Vitamin D Receptor Start Codon Polymorphism (FokI) and Prostate Cancer Progression¹

Prostate cancer is the most commonly diagnosed noncutaneous cancer in men in the United States. Incidence rates of prostate cancer increased markedly from the late 1980s to 1992, due largely to the increased testing of serum PSA³ for cancer screening, followed by a decrease to a level still higher than that of the pre-PSA era (1). In contrast, mortality rates of prostate cancer have shown less remarkable time trends (1, 2). The majority of prostate tumors never become metastatic or life-threatening. Thus, the identification of biomarkers that predict progression to advanced disease carries major public health significance (2, 3, 4).

Vitamin D maintains calcium homeostasis and regulates growth and differentiation of many types of cells (5). The active form of vitamin D, 1,25(OH)₂D₃³, exerts its activity through the intracellular VDR³ (6). The VDR, a member of the steroid/thyroid hormone nuclear receptor family, is expressed in both normal and malignant prostate cells (7, 8). 1,25(OH)₂D₃ binds to the VDR, which forms a heterodimer with the retinoid X receptor. The heterodimer in turn binds to vitamin D response elements in the promoter regions of target genes and regulates transcription of those genes (5).

Several polymorphisms have been identified in the VDR gene, and their functional significance and potential effects on disease susceptibility have been investigated (9). One of the known DNA sequence variants is a thymine/cytosine (T/C) polymorphism in the first of two potential start (ATG) codons separated by 3 codons. This polymorphism results in two alleles that can be distinguished by a RFLP³ using the endonuclease FokI (10). The F allele lacks the first ATG, and translation starts at the second ATG, with the resulting VDR being shorter by three amino acids than that of the f allele, which starts at the first ATG. Findings regarding functional significance of the FokI polymorphism have been inconsistent (11, 12, 13). However, most data indicate that the F allele is more effective than the f allele in transactivation of the 1,25(OH)₂D₃ signal (12, 13). In recent studies, FokI genotypes have been associated with risk of malignant melanoma (14) and colon adenomas (15). To our knowledge, the association between FokI genotypes and prostate cancer risk has thus far only been investigated in men in the United Kingdom (16) and in China (17). Other VDR polymorphisms located at the 3′ end of the gene have been associated with risk of prostate cancer in some studies (17, 18, 19, 20, 21) but not in others (22, 23, 24, 25, 26).

Because of the important role of vitamin D and its receptor in prostate cell growth and differentiation (27, 28, 29, 30), we hypothesized that the FokI polymorphism in the VDR gene might be associated with progression of prostate cancer. The goal of this study was to examine the association of the FokI polymorphism with histopathological features and prognosis of prostate cancer among men who had undergone radical prostatectomy. Because the percentage of Gleason grade 4/5 has been found to be a strong predictor of progression of prostate cancer after radical prostatectomy (31, 32), we focused on the relationship of the VDR FokI genotype and this variable.

A total of 227 subjects who had undergone radical prostatectomy at Stanford Medical Center between 1984 and 1992 were selected for the study (31). These subjects were those who were found to have at least 5% of Gleason grade 4/5 in their cancers. Thirty-six of the selected subjects were excluded because: (a) DNA from archived tissue failed to yield detectable PCR products in the genotyping assay for the VDR polymorphism (35 subjects); or (b) cancer was in the transition zone rather than in the peripheral zone (1 subject). These exclusions resulted in a sample size of 191 subjects. None of the subjects had received hormonal therapy before surgery. Although race was not recorded for individual patients, the source population historically had been predominantly (>95%) Caucasian.

Surgically removed prostates were routinely subjected to a comprehensive histopathological review by a single pathologist according to the method described previously (33). When two or more distinct foci existed in a single prostate, the volume of the largest (index) cancer was used in statistical analysis. The Stanford modified Gleason scale was used to estimate the proportion of each cancer that was poorly differentiated in each prostatectomy specimen (34). From the Gleason scale of 5 grades (35), grades 4 and 5 (poorly differentiated) constitute a variety of histological patterns having the common feature that they do not form intact glands. For each patient, an estimate was made of the percentage of the total cancer represented by Gleason grade 4 or 5 (percentage of Gleason grade 4/5). This measurement of tumor grade was shown to be the strongest predictor of PSA failure in a study of 379 men who had undergone surgery at the Stanford Medical Center between 1983 and 1992 (31). Information on the subject’s age at surgery and serum PSA level before surgery was abstracted from an existing database. The study protocol was approved by the Panel on Medical Human Subjects at Stanford University.

DNA was extracted from formalin-fixed, paraffin-embedded prostate tissue as described previously (36). Because we were interested in DNA sequence variation of constitutional (i.e., germ-line) DNA, rather than somatic changes in malignant cells, we selected tissue blocks that contained no tumor according to histological review. The primers and PCR³ conditions for amplifying exon 2 (containing the FokI site) of the VDR gene were described previously (37). Five μl of DNA were used for each PCR reaction. The resulting 260-bp PCR products were digested with FokI endonuclease (New England Biolabs, Beverly, MA) at 37°C for 3 h and then electrophoresed through a 3% agarose gel containing Tris-acetate-EDTA buffer and ethidium bromide. PCR products with an undigested large band were scored as FF homozygotes, those with a smaller digested band were scored as ff homozygotes, and those with a large and small band were scored as Ff heterozygotes.

PSA failure (or biochemical failure) was used as the outcome event for longitudinal analysis. Typically, follow-up PSA test results were available approximately every 6 months, although the length of follow-up and intervals between PSA tests varied among subjects. PSA failure was defined as two consecutive PSA values above a cutoff point of 0.07 ng/ml for PSA measured by the sensitive Tosoh method and 0.2 ng/ml for measurements by less sensitive methods. When a patient had experienced PSA failure, time to failure was calculated as the number of months between the date of surgery and the date of the first of the two consecutive PSA values that exceeded the cutoff point.

Twenty-nine subjects whose serum PSA levels never dropped to zero (or undetectable level) after surgery were classified as experiencing PSA failure at time 0 (i.e., at the time of surgery). Three subjects had a serum PSA level above the cutoff point only in the last follow-up PSA test available, and thus we could not confirm their failure status definitively in the absence of a subsequent PSA test also exceeding the cutoff point. We performed longitudinal analysis for PSA failure in two ways, classifying these three subjects with indeterminate failure status as (a) experiencing PSA failure or (b) not experiencing PSA failure.

All statistical analyses were performed using the SAS statistical package, release 8.1 (SAS Institute, Inc., Cary, NC). Means of age, percentage of Gleason grade 4/5, and cancer volume were compared among the three VDR FokI genotypes (FF, Ff, and ff) by ANOVA, using the GLM procedure (38). For a comparison between the two groups (e.g., FF + Ff versus ff), the t test was used (39).

The relationship of the VDR FokI genotype to the risk of PSA failure was examined by Cox proportional hazards analysis, using the PHREG procedure with the “exact” option for handling ties (40). The Cox model included the VDR FokI genotype (FF, Ff, or ff), age at surgery (<60, 60–64, 65–69, and ≥70 years), percentage of Gleason grade 4/5, cancer volume, and preoperative PSA level (quartiles based on the distribution in the study sample for the last three variables).

Table 1 summarizes the characteristics of 191 study subjects, their tumors, and their VDR FokI genotype. In this population with prostate cancer, the frequencies of the F allele and the f allele were 60% and 40%, respectively. The mean percentage of Gleason grade 4/5 cancer in subjects with the ff genotype was 30.3%, which was lower than that in subjects with the FF and Ff genotypes (42.8% and 43.8%, respectively; P = 0.015 by t test for ff versus FF + Ff). No statistically significant differences among the VDR FokI genotypes were observed for age at surgery, index cancer volume, or preoperative serum PSA level.

Among the 191 subjects, the duration of follow-up ranged from 0 to 175.1 months (median, 54.7 months). Serum PSA levels never dropped to zero or to an undetectable level after surgery in 29 subjects, who were considered to experience PSA failure at time 0 (i.e., at the time of surgery). For three subjects, PSA failure status was not definitive because their last available PSA values were above the cutoff for the failure definition without subsequent elevated PSA levels to confirm the failure status. When these three subjects were regarded as experiencing failure, the total number of PSA failures was 102; otherwise there were 99 PSA failures. For the 102 PSA failures, time to failure ranged from 0 to 149.2 months (median, 16.6 months).

Table 2 shows the HR³ estimates from the Cox proportional hazards analysis for the association between the FokI genotype and risk of PSA failure. One analysis was performed with adjustment for age at surgery alone. In another analysis, HR estimates were further adjusted for percentage of Gleason grade 4/5, cancer volume, and preoperative serum PSA level. Risk of PSA failure in subjects with the Ff genotype showed little difference from that in subjects with the FF genotype, regardless of whether the HR estimate was adjusted for age alone or for age and additional covariates. In contrast, the risk of PSA failure for subjects with the ff genotype was lower than that for men with the FF genotype, although the association was not statistically significant when the HR estimate was adjusted for age alone (HR = 0.76; 95% CI,³ 0.44–1.32). This decrease in PSA failure risk for the ff genotype disappeared when the HR was further adjusted for Gleason grade, cancer volume, and preoperative serum PSA level. HR estimates were essentially the same as those shown in Table 2 when the three subjects with indeterminate failure status were reclassified as not experiencing PSA failure (data not shown).

It is plausible to search for potential prognostic markers of prostate cancer among the genes and gene products related to the pathophysiology of prostate cells. The active form of vitamin D, 1,25(OH)₂D₃, has growth-suppressive and differentiating effects on prostate cells (27, 28, 29, 30). These effects are mediated by the nuclear receptor, VDR, in cancer cell lines as well as in nonmalignant and malignant primary cultures of prostatic epithelial cells. However, the cellular response to vitamin D does not necessarily correlate with the receptor content, suggesting that additional factors are involved in vitamin D-mediated growth regulation (5, 41). Small changes in the VDR structure, including changes in protein composition caused by DNA sequence variants, might have subtle influences on the cellular response to 1,25(OH)₂D₃. Naturally occurring DNA polymorphisms in the VDR gene are of interest as candidate prognostic markers.

The goal of our study was to determine whether the VDR FokI genotype affected outcome after radical prostatectomy to cure prostate cancer. To carry out this study, we chose men who had undergone radical prostatectomy at the Stanford Medical Center between 1984 and 1992. These men have a relatively long follow-up (median, 54.7 months). We observed a lower percentage of Gleason grade 4/5 cancer in subjects with the ff genotype than in those with the FF or Ff genotype. Percentage of Gleason grade 4/5 cancer is the primary independent predictor of biochemical (PSA) failure in men undergoing radical prostatectomy. The cure rate for men with no Gleason grade 4/5 was 93% in the Stanford series reported in 1999 (31). The critical biological change that occurs when cancers become grade 4/5 is shown by the fact that every 10% increase in the percentage of grade 4/5 cancer in radical prostatectomy specimens correlates with an almost proportionate increase in biochemical failure (31). Therefore, an association of the ff genotype with a lower percentage of Gleason 4/5 cancer suggests that this VDR variant slows the progression of prostate cancer from the less aggressive and curable grade 3 to the more aggressive grade 4/5. To our knowledge, this is the first report of an association between the VDR FokI genotype and histopathological Gleason grade of prostate cancer. Luscombe et al. (42) reported that the VDR FokI genotype was not associated with histopathology of prostate cancers in men in the United Kingdom, but cancers were classified only as moderately or poorly differentiated.

Longitudinal analysis of our cases suggested that age-adjusted risk of PSA failure was lower in subjects with the ff genotype, although the association was not statistically significant. This decrease in risk disappeared when the risk estimate was further adjusted for Gleason grade, cancer volume, and preoperative serum PSA level, which was not surprising because the ff genotype was associated with lower Gleason grade. However, in the study in the United Kingdom, the VDR ff genotype was associated with increased risk of bone metastases (42). Several years are required from first evidence of biochemical PSA failure after radical prostatectomy to positive bone scans (43, 44), therefore it is not possible to directly compare our findings with those of the British study.

Results of in vitro studies have not conclusively demonstrated whether one or the other of the FokI alleles (F versus f) leads to more efficient responses of cells to vitamin D. Gross et al. (11) did not detect allelic differences in transactivation activity of the VDR alleles. Jurutka et al. (12) found that the F variant of VDR had more potent transactivation activity than the f variant in transient transfection assays in COS-7 cells. Recent studies demonstrated that the F allele, as well as the L allelic form of a singlet (A) repeat in the 3′-untranslated region, displayed higher transcriptional activity induced by 1,25(OH)₂D₃ in human fibroblast cell lines (13). However, only when genotypes at both sites were considered simultaneously did statistically significant differences emerge.

Several studies found that the ff genotype was associated with lower bone mineral density in postmenopausal women (10, 37). In contrast, the F allele was associated with an increased risk of developing rickets in Nigerian children (45). If indeed the F allele creates a better-functioning VDR variant, then this finding, like our finding of a protective effect of the f allele on prostate cancer progression, is counterintuitive. However, the FokI genotype has not been found, at least in females, to correlate with serum levels of 25-hydroxyvitamin D₃, 1,25(OH)₂D₃, parathyroid hormone, or osteocalcin (10, 37). Because these factors are all regulated by VDR, it might be expected that their levels would depend on the level of activity of the VDR genotype. Overall, it appears that additional studies are required to comprehensively determine whether VDRs translated from F versus f alleles have differential activity. It will be important to examine this in appropriate model systems because activity of VDR variants may be cell type specific.

Some limitations of the study design in the present investigation are worth noting. First, the statistical power of the study was limited for longitudinal analysis. Therefore, confirmation of the present findings in larger study samples is needed. Second, subjects in the present study were selected because they underwent radical prostatectomy. Caution is due in extrapolating the results of this study to prostate cancer patients who choose other forms of therapy or watchful waiting. However, our present finding shows that the ff genotype is associated with lower percentage of Gleason grade 4/5 than the FF and Ff genotypes, suggesting a protective effect of the ff genotype on prostate cancer progression. The mechanistic basis of this finding is under investigation.