Discrimination and calibration of the MSKCC and sarculator nomograms in extremity soft tissue sarcoma: a single-center retrospective comparison

Accurate prognostication in extremity soft tissue sarcoma (STS) is essential for treatment decisions and patient counseling. The Memorial Sloan Kettering Cancer Center (MSKCC) nomogram and the Sarculator tool are widely used, but head-to-head comparisons within exclusively extremity-localized disease and in non-Western cohorts are scarce. In an exploratory, hypothesis-generating analysis, we compared the discrimination and calibration of the two nomograms in a single-center extremity STS cohort. We retrospectively analyzed 41 patients with localized, resectable extremity STS who underwent curative-intent surgery between 2008 and 2024. Overall survival (OS) was estimated by Kaplan–Meier analysis. For each nomogram, horizon-specific survival predictions (MSKCC 4-, 8- and 12-year; Sarculator 5- and 10-year) were used as prognostic markers, and discrimination was assessed with Harrell’s C-index over the full follow-up, with bias-corrected and accelerated (BCa) bootstrap 95% confidence intervals (CIs). A single prespecified paired comparison was performed at the closest available horizons (Sarculator 5-year vs. MSKCC 4-year); because the two tools predict at different horizons, this contrast does not evaluate identical prediction targets and was interpreted cautiously. Calibration was assessed graphically. Given only 12 events, all discrimination and calibration estimates were treated as exploratory, and uncertainty (95% CIs) was emphasized over p-values. At a median follow-up of 103 months, 12 patients (29.3%) had died, all from sarcoma; 5- and 10-year OS were 78.0% and 61.1%, respectively. The MSKCC nomogram showed a higher C-index point estimate (C-index ≈ 0.81 across horizons) than Sarculator (≈ 0.71). In the prespecified comparison, the difference in C-index was 0.096 (95% CI 0.004–0.225; Wald p = 0.042; paired bootstrap p = 0.072). Graphically, MSKCC predictions approximated observed survival most closely at 4 and 8 years, whereas Sarculator tended to overestimate survival, most notably at 10 years. In this small, single-center extremity STS cohort, the MSKCC nomogram showed a higher C-index point estimate than Sarculator, but with a wide confidence interval, discordant inferential tests, and evaluation at a different prediction horizon; together with only 12 events, these factors preclude any conclusion of comparative superiority. Sarculator tended to overestimate survival, although the small cohort prevents a precise assessment of calibration. These findings are exploratory and hypothesis-generating and require confirmation in larger, externally validated cohorts.