Clinicopathologic Characteristics and Prognosis of PDGFRA -Mutant Gastrointestinal Stromal Tumors: A Large-Scale, Multi-Institutional, Observational Study in China

Background: To evaluate clinicopathologic features, adjuvant therapy efficacy, and prognosis of post-complete resection in PDGFRA-mutant gastrointestinal stromal tumor (GIST) patients and establish a relapse-free survival (RFS) prognostic model for this GIST subgroup. Methods: This retrospective study used data from primary PDGFRA-mutant GIST patients who underwent complete resection (2005-2019) at 16 large-scale medical centers in China. Propensity score matching (PSM) was applied to adjust for confounding. Multivariate Cox regression models with stepwise were performed to build the prediction model, in which the potential predictors were available in routine clinical practice and using the risk score functions. The prediction model was cross-validated by calibration histogram and time-dependent receiver operating characteristic curves. Findings: A total of 280 PDGFRA-mutant (172 D842V-mutant; 108 non-D842V-mutant) GIST patients post-complete resection were enrolled. The 1-, 3-, and 5-year RFS rates of patients were 95.9, 91.2, and 89.5%, respectively. The RFS of the non-D842V-mutant group was superior to that of the D842V group (P=0.033). The RFS in intermediate/high-risk non-D842V-mutant GIST patients who received adjuvant therapy with or without imatinib did not significantly differ before and after PSM. Multivariate analysis revealed that D842V mutation (P=0.017), non-gastric tumor (P 5% (P=0.005) were the independent variables influencing PDGFRA-mutant GIST patient prognosis. The scoring model showed the predicted and actual cumulative 1-, 3- and 5-year follow-up relapse rates fit well. Interpretation: PDGFRA-mutant GIST patients had a favorable prognosis after surgery. Non-D842V-mutant patients might have better prognoses than D842V-mutant patients. The beneficial effect of adjuvant imatinib for non-D842V-mutant patients is negligible. The prognostic model demonstrated favorable prediction accuracy, suggesting its clinical utility. Funding Statement: This study was supported by the National Natural Science Foundation of China (No.81874184). Declaration of Interests: None to declare. Ethics Approval Statement: The institutional review boards of each participating institution approved the study and deemed that a separate informed consent was not necessary for this study.