Evidence reviews for the US Preventive Services Task Force (USPSTF) use an analytic framework to visually display key questions addressed by the review to allow the USPSTF to evaluate the effectiveness and safety of a preventive service. The questions are depicted by linkages that relate interventions to outcomes. A dashed line indicates a health outcome that precedes subsequent outcomes. A circle that is half blue and half orange indicates a key question that pertains to both benefits and harms, such as key question 4. Refer to the USPSTF procedure manual for further details.12
KQ indicates key question.
aForty-six studies in 82 publications provided data; some addressed more than 1 KQ.
The area of the square representing the risk ratio is proportional to the number of events in each subgroup. CORE indicates Continuing Outcomes Relevant to Evista; HOT, Hormone replaced therapy Opposed by low-dose Tamoxifen; IBIS, International Breast Cancer Intervention Study; MAP.3, Mammary Prevention.3; MORE, Multiple Outcomes of Raloxifene; NSABP-1, National Surgical Adjuvant Breast and Bowel Project P-1; RUTH, Raloxifene Use for the Heart.
aVeronesi (2007) and Barrett-Connor (2006) reported mean or median duration of the treatment period.
bIncludes data from both MORE (4-year treatment) and CORE (4-year additional treatment); total follow-up time is averaged over both studies for 7705 participants.
cIntended treatment duration is 5 years or until a breast, neoplastic, cardiovascular, or toxicity event.
eMethods 1. Search Strategies
eMethods 2. Quality Assessment Criteria
eMethods 3. Additional Details of Statistical Analysis
eTable 1. Quality Assessment of Diagnostic Accuracy Studies (Key Question 1)
eTable 2. Quality Assessment of Randomized Controlled Trials (Key Questions 2-4)
eTable 3. Discriminatory Accuracy Studies of Risk Stratification Models (Key Question 1)
eFigure 1. Meta-analysis of Trials: Estrogen Receptor Positive Breast Cancer (Key Question 2)
eFigure 2. Meta-analysis of Trials: Estrogen Receptor Negative Breast Cancer (Key Question 2)
eFigure 3. Meta-analysis of Trials: Noninvasive Breast Cancer (Key Question 2)
eFigure 4. Meta-analysis of Trials: Breast Cancer Mortality (Key Question 2)
eFigure 5. Meta-analysis of Trials: All-cause Mortality (Key Question 2)
eFigure 6. Meta-analysis of Trials: Vertebral Fractures (Key Question 2)
eFigure 7. Meta-analysis of Trials: Nonvertebral Fractures (Key Question 2)
eFigure 8. Meta-analysis of Trials: Timing of Initiation (Key Question 2a)
eFigure 9. Meta-analysis of Trials: Venous Thromboembolism (Key Question 3)
eFigure 10. Meta-analysis of Trials: Deep Vein Thrombosis, Pulmonary Embolism, and Superficial Phlebitis (Key Question 3)
eFigure 11. Meta-analysis of Trials: Coronary Heart Disease (Key Question 3)
eFigure 12. Meta-analysis of Trials: Stroke (Key Question 3)
eFigure 13. Meta-analysis of Trials: Endometrial cancer (Key Question 3)
eFigure 14. Meta-analysis of Trials: Cataracts (Key Question 3)
eFigure 15. Meta-analysis of Trials: Menopausal Status (Key Question 4)
eFigure 16. Meta-analysis of Trials: Family History (Key Question 4)
eFigure 17. Meta-analysis of Trials: Body Mass Index (Key Question 4)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Nelson HD, Fu R, Zakher B, Pappas M, McDonagh M. Medication Use for the Risk Reduction of Primary Breast Cancer in Women: Updated Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA. 2019;322(9):868–886. doi:10.1001/jama.2019.5780
Medications to reduce risk of breast cancer are effective for women at increased risk but also cause adverse effects.
To update the 2013 US Preventive Services Task Force systematic review on medications to reduce risk of primary (first diagnosis) invasive breast cancer in women.
Cochrane Central Register of Controlled Trials and Database of Systematic Reviews, EMBASE, and MEDLINE (January 1, 2013, to February 1, 2019); manual review of reference lists.
Discriminatory accuracy studies of breast cancer risk assessment methods; randomized clinical trials of tamoxifen, raloxifene, and aromatase inhibitors for primary breast cancer prevention; studies of medication adverse effects.
Data Extraction and Synthesis
Investigators abstracted data on methods, participant characteristics, eligibility criteria, outcome ascertainment, and follow-up. Results of individual trials were combined by using a profile likelihood random-effects model.
Main Outcomes and Measures
Probability of breast cancer in individuals (area under the receiver operating characteristic curve [AUC]); incidence of breast cancer, fractures, thromboembolic events, coronary heart disease events, stroke, endometrial cancer, and cataracts; and mortality.
A total of 46 studies (82 articles [>5 million participants]) were included. Eighteen risk assessment methods in 25 studies reported low accuracy in predicting the probability of breast cancer in individuals (AUC, 0.55-0.65). In placebo-controlled trials, tamoxifen (risk ratio [RR], 0.69 [95% CI, 0.59-0.84]; 4 trials [n = 28 421]), raloxifene (RR, 0.44 [95% CI, 0.24-0.80]; 2 trials [n = 17 806]), and the aromatase inhibitors exemestane and anastrozole (RR, 0.45 [95% CI, 0.26-0.70]; 2 trials [n = 8424]) were associated with a lower incidence of invasive breast cancer. Risk for invasive breast cancer was higher for raloxifene than tamoxifen in 1 trial after long-term follow-up (RR, 1.24 [95% CI, 1.05-1.47]; n = 19 747). Raloxifene was associated with lower risk for vertebral fractures (RR, 0.61 [95% CI, 0.53-0.73]; 2 trials [n = 16 929]) and tamoxifen was associated with lower risk for nonvertebral fractures (RR, 0.66 [95% CI, 0.45-0.98]; 1 trial [n = 13 388]) compared with placebo. Tamoxifen and raloxifene were associated with increased thromboembolic events compared with placebo; tamoxifen was associated with more events than raloxifene. Tamoxifen was associated with higher risk of endometrial cancer and cataracts compared with placebo. Symptomatic effects (eg, vasomotor, musculoskeletal) varied by medication.
Conclusions and Relevance
Tamoxifen, raloxifene, and aromatase inhibitors were associated with lower risk of primary invasive breast cancer in women but also were associated with adverse effects that differed between medications. Risk stratification methods to identify patients with increased breast cancer risk demonstrated low accuracy.
Although periodic mammography screening is currently the main approach to early detection of primary breast cancer,1 medications to reduce risk of breast cancer provide an additional prevention option for women at increased risk.2 Clinical trials indicate that the selective estrogen receptor modulators raloxifene and tamoxifen3 and the aromatase inhibitors anastrozole4 and exemestene5,6 are associated with reduced incidence of primary invasive breast cancer. However, these medications are also associated with adverse effects,3 and candidates for risk-reducing medications need to be accurately identified to optimize potential benefits and minimize harms.
In 2013, the US Preventive Services Task Force (USPSTF) issued a B recommendation for clinicians to offer to prescribe risk-reducing medications for women at increased risk for breast cancer and low risk for adverse medication effects.2 However, use of medications for breast cancer risk reduction is low in clinical practice7-9 because of women’s concerns about adverse effects and beliefs that benefits are not worth the harms,3 difficulty in identifying candidates for therapy, and primary care physicians’ unfamiliarity with tamoxifen and aromatase inhibitors because they are predominantly used for breast cancer treatment.10 This review was conducted to update evidence on the efficacy and harms of risk-reducing medications and the accuracy of clinical risk assessment methods to select candidates for therapy to inform new USPSTF recommendations.
Detailed methods are available in the full evidence report at http://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/breast-cancer-medications-for-risk-reduction1.11Figure 1 shows the analytic framework, key questions (KQs), and contextual questions that guided the review. Contextual questions provide additional information for the USPSTF but are not systematically reviewed or represented in the analytic framework.
Cochrane Central Register of Controlled Trials and Database of Systematic Reviews, Ovid EMBASE, and MEDLINE (January 1, 2013, to February 1, 2019) were searched for relevant English-language articles (eMethods 1 in the Supplement), and reference lists were manually reviewed. Studies published before 2013 were identified from prior systematic reviews for the USPSTF.3,13
Investigators reviewed abstracts and full-text articles using prespecified eligibility criteria.11 A second reviewer independently confirmed results of the initial review; discrepancies were resolved by consensus.
Studies reporting discriminatory accuracy (ie, area under the receiver operating characteristic curve [AUC]) of risk assessment methods that could be used in primary care settings to identify women at higher than average risk for breast cancer were included for KQ1. For KQ2 and KQ4 (efficacy), only double-blind, placebo-controlled or head-to-head randomized clinical trials (RCTs) of tamoxifen, raloxifene, or aromatase inhibitors for primary prevention of breast cancer that enrolled women without preexisting breast cancer and reported breast cancer incidence as a primary or secondary outcome were included. For KQ3 and KQ4 (harms), RCTs and observational studies of these medications that had nonuser comparison groups or direct comparisons between the medications were included. All adverse outcomes at all reported follow-up times were considered in an effort to capture potential short- and long-term adverse effects. Studies reporting only intermediate outcomes rather than health outcomes, such as bone density rather than fractures, were not included.
Investigators abstracted data from studies of risk assessment methods (study design; population characteristics; eligibility criteria; reference standards; risk factors included in the models; and performance measures) and trials of medications (study design; setting; population characteristics; eligibility criteria; interventions [dose and duration]; numbers enrolled and lost to follow-up; method of outcome ascertainment; and results for each outcome). A second investigator reviewed accuracy of abstracted data.
Two investigators independently applied criteria developed by the USPSTF12 to rate the quality of each study as good, fair, or poor (eMethods 2 in the Supplement). Discrepancies were resolved through a consensus process. Individual study quality ratings are provided in eTables 1 and 2 in the Supplement.
For all KQs, the overall quality of evidence was rated good, fair, or poor, based on study quality, consistency of results, precision of estimates, study limitations, risk of reporting bias, and applicability and was summarized in a table.12
Results of placebo-controlled trials were combined for each medication separately using meta-analysis that considered clinical and methodological differences. Estimates of risk ratios (RRs [rate ratio, hazard ratio, or relative risk]) and their standard errors were abstracted or calculated from each study and used as effect measures (additional information in eMethods 3 in the Supplement). Statistical heterogeneity was assessed using Cochrane χ2 tests and the magnitude of heterogeneity using the I2 statistic.14 The RRs were combined by using a profile likelihood random-effects model to account for variation among studies.15 All analyses were performed using Stata version 13.1 (StataCorp).
To determine outcomes for subgroups, analysis was performed by age (≤50 years; >50 years), family history of breast cancer (yes; no), use of menopausal hormone therapy (yes; no), menopausal status (pre; post), and body mass index (≤25; >25 [calculated as weight in kilograms divided by height in meters squared]), when at least 2 studies reported results. For outcomes of major adverse events, analyses were stratified by active vs posttreatment periods, although this was possible for tamoxifen only.
To interpret the clinical effect of medications, the numbers of events reduced for benefits or increased for harms compared with placebo per 1000 women, assuming 5 years of medication use, were estimated when the meta-analysis indicated a significant difference between treatment and placebo groups. Estimates used combined RRs from the meta-analyses and combined event rates from placebo groups of included trials. Combined event rates were determined from a meta-analysis of placebo event rates from each trial by using a random-effects Poisson model and raw data of the number of events and women-years of follow-up. This analysis was performed using PROC NLMIXED in SAS version 9.4 (SAS Institute Inc). The 95% CIs were estimated using a simulation method that assumed that the logs of both the RRs and the event rates have normal distributions and then drew 10 000 random samples from the distribution. The numbers of events reduced or increased were then estimated from each sample; 95% CIs were obtained by computing the 2.5% and 97.5% quantiles of the full sample.
A total of 46 studies4,16-60 (82 articles [>5 million participants]) met inclusion criteria for KQs (Figure 2), including 25 studies of discriminatory accuracy and 21 studies of the efficacy and harms of medications (20 RCTs; 1 observational study). In addition, 14 studies addressed the contextual question regarding clinician and patient attitudes and practices.7,61-73
Key Question 1. In adult women without preexisting breast cancer, what is the accuracy of risk assessment methods to identify women who could benefit from medications to reduce risk for primary breast cancer?
A total of 25 studies reporting results of evaluations of 18 risk assessment methods based on data from more than 5 million women met inclusion criteria (Table 1; eTable 3 in the Supplement).16-40 Although most methods shared common risk factors, they differed by including additional variables and using dissimilar reference populations. Three new studies expanded existing methods with new data by adding breast density to the Gail and Tyrer-Cuzick models21; modifying the Gail model for Asian Americans31; and adding benign breast disease to the Breast Cancer Surveillance Consortium (BCSC) model.37 A fourth new study developed models to predict estrogen receptor–positive and estrogen receptor–negative breast cancer.30 No studies evaluated the optimal age or frequency of risk assessment (KQ1a and KQ1b).
Studies reported AUC values from 0.55 to 0.65, indicating low accuracy in predicting incidence of breast cancer in individual women.16-39 Only 1 study reported AUC values above 0.70 for both the Gail-2 model (AUC, 0.74 [95% CI, 0.67-0.80]) and the Tyrer-Cuzick model (AUC, 0.76 [95% CI, 0.70-0.82]).17 However, this study was small and did not include a primary care population, limiting its clinical applicability. One study determined how well the BCSC-Tice model stratified women into low- vs high-risk groups based on the risk threshold used in the National Surgical Adjuvant Breast and Bowel Project (NSABP P-1) and the Study of Tamoxifen and Raloxifene (STAR) trials (≥1.66% 5-year breast cancer risk).36 Results indicated AUC values from 0.61 to 0.64.36
Key Question 2. In adult women without preexisting breast cancer, what is the effectiveness and comparative effectiveness of medications to reduce risk for primary breast cancer on improvement in short- and long-term health outcomes, including invasive breast cancer, noninvasive breast cancer (including ductal carcinoma in situ), breast cancer mortality, all-cause mortality, and other beneficial outcomes (such as reduced fractures caused by certain medications and improved quality of life)?
Ten RCTs (40 articles) provided results for KQ2, including 7 new publications. These included updated long-term results of the International Breast Cancer Intervention Study (IBIS-I) trial of tamoxifen,74 a placebo-controlled trial of low-dose tamoxifen,41 and placebo-controlled trials of anastrozole4,75,76 and exemestane.44,77
Trials include the STAR head-to-head trial of tamoxifen and raloxifene49,78,79; 5 placebo-controlled trials of tamoxifen, including IBIS-I,42,74,80 NSABP P-1,46,81,82 Royal Marsden Hospital Trial,47,83 Italian Tamoxifen Prevention Study,43,84-87 and the Hormone Replacement Therapy Opposed by Low-dose Tamoxifen (HOT) study41; 2 placebo-controlled trials of raloxifene, the Multiple Outcomes of Raloxifene Evaluation (MORE) with long-term follow-up in the Continuing Outcomes Relevant to Evista (CORE) study45,88-102 and the Raloxifene Use for the Heart (RUTH) trial48,103; and 2 placebo-controlled trials of aromatase inhibitors, the International Breast Cancer Intervention Study II (IBIS-II) of anastrozole4,75,76 and the Mammary Prevention.3 trial (MAP.3) of exemestane.44,77 The most recent placebo-controlled tamoxifen trial, HOT, using a lower dose than the other trials (5 mg/d vs 20 mg/d), did not indicate reduction in invasive breast cancer risk (RR, 0.83 [95% CI, 0.42-1.62]; n = 1884) and was not included in the meta-analyses of tamoxifen trials (see eFigures 1-8 in the Supplement for other results).41 Details of individual trials are provided in Table 2 and the full report.11 Trials met criteria for fair or good quality (eTable 2 in the Supplement).
Trials included large numbers of women, ranging in size from 188441 to 19 747,49 primarily from North America, Europe, and the United Kingdom. Most participants were white, and none of the trials provided outcomes specific to racial or ethnic groups. Participants ranged in age from 30s to 80s at baseline, and only the placebo-controlled tamoxifen trials enrolled premenopausal women. The Italian trial of tamoxifen exclusively enrolled women who had undergone prior hysterectomy, including some with oophorectomy.43 Participants used exogenous estrogen in the Italian (14% of women), Royal Marsden (15%-27%), IBIS-I (40%), and HOT (100%) tamoxifen trials. The raloxifene trials enrolled older women with osteoporosis45,88-102 or increased cardiovascular risk.45,88-102
Results of the meta-analysis for KQ2 are summarized in Table 3 and in eFigures 1-8 in the Supplement. In placebo-controlled trials, tamoxifen (RR, 0.69 [95% CI, 0.59-0.84]; 7 fewer cases per 1000 women over 5 years of use [95% CI, 4-12]; 4 trials [n = 28 421]), raloxifene (RR, 0.44 [95% CI, 0.24-0.80]; 9 fewer cases [95% CI, 3-15]; 2 trials [n = 17 806]), and the aromatase inhibitors exemestane and anastrozole (RR, 0.45 [95% CI, 0.26-0.70]; 16 fewer cases [95% CI, 8-24]; 2 trials [n = 8424]) (Figure 3) were associated with reduced invasive breast cancer. Risk for invasive breast cancer was higher for raloxifene than tamoxifen in the STAR head-to-head trial (RR, 1.24 [95% CI, 1.05-1.47]; n = 19 747) after long-term follow-up. All medications were associated with reduced estrogen receptor–positive but not estrogen receptor–negative invasive breast cancer. Tamoxifen was associated with reduced noninvasive cancer in the NSABP P-182 and IBIS-I74 trials but not in the meta-analysis of all 4 trials (RR, 0.72 [95% CI, 0.56-1.41]; 4 trials [n = 28 421]). Medications were not associated with reductions in breast cancer–specific and all-cause mortality.
In placebo-controlled trials, raloxifene (RR, 0.61 [95% CI, 0.53-0.73]; 2 trials [n = 16 929]) was associated with reduced vertebral fractures; tamoxifen was associated with reduced nonvertebral fractures in the NSABP P-1 trial (RR, 0.66 [95% CI, 0.45-0.98]; n = 13 388) and the aromatase inhibitors had no effect on fractures. Tamoxifen and raloxifene had similar effects on fracture incidence at multiple vertebral and nonvertebral sites in the STAR head-to-head trial.
Key Question 2a. Does the effectiveness of risk-reducing medications vary by timing of initiation or duration of use?
Eight trials reported similar breast cancer outcomes regardless of age, although age categories varied by trial. No studies specifically compared shorter vs longer regimens of medication use or initiation based on time since menopause. While most trials intended 5 years of medication use, mean exposure times varied across the trials from 3 to 5 years. However, trials of similar medications indicated general consistency in their associations with breast cancer risk reduction, despite exposure time.
Key Question 2b. Does the effectiveness of risk-reducing medications persist beyond discontinuation of use?
The IBIS-I80 and Royal Marsden83 trials provided results for invasive and estrogen receptor–positive breast cancer for both the active treatment (mean duration, 5 years) and the posttreatment (median follow-up, 13 and 16 years, respectively) periods. These results indicate continued associations with reduced risk after discontinuation of tamoxifen, providing point estimates of even larger reductions in invasive and estrogen receptor–positive breast cancer during the posttreatment period. For IBIS-I, the RR for invasive breast cancer was 0.74 (95% CI, 0.60-0.93) for the 0- to 10-year follow-up period and 0.70 (95% CI, 0.52-0.95) for the greater than 10 years follow-up period,4 although the difference between periods was not statistically significant.
Key Question 3. What are the harms of using medications to reduce risk for primary breast cancer?
A total of 22 studies (54 articles) met inclusion criteria, including updated long-term results of the IBIS-I trial of tamoxifen,74 a placebo-controlled trial of low-dose tamoxifen,41 and placebo-controlled trials of anastrozole4,75,76 and exemestane.44,77
For tamoxifen, information on adverse effects was confined to the 5 large placebo-controlled primary prevention trials41-43,46,47,74,80-84,86,87,105-108 and the STAR head-to-head trial.49,78,79,109 The HOT trial of low-dose tamoxifen indicated no statistically significant differences in outcomes compared with placebo and was not included in the meta-analyses of tamoxifen trials.41 For raloxifene, adverse effects were reported from the 2 large placebo-controlled trials, MORE/CORE and RUTH88-94,104; the STAR head-to-head trial49,78,79; 8 smaller trials (in 11 publications) evaluating either bone density, biochemical profiles, or fractures51-60,110; and 1 observational study.50 These additional studies contribute little to the evaluation of harms because they involve few women relative to the large primary prevention trials, although these results are generally consistent with those of the larger trials. Consequently, they were not included in the meta-analyses of raloxifene trials. For anastrozole and exemestane, information on adverse effects was based on the 2 large placebo-controlled primary prevention trials.4,44 Similar to tamoxifen, no other RCTs or observational studies evaluated adverse effects of aromatase inhibitors in women without breast cancer.
Results of the meta-analysis for KQ3 are summarized in Table 3 and in eFigures 9-14 in the Supplement. In placebo-controlled trials, tamoxifen (RR, 1.93 [95% CI, 1.33-2.68]; 4 trials [n = 28 421]) and raloxifene (RR, 1.56 [95% CI, 1.11-2.60]; 2 trials [n = 17 806]) were associated with increased thromboembolic events. Raloxifene was associated with fewer thromboembolic events than tamoxifen in the STAR head-to-head trial (RR, 0.75 [95% CI, 0.60-0.93]; n = 19 490). Tamoxifen, raloxifene, and aromatase inhibitors were not associated with increased coronary heart disease events or strokes.
In placebo-controlled trials, tamoxifen was associated with increased incidence of endometrial cancer (RR, 2.25 [95% CI, 1.17-4.41]; 3 trials [n = 15 421]). In the STAR head-to-head trial, raloxifene was associated with fewer cases of endometrial cancer (RR, 0.55 [95% CI, 0.36-0.83]; n = 19 490) and endometrial hyperplasia (RR, 0.19 [95% CI, 0.12-0.29]; n = 19 490) and with fewer hysterectomies (RR, 0.45 [95% CI, 0.37-0.54]; n = 19 490) than tamoxifen. Tamoxifen was associated with increased incidence of cataracts (RR, 1.22 [95% CI, 1.08-1.48]; 3 trials [n = 22 832]) and cataract surgery compared with placebo. Risks for thromboembolic events and endometrial cancer with tamoxifen were higher for older compared with younger women and returned to normal after discontinuation. All medications were associated with adverse effects, such as vasomotor or musculoskeletal symptoms, that varied by medication.
Key Question 3a. Do the harms of risk-reducing medications vary by timing of initiation or duration of use?
The NSABP P-1 placebo-controlled trial of tamoxifen reported point estimates consistent with higher risks for deep vein thrombosis, pulmonary embolus, and stroke for women 50 years and older than for women younger than 50 years, although results were not statistically significant.46 Results of the NSABP P-1 trial also indicated that the risk of thromboembolic events was elevated only during the first 3 years of tamoxifen use.111 Age older than 60 years was also an important risk factor for venous thrombosis in the Italian trial.87 The NSABP P-1 trial found that endometrial cancer was more common among women 50 years and older than among women younger than 50 years (RR, 4.01 [95% CI, 1.70-10.90]; n = 7998 for those ≥50 years vs RR, 1.21 [95% CI, 0.41-3.60]; n = 5177 for those <50 years).46 Initiation based on time since menopause was not reported.
Key Question 3b. Do the harms of risk-reducing medications persist beyond discontinuation of use?
Although tamoxifen was associated with increased thromboembolic events compared with placebo during the trials, risk returned to normal after discontinuation of tamoxifen in the 2 trials (IBIS-I80 and Royal Marsden83) that reported posttreatment data (RR, 0.98 [95% CI, 0.48-1.80]; 2 trials; n = 10 130).74 In the IBIS-I trial, risk for endometrial cancer was higher for tamoxifen compared with placebo during the first 5 years of follow-up (RR, 3.76 [95% CI, 1.20-15.56]) but declined after discontinuation (RR for 5- to 10-year follow-up, 0.64 [95% CI, 0.21-1.80]; RR for ≥10-year follow-up, 1.40 [95% CI, 0.38-5.61]).74
Key Question 4. Do the outcomes of using medications to reduce risk for primary breast cancer vary by population subgroups?
Studies included for KQ2 and KQ3 also provided results for KQ4. Medications were associated with lower risks for invasive breast cancer in all population subgroups evaluated based on menopausal status; family history of breast cancer; body mass index categories; modified Gail model risk categories; and age at menarche, parity, or age at first live birth (eFigures 15-17 in the Supplement). Tamoxifen and anastrozole were associated with reduced risk, regardless of history of previous breast lesions (lobular carcinoma in situ, atypical ductal hyperplasia, or atypical lobular hyperplasia) but demonstrated larger estimates of effect in women with previous lesions.
Factors associated with adherence and nonadherence in patient use of risk-reducing medications were examined in systematic reviews7,69,72 and additional observational studies.63,64,66-68,70,71,73 Factors associated with adherence included higher breast cancer risk; clinician recommendation; peers with good experiences using medications; belief that medications are effective; anxiety or worry about breast cancer; and history of an abnormal breast biopsy result. Factors associated with nonadherence included concern for adverse effects; estrogen contraindication; peers with poor experiences using medications; belief that medications are for treatment, not risk reduction; medication is a daily reminder of illness; preference for other risk-reducing approaches such as mastectomy; and knowledge of the benefits and harms of medication.
Prescribing risk-reducing medications is an uncommon practice among primary care physicians surveyed in 3 US studies.61,62,65 Factors associated with prescribing included more breast cancer diagnoses in clinical practice; belief that benefits outweigh harms; patients asking about medications; personal experience with breast cancer in self or a relative; and belief that eligibility for medications is easy to determine. Barriers to prescribing included lack of training, experience, or comfort with medications; belief that benefit may not be worth harms; belief that patients lack interest in medications; preference that specialists prescribe medications; lack of comfort or certainty with identifying women eligible for medications; and time constraints.
This evidence report reviewed trials of the efficacy and harms of medications to reduce the risk of primary invasive breast cancer and studies of the accuracy of clinical risk assessment methods to select patients for therapy. Table 4 summarizes the evidence included in this review. Although most results are consistent with the 2013 USPSTF review,3 this update provides additional evidence of the inaccuracy of risk assessment methods21,30,31,37; long-term follow-up of the IBIS-1 tamoxifen trial demonstrating persistent breast cancer risk reduction and normalization of endometrial cancer risk after discontinuation of tamoxifen4; and new trials of aromatase inhibitors.4,44,75-77 In addition, a placebo-controlled trial of low-dose tamoxifen indicated no reduction in risk of invasive breast cancer.41
Results of 4 recently published studies of breast cancer risk assessment methods indicated low discriminatory accuracy in predicting the probability of breast cancer in individual women,21,30,31,37 similar to previous studies. Most methods performed only slightly better than age alone as a risk predictor. Based on these studies, current practices of selecting women for risk-reducing medications according to a modified 5-year Gail score of 1.66% or higher, as used for inclusion criteria in primary prevention trials and US Food and Drug Administration approval of tamoxifen and raloxifene for risk reduction, are likely inaccurate. Most women 60 years and older without other risk factors would meet this threshold by age alone. Studies also provide no clinical guidance on optimal ages or frequencies for risk assessment because these components have not yet been evaluated.
Primary prevention trials of anastrozole4,75,76 and exemestane44,77 provide new evidence of the efficacy and harms of aromatase inhibitors for breast cancer risk reduction. However, no long-term follow-up data are available to determine whether harms demonstrated in treatment trials of women with noninvasive and early stage breast cancer, such as fractures and cardiovascular events, apply to risk reduction. An RCT of 2980 women with locally excised estrogen receptor–positive ductal carcinoma in situ compared anastrozole (1 mg/d) with tamoxifen (20 mg/d) for 5 years, with median follow-up of 7.2 years.112 Results indicated increased risk of fractures (odds ratio, 1.36 [95% CI, 1.03-1.80]) and stroke (odds ratio, 3.36 [95% CI, 1.04-14.18]) with anastrozole and increased venous thromboembolic events with tamoxifen.112 A meta-analysis of individual-level data from 31 920 postmenopausal women with estrogen receptor–positive early breast cancer in treatment RCTs of aromatase inhibitors vs tamoxifen also indicated associations with increased fractures for aromatase inhibitors but no differences for venous thromboembolic events or stroke.113 Also, 7 RCTs that compared extended aromatase inhibitor treatment with treatment followed by placebo or no treatment showed associations with increased fractures and stroke for extended aromatase inhibitors and suggested increased cardiovascular events.114 Although these trials imply associations of aromatase inhibitors with increased risk for fractures and stroke, it is unclear how well the results of treatment trials translate to women without cancer, particularly in the absence of true placebo comparison groups. For example, it is not known whether the increase in fractures reflects the direct harm of aromatase inhibitors or the protective effect of tamoxifen.
Future research to determine optimal candidates for risk-reducing medications should focus on the women mostly likely to benefit. Applying research findings to clinical selection criteria would improve identification of candidates in practice settings and clinical decision making. For example, no new studies and no studies in the 2013 review evaluated risk-reducing medications specifically in carriers of pathogenic BRCA1/2 mutations. Mutation testing was not a common practice when most of the trials were conducted, and it is not known how many BRCA1/2 carriers were enrolled. The NSABP P-1 trial of tamoxifen described results for 288 mutation carriers who developed breast cancer during the trial.115 Of the 8 women with breast cancer who had BRCA1 mutations, 5 received tamoxifen and 3 placebo (RR, 1.67 [95% CI, 0.32-10.70]). Of 11 women with breast cancer and BRCA2 mutations, 3 received tamoxifen and 8 placebo (RR, 0.38 [95% CI, 0.06-1.56]). Also, 6 of 7 women (86%) with BRCA1 mutations had estrogen receptor–negative breast cancer and 6 of 9 (67%) with BRCA2 mutations had estrogen receptor–positive cancer. Tamoxifen is only effective in reducing risk for estrogen receptor–positive breast cancer.
This review had several limitations. First, there was potential publication bias as well as biases of the literature review process, such as including only English-language articles. Second, studies of risk assessment methods varied by size, study populations, reference groups, and methods. Third, RCTs were limited by clinical heterogeneity related to different eligibility criteria, exposure durations and follow-up, adherence, and ascertainment of outcomes. The trials were not designed for subgroup analysis and may have been underpowered to demonstrate treatment effects. Furthermore, no trials directly compared the effects of timing and duration of medication use. Fourth, research is lacking for optimal doses, duration of use, persistence of effects after treatment for most medications, and outcomes in women who are nonwhite, premenopausal, have comorbidities, or are taking additional medications for other indications.
Tamoxifen, raloxifene, and aromatase inhibitors were associated with lower risk of primary invasive breast cancer in women but also were associated with adverse effects that differed between medications. Risk stratification methods to identify patients with increased breast cancer risk demonstrated low accuracy.
Corresponding Author: Heidi D. Nelson, MD, MPH, MACP, FRCP, Pacific Northwest Evidence-based Practice Center, Oregon Health & Science University, 3181 SW Sam Jackson Park Rd, Mail Code BICC, Portland, OR 97239 (firstname.lastname@example.org).
Accepted for Publication: April 17, 2019.
Author Contributions: Dr Nelson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Nelson, Zakher, McDonagh.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Nelson, Fu.
Critical revision of the manuscript for important intellectual content: Nelson, Zakher, Pappas, McDonagh.
Statistical analysis: Nelson, Fu.
Obtained funding: Nelson.
Administrative, technical, or material support: Nelson, Pappas.
Conflict of Interest Disclosures: None reported.
Funding/Support: This research was funded under contract HHSA290201500009I, Task Order 7, from the Agency for Healthcare Research and Quality (AHRQ), US Department of Health and Human Services, under a contract to support the US Preventive Services Task Force (USPSTF).
Role of the Funder/Sponsor: Investigators worked with USPSTF members and AHRQ staff to develop the scope, analytic framework, and key questions for this review. AHRQ had no role in study selection, quality assessment, or synthesis. AHRQ staff provided project oversight, reviewed the report to ensure that the analysis met methodological standards, and distributed the draft for peer review. Otherwise, AHRQ had no role in the conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript findings. The opinions expressed in this document are those of the authors and do not reflect the official position of AHRQ or the US Department of Health and Human Services.
Additional Contributions: We gratefully acknowledge the following individuals for their contributions to this project: AHRQ medical officer Tina Fan, MD, MPH; and Pacific Northwest Evidence-based Practice Center expert consultant Rachel Yung, MD, research librarian Andrew Hamilton, MLS, MS, and research assistant Lucy Stillman, BS. We also acknowledge past and current USPSTF members who contributed to topic deliberations. USPSTF members, external reviewers, and federal partner reviewers did not receive financial compensation for their contributions.
Additional Information: A draft version of this evidence report underwent external peer review from 4 content experts (Therese Bartholomew Bevers, MD, University of Texas MD Anderson Cancer Center, Houston; Jack Cuzick, PhD, FRS, CBE, Wolfson Institute of Preventive Medicine, Queen Mary University, London; Sam G. Smith, MSc, PhD, Leeds Institute of Health Sciences, London; and Diana Petitti, MD, MPH, University of Arizona, Tucson) and from 4 federal partners at the Centers for Disease Control and Prevention and National Cancer Institute. Comments from reviewers were presented to the USPSTF during its deliberation of the evidence and were considered in preparing the final evidence review.
Editorial Disclaimer: This evidence report is presented as a document in support of the accompanying USPSTF Recommendation Statement. It did not undergo additional peer review after submission to JAMA.
Create a personal account or sign in to: