To the best of the authors' knowledge, few population-based studies to date have examined the use of BRCA1/2 testing or patterns of physician recommendations for genetic testing among women diagnosed with breast cancer. The objective of the current study was to evaluate the rates and predictors of physician recommendation for BRCA1/2 testing among patients with breast cancer.
Women aged 18 years to 64 years who were diagnosed with invasive breast cancer in 2007 were identified from the Pennsylvania Cancer Registry and mailed a survey regarding their family history of cancer, physician treatment recommendations, and BRCA1/2 testing. Of the 4009 women who were sent surveys, 2258 responded (56%). Based on age at diagnosis and family history, women were categorized as being at high, moderate, or low risk of BRCA1/2 mutations.
Nearly 25% of the participants were classified as being at high risk of carrying a BRCA1/2 mutation based on their age at the time of breast cancer diagnosis and family history of breast and/or ovarian cancer. Physician recommendations for BRCA1/2 testing were found to be strongly associated with risk of carrying a mutation, with 53% of high-risk women reporting a testing recommendation compared with 9% of low-risk women. In addition, physician recommendations were strongly correlated with the use of testing in all risk groups. Among high-risk women, the lack of a recommendation for BRCA1/2 testing was more common among older, low-income, and employed women.
Testing for mutations in BRCA1 and BRCA2 can be useful among women diagnosed with breast cancer. Being found to carry a mutation may have treatment implications, such as eligibility for contralateral mastectomy, prophylactic oophorectomy, and experimental therapeutic agents such as poly(ADP-ribose) polymerase (PARP) inhibitors. In addition, BRCA1/2 testing can provide information for familial risk assessment. If a woman with cancer is found to have a mutation, her relatives may undergo testing for that mutation and make cancer risk reduction decisions based on those test results.[1-3] The risk of carrying a BRCA1/2 mutation is approximately 5% to 10% among women diagnosed with breast cancer, and the mutation risk is higher for women with early-onset disease or a family history of breast and/or ovarian cancer.[4, 5] Clinical guidelines suggest that patients with breast cancer should receive personalized risk assessment and consider genetic counseling and testing if they have early-onset disease (diagnosed at age ≤ 45 years), bilateral breast cancer, triple-negative disease (estrogen receptor-negative [ER-]/progesterone receptor-negative [PR-]/human epidermal growth factor receptor 2-negative [HER2-]), Ashkenazi Jewish ancestry, a strong family history of breast and/or ovarian cancer, or a combination of these characteristics.
Currently, relatively little is known about the use of BRCA1/2 testing among women with a breast cancer diagnosis. Although patients with breast cancer are more likely to undergo genetic testing than women without breast cancer, studies have suggested that the rate of BRCA1/2 testing among patients with breast cancer is relatively low.[7-13] Two surveys of convenience samples of breast cancer survivors found that < 15% reported undergoing testing, with higher rates of testing reported among women with a family history of breast cancer, a younger age at diagnosis, or Jewish ancestry.[14, 15]
Similar to breast cancer treatment decisions, the use of BRCA1/2 testing among patients with breast cancer is likely influenced by both patient preferences and the recommendations of their health care providers.[15-18] Although there has been a growing interest in the use of BRCA1/2 testing at the time of diagnosis among surgeons and oncologists, to the best of our knowledge the degree to which these providers are recommending this testing, particularly to patients at high risk of carrying a mutation, is currently unknown. Thus, we conducted a retrospective cohort study of patients with breast cancer residing in the state of Pennsylvania to examine provider recommendations for genetic testing, receipt of BRCA1/2 testing, and factors associated with not receiving a recommendation for testing among women at high risk of carrying a mutation.
MATERIALS AND METHODS
Study Design and Participants
Study participants were identified through the Pennsylvania Cancer Registry (PCR), which has achieved North American Association of Central Cancer Registries (NAACCR) Gold certification for the accuracy and completeness of its data. The Institutional Review Boards of the University of Pennsylvania and the PCR approved the study protocol. Women diagnosed with invasive breast cancer between ages 18 years and 64 years in Pennsylvania between January 1, 2007 and December 31, 2007 (N = 4920) were mailed an introductory letter explaining the study, followed by a second mailing with a consent form, study questionnaire, prepaid return envelope, and an unconditional incentive of $5. Nonrespondents were sent 2 additional mailings. Women were excluded if they were deceased (N = 252), had invalid addresses (N = 645), or were otherwise ineligible (reported not having cancer and/or not able to read/speak English [N = 14]). Of the 4009 women eligible for the study with valid addresses, 2258 returned the questionnaire (56%).
The study questionnaire elicited sociodemographic characteristics, detailed family history of breast and ovarian cancer, and tumor characteristics. Women were asked to list their treatment recommendations, including whether BRACAnalysis® (Myriad Genetics, Salt Lake City, UT) or BRCA1/2 testing was recommended. Women were asked about recommendations for genetic testing rather than genetic counseling, based on prior work and pilot studies indicating that women reported referral for counseling as referral for testing and were confused when asked about recommendations for genetic counseling. Women also were asked if they had undergone BRCA1/2 testing, and the approximate date of the test. Because of privacy concerns given the mailed questionnaire, the results of genetic testing were not ascertained. Participants' responses were linked to tumor characteristics from the PCR.
Women were asked whether their provider recommended BRCA1/2 testing; possible responses included “no,” “yes,” or “don't know.” Approximately 10% of respondents indicated “don't know,” and 37% of respondents left this item unanswered. We combined those who responded “no,” “don't know,” and nonresponses into a “no recommendation” group for the main analysis. Sensitivity analyses limiting the study population to those who answered “yes” or “no” to testing recommendations were performed. Women were categorized into 3 groups based on contact with a medical oncologist and treatment: 1) those who saw an oncologist and received chemotherapy; 2) those who saw an oncologist and did not receive chemotherapy; and 3) those who did not see an oncologist and did not receive chemotherapy. Based on cancer registry data, women were categorized as having positive, negative, or unknown ER or PR status. Because collection of HER2/neu status was not required by the PCR until 2010, self-reported HER2/neu status was used in this analysis.
BRCA1/2 Mutation Risk Categories
Established guidelines and other data regarding mutation prevalence[4, 6, 19] were used to categorize patients into 3 levels of risk of BRCA1/2 mutations based on their age at the time of diagnosis and family history (Fig. 1). We were conservative in our definition of high risk so as to capture women who would have been clear candidates for BRCA1/2 testing in 2007, as well as anticipated to have insurance coverage for genetic testing and therefore should have received a recommendation from their provider. High-risk women were those diagnosed with breast cancer at age ≤ 40 years, women with Ashkenazi Jewish heritage, or women diagnosed at age ≤ 50 years who met 1 of the following criteria: first-degree or second-degree female relative diagnosed with breast cancer at age ≤ 50 years, first-degree or second-degree relative diagnosed with ovarian cancer, first-degree or second-degree male relative with breast cancer, or 2 relatives on the same side of the family diagnosed with breast or ovarian cancer. Moderate-risk women were those aged 41 years to 49 years who did not meet the high-risk criteria, women diagnosed at age ≥ 50 years with a family history of breast or ovarian cancer, and women aged < 60 years with triple-negative disease. Low-risk women were those diagnosed at age ≥ 50 years who did not have triple-negative disease with no family history, or women aged 60 years to 64 years with no family history regardless of tumor biology.
Patient characteristics, rates of testing recommendations, and rates of testing were compared across risk groups using Student t tests and chi-square tests. Agreement between recommendations and BRCA1/2 testing and between physician recommendations was assessed using kappa statistics. Differences in the percentage of women with recommendations for BRCA1/2 testing who reported undergoing testing were compared using chi-square tests. Multivariable logistic regression was performed to estimate the odds of having a provider recommendation for BRCA1/2 testing by various patient and tumor characteristics. In addition, we estimated the odds of lacking a test recommendation among high-risk women by patient and tumor characteristics. All statistical tests were 2-sided with α of .05.
Of the 4009 eligible women with valid addresses, 2258 returned the questionnaire (56%). The mean age of respondents (52.1 years) was the same as that of the full Pennsylvania registry population of women diagnosed with breast cancer before age 65 years (Table 1). Respondents were slightly more likely to be white, have ER-positive/PR-positive disease, and an earlier stage of disease at the time of diagnosis than the full registry population.
Table 1. Comparison of Characteristics of Women Aged 18 to 64 Years Diagnosed With Invasive Breast Cancer From the Pennsylvania Cancer Registry in 2007 and Survey Respondents
2007 Registry Population (N=4920)
Abbreviations: ER, estrogen receptor; PR, progesterone receptor; SD, standard deviation.
Demographic and tumor characteristics and report of BRCA1/2 testing recommendations for the 2258 respondents are listed in Table 2. Approximately 90% of participants reported having health insurance, 95% were identified as having a medical oncologist, and 97% identified their surgeon (data not shown). The majority of women were diagnosed with localized disease (65%) and received chemotherapy (61%). Approximately 90% of respondents were white, > 60% reported postsecondary education, and 38% reported an annual household income > $70,000.
Table 2. Characteristics of Survey Respondents Aged 18 to 65 Years Diagnosed With Invasive Breast Cancer in Pennsylvania in 2007, Stratified by Risk of BRCA1/2 Mutation (N=2258)
In total, 26% of patients reported a recommendation for genetic testing, and a similar percentage reported undergoing BRCA1/2 testing. We validated self-report of BRCA1/2 testing by reviewing medical records of the 55 respondents who received breast cancer treatment at the University of Pennsylvania, 40 of whom reported undergoing BRCA1/2 testing and 15 of whom reported no testing. BRCA1/2 testing was confirmed in 37 of 40 women who reported testing (positive predictive value of 93%). Evidence of BRCA1/2 testing was found in 1 of the 15 women who reported not being tested (negative predictive value of 93%).
Of the 2258 respondents, 24% met our definition of being at high risk of carrying the BRCA1/2 mutation (N = 546) (Table 2), 38% were classified as being at moderate risk (N = 860), and 38% were considered to be at low risk (N = 852). We chose a conservative high-risk definition to identify patients with breast cancer who should have been clear candidates for genetic testing based on 2007 guidelines. In addition to the criteria for being at high risk listed in Figure 1, current National Comprehensive Cancer Network criteria for BRCA1/2 testing include all women diagnosed at age ≤ 45 years, women aged ≤ 60 years with triple-negative disease, and women with ≥ 2 affected relatives in the same lineage regardless of age at diagnosis. When these 3 additional guidelines were applied, 43% of the study population met these testing guidelines.
Women in the high-risk group were younger (by definition), had a higher educational level and household income, and were more likely to see an oncologist and receive chemotherapy than women in the moderate-risk or low-risk groups (P < .001). High-risk women were less likely to have ER-positive/PR-positive disease and more likely to have HER2-neu-positive disease than women in the low-risk group (P < .001). Approximately 53% of high-risk women, 25% of moderate-risk women, and 9% of low-risk women reported receiving a recommendation for BRCA1/2 testing from a provider. Among high-risk women who received a recommendation, 30% reported a testing recommendation from a medical oncologist only, 22% reported a testing recommendation from a surgeon only, and 49% reported a testing recommendation from both providers. There was fair agreement between medical oncologist and surgeon recommendations for BRCA1/2 testing among high-risk women (72% agreement; kappa, 0.43), and the characteristics of patients referred by each type of provider were similar (data not shown).
Provider recommendations and undergoing BRCA1/2 testing were found to be strongly correlated (91% agreement, kappa, 0.77). The percentage of women who underwent BRCA1/2 testing by testing recommendation is shown in Table 3. A high percentage of women who reported a provider recommendation underwent testing (83%), particularly among the high-risk group (89%), and few women without a provider recommendation reporting receiving BRCA1/2 testing (6%), regardless of risk group (3%-12%). Very few women reported undergoing BRCA1/2 testing before 2007 (< 1%) (data not shown), and therefore the majority of BRCA1/2 tests occurred after the diagnosis of breast cancer.
Table 3. Percentage of Women With Breast Cancer Undergoing BRCA1/2 Testing by Provider Recommendation (N=2258)
aP value was derived from chi-square test of percentage reporting BRCA1/2 test by risk groups.
After adjusting for age at diagnosis, tumor characteristics, demographics, and socioeconomic factors (Table 4), receiving a recommendation for BRCA1/2 testing was found to be strongly associated with risk category; women at high risk had 6 times the odds (odds ratio [OR], 5.84; 95% confidence interval [95% CI], 4.03-8.46 [P < .001]) and women at moderate risk had 3 times the odds (OR, 2.98; 95% CI, 2.19-4.04 [P < .001]) of a testing recommendation compared with low-risk women. Receiving a BRCA1/2 testing recommendation was inversely associated with age at diagnosis (P < .001) and was less common among women who did not see an oncologist or receive chemotherapy (OR, 0.48; 95% CI, 0.25-0.91 [P = .018]). Education and annual household income were found to be strongly associated with a testing recommendation, with an OR of 1.75 (95% CI, 1.21-2.54; P = .007) for an annual income of > $70,000 compared with < $30,000 and an OR of 1.54 (95% CI, 1.19-1.99; P = 0.001) for any college versus high school or less.
Table 4. Predictors of Recommendation for BRCA1/2 Testing (N=2258)a
aLogistic regression analysis adjusted for all factors in the model, stage of disease at diagnosis, and marital status.
Age at diagnosis, y
Medical oncologist care
Saw oncologist, received chemotherapy
Saw oncologist, did notreceive chemotherapy
Did not see oncologist, did not receive chemotherapy
College (2-4 y)
Annual household income
Among high-risk women, we examined the risk factors for not receiving a recommendation for a BRCA1/2 test from a provider (Table 5). Age at diagnosis was found to be the strongest predictor of not receiving a recommendation for testing. Compared with women aged ≤ 40 years, women who were aged 41 years to 50 years at the time of diagnosis had greater than twice the odds of having no testing recommendation (OR, 2.20; 95% CI, 1.47-3.30 [P < .001]), and women aged 51 years to 64 years had > 5 times the odds of having no testing recommendation (OR, 5.55; 95% CI, 2.40-12.8 [P < .001]). Low-income women (those with an annual income < $30,000) had greater than twice the odds of lacking a testing recommendation as women in the highest income group (OR, 2.36; 95% CI, 1.24-4.51 [P = .009]). In addition, women who were employed were more likely to lack a testing recommendation compared with unemployed women (OR, 1.57; 95% CI, 1.04-2.36 [P = .031]). We repeated the logistic regression analyses among women who definitively answered “yes” or “no” as to whether their physician had recommended BRCA1/2 testing, excluding those who answered “don't know” or did not respond, and the results were similar to those of the full analyses.
Table 5. Predictors of Lack of Recommendation for BRCA1/2 Testing Among High-Risk Women (N=546)
Logistic regression, odds of lack of recommendation for BRCA1/2 testing, adjusted for all factors in the table, stage of disease at diagnosis, and marital status.
Age at diagnosis, y
ER or PR positive
Medical oncologist care
Saw an oncologist and received chemotherapy
Saw an oncologist, did not receive chemotherapy
Did not see an oncologist, did not receive chemotherapy
College (2-y or 4-y)
Annual household income
To our knowledge, this is the first population-based study to examine the rates of physician recommendation for BRCA1/2 testing in patients with breast cancer. The results of the current study confirm that physician recommendations are a critical determinant of the use of genetic testing among all risk groups; few patients received genetic testing without first receiving a physician recommendation. Testing recommendations among patients with breast cancer appear to be largely and appropriately driven by patient risk factors for carrying a mutation (ie, risk group), with 53% of high-risk women reporting a physician recommendation for genetic testing compared with 9% of women at low risk of carrying a mutation. Nearly 25% of patients with breast cancer diagnosed at age < 65 years met our conservative definition of being at high risk of carrying a mutation, and > 40% met current National Comprehensive Cancer Network guidelines. Although testing guidelines vary somewhat across professional organizations and change over time, patients categorized as high risk in the current study should have been clear candidates for referral to genetic counseling and testing at the time of breast cancer diagnosis. The current study results suggest that a significant percentage of women at high risk of carrying a BRCA1/2 mutation may not receive a testing recommendation from their provider. In addition to risk group, factors found to be independently associated with a BRCA1/2 testing recommendation included age at diagnosis, ER/PR status, and higher socioeconomic status.
Among high-risk women, risk factors for failure to receive a testing recommendation included relatively late age at diagnosis, lower household income, and employment outside of the home. Although it is appropriate that early age at the time of diagnosis was used to target genetic testing recommendations among cancer patients, these results suggest that some strong candidates for genetic testing may be more likely to be overlooked by providers if they are older. This underscores the importance of a detailed family history assessment for patients with breast cancer regardless of age at diagnosis. The strong association noted between household income and testing recommendation among high-risk women was surprising given that the vast majority of the population in the current study reported having health insurance, and most health insurers cover BRCA1/2 testing in high-risk populations. However, coverage criteria and copays may continue to represent substantial barriers to testing among patients with financial concerns or inadequate insurance coverage, or may be perceived as creating such barriers by providers. In addition, women in the high-risk group and those who were employed were more likely to lack a recommendation for BRCA1/2 testing. This could be due to concerns about discrimination or loss of insurance coverage given that women in the study were diagnosed with cancer before passage of the Genetic Information Nondiscrimination Act (GINA) in 2008, or because of time constraints limiting the women's ability to participate in pretest genetic counseling.
Providers appear to be successful at limiting testing among women at low risk of carrying a mutation. The small number of women in the low-risk group who received a recommendation for testing (9%) limited our ability to identify the determinants of testing recommendations in that group. However, in studies that included women with and without breast cancer, physician recommendation for testing was associated with family history, age at diagnosis, education, and income, which have been previously identified as predictors of both receiving a BRCA1/2 test recommendation and uptake of genetic testing.[9, 11, 12, 14, 23] Minority race was not found to be significantly associated with a BRCA1/2 testing recommendation in this sample; however, the small number of minority women limited the power to identify an association.
The main strength of the current study is the population-based design with recruitment from the PCR, which provided a large sample of women with cancer diagnosed before age 65 years. Many studies of genetic testing have relied on convenience samples or patients referred to genetic testing clinics, and such sampling techniques are subject to bias.[7, 11, 23-25] Recruiting from the cancer registry provides a well-defined cohort and allows us to better assess the generalizability of the current study results. In addition, the current study elicited detailed information regarding family history, allowing us to categorize an individual's mutation risk level relative to National Comprehensive Cancer Network guidelines. We chose a conservative definition of high risk of carrying a mutation to focus on those women who should have been clear candidates for referral for genetic testing in 2007.
Several limitations of the data should be considered. First, we relied on self-report of physician recommendations and BRCA1/2 testing. We found good agreement between self-reported BRCA1/2 testing and medical records in the small subset of women treated at the study institution; however, we cannot validate physician recommendations for testing. Patients may not have remembered a recommendation for genetic testing, and therefore our rates of physician recommendation may be underestimated. Underreporting of physician recommendations is also of concern, particularly in the high-risk group, because it suggests a lack of effective communication between patients and providers. Recommendations for genetic testing should ideally be accompanied by genetic counseling. It is likely that women who participated in genetic counseling and who underwent genetic testing would be more likely to recall a physician's recommendation for BRCA1/2 testing, and this may partly explain the high concordance between physician recommendation and BRCA1/2 testing rates. The response rate to the mailed questionnaire was modest, and minority women and women with a later stage disease were slightly underrepresented, limiting our ability to evaluate genetic testing use in these groups. Finally, the current study sample was drawn only from Pennsylvania and may not be generalizable to other areas of the country.
These results from a population-based study of patients with breast cancer diagnosed before age 65 years suggest that a significant number of women who meet established testing criteria may not receive recommendations for BRCA1/2 testing. Women at high risk of mutations who were older at the time of diagnosis, had lower incomes, or were employed were less likely to report a physician recommendation for genetic testing. Interventions aimed at standardizing the use of familial risk assessment, increasing physician referrals for genetic counseling and testing for high-risk women, and facilitating provider-patient communication regarding genetic testing may be warranted to help identify those patients with breast cancer who are most likely to benefit from BRCA1/2 testing.
Supported by a National Institutes of Health grant from the National Cancer Institute (5-R01-CA133004-3).
CONFLICT OF INTEREST DISCLOSURES
Ms. Bristol owns stock in Johnson and Johnson. Dr. Domchek has received funding from AbbVie and Astra Zeneca for poly(ADP-ribose) polymerase (PARP) inhibitor trials related to BRCA1/2.