Which is the optimal risk stratification system for surgically treated localized primary GIST? Comparison of three contemporary prognostic criteria in 171 tumors and a proposal for a modified armed forces institute of pathology risk criteria

Published online June 11, 2008.BACKGROUND: This study aims to validate and compare the performance of the National Institute of Health (NIH) criteria, Huang modified NIH criteria, and Armed Forces Institute of Pathology (AFIP) risk criteria for gastrointestinal stromal tumors (GISTs) in a large series of localized primary GISTs surgically treated at a single institution to determine the ideal risk stratification system for GIST. METHODS: The clinicopathological features of 171 consecutive patients who underwent surgical resection for GISTs were retrospectively reviewed. Statistical analyses were performed to compare the prognostic value of the three risk criteria by analyzing the discriminatory ability linear trend, homogeneity, monotonicity of gradients, and Akaike information criteria. RESULTS: The median actuarial recurrence-free survival (RFS) for all 171 patients was 70%. On multivariate analyses, size >10 cm, mitotic count >5/50 high-power field, tumor necrosis, and serosal involvement were independent prognostic factors of RFS. All three risk criteria demonstrated a statistically significant difference in the recurrence rate, median actuarial RFS, actuarial 5-year RFS, and tumor-specific death across the different stages. Comparison of the various risk-stratification systems demonstrated that our proposed modified AFIP criteria had the best independent predictive value of RFS when compared with the other systems. CONCLUSION: The NIH, modified NIH, and AFIP criteria are useful in the prognostication of GIST, and the AFIP risk criteria provided the best prognostication among the three systems for primary localized GIST. However, remarkable prognostic heterogeneity exists in the AFIP high-risk category, and with our proposed modification, this system provides the most accurate prognostic information.Brian K. P. Goh, Pierce K. H. Chow, Wai-Ming Yap, Sittampalam M. Kesavan, In-Chin Song, Pradeep G. Paul, Boon-Swee Ooi, Yaw-Fui A. Chung, and Wai-Keong Won