Racial Differences in Prediction of Time to Prostate Cancer Diagnosis in a Prospective Screening Cohort of High-Risk Men: Effect of TMPRSS2 Met160Val

Abstract

Introduction

The TMPRSS2-ERG gene fusion occurs in >50% of prostate tumors and has been associated with poor outcomes. The T-allele (Valine) of the Met160Val (rs12329760) in TMPRSS2 has been associated with this fusion. We evaluated this polymorphism with respect to self-identified race or ethnicity (SIRE), time to prostate cancer (PCA) diagnosis, and screening parameters in the Prostate Cancer Risk Assessment Program, a prospective screening program for high-risk men.

Patients and Methods

631 men ages 35-69 years were studied. “High-risk” was defined as ≥ one first degree or two second degree relatives with PCA, any African American (AA) man regardless of familial PCA, and men with BRCA1/2 mutations. Men with elevated PSA or other indications for PCA underwent biopsy. Men were followed from time of study entry to PCA diagnosis. Cox models were used to evaluate time to PCA diagnosis by genotype.

Results

Genotype distribution differed significantly by SIRE (CT/TT vs. CC, p<0.0001). Among 183 Caucasian men with at least one follow-up visit, PCA was more than doubled in men carrying CT/TT vs CC genotypes (HR= 2.55, 95% CI=1.14-5.70) after controlling for age and PSA. No association was seen among AA men by TMPRSS2 genotype.

Conclusions

The T-allele of the Met160Val variant in TMPRSS2, which has been associated with the TMPRSS2-ERG fusion, may be informative of time to PCA diagnosis for a subset of high-risk Caucasian men who are undergoing regular PCA screening. This variant along with other genetic markers warrant further study for personalizing PCA screening.

Introduction

Prostate cancer (PCA) is the second leading cause of cancer-related deaths in US men [1]. Nevertheless, the benefit of screening for PCA remains controversial for the general population [2, 3]. Men with a family history (FH) of PCA and African American (AA) men are at 2-7 fold increased risk for PCA diagnosis [1, 4]. Unfortunately, further risk-stratification measures for high-risk men are not well-characterized, and the optimal surveillance protocol for them remains unknown. Ultimately, the majority of high-risk men still may undergo unnecessary and potentially morbid evaluations to identify PCA. Even in high-risk men, PCA detection rates have been reported to be in the range of 10-17% [5, 6], suggesting that only a subset of high-risk men are diagnosed with PCA when current screening strategies are used. Genetic markers that may more precisely identify which high-risk men develop early onset PCA or aggressive PCA are needed in order to better match intensive PCA screening to those high-risk marker-positive men and similarly avoid overly aggressive approaches in other high-risk men who are marker negative. Thus, PCA early detection may become personalized for men having genetic evidence for high risk.

Recently, fusion between the TMPRSS2 and ERG (TMPRSS2-ERG) genes has been reported to occur in over 50% of prostate tumors[7]. ERG (ETS-related factor gene) is a proto-oncogene overexpressed in prostate tumors [8]. TMPRSS2 (transmembrane serine protease 2) is mostly localized in the prostate gland and is under the regulation of androgens [9]. The TMPRSS2-ERG fusion has been characterized in multiple studies and has been found to be associated with poorly differentiated prostate tumors and with worse outcomes [10-12]. After examining several polymorphisms in TMPRSS2 and ERG, Fitzgerald et al reported that the T-allele of the germline single nucleotide polymorphism (SNP) Met160Val in TMPRSS2 was associated with multiple copies of the TMPRSS2-ERG fusion in prostate tumors [13]. This polymorphism leads to a Valine-to-Methionine substitution at codon 160 (Met160Val), and genetic markers with biologic relevance to PCA may be informative in risk-stratifying high-risk men and personalizing PCA early detection protocols.

The purpose of this analysis is to evaluate whether TMPRSS2 Met160Val predicts PCA in prospectively screened high-risk men enrolled in the Prostate Cancer Risk Assessment Program (PRAP) at Fox Chase Cancer Center (FCCC) (Figure 1). Factors analyzed here were baseline distribution of the TMPRSS2 Met160Val genotype by self-identified race or ethnicity (SIRE), correlation to Gleason scores for men diagnosed with PCA, and time to PCA diagnosis by genotype.

Figure 1.

Rationale

Patients and Methods

Patients

PRAP was established in 1996 with a goal of providing screening for the early detection of PCA for high-risk men. The details of the PRAP study have been described previously, as well as the cancer detection rate in this cohort [6, 14]. Eligibility criteria for PRAP include men without a previous or current diagnosis of PCA who are ages 35-69 years and (1) have a family history (FH) positive for PCA defined as having at least one first degree relative with PCA, or two second degree relatives with PCA on the same side of the family, or (2) are AA, regardless of a FH of PCA, or (3) have deleterious mutations in BRCA1 or BRCA2 (less than 1% of the PRAP cohort). Mean follow-up in the PRAP cohort as a whole is 44.7 months (range 0.3-141.6 months).

Criteria for prostate biopsy have also been described previously [6]. Briefly, until November 2005, criteria for recommending a prostate biopsy included (1) PSA > 4.0 ng/mL, (2) PSA 2.0-4.0 ng/mL with a percent free PSA less than 27%, (3) any abnormality on digital rectal examination (DRE) of the prostate, or (4) PSA velocity of 0.75 ng/mL/year. Based on emerging evidence in the literature of high PCA detection rates with these criteria, the biopsy criteria were modified to detect earlier changes in screening parameters [5]. After November 2005, the criteria for recommending a prostate biopsy included (1) PSA > 2.0 ng/mL, (2) PSA 1.5-2.0 ng/mL with a percent free PSA of ≤ 25%, (3) any abnormality on (DRE), or (4) PSA velocity of 0.75 ng/mL/year. All biopsies were transrectal ultrasound-guided five-region biopsies with additional biopsies obtained at physician discretion. Men with initial negative biopsies or those who decline a biopsy are recommended to return in 3-6 months for follow-up screening assessments and potential additional biopsies. Informed consent was obtained from all participants upon enrollment. The PRAP study is approved by the Institutional Review Board at FCCC.

Genotyping

The TMPRSS2 polymorphism Met160Val was genotyped from germline DNA using a fluorogenic 5’ nuclease allelic discrimination assay (TaqMan® SNP Genotyping Assay C 25622353_20, Applied Biosystems). Reactions were prepared using TaqMan Universal PCR Mastermix, No AmpErase UNG (Applied Biosystems) according to manufacturer’s instructions. Thermal cycling and analysis were performed using an ABI7900 Sequence Detection System (Applied Biosystems). Control DNA samples with known genotypes were included in each run. In addition, a no template control was included to assess DNA contamination. SNP assignment was achieved automatically with the SDS software (Applied Biosystems) using a proprietary algorithm. In addition, genotypes were confirmed on a random selection of 2% of the samples with a 100% concordance.

Statistical methods

We compared genotype distribution by self-identified race or ethnicity (SIRE) using a Chi-square test. Men with at least one follow-up visit were included in analyses examining genotype with time to PCA diagnosis. This included 79.2% of the Caucasian men and 57.3% of the AA men with follow-up or were eligible for having a follow-up visit (i.e. were enrolled over a year prior to this analysis). The Kaplan-Meier product-limit method was used to estimate the survival functions for freedom from PCA by genotype. Differences in the curves were assessed using the log rank test. Results were considered statistically significant with a p<0.05. Cox proportional hazards analyses were used to estimate the hazard ratio comparing CT/TT genotypes to CC based on previous reports of the association of the T-allele to TMPRSS2-ERG fusion (13), adjusting for age and PSA at time of study enrollment for each SIRE group. All analyses were conducted using SAS statistical software, and Kaplan Meier plots were generated using R, version 2.5.1.

Results

Out of 657 PRAP participants, 631 were analyzed in this study. Twenty-six men were excluded for the following reasons: undetermined genotype results (n=15), refused blood draw (n=1), “other” SIRE groups (n=5), and missing baseline PSA (n=5). Table 1 shows the baseline demographics of these 631 men by SIRE. AA men and Caucasian men were similar with respect to age at entry into PRAP, baseline PSA, baseline percent free PSA, DRE results, and prior biopsy history. There was no significant difference in biopsy rates between Caucasian and AA participants (27% vs 21% respectively, p=0.12) or in the mean number of biopsies performed as a result of the PRAP protocol (Table 1).

Table 1.

Demographics and Prostate Cancer Characteristics of 631 PRAP Participants by Self-reported Race

	African American (n=400)			Caucasian (n=231)
	N	Mean	Range	N	Mean	Range
Age at entry (years)	400	49.7	34-69	231	49.8	35-69
Duration of follow-up (months)	229	41.8	0.3-129.8	183	51.1	0.6-137.5
PSA at baseline (ng/mL)	400	1.6	0.1-27.2	231	1.7	0.1-22.5
Percent Free PSA at baseline+	81	16.8	3.5-39.4	53	16.9	4.6-40
DRE at baseline
Normal/BPH	379 (95.7 %)			216 (96.0%)
Abnormal	17 (4.3 %)			9 (4.0 %)
Biopsy history (reported at baseline)
No Prior Biopsy/Unknown	293 (92.4%)			199 (93.0%)
Had Prior Negative Biopsy	24 (7.6 %)			15 (7.0)
Number of Biopsy Sessions while in PRAP	85	1.5	1.0-6.0	62	1.6	1.0-7.0
PCA diagnosis *	39 (9.8%)			28 (12.1%)
Last PSA prior to PCA diagnosis (ng/mL)	39	4.6	0.9-31.6	28	4.6	1.1-22.5
Gleason Score	39	6.23	5-8	28	6.18	5-7

⁺Note: Percent free PSA is only performed for men with a PSA 2.0-4.0 ng/mL by the previous criteria or a PSA 1.5-2.0ng/mL by the current criteria in PRAP. Therefore, not all men have a percent free PSA performed at baseline.

^*Percent of the race group;

Approximately 12.1% of the Caucasian men and 9.8% of the AA men were diagnosed with PCA as a result of the PRAP screening protocol. The mean PSA prior to diagnosis was identical at 4.6 ng/mL in both SIRE groups. There was no significant difference in median follow-up between Caucasian and AA men diagnosed (15.1 vs. 8.0 months respectively, p=0.322). Among those PRAP participants not yet diagnosed with PCA, Caucasian men had longer median follow-up than AA men (49.2 vs. 36.9 months respectively, p=0.015).

Time to PCA diagnosis was evaluated in 183 of 231 (79.2%) Caucasian men and 229 of 400 (57.3%) AA men in PRAP with at least one follow-up visit. Among the Caucasian men, no difference in follow-up was found by age at entry, PSA at baseline, marital status, level of education, or employment status. Among AA men, older age at enrollment (p<0.0001), higher baseline PSA (p=0.003), and full-time employment/retired (p=0.0059) were associated with a greater rate of follow-up (data not shown).

Genotype frequencies at Met160Val differed significantly by SIRE (p<0.0001). Genotypes were in Hardy-Weinberg equilibrium within SIRE groups. Among the 231 Caucasian men at baseline, 67% had the CC genotype and 33% had the CT/TT genotypes. Among the 400 AA men at baseline, 49% had the CC genotype and 51% had the CT/TT genotypes.

Figures 2 and 3 show the Kaplan-Meier plots for time to diagnosis in these men by SIRE according to genotype. No significant difference was observed in time to iagnosis in AA men. However among Caucasian men with a FH of PCA (defined as having at least one first degree relative with PCA or two second degree relatives with PCA on the same side of the family), there was a statistically significant difference in time to PCA diagnosis with earlier time to diagnosis in men with the CT/TT genotypes compared to the CC genotype (p=0.006). The hazard ratio for PCA for Caucasian men with CT/TT genotypes vs. the CC genotype was significant (HR=2.85, 95% CI=1.35-6.46, p=0.007) adjusting for age at enrollment only. After further controlling for first PSA at enrollment, the association remained significant (HR = 2.55, 95% CI=1.14-5.70, p=0.02). Gleason score was not significantly associated with Met160Val genotype (p=0.35 for Caucasian men and p=0.44 for AA men), although the sample size was small. Interestingly, of the Caucasian men diagnosed with PCA with the CT/TT genotypes, 29% had a Gleason score of 7 compared to 9% of the men with the CC genotype although this was not significant (p=0.35) (Table 2).

Figure 2.

Time to PCA Diagnosis in 183 Caucasian Men with a FH Of PCA in PRAP by TMPRSS2 Met160Val Genotype

Figure 3.

Time to PCA Diagnosis in 229 African American Men in PRAP by TMPRSS2 Met160Val Genotype

Table 2.

Gleason scores among PRAP Participants diagnosed with PCA by TMPRSS2 Met160Val genotype

Gleason score	Caucasian PCA (n=28)		African American PCA (n=39)
	CC (n=11)	CT/TT (n=17)	CC (n=17)	CT/TT (n=22)
5	0	1	1	2
6	10	11	10	16
7	1 (9%)	5 (29%)	4	4
8	-	-	2	0

To assess whether results for time to PCA diagnosis may have been confounded by preexisting, undiagnosed PCA at presentation, we repeated the analysis for time to PCA diagnosis after excluding men who were diagnosed with PCA within 9 months of enrollment into PRAP. This analysis therefore excluded 12 out of 183 (6.6%) Caucasian men and 19 out of 229 (8.3%) AA men due to PCA diagnosis soon after enrollment into PRAP. Among the remaining 171 Caucasian men, a trend toward a difference in time to PCA diagnosis by TMPRSS2 Met160Val genotype in favor of CT/TT vs. CC was found after six years of follow-up (n=15 PCA cases, p=0.055). No differences were seen in time to PCA diagnosis by TMPRSS2 Met160Val genotype among the 210 AA men (n=19 PCA cases, p=0.81).

Discussion

Identifying which high-risk men will develop aggressive and early-onset PCA is key to personalized risk assessment. Stratifying high-risk men by genetic markers may help in tailoring recommendations for PCA early detection. More aggressive recommendations may be suitable in men who carry genetic markers found to be associated with early time to PCA diagnosis or high-grade disease while less intensive approaches would be appropriate for men not found to harbor risk-associated genetic variants. The TMPRSS2-ERG gene fusion is a potential target to characterize for risk assessment for PCA in men at high-risk for the disease due to the multiple associations with aggressive disease and poor outcomes [10-12].

This study characterized a particular genetic variant, Met160Val, in the TMPRSS2 gene which has been shown to be associated with multiple copies of the TMPRSS2-ERG fusion in prostate tumors [13]. The presence of multiple copies of this gene fusion has been associated with a significantly decreased PCA-specific survival [12]. This polymorphism leads to a Valine-to-Methionine substitution at codon 160, which has been reported to affect the catalytic function of TMPRSS2 [15]. Due to its association with multiple copies of the TMPRSS2-ERG fusion (which has poor prognostic characteristics), we decided to study the Met160Val polymorphism in high-risk men presenting for PCA screening for the early detection of PCA in PRAP to evaluate its role in risk assessment.

A unique aspect of the PRAP cohort is the over 60% AA representation, allowing for characterization of genetic variants by SIRE. Indeed we found a significant difference in the distribution of the Met160Val genotype by SIRE, with proportionately more Caucasian men carrying the CC genotype than AA men. Furthermore, due to the ongitudinal follow-up information in the PRAP cohort, we were able to evaluate time to PCA diagnosis by genotype status. While the genotype status at Met160Val was not informative of time to diagnosis in AA men, we did find that proportionately more Caucasian men with a FH of PCA carrying the CT/TT genotypes were diagnosed with PCA during follow-up compared to the CC genotype. Hazard ratios also correlated with an over 2.5-fold increased risk for PCA in Caucasian men carrying the CT/TT genotypes. Overall these findings support that the T-allele at Met160Val, which was reported to be associated with multiple copies of the TMPRSS2-ERG fusion [13], is informative regarding early PCA diagnosis in Caucasian men with familial PCA. Since PCA is a genetically complex disease, a single genetic marker likely has limited ability to provide comprehensive risk stratification. However clinical characterization of genetic markers, whether alone or in combination, will hopefully lead to a greater understanding of how to incorporate genetics into PCA risk assessment for high-risk men. One possibility of how genetic markers may impact PCA risk assessment is that Caucasian men with familial PCA who present for PCA screening and who carry markers associated with early time to PCA diagnosis (such as the T-allele at Met160Val) could be recommended for more intensive screening measures. These types of tailored screening approaches based on genetic marker status warrant further study.

Lubieniecka et al. reported on the risk for PCA in a population-based case-control study from Seattle of the Met160Val marker in TMPRSS2 previously [16]. This study found that men with the GG genotype and a first-degree relative with PCA had a higher risk for PCA. However, the interaction between FH of PCA and genotype status of Met160Val was not significant. Our study did find that more Caucasian men (all of whom are required to have a FH of PCA for entry into PRAP) carried the CC (or GG) genotype compared to the CT/TT genotypes (67% vs. 33%, respectively). However, our study found that the opposite genotypes were associated with shorter time to PCA diagnosis and risk for PCA. A possible explanation for these apparently disparate results is that while more Caucasian men with familial PCA may carry the CC genotype, the T-allele may be more clinically informative in the risk assessment setting. These findings deserve further study in larger patient populations.

There are some limitations to our study. Our goal was not to determine the mechanism of how the TMPRSS2 Met160Val polymorphism is related to TMPRSS2-ERG fusion or how the gene fusion is associated with PCA. While this is a worthwhile endeavor, the main limitation is prostate tumor tissue availability. Unlike other cancers, prostate tissue availability for research is limited as many patients opt for radiation therapy after a PCA diagnosis is made on core biopsies. In addition, PCA tends to be multifocal and individual lesions can be heterogenous in genetic and pathologic features [17]. There is precedent, however, for studying the clinical value of genetic markers prior to a full understanding of the mechanism of how such markers may be leading to PCA susceptibility. For example, genetic markers identified from genomewide association studies frequently are in genomic regions which are gene-poor and in which further understanding of mechanism is needed [18-20]. However, the clinical utility of many such genetic markers is being explored prior to a full understanding of the mechanism of susceptibility for PCA in order to make progress in personalizing PCA risk assessment.

Another limitation of this study is that the overall follow-up rate in PRAP is approximately 60%, which can hinder the interpretation of PCA development over time. This is a known challenge in screening studies and warrants further efforts and resources to improve adherence to screening protocols. However, our follow-up rate of 60% is similar to the adherence rates in other high-risk screening studies (5). Caucasian participants did have higher follow-up rates than AA men, which may account for some of the differences seen by SIRE. Nevertheless, the differences seen in the association of TMPRSS2 Met160Val genotype by SIRE was unlikely related to sample size differences. The sample size of Caucasian PRAP participants (where an association was found between TMPRSS2 genotype and time to PCA diagnosis) was substantially smaller than sample size of the AA participants where no association to time to PCA diagnosis was found. Another limitation is that not all PRAP participants undergo prostate biopsy, only if they meet biopsy criteria per PRAP guidelines. Thus some high-risk men may have undiagnosed PCA which can also hinder the interpretation of time to PCA diagnosis and hazard ratio estimates for PCA risk.

Finally, we only studied one genetic variant in the TMPRSS2 gene in this study. There may be other important genetic variants in TMPRSS2 or ERG to study that may be informative in high-risk men for PCA risk assessment. In addition, TMPRSS2 also has been found in fusion with ETV genes, in which polymorphisms should also be studied.

Conclusions

The T-allele of the Met160Val polymorphism in TMPRSS2 appears to be informative of time to PCA diagnosis in Caucasian men with a FH of PCA who are undergoing screening for the disease. This variant, along with other variants in the TMPRSS2 gene and other genetic and non-genetic prognostic variables, deserve further study for their role in clinical risk assessment for PCA.