Comparison of paired bone marrow and peripheral blood measurable residual disease in pediatric acute lymphoblastic leukemia by next-generation sequencing: A report from the dana farber ALL consortium protocol 16-001.

Abstract Introduction: Current protocols treating acute lymphoblastic leukemia (ALL) in children, adolescents, and young adults utilize bone marrow (BM) measurable (minimal) residual disease (MRD) to assess early treatment response and risk stratify post-induction therapy. Measuring MRD in peripheral blood (PB) may provide a less invasive way to quantitate leukemia burden facilitating more frequent and nuanced assessment of disease status and may provide additional insight on tumor biology. Methods: Next-generation sequencing of immunoglobulin and T-cell receptor loci by the clonoSEQ Assay (Adaptive Biotechnologies Corporation, Seattle, USA) was used to evaluate paired BM and PB samples from participants enrolled on a front line ALL clinical trial DFCI 16-001 (NCT03020030). Participants were selected by having paired samples available. Participants on 16-001 were stratified into initial risk groups by age, presenting white blood cell count, immunophenotype, CNS status, and adverse leukemia biology. MRD was evaluated at End-induction IA, 4 weeks after starting therapy (TP1, 138 pairs after QC) and if the TP1 result was greater than 100 per million cells (10^-4) MRD was evaluated again at end-induction IB, 10 weeks after the start of treatment (TP2, 42 pairs after QC). MRD was assessed in patients classified as very high risk at a third timepoint approximately 19 weeks after the start of therapy (TP3, 26 pairs after QC). BM MRD value by clonoSEQ if available determined final risk group; PB MRD was not used for clinical decision-making. Scatterplots were created and Spearman correlation coefficients were calculated to assess correlation between BM and PB MRD. Results: This analysis includes 213 paired BM and PB samples from 149 participants with median age at diagnosis 6.1 years (1-19.7), 54% male, 74% White, 77% Non-Hispanic, and 87% with B-cell immunophenotype. In this cohort, 50% were categorized as initial low risk, 29% initial high risk, and 21% initial very high risk. Of the 213 paired samples, 7 PB samples failed quality control (QC) leaving 206 for the correlation analysis. BM and PB MRD were highly correlated across all samples (R=0.78, p<2.2x10-16) and at each time point including TP1 (R=0.79, p<2.2x10-16), TP2 (R=0.71, p=1.7x10-7), and TP3 (R=.83, p=1.2x10-7). BM and PB MRD were highly correlated for both B-ALL (N=184, R=0.76, p<2.2x10-16) and T-ALL samples (N=29, R=0.88, p=6.1x10-10). Concordance was similar to the whole cohort for B-ALL samples at TP1(R=0.76, p<2.2x10-16), TP2 (R=0.7, p=3.2x10-6), and TP3 (R=0.89, p=8.9X10-9). For 44 paired samples with protocol defined high BM MRD (more than 100 per million cells for TP1, more than 1000 per million cells for TP2 or TP3), correlation with PB was also strong (R=0.50, p=5.0x10-4) however there is a discordance rate of 50% with 22 having a PB value below the protocol cut-off value. Using the whole cohort, samples with BM count in the range that may have effected management on 16-001 between 100 and 3000 per million cells were not correlated with PB (R=0.056, p=0.76). In subgroup analysis by initial risk group (low, high, and very high) and favorable tumor biology (presence of double trisomies 4 and 10 or ETV6::RUNX1) correlation remained strong. Conclusions: In this analysis, BM and PB MRD during treatment for pediatric ALL were highly correlated, particularly at high and low levels of detectable BM disease, across protocol defined risk groups, immunophenotype, and with favorable leukemia biology. At moderate levels of BM detectable disease, correlation with PB detectable disease was less strong limiting immediate direct replacement of BM sampling with PB. PB represents a desirable sample for assessing MRD allowing more frequent and less invasive evaluation of ALL disease status particularly for children, allowing for less painful modality to monitor disease status and less exposure to sedation/anesthesia. High correlation between BM and PB suggests that PB may represent a valuable adjunct for patient monitoring. With regard to specific levels for risk stratification, more work is needed. Correlation of PB MRD with ALL outcomes including event free and overall survival in the future will be essential in determining utility of PB MRD assessments.