Health-related quality of life (QoL) in patients with advanced melanoma receiving immunotherapies in real-world clinical practice settings

Melanoma is a disease associated with clinically significant levels of distress at the time of diagnosis and treatment initiation, decreasing over time [1‐3]. However very little is known about the impact of advanced melanoma (AM) on quality of life (QoL) outside the clinical trial setting. Patient-reported outcomes (PROs) specifically those that focus on health-related QoL, are used in melanoma clinical trials to supplement clinical outcomes [4, 5]. Assessing PROs in oncology is important as adverse events (AEs) and cancer symptoms affect subjective experience and functioning [6]. In melanoma trials, PROs are generally collected during therapy and for an abbreviated period after patients conclude treatment with investigational agents [4, 5]. This provides limited insight into real-world therapy effectiveness in patients more heterogeneous than are included in clinical trials.

Most evidence to support the use of immunotherapies derives from trials. Pembrolizumab (PEMBRO) and ipilimumab + nivolumab (IPI + NIVO) are approved immunotherapies for AM [7, 8]. In KEYNOTE-006, PEMBRO demonstrated prolonged progression-free survival (PFS) and overall survival (OS) compared to ipilimumab (IPI) [9]. This study also showed patients treated with PEMBRO experienced less QoL deterioration at week 12 compared to IPI. Subjects receiving IPI experienced more grade 3–5 AE, (59% IPI vs 17% PEMBRO), which may impact QoL [5, 9]. In CheckMate-067, which examined patients with AM treated with nivolumab (NIVO) or IPI + NIVO, and in CHECKMATE-069, which included patients treated with either IPI or IPI + NIVO, no treatments were associated with clinically meaningful change in QoL when measured by The European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire C30 (QLQ-C30) or the EQ-5D [4, 10]. There are no studies reporting QoL in real-world settings in patients who receive PEMBRO or IPI + NIVO. Thus, the aim of this study was to examine QoL outcomes associated with PEMBRO treatment compared to IPI + NIVO when administered as first-line therapy in AM patients.

This study examined QoL in real-world patients. Real-world studies do not exclude those who may not qualify for clinical trials due to factors influencing QoL such as poorer performance status and significant comorbidity [4, 5]. In contrast to clinical trials, we collected longitudinal data on enrolled patients through 24 weeks after initiating therapy without removing those who fail to achieve an objective treatment response. In our study, PEMBRO was associated with greater improvement in scores indicating better QoL from baseline to week 24 for both the QLQ-C30 GHS/QoL and EQ-VAS, compared to IPI + NIVO. Although statistically significant, the mean differences in QoL for either instrument did not meet the criteria for clinically significant superiority. These results align with KEYNOTE-006 (PEMBRO vs IPI), which demonstrated a QLQ-C30 GHS/QoL score decrease of − 10.0 for IPI vs a more modest score decrease of − 1.9 and − 2.5 for the two PEMBRO treatment arms; p < 0.001.

In the real-world patients studied, we found that baseline mean QLQ-C30 GHS/QoL scores were lower than observed in clinical trials [4, 5]. In the PEMBRO arm, baseline means (SD) QLQ-C30 GHS/QoL was 58.8 (22.3) and 59.5 (22.6) for the IPI + NIVO arm. This contrasts with baseline mean (SD) QLQ-C30 GHS/QoL scores in KEYNOTE-006 for PEMBRO of 71.4 (20.4); 70.5 (21.9); and IPI: 67.4 (24.0). In CHECKMATE-067, baseline mean (SD) QLQ-C30 GHS/QoL for NIVO was 74.7 (19.4) and 70.7 (22.3) for IPI + NIVO. In CHECKMATE-069, NIVO + IPI and IPI had similar mean baseline QLQ-C30 GHS/QoL scores (76.9 vs 80.9), respectively. The findings may be related to differences between real-world and trial populations, as these trials exclude poor performance status subjects.

In our study, if patients responded to treatment, both treatments had improved QoL at week 24 compared to baseline, slightly greater but not statistically significant with PEMBRO. However, there was statistically significant worsening of QoL in IPI + NIVO vs PEMBRO patients not achieving an objective therapy response according to both the QLQ-C30 GHS/QoL and EQ-VAS. PEMBRO non-responders did not worsen. It is possible that when controlling for baseline clinical differences in treatments in the real-world in those not achieving response to therapy, QoL may be negatively influenced by the impact of differing risk–benefit profiles of the treatments. Interpreting QoL in cancer patients receiving treatment is challenging as it is difficult to separate the impact of QoL from the treatment versus the disease, given relevant patient characteristics. When we analyzed the QoL of the two groups of patients overall controlling for disease and patient characteristics that differed significantly between the cohorts, we did not see a large difference in QoL. However, when we specifically focused on patients who were not benefiting from therapy, we found that the IPI + NIVO group’s QoL declined more than the PEMBRO group suggesting that the cause of the decline was at least partially related to the treatment. While some of the decline in QoL could be attributed to differences in patient characteristics, we believe this is less likely given that the PEMBRO group had worse baseline features. However, clinical relevance of a two-year difference in mean age, and a 0.4 difference in comorbidity index is not known.

The results of this study highlight that QoL is a multidimensional construct, and that a single summary score complements the information provided through dimension specific QoL scores. If treatments have different AEs, the differential impact on various aspects of QoL are potentially captured by QoL measures with dimension specific scores. Global QoL ratings or summary scores reveal the overall impact, either through explicit or implicit weighting of the various dimensions of health relevant to an individual. For these reasons, rather than view the issue as a matter of accuracy or bias, the results demonstrate the importance of reporting each relevant outcome that differentiates the impact of treatment—AEs, dimension-specific impact, and overall QoL impact. If all of these outcomes favor a specific treatment, the more robust the results. Otherwise, it is important to note that how the treatments have differential impact on specific aspects of QoL potentially due to clinical heterogeneity of response to treatment, and/or different preferences for various aspects of health. CHECKMATE-069 reported treatment-related grade 3–4 adverse events in 54% of patients receiving IPI + NIVO compared with 20% of patients who received IPI alone. In this study, however, treatment arms similarly maintained QLQ-C30 GHS/QoL at week 7 (69.2 vs 74.5) and at week 13 (78.5 vs 72.2), IPI + NIVO and IPI, respectively, despite differences in AEs [10]. CHECKMATE-067 also reported that despite differences in the rate of grade 3–4 adverse events (IPI + NIVO [58.5%], NIVO [20.8%], and IPI [27.7%]) [6], this did not result in differences in QoL. Neither CHECKMATE-067 or CHECKMATE-069 examined objective responder subgroups, however.

QoL as an outcome has been gaining importance within clinical trials to understand the patient’s perspective [15]. A recent systematic review of 49 oncology treatments seeking 64 indications indicated that 70.3% included PRO data in their regulatory submissions [16]. Many recent clinical trials in AM immunotherapies have specifically collected QoL data [4, 5, 17‐23]. To-date, there is a dearth of published real-world PRO evidence, with only one small cross sectional study identified, assessing a small sample (n = 41) at one timepoint in the course of their therapy, either just prior to IPI initiation, or at an interval thereafter [24]. Our study is one of the first to generate longitudinal evidence on the impact of immuno-oncology treatments on QoL in AM treatment in the real-world.

An unusual feature of this study is the extremely high rate of retention, completion and compliance (> 90%). This is due in part to the study design, in that patients were not excluded from long term follow-up if progressing on therapy. Also in recent trials, patients were required to complete PRO instruments in the clinic prior to the study visit so that PRO responses would not be influenced by information regarding response to therapy during the visit [5]. No attempt to encourage subjects to complete instruments was made after the study visit if the PRO could not be completed before the consultation. In our study, patients were given a 10-day window around measurement timepoints with reminders in order to complete instruments, enhancing completion and compliance. It is also possible that treating clinicians introduced selection bias in enrolling patients they believed would be most likely to complete study measurements.

This study has several limitations. Notably, real-world treatment comparisons are potentially subject to selection bias, as random allocation of patients to treatments cannot be employed to achieve balanced characteristics across treatment arms. The PEMBRO cohort was older and had more comorbidity, than IPI + NIVO patients, which may have impacted QoL. As sites did not capture information about number and characteristics of eligible patients that declined to participate in the study, generalizability is limited. Data collection of clinical characteristics and objective treatment response relied on retrospective medical chart review. As such, results are limited by the completeness of information that was recorded in those charts and data that was collected. Study parameters important in potentially impacting QoL such as ECOG performance status are not always assessed or recorded in routine medical practice. However, as an assessment of frailty can be made using clinical judgement and patient objective presentation during a clinical consultation, this may or may not be a limitation in extrapolating our findings to routine practice. Baseline assessment of HRQL was not true baseline as there may have been a delay between questionnaire completion and initiation of treatment; the questionnaire was mailed within up to ten days of starting therapy, not completed at initiation of treatment. This study did not capture the occurrence of adverse events, as these events may be unreliably captured outside of a clinical trial setting. Our study only examined QoL associated with PEMBRO as a single agent PD-1 inhibitor, although nivolumab was also approved for monotherapy in this indication; as such, no real-world data are available on QoL for patients treated with nivolumab as a monotherapy. This was a real-world observational study of patients whose treating clinician had made the decision to initiate either PEMBRO or IPI + NIVO. We did not capture the clinician’s reasoning for choosing one treatment over the other and did not intervene in treatment selection. As such, we expected patient specific characteristics may have impacted treatment selection Although we captured QoL improvement, deterioration or stability at each measurement point, we did not attempt to correlate that with treatment response.

In conclusion, these exploratory PRO analyses from the real-world indicate that QoL was better maintained with PEMBRO than with IPI + NIVO when administered as first line therapy in patients with advanced melanoma. Further real-world studies are warranted to replicate our findings. Combined with the understanding of PFS, OS and toxicity risk with each treatment derived from trial data, this data can support treatment decision making.