Skip to main content
Premium Trial:

Request an Annual Quote

NGS-Based Post-Transplant Acute Myeloid Leukemia Assay Predicts Relapse Risk


NEW YORK (GenomeWeb) – An international group, led by researchers from the University of Toronto, has developed a next-generation sequencing-based assay to predict the risk of relapse in patients with acute myeloma leukemia (AML) who have received chemotherapy and a bone marrow transplant.

The team said that the tool could offer clinicians a prognostic indicator for long-term patient survival following transplant surgeries, insights that could be used to refine drug treatment.

As one of the most common types of blood cancer, AML normally begins in a patient's bone marrow, where cells rapidly develop into immature white blood cells. The cells can rapidly spread to the blood and damage parts of the body, including the brain, liver, spleen, testicles, and lymph nodes.

While chemotherapy can kill most leukemia cells, individuals who have undergone both chemotherapy and hematopoietic cell transplantation (HCT) still have about a 33 percent chance of relapse three to six months after receiving the transplant. Due to the cancer's heterogenous nature, researchers often struggle to track changes in the mutational profile of a patient's cancer.

In a study published last month in Blood, University of Toronto Molecular Genetics professor Zhaolei Zhang and his colleagues aimed to assess the feasibility of an NGS assay to monitor mutations in AML patients after receiving HCT. Zhang, one of the study's leaders, and his team collected 529 bone marrow samples from 104 AML patients at different time points, including the day of diagnosis, during chemotherapy-induced remission, and 21 days after the bone marrow transplant.

In addition to performing sequencing, "we [sought to] develop a statistical tool to predict which mutations would occur in different patients, based on mutations cleared after chemotherapy and transplants," Zhang explained. "We [also] wanted to see how many deleterious clones exist in the patient's body, and how likely they are to come back."

A smaller subset of patients also provided samples at three, six, and 12 months after receiving a bone marrow transplant. In addition, the researchers collected samples from bone marrow donors to verify that they did not carry leukemia cells.

According to Zhang, he and his colleagues performed targeted sequencing at a depth of over 1700x on a panel of 84 genes compiled from previous large-scale mutation profiles that are often linked to AML. The researchers detected 256 mutations in 90 of 104 patients during initial diagnosis, 15 mutations in 23 patients who relapsed, and one additional mutation in a patient before HCT.

Initially examining the changes in mutations from diagnosis to pre-HCT (and after chemotherapy), the team saw that several genes with mutations — including DNMT3A, IDH2, FTL3-ITD — were persistent in AML patients. They noted that in general, mutations in genes involved in DNA methylation occurred at a higher rate than mutations in other genes. Overall, 142 of the 256 mutations persisted pre-HCT.

Observing the changes in mutations detected post-HCT, however, the team saw that the procedure cleared an additional 110 mutations.

Post-HCT, targeted sequencing detected a total of 41 mutations, of which 19 had a variant allele frequency of 0.2 percent or higher. Overall, the researchers found 23 high VAF mutations in 16 patients — 14 persistent through chemotherapy and HCT; five mutation-cleared pre-HCT but reappearing post-HCT; two mutations from a relapse clone; one mutation from a nonleukemic clone; and one mutation from donor marrow.

Afterward, the team traced the samples' allelic burdens — the fraction of mutations compared to wild type — and compared the measurements for patients with available sample at relapse or at the three-month follow-up after HCT.

They saw that the post-HCT allelic burden in relapsed patients was higher compared to that in non-relapsed patients. In addition, all post-HCT mutations in relapsed patients increased at relapse. However, post-HCT mutations initially detected from non-relapsed patients were cleared in three to six months.

Assessing the genes' VAF, the team found that an overall higher VAF post-HCT was linked to an increased risk of relapse and worse survival for patients. By performing multivariate analysis, the team confirmed that VAF post-HCT was an adverse prognostic factor for overall survival and relapse incidence.

Zhang explained that "by comparing allele frequencies in leukemia bone marrow and the T-cells … we can distinguish polymorphisms from mutations, and we can figure out, with each follow-up, the allelic burdens among the mutations in the sample."

The study authors concluded that posttransplant NGS monitoring at three weeks post-HCT "is [critical] to provide any therapeutic intervention early when impending relapse is highly suspected." Potential intervention includes targeted therapy, which target on specific genes or mutations.

Study challenges

Zhang acknowledged that his team struggled to maintain patient participation over the course of the study due to health issues and patient death.

In addition, the researchers dealt with challenges identifying useful sequencing information with the high number of samples collected from patients at initial diagnosis. They therefore developed new software to identify leukemia-related information from the data they sequenced.

"We needed to use complicated software to make sure to see the real mutation, [and] then derive an accurate [model] to examine the data," Zhang explained.

"With a molecular timepoint, you can de-convolute the data into subclones in order to figure out what fraction of patients have a point mutation or germline mutation," he said.

In an email, first author TaeHyung Kim said that existing software was not sufficient for the team’s purposes, requiring the group to tweak certain tools or write its own methods.

“For example, we impelled our own variant calling methods, which incorporate information from several public databases of natural variants and cancer-associated variants as well as samples of the patient taken at other timepoints,” Kim explained.  

Kim noted that the statistical model his team developed identified a low residual mutation frequency of 0.2 percent in the patient population. However, he stressed that the percentage his team found is not the only threshold and instead should be interpreted as "low residual mutational burden" that could to use as a surrogate marker for a personal chance of relapse.

"I think a lot of issues [with monitoring AML] stem from deciding which genes you want to include on the panel, especially since AML is a heterogeneous disease," Mary-Elizabeth Percival, an assistant member of the Clinical Research Division at the Fred Hutchinson Cancer Research Center, explained. "As done in this study, you need to look at what [the] baseline characteristics of the person's leukemia are."

However, Percival, who is unaffiliated with the Blood study, also noted that examining baseline characteristics may not be enough, since the patient's individual leukemia may change, either through "losing or acquiring new mutations."

While the researchers reported that almost everyone in the study had "complete remission before moving to HCT," Percival was unsure whether Zhang and his team were fully aware of the patients' measurable residual disease (MRD) status after chemotherapy. Importantly, she noted that the outcome of patients with MRD, usually evaluated by a variety of challenging multiparameter flow cytometry and molecular protocols, are usually "inferior after transplants."

Jeffery Klco, assistant member of the pathology department at St. Jude Children's Research Hospital, noted that the AML detection tool will encounter challenges when examining different species of AML. For example, he said that certain AMLs have several fusions that drive leukemia.

"These types of approaches do not track the fusion events yet, and if you're looking at single nucleotide variations, you won't find fusions." Klco explained. "The tools can work, but we need to revise the genomic targets we're after with these approaches."

Zhang believes that the study is the first of its kind to examine the mutational burden in patients' genes after they undergo HCT. By following up with patients and recording their mutational profile several months after surgery, the researchers sought to identify potential timepoints of relapse.

Zhang argued that future sequencing-based mutational analysis of AML patients "will cost more and require more computational expertise due to its complexity," so international collaboration will be crucial to help replicate the study's results.

The team believes that future studies comparing the results from NGS and multicolor flow cytometry (MFC) — a technique that uses multiple fluorescent marker to identify and characterizes cell subpopulations of interest — in a larger sample size may also confirm the use of NGS for monitoring AML patients. According to Kim, MFC can be used to detect leukemia-associated immunophenotypes that are rare in normal bone marrow. Zhang envisions his team's software being used routinely to handle AML patients. However, he said that his team is still in the early stages of deciding the best path to commercialize the method for clinical use.

Kim noted that the team is planning to replicate the study in several hospitals and potentially receive approval from regulatory agencies such as the US Food and Drug Administration and Korean Ministry of Food and Drug Safety. He noted that Korean hospitals involved in the study are currently using the tool as part of research.

However, Percival argued that the clinical space is currently unprepared to handle irregular or unexpected results from platforms that monitor AML progression or relapse after bone marrow transplants. She explained that when clinicians detect a relapse in patients, they have pancytopenia, and treatment "is clearly indicated." If they detect the mutations on a NGS panel every three months or so, however, doctors would not know the next course of action from a clinical perspective.

In addition, Percival questions if the "early intervention for the preclinical relapse [would actually] help patients … or is it just stressful to know about the information and not know what to do with it? We're not at a point where we are sure what to do if we detect abnormalities by an NGS platform like this study."

At the same time, Klco believes that NGS-based methods like the workflow in the study will replace standard approaches normally used to track AML patients. 

"With this tool, you'd be able to detect earlier if a patient is going to truly relapse," Klco said. "If you picked up low levels of a mutation, that would probably make the clinicians be a bit more aggressive early on." he explained. "If you detect a mutation that has a targeted therapy, then it would be reasonable to tailor the therapy based on that mutation, since it may or may not have been there at the time of original diagnosis."

However, the study authors wrote that the overall data "demonstrates that NGS-based posttransplant monitoring in AML patients receiving allogenic HCT provides valuable information and needs to be combined with the baseline mutational profile and clinical evaluation to predict posttransplant outcome and mortality."