Prostate Cancer Risk in Relation to Selected Genetic Polymorphisms in Insulin-like Growth Factor-I, Insulin-like Growth Factor Binding Protein-3, and Insulin-like Growth Factor-I Receptor

We conducted a nested case-control study within a cohort of elderly Americans to examine the role of the insulin-like growth factor (IGF) signaling pathway in prostate cancer etiology. The distribution of genotypes of IGF-I (CA)_n, IGF binding protein-3 (IGFBP-3) A-202C, and of the 2-bp deletion and (AGG)_n polymorphisms in IGF-I receptor (IGF-IR) was compared between men with prostate cancer (n = 213) and equal number of controls matched on year of blood draw, survival until the date of diagnosis, race, and age. Among controls, the number of CA repeats in IGF-I was not correlated to any appreciable degree with plasma IGF-I concentration, whereas the IGFBP-3 CC genotype was associated with a relatively low level of plasma IGFBP-3. There was no association between prostate cancer risk and the number of CA repeats in IGF-I, IGFBP-3 genotype, or the presence of the 2-bp deletion in IGF-IR. There was a small increased risk among men who did not carry two copies of the (AGG)₇ allele of IGF-IR. These results add to the evidence that the number of IGF-I CA repeats is not associated with prostate cancer risk. Our observation that men who do not carry two copies of the IGF-IR (AGG)₇ allele are at increased risk of prostate cancer merits further investigation. (Cancer Epidemiol Biomarkers Prev 2006;15(12):2461–6)

Insulin-like growth factor-I (IGF-I), produced primarily by the liver, plays an important role in the regulation of cell proliferation, differentiation, and apoptosis (1-3). Prostatic epithelial and stromal cells express IGF-I receptors (IGF-IR) and are responsive to IGF-I (4-6). In addition to their involvement in normal cell growth, IGF-IR seem to play a crucial role in tumor transformation and the survival of tumor cells through their interactions with oncogenes and tumor suppressor genes (7), and through regulation of metalloproteinases synthesis and tumor invasion and metastasis (8). The ability of circulating IGF-I to bind to IGF-IR is modulated by at least six high-affinity IGF-binding proteins (IGFBP; refs. 9, 10). More than 95% of circulating IGF-I is complexed with IGFBP-3 and an acid-labile subunit (11). In addition to modulating circulating IGF-I levels, IGFBP-3 has been shown to have IGF-I–independent apoptotic effects (12).

A number of studies have examined the levels of IGF-I and IGFBP-3 in blood samples taken from men before a diagnosis of prostate cancer and compared these results with the corresponding levels in controls (13-26); however, the results have been inconsistent. In each of these studies, IGF-I and IGFBP-3 levels were measured on blood samples obtained at just one point in time. Because a man's genotype might be a relatively better indicator of endogenous exposure over his lifetime, we sought to examine the possible influence of genes encoding for IGF-I, IGFBP-3, and IGF-IR on the occurrence of prostate cancer using samples and data from a large cohort study of elderly Americans. The polymorphisms we examined included the CA dinucleotide repeats in the promoter region of IGF-I, the A-202C polymorphism in the promoter region of IGFBP-3, and the AGG repeat polymorphism and a 2-bp deletion polymorphism in the 3′ untranslated region of IGF-IR. Allelic variation in the promoter region may influence the transcription of the gene, and a mutation in the 3′ untranslated region may influence the stability of the transcribed RNA. We also evaluated the correlations between genotypes of IGF-I CA repeat and IGFBP-3 A-202C polymorphisms and concentrations of circulating IGF-I and IGFBP-3, respectively, among healthy controls to examine the potential functional impact of these polymorphisms.

This study, described in detail previously (19), was conducted among participants in the Cardiovascular Health Study (CHS; ref. 27). The CHS recruited 5,888 individuals 65 years of age or older from 1989 to 1993 using Medicare eligibility lists from four U.S. communities: Sacramento County, CA; Forsyth County, NC; Pittsburgh, PA; and Washington County, MD (28). The institutional review board at each participating center approved the study, and each participant gave informed consent. Participants completed up to 10 annual clinical examinations from baseline until 1998 or 1999. A fasting blood specimen was collected during these exams, as well as demographic data and medical history using a standardized questionnaire. Surveillance for all hospitalizations with extensive medical record collection was initiated and continues to be carried out (29).

Incident prostate cancer cases were identified by linking CHS data to available records from 1989 to 1999 of population-based cancer registries serving the four CHS communities. The registries include the California Cancer Registry; North Carolina Central Cancer Registry; Pennsylvania Department of Health, Bureau of Health Statistics and Research; Maryland Cancer Registry; and Johns Hopkins Training Center for Public Health Research. Cancer registry records were estimated to be 93% to 100% complete as measured by the standards of the North American Association of Central Cancer Registries. Cases in this study had either a registry-confirmed diagnosis of prostate cancer or had documentation of prostate cancer by both self-report (at the time of an annual CHS examination) and a hospital discharge diagnosis code in CHS records.

Men were eligible to serve as controls only if they had no documentation of prostate cancer in either the cancer registries, in self-report at the examinations, or in the abstracted medical records for all hospitalizations. Both cases and potential controls (∼10% of each group) were excluded from this study if they had an incident myocardial infarction or stroke and, thus, whose blood specimens were reserved for studies of cardiovascular disease. Controls were individually matched to cases on race, year of entry, age at enrollment (within 4 years), and clinic. Cases with plasma were also matched to controls by year of blood draw. Controls were also required to have survived at least to the age at diagnosis of the matched case to ensure that cases and controls experienced a comparable risk period for the development of prostate cancer.

Assay procedures for plasma concentrations of IGF-I and IGFBP-3 had been reported previously (19). DNA samples were prepared at the CHS Central Laboratory in the University of Vermont and sent to the Chen Laboratory in the Fred Hutchinson Cancer Research Center for genotype analyses.

PCR primers and thermal cycling conditions that we used to determine the genotypes of polymorphisms of interest are listed in Table 1. The number of IGF-I (CA)_n repeats and the number of IGF-IR (AGG)_n repeats were assessed by using primers described by Rosen et al. (30) and Meloni et al. (31), respectively. Samples that failed to amplify IGF-IR (AGG)_n repeats were subjected to an initial PCR using nested primers (Table 1). For the IGF-IR 2-bp deletion, we used the antisense primer described by Poduslo et al. (32). The labeled PCR products of the IGF-I (CA)_n, IGF-IR (AGG)_n, and IGF-IR 2-bp deletion were mixed with GeneScan-500 TAMRA size standard, denatured in Hi-Di formamide (Applied Biosystems, Inc., Foster City, CA) at 95°C for 2 min, and subjected to capillary electrophoresis on an Applied Biosystems Genetic Analyzer. The Applied Biosystems GeneScan and Genotyper software was used for allele/genotype calling. Any samples with anomalous peak morphologies were sequenced to confirm repeat number or deletion status. Positive controls were purified, and plasmids containing 18, 20, and 21 IGF-I CA repeats; 6 or 7 IGF-IR (AGG) repeats; or genomic DNA samples with known IGF-IR 2-bp-deletion genotypes were sequenced. Negative controls were prepared identically, but without the DNA template. To test for the presence or absence of a GA dinucleotide (rs11829707³

) at the end of the IGF-I (CA)_n repeat, a single nucleotide primer extension protocol based on the Applied Biosystems SNaPshot Multiplex System was used. The SNaPshot reaction extended the sequence with 1 bp labeled with fluorescent dye(s) and revealed whether the genomic DNA contained a GA or CA dinucleotide at the end of the CA repeats. DNA from individuals heterozygous for both the GA and CA dinucleotide and the IGF-I (CA)_n repeats were sequenced to determine which CA-repeat allele contained the GA dinucleotide. Positive controls for this procedure were cloned, purified and sequenced plasmids of known GA status and CA repeat length.

The DNA fragment containing the IGFBP-3 A-202C polymorphism was amplified using the primer sequences of Deal et al. (33), followed by restriction digestion by BsiHKA I (New England Biolabs, Beverly, MA) and agarose gel electrophoresis. Genomic samples of known genotypes served as positive controls. Negative controls contained no DNA template.

DNA samples were available for 214 cases and 220 controls. Seven controls were excluded because their matched case did not have a DNA sample. One case was excluded because his matched control did not have a DNA sample. This left 213 matched sets for analyzing the relationship of genotype to prostate cancer.

The matched case-control sets were used to calculate odds ratios (as estimates of the relative risks) and to compute 95% confidence intervals using conditional logistic regression (Stata Statistical Software: Release 8, Stata Corporation, College Station, TX).

There were 163 controls with plasma and DNA. Among controls, 4.3% of the Caucasians and 7.5% of the African Americans had one GA at the end of the CA repeats in IGF-I. The presence of the GA at the end of the CA repeats was not associated with plasma IGF-I level. Therefore, we counted the GA as one of the CA repeats to allow us to compare results from our current study with those of others who inferred the CA repeat numbers by the analysis of fragment size of PCR products without investigating the presence or absence of GA. Using the PCR primers described above, the 192-bp PCR product correspond to a (CA)₁₉ allele or a (CA)₁₈ GA allele. The associations of genotypes with plasma levels of IGF-I and IGFBP-3 were assessed using the ANOVA F test for equality of means. (We decided to present adjusted values in the tables.) The Kuskal-Wallis nonparametric test was used to compute P values when variances were unequal.

Among the study participants, 81.7% were Caucasians; 17.8% were African Americans; and 0.5% were other races/ethnicities. The mean age for the cases and controls was 72.8 years (SE 0.335). Twenty-eight Caucasians and four African Americans had a prostate cancer diagnosis within a year after enrollment and one Caucasian and one African American had diagnosis within a month after enrollment. The interval between the time of enrollment to the time of diagnosis ranged from 1 to 113 months (mean, 42.9 months) among Caucasians and from 1 to 118 months (mean, 38.6 months) among African Americans.

The number of CA repeats in IGF-I ranged between 11 and 22, with 19 repeats being the most common number. We did not observe a monotonic relation between the number of CA repeats and the plasma IGF-I concentration among controls. However, there was a suggestion that men having two (CA)₁₉ repeat alleles had lower concentrations of plasma IGF-I than men who had no or one (CA)₁₉ repeat allele (Table 2). Among controls, men with the CC genotype of the IGFBP-3 A-202C polymorphism tended to have a lower circulating concentration of IGFBP-3 than men with the AA or AC genotypes (Table 3). Neither these associations, nor the mean levels of IGF-I and IGFBP-3 among controls, differed between Caucasians and African Americans (P = 0.48 and 0.19 for mean IGF-I and IGFBP-3, respectively, adjusted for age). There was a suggestion of an inverse relationship between the number of AGG repeats in IGF-IR and circulating concentrations of IGF-I (Table 4).

There was no association between prostate cancer risk and the number of CA repeats in IGF-I, the IGFBP-3 A-202C genotype, or the 2-bp deletion in IGF-IR (Table 5). We did, however, observe a slight increase in risk among individuals who did not carry two copies of the (AGG)₇ repeat allele of IGF-IR compared with those who did. The odds ratios ranged between 1.25 and 2.18, depending on the particular genotype (Table 4). The odds ratios [matched for race, year of entry, age at enrollment (within 4 years), and clinic] associated with the various genotypes were similar when cases were defined as men with “aggressive” tumors (i.e., stage C or D or poorly differentiated; n = 68; data not shown).

We examined the correlation between polymorphisms in the regulatory regions of the IGF-I, IGFBP-3, and IGF-IR genes with circulating levels of IGF-I and IGFBP-3, as well as the association of these polymorphisms with prostate cancer risk among CHS participants. A number of studies have evaluated the association between the presence of one or two (CA)₁₉ repeat alleles in IGF-I and plasma IGF-I concentration. Although some studies reported no genotypic association, other studies showed either a positive or a negative association (34-37). Results of the current study add to the evidence that there is no association between the IGF-I (CA)₁₉ allele and plasma IGF-I concentration.

The handful of studies that evaluated the association of IGF-I (CA)_n genotype and prostate cancer risk have produced conflicting results. A study of the Japanese population found a positive association between carriers of one or two (CA)₁₉ repeat alleles and prostate cancer risk (38); a study among samples drawn from the U.S. population found a negative association (38); three other U.S. studies, in addition to the present one, reported no association (39-41). The lack of influence of the IGF-I (CA)_n genotype on the risk of prostate cancer is not surprising given our finding (and that of others as described above) that the IGF-I (CA)_n genotype is unrelated to levels of IGF-I in plasma. Results of a recent study conducted within the Multiethnic Cohort suggest that inherited variation in the coding region of this gene may play a role in prostate cancer risk (42). Using 29 tagging single nucleotide polymorphisms to capture a large part of the genetic variation of the entire coding sequence of IGF-I, that study found a statistically significant 20% to 25% increased risk of prostate cancer associated with four haplotypes and 1.2- to 2.7-fold increased risk among men who carried variant genotypes of two perfectly correlated single nucleotide polymorphisms (42). Studies using similar approaches may help further elucidate the role of genetic variation in IGF-I in prostate cancer development.

Our results suggest that the IGFBP-3 A-202C polymorphism is associated with circulating IGFBP-3 concentration, with the CC genotype associated with the lowest levels. This is consistent with results from other published studies to date among men and women, and among Caucasians and Asians (33, 35, 37, 43-48), suggesting that this polymorphism may be a good surrogate in the evaluation of exposure of target organs to endogenous IGFBP-3 and a good marker for the evaluation of cancer risk. Two prior studies observed a weak and statistically imprecise elevation in risk of prostate cancer associated with having the C allele (39, 49). We found no evidence of any such increase, whether a man has one or two C alleles. Nonetheless, given the consistent observations of a genotype-phenotype correlation and the possibility that circulating levels of IGFBP-3 are associated with prostate cancer, further evaluation of this polymorphism in relation to the incidence of prostate cancer risk is warranted.

Recent evidence suggests that the IGF-IR may play an important role in cancer biology (50). It is involved in cell proliferation in response to IGF stimulation. In experimental systems, IGF-IR has been shown to be involved in cell invasion; activation of Ras proto-oncogene by IGF-IR signaling is required for the invasion of a pancreatic carcinoma cell line in matrigel (51). IGF-IR induces actin filament disassembly in MCF-7 breast epithelial cells (52). It also seems to be involved in the regulation of cell-cell and cell-matrix adhesion in a cell type–specific manner, with implications for cell survival, tumor invasion, and metastasis (8, 53, 54). Androgens, such as dihydrotestosterone and synthetic androgen R19881, up-regulate IGF-IR in androgen receptor–positive prostate cancer cells LNCaP (55). The increased IGF-IR expression has been shown to be involved in androgen-independent antiapoptotic and mitogenic IGF signaling in a LNCaP human prostate cancer progression model (56). IGF-IR signaling is associated with the acquired resistance of prostate cancer cell DU145 to the antitumor drug gefitinib (ZD1839, Iressa; ref. 57). For these reasons, the alteration of IGF-IR function could influence prostate cancer risk. To our knowledge, there have been neither reports of the correlation of the polymorphisms of IGF-IR evaluated in the current study with IGF-IR function, nor reports of the association of these polymorphisms with cancer risk. Our results suggest that men with fewer than two copies of the IGF-IR (AAG)₇ allele have lower plasma levels of IGF-I and a 1.25- to 2.18-fold increase in the incidence of prostate cancer. However, given the uncertain functional impact of these genotypes, the relatively modest alterations in risk seen, and the multiple hypotheses evaluated here, these associations need to be interpreted cautiously, pending the availability of additional data on the subject.