AML Relapse Predictive Ability of ctDNA Comparable With Bone Marrow Mutation Persistence

October 15, 2019

Article

Among patients with acute myeloid leukemia and myelodysplastic syndrome who completed myeloablative allogeneic hematopoietic stem cell transplantation, the predictive utility of testing circulating tumor DNA was comparable with that of mutation persistence evaluation in matched bone marrow samples, according to a study published recently in Blood.

Among patients with acute myeloid leukemia and myelodysplastic syndrome (AML/MDS) who completed myeloablative allogeneic hematopoietic stem cell transplantation (alloSCT), the predictive utility of testing circulating tumor DNA (ctDNA) was comparable with that of mutation persistence evaluation in matched bone marrow samples, according to a study published recently inBlood.

The authors, led by Sousuke Nakamura, MD, of the University of Tokyo, found that patients with persistent ctDNA-positive status at either 1 or 3 months following transplant were significantly more likely to relapse than those whose ctDNA status was negative. Additionally, they identified increasing ctDNA levels between 1 and 3 months post-transplant as the specific predictor of relapse.

“Our study, for the first time, provides proof of concept that driver mutation persistence based on serum ctDNA can serve as a comparable prognostic biomarker for patients with AML and MDS undergoing alloSCT,” the study authors wrote. “Our results also indicated that ctDNA testing may replace conventional MRD [minimal residual disease] testing with BM [bone marrow] as a noninvasive biomarker for prediction of impending relapse in these patients.”

The study population consisted of 53 patients with AML and MDS who received alloSCT at the University of Tokyo, but 51 patients were used in the ctDNA analysis. Of these, 15 patients hadde novoAML, 22 patients had secondary AML, and 14 patients had MDS. Their median age was 53 (range, 17-68) and all patients had received myeloablative conditioning. Tumor and serum samples taken at diagnosis and either 1 or 3 months following alloSCT were available for all patients. Half the patients had relapsed or refractory disease at the time of their alloSCT, and most patients received cord blood as their stem cell source.

The authors used next-generation sequencing (NGS) to identify driver mutations, which they found in 51 patients. With 37 mutated genes total, the median number of mutations per-patient was 2, although some patients had as many as 5 mutations. The most frequent mutations found involved epigenetic regulators (TET2/ASXL1/DNMT3A), which were mutated in 32.1% of the samples. This was followed by signal transduction proteins (NRAS/FLT3, 31.4%) and spliceosome factors (U2AF1/SF3B1/SRSF2, 21.6%). Approximately one-third of patients had a normal karyotype, while nearly 40% of patients had a conventional cytogenetic risk category of adverse or high risk.

Next, Nakamura et al designed at least 1 personalized digital polymerase chain reaction assay per case (droplet digital PCR, ddPCR). “As expected, all ctDNAs, for which we could design ddPCR assays, were identified in all available diagnostic serum samples (n = 51),” they wrote. “Diagnostic ctDNA and matched tumor DNA exhibited excellent correlations with variant allele frequencies.”

About one-third of patients experienced disease relapse (16 of 51). The median follow-up after transplant was 32 months, and the median time to relapse was 7 months (range, 1.9-53.6). The authors investigated an association between the molecular MRD status and patient outcome.

“As expected, patients with MP1 [mutation positive at 1 month] and MP3 [mutation positive at 3 months], and correspondingly CP1 and CP3 [ctDNA persistence at 1 month and 3 months], were associated with an increased risk of relapse and death compared with those with a molecular MRD-negative status,” Nakamura et al wrote. “Overall, these results supported the relevance of serum ctDNA in molecular MRD testing as a noninvasive alternative to BM mutation status.”

The 3-year cumulative incidence of relapse (CIR) rate in patients with MP1 was 72.9% compared with 13.8% in patients who were not mutation positive at 1 month (P= .0012), and the 3-year overall survival (OS) rates were 50.0% and 88.0%, respectively (P= .0304). Among patients with CP1, the 3-year CIR rate was 65.6% and the 3-year OS rate was 45.8% compared with rates of 9.0% and 91.7%, respectively (P= .0002 andP= .0014), in patients who did not have ctDNA persistence at 1 month.

Among patients with MP3, the 3-year CIR rate was 80.0% compared with 11.6% in patients who were not mutation positive at 3 months (P= .0002), and the 3-year OS rates were 30.0% versus 94.1%, respectively (P= .0007). In patients who had CP3, the 3-year CIR rate was 71.4% versus 8.4% in patients who did not have ctDNA persistence at 3 months (P<.0001), and the 3-year OS rates were 53.4% and 92.5%, respectively (P= .0021).

Nakamura et al also explored stratification of relapse risk based on ctDNA kinetics. They divided 40 patients who had ctDNA information available at both 1 and 3 months into either “increasing,” “both negative,” or “decreasing/stable” groups. Results showed that patients in the increasing group were associated with the highest risk of relapse compared with those in decreasing/stable and both-negative groups (P= .0027 and P <.0001, respectively).

“Although not statistically significant, patients in the decreasing/stable group exhibited a clear trend of relapse compared with those in negative ctDNA (P= .087),” Nakamura et al wrote. “Based on these results, it is recommended that clinicians monitor ctDNA kinetics as well when predicting relapse.”

The study authors also found an association of ctDNA persistence in patients with mutations in epigenetic regulators, which had a prognostic impact on relapse and survival.

Nakamura et al noted several study limitations, including the fact that the most useful checkpoint for predicting patient outcomes needs further testing. Additionally, their approach requires the construction of a personalized ddPCR assay for each patient. Another limitation is that their patient pool was relatively young and had received myeloablative conditioning prior to transplant, perhaps limiting the findings’ relevance for elderly patients who must receive reduced-intensity conditioning. Finally, the small sample and retrospective design must be tested in larger prospective trials.

The study authors are currently conducting a prospective study to address these issues.

Reference:

Nakamura S, Yokoyama K, Shimizu E, et al. Prognostic impact of circulating tumor DNA status postallogeneic hematopoietic stem cell transplantation in AML and MDS. Blood. 2019;133(25):2682-2695. doi: 10.1182/blood-2018-10-880690.