• BRCA
• Ashkenazi Jews
• prostate cancer
• pathology review

• PSA, prostate specific antigen

The magnitude of the risk of prostate cancer among Ashkenazi Jewish carriers of the common mutations in the BRCA1 and BRCA2 genes, the so-called Ashkenazi BRCA founder mutations (BRCA1 185delAG and 5382insC, and BRCA2 6174delT), remains uncertain. Two large epidemiological studies have suggested that the cumulative incidence of prostate cancer among male first degree relatives of individuals with a founder mutation may approach 30% by 80 years of age.^1,2 Smaller clinical studies have, in general, not demonstrated an increased frequency of founder mutations among Jewish men with prostate cancer.^3–6 Only one study to date has examined clinical characteristics of prostate cancer patients with founder mutations.³ In this study, the three BRCA associated tumours that were identified appeared to be associated with biologically more aggressive disease, presenting with higher PSA levels, higher Gleason scores and more advanced stage at diagnosis, than did prostate cancers diagnosed in men without mutations.

Founder BRCA mutations occur in more than 2% of Ashkenazi Jews.^1,7,8 Thus, an increased risk of prostate cancer among BRCA carriers, or an aggressive phenotype in BRCA associated prostate cancers, could constitute a significant cancer burden in this group, and would have substantial management implications for male carriers of Ashkenazi founder mutations.

We undertook the current study to obtain a more stable estimate of the Ashkenazi founder mutation rate among Jewish men diagnosed with prostate cancer, and to determine if prostate cancer occurring in carriers of the common Ashkenazi BRCA1 or BRCA2 mutations is associated with distinctive pathological or clinical features. The precedent for this hypothesis is best illustrated by the hereditary renal cancer syndromes, in which recognising specific histological subsets of subjects has facilitated the identification of genetically distinct disorders.^9,10

All men newly diagnosed with prostate cancer were identified at sixteen of the 23 hospital based and reference pathology laboratories that provided histological diagnosis of prostate cancer in Israel from 1994 to 1995. This time period was selected because it antedated the widespread use of prostate specific antigen (PSA) screening in Israel. One laboratory declined participation, and a second had no Ashkenazi cases during the period of interest. To represent the remaining five small laboratories, a larger centre, which drew its cases from the same geographic region, was selected for study. Only half of the cases diagnosed at one large reference laboratory were enrolled, by selecting all cases diagnosed at that facility every other month. Each pathology report was reviewed to verify the diagnosis of prostate cancer, and to exclude cases with a previous diagnosis of prostate cancer.

All identified prostate cancer subjects were matched to the Israeli Population Registry. Patients were considered to be of Ashkenazi heritage if they were born in Europe or the United States or, if Israeli born, if both parents were born in Europe or the United States. The diagnostic pathology block was retrieved from Ashkenazi prostate cancer cases when available. Minimal information on cases was retained prior to sample anonymisation including: age at diagnosis, hospital of registration, tissue acquisition method (core biopsy, prostatectomy, or transuretheral resection of the prostate). Participants were then assigned a unique study identifier linked to the pathology specimens. The pathology material was anonymised prior to pathology review and data analysis. Due to the anonymous nature of the study design, medical records were not abstracted, and no information concerning stage or PSA levels at the time of diagnosis was obtained. The study was done under a waiver of the requirement for Institutional Review Board review granted by the National Institutes of Health Office of Human Subjects Research.

We estimated the population founder mutation carrier frequency in Ashkenazi men over 50 years old using data from two study populations: (a) Ashkenazi Jewish male volunteers with no personal history of prostate cancer in the Washington Ashkenazi Study (WAS) (Struewing et al¹ and J Struewing, unpublished data), and (b) Israeli men enrolled in the Molecular Epidemiology of Colorectal Cancer (MECC) study, an Israeli case-control study with phenotype and genotype data available for 901 population based controls (Gruber et al¹¹ and S Gruber and G Rennert, unpublished data). Controls were individually matched by sex, year of birth, and clinic to incident cases of colorectal cancer diagnosed in northern Israel between March 1998 and December 2002.¹¹

For the pathology review, each prostate cancer case identified as positive for one of the three Ashkenazi founder mutations was matched to five non-carrier prostate cancer cases by tissue acquisition procedure (core needle biopsy, transuretheral resection of the prostate, or prostatectomy). The best available match for each carrier case was made using a four tiered system: (a) an exact match by age and pathology laboratory; (b) the closest age match (within 5 years) from the same pathology laboratory; (c) the closest age match (within 5 years) from the same type of laboratory (hospital based or referral laboratory); or (d) the closest age match (within 3 years) from any laboratory type. To select five controls per case, we first selected one control per case, selecting a control from the most optimal tier. A second control for each case was then selected, and so forth. Controls were selected without replacement. Thus, once a control was selected as a match for a case, it was unavailable for selection for another case. Histology slides from carrier and non-carrier cases were randomly arranged, and reviewed by a single pathologist (PD), who was blinded to the mutation status of the cases.

Unstained 5 μm paraffin sections were treated with xylene, rehydrated, and digested with proteinase K, followed by phenol extraction and ethanol precipitation. Using 2−4 μl of the sample, a multiplex PCR was performed to detect the three founder mutations, as described previously.¹²

Mutation rates were estimated as simple proportions. These rates were compared with the expected rates in controls using the χ² test with Yates’ continuity correction. An extended McNemar test was employed to compare binary outcomes in mutation related cases with their matched controls. Here, k_i=number of controls in matched set i, x_i=case result (0 or 1) in matched set i, m_i=number of controls in matched set i with positive result (equal to 1) and n=number of matched sets (equal to 27 in our study). The estimate of the odds ratio was obtained from the sum of x_i (k_i−m_i) over each matched set divided by the sum of m_i(1−x_i). To test whether the odds ratio was different from 1, for each matched set the observed minus the expected value of x, (O_i−E_i) was equal to

and its variance, V_i was equal to

The sum of O_i−E_i over all matched sets was denoted by O−E, and the sum of V_i over all matched sets by V. The test statistic was then calculated as (O−E)²/V. Because of the small sample size in the case-control study, the usual χ² approximation did not provide a reliable reference distribution for this statistic under the null hypothesis. We therefore estimated the p value and confidence limits for the odds ratio by Monte Carlo simulation, using runs of 10 000 simulations.

To compare ordered outcomes, such as Gleason’s grade, between carriers and their matched controls, the stratified Wilcoxon test was used as the test statistic¹³ and, again, the p values and confidence limits for the difference in mean tumor grade were estimated from runs of 10 000 Monte Carlo simulations. A three way test for the difference in the least squared mean age at diagnosis was performed among BRCA1 positive, BRCA2 positive and mutation negative men, using SAS PROC GLM (SAS Institute Inc., Carry, NC, USA). Associations with p<0.05 were considered to be significant.

Among Ashkenazi men, 992 eligible cases of prostate cancer were identified, representing approximately two thirds of all prostate cancer cases diagnosed in Israel during the study period. In 13 cases (1%), no pathology material could be retrieved. In the remaining 979 cases, a haematoxylin and eosin stained slide was obtained for review, and five unstained 5 μm paraffin fixed sections were obtained as the DNA source for mutation testing. DNA was extracted and mutation testing was successfully completed for all three of the founder BRCA mutations on 940 (96%) of these cases.

In the entire cohort, although the BRCA2 related cases were slightly younger at diagnosis than the non-carriers and the BRCA1 related cases, there were no statistical differences in the mean ages of these three groups (BRCA2 71.6 years, non-carriers 73.6 years, and BRCA1 74.2 years). One prostate cancer case, a non-carrier, was diagnosed at the age of 49 years. All other cases were diagnosed at 50 years of age or later. Among the prostate cancer cases, mutation carriers were more frequently diagnosed by core needle biopsy (29/30 (93%) v 625/910 (69%)) than non-carriers. This difference in the mode of diagnostic tissue acquisition was significantly different between mutation carriers and non-carriers (χ²=9.03, p=.01).

The overall founder mutation prevalence among Israeli prostate cancer cases was 30/940 (3.2%). Of these, 16 (1.7%) had a BRCA1 mutation [185delAG, 14 (1.5%); 5382insC, 2 (0.2%)] and 14 (1.5%) had a BRCA2 6174delT mutation. In the WAS and MECC studies, the mutation frequencies found among Ashkenazi men over the age of 50 years with no history of prostate cancer were 1.49 and 1.69 respectively (table 1). The difference between these two frequencies was not statistically significant (OR=0.88; 95%CI=0.3–2.3). When the two comparison groups were combined (WAS+MECC), the increased frequency of any of the founder mutations and of the BRCA1 185delAG founder mutation among Israeli PrC cases reached statistical significance (table 1).

Cases and controls in the pathology review were well matched. An exact match was found for 85% of the first selected controls and for 62%, 54%, 50% and 39% of the second to fifth controls, respectively. In only four cases was a fourth tier control selected for inclusion. One of the 30 identified mutation carriers (with a BRCA1 185delAG mutation) and his five matched controls were eliminated from the pathology case-control study, because the reviewing pathologist was not able to confirm the diagnosis of prostate cancer based on the limited material available for review. The remaining 29 confirmed prostate cancer cases with an identified BRCA1 or BRCA2 founder mutation, and a total of 145 individually matched non-carrier cases constituted the pathology review group.

Mutation carriers and non-carriers were comparable with respect to all the histological characteristics evaluated. While prostate cancer among carriers was more likely to arise in the setting of chronic prostatitis and to be associated with neuroendocrine features than was prostate cancer in non-carriers, these findings were not statistically significant. No significant differences were noted in either the mean primary or secondary Gleason grades or in the Gleason sum (table 2).

A number of case series have examined the carrier frequency among Ashkenazi men with prostate cancer^4–6 (table 3). While the numbers of mutation carriers included in these studies have been very small, these studies have not suggested a large increase in prostate cancer risk associated with founder mutations. One small case-control study of Israeli prostate cancer cases and matched controls did not suggest an increased carrier frequency among cases, but provide a suggestion of a more aggressive phenotype among mutation carriers.

Two larger studies have estimated risk of prostate cancer indirectly, based on the history of prostate cancer in relatives of founder mutation carriers and non-carriers.^1,2 Struewing and colleagues used a kin cohort analytic approach to estimate the cumulative incidence of cancer associated with Ashkenazi founder mutations in a study of over 5000 American Ashkenazi Jewish volunteers from the Washington DC area [the Washington Ashkenazi Study (WAS)].¹ In this study, the founder mutations were found at a similar frequency to that reported in previous studies of Ashkenazi Jews. The risk of prostate cancer by 70 years of age was estimated to be 16% (95% CI 4%–30%) among male carriers of any founder mutation, compared with 3.8% among non-carriers (95% CI 3.3%–4.4%), a risk equal to that of ovarian cancer among female founder mutation carriers. The incidence of prostate cancer in carriers in the WAS was elevated by the age of 50 years, was statistically significantly elevated by 67 years, and increased thereafter with age, approaching 30% by 80 years, suggesting both an overall excess in prostate cancer risk, and an increased risk of early onset disease among founder mutation carriers. While confidence limits were widely overlapping in the WAS analysis, the risk of prostate cancer appeared higher among individuals with BRCA1 mutations than those with BRCA2 mutations, and there was some suggestion of early onset disease among those with a BRCA1 5382insC mutation. In that study, the number of individuals with a family history of prostate cancer was 5/40 (12%), 5/20 (25%), 6/54 (11), 16/114 (14%), and 364/4759 (8%) for 185delAG, 5382insC, 6174delT, any mutation, and non-carriers, respectively. Compared with non-carriers, those with a BRCA1 5382insC mutation and those with any founder mutation were at significantly increased risk of PrC (p<0.01).

Warner and colleagues studied the risk of breast, ovarian and other cancers in the first degree relatives of 412 Ashkenazi Jewish women with breast cancer unselected for age or family history, who elected to undergo mutation testing.² Among first degree relatives of carriers of any founder mutation, the risk of prostate cancer was estimated to be 34% by by the age of 85 years compared with 12.6% among non-carriers (p=.05). Interestingly, this study appeared to enrich for carriers of the BRCA1 185delAG mutation, despite the nearly equal frequencies of the BRCA1 185delAG and the BRCA2 6174delT mutations in the general Ashkenazi population. Only 15 (3.6%) of breast cancer patients in this study carried a 6174delT mutation compared with 26 (6.3%) patients who carried a 185delAG mutation, perhaps reflecting variable penetrance of these two mutations for breast cancer. Differences in subject ascertainment could account for the observed differences in the estimated risk of prostate cancer in these two studies.

In our population based series, 3.2% of Israeli Ashkenazi men with an incident diagnosis of prostate cancer, unselected for family history of breast and/or ovarian cancer, carried an Ashkenazi founder mutation. No population controls were obtained as part of this study. This represents a major limitation of our study design. We have attempted to address this limitation though comparison with existing Ashkenazi groups for which age and sex specific estimates of the founder mutation rate are available, i.e., the WAS and the MECC.

The MECC data are derived from a population based study of Israeli men and, while relatively small, is the best available Israeli comparison group. There is no statistical difference in the mutation frequency among participants in the MECC (1.69%), which may over-represent Ashkenazi Jews immigrating from the former Soviet Union, and in that observed in the WAS (1.49%), composed predominantly of immigrants from Germany, Poland and Austria. Estimates of the Ashkenazi BRCA founder mutation prevalence have been reported for several populations. These rates are strikingly similar across Ashkenazi populations living in various geographical regions.^1,7–8,14 We therefore thought it reasonable to combine the data from the MECC and WAS studies.

In the current study, we did observe an overall twofold risk of prostate cancer among founder mutation carriers. While increased risk of prostate cancer among carriers of the BRCA1 185delAG and the BRCA2 6174delT mutations appeared to be comparable (within the constraints of small numbers), only the estimate for the all mutations combined achieved statistical significance (p<0.05). No increased risk was associated with the 5382insC mutation in this study. The mean age at diagnosis among BRCA1 mutation carriers did not differ significantly from that of the BRCA2 mutation carriers, and neither differed significantly from the age at diagnosis among non-carriers. We found no evidence in this study to suggest an earlier than usual age of diagnosis in BRCA associated prostate cancer.

Interestingly, data from both the WAS and the MECC suggest that the BRCA founder mutation prevalence rate may decrease with age. This would suggest that employing the overall BRCA founder mutation frequency of approximately 2.4%^7–8,14,1 as the comparison, rather than an age specific prevalence estimate, will result in an underestimation of the risk of prostate cancer. If one takes into account a possible decline in founder mutation prevalence with age, our findings are consistent with at most a 2–4 fold increase in the risk of prostate cancer among carriers of an Ashkenazi BRCA founder mutation.

Differences in the reported mutation specific risks probably represent statistical fluctuation related to small sample sizes. While a well designed, population based case–control study testing the hypotheses we explore here would be desirable, a study sufficiently powered to detect mutation specific risks is probably unfeasible. To detect reliably (with a power of 90%) an odds ratio of 1.5 for the separate mutations 185delA, 5382 insC, and 6174delT, one would need total sample sizes of approximately 25 000, 100 000 and 25 000 respectively, assuming equal numbers of cases and controls. However, there would be little clinical interest in such small odds ratios, as risks at this level would result in a very low population attributable risk (%), (calculated by 100.P.OR/(P.OR + 1−P), where P=prevalence of mutation in the general population and OR=odds ratio) associated with BRCA founder mutation status. In our study, P=0.0156 and OR=2.08, so that the population attributable risk=3.2%. Thus, although the relative risk of prostate cancer among carriers of BRCA Ashkenazi Jewish founder mutations is significantly elevated, founder mutations explain only a small percentage of all prostate cancers occurring among Ashkenazi men. In view of the moderately high incidence of prostate cancer among Ashkenazi Jews, the absolute number of cases attributable to the founder mutations is not inconsequential, but alternative aetiological explanations must be sought for the majority of prostate cancers.

This study represents the largest reported series of BRCA related prostate cancer to have undergone detailed, blinded histopathological review. Only one study to date has examined the clinical characteristics of prostate cancer diagnosed in Ashkenazi mutation carriers. In a small case-control study of Israeli Ashkenazi Jewish men with prostate cancer, Hubert and colleagues reviewed the pathology of three BRCA associated prostate cancers (two patients with BRCA1 185delAG mutations and one patient with a BRCA2 6174delT mutation). They suggested that these tumours appeared to be associated with biologically more aggressive disease. These cancers presented with higher PSA levels, higher Gleason scores and more advanced stage at diagnosis than did prostate cancers diagnosed in men without mutations.³

While data on initial stage and PSA at diagnosis were not available in our study, we found no evidence for distinctive clinical or pathological features among prostate cancer cases associated with an Ashkenazi founder mutation compared with prostate cancers occurring among non-carriers. It is interesting to note that the majority of carriers were diagnosed by needle biopsy rather than at prostatectomy. We have insufficient information to address the basis for this association. We selected prostate cancer cases diagnosed from 1994–1995, a time period antedating the widespread use of PSA screening in Israel. Therefore, we believe it unlikely that a detection bias due to differential PSA screening among men with a family history of breast and/or ovarian cancer could account for this association. Some clustering of chronic prostatitis was noted among cases with mutations, which may suggest a predilection for BRCA associated prostate cancers to arise in this setting. We did not find a higher Gleason Score among mutation positive cases.

In summary, our findings are consistent with previous reports of a modest increase in the risk of prostate cancer associated with the Ashkenazi founder mutations. This risk is unlikely to exceed a fourfold increase. This level of risk may warrant consideration of male carriers of Ashkenazi founder mutations as candidates for inclusion in high risk prostate cancer screening and/or risk reduction studies. Our review of the pathological characteristics of founder mutation BRCA associated prostate cancer, while based on a relatively small number of cases, does not suggest that these cases display a more aggressive phenotype.

Members of the Israeli BRCA Prostate Cancer Working Group The following are the contributing centers and the names of the pathologists contributing cases included in this analysis: I Avinoach MD, Wolfson Medical Center, Holon; E Felldberg, Kaplan Medical Center, Rehovot; E Goldshtein MD, Soroka Medical Center, Beer Sheba; G Grossman MD, Hillel Jaffe Medical Center, Hadera; I Kohen MD, West Galil Medical Center, Nahariya; M Konichezky MD, Rabin Medical Center, Beilinson Campus, Petah Tikva; S Lew MD, Meir Medical Center, Kfar Saba; B Lifschitz-Mercer MD, Tel Aviv-Jaffo – Ichilov Medical Center, Tel-Aviv; I Mislevich MD, Bnei Zion Medical Center, Haifa; M Munichor MD, Rambam Medical Center, Haifa; M Resnick MD, Carmel Medical Center, Haifa; A Rozenman MD, Sharei Tzedek Medical Center, Jerusalem; I Sapir MD, Emek Medical Center, Afula; S Szvald MD, Zieff Medical Center, Zefat; A Tchernobilsky MD, PathoLab, Rehovot.