Germline BRCA mutations denote a clinicopathologic subset of prostate cancer

INTRODUCTION

Genome-wide association studies have identified common susceptibility loci in more than a dozen independent chromosomal regions associated with an increased risk for prostate cancer,¹ but candidate genes at these loci have yet to be functionally implicated. While present in only a small proportion of prostate cancer cases overall, mutations of the BRCA2 and BRCA1 genes are commonly observed in families with multiple cases of breast cancer. Whether detected in affected families or in the general population, BRCA mutations cause an increased risk for breast and ovarian cancer in women, and breast and prostate cancer in men.^{2-5 6-7 8}

Histologic differentiation is an established surrogate for survival in prostate cancer, and Gleason scoring is the most widely used grading system.^9-10 Clinical management of this malignancy depends on assessment of clinical features, age of the patient, as well as histopathology of the tumor. There remains a critical need for prognostic markers to tailor therapeutic strategies, which may range from “watch and wait” to surgery or radiation therapy.¹¹ BRCA mutations may constitute one such prognostic marker, however, few studies have investigated the tumor characteristics of BRCA-associated prostate cancer. A more aggressive phenotype has been suggested by four analyses of BRCA-associated prostate cancer,^12-15 but the largest study found no difference in histopathology between carriers and non-carriers of BRCA mutations.¹⁶ Furthermore, loss of heterozygosity has been identified in prostate tumors of BRCA2 carriers, and has been associated with higher Gleason score¹⁷. Prostate cancer-specific mortality is a more robust endpoint and an association between BRCA2 mutation and earlier death due to prostate cancer has been reported for patients with advanced disease¹⁸. Here, we confirm a more aggressive histology associated with BRCA2-associated prostate cancer, and for the first time demonstrate the independent prognostic significance of this genetic marker in early stage prostate cancer, taking into account known prognostic factors for the disease, including prostate-specific antigen (PSA) and clinical features.

MATERIALS AND METHODS

DNA was extracted from blood samples of 894 prostate cancer cases treated at Memorial Sloan-Kettering Cancer Center between June 1988 and December 2007. DNA samples were also acquired from 454 non-cancer controls who provided informed consent as part of a prospective study of healthy people conducted from 2000-2002 in New York City.¹⁹ Both cases and controls identified themselves as being of Ashkenazi Jewish background. The inclusion of only Ashkenazi Jewish men enabled targeted genotyping for founder mutations, and all blood samples were genotyped for the 2 major Ashkenazi BRCA founder mutations BRCA1*185delAG and BRCA2*6174delT. The third founder mutation BRCA1*5382insC was not genotyped. The TaqMan™ (fluorogenic 5’ nuclease) assay was used for SNP genotyping. The primers and probes were obtained from Applied Biosystems (Foster City, CA, USA). PCR was conducted with both primers and probes added in ABI 9700 thermocycler, and the endpoint results were scored using the ABI 7900HT Sequence Detection System (ver 2.3). In each 384-well plate, one reference sample and one negative control were included for quality control. Cases’ medical records and pathology reports were collected as part of an institutional prostate cancer research database utilizing standardized questionnaires and chart abstraction forms, created and maintained by the prostate cancer disease management team at MSKCC. Analysis for this study was based on data on clinical stage, Gleason score (from needle biopsy), PSA levels, and age at diagnosis of the primary prostate tumor as well dates of biochemical recurrences, development of castrate metastasis, death due to prostate cancer, and overall survival. Biochemical recurrence refers to the detection of rising PSA after local therapy, and castrate metastasis refers to time to progression of disease following initiation of anti-androgen therapy. Biochemical recurrence was defined as PSA ≥ 0.2 ng/ml post radical prostatectomy, and a value of ‘nadir+2’ post other therapy.^20-21 Castrate metastasis was defined as progression of disease despite castrate levels of testosterone. Data were collected on 894 genotyped cases but only patients with localized disease at presentation were included in the analysis of clinical outcome; 62 cases presented with metastatic prostate cancer and were excluded for the survival analysis. In accordance with an Institutional Research Board approved protocol, patient identifiers were removed at time of genetic analysis. Therefore, 832 prostate cancer cases and 454 controls were included in subsequent analyses.

Gleason scores of the prostate tumors were categorized into <7, ≥7 subgroups. This cutoff was chosen based on clinical experience and prior literature suggesting that clinical outcome for Gleason score 7 prostate cancer is more similar to that for Gleason score 8-10 than for Gleason score < 7 disease²². Among cases, Kruskal-Wallis tests were used to compare distribution of age at diagnosis and Gleason score by carrier status. Logistic regression models, adjusted for age and clinical stage, were used to calculate odds ratios (OR) and 95% confidence intervals (CI) for the association between carrier status and prostate cancer risk, and between carrier status and Gleason score (case only model). Unconditional logistic regression was used as controls and cases were not matched. Time at risk was calculated from the date of diagnosis of prostate cancer to the date of death or date of last contact. As a secondary analysis, time at risk intervals were calculated from date of blood draw for genetic testing (left truncation). The Kaplan-Meier method stratified by genotype was used to generate survival curves.²³ Hazard ratios (HR) and 95% CI for mortality, biochemical recurrence and castrate metastases associated with mutation status were estimated using Cox proportional hazard models,²⁴ adjusting for age, clinical stage, Gleason score, PSA levels of primary tumor, and treatment type. We checked for violations of the proportional hazards assumption for all variables by visual inspection of the log (-log) plots and analytically using Schoenfeld residuals.²⁵ All statistical analyses were conducted using Stata 10.0 (StataCorp LP, College Station, TX).

RESULTS

The median age for controls was 42 years old (interquartile range: 42 to 57), while the median age for cases was 68 years old (interquartile range: 62 to 73). Prostate cancer cases were more likely to carry a BRCA2 mutation than controls (2.4% vs. 0.7%, p=0.002). Men who carried a BRCA2 mutation had a 3.18 times higher risk of prostate cancer than non-carriers (OR=3.18, 95% CI 1.52-6.66; Table 1). The frequency of BRCA1 mutation was similar for cases and controls (p=0.34); 3.1% of prostate cancer cases in this series carried either a BRCA1 or BRCA2 mutation.

Among the total of 832 cases eligible for analysis of clinical outcome with a median follow-up in excess of 8 years, the age of diagnosis was slightly greater (median age of 68.2 years) for 806 patients with no BRCA mutations compared to 20 cases with BRCA2 mutations (median age of 62.0 years), although this difference was not statistically significant (p=0.1). After adjusting for Gleason score and stage, there was no difference in age at presentation between BRCA2 mutation carriers and non-carriers (HR = 1.23, 95% CI 0.76-2.00, P = 0.41), nor between BRCA1 mutation carriers and non-carriers (HR = 0.67, 95% CI: 0.28-1.61, P = 0.37). Mutation carriers’ dates of diagnoses were spread across the duration of the study and carriers do not represent a more contemporary cohort that could have received higher Gleason scoring. Poorly differentiated prostate cancers (Gleason score 7-10) (Figure 1) were found in 57% of non-carrier cases and in 85% of BRCA2 mutation-associated cases (P = 0.0002), but there were no differences between BRCA1 mutation carriers and non-carriers (P = 0.71). Adjusting for age and stage, BRCA2 mutation carriers were more than twice as likely to have a Gleason score of 7-10 (hazard ratio 2.63, 95% CI 1.23 – 5.6, P = 0.01) than non-carriers, but the distribution of Gleason scores did not differ for BRCA1 mutation carriers and non-carriers (HR= 0.53, 95% CI 0.09 – 3.29, P = 0.50) (Table 2). Younger patients were more likely to have a Gleason score of 7-10, compared to lower scores (HR= 0.76; 95% CI 0.65 – 0.89; P = 0.001). Distribution by type of treatment received is shown in Table 2; half of BRCA2 mutation carriers were treated by radical prostatectomy, compared to 30% of BRCA wild-type cases.

After 7,254 person-years of follow-up for the entire cohort, the following outcomes occurred: 369 biochemical recurrences, 158 castrate metastases, and 184 deaths of which 98 were due specifically to prostate cancer (Table 3, Figure 2). Compared to non-carriers, prostate cancer patients, who carried mutations in BRCA1/2, were at higher risk of recurrence and death (Figure 2). Results of a Cox proportional hazards model (Table 3) analysis taking into account age, clinical stage, Gleason score, PSA and treatment showed that both BRCA1 and BRCA2 mutation carriers had an increased risk of developing biochemical recurrence (HR = 4.32; 95% CI 1.31-13.62, p-value=0.016 and HR=2.41; 95% CI 1.23-4.75, p-value=0.011, respectively). BRCA2 mutation carriers had a greater risk of castrate metastases (HR=3.01; 95% CI 1.26- 7.14, p-value=0.013), compared with non-carriers. This translated into a greater risk of death due to prostate cancer (HR = 5.16; 95% CI 1.09- 24.53, p value = 0.039 and HR=5.48; 95% CI 2.03 - 14.79, p value =0.001) for BRCA1 and BRCA2 carriers, respectively.

The median time from diagnosis to blood collection for genotyping in this cohort was 4.6 years. When we left truncated time at risk, HR estimates were in a similar direction and magnitude, the hazard ratio for risk of death due to prostate cancer in BRCA2 mutation carriers was 11.21 with 95% CI 3.6-34.53 (other data not shown).

A separate analysis was made comparing time from biochemical recurrence to castrate metastasis. There were 369 men with biochemical recurrence and 158 of those had castrate metastasis. There was a trend towards a higher frequency of biochemical recurrence progressing to castrate metastasis in BRCA2 carriers than non-carriers (HR [95% CI] = 2.01 [0.85-4.79]). There was no difference in the frequency of castrate metastasis after biochemical recurrence in BRCA1 carriers compared to controls (HR [95% CI] = 1.08 [0.15-7.85]).

DISCUSSION

Although early studies of smaller subsets of patients were inconclusive,^{7, 26-29} recent studies have associated BRCA1 and BRCA2 mutations with a hereditary predisposition to prostate cancer, noting up to a 33% cumulative risk by age 80.^{3-4, 16} It has previously been suggested that BRCA mutation confers a greater risk of early onset prostate cancer (<65 years),³⁰ although Giusti et al reported no difference in mean age of presentation between men with and without BRCA mutations;¹⁶ such mutations were found in only 0.78% of 290 men diagnosed with prostate cancer at <55 years of age.³¹ Our study, one of the largest to date, again demonstrates no significant difference in age of presentation between BRCA mutation carriers and non-carriers,

The proportion of high grade prostate cancer in our population was significantly higher in the subset with inherited BRCA2 mutations compared to other subsets analyzed. The findings were striking with 17/20 (85%) of BRCA2 mutation carriers demonstrating Gleason 7 or higher disease, representing a group with an aggressive phenotype and confirming this recently reported association.¹⁵ However, the proportion of cases with high grade prostate cancer was higher than expected for both cases and controls. While the Prostate Cancer Prevention Trial included only patients with a PSA ≤ 4ng/mL, only 14.9 % of cases had Gleason 7 or higher disease,³² compared to 53% in controls in a large study in Iceland,¹² and 57% of our population. It is probable that differences in these populations could be due to differences in PSA or stage. Cases referred for care at a specialty center, for example, may be weighted toward more aggressive disease. The 57% of cases in this series with Gleason scores ≥ 7, is comparable to the annual proportion (in 2007) of 59% of all cases biopsied at our institution, and the 50% of cases biopsied outside the institution whose pathology showed Gleason scores ≥ 7. It is also possible that the higher baseline Gleason scores in some series are population (e.g. Ashkenazi, Icelandic) specific. Racial variation in both Gleason score at presentation and clinical outcome has been reported. ^33-34 However, in two prospective studies of over 800,000 men, prostate cancer death rates were significantly lower among Jewish men than non-Jewish men,³⁵ suggesting that the clinicopathologic associations observed here may result from genetic factors in a subset of cases.

When results were stratified by gene, BRCA2 mutations were associated with more poorly differentiated disease, but BRCA1 mutations were not. We found no differences in Gleason scores between BRCA1 mutation carriers and non-carriers, although we genotyped for only one of two BRCA1 founder mutations, and may have been underpowered to observe such differences. The “threshold” Gleason score of 7 or greater was chosen based a priori on previous literature and on clinical experience. Re-analysis of the data using an alternative definition of Gleason 8 or higher revealed essentially the same correlation, with a HR greater than 3 (95% CI 1.38-7.53) for BRCA2 mutation carriers. The association between BRCA mutation and aggressive histology has now been repeatedly reported but remains unexplained. Recent data suggest that double strand DNA breaks, acting through the androgen receptor, produce the TMPRSS2-ERG gene fusion that is found in 50% of prostate cancer^36-37. The TMPRSS2-ERG gene fusion has been associated with aggressive disease, and it is possible that, in the absence of BRCA, accumulating double stranded breaks facilitate this translocation.

BRCA mutations were not only associated with a more aggressive histology but also independently predicted increased risk of disease progression through the prostate cancer clinical states model. ³⁸ A Cox proportional hazards model incorporating the clinical predictors included in commonly used prostate cancer nomograms (Gleason score, PSA at diagnosis, stage)³⁹ and treatment type, demonstrated a significance of BRCA mutation status that was independent of these known prognostic factors; there was an increased risk of biochemical recurrence and the development of castrate metastases for BRCA carriers compared with non-carriers. As expected, the more advanced disease in BRCA mutation carriers was associated with worse overall and disease-specific survival. In a previous report suggesting an inferior survival for BRCA-associated prostate cancer,¹² 16/29 (55%) cases with the Icelandic founder mutation (BRCA2 999del5) had metastatic prostate cancer at presentation, whereas all men in our study had localized disease. In the prior study, the 13 cases with localized disease had a greater risk of dying from prostate cancer compared with non-carriers, a finding that is supported by our larger study. A recent consortium of prostate cancer patients in BRCA1- and BRCA2- linked families noted an inferior survival of BRCA2 compared to BRCA1-associated prostate cancer cases, however direct genotyping of all cases was not performed, there was no available BRCA wild-type comparison group, and no information available on pathology or clinical features of the disease.¹⁴ In contrast to that series, we observed an association with aggressive disease in both BRCA1 and BRCA2 mutation carriers.¹⁴ Although the size of our BRCA1 cohort was small, carriers had more aggressive disease with earlier biochemical recurrence and more prostate-specific deaths compared with non-carriers despite similar Gleason scores, suggesting that the effect of BRCA1 mutation on clinical outcome may be independent of aggressive histology.

The current series of BRCA-genotyped prostate cancer cases is among the largest to date, but there is the general concern that combining incident and prevalent cases could have introduced a survival bias in our results. Recent studies of genetic risk factors have noted that inclusion of prevalent cases did not result in a bias of the hazards ratio estimates with or without left-truncation of the data, as was observed in our study.^40-41 In addition, a survival bias due to a higher hazard ratio in the years following diagnosis would tend to obscure adverse prognostic markers (a bias toward the null) and would lead to an underestimation of the adverse prognostic impact of BRCA mutations in the current study. Another limitation of this study is the use of a non-concurrent, albeit more recent, cohort of controls which may not have had the same opportunity for being diagnosed with prostate cancer due to secular trends. However, the prevalence of subclinical prostate cancer in a younger control group would have resulted in a bias towards the null in the odds ratio analysis, and would not have affected the prognostic findings which are based on a case-case analysis.

If it is confirmed that BRCA mutations confer a more aggressive clinical course for men with localized prostate cancer, tailored interventions may be warranted. However, due to the small number of mutation carriers in this study the strong effect reported must be viewed as hypothesis generating and requires confirmation in a larger cohort. The additional finding that the interval to castrate metastases after biochemical relapse appears shorter in BRCA2 mutation carriers may be an appropriate factor to include in the consideration of anti-androgens in this genetically defined subset. Recent therapeutic strategies such as small molecules inhibiting Poly(ADP-Ribose) polymerase (PARP), may biologically target a spectrum of BRCA-deficient tumors, with evidence of response in hereditary breast, ovarian, and prostate cancers.⁴² Given the estimated .0069 BRCA2 mutation frequency in the population and the relative risk for prostate cancer of 3.18 observed here, this translates to ~1.5% of the 180,000 prostate cancer cases in the U.S. attributable to BRCA2 mutations, or 2790 cases each year which could benefit from treatments tailored to BRCA2 status. PARP inhibitors may improve outcome for patients with advanced BRCA-associated prostate cancer, but due to the aggressive nature of this disease, investigation of PARP inhibitors in the neoadjuvant setting may also be warranted.

Further large scale studies will be needed to determine if the findings shown here, in a genetic isolate (Ashkenazi Jews), can be generalized to the population at large. The types of protein-truncating mutations of BRCA studied here are representative of BRCA mutations in the general population, and the clinical findings regarding risk and outcome of BRCA-associated malignancies in Ashkenazim have been applied to other populations.^{3, 43} This is further supported by concordant findings in an Icelandic population with advanced BRCA2-associated prostate cancer¹⁸. Nonetheless, due to the small number of BRCA carriers in this study, large consortia will be needed to confirm the associations with histology and clinical outcome reported here. Pending such studies, recommendations for screening for and treatment of BRCA-associated prostate cancer should be individualized based on established clinical factors and prognostic markers.