Vitamin D Receptor Gene Polymorphisms, Insulin-like Growth Factors, and Prostate Cancer Risk: A Population-based Case-Control Study in China

Laboratory studies demonstrate a strong and consistent prodifferentiative and growth-inhibitory effect of the steroid hormone 1,25(OH)₂D₃² and its analogues on prostate cancer both in vitro and in vivo (1). Because the effects of 1,25(OH)₂D₃ are mediated by the VDR (2) and because both normal and malignant prostate cells express active VDR (2), it has been suggested that polymorphic markers within the VDR gene may be related to prostate cancer risk. To date, a number of such markers have been identified, including the 3′ BsmI and the 5′ FokI markers; the latter have been shown to have functional effects. However, epidemiological studies of these two VDRpolymorphisms have been inconsistent, with both positive (3) and null (4, 5, 6) results. In recent laboratory studies, the inhibition of prostate cancer cell growth by 1,25(OH)₂D₃ and its analogues has been associated with increased expression of IGFBPs and decreased expression of IGF-II (7, 8, 9). Because IGFBPs inhibit the mitogenic actions of IGFs on cancer cells (10, 11) and because we and others have found previously that systemic levels of IGFBP-3 and IGFBP-1 are inversely associated with prostate cancer risk (12, 13), it has been suggested that the growth inhibition of prostate cancer cells induced by vitamin D through the VDR occurs via a pathway involving the IGF axis (9). To date, no epidemiological studies have addressed the potential combined effects of the IGF axis and vitamin D regulatory pathway. In the current population-based investigation, we examined the a priori hypothesis that VDR gene polymorphisms affect prostate cancer risk, both independently and in conjunction with plasma levels of IGFBP-1, IGFBP-3, IGF-I, and IGF-II, among newly diagnosed prostate cancer cases and healthy controls randomly selected from the population in Shanghai, China.

Details of this study have been described previously (14, 15). Briefly, histologically confirmed cases of primary prostate cancer newly diagnosed in urban Shanghai between 1993 and 1995 were identified through a rapid reporting system established by the Shanghai Cancer Registry. On the basis of a regional registry of all persons >16 years in urban Shanghai, male population controls were selected at random from the 6.5 million permanent residents and frequency-matched to the expected age distribution (by 5-year age category) of the cases. Using a structured questionnaire, trained interviewers elicited information on epidemiological risk factors from cases and controls within 30 days after selection. Anthropometric measures were taken during the interview. Of the 268 cases who were permanent residents of Shanghai (95% of cases diagnosed in Shanghai during the study period), 243 (91%) were interviewed. Of the 495 selected controls, 472 (95%) were interviewed. This study was approved by the Office of Human Subjects Research, NIH, and the Institutional Review Board, Shanghai Cancer Institute, Shanghai, China.

Two hundred cases and 330 controls (82 and 70% of those interviewed, respectively) provided 20 ml of fasting blood for the study. Samples were processed within 3 h of collection at a central laboratory in Shanghai. DNA extracted from the buffy coat fractions was used for VDR genotyping, whereas IGFs and IGFBPs were quantified from the plasma samples. All laboratory personnel were masked to case-control status, and samples from cases and controls were physically arranged and assayed in an alternating fashion to minimize bias attributable to day-to-day laboratory variation.

Plasma levels of IGF-I, IGF-II, IGFBP-1, and IGFBP-3 were determined using ELISA kits from Diagnostic Systems Laboratories (Webster, TX). Analytical sensitivities of the assays are 0.03, 2.4, 0.04, and 0.04 ng/ml, respectively. For all four analytes, each sample was assayed twice, and the mean of the two determinations was used for data analysis.

We genotyped the study subjects for both the 5′ FokI and the 3′ BsmI VDR markers; the latter were in linkage disequilibrium with other 3′ markers, including TaqI, and poly(A) (2). For each VDR gene marker, 50 ng of genomic DNA were amplified by PCR in a total volume of 10 μl using the primers described by Morrison et al. (16) for the BsmI marker and by Gross et al. (17) for the FokI marker. After sequence-specific digestion with either 2 units of BsmI or 0.8 unit of FokI restriction enzyme (New England Biolabs), the samples were electrophoresed through a 2% agarose gel containing ethidium bromide and scored for genotypes of the BsmI (bb, Bb, or BB) and FokI (FF, Ff, or ff) markers, where lowercase and uppercase letters indicate alleles in which the restriction sites are present and absent, respectively.

We performed Mantel-Haenszel χ² analyses to assess the association of the FokI and BsmI markers with prostate cancer. To avoid a potential treatment effect among cases, only cases whose blood samples were collected at least 1 day prior to treatment were included in analyses involving plasma levels of IGFs (n = 120). Stratified analyses were used to identify potential confounding factors of the BsmI and FokI markers and to assess potential combined effects with IGFs and IGFBPs. Using pairwise t tests from linear regression models, we compared the age-adjusted mean plasma levels of IGFBP-1, IGFBP-3, IGF-I, and IGF-II across VDR FokI genotypes among controls. Unconditional logistic regression was used to generate ORs and 95% CIs estimating the combined effects of the VDR FokI marker with either IGFs or IGFBPs on prostate cancer after adjustment for other potential risk factors, including age, anthropometric factors, and dietary intake (18). Levels of IGF-I, IGF-II, IGFBP-1, and IGFBP-3 among cases and controls were categorized according to medians and tertiles defined by the distributions among controls. All presented Ps are two-sided.

One-hundred ninety-one prostate cancer cases and 304 controls were assayed for the BsmI and FokI markers. Compared with controls, cases were more likely to have an education level higher than middle school, less likely to be married, and less likely to smoke or consume alcohol.

The BsmI and FokI genotyping results are shown in Table 1. Among the population controls, the prevalences of each of the bb, Bb, and BB BsmI genotypes were 87.2, 10.4, and 2.4%, respectively, yielding only 7 controls with the BB genotype. For the FokI marker, the prevalences of the FF, Ff, and ff FokI genotypes were 28.8, 50.7, and 20.5%, respectively. Both markers were in Hardy-Weinberg equilibrium. We found no significant association of either marker with total prostate cancer risk or stage-specific cancer (clinical staging: early versus late). However, despite the fact that early-stage cancers made up only one-third of the total case group, all four cases with the BB BsmI genotype had early-stage disease.

To assess whether systemic levels of IGFBPs and IGFs are associated with VDR genotype, we compared age-adjusted means of IGFBP-1, IGFBP-3, IGF-I, and IGF-II in each genotype of the FokI marker among the 304 population controls (Table 2). Because of the rarity of the B allele of the BsmI marker in this study population, we did not assess this marker in relation to IGF/IGFBP levels or to their combined effects on prostate cancer risk. Those with the ff genotype of FokI had higher IGFBP-1 levels (P = 0.07) and significantly lower IGF-II levels (P = 0.03) than those with the FF and Ff genotypes. However, mean levels of IGFBP-3 and IGF-I did not differ significantly across FokI genotypes.

Among those with the FF and Ff FokI genotypes, IGFBP-1 was not significantly associated with prostate cancer risk after adjusting for age and IGF-I (Table 3). In contrast, among those with the ff genotype, those in the highest tertile of IGFBP-1 had a 75% decreased risk compared with those in the lowest tertile, with a significant trend (OR, 0.25; 95% CI, 0.07–0.85; P_trend = 0.02). Similar results were observed for the combined effects of FokI with IGFBP-3 levels. There was no significant association of IGFBP-3 with prostate cancer among those with the FF and Ff FokI genotypes, but among those with the ff genotype there was a significant 86% decreased risk in the highest tertile of IGFBP-3, along with a significant trend (OR, 0.14; 95% CI, 0.04–0.56; P_trend < 0.01). However, no differences were observed in the association of either IGF-I or IGF-II with prostate cancer across strata of FokI genotypes.

In addition, the risk of disease associated with the ff FokI genotype relative to the FF and Ff genotypes differed by tertiles of systemic IGFBP-1 and IGFBP-3. Among those in the lowest IGFBP-1 tertile, the ff genotype was associated with increased risk relative to the FF and Ff genotypes (OR, 2.08; 95% CI, 0.86–5.01), whereas among those in the highest tertile of IGFBP-1, the ff genotype was associated with a reduced risk (OR, 0.39; 95% CI, 0.14–1.11). Similarly, among those in the lowest tertile of IGFBP-3 levels, the ff genotype was associated with increased risk of prostate cancer relative to the FF and Ff genotypes, whereas among those with the highest IGFBP-3 levels (OR, 2.20; 95% CI, 0.94–5.11), the ff genotype was associated with a reduced risk (OR, 0.39; 95% CI, 0.12–1.25). In contrast, the effect of the ff genotype relative to the FF and Ff genotypes did not differ markedly by tertiles of IGF-I or IGF-II.

In this low-risk Chinese population, there was no evidence linking either the BsmI or FokI polymorphisms of the VDR gene with prostate cancer risk, although small to moderate effects of BsmI marker cannot be ruled out because of the low prevalence of its B allele. We have reported previously that systemic levels of IGFBP-1 and IGFBP-3 in this population were inversely associated with prostate cancer risk, whereas plasma levels of IGF-I showed a positive relation to risk (12). In the current study, we found that the inverse associations of IGFBP-1 and IGFBP-3 levels were mainly limited to subjects with the ff genotype of the FokI marker, whereas the effects of IGF-I did not differ across strata of FokI genotypes.

Previous studies of the BsmI polymorphism among African-Americans (4) and Caucasian-Americans (5) have also shown no association with prostate cancer risk, although a study among Japanese men (also a low-risk population) found a significant reduction in risk associated with the B allele (3). The higher prevalence of the B allele in the Japanese study (3) relative to the current investigation (27% versus 8% among controls, respectively) is unlikely to explain the difference in results, because both United States studies had even higher B allele prevalences (32–41%; Refs. 4 and 5).

In contrast to the BsmI marker and other 3′ polymorphisms of the VDR gene [including ApaI, TaqI, and poly(A)], polymorphism of the 5′ FokI site alters the VDR amino acid sequence; the F and f alleles of the FokI marker encode VDR proteins of 424 and 427 amino acids, respectively (19). Recent data suggest that the VDR coded by the F allele of the FokI polymorphism is more responsive to vitamin D than that coded by the f allele (19). However, we found no significant association of the FokI polymorphism with prostate cancer risk, despite high prevalence of the f allele (46%). Similarly, the only other published study of FokI and prostate cancer, conducted in a European population with an f allele prevalence of 34%, also showed no association (6).

In cross-sectional analyses, we found that controls with the ff genotype of the FokI marker had significantly higher mean IGFBP-1 levels and significantly lower mean IGF-II levels than those with the FF or Ff genotypes. Indeed, systemic IGFBP-1 levels were elevated in parallel with increasing copies of the f FokI allele. Given that we previously found that IGFBP-1 levels were inversely related to prostate cancer risk in this population (12), our results suggest that the reduced risk associated with the ff genotype might be mediated through elevated levels of IGFBP-1. However, the overall null results for FokI polymorphism genotypes on prostate cancer risk seem inconsistent with this notion, calling for further studies to elucidate the relationships observed. Furthermore, because systemic IGF-II levels have not been found to be associated with prostate cancer risk in our study and in others (11, 12), the implication of the significant difference in systemic IGF-II levels between the FF/Ff and the ff genotypes is unclear.

In the current study, we found that levels of IGFBP-1 and IGFBP-3 were associated with reduced prostate cancer risk only among men with the ff genotype of the FokI polymorphism, but neither of the IGFBPs was related to prostate cancer risk among those with the FF and Ff genotypes. Given that the shorter VDR encoded by the F allele may be more effective at exerting vitamin D effects than that coded by the f allele (19), our results suggest that systemic levels of IGFBP-3 and IGFBP-1 are associated with risk reduction only among men with presumably lower VDR function. This issue should be investigated further, particularly because increases in IGFBP levels after vitamin D administration in laboratory studies occur at the local rather than systemic level (7, 8, 9), and it is as yet unclear how systemic and local levels of IGFBPs and IGFs are related.

The association of the ff FokI genotype with prostate cancer relative to the FF and Ff genotypes differed by systemic IGFBP-3 and IGFBP-1 levels. Indeed, the ff genotype was associated with increased risk of prostate cancer in the lowest IGFBP-1 and IGFBP-3 tertiles and decreased risk in the highest tertiles. Reasons for these findings are unclear.

The biological mechanism underlying the differential effects of IGFBP-3 and IGFBP-1 according to FokI genotype, if confirmed in other studies of prostate cancer, is as yet unclear. One possible mechanism is that, if the VDR coded by the F FokI allele (F-VDR) is indeed more transcriptionally active than that coded by the f allele (19), men with the F allele (either the FF or Ff genotypes) may more easily activate the local, prostatic IGFBP expression that occurs in rats and in cultured human cells in response to vitamin D administration (7, 8, 9), thus inhibiting prostate cellular proliferation irrespective of systemic IGFBP levels (Fig. 1,A). In contrast, men without the VDR coded by the F allele (those with the ff genotype) would have a lowered production of local IGFBPs so that inhibition of IGF-mediated cellular proliferation would be more dependent on systemic IGFBP levels (Fig. 1 B). It should be noted, however, that it is as yet unclear whether systemic IGFBPs can influence local IGFBP levels. Other possible mechanisms exist as well; because IGFBP-3 can bind to the retinoid X receptor α (an obligatory cofactor of the VDR) to induce apoptosis in prostate cancer cells (20), IGFBPs may also interact directly with the VDR, or even replace it under certain circumstances, to effect transcriptional change. Such mechanisms are speculative at this point but may provide directions for further research to explain the interactions observed in our study. It is also noteworthy that although the FokI genotypes appeared to interact with systemic levels of both IGFBP-1 and IGFBP-3, the correlation of FokI genotype with levels of IGFBP-1, but not IGFBP-3, suggest that different mechanisms may be involved for each observed interaction.

Because the plasma samples from the cancer cases in this study were collected after diagnosis, it is possible that the presence of disease among cases affected the results of this study. However, this potential disease effect, if present, is likely to be minimal, because the age-adjusted mean levels of IGFBP-3, IGF-I, and IGF-II did not differ appreciably between localized and advanced stage cancer cases. Levels of IGFBP-1 were slightly higher among localized relative to advanced stage cases (112.0 versus 101.0 ng/ml, respectively), but not significantly so (P = 0.35), and it is unclear whether this difference is biologically meaningful. In addition, the associations with prostate cancer of IGF-I and IGFBP-3 reported previously in this study population (12) are similar in direction and magnitude to those reported in a prospective study (13), which is not subject to disease effects. Finally, despite the fact that the combined effects of systemic IGFBP levels and FokI genotypes were based on small numbers, it is noteworthy that no such effects were seen for the IGFs, a very different class of proteins.

In summary, the results of this study suggest that the BsmI VDR gene polymorphism is not related to prostate cancer (though a small effect cannot be ruled out), and that the ff genotype of the FokI VDR gene polymorphism may combine with systemic levels of IGFBP-3 and IGFBP-1 to modulate disease risk. Indeed, it appears that the inverse association of both IGFBP-3 and IGFBP-1 with prostate cancer we observed previously in this population (12) is confined to men with the ff FokI genotype. These results suggest that the vitamin D regulatory system and the IGF axis may interact to influence prostate cancer risk. These findings should be explored in large prospective investigations of populations at varying risk of prostate cancer that include measurements of circulating vitamin D metabolites.