Patient-Reported Outcomes Bridge an Important Gap in Identifying Risk for Early Endocrine Therapy Discontinuation
Year: 2021
Type: article
Abstract: I have had the following experience dozens of times while seeing patients alongside a mentor and colleague, who is a highly admired and well-seasoned oncologist. We enter the exam room and my colleague asks, “Mrs Smith, how are you?” The patient smiles and politely responds, “I’m fine.” My colleague begins his examination by reviewing her symptom ratings on the questionnaire she completed prior to our entering the room. “Your pain is a 4, fatigue an 8, and anxiety a 6!” After completing this clinical encounter, my colleague exits the room, and his puzzled expression clearly conveys his frustration. He wonders aloud, “Why do patients say they’re fine? Why don’t they really tell me how they’re doing? I’ve been treating Mrs Smith for 10 years. She can tell me anything.” This scenario is all too common, as evidenced by accumulating research documenting the lack of concordance between clinician- and patient-rated symptoms and quality of life (1-3). This discrepancy reflects the complexity of patient–provider communication, particularly in the context of cancer treatment. Patients face inherent disincentives to reporting symptoms and toxicities. Patients do not want to distract their oncology team from treating their cancer, and many understand that reports of treatment-related symptoms may result in dose reductions, delays, or discontinuation. Patients are understandably reluctant to spontaneously report symptoms, and providers assume patients will initiate these conversations. Fortunately, patient-reported outcomes (PROs) measures, the gold standard for quantifying symptoms from the patient’s perspective (4,5), provide an opportunity to bridge this communication gap and, in doing so, measurably improve care. In a randomized controlled trial, routine PRO assessment resulted in a survival advantage over usual care among patients receiving treatment for advanced cancer (6). Given increased availability of oral therapies, which are often taken on a long-term basis, extending this research to patients with elevated risk for elective discontinuation would generate opportunities to intervene to promote adherence. Patients receiving therapies assumed to be well tolerated may not be routinely assessed for side effects, thus they may be uniquely at risk for nonadherence and not receiving maximum benefit of treatment. In this issue of the Journal, Hershman and colleagues (7) report on the value of PROs in predicting long-term adherence to aromatase inhibitor (AI) therapy using data from the large multisite SWOG S1105 trial. The primary result from this trial has been reported (8), and the finding that a text-messaging intervention did not improve AI therapy adherence points to the complex nature of health behaviors, such as daily oral therapy adherence. The deeper understanding of predictors of long-term AI adherence provided by Hershman et al. (7) helps target interventions for this challenging and common clinical problem to patients at greatest risk for early discontinuation. A considerable proportion, 35.9%, were nonadherent at 36 months, which underscores the need to validate predictors of elective discontinuation. We have previously demonstrated that the risk of long-term nonadherence to AI therapy can be predicted using patient report (9). We found that with every 1-point increase on a 5-point scale using a Functional Assessment of Cancer Therapy-General (FACT-G GP5) item that assesses bother by treatment side effects (10), the risk of early AI discontinuation increased by 29% (9). At 48 months, 53.6% of women reporting moderate-severe side effect bother discontinued AI therapy compared with 70% of women who reported no to little bother on the GP5 item. Similar to Hershman et al. (7), we observed a high rate of early discontinuation. Initially, we were surprised that side effect bother at the time of AI initiation was predictive of long-term adherence. In examining this further, prior chemotherapy, prior radiation therapy, and 8 or more medications per day were associated with side effect bother prior to starting an AI. We interpreted our findings as evidence for “treatment fatigue,” which we conceptualize as weariness or feeling worn down because of chronic symptom burden, which results in a diminished capacity to tolerate ongoing treatment side effects. Among women enrolled on the TAILORx trial PRO substudy (11,12), depression and physical and social well-being PRO scores at baseline predicted early endocrine therapy discontinuation looking at adherence up to 4 years, although we observed an effect in the opposite direction with prior chemotherapy (13). Collectively, our findings indicate that the threshold for treatment tolerability is influenced by prior treatment, comorbid conditions, and health-related quality of life and can be measured using PROs. The predictive value of patient-rated side effect bother (9), depression, and lower health-related quality of life (13) at treatment initiation is a powerful lesson that our definition of baseline in the context of a clinical trial may be far from baseline from the patient’s perspective, particularly those who received prior cancer therapies. Hershman et al. (7) provide additional confirmation of the predictive value of PROs using a composite PRO score to predict AI adherence at 36 months. Their design and primary findings converge diverge from ours in a few interesting ways that advance our understanding of AI adherence. Hershman et al. (7) required prior AI therapy for 30 or more days, whereas we examined PROs at AI initiation (9). Their results demonstrate the value of ongoing PRO administration throughout AI therapy to identify women with emerging risk for early discontinuation. Although they did not find prior chemotherapy or radiation therapy as predictors of AI adherence, the additional time since treatment completion may have afforded time for side effects from prior therapy to abate. In our sample, approximately 75% recently completed chemotherapy and/or radiation therapy (9). This lends further support to our conceptualization of treatment fatigue as an important influence on one’s threshold for tolerating treatment side effects and that fluctuates in tandem with symptom burden from other therapies. Both trials assessed long-term adherence, and the use of biochemical validation to operationalize adherence is a major strength of the Hershman et al. (7) multisite trial, whereas we relied on clinician report (9). Hershman and colleagues (7) employ a broad approach to understanding contributors to nonadherence by examining the role of social determinants of health, another major strength of this trial even though they did not find associations. In contrast, women enrolled on TAILORx (11) who were uninsured or underinsured had higher rates of endocrine therapy nonadherence (14). The contribution of social determinants to AI therapy persistence clearly requires much more in-depth exploration. Hershman and colleagues (7) have a solid foundation on which to advocate for the use of PROs to identify patients vulnerable to nonadherence who may benefit from targeted intervention. Although longitudinal, multidimensional PRO assessment is increasingly recognized as an integral component of high-quality cancer care (15), implementation may be challenging for busy, resource-constrained clinical practices. A brief, precise assessment of treatment tolerability may offer a viable approach for the identification of patients at risk for elective discontinuation before gaining the full benefit of therapy. The National Cancer Institute, through the Cancer Moonshot Initiative, established a Cancer Treatment Tolerability Consortium (16) to analyze an expansive array of clinical trial data to improve our understanding of cancer treatment tolerability and identify indicators that signify patients vulnerable to treatment-related tolerability issues and early treatment discontinuation (17–21). Through this Consortium, our research team demonstrated increasing bother by treatment side effects [FACT-GP5 (10)] predicted early discontinuation of myeloma therapy (22), and we partially replicated the association between baseline bother and early treatment discontinuation across a diverse range of clinical trials (23). In conclusion, PROs provide a methodologically robust strategy to bring the patient’s voice to the clinical encounter. Clinicians can proceed with confidence that PROs will bypass the “I’m fine” response and will point to patients who may benefit from additional clinical attention to ensure that they gain maximum benefit from treatment. Hershman and colleagues (7) should be commended for this large well-designed, multisite trial that paves the way for routine use of PROs to optimize outcomes for our patients. None. Role of the funder: Not applicable. Disclosures: Lynne Wagner, PhD, has reported consulting income from Celgene, Inc, and Athenex, unrelated to this work. Acknowledgements: For contributing to this work, I gratefully acknowledge patient advocate Mary Lou Smith, Drs David Cella, Steven Rosen, members of the NCI Cancer Treatment Tolerability Consortium, ECOG-ACRIN Cancer Research Group collaborators, cancer survivors who have entrusted me with their psychosocial care, and the countless number of patients who have generously shared their time and experiences by completing cancer trial PRO questionnaires. Author contributions: Writing, original draft—LW; writing, revision and editing—LW. Not applicable.
DOI: 10.1093/jnci/djab024
Author: Lynne I Wagner
Language: English
Topic: Cancer survivorship and care
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W3132467283