Prospective Validation of CD34+CD38- Leukemic Stem Cell Frequency in the HOVON-SAKK 132 Trial: Perspectives for Future Improvements

Clinical implementation of measurable residual disease (MRD) requires accurate MRD detection to assess relapse risk for acute myeloid leukemia (AML) patients. One possibility to further improve the prognostic ability of the current MRD assay is to include the quantification of relapse initiating cells such as the CD34+CD38- leukemic stem cells (LSC) in the assay. In the HOVON-SAKK 102 trial, detection of LSC by multiparameter (MFC) flow MRD at diagnosis and after 2^nd chemotherapy cycle (C2) has shown prognostic value (Zeijlemaker et al. Leukemia 2019). In addition, together with MRD detection, LSC could identify a category of patients with very high risk of relapse. Based on those data, improvements for LSC assessment measurements have been suggested, such as measuring more white blood cells (WBC) with the purpose of enhancing the sensitivity for the low frequency of CD34+CD38- cells. In the HOVON-SAKK 132 trial (Lowenberg et al. Blood Adv 2021), the one tube LSC assay was used that included LSC markers CD45RA, CD123, CD33, CD44, CLL-1, TIM-3, CD56, CD7, CD22 and CD11b (Zeijlemaker et al. Leukemia 2016). The one tube approach allows acquiring the suggested 4 million WBC for LSC-load estimation.

In the current study, we prospectively validated the prognostic value of LSC burden at the following time points, i.e. at diagnosis (764/905 patients) and after two cycles of chemotherapy (C2; 357/706 patients). The median WBC acquired was 2x10⁶ (range: 3x10⁴- 6x10⁶) at diagnosis and 3x10⁶ (range: 4x10⁵-5x10⁶) after C2. Median percentages of LSC at diagnosis and after C2 were respectively 0.004% (range: 0.0000%-26.85%) and 0.00002% (range: 0.0000%-1.36%). Similar to the HOVON-SAKK102 study, three prognostic groups based on the presence or absence of CD34 and/or LSC, showed significant associations between LSC burden at diagnosis and overall survival (OS) or relapse- free survival (RFS) (Figure 1A): 1. CD34negative patients without any measurable LSCs and Leukemia Associated Immunophenotypes on the CD34+ blasts fraction (CD34^negLSC=0 3 years (yrs) OS: 76.4% standard error (SE) 5; median OS not reached). 2. Patients with low percentages of CD34+CD38-LSCs < 0.03% (LSC^low; 3 yrs OS 62.8% SE 2; median OS not reached) 3. Patients with high percentages of CD34+CD38-LSCs ≥0.03% (LSC^high, 3 yrs OS 44.5% SE 3; median 25.6 months; 95% confidence interval (CI): 18.7-49.1). Multivariate analysis adjusted for age, WBC at diagnosis, AML type, ELN2017 risk group and -only for RFS- the number of chemotherapy cycles to attain complete remission, showed that LSC burden at diagnosis is a strong independent prognostic factor for OS and RFS (Figure 1B). After C2, before the consolidation treatment, a cut-off value of 0.000075% of LSC showed the best prognostic value of LSC (univariate, LSC^neg versus LSC^pos: OS, HR 1.79 95% CI: 1.25-2.56, P < 0.01; RFS, HR 1.55, 95% CI: 1.12-2.13, P < 0.01). This cut-off appeared also highly significant as a prognostic factor in former trials (Terwijn et al. PLoS One 2014; Zeijlemaker et al. Leukemia 2018). Multivariate analysis showed that also LSC status after C2 was an independent prognostic factor for OS and RFS (Figure 1B). By combining the LSC status after C2 (≥0.000075%) together with the MRD status as used in the HOVON-SAKK 132 (based on ≥0.1% MFC-MRD and/or NPM1-MRD), MRD^pos patients who are LSC^pos have worse prognosis in OS and RFS (univariate: MRD^negLSC^neg vs MRD^posLSC^pos; OS HR 3.2, 95% CI: 1.94-5.28, P < 0.01, RFS: HR 2.61 95% CI: 1.65-4.12, P < 0.01, multivariate: Figure 1B). Further analyses of the data are now directed towards potential influences of the cell amount of acquisition, ELN risk groups and MRD guidance in this trial. In conclusion, this prospective validation confirmed that assessing LSC burden, at diagnosis and after C2, remains to have additional prognostic value to MRD.