Background

Measurable residual disease (MRD) detection in acute myeloid leukemia (AML) patients in complete morphological remission (CR) is associated with an increased risk of relapse and mortality. While molecular methods such as qPCR offer high sensitivity, most patients lack a molecular marker and are monitored using flow cytometry (FC). Next-generation sequencing (NGS) and digital PCR (ddPCR) represent a potential alternative for MRD assessment in AML, given that they offer high sensitivity and broad applicability.

In this prospective study, we aimed to compare the clinical performance of NGS and ddPCR for MRD assessment in a cohort of intensively treated AML patients.

Methods

Adult patients with newly diagnosed AML treated with intensive chemotherapy (IC) according to CETLAM 2022 protocol were included prospectively in this study. Bone marrow (BM) samples of patients in CR were collected for MRD analysis at different time points: after two courses of IC (all), at the end of IC (favorable patients), at the time of hematopoietic stem cell transplant (HSCT) and 1 month after (intermediate/adverse patients).

A custom NGS panel for MRD detection (NGS-MRD) was designed including 30 genes frequently mutated in AML and a specific assay to detect FLT3-ITD. NGS-MRD was performed on DNA extracted from BM. ddPCR assays were selected personally based on the mutational profile of each patient and performed on the same DNA samples. Dilution studies were conducted to assess the level of detection of both techniques, establishing a positivity threshold of 0.2% VAF for NGS-MRD and 0.1% VAF for ddPCR. MRD status was defined according to the European LeukemiaNet (ELN) guidelines (Heuser M et al, Blood, 2021).

Results

Forty-nine BM samples in CR from 23 AML patients were included in this study. The median age at diagnosis was 61 (range 21-73), and 52.2% were male. Fifteen (65.2%) patients had HSCT indication based on ELN 2022 intermediate or adverse-risk, while 8 (34%) were classified as favorable-risk. Standard of care (SoC) methods for MRD evaluation were flow cytometry in 16 (69.5%) of them, NPM1-qPCR in 6 (21.8%), and CBF-qPCR in 1 (4.3%). With a median follow-up of 17.9 (range 3.7-24.9) months, there were 4 (17.4%) relapses and 2 (8.7%) deaths.

All 49 samples were analyzed with NGS-MRD. A median of 3 mutations (range 1-8) could be tracked for each patient with a median read depth of 6336x (range 576-26402x). Twenty-five (51%) samples had a targetable mutation for ddPCR analysis including IDH1, IDH2, NRAS, and FLT3-TKD. In the 25 samples studied by NGS and ddPCR, we found a strong correlation (r=0.925) between the VAF values obtained by both methods. The median VAF for positive MRD samples was 4.2% (range 0.1-42.1%). All NGS-MRD-positive samples were also considered positive by ddPCR, while 5 NGS-MRD-negative had low-level detectable ddPCR-MRD (range 0.1-0.5% VAF).

Next, we compared our results with standard methods. After two cycles of IC, 33.7% of patients had detectable MRD by SoC techniques, while 52.1% by NGS and 63.6% by ddPCR. All SoC-MRD+ were also positive by NGS and ddPCR, but we could identify MRD persistence in 50% of patients with undetectable MRD by FC or qPCR. Moreover, in our cohort, NGS-MRD status after the second induction significantly predicted RFS (22.2 vs not reached; p=0.025). ddPCR-MRD status showed a positive trend (15.2 vs NR; p=0.191) but did not reach statistical significance.

At the end of chemotherapy, all favorable-risk patients had undetectable MRD by SoC tests, but we could detect 3 (37.5%) MRD+ cases by both NGS and ddPCR. Two of these cases relapsed within 6 months.

Regarding HSCT recipients, pre-transplant MRD was detectable in 35.7% of patients by FC, 50% by NGS, and 75% by ddPCR. However, pre-HSCT MRD by any method did not show an impact on RFS. This could be due to the high MRD clearance rate after HSCT: all FC-MRD+ patients converted to MRD- and 92.9% of MRD+ patients with NGS or ddPCR.

Conclusions

NGS and ddPCR showed a high correlation as MRD assessment methods. While ddPCR offered higher sensitivity, NGS-MRD status was more predictive of relapse-free survival. Also, NGS allowed us to track multiple mutations and even new clones, while ddPCR was restricted to 1-2 targets per patient. Applying these techniques in a prospective AML cohort, we could detect MRD persistence in a substantial number of MRD-negative patients by standard methods.

Disclosures

Ramil López:Abbvie: Other: Travel grant to attend conference; Johnson & Johnson's: Other: Travel grant to attend conference; Astellas: Other: Travel grant to attend conference. Díaz-Beyá:BMS: Consultancy; Abbvie: Consultancy; Takeda: Consultancy; Novartis: Consultancy; Jazz Pharmaceuticals: Consultancy. Tormo:SOBI: Other: Data Safety Monitoring Board; Janssen, AbbVie, Jazz: Other: Travel grant for attending meetings; AbbVie, Gilead, Pfizer, Astellas, BMS: Honoraria. Salamero:Jazz, Abbvie: Honoraria; Astellas, Jazz, BMS: Consultancy.

This content is only available as a PDF.
2024
Sign in via your Institution