Health-Related Quality of Life Impact from Adding Bevacizumab to Cisplatin-Pemetrexed in Malignant Pleural Mesothelioma in the MAPS IFCT-GFPC-0701 Phase III Trial

Pleural mesothelioma causes over 43,000 deaths worldwide each year (1). Malignant pleural mesothelioma (MPM) is primarily associated with occupational exposure to asbestos. The incidence of MPM could potentially continue to grow in the future (2).

Vascular endothelial growth factor (VEGF) is one of the most relevant endothelial angiogenic factors, expressed by virtually all tumors. In pleural mesothelioma, high levels of VEGF expression are associated with poor prognosis (3). Bevacizumab is a mAb with a high affinity for VEGF that has demonstrated activity in other types of cancers (4).

The French Cooperative Thoracic Intergroup (IFCT; Intergroupe Francophone de Cancérologie Thoracique) conducted a multicenter phase III trial (MAPS IFCT-GFPC-0701) evaluating the effects of adding bevacizumab to the standard first-line chemotherapy regimen (cisplatin-pemetrexed; ref. 5) in advanced MPM. This study reported significant improvement in overall survival (OS), with a median OS of 18.8 months in the bevacizumab-cisplatin-pemetrexed (PCB) arm versus 16.1 months in the cisplatin-pemetrexed arm (PC), and an adjusted HR of 0.77 [95% confidence interval (CI), 0.62–0.95; P = 0.0167]. As expected, several grade 3 to 4 adverse events were found to be increased by adding bevacizumab, namely low-grade hemorrhages (mainly epistaxis), hypertension, and arterial and venous thromboembolic events.

Health-related quality of life (HRQoL) analysis in cancer research is critical to better understand patients' perception of the impact of cancer and cancer therapy on their life. The FDA considers HRQoL to be a critical endpoint for assessing direct clinical benefits (6). Contrasting results of HRQoL analyses have been reported with bevacizumab in other tumor localizations, however. For example, bevacizumab combined with standard chemotherapy, following debulking surgery for FIGO high-risk stage I to IV epithelial ovarian cancer, significantly decreased global HRQoL (EORTC-QLQ-C30 analysis), despite markedly improving progression-free survival (PFS; ref. 7). Therefore, it was of utmost importance to verify that the increased toxicity associated with bevacizumab did not cause impaired HRQoL. We thus sought to evaluate the impact of adding bevacizumab to cisplatin-pemetrexed in terms of HRQoL in the IFCT-GFPC-0701 MAPS trial.

In this multicenter, randomized, controlled, open-label, phase III trial, patients were eligible if they had a previously untreated, histologically proven, MPM. Our detailed inclusion criteria can be found in a previous publication (5).

The research protocol was approved by a National Ethics Committee. The study was conducted according to the Declaration of Helsinki of 1964 and Good Clinical Practice guidelines. All patients provided written informed consent to participate.

HRQoL was assessed using the European Organisation for Research and Treatment of Cancer (EORTC) QLQ-C30 questionnaire (8) and the lung cancer–specific module QLQ-LC13 (9) questionnaire, validated in patient treated with chemotherapy in mesothelioma (10). Both were completed at randomization and then every 9 weeks until disease progression. Paper questionnaires were given to the patients and filled in, by the patients themselves, at the hospital, before consultation with the oncologist.

The QLQ-C30 is a 30-item cancer-specific tool (11), enabling assessment of 5 functional scales (physical, role, cognitive, social, and emotional), a global QoL scale, and nine symptom scales (nausea and vomiting, pain, fatigue, dyspnea, sleeping disturbances, appetite loss, constipation, diarrhea, and financial difficulties; ref. 8).

The QLQ-LC13 was developed and validated specifically for patients with lung cancer (9). It includes 13 items that address lung cancer symptoms (cough, hemoptysis, dyspnea, and site-specific pain), treatment-related adverse effects (sore mouth, dysphagia, peripheral neuropathy, and alopecia), and pain medication.

We computed the HRQoL scores using the EORTC Scoring Manual and standardized them on a 0 to 100 scale. A high score reflected a high functional level, high global quality of life level, and high symptomatic level (12).

The primary objective of MAPS was to demonstrate if significant improvement in OS was achieved by adding bevacizumab to standard first-line chemotherapy (cisplatin plus pemetrexed; ref. 5) in advanced MPM. HRQoL was a secondary endpoint of the trial. This was analyzed on the modified intent-to-treat (mITT) population, defined as all intent-to-treat patients for whom at least one baseline HRQoL score was available.

The analysis was conducted according to the patient-reported outcome Consolidated Standards of Reporting Trials (CONSORT) statement (13). Each dimension of the QLQ-C30 and QLQ-LC13 was studied. Baseline characteristics including HRQoL scores of the patients were described by treatment arms in the mITT population. Qualitative variables were described using frequencies and percentages and compared using a Chi-square or a Fisher exact test. Quantitative variables were described using mean and SD and compared using a T test or a nonparametric Mann–Whitney test.

HRQoL deterioration-free survival (QFS) was used as the modality of longitudinal analysis. It was defined as the interval between randomization and the occurrence of the first deterioration representing a minimal clinically important difference (MCID) of 5 points (14) compared with the HRQoL score at baseline, with no further improvement in HRQoL score ≥5 points as compared with baseline score, or death (15). Patients still alive at the last follow-up were censored if no significant definitive deteriorations in their HRQoL score were observed before. Patients with baseline scores but no follow-up score were censored immediately after baseline (Day 1). Only questionnaires completed at the planned measurement times according to protocol were included in the analysis. Thus, in case questionnaires had been completed on the visit at which progression was determined or after, they were excluded from the analysis.

The QFS curves were calculated using the Kaplan–Meier estimation method, described using medians with 95% CIs, and compared between treatment arms using log-rank tests. The univariate Cox model was applied to estimate HRs with their 95% CIs. We assessed the impact of following baseline variables in univariate analysis: treatment arm, age, gender, performance status, smoking status, weight loss, hemoglobin level, and histology. All variables with a univariate P-value <0.1 were eligible for a multivariate Cox regression model in order to identify factors independently associated with QFS. The variables for multivariate analysis were selected with consideration of colinearity between variables. The treatment arm variable was forced into the multivariate model.

In case of a missing questionnaire between 2 available measurement times, the last HRQoL scores obtained before the missing evaluation were used. It means that HRQoL level remains constant between 2 available assessment times. For missing items, according to recommendations of the EORTC scoring manual for calculating the scale scores, if at least half of the items from the scale had been answered, missing items were given values equal to the average of the present items (12). If there were not half of the items, the score was considered as missing. For single-item measures, the score was considered as missing. Compliance to HRQoL questionnaires was assessed by treatment arms, for each measurement time. In case of a difference in compliance between treatment arms, a comparison between completers and noncompleters was made according to baseline characteristics, PFS, and OS of noncompleters.

A P-value of 0.05 or lower was considered statistically significant. No adjustment for multiple testing was made since HRQoL was a secondary endpoint. All analyses were performed using SAS software, Version 9.3 (SAS Institute), and R software (Version 3.4) using the QoLR package (16).

A total of 448 patients with advanced MPM were included in the IFCT-GFPC-0701 MAPS study between February 2008 and January 2014. At baseline, 217 of 225 patients (96.4%) in the PC group had at least 1 HRQoL score available, versus 208 of 223 (93.3%) in the PCB group (Fig. 1). The baseline characteristics of these patients were well balanced. They comprised 324 males (76.2%), the mean age was 65.1 years (SD = 7), 410 exhibited performance status 0 to 1 (96.5%), 180 nonsmokers (42.4%), the mean weight loss was 3.6% (SD = 4), and 346 patients (81.4%) had epithelioid mesothelioma. The baseline HRQoL scores did not differ between arms. The mean score for global health status at baseline was 59 (SD = 20.1) and 60 (21.4), respectively (P = 0.52; Table 1).

Compliance to HRQoL questionnaires over time is presented in Fig. 1. The total numbers of questionnaires completed with at least 1 HRQoL score available were: 146 in 255 patients (57.25%) at 27 weeks, and 37 in 71 patients (52.11%) at 54 weeks. From week 27, compliance was higher in the PC arm. The difference was statistically significant at weeks 27 (P value Chi-square 0.003), 36 (P = 0.003), 45 (P = 0.01), and 72 (P = 0.004), but not at weeks 54 and 63, whereas the low numbers of available questionnaires precluded any calculation after week 72. There was no difference between compliant and noncompliant patients in the whole population, nor in the PCB arm, in terms of patient characteristics, baseline HRQoL scores, PFS and OS (Supplementary Table S1).

In univariate analysis, QFS was significantly longer in the PCB group for one HRQoL dimension: peripheral neuropathy (Fig. 2A and B). The median QFS of peripheral neuropathy or death was 7.59 months (95% CI, 6.57–8.61) in the PC group, and 12.09 months (95% CI, 9.59–13.67) in the PCB group [HR (PCB vs. PC) = 0.74; 95% CI, 0.61–0.91; P = 0.004; Fig. 3A]. QFS was longer with a P < 0.1 for pain or death (median: 9.79 months; 95% CI, 8.34–11.6), and 11.73 months (95% CI, 9.10–13.60), respectively [HR (PCB vs. PC) = 0.85; 95% CI, 0.69–1.03; P = 0.097; Fig. 3B].

Multivariate Cox analysis revealed that adding bevacizumab was independently associated with longer QFS for peripheral neuropathy (HR = 0.74; 95% CI, 0.60–0.90; P = 0.003). There was a trend towards improved QFS in the PCB arm for the pain dimension (HR = 0.84; 95% CI, 0.69–1.02; P = 0.08). Epithelioid histology was associated with a significantly longer QFS for all HRQoL dimensions except alopecia. No variable significantly influenced the QFS for alopecia.

Our analysis of the phase III IFCT-GFPC-0701 MAPS trial revealed that the increase in bevacizumab-related toxicities did not translate to HRQoL deterioration. Moreover, longitudinal analysis of HRQoL showed that the addition of bevacizumab to cisplatin and pemetrexed significantly improved QFS with a MCID of 5 points for 1 HRQoL dimension, peripheral neuropathy, with a clinically meaningful increase of 4 months in the median QFS (HR = 0.74; 95% CI, 0.61–0.91; P = 0.004), and a trend towards improved QFS for the pain dimension.

In a phase II trial of 53 patients treated with cisplatin-gemcitabine for MPM, HRQoL was evaluated using the same questionnaires as in our study. This study supported the validity of the EORTC QLQ-C30 and QLQ-LC13 questionnaires as outcomes for MPM. Pain and coughing were significantly improved with chemotherapy (10). There are no guidelines currently recommending the use of a specific questionnaire for HRQoL studies for patients with MPM. The most frequently used questionnaires in the literature include the EORTC QLQ-C30 and QLQ-LC13, the Lung Cancer Symptom Scale (LCSS)-meso and the FACT-L questionnaires (17). The EORTC QLQ-C30 and the QLQ-LC13 questionnaires have been specifically validated in patients with MPM treated with chemotherapy by Nowak and colleagues (10). The LCSS-meso includes the same items as the LCSS except for the deletion of one item, hemoptysis, and was published in 2004 (18). A drawback of the LCSS is that it ignores components of HRQoL, such as the social and emotional aspects (19). The comparison between LCSS and other HRQoL instruments demonstrated a reasonably good reliability and validity for the other components. Actually, the LCSS-meso was also completed by patients in this study, but did not add any information (data not shown). The acceptability of the FACT-L questionnaire, although frequently used, has been compared with the EORTC questionnaire in patients with MPM and was shown to be significantly lower (20).

In a Phase III trial of cisplatin with or without raltitrexed in 250 patients with MPM, HRQoL analysis using the same questionnaires showed an improvement in dyspnea over time (21). In a series of 63 consecutive patients, including 58 patients treated with first-line platinum-pemetrexed and 15 patients who opted for best supportive care, analysis of HRQoL using the EORTC QLQ-C30 and QLQ-LC13 questionnaires revealed deterioration in global HRQoL, dyspnea, and pain in the best supportive care group (22). The trend observed in our study with a nonsignificant improvement of 2 months in the median QFS for pain or death in the PCB arm may indicate further improvement of a key symptom with the addition of bevacizumab, and is understandable given the context of improved response and progression-free survival in the PCB arm (5). The relationship between peripheral neuropathy and bevacizumab is, however, puzzling, with increased median QFS over 4 months observed for peripheral neuropathy in the PCB arm. There was no difference between the groups in terms of physician-reported frequency and peripheral neuropathy severity (5). In non–small cell lung cancer, adding bevacizumab to carboplatin-paclitaxel, a more neurotoxic chemotherapy regimen compared with cisplatin-pemetrexed, did not influence the risk of peripheral neuropathy (4). Bevacizumab has been occasionally reported to improve neuropathy symptoms in polyneuropathy, organomegaly, endocrinopathy, M-protein, and skin changes (POEMS) syndrome (23). Recently, bevacizumab was proven protective against experimental diabetic neuropathy by its capacity to neutralize Schwann cell-secreted VEGF, restoring neurite outgrowth (24).

An important challenge of the analysis of HRQoL that could influence our results is the multiplicity of tests performed, due to the multidimensionality of HRQoL. Because HRQoL was a secondary endpoint in this study, no adjustment on multiple testing was performed. Moreover, due to the number of dimensions analyzed (25 dimensions), a Bonferroni adjustment would be too much conservative. To date, there is no satisfying solution for this issue, except to plan to target specific HRQoL dimensions at the time of study design (25).

In our trial, compliance levels were high at baseline, with around 5% missing data. The quality of our longitudinal analysis consisted of frequent HRQoL evaluations covering a long period. Long-term compliance was good, with over 50% patients answering the questionnaires at 54 weeks, which compares favorably with published compliance rates in MPM (10, 21, 22). Compliance to HRQOL questionnaires over time was better in the PC arm with a significant difference since 27 weeks. In the literature, it is generally admitted that noncompliant patients are usually the ones with deterioration in their HRQoL (26). To the best of our knowledge, imputation techniques for missing data are only exploratory and not validated or recommended yet, for longitudinal HRQoL calculations using survival analyses such as QFS. Because QFS was calculated with death considered as an event, QFS allows capturing the HRQoL deterioration of nonresponders when such patients are close to death. There was no significant difference between compliant and noncompliant patients at week 27 neither in baseline characteristics nor in PFS or OS. Although we cannot exclude that this discrepancy in compliance between both groups from week 27 could have influenced the results of our HRQoL analysis, we therefore consider that it is unlikely.

Increased cardiovascular adverse events, including low-grade hemorrhage, hypertension, and thromboembolic events reported in the PCB arm did not translate into HRQoL impairment. Except for hemoptysis, which is a rare event in pleural mesothelioma, these symptoms were not evaluated by a specific HRQoL dimension. The more general HRQoL dimensions such as overall health status, physical functioning, or fatigue are probably not sensitive enough to capture these events. Moreover, a proportion of patients who experienced these bevacizumab-induced adverse events might have been asymptomatic.

In conclusion, not only has adding bevacizumab to standard first-line chemotherapy regimen (cisplatin-pemetrexed) proven able to enhance progression-free survival and OS, it also caused no deterioration in HRQoL. The result was quite the contrary, with a trend towards improved QFS for pain and a significantly longer QFS for peripheral neuropathy; with the latter effect being potentially the consequence of an actual biological effect of bevacizumab.

A. Anota reports receiving speakers bureau honoraria from Roche and other remuneration from Bristol-Myers Squibb and AstraZeneca. A. Scherpereel is a consultant/advisory board member for Roche, Bristol-Myers Squibb, MSD, and AstraZeneca. C. Audigier-Valette is a consultant/advisory board member for Roche and Lilly. D. Moro-Sibilot is a consultant/advisory board member for Lilly, Roche, Bristol-Myers Squibb, MSD, AstraZeneca, Boehringer, Pfizer, Takeda, and Novartis. O. Molinier is a consultant/advisory board member for Bristol-Myers Squibb, Boehringer Ingelheim, and AstraZeneca. H. Lena is a consultant/advisory board member for Lilly and Roche. V. Gounant is a consultant/advisory board member for Roche. G. Zalcman reports receiving speakers bureau honoraria from Roche and is a consultant/advisory board member for Roche and Lilly. V. Westeel reports receiving speakers bureau honoraria from Roche, Lilly, and Boehringer Ingelheim, and is a consultant/advisory board member for and reports receiving other remuneration from Roche and Boehringer Ingelheim. No potential conflicts of interest were disclosed by the other authors.

Conception and design: G. Eberst, G. Zalcman, V. Westeel

Development of methodology: G. Eberst, A. Anota, J. Mazieres, G. Zalcman, V. Westeel

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Eberst, A. Scherpereel, J. Mazieres, J. Margery, L. Greillier, D. Moro-Sibilot, H. Léna, F. Rivière, I. Monnet, V. Gounant, R. Gervais, C. Locher, F. Morin, G. Zalcman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Eberst, A. Anota, A. Scherpereel, J. Mazieres, L. Greillier, E. Charton, F. Morin, G. Zalcman, V. Westeel

Writing, review, and/or revision of the manuscript: G. Eberst, A. Anota, A. Scherpereel, J. Mazieres, L. Greillier, C. Audigier-Valette, D. Moro-Sibilot, O. Molinier, H. Léna, F. Rivière, H. Janicot, E. Charton, F. Morin, G. Zalcman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G. Eberst, D. Moro-Sibilot, F. Morin, G. Zalcman, V. Westeel

Study supervision: G. Eberst, A. Anota, A. Scherpereel, D. Moro-Sibilot, F. Morin, G. Zalcman, V. Westeel

This work was supported by the Intergroupe Francophone de Cancérologie Thoracique (IFCT: French Cooperative Thoracic Intergroup) and received an unrestricted grant from the French Ministry of Health (Programme Hospitalier de Recherche Clinique, 2007) and F. Hoffmann-La Roche. F. Hoffmann-La Roche also supplied bevacizumab but played no role in trial design, data collection, data interpretation, discussion, or manuscript writing.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.