Influence of patient and tumor characteristics on therapy persistence with letrozole in postmenopausal women with advanced breast cancer: results of the prospective observational EvAluate-TM study

Endocrine therapy (ET) is the recommended treatment in patients with hormone receptor-positive, advanced breast cancer. While, according to current guidelines, premenopausal women should receive tamoxifen as first-line therapy, aromatase inhibitors (AIs) or fulvestrant are preferred in postmenopausal metastatic breast cancer (MBC) patients [1, 2].

ET has recently become the focus of MBC treatment as novel combination therapies are being developed for hormone receptor-positive, advanced breast cancer patients to overcome endocrine resistance [3]. Adding the mTOR inhibitor everolimus to a therapy with the AI exemestane, for instance, improved progression-free survival (PFS) [4]. Furthermore, a combination of the cyclin-dependent kinase 4/6 (CDK 4/6) inhibitors palbociclib, ribociclib, or abemaciclib with ET has been investigated and has consistently shown a clinically relevant improvement in PFS [5‐7].

To ensure ET efficacy, patient compliance and treatment persistence are needed in both the adjuvant and advanced setting. In the adjuvant setting, which has been investigated in several trials, compliance and persistence of therapy with tamoxifen or AIs in postmenopausal breast cancer patients decrease over the course of treatment [8‐11], which, in turn, is associated with reduced disease-free survival (DFS) [12]. Some baseline patient and tumor characteristics such as age, socioeconomic factors, or tumor stage have been reported to have an influence [8‐14]. In the advanced setting, in contrast, only few studies have analyzed patient compliance with ET, and thus not much is known about possible risk factors [15, 16].

Data concerning the persistence with regard to AI might not only be helpful for patients treated with a monotherapy but also for comparing persistence regarding endocrine combination therapies. Aim of this study was, therefore, to describe therapy persistence and to identify predictors for therapy persistence among those patient and tumor characteristics known at the start of treatment in a prospective, noninterventional study in patients receiving letrozole monotherapy in the metastatic setting. The hypothesis was that side effects and patient characteristics result in patients with different adherence rates.

Between 01/2008 and 12/2009 a total of 5045 patients were enrolled in the study, of whom 252 had advanced breast cancer. Of these, 52 women were excluded, out of which 28 patients were excluded because endocrine therapy started more than 30 days before signing the informed consent and 14 patients because treatment was started more than 30 days after the informed consent. In 4 patients the follow-up was too short (< 30 days) for side effect evaluation and in 6 patients data on disease progression or therapy compliance was missing. Thus, the current analysis consists of data of 200 MBC patients (Additional file 1: Figure S1).

After 12 months of observation, patients who were nonpersistent for reasons other than disease progression were still under AI treatment, with a persistence rate of 85.5%. In these patients, persistence was clearly compromised when AEs were reported within the first 30 days of treatment. Furthermore, statements about noncompliance in the past could also predict lower persistence.

ET with an AI not only reduces the recurrence rate of hormone receptor-positive breast cancer in the adjuvant setting [23], it also prolongs overall survival (OS) in those patients with advanced breast cancer [24]. As in the adjuvant setting adherence to AI therapy seems to have a direct influence on DFS [12], an important role in the treatment of MBC can be hypothesized.

In early breast cancer age [9, 13], comorbidities [9], prior chemotherapy or radiation [25, 26], tumor size [13], as well as socioeconomic factors [25] have been reported to have an influence on adherence to ET. In a Brazilian cohort of breast cancer patients, those who were diagnosed at a noncurable stage were less adherent to ET [25], while other contradictory analyses report that the stage at diagnosis seems to be associated with persistence, but not compliance [27]. Among MBC patients, there are only few analyses of adherence to ET [15, 16, 28]. An Italian investigator group observed among 285 postmenopausal MBC patients treated with exemestane that those who were married or had a university degree were less likely to not adhere to ET. Furthermore, older age at diagnosis, a higher number of comorbidities, as well as a lower receptivity toward therapy seemed to be associated with nonadherence. After 6 months of treatment, the adherence rate was 78% [15]. A recent analysis from Switzerland shows that out of 165 women who started palliative ET, a total of 12.8% did not persist (therapy termination or therapy change) with therapy due to side effects or for reasons other than disease progression. Those who were naïve to ET showed a higher persistence with palliative ET, while those with more metastatic lesions at diagnosis were less persistent [16]. In the FALCON study, 78.9% of the postmenopausal MBC patients receiving anastrozole discontinued treatment, among whom only 10.8% were for reasons other than disease progression. Of these treatment terminations, 4.7% were reported to be due mainly to adverse events. The median duration of actual exposure to anastrozole was 13.9 months [28].

Some of the aforementioned patient and tumor characteristics were also investigated in the present analysis. While none of the analyzed patient characteristics such as age, BMI, ECOG, or the number of concomitant medications had a significant influence on therapy persistence, AEs in the early treatment phase and patient noncompliance due to illness and forgetfulness correlate with nonpersistence. As mentioned before, literature on predictors of persistence with palliative ET is scarce, and thus it is difficult to bring these results in line with others. AI-induced side effects, which are described as the main reason for nonpersistence in the present work, represent one reason often given for noncompliance and an associated premature end of treatment [26, 29]. In the adjuvant setting, the COMPAS trial could demonstrate that compliance with AIs improves side effects, while noncompliant women were more likely to experience a deterioration of AE and might therefore discontinue treatment prematurely [30]. This might explain the vicious circle and why, in the current analysis, adverse events and noncompliance are in turn associated with a higher risk for nonpersistence.

For novel combination therapies, compliance rates are only available from prospectively randomized clinical trials. In the PALOMA-2 study with a median follow-up time of 23 months, an overall permanent discontinuation of study treatment as a result of AEs was reported in 43 patients (9.7%) in the palbociclib–letrozole group and in 13 patients (5.9%) in the placebo–letrozole group [5]. In the MONALEESA-2 study, at a median duration since randomization of 15.3 months, discontinuation due to AEs was reported in 87 patients (26.0%) in the ribociclib group and in 146 (43.7%) in the placebo group [7]. These figures are lower than those from our data. However, in clinical trials, compliance is generally thought to be higher for several reasons. Therefore, it will be important to observe real-world data that will capture this figure for patients on these novel combination therapies. However, it can be assumed that the rate will be lower than the 85.5% persistence rate that we reported for monotherapy.

A strength of this study is that, due to the nationwide patient recruitment, a broad MBC patient population is represented. Interesting is the high rate of MBC at first diagnosis, namely, 56.0%, which in the literature is reported to be only 5–10% [1], but is similar to the percentage in recent studies in that patient population [6]. Further trials report lower rates [4, 5, 7], which nevertheless, in comparison to epidemiologically known data, are high, so that the question arises as to whether there is a general increase in MBC at first diagnosis or whether this is based on a study selection bias. A weakness of our analysis is that, due to the small number of events (n = 26), the possible predictors for TTEOT were split into two competing risk regression models to achieve convergence and obtain robust results. Therefore, the results of the two models must be interpreted carefully by taking into account the separation of the predictors. A further weakness of the study is that data regarding compliance were collected by evaluating patient questionnaires and physicians’ assessments only at the time of enrollment and after 6 and 12 months. Since the median observation time in this study was only 10.6 months, but the median PFS of an ET with an AI is about 14.0 months according to the literature [28], it can be assumed that therapy persistence would even continue to decrease over the following months. As patients were not observed after the end of treatment, it also remains unknown whether women nonpersistent to letrozole treatment switched to another ET or were nonpersistent in general and discontinued therapy altogether. Importantly it has also to be noted that for our predictor women would have to be observed for 30 days with regard to the occurrence of side effects. Our findings can only be used for these women. For women who terminate the therapy before that time our findings are not applicable.

The analysis suggests that the presence of AEs and statements about previous noncompliance can predict those women who will terminate palliative therapy with an AI. Despite suffering from a life-threatening disease and receiving a treatment that is generally considered as being well tolerated and thus the treatment of choice, AEs of an AI and a behavioral pattern related to noncompliance will result in a significant proportion of patients who prematurely terminate treatment. Therefore, further analyses are necessary to find predictive factors and identify MBC patients who are at risk for early treatment discontinuation and could benefit from supporting compliance programs. For example that up to 44% of women with side effects would terminate the therapy within 12 months of treatment compared to about 11% without side effects, makes this population a group of patients of interest who should be part of an intensified treatment management program. Furthermore, it should be investigated whether compliance and persistence patterns are the same with novel endocrine combination therapies.