S167: prediction of relapse after allogeneic stem cell transplantation using individualized measurable residual disease markers; the prospective nordic study nmdsg14b

Background: One third of patients with myelodysplastic syndrome (MDS) relapse after allogeneic stem cell transplantation (HCT). Early detection of impending relapse would enable pre-emptive treatment and potentially reduce relapse risk but is limited by the lack of sensitive markers for measurable residual disease (MRD). We developed a pipeline where patient-specific mutations, as determined by a myeloid next generation sequencing (NGS) panel are tracked using digital droplet PCR (ddPCR). Aims: To evaluate if personalized MRD detection by ddPCR can predict clinical relapse earlier than conventional methods. Methods: The prospective study (NCT02872662) enrolled patients with MDS, MDS/MPN or MDS-AML with < 30% marrow blasts undergoing HCT. Patients were included before HCT, and serial bone marrow (BM) samples were collected every third month post-HCT for 2 years. Peripheral blood (PB) samples were collected monthly. MRD results were not available for the treating physician. Results: We screened 286 pts between 2016 and 2020, whereof 20 were excluded mainly due to lack of genetic aberration or no HCT performed. 266 pts were included from 12 HCT centers. Median age was 64 (18-78) years and 59% were male. Myeloid panel NGS screening identified a median of 2 (0-9) mutations. The most common mutations were TET2 (n=85), ASXL1 (n=73) and SRSF2 (n=59). Median time of follow up was 886 (4-1934) days. Sixty pts relapsed after a median of 189 (53-1281) days and 46 died due to non-relapse mortality after a median of 121 (4-1036) days. Remaining pts (n=160) were in continuous complete remission (CCR) after a median follow-up of 1053 (479-1934) days. Estimated 1 and 2y overall survival was 79%, and 71%, respectively, while estimated 1 and 2y relapse-free survival (RFS) was 75% and 66%, respectively. MRD data was missing in 46 pts; no post-HCT samples available (n=15), no mutation detected (n=14) and difficulties to design ddPCR primers (n=11). 221 pts were available for MRD analysis with a median number of 4 (0-13) and 5 (0-23) samples from BM and PB, respectively. Of 53 clinical relapses with MRD results available, 42 were preceded by pos MRD (>0.1%) with a median of 70 (range 20-425) days between first pos MRD and clinical relapse. For the 11 remaining pts, 8 were inadequately sampled with a median time of 189 (82-397) days between last sampling and clinical relapse. One patient had an extramedullary relapse only. Of 31 pts who died without relapse, 19 were consistently MRD neg, while 5 were borderline positive (MRD > 0.1% and <0.5%) during the first 100 days but negative thereafter. Four MRD+ patients died without clinical relapse. Three pts were initially MRD+ but turned negative, all of which had chronic GVHD (cGVHD). Of 136 CCR patients, 94 were consistently MRD neg; 26 were borderline pos (MRD > 0.1% and <0.5%) during the first 100 days followed by neg samples; 16 were MRD positive (either > 0.5% during the first 100d or > 0.1% after 100d) of which 10 had a transition from pos to neg samples (all had cGVHD); one patient was treated for a molecular relapse detected by clinical routine method (FISH) and five patients were MRD positive at time of last follow-up. MRD used as a time-dependent co-variate was negatively associated with RFS (HR 7.1, p<0.01). Estimated cumulative incidence of relapse and non-relapse mortality 2y after pos MRD was 60% and 7% respectively (see figure). Image:Summary/Conclusion: We report the development of a highly functional personalized MRD pipeline based on patient-specific mutations showing a high sensitivity to predict relapse and relapse-free survival.