CT-Defined Low Skeletal Muscle Mass, Prognostic Nutritional Index, and Anemia in Relation to In-Hospital Major Morbidity After Gastrectomy for Gastric Carcinoma: A Retrospective Cohort Study

Background and Objectives: Computed tomography (CT)-defined skeletal muscle mass, nutritional status, and anemia are widely proposed markers of perioperative vulnerability before gastrectomy, but published estimates are heterogeneous and derive largely from East Asian, upfront-surgery cohorts. It is unclear how frequently these abnormalities occur in a predominantly neoadjuvant-treated cohort, and how precisely a single-center series can relate them to severe early morbidity. Materials and Methods: This retrospective, single-center observational cohort included 107 consecutive patients who underwent gastrectomy for histologically confirmed gastric carcinoma. The outcome was an in-hospital Clavien–Dindo grade ≥III complication. Exposure prevalence is reported with Wilson 95% confidence intervals (CIs). Exploratory unadjusted logistic regression was used to estimate odds ratios (ORs), and Firth bias-reduced logistic regression was used as a sparse-event sensitivity analysis. Sensitivity analysis applied gastric-cancer-specific muscle thresholds and excluded completion gastrectomies. Results: CT-defined low skeletal muscle mass, low Prognostic Nutritional Index (PNI < 46), and anemia were present in 67 (62.6%; 95% CI 53.2–71.2), 45 (42.1%; 33.1–51.5), and 71 (66.4%; 57.0–74.6) patients, respectively. Fifteen patients (14.0%; 95% CI 8.7–21.8) had major complications, of whom 14 (93.3%) had undergone open surgery. All estimates were unadjusted and imprecise, with confidence intervals spanning potentially clinically important associations in both directions: low skeletal muscle mass (OR 1.23, 95% CI 0.39–3.89), SMI per 5-cm2/m2 decrease (OR 1.20, 0.88–1.65), PNI per 5-unit decrease (OR 1.18, 0.80–1.75), low PNI (OR 1.24, 0.42–3.72), and anemia (OR 1.02, 0.32–3.23). Firth penalized estimates were materially similar to the conventional estimates. Substituting gastric-cancer-specific muscle thresholds lowered the prevalence of low muscle mass from 62.6% to 21.5% without resolving the estimate (OR 1.40, 0.40–4.88). Excluding the 20 completion gastrectomies (87 patients, 11 events) moved the low-muscle-mass estimate below unity (OR 0.74, 0.21–2.65) without improving precision. Conclusions: These abnormalities were highly prevalent, but with only 15 events, the confidence intervals did not exclude clinically important associations in either direction. The concentration of major events among patients undergoing open surgery underscores the potential influence of operative case mix and residual confounding. The results support neither causal nor independent-effect inference whether assessment of these parameters adds to perioperative risk stratification remains a hypothesis to be tested in larger, multicenter, well-characterized cohorts.