Why This Study Matters
Biliary tract cancer (BTC) carries a high risk of recurrence even after curative resection. The source notes that only fluoropyrimidine-based regimens have demonstrated clinically meaningful adjuvant benefit, with BILCAP supporting capecitabine. The original ASCOT report showed that adjuvant S-1, an oral fluoropyrimidine, improved 3-year OS from 67.6% to 77.1%. This paper reports the prespecified final analysis 5 years after the completion of accrual.
Study Design
ASCOT is a multicenter, open-label, randomized phase 3 trial. Patients were randomly assigned 1:1 to observation (n = 222) or adjuvant S-1 (n = 218), stratified by primary tumor site and lymph node status. S-1 was given at 40 mg/m² orally twice daily for 4 weeks followed by a 2-week rest, for four cycles (24 weeks). The primary end point was OS in the intention-to-treat population; secondary end points included relapse-free survival (RFS) and safety. One-sided P values were used for the primary end point per the prespecified design; two-sided P values were used for secondary and exploratory analyses.
440 patients randomly assigned at 38 Japanese institutions; 1:1 allocation to observation or S-1; median follow-up 61.5 months; data cutoff June 2023.
Patient Population
Eligible patients were aged 20-80 years with histologically confirmed BTC (intrahepatic and extrahepatic bile duct, gallbladder, or ampullary carcinoma) after curative-intent resection (R0 or R1), with ECOG performance status 0 or 1. Enrollment ran 2013 to 2018.
| Characteristic | Observation (n = 222) | S-1 (n = 218) |
|---|---|---|
| Age, median (range) | 70 (40-80) | 68 (33-80) |
| Male, No. (%) | 152 (68) | 162 (74) |
| ECOG PS 0, No. (%) | 193 (87) | 191 (88) |
| N1, No. (%) | 90 (41) | 86 (39) |
| R1 resection, No. (%) | 33 (15) | 31 (14) |
Primary Endpoint Results
Median OS was 8.2 years with S-1 and 5.9 years with observation. The 5-year OS rates were 64% and 52% (HR, 0.72 [95% CI, 0.55 to 0.95]; one-sided P = .01). The authors conclude that the analysis confirms that adjuvant S-1 improves OS in patients with resected BTC.
ASCOT demonstrated a clear survival benefit in the intention-to-treat population, providing robust evidence for adjuvant S-1.
— Discussion, J Clin Oncol (2026)
Secondary Endpoint: Relapse-Free Survival
| Analysis | Median RFS (S-1 vs observation) | HR (95% CI) | P |
|---|---|---|---|
| Stratified | 6.1 vs 3.5 years | 0.76 (0.59 to 0.98) | .04 (two-sided) |
| Unstratified | — | 0.80 (0.62 to 1.03) | .08 (two-sided) |
The authors attribute the difference between the stratified and unstratified analyses to likely heterogeneity in baseline hazards across random assignment strata. RFS curves continued to decline and converged between the arms over time.
Subgroup Analyses
Subgroup-level hazard ratios appear only in the Data Supplement, so numerical subgroup results are not specified in the source main text. Qualitatively, OS subgroup analyses generally favored S-1, although the benefit appeared attenuated in some subgroups, including patients with ECOG PS 1 and those with R1 resection. For RFS, estimated HRs for S-1 versus observation were ≤1 in 23 of 24 subgroups. Multivariable analyses identified T3-4, N1, and R1 as predictors of worse survival; for RFS, T3-4, N1, R1, and CA19-9 >37 U/mL were significant predictors of poorer outcome. The authors state subgroup analyses were exploratory, and that a potential OS benefit in ampullary cancer is preliminary.
Safety Profile
Arm-level adverse event rates are reported in the Data Supplement (Table S4) and are not specified in the source main text. The report does state that non–cancer-related deaths without recurrence occurred in 10 of 222 patients (5%) with observation and 15 of 218 (7%) with S-1, including second primary malignancies and pneumonia, and that S-1 tolerability might have supported sustained treatment delivery.
Interpretation and Broader Context
The authors note that, unlike BILCAP, which supported capecitabine on a per-protocol basis despite a nonsignificant result, ASCOT showed benefit in the intention-to-treat population. Future research is directed at immune checkpoint inhibitors in the adjuvant setting and neoadjuvant strategies. The authors conclude that the data support S-1 as a standard postoperative treatment for resected BTC.
Limitations
Limitations listed in the source: the population was entirely Japanese, so pharmacokinetics and tolerability may differ in Western populations; no quality-of-life evaluation was performed; and the trial was conducted before widespread use of immune checkpoint inhibitors. The trial was open-label. The RFS result depends on the analysis: the stratified hazard ratio was significant (two-sided P = .04) but the unstratified one was not (two-sided P = .08). Subgroup analyses were exploratory, subgroup-level hazard ratios and arm-level adverse event rates appear only in the Data Supplement, and the potential OS benefit in ampullary cancer is preliminary.