The role of obesity in prostate cancer risk is not clearly understood, but hyperinsulinemia-related increases in bioactive IGF-1 levels, associated with obesity, could be a component of the relationship between the IGF-axis and prostate cancer. We conducted a nested case–control study in the prospective Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial to examine associations between IGF-1 and IGFBP-3 and risk of prostate cancer. A total of 727 incident prostate cancer cases and 887 matched controls were selected for this analysis. There was no clear overall association between IGF-1, IGFBP-3 and IGF-1:IGFBP-3 molar ratio (IGFmr) and prostate cancer risk, however, IGFmr was associated with risk in obese men (BMI > 30, p-trend = 0.04), with a greater than 2-fold increased risk in the highest IGFmr quartile (OR 2.34, 95% CI 1.10–5.01). Risk was specifically increased for aggressive disease in obese men (OR 2.80, 95% CI 1.11–7.08). In summary, our large prospective study showed no overall association between the insulin-like growth factor axis and prostate cancer risk, however, IGFmr was related to risk for aggressive prostate cancer in obese men. © 2007 Wiley-Liss, Inc.
Prostate cancer is the most commonly diagnosed non-skin cancer and the second leading cause of cancer mortality among men in the United States.1 Age, race and family history of prostate cancer are well established risk factors for this disease (reviewed in Ref.2), while genetic factors, steroid hormones, growth factors, obesity and dietary factors are also potentially important components in the etiology of this disease (reviewed in Refs.2, 3, 4, 5, 6, 7).
The role of the IGF axis in prostate cancer is uncertain,8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26 including its relative importance in prostate cancer initiation and progression towards aggressive disease.27 Insulin-like growth factors (IGF) are involved in cellular metabolism, differentiation, proliferation, transformation and apoptosis, during both normal and malignant growth of the prostate.28 Insulin-like growth factor-1 (IGF-1) has mitogenic and anti-apoptotic effects, increases transcription of IGF-1, and has been related to prostate cancer progression.29, 30 The majority of circulating IGF-1 is bound to one of the 6 insulin-like growth factor binding proteins (IGFBPs), which regulate the availability and ligand function of IGF-1 for its receptor, IGF-1R.31, 32 IGFBP-3, the most abundant of these proteins,33 has independent anti-mitogenic and pro-apoptotic activity in prostate cells.34, 35
There are also questions about the role of obesity in prostate carcinogenesis, with several large prospective cohort studies suggesting that obesity is associated particularly with aggressive prostate cancer and prostate cancer mortality.36, 37, 38, 39, 40 Obesity is a chronic hyperinsulinemic state, in which growth hormone (GH) synthesis and secretion are down-regulated,41, 42 while, among other factors,43, 44, 45, 46, 47, 48 GH is a primary regulator of hepatic IGF-1 production and associated serum levels. Insulin, however, also inhibits the production of IGF-binding proteins-1 and 2, which together with IGFBP-3, regulate the free fraction of IGF-1.49, 50, 51, 52 Further, increased free IGF-1 exerts negative feedback on pituitary GH secretion, thus leading to decreased hepatic IGF-1 synthesis.53 Therefore, the role of obesity in IGF bioavailability and function is complex, as is the potential interrelationship of these factors with prostate cancer.
We conducted a nested case–control study, using prediagnostic samples, in the prospective Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial, to examine associations between IGF-1 and IGFBP-3 and risk of prostate cancer.
Material and methods
The PLCO Cancer Screening Trial
This nested case–control study was conducted within the screening arm of the PLCO Screening Trial, which is a two-armed, randomized controlled trial designed to evaluate the effectiveness of prostate, lung, colorectal and ovarian cancer screening and to investigate etiologic factors and early markers of cancer.54, 55 Approximately 150,000 U.S. men and women, ages 55–74 at enrollment, were randomly assigned to the screening or nonscreening arm of the study. PLCO participants were recruited from 10 screening centers in the United States (Birmingham, AL; Denver, CO; Detroit, MI; Honolulu, HI; Marshfield, WI; Minneapolis, MN; Pittsburgh, PA; Salt Lake City, UT; St. Louis, MO and Washington, DC) between September 1993 and June 2001.
Men randomized to the screening arm of the trial (N = 38,350) were offered prostate cancer screening by serum prostate-specific antigen (PSA) and digital rectal exam (DRE) at study entry and then annually for 5 and 3 years, respectively. If the PSA test result was >4 ng/ml or DRE was suspicious for prostate cancer, men were referred to their healthcare providers for prostate cancer diagnostic evaluation. In addition, annual follow-up for recent diagnosis of cancer, loss to follow-up, or death was carried out by mailed questionnaires and through periodic search of the National Death Index. All medical and pathologic records related to diagnosis were obtained for participants suspected of having prostate cancer through screening or annual questionnaire. Furthermore, death certificates and supporting medical/pathologic records were collected. All data related to cancer diagnosis and death were abstracted by trained medical record specialists. All participants of the trial are followed for the incidence of cancer and all causes of mortality for at least 13 years from their randomization date. Screening arm participants were asked to provide a blood sample at each of the screening visits. All participants provided written informed consent. The trial was approved by the institutional review boards of the U.S. National Cancer Institute and the 10 study centers.
Of the 38,350 men randomized to the screening arm, we excluded men who did not have at least one valid screening for prostate cancer (PSA and/or DRE) before October 1, 2001 (the censor date for this analysis), men with a prior history of prostate cancer, men who did not complete the baseline risk factor questionnaire, men who refused to provide a blood sample, men with an ethnic/racial background other then non-Hispanic white or non-Hispanic black, and men who did sign the informed consent for etiologic studies on cancer. After exclusion, the analytic cohort included 28,243 men. All men were followed from their initial valid prostate cancer screen (PSA and/or DRE) to first occurrence of prostate cancer, loss to follow-up, death, or October 1, 2001, whichever came first.
Cases were defined as men diagnosed with adenocarcinoma of the prostate. Clinical stage grouping was assigned on the basis of clinical and pathological assessment of the extent of tumor involvement using the TNM system.56 Tumor (T) stage was categorized according to the fourth or fifth edition of the AJCC (American Joint Committee on Cancer) Cancer Staging Manual,57, 58 depending on the date of diagnosis. Clinical information for nodal (N) and metastatic (M) staging was recorded when available. To minimize the potential effects of undetected pre-existing disease on serum IGF-1 and IGFBP-3 levels, we excluded cases diagnosed within the first year of follow-up.
Among the 28,243 eligible men, 1,320 prostate cancer cases were identified. For the current analysis, we further excluded non-Hispanic black cases, leaving 803 cases. Controls (n = 947) were selected by incidence-density sampling,59 with a case–control ratio of 1:1.2, by age at entry (5-year intervals), time since initial screen (1-year time windows) and year of blood draw (study entry). Laboratory results could not be obtained for 8% of the 1,750 eligible men because of the lack of sufficient sample volume (n = 134) or assay failure (n = 2), resulting in a study population of 727 cases and 887 controls.
Laboratory measurement of IGF-1 and IGFBP-3
Nonfasting blood samples were collected at baseline, aliquoted to 1.8 ml storage vials within 2 hr of collection and stored at −70°C. Serum samples for our study were aliquoted at NCI and shipped to the IARC for IGF-1 and IGFBP-3 measurement by ELISA assay (Diagnostic Systems Laboratories, Webster, TX), with cases and their matched controls analyzed in the same batch. The IGF-1 assay protocol included acid–ethanol precipitation of the binding proteins, to avoid interference with IGF-1 measurement.
First, we conducted a pilot study with sera from twenty PLCO male controls to assess assay reproducibility for IGF-1 and IGFBP-3, within and across batches, over a 4-week period; the within-batch coefficients of variation (CV) for serum measurements were 9 and 5% for IGF-1 and IGFBP-3, respectively, with corresponding intraclass correlation coefficients of 90 and 92%. Some other studies relating these analytes to prostate cancer risk have used plasma samples,10, 18, 21 we also included paired plasma samples from twenty male volunteers in the PLCO trial in the pilot study, which revealed strong correlation between results for these 2 blood components (IGF-1, r = 0.97 and IGFBP-3, r = 0.99).
For the case and control sample analysis, we included blinded quality control samples (2 duplicates from each of 2 subjects) showing overall CVs of 9% for IGF-1 and 9% for IGFBP-3. We also determined temporal intrasubject variability for these analytes (r = 0.72 and r = 0.70 for IGF-1 and IGFBP-3, respectively), by comparing results for study (baseline) sera with serum samples collected 1 year later from the same control subjects (n = 49).
Assessment of questionnaire-based covariates
Participants completed self-administered questionnaires at enrollment, including information on age, ethnicity, height, weight, education, current and past smoking behavior, history of cancer and other diseases, use of selected drugs, recent history of screening exams and prostate-related health factors.
Using conditional logistic regression analysis, odds ratios (ORs) were calculated for the relation between IGF-1, IGFBP-3 and IGF-1:IGFBP-3 molar ratio (IGFmr) and incident adenocarcinoma of the prostate. IGF-1, IGFBP-3 and IGFmr were modeled as categorical variables defined by the quartile and tertile distributions among controls. Other factors [body mass index (BMI), height, diabetes, family history of prostate cancer, physical activity, cigarette smoking, intake of various nutrients and study center] were evaluated as potential confounders by assessing whether their inclusion altered risk estimates by at least 10%; none of these factors remained in the final models. Missing values (1–2%) for continuous and categorical confounding variables were replaced by the median and mode values, respectively, of these variables in the study population.60 Final models included mutual adjustment for IGF-1 and IGFBP-3, although inclusion of either into the model did not materially alter risk estimates for the other. IGFmr was calculated using conversion factors of 0.13 for IGF-1 and 0.036 for IGFBP-3.61 To test for linear trends in ORs across groups, we created score variables representing the mean of each quartile or tertile, including these into the logistic regression models as continuous variables. To determine whether the relationships with prostate cancer would better be represented with a semiparametric approach, IGF-1 and IGFBP-3 were also modeled flexibly using restricted cubic spline functions.
To determine if the relationship between IGF-1 and IGFBP-3 differed by these factors, stratification was performed by age (<65 vs. 65+ years), BMI (<30 kg/m2vs. 30+ kg/m2) and time between blood draw and diagnosis (1–2 vs. 3–7 years). Likelihood ratio tests were performed to test for multiplicative interactions. Polytomous logistic regression was used to asses the relationship between IGF-1, IGFBP-3, their molar ratio and nonaggressive versus aggressive disease. Aggressive disease was defined as Gleason score ≥ 7 or Stage III/IV. To test for the homogeneity of ORs across categories of tumor aggressiveness by levels of each exposure variable, we compared constrained regression models using the likelihood ratio test. All analyses were performed using the STATA statistical package, Version 9 (STATA, College Station, TX).
In controls, serum levels of IGF-1 and IGFBP-3 decreased with increasing age; serum levels of IGF-1 decreased with increasing BMI; and, serum levels of IGFBP-3 decreased with increasing height (Table I). Also, IGF-1 levels were higher in controls with elevated serum PSA (≥4 ng/ml). The mean molar ratio of IGF-1 to IGFBP-3 (IGFmr) was constant across all these factors. With respect to the cases, diagnoses were made after an elevated PSA (57%), abnormal DRE (14%), both an elevated PSA and abnormal DRE (17%), or for another reason (11%). Cases presented with median PSA levels of 5.2 and tumors of Gleason score 6 and Stage III. Cases and controls were similar, at study entry, with respect to age, height and BMI (data not shown).
|IGF-1 (ng/ml)||IGFBP-3 (ng/ml)||IGFmr (ng/ml)|
|<60||224.0 ± 90.6||4840.9 ± 1096.0||0.16 ± 0.05|
|60–65||210.4 ± 78.9||4809.6 ± 1025.4||0.16 ± 0.05|
|65–70||207.8 ± 90.2||4503.4 ± 1102.0||0.16 ± 0.05|
|≥70||189.5 ± 84.1||4104.0 ± 1095.4||0.16 ± 0.05|
|<25||207.9 ± 91.7||4493.0 ± 1088.3||0.16 ± 0.05|
|25–30||212.8 ± 85.8||4660.0 ± 1108.8||0.16 ± 0.05|
|30–35||196.8 ± 79.3||4486.2 ± 1076.4||0.16 ± 0.05|
|≥35||185.4 ± 72.3||4453.6 ± 1226.8||0.15 ± 0.04|
|<68||204.4 ± 84.9||4705.3 ± 1174.3||0.16 ± 0.05|
|68–70||201.5 ± 83.2||4414.6 ± 1000.0||0.16 ± 0.05|
|70–72||211.6 ± 86.5||4635.2 ± 1131.5||0.16 ± 0.04|
|≥72||209.8 ± 88.5||4574.7 ± 1114.9||0.16 ± 0.04|
|<4||205.9 ± 85.0||4565.1 ± 1101.8||0.16 ± 0.05|
|≥4||225.1 ± 96.3||4693.0 ± 1155.5||0.17 ± 0.05|
There was no clear association between serum levels of IGF-1, IGFBP-3 or IGFmr and risk for all prostate cancer combined (Table II). Risks for aggressive prostate cancer (Gleason score ≥ 7 or Stage III/IV) tended to be greater with greater serum IGF-1 (p-trend = 0.06) and IGFmr (p-trend = 0.07); these trends were due, however, to risks associated only with the highest levels of IGF-1 and IGFmr. Similar results were observed when tumors were classified solely by Gleason score or stage of disease (IGFmr, highest quartile: OR 1.33, 95% CI 0.91–1.91 and OR 1.69, 95% CI 0.94–3.04, respectively). Spline analyses supported these overall results.
|Median ± SD in controls||113.8 ± 25.5||170.0 ± 13.0||220.8 ± 17.1||323.4 ± 64.7|
|OR (95% CI)2||Reference||0.85 (0.64–1.13)||0.96 (0.72–1.27)||1.14 (0.86–1.51)||0.18|
|OR (95% CI)3||Reference||0.81 (0.60–1.10)||0.93 (0.67–1.28)||1.12 (0.79–1.60)||0.28|
|N (%)||115 (25.6)||107 (23.8)||108 (24.0)||120 (26.7)|
|OR (95% CI)4||Reference||0.94 (0.68–1.30)||0.96 (0.69–1.32)||1.05 (0.76–1.45)||0.69|
|OR (95% CI)5||Reference||0.90 (0.64–1.27)||0.89 (0.62–1.30)||0.94 (0.63–1.41)||0.85|
|N (%)||71 (25.6)||50 (18.1)||71 (25.6)||85 (30.7)|
|OR (95% CI)4||Reference||0.69 (0.46–1.05)||0.98 (0.66–1.43)||1.23 (0.85–1.78)||0.08|
|OR (95% CI)5||Reference||0.64 (0.42–1.00)||0.96 (0.63–1.49)||1.32 (0.83–2.11)||0.06|
|Median ± SD||3175.3 ± 527.7||4198.9 ± 221.2||4897.2 ± 197.5||5957.4 ± 612.7|
|OR (95% CI)2||Reference||1.27 (0.96–1.68)||1.02 (0.76–1.37)||1.20 (0.89–1.61)||0.44|
|OR (95% CI)3||Reference||1.30 (0.97–1.76)||1.00 (0.71–1.41)||1.10 (0.76–1.59)||0.95|
|N (%)||106 (23.6%)||118 (26.2)||102 (22.7)||124 (27.6)|
|OR (95% CI)4||Reference||1.12 (0.82–1.56)||0.99 (0.71–1.39)||1.20 (0.86–1.67)||0.39|
|OR (95% CI)5||Reference||1.17 (0.83–1.66)||1.04 (0.71–1.52)||1.25 (0.82–1.89)||0.41|
|N (%)||56 (20.2%)||88 (31.8)||63 (22.7)||70 (25.3)|
|OR (95% CI)4||Reference||1.57 (1.07–2.32)||1.09 (0.72–1.65)||1.24 (0.82–1.87)||0.67|
|OR (95% CI)5||Reference||1.62 (1.07–2.46)||1.00 (0.62–1.61)||1.00 (0.60–1.67)||0.45|
|Median ± SD||0.11 ± 0.02||0.14 ± 0.01||0.17 ± 0.01||0.22 ± 0.03|
|OR (95% CI)2||Reference||0.76 (0.57–1.01)||0.78 (0.58–1.03)||1.13 (0.86–1.49)||0.27|
|N (%)||128 (28.4)||97 (21.6)||101 (22.4)||124 (27.6)|
|OR (95% CI)4||Reference||0.77 (0.55–1.06)||0.78 (0.56–1.08)||0.96 (0.70–1.32)||0.97|
|N (%)||75 (27.1)||54 (19.5)||53 (19.1)||95 (34.3)|
|OR (95% CI)4||Reference||0.71 (0.48–1.06)||0.72 (0.48–1.08)||1.31 (0.91–1.89)||0.07|
In obese men (BMI ≥ 30), serum IGF-1 levels greater than the lowest quartile (cutpoint, 146.7 ng/ml) were associated with about 2-fold increased risk for prostate cancer, although risks did not increase in a dose-response fashion, with increasing IGF-1 levels (p-trend = 0.20 and p-interaction = 0.12, Table III). No association was observed, in this group, between IGFBP-3 and prostate cancer (p-interaction = 0.70). Greater IGFmr tended to be associated with increased prostate cancer risk in obese men (p-trend = 0.04), with greater than 2-fold risks for men in the highest IGFmr quartile (OR 2.34, 95% CI 1.10–5.01; Table III), but the interaction was not statistically significant (p-interaction = 0.58). The increased risk associated with IGFmr in obese men was limited largely to those with aggressive cancer (highest quartile OR 2.80, 95% CI 1.11–7.08; Table IV). No risk relationships of interest were found for men with BMI < 30 (Tables III and IV). Similar results were observed when tumors were classified separately by Gleason score or stage of disease (IGFmr, highest quartile: OR 2.83, 95% CI 1.05–7.64 and OR 2.27, 95% CI 0.39–13.31, respectively).
|Body mass index < 30 kg/m2|
|OR (95% CI)1||Reference||0.71 (0.50–1.01)||0.78 (0.55–1.12)||0.95 (0.64–1.42)||0.80|
|OR (95% CI)1||Reference||1.37 (0.97–1.93)||1.05 (0.71–1.54)||1.16 (0.76–1.78)||0.92|
|OR (95% CI)1||Reference||0.73 (0.53–1.00)||0.72 (0.52–0.99)||0.98 (0.72–1.34)||0.88|
|Body mass index ≥ 30 kg/m2|
|OR (95% CI)1||Reference||2.17 (1.00–4.70)||2.24 (0.91–5.51)||2.11 (0.82–5.45)||0.20|
|OR (95% CI)1||Reference||0.97 (0.44–2.15)||0.74 (0.30–1.84)||1.06 (0.39–2.88)||0.92|
|OR (95% CI)1||Reference||1.22 (0.59–2.54)||1.06 (0.51–2.17)||2.34 (1.10–5.01)||0.04|
|Body mass index < 30 kg/m2|
|N (%)||117 (32.1)||123 (33.7)||125 (34.2)|
|OR (95% CI)2||Reference||1.01 (0.72–1.43)||0.92 (0.62–1.35)||0.61|
|N (%)||76 (32.8)||68 (29.3)||88 (37.9)|
|OR (95% CI)2||Reference||0.87 (0.58–1.31)||1.19 (0.76–1.86)||0.32|
|N (%)||116 (31.8)||116 (31.8)||133 (36.4)|
|OR (95% CI)2||Reference||0.99 (0.70–1.40)||1.17 (0.79–1.74)||0.40|
|N (%)||72 (31.0)||83 (35.8)||77 (33.2)|
|OR (95% CI)2||Reference||1.05 (0.70–1.58)||0.91 (0.57–1.46)||0.60|
|N (%)||126 (34.5)||109 (29.9)||130 (35.6)|
|OR (95% CI)2||Reference||0.85 (0.62–1.17)||0.95 (0.69–1.29)||0.79|
|N (%)||82 (35.3)||56 (24.1)||94 (40.5)|
|OR (95% CI)2||Reference||0.70 (0.47–1.03)||1.12 (0.79–1.60)||0.38|
|Body mass index ≥30 kg/m2|
|N (%)||26 (30.6)||35 (41.2)||24 (28.4)|
|OR (95% CI)2||Reference||1.70 (0.84–3.42)||1.18 (0.51–2.74)||0.82|
|N (%)||12 (26.7)||14 (31.1)||19 (42.2)|
|OR (95% CI)2||Reference||1.13 (0.45–2.84)||1.97 (0.71–5.41)||0.16|
|N (%)||30 (35.3)||22 (25.9)||33 (38.8)|
|OR (95% CI)2||Reference||0.73 (0.35–1.55)||1.23 (0.55–2.74)||0.56|
|N (%)||12 (26.7)||18 (40.0)||15 (33.3)|
|OR (95% CI)2||Reference||1.31 (0.53–3.26)||0.99 (0.34–2.93)||0.94|
|N (%)||31 (36.5)||23 (27.1)||31 (36.5)|
|OR (95% CI)2||Reference||0.87 (0.44–1.72)||1.47 (0.75–2.86)||0.24|
|N (%)||9 (20.0)||18 (40.0)||18 (40.0)|
|OR (95% CI)2||Reference||2.00 (0.82–4.89)||2.80 (1.11–7.08)||0.03|
Risks associated with the IGF axis in obese men were similar for cases diagnosed 1–2 versus 3–7 years after study entrance. No associations were noted on stratification by height. Although risks associated with IGF-1 and IGFmr tended to be stronger in younger (age < 65 years) men, the associations were not significant (results not shown).
We did not observe an overall association between IGF-1, IGFBP-3 or their molar ratio and prostate cancer risk, over the full range of these analytes in serum in the PLCO Trial. However, greater IGF-1 and particularly IGFmr levels were associated with increased risk of prostate cancer in obese men.
Our results are broadly consistent with findings from the previous largest study (530 cases and 534 controls), the prospective Physicians Health Study (PHS),9 which showed no overall associations between IGF-1 and IGFBP-3 and prostate cancer, but reported increased risks of aggressive disease for men in the highest quartile of IGF-1 (OR 5.1, 95% CI 2.0–13.2) and IGFmr (OR 2.5, 95% CI 1.2–5.2).9 Associations between IGF-1 and IGFBP-3 and disease risk were limited to, or more pronounced among, men with aggressive disease in other cohort studies as well10, 18, 20 The interrelationship of the IGF axis and BMI in prostate cancer etiology remains uncertain, however, as other prospective studies have not shown such effects.11, 21, 25
The observed increase in risk of aggressive prostate cancer in obese men with elevated IGFmr may operate through increased levels of estradiol (E2)62 and cytokines (leptin, IL-6)63 produced in peripheral adipose tissue. IGF-1 and leptin cooperate in the stimulation of androgen-dependent prostate cancer cell progression in vitro.64 The mitogenic and anti-apoptotic effects of IGF-1, together with the proliferative effects of excess estradiol and cytokines, promote tumor growth and development.
Differences in study design and populations, modest sample sizes and laboratory variability could account for some of the inconsistencies observed across studies. Questions have been raised concerning the reported tendency of IGFBP-3 assays to underestimate serum concentrations, particularly at higher levels.65 There are also concerns about lack of comparability of kits differing in sensitivity to detect intact and fragmented forms,21, 66 with laboratory results potentially depending on differential IGFBP-3 susceptibility to proteolytic cleavage, because of glycosylation, phosphorylation and other secondary aspects of protein structure.66, 67 The DSL ELISA kit used for the current assays measures total IGFBP-3,66 circumventing many of these problems. Others have raised concerns about comparability of results between plasma and serum-based studies. Our pilot study showed that mean IGF-1 measurements from serum and plasma did not differ, but that mean IGFBP-3 measurements from serum were significantly greater than the plasma measurements. The correlations, both Pearson and Spearman, were close to one so any case/control comparisons of IGF-1, IGFBP-3 or their molar ratio (IGFmr) would not likely be changed in an important way by the use of serum rather than plasma.
The current study has a number of design strengths: it is the largest prospective study to examine the relationship between IGF-1 and IGFBP-3 and risk of prostate cancer. Samples were collected prior to diagnosis and our substudy of bloods collected at 2 time points showed that a single measure of these analytes gave a good representation of a given individual's exposure level.
Limiting cases for our study to men diagnosed at least 1 year after blood draw does not, however, eliminate the possibility of reverse-causation bias, as undetected prostate cancers were likely prevalent at the initial blood draw. Nevertheless, we found no differential patterns for men diagnosed 1–2 versus 3–7 years after blood draw. Our study was carried out in a population under surveillance for prostate cancer by PSA and DRE screening. Both BMI68, 69, 70, 71, 72 and IGF-1 can influence PSA levels, which could have lead to differential rates of diagnosis for total prostate cancer in our study; however, these influences would not lead specifically to differential rates for aggressive cancers in men under serial PSA screening.
Linear dose-response relationships were not observed across quartiles of IGF-1, IGFBP-3 or IGFmr measurements. Restricted cubic spline analyses did not provide additional insight into the relationships between these analytes and prostate cancer. It is plausible that there is a threshold level above which disease risk is altered, but the presence of a dose-response relationship would be more suggestive of a causal relationship, and no mechanistic data directly indicate a threshold effect.
In summary, we observed only modest overall associations between IGF-1 and IGFmr and risk of prostate cancer in a screened population of Caucasian men being followed prospectively over time, however, 2-fold increases in risk were observed in obese men with the greatest IGFmr levels, and these associations were particularly shown for aggressive disease. Our findings suggest an interplay between obesity and the IGF axis in prostate carcinogenesis.
- 1American Cancer Society. Cancer facts and figures. Atlanta, GA: American Cancer Society, 2007.
- 57American Joint Committee on Cancer staging manual. 5th editon. Philadelphia: J.B. Lippincott 1997; 219–22., , , , , .
- 58American Joint Committee on Cancer staging manual. 4th editon. Philadelphia: J.B. Lippincott 1992; 181–6., , , .
- 59Modern epidemiology. Philadelphia, PA: Lippincott Raven, 1998., .
- 60Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. New York: Springer-Verlag; 2001; 41–52.Missing data.