Relative benefit of perioperative chemotherapy after resection for stage III colon cancer in the 70s age range: SEER 2004–2015.

Abstract

e13778 Background: The magnitude of survival benefit associated with perioperative chemotherapy after resection for stage III colon cancer may vary by age in real-world practice due to differential treatment selection. We examined age-stratified associations of chemotherapy with overall and cause-specific mortality in a population-based cohort. Methods: SEER 2004–2015 resected AJCC6 stage III colon cancer. Treatment: surgery+chemotherapy vs surgery only (no/unknown chemo). Multivariable Cox models with age interaction estimated age-stratified HRs for OS. Cause-specific and non–cancer-specific death were evaluated using cause-specific Cox models. Results: Among 72,297 patients, 57.5% received surgery+chemo (41,584) and 42.5% surgery-only (30,713). The chemo group was markedly younger (age ≥80: 9.2% vs 43.0%) and less often widowed (12.6% vs 30.5%); sex and grade were similar, and stage mix was comparable (IIIB most common; IIIC ~35% vs 32%). Chemotherapy was associated with improved OS in all age groups with a consistent U-shaped pattern: HRs were higher in younger patients (45–49: 0.82, 95% CI 0.72–0.92; 50–54: 0.80, 0.73–0.88), decreased to a nadir in the 70s (70–74: 0.56, 0.53–0.58; 75–79: 0.56, 0.53–0.58), then attenuated in older strata (80–84: 0.60, 0.57–0.63; ≥90: 0.64, 0.52–0.78). This non-monotonic age pattern was preserved across AJCC6 IIIA/IIIB/IIIC subgroups (lowest HR typically ~70–79, with wider CIs at extremes). In competing-risk analyses, surgery+chemo was associated with lower hazards of both cancer death (HR range ~0.59–0.86; e.g., 75–79: 0.58, 0.55–0.62; ≥90: 0.75, 0.59–0.96) and non-cancer death (HR range ~0.51–0.62; e.g., 70–74: 0.52, 0.48–0.56; 75–79: 0.52, 0.49–0.56). Conclusions: Perioperative chemotherapy was associated with improved OS and reduced cancer and non-cancer mortality across ages, but the magnitude was non-monotonic—strongest in patients ~70–79 and attenuated at younger and very old extremes—consistent with substantial age-dependent treatment selection in observational data.