AUTHORS
Alexa Rane Batingana, Helen Tian2, Denise Masli3, Jeevitha D'Souza, Postdoctoral fellow4, Rhonda E. Ries, MA5, Matthew A. Kutny, MD6, Jessica A. Pollard I, MD7, Todd Cooper I, DO8, Xiaotu Ma, PhD9, Soheil Meshinchi, MD, PhD10, Alexandra McLean Stevens11, Katherine Tarlock I, MD12, Benjamin J. Huang, MD13
AFFILIATIONS
1 University of California, San Francisco, San Francisco, California, United States
2 University of California, San Francisco, San Francisco, California, United States
3 University of California, San Francisco, San Francisco, California, United States
4 University of California, San Francisco, San Francisco, California, United States
5 Fred Hutchinson Cancer Center, Seattle, Washington, United States
6 University of Alabama at Birmingham, Birmingham, Alabama, United States
7 Dana Farber Cancer Institute, Boston, Massachusetts, United States
8 Seattle Children’s Hospital, Seattle, Washington, United States
9 St Jude Children's Research Hospital, Memphis, Tennessee, United States
10 Fred Hutchinson Cancer Center, Seattle, Washington, United States
11 Texas Children's Hospital, Houston, Texas, United States
12 Seattle Children’s Hospital, Seattle, Washington, United States
13 University of California, San Francisco, San Francisco, California, United States
ABSTRACT
Acute myeloid leukemia (AML) remains a therapeutic challenge with high mortality rates despite intensive and myeloablative therapies. A major challenge to improving outcomes is distinguishing patients who are cured with frontline therapies from others who will eventually relapse and may benefit from additional therapies. Measurable residual disease (MRD) has become an important predictor of disease recurrence and mortality. However, current MRD detection approaches fail to accurately risk-stratify a substantial number of patients, which underscores the need to identify more accurate and sensitive MRD approaches for AML.
Over the last two decades, multiparametric flow cytometry (MFC) has been the “gold standard” MRD assay for AML. While MFC improved upon prior definitions of remission, this technology is associated with several limitations: (1) MFC MRD in AML cannot consistently exceed a limit of detection (LOD) of 1:1,000; (2) LOD varies based on leukemia immunophenotype; and, (3) myeloid leukemias often display overlapping immunophenotypes with normal myeloid populations. For pediatric AML, these limitations have restricted the prognostic value of MFC MRD to patients with standard-risk AML and prior to HSCT. Additionally, patients that have achieved undetectable MFC MRD prior to HSCT have a disappointingly high cumulative incidence of relapse (CIR) of 32% ± 8% at 2 years. Importantly, extending the MRD LOD for B-ALL has resulted in significant improvements in outcomes. Specifically, by using an NGS assay that targets TCR/Ig rearrangements, achieving an MRD level below one leukemia cell per million (LOD 10E-6) prior to HSCT was associated with 0% CIR at 2 years. Therefore, our hypothesis is that extending the LOD for AML will have a similar impact.
To address these limitations, we developed digital droplet PCR (ddPCR) assays to target fusion positive AML, with the goal of achieving a 10E-6 LOD. We focused our initial assay development on KMT2A-rearranged (KMT2Ar) and NUP98-rearranged (NUP98r) AML based on their high prevalence, the frequent indication for HSCT in CR1 for these leukemias, and the discovery of menin inhibitors for KMT2Ar and NUP98r AML.
First, we performed bulk RNA-seq on diagnostic samples obtained from 1,503 patients enrolled on the past three consecutive COG frontline trials for de novo AML: AAML03P1 (NCT00070174), AAML0531 (NCT00372593), and AAML1031 (NCT01371981). Each protocol was approved by the NCI central IRB and the local IRB for each participating institution; and patients and/or families provided informed consent or assent as appropriate. We performed gene expression and fusion calling as previously described. Based on fusion calling, we identified 367 cases of KMT2Ar AML and 146 cases of NUP98r AML. Based on gene expression calling, we found that these fusions were expressed at lower levels than other fusion subtypes.
We next validated our ddPCR assays using cDNA from a recent RT-qPCR study and confirmed that the two platforms generated concordant results (R2 = 0.9052). Notably, these RNA/cDNA were extracted and converted using standard protocols, which restricts LOD. To optimize our ddPCR assays, we utilized cell lines with KMT2Ar whenever possible (e.g., MOLM-13, MV-4-11, OCI-AML2, NOMO-1, THP-1) and gene blocks otherwise. We performed cell line or gene block serial dilution experiments. Compared to RT-qPCR (10E-4 LOD) and standard ddPCR protocols (10E-5 LOD), we refined and optimized RNA/cDNA processing and ddPCR protocols to consistently achieve 10E-6 LOD. Based on the our RNA-seq fusion calling results, our current ddPCR assays cover 75% and 97% of recurrent fusion breakpoints for KMT2Ar and NUP98r AML, respectively.
We are evaluating the performance of our ddPCR assays on bone marrow aspirate samples obtained at diagnosis and various treatment time points on AAML1031. Additionally, we are incorporating our ddPCR assays into the next COG AML frontline clinical trial. Our overall goals are to improve disease detection and outcome prediction for children, adolescents, and young adults diagnosed with AML.
DISCLOSURES
Alexa Rane Batingana: No relevant conflicts of interest to declare.