We identified 6 cases of distant metastatic recurrence in a cohort of 194 low risk Oncotype breast cancer patients with a median follow-up period of 7.7 years, corresponding to an overall recurrence rate of 3%. For persistent patients the recurrence rate was only 0.7% (1/144) and for the non-persistent patients was 8.8% (5/57). This recurrence rate is lower than previously reported by Paik et al, 2004 (>6.8%) and Sparano et al 2018 (5.8%). Possible explanations for this including pre-selection of patients with favorable prognosis to receive Oncotype-DX testing, a shorter follow-up period, and increased use of aromatase inhibitors over the last 10 years, which confer better disease-free survival compared with tamoxifen [48].
Among the 6 cases of recurrence, only one patient completed the recommended course of therapy. It is important to note that despite her compliance, there were several factors that conferred higher recurrence risk in this case. Firstly, her Oncotype score was on the high end of the low-risk range at 24. Her risk is further increased given her high grade tumor that measured 4.0cm in size. These clinicopathologic factors are better captured by the Recurrence Score Pathologic Clinical (RSPC) calculator, which would estimate her 10-year risk of recurrence to be 30% (compare to 16% risk estimated by Oncotype-DX). Whether to categorize this patient as ‘low risk’ within this context is an important question [49], [50].
In the remaining 5 cases of low-risk breast cancer that progressed to metastatic disease, each of those patients either did not initiate or did not persist with endocrine therapy as was recommended. Our findings raised the question of whether most early distant recurrences in women with low-risk Oncotype breast cancer occur in the setting of failure to complete a 5-year course of endocrine therapy. This prompted further work to understand non-persistence in our population.
We found the rate of non-persistence with endocrine therapy in our population to be 25% which is consistent with the United States literature. However, for reasons discussed below and as has been pointed out by other groups, it is likely our data are underestimating the true rate of non-persistence.
Similar to others [31], we found that social support is a key factor associated with persistence to endocrine therapy, with 92% of persistent patients reporting strong or satisfactory support, compared with 84% of non-persistent patients (p=0.002). It has been hypothesized that clinical social support can mitigate low personal social support [51], although we did not measure this directly in our study, a surrogate may have been participation in RN/NP teaching prior to surgery. We found that a higher percentage of persistent patients engaged in teaching compared with non-persistent patients, and this difference approached statistical significance (63% vs 45%, respectively, p=0.07).
We hypothesized that if persistence with medication is motivated in part by an understanding of risk, the 10-year recurrence risk as estimated by Oncotype-DX may be an influential factor in the decision to continue endocrine therapy for the recommended 5-year course. Others have shown the influence of recurrence score on initiation of endocrine therapy [46] however did not find a significant effect on early discontinuation. In contrast, we found that patients who persisted with endocrine therapy despite bothersome side effects had significantly higher Oncotype Scores compared to patients who discontinued early, although the difference was small (15.9 vs 13.7, p=0.028). To our knowledge, this study is the first to show data suggesting a potential influence of Oncotype-DX score on patient decisions regarding persistence to endocrine therapy. Ensuring that patients fully understand Oncotype-DX score reports may be important when guiding patients in making informed decisions regarding whether to continue with endocrine therapy or not.
Previous studies that have shown higher tumor stage and size are associated with persistence with endocrine therapy [21], [52], [53]. In contrast, we found no significant difference between persistent and non-persistent patients in terms of tumor size, grade, lymph node involvement, or need for revision surgery. Lymph node involvement has produced mixed results in the literature [54], [55] and in our study this was the only clinicopathologic factor that trended toward having an effect with 26% of persistent patients having lymph node involvement compared with 18% of non-persistent patients, but this did not reach statistical significance (p=0.20). Additionally, there was no difference between groups in terms of 10-year risk as estimated by the RSPC calculator, which incorporates tumor size and grade (6.9% vs 7.6%, p=0.475), although it is notable that this calculator was not in routine use clinically during this study time period.
There was no difference in the median number of side effects reported between persistent and non-persistent patients; however, non-persistent patients were more likely to report severe symptoms compared to persistent patients (OR 4.11, 95% CI 2.1-7.9, p=0.002). Fatigue was the only side effect more frequently reported in non-persistent patients compared to persistent patients (33% vs 12%, p=0.01). To our knowledge, this is the first report implicating fatigue as being significantly associated with non-persistence. Based on our review, the most common side effects associated with non-persistence are gynecologic symptoms [13] and arthralgia [5]. Attributing non-persistence to side effects alone, however, may be an over-simplification [4]. Further, it may be difficult to discern side effects of endocrine therapy from symptoms and perceptions of normal aging. Analysis of the International Breast Cancer Intervention Study (IBIS 1) showed no significant difference between symptom effect size on adherence between patients on tamoxifen vs placebo [56]. Additional support for this concept arose from IBIS 2, where investigators found no difference in adherence rates for women on anastrozole compared to women on placebo (65.7% vs 65.9%, respectively). In fact, in that study adherence rates were lower in the placebo group than in the treatment group for women with arthralgia [57]. Another group found that, after adjusting for joint pain severity, women with high levels of ageing perceptions were at greater risk of non-adherence than women with low levels of ageing perception, suggesting that the association of joint pain with older age compounds the issue of adherence [58].
The IBIS 2 study showed a significantly greater tendency toward non-adherence in women on anastrozole who reported gynecologic symptoms. Sexual side effects of endocrine therapy are thought to be underreported by both patients and clinicians [4], [59] and remains an important barrier to persistence with endocrine therapy. Interestingly, we found that patients who complained of vaginal dryness were more likely to persist with tamoxifen vs AI, although this difference did not reach statistical significance (OR 6.2; 95% CI 0.78-49.2; p=0.05). Tamoxifen has a tendency to increase vaginal discharge[60]; it is possible this effect is mitigating vaginal dryness and increasing quality of life for some women, thus enabling persistence with tamoxifen.
Switching endocrine therapy has been associated with poorer persistence [16], [19]. Our study showed that non-persistent patients were 3.8 times more likely to be offered a switch in endocrine therapy upfront rather than first trying symptom-relieving medications (OR 3.8, 95% CI 1.44-10.1; p=0.006). Moreover, compared with the persistent group, a greater proportion of non-persistent patients declined a switch, and this difference approached statistical significance (26% vs 10%, p=0.05). In contrast, persistent patients were 10 times more likely to have been offered symptom alleviating medications compared with non-persistent patients (OR 10.3, 95% CI 3.8-27.4, p<0.001). In both groups, nearly all patients who were offered symptom alleviating medications went on to accept a prescription (100% of non-persistent, and 96% of persistent patients).
A possible explanation for these findings may be the perception that patients experiencing certain side effects are more likely to benefit from medication management vs a switch. A subset analysis (data not shown) revealed that side effect profiles differed between patients who were offered symptom-management vs a switch, specifically, patients with hot flashes were more likely to have been offered side effect management. Patients who switched medications reported a slightly higher proportion of arthralgias and a significantly greater proportion ‘other’ less common side effects, perhaps for which evidence-based management is not readily available. Further, it is possible that when patients are engaged in side effect management with their provider they may have a more positive communication experience compared with those being offered a switch in therapy. Numerous studies have reported on the association of patient-centered communication and persistence with endocrine therapy [34], [35], [61], [62]. Finally, it is possible that some patients had indeed expressed their disinterest in a trial of symptom-relieving medications, and these discussions simply were not documented in the office notes.
There are several limitations to our study. Firstly, this is a retrospective clinical study with data based entirely on chart review, and thus caution must be exercised when drawing conclusions. Errors in documentation and patient reporting of medication-taking behavior and side effects are surely present to some extent. Records of treatment duration would ideally have been confirmed by pharmacy records which were not available to us during this time period. For these reasons, it is likely that we have underestimated the rate of non-persistence with endocrine therapy in our population.