Health-related quality of life in the randomized phase 3 study of ramucirumab plus docetaxel versus placebo plus docetaxel in platinum-refractory advanced urothelial carcinoma (RANGE)

The RANGE trial design and the outcome of its primary endpoint were previously published [2]. The trial was conducted during July 2015 and April 2017 and approved by appropriate review boards/ethics committees and followed the principles outlined in the Declaration of Helsinki. All patients provided written informed consent. This study adhered to the CONSORT guidelines.

Eligibility criteria included patients with Eastern Cooperative Oncology Group (ECOG) performance status of zero or one and whose advanced disease progressed during or ≤ 14 months after first-line platinum-based chemotherapy. Patients may have also received one checkpoint inhibitor therapy prior to enrollment onto RANGE, in which case prior platinum-containing chemotherapy in ≤ 24 months was allowed. Patients were excluded if they had received a taxane previously [2].

Eligible patients were randomized to receive docetaxel (75 mg/m², intravenously [IV]) plus ramucirumab (10 mg/kg, IV) or docetaxel plus placebo, administered on day 1 of each 21-day cycle. Randomization was stratified by geographic region (North America, East Asia, or Europe/other), ECOG performance status at baseline (0 or 1), and visceral metastasis (present or absent). Ramucirumab treatment continued until there was documented disease progression, toxicity or intolerance requiring discontinuation, withdrawal of consent, or noncompliance. Docetaxel could continue, if no prespecified discontinuation criteria were met, for up to six cycles with an additional four cycles allowed if an adequate treatment response was observed. Tumor responses were assessed using Response Evaluation Criteria in Solid Tumors (RECIST) v1.1 [13].

Assessment of PROs, a secondary endpoint of the RANGE trial, was conducted through use of the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire C30 (EORTC QLQ-C30), version 3 and the EuroQoL five-dimensions, 5 level (EQ-5D-5L) questionnaire, which measure QoL (functional domains and symptoms) and health status, respectively [14, 15]. Patients completed the questionnaires only if there were cross-culturally validated translations in which they were fluent. Prior to any extensive contact with clinical personnel, patients were asked to complete the questionnaires at baseline, prior to the start of each cycle, and within 30 days after treatment discontinuation (follow-up). QLQ-C30 data were scored according to the EORTC guidelines, using a 0–100 scale with higher scores for global QoL and functional scales representing better QoL and lower scores for symptom scales representing less burden. A ≥ 10-point change was considered clinically meaningful [16]. EQ-5D-5L index values were based on the value set for England with a range of -0.281 to 1, with zero representing death and one representing perfect health [17]. Additionally, patients indicated their current health status by marking on a visual analogue scale (VAS) ranging from 0 (worst-imaginable health state) to 100 (best-imaginable health state) [15].

Analyses were conducted in the intention-to-treat (ITT) population except for descriptive statistics that were limited to the number of patients who provided data at a given assessment. Compliance rates were calculated based on the number of patients expected to provide data at a given assessment (i.e., those patients still receiving study therapy). Descriptive statistics were used to summarize data, including change from baseline as well as for the EQ-5D-5L scores. QLQ-C30 scores were classified as improved or worsened if change from baseline was ≥ 10 points; change < 10 points was classified as stable. For each scale at each postbaseline assessment, the proportion of patients with improved or stable scores was compared using the Fisher’s exact test. Time to sustained deterioration (TtD) was defined as time from randomization date to first worsening of ≥ 10 points with no subsequent non-worsened assessment relative to baseline. If there were no subsequent assessments, the patient was classified as deteriorated. The follow-up assessment after treatment discontinuation could not be considered as a subsequent non-worsened assessment. Patients without deterioration were censored at the last non-deteriorated assessment. Kaplan–Meier method was used to estimate the probability of TtD, and unstratified log-rank test was used to investigate significance between treatment groups. Univariable Cox regression analysis was performed to test the association of treatment with TtD. No adjustments were made for multiplicity, but p-values < 0.05 were used to identify potential trends. These analyses discussed here were performed on the ITT population at the time of the OS database lock [3]. All the analyses were conducted using SAS software (SAS, Version 9.1.2 or higher).

In the randomly assigned patients (N = 530, ITT population) from the 727 screened for eligibility (Fig. 1), baseline characteristics were similar between treatment arms (Table 1). At baseline, 254 (96.6%) of patients in the ramucirumab arm and 260 (97.4%) in the placebo arm provided QLQ-C30 and EQ-5D-5L data (Table 2). For the first 8 cycles, compliance was ≥ 88% for both the QLQ-C30 and EQ-5D-5L, (Table 2). The most common reasons for noncompliance were failure by the site to administer and patient refusal to complete the questionnaires (Fig. 1). At the post-discontinuation follow-up visit, compliance was lower overall, with 52–54% of patients providing data. When grouped by those who provided data at follow-up and those who did not, the groups were similar in baseline patient characteristics, time on therapy, BOR, and reasons for discontinuation.

Mean baseline scores for both questionnaires were similar between treatment arms (Table 3). With high scores being favorable for global QoL and functional domains of the QLQ-C30, the lowest mean scores were reported for global QoL, indicating the domain with greatest impairment at baseline. With low scores being favorable for symptoms, the highest mean scores were reported for fatigue, pain, and insomnia, indicating the symptoms of greatest burden at baseline.

When considering high-level assessment of QoL and health status, no clear differences were observed between the treatment arms for changes in mean scores from baseline of the QLQ-C30 global QoL scale, EQ-5D-5L index, and EQ-5D-5L VAS (Fig. 2a–c). Although scores within arms worsened over time, the mean changes for on-treatment assessments were small. Similar patterns of no differences between arms in changes from baseline were observed for the other QoL scales, with the exception of pain (Additional file 1: Figure S1). For the functional scales, worsened scores over time were observed for physical and role functioning. For symptom scales, fatigue, dyspnea, and diarrhea had the greatest worsening over time. In the case of pain, improved scores were observed in the ramucirumab arm over the first five cycles, while scores worsened in the placebo arm. Additional file 1: Figure S2 summarizes the distribution of EQ-5D-5L responses over time. At baseline, the highest levels of impairment (i.e., moderate or more severe) were reported for pain/discomfort. In general, the distributions of responses for all dimensions were similar over time for on-treatment assessments (Additional file 1: Figure S2A-E). For all scales, change from baseline was worst at the post-discontinuation follow-up assessment. Thus, in general, findings of the QLQ-C30 and the EQ-5D-5L were consistent.

Results for TtD of QLQ-C30 scales are summarized in Fig. 3. The HR was < 1 for all scales (range 0.76–0.96), indicating a trend towards longer TtD in the ramucirumab arm. Kaplan–Meier figures are presented for those scales with ˂65% censoring of events (Additional file 1: Figure S3A-I). Median TtD for global QoL was 6.9 months (95% CI 4.2 months—8.9 months) vs. 4.6 months (95% CI 3.5 months—5.5 months) (HR 0.88, 95% CI 0.7–1.2).

The proportion of patients whose QoL scores improved, remained stable, or deteriorated from baseline at each postbaseline assessment are shown in Fig. 4, for the scales with the greatest impairment at baseline; Additional file 1: Figure S4 depicts these results for all other scales. For each scale, rates of improved/stable were similar between treatment arms in early cycles. Of note, a higher proportion of patients in the ramucirumab arm had either improved or stable pain scores at cycles 4, 5, and 7 compared to the placebo arm (Fig. 4d). Indeed, across all cycles, the pain scale, relative to the other scales (Fig. 3a and Additional file 1: S4), exhibited consistency with respect to having the greatest proportion of patients with improvement in pain within the ramucirumab arm.

Considering all prespecified analyses, the consistency of the pain results prompted an exploratory analysis evaluating the association with best overall response to treatment (Fig. 5). Because the global QoL scale provides the most holistic assessment, it was similarly evaluated. During the first four cycles, the range of patients achieving complete response (CR)/partial response (PR) with improved global QoL was 17.6–25.0% for ramucirumab and 10.8–21.6% for placebo; the range for patients with stable disease (SD) was 9.6–17.3% for ramucirumab and 10.0–21.8% for placebo (Fig. 5a). During the first four cycles, the range of patients achieving CR/PR with improved pain was 29.4–30.9% for ramucirumab and 13.5–27.0% for placebo; the range for patients with SD was 17.3–28.8% for ramucirumab and 16.4–29.1% for placebo (Fig. 5b).

In general, a higher proportion of patients who achieved CR/PR had improved pain scores in the ramucirumab arm versus the placebo arm. No apparent difference in pain palliation was seen between arms for patients who achieved SD. Similar results were observed for improved global QoL between ramucirumab versus placebo arms in patients who achieved CR/PR and SD (Fig. 5).