Myeloid Progenitor Cells with Preleukemic Immunophenotypes Are Common in Myelodysplasia-Related and NPM1-mutated Acute Myeloid Leukemia and Should Not be Interpreted As Measurable Residual Disease

Background: Measurable residual disease (MRD) status is a robust indicator of outcomes in patients (pts) with acute myeloid leukemia (AML). Flow cytometric immunophenotyping (FCI) is among the most utilized methods for MRD assessment, particularly in AML without molecular markers of MRD (i.e. FLT3, NPM1, core binding factor fusions). The two main approaches used for MRD assessment by FCI include 1) identifying leukemia-associated immunophenotypes (LAIP); and 2) identifying deviation from normal (DFN). The latter may be confounded by the presence of persistent clonal hematopoiesis (CH) or underlying myelodysplastic (MDS) clones (Loghavi S, Br J Haematol. 2021), leading to non-definitive interpretation of results.

Methods: We searched for pts with AML treated at our institution between 2017 and 2022 who achieved CR/CRi and had at least one FCI report interpreted as “indeterminate” or “negative for MRD, but with immunophenotypically atypical myeloid blasts”. These were negative for LAIP but had CD34+ myeloid progenitors with DFN insufficient for a diagnosis of AML MRD, including alterations in intensity of of CD13, CD34, CD38, CD54, CD117, CD123 and HLA-DR; abnormal CD13 vs CD33 pattern; and/or dim/partial expression of CD5, CD7 and/or CD56 herein referred to as “DFN phenotype”. We divided these cases into “persistent” DFN if this phenotype was identified in the first adequate ( >200 CD34+ events) MRD-negative sample and “emergent” if DFN was identified after a previously definitive MRD-negative assessment in an adequate sample. We correlated these findings with baseline and post-remission cytogenetic and molecular features. Routine karyotyping and panel-based next-generation sequencing were performed on all samples at baseline and in the most pts at remission.

Results: At least 1 indeterminate MRD assessment was identified in 36/398 pts (9%) treated with intensive therapy (median age: 53 years [range, 25-68]) and in 60/276 pts (20%) treated with lower-intensity therapy (median age: 69.5 years [range, 38-81]). Upon retrospective review, 10 pts had molecular or cytogenetic evidence of AML MRD (e.g. FLT3, NPM1 mutation or CBFB, MECOM or KMT2A rearrangement) and were excluded from the analysis. Baseline AML subtypes for the rest of the cohort (n=86) included myelodysplasia-related (MR) AML (n=30, 35%), AML with NPM1m (n=30, 35%; 8 with MR co-mutations), TP53m (n=8, 9%), KMT2Ar (n=5, 6%), with germline predisposition (n=3, 3%; all with DDX41m), MECOMr (n=1, 1%) and AML, not otherwise specified (n=9, 10%). 72 (84%) pts had persistent and 14 (16%) had emergent DFN. The median time from the start of therapy to the initial identification of DFN was 27 days (range, 14-358 days) in the entire cohort and 148 days for the emergent cohort. The median percentage of aberrant cells was 1% (range, 0.01-3.8%). 79/86 (88%) and 63/86 (73%) pts had repeat cytogenetic/molecular testing at the time of identification of the DFN, respectively. 46/63 samples with DFN phenotype had residual mutations (median= 2; range, 1-6). The most common mutations involved DNMT3A (43%), TET2 (22%), SRSF2 (16%), ZRSR2 (14%), ASXL1 (13%), RUNX1 (10%), IDH1 and IDH2 (10% each), NRAS and U2AF1 (8% each), and TP53 (5%). 17/63 (27%) had no residual mutations; among these, 2 had -Y, and the 15 others had no clonal cytogenetic abnormalities. Among pts with relapsed disease (n=27/86, 31%), 1/17 (6%) in the low-intensity and 2/10 (20%) in the high-intensity group relapsed with the DFN phenotype, suggesting the other DFN phenotypes may represent pre-leukemic or MDS clones. A landmark analysis to compare survival outcomes of the study cohort compared with pts with definitive MRD positive or negative status is underway.

Conclusion: Atypical CD34+ myeloid blasts with deviation from normal (DFN) phenotypes are commonly observed in myelodysplasia-related AML and AML with NPM1 (with or without MR gene co-mutations) and are more commonly seen in pts treated with low-intensity regimens. Samples with blasts with DFN phenotypes may represent AML MRD with phenotypic shift or pre-leukemic and myelodysplastic clones. DFN is commonly associated with residual myelodysplasia-related gene mutations (e.g DNMT3A, TET2, ASXL1 (DTA), and IDH1/2). It is important to be aware of these findings to avoid overinterpretation as AML MRD, particularly as more MRD-directed therapies become available in the context of clinical trials for pts with AML.