Multigene Measurable Residual Disease Assessment Improves Acute Myeloid Leukemia Relapse Risk Stratification in Autologous Hematopoietic Cell Transplantation

We report here the largest study to date of adult patients with acute myeloid leukemia (AML) tested for measurable residual disease (MRD) at the time of autologous hematopoietic cell transplantation (auto-HCT). Seventy-two adult patients who underwent transplantation between 2004 and 2013 at a single academic medical center (University of California San Francisco) were eligible for this retrospective study based on availability of cryopreserved granulocyte colony–stimulating factor (GCSF)–mobilized autologous peripheral blood progenitor cell (PBPC) leukapheresis specimens (“autografts”). Autograft MRD was assessed by molecular methods (real-time quantitative PCR [RQ-PCR] for Wilms tumor 1 (WT1) alone or a multigene panel) and by multiparameter flow cytometry (MPFC). WT1 RQ-PCR testing of the autograft had low sensitivity for relapse prediction (14%) and a negative predictive value of 51%. MPFC failed to identify MRD in any of 34 autografts tested. Combinations of molecular MRD assays, however, improved prediction of post–auto-HCT relapse. In multivariate analysis of clinical variables, including age, gender, race, cytogenetic risk category, and CD34⁺ cell dose, only autograft multigene MRD as assessed by RQ-PCR was statistically significantly associated with relapse. One year after transplantation, only 28% patients with detectable autograft MRD were relapse free, compared with 67% in the MRD-negative cohort. Multigene MRD, while an improvement on other methods tested, was however suboptimal for relapse prediction in unselected patients, with specificity of 83% and sensitivity of 46%. In patients with known chromosomal abnormalities or mutations, however, better predictive value was observed with no relapses observed in MRD-negative patients in the first year after auto-HCT compared with 83% incidence of relapse in the MRD-positive patients (hazard ratio, 12.45; P = .0016). In summary, increased personalization of MRD monitoring by use of a multigene panel improved the ability to risk stratify patients for post–auto-HCT relapse. WT1 RQ-PCR and flow cytometric assessment for AML MRD in autograft samples had limited value for predicting relapse after auto-HCT. We demonstrate that cryopreserved autograft material presents unique challenges for AML MRD testing because of the masking effects of previous GCSF exposure on gene expression and flow cytometry signatures. In the absence of information regarding diagnostic characteristics, sources other than GCSF-stimulated PBSC leukapheresis specimens should be considered as alternatives for MRD testing in AML patients undergoing auto-HCT.