Introduction

Advanced upper gastrointestinal malignancies, such as esophageal cancer (EC), esophagogastric junction cancer (EGJC), and gastric cancer (GC), are highly lethal. GC is the fifth-most common cancer and is the fourth-highest cause of cancer mortality worldwide [1]. The proportion of patients with cancer involving the upper third of the stomach has gradually increased in Asia, including Japan [2]. The number of patients with lower region EC and EGJC is increasing in Japan, similar to Western countries [3, 4]. The esophagogastric junction represents an anatomical site with a remarkably high and rapidly rising incidence of adenocarcinoma. In Western countries, the incidence of esophagogastric junction adenocarcinoma gradually increased over the past few decades. In Eastern countries, the incidence of non-cardiac GC has decreased with the declining prevalence of Helicobacter pylori infection [5]. In Japan, the number of operations for EGJC increased by 2.4-fold during the targeted decade in a previous study [6].

Regarding multimodality treatment for EGJC, its role remains unknown because of a lack of studies focused specifically on this malignancy. Most clinical trials of the multimodality treatment of EGJC predominantly included patients with EC or GC [7,8,9]. Only one prospective trial limited to EGJC, namely the POET trial of neoadjuvant chemoradiotherapy (NACRT) versus neoadjuvant chemotherapy (NAC), has been conducted [10]. Unfortunately, this trial was closed prematurely because of low accrual, and it failed to demonstrate statistically significant benefits for NACRT. Therefore, the best strategy for multimodality therapy has not yet been established despite the increasing incidence of EGJC globally.

NAC is considered a standard treatment for locally advanced EC, EGJC, and GC in Western countries. In Japan, some S-1-containing regimens, which were substantiated by clinical trials for unresectable EGJC and GC [11], have been evaluated in the NAC setting [12]. Conversely, the potential disadvantages of NAC include an increase of risk of morbidity after surgery. Therefore, we should seek a balance between efficacy against the tumor and the safety of surgery when we select an optimal regimen for NAC.

S-1 plus oxaliplatin (SOX100: 80–120 mg/day S-1 for 2 weeks with 100 mg/m2 oxaliplatin on day 1, every 3 weeks) has been proposed as an alternative to cisplatin plus S-1 (CS) as the first-line chemotherapy for advanced GC given its similar efficacy and favorable safety profile. Furthermore, SOX100 is an effective and feasible therapy for both non-elderly and elderly patients with advanced GC [13]. Several clinical trials have been conducted to determine the efficacy and safety of SOX130 in the perioperative adjuvant setting [14,15,16,17]. The current phase II study, named KSCC1601, was conducted to investigate the efficacy and safety of SOX130 as NAC for locally advanced gastric cancer (LAGC) and EGJC.

Patients and methods

This multicenter, open-label, single-arm, phase II trial was conducted at 19 institutions in Japan. The Kyushu University Certified Institutional Review Board approved this study protocol. The study was conducted in accordance with the precepts established in the Declaration of Helsinki and Clinical Trials Act. Written informed consent was obtained from all eligible patients prior to registration. This trial was registered with the University Hospital Medical Information Network Clinical Trials Registry (https://www.umin.ac.jp/ctr/) as UMIN000021061 and the Japan Registry of Clinical Trials as jRCTs071180064.

Patients

Patients with histologically confirmed adenocarcinoma of the gastric or esophagogastric junction (Siewert Type II and III) were eligible for registration. The inclusion criteria were as follows: (a) clinical T3/T4a/T4b or N1/2/3a/3b GC or clinical T3/T4a/T4b or any N esophagogastric junction cancer according to the Japanese Classification of Gastric Carcinoma [18]; (b) no evidence of distant metastasis; (c) absence of carcinoma cells in the peritoneal cytology test, absence of peritoneal dissemination (P0), and absence of liver metastasis (H0) as confirmed via staging laparoscopy before entry into the study; (d) no remnant GC; (e) no previous treatment (chemotherapy, radiotherapy, or endocrine therapy); (f) no history of surgery for GC or EGJC excluding endoscopic resection; (g) age ≥ 20 years; (h) Eastern Cooperative Oncology Group performance status ≤ 1; (i) sufficient oral intake; (j) adequate organ function (bone marrow function [white blood cell count ≤ 12,000/mm3; neutrophil count ≥ 1500/mm3; platelet count ≥ 100,000/mm3; hemoglobin ≥ 8.0 g/dL], adequate liver function [serum bilirubin level ≤ 1.5 mg/dL, serum transaminase level ≤ 100 IU/L, and serum albumin ≥ 2.5 g/dL], and adequate renal function [serum creatinine level ≤ 1.5 mg/dL and creatinine clearance ≥ 60 mL/min]).

Treatment

SOX130 was administered every 3 weeks for three cycles. Oxaliplatin was administered intravenously at a dose of 130 mg/m2 on day 1 of the 21-day cycle (SOX130). S-1 was administered orally at 80 mg/m2 (80–120 mg/day total dose depending on the patient’s body surface area as follows: < 1.25 m2, 80 mg; 1.25–1.5 m2, 100 mg; and > 1.5 m2, 120 mg). S-1 was administered daily for 14 days, followed by a 7-day rest period. Tumor resectability was assessed comprehensively via computed tomography, upper gastroenterological endoscopy, and barium meal study within 7–56 days after the completion of three cycles of SOX130. Adverse events associated with chemotherapy were evaluated using the Common Terminology Criteria for Adverse Events (version 4.0).

Surgical resection was performed within 56 days after the last dose of S-1. R0 resection was attempted via gastrectomy with D2 lymph node dissection, depending on the tumor location, according to the Japanese gastric cancer treatment guideline 4th edition [19]. The resection criteria after NAC were as follows: (1) R0 resection was possible via gastrectomy with D2 lymph node dissection and (2) sufficient organ function (white blood cell count ≥ 3000/mm3; platelet count ≥ 100,000/mm3). Morbidity was evaluated using the Clavien–Dindo (CD) classification in the safety analysis set [20]. Adjuvant chemotherapy with S-1 monotherapy, capecitabine plus oxaliplatin (CapeOX), or SOX was recommended according to Japanese gastric cancer treatment guidelines [19]. The rate of protocol completion was defined as the proportion of patients who completed NAC and underwent gastrectomy. The completion of NAC was defined as the completion of three cycles of SOX130. However, the protocol completion included patients who received a minimum of 1 cycle of SOX130 if patients underwent curative surgery after NAC.

Statistical methods

The primary endpoint was the pathological response rate (pRR) among patients who underwent gastrectomy. The pRR was defined as the ratio of grade 1b–3 for primary tumors according to the Japanese classification of gastric carcinoma 3rd English edition [18]. The threshold and expected response rates of patients who received SOX130 were estimated to be 30 and 50%, respectively, according to previous studies [21,22,23]. At a one-sided alpha of 0.05 and 80% power based on application of the Clopper–Pearson exact method to the binomial distribution, 43 patients were required for the study. Assuming a loss of 10% based on patient ineligibility or withdrawal, we set the target number of patients as 46. Furthermore, we capped the number of patients with LAGC at 22 in the study protocol. The key secondary endpoint was the rate of anastomotic leakage after esophagojejunal anastomosis for Siewert types II and III tumors. The threshold and expected rates of anastomotic leakage of patients who underwent esophagojejunal anastomosis were estimated to be 30 and 6%, respectively, according to previous studies [24]. At a one-sided alpha of 0.05 and 80% power based on application of the Clopper–Pearson exact method to the binomial distribution, 19 patients were required for the study. The incidence of esophagojejunal anastomotic leakage was investigated in 20 patients with EGJC. Other secondary endpoints were the rates of overall survival (OS), relapse-free survival (RFS), complete resection (R0), resection, response, adverse events, surgical complications, and completion of NAC. The clinical tumor response was assessed according to Response Evaluation Criteria in Solid Tumors version 1.1. OS and RFS were calculated using the Kaplan–Meier method. OS was defined as the time from enrollment to death from any cause. RFS was defined as the time from the date of surgery to the first date of relapse or death from any cause. The survival and safety analyses were assessed in 47 patients in the full analysis set (FAS). The pRR was determined for the 42 patients who underwent gastrectomy. All statistical analysis was performed using SAS version 9.4 (SAS Institute, Cary, NC, USA).

Results

Patient characteristics

From April 2016 through July 2017, 47 patients were enrolled at 19 institutions in Japan. The baseline characteristics of the patients are presented in Table 1. In the FAS, 23 patients (48.9%) had LAGC, and 24 patients (51.1%) had EGJC (Fig. 1).

Table 1 Patients characteristics
Fig. 1
figure 1

Patient flowchart. FAS final analysis set; EGJC esophagogastric junction cancer, LAGC locally advanced gastric cancer

NAC

Of 47 patients who received SOX130 NAC was terminated in three patients. The clinical response rate was 40.4% (95% confidence interval [CI] = 26.4–55.7). The number of patients who completed NAC was 42 (89.4%, 95% CI = 76.9–96.5). Table 2 presents the details of chemotherapy-related toxicities during SOX130 among the 47 treated patients. The most frequently observed toxicity (all grades) was appetite loss in 36 patients (76.6%), followed by peripheral neuropathy in 33 patients (70.2%). The respective major grade 3 or 4 hematological toxicity was thrombocytopenia in five patients (10.6%), followed by neutropenia in three patients (6.4%). No cases of febrile neutropenia were observed. The respective major grade 3 or 4 non-hematological toxicities were appetite loss in six patients (12.8%) and diarrhea in three patients (6.4%). There were no chemotherapy-related deaths.

Table 2 Toxicities of neoadjuvant chemotherapy

Surgical and pathological findings

Five patients did not undergo gastrectomy because disease progression (four patients) and patient refusal (one patient, Fig. 1). Consequently, 42 patients received per-protocol treatment (EGJC, 20; LAGC, 22). The rate of transition to surgery (i.e., protocol completion) was 89.4% (95% CI = 76.9–96.5). The average interval between NAC and surgery was 29 days (11–55 days). Meanwhile, 41 patients underwent R0 resection (87.2%, 95% CI = 74.3–95.2, Table 3). The pRR in the eligible patients was 59.5% (90% CI = 45.7–72.3), including complete remission in four patients (9.5%, Table 4). This pRR was higher than the estimated threshold and expected rates for patients who received SOX130.

Table 3 R0 resection rate
Table 4 Proportion of necrosis in the tumor

Operative morbidity data are summarized in Supplemental Table S1. The most frequently observed morbidity (all grades) was anastomotic leakage in six patients (12.8%), followed by dumping syndrome in five patients (10.6%). Esophagojejunal anastomosis was performed in 20 patients with Siewert types II and III tumors. A total of six patients (30%) had anastomotic leakage. The rate of severe anastomotic leakage (greater than CD grade III) after esophagojejunal anastomosis was 25.0% (90% CI = 10.4–45.6, Table 5). The laparoscopic procedure was significantly associated with severe anastomotic leakage (Supplemental Table S2). Among the 42 patients who completed NAC, 33 (78.6%) subsequently received adjuvant chemotherapy (Supplemental Table S3). Seventeen patients received second-line chemotherapy.

Table 5 Anastomosis leakage rate of EGC

Survival outcomes

At the cutoff date of June 2020, the median duration of follow-up for the OS analysis was 32.9 months (range, 2.0–45.5 months). There were 17 deaths, including 14 deaths from progressive disease. The 3-year OS rate was 62.9% (95% CI = 47.2–75.1, Fig. 2A). The 3-year RFS rate was 53.2% (95% CI = 38.1–66.2, Fig. 2B). The effect of severe anastomotic leakage on RFS is presented in Supplemental Fig. S1. There was no difference in the 3-year RFS rate between patients with and without severe anastomotic leakage. Twenty-three patients (48.9%) experienced cancer recurrence.

Fig. 2
figure 2

A Kaplan–Meier curve of overall survival in all eligible patients. B Kaplan–Meier curve of relapse-free survival in all eligible patients. OS overall survival; RFS relapse-free survival, CI confidence interval

Discussion

This phase II study demonstrated that NAC with SOX130 followed by surgery is acceptable method with a high pRR in patients with EGJC and LAGC. NAC was completed in approximately 90% of patients with acceptable adverse event rates except anastomotic leakage for EGJC.

In Western countries, NAC using 5-fluorouracil plus cisplatin (FP) or FP plus epirubicin (ECF) has been the standard regimen [7]. Most recently, the German FLOT4 trial established the perioperative docetaxel/oxaliplatin/5-fluorouracil (FLOT) regimen as the new standard treatment for resectable EGJC and GC as a substitute for ECF/ECX [25]. However, the FLOT regimen has been criticized for its toxicity. In Eastern countries, a recent study found that NAC with CS followed by D2 gastrectomy for type 4 and large type 3 GC can be safely performed without increasing the risk of morbidity or mortality [12]. Compared with the FLOT regimen, few severe adverse events, such as neutropenia and febrile neutropenia, were observed for the CS regimen. To enhance the efficacy of NAC, Hosoda et al. conducted a phase II trial (KDOG1001) to reveal the efficacy of the addition of docetaxel to CS (DCS) [26]. A higher pRR of approximately 60% was observed. Conversely, the incidence of grade 3 or higher neutropenia was 55% with DCS, versus 29% with CS, and that of grade 3 or higher febrile neutropenia was 7.5% in the DCS arm, compared with 0.7% for the CS group. A phase II study of perioperative CapeOX was also conducted for patients with clinical SS/SE N1–3 GC [27]. Perioperative CapeOX has good feasibility and favorable efficacy with a higher pRR (54.1%), although peripheral neuropathy or hand foot skin reaction was frequently observed. In the current study, the pRR, the primary endpoint, was 59.5% (90% CI = 45.7–72.3), including complete remission in four patients (9.5%). In addition, the incidence of grade 3 or higher neutropenia with the SOX130 regimen was 6.4%, and that of any-grade febrile neutropenia was 0% (Supplemental Table 4). The rate of R0 resection was equivalent that in previous studies despite the slightly lower rate of transition to surgery (89.4%) [12, 26]. Furthermore, the rate of disease progression was 8.5%, which compares favorably with other studies. Therefore, SOX130 appears to be more effective and feasible, as well as less toxic and more convenient clinically. In addition, forced hydration is not needed, unlike the CS regimen.

In the current study, we also highlighted the incidence of esophagojejunal anastomotic leakage after total gastrectomy and esophagocardial resection for EGJC. These types of gastrectomy are well known as difficult procedures that carry significant risks of postoperative morbidity and mortality. Esophagojejunal anastomotic leakage, which is the most frequent complication in gastrectomy, can be fatal. In another study, serious morbidity and mortality rates were significantly higher in the total gastrectomy group than in the partial gastrectomy group [28]. Proximal or total gastrectomy with transhiatal resection of the distal esophagus, as well as lymphadenectomy, was required for EGJC. Classen et al. analyzed the influence of NAC on surgical morbidity and mortality in the CRITICS trial (three cycles of epirubicin, cisplatin or oxaliplatin, and capecitabine), finding that complications following anastomotic leakage represented the most important factor for postoperative mortality. Total gastrectomy and esophagocardial resection were associated with a greater risk of morbidity than subtotal gastrectomy [29]. In the CRITICS trial, the rates of anastomotic leakage were 2.0% in subtotal gastrectomy, 10.1% in total gastrectomy, and 12.7% in esophagocardial resection. Conversely, in a phase II trial conducted in Japan, the rate of anastomotic leakage after NAC treatment was 0.7%–3% across all types of gastrectomy [12, 26, 27]. Our trial illustrated that the rate of severe anastomotic leakage (greater than CD grade III) after esophagojejunal anastomosis was 25.0%, which was significantly associated with the laparoscopic procedure. A nationwide study in Japan revealed that the incidence of anastomotic leakage after laparoscopic gastrectomy is significantly higher than that after open total gastrectomy [30]. Tokunaga et al. revealed that postoperative intra-abdominal infectious complications can lead to adverse effects, affecting both OS and disease-free survival in patients with GC, even with curative resection [31]. Although our data indicated that severe esophagojejunal anastomotic leakage does not affect RFS, it is critical to clarify whether the survival benefit of NAC over surgery alone outweighs this potential disadvantage.

Our study had some limitations. First, this trial had a single-arm, phase II design and a limited number of patients. The small number of patients recruited in this study might explain the high rates of severe anastomotic leakage. The incidence of esophagojejunal anastomotic leakage was higher than that reported in previous studies [12, 26, 27]. Also, variability among surgeons and hospitals can influence the outcome of surgical multicenter trials. A nationwide study in Japan revealed that surgeons and hospital volume had strong effects on postoperative morbidity and mortality after gastrectomy [32, 33]. Therefore, we should consider assessing the impact of quality assurance for esophagojejunal anastomosis. Second, some cases underwent surgery a very short time after the end of NAC because we did not set a minimum interval. Most previous studies set a minimum interval of at least 2 weeks [12, 25, 26]. However, an appropriate time to surgery after NAC remains controversial because of issues related to the pathologic response and surgical safety [34].

In conclusion, SOX130 demonstrated substantial benefit for LAGC and EGJC. However, special attention should be paid to anastomotic leakage during surgery for EGJC. A phase III study is required to clarify whether the SOX130 regimen is more beneficial than other NAC regimens for LAGC and EGJC.