Clinical feasibility of peripheral blood–based MRD monitoring using the miselect R AML MRD assay in Acute Myeloid Leukemia

Abstract Background Monitoring measurable residual disease (MRD) is critical for guiding post-remission therapy in acute myeloid leukemia (AML), but current MRD testing relies on invasive bone marrow (BM) sampling and conventional multiparameter flow cytometry with limited sensitivity. Peripheral blood (PB) offers a less invasive alternative for longitudinal MRD assessment but has been limited by lower blast burden. The MiSelect R II System is an automated rare-cell enrichment and single-cell imaging platform that isolates CD117⁺ cells and quantifies leukemia-associated immunophenotype (LAIP)–positive myeloblasts with analytical sensitivity near 10⁻⁷. We evaluated the analytical performance and clinical feasibility of PB MRD monitoring using this system, its concordance with BM MRD, and its association with progression-free survival (PFS). Methods: Analytical validation used contrived peripheral blood (PB) samples spiked with diagnostic AML specimens at defined frequencies to determine limit of detection (LoD), precision, and reproducibility. Clinical feasibility was assessed in 32 AML patients (median age 68.7 y; 59% male); 31% received intensive 7+3–based induction and 69% received venetoclax- or hypomethylating agent–based low-intensity regimens. Longitudinally, 110 PB and 70 bone marrow (BM) samples were collected at diagnosis, post-induction, and during follow-up. MRD was quantified by LAIP-based analysis of enriched CD117⁺CD34⁺CD45^dim/− myeloblasts. PB and BM MRD levels were compared by Pearson correlation, and associations with progression-free survival (PFS) were evaluated using Kaplan–Meier analysis and log-rank testing. Results: The MiSelect R AML MRD Assay achieved an LoD of 4.5 × 10⁻⁷ with 100% detection (25/25) at the lowest tested frequency. Intra-assay, inter-assay, and inter-operator CVs were 6.2–10.9%, 0–16.1%, and 5.3–11.1%, respectively, confirming robust analytical performance. Across 62 paired PB and BM samples, MRD levels correlated strongly (r = 0.511, p < 0.001), indicating that PB reflects BM disease burden. Among 32 patients (median follow-up 198 d), 11 (34%) achieved CR, 3 (9%) CRi, 6 (19%) were refractory, 1 (3%) died prior to assessment, and 11 (34%) lacked BM response data. During follow-up, 2 (6%) relapsed and 11 (34%) died. PB MRD negativity at last follow-up among CR/CRi patients (n = 14) was associated with longer PFS (p = 0.011); BM MRD showed a similar but non-significant trend (p = 0.142). In the entire cohort, PB MRD negativity at subsequent post-treatment assessments trended toward improved PFS (p = 0.051), including at Day 28 ± 7 (p = 0.12) and Day 90 ± 30 (p = 0.065). MRD kinetics analysis revealed heterogeneous outcomes. Sustained PB MRD negativity was associated with the most favorable OS and RFS (0% relapse and death). Patients with sustained MRD positivity had the highest mortality (45.5%). No relapses or deaths were observed in the re-emergence group during the limited follow-up. Conclusions: The MiSelect R AML MRD Assay is a sensitive, standardized PB-based MRD test that correlates with BM MRD and shows potential prognostic value. PB MRD monitoring offers a less invasive approach for longitudinal disease assessment and may enable earlier relapse detection. Larger prospective studies with longer follow-up are warranted to confirm its prognostic significance and clinical utility in AML management.