Background: Relapse is the most common reason for failure of allogeneic stem cell transplant (SCT). Identifying patients at increased relapse risk is crucial for improving outcomes. Prognostic somatic mutations have been identified in AML but their prognostic value after SCT is unknown. We hypothesize that next generation sequencing (NGS) can predict SCT outcome in AML. Methods: We performed NGS for 33 hematologic malignancy associated genes in 62 adult patients treated with SCT for de novo AML (2003-2013). Relapse free survival (RFS) was compared by Cox regression with adjustment for cytogenetic risk, conditioning intensity, donor type and graft source. Results: At least 1 mutation was identified in 56 (90%) patients (median 2, range 0-6) including 35 of 36 (97%) patients with intermediate cytogenetics (median 3, range 0-6). Among patients with intermediate cytogenetics transplanted in CR, poor prognostic mutations were found in 9 of 15 FLT3ITD negative patients (2 DNMT3A; 3 DNMT3A + IDH; 1 DNMT3A + TET2 + IDH2; 1 TET2; 1 TET2 + PHF6; 1 ASXL1); among FLT3ITD positive patients, 4 of 9 were additionally DNMT3A mutant which may confer added risk. Among patients with favorable cytogenetics transplanted in CR2, 2 of 8 patients had kit mutations; there were no mutations in DNMT3A, IDH, ASXL1, TET2, or PHF6. Among patients with unfavorable cytogenetics, 4 of 13 had additional poor prognostic mutations (1 FLT3ITD +TET2, 1 TET2, 2 TP53). In FLT3ITD negative patients with any cytogenetic risk, DNMT3A mutant status significantly worsened RFS (adjusted HR=4.76 [95% CI 1.34-16.9]; p=0.016). Conclusions: NGS identifies a high frequency of potentially prognostic mutations in SCT patients. Mutant DNMT3A may negatively impact prognosis in FLT3ITD negative AML post SCT. NGS testing of a larger cohort is warranted and underway. Identifying patients at high risk for relapse by molecular profiling may allow early intervention.

Abbreviations: VUS, variant of uncertain significance.