Abstract
Androgens and growth factors are thought to be associated with prostate cancer risk, though past research has produced mixed results. We conducted a nested case-control study of biomarkers of prostate cancer risk within the Multiethnic Cohort. We compared prediagnostic levels of testosterone, dihydrotestosterone (DHT), sex hormone-binding globulin (SHBG), 3α-androstanediol glucuronide (3α-diol G), insulin-like growth factor I (IGF-I), IGF-II, insulin-like growth factor binding protein 1 (IGFBP-1), and IGFBP-3 in serum from 467 incident prostate cancer cases and 934 cancer-free controls. Controls were matched to the cases on geographic site (HI, LA), ethnicity, age at specimen collection (± 1 year), date (± 1 month) and time of day (± 2 hours) of sample collection, and fasting status (<6, 6-7, 8-9, 10+ hours). Multivariate conditional logistic regression models were used to compute adjusted odds ratios (ORs) and 95% confidence intervals (CIs). Serum concentrations of testosterone, DHT, SHBG, 3α-diol G, IGF-I, IGF-II, IGFBP-1, and IGFBP-3 were not associated with risk of prostate cancer. Tests for trend of quartiles of serum concentrations also did not show any association. Results were relatively unchanged for men with advanced prostate cancer and their matched controls. However, the follow-up period was relatively short (mean of 1.9 years). Analysis by ethnic group showed an increased risk for Latino men in the second (OR=3.67, 95% CI: 1.63-8.24) and third (OR=2.96, 95% CI:1.19-7.40) tertiles of IGF-I serum levels compared to the first tertile. The suggested increased risk for IGF-I in Latino men merits further study, with greater statistical power.
Keywords: IGFs, androgens, prostate cancer, risk, ethnicity
Introduction
Although prostate cancer is the leading cancer among males in the US, there are few established risk factors for this neoplasm. Because hormones and growth factors stimulate cell proliferation and have been linked to other cancers, many research studies have examined the association of androgens and/or growth factors with the risk of prostate cancer. The results, however, have been inconsistent with regard to which, if any, of the androgens and growth factors alter prostate cancer risk. Two recent pooled analyses of 12 (1) and 18 (2) prospective studies, respectively, found no overall association of testosterone, dihydrotestosterone (DHT), 3α-androstanediol gluconoride (3α-diol G), and insulin-like growth factor II (IGF-II) with the risk of prostate cancer. However, they did observe a statistically significant negative association of sex hormone-binding globulin (SHBG) with prostate cancer risk and statistically significant positive associations of insulin-like growth factor I (IGF-I) and insulin-like growth factor binding protein 3 (IGFBP-3) with prostate cancer risk.
In this analysis, we examined serum levels of several androgens and growth hormones-- 3α-diol G, SHBG, testosterone, DHT, IGF-II, IGF-I, IGFBP-1, and IGFBP-3-- in a nested case-control study of prostate cancer. Cases and controls were identified through the prospective Multiethnic Cohort Study of African-Americans, Caucasians, Japanese-Americans, Latinos and Native Hawaiians.
Materials and Methods
Study Population
Details of the Multiethnic Cohort (MEC) were described previously (3). In brief, data were collected between 1993-1996 using a 26-page self-administered mail questionnaire sent to residents of Hawaii and California, mainly Los Angeles County. Subjects were identified through drivers' license records in both locations, supplemented with voter registration records in Hawaii and Health Care Financing Administration (Medicare) files in California. African-Americans, Caucasians, Japanese-Americans, Latinos and Native Hawaiians were the primary targets for recruitment, but a small number of persons of other ethnicities were also enrolled in the study. Participation in the cohort was limited to people between the ages of 45-75 years in 1993, except for Native Hawaiians who were recruited at 42 years and older. The MEC dataset consists of 215,251 people, including 96,382 men. The Institutional Review Boards of both the University of Hawaii and the University of Southern California approved the study.
Biospecimen Subcohort
Participants for this nested case-control study were men from the MEC who had provided prediagnostic blood specimens primarily between 2001 and 2006. Cohort members were contacted by letter, and then by phone, to request biological specimens (blood and urine). For those who agreed, a short screening questionnaire (use of anticoagulants, blood clotting disorders, etc.) and updated information on a few items (including current smoking habits, weight, vitamin supplement use, PSA screening) was administered. Specimens were collected at a clinical laboratory or in the subjects' home and were processed within four hours of collection. Blood samples were drawn in a fasting state of 8 or more hours for most cases (94%), and were separated into components (serum, plasma, buffy coat, red cells) and stored in multiple 0.5 cc aliquots in the vapor phase of liquid nitrogen freezers.
Selection of Cases and Controls
Cases of invasive prostate cancer, diagnosed after specimen collection, were identified through linkages with the Los Angeles County Cancer Surveillance Program, the State of California Cancer Registry, and the Hawaii Tumor Registry, all members of the Surveillance, Epidemiology, and End Results program supported by the National Cancer Institute. Advanced prostate cancer cases were defined as: 1) having either regional or distant spread and/or 2) having a Gleason score ≥ 7 irrespective of tumor stage. A total of 467 prostate cancer cases were identified for this study. Controls were selected among the male biorepository participants who were alive and free of prostate cancer at the age of diagnosis of the case. A control pool that met the matching criteria was created for each case, from which two controls were randomly selected. Matching criteria included geographic site (HI, LA), ethnicity, age at specimen collection (± 1 year), date (± 1 month) and time of day (± 2 hours) of sample collection, and fasting status (<6, 6-7, 8-9, 10+ hours).
Assay Methods
Serum levels of testosterone and DHT were measured using radioimmunoassay (RIA) methods(4,5). After adding appropriate internal standards (3H-testosterone and 3H-DHT) to each sample to follow procedural bases, testosterone and DHT were extracted from the serum using ethyl acetate: hexane (1:1). Following evaporation of the organic solvents, the extract was redissolved in isooctane and applied on a Celite partition column impregnated with ethylene glycol. DHT was eluted from the column with 10% toluene in isooctane, whereas 40% toluene in isooctane was used to elute testosterone. After evaporating the eluates, the residues were reconstituted in assay buffer and aliquots were taken for the RIA procedure and for determining procedural losses. The RIAs for testosterone and DHT utilized specific antisera in conjunction with the appropriate iodinated testosterone or DHT derivative as the radioligand. Testosterone and DHT were incubated for 16-18 hrs. Subsequently, separation of antibody-bound testosterone or DHT was achieved by use of a second antibody. After centrifugation, the antibody-bound fraction was counted in a gamma counter and the concentration of each androgen was calculated and corrected for procedural losses.
3α-Diol G, SHBG and PSA were quantified by direct immunoassays. 3α-Diol G was measured by RIA using a commercial kit obtained from Beckman-Coulter Diagnostic Systems Laboratories (Webster, Texas) (6), whereas SHBG and prostate specific antigen (PSA) were measured by chemiluminescent immunoassay on the Immulite analyzer (Siemens Medical Solutions, Los Angeles, CA).
Insulin-like growth factors, including IGF-I and IGF-II, and IGF binding proteins 1 and 3 were measured by an enzyme-linked immunoabsorbent assay (ELISA) method. The analytic laboratory has validated the IGF I, II and IGFBP 1, 3 ELISA methodology using reagents from Diagnostic Systems Laboratories (Webster, Texas) against regular radioimmunoassays following acid extraction. The results obtained with the ELISA methodology, which is more suitable for large numbers of specimens, were highly correlated with the more labor-intensive methods (r=0.82). All assays were done with standards and controls that include recombinant proteins, as well as a pooled serum sample. For IGF-I and IGF-II, the absolute sensitivity of the assay was 0.3 ng/ml.
Serum analysis was performed on 394 (84%) cases, 792 (85%) controls for 3α-diol G, 463 (99%) cases, 931 (99%) controls for SHBG, testosterone, and DHT, 326 (70%) cases, 657 (70%) controls for IGF-II and 386 (83%) cases, 777 (83%) controls for IGF-I, IGFBP-1 and IGFBP-3. The percentages vary because of the availability of sample and the requirement of a fasting sample for the growth factor measures. Additionally, IGF-II was not performed on all samples because of the temporary cessation of the assay material by the manufacturer. The intra-assay coefficients of variation were 3.4% for 3α-diol G, 3.0% for SHBG, 3.5% for testosterone, 3.8% for DHT, 1.8% for IGF-II, 2.1% for IGF-I, 2.2% for IGFBP-I, and 2.5% for IGFBP-3.
Statistical Analysis
We applied multivariate conditional logistic regression models of prostate cancer incidence, with case-control matched sets as the strata variable, to estimate odds ratios (ORs) and 95% confidence intervals (95% CIs). We created quartiles for each variable based on the distribution of cases and controls combined, and represented them with three indicator variables. Individual trend variables were created by assigning them the median values of each quartile grouping. We adjusted for the following covariates in our models: body mass index (≤ 25, >25 to ≤ 30, >30 kg/m2), family history of prostate cancer in father and/or brother(s) (yes, no), years of education (continuous), age at blood draw (continuous), and number of fasting hours prior to blood draw (continuous). The first 3 variables were selected because they were found to be related to prostate cancer risk or screening in the MEC. The last two variables accounted for any systematic differences in these variables within matched sets. We repeated the analyses using only controls with PSA values less than or equal to 4.0 ng/ml and their matched cases to minimize any potential bias due to disease misclassification. We also performed analyses using only advanced prostate cases and their matched controls. Additionally, we performed analyses using tertiles of each variable for the three ethnic groups with adequate sample size (African-Americans, Japanese-Americans and Latinos). We tested the interaction of ethnicity and BMI (≥ 25 kg/m2 versus < 25 kg/m2) with the trend variable for each androgen/growth factor variable using the Wald test for the cross-product terms.
Results
Cases and controls had similar median values for all the analytes (Table 1), though there were slight variations in median values of IGFBP-3 and IGF-II.
Table 1.
Characteristics of cases and controls#
| Cases | Controls | |
|---|---|---|
| Covariates | n=467 | n=936 |
| Body mass index (kg/m2), mean (SD) | 26.2 (4.0) | 26.5 (4.1) |
| Age at blood draw (years), mean (SD) | 68.9 (7.1) | 68.7 (7.1) |
| Fasting hours prior to blood draw, mean (SD) | 11.8 (4.8) | 11.9 (4.9) |
| High school education or less, % | 34.0 | 34.4 |
| Family history of prostate cancer, % | 12.6 | 8.3 |
| Ethnicity, % | ||
| African-American | 46.9 | 46.8 |
| Caucasian | 13.1 | 13.1 |
| Japanese-American | 18.8 | 18.8 |
| Latino | 17.8 | 17.7 |
| Native Hawaiian | 3.4 | 3.5 |
| Analytes median (interquartile range) | ||
| Testosterone (ng/dl) | 550 (422 - 689) | 541 (425 – 683) |
| Dihydrotestosterone (ng/dl) | 58 (42 – 78) | 57 (43 – 75) |
| 3α-androstanediol glucuronide (ng/ml) | 4.71 (3.15 - 6.69) | 4.47 (3.18 – 6.63) |
| Sex hormone-binding globulin (nmol/L) | 36 (28 - 45) | 35 (27 – 46) |
| Insulin-like growth factor I (ng/ml) | 182 (148 – 217) | 180 (151 – 222) |
| Insulin-like growth factor II (ng/ml) | 876 (728 – 1039) | 900 (744 – 1062) |
| Insulin-like growth factor binding protein 1 (ng/ml) | 21 (11 – 36) | 20 (11 – 38) |
| Insulin-like growth factor binding protein 3 (ng/ml) | 3892 (3201-4664) | 3971 (3387 – 4591) |
Cases and controls were matched on geographic site (Hawaii/ Los Angeles, Ca), ethnicity, age at specimen collection (± 1 year), date (± 1 month) and time of day (± 2 hours) of sample collection, and fasting status (<6, 6-7, 8-9, 10+ hours).
We observed no association between serum 3α-diol G or SHBG and prostate cancer risk (Table 2); the odds ratios for the fourth quartile compared to the first quartile of serum concentration were 1.03 (95% CI: 0.71-1.48) and 0.83 (95% CI: 0.59-1.18), respectively. The odds ratios for serum concentrations of testosterone and DHT were similar and both were not associated with prostate cancer risk. The odds ratio were: OR=0.87 (95% CI: 0.61-1.22) for testosterone, OR= 0.81 (95% CI: 0.57-1.15) for DHT and OR=1.04 (95% CI: 0.74-1.48) for the ratio of testosterone to DHT for the fourth quartile compared to the first quartile. Risk estimates for testosterone did not change materially with further adjustment for serum SHBG levels. We observed no trends in risk for any of the hormones or SHBG.
Table 2.
Odds ratios and 95% confidence intervals for risk of prostate cancer across quartiles of serum androgens, SHBG and growth factors
| Variable | Quartiles of serum concentration levels | ||||
|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | P trend | |
| Testosterone | |||||
| Median level (ng/dl) | 348 | 486 | 608 | 832 | |
| No. of cases† | 111 | 119 | 109 | 113 | |
| Unadjusted OR* | 1 | 1.02(0.73-1.41) | 0.93(0.66-1.3) | 0.98(0.70-1.36) | |
| Multivariate OR# | 1 | 0.98 (0.70-1.37) | 0.84(0.59-1.18) | 0.87(0.61-1.22) | 0.38 |
| Dihydrotestosterone | |||||
| Median level (ng/dl) | 33 | 50 | 66 | 94 | |
| No. of cases | 110 | 120 | 117 | 105 | |
| Unadjusted OR | 1 | 1.15(0.83-1.59) | 1.05(0.75-1.46) | 0.91(0.66-1.27) | |
| Multivariate OR | 1 | 1.12(0.80-1.56) | 0.97(0.69-1.37) | 0.81(0.57-1.15) | 0.14 |
| Testosterone/ Dihydrotestosterone ratio |
|||||
| Median level | 6.5 | 8.6 | 10.3 | 13.2 | |
| No. of cases | 113 | 120 | 106 | 113 | |
| Unadjusted OR | 1 | 1.12(0.81-1.53) | 0.93(0.67-1.29) | 1.02(0.72-1.43) | |
| Multivariate OR | 1 | 1.13(0.82-1.55) | 0.94(0.67-1.3) | 1.04(0.74-1.48) | 0.95 |
| 3α-diol G | |||||
| Median level (ng/ml) | 2.30 | 3.94 | 5.52 | 8.38 | |
| No. of cases | 95 | 110 | 89 | 91 | |
| Unadjusted OR | 1 | 1.30(0.91-1.84) | 0.93(0.65-1.34) | 0.97(0.68-1.39) | |
| Multivariate OR | 1 | 1.31(0.92-1.87) | 1.00(0.69-1.44) | 1.03(0.71-1.48) | 0.72 |
| SHBG | |||||
| Median level (nmol/L) | 23 | 32 | 40 | 55 | |
| No. of cases | 122 | 107 | 109 | 115 | |
| Unadjusted OR | 1 | 0.84(0.61-1.15) | 0.88(0.63-1.22) | 0.94(0.67-1.31) | |
| Multivariate OR | 1 | 0.80(0.58-1.11) | 0.81(0.58-1.13) | 0.83(0.59-1.18) | 0.42 |
| IGF-I | |||||
| Median level (ng/ml) | 124 | 166 | 197 | 248 | |
| No. of cases | 87 | 108 | 84 | 97 | |
| Unadjusted OR | 1 | 1.43(0.99-2.05) | 0.97(0.67-1.41) | 1.19(0.83-1.72) | |
| Multivariate OR | 1 | 1.43(0.99-2.06) | 0.97(0.67-1.41) | 1.23(0.85-1.78) | 0.62 |
| IGFBP-1 | |||||
| Median level (ng/ml) | 6.3 | 15.8 | 27.9 | 51.1 | |
| No. of cases | 92 | 99 | 84 | 101 | |
| Unadjusted OR | 1 | 1.06(0.73-1.52) | 0.85(0.58-1.25) | 1.11(0.75-1.66) | |
| Multivariate OR | 1 | 1.00(0.69-1.44) | 0.78(0.52-1.16) | 0.98(0.65-1.49) | 0.94 |
| IGF-II | |||||
| Median level (ng/ml) | 627 | 810 | 962 | 1176 | |
| No. of cases | 78 | 74 | 78 | 88 | |
| Unadjusted OR | 1 | 0.94(0.63-1.38) | 1.05(0.70-1.56) | 1.21(0.81-1.81) | |
| Multivariate OR | 1 | 0.91(0.61-1.35) | 1.05(0.70-1.57) | 1.17(0.78-1.76) | 0.36 |
| IGFBP-3 | |||||
| Median level (ng/ml) | 2792 | 3626 | 4295 | 5162 | |
| No. of cases | 79 | 101 | 109 | 87 | |
| Unadjusted OR | 1 | 1.42(0.99-2.04) | 1.67(1.15-2.41) | 1.15(0.78-1.70) | |
| Multivariate OR | 1 | 1.35(0.94-1.95) | 1.66(1.14-2.41) | 1.15(0.77-1.71) | 0.35 |
| IGF-1/IGFBP-3 | |||||
| Median level | 0.036 | 0.043 | 0.050 | 0.059 | |
| No. of cases | 93 | 110 | 89 | 84 | |
| Unadjusted OR | 1 | 1.27(0.89-1.81) | 0.94(0.66-1.35) | 0.85(0.58-1.25) | |
| Multivariate OR | 1 | 1.28(0.90-1.83) | 0.93(0.64-1.33) | 0.90(0.61-1.33) | 0.34 |
Adjusted by conditional logistic regression for matched sets as strata and age at blood draw and fasting hours prior to blood draw as continuous measures.
Adjusted by conditional logistic regression for matched sets as strata and for age at blood draw, fasting hours prior to blood draw, body mass index, family history of prostate cancer, and education as covariates in the log-linear model.
Sample size for cases varies by analyte due to differences in the number of samples analyzed for each analyte and due to missing values for some covariates.
Serum IGF-I concentrations were not associated with prostate cancer risk. However, the ORs across quartiles are not close to the null. Risk estimates for the second, third and fourth quartiles compared to the first quartile were respectively: 1.43 (95% CI: 0.99-2.06), 0.97(0.67-1.41) and 1.23 (95% CI: 0.85-1.78). We observed no association of serum IGFBP-I levels and prostate cancer risk. The OR for the fourth quartile compared to the first quartile was close to the null (OR= 0.98, 95% CI: 0.65-1.49). The risk estimates for IGF-II increased somewhat across the quartiles, but there was no statistically significant trend (P trend = 0.36). The OR for IGF-II was 1.17 (95% CI: 0.78-1.76) for the fourth quartile compared to the first quartile. The results for IGFBP-3 were interesting. We observed a 66% increased risk of prostate cancer (95% CI: 1.14-2.41) for men in the third quartile of serum IGFBP-3 compared to the first quartile. However, the OR decreased for men in the fourth quartile of serum concentration, OR= 1.15 (95% CI: 0.77-1.71) and there was no trend (P trend= 0.35). The ratio of IGF-I to IGFBP-3 was not associated with risk of prostate cancer; the odds ratio was 0.90 (95% CI: 0.61-1.33) for the fourth quartile compared to the first quartile.
When all of the analyses were restricted to control subjects with PSA values less than or equal to 4.0 and their matched cases, our conclusions were unchanged (data not shown). We also examined effect modification by BMI (< 25 and ≥ 25 kg/m2) and found no statistical evidence for differences across the strata (data not shown).
We repeated the analyses by ethnic group (Table 3) to see whether the findings in Table 2 appeared consistent. Due to a limited sample size, we were unable to perform analyses on Native Hawaiians and Caucasians. As shown in Table 2, serum concentrations of testosterone, DHT, 3α-diol G and SHBG, were not associated with risk of prostate cancer for African-Americans, Japanese-Americans or Latinos. However, we observed an increase in risk of prostate cancer with serum IGF-I levels for Latino men; in contrast, African-American men had a decrease in risk of prostate cancer (P interaction for ethnicity < 0.001). Latino men in the second tertile of IGF-I had an OR=3.67 (95% CI: 1.63 – 8.24) compared to men in the first tertile of serum IGF-I. The corresponding OR was close to three-fold greater for Latino men in the third tertile of IGF-I (OR= 2.96, 95% CI: 1.19 – 7.40). For African-American men, those in the third tertile of IGF-I serum concentration had a 46% lower risk of prostate cancer (95% CI: 0.32 – 0.90) compared to men in the first tertile of IGF-I. We observed statistically significant trends in risk estimates for Latino men and African-American men (P trend = 0.02 for both groups). Latino men also had an increased risk of prostate cancer, as IGFBP-3 levels increased (P trend = 0.02). Men in the highest tertile of IGFBP-3 had 2.75 times the risk (95% CI: 1.20 – 6.32) of prostate cancer compared to men in the first tertile. Like the Latino men, Japanese-American men also had risk estimates above 1, but none were statistically significant. We observed no statistically significant interaction between IGFBP-3 levels and ethnicity, though there was a suggestion of a possible interaction (P interaction = 0.09).
Table 3.
Odds ratios and 95% confidence intervals for risk of prostate cancer by ethnicity, across tertiles of serum androgens, SHBG and growth factors
| Variable | Tertiles of serum concentration levels | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| African-Americans | Japanese-Americans | Latinos | P interaction* |
|||||||
| 1 | 2 | 3 | 1 | 2 | 3 | 1 | 2 | 3 | ||
| Testosterone | ||||||||||
| No. of cases | 75 | 68 | 63 | 24 | 28 | 36 | 22 | 30 | 26 | |
| Multivariate OR# | 1 | 0.99(0.64-1.54) | 0.86(0.54-1.38) | 1 | 0.61(0.31-1.20) | 0.81(0.42-1.55) | 1 | 1.68(0.79-3.57) | 1.34(0.61-2.94) | 0.21 |
| P-trend | 0.51 | 0.76 | 0.55 | |||||||
| Dihydrotestosterone | ||||||||||
| No. of cases | 65 | 66 | 75 | 28 | 26 | 34 | 23 | 34 | 21 | |
| Multivariate OR | 1 | 1.02(0.65-1.61) | 0.93(0.59-1.47) | 1 | 0.73(0.37-1.45) | 1.16(0.61-2.18) | 1 | 1.81(0.87-3.79) | 1.17(0.56-2.47) | 0.66 |
| P-trend | 0.71 | 0.55 | 0.82 | |||||||
| Testosterone/ Dihydrotestosterone ratio |
||||||||||
| No. of cases | 84 | 57 | 65 | 25 | 31 | 32 | 19 | 33 | 26 | |
| Multivariate OR | 1 | 0.89(0.59-1.36) | 1.12(0.72-1.72) | 1 | 0.88(0.45-1.72) | 0.72(0.36-1.47) | 1 | 1.53(0.73-3.21) | 1.21(0.54-2.74) | 0.45 |
| P-trend | 0.63 | 0.36 | 0.81 | |||||||
| 3α-diol G | ||||||||||
| No. of cases | 57 | 58 | 59 | 44 | 22 | 12 | 15 | 18 | 24 | |
| Multivariate OR | 1 | 0.84(0.52-1.37) | 0.87(0.54-1.40) | 1 | 0.78(0.41-1.47) | 0.96(0.43-2.17) | 1 | 1.40(0.57-3.44) | 1.03(0.47-2.25) | 0.56 |
| P-trend | 0.63 | 0.76 | 0.82 | |||||||
| SHBG | ||||||||||
| No. of cases | 66 | 67 | 73 | 24 | 27 | 37 | 29 | 29 | 20 | |
| Multivariate OR | 1 | 0.90(0.58-1.39) | 0.99(0.62-1.58) | 1 | 0.80(0.40-1.61) | 0.84(0.38-1.85) | 1 | 1.21(0.62-2.36) | 0.82(0.39-1.71) | 0.99 |
| P-trend | 0.96 | 0.73 | 0.58 | |||||||
| IGF-I | ||||||||||
| No. of cases | 58 | 55 | 39 | 30 | 25 | 27 | 15 | 33 | 21 | |
| Multivariate OR | 1 | 0.77(0.47-1.26) | 0.54(0.32-0.90) | 1 | 0.91(0.46-1.83) | 1.11(0.53-2.35) | 1 | 3.67(1.63-8.24) | 2.96(1.19-7.40) | <0.001 |
| P-trend | 0.02 | 0.75 | 0.02 | |||||||
| IGFBP-I | ||||||||||
| No. of cases | 64 | 54 | 34 | 15 | 18 | 49 | 27 | 25 | 17 | |
| Multivariate OR | 1 | 1.26(0.76-2.08) | 0.77(0.43-1.39) | 1 | 0.41(0.17-1.01) | 0.95(0.41-2.19) | 1 | 0.68(0.33-1.44) | 0.60(0.25-1.42) | 0.21 |
| P-trend | 0.34 | 0.30 | 0.28 | |||||||
| IGF-II | ||||||||||
| No. of cases | 53 | 43 | 32 | 18 | 28 | 27 | 15 | 18 | 15 | |
| Multivariate OR | 1 | 1.05(0.62-1.79) | 0.91(0.50-1.67) | 1 | 1.53(0.71-3.28) | 1.10(0.51-2.37) | 1 | 1.78(0.75-4.21) | 1.52(0.59-3.88) | 0.15 |
| P-trend | 0.80 | 0.94 | 0.34 | |||||||
| IGFBP-3 | ||||||||||
| No. of cases | 57 | 55 | 40 | 20 | 31 | 31 | 22 | 21 | 26 | |
| Multivariate OR | 1 | 1.02(0.64-1.64) | 1.00(0.57-1.76) | 1 | 1.56(0.75-3.25) | 1.34(0.64-2.78) | 1 | 1.72(0.77-3.83) | 2.75(1.20-6.32) | 0.09 |
| P-trend | 0.98 | 0.53 | 0.02 | |||||||
| IGF-I/IGFBP-3 | ||||||||||
| No. of cases | 40 | 62 | 50 | 30 | 30 | 22 | 21 | 24 | 24 | |
| Multivariate OR | 1 | 1.73(1.00-2.97) | 0.87(0.50-1.51) | 1 | 0.90(0.46-1.74) | 1.26(0.61-2.62) | 1 | 1.23(0.58-2.62) | 2.13(0.83-5.44) | 0.03 |
| P-trend | 0.35 | 0.54 | 0.12 | |||||||
Adjusted by conditional logistic regression for matched sets as strata and for age at blood draw, fasting hours prior to blood draw, body mass index, family history of prostate cancer, and education as covariates in the log-linear model.
P-value for Wald test of interaction between the three ethnic groups and each biomarker trend variable.
Table 4 shows results of an analysis restricted to men with advanced prostate cancer and their matched controls. The lack of association between any of the biomarkers and risk of prostate cancer persisted though the risk estimates were higher for DHT, 3α-diol G, IGF-I and the ratio of IGF-I/IGFBP-3 and lower for IGF-II compared to those in Table 2. Risk estimates remained relatively similar for SHBG, testosterone, testosterone/DHT ratio. For IGFBP-3, the risk estimate decreased when restricted to the advanced prostate cancer cases and their matched controls; the lack of association remained. Although we observed changes in the values of risk estimates for the advanced prostate cancer cases, we observed no statistically significant trends.
Table 4.
Odds ratios and 95% confidence intervals for risk of advanced prostate cancer across quartiles of serum androgens, SHBG and growth factors
| Variable | Quartiles of serum concentration levels | ||||
|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | P trend | |
| Testosterone | |||||
| No. of cases | 29 | 38 | 33 | 25 | |
| Multivariate OR# | 1 | 1.24(0.68-2.26) | 1.06(0.55-2.05) | 0.96(0.48-1.95) | 0.79 |
| Dihydrotestosterone | |||||
| No. of cases | 29 | 34 | 33 | 29 | |
| Multivariate OR | 1 | 1.48(0.76-2.86) | 1.66(0.83-3.29) | 0.94(0.47-1.90) | 0.73 |
| Testosterone/ Dihydrotestosterone ratio |
|||||
| No. of cases | 34 | 35 | 31 | 25 | |
| Multivariate OR | 1 | 1.17(0.64-2.15) | 1.00(0.53-1.92) | 0.84(0.42-1.68) | 0.52 |
| 3α-diol G | |||||
| No. of cases | 13 | 22 | 19 | 19 | |
| Multivariate OR | 1 | 2.22(0.91-5.43) | 1.74(0.75-4.06) | 1.46(0.60-3.53) | 0.75 |
| SHBG | |||||
| No. of cases | 38 | 29 | 32 | 26 | |
| Multivariate OR | 1 | 0.75(0.40-1.42) | 0.92(0.50-1.69) | 0.71(0.36-1.40) | 0.44 |
| IGF-I | |||||
| No. of cases | 28 | 26 | 25 | 28 | |
| Multivariate OR | 1 | 1.47(0.72-3.02) | 1.24(0.63-2.45) | 1.39(0.71-2.72) | 0.38 |
| IGFBP-I | |||||
| No. of cases | 30 | 24 | 22 | 31 | |
| Multivariate OR | 1 | 0.88(0.42-1.84) | 0.71(0.33-1.54) | 1.44(0.68-3.07) | 0.20 |
| IGF-II | |||||
| No. of cases | 21 | 12 | 14 | 11 | |
| Multivariate OR | 1 | 0.35(0.13-0.94) | 0.77(0.28-2.13) | 0.40(0.14-1.14) | 0.16 |
| IGFBP-3 | |||||
| No. of cases | 36 | 28 | 30 | 13 | |
| Multivariate OR | 1 | 1.15(0.61-2.19) | 1.95(0.97-3.92) | 0.53(0.23-1.23) | 0.57 |
| IGF-I/IGFBP-3 | |||||
| No. of cases | 17 | 34 | 23 | 33 | |
| Multivariate OR | 1 | 2.33(1.08-5.00) | 1.40(0.67-2.89) | 1.75(0.79-3.87) | 0.44 |
Adjusted by conditional logistic regression for matched sets as strata and for age at blood draw, fasting hours prior to blood draw, body mass index, family history of prostate cancer, and education as covariates in the log-linear model.
Discussion
In this study we observed no clear associations between serum levels of testosterone, DHT, 3α-diol G, SHBG, IGF-I, IGF-II, IGFBP-1, and IGFBP-3 and the risk of prostate cancer. We did observe an interaction between ethnicity and levels of IGF-I; Latino men were at increased risk of prostate cancer and African-American men had a decreased risk of prostate cancer. Latino men also had a higher risk of prostate cancer with increasing levels of IGFBP-3. Our overall findings were relatively unchanged when analyses were restricted to controls with normal PSA values and their matched cases. Analyses restricted to advanced prostate cancer cases and their matched controls did not change the overall conclusions though some risk estimates were higher.
Our null results for 3α-diol G agree with several prospective studies (7-13) that also found no association with prostate cancer. One study (13) did find an inverse association with aggressive prostate cancer cases ≥ 65 years old (OR = 0.52, 95% CI: 0.28 - 0.97). In contrast, the risk estimates for our advanced cases were all greater than 1.0, though none were statistically significant. Serum testosterone, DHT and SHBG levels were not associated with the risk of prostate cancer in our study, a finding that is consistent with the results of several prospective studies of testosterone (7-18) , DHT (7,9,11,16) and SHBG (7-15,17,18),. Three prospective studies also examined the ratio of testosterone to DHT. Two studies (7,16) reported no association with prostate cancer, in concordance with our results, while the Physician's Health Study (11) observed an increase in risk of prostate cancer (OR= 2.35, 95% CI:1.22-4.53 for the fourth quartile compared to the first). The discrepancy of this finding compared to other studies, including ours, could be due to differences in the hormone levels of the study populations. For example, the median value for the fourth quartile of testosterone in the Physician's Health Study was slightly lower than the median value in our study (7.02 ng/mL and 8.32 ng/mL, respectively) and the median value of the fourth quartile of DHT was much lower in the Physician's Health Study (0.68 ng/mL) compared to our study (0.94 ng/mL). Furthermore, this difference remained when we restricted our comparison to white cases in our study (mean DHT level of 0.59 ng/mL compared to a mean of 0.34 ng/mL in the Physician's Health Study). Therefore, the values for the testosterone to DHT ratios would be very different. One possible explanation for the differences may be attributed to differences in assay methodology. In our study testosterone and DHT were measured in the same aliquot of serum by RIA with preceding extraction and chromatography steps whereas in the Physicians' Health Study, testosterone was measured by direct RIA with a commercial kit and DHT was measured by RIA with preceding purification steps similar to the ones we used.
Our study observed no overall association between IGF-I levels and risk of prostate cancer. Other prospective studies (19-24) are in agreement with our null results. However, the Physician's Health Study (25) results were again discrepant, with an increased risk of advanced prostate cancer (RR= 5.1, 95% CI : 2.0 -13.2) for men in the fourth quartile compared to men in the first quartile. Our risk estimates for IGF-I were greater than 1.0 for men with advanced prostate cancer, however, none of the estimates were statistically significant. We did, however, observe a statistically significant positive association of IGF-I and prostate cancer among Latino men and a statistically significant inverse association among African-American men. We did perform the analysis restricting cases (and their matched controls) to those with blood drawn more than a year before diagnosis, but the results did not change materially. A case-control study (26) of African-American men and prostate cancer reported a non-statistically significant inverse association (OR= 0.67, 95% CI: 0.29 – 1.50) for the fourth quartile of serum IGF-I compared to the first quartile. However, since the blood was collected post-diagnostically, this may reflect reverse causation. No other prospective research studies with results for African-American or Latino men were found. Therefore, our results for IGF-I should be interpreted with caution until more research is done.
The role of IGF-I in the development of prostate cancer remains debatable. It may be that early studies, such as the Physician's Health Study, which took place before PSA screening was commonplace observed the association of growth factors with risk for progression to clinical significance rather than risk of prostate cancer diagnosis. A recent pooled analysis (1) concluded that high circulating IGF-I levels are associated with an increase in risk of prostate cancer (OR= 1.38, 95% CI : 1.19 – 1.60) , but results from most prospective studies, taken separately, showed no association implying that the association may be weak (19-24). Our observed increased risk for Latino men is interesting. The Latino men in our study had the lowest mean IGF-I levels of all ethnic groups (176 ng/ml compared to 186 ng/ml for African-American men and 185 ng/ml for Japanese-American men) which is contrary to the current belief that high IGF-I levels should be related to increased risk of prostate cancer because of IGF-I's cell proliferative and anti-apoptotic effects. Furthermore, it was the African-American group, not the Latinos, that made up the majority of our advanced prostate cancer cases, another inconsistency that is hard to reconcile with the strong positive results from the Physicians Health Study, (25) given that IGF-I had a negative association with prostate cancer for the African-Americans.
We observed no association between IGFBP-3 and the risk of prostate cancer. Several prospective studies support our null findings (19,20,23,24,27). However, two studies (21,22) reported a positive association between prostate cancer and IGFBP-3 levels. The Health Professionals Follow-up Study reported a 62% increased risk of prostate cancer (95% CI: 1.07 – 2.46) for men in the fourth quartile of serum IGFBP-3 compared to men in the first quartile (21). Results from the Melbourne Collaborative Cohort study (22) also showed an increased risk of prostate cancer for men in the fourth quartile of serum IGFBP-3 compared to the first quartile (HR: 1.49, 95% CI: 1.11 – 2.00). The reasons for the discrepancies in research of IGFBP-3 and prostate cancer are difficult to pinpoint. Both positive studies reported mean or median IGFBP-3 plasma levels lower than ours, but that is probably not the source of the differences in results since studies that have null results had similar IGFBP-3 values to the studies that found associations (19,20,24,27).
Few prospective studies have examined levels of IGFBP-I and IGF-II. A prospective study of Chinese men with prostate cancer (28) reported no associations of IGFBP-I or IGF-II and risk of prostate cancer, a finding confirmed by our results.
Our study had several strengths. It is a prospective study and specimens were collected before prostate cancer diagnosis. We were able to examine the consistency of risk estimates for three ethnic groups: African-Americans, Latinos and Japanese-Americans. Other prospective studies consisted mainly of Caucasian men. A limitation of our study was the lack of power in the analysis of advanced prostate cancer and the analysis by ethnic groups.
Overall, our study found no association of testosterone, DHT, 3α-diol G, SHBG, IGF-I, IGF-II, IGFBP-1, and IGFBP-3 and the risk of prostate cancer. The observed negative association of IGF-I with prostate cancer in African-Americans and the positive association of IGF-I with prostate cancer in Latino men are interesting, but need to be validated in other studies.
Grant Acknowledgements
The Multiethnic Cohort Study has been supported by USPHS (National Cancer Insitute) Grant R37 CA 54281 (PI: Dr. L.Kolonel).
Contributor Information
Jasmeet K Gill, Department of Health Science, California State University, Fullerton.
Lynne R Wilkens, Department of Epidemiology, University of Hawaii, Cancer Research Center.
Michael N Pollak, Cancer Prevention Research Unit, McGill University.
Frank Z. Stanczyk, Keck School of Medicine, University of Southern California.
Laurence N Kolonel, Department of Epidemiology, University of Hawaii, Cancer Research Center.
References
- 1.Roddam AW, Allen NE, Appleby P, Key TJ, Ferrucci L, Carter HB, Metter EJ, Chen C, Weiss NS, Fitzpatrick A, Hsing AW, Lacey JV, Jr., Helzlsouer K, Rinaldi S, Riboli E, Kaaks R, Janssen JA, Wildhagen MF, Schroder FH, Platz EA, Pollak M, Giovannucci E, Schaefer C, Quesenberry CP, Jr., Vogelman JH, Severi G, English DR, Giles GG, Stattin P, Hallmans G, Johansson M, Chan JM, Gann P, Oliver SE, Holly JM, Donovan J, Meyer F, Bairati I, Galan P. Insulin-like growth factors, their binding proteins, and prostate cancer risk: analysis of individual patient data from 12 prospective studies. Ann Intern Med. 2008;149(7):461–471. doi: 10.7326/0003-4819-149-7-200810070-00006. W483-468. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Roddam AW, Allen NE, Appleby P, Key TJ. Endogenous sex hormones and prostate cancer: a collaborative analysis of 18 prospective studies. J Natl Cancer Inst. 2008;100(3):170–183. doi: 10.1093/jnci/djm323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kolonel LN, Henderson BE, Hankin JH, Nomura AM, Wilkens LR, Pike MC, Stram DO, Monroe KR, Earle ME, Nagamine FS. A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol. 2000;151(4):346–357. doi: 10.1093/oxfordjournals.aje.a010213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Goebelsmann U, B GS, Gale JA, Kletzky OA, Nakamura RM, Coulson AH, Korelitz JJ. Serum gonadotropin, testosterone, estradiol and estrone levels prior to and following bilateral vasectomy. In: Lepow IH, C R, editors. Vasectomy: Immunologic and pathophysiologic effects in animals and man. Academic Press; New York: 1979. pp. 165–175. [Google Scholar]
- 5.Serafini P, Ablan F, Lobo RA. 5 alpha-Reductase activity in the genital skin of hirsute women. J Clin Endocrinol Metab. 1985;60(2):349–355. doi: 10.1210/jcem-60-2-349. [DOI] [PubMed] [Google Scholar]
- 6.Narang RRJ, Savjani G, Peterson J, Gentzschein E, Stanczyk FZ. Radioimmunoassay kit for the quantitative measurement of androstanediol glucuronide in unextracted serum; The 17th National Meeting of the Clinical Ligand Assay Society; Chicago, IL. 1991. [Google Scholar]
- 7.Dorgan JF, Albanes D, Virtamo J, Heinonen OP, Chandler DW, Galmarini M, McShane LM, Barrett MJ, Tangrea J, Taylor PR. Relationships of serum androgens and estrogens to prostate cancer risk: results from a prospective study in Finland. Cancer Epidemiol Biomarkers Prev. 1998;7(12):1069–1074. [PubMed] [Google Scholar]
- 8.Severi G, Morris HA, MacInnis RJ, English DR, Tilley W, Hopper JL, Boyle P, Giles GG. Circulating steroid hormones and the risk of prostate cancer. Cancer Epidemiol Biomarkers Prev. 2006;15(1):86–91. doi: 10.1158/1055-9965.EPI-05-0633. [DOI] [PubMed] [Google Scholar]
- 9.Mohr BA, Feldman HA, Kalish LA, Longcope C, McKinlay JB. Are serum hormones associated with the risk of prostate cancer? Prospective results from the Massachusetts Male Aging Study. Urology. 2001;57(5):930–935. doi: 10.1016/s0090-4295(00)01116-x. [DOI] [PubMed] [Google Scholar]
- 10.Chen C, Weiss NS, Stanczyk FZ, Lewis SK, DiTommaso D, Etzioni R, Barnett MJ, Goodman GE. Endogenous sex hormones and prostate cancer risk: a case-control study nested within the Carotene and Retinol Efficacy Trial. Cancer Epidemiol Biomarkers Prev. 2003;12(12):1410–1416. [PubMed] [Google Scholar]
- 11.Gann PH, Hennekens CH, Ma J, Longcope C, Stampfer MJ. Prospective study of sex hormone levels and risk of prostate cancer. J Natl Cancer Inst. 1996;88(16):1118–1126. doi: 10.1093/jnci/88.16.1118. [DOI] [PubMed] [Google Scholar]
- 12.Travis RC, Key TJ, Allen NE, Appleby PN, Roddam AW, Rinaldi S, Egevad L, Gann PH, Rohrmann S, Linseisen J, Pischon T, Boeing H, Johnsen NF, Tjonneland A, Overvad K, Kiemeney L, Bueno-de-Mesquita HB, Bingham S, Khaw KT, Tumino R, Sieri S, Vineis P, Palli D, Quiros JR, Ardanaz E, Chirlaque MD, Larranaga N, Gonzalez C, Sanchez MJ, Trichopoulou A, Bikou C, Trichopoulos D, Stattin P, Jenab M, Ferrari P, Slimani N, Riboli E, Kaaks R. Serum androgens and prostate cancer among 643 cases and 643 controls in the European Prospective Investigation into Cancer and Nutrition. Int J Cancer. 2007;121(6):1331–1338. doi: 10.1002/ijc.22814. [DOI] [PubMed] [Google Scholar]
- 13.Weiss JM, Huang WY, Rinaldi S, Fears TR, Chatterjee N, Hsing AW, Crawford ED, Andriole GL, Kaaks R, Hayes RB. Endogenous sex hormones and the risk of prostate cancer: a prospective study. Int J Cancer. 2008;122(10):2345–2350. doi: 10.1002/ijc.23326. [DOI] [PubMed] [Google Scholar]
- 14.Barrett-Connor E, Garland C, McPhillips JB, Khaw KT, Wingard DL. A prospective, population-based study of androstenedione, estrogens, and prostatic cancer. Cancer Res. 1990;50(1):169–173. [PubMed] [Google Scholar]
- 15.Heikkila R, Aho K, Heliovaara M, Hakama M, Marniemi J, Reunanen A, Knekt P. Serum testosterone and sex hormone-binding globulin concentrations and the risk of prostate carcinoma: a longitudinal study. Cancer. 1999;86(2):312–315. [PubMed] [Google Scholar]
- 16.Hsing AW, Comstock GW. Serological precursors of cancer: serum hormones and risk of subsequent prostate cancer. Cancer Epidemiol Biomarkers Prev. 1993;2(1):27–32. [PubMed] [Google Scholar]
- 17.Stattin P, Lumme S, Tenkanen L, Alfthan H, Jellum E, Hallmans G, Thoresen S, Hakulinen T, Luostarinen T, Lehtinen M, Dillner J, Stenman UH, Hakama M. High levels of circulating testosterone are not associated with increased prostate cancer risk: a pooled prospective study. Int J Cancer. 2004;108(3):418–424. doi: 10.1002/ijc.11572. [DOI] [PubMed] [Google Scholar]
- 18.Tsai CJ, Cohn BA, Cirillo PM, Feldman D, Stanczyk FZ, Whittemore AS. Sex steroid hormones in young manhood and the risk of subsequent prostate cancer: a longitudinal study in African-Americans and Caucasians (United States) Cancer Causes Control. 2006;17(10):1237–1244. doi: 10.1007/s10552-006-0052-4. [DOI] [PubMed] [Google Scholar]
- 19.Allen NE, Key TJ, Appleby PN, Travis RC, Roddam AW, Rinaldi S, Egevad L, Rohrmann S, Linseisen J, Pischon T, Boeing H, Johnsen NF, Tjonneland A, Gronbaek H, Overvad K, Kiemeney L, Bueno-de-Mesquita HB, Bingham S, Khaw KT, Tumino R, Berrino F, Mattiello A, Sacerdote C, Palli D, Quiros JR, Ardanaz E, Navarro C, Larranaga N, Gonzalez C, Sanchez MJ, Trichopoulou A, Travezea C, Trichopoulos D, Jenab M, Ferrari P, Riboli E, Kaaks R. Serum insulin-like growth factor (IGF)-I and IGF-binding protein-3 concentrations and prostate cancer risk: results from the European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol Biomarkers Prev. 2007;16(6):1121–1127. doi: 10.1158/1055-9965.EPI-06-1062. [DOI] [PubMed] [Google Scholar]
- 20.Lacey JV, Jr., Hsing AW, Fillmore CM, Hoffman S, Helzlsouer KJ, Comstock GW. Null association between insulin-like growth factors, insulin-like growth factor-binding proteins, and prostate cancer in a prospective study. Cancer Epidemiol Biomarkers Prev. 2001;10(10):1101–1102. [PubMed] [Google Scholar]
- 21.Platz EA, Pollak MN, Leitzmann MF, Stampfer MJ, Willett WC, Giovannucci E. Plasma insulin-like growth factor-1 and binding protein-3 and subsequent risk of prostate cancer in the PSA era. Cancer Causes Control. 2005;16(3):255–262. doi: 10.1007/s10552-004-3484-8. [DOI] [PubMed] [Google Scholar]
- 22.Severi G, Morris HA, MacInnis RJ, English DR, Tilley WD, Hopper JL, Boyle P, Giles GG. Circulating insulin-like growth factor-I and binding protein-3 and risk of prostate cancer. Cancer Epidemiol Biomarkers Prev. 2006;15(6):1137–1141. doi: 10.1158/1055-9965.EPI-05-0823. [DOI] [PubMed] [Google Scholar]
- 23.Weiss JM, Huang WY, Rinaldi S, Fears TR, Chatterjee N, Chia D, Crawford ED, Kaaks R, Hayes RB. IGF-1 and IGFBP-3: Risk of prostate cancer among men in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial. Int J Cancer. 2007;121(10):2267–2273. doi: 10.1002/ijc.22921. [DOI] [PubMed] [Google Scholar]
- 24.Woodson K, Tangrea JA, Pollak M, Copeland TD, Taylor PR, Virtamo J, Albanes D. Serum insulin-like growth factor I: tumor marker or etiologic factor? A prospective study of prostate cancer among Finnish men. Cancer Res. 2003;63(14):3991–3994. [PubMed] [Google Scholar]
- 25.Chan JM, Stampfer MJ, Ma J, Gann P, Gaziano JM, Pollak M, Giovannucci E. Insulin-like growth factor-I (IGF-I) and IGF binding protein-3 as predictors of advanced-stage prostate cancer. J Natl Cancer Inst. 2002;94(14):1099–1106. doi: 10.1093/jnci/94.14.1099. [DOI] [PubMed] [Google Scholar]
- 26.Hernandez W, Grenade C, Santos ER, Bonilla C, Ahaghotu C, Kittles RA. IGF-1 and IGFBP-3 gene variants influence on serum levels and prostate cancer risk in African-Americans. Carcinogenesis. 2007;28(10):2154–2159. doi: 10.1093/carcin/bgm190. [DOI] [PubMed] [Google Scholar]
- 27.Li L, Yu H, Schumacher F, Casey G, Witte JS. Relation of serum insulin-like growth factor-I (IGF-I) and IGF binding protein-3 to risk of prostate cancer (United States) Cancer Causes Control. 2003;14(8):721–726. doi: 10.1023/a:1026383824791. [DOI] [PubMed] [Google Scholar]
- 28.Chokkalingam AP, Gao YT, Deng J, Stanczyk FZ, Sesterhenn IA, Mostofi FK, Fraumeni JF, Jr., Hsing AW. Insulin-like growth factors and risk of benign prostatic hyperplasia. Prostate. 2002;52(2):98–105. doi: 10.1002/pros.10096. [DOI] [PubMed] [Google Scholar]