External Validation of a Score Predictive of Recurrence after Radical Surgery for Non-Cardia Gastric Cancer: Results of a Follow-Up Study

• Published in: Journal of the American College of Surgeons Vol. 221; no. 2; pp. 280 - 290
• Main Authors: Marrelli, Daniele, MD, Morgagni, Paolo, MD, de Manzoni, Giovanni, MD, Marchet, Alberto, MD, Baiocchi, Gian Luca, MD, Giacopuzzi, Simone, MD, Coniglio, Arianna, MD, Mocellin, Simone, MD, Saragoni, Luca, MD, Roviello, Franco, MD
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.08.2015
• Subjects: Adult; Aged; Cardia; Decision Support Techniques; Female; Follow-Up Studies; Gastrectomy; Humans; Kaplan-Meier Estimate; Logistic Models; Male; Middle Aged; Neoplasm Recurrence, Local - diagnosis; Neoplasm Recurrence, Local - etiology; Prognosis; Prospective Studies; Risk Assessment; Risk Factors; ROC Curve; Sensitivity and Specificity; Stomach Neoplasms - surgery; Surgery
• ISSN: 1072-7515; 1879-1190
• DOI: 10.1016/j.jamcollsurg.2015.03.042

Background A score predictive of tumor recurrence after radical surgery for non-cardia gastric cancer was previously developed in Italian centers. The aim of this study was to validate the score in a consecutive cohort with prospectively collected follow-up data. Study Design Of 1,178 patients surgically treated between 1998 and 2006, six hundred and thirty-five patients who fulfilled the selection criteria and completed the follow-up program were available for analysis. The score value for each patient was calculated using the formula obtained from a logistic regression model. Discrimination and calibration of the score in the validation group were evaluated and compared with the data of 438 patients in the study group where the score was developed. Results Most patients in both groups had very low or very high score values. In the validation group, the observed recurrence rates ranged from 5% to 92% in different score strata. The area under the receiver operating characteristic curve was 0.889 (95% CI, 0.864–0.914; p < 0.001), indicating a high discrimination value of the score for recurrence. A good calibration was observed by comparing the predicted risk with the actual risk of recurrence. With a score cut-off value of 50, sensitivity, specificity, and overall accuracy were 74%, 86%, and 81%, respectively. An inverse correlation between the time to recurrence and score level was also estimated ( R2  = 0.119; p < 0.001). Conclusions The high predictive value of the score was validated in a consecutive cohort. These results might allow the introduction of the score in clinical practice for Western patients.