Beyond regimen selection: Systemic therapy completion and host determinants as predictors in localized upper gastrointestinal adenocarcinoma.

315 Background: Perioperative systemic therapy is standard in localized upper gastrointestinal (UGI) cancers, yet real-world outcomes are inferior to trials, with many patients unable to complete both neoadjuvant and adjuvant phases. While tumor grade and regression response are recognized prognostic factors, the relative contributions of systemic therapy completion versus intrinsic tumor biology, and the determinants of treatment completion, remain poorly defined. Methods: We retrospectively analyzed 248 patients with localized UGI adenocarcinoma treated with curative intent across multiple institutions (2006–2025). Demographic, clinical, pathologic, and treatment variables were abstracted. Logistic regression models evaluated predictors of recurrence and early mortality (≤24 months). Covariates included systemic therapy exposure (neoadjuvant ± adjuvant), tumor grade, regression (TRG), and patient factors (age, ECOG, albumin, LDH, neutrophil-to-lymphocyte ratio NLR). Separate models assessed predictors of completing ≥4 cycles of neoadjuvant therapy, ≥4 cycles of adjuvant therapy, and receipt of both neoadjuvant+adjuvant (Neo+Adj). Results: Median age was 64; 85% were male, and most presented with cT3N1 disease. 98% received neoadjuvant therapy, 67% received adjuvant therapy, and 66% completed both phases. Recurrence occurred in 46%. Median overall survival (OS) was 50.3 months, and 34 months for those who recurred. In multivariable analysis, completion of both systemic therapy phases was the strongest predictor of improved OS and reduced recurrence, independent of age, ECOG, stage, grade, and TRG (OR for death ≤24 months 0.28, p=0.004). While high grade and poor TRG retained adverse impact (OR ~2–4), their effect size was smaller than systemic therapy exposure. Importantly, the specific chemotherapy regimen (DCF, FLOT, DCF+IO, others) did not independently influence recurrence or OS once completion was considered. Predictors of treatment completion were not tumor-related: stage, grade, and TRG were not significant. Instead, patient factors dominated: higher albumin favored neoadjuvant completion (OR 1.09 per g/L, p=0.017); elevated NLR predicted failure to complete both phases (OR 0.88 per unit, p=0.035); ECOG ≥1 consistently reduced odds of completion. Conclusions: In localized UGI cancer, completing planned systemic therapy—not regimen choice—most strongly predicts recurrence and survival, outweighing tumor grade and regression. The main determinants of therapy completion are host factors rather than tumor characteristics. These findings suggest that future research and quality improvement should prioritize optimizing patient resilience (nutrition, inflammation, functional status) to enable therapy delivery, rather than focusing solely on intensifying or substituting regimens.