Adherence to Endocrine Therapy in Breast Cancer Adjuvant and Prevention Settings

Adherence to oral endocrine therapy for adjuvant breast cancer treatment and for breast cancer prevention are substantial problems for clinicians and healthcare systems (1–3). In the adjuvant setting, adherence to endocrine therapy now takes on greater importance given reports that adjuvant tamoxifen of greater than 5 years duration is associated with lower recurrence risk (4, 5).

Adherence is defined as a composite of compliance (how well physician's orders are followed) and persistence (how long an individual continues on prescribed therapy; ref. 6). Currently, no gold-standard method exists for adherence measurement. Adherence can be estimated from prescription and medical claims and pharmacy databases, medical record review, hospital databases, pill counts, patient self-reports, prospective studies, and, rarely, pharmacologic assessments of drug concentrations. Methodologic concerns were raised by a study of 242 patients in which correlations of only 0.2 to 0.4 were seen comparing endocrine therapy adherence estimates from self-report, physician rating, refill records, and anastrozole concentrations (7). Other studies found overestimated adherence of patients with breast cancer to tamoxifen based on prescription checks (8) or microelectronic monitoring (9). In this regard, in a report comparing nonadherence with adjuvant anastrozole using three separate databases in the same population, estimates of nonadherence varied from 32% to 50% (10); however, subjects in these databases had variable medical insurance coverage, which may partially explain adherence differences. Despite these concerns, consistent general conclusions have emerged from studies using various methods of adherence assessment.

Clinical adjuvant endocrine therapy trials, in which adherence is commonly closely monitored, did not suggest a major adherence problem. In the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-14 trial in patients with breast cancer receiving adjuvant tamoxifen or placebo, discontinuation rates were 23% in both groups at 60 months median follow-up (11). In the NSABP B-24 adjuvant intraductal breast cancer trial, 60-month discontinuation rates for placebo were 30% compared with 33% for tamoxifen (12).

The seminal reports by Partridge and colleagues (1, 2) brought attention to the issue of poor adjuvant tamoxifen adherence in clinical practices. In a retrospective analysis of prescription claims (Medicaid and Pharmaceutical Assistance to Aged and Disabled) databases from the years 1990 to 1996, adherence to adjuvant tamoxifen was 83% after 1 year, 68% after 2 years, 61% after 3 years, and only 50% after 4 years. Other studies also found that more than half of patients with breast cancer discontinue endocrine therapy before completion of a recommended 5-year treatment (13, 14). For aromatase inhibitors, the commonly experienced arthralgias raise particular adherence concerns (15–17). However, in several adjuvant clinical trials, adherence to aromatase inhibitors was closely comparable, or even superior, with tamoxifen. In the Arimidex, Tamoxifen, Alone and Combined (ATAC) trial after 5 years, 2.1% of the anastrozole-treated patients and 14.3% of the tamoxifen-treated patients had discontinued use due to adverse events (18). Similar adherence was seen in both treatment groups in adjuvant trials comparing the aromatase inhibitor, exemestane, with tamoxifen (14% discontinued therapy in both arms; ref. 19) and the aromatase inhibitor, letrozole, with placebo (only 10% discontinued therapy in both arms; ref. 20).

Recent systematic reviews on adherence and/or persistence to adjuvant endocrine therapy in clinical practice settings identified 29 reports. Adherence in tamoxifen users ranged from 41% to 88%, whereas adherence in aromatase inhibitor users ranged from 50% to 91% (21). These findings were extended by Huiart and colleagues (22) who conducted meta-regression analyses to provide summary estimates of nonpersistence in 17 trials. For tamoxifen, 5-year nonpersistence was 47.2% (95% confidence interval, CI, 41.1%–53.5%) compared with 31.0% (95% CI, 25.9%–37.5%) for aromatase inhibitors (Table 1).

The findings are somewhat mixed considering aromatase inhibitor adherence in clinical practices (23, 24); however, in the United Kingdom general practice database, the 1-year discontinuation rate for adjuvant aromatase inhibitor use was 5% compared with about 10% for tamoxifen in women >49 years old and 20% for tamoxifen in women <40 years old (25). Similarly, in the Disease Analyses database (IMS Health), among 16,865 patients with breast cancer, 3-year discontinuation rates were 52% for tamoxifen, 47% for anastrozole, and 44% for letrozole (26). A randomized adjuvant adherence trial found shorter time to treatment discontinuation for exemestane, compared with letrozole (HR, 1.5; 95% CI, 1.1–2.1; ref. 27).

In summary, a substantial problem about adherence and persistence to adjuvant endocrine therapy remains in clinical practice. Somewhat surprisingly, adherence to aromatase inhibitors has been similar or superior to adherence to tamoxifen in several settings.

Evidence that adherence to adjuvant endocrine therapy could influence clinical outcomes came from a series of randomized adjuvant breast cancer trials evaluating duration of tamoxifen use. As summarized in Early Breast Cancer Trialist Cooperative Group (EBCTCG) analyses, compared with no therapy/placebo, with tamoxifen for 1 year recurrence, reduction was 27%; for 2 years, reduction was 33%; and for 5 years, reduction was 47%; P trend < 0.00001 (28).

Adherence to adjuvant tamoxifen therapy and breast cancer outcome has been examined in several cohort studies. In a U.S. cohort of 1,837 older women with early-stage breast cancer, those who used tamoxifen less than 1 year had substantially higher breast cancer mortality than those who used the drug for 5 or more years (HR, 6.26; 95% CI, 3.10–12.64; ref. 29). Similar findings were reported from a Scottish cohort of 2,080 patients with early-stage breast cancer. In that study, tamoxifen adherence <80% was associated with increased mortality (HR, 1.100; 95% CI, 1.001–1.21; ref. 30).

In the managed care Kaiser Permanente Northern California population, among 8,769 patients with breast cancer, 2,761 (31%) discontinued therapy within 6 months of diagnosis (based on automated pharmacy records); of those who continued, 1,684 (28%) were nonadherent (possession ratios <80%; defined as days with index prescription supplies/total days of follow-up). The survival at 10 years was 80.7% and 73.6% for those who continued therapy compared with those who discontinued therapy, respectively (P < 0.001; ref. 31). Of those who continued therapy, survival was 81.7% in those adherent to therapy, compared with 73.6% in those nonadherent. In a similar study in Kaiser Permanente Southern California, although breast cancer recurrence was lowest in women with greater adherence (possession ratios >80%), the rates were not markedly different from women with less regular use (32). In a retrospective cohort study of 3,361 Scottish patients with breast cancer, low adherence of <80% to adjuvant tamoxifen in aromatase inhibitor was associated with poor survival (HR, 1.20; 95% CI, 1.03–1.40; P = 0.019; ref. 33).

In a prospective cohort of 417 localized patients with breast cancer in Sweden, nonadherence at 1 year was associated with increased early breast cancer events (HR, 2.97; 95% CI, 1.08–8.15; ref. 34). In a study with 857 low-income women with early breast cancer, more recurrences and cancer deaths were observed in women nonadherent to endocrine therapy, but the results were not statistically significant (35). Similarly, in 690 women participating in International Breast Cancer Study Group Trials 13–39 and 14–93, those with ≥4-year selective estrogen receptor modulator (SERM) use had longer disease-free survival compared with those with <4-year use (71% vs. 64%; HR, 1.31; 95% CI, 0.86–1.98; P, 0.20; ref. 36; Table 2). In a small study of 116 men with breast cancer, overall survival was greater in those adherent to tamoxifen adjuvant therapy (37).

Thus, the lack of adherence and persistence to prescribed endocrine adjuvant therapy represents a barrier to achieving favorable outcomes for patients with breast cancer. The magnitude of the benefit of being adherent to adjuvant endocrine therapy is comparable with that seen with the addition of adjuvant chemotherapy.

Emerging data suggest that a substantial proportion of women who qualify for adjuvant endocrine therapy are not receiving this intervention. In a population of 13,753 patients with early-stage hormone receptor–positive breast cancer in the managed care Kaiser Permanente Northern California group, studied within 1 year of diagnosis, 30% of women did not initiate endocrine adjuvant therapy defined as having <2 prescriptions for tamoxifen or aromatase inhibitor filled within the first year after the cancer diagnosis (38). In the Kaiser Permanente Southern California population of breast cancer survivors, nearly 24% (3,237 of 13,412) of patients with estrogen receptor–positive disease did not use endocrine therapy (or had discontinued treatment within 6 months) despite having pharmacy coverage (32). This finding stimulated that organization to implement a medication adherence tool in the electronic medical records to potentially improve adherence. In the Women's Health Initiative cohort, in 3,588 patients with hormone receptor–positive, early-stage invasive breast cancer evaluated within 5 years of diagnosis by survey questionnaire, whereas adjuvant endocrine therapy use was reported by 83%, 17% reported no use. In their response, women cited “lack of physician recommendation” as the most common reason for nonuse. (39). Finally, 743 patients, identified from Surveillance, Epidemiology, and End Results (SEER) registries, eligible for adjuvant endocrine therapy were surveyed 4 years after diagnosis, surprisingly 10.8% never initiated therapy, and 15.1% started therapy but discontinued before 4 years (40; Table 3). Although detailed information on the characteristics of those not initiating adjuvant endocrine therapy is not currently available, further exploration of this issue is warranted.

Data are sparse about the communication between oncologists and patients with breast cancer on their therapeutic plan, as it is difficult to conduct linguistic communication studies. However, one study videotaped the initial breast cancer adjuvant therapy discussion in a series of 28 early-stage patients and found the issue of adherence to be poorly addressed. Much of the discussions on endocrine therapy focused on side effects and trial findings rather than on the importance of adherence (41).

Finally, a recent study found substantial discordance in adherence to adjuvant endocrine therapy when comparing results among prescription refill information, patient self-report, and oncologists' estimates. The oncologists estimated their patients' adherence at over 94%, which was less than estimated by telephone questionnaire self-report (P = 0.003), or by the pharmacy database in which only 67% of women >65 years old were identified as having drug available (P = 0.0001; ref. 42).

Interest in adherence to long-term adjuvant endocrine therapy regimen was enhanced by the recent report from the worldwide Adjuvant Tamoxifen Longer Against Shorter (ATLAS) adjuvant trial in which continued tamoxifen use for longer than 5 years reduced breast cancer recurrence (P = 0.002) and overall mortality (P = 0.01; ref. 4). Based on self-report, 5-year adherence was an excellent 84% for continued tamoxifen users. In contrast, the Investigation on the Duration of Extended Adjuvant Letrozole treatment (IDEAL) trial entered 1,250 patients with early breast cancer comparing 2.5 years with 5 years of extended letrozole use after 5 years of adjuvant endocrine therapy found overall nonadherence was 18.4% at 2.5 years (43). It is not clear whether this apparent difference between long-term continued tamoxifen and continued aromatase inhibitor use represents real differences in tolerability, or is the result of the limited data on the aromatase inhibitors currently available. In any event, more information is needed about persistence to long-term aromatase inhibitor adjuvant use.

Available evidence suggests that adherence to endocrine therapy in primary breast cancer prevention trial participants is similar to that seen in the adjuvant setting. In the NSABP P-1 prevention trial, discontinuation rates after 54.6 months mean follow-up were 23.7% on tamoxifen, and 19.7% on placebo (11). Discontinuation rates were somewhat higher in the International Breast Intervention Study-1(IBIS-1) in which, in a primary prevention setting, the 50-month median follow-up discontinuation rate for tamoxifen was 36% compared with 26% for placebo (44). In the longer intervention duration Royal Marsden Hospital trial comparing tamoxifen with placebo, therapy was prematurely discontinued at a median of 70 months in 46% of tamoxifen and 36% of placebo participants, respectively (45). In the NSABP Study of Tamoxifen and Raloxifene (STAR) prevention trial, 5-year adherence was 70.8% for tamoxifen and 73.9% for raloxifene (P < 0.001; ref. 46).

The aromatase inhibitor, exemestane, has been compared with placebo for primary breast cancer prevention in the Mammary Prevention (MAP).3 trial. After median 35 months follow-up, a 65%, statistically significant, relative reduction in invasive breast cancer incidence was seen for exemestane (47). During the study, exemestane was discontinued because of “intolerable side effects” by 15.4% of participants but, surprisingly, 10.8% of placebo participants discontinued study pills for the same reason. With only a net 5.3% difference, a major influence of factors other than drug side effects likely influenced the adherence results seen. A similar result was seen in the MA.17 adjuvant trial, in which about 20% of patients with breast cancer in the placebo group reported climactic symptoms (48). These results point to the importance of placebo controls to generate the most reliable tolerability information.

In a prevention study, adherence was related to outcome in the Women's Health Initiative (WHI) trial of estrogen alone. In this study, when 10,739 postmenopausal women with prior hysterectomy were randomized to conjugated equine estrogen alone or placebo, surprisingly, a statistically significant, lower breast cancer incidence was seen in the estrogen-alone group in intent-to-treat analyses (HR, 0.77; 95% CI, 0.62–0.95; ref. 49). However, in sensitivity analyses, censoring participants with less than 80% adherence to the pill-taking regimen, an even stronger association between estrogen-alone use and lower breast cancer incidence was seen (HR, 0.68; 95% CI, 0.49–0.95).

Factors predictive of tamoxifen chemoprevention nonadherence were examined in the P-1 breast cancer prevention trial. Current smokers and heavy alcohol users had lower tamoxifen adherence, whereas obesity and lower physical activity were unrelated to adherence (50). Similar findings were seen in 100 participants in the IBIS-1 study where women with smoking history also were less likely to persist with their randomized drug (51). In addition, in the IBIS-1 trial, use of additional prescribed medication was an important factor in predicting successful completion of therapy (P = 0.04; ref. 51). The latter findings suggest that women already using other prescription medications may represent a potentially favorable population, and thus, be more likely to accept and adhere to endocrine chemoprevention regimens. Lack of influence of obesity and low physical activity on adherence suggests that factors other than an unhealthy lifestyle are related to medication discontinuation.

Currently, use of the two drugs approved for chemoprevention in the United States (tamoxifen and raloxifene) continues to be low (52), and, for this reason, information on adherence in clinical practice settings is not available. However, a review of a clinical experience from the Partners HealthCare System identified 2,938 women with breast lesions with atypia. Women who received no chemoprevention had 10-year breast cancer incidence of 21.3% compared with 7.5% (P < 0.001) in women who did receive chemoprevention (53).

Factors associated with nonadherence to adjuvant hormonal therapy include lack of physician recommendation (32), patient perception of low risk for recurrence (54), adverse effects of therapy (55, 56, 57), age extremes: older age (23, 58) and younger age (23, 59), medication costs (60, 61, 62), low social economic status (63), suboptimal patient–physician communication (64), higher comorbidity (23, 59, 62), cigarette smoking (50, 51), and lack of social support (65; Table 4). Similar factors were associated with adherence in a low-income population in California (66). Findings about adherence by race/ethnicity have produced mixed results (23, 38, 67).

Many oncologists likely consider endocrine therapy side effects to be a major factor influencing therapy adherence. However, the available evidence identifies a less-straightforward relationship. In patients with breast cancer in the Commonly used Medications and Breast Cancer outcomes (COMBO) study, among 538 participants, 18.2% discontinued use before completing 5 years of therapy, whereas 25% discontinued after <1-year use (68). As in several prior reports, women who discontinued therapy were more likely to have been tamoxifen (43.9%) compared with aromatase inhibitor users (22.4%). Of interest, the only adverse effect significantly associated with discontinuation of both aromatase inhibitor and tamoxifen was headaches, an adverse event not commonly associated with these therapies. Such findings suggest that, although control of adverse effects is an important clinical consideration, adverse effects of endocrine therapy use may not play a major role in determining adherence and persistence to adjuvant endocrine therapy.

Factors adversely influencing adherence, perhaps in unexpected ways, are anxiety and depression. Following a breast cancer diagnosis, anxiety and depression decrease from about 50% in year one to about 15% in year five (69), a reciprocal to endocrine therapy adherence over the same period (1, 3, 70; Fig. 1). Supporting the concept that greater patient anxiety correlates with better adjuvant hormone therapy adherence are findings from the prospective COMPAS study in which patients with breast cancer with higher anxiety levels had better adherence to adjuvant endocrine therapy (P = 0.028; ref. 71). In an extremely large breast cancer population from IMS Health with 17,512 patients, depression (P < 0.002) was also associated with decreased risk of treatment discontinuation (26). As anxiety and depression can be linked in a cancer population, unraveling the relative contribution of these two factors on adjuvant endocrine therapy adherence requires further study. In this regard, despite early concerns, evidence from the NSABP placebo-controlled clinical prevention trial found depression was not increased by tamoxifen use (72, 73).

Few prospective studies have evaluated interventions designed to improve adherence to endocrine adjuvant therapy. However, to guide future study designs, theoretical models of factors influencing adherence and persistence have been proposed (74).

Although adherence to endocrine therapy in breast cancer adjuvant and prevention settings remains problematic, there are limitations to the currently available information. As reviewed (59, 75), only modest information about factors associated with continued hormone therapy use is known and, importantly, few of the factors identified are easily modifiable. In addition, current medical claims databases, commonly used in adherence analyses, contain limited information on healthcare practice patterns or patient characteristics needed to identify new potentially modifiable factors.

Despite the important influence of adjuvant endocrine therapy adherence on clinical outcome, there has only been one full-scale, randomized intervention trial designed to improve adherence completed to date. The Patient's Anastrozole Compliance to Therapy (PACT) program was a randomized, prospective, multicenter study designed to improve persistence and compliance to adjuvant endocrine therapy (76). In this trial, 4,844 patients were randomly assigned to standard therapy or standard therapy plus mailed educational materials including monthly reminders on persistence and additional letters and brochures. Questionnaires were completed before therapy was initiated, at 12 and 24 months and at treatment discontinuation. At 1 year, there was no difference in the primary endpoint of compliance (88.5% vs. 88.8 %, respectively; P = 0.81). Thus, provision of education materials did not increase adherence to adjuvant endocrine therapy (76).

A more promising result was seen in a smaller COMPAS study of 181 patients receiving adjuvant aromatase inhibitor therapy. The randomization was either to a control condition, a letter group in which participants received five mailings in the first year and three in the second, and a telephone group in which participants were contacted by a study nurse using a semistructured interview technique at the same intervals as in the letter group (77). Adherence was determined as a composite of self-report using a standardized questionnaire plus medication possession ratios calculated from pharmacy prescription refill information. At 12 months, 48% in the control group, 63% in the telephone group, and 65% in the letter group were judged adherent. Although the differences between the groups were not statistically significant, a post hoc analysis pooling both interventions versus control indicated a significant difference favoring intervention (P = 0.039). These encouraging results provide a foundation for a future confirmation trial as either intervention would be feasible for implementation in clinical practice settings.

There is emerging evidence that a patient's expectation about the benefits and drawbacks of a therapy can influence adverse effects and persistence with therapy. A meta-analysis identified significant associations between cancer patients' expectation of developing adverse effects and the actual adverse effect experience (78). When 597 patients with early-stage breast cancer who were prescribed tamoxifen were followed for 2 years, 17% discontinued tamoxifen use. Of these, women with neutral or negative beliefs about tamoxifen efficacy were significantly more likely to discontinue than those with more positive beliefs. Based on these and similar findings, several strategies to enhance endocrine therapy adherence are now focused on the development and testing of structured educational sessions implementing at the beginning of therapy with the goal of optimization of expectations. In another study, the balance between efficacy and side effects was assessed in women receiving adjuvant endocrine therapy with an Adaptive Conjoint Analysis customized to each patient. Using such information, a benefit/drawback ratio was calculated and the 16% of women who valued the efficacy less than the adverse effects had substantially lower adherence (79). Based on such findings, an ongoing randomized, controlled trial is evaluating a three-session program of cognitive, behavioral training designed to provide a realistic and balanced view of endocrine therapy (80).

Despite the paucity of full-scale clinical trial evidence, there are strategies for implementation in current clinical practice which would likely have a favorable effect on endocrine therapy adherence resulting in more favorable clinical outcome.

Adherence to endocrine adjuvant therapy has been higher in clinical trials where patient contacts are commonly greater than in clinical practice settings, and where concerned attention is directed at encouraging the maintenance of adherence. Strategies to increase patient contacts, which incorporate emerging technologies such as email reminder programs and use of cell phone apps (81–84) shown to improve adherence in other disease settings, seem promising to evaluate in breast cancer trials. Strategies to increase contacts with patients in practice settings include use of automated telephone refill reminders and implementing medication adherence tools in electronic medical records. The concept that increased contacts with patients would increase endocrine therapy adherence is strengthened by the findings from the COMPAS trial in which both additional mailings and telephone contacts seem to influence favorable adherence (77).

While awaiting results of ongoing clinical studies, one could reasonably conclude that attention to endocrine therapy patient education to optimize realistic patient expectations for adjuvant endocrine therapy, besides being good medical practice, could also improve therapy adherence. For infusional chemotherapy, in many practices, the benefits and risks of therapy, originally discussed by the oncologist, are reinforced in formal chemotherapy education sessions by a mid-level provider. A similar approach to improve adjuvant endocrine therapy patient education could be considered. Educational interventions should focus on increasing patients' understanding of the benefits and risks of therapy including the relationship between therapy adherence and persistence and higher efficacy of the therapy in reducing cancer recurrence. Implementation of these recommendations is likely to favorably affect adherence in clinical practice at this time. More definitive evidence must come from future activity in the research arena.

R.T. Chlebowski received honoraria from the speakers' bureau of Novartis, and is a consultant/advisory board member for Pfizer, Novartis, and Amgen.

Studies from the WHI program reported here were funded by the National Heart, Lung, and Blood Institute with additional support from the National Cancer Institute.