Drs. Kopans and Hughes and Mr. Moore are shareholders in Mistsoft Corporation, a software company involved with breast risk factor collection.
Prevalence of hereditary breast/ovarian carcinoma risk in patients with a personal history of breast or ovarian carcinoma in a mammography population
Article first published online: 26 SEP 2005
Copyright © 2005 American Cancer Society
Volume 104, Issue 9, pages 1849–1853, 1 November 2005
How to Cite
Dominguez, F. J., Jones, J. L., Zabicki, K., Smith, B. L., Gadd, M. A., Specht, M., Kopans, D. B., Moore, R. H., Michaelson, J. S. and Hughes, K. S. (2005), Prevalence of hereditary breast/ovarian carcinoma risk in patients with a personal history of breast or ovarian carcinoma in a mammography population. Cancer, 104: 1849–1853. doi: 10.1002/cncr.21393
- Issue published online: 17 OCT 2005
- Article first published online: 26 SEP 2005
- Manuscript Accepted: 24 MAY 2005
- Manuscript Revised: 4 MAY 2005
- Manuscript Received: 4 MAR 2005
- BRCA1 gene;
- BRCA2 gene;
- genetic counseling;
- genetic screening;
- risk assessment;
- mass screening;
- hereditary neoplastic syndromes;
- breast neoplasm;
- ovarian neoplasm
Identifying BRCA1 and BRCA2 mutation carriers is increasingly important as new management options show promise in decreasing morbidity and mortality in these women. The authors sought to determine the prevalence of family histories suggestive of a hereditary breast carcinoma syndrome in a cohort of patients with a personal history of breast and/or ovarian carcinoma presenting for mammography.
The authors reviewed the family histories of all women with a history of breast or ovarian carcinoma presenting for mammography over a 37-week period. Using the Myriad model, the authors evaluated the prevalence of family histories with a ≥ 10% risk of a BRCA1 or BRCA2 mutation.
During the period of the current study, 14,597 women completed a family history questionnaire. Of these women, 1764 had a personal history of breast or ovarian carcinoma, 86.6% had unilateral breast carcinoma, 4.6% had bilateral breast carcinoma, 8.2% had ovarian carcinoma, and 0.5% had both breast and ovarian carcinoma. Overall, 20.6% met the criteria for a ≥ 10% risk of mutation according to the Myriad model. This incidence was higher among Ashkenazi women (47.3%) and among patients with a personal history of ovarian carcinoma (35.9%).
Application of the Myriad model to women with a personal history of breast and ovarian carcinoma suggested that approximately 1 in 5 of these women (20.6%) will have family histories suspicious for a genetic mutation. This risk was higher for Ashkenazi women and for those with a personal history of ovarian carcinoma. This prevalence was considerably higher than the rate reported among women with no personal history of cancer, and has significant implications for their management, as well as for the capacity for risk assessment and testing. Cancer 2005. © 2005 American Cancer Society.
Approximately 7% of patients with breast carcinoma and 10% of patients with ovarian carcinoma are estimated to be carriers of 1 of the BRCA breast/ovarian carcinoma susceptibility genes.1 Women who carry a deleterious BRCA1 or BRCA2 mutation have a 50–80% lifetime risk of breast carcinoma and a 10–40% lifetime risk of ovarian carcinoma.2 Identifying these women is increasingly important because new management options have demonstrated promise in decreasing the incidence of breast and ovarian carcinoma or in finding these tumors at an earlier, more treatable stage. Breast carcinoma screening with magnetic resonance imaging (MRI) scans,3, 4 chemoprevention of breast carcinoma with tamoxifen,5, 6 chemoprevention of ovarian carcinoma with oral contraceptives,7, 8 prophylactic mastectomy,9, 10 and prophylactic oophorectomy11, 12 are examples of proven effective strategies. In addition, the identification of carriers has benefits beyond the patient herself, by initiating an intensive effort to identify carriers in her extended family.
Currently, primary care providers must recognize that their patient has a significant risk of hereditary cancer to refer for cancer risk assessment. Under the time constraints and pressures of modern medicine, this places an undue burden on the clinician. A more universal, simple, and robust system of identification needs to be developed. This will require a better understanding of the magnitude of this problem, a systematic approach to collecting family histories, and the use of risk models to identify those patients who require further evaluation. Using several currently available models, it has been shown that 3.3–6.0% of female patients without a history of breast or ovarian carcinoma have family histories suggestive of a mutation, and are therefore eligible for additional evaluation.13, 14 Among patients with a personal history of breast or ovarian carcinoma, the proportion with significant family histories may be much higher. Shannon et al.15 reported a rate of 22% of high-risk patients among a cohort of 50 women with a history of breast or ovarian carcinoma, as determined by a genetic counselor.
In the current study, we examined the prevalence of family histories suggestive of a hereditary breast or ovarian carcinoma syndrome in a cohort of patients with a personal history of breast and/or ovarian carcinoma presenting for mammography, to better understand the magnitude of the problem in this population.
MATERIALS AND METHODS
With the approval of our institutional review board, we undertook a retrospective review of family history data collected for all women with a personal history of breast or ovarian carcinoma presenting to the Avon Comprehensive Breast Evaluation Center at the Massachusetts General Hospital between May 13, 2003 and January 28, 2004. The data were prospectively collected using a self-administered questionnaire that included information regarding personal cancer history, family history of breast or ovarian carcinoma, age at the time of the diagnosis of cancer (age < 36 yrs, 36–39 yrs, 40–50 yrs, and > 50 yrs), and Ashkenazi Jewish ancestry.
This questionnaire was administered using two systems: a self-administered scannable questionnaire and a tablet computer system. Both systems immediately downloaded the data into a database (in compliance with all provisions of the Health Insurance Portability and Accountability Act [HIPAA]) and the data were available to clinicians the same day for patient care.
Using the Myriad mutation prevalence tables to estimate mutation risk,16 we evaluated the prevalence of family histories with a ≥ 10% risk of a BRCA1 or BRCA2 mutation, according to the recommendations of the American Society of Clinical Oncology.17, 18 It should be noted that these tables are constantly being updated. We used the tables as they existed in the spring of 2004. These tables are based on the actual results of genetic testing at the Myriad laboratories, as correlated with family history, but they have a few weaknesses that need to be considered. In this model, all relatives, regardless of degree, are counted equally. More importantly, only one breast carcinoma per patient is considered, so patients with bilateral breast carcinoma are not considered to be at a higher risk than those with unilateral breast carcinoma. To correct for this weakness, we made the assumption that bilateral breast carcinoma would be counted as two relatives with breast carcinoma, and that breast plus ovarian carcinoma in a single individual would be counted as two relatives, one with breast carcinoma and one with ovarian carcinoma. For purposes of data analysis, we assumed the age of diagnosis provided for patients or relatives with both breast and ovarian carcinoma to be the age at the time of the breast carcinoma diagnosis, and an unreported age of diagnosis was classified as age ≥ 50 years. Because we collected age data as age ≤ 50 years versus age > 50 years, and the tables grouped patients as age < 50 years or age ≥ 50 years, we assumed that these 2 groupings were the same (realizing it would slightly overestimate the risk for our patients who were exactly age 50 yrs).
We studied the prevalence of family histories suggestive of a hereditary breast carcinoma syndrome in several subgroups (Ashkenazi vs. non-Ashkenazi) and according to the type of carcinoma in the proband (unilateral breast, bilateral breast, breast plus ovarian, and ovarian carcinoma).
Between May 13, 2003 and January 28, 2004, the family histories of 14,597 women were collected. Of these, 1764 women had a personal history of breast or ovarian carcinoma. Among these patients, 86.7% had unilateral breast carcinoma, 4.6% had bilateral breast carcinoma, 8.2% had ovarian carcinoma, and 0.5% had both breast and ovarian carcinoma (Table 1).
|Characteristics||No. of patients||Percent|
|Patients with breast carcinoma|
|Median age at diagnosis in yrs (range)||53 (22–97)|
|Age ≤ 50 yrs||671||41.40%|
|Age > 50 yrs||948||58.60%|
|Patients with ovarian carcinoma|
|Median age at diagnosis in yrs (range)||47 (−18–81)|
|Type of carcinoma|
|Breast and ovarian||8||0.5%|
Of the 1764 patients with a personal history of breast and/or ovarian carcinoma, 20.6% had a ≥ 10% risk of carrying a BRCA1 or BRCA2 mutation. In comparing carcinoma types, 35.9% of women with ovarian carcinoma versus 18.9% of women with unilateral breast carcinoma were considered to have a ≥ 10% risk for carrying a mutation. All the patients with a personal history of both breast and ovarian carcinoma were considered to have a ≥ 10% risk according to the Myriad tables, in which we assumed this was the same as 1 relative with breast carcinoma plus a second relative with ovarian carcinoma.
When analyzed by Ashkenazi ancestry, 47.3% of Ashkenazi women versus 18.5% of non-Ashkenazi women had a ≥ 10% risk of carrying a mutation. All the patients (100%) with Ashkenazi ancestry and a personal history of ovarian carcinoma were considered to have a ≥ 10% by definition, in contrast with only 30.6% of the non-Ashkenazi women in that group. This increased genetic risk among the Ashkenazi women was observed among all subgroups of personal cancer history (Table 2).
|Characteristics||All patients (%)||Ashkenazi (%)||Non-Ashkenazi (%)|
|Breast||288/1529 (18.9)||47/108 (43.5)||241/1421 (16.9)|
|Breast bilaterala||15/82 (18.3)||3/10 (30)||12/72 (16.6)|
|Ovarian||52/145 (35.9)||11/11 (100)||41/134 (30.6)|
|Breast/ovarian||8/8 (100.0)||N/A||8/8 (100)|
|Total||363/1764 (20.6)||61/129 (47.2)||302/1635 (18.5)|
Retrospective analysis of this large series of patients with a personal history of breast and/or ovarian carcinoma presenting for mammography shows that 1 in 5 patients (20.6%) has a family history suggestive of a BRCA1 or BRCA2 genetic mutation. This incidence is highest among the patients with Ashkenazi ancestry (47.3%) and among those with a personal history of ovarian carcinoma (35.9%). This prevalence of high-risk family histories is considerably higher than that reported among women with no personal history of breast or ovarian carcinoma (20.6% vs. 3.3–6.0%),13, 14 and similar to the 22% result reported by Shannon et al.15 among patients with breast carcinoma.
Recognition of a BRCA mutation is often valuable in the decision making for the patient with newly diagnosed breast carcinoma. Although breast conservation appears to remain a viable option, with minimal effect on breast ipsilateral recurrence as compared with noncarriers in the first 10 years, many patients and clinicians prefer mastectomy. In addition, these patients are at a very high risk for the development of contralateral breast carcinoma,19–22 and the patients or their clinicians often consider a contralateral prophylactic mastectomy at the time of surgery. The extrapolation of data from unaffected carriers would suggest that this would markedly decrease contralateral risk.7, 9, 22 The timing of risk assessment and testing related to definitive surgery is thus an important issue to consider due to the influence the genetic testing results may have on the final surgical recommendations.23
After surgery, BRCA mutation status may be an important component of adjuvant therapy discussions, as the contribution and timing of oophorectomy are considered. Prophylactic oophorectomy, generally recommended after childbearing in BRCA mutation carriers, dramatically reduces the incidence of ovarian carcinoma in these women, and appears to significantly reduce the risk of breast carcinoma as well (53% and 68%, respectively, in 2 recent studies).11, 12 In premenopausal women with estrogen-responsive tumors, ovarian ablation is reported to improve the disease recurrence-free and overall survival with a magnitude of benefit similar to that noted with cyclophosphamide, methotrexate, and 5-fluorouracil chemotherapy or tamoxifen,24–26 and ovarian ablation by medical or surgical means is the focus of several international trials currently evaluating its role in relation to other therapies. Therefore, for BRCA mutation carriers with estrogen-responsive tumors, oophorectomy is not only prophylactic but also a key component of their cancer therapy.
Genetic testing may influence nonsurgical management as well. Recently publications by Kriege et al.3 and Warner et al.4 have shown that breast MRI scans are significantly more sensitive than mammography and may detect breast carcinomas at an earlier stage in these patients when compared with other modalities. Therefore, these women, if they opt for close follow-up rather than prophylactic mastectomy, are often placed on an intensified screening schedule involving annual MRI scans and mammograms at 6-month intervals.
If we use this self-reported family history as a screening method, these women can then be easily identified and referred for further evaluation. Both the self-administered scannable questionnaire and the tablet computer system were shown to be effective methods of data collection. They allowed us to capture a family history for virtually every patient presenting to our center without the need of additional personnel in a simple, fast, and private way. This information was immediately available to our health care providers for patient evaluation and care.
Hughes et al.13 and Westman et al.27 reported the utility, feasibility, and accuracy of a self-administered questionnaire in primary care practice as a method of detecting high-risk patients based on their family history. In addition, this method allows updating of the family history at every visit, as opposed to a static family history that serves only as a snapshot of the time it was obtained.
A possible criticism of the current study is the accuracy of self-reported family history data. However, several studies have shown that women can provide accurate information regarding their family history of breast carcinoma. Love et al.28 compared patient-reported breast carcinoma information with medical records, and found that patients were correct in reporting a breast carcinoma diagnosis in 94% of first-degree relatives and in 88% of second and third-degree relatives (91% overall). Parent et al.29 found that 90.5% of familial breast carcinomas reported by affected women and 97% of those reported by unaffected women were correct when compared with pathology reports. Finally, Kerber and Slattery30 have stated that any recall bias operating in the self-reporting of breast carcinoma family histories would have a negligible effect on risk estimates.
Another criticism may be the choice of the risk assessment model utilized. Various models currently are applied to evaluate the patient's risk and to assist in the decision of whether to recommend testing, but ample discrepancies exist between them and the risk probabilities they generate.14, 15 We believe that the Myriad model is a reasonable model to use as a first screen for high-risk women, although we have tried to compensate for some of its weaknesses. As a second breast carcinoma is not considered in the Myriad model, the group of patients with bilateral breast carcinoma may be an underdetected group. In addition, nonwhite women may not be accurately represented by this method because they represent only 1.6% of individuals analyzed by Myriad laboratories.18 Finally, the results obtained in the current study represent the population presenting to a large academic medical center, and may not be applicable to other settings.
Shannon et al.15 showed that 22% of a cohort of 50 patients with breast carcinoma were at high risk according to 3 different models (U Penn, BRCAPRO, and Myriad), using a detailed 3-generation pedigree collected by a genetic counselor. We obtained a similar result using the same threshold (≥ 10%) but using only the Myriad model and a self-administered questionnaire. This strongly suggests that our system has a similar ability to identify those at high risk. In addition, this system is far less labor intensive (requiring no direct face-to-face interaction or data entry) and therefore is sustainable. We continue to collect approximately 100 family histories every day.
Our current method opens the possibility of identifying high-risk women at an industrial strength level and then informing these women of that risk. Although this raises significant logistical and operational issues that we must address, it does not fall under restrictions from HIPAA. HIPAA specifically allows the use of specific clinical information to communicate to appropriate patients the information regarding services that may be of value to them without previous consent.31
The identification of this large number of women at risk has complex implications concerning the current capacity to manage their needs for risk assessment and genetic testing. We believe that only a small proportion of patients at risk are currently being identified and correctly referred. There are few high-risk clinics, and these clinics do not see the number of patients that would translate to this volume. If patients are identified and referred at this volume, it will clearly exceed the capacity of the current resources of the health care system. We will need to develop new approaches to this problem to manage the volume anticipated.
In conclusion, we have developed a simple, fast, and effective method of detecting a large number of patients at high risk for hereditary breast/ovarian carcinoma syndrome in a mammography population. We now need to improve the accuracy and sensitivity of these methods, evaluate the reproducibility of these results in the setting of the general population, and determine how to manage this large number of patients needing risk assessment.
- 16Myriad Genetics, Inc. Identifying patients at risk for HBOC. Available from URL: http://www.myriadtests.com/provider/id_hboc.htm. [accessed Spring 2004].
- 31United States Deparment of Health and Human Services. The health portability and accountability act of 1996 (HIPPAA). Available from URL: http://answers.hhs.gov/ [accessed December 21, 2004].