Abstract
Background
In the intent‐to‐treat (ITT) population of the RAINBOW study, objective response rate (ORR) was 28% and 16% in the ramucirumab and control arms, respectively. To further characterize tumor response, we present details on timing and extent of tumor shrinkage, as well as associations with symptom palliation.
Materials and Methods
Tumor response was assessed with RECIST v1.1, and quality of life (QoL) was assessed with the European Organization for Research and Treatment of Cancer Quality‐of‐Life Questionnaire‐Core 30 (EORTC QLQ‐C30) v3.0. Prespecified and post hoc analyses were conducted in the ITT population, patients with measurable disease, or responders, and included best overall response (BOR), ORR, disease control rate (DCR), duration of response, time to response (TtR), change in tumor size, and associations of QoL with tumor shrinkage and BOR.
Results
In both treatment arms, median TtR was 1.5 months. Responses were more durable in the ramucirumab versus control arm (median 4.4 vs. 2.8 months). In patients with measurable disease (78% of ITT), ORR was 36% versus 20%; DCR was 81% versus 61% in the ramucirumab versus control arms. Waterfall plots demonstrated more tumor shrinkage in the ramucirumab versus control arm. Regardless of treatment, tumor response and stable disease were associated with improved or stable QoL, with more tumor shrinkage associated with greater symptom palliation.
Conclusion
Treatment with ramucirumab plus paclitaxel yielded the highest ORR reported to date for patients with previously treated advanced gastric or gastroesophageal junction adenocarcinoma. Additional details demonstrate robustness of tumor response results. The extent of tumor shrinkage is directly associated with symptom palliation and should be considered when evaluating patient needs and treatment selection. Clinical trial identification number. NCT01170663.
Implications for Practice
Ramucirumab plus paclitaxel is a recognized standard of care as it improves survival for patients with advanced gastric or gastroesophageal junction adenocarcinoma who have been previously treated with recommended first‐line therapy. These additional data on tumor response demonstrate a positive association between tumor shrinkage and symptom palliation in a patient population that is often symptomatic. These observations included patients with nonmeasurable disease, a group of patients often underrepresented in clinical trials. This knowledge can inform treatment decisions, which align individual patient characteristics and needs with demonstrated benefits.
Keywords: Ramucirumab, Paclitaxel, Gastric cancer, Gastroesophageal junction adenocarcinoma
Short abstract
To further characterize tumor response and disease‐related symptoms from the RAINBOW study, this article presents details on timing and extent of tumor shrinkage and the association of response with symptom palliation. The reported results provide additional context for considering second‐line options for patients with advanced gastric/gastroesophageal cancer.
Introduction
Gastric cancer is the fifth most common cancer worldwide, with an annual incidence of more than 1,000,000. With an estimated 783,000 deaths in 2018, it is the third most common cause of cancer‐related death [1]. In the first‐line setting, chemotherapy and/or chemoradiation may extend survival and improve quality of life (QoL) of patients with locally advanced, metastatic, and/or unresectable gastric or gastroesophageal junction (GEJ) adenocarcinoma, but durable responses are rare [2]. Even with the advances in chemotherapy in the last 20 years and the introduction of anti‐HER2 therapy for patients with HER2‐positive disease, patients with locally advanced or metastatic gastric or GEJ cancer have a median survival of less than 2 years [2].
With initial therapy for advanced disease, extending survival is generally the primary goal. However, as disease progresses, symptom palliation and preservation of QoL gain prominence as additional goals of treatment. Control of tumor progression is likely key in symptom control. A recent study of QoL data from two large phase III studies in patients previously treated for gastric or GEJ cancer confirmed that disease control is important for maintaining or improving QoL [3]. Chau et al. observed that fatigue, pain, and appetite loss were the most prominent patient‐reported baseline symptoms. For patients who achieved an objective response, scores improved for global QoL, emotional functioning, pain, appetite loss, and nausea and/or vomiting. Patients achieving stable disease maintained their QoL with treatment. Disease progression resulted in worsening QoL across all functional domains and most symptoms [3]. The negative QoL impact of disease progression has been reported for other tumor types [4], but reports on QoL and symptom palliation with tumor response are more limited.
Although tumor response criteria are generally standardized across clinical trials, differences in eligibility criteria and analysis populations may impact applicability of some results to the clinical practice setting. Advanced gastric cancer is often accompanied by peritoneal metastases or ascites, which are considered nonmeasurable by the RECIST standards [5]. In clinical studies in which response is the primary endpoint, patients with only nonmeasurable disease are excluded, and all eligible patients are expected to be included in the response calculation [5]. When response is a secondary endpoint and nonmeasurable‐only disease is not an exclusion criterion, the protocol must prespecify the analysis population for calculating response rates [5]. Thus, depending on the study, patients with only nonmeasurable disease may be excluded from secondary response analyses despite being included in the intent‐to‐treat (ITT) population [5, 6]. Consequently, it would be unclear if patients with only nonmeasurable disease who were excluded from the response analyses derive any of the benefits of disease control from the tested interventions. Reports of response calculated in this manner may limit the applicability to the real‐world clinical population.
Patients with advanced gastric cancer whose disease progresses on or after first‐line therapy typically have low response rates to subsequent lines of treatments. In the second‐line setting, response rates from 0% to 21% have been reported for chemotherapy regimens without targeted agents recommended by current guidelines for advanced or metastatic disease [7, 8, 9, 10, 11, 12]. Among these studies, one study excluded patients without measurable disease [8], and three calculated response based on a subset of patients [7, 10, 11].
In the global phase III RAINBOW trial, ramucirumab plus paclitaxel improved overall survival (OS) and progression‐free survival (PFS) in patients with advanced, fluoropyrimidine‐ and platinum‐resistant gastric or GEJ adenocarcinoma versus placebo plus paclitaxel [13]. As a tumor‐based outcome, the hazard ratio for PFS was 0.635 (95% confidence interval, 0.536–0.752), with early separation of the Kaplan‐Meier curves and 6‐week PFS rates of 88% in the ramucirumab arm and 75% in the placebo arm. Ramucirumab plus paclitaxel demonstrated the highest objective response rate (ORR) reported in the second‐line setting for an unselected patient population, with an ORR of 28% versus 16% in the ramucirumab versus placebo arm (p = .0001) [13]. This is noteworthy because the trial enrolled patients with RECIST v1.1 evaluable disease, regardless of whether it was measurable or nonmeasurable only (approximately 20% of those enrolled in each arm for the latter), and ORR was estimated in the ITT population [13]. RAINBOW also demonstrated that the addition of ramucirumab to paclitaxel maintained patient‐reported QoL while delaying symptom worsening and functional status deterioration [14]. Ramucirumab is approved worldwide in combination with paclitaxel and as monotherapy for previously treated gastric or GEJ adenocarcinoma.
To further characterize the relationship between tumor response and symptoms from the RAINBOW study, we present details on timing and extent of tumor shrinkage, and the association of response with symptom palliation. Published data examining the benefits of disease control and tumor reduction in relation to QoL and symptom control in advanced gastric/GEJ cancer are limited. Results from this study will provide additional context when considering second‐line options for patients with advanced gastric/GEJ cancer.
Materials and Methods
Study Design
Details for the RAINBOW trial have been published elsewhere [13]. Eligibility included patients 18 years of age and older with unresectable or metastatic disease that progressed on or within 4 months of last dose of first‐line therapy and who had an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 or 1. Patients were randomized 1:1 to receive paclitaxel intravenously on days 1, 8, and 15 in combination with either ramucirumab (8 mg/kg) or placebo administered intravenously on days 1 and 15 of a 28‐day cycle. Randomization was stratified by geographic region (region 1, Europe, Australia, and the U.S.; region 2, rest of world including South America; and region 3, East Asia), time to progression after first dose of first‐line therapy (<6 months vs. ≥6 months), and disease measurability (measurable vs. nonmeasurable only). The primary outcome was OS in the ITT population. Secondary outcomes included ORR and QoL. The ITT population consisted of 330 and 335 patients treated in the ramucirumab and control arms, respectively (supplemental online Fig. 1; supplemental online Table 1).
Assessments
Images of the chest, abdomen, and pelvic areas were obtained (computed tomography scan or magnetic resonance imaging) on all patients at baseline and every 6 weeks for the first 6 months and every 9 weeks thereafter. Tumor response was assessed according to the RECIST version 1.1 [5]. Quality of life was assessed using the European Organisation for Research and Treatment of Cancer QLQ‐C30 (version 3.0), which patients completed at baseline, every 6 weeks, and at end of therapy.
Prespecified Analyses
ORR in the ITT population was defined as the proportion of patients who achieved a best overall response (BOR) of complete response (CR) or partial response (PR). Disease control rate (DCR) was defined as the proportion of patients in the ITT population who achieved a BOR of CR, PR, or stable disease (SD). Additional analyses for those with measurable disease included ORR and DCR estimations.
Duration of response (DOR) was measured in the ITT population and defined as the time from CR or PR until progressive disease (PD) or death. DOR was compared between arms with an unstratified log‐rank test, and median DOR was estimated using the Kaplan‐Meier method.
For each patient in the ITT population, changes from baseline in QLQ‐C30 scores were calculated for each postbaseline assessment. Changes of ≥10 points (on 100‐point scale) were considered meaningful [15]. Changes were classified as improved if ≥10 points favorable, worsened if ≥10 points deteriorated, or stable if <10 points. Best change from baseline was also calculated and classified. Proportions of patients with improved or stable scores were compared between arms using Fisher's exact test.
Post Hoc Analyses
Baseline patient and disease characteristics were reported by BOR and by disease measurability. Waterfall plots were constructed to display the best percentage change in tumor size for each patient by treatment arm. Because a waterfall plot uses a calculation of change from baseline, the analysis is limited to patients with measurable disease who had both a baseline and at least one postbaseline assessment. Fatigue, pain, and appetite loss were previously identified as the three most prominent patient‐reported baseline symptoms for previously treated patients with advanced gastric or GEJ cancer [3]. Thus, the association of best change from baseline for these three symptoms with best percentage change in tumor size was explored by plotting symptom improvement for patients included in the waterfall plot.
Descriptive statistics were used to estimate the median time to response, defined as time from randomization to first response estimated for patients with a BOR of CR or PR. A swimmer's plot was constructed to describe the following outcomes for each individual patient with BOR of CR or PR: time to objective response, duration of response, and time on study.
Best change in QoL scores of fatigue, pain, and appetite loss, and global QoL were plotted by BOR for the ITT population. Within each BOR group, proportions of patients with improved or stable scores were compared between arms.
Results
Baseline Characteristics and Tumor Response
As previously reported in the RAINBOW ITT population, the ramucirumab arm had an ORR of 28%; 2 (<1%) patients had a CR and 90 (27%) patients had a PR. Additionally, 172 (52%) patients achieved a BOR of SD, resulting in a DCR of 80%. The control arm had an ORR of 16%, 1 (<1%) patient had a CR, and 53 (16%) patients had a PR; 159 (47%) patients had a BOR of SD, for a DCR of 64% [13]. Tables 1 and 2 show baseline characteristics for patients by BOR. In the BOR groups of CR/PR or SD, a higher proportion of patients were Asian or from East Asia compared with the PD group (independent of treatment). Relative to those in the SD and PD groups, patients in the CR/PR group all had measurable disease; they also had a higher proportion of liver metastasis and a lower proportion of ascites and peritoneal disease, regardless of treatment arm. Examination of those achieving BOR of CR/PR in each treatment group revealed patient and tumor characteristics were generally equally distributed, although there were higher percentages of patients with ECOG PS 1 and ascites in the ramucirumab arm. In patients with a BOR of SD, 67% had measurable disease in each treatment arm. For those with a BOR of PD, 72% in the ramucirumab arm had measurable disease versus 87% in the control arm.
Table 1.
Patient and disease characteristics by best overall response
| Characteristics | CR/PR | SD | PD | NE/other | ||||
|---|---|---|---|---|---|---|---|---|
| RAM + PAC (n = 92) | PBO + PAC (n = 54) | RAM + PAC (n = 172) | PBO + PAC (n = 159) | RAM + PAC (n = 43) | PBO + PAC (n = 83) | RAM + PAC (n = 23) | PBO + PAC (n = 39) | |
| Male sex, n (%) | 73 (79) | 45 (83) | 110 (64) | 108 (68) | 25 (58) | 61 (73) | 21 (91) | 29 (74) |
| Median age (range), yr | 61 (30–77) | 63 (28–78) | 60 (25–81) | 63 (30–84) | 61 (35–78) | 59 (26–77) | 60 (43–83) | 58 (24–81) |
| Race, n (%) | ||||||||
| White | 55 (60) | 27 (50) | 104 (60) | 89 (56) | 32 (74) | 55 (66) | 17 (74) | 28 (72) |
| Asian | 37 (40) | 25 (46) | 60 (35) | 64 (40) | 8 (19) | 25 (30) | 5 (22) | 7 (18) |
| Other | 0 | 2 (4) | 8 (5) | 6 (4) | 3 (7) | 3 (4) | 1 (4) | 4 (10) |
| Geographic region, n (%) | ||||||||
| Eur/N Am/Aus | 53 (58) | 26 (48) | 99 (58) | 87 (55) | 32 (74) | 57 (69) | 14 (61) | 30 (77) |
| Rest of world including S Am | 2 (2) | 5 (9) | 13 (8) | 10 (6) | 4 (9) | 3 (4) | 4 (17) | 3 (8) |
| East Asia | 37 (40) | 23 (43) | 60 (35) | 62 (39) | 7 (16) | 23 (28) | 5 (22) | 6 (15) |
| ECOG PS, n (%) | ||||||||
| 0 | 34 (37) | 29 (54) | 67 (39) | 76 (48) | 15 (35) | 26 (31) | 1 (4) | 13 (33) |
| 1 | 58 (63) | 25 (46) | 105 (61) | 83 (52) | 28 (65) | 57 (69) | 22 (96) | 26 (67) |
|
Primary tumor location, n (%) |
||||||||
| Gastric | 70 (76) | 43 (80) | 141 (82) | 133 (84) | 34 (79) | 57 (69) | 19 (83) | 31 (79) |
| GEJ | 22 (24) | 11 (20) | 31 (18) | 26 (16) | 9 (21) | 26 (31) | 4 (17) | 8 (21) |
| Disease measurability a , n (%) | ||||||||
| Measurable | 92 (100) | 54 (100) | 115 (67) | 107 (67) | 31 (72) | 72 (87) | 18 (78) | 32 (82) |
| Nonmeasurable Only | 0 | 0 | 57 (33) | 52 (33) | 12 (28) | 11 (13) | 5 (22) | 6 (15) |
| Time to PD on 1L <6 mo, n (%) | 50 (54) | 34 (63) | 111 (65) | 96 (60) | 31 (72) | 45 (54) | 16 (70) | 25 (64) |
| Weight loss over prior 3 mo ≥10%, n (%) | 17 (18) | 7 (13) | 15 (9) | 20 (13) | 11 (26) | 14 (17) | 10 (43) | 6 (15) |
| Ascites present, n (%) | 26 (28) | 8 (15) | 74 (43) | 63 (40) | 20 (47) | 21 (25) | 10 (43) | 15 (38) |
As reported in case report forms.
Abbreviations: 1L, first‐line; Aus, Australia; CR, complete response; ECOG PS, Eastern Cooperative Oncology Group performance status; Eur, Europe; GEJ, gastroesophageal junction; N Am, North America; NE, not evaluable (missing baseline and/or postbaseline tumor assessment); ORR, objective response rate; PAC, paclitaxel; PBO, placebo; PD, progressive disease; PR, partial response; RAM, ramucirumab; S Am, South America; SD, stable disease.
Table 2.
Additional tumor characteristics by best overall response
| Characteristics | CR/PR | SD | PD | NE/other | ||||
|---|---|---|---|---|---|---|---|---|
| RAM + PAC (n = 92) | PBO + PAC (n = 54) | RAM + PAC (n = 172) | PBO + PAC (n = 159) | RAM + PAC (n = 43) | PBO + PAC (n = 83) | RAM + PAC (n = 23) | PBO + PAC (n = 39) | |
| Histological/ pathological type, n (%) | ||||||||
| Intestinal | 43 (47) | 20 (37) | 73 (42) | 59 (37) | 22 (51) | 42 (51) | 7 (30) | 14 (36) |
| Diffuse | 24 (26) | 21 (39) | 66 (38) | 72 (45) | 15 (35) | 25 (30) | 10 (43) | 15 (38) |
| Mixed | 12 (13) | 5 (9) | 7 (4) | 5 (3) | 1 (2) | 2 (2) | 1 (4) | 2 (5) |
| Unknown | 13 (14) | 8 (15) | 26 (15) | 23 (14) | 5 (12) | 14 (17) | 5 (22) | 8 (21) |
| Grade, n (%) | ||||||||
| Well differentiated | 6 (7) | 2 (4) | 19 (11) | 9 (6) | 3 (7) | 9 (11) | 0 | 2 (5) |
| Moderately differentiated | 29 (32) | 22 (41) | 47 (27) | 45 (28) | 11 (26) | 29 (35) | 9 (39) | 10 (26) |
| Poorly differentiated | 52 (57) | 26 (48) | 96 (56) | 98 (62) | 27 (63) | 39 (47) | 11 (48) | 23 (59) |
| Number of metastatic sites, n (%) | ||||||||
| 0 | 0 | 1 (2) | 0 | 5 (3) | 1 (2) | 0 | 0 | 1 (3) |
| 1 | 22 (24) | 16 (30) | 50 (29) | 46 (29) | 9 (21) | 23 (28) | 4 (17) | 6 (15) |
| 2 | 35 (38) | 19 (35) | 68 (40) | 65 (41) | 15 (35) | 32 (39) | 5 (22) | 18 (46) |
| ≥ 3 | 35 (38) | 18 (33) | 54 (31) | 43 (27) | 18 (42) | 28 (34) | 14 (61) | 14 (36) |
| Sites of metastases, n (%) | ||||||||
| Lung | 19 (21) | 17 (31) | 34 (20) | 28 (18) | 14 (33) | 17 (20) | 10 (43) | 8 (21) |
| Liver | 57 (62) | 33 (61) | 62 (36) | 44 (28) | 16 (37) | 45 (54) | 15 (65) | 16 (41) |
| Bone | 7 (8) | 2 (4) | 10 (6) | 11 (7) | 3 (7) | 9 (11) | 5 (22) | 6 (15) |
| Lymph nodes | 70 (76) | 38 (70) | 108 (63) | 90 (57) | 24 (56) | 53 (64) | 13 (57) | 24 (62) |
| Pleural | 9 (10) | 3 (6) | 15 (9) | 17 (11) | 4 (9) | 9 (11) | 4 (17) | 5 (13) |
| Peritoneal | 33 (36) | 16 (30) | 93 (54) | 90 (57) | 26 (60) | 29 (35) | 11 (48) | 17 (44) |
| Other | 18 (20) | 4 (7) | 43 (25) | 36 (23) | 13 (30) | 20 (24) | 6 (26) | 16 (41) |
Abbreviations: CR, complete response; NE, not evaluable (missing baseline and/or postbaseline tumor assessment); ORR, objective response rate; PAC, paclitaxel; PBO, placebo; PD, progressive disease; PR, partial response; RAM, ramucirumab; SD, stable disease.
Additional Outcomes in Patients With Tumor Response
For those achieving an objective response, the median time to first response was 1.5 months (interquartile range, 1.4–2.8) for both treatment arms. The median duration of response was 4.4 months in the ramucirumab arm versus 2.8 months in the control arm (Fig. 1).
Figure 1.
Duration of response for patients with objective tumor response. Kaplan‐Meier analysis of the duration of response for patients from the RAINBOW study who had a complete or partial response in the ramucirumab (red line) versus the control (blue line) arms. Abbreviations: CI, confidence interval; HR, hazard ratio; PAC, paclitaxel; PBO, placebo; RAM, ramucirumab.
For patients with an objective response, regardless of treatment arm, more than half of patients achieved a tumor response at 6 weeks (53/92 [58%] in the ramucirumab arm and 33/54 [61%] in the control arm; Fig. 2). Among those who responded within 6 weeks, those treated with ramucirumab maintained a more durable response than those treated in the control arm. Sustained response at 6 months was 36% (19/53) in the ramucirumab arm and 18% (6/33) in the control arm. This trend continued at 12 months, with rates of 11% and 6% in the ramucirumab and control arms, respectively.
Figure 2.
Time to and duration of tumor responses for patients with an objective response. The swimmer's plot characterizes the time to and duration of response (weeks) to ramucirumab plus paclitaxel (n = 92) (left panel) or to placebo plus paclitaxel (n = 54) (right panel) for each patient who achieved a complete or partial response in the RAINBOW trial.
Similar trends were observed when considering all patients that had responded by 12 weeks. In the ramucirumab arm, 79 of 92 (86%) demonstrated response by 12 weeks similar to 47 or 54 (87%) in the control arm. After 6 months, 48% (38/79) versus 23% (11/47) had sustained response in the ramucirumab and control arms, respectively. This trend persisted at 12 months, with rates of 13% versus 6% in the ramucirumab and control arms, respectively.
Outcomes in Patients With Measurable Disease
From the ITT population, the proportion of patients with measurable disease was similar in the ramucirumab and control arms (78% [256/330] vs. 79% [265/335], respectively). When dichotomized by disease measurability, patient and disease baseline characteristics were similar between treatment arms; however, a higher proportion of patients with nonmeasurable‐only disease were female, Asian, and had ascites, whereas a lower proportion had GEJ disease (supplemental online Table 2).
Tumor Response and Disease Control Rates
In the subgroup of patients with measurable disease, both ORR and DCR were higher in the ramucirumab arm. The ORR was 36% (92/256) in the ramucirumab arm versus 20% (54/265) in the control arm. The DCR was 81% (207/256) in the ramucirumab arm and 61% (161/265) in the control arm.
Waterfall Plot of Best Percentage Change in Tumor Size
Figure 3 demonstrates the best percentage change from baseline in the sum of tumor diameters from the subgroup of patients with measurable disease that also had both baseline and postbaseline assessments. Thus, 238 of 330 (72%) patients from the ramucirumab arm and 236 of 335 (70%) patients from the control arm are included in the waterfall plot. Of the patients represented in the plot, a larger proportion in the ramucirumab arm experienced any tumor shrinkage compared with the control arm (185/238 [78%] vs. 141/236 [60%], respectively; Fig. 3).
Figure 3.
Best percentage change in tumor size from baseline. The waterfall plot shows the best percentage change in the sum of tumor diameters measured at baseline and reassessed at least once postbaseline. Blue bars denote a sum of tumor diameters that decreased less than 30% from baseline. Red bars denote a sum of tumor diameters that decreased ≥30% from baseline, which is categorized as a complete or a partial response per RECIST v1.1. n = 238 in the ramucirumab plus paclitaxel arm, and n = 236 in the placebo plus paclitaxel arm. Note: from the ramucirumab arm and the control arm in the intent‐to‐treat population, 92 of 330 (28%) and 99 of 335 (30%) patients were not included in this figure. The most frequent reason for exclusion was nonmeasurable disease in 74 of 330 (22%) patients in the ramucirumab arm and 69 of 335 (21%) in the control arm. The second most common reason for exclusion was no postbaseline tumor assessment in 17 of 330 (5%) ramucirumab patients and 28 of 335 (8%) control patients. Less than 1% of patients in each treatment arm had no baseline tumor assessment.
Association of Tumor Response and Quality of Life
Fatigue, pain, and appetite loss were previously identified as the three most prominent patient‐reported baseline symptoms for previously treated patients with advanced gastric or GEJ cancer [3]. Figure 4 presents the association between best percentage change in tumor size and best improvement in these selected symptoms, using the same patient population as in Figure 3. Baseline symptom levels were similar between arms (supplemental online Table 3). The number of nonevaluable patients (those without both baseline and postbaseline QoL) was low (18/330 [5%] in the ramucirumab arm, 30/335 [9%] in the control arm) and more common among patients with increased tumor size from the control arm. For pain and appetite loss, approximately two‐thirds of patients were symptomatic at baseline. Regarding fatigue, the vast majority (85%) of the patients were symptomatic at baseline, indicating the potential for improvement.
Figure 4.
Association of best percentage change in tumor size with best improvement in selected symptoms. Patient population is the same as that shown in the waterfall plots in Figure 3. Red bars are improved symptom scores, gray bars are stable or worsened symptom scores, white bars indicate patients who were not symptomatic at baseline, and black bars are patients who were not evaluable (without both a baseline and postbaseline assessment). Abbreviations: PAC, paclitaxel; PBO, placebo; RAM, ramucirumab.
Improvement in these three symptoms was more common among patients with tumor shrinkage and in the ramucirumab arm. For each symptom, approximately 40% of patients in the ramucirumab arm reported improvement and nearly 85% of these patients experienced tumor shrinkage. Approximately 30% of patients in the control arm reported symptom improvement and nearly 70% of these patients experienced tumor shrinkage. Of note, among those patients with a BOR of SD who were symptomatic at baseline, were evaluable for symptom change, and experienced tumor shrinkage, at least 60% reported improved pain and appetite loss, and more than 50% reported improved fatigue, regardless of treatment arm.
Figure 5 presents best change in global QoL and selected symptoms by BOR for each treatment arm. In both arms, for global QoL, nearly half of patients reported stable scores and nearly 25% reported improved scores. Those who improved most commonly had a BOR of CR/PR or SD. For each of the symptoms, improvement was reported in approximately 35%–40% of patients in the ramucirumab arm and 24%–31% of patients in the control arm, with almost all of these patients having a BOR of CR/PR or SD. For fatigue, approximately 25% of patients in each arm reported stable scores; regardless of treatment arm, approximately twice as many patients with a BOR of CR/PR reported improved versus stable fatigue scores. For pain, approximately 40% of patients in each arm reported stable scores; within each treatment arm, the numbers of patients with a BOR of CR/PR or of SD were similar between the improved and stable score groups. For appetite loss, more patients in each arm reported stable scores compared with improved scores; within the ramucirumab arm, the number of patients with a BOR of CR/PR or of SD was similar between the improved and stable score groups, but within the control arm, the number of patients with a BOR of CR/PR with stable scores was nearly twice that of those with CR/PR with improved scores. Less than 20% of patients reported deterioration as best change for global QoL, pain, and appetite loss. Among those whose symptoms worsened, very few had a BOR of CR/PR. Patients classified as nonevaluable most commonly had a BOR of PD or nonevaluable for each of these scales. Within each of the BOR groups, the largest numerical difference observed in proportions of patients with improved or stable scores was for pain in the CR/PR group (92% for ramucirumab vs. 80% for placebo, p = .0353).
Figure 5.
Best improvement of global QoL and selected symptoms by best overall tumor response. All patients from the intent‐to‐treat population with best change in global QoL or symptoms (fatigue, pain, appetite loss) classified as improved, stable, deteriorated, or not evaluable are shown by their best overall tumor response. Best overall tumor response is denoted with blue bars for a CR or PR, orange bars for SD, gray bars for PD, and yellow indicates NE or not available. Not evaluable indicates that baseline and/or postbaseline assessment was not available. Abbreviations: CR, complete response; NE, not evaluable; PAC, paclitaxel; PBO, placebo; PD, progressive disease; PR, partial response; QoL, quality of life; RAM, ramucirumab; SD, stable disease.
Discussion
To the best of our knowledge, the phase III RAINBOW study demonstrated the highest second‐line response rate to date in patients with advanced gastric or GEJ adenocarcinoma. This response rate was seen in a trial that included patients with either measurable or nonmeasurable‐only disease, which is more representative of a real‐world clinical patient population [13]. In the RAINBOW study subpopulation of patients with measurable disease, more than one‐third of patients in the ramucirumab arm achieved a tumor response, compared with one‐fifth in the control arm. The data in the analyses presented here provide additional context when evaluating other studies that entirely exclude patients with nonmeasurable‐only disease from the study or from tumor response analyses [7, 8, 9, 10, 11, 16, 17].
In prespecified and post hoc analyses of patients with objective responses in the RAINBOW study, just over half of the responses occurred within the first 6 weeks of the study treatment. By 12 weeks, just over 80% of patients in both arms had responded, with ongoing responses still observed at 6 and 12 months. The durability of responses was longer in the ramucirumab arm than with paclitaxel alone. The demonstrated improvements to OS and response rates were unlikely to be driven by the underperformance of the control arm in RAINBOW, as the observed results were in line with previous reports for paclitaxel [9, 10, 11, 13, 16].
Waterfall plots are a useful tool to visualize tumor shrinkage in a study, but there are limitations that need to be considered when interpreting data between treatment arms or across different clinical trials. Waterfall plots do not always represent the ITT population; both baseline factors and treatment outcomes will determine which patients are included in the waterfall plot population. They exclude patients who have only nonmeasurable disease at baseline and patients who discontinue the study early because of rapid disease progression or toxicity and thus may have no postbaseline tumor assessment. Furthermore, if patients are not stratified by disease measurability, or if there are differential rates of early discontinuation, the patients who are excluded may be disproportionate between the treatment arms. These factors may introduce a bias to the visual presentation of tumor shrinkage for a given treatment, potentially presenting a more favorable picture of a treatment by excluding those patients with worse outcomes [18]. To interpret data from clinical trials, one must consider the number of patients excluded and the associated reasons for exclusion. The majority of the patients excluded from the RAINBOW study waterfall plots presented here had nonmeasurable‐only disease at baseline, and only a small percentage lacked a postbaseline assessment; furthermore, these exclusions were mostly balanced between treatment arms. The majority of patients included in the waterfall plots experienced at least some degree of tumor shrinkage, regardless of treatment arm, even when defined as stable disease by RECIST.
The additional waterfall plots in this report suggest an association of tumor shrinkage with palliation of fatigue and pain, symptoms commonly observed in patients with advanced gastric or GEJ cancer. Not unexpectedly, greater symptom improvement was observed in patients achieving a complete or partial response, although this report demonstrates that symptom control, and even improvement, is achievable when the best response is only that of stable disease. Notably, more patients in the ramucirumab arm experienced tumor shrinkage, greater symptom control, and maintained QoL. These data are in alignment with previous reports demonstrating improved OS and PFS, acceptable toxicity, and delayed functional status deterioration from the addition of ramucirumab to paclitaxel [13, 14]. In addition, for time‐to‐deterioration analyses in QLQ‐C30 scores, hazard ratios for fatigue, pain, and appetite loss were among the most favorable for ramucirumab plus paclitaxel (<0.85; however, 95% confidence intervals included 1) [14].
Taken as a whole, these analyses show consistent findings of greater symptom palliation associated with tumor shrinkage. However, palliation was also observed in some patients who did not experience tumor shrinkage. Potential explanations include supportive care interventions and variable symptom manifestations of tumor burden, which could not be addressed in these analyses. Although tumor response may not necessarily correlate with improved survival, the benefits of tumor shrinkage include reduction in symptoms and maintenance or improvement in QoL [19]. A similar relationship between symptom palliation and tumor response has been observed in patients with metastatic breast cancer receiving chemotherapy [20].
Conclusion
Ramucirumab plus paclitaxel is a recognized standard of care for patients with advanced gastric or GEJ adenocarcinoma who have been previously treated with recommended first‐line therapy for advanced disease. In addition to the survival benefit with acceptable toxicity, these supplemental data demonstrate that patients with tumor response as well as those with stable disease experience improved or stable fatigue, pain, appetite loss, and global QoL with ramucirumab treatment, and tumor response was associated with more improvement. This information can inform treatment decisions, which align individual patient characteristics and needs with demonstrated benefits.
Author Contributions
Conception/design: Astra M. Liepa
Provision of study material or patients: Sefano Cascinu, Kei Muro, Eric Van Cutsem, Sang Cheul Oh, Gunnar Folprecht, Sumitra Ananda, Gustavo Girotto, Zev A. Wainberg, Maria Luisa Limon Miron
Collection and/or assembly of data: Sefano Cascinu, Kei Muro, Eric Van Cutsem, Sang Cheul Oh, Gunnar Folprecht, Sumitra Ananda, Gustavo Girotto, Zev A. Wainberg, Maria Luisa Limon Miron
Data analysis and interpretation: György Bodoky, Sumitra Ananda, Gustavo Girotto, Astra M. Liepa, Roberto Carlesi, Michael Emig, Maria Luisa Limon Miron, Jaffer Ajani, Atsushi Ohtsu, Eric Van Cutsem, Zev Wainberg, Ran Wei
Manuscript writing: Stefano Cascinu, György Bodoky, Kei Muro, Eric Van Cutsem, Sang Cheul Oh, Gunnar Folprecht, Sumitra Ananda, Gustavo Girotto, Zev A. Wainberg, Maria Luisa Limon Miron, Jaffer Ajani, Ran Wei, Astra M. Liepa, Roberto Carlesi, Michael Emig, Atsushi Ohtsu, Gunnar Folprecht
Final approval of manuscript: Stefano Cascinu, György Bodoky, Kei Muro, Eric Van Cutsem, Sang Cheul Oh, Gunnar Folprecht, Sumitra Ananda, Gustavo Girotto, Zev A. Wainberg, Maria Luisa Limon Miron, Jaffer Ajani, Ran Wei, Astra M. Liepa, Roberto Carlesi, Michael Emig, Atsushi Ohtsu
Disclosures
Stefano Cascinu: Celgene, Eisai (RF), Amgen, Eli Lilly & Co, Bayer, Baxter, Merck Sharpe & Dohme, Servier, Celgene, Novartis (C/A), Amgen, Bayer, Eli Lilly & Co, Servier (Other [speaker's fees]); György Bodoky: BVS, Janssen, Novartis, Pfizer, Roche (SAB), Roche, Pfizer, Servier, Janssen (Other); Kei Muro: ONO Pharmaceutical CO., LTD., Sanofi, Daiichi Sankyo, Parexel International, Shionogi Pharmaceutical, Sumitomo Dainippon Pharma, Merck Sharpe & Dohme, Pfizer, Mediscience Planning, Solasia Pharma, Merck Serono (RF), Bayer, Bristol‐Myers Squibb, Taiho Pharmaceutical, Takeda Pharmaceutical, Chugai Pharmaceutical, Eli Lilly & Co, ONO Pharmaceutical CO., LTD., Sanofi (Other); Eric Van Cutsem: Array, AstraZeneca, Bayer, Biocartis, Bristol‐Myers Squibb, Celgene, Daiichi, Halozyme, GlaxoSmithKline, Pierre‐Fabre, Incyte, Ipsen, Eli Lilly & Co, Merck Sharp & Dohme, Merck KGaA, Novartis, Roche, Servier, Sirtex, Taiho (SAB), Bayer, Boehringer Ingelheim, Celgene, Ipsen, Lilly, Roche, Merck Sharp & Dohme, Merck KGaA, Novartis, Roche, Servier (RF‐institution); Gunnar Folprecht: Merck, Roche, Bristol Myers Squibb, Merck Sharpe & Dohme, Pierre Fabre, Servier, Bayer, Amgen, Sanofi and Laboratoire HRA (Other [advisory board participation or lectures]); Merck (RF [institution]); Gustavo Girotto: Eli Lilly & Co, Merck Sharpe & Dohme, Bristol‐Myers Squibb, Merck and Boston Biomedical (RF); Zev A. Wainberg: Array BioPharma, Five Prime Therapeutics, Novartis, Eli Lilly & Co, Merck, Merck KGaA, Bristol‐Myers Squibb, Genentech, Bayer, and AstraZeneca/MedImmune (C/A), Novartis, Plexxikon, Pfizer, Merck, Five Prime Therapeutics, Eli Lilly & Co (RF); Astra M. Liepa: Eli Lilly & Co (E, OI), Roberto Carlesi: Eli Lilly & Co (E); Michael Emig Lilly Deutschland GmbH (E); Atsushi Ohtsu RF from BVS (RF), Bristol Myers Squibb, Ono, Taiho, Chugai (Other). The other authors indicated no financial relationships.
(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board
Supporting information
See http://www.TheOncologist.com for supplemental material available online.
Supplemental Figure 1 RAINBOW CONSORT Diagram
Supplemental Table 1 Analysis Populations
Supplemental Table 2. Baseline Characteristics by Disease Measurability
Supplemental Table 3. Baseline Symptom Level for Patients Included in Waterfall Plots
Acknowledgments
We thank the patients, their families, the study sites, and the study personnel who participated in this clinical trial. Eli Lilly and Company contracted with Syneos Health for writing support provided by Andrea D. Humphries, PhD, and editing support provided by Antonia Baldo. Funding: This study was funded by Eli Lilly and Company.
S.C. is currently affiliated with the Department of Oncology, Università Vita‐Salute; IRCCS Ospedale San Raffaele Milano, Italy
Disclosures of potential conflicts of interest may be found at the end of this article.
No part of this article may be reproduced, stored, or transmitted in any form or for any means without the prior permission in writing from the copyright holder. For information on purchasing reprints contact commercialreprints@wiley.com. For permission information contact permissions@wiley.com.
References
- 1. Bray F, Ferlay J, Soerjomataram I et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018;68:394–424. [DOI] [PubMed] [Google Scholar]
- 2. National Comprehensive Cancer Network . NCCN clinical practice guidelines in oncology: Gastric cancer. Version 1. Plymouth Meeting, PA: National Comprehensive Cancer Network, 2019. [Google Scholar]
- 3. Chau I, Fuchs CS, Ohtsu A et al. Association of quality of life with disease characteristics and treatment outcomes in patients with advanced gastric cancer: Exploratory analysis of RAINBOW and REGARD phase III trials. Eur J Cancer 2019;107:115–123. [DOI] [PubMed] [Google Scholar]
- 4. Marschner N, Zacharias S, Lordick F et al. Association of disease progression with health‐related quality of life among adults with breast, lung, pancreatic, and colorectal cancer. JAMA Netw Open 2020;3:e200643. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Eisenhauer EA, Therasse P, Bogaerts J et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228–247. [DOI] [PubMed] [Google Scholar]
- 6. Song H, Zhu J, Lu D. Molecular‐targeted first‐line therapy for advanced gastric cancer. Cochrane Database Syst Rev 2016;7:CD011461. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Ford HE, Marshall A, Bridgewater JA et al. Docetaxel versus active symptom control for refractory oesophagogastric adenocarcinoma (COUGAR‐02): An open‐label, phase 3 randomised controlled trial. Lancet Oncol 2014;15:78–86. [DOI] [PubMed] [Google Scholar]
- 8. Thuss‐Patience PC, Kretzschmar A, Bichev D et al. Survival advantage for irinotecan versus best supportive care as second‐line chemotherapy in gastric cancer–A randomised phase III study of the Arbeitsgemeinschaft Internistische Onkologie (AIO). Eur J Cancer 2011;47:2306–2314. [DOI] [PubMed] [Google Scholar]
- 9. Pauligk C, Lorenzen S, Goetze T et al. A randomized, double‐blind, multi‐center phase III study evaluating paclitaxel with and without RAD001 in patients with gastric or esophagogastric junction carcinoma who have progressed after therapy with a fluoropyrimidine/platinum‐containing regimen (RADPAC). Ann Oncol 2017;28:670Pa. [Google Scholar]
- 10. Hironaka S, Ueda S, Yasui H et al. Randomized, open‐label, phase III study comparing irinotecan with paclitaxel in patients with advanced gastric cancer without severe peritoneal metastasis after failure of prior combination chemotherapy using fluoropyrimidine plus platinum: WJOG 4007 trial. J Clin Oncol 2013;31:4438–4444. [DOI] [PubMed] [Google Scholar]
- 11. Bang YJ, Xu RH, Chin K et al. Olaparib in combination with paclitaxel in patients with advanced gastric cancer who have progressed following first‐line therapy (GOLD): A double‐blind, randomised, placebo‐controlled, phase 3 trial. Lancet Oncol 2017;18:1637–1651. [DOI] [PubMed] [Google Scholar]
- 12. Sym SJ, Hong J, Park J et al. A randomized phase II study of biweekly irinotecan monotherapy or a combination of irinotecan plus 5‐fluorouracil/leucovorin (mFOLFIRI) in patients with metastatic gastric adenocarcinoma refractory to or progressive after first‐line chemotherapy. Cancer Chemother Pharmacol 2013;71:481–488. [DOI] [PubMed] [Google Scholar]
- 13. Wilke H, Muro K, Van Cutsem E et al. Ramucirumab plus paclitaxel versus placebo plus paclitaxel in patients with previously treated advanced gastric or gastro‐oesophageal junction adenocarcinoma (RAINBOW): A double‐blind, randomized phase 3 trial. Lancet Oncol 2014;15:1224–1235. [DOI] [PubMed] [Google Scholar]
- 14. Al‐Batran SE, Van Cutsem E, Oh SC et al. Quality‐of‐life and performance status results from the phase III RAINBOW study of ramucirumab plus paclitaxel versus placebo plus paclitaxel in patients with previously treated gastric or gastroesophageal junction adenocarcinoma. Ann Oncol 2016;27:673–679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Osoba D, Rodrigues G, Myles J et al. Interpreting the significance of changes in health‐related quality‐of‐life scores. J Clin Oncol 1998;16:139–144. [DOI] [PubMed] [Google Scholar]
- 16. Shitara K, Özgüroğlu M, Bang YJ et al. Pembrolizumab versus paclitaxel for previously treated, advanced gastric or gastro‐oesophageal junction cancer (KEYNOTE‐061): A randomised, open‐label, controlled, phase 3 trial. Lancet 2018;392:123–133. [DOI] [PubMed] [Google Scholar]
- 17. Kang YK, Boku N, Satoh T et al. Nivolumab in patients with advanced gastric or gastro‐oesophageal junction cancer refractory to, or intolerant of, at least two previous chemotherapy regimens (ONO‐4538‐12, ATTRACTION‐2): A randomised, double‐blind, placebo‐controlled, phase 3 trial. Lancet 2017;390:2461–2471. [DOI] [PubMed] [Google Scholar]
- 18. Kim MS, Prasad V. Assessment of accuracy of waterfall plot representations of response rates in cancer treatment published in medical journals. JAMA Netw Open 2019;2(5):e193981. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Markman M. What does tumor shrinkage mean to the patient receiving chemotherapy? Cleve Clin J Med 1996;63:301–302. [DOI] [PubMed] [Google Scholar]
- 20. Geels P, Eisenhauer E, Bezjak A et al. Palliative effect of chemotherapy: objective tumor response is associated with symptom improvement in patients with metastatic breast cancer. J Clin Oncol 2000;18:2395–2405. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
See http://www.TheOncologist.com for supplemental material available online.
Supplemental Figure 1 RAINBOW CONSORT Diagram
Supplemental Table 1 Analysis Populations
Supplemental Table 2. Baseline Characteristics by Disease Measurability
Supplemental Table 3. Baseline Symptom Level for Patients Included in Waterfall Plots