Abstract
Testing for mutations in BRCA1 and BRCA2can provide important information about breast and ovarian cancer risk to a small but identifiable subgroup of women. Women who test positive for a BRCA1/2 mutation can pursue more aggressive cancer surveillance and prevention regimens. Among families with known mutations, women who test negative may avoid unnecessary interventions. Currently, little is known about the factors associated with the use of clinical BRCA1/2 testing. The objective of this study was to determine the factors associated with decisions about clinical BRCA1/2 testing among women undergoing clinical BRCA1/2 counseling through a retrospective cohort study of women who participated in a university-based clinic offering breast cancer risk assessment, genetic counseling, and BRCA1/2testing between January 1996 and April 1998. From the 251 eligible women who responded to a follow-up survey, 125 (50%) had undergone or were undergoing BRCA1/2 testing, 86 (34%) had decided not to undergo testing, and 40 (16%) were undecided about testing. After multivariate adjustment, we found that women who chose to undergo BRCA1/2 testing were more likely to have a known familial mutation [odds ratio (OR), 7.46; 95% confidence interval (CI), 0.97–62.16], more likely to be Ashkenazi Jewish (OR,6.37; 95% CI, 2.68–15.12), more likely to want cancer risk information for family members (OR, 1.93; 95% CI, 0.99–4.14), more likely to want information about ovarian cancer risk (OR, 1.69; 95%CI, 1.18–3.69), and less likely to be concerned about insurance or job discrimination (OR, 0.45; 95% CI, 0.21–0.94). These associations were also found in the subgroup of women with a predicted probability of a BRCA1 mutation of ≥5%. Our study suggests that approximately half of eligible women choose to undergo clinical BRCA1/2 testing after participating in counseling. Women who have the highest risk of carrying a mutation, and thus the greatest probability of gaining some useful information from the test results,are most likely to undergo testing. Women who undergo testing are also more interested in ovarian cancer risk information and less concerned about job and insurance discrimination.
Introduction
Mutations in the cancer susceptibility genes BRCA1 and BRCA2 are associated with a significantly increased lifetime risk of breast and ovarian cancer (1, 2). Although interest in genetic testing for cancer susceptibility has grown quickly in the medical community, deciding about BRCA1/2 testing remains a potentially complex and difficult process.
The primary benefit of BRCA1/2 testing is the information that can be gained about individual and familial breast and ovarian cancer risk. This information may have significant implications for decisions about cancer surveillance and cancer prevention (3, 4). The limitations and risks of BRCA1/2 testing are complex (4, 5, 6). Currently available options for cancer surveillance and prevention have limited efficacy and/or involve significant trade-offs (4). Furthermore, the cancer risk information gained from testing is limited in most contexts. Outside of families with known mutations, most women test negative and have little change in their predicted risk of breast or ovarian cancer (3). For these women, testing may be unlikely to affect their surveillance or risk reduction regimens. The adverse psychological consequences of positive or negative tests and employment, social, or insurance discrimination are often cited as potential drawbacks to undergoing BRCA1/2 testing (5, 6). In addition, full BRCA1/2 testing currently costs over $2,500, and insurance coverage is variable (7).
Currently, little information is available regarding the uptake of BRCA1/2 testing in a clinical setting or the reason women decide against undergoing testing. To date, most studies have focused on high-risk families offered testing through research protocols (8, 9). The aims of our study were to determine the proportion of women who undergo BRCA1/2 testing and the factors associated with decisions about BRCA1/2 testing among women undergoing BRCA1/2 counseling at a clinical breast cancer risk assessment program that offers genetic testing as a clinical service.
Materials and Methods
Study Setting.
The University of Pennsylvania BCREP3is a multidisciplinary clinical program that provides breast cancer risk assessment, genetic counseling, and genetic testing for BRCA1/2 mutations. The program has provided clinical testing for BRCA1/2 mutations to women without cancer since October 1996. Although research testing is offered selectively based on eligibility criteria, clinical testing is provided to any individual who chooses to undergo testing after participating in genetic counseling. Women with an estimated probability of a BRCA1/2mutation of <5% are counseled that they are unlikely to gain information from testing. During this study, estimates of the probability of BRCA1 mutation were provided using a prediction model developed by Couch et al(10) . A similar BRCA2 prediction model did not exist at the time of the study. On the basis of the population genetics of BRCA1/2, non-Ashkenazi women were told their risk of BRCA2 mutation was about half that of BRCA1, whereas Ashkenazi women were told their risk of BRCA2 mutation was equivalent to that of BRCA1 (11). Women who were not considering undergoing BRCA1/2 testing at the time of their visit to BCREP received individualized information about breast and ovarian cancer risk and surveillance recommendations but did not undergo full pretest genetic counseling.
Study Design and Subject Selection.
A total of 518 individuals participated in the BCREP between January 1995 and April 1998. Women who had previously requested not to participate in further research (n = 22) and men(n = 6) were excluded. In October 1998, all of the eligible subjects (n = 490) were mailed a questionnaire, a letter, and a stamped, addressed envelope. Subjects who did not respond were mailed two reminder letters, including questionnaires. The study protocol was approved by the Institutional Review Board of the University of Pennsylvania.
Data Collection.
To identify factors that were associated with decisions about genetic testing, four focus groups of women (n = 16) who had participated in the BCREP were held. In each group, women were asked to list all of the issues that had influenced their decision about genetic testing. A questionnaire was developed that asked respondents to rate the importance of each factor identified in the focus groups on a four-point Likert response scale (very important, moderately important,a little important, and not at all important). These factors are listed in Table 2. In addition, the questionnaire asked subjects if they had already undergone testing, had decided to undergo testing in the future, were undecided about testing, or had decided not to undergo testing. Sociodemographic characteristics and family history of breast cancer were obtained from clinical records.
Statistical Analysis.
Predicted lifetime risks of breast cancer for subjects without a diagnosis of breast cancer were calculated from prediction tables developed by Claus et al (12) .Predicted risks of BRCA1 mutation for the BCREP population were calculated from tables developed by Couch et al.(10). Because these risks had skewed distributions, Wilcoxon’s rank-sum test was used in confirmatory analyses. For the primary analysis, women were characterized by self-report as having decided not to undergo testing (declined testing group) or undergoing/having undergone testing (tested group). Women who were undecided about testing were excluded. Associations between each variable and the testing decision were examined using Wilcoxon’s rank-sum test for ordered variables (i.e., importance rated on a four-point scale) and the ordinary χ2 test for dichotomous variables (e.g., very important versus other). Multivariable analyses were conducted using multiple logistic regression. Because of correlations between concerns about health insurance, life insurance, and job discrimination and between the importance of ovarian cancer risk information and the importance of help deciding about prophylactic oophorectomy, composite variables were constructed to represent concern about discrimination from testing and interest in information about ovarian cancer risk. No other significant correlations were identified between variables associated with testing in this sample, including Ashkenazi background and presence of familial mutation. Each variable associated with testing in bivariate analysis at P ≤ 0.10 was tested for inclusion in the model. The final model included all of the variables whose inclusion altered the odds ratio for another variable by ≥10%. Because of concern that women might perceive the factors that influenced their decisions differently over time and according to their test results, we tested interaction terms for calendar time since counseling and BRCA1/2 test results. To understand the factors that affected testing decisions among women who had an elevated risk of carrying a mutation, we repeated our analyses in the subgroup of women with a predicted probability of BRCA1 mutation of≥5%.
Results
Of the 490 women to whom surveys were mailed, 10 women had died,and 36 women had moved. A total of 353 women returned completed surveys for a response rate of 80%. Nonresponders did not differ from responders in age, predicted risk of breast cancer, or predicted risk of a BRCA1 mutation in the family (Ps >0.1). Eighteen women who were not considering undergoing BRCA1/2 testing at the time of their visit, 76 women who were seen before BRCA1/2 testing was offered to women without cancer outside of a research protocol, and 8 women who pursued testing through a research protocol were excluded from these analyses. Of the remaining 251 eligible women, 125 (50%) women had undergone BRCA1/2 testing or were undergoing testing, 86 (34%) had decided not to undergo testing, and 40 (16%) were undecided about testing (including 14 women who had a family member pursuing testing).
The characteristics of women who underwent testing and women who decided not to undergo testing are reported in Table 1. Women who underwent testing were older and more likely to be Ashkenazi Jewish, to have a diagnosis of breast cancer, and to have a known familial BRCA1 or BRCA2 mutation than women who declined testing. Women who underwent testing had a slightly higher risk of breast cancer and a substantially higher risk of carrying a BRCA1 mutation than women who declined testing.
Women who underwent testing were significantly more likely to rank several potential benefits of testing as very important in their decision (Table 2). These benefits included providing cancer risk information for family members, learning information about ovarian cancer risk, and obtaining help in deciding about prophylactic oophorectomy and prophylactic mastectomy. Conversely, concerns about life insurance and job discrimination were more likely to be considered very important by women who declined testing. After multivariable analyses, Ashkenazi background, known familial mutation, fear of insurance discrimination,importance of information for family members, and importance of information about ovarian cancer risk remained associated with use of testing (Table 3). No interaction was found between the effects of these factors and calendar time since counseling or BRCA1/2 test results(Ps > 0.2).
Among the subgroup of women (n = 206) with a predicted probability of a BRCA1 mutation of ≥5%, 60 (29%) women had declined testing, 116 (56%) women had chosen to undergo testing,and 30 (15%) women were undecided (including 11 women who had a family member pursuing testing). After multivariable adjustment, there were no substantial differences between the associations with testing decisions in this subgroup and those associations found in the entire sample(data not shown).
Discussion
This study suggests that approximately two-thirds of women considering BRCA1/2 testing at the time of their visit to a clinical cancer risk evaluation program decide to undergo testing after participating in counseling. Women who undergo testing are at higher risk of carrying a BRCA1 mutation, more likely to want information about ovarian cancer risk for themselves and about breast and ovarian cancer risk for family members, more likely to be Ashkenazi Jewish, more likely to have a known familial mutation, and less likely to be concerned about insurance or job discrimination. The association with risk of BRCA1 mutation is present whether measured by predicted probabilities, the presence of familial mutation,or the presence of risk factors, i.e., Ashkenazi Jewish heritage.
The associations between the risk of carrying a mutation, a known familial mutation, and gaining risk information for family members and decisions about BRCA1/2 testing are reassuring. Most experts agree that BRCA1/2 testing should be targeted to women who are most likely to gain useful information from testing (13, 14). Women at higher risk of carrying a mutation are more likely to be found to carry a mutation, more likely to gain useful information, and should be more likely to decide to get tested. Women with a familial mutation will also gain more information from a negative test, because the cause of their familial predisposition has been identified. Furthermore, because of the potential implications of genetic testing for family members, more information is gained from BRCA1/2 testing when the results are salient to other family members.
The relatively greater importance of ovarian cancer risk information is likely to be multifactorial. First, prophylactic oophorectomy may appeal to more women than prophylactic mastectomy, both because prophylactic mastectomy is a more extensive and potentially disfiguring procedure and because substantially more evidence exists supporting the efficacy of breast cancer surveillance than that of ovarian cancer surveillance (15, 16, 17). Second, for the majority of women concerned about their increased breast cancer risk at the time they seek BRCA1/2 counseling, finding a BRCA1/2mutation only confirms their belief in their increased risk. The information that testing may bring about ovarian cancer risk may seem like the bigger change. Third, BRCA1/2 testing was the only method available to assess individual ovarian cancer risk at the time of this study, whereas several models were available to predict breast cancer risk (18).
Although there is little evidence suggesting that insurance discrimination is occurring at present, the association between fear of insurance or job discrimination and decisions about BRCA1/2testing is disconcerting. Because genetic information cannot be taken back once received, many women are reluctant to pursue testing without assurance that discrimination could not occur in the future. This situation is particularly paradoxical if women who would have been found to carry a mutation and taken steps to lower their cancer risk decline testing because of fear of insurance discrimination. Information gained from BRCA1/2 testing that results in women choosing interventions that lower their risk of cancer is good for everyone concerned, including life and health insurers.
This study both extends and supports the findings of prior studies of decisions about BRCA1/2 testing. Prior studies using hypothetical scenarios generally found a majority of women reported interest in testing, and interest in testing was higher among women with a higher perceived risk of carrying a mutation, greater concerns about cancer risk, and more interest in getting information for family members (19, 20, 21, 22, 23). Conversely, studies of research family members found that <50% of participants requested their genetic test results; however, participants requesting results also rated the benefits of testing more highly, knew more about BRCA1 testing, and had more first-degree relatives with breast cancer (8, 9).
Because this study was conducted retrospectively, the decision about testing may have influenced the perceptions and reporting of the factors that were important in that decision. We cannot determine to what degree women may have adopted beliefs after they made their decision to support or justify their behavior (24). In addition, the factors that women felt were most important in their decision about BRCA1/2 testing may have changed over time. Establishing a single time when decisions are made about testing is difficult. In our sample, almost a fifth of women were still undecided about testing up to 2 years after counseling. The time point for this study was selected to minimize the number of women who were undecided about testing while maintaining reasonable proximity to the date of counseling. Although the cost of testing was not an important factor in our study, our sample was highly educated and thus likely to be relatively affluent. Cost may be an important barrier to testing in less affluent populations. Finally, the generalizability of these results to women currently participating in similar programs is unknown.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
K. A. is supported by American Cancer Society Clinical Research Training Grant CRTG9902301 and Department of the Army Breast Cancer Research Program Grant BC971623. B. W. is supported by the Breast Cancer Research Foundation and National Cancer Institute Grant CA57601.
The abbreviation used is: BCREP, Breast and Ovarian Cancer Risk Evaluation Program.
Subject characteristics
. | Overall (n = 211) . | Testing (n = 125) . | Declined testing (n = 86) . | Two-tailed P . |
---|---|---|---|---|
Mean age, yrs (range) | 44.6 (24–73) | 45.8 (24–73) | 42.7 (27–73) | 0.04 |
Caucasian (%) | 96.8 | 98.1 | 94.6 | 0.21 |
Ashkenazi (%) | 29.9 | 42.9 | 13.3 | 0.0005 |
College education (%) | 73.6 | 77.6 | 71.7 | 0.49 |
Employed (%) | 74.0 | 74.1 | 77.3 | 0.72 |
Breast cancer (%) | 30.9 | 36.5 | 22.7 | 0.04 |
Known familial mutation | 6.2 | 9.7 | 1.1 | 0.04 |
Predicted breast cancer risk, mean (SD)a | 0.24 (0.13) | 0.26 | 0.21 | 0.03 |
Predicted BRCA1 risk, mean (SD) | 0.18 (0.20) | 0.24 (0.23) | 0.10 (0.07) | <0.0005 |
. | Overall (n = 211) . | Testing (n = 125) . | Declined testing (n = 86) . | Two-tailed P . |
---|---|---|---|---|
Mean age, yrs (range) | 44.6 (24–73) | 45.8 (24–73) | 42.7 (27–73) | 0.04 |
Caucasian (%) | 96.8 | 98.1 | 94.6 | 0.21 |
Ashkenazi (%) | 29.9 | 42.9 | 13.3 | 0.0005 |
College education (%) | 73.6 | 77.6 | 71.7 | 0.49 |
Employed (%) | 74.0 | 74.1 | 77.3 | 0.72 |
Breast cancer (%) | 30.9 | 36.5 | 22.7 | 0.04 |
Known familial mutation | 6.2 | 9.7 | 1.1 | 0.04 |
Predicted breast cancer risk, mean (SD)a | 0.24 (0.13) | 0.26 | 0.21 | 0.03 |
Predicted BRCA1 risk, mean (SD) | 0.18 (0.20) | 0.24 (0.23) | 0.10 (0.07) | <0.0005 |
Among women without a breast cancer diagnosis.
Benefits, risks, and limitations of BRCA1/2 testing(reported as the percentage of subjects rating a factor very important)
Factors . | Testing (n = 125) . | Declined testing (n = 86) . | Two-tailed P . |
---|---|---|---|
Learning about my breast cancer risk | 76.3 | 73.8 | 0.69 |
Learning about my ovarian cancer risk | 76.1 | 57.5 | 0.005 |
Providing information for family members | 75.8 | 56.3 | 0.003 |
Help deciding about prophylactic mastectomy | 38.7 | 21.5 | 0.01 |
Help deciding about prophylactic oophorectomy | 59.1 | 29.5 | 0.0001 |
Help deciding about estrogen replacement | 29.9 | 28.8 | 0.89 |
Desire to be reassured if test was negative | 73.9 | 69.7 | 0.52 |
Concern about my anxiety if test was positive | 36.7 | 46.3 | 0.17 |
Fear of health insurance discrimination | 36.1 | 47.1 | 0.11 |
Fear of life insurance discrimination | 28.1 | 42.2 | 0.04 |
Fear of job discrimination | 12.4 | 27.7 | 0.006 |
Cost of the test | 22.3 | 22.9 | 0.74 |
My doctor’s recommendation | 39.3 | 32.1 | 0.30 |
My family’s recommendation | 30.7 | 30.0 | 0.96 |
Desire to help advance research | 46.3 | 40.0 | 0.37 |
Factors . | Testing (n = 125) . | Declined testing (n = 86) . | Two-tailed P . |
---|---|---|---|
Learning about my breast cancer risk | 76.3 | 73.8 | 0.69 |
Learning about my ovarian cancer risk | 76.1 | 57.5 | 0.005 |
Providing information for family members | 75.8 | 56.3 | 0.003 |
Help deciding about prophylactic mastectomy | 38.7 | 21.5 | 0.01 |
Help deciding about prophylactic oophorectomy | 59.1 | 29.5 | 0.0001 |
Help deciding about estrogen replacement | 29.9 | 28.8 | 0.89 |
Desire to be reassured if test was negative | 73.9 | 69.7 | 0.52 |
Concern about my anxiety if test was positive | 36.7 | 46.3 | 0.17 |
Fear of health insurance discrimination | 36.1 | 47.1 | 0.11 |
Fear of life insurance discrimination | 28.1 | 42.2 | 0.04 |
Fear of job discrimination | 12.4 | 27.7 | 0.006 |
Cost of the test | 22.3 | 22.9 | 0.74 |
My doctor’s recommendation | 39.3 | 32.1 | 0.30 |
My family’s recommendation | 30.7 | 30.0 | 0.96 |
Desire to help advance research | 46.3 | 40.0 | 0.37 |
Adjusted associations with undergoing testing (n = 169)
. | ORa . | 95% CIb . | Two-tailed P . |
---|---|---|---|
Familial mutation | 7.46 | 0.97–62.16 | 0.06 |
Ashkenazi background | 6.37 | 2.68–15.12 | 0.0005 |
Importance of | |||
Information for family members | 1.93 | 0.99–4.14 | 0.05 |
Information about ovarian cancer risk | 1.69 | 1.18–3.69 | 0.009 |
Fear of insurance discrimination | 0.45 | 0.21–0.94 | 0.03 |
. | ORa . | 95% CIb . | Two-tailed P . |
---|---|---|---|
Familial mutation | 7.46 | 0.97–62.16 | 0.06 |
Ashkenazi background | 6.37 | 2.68–15.12 | 0.0005 |
Importance of | |||
Information for family members | 1.93 | 0.99–4.14 | 0.05 |
Information about ovarian cancer risk | 1.69 | 1.18–3.69 | 0.009 |
Fear of insurance discrimination | 0.45 | 0.21–0.94 | 0.03 |
OR, odds ratio.
CI, confidence interval.