Distinguishing optimal esophagectomy candidates in elderly patients: A nomogram based on propensity score matching.

Background: A survival benefit from esophagectomy was observed in elderly patients. But it's unclear how to identify specific patients who can benefit. Thus, we aimed to establish a predictive model to identify optimal candidates for esophagectomy.

Methods: Patients (age ≥75 years) with esophageal cancer in Surveillance, Epidemiology and End Results (SEER) database were used to establish the predictive model. Propensity-score matching (PSM) was applied to eliminate the imbalance between esophagectomy group and non-esophagectomy group. We hypothesized that elderly patients could benefit from esophagectomy with longer cancer specific survival (CSS) time than those who did not receive esophagectomy. Patients received surgery were divided into beneficial group and non-beneficial group according to the median CSS time of non-esophagectomy group. Prognostic factors affecting patients' long-term survival were identified. Among esophagectomy group, a logistic regression model based on these factors was established to build a nomogram.

Results: A total of 7,025 eligible patients were extracted from the SEER database, with 831 patients received esophagectomy. Surgery was independently associated with better long-term survival (median CSS time in the matched population: 35 vs. 8 months, p < 0.001). As a result, 361 (68.6%) patients were divided into beneficial group (CSS >8 months). Factors including age, tumor site, histology, differentiation grade, TNM stage, and tumor size were used to formulate the nomogram, which was named as esophagectomy candidates screening score (ECSS). The validation from two aspects showed the model a useful and stable one.

Conclusion: A predictive model was established to distinguish optimal candidates for esophagectomy among elderly patients with EC.