Optimizing Donor Chimerism Threshold for Next-Generation Sequencing Monitoring of Measurable Residual Disease Post-Allogeneic Stem Cell Transplantation for Myeloid Neoplasms

Abstract

Next-generation sequencing (NGS) is used to monitor genetically measurable residual disease (gMRD) following allogeneic stem cell transplantation (aSCT). It is unknown whether an upper limit of chimerism exists such that gMRD NGS testing can be safely forgone. We reviewed 61 patients with acute myeloid leukemia and 24 patients with myelodysplastic syndrome who had at least 1 NGS panel before and after aSCT between 2016 and 2020. Donor chimerism was quantified. Logistic regression characterized which factors predicted gMRD. Receiver operating characteristic (ROC) curves were used to determine the optimal chimerism threshold for which gMRD would not be detected. Data from an additional 22 patients with follow-up NGS testing in 2022 were also analyzed to validate our proposed threshold. A: s expected, donor chimerism was a significant predictor of gMRD (odds ratio, .38; 95% confidence interval, .10 to .62; P = .02). Age, sex, conditioning regimen, presence of a related donor, and diagnosis were not associated with gMRD. A chimerism threshold of 92.5% optimized sensitivity (97.7%) and specificity (95.4%) such that values >92.5% strongly predicted the absence of gMRD (area under the ROC curve [AUC], .986). The validation cohort demonstrated similarly strong predictive capability (AUC, .974) with appropriate sensitivity (100%) and specificity (90.9%). NGS monitoring of gMRD is redundant at chimerism values greater than a more conservative threshold of 92.5% after aSCT.

Keywords: MRD; chimerism; molecular diagnostics; sequencing; transplant.