Results

Heidi D Nelson; Laurie Hoyt Huffman; Rongwei Fu; Emily L Harris; Miranda Walker; Christina Bougatsos

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Nelson HD, Huffman LH, Fu R, et al. Genetic Risk Assessment and BRCA Mutation Testing for Breast and Ovarian Cancer Susceptibility [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2005 Sep. (Evidence Syntheses, No. 37.)

See 2013 update

This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Genetic Risk Assessment and BRCA Mutation Testing for Breast and Ovarian Cancer Susceptibility

Genetic Risk Assessment and BRCA Mutation Testing for Breast and Ovarian Cancer Susceptibility [Internet].

Show details

Contents

< Prev Next >

3Results

Key Question 1. Does risk assessment and BRCA mutation testing lead to a reduction in the incidence of breast and ovarian cancer and cause-specific and/or all cause mortality?

Although several studies describe risk assessment methods that are relevant to primary care, none demonstrate that a screening approach enlisting risk assessment in a primary care setting followed by BRCA mutation testing and preventive interventions for appropriate candidates ultimately leads to a reduction in the incidence of breast and ovarian cancer and cause-specific and/or all cause mortality.

Key Question 2. What are the ethical, legal, and social implications of genetic screening for breast and ovarian cancer susceptibility?

A total of 229 studies of ethical, legal, and social implications (ELSI) of genetic screening were identified by the literature searches, reference lists, and experts. Studies pertinent to specific key questions are described in appropriate sections of this report (Key Questions 3b, 4, and 6).

Key Question 3a. How well does risk assessment for cancer susceptibility by a clinician in a primary care setting select candidates for BRCA mutation testing?

Determination of Family History

Family history of breast and ovarian cancer is the most important factor for determining risk for a deleterious BRCA mutation in a woman without cancer or known family mutation. For women with first-degree relatives with cancer, the relative risks for cancer have been estimated in meta-analyses as 2.1 (2.0–2.2) for breast cancer⁸ and 3.1 (2.6–3.7) for ovarian cancer.⁹⁰ Decisions about referral, testing, and prevention interventions are often based on self-reports of family histories that include types of cancers, relationships within the family, and ages of onset. Appropriate decisions rely on family histories that are accurately reported by women and correctly obtained by clinicians.

The accuracy of family cancer history information was addressed in a systematic review of studies of validated self-reported family histories.⁹¹ One study determined the sensitivity and specificity of a family history of breast or ovarian cancer in first-degree relatives reported by healthy individuals.⁹² A report of breast cancer in a first-degree relative had a sensitivity of 82%, specificity of 91%, positive likelihood ratio of 8.9 (5.4–15.0), and a negative likelihood ratio of 0.20 (0.08–0.49).⁹² A report of ovarian cancer in a first-degree relative was less reliable, and had a sensitivity of 50%, specificity of 99%, positive likelihood ratio of 34.0 (5.7–202.0), and a negative likelihood ratio of 0.51 (0.13–2.10).⁹² Overall, accuracy was better in studies concerning first-degree rather than second-degree relatives.⁹¹

Tools to Assess Risk of BRCA Mutation

Although several tools to predict risk for deleterious BRCA mutations have been developed from data on previously tested women, no studies of their effectiveness in a screening population in a primary care setting are available.⁹³ Much of the data used to develop the models are from women with existing cancer. Models with potential clinical applications are described in Table 2.

Table 2

Tools to Assess Risk of BRCA Mutation.

Myriad Genetic Laboratories Models

Logistic regression models have been developed by the Myriad Genetic Laboratories,^34, ^35, ⁹⁴ a commercial laboratory in the U.S. providing DNA full sequence testing for BRCA mutations. One model predicts risk for BRCA1 mutation and is based on a population of women with either early-onset breast cancer or ovarian cancer, or with a family history of breast or ovarian cancer.³⁵ This model also takes into account bilateral breast cancer, age of diagnosis, and Ashkenazi Jewish ancestry, and is not dependent on affected relatives. A second Myriad model predicts risk for both BRCA1 and BRCA2 mutations and is based on a population of women with breast cancer under age 50 or ovarian cancer who have at least one first- or second-degree relative with early breast or ovarian cancer.^34, ⁹⁴ This model considers bilateral breast cancer, concurrent breast and ovarian cancer, and breast cancer under age 40.

Couch Model. The Couch Model is based on logistic regression of data from a population of women with breast cancer and a family history of breast and/or ovarian cancer, and predicts risk for BRCA1 mutation.³⁶ Mutations were originally determined by conformation sensitive gel electrophoresis (CSGE) rather than DNA full sequencing, potentially underestimating mutation prevalence. A refined model now includes both BRCA1 and BRCA2 mutations using DNA full sequencing.⁹⁵ In this model, the individual may or may not have breast or ovarian cancer, but the family must have more than two cases of breast cancer. Predictors include the number of women diagnosed with breast cancer under age 50, concurrent breast and ovarian cancer, ovarian cancer, male breast cancer, and Ashkenazi Jewish ancestry.

BRCA PRO. BRCAPRO is a Bayesian model providing estimates of risk for BRCA1 and BRCA2 mutations⁹⁶^–⁹⁸ that has been validated in populations of women with increased prevalence of specific mutations.^97, ⁹⁹ The performance of BRCAPRO was compared with evaluations by cancer risk counselors in 148 pedigrees with women affected by breast or ovarian cancer who had BRCA mutation analysis. Using a greater than 10% BRCA gene mutation probability threshold, the sensitivity for identifying mutation carriers was 94% for counselors and 92% for BRCAPRO, and specificity was 16% for counselors and 32% for BRCAPRO.⁹⁹ Studies are currently under way to evaluate BRCAPRO estimates compared with other models.¹⁰⁰ In BRCAPRO, the individual may or may not have breast or ovarian cancer, and it considers current age, age at diagnosis, bilateral breast cancer, concurrent breast and ovarian cancer, all first- and second-degree relatives with and without cancer, males with breast cancer, and Ashkenazi Jewish ancestry. It includes information on both affected and unaffected relatives. CancerGene is a user-friendly software program¹⁰¹ that provides prior probabilities from BRCAPRO, but is much easier to use. A new version of Cyrillic software (Cyrillic 3) also includes BRCAPRO.¹⁰²

Tyrer Model. This model integrates personal risk factors with a genetic analysis to provide a comprehensive risk estimate.¹⁰³ Personal risk factors include current age, age at menarche, parity, age at first childbirth, age at menopause, atypical hyperplasia, lobular carcinoma in situ, height, and body mass index (BMI). The individual may or may not have breast or ovarian cancer. Genetic analysis incorporates the high-risk, high-penetrance BRCA1 and BRCA2 germline mutations with the addition of a low-penetrance gene. This was created as a stand-in to account for the effect of all other unidentified genes. Methods include segregation analysis techniques based on Bayes' theorem. This model is accessible through a computer program that is not yet widely distributed, but available from the investigators.

Tools for Primary Care That Assess Risk and Guide Referral

The family history assessment tool (FHAT) was developed to assist clinicians in selecting patients for referral to genetic counseling.¹⁰⁴ The referral threshold was doubling of the general population lifetime risk for breast cancer or ovarian cancer (22%) as estimated by Claus¹⁰⁵ and BRCAPRO methods. With FHAT, points are assigned according to the number of relatives, third-degree relatives or closer, diagnosed with breast, ovarian, colon, or prostate cancer, and the relationship to the proband, age at diagnosis, and type and number of primary cancers. Patients with scores of 10 or more points warrant referral. Results of FHAT were compared with Claus and BRCAPRO estimates for 184 women with incident familial and non-familial breast cancer.¹⁰⁴ The sensitivity and specificity of FHAT for a clinically significant BRCA1 or BRCA2 mutation were 94% and 51%, respectively. This compares with sensitivity and specificity of 74% and 79% for a 20% threshold for having a clinically significant BRCA1 or BRCA2 mutation using BRCAPRO, and 74% and 54% using Claus.

The Manchester scoring system was developed in the U.K. to predict deleterious BRCA1 or BRCA2 mutations at the 10% likelihood level.¹⁰⁶ Points are assigned depending on type of cancer (breast, ovarian, pancreatic, or prostate), affected family members, and age at diagnosis and provide scores for BRCA1 and BRCA2 mutations separately. The scoring system was validated in three sample sets in other regions of the U.K. and compared with other existing models. The Manchester model (combined BRCA1 and BRCA2) had 87% sensitivity and 66% specificity, comparing well with other models tested, including BRCAPRO with 61% sensitivity and 44% specificity.

Risk Assessment in Genetics (RAGs) is a computer program designed to support assessment and management of family breast and ovarian cancer in primary care settings.¹⁰⁷ It generates pedigrees after information about the proband and relatives are entered, categorizes risks of breast and ovarian cancer, generates referral guidelines based on the Claus model, and suggests appropriate management. Scores from RAGs are based on family history of affected relatives and the age of the presenting patient. One of three risk levels is assigned: low (<10% risk of having a clinically significant BRCA1 or BRCA2 mutation), in which the patient is reassured and managed in primary care; moderate (10–25% risk), in which the patient is referred to a breast clinic; and high (>25% risk), in which the patient is referred to a clinical geneticist.¹⁰⁸ A study of a small random sample of general practitioners in the U.K. compared how well they managed 18 simulated cases using RAGs, Cyrillic, and pen and paper approaches.¹⁰⁹ RAGs resulted in significantly more appropriate management decisions and more accurate pedigrees, and was the preferred approach.¹⁰⁹ RAGs took 178 seconds (mean) to administer, which was longer than pen and paper but shorter than Cyrillic.¹⁰⁹

Referral Guidelines

Referral guidelines have been developed by health maintenance organizations (HMOs),¹¹⁰ professional organizations,^31, ³² cancer programs,^111, ¹¹² state and national health programs,¹¹³^–¹¹⁶ and investigators¹¹⁷ to assist primary care clinicians in identifying women at potentially increased risk for BRCA mutations (Table 3). Although specific items vary among the guidelines, most include questions about personal and family history of BRCA mutations, breast and ovarian cancer, age of diagnosis, bilateral breast cancer, and Ashkenazi Jewish heritage. Most guidelines are intended to lead to a referral for more extensive genetic evaluation and counseling, not directly to testing. There is currently no consensus or gold standard about the use of guidelines. The effectiveness of this approach has not been evaluated.

Table 3

Criteria for Referral for Breast and Ovarian Cancer Genetic Counseling and Testing*.

Key Question 3b. What are the benefits of genetic counseling prior to testing?

No studies describe cancer or mortality outcomes related to genetic counseling, although 10 randomized controlled trials reported psychological and behavioral outcomes (Appendix G). Of these, 4 met criteria for good quality^41, ¹¹⁸^–¹²⁰ and 6 for fair quality (Appendix H).³⁸^–^40, ⁴²^–⁴⁴ Trials examined the impact of genetic counseling on breast cancer worry, anxiety, depression, perception of cancer risk, and intent to participate in genetic testing. Trials were conducted in highly selected samples of women, and results may not generalize to a screening population.

The trial most applicable to primary care practice randomized women at risk for clinically significant BRCA mutations to two groups and compared the effectiveness of a computer-based decision aid with standard genetic counseling.¹²⁰ The decision aid could potentially be used in primary care settings. Although knowledge scores increased in both groups, the decision aid was more effective than standard genetic counseling for increasing knowledge of breast cancer and genetic testing among women at low risk (<10% chance of deleterious BRCA mutation) (p=0.03), but not among women at high risk (≥10% chance). Perception of risk and intention to test were significantly lower for low-risk women using either method. The numbers of women undergoing mutation testing 1 month or 6 months after the intervention did not differ by type of intervention. Standard genetic counseling was more effective than the decision aid at reducing anxiety, although anxiety scores were within normal ranges for both groups at baseline and after either intervention.

Psychological Benefits

Results of nine trials indicated either decreased measures of psychological distress or no effect after genetic counseling (Table 4). Five of seven trials showed decreased breast cancer worry after genetic counseling,^38, ⁴²^–^44, ¹¹⁹ and two showed no significant effect.^40, ¹¹⁸ Three studies reported decreased anxiety after genetic counseling,^38, ^118, ¹²⁰ and three reported no significant effect.^41, ^43, ¹¹⁹ One study reported decreased depression after genetic counseling,¹¹⁸ and four found no significant effect.^38, ^41, ^43, ¹¹⁹ These findings are consistent with a recent meta-analysis of 12 published studies on genetic counseling for breast cancer with randomized controlled trial or prospective study designs.¹²¹ Results indicated that genetic counseling led to significant decreases in generalized anxiety (average weighted effect; r = -0.17, p<0.01), although the reduction in psychological distress was not significant (r = -0.074, p<0.052).

Table 4

Randomized Controlled Trials of Genetic Counseling: Benefits, Adverse Effects, and Impact on Risk Perception.

Perception of Cancer Risk

Women often overestimate their risk of breast cancer and/or deleterious BRCA mutations.^43, ^89, ¹²² Most women responding to surveys, including those at average and moderate risk, report a strong desire for genetic testing^38, ¹²³ even though only those at high risk would potentially benefit.

Five trials reported increased accuracy of cancer risk perception among women who received genetic counseling,^38, ^40, ^44, ^119, ¹²⁰ implying that genetic counseling may improve the predictive value of testing by reducing testing in moderate- or average-risk individuals. One study showed less accurate risk perception after genetic counseling, ¹¹⁸ and one had mixed results.⁴¹ Three studies examining the intention to participate in genetic testing after counseling reported inconsistent results. One study indicated a decrease in intention,³⁹ another showed an increase in intention among African American, but not Caucasian women,⁴² and the third study showed decreased intention among low-risk but not high-risk women.¹²⁰

Key Question 3c. Among women with family histories predicting either an average, moderate, or high risk for a deleterious mutation, how well does BRCA mutation testing predict risk of breast and ovarian cancer?

Cancer risk in family history risk groups can be estimated by determining the prevalence of the mutation and its penetrance for breast and ovarian cancer for each risk group. A total of 38 studies of prevalence and penetrance were identified as relevant to this question (Appendixes I–L). These studies could not be rated for quality because their study designs are not addressed by the USPSTF criteria. Most studies used research laboratory techniques to detect clinically significant mutations that differ from the DNA sequencing available clinically. The prevalence of clinically significant mutations may be underestimated by one-third using these techniques.¹²⁴

Prevalence

General Population

No direct measures of the prevalence of clinically significant BRCA1 or BRCA2 mutations in the general, non-Jewish U.S. population have been published. Models estimate it to be about 1 in 300 to 500.²⁷^–³⁰ For BRCA1, one model estimates a 0.12% prevalence rate.²⁵ The prevalence among those with a strong family history of cancer is estimated to be 8.7% based on one report of clinical referral populations that considered both BRCA1 and BRCA2 mutations together.³³ Additional prevalence estimates for individuals from referral populations with various levels of family history range from 3.4% (no breast cancer diagnosed in relatives younger than age 50, no ovarian cancer) to 15.5% (breast cancer diagnosed in a relative younger than age 50 and ovarian cancer diagnosed at any age).⁴⁶ Based on these estimates, the prevalence of BRCA1 and BRCA2 mutations in women at average risk could be considered as up to 0.24%, moderate risk from 0.24% to 3.4%, and high risk as 8.7% and above (Table 5). In the absence of direct measures, it can be assumed that one-half of the mutations would be in BRCA1 and one-half in BRCA2.

Table 5

Results of Meta-Analysis of Prevalence Studies.

Ashkenazi Jewish Population

For the Ashkenazi Jewish population unselected by family history, five studies provided data about prevalence of BRCA1 mutations,^11, ^12, ¹²⁵^–¹²⁷ six for BRCA2 mutations, ^11, ^12, ¹²⁵^–¹²⁸ and four for mutations in the two genes combined.^11, ^12, ^125, ¹²⁷ Results of the meta-analysis indicate an estimated prevalence of founder mutations of 1.9% (95% CI, 1.3%–2.8%), including 0.8% (0.5%–1.3%) BRCA1 and 1.1% (0.9%–1.4%) BRCA2 (Table 5).

For Ashkenazi Jews with a family history of breast or ovarian cancer, two studies provided prevalence data about BRCA1 mutations,^11, ¹²⁹ two for BRCA2 mutations,^11, ¹²⁹ and three for mutations in the two genes combined.^11, ^33, ¹²⁹ Results of the meta-analysis indicated an estimated prevalence of founder mutations of 10.2% (4.2%–22.9%) including 6.4% (1.1%–29.1%) BRCA1 and 1.1% (0.6%–2.0%) BRCA2 (Table 5).

Penetrance

Penetrance is the probability of developing breast or ovarian cancer among women who have a clinically significant BRCA1 or BRCA2 mutation. Published reports of penetrance describe estimates of BRCA1 and BRCA2 mutations ranging from 35% to 84% for breast cancer and 10% to 50% for ovarian cancer, calculated to age 70 years, for non-Ashkenazi Jewish women or those unselected for ethnicity.^21, ^27, ^28, ¹³⁰^–¹³³ Among Ashkenazi Jewish women, penetrance estimates range from 26% to 81% for breast cancer and 10% to 46% for ovarian cancer.^11, ^125, ¹³⁴^–¹³⁸

Limitations and Biases of Studies

Breast and ovarian cancer risk estimates are higher for relatives of women with breast cancer diagnosed at younger ages,¹³⁰ and for women from families with a greater number of affected relatives.^21, ¹³¹ Penetrance estimates are highest when based on data from families selected for breast or ovarian cancer—the selection approach used for genetic linkage studies and for clinical referrals. In addition to family history of cancer, penetrance may be influenced by the mutation's location with the gene.^133, ¹³⁹ Most studies do not have sufficient data to assess such heterogeneity.

For many published studies, penetrance was estimated from families without the benefit of genetic testing of all family members.^21, ^27, ^28, ¹³⁰^–^135, ¹³⁷ Studies used genetic segregation analysis in which the probability of having a clinically significant BRCA1 and BRCA2 mutation is estimated for each relative of an individual who has an identified mutation. Penetrance is estimated from the occurrence of breast or ovarian cancer and the a priori mutation carrier probability for each relative. Such estimates are typically based on family members of women who have breast or ovarian cancer (probands). Even when unselected for family history of breast and/or ovarian cancer, estimates from this study design can result in biased estimates of penetrance because the probands, and thus their family members, are more likely to have other risk factors for breast cancer that may affect penetrance.¹⁴⁰

Many studies focus on women with existing breast and ovarian cancer, introducing bias, since breast or ovarian cancer survivors may have different mutation frequencies compared with women with newly diagnosed cancer. Also, mutations are underestimated by most research studies because they employ a 2-step process in testing. This involves an initial test to detect clinically significant mutations followed by direct DNA sequencing for positive specimens only, rather than complete DNA sequencing of all specimens.

Meta-analysis-General Population

The probabilities of having a mutation if breast or ovarian cancer is present were combined with mutation prevalence among women without cancer, and a range of estimates of breast and ovarian cancer risk in average-, moderate-, and high-risk groups to estimate penetrance in the general population. (Methods are described in Appendix D.)

Breast cancer penetrance. Nine studies provided data of the probability of a mutation if breast or ovarian cancer is present for women at average risk,^27, ^29, ^36, ^132, ¹⁴¹^–¹⁴⁵ five studies for moderate risk,^27, ^29, ¹⁴²^–¹⁴⁴ and six for high risk.^28, ^33, ^36, ^141, ^144, ¹⁴⁶ Breast cancer penetrance estimates to ages 40 and 75, respectively, for clinically significant BRCA1 and BRCA2 mutations were 8.5% (6.7%–10.6%) and 31.6% (20.4%–45.4%) in average-risk, 3.4% (2.0%–5.5%) and 19.0% (1.0%–32.6%) in moderate-risk, and 7.7% (6.5%–9.1%) and 59.1% (44.4%–72.3%) in high-risk groups (Table 6).

Table 6

Results of Meta-Analysis of Penetrance Studies of Breast Cancer.

Ovarian cancer penetrance. Three studies provided data of the probability of having a mutation if breast or ovarian cancer is present for women at average risk,^133, ^147, ¹⁴⁸ three for moderate risk,^133, ^144, ¹⁴⁷ and three for high risk.^33, ^141, ¹⁴⁹ Ovarian cancer penetrance estimates to ages 50 and 75 were 13.0% (9.6%–17.4%) and 19.3% (13.7%–26.4%) in average-risk, no data for age 50 and 18.6% (14.0%–24.3%) in moderate-risk, and 4.0% (3.1%–5.2%) and 15.6% (12.9%–18.9%) in high-risk groups (Table 7).

Table 7

Results of Meta-Analysis of Penetrance Studies of Ovarian Cancer.

These penetrance estimates are similar to those published for a combined analysis of 22 studies based on case series data from women unselected for cancer family history.¹³⁰ Breast and ovarian cancer risk estimates to age 70 years for women who have a BRCA1 mutation were 65% (44%–78%) and 39% (18%–54%), respectively; for BRCA2 mutation carriers, estimated breast and ovarian cancer risks were 45% (31%–56%) and 11% (2%–19%), respectively.

Meta-analysis-Ashkenazi Jewish Population

Breast cancer penetrance. Ten studies provided data of the probability of having a mutation if breast or ovarian cancer is present for Ashkenazi Jewish women without a family history,^86, ^125, ^134, ^137, ^138, ^144, ¹⁵⁰^–¹⁵³ and nine for those with a family history.^33, ^86, ^125, ^128, ^134, ¹⁵⁰^–¹⁵³ Among Ashkenazi Jewish women without a family history of breast or ovarian cancer, penetrance estimates to ages 40 and 75 were 5.0% (3.0%–8.3%) and 33.7% (24.1%–44.9%) (Table 6). For those with a family history, penetrance estimates to ages 40 and 75 were 4.9% (1.9%–12.0%) and 34.7% (17.6%–57.0%) (Table 6).

Ovarian cancer penetrance. Five studies provided data to determine ovarian cancer penetrance for women without a family history,^127, ^133, ^135, ^152, ¹⁵⁴ and two for those with a family history.^33, ¹³⁵ Among Ashkenazi Jewish women without a family history of breast or ovarian cancer, penetrance estimates to ages 50 and 75 were 7.5% (4.9%–11.3%) and 21.4% (14.9%–29.7%) (Table 7). For those with family history, penetrance estimates to ages 50 and 75 were 3.3% (1.3%–7.9%) and 18.1% (7.6%–37.3%) (Table 7). These penetrance estimates are consistent with those published in individual studies of Ashkenazi Jewish women.

Key Question 4. What are the adverse effects of risk assessment, counseling, and testing?

Adverse effects of risk assessment, including genetic counseling, and testing include the potential for false positive and false negative results at each step of the process leading to false reassurance or inappropriate interventions. No studies directly addressed these issues. However, several studies described potential emotional distress.

A total of 57 studies, including 10 randomized controlled trials and 47 observational studies, were identified as relevant. Of these, 40 studies using non-standardized measures were excluded from further analysis. Nine fair to good quality studies assessing emotional distress were included (Appendixes M and N), and results are summarized in Table 8.¹⁵⁵^–¹⁶³ Eight poor-quality studies were excluded because of high or differential loss to follow-up, attrition, contamination, failure to consider important outcomes, lack of adjustment for potential confounders, or poorly defined interventions.¹⁶⁴^–¹⁷¹

Table 8

Distress Due to Adverse Effects of Risk Assessment and Testing.

One randomized controlled trial¹⁵⁶ and eight observational trials with pre-post,¹⁶³ case series,¹⁵⁵ longitudinal,¹⁶⁰ prospective cohort,^157, ^159, ^161, ¹⁶² and non-comparative¹⁵⁸ designs assessed breast cancer risk assessment, genetic testing, or both and their subsequent impact on distress measured as breast cancer worry, anxiety, or depression. All studies included genetic counseling. Studies varied in the number of distress indicators reported. Follow-up periods also varied; the first follow-up was defined as immediate to 2 weeks in three studies,^155, ^156, ¹⁶² 4 weeks in four studies,^157, ^159, ^160, ¹⁶³ and 4 months in one study.¹⁶¹ Final follow-up was defined as 6 months for all studies but two.^158, ¹⁶¹

Overall, more studies showed decreased rather than increased distress indicators after risk assessment and testing (Table 8). However, generalizability is limited, and only two studies distinguished between mutation carriers and non-carriers.^159, ¹⁶¹

Breast Cancer Worry

Two studies reported decreased breast cancer worry at the first follow-up evaluation,^156, ¹⁵⁷ and one at both the first and final evaluations.¹⁵⁷ These studies included women from high-risk breast cancer families,¹⁵⁶ and a mixed group of women at average and high risk who tested negative for BRCA mutations.¹⁵⁷ One study of women from high-risk breast cancer families showed increased breast cancer worry at first follow-up but had no additional follow-up data.¹⁶⁰ Increased breast cancer worry for mutation carriers¹⁶¹ was seen at the final follow-up (12 months) in one study.¹⁶¹

Anxiety

One study reported decreased anxiety for mutation carriers at the final 12-month follow-up evaluation, and decreased anxiety at the first 4-month follow-up for non-mutation carriers.¹⁶¹ A study of women in the largest known kindred identified with a deleterious BRCA1 mutation showed increased anxiety 1 to 2 weeks after testing, especially in carrier women who were tested first in their families and whose tested siblings were non-carriers.¹⁶² In contrast, three other studies of women with a family history of breast cancer,^157, ^160, ¹⁶³ including women from high-risk families,¹⁶⁰ showed decreased anxiety at the first 1-month follow-up evaluation, and one showed continued decreased anxiety at 6-month evaluation¹⁵⁷

Depression

Three studies with depression outcomes showed mixed results. Members of extended hereditary breast or ovarian cancer families, 27 with deleterious BRCA1 mutations and 6 with deleterious BRCA2 mutations, reported an increase in depression at the first and final follow-up evaluations for those who had cancer-related stress symptoms and declined testing, and a decrease in depression among non-carriers who were tested.¹⁵⁹ Another study of women with family histories of breast or ovarian cancer showed decreased depression in non-carriers at the first 4-month follow-up.¹⁶¹ A study of women from high-risk families that did not distinguish impact on carriers vs. non-carriers found a decrease in depression after the first 1-month follow-up.¹⁶⁰

Differential Impact of Risk Assessment, Testing, or Both on Distress

Distress varied by whether studies evaluated risk assessment, genetic testing, or both. When risk assessment was evaluated in four studies, one showed an increase in breast cancer worry.¹⁶⁰ There were no increases in other distress measures,^156, ^158, ¹⁶³but decreases in breast cancer worry, ¹⁵⁶ anxiety,^160, ¹⁶³ and depression.¹⁶⁰

When genetic testing was evaluated in three studies,^155, ^159, ¹⁶² results indicated no increased breast cancer worry, but in one study results indicated increased anxiety at the first follow-up evaluation ¹⁶² A study evaluating carriers and non-carriers showed increased depression at the first and final follow-up evaluations for those with high cancer-related stress who declined testing, and decreased depression for non-carriers who were tested.¹⁵⁹

In the two studies including both risk assessment and genetic testing, results were mixed. One study showed increased breast cancer worry at both follow-up evaluations for mutation carriers,¹⁶¹ while the other, which evaluated only those who tested negative, showed decreased breast cancer worry.¹⁵⁷ The first study showed decreased anxiety at the first follow-up for non-carriers and at the final follow-up for mutation carriers. It also showed decreased depression at the first follow-up for non-carriers.¹⁶¹ The second study showed decreased anxiety at first and final follow-up, and did not assess depression.¹⁵⁷

Key Question 5. How well do interventions reduce the incidence and mortality of breast and ovarian cancer in women identified as high-risk by history, positive genetic test results, or both?

Intensive Screening

Breast Cancer

Intensive screening for breast cancer in BRCA mutation carriers is recommended by expert groups,¹⁷² and is based on favorable results of programs designed for women with familial risk (Table 9).^55, ¹⁷³^–¹⁸¹ However, there are no trials of the effectiveness of intensive screening for BRCA mutation carriers in reducing mortality. Recent descriptive studies report increased risks for interval cancers (those occurring between mammograms) in BRCA mutation carriers with and without prior cancer undergoing annual mammographic screening.^70, ^173, ^182, ¹⁸³ These data imply that yearly mammograms may miss highly proliferate cancers that are more common in BRCA mutation carriers.¹⁸⁴^–¹⁸⁶

Table 9

Intensive Screening Studies in Women With Familial Breast Cancer Risk*.

In one study, high-risk women, including 113 BRCA1 and 15 BRCA2 mutation carriers without prior breast cancer, were followed in an intensive screening program at a family cancer clinic in the Netherlands that included monthly breast self-examination, twice-yearly clinical breast examinations, yearly mammography with MRI for those with dense breast tissue and/or BRCA gene mutations, and ultrasonography and fine-needle biopsy when indicated.¹⁷³ Sensitivity of this approach for detecting breast cancer was 74% overall, but dropped to 56% for BRCA mutation carriers, and four of the nine breast cancer cases among mutation carriers were detected during the period between mammograms (44%).¹⁷³

Additional studies of BRCA mutation carriers, including both women with and without previous breast cancer diagnoses, enrolled in similar programs also report high proportions of interval cancers. Four of 13 mutation carriers undergoing intensive screening at the Columbia-Presbyterian Comprehensive Breast Center in New York had breast cancer detected at the time of their annual mammograms, and 6 women had interval malignancies that presented as palpable masses from 2 to 9 months (mean 5 months) after the last mammogram (60%).¹⁸² A prospective study of BRCA carriers undergoing either preventive surgery or intensive screening reported that 6 of 12 mutation carriers who developed breast cancer while undergoing intensive screening were interval cases (50%).¹⁸³

To improve detection of early breast cancer in BRCA mutation carriers, a comparison of four intensive screening modalities was conducted in 236 Canadian women with BRCA1 or BRCA2 mutations aged 25 to 65.⁵⁵ Women underwent one to three annual screening examinations including MRI, mammography, and ultrasound with clinical breast examinations provided every 6 months. MRI was more sensitive for detecting breast cancers (sensitivity 77%, specificity 95.4%) than mammography (sensitivity 36%, specificity 99.8%), ultrasound (sensitivity 33%, specificity 96%), or clinical breast examination alone (sensitivity 9%, specificity 99.3%). Use of MRI, ultrasound, and mammography together had a sensitivity of 95%. Only one interval cancer was reported, and 14% of women had a biopsy that proved to be benign. MRI has advantages over mammography for detecting lesions in denser breast tissue and BRCA1-related cancers that have morphologic features suggesting a more benign mammographic image.

Ovarian Cancer

Data are limited regarding benefits of intensive screening strategies for ovarian cancer in BRCA mutation carriers. One study using transvaginal ultrasound to screen 1,610 women with a family history of ovarian cancer found 3.8% abnormal scans, and only 3 of 61 women with abnormal scans had ovarian cancer.¹⁸⁷

Chemoprevention

Selective Estrogen Receptor Modulators (SERMs)

Four randomized placebo-controlled prevention trials of tamoxifen, three rated fair to good quality⁵⁹^–⁶¹ and one rated fair quality,⁶² and one good quality trial of raloxifene⁶⁴ with breast cancer incidence and mortality outcomes have been published (Appendixes O and P), and a trial comparing these agents is in progress.^65, ¹⁸⁸ None of the trials specifically evaluated chemoprevention for women with BRCA mutations, although a genomic analysis of women developing breast cancer in one tamoxifen trial has been published.¹⁸⁹ No trials of chemoprevention using SERMs for ovarian cancer have been published. All trials reported high loss to follow-up (60% to 96% at 60 months), and three trials reported more loss from treatment than placebo groups due to side effects.^59, ^61, ⁶⁴

Three tamoxifen trials had inclusion criteria based on assessment of risk for breast cancer, including the Royal Marsden Hospital Trial, International Breast Cancer Intervention Study (IBIS-I), and National Surgical Adjuvant Breast and Bowel Project Breast Cancer Prevention Trial P-1 (BCPT P-1) (described in Appendix O).⁵⁹^–⁶¹ Two other trials did not assess subjects for breast cancer risk, and women in these studies could have lower risks of breast cancer than the general population. The Italian Tamoxifen Prevention Study included women who had had hysterectomies for benign conditions, with nearly half reporting previous bilateral oophorectomies, potentially reducing their risks for breast cancer.⁶² The Multiple Outcomes of Raloxifene Evaluation (MORE) study was primarily a fracture prevention trial that evaluated breast cancer as a secondary outcome, and included postmenopausal women with osteoporosis.⁶⁴ Osteoporosis may be a marker for non-use of postmenopausal hormone therapy and low endogenous estrogen production lowering risk for breast cancer.^190, ¹⁹¹

All trials enrolled healthy women without previous breast cancer; measured incident breast cancer cases and deaths; were multicenter, double-blind, and placebo controlled; and used the same dose of tamoxifen (20 mg per day), except for MORE, which used raloxifene (60 or 120 mg per day). The smallest trial enrolled approximately 1,200 women in each arm of the study,⁶¹ and the largest enrolled over 6,500 in each arm.⁶⁰ Mean follow-up ranged from 40 months in MORE to 70 months in the Royal Marsden Hospital Trial. Estrogen use during the study varied from 40% of women in IBIS-I, 26% in the Royal Marsden Hospital Trial, 14% in the Italian Tamoxifen Prevention Study, to 10% or less in the BCPT P-1 and MORE. Data were not provided to distinguish what proportion of estrogen users were using progestin as well.

For the two largest trials, tamoxifen significantly reduced the overall risk for breast cancer (Table 10).^59, ⁶⁰ Tamoxifen reduced risk for users in all age groups, and reduced estrogen receptor-positive but not estrogen receptor-negative tumors.^59, ⁶⁰ The Royal Marsden Hospital Trial and Italian Tamoxifen Prevention Study reported nonstatistically significant reductions in risk.^61, ⁶² The MORE trial reported significant reductions in risk among raloxifene users at both the 60 and 120 mg per day doses for total cases as well as invasive and estrogen receptor-positive cases (Table 10).⁶⁴ These results persisted with an additional year of treatment and follow-up.¹⁹²

Table 10

Results of Chemoprevention Trials.

Combining all trials in a meta-analysis resulted in a relative risk for total breast cancer of 0.62 (0.46–0.83) (Figure 3). Results were similar when including only the three tamoxifen trials that used family history of breast cancer as inclusion criteria, and when including the four tamoxifen trials only (Figure 3). Few deaths from breast cancer were reported in all the trials, and there were no differences between treatment and placebo groups. The relative risk was further reduced for estrogen receptor-positive breast cancer (4 trials; 0.39; 0.20–0.79) (Figure 4).

Figure

Figure 3. Relative Risk (RR) of Breast Cancer in Chemoprevention Trials.

Figure

Figure 4. Relative Risk (RR) of Estrogen Receptor (ER) Positive Breast Cancer in Chemoprevention Trials.

Treatment effects of tamoxifen could vary depending on the type of mutation. BRCA mutation status was determined in some women who developed breast cancer in BCPT P-1. Genomic analysis of 288 women indicated¹⁸⁹ 6 of 7 cases with BRCA1 mutations were estrogen receptor-negative (86%), and 6 of 9 cases with BRCA2 mutations were estrogen receptor-positive (67%), consistent with known distributions.189 The point estimate for breast cancer for BRCA2, but not BRCA1, carriers using tamoxifen approximated that of the total population of tamoxifen users for estrogen receptor-positive tumors (RR 0.31; 0.22–0.45).¹⁸⁹

Oral Contraceptives

No randomized controlled trials of oral contraceptives to prevent breast or ovarian cancer have been published. Observational studies indicate associations between oral contraceptives and reduced ovarian cancer in the general population¹⁹³^–¹⁹⁵ as well as BRCA mutation carriers,¹⁹⁶ but increased breast cancer among women with family histories of breast cancer.¹⁹⁷

Breast cancer. A retrospective cohort study of families of breast cancer probands diagnosed between 1944 and 1952 at the University of Minnesota collected follow-up data on families 40 years later.¹⁹⁷ Use of oral contraceptives was associated with a significantly increased risk of breast cancer among sisters and daughters of the probands (RR 3.3; 1.6–6.7), but not among granddaughters and nieces or non-blood relatives.¹⁹⁷ Risk was highest for women using oral contraceptives prior to 1975, when higher dosages of estrogen and progestins were used. Small numbers of cases using oral contraceptives after 1975 and younger ages of granddaughters restrict these estimates.

A small study of women with breast cancer compared past oral contraceptive use of mutation carriers with non-carriers.¹⁹⁸ Results indicated that more mutation carriers than non-carriers used oral contraceptives for more than 48 months before a first full-term pregnancy (OR 7.8; 1.1–55.0).¹⁹⁸

Ovarian cancer. A case-control study of BRCA mutation carriers with ovarian cancer and their sisters without ovarian cancer (both mutation carriers and non-carriers) indicated reduced risk among those with any past use of oral contraceptives (OR 0.5; 0.3–0.8).¹⁹⁶ Risk decreased with increasing duration of use and was protective for carriers of either BRCA1 or BRCA2 mutations.¹⁹⁶ A population-based case-control study of ovarian cancer among BRCA1 and BRCA2 mutation carriers among Jewish women in Israel indicated that risk of ovarian cancer decreased with each birth, but not with increased duration of oral contraceptives.¹²⁷ A study of risk factors for ovarian cancer included BRCA1 mutation carriers and non-carriers with ovarian cancer identified through registries compared with matched controls identified randomly in the San Francisco Bay Area.¹⁹⁹ Results indicated associations between reduced risk for ovarian cancer and ever use of oral contraceptives, duration of oral contraceptive use, history of tubal ligation, and increasing parity. Risk reduction was similar between mutation carriers and non-carriers. Differences between results of studies may be due to discrepancies in populations, methods, and confounders, or chance and other factors.

Prophylactic Surgery

No randomized controlled trials of prophylactic surgery have been conducted, and investigators acknowledge that this approach would not be ethical. Cohort studies of prophylactic surgery present several methodologic limitations to consider when interpreting their results.²⁰⁰

Biases Leading to Overestimation of Effect

For subjects selecting surgery, the course of events leading to surgery progresses in a sequence that can be easily captured in cohort studies, i.e., women obtain test results, make a decision about surgery, and then undergo the procedure. This course of events is less clear for women in nonsurgical comparison groups, particularly if they are not enrolled prospectively. Many of these women underwent testing after receiving a diagnosis of cancer. Risk reduction from surgery would be overestimated because the comparison group would be weighted with women with cancer.

This bias may be even more pronounced in studies enrolling women who are related to each other.^70, ⁷² A woman with cancer who then undergoes testing may be selected for the nonsurgical comparison group. In the meantime, she may have influenced her sister without cancer to undergo testing and surgery and become part of the intervention group. Subjects in the comparison group should be free of cancer at the point of follow-up in order to establish a similar baseline risk for both surgical and nonsurgical groups.

Women who choose prophylactic mastectomy may be more likely to also choose prophylactic oophorectomy^70, ⁷⁴ and experience a cumulative reduction in risk for breast cancer that would be attributed to only the mastectomy. Prophylactic oophorectomy may act as both a confounder and an effect modifier for breast cancer.

Parity and young age at first birth are associated with decreased breast and ovarian cancer risk, and women who experienced childbirth at early ages and multiple times may have more benefit from prophylactic oophorectomy.⁷² Parous mutation carriers may be more likely to elect prophylactic oophorectomy and at younger ages than nonparous women. This could lead to an overestimate of effect. Similarly, other important confounders should also be considered, such as increased use of hormone therapy in women undergoing oophorectomy.^72, ⁷³

Biases Leading to Underestimation of Effect

Selection of comparison groups is problematic because even if subjects are well matched on type of mutation, age, and other factors, it is currently not possible to match unrelated subjects on expected penetrance. Penetrance varies widely, and members of families with more cases of cancer, and likely higher prevalence and higher risk than the comparison group, may be more likely to choose prophylactic surgery for themselves.

Prophylactic surgery may reveal clinically undetected tumors. Studies that include this event in the surgery group could underestimate the efficacy of surgery for incidence outcomes, and overestimate it for mortality outcomes. Excluding these tumors entirely, however, would bias survival outcomes because the surgery may have increased life expectancy.

The type of prophylactic procedure could also influence outcomes. Patients undergoing mastectomy at times when subcutaneous mastectomies were performed, rather than total mastectomies, may have higher subsequent breast cancer rates because more residual breast cancer tissue remained after surgery than women undergoing total mastectomy. Most of the women in a retrospective study at the Mayo Clinic had subcutaneous mastectomies.²⁰¹ Similarly, results from an oophorectomy could be less optimal than from a salpingo-oophorectomy or salpingo-oophorectomy with hysterectomy because of residual tissue at continued risk for cancer.

Bilateral Mastectomy

Four studies of prophylactic bilateral mastectomy in high-risk women have been published, including two retrospective cohort studies based on medical records at the Mayo Clinic,^201, ²⁰² a prospective cohort study of mutation carriers in the Netherlands,⁷⁰ and a study of mutation carriers with prospective and retrospective cohort data from multiple centers in North America and Europe⁷¹ (Appendix Q). Studies of mutation carriers ranged from 26 to 483 subjects and follow-up for 3 to 14 mean years postmastectomy. Study quality was fair for two studies,^70, ²⁰² and two studies had designs that did not fit USPSTF criteria (Appendix R).^71, ²⁰¹

Study results were consistent, indicating an 85% to 100% risk reduction for breast cancer, despite differences in study designs and comparison groups ranging from sisters,²⁰¹ matched controls,⁷¹ a surveillance group,⁷⁰ and penetrance models.²⁰²

Bilateral Oophorectomy

Four studies of prophylactic oophorectomy met inclusion criteria, including a retrospective study of families with members with breast and ovarian cancer,²⁰³ two retrospective cohort studies of mutation carriers undergoing oophorectomy compared with matched comparison groups in North America and Europe,^72, ⁷³ and a prospective cohort study of mutation carriers undergoing elective oophorectomy or surveillance⁷⁴ (Appendix Q). Average follow-up time in the retrospective studies was from 5 to 11 years and in the prospective study 2 years. Study quality was fair for the prospective study⁷⁴ and the retrospective study of family members,²⁰³ and two studies had designs that did not fit USPSTF criteria (Appendix R).^72, ⁷³

All studies reported reduced risks for ovarian and breast cancer with prophylactic oophorectomy, although numbers of cases were small and the confidence intervals for the only prospective study crossed 1.0 for both outcomes.⁷⁴ Overall, the risk reduction for ovarian cancer ranged from 85% to 100%, and for breast cancer from 53% to 68%. One study found that oophorectomy after the age of 50 years was not associated with substantial breast cancer risk reduction,⁷² consistent with other studies of oophorectomy in the general population.²⁰⁴^–²⁰⁷

Tubal Ligation

Tubal ligation has been associated with a decreased risk of invasive epithelial ovarian cancer in observational studies.^194, ^208, ²⁰⁹ A matched case-control study of mutation carriers with and without ovarian cancer indicated a reduced odds ratio among controls who underwent previous tubal ligation when adjusted for oral contraceptive use, parity, history of breast cancer, and ethnic group (OR 0.39; 0.22–0.70).²¹⁰ This protective effect was present only among BRCA1 mutation carriers, although the number of BRCA2 carriers was small in this study.