Neoadjuvant versus Postoperative Chemoradiotherapy in Gastric Cancer

Abstract

The optimal timing of chemoradiotherapy (CRT) in patients with gastric cancer remains unclear. We sought to compare the survival outcomes between neoadjuvant CRT (NCRT) vs. postoperative CRT (postCRT) in patients with gastric cancer. We hypothesized that NCRT would be associated with better survival compared to postCRT. We retrospectively reviewed patients with gastroesophageal junction (GEJ) or gastric adenocarcinoma who underwent surgical resection and NCRT or postCRT between 2005-2017 at a single institution. Clinical parameters such as sex, age, performance status, histology, stage, surgical outcomes, chemotherapy (CTX), and radiotherapy (RT) regimen were analyzed. CRT-related toxicity was graded according to CTCAE v5.0. The primary endpoint was overall survival (OS), assessed from the time of diagnosis. Survival analysis was conducted using Cox proportional hazards regression and Kaplan Meier estimates. We identified 152 patients, and median follow-up was 37.5 months. Median age was 63. 77% were male. 93% had an ECOG <2. Tumor location was gastroesophageal junction (GEJ) in 58% and gastric in 42%. Clinical stage was mostly II (44%) or III (44%). 102 (67%) patients underwent NCRT while 50 (33%) underwent postCRT. Patients who received NCRT were more likely to be male (83% vs. 64%; p = 0.013) and have a GEJ tumor (76% vs. 22%, p<0.001), greater number of involved lymph nodes (median 1 vs. 0; p = 0.005), and higher clinical stage (p = 0.002). Median RT dose was 50.4 Gy for neoadjuvant RT and 45.0 Gy for postoperative RT (p<0.001). Concurrent CTX was carboplatin/taxol (47%), cisplatin/fluorouracil (FU; 17%), or FOLFOX (folinic acid/FU/oxaliplatin; 14%) in the neoadjuvant setting and single-agent with FU (86%) in the postoperative setting (p<0.001). The NCRT group had a pathologic complete response (pCR) rate of 26% and had greater rate of R0 resection compared to the postCRT group (95% vs. 76%; p = 0.002). NCRT vs. postCRT was associated with a lower rate of any Grade 3 or higher toxicity (10% vs. 54%; p<0.001). Only 1% had treatment break(s) in the NCRT setting vs. 10% in the postCRT setting (p = 0.015). On multivariable analysis of OS, NCRT vs. postCRT (HR = 0.57 [95% CI 0.36-0.91]; p = 0.020) and R0 resection (HR = 0.50 [95% CI 0.27-0.90]; p = 0.021) were independently associated with lower hazards of death. The estimated 5-year OS was 67% (95% CI 0.55-0.76) in the NCRT group vs. 40% (95% CI 0.26-0.54) in the postCRT group (log-rank p = 0.003). NCRT was associated with a higher rate of R0 resection and longer OS with a lower toxicity compared to postCRT. Our findings suggest NCRT is superior to postCRT and support randomized trials to establish the optimal timing of CRT in gastric cancer.