Abstract:
Objective
To retrospectively analyze clinical and pathological factors related to the prognosis of gastrointestinal stromal tumor (GIST) patients, construct a dynamic risk assessment model for different treatment modalities, and provide evidence-based support for optimizing postoperative adjuvant treatment decisions.
Methods
A total of 250 GIST patients diagnosed between May 2009 and June 2019 were included. Univariate Kaplan-Meier survival analysis (Log-Rank test) was used to screen factors related to recurrence-free survival (RFS), and a prognostic prediction model and risk stratification were constructed based on the random survival forest (RSF) algorithm.
Results
In the group not receiving postoperative adjuvant therapy, male sex, intestinal origin, tumor size > 5cm, mitotic count > 5/50HPF, elevated Ki-67 proliferation index, epithelioid cell type, and KIT exon 11 codon 557/558/559 deletion mutation were significantly associated with recurrence. In the imatinib adjuvant therapy group, only intestinal origin, mitotic count > 5/50HPF,elevated Ki-67 proliferation index, and cell morphology were related to recurrence risk. Epithelioid cell type had a better survival outcome in the imatinib group, suggesting potential sensitivity to tyrosine kinase inhibitor therapy.
Conclusion
The constructed RSF dynamic prediction model has good performance, compensating for the NIH classification deficiency in the imatinib group. The dual characteristics of epithelioid cell type offer new evidence for histological subtype-guided precise treatment.
Key words:
Gastrointestinal Stromal Tumors,
Recurrence-Free Survival,
Random Survival Forest Model
Qiang Yuan, Wenrui Xiang, Yuan Lv, Zhangzhe Yan, Hanfang Zhu, Guang Chen, Liang Sun, Gang Chen, Suo Zhao. Study on prediction model of recurrence-free survival for gastrointestinal stromal tumors based on random survival forest[J]. Chinese Journal of Operative Procedures of General Surgery(Electronic Edition), 2025, 19(04): 438-441.