[A model to predict the recurrence of middle-high risk gastrointestinal stromal tumors based on preoperative fibrinogen and peripheral blood inflammatory indexes]

Objective: At present, the modified NIH classification commonly used in clinical practice is still insufficient for assessing the risk of postoperative recurrence in some patients with intermediate-high risk gastrointestinal stromal tumors (GIST). Through exploring risk factors for recurrence of intermediate-high risk GIST, this study establishes a predictive model for recurrence with more convenience and more precision in order to guide adjuvant therapy for intermediate-high risk GIST patients. Methods: A retrospective case-control study was carried out. Clinical and pathological data of 432 GIST patients who did not receive preoperative targeted treatment, underwent complete resection in the Union Hospital of Tongji Medical College of Huazhong University of Science and Technology from January 2005 to June 2018, and were diagnosed as intermediate- or high-risk based on modified NIH classification by postopertive pathology, were retrospectively analyzed. Cox regression model was used to idenitify independent risk factors of recurrence, and a recurrence risk scoring model was established. The receiver operating characteristic curve (ROC curve), consistency index (C-index) and calibration curve were used to evaluate the accuracy of the scoring model in predicting the recurrence of moderate-risk and high-risk GIST patients. Results: Among 432 GIST patients, 332 were diagnosed as high-risk and 100 as moderate-risk; 237 were males and 195 females with average age of (57.4±12.4) years. Of 432 patients, 211 cases (48.8%) had fibrinogen (FIB) >3.5 g/L; 85 cases (19.7%) had platelet to lymphocyte ratio (PLR)>272.5; 122 cases (28.2%) had neutrophil to lymphocyte ratio (NLR) > 4.2; 102 cases (23.6%) had systemic inflammatory reaction index (SIRI)> 2.7; 198 cases (45.8%) had tumor long diameter >8 cm and 108 cases (25.0%) had mitotic counts > 8/50 HPF. Cox multivariable analysis showed that FIB (HR=1.789, 95% CI: 1.058-3.027, P=0.030), PLR (HR=1.862, 95% CI: 1.067-3.249, P=0.029), SIRI (HR=1.790, 95% CI: 1.039-3.084, P=0.036), tumor long diameter (HR=1.970, 95% CI: 1.105-2.925, P=0.017) and mitotic counts (HR=2.187, 95% CI:1.211-3.950, P=0.009) were independent risk factors for recurrence in patients with middle-risk and high-risk GIST. These 5 factors were included in the risk scoring model, which was given a weight score of 58 points, 62 points, 58 points, 63 points, and 78 points, respectively. Patients with a total score of ≤ 78 points were classified as moderate-risk recurrence (group I), those of 78 to 136 points as high-risk recurrence (group II) and those of >136 points as very high-risk recurrence (group III). ROC curve showed that the area under the curve (AUC) of the scoring model was 0.730 and the C-index was 0.724 (95% CI:0.687-0.787). The calibration curves and the Kaplan-Meier curves of patients in the three groups revealed that this model had a good predictive accuracy. Conclusions: For intermediate-risk and high-risk GIST patients, the preoperative FIB >3.5 g/L, PLR > 272.5 and SIRI > 2.7 are independent risk factors of recurrence after surgery. The recurrence risk scoring model established by combining tumor long diameter, mitotic counts, FIB, PLR and SIRI can effectively predict the risk of postoperative recurrence and metastasis in moderate-risk and high-risk GIST patients.