Risk stratification systems for surgically treated localized primary Gastrointestinal Stromal Tumors (GIST). Review of literature and comparison of the three prognostic criteria: MSKCC Nomogramm, NIH-Fletcher and AFIP-Miettinen.

PURPOSE The discovery of Imatinib mesylate (Gleevec®) has revolutionized the treatment of GIST, increasing diseasefree survival (DFS) after complete surgical resection of a primary localized GIST and extending overall survival in metastatic disease. The definition of an accurate prognostic system is critical for the therapeutic decision making process. In literature, there are three main prognostic criteria F/NIH consensus, AFIP standards and modified NIH standards. In recent years were added various risk identification methods applying mathematical calculation model, including MSKCC risk nomogram, Rossi nomogram and Joensuu high Hotline Dengjun. Despite all these attempts, it seems that the recurrence risk probability still cannot be predicted accurately. The aim of our study was to assess and compare the real ability of these prognostic instruments in our single-centre clinical experience, and to define if the use of the MSKCC nomogram can bring benefits in the therapeutic decision. METHODS All data regarding 37 GIST, who underwent surgical resection from 1996 to 2011 in our institution were retrospectively reviewed. We selected only primary GIST without metastatic disease who underwent a radical resection (R0) but no other therapy. The literature data concerning GISTs prognostication criteria were reviewed. All patients were classified according to the three prognostic criteria (NIH, AFIP and Nomogram MSKCC) and the three instruments were compared with the Kaplan-Meier method. Then we compared the three criteria for their c-index value and we assessed the performance of the nomogram with the calibration test. RESULTS We observed 9 recurrences (24%) with an average time to relapse of 43 months; the median follow-up was 65 months. In the study selected sample occurred 5 relapses. The probability of relapsing after radical surgery was 7.9% (95% CI 0 - 17.3) at 2 years and 13.3% at 5 years (95% CI 0 - 26.4). The C-Index of the three risk assessment tools was 0.93 (95% CI 0.83-1) for the Nomogram at 5 years, 0.86 (95% CI 0.76-0.95) for the NIH risk criteria and 0.88 (95% CI 0.74-1) for the AFIP risk criteria. The calibration analysis of the nomogram showed an overestimating trend both at 2 and 5 years. CONCLUSION MSKCC nomogram seems to perform better than NIH, NIH modified and AFIP in our sample and can be used in clinical practice to predict the risk of recurrence, being especially helpful for the therapeutic decision making since it is at the same time simple to use and accurate. As showed from calibration, MSKCC doesn't seem to neglect relapses, even though it is not impeccable in predicting the RFS. Among the 2 older criteria AFIP was more precise than NIH, but considering size in not linear way represented a limit in comparison with the MSKCC Nomogram. All the three risk assessement tools criteria con sidered are capable to predict recurrence in high-risk GISTs while they performed worse in those with lower risk. MSKCC nomogram main limit remains the not linear consideration of mitotic count.