Transplant Outcomes for IDH-Mutated AML: Good Outcomes Thanks to Keeping Good Company

Background: Mutations in the isocitrate dehydrogenase (IDH) 1 and 2 genes are found in approximately 20% of patients with acute myeloid leukemia (AML). Although independent prognostic data are lacking, our program often moves patients with IDH-mutated AML (mIDH) towards allogeneic bone marrow transplant (BMT). Reports on the impact of mIDH on BMT outcomes are conflicting. One study (Quek L, et al. Blood Adv. 2016; 1(3):19320) found improved rates of relapse (RR), relapse free survival (RFS) and overall survival (OS) for patients with mIDH1, whereas another study (Salhotra A, et al. Clin Lymphoma, Myeloma Leuk. 2019 Jul 1;19(7):e400-5) reported higher RR, but no differences in RFS or OS for patients with mIDH1/2. In light of the approved IDH1/2 inhibitors, the role of early BMT for such patients requires additional study. We examined the BMT outcomes of patients with mIDH AML compared to wildtype (wt)-IDH AML.

Patients and methods: Patients with AML who underwent a first allo-BMT at Johns Hopkins between 2014 and 2020 and had molecular testing for IDH mutation status were included. Patients were grouped according to IDH mutation status (mIDH versus wt-IDH). Patient demographics, molecular genetics, initial treatment response, and BMT characteristics were compared. Post-BMT RR, RFS and OS were then compared between groups. Multivariate models were used to assess the prognostic significance of mIDH after controlling for co-variates. We also examined the relationship between peri-BMT IDH inhibitor use and IDH mutation status at the time of post-BMT relapse.

Results: We identified 109 patients with mIDH AML, 70 (64%) of whom underwent BMT. 210 BMT patients with wt-IDH AML were used as a comparison. Outcomes between patients with mIDH1 and mIDH2 AML were similar and they were combined for all further analyses. Patients with mIDH were generally older (mean age at diagnosis 61 (range 28-75) vs. 54 (range 19-76), p < .001), less likely to have poor-risk cytogenetics, more likely to have a haploidentical donor, and more likely to receive non-myeloablative conditioning (Table 1). mIDH AML was associated with co-mutations in SRSF2 and DNMT3A mutations and less likely to have TET2 and KRAS co-mutations.

Patients with mIDH had significantly improved post-BMT RR, RFS and OS compared to patients with wt-IDH (log rank p = 0.04, .01, and .01 respectively, Figure 1). In a multivariate model controlling for ELN risk classification, mIDH was associated with improved OS with a trend towards significance (HR 0.51, 95% CI 0.24 – 1.10, p = 0.085). Recognizing that significant heterogeneity exists within ELN risk groups, we examined the specific components of poor ELN risk classification in each group and found that poor risk in the wt-IDH group was primarily driven by complex karyotypes and TP53 mutations, whereas the ELN poor risk was driven by RUNX1 or ASXL1 co-mutation in the mIDH group. In a model controlling for specific genetic abnormalities (complex karyotype, TP53 mutation, and KRAS mutation) IDH mutation status was not an independent prognostic factor (HR 0.66, 0.34 – 1.28, p = 0.2, Table 2). We also assessed whether improved outcomes might be associated with use of IDH inhibitors in the pre or post-BMT setting. 22 (31%) of the 70 patients with mIDH AML received pre-BMT IDH inhibitors and an additional 4 patients received post-BMT maintenance. Of the 15 patients with mIDH who relapsed post-BMT, 13 had mIDH at the time of relapse. 11 of the 11 patients who had never received an IDH inhibitor had mIDH at relapse compared to 2 of 4 (50%) patients who had received IDH inhibitors (p = .057).

Conclusion: In our cohort, patients with mIDH AML had significantly better OS, RFS, and RR following BMT compared to patients with wt-IDH. However, our analyses suggest that IDH mutations are associated with favorable genetic abnormalities and that the differences in BMT outcome are likely attributable to these differing genetic profiles as opposed to the presence of the mIDH. Differences in outcome were specifically driven by a lower prevalence of complex cytogenetics, TP53 mutations, and KRAS mutations amongst mIDH AML and were incompletely captured by ELN risk classification. In addition, the frequency with which post-BMT relapsed AML bears the IDH-mutant clone suggests that incorporation of IDH inhibitors into the peri-BMT setting, as is currently being studied in large randomized trials (NCT03564821, NCT03515512), may further improve outcomes.