Real-life use of ramucirumab in gastric cancer in Spain: the RAMIS study

Gastric cancer (GC) is the fifth most frequently diagnosed cancer and the third leading cause of cancer death [1]. Worldwide, it accounts for 5.7% [2] of all new cancer cases and 8.2% of all cancer-related deaths [1]. Although a gradual decline in its incidence has been globally registered in recent years, it remains a major health concern with 140,000 new cases diagnosed across Europe in 2012, resulting in 107,000 deaths [3]. In Spain, prevalence of gastric adenocarcinomas is 7810 cases per year, resulting in more than 5675 deaths per year [4]. Approximately 90% of GCs are adenocarcinomas that arise from the glands of the mucosa in the stomach [5]. The most common risk factors associated with GC include male sex (incidence is twice as high), advanced age, tobacco smoking, family history and possibly dietary habits, such as low consumption of fruits and vegetables or high intake of salty and smoked food [6,7].

Surgery (subtotal or total gastrectomy) is the primary treatment option for GC, with recurrence common in up to 70% of patients [8]. Patients with inoperable locally advanced and/or metastatic (stage IV) disease receive systemic chemotherapy, the only treatment option that has shown a superior survival rate and improved quality of life (QoL) compared with best supportive care [7,9]. However, most patients relapse after first-line therapy. Literature suggests that 20–64% patients receive further treatment with second-line chemotherapy [8,10]. A second-line treatment is always justified in physically fit patients who are willing to receive treatment because it improves overall survival (OS), as well as symptom control and QoL [11].

Ramucirumab (Cyramza^®) is a second-line treatment option that selectively targets VEGF receptor (VEGFR)-2. It is a human IgG1 monoclonal antibody and the first biological agent for angiogenesis approved for GC by the US FDA and the EMA based on two randomized, double-blind, placebo-controlled Phase III trials (REGARD and RAINBOW) [12,13]. It demonstrated an ability to extend the life span and slow down tumor growth in patients with metastatic gastric or gastroesophageal junction (GEJ) adenocarcinoma progressing after first-line chemotherapy either as monotherapy or in combination with paclitaxel [12,13]. The REGARD trial compared ramucirumab with placebo and demonstrated improvement in OS (median OS of 5.2 vs 3.8 months), progression-free survival (PFS) (2.1 vs 1.3 months) and disease control rate (DCR) (49% vs 23%) [12]. The RAINBOW study demonstrated a longer OS (9.6 months vs 7.4 months) with ramucirumab plus paclitaxel compared to placebo plus paclitaxel along with improvement in PFS (4.4 months vs 2.9 months) and objective response rate (ORR) (28% vs 16%) [13].

In Europe, available data outside the clinical trial setting come from the RAMoss study [14], which assessed the safety and effectiveness of ramucirumab in advanced GC (AGC) patients treated in Italy within a compassionate-use program before ramucirumab reimbursement authorization. Results from RAMoss study on OS (8 months), PFS (4.3 months) and ORR (20.2%) were consistent with those previously reported in the clinical trials, confirming the safety and efficacy of ramucirumab [14]. Outside of Europe, additional evidence comes from a real-world evidence study conducted in the USA [15] and from a study based on an expanded access program conducted in South Korea [16]; the authors also suggested in the conclusions that clinical outcomes observed in these settings were similar to data from Phase III clinical trials [12,13].

Ramucirumab, as a single agent or combined with paclitaxel, has been marketed in Spain since December 2015 for the treatment of patients with AGC or GEJ adenocarcinoma with disease progression on or after prior fluoropyrimidine- and/or platinum-containing chemotherapy. The RAMIS observational study aimed to obtain real-world data on ramucirumab utilization in clinical practice in Spain as monotherapy or in combination with paclitaxel.

Methods

Study objectives

The primary objectives of the study were to describe the clinical characteristics and the treatment patterns of the patients with AGC or GEJ adenocarcinoma treated with ramucirumab in clinical practice in Spain. The secondary objective aimed to describe the effectiveness of ramucirumab as a combination (with paclitaxel) or as monotherapy in current clinical practice. Further, the data were used to identify the predictive factors for effectiveness and assess effectiveness in specific subgroups of interest.

Study design

This was an observational, retrospective single-cohort study, based on medical chart review of patients with AGC or GEJ adenocarcinoma treated with ramucirumab in routine clinical practice. Twenty Spanish hospitals participated in the study.

The observational study period was from December 2015 (date of ramucirumab commercialization in Spain) to the study initiation in each hospital (October–December 2018). Oncologist investigators or their collaborators collected the study variables of each patient from the time of ramucirumab treatment initiation until the index date (site study initiation) or death, whichever occurred first.

The study was approved by the Ethics Committee of Hospital Ramón y Cajal in Madrid, Spain. It was conducted with the support of the Clinical Research Organization IQVIA for the following activities: protocol development, site monitoring and statistical analysis.

Study population

All patients with diagnosis of AGC or GEJ adenocarcinoma who had initiated treatment with ramucirumab since December 2015, aged ≥18 years and with complete medical record at the participating site or at least since the AGC or GEJ adenocarcinoma diagnosis, were included in the study. Patients who had participated in a clinical trial during the observational study were excluded.

Study variables

Demographic and clinical characteristics of patients included demographic variables, initial diagnosis of AGC or GEJ adenocarcinoma, tumor stage, diagnostic procedures, Eastern Cooperative Oncology Group (ECOG) performance status, histological and molecular characteristics of the tumor, comorbidities, history of previous treatments and laboratory variables. Descriptive data on treatment with ramucirumab were also collected, including the date of initiation of treatment with ramucirumab, regimen, dosage, number of cycles (as monotherapy or combination), regimen changes and dose adjustments.

Effectiveness of ramucirumab was described in terms of median PFS and PFS rates at 3, 6, 9 and 12 months; median OS and OS survival rates at 3, 6 and 12 months; median time to progression (TTP); and tumor response for both measurable versus nonmeasurable disease (following RECIST v1.1 criteria) [17].

For measurable tumors, responses were classified as complete response (CR), partial response (PR), progressive disease (PD) and stable disease (StD) (when an evaluation by imaging was available). In the case of nonmeasurable tumors, the response evaluation criteria used for classification were as follows: CR, noncomplete response/nonprogression and progression (as described in the medical record).

Additionally, ORR, DCR, time to response and duration of response (DOR) were calculated for patients with measurable tumors evaluated for imaging.

Statistical analysis

A descriptive analysis of all included study variables was performed. Continuous variables were described as the number of patients with valid/missing observations, mean, standard deviation (SD), median, 25 and 75 percentiles (P25 and P75, respectively), minimum and maximum. Categorical variables were described by frequencies and related percentages.

Descriptive analysis was also performed to describe the effectiveness variables in a set of subgroups of patients. Exploratory analysis of predictors of effectiveness was performed. Univariate Cox proportional hazard models among baseline characteristics and effectiveness variables were used to select the covariates to be introduced in the multivariate models. Multivariate Cox regression models were used to explore factors associated with clinical outcomes and was reported as hazard ratios (HR) with 95% CI and corresponding p-values. Exploratory analysis was performed to compare the results between different categories defined in each subgroup by parametric tests (Student's t-test) or nonparametric test (Mann–Whitney) for quantitative variables and X²-test for categorical variables. All analyses were performed using Kaplan–Meier (KM) survival analysis and the plots of survival curve were produced. DCR was calculated by adding the CR, PR and StD, as the best response during the study period. The method of KM was used to estimate the OS and PFS of patients stratified by therapy at 3, 6, 9 and 12 months. All effectiveness variables were analyzed separately for patients treated with monotherapy and combination therapy. Statistical analysis was performed using SAS version 9.4.

Results

Study population

A total of 318 patients treated with ramucirumab beginning in December 2015 were included in the database. One patient was excluded from final analysis because the treatment was initiated after the start of the study data collection; thus, the final number of patients eligible for full analysis set (FAS) were 317. Among the FAS, 297 (93.7%) initiated ramucirumab as combination regimen (first cycle of ramucirumab received in combination with paclitaxel), and 20 (6.3%) were prescribed monotherapy regimen (first cycle of ramucirumab received as monotherapy).

Patient & clinical characteristics

Patient characteristics

The mean (SD) age of enrolled patients was 62.5 (11.3) years and majority of the patients were males (66.9%) with mean (SD) BMI of 25.4 (29.4) kg/m² at ramucirumab treatment initiation. Clinically relevant chronic comorbidities at ramucirumab treatment initiation were present in 62.5% of patients. The most common comorbidities were hypertension (29.3%), dyslipidemia (23.0%), diabetes mellitus (12.0%) and other cardiovascular diseases (9.5%) (Table 1).

Tumor characteristics

The majority of the patients had gastric adenocarcinoma (77.6%); primary tumor was still present in 64.0% of the patients. Tumor grade was poorly (43.2%), moderately (24.0%) or well differentiated (6.9%). Histological subtype distribution was intestinal (41.0%), diffuse (30.3%) and mixed (3.8%). TNM (tumor node metastasis) staging revealed a high percentage of patients with T3 (32.9%) and T4 (29.4%), N2 (18.8%) and N3 (18.8%) and M1 (62.2%) disease. A majority of the patients had measurable tumor (77.9%) (Table 1).

Diagnostic characteristics

The most commonly used diagnostic procedures were computerized tomography (CT) (97.2%) and endoscopy (94.6%).

The mean time since initial diagnosis of AGC or GEJ adenocarcinoma to start of ramucirumab treatment was 18.6 (SD: 22.2) months. Metastatic/locally advanced unresectable disease was detected at initial diagnosis in 66.8% patients and occurred after radical surgery in 33.2%. More than three-quarters of the patients had one or two metastatic sites (77.6%) at the start of the treatment with ramucirumab. A total of 14.2% of patients with AGC/GEJ adenocarcinoma were HER2 positive, and ascites were present in 28.7% of patients (Table 1).

ECOG Performance Status

Patients included in the study were assessed as ECOG Performance Status 0–3 [18]. Higher proportion of patients in both groups (combination and monotherapy) belonged to ECOG 1 category (63.6% and 55.0%, respectively). Patients in monotherapy group had a higher proportion of patients in the ECOG categories 2 and 3 compared with the combination group (Table 1).

Laboratory values

Various laboratory parameters were analyzed at treatment initiation. The mean (SD) neutrophil-to-lymphocyte ratio was 4.5 (14.02). The mean (SD) hemoglobin level was 11.7 (1.6) g/dl and albumin level was 3.7 (0.6) g/dl; 30.0% of patients had values <3.5 g/dl. The mean (SD) values for C-reactive protein was 0.5 (0.6) mg/dl and 56.3% showed levels >0.2 mg/dl. The mean (SD) level of carcinoembryonic antigen (CEA) was 64.0 (287.4) ng/ml (Table 2).

Treatment pattern

Treatment history

A total of 97.2% patients received chemotherapy, 47.3% underwent surgery, 13.9% received radiotherapy and 10.4% received targeted therapy before ramucirumab initiation. Chemotherapy was administered as first line in 75.6% patients (73.4% and 26.6% patients received doublet and triplet regimen, respectively), neoadjuvant in 28.9% (30.4% and 69.6% patients received doublet and triplet regimen, respectively) and adjuvant in 21.4% patients (61.0% and 39.0% patients received doublet and triplet regimen, respectively). Out of the 26.6% of patients who received triplet chemotherapy in first line, 13.3% included taxanes. Mean duration of taxanes treatment in those patients was 147 (126.7) days (Table 3).

Treatment with ramucirumab

Most (93.7%) patients initiated ramucirumab treatment in combination; 20 (6.3%) in monotherapy. The median duration of ramucirumab treatment according KM estimator was 3.2 months (95% CI: 2.8–3.4 months). The median number of cycles (28 days) received was 2.5 for patients initiating monotherapy and 4.0 for combination therapy.

According to ramucirumab treatment, 297 (93.7%) patients received combination therapy (always with paclitaxel); 265 (83.6%) of total patients received it during all treatment duration (median of three cycles); 32 (10.1%) received combination followed by monotherapy (three cycles in combination and three cycles in monotherapy) with a mean (SD) time of 4.1 (3.7) months in combination before changing to monotherapy. Nineteen patients (6.0%) received ramucirumab monotherapy (median of three cycles). One patient received monotherapy and then combination therapy. A total of seven patients (2.2%), all treated with a combination regimen, needed dose adjustment of ramucirumab, and the mean (SD) time to first dose adjustment was 5.5 months (3.4 months).

The dose of ramucirumab combination and monotherapy was 8 mg/kg on days 1 and 15 of a 28-day cycle during all treatment in 98.5% and 100% patients, respectively.

Effectiveness with ramucirumab treatment

Overall survival

The median OS was 7.2 months (95% CI: 6.2–8.5 months) for overall population. OS rate at 3 months was 83.1%, 58.7% at 6 months and 33.4% at 12 months (Figure 1). The median OS for patients treated with ramucirumab in a combination regimen was 7.4 months (95% CI: 6.4–8.9 months) and 4.3 months (95% CI: 1.9–7.3 months) for patients with a monotherapy regimen.

Progression-free survival

The median PFS was 3.8 months for overall population (3.9 months for combination therapy and 2.0 months for monotherapy group) (Table 4). PFS rate at 3 months was 58.8%, 29.7% at 6 months and 11.1% at 12 months (Figure 2). PFS rates for combination and monotherapy, respectively, were 60.9% and 26.3% at 3 months, 30.6% and 15.8% at 6 months and 11.6% and 0% at 12 months.

Time to progression

Median TTP was 4.3 months for overall population (95% CI: 3.7–4.9 months) with 4.5 months (95% CI: 3.8–5.0) for combination regimen and 2.5 months (95% CI: 1.2–4.3 months) for monotherapy.

Objective tumor response

For patients with measurable disease (n = 239), CR and PR was observed in 2.9% (n = 7) and 17.6% (n = 42) of the patients, respectively (ORR: 19.2%). For nonmeasurable disease (n = 68), no patients obtained a CR, but 35.3% of patients obtained a best overall response of noncomplete response/nonprogression.

Time to response

Response was achieved by 46 patients for whom mean (SD) time to first response was 2.95 (2.37) month or 2.5 months as median.

Duration of response

Forty-six patients obtained response (CR/PR). During follow-up period, a total of 28 patients (60.9%) progressed after PR/CR response. The median time for DOR (since response to first progression) was 5.6 months (95% CI: 3.1–7.9 months). The probability of being progression free after response was 75.7% at 3 months decreasing to 44.7% and 20.6% at 6 and 12 months, respectively.

Subgroup analysis

Table 4 describes KM survival analysis of PFS and OS according to different sociodemographic and clinical characteristics. Median time of OS was longer for patients with ECOG 0 compared with ECOG ≥2 (10.2 [95% CI: 8.5–20.9] vs 4.0 [95% CI: 2.0–5.2] months) and for patients with measurable versus nonmeasurable disease (8.9 [95% CI: 7.0–10.3] vs 5.0 [95% CI: 4.0–6.4] months). Median time of OS was shorter for patients with peritoneal metastatic sites versus no such sites (6.1 [95% CI: 5.2–6.8] vs 9.6 [95% CI: 7.3–12.4]) and for patients with presence of ascites versus no presence (5.7 [95% CI: 4.3–6.0] vs 9.5 [95% CI: 7.7–11.9]). The median OS in patients treated with ramucirumab in combination, ECOG PS-0/1 and effectiveness variables assessed by imaging was 10.3 months (95% CI: 8.5–12.3 months).

Median time of PFS in patients with ECOG 0 was 5.55 (95% CI: 4.9–6.3 months), with measurable disease was 4.1 (95% CI: 3.6–4.7) and in patients without presence of ascites was 4.1 (95% CI: 3.5–4.9). The median PFS according to patients with tumor response evaluated by imaging, treated with combination regimen and ECOG 0/1 was 4.9 months (95% CI: 3.9–5.4 months).

Predictor of response

According to multivariable Cox proportional hazards regression model, higher hazard of death was related to nonmeasurable disease compared with measurable disease (HR: 1.7 [1.2–2.3; 95% CI]), ECOG 1 or ECOG ≥2 status compared with ECOG 0 performance status (HR: 1.7 and 2.6, respectively), and patients with ascites compared with patients without ascites (HR: 1.5; 95% CI: 1.1 to 2.1). Higher hazard of progression (or death) was related to monotherapy versus combination therapy (HR: 2.1; 95% CI: 1.2–3.4), nonmeasurable versus measurable disease (HR: 1.8 [95% CI: 1.4–2.4]), ECOG 1 or ECOG ≥2 versus to ECOG 0 (HR: 1.6; 95% CI: 1.2–2.2 and HR: 2.3; 95% CI: 1.5–3.7) and patients with 3 or more versus ≤2 metastatic sites (HR: 1.5; 95% CI: 1.1–2.0). Additional information is included in Supplementary Tables 1 & 2).

Discussion

Real-world evidence about the treatment patterns and effectiveness of drugs is increasingly being requested by regulatory bodies, payers and healthcare professionals to obtain additional data beyond predefined criteria in clinical trials.

Several observational studies have confirmed ramucirumab's safety and effectiveness in real life, with survival data similar to that obtained in Phase III trials [14–16]. However, no observational studies had been performed in Spain. The present study is the largest observational study with ramucirumab in gastric cancer conducted in Europe and the second worldwide, after the North American study [15], in terms of number of patients. As occurs with other real-life studies, its broader inclusion criteria with respect to Phase III trials [12,13] allow the generation of evidence in patients not represented or underrepresented in the latter, such as patients with comorbidities. Although some baseline characteristics in the RAMIS study were similar to patients participating in clinical trials regarding mean values of age, gender, AGC diagnosis with primary tumor still present and percentage of patients with ECOG 1, in general, the patient population in the present study had a worse prognosis, with a higher presence of proven risk factors such as ECOG 2/3, peritoneal metastases, clinically relevant chronic comorbidities (e.g., hypertension, dyslipidemia), weight loss >10% and nonmeasurable disease [19].

In first line, 73% of patients had been treated with doublet (without taxanes) and 27% with triplet chemotherapy (i.e., taxanes in combination with a platinum and 5-fluorouracil-based regimen). Triplet chemotherapy with taxanes can be a good treatment option for patients with high tumor burden and strong pressure to achieve remission, but they should only be administered in select patients with good general health because this modality lacks survival benefit and confers high toxicity [20]. On the other hand, clinical practice seems to reflect physicians' perception that a small number of patients with GC may benefit clinically from reexposure to taxanes, similar to other cancer treatment approaches (i.e., hormone-sensitive breast cancer or platinum-sensitive ovarian cancer). The RAMIS study suggests that ramucirumab outcomes after a doublet chemotherapy were similar to those observed after triplet chemotherapy, in both median PFS (3.8 vs 3.4 months) and OS (7.5 vs 7.2 months), thus being an appropriate choice in either setting.

After first-line chemotherapy, 93.7% patients started ramucirumab treatment in combination with paclitaxel, and 89.2% of these remained in combination during all treatment period. Only 2.2% of patients needed a ramucirumab dose adjustment, which was comparable to dose adjustments (5%) needed in the RAINBOW trial [13], suggesting good tolerability of ramucirumab in combination therapy in real-life conditions.

Although cross-trial comparisons should be made with caution, it seems that ramucirumab effectiveness results in both median time and rates were in line with previous trials [12,13], finding better outcomes with a combination regimen versus monotherapy (median PFS 3.9 months [95% CI: 3.4–4.3 months] vs 2.0 months [95% CI: 1.1–2.8 months] and median OS 7.4 months [95% CI: 6.4–8.9 months] vs 4.3 months [95% CI: 6.4–8.9 months]).

Similarly, ramucirumab treatment in combination with paclitaxel showed 4.4 months for PFS and 9.6 months for OS in the RAINBOW trial [13] and 2.1 months for PFS and 5.2 months for OS in REGARD trial [12]. Other treatments such as irinotecan or docetaxel have shown a median OS of 4.0 months in the AIO20 and of 5.2 months in the COUGAR-02 clinical trials, respectively [21,22].

All observed responses in our study were in the combination group; results are also in line with clinical trials (3% of patients obtained a PR/CR in the REGARD trial and a 28% response was observed in the RAINBOW trial) [12,13]. ORR values in RAMIS (19.2%) were comparable to the similar observational study RAMoss (20.2%) and slightly lower than in RAINBOW (28%). The difference with RAINBOW could be related to including ECOG PS 2–3 patients.

The patients more similar to those included in RAINBOW trial (173 patients with ECOG PS 0/1, treated with ramucirumab combined with paclitaxel and effectiveness assessed by imaging) showed better results than the total sample, with median PFS and OS values of 4.9 and 10.3 months, respectively.

Predictive factors for effectiveness, consistent with previous studies, were ramucirumab in combination regimen, ECOG PS 0, few metastatic sites and absence of ascites [12–16]. Presence of measurable disease was found to be a positive factor, unlike previous studies [12–14]. HER2+ had not been evaluated in multivariate analysis in previous studies [12–16] and in RAMIS was found to be a predictor for survival (HER2+ versus test not available).

The RAMIS study allowed evaluation of some subgroups of patients on which there is limited information in the literature, such as HER2+ patients. There are published data on a small cohort study of ten HER2+ patients who responded well and for long periods of time when receiving ramucirumab-based therapy in the second-line setting. The authors argued that this observation was consistent with an interaction between angiogenesis and HER2 signaling and that trastuzumab resistance might be reversed with the inhibition of the angiogenesis pathway [23].

Results in the RAMIS and RAMoss studies were consistent in this subpopulation, with median PFS were 4.9 versus 4.4 months, respectively, and median OS were 9.7 versus 7.9 months, respectively.

The present study has several limitations, mainly those related to its retrospective design, such as the possible lack of information registered in the medical charts and the fact that the effectiveness variables could not be complemented with quality-of-life data or other patient-reported outcomes not systematically collected in clinical practice. Nevertheless, this design guarantees that the data followed the usual clinical practice, without additional intervention. On the other hand, no safety information was collected; however, data coming from an ongoing ramucirumab post-authorization safety and effectiveness study will help physicians reveal this important information [24].

Additionally, in RAMIS study, 13.9% of patients were still receiving treatment at the time of data collection, so final figures could vary slightly. Finally, the effectiveness results presented by subgroups must be interpreted with caution because no formal sample size calculation was done to make any comparison, including the analysis between ramucirumab in monotherapy versus ramucirumab plus paclitaxel. In addition, it must be taken into consideration that the rest of the subgroup analysis included both RAM-treated and RAM + PTX-treated patients, with the potential impact on the results of the distribution of patients by treatment schedule.

In conclusion, ramucirumab use and effectiveness are consistent with available real-world studies and randomized clinical trials, given the presence of poor outcome indicators in the studied population.