Antigen Expression Varies Significantly between Molecular Subgroups of Acute Myeloid Leukemia Patients: Clinical Applicability Is Hampered by Establishment of Relevant Cutoffs

Abstract

<b><i>Introduction:</i></b> In this single-center study of 268 acute myeloid leukemia (AML) patients, we have tested if a subset of 4 routinely employed immunophenotypic stem cell-associated markers correlated with the presence of recurrently mutated genes and if the markers were predictive for mutational status. <b><i>Methods:</i></b> Immunophenotypic data from 268 diagnostic AML samples obtained in 2009–2018 were analyzed retrospectively for the antigens CD34, CD117, CD123, and CLEC12A. Correlation between immunophenotypes and mutations was analyzed by Fischer’s exact test. Clinical applicability of the markers for predicting mutational status was evaluated by receiver operating characteristics analyses, where an area under the curve (AUC) of at least 0.85 was accepted as clinically relevant. <b><i>Results:</i></b> For a number of genes, the antigen expression differed significantly between mutated and wild-type gene expression. Despite low AUCs, CD123 and CLEC12A correlated with <i>FLT3</i>+<i>NPM1−</i> and <i>FLT3</i>+<i>NPM1</i>+. Three subsets met the AUC requirements (CD34+, CD34+CD117+, and CD34−CD117+) for predicting <i>FLT3−NPM1</i>+ or <i>FLT3</i>+<i>NPM1</i>+. <b><i>Conclusion:</i></b> The value of immunophenotypes as surrogate markers for mutational status in AML seems limited when employing CD123 and CLEC12A in combination with CD34 and CD117. Defining relevant cutoffs for given markers is challenging and hampered by variation between laboratories and patient groups.