Importance of endocrine treatment adherence and persistence in breast cancer survivorship: a systematic review

The inclusion criteria were prospective and retrospective studies that investigated how patient adherence and/or persistence to AET were correlated with breast cancer recurrence and survival among patients with early-stage luminal breast cancer. We excluded studies on neoadjuvant treatment, prophylactic treatment, or subgroups of patients with ductal carcinoma in situ (DCIS), male only breast cancer, and metastatic breast cancer.

Previous publications have applied heterogeneous definitions of AET adherence and persistence [22]. From the included studies, we divided the results into two subgroups: non-adherence and non-persistence. Studies were included if they investigated the effects of AET non-adherence and/or non-persistence in women with early breast cancer. Moreover, the outcomes had to include breast cancer recurrence, disease-free survival, breast cancer-specific survival, or overall survival, with the former three end-points grouped as ‘event-free survival’.

A commonly used definition of adequate adherence is MPR ≥ 80%, which was used in most of the included studies. Although the optimal MPR cut-off point is not known, it is evident that relapses and mortality increase when MPR is below 80% [16, 23].

Persistence to medication use is calculated from the start of treatment until the patient completes the treatment or prematurely ends the treatment. Prescription databases are often used to monitor persistence, and a commonly used definition of non-persistence is the registration of a gap period without new prescriptions for several months (90–180 days) before the scheduled end of treatment [19, 24].

Our systematic literature search was conducted on November 12^th 2020. Studies were retrieved from the following electronic databases: Medline (Ovid), Embase (Ovid), the Cochrane Library (CDSR and CENTRAL), CINAHL (EBSCOhost), and Web of Science. The search strategy was initially developed for Medline (Ovid) and appropriately adapted for other databases. The keywords of the Boolean search were "breast cancer" AND "endocrine treatment" AND "adherence OR persistence" AND "recurrence OR survival". Synonyms were used for all keywords (eg, compliance, refusal, etc.), and controlled vocabularies in the databases were used to extract all relevant search terms. Attachment 1 (Medline search strategy) presents a fully reproducible search strategy for Medline (Ovid). The search results were limited to English language articles published between 2010 and the search date. Conference abstracts and unpublished clinical trials were excluded.

The search results were imported into EndNote X9, and duplications were removed. The final set of citations was imported into COVIDENCE [25] for further deduplication and eligibility screening of the titles and abstracts. Thus, the literature search, study design, and data analysis process followed the PRISMA guidelines [26]. All search results were uploaded, and titles and abstracts were independently screened by three authors (FE, HS, and TL), based on the predefined inclusion and exclusion criteria. These authors then conducted a full-text screening. Any conflicts were resolved by majority vote. Figure 1 shows a flow diagram of the selection process. Table 1 presents an overview of the study characteristics of the included studies. Table 2 includes an overview of the patient characteristics of the included studies.

The included studies were evaluated for risk of bias using the Cochrane Risk of Bias Comparison in Covidence. The individual studies were graded as having high, low, or unclear risk of bias in the following categories: sequence generation, allocation concealment, blinding of participants and personnel for all outcomes, blinding of outcome assessors for all outcomes, incomplete outcome data for all outcomes, selective outcome reporting, and other sources of bias. Table 3 presents a full description of the bias analysis.

Hazard ratios (HRs) were used to describe the relative risk of non-adherence and non-persistence to AET. Forest plots were generated to present HRs with calculated 95% confidence intervals (CIs) and the number of patients. Variables were coded such that the HR represented the relative risk of non-adherent/non-persistent patients versus adherent/persistent patients. We separately analyzed event-free survival and overall survival. The analyses were performed using R version 4.2.1 [27]. The studies exhibited substantial heterogeneity in study design, analysis approaches, definitions of adherence and persistence, whether adherence was treated as a time-dependent covariate, definitions of outcome variables, and adjustment for specific factors. This implies that the HRs reported in the different studies are not directly comparable, as they measure different quantities and have different interpretations. Consequently, it was not appropriate to perform a meta-analysis, even though the individual HR values were independently valid.

There is a close relationship between adverse events, and patient adherence and persistence. This point has been raised in several large multi-centre studies such as BIG-1–98 [31] and MA.27 [43]. Medication side-effects are believed to play a major role, where symptoms that are caused by depletion of circulating estrogens such as hot flashes [44], muscle pains and vulvovaginal symptoms are commonplace [45]. The women may seek to minimalize these side-effects by either reducing adherence or persistence. Despite this fact, there is a history of underreporting side-effects by clinicians [46]. When asked, 81% of all breast cancer patients on AET report at least one major side-effect, and the majority report more than one.

The patients who report side-effects have a significantly increased risk of non-persistence to treatment [47]. Musculoskeletal symptoms are one of the most reported side-effects among patients undergoing AI treatment. It is suggested that these side-effects cause premature discontinuation, and a study of 3887 patients showed 15% non-persistence after 12 months of treatment, with 63% citing musculoskeletal symptoms as the cause of non-persistence. Despite several high quality studies showing a significant relationship between AET side-effects and non-adherence and non-persistence, the large variation in study characteristics and definitions of adherence makes it difficult to draw any definite conclusions as to the causes of non-adherence [48].

An important factor to take into consideration is that despite it being understandable that some breast cancer survivors may stop adhering to their treatment if the side effects are intolerable, there may be ways to reduce the side effects, for example vasomotor drug treatment, switching between different AETs, or non-pharmacological approaches [44, 49]. Although there are many treatments available to reduce side-effects, only 44% of patients with side-effects use medications to overcome them [47]. There is evidently a large variation in the intensity and type of side-effects in this patient group. Having treatment guidelines in place, with individualized treatment plans for managing these side effects, can be one factor in improving patient adherence.

There may be a direct link between patient side-effects and the effect of AET on survival outcomes, where patients with vasomotor side-effects when undergoing AET have been shown to have a greater decrease in breast cancer recurrence than patients with no side-effects [50]. Similarly in a study of the WHEL trial, where 864 patients were taking tamoxifen, the patients who reported hot flashes at baseline were less likely to develop recurrent breast cancer after 7 year follow up [51]. This paradoxical effect of higher survival rates in a patient group with lower expected adherence may indicate a greater response to AET in the group of patients that experience larger side effects of medication. Building on this fact, individual assessments of active tamoxifen metabolites could be used to predict side effects [52] and survival [45, 53]. This point further highlights the importance of continuing AET despite its troublesome side-effects.

The four-way link between serum levels of active tamoxifen metabolites, medication side-effects, non-adherence and persistence, and survival outcomes among these patients is therefore a better approach for increased knowledge of the factors that can improve patient outcome in breast cancer patients undergoing AET.

Planned sequential treatment (two years AI followed by three years tamoxifen, or two years tamoxifen followed by three years of AI) have been studied in a large randomized trial against five years of AI and five years of tamoxifen. The rates of non-persistence have been shown to be poorest among the patients who have received sequential treatment [31]. This suggests that patients find it difficult to handle a new set of side-effects after a switch in treatment. In long-term follow up studies on this cohort of patients, the AI arm has shown a relative risk reduction of 9% after 12.6 years [54]. Despite this, for a subset of patients with a high burden of adverse effects, switching from AI to tamoxifen may be preferable to stopping all treatment [54]. Several other studies have found that switching regimens from tamoxifen to AI may increase compliance, and can be beneficial to survival [55]. Individual treatment plans, and an understanding of each patients challenges with adverse events and side-effects is therefore important when determining the treatment plan.

Demographic variables and patients’ psychological and financial situations are reported to be important predictors of poor adherence [56, 57]. In a nested cohort study including 2,616 patients from Denmark, patients with low socioeconomic position (SEP) were found to have increased 5-year cumulative incidence rates of disease relapse (13%) and breast cancer-related death (11%). This study also found that single women had increased 5-year rates of recurrence (RR, 1.45; 95% CI, 1.11 to 1.89) and mortality (RR, 1.83; 95% CI, 1.32 to 2.52) [58]. The socioeconomic disparity was most evident within the group of patients with ER + tumors, suggesting that lower treatment adherence in these groups may partly explain the differences in survival and recurrence. Further studies of how socioeconomic factors are related to outcome could potentially enable the identification of subsets of patients who have an increased risk of non-adherence, and who thus require closer follow-up throughout their treatment.

Medicine availability remains a global challenge. Breast cancer is a major health concern in poorer areas of the world and, for many people, the lack of availability and prohibitive cost of medication leads to poor adherence and non-persistence. Women must shoulder the double burden of disease and inadequate healthcare, which broadens the health gap and increases disparities. In the United States, it has been reported that racial and ethnic diversity impacts treatment and survival [59]. Since these disparities are multifactorial, the strategies aimed at reducing them must involve advocacy, research, education, and healthcare services. The success of these strategies depends on their support at the federal and state levels, and the involvement of local communities in developing programs and policies that are culturally and linguistically appropriate for their communities, in order to ensure both the utility and the duration of these efforts [59].

The present study has several limitations. First, only a small number of studies were included in this analysis, despite over 2,000 studies being reviewed. This may be due to a lack of research focused on adherence to adjuvant endocrine treatment in the breast cancer research community and may be an indication of the complexity of endocrine treatment adherence. Additionally, studies were excluded from the analysis due to problems related to bias and study protocols that made the results unreliable. Despite the high quality of the 14 included studies, the heterogeneity was too great to allow us to perform a meta-analysis. This was mainly due to the use of different definitions for adherence and persistence, differing outcomes and analysis methods, and differences in follow-up time and the treatment received, which make it difficult to provide a summated HR with 95% CI.

Implementing common definitions of adherence and persistence would expand the possibilities of studying the effects of poor adherence to AET. There is presently no gold standard for assessing adherence to AET therapy. Using an MPR cut-off of 80% adherence seems to be the most widely used definition of adequate adherence to AET and can be feasibly measured using pill counting or prescription records. However, this is a measurement with many potential sources of error. On the other hand, more direct approaches, such as serum measurements of active metabolites, can provide a more objective status, but have yet to be accessible for large-scale studies [34, 60, 61].

Reporting of tumor size and nodal status was lacking in several of the studies. This gives limitations as to which patients are most affected by non-adherence and non-persistence. Standardized reporting of patient characteristics would allow for better comparisons of the patient groups included in different studies and can give evidence as to which patient groups suffer more from non-adherence and non-persistence.

Our present review indicated that non-adherence and non-persistence to AET has been a neglected research topic for decades. Importantly, improving adherence and persistence represents a low-hanging fruit for increasing survival in luminal breast cancer. There is a need for tailored, interdisciplinary, long-term follow-up strategies, including identification and treatment of side effects. Moreover, serum detection of tamoxifen and AI metabolites during the whole treatment period would enable identification of non-adherent patients, and allow doctors to adjust treatment to more functional levels to reduce side effects [45, 61]. Establishing AET units in oncological/surgical departments to promote cooperation between clinicians, family doctors, oncology nurses, psychiatrists, psychologists, and social workers may help to ensure that patients receive the care that they need to improve adherence [58]. However, randomized studies are needed to evaluate how such AET units impact adherence and survival. If effective, AET units could be a low-cost, high-reward means of improving the survival of women with breast cancer.