Background. Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the gastrointestinal tract. Radical surgery is the primary treatment for GIST. Unfortunately, 40–50% of patients relapse, mainly due to hepatic and peritoneal metastases. Currently, the treatment of choice for locally advanced, inoperable or metastatic GIST is the use of tyrosine kinase inhibitors, including imatinib. GISTs are a group of tumors with various morphological, pathological and molecular features as well as different clinical courses, therefore their biological course is difficult to determine. Nevertheless, we currently have 5 classifications that assess the risk of relapse after surgery. The aim of this study was to analyze prognostic factors with regard to the risk of recurrence and overall survival, and to compare the clinical reliability of the recurrence risk classifications developed so far with an attempt to present a new classification including the genotype of primary GIST.

Patients and methods. The material consisted of a group of 697 patients with primary GIST treated with the intention to cure, collected prospectively as part of the GIST clinical registry, Department of Melanoma and Soft Tissue and Bone Sarcomas, Maria Sklodowska-Curie National Research Institute of Oncology in Warsaw. All patients were classified based on 5 existing recurrence risk classifications. Univariate and multivariate analysis were performed for disease-free survival (DFS) and overall survival (OS). The relationships of the following factors with DFS and OS were assessed: sex, age, primary tumor mutational status, primary tumor location, primary tumor size, number of mitoses/50 HPF, surgical margins and the presence of tumor rupture. The next analysis concerned the comparison of the accuracy of existing recurrence risk classifications. The analysis was performed using ROC curves and a new classification model was proposed including mutation analysis as well as factors such as gender and age for selected existing recurrence risk assessment models. Results. Univariate and multivariate analyses showed statistical significance of variables such as male sex (P = 0.02), mitotic index 5–10/50 HPF and > 10/50 HPF (P < 0.001), primary tumor size 5–10 cm and > 10 cm (P < 0.001), primary tumor location outside of the stomach (P < 0.001), R1 surgery (P < 0.001), tumor rupture (P < 0.001), and the presence of mutations in the KIT gene exon 11 including deletion 557–558 and the KIT gene exon 9 (P = 0.009) as negative prognostic factors affecting disease recurrence. Five-year disease-free survival rate was 57.3%. Median DFS was 76 months. Negative prognostic factors for OS are: age < 40 (P = 0.045), mitotic index 5–10/50 and > 10/50 HPF (P < 0.001), primary tumor size 5–10 cm and > 10 cm (P < 0.001), R1 surgery and tumor rupture (P < 0.001). All existing recurrence risk classifications showed prognostic value for assessing differences in DFS and OS, no significant differences were found between individual recurrence risk classifications. In addition, the reliability of all these classifications was improved by adding gender, age and mutation status. The value added of mutation status for better risk assessment was most significant when used in intermediate risk groups according to different classifications (P < 0.01). Conclusion. All current GIST recurrence risk classifications allow for reliable assessment of recurrence risk. Mutations involving deletions (557–558) in the KIT gene exon 11 are most often present in the group at high risk of recurrence. Patients with confirmed mutations in the PDGFRA gene exon 18 and wild-type genotype have a favorable prognostic effect. The reliability of existing classifications for assessing the risk of relapse after GIST resection can be improved by adding mutation status, especially in groups at intermediate risk of relapse, which should facilitate therapeutic decisions in the context of adjuvant therapy.