Absence of leukaemic CD34+ cells in acute myeloid leukaemia is of high prognostic value: a longstanding controversy deciphered.

Primary resistance and relapses after initial successful treatment are common in acute myeloid leukaemia and therefore outcome remains poor. More accurate risk group stratification and effective personalized risk adapted treatment are necessary to improve outcome. In the last two decades, controversial results have been published concerning the prognostic relevance of CD34 expression. In this study of 706 acute myeloid leukaemia patients, we established a new flow cytometric-based CD34-definition, without use of cut-off values. We discriminated CD34-positive (n = 548) and CD34-negative (n = 158) patients by the presence or absence of neoplastic CD34+ cells, respectively. CD34-status was defined using aberrant immunophenotypes and validated using molecular phenotypes. This new definition of CD34 enables strong prediction of treatment outcome in the entire patient group and in several risk subgroups. Previously observed discrepancies in prognostic impact of CD34 protein expression using cut-offs (5-20%) can now entirely be explained by considering the number of CD34-negative cases. In the total patient group, the absence of neoplastic CD34-positive cells is paralleled by low levels of minimal residual disease, suggesting relative therapy sensitivity and explaining longer survival. Overall, we present CD34 surface expression as a relatively simple, powerful and independent predictor of clinical outcome, now warranting incorporation in acute myeloid leukaemia risk stratification.