Poster Session I: Acute lymphoblastic leukemia - Clinical
In 2014, we used flow-cytometry (FC) to describe a subtype of pediatric B cell precursor acute lymphoblastic leukemia (BCP ALL) with a lineage switch of leukemic blasts into monocytoid cells (swALL) during early treatment (Slamova et al., 2014). Although part of the swALL cases harbored deletion of ERG (ERGdel) and/or IKZF1 alterations, the genetic background remained unknown in the majority of cases. Recently, DUX4-rearranged (DUX4r) ALL was described as a new subset of BCP ALL. Based on its association with ERGdel, we hypothesized that DUX4r ALL may represent a BCP ALL subtype highly prone to swALL. After the lineage switch, the monocytoid cells lose BCP markers (typically CD19) while they retain Immunoglobulin and T-cell receptor gene rearrangements (IG/TR), which results in a discrepancy between BCP-immunophenotype oriented FC-based and IG/TR-oriented quantitative (q) PCR-based minimal residual disease (MRD) detection.
In the current study we aimed to describe the genotypic composition of swALL. Next, we wanted to evaluate in detail the impact of swALL phenomenon on discrepancies between FC-based and PCR-based MRD assessment during the induction and consolidation phases of treatment and its homogeneity across genetic/biological ALL subtypes.
Occurrence of swALL phenomenon was assessed in 738 patients consecutively diagnosed between 09/2007-02/2019 via 8-color FC using a panel of monoclonal antibodies against BCP and myelomonocytic markers as described previously (Slamova et al., 2014). RNA sequencing (RNAseq) was performed in 126 ALL negative for routinely assessed (cyto)genetic markers in order to classify them into additional genetic/biological subtypes. MRD was assessed via FC and qPCR in peripheral blood at day (d) 8 and in bone marrow at d15, d33 and week (w) 12.
We identified swALLs in 59/738 (8%) of ALL cases. The frequency of swALL ranged from 0 to 100% across individual genetic/biological subtypes (Table 1). Compared to non-swALL, swALL was enriched mainly for DUX4r, PAX5 P80R and ZNF384 subtypes (Figure 1).
We observed a poorer MRD correlation in swALL compared to non swALL at d8 PB (0.55 vs 0.81, respectively), d15 BM (0.61 vs 0.9, respectively) and d33 BM (0.35 vs 0.69, respectively). Concordance between MRD results at d15 at the level of 1e-3 was 97% in DUX4r swALL, but only 17% in PAX5-P80R swALL. The main cause of discordance was the underestimation of MRD by FC compared to qPCR. The poor correlation of both methods at d15 in PAX5-P80R indicates rapid loss of B cell phenotype. Correlation of both methods was uneven across ALL subtypes. At d33, 41% and 33% of patients with DUX4r and PAX5-P80R swALL, respectively, were discordantly called to have MRD ≥ 10-3 by PCR but not by BCP FC (which could theoretically lead to discordant risk stratification). On the contrary, the MRD categorization using the same cut-off level at d15 was fully concordant in high-hyperdiploid and ZNF384r swALLs.
We identified DUX4r and PAX5-P80R ALL as the most prevalent subtypes among swALL. We showed that the correlation between FC- and qPCR-based MRD is influenced by the treatment time point and genetic background of ALL. Blasts in PAX5-P80R mutated swALL cases lose B cell antigens early and already at day 15 often majority of blasts is of monocytic phenotype. Loss of B cell phenotype in DUX4r cases is more gradual and at day 15 we observe blasts with decreased expression of CD19 but still clearly positive.
Supported by NV18-03-00343, NV18-07-00430, 16-32568A, UNCE/MED/015