A novel differentiation response with combination IDH inhibitor and intensive induction therapy for AML

TO THE EDITOR:

Somatic mutations in the isocitrate dehydrogenase 1 and 2 (IDH1 and IDH2) genes occur in up to 20% of patients with acute myeloid leukemia (AML).^1,2  Ivosidenib (AG-120) and enasidenib (AG-221) are mutant-specific oral inhibitors of the IDH1 and IDH2 proteins, respectively, which are US Food and Drug Administration (FDA) approved as single agents for the treatment of newly diagnosed (ivosidenib) and/or relapsed/refractory (ivosidenib and enasidenib) AML. The complete results of a phase 1 study (NCT02632708) examining ivosidenib or enasidenib in combination with intensive induction chemotherapy (IDHi/“7+3”) in newly diagnosed mutant IDH1 or IDH2 (mIDH) AML have recently been reported.³  As per National Comprehensive Cancer Network (NCCN) guidelines, study patients receiving combination IDHi/7+3 underwent bone marrow (BM) aspirate and trephine core biopsy (BMBx) ∼14 days after initiation of therapy. Although differentiation of myeloid blasts is known to occur with ivosidenib or enasidenib monotherapy, often without a period of BM aplasia,^4-7  the effects of IDHi therapy on blast differentiation and day 14 BM morphology when these agents are used in combination with 7+3 induction chemotherapy is unclear. Here, we describe a distinct pattern of BM response in a subset of patients receiving IDHi/7+3. Recognition of this phenomenon could be of critical importance to avoid misinterpretation of midtreatment BMBx patterns as therapeutic failures.

Investigators from participating sites for NCT02632708 were invited to contribute cases to a post hoc comprehensive BM biopsy review. Thirty-six patients from 4 participating sites consented to participate and were evaluated (additional details are available in supplemental Methods). Three patients did not have day 30 BMBx material available and were excluded from further analysis. Ivosidenib or enasidenib was started on day 1 and continued daily until patients were removed from the study, developed toxicity, or proceeded to stem cell transplant. Clinicopathologic data were collected from electronic medical records and from the clinical study database.

For each patient, the diagnostic BMBx and all subsequent BMBx performed until recovery from induction or removal from study were retrospectively reviewed at individual sites by the hematopathologists (E.F.M., O.P., M.R., and R.P.H.). Patients were categorized into 3 groups based on cellularity and blast composition on the day 14 BMBx: (1) aplasia (D14A): ≤10% cellular and <5% blasts; (2) differentiation (D14D): ≥10% cellular and ≥5% blasts at day 14, and with morphologic and/or flow cytometric evidence of blast differentiation at day 14 or day 21, as described herein; or (3) persistent AML (D14P): ≥10% cellular and ≥5% blasts, andwith no differentiation or clearance of blasts at day 14 or later time points. Pathologists were aware of subsequent BMBx results when classifying patients as D14A, D14D, or D14P. For 1 D14D patient, only aspirate smears were performed at day 14; therefore, day 14 cellularity could not be determined. The presence of IDH1, IDH2, and FLT3–internal tandem duplication mutations at diagnosis was determined by local study-site testing. Next-generation sequencing and mIDH variant allele frequency testing were performed as described in supplemental Methods.

Of the 33 patients studied, 15 were mIDH1 patients who received ivosidenib/7+3, and 18 were mIDH2 patients who received enasidenib/7+3. One patient in each treatment group had mutations in both IDH1 and IDH2. Clinical and baseline characteristics of the cohort are summarized in supplemental Table 1 and were similar to those for the larger NCT02632708 cohort³  (see supplemental Results). At the end of induction evaluation, 30 of 33 patients achieved a morphologic response (complete response [CR], CR with incomplete hematologic recovery/CR with incomplete platelet recovery [CRi/p], morphologic leukemia-free state). Of these, 20 of 30 (67%) had D14A; the remaining 10 of 30 (33%) exhibited increased BM cellularity (median, 20%; range, 10% to 20%) and an increased BM blast percentage (median, 39.5%; range, 5% to 80%) with morphologic or flow cytometric evidence of leukemic blast differentiation (D14D) (Figure 1). The 3 patients with persistent disease (including partial response and stable disease; D14P) showed median BM cellularity (median, 20%; range, 10% to 50%) and blast percentage (median, 35%; range, 20% to 58%) similar to D14D patients.

The majority (7 of 10) of D14D cases showed features of monocytic differentiation. Evidence of monocytic differentiation included the presence of an expanded population of immature monocytic cells, including monoblasts and promonocytes, in patients who showed no evidence of monocytic AML at diagnosis (Figure 2A; Table 1). The immature monocytic cells were larger in size than the diagnostic blasts, with round to folded nuclei and abundant cytoplasm, often with cytoplasmic vacuolization. Monocytic differentiation was also seen by flow cytometric analysis, with persistent immature cells in D14D samples showing a shift to a monocytic immunophenotype, including expression of CD11b, CD14, bright CD33, and/or CD64. Two D14D cases showed evidence of granulocytic differentiation, with an expanded population of differentiating cells resembling abnormal promyelocytes⁸  with abundant cytoplasmic granulation, which were not present at diagnosis, as well as a persistent increased population of cells categorized as blasts (Figure 2B-C; Table 1). The remaining D14D case showed both granulocytic and monocytic differentiation (Figure 2D). Differentiation was most commonly seen at day 14 (8 of 10 patients). However, in 2 patients (D14D-3 and D14D-10; Table 1), differentiation was not apparent at day 14 and only emerged at day 21 (Figure 2E-G); both patients were originally diagnosed with persistent AML based on the day 14 BMBx. Despite the elevated blast percentages at day 14, 7 of 10 D14D patients achieved CR or CRi/p by day 43 postinduction in the absence of additional intensive therapy. The remaining 3 D14D patients received a second cycle of induction and all subsequently achieved CR or CRi/p.

The 3 D14P patients showed no definitive evidence of monocytic or granulocytic differentiation or change in blast morphology at day 14 or day 21 (Figure 2H; supplemental Table 2) and showed no clearance of blasts on subsequent BMBx. Flow cytometric data were available for only 1 D14P patient, who showed no definitive change in blast phenotype over time.

There were no significant differences between D14A and D14D patients with respect to clinicopathologic characteristics or time to count recovery (assessed in the 7 of 10 D14D patients achieving CR or CRi/p after 1 induction cycle) (supplemental Table 1). Identification of significant clinicopathologic differences between the D14D and D14P patients was precluded by the small number of D14P patients. D14D was more common with enasidenib (7 of 18 [38.8%]) than ivosidenib (3 of 15 [20%]) and was seen in 4 of 6 patients with mutations at the arginine 172 position of IDH2 (IDH2^R172) (supplemental Table 1).

Only 1 NPM1 mutation and no RAS-pathway mutations were seen in the 10 D14D patients, whereas NPM1 and RAS-pathway (NRAS and PTPN11) mutations were seen in 8 of 20 and 6 of 20 D14A patients, respectively (supplemental Figure 1; supplemental Table 3). Similar rates of mIDH BM mutation clearance were seen in D14A (4 of 20) and D14D (2 of 10) patients. However, patients with D14D showed delayed reduction of mIDH variant allele frequency compared with D14A patients (supplemental Figure 2).

Although IDHi monotherapy is known to cause differentiation,^4-7  BMBx is not routinely performed early in the course of therapy to assess response in this context. By contrast, day 14 BMBx findings routinely impact management in the context of intensive induction therapy, and the effect of IDHi on BM morphology when used in combination with standard induction chemotherapy has not been previously evaluated. The morphologic findings described here are critical for the clinician and hematopathologist to understand when making therapeutic decisions at this important timepoint.

Approximately one-third of patients receiving IDHi/7+3 showed leukemic cell differentiation. Importantly, although the morphologic findings seen in the D14D patients might be interpreted as refractory AML in the setting of standard induction chemotherapy, 7 of 10 D14D patients achieved CR or CRi/p by day 43 postinduction in the absence of additional cytotoxic therapy; the remaining 3 achieved CR/CRi/p after an additional induction cycle. Therefore, in the context of IDHi/7+3, relatively high BM cellularity and residual blasts at day 14 may not always warrant a second course of induction chemotherapy. Based on these data, delaying additional chemotherapy and repeating a BMBx in 1 to 2 weeks in patients with D14D is reasonable. Persistence of differentiation and/or a further decrease in blast count on repeat BMBx should argue against the administration of additional anthracycline-based chemotherapy.

This study is a post hoc analysis of a nonrandomly selected group of patients from a safety study of IDH inhibitors plus chemotherapy for newly diagnosed AML (NCT02632708). The sample size limits our ability to distinguish features associated with D14D. D14D was more common in patients receiving enasidenib (39%) than in patients receiving ivosidenib (20%) and was seen in 4 of 6 patients with an IDH2^R172 mutation, but the significance of these differences is uncertain. NPM1 and RAS-pathway comutations were relatively common in D14A patients but were rare in D14D patients. Induction chemotherapy is associated with high response rates in NPM1-mutated AML^9,10  and has been reported to induce high rates of RAS-pathway mutation clearance.¹¹  Therefore, D14A may represent leukemic clones that are rapidly eliminated by IDHi/7+3, whereas D14D may reflect clones more likely to differentiate in this context, although future studies are needed to address this possibility.

Our data suggest a nuanced approach to the use of posttherapy BMBx for patients receiving IDHi/7+3. The use of these and other targeted agents¹²  that can induce a differentiation response is likely to increase, underscoring the importance of appropriate management of posttreatment evaluation timepoints.

Contribution: E.F.M., M.C., R.P.H., and M.R.S. conceived and designed the study; A.T.F., E.M.S., R.M.S., and M.R.S. were responsible for patient enrollment and care of patients for NCT02632708; E.F.M., O.P., M.R., A.T.F., E.M.S., P.B.F., A.C.S., M.F., H.W., L.H., J.M., S.C., R.X., C.A., M.C., R.M.S., R.P.H., and M.R.S. collected, analyzed, and interpreted data; and E.F.M., O.P., M.R., A.T.F., E.M.S., P.B.F., A.C.S., M.F., H.W., L.H., J.M., S.C., C.A., M.C., R.M.S., R.P.H., and M.R.S. wrote, reviewed, and/or revised the manuscript.

Conflict-of-interest disclosure: M.R. has provided services to Celgene and Physicians’ Education Resource, and has equity ownership in and has provided services to Auron Therapeutics. A.T.F. is a consultant to and/or participates in advisory board meetings for Agios, AbbVie, Astellas, Daiichi Sankyo, Celgene (Bristol Myers Squibb [BMS]), Takeda, Trovagene, Jazz, Boston Biomedical, and Amgen, and has received research funding from Agios and Celgene (BMS). E.M.S. is a stockholder in/has ownership of Auron Therapeutics; is a consultant for/advisor to AbbVie, Agios, Astellas, Bayer, BioLineRx, Celgene, Daiichi Sankyo, Genentech, Novartis, Pfizer, PTC Therapeutics, and Syros; received research funding from Agios, Amgen, Bayer, Celgene, and Syros; and has been reimbursed for travel expenses by AbbVie, Astellas Pharma, Biotheryx, Celgene, Daiichi Sankyo, Novartis, Syndax, and Syros. P.B.F. received research funding from Incyte, Forma Therapeutics, and Astex Pharmaceuticals, and is a consultant to Agios. M.F. is an employee of and has equity ownership in Celgene Corporation. H.W., L.H., J.M., S.C., R.X., C.A., and M.C. are employees of and stockholders in Agios Pharmaceuticals, Inc. R.M.S. is on steering committees for AbbVie and Agios; is on advisory boards for Actinium, Amgen, Astellas, BioLineRx, GEMoaB, Janssen, Macrogenics, Novartis, Takeda, and Trovagene; received research funding from Arog and Novartis; is on a data and safety monitoring board for Celgene; and is a consultant to Daiichi-Sankyo and Novartis. M.R.S. received research support from ALX Oncology, Astex, Incyte, Takeda, and TG Therapeutics; has equity in Karyopharm; has provided advisory/consultancy services to AbbVie, BMS, Celgene, Geron, Ryvu, Sierra Oncology, Taiho Oncology, and TG Therapeutics; and holds patents with/receives royalties from Boehringer Ingelheim. The remaining authors declare no competing financial interests.

Correspondence: Michael R. Savona, Vanderbilt-Ingram Cancer Center, Vanderbilt University School of Medicine, Preston Research Building 777, 2200 Pierce Ave, Nashville, TN 37232; e-mail: michael.savona@vanderbilt.edu.