Personalized Transplant Decision Making for Myelofibrosis in the Era of Molecular Genetics and JAK Inhibition

Le résumé fourni par la source

Background Over the past 15 years, the field of myelofibrosis (MF) has seen remarkable improvements in terms of risk stratification, with the development of clinical, clinico-genomic, as well as treatment-specific models. JAK inhibitors revolutionized the treatment of MF, while allogeneic hematopoietic cell transplantation (HCT) is the only potentially curative treatment for patients with MF. Several issues must be considered when evaluating the benefits and risks of treatments for patients with MF, with the timing of HCT being a crucial question. Here, we aimed to develop a uniform decision model to define the optimal timing of treatment in MF patients based on clinical and genomic information in the modern era of JAK inhibition. Patients and Methods We studied a retrospective, international cohort of 1913 patients with primary (PMF, 58%) or secondary MF (SMF, 42%), of whom 1010 patients underwent HCT. All patients who did not undergo HCT were treated with ruxolitinib. First, we developed a multi-state model to estimate transition hazards between different disease-specific categories, using the DIPSS and MIPSS70 (for PMF), and MYSEC-PM (for SMF). Next, we simulated a randomized clinical trial, using Bayesian-guided artificial intelligence algorithms, where subjects are selected to receive HCT at different time points, stratified according to transplant-specific risk (using MTSS) and response to ruxolitinib (using the RR6 model). Results were used to estimate the average survival time over a 6-year time horizon. Results When considering patient life expectancy for patients who received an HCT in traditional modelling, earlier transplantation was associated with better outcome for higher risk disease, irrespective of PMF or SMF. However, considering treatment-specific risk significantly influenced decision modelling towards earlier or delayed transplantation (P<0.001). Considering the scenario of PMF, a low/intermediate MTSS was associated with better outcomes for early transplantation in all DIPSS categories as well as MIPSS70 intermediate/high risk. Ideal timing of HCT for achieving prolonged life expectancy of patients was suggested within first 3 years from diagnosis for DIPSS low/intermediate-1, 2 years for MIPSS70 intermediate, and within 1 year from diagnosis for both DIPSS intermediate-2/high and MIPSS70 high. In contrast, considering a scenario of MTSS high/very high risk, prolonged life expectancy for transplantation was only found for MIPSS70 intermediate/high and DIPSS intermediate-2/high, suggesting delay HCT by an average of 0.5 years. Considering SMF, a scenario of low/intermediate MTSS was associated with better outcomes for transplantation only for a MYSEC-PM intermediate-2/high risk constellation. Ideal timing of HCT for achieving prolonged life expectancy of patients was suggested within first 1.5 years from diagnosis. In a next step considering response to ruxolitinib using the RR6 model in combination with the MTSS significantly affected timing of transplantation (P<0.001), suggesting earlier transplantation by an average of 0.5 years for RR6 high risk, while RR6 low risk suggested delayed transplantation by an average of 1 year. Importantly, high risk RR6 suggested earlier HCT for patients with intermediate DIPSS and MYSEC-PM, traditionally populations of unmet need for optimal decision making. Furthermore, additional genomic constellations significantly affected decision modelling (P<0.001). First, present ASXL1 drove the decision towards delayed transplantation for patients with MYSEC-PM low/intermediate-1 and MTSS low/intermediate risk. Second, patients with TP53 mutation (irrespective of disease type and risk) benefitted from early transplantation within 1 year of diagnosis. Last, we developed an online tool (with an extension for mobile application) as a uniform platform incorporating all revelant information, helping clinicians and patients alike navigate through the decision-making process for both PMF and SMF, with respect to onset and treatment-specific prognosis. Conclusion We provided evidence for the relevance of combining all clinico-genomic and treatment-specific features into the treatment decision making process, specifically regarding the optimal timing of HCT, allowing personalized decision-making in MF.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.