Background: In various malignancies, detection of measurable residual disease (MRD) is associated with adverse outcomes. Bone marrow (BM) is a widely established source for MRD detection by multiparametric flow cytometry (MFC) in patients (pts) with acute myeloid leukemia (AML) due to its presumably more favourable sensitivity compared to peripheral blood (PB). However, frequent BM aspirations are hampered by the invasiveness and the discomfort of the procedure. Moreover, assessment of MRD in the PB may add additional prognostic information.
Recently we established a MFC approach that allows a fast, standardised and reproducible MRD assessment in BM samples (Röhnert et al. Leukemia 2022). The eight colour panel contains the core antigens for MRD detection recommended by ELN. The approach tracks 32 populations with aberrant immunophenotypes utilizing a hierarchical gating strategy with mainly fixed gates. The aberrant populations are grouped according to four distinct aberrancies (deficiency of CD13 or CD33, cross-lineage expression of CD7 or CD56). This approach screens for the reoccurrence of an aberrant leukemia associated immunophenotype already present at diagnosis (LAIP, affecting ≥10% of the myeloid progenitors/monocytes) and a de-novo occurrence of an aberrant group (different from normal, DfN).
Aims: We aimed to adapt this MFC approach to be used for PB samples.
Methods: For PB-samples, the fixed gates of the BM approach had to be slightly adjusted after review of leukemia-free reference measurements and according to non-leukemic populations within PB AML samples. Reference values for each aberrant population were established by using PB from blood donors (n=30) and pts with malignancies who had undergone chemotherapy prior to PB collection: lymphoma pts (n=21), pts with acute lymphoblastic leukemia in molecular remission (n=10), and pts with non-haematological neoplasms (n=22). The upper limit of the one-sided 97.5% confidence interval of each aberrant population was set as reference value (median 0.0026% of CD45+, range 0.0001%-2.2324%).
To compare the new PB approach with the established BM approach, we measured paired samples of BM and PB (187 pts at diagnosis and 489 samples at follow up from 231 pts).
Results: At diagnosis, a LAIP could be detected in 55% of PB samples, compared to 69% of matching BM samples. Two thirds of these paired specimens showed comparable aberrancies. The most frequent disparity was a lack of detection for deficiency of the myeloid markers CD13 and CD33 in PB compared to BM, except in samples with expression of CD7, which showed more often a concurrent deficiency of CD33 in PB than in BM (see figure).
In 489 paired follow up samples at various time points, concordant MRD results were observed in 77% of measurements (44% MRDpos, 33% MRDneg). The most common disparity was MRDneg in PB and MRDpos in BM in 15% of cases, mostly due to detection of populations with deficiency of myeloid markers in the BM but not in PB (as at initial diagnosis). Two thirds of those samples were only low-level BM-MRDpos (<0.1% above reference value), suggesting a lower sensitivity of the PB-MRD-approach. This is supported by the higher BM-MRD load compared to PB-MRD load in paired MRDpos samples (median 0.98% vs. 0.31% of CD45+, p<0.0001). PB-MRDpos paired with BM-MRDneg was seen in only 8% of samples.
Clinical outcome data are not available so far due to a short follow-up time (median 10 months).
Summary/Conclusion: The proposed LAIP based DfN approach for MRD detection in PB shows a promising high concordance compared to BM samples with the same level of standardization and analytical speed. However, the prognostic relevance of the proposed approach needs to be evaluated in prospective clinical outcome data. Updated results will be presented.
Keywords: Peripheral blood, AML, Minimal residual disease (MRD), Flow cytometry