0022-5347/03/1691-0240/0 THE JOURNAL OF UROLOGY® Copyright © 2003 by AMERICAN UROLOGICAL ASSOCIATION
Vol. 169, 240 –244, January 2003 Printed in U.S.A.
FAMILIAL PROSTATE AND BREAST CANCER IN MEN TREATED WITH PROSTATECTOMY FOR PROSTATE CANCER: A POPULATION BASED CASE-CONTROL STUDY PIERRE I. KARAKIEWICZ, WILLIAM FOULKES, SIMON TANGUAY, MOSTAFA M. ELHILALI AND ARMEN G. APRIKIAN* From the Departments of Surgery (Urology), Medicine, Human Genetics and Oncology McGill University, Montreal, Quebec, Canada
Purpose: We assessed familial prostate and breast cancer in Quebec. Materials and Methods: Using a self-administered mail survey we assessed the prevalence of prostate and breast cancer in first degree relatives of 1,633 men treated with prostatectomy for prostate cancer in the province of Quebec and in first degree relatives of 1,386 spouse controls. Results: The OR of familial breast cancer was 1.1 (95% CI 0.9 to 1.4). The OR of 3.0 (95% CI 2.5 to 3.6) recorded for prostate cancer was modified by francophone versus anglophone linguistic preference (OR 3.2, 95% CI 2.6, 3.9 versus 1.5, 95% CI 0.8 to 2.7, p ⫽ 0.02). Male sibship size was a statistically significant parameter modifying this association (p ⫽ 0.02), namely no brothers (OR 1.7, 95% CI 1.0 to 2.8), 1 or 2 (OR 3.1, 95% CI 2.2 to 4.3) and 3 or more (OR 3.9, 95% CI 2.9 to 5.2). Geographic regions of the province including and neighboring greater Montreal showed a lower OR than more peripheral regions (2.5, 95% CI 2.0 to 3.2 versus 4.1, 95% CI 2.9 to 5.7, p ⫽ 0.02). Conclusions: Francophone men with large male sibships residing in remote areas may be at higher risk for familial prostate cancer and represent the ideal target for further efforts to determine the genetic component of prostate cancer in Quebec. KEY WORDS: prostate, prostatic neoplasms, family, breast neoplasms, questionnaires
Prostate cancer is the most frequently diagnosed malignancy and the second leading cause of cancer death in North American men.1, 2 Studies consistently show that prostate cancer in a family is associated with disease development.3⫺7 Although several population based series have provided general risk estimates, these estimates may vary according to ethnicity and geography.3⫺11 Exposure to different genetic and environmental risk factors may also vary according to geographic region and be population specific. A greater prevalence of familial aggregation may be present in remote, geographically isolated regions, where less extensive gene mixing may explain persistent founder mutations. Population specific differences in the context of the province of Quebec may stem from ethnically heterogeneous French and English populations.8 To address these issues we initiated a population based self-administered questionnaire study to assess the prevalence of familial prostate and breast cancer in Quebec men with prostate cancer and explore potential differences in familial cancer prevalence according to ethnicity and geographic location. MATERIALS AND METHODS
The target population included 4,997 men treated with radical prostatectomy in Quebec between January 1, 1988 and January 16, 1996. The cases represented all radical prostatectomy claims submitted to the Quebec Health Care Plan data base using the radical prostatectomy specific billing code. Since the Quebec Health Plan represents the exclusive medical insurer in the province, billing code based rad-
ical prostatectomy identification results in virtually complete ascertainment. According to the Quebec Health Care Plan Database 4,546 of the 4,997 men treated with radical prostatectomy during this period were alive on August 1, 1997. These men were mailed a self-administered mail survey. The study was approved by the patient privacy protection branch and ethics board of the Quebec Health Care Plan. The survey focused on familial history of prostate and breast cancer in addition to proband age, education, household income and living with a spouse or partner. For prostate and breast cancer positive family history was defined as cancer in 1 or more first degree relatives. For breast cancer familial history was defined as the prevalence of the disease in the mother and/or sisters. The survey also assessed the total number of male and female sibs regardless of cancer status. Also, each spouse was asked to answer the same questions in a separate section, serving as controls. In addition to assessing the prevalence of familial prostate and breast cancer in the province, we examined the prevalence according to the proband geographic region of residence, as defined by the 1989 government administrative regional grouping.12 The geographic region of residence was defined according to mailing address records. Furthermore, we compared the familial prostate cancer prevalence in anglophone and francophone probands. The distinction in anglophones and francophones was made according to proband linguistic preference of communication with the Quebec Health Care Plan and, therefore, surveys in the proband language of preference was used. Based on a previously established concordance of linguistic preference in husbands and wives13 spouses were categorized in a manner similar to the proband. In addition, prevalence values and ORs were compared according to male sibship size. All statistical anal-
Accepted for publication August 16, 2002. Supported by the Fonds de la Recherche en Sante du Quebec, American Foundation for Urologic Diseases, National Cancer Institute of Canada, Medical Research Council of Canada and AstraZeneca Pharmaceuticals. * Requests for reprints: MUHC, L8 –309, 1650 Cedar Ave., Montreal, Quebec, Canada H3G-1A4l 240
FAMILIAL PROSTATE AND BREAST CANCER IN MEN TREATED WITH PROSTATECTOMY
yses were performed by commercially available statistical software. All tests were 2-sided and a 0.05 level of significance was used.
TABLE 2. Prostate cancer in families of men treated with prostatectomy for prostate cancer and families of spouse controls overall and stratified according to French or English preference Familial Prostate Ca
Of 4,546 probands alive at the time of survey mailing 1,633 (35.9%) responded and 1,386 of their spouses (95.8%) participated (table 1). The overall odds of familial prostate cancer was 3.0 in cases versus controls (95% CI 2.5 to 3.6, p ⬍0.001, table 2). Stratum specific ORs according to linguistic preference of the proband showed a low 1.5 OR of familial prostate cancer in English speaking participants (table 2). Conversely the OR was significantly greater in French speaking participants (table 2). The Mantel-Haenszel linguistic preference adjusted OR of 3.0 (95% CI 2.5 to 3.6, p ⬍0.001) was not different from the crude OR of 3.0, suggesting a lack of confounding. However, when stratified according to linguistic preference, there was a significantly different OR in anglophones and francophones (Mantel-Haenszel test of homogeneity of odds chi-square homogeneity of odds 5.4, p ⫽ 0.02). Consequently further analyses were performed separately in the francophone and anglophone subgroups. Stratum specific OR according to male sibship size. The effect of male sibship size on familial prostate cancer in francophone probands showed a lack of confounding (MantelHaenszel combined versus crude OR 3.1 versus 3.2). When the ratio of male sibs in patients with prostate cancer relative to those in controls was stratified into 3 categories (0 versus 1 or 2 versus 3 or more brothers), the stratum specific OR of familial prostate cancer showed significant effect modification (Mantel-Haenszel chi-square homogeneity of odds 7.6, p ⫽ 0.02, table 3). Testing anglophone probands failed to reveal confounding by sibship size (crude versus adjusted OR 1.5 versus 1.6). In contrast to francophones, strata homogeneity could not be rejected in anglophone survey participants (data not shown, Mantel-Haenszel chi-square homogeneity of odds 2.3, p ⫽ 0.3). The mean age of probands did not differ significantly with varying sibship size (fig. 1).
TABLE 1. Characteristics of 1,633 probands and 1,386 controls treated with radical prostatectomy for prostate cancer in Quebec between January 1, 1988 and January 16, 1996 Probands
Age: Range Mean SD % Education level: Grade school or less High school Post-secondary school % Household income/yr. (Canadian $): 0–30,000 30,001–50,000 Greater than 50,000 % Living with spouse or partner No. survey participation: Targeted population Participants % Participation: Vs. targeted population/column Vs. total 1,633 participants % Male sibship size distribution: 0 1–2 3 or More % Control male sibship size distribution: 0 1–2 Male siblings 3 or More No. 1996 Quebec population male ⫹ 45 yrs. or older (%)
43–80 63.3 5.8
49–78 64.9 5.5
43–80 63.4 5.8
29.7 37.2 33.1
9.7 34.7 55.6
27.9 37.0 35.1
50.1 28.4 21.5 88.4
24.8 20.4 54.7 90.9
47.8 27.7 24.5 88.6
18.2 37.4 44.4
37.3 50.7 12.0
20.0 38.6 41.5
22.8 36.9 40.3 900,090 (83.9)
33.1 23.7 53.1 38.5 13.8 37.5 93,095 1,109,735 (8.4) (100)
No. overall: Cases Controls OR (95% CI) p Value No. English: Cases Controls OR (95% CI) p Value No. French: Cases Controls OR (95% CI) p Value
150 130 1.5 (0.8–2.8) 0.2
1,483 1,256 3.2 (2.6–3.9) 0.001
1,633 1,386 3. (2.5–3.6) ⬍0.001
Stratum specific OR according to geographic region of residence. At least 1 patient and 1 control with and without familial prostate cancer were identified in 11 of the 16 geographic regions of the province, allowing the assessment of geographic region of residence specific OR. These values were 1.5 to 5.0 (table 4.). No individual region of residence was identified as a significant confounder of the main effects (crude and Mantel-Haenszel combined OR 2.9) or as an effect modifier (Mantel-Haenszel homogeneity chi-square 9.7, p ⫽ 0.5). However, inspection of region specific ORs suggested that the regions in the southwest including and neighboring greater Montreal had a lower familial prostate cancer OR than regions in the north and east (fig. 2). Testing to this grouping resulted in an OR of 2.5 (95% CI 2.0 to 3.2) in the southwestern regions and 4.0 (95% CI 2.9 to 5.7) in the other regions. These stratum specific ORs differed significantly, suggesting effect modification by grouped regions (MantelHaenszel homogeneity chi-square 5.1, p ⫽ 0.02). The mean age of probands did not differ significantly among individual regions (fig. 3). Combined effect of geographic region grouping and male sibship size. A statistical correlation of male sibship size with low and high OR geographic region was observed (2.15 and 2.72 brothers, respectively, t test p ⬍0.001). The MantelHaenszel combined OR test for the potential combined effect of high versus low OR region grouping with male sibship size showed that after accounting for effect modification by region male sibship size was not a confounder. Effect modification by male sibship size within the low OR region stratum showed a significant difference in ORs (Mantel-Haenszel test of homogeneity of odds chi-square 9.1, p ⫽ 0.01). The ORs were 1.27, 2.34 and 3.8 for 0, 1 or 2 and 3 or more male sibs, respectively. Within the high OR region no confounding by sibship size was noted. ORs according to sibship size were 2.9, 4.7 and 4.1 for 0, 1 or 2 and 3 or more male sibs, respectively. Male sibship size within the high OR region was not a statistically significant effect modifier (MantelHaenszel test of homogeneity of odds chi square 0.75, p ⫽ 0.7). Table 3 shows the combined effect of sibship size and high versus low OR regions. No confounding by region grouping was noted (Mantel-Haenszel combined versus crude ORs) and similarly no significant effect modification was observed (Mantel-Haenszel test of homogeneity of odds). Familial breast cancer. Analysis of breast cancer family history in men treated with prostatectomy for prostate cancer showed an OR of 1.1 (95% CI 0.9 to 1.4, p ⫽ 0.2), suggesting no difference in familial breast cancer prevalence in probands and spouse controls. Similarly the MantelHaenszel linguistic preference adjusted OR of 1.1 (95% CI 0.9 to 1.4) was not different from the crude estimate and there
FAMILIAL PROSTATE AND BREAST CANCER IN MEN TREATED WITH PROSTATECTOMY
TABLE 3. Prostate cancer in families of men treated with prostatectomy for prostate cancer and spouse controls stratified according to male sibship size No. Male Sibs
Cases Controls: OR (95% CI) Age adjusted OR (95% CI) Low risk regions OR (95% CI) High risk regions OR (95% CI) Mantel-Haenszel combined high and low risk regions OR (95% CI) Chi-square high vs. low risk regions Mantel-Haenszel odds homogeneity Cases Controls: OR (95% CI) Age adjusted OR (95% CI) Low risk regions OR (95% CI) High risk regions OR (95% CI) Mantel-Haenszel combined high and low risk regions OR (95% CI) Chi-square high vs. low risk regions Mantel-Haenszel odds homogeneity Cases Controls: OR (95% CI) Age adjusted OR (95% CI) Low risk regions OR (95% CI) High risk regions OR (95% CI) Mantel-Haenszel combined high and low risk regions OR (95% CI) Chi-square high vs. low risk regions Mantel-Haenszel odds homogeneity
Familial Prostate Ca Yes
No. 0 228 258
Totals 270 286 1.7 (1.02–2.8) 1.7 (1.03–2.9) 1.3 (0.7–2.4) 2.9 (1.1–7.3) 1.7 (1.01–2.8) 2.0
No. 1–2 397 411
554 464 3.1 3.0 2.3 4.7 3.0 3.4
No. 3 or greater 399 432
654 503 3.9 3.9 3.8 4.1 3.9 0.07
(2.2–3.3) (2.2–4.3) (1.5–3.6) (2.6–8.5) (2.1–4.2)
0.04 0.04 0.4 0.02 0.15
⬍0.001 ⬍0.01 ⬍0.001 ⬍0.001 0.06
(2.9–5.3) (2.9–5.3) (2.6–5.6) (2.5–6.6) (2.9–5.3)
⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 0.8
participant and 2,913 nonparticipants. The binary logistic regression model showed a significant contribution of age category for predicting survey nonparticipation after controlling for year of surgery and the province region where surgery was performed. Men 70 years or older at surgery were 5 times less likely to participate than those 54 years and younger. Response rates were similar in anglophone and francophone participants. DISCUSSION
FIG. 1. Proband age distribution according to sibship size. Boxes represent 25th to 75 percentile distribution. Whiskers represent 1.5th to 97.5th percentile distribution. Horizontal lines within boxes represent mean. Top and bottom lines indicate ⫹1 SD of overall group mean (dashed line).
was no evidence of effect modification (Mantel-Haenszel homogeneity of odds chi-square 1.2, p ⫽ 0.3). Analysis according to female sibship size revealed no confounding (adjusted OR 1.1, 95% CI 0.9 to .4). Female sibship size stratum specific ORs failed to indicate significant differences (MantelHaenszel homogeneity chi-square 0.7, p ⫽ 0.7). Furthermore, sufficient data were available for analysis of 12 of 16 region specific ORs of 0.2 (95% CI 0.0 to 1.3) to 2.6 (95% CI 0.9 to 7.1). There was no confounding of the main effects model by region of residence, as evidenced by the adjusted OR of 1.2 (95% CI 0.9 to 1.5). The homogeneity of region specific ORs was not rejected (chi-square 12.4, p ⫽ 0.6). Nonresponse bias. Assessment of survey participation showed a 64.1% proportion of nonresponders. Missing data analysis focused on differences in the profile of the 1,633
Our results corroborate previous reports of a 3-fold increase in the odds of prostate cancer in probands with a positive prostate cancer family history relative to probands without familial prostate cancer.3⫺11 Furthermore, we identified appreciable effect modification by proband linguistic preference, as evidenced by significantly different OR in francophone and anglophone men (OR ⫽ 3.2 versus 1.5, p ⫽ 0.02). This lack of a significantly stronger association in familial prostate cancer in cases versus controls in anglophone probands contradicts previous reports.7, 9, 10 Selection bias and small sample size of the anglophone group represent significant potential contributors to this discrepancy. In addition, linguistic preference is not a direct assessment of ethnicity and thus, these data should serve to promote a more in-depth analysis of familial cancer and ethnicity. Previous reports have shown that the risk of prostate cancer increases in proportion to the number of first degree relatives with prostate cancer. We explored whether the risk varied according to the total number of men in a family irrespective of prostate cancer diagnosis. The OR was compared in familial prostate cancer cases and controls for 3 strata defined by the number of brothers as 0, 1 or 2 and 3 or more. Across these strata the odds of familial prostate cancer was significantly higher in cases than in controls (table 3). These observations suggest that in men with prostate cancer large male sibship size may contribute to an increased risk of familial prostate cancer. Several significant biases may explain this finding, including possible differences in reporting or recall in cases and spouse controls with respect to the health of brothers. It is tempting to assume that large male families have greater familial prostate cancer due only to the
FAMILIAL PROSTATE AND BREAST CANCER IN MEN TREATED WITH PROSTATECTOMY TABLE 4. Families with and without familial prostate cancer according to geographic region of residence in Quebec Administrative Region No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 * Southwest Quebec.
Bas-Saint-Laurent Saguenay–Lac-Saint-Jean Quebec Mauricie–Bois-Francs Estrie* Montreal* Outaouais* Abitibi–Temiscamingue Cote-Nord Nord-du-Quebec Gaspesie–Iles-de-la-Madelaine Chaudiere–Appalaches Laval* Lanaudiere Laurentides* Monteregie*
No. Familial Prostate Ca Cases
No. Familial Prostate Ca Controls
12 24 64 59 18 105 6 9 5 0 4 16 23 23 36 83
32 53 106 97 39 317 19 13 13 2 8 31 63 53 119 181
2 5 28 11 8 70 3 0 2 0 0 0 13 0 11 19
22 45 188 91 47 529 17 9 8 2 0 0 52 4 82 120
OR (95% CI)
Age Adjusted OR (95% CI)
4.1 (0.8–20.3) 4.1 (1.4–11.6) 4.1 (2.4–6.7) 5.0 (2.5–6.7) 2.7 (1.0–6.9) 2.5 (1.7–3.5) 1.8 (0.4–8.3) – 1.5 (0.2–9.9) – – – 1.5 (0.7–3.2) – 2.3 (1.1–4.7) 2.9 (1.7–5.0)
4.0 (0.8–19.9) 4.1 (1.5–11.8) 4.0 (2.4–6.7) 5.0 (2.5–10.2) 2.7 (1.1–6.9) 2.5 (1.8–3.5) 1.9 (0.4–8.7) – 1.9 (0.3–13.0) – – – 1.4 (0.6–3.1) – 2.3 (1.1–4.7) 3.0 (1.7–5.2)
FIG. 3. Proband age distribution according to Quebec geographic region. Boxes represent 25th to 75 percentile distribution. Whiskers represent 1.5th to 97.5th percentile distribution. Horizontal lines within boxes represent mean. Top and bottom lines indicate ⫹1 SD of overall group mean (dashed line).
FIG. 2. Province of Quebec map showing administrative regions (01 to 16) used in analysis. Inset, Montreal region and neighboring City of Laval.
greater number of men at risk, although one must consider that family size was controlled for among the spouses in this analysis. While this observational study did not address etiological explanations for this phenomenon, it suggests that this association may warrant further attention. We noted a relatively wide range of geographic region specific ORs (1.5 to 5.0, table 4) in the province. These differences were based on administrative boundaries and did not reflect ethni-
cally distinct regions.12 When these regions were grouped into 2 larger zones, there appeared to be significantly greater odds of familial prostate cancer in northeastern Quebec, which includes more rural and remote regions, than in southwestern Quebec, which includes metropolitan Montreal and more urban areas (OR 2.5 versus 4.0). However, notably such grouping substantially increased the power of the test by decreasing the number of strata from 11 to 2. Furthermore, there was a slightly larger sibship size families in remote regions (2.7 versus 2.1 brothers). However, multivariate analysis suggested that sibship size and geographic region were somewhat independent. In addition, it is possible that there may have been differences in reporting by rural and urban cases and controls. Nevertheless, our data suggest that more remote areas of the province represent the principal contributors to the observed prevalence of familial prostate cancer. This observation may be due to genetic and/or environmental factors. Although this finding requires corroboration, it suggests that the risk of familial prostate cancer may be higher in more remote areas with a stronger founder effect relative to metropolitan centers, where a greater extent of gene mixing likely occurs. Also, remote areas may be exposed to a more agrarian economy, different dietary habits and as yet unidentified factors. Based on this hypothesis the more remote
FAMILIAL PROSTATE AND BREAST CANCER IN MEN TREATED WITH PROSTATECTOMY
areas of the province may represent a useful region for further studies of familial prostate cancer. Despite previous reports suggesting the association of breast with prostate cancer14⫺18 this relationship was not confirmed in our study. Alternatively our data corroborate the results of a population based cohort study of first degree relatives of men with prostate cancer in Sweden that showed a higher than expected incidence of prostate cancer but no difference in the incidence of other malignancies.19 Others have reported a relative risk of 1.2 (95% CI 1.01 to 1.33) for fatal prostate cancer in the presence of familial breast cancer and an OR of 1.5 for other malignancies, including breast cancer, in the first degree relatives of men with prostate cancer.15, 20 These estimates suggest a markedly weaker association of breast with prostate cancer than those in studies addressing more select populations. For example, Warner et al focused on women of Jewish ancestry with and without BRCA mutations and identified a prostate cancer relative risk of 3.4 (95% CI 1.5 to 7.6).14 Therefore, a stronger relationship of the 2 cancers may exist in high risk or more specific populations. Our study contains several methodological limitations that require attention. A significant potential bias exists when only a minority of potential cases participates in a study. It is possible that men with prostate cancer are more likely to respond to such a survey if they have relatives who also have or had prostate cancer. In addition, our analysis indicates that older men were not as well represented as younger men, which could have introduced selection and reporting biases, especially in a disease related to increasing age. Also, it is known that familial disease is greater in younger men. Moreover, since our survey did not ascertain the age of spouse controls or respective sibs, one cannot exclude that age discrepancies in controls and sibs may have biased results. Furthermore, although spouses have served as controls in several epidemiological studies, they do not represent ideal controls in such studies and could have introduced reporting bias. In addition, our cases included only men with presumed localized prostate cancer who underwent radical prostatectomy and did not represent the whole range of prostate cancer cases. CONCLUSIONS
Our data suggests that francophone men with large male sibships who live in remote areas of Quebec may represent those at highest risk for familial prostate cancer. These findings may help increase the effectiveness of studies aimed at determining the genetic epidemiology of prostate cancer. Drs. Michel Bazinet and Mike Kattan provided assistance. REFERENCES
1. Levy, I. G., Iscoe, N. A. and Klotz, L. H.: Prostate cancer: 1. The descriptive epidemiology in Canada. CMAJ, 159: 509, 1998 2. Karakiewicz, P. I. and Aprikian, A. G.: Prostate cancer: 5.
Diagnostic tools for early detection. CMAJ, 159: 1139, 1998 3. Carter, B. S., Beaty, T. H., Steinberg, G. D., Childs, B. and Walsh, P. C.: Mendelian inheritance of familial prostate cancer. Proc Natl Acad Sci USA, 89: 3367, 1992 4. Carter, B. S., Bova, G. S., Beaty, T. H., Steinberg, G. D., Childs, B., Isaacs, W. B. et al: Hereditary prostate cancer: epidemiologic and clinical features. J Urol, 150: 797, 1993 5. Gronberg, H., Isaacs, S. D., Smith, J. R., Carpten, J. D., Bova, G. S., Freije, D. et al: Characteristics of prostate cancer in families potentially linked to the hereditary prostate cancer 1 (HPC1) locus. JAMA, 278: 1251, 1997 6. Keetch, D. W., Rice, J. P., Suarez, B. K. and Catalona, W. J.: Familial aspects of prostate cancer: a case control study. J Urol, 154: 2100, 1995 7. McLellan, D. L. and Norman, R. W.: Hereditary aspects of prostate cancer. CMAJ, 153: 895, 1995 8. Ghadirian, P., Cadotte, M., Lacroix, A. and Perret, C.: Family aggregation of cancer of the prostate in Quebec: the tip of the iceberg. Prostate, 19: 43, 1991 9. Fincham, S. M., Hill, G. B., Hanson, J. and Wijayasinghe, C.: Epidemiology of prostate cancer: a case-control study. Prostate, 17: 189, 1990 10. Ghadirian, P., Howe, G. R., Hislop, T. G. and Maisonneuve, P.: Family history of prostate cancer: a multi-center case-control study in Canada. Int J Cancer, 70: 679, 1997 11. Aprikian, A. G., Bazinet, M., Plante, M., Meshref, A., Trudel, C., Aronson, S. et al: Family history and the risk of prostatic carcinoma in a high risk group of urological patients. J Urol, 154: 404, 1995 12. Cartier, Y.: Les regions administratives du Quebec. Quebec City, Quebec, Canada: Publications du Quebec, 1990 13. Foulkes, W. D., Brunet, J. S., Sieh, W., Black, M. J., Shenouda, G. and Narod, S. A.: Familial risks of squamous cell carcinoma of the head and neck: retrospective case-control study. BMJ, 313: 716, 1996 14. Warner, E., Foulkes, W., Goodwin, P., Meschino, W., Blondal, J., Paterson, C. et al: Prevalence and penetrance of BRCA1 and BRCA2 gene mutations in unselected Ashkenazi Jewish women with breast cancer. J Natl Cancer Inst, 91: 1241, 1999 15. Rodriguez, C., Calle, E. E., Tatham, L. M., Wingo, P. A., Miracle-McMahill, H. L., Thun, M. J. et al: Family history of breast cancer as a predictor for fatal prostate cancer. Epidemiology, 9: 525, 1998 16. Cerhan, J. R., Parker, A. S., Putnam, S. D., Chin, B. C., Lynch, C. F., Cohen, M. B. et al: Family history of prostate cancer risk in a population-based cohort of Iowa men. Cancer Epidemiol Biomarkers Prev, 8: 53, 1999 17. Lehrer, S., Fodor, F., Stock, R. G., Stone, N. N., Eng, C., Song, H. K. et al: Absence of 185delAG mutation of the BRCA1 gene and 6174delT mutation of the BRCA2 gene in Ashkenazi Jewish men with prostate cancer. Br J Cancer, 78: 771, 1998 18. Vazina, A., Baniel, J., Yaacobi, Y., Shtriker, A., Engelstein, D., Leibovitz, I. et al: The rate of the founder Jewish mutations in BRCA1 and BRCA2 in prostate cancer patients in Israel. Br J Cancer, 83: 463, 2000 19. Damber, L., Gronberg, H. and Damber, J. E.: Familial prostate cancer and possible associated malignancies: nation-wide register cohort study in Sweden. Int J Cancer, 78: 293, 1998 20. McCahy, P. J., Harris, C. A. and Neal, D. E.: Breast and prostate cancer in the relatives of men with prostate cancer. Br J Urol, 78: 552, 1996