expression with high rale of cell proliferation, increased risk of visceral metastasis and poor long-term survival in breast cancer. Int J Cancer 49:650-655, 1991 (//)

KALLJONIEMI O-P,

VISAKORPI T, PARONEN I,

ET AL: A new algorithm to compensate for the effects of sliced nuclei on DNA histogram improves the prognostic impact of S-phase values from paraffin-embedded breast and prostatic carcinomas. Cytometry 12:413-421, 1991 (12) IWAYA K, TSUDA H, HlRAIDE H, ET AL: Nuclear p53 immunoreaction associated with poor prognosis of breast cancer. Jpn J Cancer Res 82:835-840, 1991 (13)

THOR AD, MOORE DH, EDGERTON SM, ET AL:

Accumulation of p53 rumor suppressor gene protein is an independent marker of prognosis in breast cancer. J Natl Cancer Inst. In press (14)

CATTORETTI G, RUJCE F, ANDREOLA S, ET AL:

p53 expression in breast cancer. Int J Cancer 41:178-183, 1988 MARTINEZ J, GEORGOFF I, MARTINEZ J, ET AL:

Cellular localization and cell cycle regulation by a temperature-sensitive p53 protein. Genes Dev 5:151-159, 1991 (16)

CHEN L-C,

NEUBAUER A, KURISU W, ET AL:

Loss of heterozygosity on the short arm of chromosome 17 is associated with high proliferative capacity and DNA aneuploidy in primary human breast cancer. Proc Natl Acad Sci USA 88:3847-3851, 1991 (17)

BORG A, BALDETORP B, FERNO M, ET AL:

ERBB2 amplification in breast cancer with high rate of proliferation. Oncogene 6:137143, 1991

Breast Cancer Risks in Relatives of Male Breast Cancer Patients David E. Anderson,* Michael D. Badzioch Background: Previous studies have provided conflicting results concerning the familial effect of male breast cancer on breast cancer risks in female relatives. Purpose: We studied breast cancer risks in flrst-degree relatives of male patients and compared them with relatives of female patients. Methods: Our study included 88 consecutively ascertained male patients and 320 of their first-degree relatives as well as 186 consecutively ascertained female patients and 633 of their first-degree relatives. Observed numbers of breast cancers in relatives were compared with the expected number derived from the Connecticut Tumor Registry. Multiple logistic regression analysis was also performed. Results: Relatives of male patients exhibited a significant twofold increased risk when compared 1114

Previous studies have provided conflicting results concerning breast cancer risks in female relatives of male breast cancer patients. Casagrande et al. (7) and Rosenblatt et al. (2) reported twofold to threefold increased risks in female relatives of male patients compared with female relatives of male controls, similar to the twofold to threefold risks long reported for relatives of female patients. Olsson and Ranstam (J) and LenfantPejovic et al. (4), however, found significant fourfold to sevenfold increased risks of breast cancer in men when their first-degree female relatives were reported to have had the disease. These high risks are consistent with the proposal of Jacobsen (5) and Williams et al. (6) that men affected with breast cancer are indicative of an inherited high liability to the disease, resulting in high risks to their relatives. To resolve these conflicting results, we conducted a family study of male breast cancer patients and paid special attention to the breast cancer risks in their female relatives.

Patients and Methods Patients for this study were obtained from listings of all men admitted to the M. D. Anderson Cancer Center from 1958 through 1989 with a diagnosis or a provisional diagnosis of breast cancer. Medical records were reviewed to identify patients with microscopically con-

firmed breast cancer, who were White and who lived in the United States. These criteria were imposed to reduce potential heterogeneity in the data and to avoid difficulties in locating patients and their families and in obtaining physician reports, hospital records, and death certificates. Of the 260 medical records, 145 were excluded for the following reasons: 30 were foreign residents; 21 were Black; 17 were Hispanic; 42 had a diagnosis of gynecomastia or no evidence of breast cancer; eight lacked requisite pathological reports; 25 were incorrectly labeled as male; and two had been adopted. Eligibility criteria were thus met in 115 male patients. The collection of family and medical histories of patients and their relatives and the confirmation of reported tumors and cancers followed the procedures previously detailed (7). Extended pedigrees were developed using a sequential sampling scheme to allow correction for ascertainment bias in segregation analysis. In this study, however, only breast cancer risks to the patient's first-degree female relatives are presented because of their relevance to counseling. No breast cancers developed among male relatives. We had to exclude 27 families because of lack of information on dates of birth and/or death and causes of death of siblings, parents, and/or children and because reported cancers had not been confirmed. The excluded patients were compared with the 88 retained patients regarding residence; average age at diagnosis; average age at birth of first child; parity; laterality of disease; average number of sisters, daughters, and mothers; and frequency of family histories of reported breast cancer. Retained patients were slightly older (58.1 years)

Received November 29, 1991; revised April 7, 1992; accepted April 23, 1992. Supported in part by Public Health Service grant CA-40173 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. Department of Molecular Genetics, The University of Texas M. D. Anderson Cancer Center, Houston, Tex. We thank the Department of Patient Studies, The University of Texas M. D. Anderson Cancer Center, for providing patient listings and Mary McCabe for her valuable technical assistance. 'Correspondence to: David E. Anderson, Ph.D., Department of Molecular Genetics, The University of Texas M. D. Anderson Cancer Center, 1100 Holcombe Blvd., HMB Box 209, Houston, TX 77030.

Journal of the National Cancer Institute

Downloaded from http://jnci.oxfordjournals.org/ at University of California, San Fransisco on April 1, 2015

(15)

with expected rates and no difference in risk when compared with that of relatives of female patients. Prostate cancer in the family of a male patient resulted in a significant fourfold increased breast cancer risk compared with a risk of 1.4 in families with no history of prostate cancer. A family history of lung cancer, colon cancer, or melanoma bad no effect on increasing risks of breast cancer. Conclusion: The familial effect of male breast cancer is the same as that of female breast cancer. Implications: Any estimates of breast cancer risk provided to individuals should also consider the occurrence of prostate cancer in the family, since prostate cancer appears capable of at least doubling the underlying twofold risk. [J Natl Cancer Inst 84:1114-1117, 1992]

Vol. 84, No. 14, July 15, 1992

observation, or age 80 years. 2) Multiple logistic regression analysis was performed by the LOGRESS computer program described by McGee (9) to provide odds ratios of the risks of breast cancer in female relatives of male patients compared with those of female patients.

Results Twenty-five (22%) of the 115 male patients had nonbreast multiple primary tumors, including four nonmelanoma skin cancers, four melanomas, three oropharyngeal cancers, three thyroid cancers, five prostate cancers, two lung cancers, two colon cancers, one bladder cancer, and one lymphoma. The four melanomas were significantly in excess of the 0.3 expected number computed from the Connecticut Tumor Registry data, as were the three thyroid cancers (0.1 expected). Three of the melanomas preceded the diagnosis of breast cancer. The three thyroid cancers were follicular or papillary carcinomas and developed in individuals who had been exposed to radiation. Seven other patients received radiation to the chest wall in childhood or adolescence—either as therapy to the thymus or for treatment of gynecomastia, tuberculosis, or acne of the chest and back. One of these patients was exposed to atomic bomb radiation in Nevada during military service. Prior to developing breast cancer, five additional patients had received endocrine therapy for gynecomastia or following prostate surgery. Because it was not known which, if any, of the radiation-exposed or endocrine-treated patients developed breast cancer as a consequence of their exposure or treatment and because we wanted to avoid the possible inclusion of any prostate cancers metastatic to the breast, analyses were performed twice—once including all families and once excluding families of exposed and treated patients. Since only minimal effects on breast cancer risks resulted from exclusion of fam-

ilies of exposed and treated patients, the present results pertain to the total sample of 88 families. Table 1 summarizes the observed and expected frequencies of confirmed breast cancer in first-degree relatives of male breast cancer patients compared with expected frequencies derived from the Connecticut Tumor Registry data. All categories of relatives showed an excess of breast cancer compared with expected rates. The risks were significant for sisters (P = .009) and for all first-degree female relatives (P = .002). The addition of two unconfirmed breast cancers to the sister category increased their odds ratio to 2.40 (95% confidence interval [CI], 1.37-3.90) and increased the odds ratio for all first-degree relatives to 2.05 (95% CI, 1.36-2.96). Results from the multiple logistic regression analysis are given in Table 2. The odds ratios of confirmed breast cancer developing in female relatives of male patients compared with female relatives of female patients were variable, ranging from 0.96 to 7.67. None of the ratios were significantly different from the expected ratio of 1.0. The high ratio for daughters is probably not meaningful, since breast cancer was observed only in four daughters in the male series and in one daughter in the female series. The addition of unconfirmed breast cancers to both groups of relatives reduced the odds ratios, again indicating no difference between the two groups. Risks were calculated to determine whether breast cancer risks were influenced by age at diagnosis in male patients. Results were highly variable for individual classes of relatives and were only suggestive of a difference in all first-degree relatives (risks of 1.87 [95% CI, 0.83-4.22] for diagnoses before age 55 years and 1.26 [95% CI, 0.59-2.68] for diagnoses at later ages). We also investigated some of the cancers that occurred as multiple primary tumors in the male patients to determine

Downloaded from http://jnci.oxfordjournals.org/ at University of California, San Fransisco on April 1, 2015

than excluded patients (52.2 years), but none of the differences were significant. Since many of the excluded patients were ascertained in the latter stages of the study (1984-1989), retained and excluded patients ascertained in this period were compared. Retained patients were again slightly older (55.9 years versus 51.8 years) but not significantly so; significant differences were also not detected for the other variables. Controls for a portion of the study were the first-degree female relatives of a consecutive series of female patients admitted to M. D. Anderson Cancer Center in 1978 to 1979 who had had a histological diagnosis of breast cancer. During 1978 to 1979, 949 women were admitted with a diagnosis of breast cancer by microscopic examination. To provide a sample of approximately 200 patients, we selected every fourth patient for study. The medical records of 267 patients were reviewed. Sixty-eight of these were excluded because of race, residence, diagnosis, or adoption. The 199 remaining patients and their families were followed in the same manner and at the same time as the male breast cancer patients and their families. For analytical purposes, 13 families were excluded because of incomplete information on age, cause of death, age at death, and cancer documentation, leaving 186 patients and their families. There was evidence of underreporting in this group of families, since only 34 of 52 of reported breast cancers in first-degree relatives were confirmed, whereas in the 88 families of male patients 26 of 28 reported breast cancers were confirmed by a medical or vital statistics report. Because of this disparity in confirmed cancers, analyses were conducted in two ways. One used only confirmed breast cancers in relatives of male and female patients, and the other used both confirmed and unconfirmed breast cancers. Two analytical approaches were used. 1) The observed number of breast cancers in first-degree female relatives of male patients was compared with the expected number derived by applying sex-specific, 5-year age-specific, and 5-year calendar time-specific rates from the Connecticut Tumor Registry with the corresponding person-years of observation, using the computer program described by Monson (8). Person-years at risk were determined from the date of birth to the date of death, date of first diagnosis, date of last

Table 1. Observed and expected frequency (risk) of breast cancer in first-degree relatives of 88 male breast cancer patients

First-degree relative Mother Sister Daughter Total

Breast cancer

No. of relatives

Person-years at risk

Observed

Expected

Observed/expected

95% CI

88 134 98 320

6426.4 8487.5 4061.8 18975.7

8 14 4 26

5.63 6.71 1.38 13.72

1.42 2.09 2.91 1.90

0.61-2.80 1.14-3.50 0.78-7.45 1.24-2.78

REPORTS

1115

Table 2. Odds ratios of breast cancer in relatives of 88 male breast cancer patients versus relatives of 186 female patients No. of relatives of Relative

Male patients

Female patients

Adjusted* odds ratio

95% a

Confirmed breast cancers Mother SisteT Daughter Total

88 134 98 320

Mother Sister Daughter Total

88 134 98 320

184 271 178 633

2.23 0.96 7.67 1.50

0.81-6.18 0.43-2.16 0.56-105.5 0.85-2.60

Confirmed and unconfirmed breast cancers 184 271 178 633

0.97 0.77 2.80 1.00

0.39-2.39 0.37-1.59 0.47-17.6 0.60-1.67

•Adjusted for age at last observation of relatives and year of birth of proband.

Discussion The aim of this study was to determine the breast cancer risks in first-degree

female relatives of male breast cancer patients. The results suggested that male breast cancer has the same effect on breast cancer risks in first-degree female relatives as does female breast cancer, since the risk to female relatives of male patients was approximately two times higher than that expected from the Connecticut Tumor Registry data, in agreement with the twofold to threefold risks generally cited for first-degree relatives of female patients (70). Furthermore, no differences in risk were observed between relatives of male and female patients. Risks for relatives of male patients whose disease was diagnosed before the age of 55 years were only slightly higher than those for relatives of male patients diagnosed at or after this age. Rosenblatt et al. (2) reported a similar age effect. Of the five studies of male breast cancer concerned with familial risks, three provided risks in agreement with the twofold to threefold increased risks usually cited for first-degree relatives of female breast cancers (10). Rosenblatt et

Table 3. Risks of breast canceT in first-degree female relatives of male breast cancer patients with or without a family history of prostate cancer or other cancers No. of relatives at risk

Observed/expected*

95% a

Prostate cancer No prostate cancer

65 255

3.99 1.37

1.99-7.14 0.77-2.26

Lung cancer No lung cancer

72 248

1.25 2.09

0.34-3.19 1.31-3.17

Colon cancer No colon cancer

45 275

1.36 2.00

0.27-3.97 1.27-3.00

Melanoma No melanoma

37 283

1.35 1.96

0.15-4.88 1.26-2.92

Family history

•Observed frequencies of breast cancer were compared with expected frequencies from the Connecticut Tumor Registry.

1116

Journal of the National Cancer Institute

Downloaded from http://jnci.oxfordjournals.org/ at University of California, San Fransisco on April 1, 2015

their possible influence on breast cancer risk (Table 3). Prostate cancer in the patient and/or relative resulted in a significantly increased breast cancer risk, while the risk in the absence of a family history of prostate cancer was insignificant. A family history of lung cancer, colon cancer, or melanoma had no effect on increasing risks. The significant risks in families without these tumors stem from including the majority of remaining relatives at risk for whom the risks were expected to approach a value of 2.0 (Table 1). Logistic regression analysis, comparing first-degree female relatives of male patients who had any family history of prostate cancer with those not having a family history of prostate cancer, yielded an odds ratio of 3.68 (95% CI, 1.35-10.0), adjusted for the age of the first-degree relative, year of birth of the patient, and family size.

al. (2) found a risk of 2.95 for sisters and mothers of male breast cancer patients when compared with sisters and mothers of male control patients. Casagrande et al. (1) reported a nonsignificant risk of 1.79 in first-degree female relatives of male patients when compared with relatives of male controls. The present risk estimates of 1.90 and 2.04, respectively, are in general agreement with these estimates. The high risks reported by Olsson and Ranstam (3) and Lenfant-Pejovic et al. (4) may have been influenced by study design and small sample size, since they evaluated the risks in cases versus controls according to a reported family history in first-degree female relatives. Olsson and Ranstam (3) reported breast cancer in five of 95 patients with a family history of breast cancer compared with five of 383 controls, and LenfantPejovic et al. (4) reported breast cancer in five of 91 patients with a family history of breast cancer compared with two of 255 controls. Thus, their high risks seem to be influenced by a deficit of breast cancer among the controls rather than by an increase among the cases. In contrast to the previous case-control studies in which family histories and breast cancer occurrences were based on anamnestic data (2-4), present family histories were collected directly from multiple relatives over the age of 20 years. Also, cases and controls were ascertained from the same patient population, and all patients were subjected to the same documentation procedures within the same follow-up period. Cancers in the families of the female patients, however, were underreported. When unconfirmed cancers were included to allow for underreporting, the results were similar to those using confirmed cancers alone in indicating no differences in risk between relatives of male and female patients. Assuming equal risks in the male and female groups (Table 2), any underreporting would have increased rather than decreased the differences between the two groups. Another deficiency of the present study was the small sample. The exclusion of certain families further reduced sample size but was necessary because required age and documentation information was lacking. The possible bias resulting from this exclusion is not known, but it is not thought to be large because a comparison of excluded and retained patients indicated no significant differences.

(16) found a significantly increased risk of prostate cancer in all relatives (P

Breast cancer risks in relatives of male breast cancer patients.

Previous studies have provided conflicting results concerning the familial effect of male breast cancer on breast cancer risks in female relatives...
2MB Sizes 0 Downloads 0 Views