[External validation and incremental value assessment of the RR6 prognostic model in Chinese myelofibrosis patients treated with ruxolitinib: focusing on overall survival].

Objective: To validate the Response to Ruxolitinib After 6 Months (RR6) prognostic model in Chinese patients with myelofibrosis (MF) treated with ruxolitinib and assess its incremental prognostic value when integrated with high molecular risk mutations (HMR(mt)) and/or RAS pathway mutations (RASp(mt)) . Methods: Altogether, 153 patients with myeloproliferative neoplasm-associated MF treated with ruxolitinib at our hospital from May 2016 to February 2024 were retrospectively enrolled. The RR6 prognostic model was applied for prognostic stratification and OS analysis, and calibration curves were plotted to evaluate the model's calibration. The predictive efficacy and clinical benefit of the RR6 model and its integrated models were comprehensively assessed and compared using the concordance index (C-index), time-dependent area under the curve (AUC), net reclassification improvement (NRI), and decision curve analysis (DCA) . Results: Altogether, 153 patients were diagnosed with primary myelofibrosis (n=114), post-polycythemia vera (post-PV) MF (n=17), and post-essential thrombocythemia (post-ET) MF (n=22). Patients had a median follow-up time from the initiation of ruxolitinib treatment of 44.6 (IQR: 27.3-64.5) months; 106 (69.3%) patients survived, 47 (30.7%) died, and 94 (61.4%) were still receiving ruxolitinib as of the last follow-up. According to the RR6 model, 28 (18.3%), 92 (60.1%), and 33 (21.6%) patients were classified as having low, intermediate, and high risks, respectively, with the estimated median overall survival (mOS) not reached for both the low- and intermediate-risk groups and 32 (95% CI: 24.0-39.5) months for the high-risk group (P=0.012). The calibration curves indicated that the RR6 model demonstrated good calibration performance at 1, 2, and 3 years. Among 102 patients with next-generation sequencing data, the integrated RR6+HMR(mt)+RASp(mt) model demonstrated the best predictive efficacy, with the highest C-index (0.735) and AUC value. The NRI also confirmed that this model significantly improved risk reclassification at 1 and 2 years (P=0.016, P<0.001). Moreover, DCA showed significant net clinical benefit. Conclusion: The present study confirms that the RR6 prognostic model has favorable prognostic predictive ability in Chinese patients with MF and can be utilized for the early identification of high-risk patients. The integration of HMR(mt) and RASp(mt) may enhance the predictive power of the RR6 model.