Development and validation of prognostic nomograms for metastatic gastrointestinal stromal tumour treated with imatinib.

PURPOSE Metastatic gastrointestinal stromal tumour (GIST) is generally an incurable disease with variable response to imatinib. We aimed to develop prognostic nomograms to predict overall survival (OS) and progression-free survival (PFS) for patients treated with imatinib. METHODS Nomograms were developed in a training cohort (n=330) of patients treated in a randomised trial (EORTC-ISG-AGITG 62005 phase III study) using Cox regression models, and validated in patients (n=236) treated in routine clinical care from six referral centres. Nomogram performance was assessed by calculating the c statistic. A classification based on the nomograms' scores was generated to group patients according to risk. RESULTS Nomogram risk factors for OS and PFS were size of the largest metastasis, tumour genotype, primary tumour mitotic count, haemoglobin and blood neutrophil count at commencement of imatinib. The nomograms predicted survival with a c statistic of 0.75 (training) and 0.62 (validation) for OS, and 0.69 (training) and 0.62 (validation) for PFS. When tested in the validation cohort, the nomograms discriminated well the high and intermediate risk from low risk patients (hazard ratio [HR] for OS 3.83, 95% confidence interval [CI] 1.71-8.56; and 2.48, 95% CI 1.12-5.50; for PFS 2.84, 95% CI 1.66-4.87; and 1.45, 95% CI 0.87-2.41, respectively). CONCLUSION The nomograms predicted the risk of GIST progression and death with good discrimination of risk groups, and may be of value for patient counselling and risk stratification.