Leukemic Stem Cells Persistence Measured By Multiparametric Flow Cytometry Is a Biomarker of Poor Prognosis in Adult Patients with Acute Myeloid Leukemia

Introduction: Despite the recent advances in chemotherapy regimens, relapse still substantially affects prognosis of intensively treated adult acute myeloid leukemia (AML) patients. There is growing evidence that a residual populations of leukemic cells may survive chemotherapy and outgrow, eventually causing relapse. These chemo-resistant cells are particularly abundant in the fraction of leukemic stem cells (LSC), which are endowed with pronounced self-renewal properties allowing to initiate and maintain leukemic clone. These cells can be detected, by high sensitivity multiparametric flow-cytometry (MFC), in the CD34⁺/CD38^- fraction of the leukemic populations and can be distinguished from normal hematopoietic stem cells by the expression of specific markers. In recent clinical trials, LSC have been demonstrated to represent a biomarker of poor prognosis when detected at diagnosis but also during treatment course. Moreover, the combined estimate of measurable residual disease (MRD) and LSC refines the prognostic assessment as determined by the sole application of MRD detection.

Aim: We analyzed a series of patients (pts) treated in the context of GIMEMA trials, in whom the LSC frequency was assessed by MFC at diagnosis. Pts with measurable levels of LSC were tested again after the consolidation cycle. At the same timepoint "standard" MRD was also determined. The purpose of the study was to demonstrate a correlation between LSC burden at baseline and prognosis in terms of overall (OS) and disease-free survival (DFS). Furthermore, we wanted to investigate the relationship between LSC and "standard" MRD persistence (>0.035%) after consolidation, and possible correlation with outcome.

Methods: LSC were evaluated by MFC as described elsewhere (Terwijn, PLoS 2014). LSC were quantified exploiting the expression of the C-type lectin-like molecule-1 (CLL1) and applying a sequential gating strategy that contained the CD34+/CD38- population. Pts were defined as LSC negative (LSC^neg) in case of zero LSC count, LSC^low or LSC^high when LSC were >0<0,03% or >0.03%, respectively. After consolidation, any level >0 was considered as a LSC persistence. Methods of analysis and thresholds were set according to previous publications (Zeijlemaker, Leukemia 2019).

Results: We analyzed 130 pts with de novo AML, in whom LSC determination was available at the baseline. Fifty-nine (45,4%) pts were LSC^neg, 49 (37,7%) LSC^low, 22 (16,9%) LSC^high. We did not observe any correlation between baseline LSC level and genetic/cytogenetic risk at diagnosis. There was not a significant difference in terms of OS duration according to the 3 LSC levels, however, pts who were LSC^high had the shortest OS (36-month estimate OS of 71.5% vs. 65.4 % vs 52.4 % for the LSC^neg, LSC^low and LSC^high categories respectively; p=0.21). A statistically significant difference, regardless of the belonging to the LSC^low or LSC^high category was observed when we focus on the subgroup of 30 pts with intermediate-risk AML, with a 36-month estimate OS of 76% vs. 77.8% vs 25% for the LSC^neg, LSC^lowand LSC^high categories respectively (p=0.023) (Figure 1A). In 19 patients, LSC persistence was assessed at the post-consolidation time-point. Nine LSC^high pts who failed to eradicate residual LSC at this timepoint had a worse outcome as compared to those belonging to the same category but achieving a LSC clearance or those who were LSC^low (36-month OS of 62.5% vs. 59.2% vs. 66.7% vs. 25% for the LSC^low converted into LSC^neg, LSC^low not converted into LSC^neg, LSC^high converted into LSC^neg and LSC^high not converted into LSC^neg categories, respectively; P=0.062) (Figure 1B). In 27 pts LSC and "standard" MRD determination was available. LSC persistence determined a worse 3-years OS both in MRD negative (66.7% vs 85.7%, p=0.44) and MRD positive pts (<20% vs 75.0%, p=0.041).

Conclusions: In line with the experience of other European groups, we demonstrated that MFC monitoring of LSC is feasible and provides prognostic information when performed at diagnosis and during treatment course. MFC assessment of LSC also offers the opportunity to monitor pts who lack aberrant phenotypes suitable for "standard" MRD investigation. When the 2 approaches - standard "MRD" and LSC assessment - are combined together, the prognosis prediction of AML can be further refined. Finally, LSC assessment can potentially represent an effective tool to monitor the effect of LSC targeting agents.

View largeDownload slide

View largeDownload slide

Close modal