Introduction

Lung cancer is currently the leading cause of cancer-related mortality worldwide, resulting in more than 25% of all cancer deaths [1]. Non-small cell lung cancer (NSCLC) accounts for a high proportion, approximately 85–90%, of lung cancer cases [2]. Many patients with NSCLC have advanced disease at diagnosis. The combination of platinum plus third-generation cytotoxic drugs has been the gold standard first-line chemotherapy for patients with advanced NSCLC. However, most patients experience disease progression during or after first-line treatment.

The standard second-line therapy for non-squamous NSCLC had been single-agent chemotherapy with docetaxel or pemetrexed. The response rate (RR), median progression-free survival (PFS), and median overall survival (OS) with these drugs range 7.1–9.1%, 2.2–2.9 months, and 6.7–8.3 months, respectively [3, 4].

S-1 is an oral medication that contains tegafur, gimeracil, and oteracil. It is widely used to treat various kinds of cancer. We previously recommended a tailored dose of S-1 adjusted by individual body surface area (BSA) and creatinine clearance (Ccr). Our pharmacokinetic data supported the validity of using Ccr in addition to BSA [5]. Our study and three other phase II trials showed that RR, median PFS, and median OS for S-1 monotherapy as second-line therapy NSCLC range 12.5–26.7%, 2.5–4.2 months and 8.2–16.4 months, respectively [6,7,8,9]. The survival benefit of these non-platinum second-line monotherapies was not satisfactory.

Bevacizumab is a monoclonal antibody that binds to vascular endothelial growth factor. In non-squamous NSCLC, bevacizumab conferred a survival benefit when combined with carboplatin and paclitaxel as first-line treatment [10]. However, the benefit of adding bevacizumab to non-platinum cytotoxic monotherapy is not clear as subsequent treatment. Therefore, we conducted a multi-center, a single-arm phase II study to evaluate the safety and efficacy of combination therapy of tailored-dose S-1 plus bevacizumab in patients with recurrent non-squamous NSCLC.

Patients and methods

Patient eligibility

Patients were enrolled if they met the following eligibility criteria: (1) histologically or cytologically confirmed non-squamous NSCLC; (2) stage III or IV disease, or postoperative recurrence; (3) at least one measurable lesion; (4) more than 4 weeks since the last administration of prior chemotherapy; (5) epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs), anaplastic lymphoma kinase (ALK) inhibitors, and mesenchymal epithelial transition factor (MET) inhibitors were not considered chemotherapy; (6) age ≥20 years; (7) Eastern Cooperative Oncology Group performance status of 0–2; (8) ability to tolerate oral intake; (9) adequate organ function, defined as adequate bone marrow function (white blood cell count ≥3500/μL, neutrophil count ≥1500/μL, platelet count ≥100,000/μL, and hemoglobin ≥9.0 g/dL), normal renal function (serum creatinine ≤1.5 mg/dL, urine protein ≤1+ or ≤2 g/24 h in pooled urine), normal liver function (aspartate aminotransferase and alanine aminotransferase ≤100 IU/L, serum total bilirubin ≤1.5 mg/dL), and normal respiratory function (SpO2 ≥90% or PaO2 ≥60 Torr on room air); (10) more than 2 weeks after radiotherapy (except thoracic irradiation); (11) recurrence more than 4 weeks after first-line chemotherapy (anti-cancer therapy given for recurrence of NSCLC within 180 days of previous preoperative or postoperative chemotherapy); (12) estimated life expectancy of more than 3 months; and (13) written informed consent.

Exclusion criteria were: (1) history of S-1 treatment; (2) interstitial lung disease on chest radiography; (3) high risk of bleeding, defined as intrabronchial invasion observed with bronchoscopy, invasion of major vessels on computed tomography (CT), history of hemoptysis (>2.5 mL), or anticoagulant or anti-platelet drug therapy; (4) clinically significant complications or unstable medical condition such as symptomatic brain metastasis, pleural effusion requiring drainage, severe heart disease, uncontrolled diabetes mellitus, active infection, or other active neoplasm; (5) thoracic irradiation within 2 weeks; (6) pregnancy, lactation, and possibility of being pregnant; and (7) current flucytosine therapy. This study followed the ethical principles of the Declaration of Helsinki. The protocol was approved by the institutional review board at each of the three participating institutions.

Study treatment

Patients received S-1 adjusted on the basis of Ccr and BSA. Based on our previous studies [5, 8] and a post-marketing survey [11], we used the S-1 doses, as shown in Table 1. Ccr values were calculated using the Cockcroft–Gault formula. Patients took oral S-1 on days 1–14. Bevacizumab (15 mg/kg) was given intravenously on day 1. This schedule was repeated every 3 weeks until disease progression, unacceptable toxicity, or withdrawal of consent. The starting criteria were as follows: neutrophil count ≥1500/μL, platelet count ≥100,000/μL, aspartate aminotransferase and alanine aminotransferase ≤100 IU/L, total bilirubin ≤1.5 mg/dL, and non-hematologic toxicities of Grade 1 or lower. If a patient did not meet these criteria on day 1, the cycle was postponed. Criteria for discontinuation of protocol treatment included: (1) disease progression; (2) Grade 2 or higher interstitial pulmonary fibrosis or pneumonitis; (3) Grade 4 non-hematologic toxicity; (4) Ccr <30 mL/min; (5) delay of the start of the next course for more than 3 weeks; (6) need for a second dose reduction; and (7) withdrawal of consent. One-level dose reduction of S-1 was permitted if Grade 3 or 4 neutropenia, thrombocytopenia, or non-hematological toxicity occurred in the previous cycle.

Table 1 S-1 dose (mg/day) based on body surface area and creatinine clearance
Full size table

Assessment of response and toxicity

Baseline assessments included medical history, physical examination, complete blood count, serum biochemical analyses, chest radiography, electrocardiography, and urinalysis. All patients underwent chest and abdominal CT, brain CT or magnetic resonance imaging, and positron emission tomography or bone scintigraphy at most 4 weeks before enrollment. Response was graded as complete response (CR), partial response (PR), stable disease (PR), progressive disease (PD), or not evaluable (NE) according to the Response Evaluation Criteria in Solid Tumors (RECIST) criteria, version 1.1. All patients were evaluated with CT every 6 weeks. Toxicity was evaluated every 2 weeks according to the Common Terminology Criteria for Adverse Events (CTCAE), version 4.0.

Statistical analyses

The primary endpoint was RR. Secondary endpoints consisted of safety, PFS, and OS. In the V15-32 trial for Japanese patients with previously treated NSCLC, the RR for docetaxel as second-line therapy was 12.8% (95% confidence interval (CI): 8.8–17.5) [12]. In a phase II study of bevacizumab in addition to docetaxel or pemetrexed, the RR of combination therapy ranged from 12.5% (95% CI 5.3–25.5%) for docetaxel to 17.9% (95% CI 9.0–32.7%) for pemetrexed [13]. Based on these results, we assumed the threshold for RR, expected RR, two-sided α error, and power to be 9, 25, 5, and 80%, respectively. The required sample size was 27 patients. Taking into account patients who might be withdraw consent or be excluded from analysis, we needed 30 patients. OS and PFS were defined as the time from study enrollment until death or last observation and disease progression or death from any cause, respectively. OS and PFS were calculated using the Kaplan–Meier method. All statistical analyses were performed using JMP® Pro 12.0.0 for Mac (SAS Institute, Cary, NC, USA).

This trial has been registered under the University Medical Hospital Information Network (UMIN) Clinical Trial Registry Identifier UMIN000008158.

Results

Patients

Thirty patients were enrolled from March 2013 to March 2015. We followed them until June 30, 2016. At the cut-off date, one patient was lost to follow-up, 20 had died, and nine were alive. Table 2 shows the patients’ baseline characteristics. Regarding prior regimens, one patient had received pemetrexed monotherapy and 29 patients had received platinum-based chemotherapy, of whom one had received pemetrexed maintenance therapy. Four patients had received EGFR-TKIs.

Table 2 Patient characteristics
Full size table

Treatment

Thirty patients received a total of 203 cycles (median 5 cycles, range 1–24 cycles). Nineteen patients withdrew because of PD, five because of toxicity, and four withdrew consent. Two patients were still on this regimen at the cut-off date. One patient had to quit bevacizumab after 18 cycles because of proteinuria but could continue S-1 alone for another four cycles until PD. Fourteen patients received post-protocol chemotherapy and 16 received palliative radiotherapy or best supportive care.

Efficacy

All patients were evaluable for response. One achieved CR, 1 PR, 22 SD, and 6 PD. Thus, the RR and disease control rate (DCR) were 6.7% (95% CI 1.8–21.3%) and 80% (95% CI 62.7–90.5%), respectively. On the other hand, the RR and DCR of 25 patients with wild-type EGFR were 12% (95% CI 4.1–30.0%) and 76% (95% CI 56.7–88.5%), respectively. Median PFS was 4.8 months (95% CI 2.7–6.4 months) (Fig. 1). Median OS was 13.8 months (95% CI 8.4 months–not available) (Fig. 2). Four patients had continued this regimen for more than 1 year with tolerable toxicity. Six patients had received bevacizumab as first-line chemotherapy. The RR of bevacizumab-naïve and bevacizumab-pretreated patients was 8.3% (95% CI 2.3–2.6%) and 0%, respectively. On the other hand, prior use of bevacizumab did not affect PFS (5.4 months for bevacizumab-naïve vs. 3.3 months for bevacizumab-pretreated, p = 0.08). EGFR mutation status did not affect PFS for this regimen (wild type vs. mutant; 4.8 months vs. 5.8 months, p = 0.76). Median PFS of patients with CCr ≥60 mL/min was longer than that of patients with Ccr <60 mL/min (2.6 months vs. 5.4 months, p = 0.02) (Fig. 3).

Fig. 1
figure 1

Kaplan–Meier curve of progression-free survival in patients treated with S-1 plus bevacizumab. PFS progression-free survival, CI confidence interval

Full size image
Fig. 2
figure 2

Kaplan–Meier curve of overall survival in patients treated with S-1 plus bevacizumab. OS overall survival, CI confidence interval

Full size image
Fig. 3
figure 3

Kaplan–Meier curve of progression-free survival in patients treated with S-1 plus bevacizumab by creatinine clearance. PFS progression-free survival, Ccr creatinine clearance, CI confidence interval

Full size image

Safety

All patients were assessed for toxicity (Table 3). No grade 4 severe toxicities or treatment-related deaths were observed. The only grade 3 hematologic toxicity was anemia. Common grade 3 non-hematologic toxicities included diarrhea, anorexia, and infection. One patient withdrew from the protocol treatment due to grade 2 gastric perforation, likely caused by non-steroidal anti-inflammatory drugs. This patient did not require surgery.

Table 3 Toxicities
Full size table

Discussion

This is the first study of combination therapy with tailored-dose S-1 and bevacizumab. The RR of 6.7% did not meet the threshold for the primary endpoint of this study. Our study did not demonstrate that bevacizumab has an effect when given in addition to S-1. There were two studies of S-1 plus bevacizumab combination therapy in patients previously treated for non-squamous NSCLC. The results of our study were consistent with those two studies. Nishino et al. compared docetaxel plus bevacizumab with S-1 plus bevacizumab. They reported that median RR and PFS were 2.2% and 3.5 months with S-1 plus bevacizumab and 22.3% and 3.9 months for docetaxel plus bevacizumab. Both regimens were associated with modest PFS, but did not warrant further investigation because of low RR for S-1 plus bevacizumab and severe toxicity with docetaxel plus bevacizumab [14]. Tokito et al. conducted a phase II trial of S-1 plus bevacizumab after platinum-based chemotherapy in patients with advanced non-squamous NSCLC. The RR, DCR, median PFS, and OS were 14.3, 85.7%, 3.2, and 11.4 months, respectively [15].

Our study suggested that S-1 with dose tailored to renal function reduces severe adverse events, even when combined with bevacizumab. We previously reported that S-1 with dose tailored to renal function was reasonable for elderly patients in terms of pharmacokinetics [5]. This study showed that, compared with patients with Ccr ≥60 mL/min, patients with Ccr <60 mL/min had shorter PFS but similar adverse event profiles. Regarding safety, the toxicity of tailored-dose S-1 plus bevacizumab was manageable. Neutropenia of Grade 3 or higher was not observed in our study, although it occurred in 4.4 and 2.1% of patients in the previous two studies [14, 15]. As a result of the latest phase III trial (EAST-LC) demonstrating non-inferiority of S-1 to docetaxel [16], S-1 monotherapy is an alternative second-line regimen. We believe that tailored-dose S-1 is also reasonable in the second-line setting.

In conclusion, the addition of bevacizumab to S-1 was not beneficial in patients with previously treated non-squamous NSCLC. We do not recommend any further study of this regimen.