BACKGROUND: Treatment decisions for gastrointestinal stromal tumors (GIST) are frequently guided by tumor characteristics. An accurate prediction of recurrence is important to determine the benefit from targeted therapy. Our goal was to compare the concordance of three validated risk stratification schemes with observed outcomes in patients undergoing resection for GISTs. METHODS: Patients who underwent surgery for GISTs from 2001 to 2011 at a tertiary centre were identified. Survival was evaluated using the Kaplan-Meier product-limit method. Cox proportional hazard models were used to obtain predicted recurrence for each system and concordance indices were calculated. RESULTS: Of 110 patients identified, 77 (70.0%) had surgery and 29 (26.4%) also received adjuvant therapy. The majority of patients had tumors that were very low (4.5%), low (32.7%), or intermediate (22.7%) in terms of malignant potential. R0 resection was achieved in 89.1% of cases. Observed 2-year and 5-year recurrence rates were significantly lower than those predicted by the Memorial Sloan Kettering Cancer Center nomogram (7.6% vs. 19.3% and 18.4% vs. 27.0%); however, it was the most favorable tool compared to the US National Institutes of Health (NIH)-consensus (P = 0.0017) and modified NIH-consensus (P < 0.001), with a concordance index of 0.811. CONCLUSION: Development of a novel predictive tool that includes additional prognostic factors may better stratify recurrence following resection for GIST.
BACKGROUND: Treatment decisions for gastrointestinal stromal tumors (GIST) are frequently guided by tumor characteristics. An accurate prediction of recurrence is important to determine the benefit from targeted therapy. Our goal was to compare the concordance of three validated risk stratification schemes with observed outcomes in patients undergoing resection for GISTs. METHODS:Patients who underwent surgery for GISTs from 2001 to 2011 at a tertiary centre were identified. Survival was evaluated using the Kaplan-Meier product-limit method. Cox proportional hazard models were used to obtain predicted recurrence for each system and concordance indices were calculated. RESULTS: Of 110 patients identified, 77 (70.0%) had surgery and 29 (26.4%) also received adjuvant therapy. The majority of patients had tumors that were very low (4.5%), low (32.7%), or intermediate (22.7%) in terms of malignant potential. R0 resection was achieved in 89.1% of cases. Observed 2-year and 5-year recurrence rates were significantly lower than those predicted by the Memorial Sloan Kettering Cancer Center nomogram (7.6% vs. 19.3% and 18.4% vs. 27.0%); however, it was the most favorable tool compared to the US National Institutes of Health (NIH)-consensus (P = 0.0017) and modified NIH-consensus (P < 0.001), with a concordance index of 0.811. CONCLUSION: Development of a novel predictive tool that includes additional prognostic factors may better stratify recurrence following resection for GIST.