Measurable residual disease (MRD) assessment is a standard practice in B-lymphoblastic leukemia (B-ALL) [1] and serves as a prognostic marker [2]. Multiparametric flow cytometry (MFC)-MRD analysis enables a rapid turnaround time with wide applicability [3, 4]. The sensitivity of MFC-MRD analysis was improved to 10⁻⁵ using a two-tube, eight-color next-generation flow MRD, as proposed by the EuroFlow Consortium [5]. Next-generation sequencing (NGS) showed higher sensitivity than MFC-MRD in assaying immunoglobulin gene clonality, was highly concordant with quantitative PCR in detecting immunoglobulin gene rearrangements, and was less laborious while providing comprehensive results [6]. Various MFC-MRD methods, involving different antibody–fluorophore combinations, sample-processing steps, and data analysis, have been developed; however, relatively few NGS-MRD assays, such as the LymphoTrack (Invivoscribe, San Diego, CA, USA) [7] and clonoSEQ (Adaptive Biotechnologies, Washington, USA) assays, are available for detecting immunoglobulin gene rearrangements [8]. MFC-MRD and NGS-MRD assays are complementary, and patients who are MRD-negative by both methods have better treatment outcomes [6, 9]. However, few studies have compared the MFC-MRD and NGS-MRD assays [6, 10]. Therefore, we compared MFC-MRD and NGS-MRD results in pediatric patients with B-ALL and evaluated the impact of hemodilution, especially in peripheral blood samples.

We examined bone marrow (BM) follow-up samples from pediatric patients with B-ALL via simultaneous MFC-MRD and LymphoTrack immunoglobulin heavy chain gene (IGH) analysis (Invivoscribe), using samples obtained in Seoul National University Hospital and Seoul National University Bundang Hospital from February 2022 to December 2022. The eight-color, two-tube MFC-MRD panel was modified based on EuroFlow B-ALL MRD guidelines [4], with common markers used in both tubes (CD34-PerCPCy5.5, CD19-PECy-7, CD10-APC, CD20-V450, CD38-APC-H7, and CD45-V500c). CD81-FITC and C73-PE were included in tube one only, and CD58-FITC, CD66c, and CD123-PE were included in tube two only. The antibodies were obtained from Becton Dickson (BD) and Company (San Diego, CA, USA), and validation results are presented in the Supplemental Materials. For patients treated with anti-CD19 therapy, CD22-FITC, CD66b-PE, and CD24-PECy7 were added to the second tube [11]. We conducted bulk lysis following the EuroFlow guidelines, acquiring 5 million events for each tube or as many events as possible (when minimal cells were available), using a BD FACSCanto (BD) instrument and Infinicyt Software 2.0.5 (Cytognos, Salamanca, Spain). The limit of detection (LOD) was defined as 60 events/100 total nucleated cells (TNCs), with a lower limit of quantitation (LLOQ) as 100 events/100 TNCs. Hemodilution was assessed based on quantitating nucleated red blood cells (RBCs) and identifying hematogones [12]. To identify nucleated RBCs, we identified cells that lacked all included markers and exhibited low forward and side scatter. Hematogones were identified as CD19⁺, CD38⁺, CD45dim, CD34^+/−, and CD10⁺ cells, and their percentages were determined based on the number of TNCs in the sample. This study was approved by our local ethics committee (approval number B-2301-805-104).

For NGS-MRD, we extracted DNA from BM aspirates with a QIAsymphony DSP DNA Midi Kit using the QIAsymphony system (Qiagen, Hilden, Germany). IGH rearrangement and clones were assessed using the LymphoTrack MRD-NGS assay (Invivoscribe). We tested 24 samples in a single run with 800 ng of DNA, achieving a minimum sensitivity of 10⁻⁴. The data were processed using LymphoTrack Analysis Software and the MRD Data Analysis Tool (Invivoscribe). NGS-MRD provided clonal/TNC (%) for MRD among TNCs and clonal/total IGH read count (%) for MRD among B cells. The TNC-based percentage was calculated using the LymphoQuant B-cell internal control (Invivoscribe).

We compared the patients’ clinical characteristics using the Mann–Whitney U-test and analyzed the correlation between the MFC-MRD and NGS-MRD results using Spearman’s correlation coefficient. To compare the median positive MRD values, we performed the Wilcoxon matched-pairs signed-rank test. Statistical analyses were performed using GraphPad Prism 9.5.1 (GraphPad Software Inc., San Diego, CA, USA) and R Software version 3.6.1.

We assessed 139 samples from 54 pediatric patients with B-ALL (median age: 9 yrs; range: 1–18 yrs). Table 1 summarizes the patient and sample characteristics. Among these patients, 40.7% (N=22) had B-ALL with recurrent genetic abnormalities, and all patients had clonal immunoglobulin gene rearrangements identified at diagnosis, with a median of two types of clones (range: 1–4). Among the 139 samples, MRD assessment was conducted after anti-CD19 therapy (such as blinatumomab or chimeric antigen receptor T cell therapy). Alternative B-cell markers were used for 18.0% (N=25) samples from seven patients.

Table 1 . Patient and sample characteristics

Characteristic	N (%)
Patients	54
Age, yr, median (range)	9 (1–18)
Sex, Male:Female	1:1.1
Genetic abnormalities
BCR::ABL1	8 (14.8)
TCF3::PBX1	7 (13.0)
ETV6::RUNX1	6 (11.1)
KMT2A rearrangement	1 (1.9)
Other karyotypic abnormalities	25 (46.3)
Normal karyotype	7 (13.0)
Samples	139
Post-induction	35 (25.2)
Post-consolidation	34 (24.5)
Post-maintenance	14 (10.1)
Post-chimeric antigen receptor T cell therapy	16 (11.6)
Post-immunotherapy (blinatumomab or inotuzumab)	6 (4.3)
Post-hematopoietic stem cell transplantation	34 (24.5)

Thirty-three (23.7%) samples from 15 patients were positive for MRD using MFC-MRD (median: 0.0180% [0.0007%–19.6%]), and 55 (39.6%) samples from 25 patients were positive for MRD using NGS-MRD (median: 0.0229% [0.0001%–112.8%]). Regarding the MFC-MRD analysis, a median of 6,900,221 (916,235–9,302,173) TNC events were analyzed, with a median LOD of 0.0009% (0.0006%–0.0066%) and a median LLOQ of 0.0014% (0.0002%–0.0109%). Regarding the NGS-MRD analysis, a median total IGH read count of 1,087,179 (1,510–2,625,491) was analyzed.

The concordance rate between both methods was 79.9% (N=111) with the 139 samples analyzed; 30 and 81 samples were MRD-positive and MRD-negative, respectively, based on both methods. Further, 25 (18.0%) and three (2.2%) samples were positive based on NGS-MRD only (MFC⁻NGS⁺MRD) and MFC-MRD only (MFC⁺NGS⁻MRD), respectively (Fig. 1). A moderate correlation was found between both methods for MRD values from TNCs (r=0.736 [0.647–0.806], P<0.001) and for MRD percentages among B cells (r=0.748 [0.662–0.815], P<0.001). Additionally, a strong correlation was identified between the MRD values based on TNCs and B cells obtained via MFC analysis (r=0.915 [0.881–0.939], P<0.001) or NGS analysis (r=0.991 [0.987–0.994], P<0.001). When considering only positive MRD values, a significant difference was noted in the median values, with MFC-MRD analysis revealing a 0.120% lower median MRD value than NGS-MRD analysis (P<0.001).

Figure 1. Comparison of MRD data obtained using MFC-MRD and NGS-MRD. The grid in the lower right of the graph shows the numbers of NGS⁺/MFC⁻, NGS⁺/MFC⁺, NGS⁻/MFC⁺, and NGS⁻/MFC⁻MRD samples.
Abbreviations: MRD, measurable residual disease; MFC, multiparametric flow cytometry; NGS, next-generation sequencing; TNC, total nucleated cells.

Among the 28 samples with discordant MRD values, 25 MFC⁻NGS⁺MRD samples (N=20 patients) showed a median MRD value of 0.0035% (0.0001%–0.5250%) by NGS-MRD, with 24.0% of the MFC⁻NGS⁺MRD samples showing an NGS-MRD value below the LLOQ of the MFC-MRD assay. Ten MFC⁻NGS⁺MRD samples showed negative MFC-MRD results in the interval before the NGS-MRD results were reported as being negative. Regarding the three MFC⁺NGS⁻MRD samples, the median MRD value was 0.0026% (0.0025%–0.0089%) by MFC-MRD. Notably, one sample with MFC⁺NGS⁻MRD showed a positive BCR::ABL1 quantitative PCR result, consistent with the MFC-MRD findings (Supplemental Data Table S1).

We assessed whether MFC-MRD analysis involving gating for alternative B-cell markers would result in more discrepant cases and found discrepant MRD values for 28.0% (N=7) and 18.4% (N=21) of samples from patients treated with anti-CD19 therapy and conventional therapies, respectively. However, these discrepancies did not significantly differ between the two groups (P=0.281).

A median of 0.0051% (0–4.3300%) of hematogones was present in the MFC-MRD-positive samples. The median percentage of hematogones was 0.0000% (N=28) in discrepant cases and 0.0109% (N=111) in concordant cases (P=0.021). The optimal cutoff value for predicting concordant MRD was 0.0555% for the hematogones, with an area under the curve of 0.637 (P<0.001).

MFC-MRD has a history of use, whereas NGS-MRD is a more recent option for detecting B-ALL. We compared the performance of the MFC-MRD assay with that of a commercial NGS-based LymphoTrack IGH assay for pediatric patients with B-ALL. Concordance between the MFC-MRD and NGS-MRD assays was high and similar to previous findings showing more frequency and sensitive MRD detection with NGS-MRD [8, 9, 13].

Many discordant cases (85.7%, N=24) had MRD values below 0.01%, and the concordance rate was higher (92.8%) when the cutoff was adjusted to 0.01%, a well-validated value [2]. Regarding the MFC⁻NGS⁺MRD cases, the MRD values detected using NGS-MRD were below the LOD of the MFC-MRD assay in 14.3% (N=4) of the cases, rendering them undetectable by MFC-MRD. Most MFC⁻NGS⁺MRD cases were detected after the MFC-MRD results had turned negative, often within one measurement interval, indicative of a more sensitive MRD detection capability. Although several MFC⁻NGS⁺MRD cases were within the detection range of the MFC-MRD assay, immunophenotypic changes may occur, and the emergence of leukemic cells with a more mature immunophenotype following therapy may hinder detection by MFC-MRD [8, 14].

The MRD values were higher with the NGS-MRD assay, possibly because of normalization with the number of TNCs as an internal control [15], whereas MFC-MRD analysis might have inadvertently included cellular debris as nucleated cell events, as some methods incorporate nucleated acid dye [16]. MRD values based on TNC or B-cell normalization correlated significantly, and the clinical significance of the normalized values obtained with the LymphoTrack NGS-MRD [15] and clonoSEQ [8] has been reported, although further validation might be required.

MRD results are significantly influenced by the sample quality and pre-analytic conditions, with more notable effects observed on MFC-MRD results than on NGS-MRD results [17]. We assessed hemodilution via MFC-MRD analysis [12], where the percentage of hematogones, rather than the percentage of nucleated RBCs, was used to assess hemodilution. In the future, combinations of different BM cell types may be used to assess hemodilution [18].

However, as with many studies regarding MFC-MRD and NGS-MRD analysis of B-ALL [8, 13], some patients lack identifiable clones with NGS at diagnosis [19]; thus, NGS-MRD may not be feasible, and IGH-clonality results may not be available for referred patients, making MFC-MRD more reliable.

A major limitation of this study is the relatively short follow-up period (median: 14.3 months; range: 5.1–88.5 months). Survival analysis could not be performed, rendering it difficult to determine the false-positive and false-negative rates of both assays. Future studies should be conducted to assess the significance of low-level MRD positivity. Although some studies have compared NGS-MRD analysis with clonoSEQ and MFC-MRD analysis [7, 13], relatively few studies have compared LymphoTrack NGS-MRD and MFC-MRD assays using samples from patients with B-ALL [15].

In conclusion, the MFC-MRD and NGS-MRD results correlated well when the MRD values were based on either B cells or TNCs. NGS-MRD analysis showed a higher rate of MRD detection than MFC-MRD using samples from pediatric patients with B-ALL. Further, evaluating the percentages of hematogones can help in assessing the effects of hemodilution during MFC-MRD analysis.

Supplementary materials can be found via https://doi.org/10.3343/alm.2023.0412

None.

Hwang SM conceived the study, designed the experiment, analyzed the data, and wrote the manuscript. Oh I, Kwon SR, Lee JS, and Seong MW analyzed the data, and Lee JS analyzed the data and wrote the manuscript. All authors contributed to the manuscript and approved the submitted version.

None declared.

This study was supported by the SNUBH research fund (grant number 02-2021-0051).