Back

Personalized Circulating Tumor DNA (ctDNA) Profiling Enables Superior and Universal Measurable Residual Disease (MRD) Detection in Acute Myeloid Leukemia (AML)

Gunaratne, R.; Zhou, C.; Rajaram, S.; Tai, J. W.; Tanaka, K.; Tiwari, C.; Yang, E.; Kim, S.; Gao, G.; Yin, R.; Carleton, M.; Alkaitis, M. S.; Schwede, M.; Sworder, B. J.; Mannis, G. N.; Khodadoust, M. S.; Majeti, R.; Kurtz, D. M.; Zhang, T. Y.

2026-02-04 hematology
10.64898/2026.01.28.26344873 medRxiv
Show abstract

Relapsed and/or refractory disease remains the leading cause of death in AML, highlighting the need for broadly applicable, high-sensitivity approaches to MRD detection. We developed AML-CAPP-Seq (Cancer Personalized Profiling by Deep Sequencing), a personalized hybrid-capture assay that tracks both canonical AML drivers and patient-specific variants identified by whole-exome sequencing. In 56 patients with longitudinal plasma and matched peripheral blood and bone marrow samples, AML-CAPP-Seq enabled universal MRD assessment and resolution of clonal dynamics using a median of 30.5 variants per patient. Plasma ctDNA outperformed cellular compartments for MRD detection and more strongly predicted relapse-free (HR 17.8, p<0.0001) and overall survival (HR 17.0, p<0.0001) than standard-of-care MRD methods. Among 29 allogeneic transplant recipients, peri-transplant ctDNA-MRD dynamics markedly improved relapse risk stratification (HR 36.0, p=0.0009). Together, these results establish personalized ctDNA profiling as a minimally invasive, highly sensitive, and generalizable platform for enhanced clinical MRD detection and clonal surveillance in AML. Significance StatementWe present a personalized blood test for acute myeloid leukemia that tracks patient-specific circulating tumor DNA, enabling sensitive, universal, noninvasive detection of residual disease. It outperforms standard-of-care marrow and cell-based methods for predicting relapse and survival, including after transplant, reveals clonal dynamics, and supports individualized disease monitoring and risk-adapted treatment.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.