Acquired Clinical Resistance to IDH2 Inhibitors: Leukemia Finds Multiple Avenues for Escape

In approximately 20 percent of patients with acute myeloid leukemia (AML), heterozygous somatic mutations affect the conserved arginine residues R140 and R172 in isocitrate dehydrogenase 2 (IDH2). As a consequence of these mutations, myeloid cell differentiation is impaired. Elegant previous work has shown that this is due to the altered enzymatic function of IDH2 dimers, which aberrantly generate 2-hydroxyglutarate (2-HG). High concentrations of 2-HG interfere with the epigenetic regulation of the transcriptional program of hematopoietic differentiation. Inhibition of mutant-containing IDH2 enzymatic dimers by the specific inhibitor, enasidenib, reduces 2-HG and can restore differentiation in IDH2-mutant leukemia cells in vitro and in patients. Enasidenib is U.S. Food and Drug Administration-approved for use in patients with relapsed or refractory AML carrying a mutated IDH2. Response occurs in about 40 percent of patients and is associated with evidence of differentiation of the leukemia, but progression is common. The mechanisms underpining acquired resistance have been a hot topic of investigation, and answers are now emerging through two recent publications.

Dr. Andrew M. Intlekofer and colleagues studied serial samples from two patients with IDH2-mutant AML who had a clinical response to enasidenib followed by disease progression and a recurrent increase in circulating levels of 2-HG. During the response, peripheral blood counts improved and blast counts were reduced. Consistent with previous observations that enasidenib promotes differentiation of IDH2-mutant blasts, the variant allele frequency for the IDH2^R140Q mutation did not substantially diminish, but remained at levels similar to those documented at the commencement of therapy. In both cases, new mutations in the IDH2 gene were identified at the time of acquired resistance. These missense mutations resulted in a substitution of glutamine 316 with glutamate (Q316E) in the first patient and in substitution of isoleucine 319 with methionine (I319M) in the second patient. Neither of these “second site” mutations were detectable by highly sensitive digital drop polymerase chain reaction assays prior to the initiation of enasidenib. Intriguingly, these mutations affected the other IDH2 allele (i.e., the previously normal allele of IDH2 that did not carry the leukemogenic mutation). When tested in cell lines as the sole mutations in IDH2, these “second-site” mutations did not significantly affect IDH2 function, but when co-expressed with IDH2^R140Q, resistance to enasidenib was observed. Modelling suggested that while these mutations did not affect the catalytic site (as does R140Q), they do appear to perturb the dimer interface where the drug interacts with its target. Second-site mutations were found in only two of nine cases of acquired resistance and were not observed in any of the 14 cases of primary resistance.

In contrast to Dr. Intlekofer and colleagues, Dr. Lynn Quek and colleagues did not find any examples of second-site IDH2 mutations at relapse in 16 patients treated with enasidenib. Their investigation of the clonal basis of response and acquired resistance to enasidenib treatment provides insight into other mechanisms of treatment failure. Using sequential patient samples and applying advanced flow cytometric analyses, clonal culture, and next-generation sequencing, they determined the clonal structure of hematopoietic cell populations at different stages of differentiation. The authors found that in pretreatment samples, the ability of mutant IDH2 to impose differentiation block was dependent on the context of co-associated mutations within a leukemic clone. Response to enasidenib was associated with relief of differentiation arrest and normalization of the balance between stem, progenitor, and precursor cell compartments. There was considerable heterogeneity as to which clones dominated the differentiated populations during response. In only a minority of patients was normal hematopoiesis, free of mutations, restored. In the majority, the differentiated clone was either the presenting leukemic clone or an ancestral clone that had given rise to the presenting leukemic clone. Their data suggested that the efficacy of enasidenib-induced differentiation was also likely to depend on the mutational landscape within a clone.

In 14 of 16 patients, 2-HG remained suppressed at relapse, suggesting that enasidenib remained active on target. In these 14 cases, relapse arose either by genetic evolution with acquisition of additional mutations in other driver genes, or by selection of pre-existing clones already carrying oncogenic mutations in other genes. In some instances, these emergent clones represented ancestral clones, while in other cases, the clones were the presenting terminal leukemia. Fascinatingly, in the two patients in whom 2-HG levels rose despite ongoing enasidenib therapy, acquired mutations in IDH1 were found. In all, seven different patterns of genetic evolution were observed among 16 cases, highlighting the multitude of paths for escape available for AML cases bearing an IDH2 mutation.