Clonal Relapse Dynamics in Acute Myeloid Leukemia Following Allogeneic Hematopoietic Cell Transplantation

• NGS-MRD monitoring from PB after alloHCT detects up to 38% and 64% of relapses with a monitoring interval of every 3 or 1 months.

• Known mutations from diagnosis including DTA mutations show the longest lead time to relapse post-alloHCT and are useful for MRD monitoring.

Patients with acute myeloid leukemia (AML) who experience relapse following allogeneic hematopoietic cell transplantation (alloHCT) face unfavorable outcomes regardless of the chosen relapse treatment. Early detection of relapse at the molecular level by measurable residual disease (MRD) assessment enables timely intervention, which may prevent hematological recurrence of the disease. It remains unclear whether molecular MRD assessment can detect MRD before impending relapse and, if so, how long in advance. This study elucidates the molecular architecture and kinetics preceding AML relapse by utilizing error-corrected next-generation sequencing (NGS) in 74 AML patients relapsing after alloHCT evaluating 140 samples from peripheral blood collected 0.6 to 14 months before relapse. At least one MRD marker became detectable in 10%, 38%, and 64% of patients at 6, 3, and 1 months prior to relapse, respectively. By translating these proportions into monitoring intervals, 38% of relapses would have been detected through MRD monitoring every 3 months, while 64% of relapses would have been detected with monthly intervals. The relapse kinetics after alloHCT are influenced by the functional class of mutations and their stability during molecular progression. Notably, mutations in epigenetic modifier genes exhibited a higher prevalence of MRD positivity and greater stability before relapse, while mutations in signaling genes demonstrated a shorter lead-time to relapse. Both DTA and non-DTA mutations displayed similar relapse kinetics during the follow-up period after alloHCT. Our study sets a framework for MRD monitoring after alloHCT by NGS supporting monthly monitoring from peripheral blood using all variants that are known from diagnosis.