Response and outcomes of patients with IDH1/2- mutated accelerated/blast-phase myeloproliferative neoplasms

TO THE EDITOR:

Philadelphia chromosome–negative (Ph^–) myeloproliferative neoplasms (MPNs) carry a variable risk of progression toward accelerated phase (AP) or blast phase (BP).^1,2 Progression to MPN-AP/BP is associated with poor outcomes; our group previously characterized outcomes of MPN-AP/BP in the current era of myeloid therapies and reported a median overall survival (OS) of 0.86 years with no difference in OS based on first-line (1L) therapy.³ One subset of MPN-AP/BP who has several treatment options are those with an IDH1 or IDH2 mutation; the incidence of IDH1/2 mutations in MPN-AP/BP is 20% to 25%.⁴ Given the role of isocitrate dehydrogenase (IDH) inhibition in myelodysplastic syndromes and de novo acute myeloid leukemia (AML) and the efficacy of DNA methyltransferase inhibitor (DNMTi) plus venetoclax (VEN) in IDH1/2-mutated (IDH1/2m) AML,^5,6 characterization of these approaches in IDH1/2m MPN-AP/BP is of particular relevance.7, 8, 9 Here, we characterize the clinical features and outcomes of IDH1/2m MPN-AP/BP compared with MPN-AP/BP without IDH1/2 mutations within our modern cohort of patients.

Our multicenter, retrospective study included adult patients with MPN-AP/BP diagnosed in 2017 or later who also had molecular testing performed at the time of MPN-AP/BP diagnosis. Patient demographics, disease characteristics, treatment, and survival data were collected. Response was characterized by European LeukemiaNet (ELN) 2017 criteria¹⁰ and 2012 MPN-BP criteria.¹¹ Of note, 1 of 9 participating centers was only able to characterize response by ELN 2017. OS was evaluated with Kaplan-Meier analysis. This multicenter retrospective study was approved by each institution's institutional review board (IRB) in accordance with the Declaration of Helsinki.

A total of 166 patients were identified; 35 had a mutation in either IDH1 or IDH2. Among IDH1/2m patients, 11 had an IDH1 mutation, and 24 had an IDH2 mutation. Nine patients had AP at the time of progression, whereas 26 had BP. Clinical characteristics, disease features, and molecular mutations are summarized in Table 1. Disease features were fairly comparable across both groups; however, patients with IDH1/2 mutations had a higher peripheral blast percentage. The most common nondriver comutations with frequency >10% in the IDH1/2m cohort were SRSF2 (43%), ASXL1 (37%), RUNX1 (34%), DNMT3A (17%), TP53 (17%), TET2 (14%), and EZH2 (11%). IDH1/2m patients more frequently carried SRSF2 (43% vs 18%; P = .002) and RUNX1 mutations (34% vs 13%; P = .003) than IDH1/2 wild-type (IDH1/2wt) patients. Analysis of high-risk mutations implicated by Mutation-Enhanced International Prognostic Scoring System 70+ version 2.0 (v2.0) and ELN 2022 adverse-risk criteria revealed no significant differences between these cohorts (Table 1).^12,13

Table 1.

Patient demographics and clinical features at the time of MPN-AP/BP diagnosis

Variable	IDH1/2-mutated n = 35	IDH1/2wt n = 131	P value
Patient demographics and laboratory data
Age at diagnosis, median (IQR)	69.5 (14.1)	67.8 (12.4)	.57∗
White blood count, median (IQR), ×103/μL	11.4 (18.3)	8.7 (25.2)	.32∗
Platelet count, median (IQR), ×103/μL	72.0 (159.0)	110.0 (51.0, 162.0)	.10∗
Hemoglobin, median (IQR), g/dL	8.3 (2.2)	8.5 (2.2)	.59∗
Peripheral blast, median (IQR), %	21.0 (25.0)	10.0 (20.0)	.0001∗
Marrow blast, median (IQR), %	22.8 (37.5)	22.0 (25.0)	.57∗
Total bilirubin, median (IQR), g/dL	0.7 (0.3)	0.7 (0.6)	.27∗
Creatinine, median (IQR), g/dL	0.9 (0.5)	1.0 (0.5)	.53∗
Palpable splenomegaly, n (%)	14 (41.2)	69 (53.9)	.19∗∗
Chronic-phase MPN type, n (%)			.28∗∗
Polycythemia vera	7 (20.0)	25 (19.1)
Essential thrombocythemia	9 (25.7)	37 (28.2)
Primary myelofibrosis	9 (25.7)	49 (37.4)
Other MPN	10 (28.6)	20 (15.3)
MPN driver mutation, n (%)
JAK2	26 (74.3)	78 (59.5)	.11∗∗
CALR	3 (8.6)	25 (19.1)	.20∗∗∗
MPL	3 (8.6)	13 (9.9)	1.00∗∗∗
Triple negative	4 (11.4)	17 (13.0)	1.00∗∗∗
ELN22 adverse-risk mutations, n (%)
ASXL1	13 (37.1)	39 (29.8)	.40∗∗
EZH2	4 (11.4)	11 (8.4)	.52∗∗∗
SRSF2	15 (42.9)	24 (18.3)	.002∗∗
STAG2	2 (5.7)	5 (3.8)	.64∗∗∗
TP53	6 (17.1)	37 (28.2)	.18∗∗
U2AF1	1 (2.9)	11 (8.4)	.46∗∗∗
BCOR	0 (0.0)	5 (3.8)	.59∗∗∗
RUNX1	12 (34.3)	17 (13.0)	.003∗∗
SF3B1	2 (5.7)	10 (7.6)	1.00∗∗∗
ZRSR2	1 (2.9)	4 (3.1)	1.00∗∗∗

ELN22, European LeukemiaNet 2022; IQR, interquartile range.

^∗Median test.

^∗∗χ² test.

^∗∗∗Fisher exact test.

We also analyzed responses in the IDH1/2m cohort based on 1L treatment approach (supplemental Table 1). Seven patients were treated on a clinical trial. The most common 1L treatment approaches were intensive chemotherapy (IC) (n = 10), DNMTi + VEN–based therapy (n = 9), other DNMTi-based (DNMTi monotherapy or DNMTi + JAK inhibitor) therapy (n = 8), and IDH inhibitor (IDHi)–based (DNMTi + IDHi or IDHi monotherapy) therapy (n = 6). Patients with IDH1 mutations received ivosidenib, and patients with IDH2 mutations received enasidenib. By ELN 2017, the overall response rate (ORR) was defined as complete remission (CR) + CR with incomplete count recovery (CRi) + partial remission (PR). The ORR for IC was 40% (3 CR and 1 CRi), 67% for DNMTi + VEN–based therapy (2 CR and 4 CRi), 50% for other DNMTi-based therapy (3 CRi and 1 PR), and 33% for IDHi-based therapy (1 CRi and 1 PR). Response by 2012 MPN-BP was also reported in patients with available data. ORR was defined as complete molecular response + complete cytogenetic response + acute leukemia response (ALR)-complete (ALR-C) + ALR-partial (ALR-P). The ORR was 57% (1 complete cytogenetic response and 3 ALR-C) with IC (n = 7), 57% (3 ALR-C and 1 ALR-P) with DNMTi + VEN–based therapy (n = 7), 50% (2 ALR-C and 2 ALR-P) with other DNMTi-based therapy (n = 8), and 67% (1 ALR-C and 3 ALR-P) with IDHi-based therapy (n = 6).

After characterizing response in the IDH1/2m cohort, we analyzed survival outcomes. The median OS for IDH1/2m patients (n = 35) was 2.2 years (95% confidence interval [CI], 0.65-3.29), whereas the median OS was 0.79 years (95% CI, 0.56-0.99) for the IDH1/2wt group (n = 131; P = .06; Figure 1A). There was no significant difference in median OS between IDH1m patients (n = 11) and IDH2m patients (n = 24; 1.47 years vs 2.56 years; P = .24; Figure 1B). We also looked at median OS specifically in MPN-AP and MPN-BP; there was no significant difference between the IDH1/2m MPN-AP (n = 9) and IDH1/2wt (n = 40) groups (0.48 years vs 1.05 years; P = .66). When focusing on MPN-BP, however, the median OS was 3.29 years for the IDH1/2m group (n = 26) and 0.28 years for the IDH1/2wt group (n = 90; P = .02; Figure 1C). Characteristics for those with MPN-BP are noted in supplemental Table 2.

Figure 1.

Survival outcomes in patients with IDH1- or IDH2-mutated MPN-AP/BP. (A) OS in patients with IDH1/2m MPN-AP/BP compared with those with IDH1/2wt MPN-AP/BP. (B) OS in patients with IDH1m MPN-AP/BP compared with those with IDH2m MPN-AP/BP. (C) OS in patients with IDH1/2m MPN-BP compared with those with IDH1/2wt MPN-BP. (D) OS in patients with IDH1/2m MPN-AP/BP based on 1L treatment approach. JAKi, JAK inhibitor.

We next investigated survival outcomes by 1L therapy within the IDH1/2m cohort. The median OS was 1.5 years for IC, not reached for DNMTi + VEN–based therapy, 2.2 years for other DNMTi-based therapy, and 2 years for IDHi-based therapy (P = .9976; Figure 1D). Of note, the median follow-up for the DNMTi + VEN group was 1.1 years. In addition, we investigated progression-free survival from the initiation of 1L therapy (progression defined as death or initiation of second-line therapy, whichever occurred sooner) by each treatment approach. The median progression-free survival was 0.48 years for IC, not reached for DNMTi + VEN–based therapy, 1.46 years for other DNMTi-based therapy, and 0.55 years for IDHi-based therapy (P=.59). The median follow-up for the DNMTi + VEN group on 1L therapy was 1.1 years as well. We also performed a univariate and multivariate analysis of factors potentially affecting OS in our cohort of IDH1/2m patients. Univariate analyses included Cox proportional hazards regression with 1 predictor. Multivariate analyses used stepwise Cox proportional hazards regression for OS with all predictors, and the P value for entry in the model was .5. The variables are summarized in supplemental Table 3A-B. Platelet, hemoglobin, and white blood count thresholds were set using Mutation-Enhanced International Prognostic Scoring System 70+ v2.0, and blasts were analyzed as a continuous variable because ≥5% blasts have been previously noted as reflecting high-risk disease biology.^13,14 On multivariate analysis the presence of a CALR mutation was significantly associated with inferior OS.

Ten of 35 patients (29%) in the IDH1/2m cohort underwent allogeneic hematopoietic stem cell transplant (allo-HCT). The median OS from the time of allo-HCT was 3.18 years (95% CI, 2.30 to not reached). Of note, 41 of 131 patients (31%) in the IDH1/2wt cohort underwent allo-HCT, and the median OS from the time of allo-HCT was 1.76 years. We also analyzed response and survival in patients who received an IDHi in the relapsed/refractory (R/R) setting. Eight patients (1 IDH1m and 7 IDH2m) received an IDHi, with an ORR of 25% (2 CRi) by ELN2017 criteria and 50% (2 ALR-C and 2 ALR-P) by 2012 MPN-BP criteria. The median OS from the time of IDHi initiation in the R/R setting was 1.1 years.

Although the approval of several novel myeloid malignancy therapies has not translated into markedly improved outcomes in MPN-AP/BP, patients with IDH1 or IDH2 mutations may particularly benefit from these advances. The median OS in the IDH1/2m MPN-AP/BP cohort was 2.2 years, compared with 0.79 years in the IDH1/2wt MPN-AP/BP cohort, with no significant statistical difference. However, when focusing on MPN-BP, IDH1/2m patients had significantly longer median OS than IDH1/2wt patients. Intensive 1L approaches did not result in a significantly different median OS in IDH1/2m patients compared with lower-intensity approaches, similar to what has been noted in previous MPN-AP/BP cohorts.^3,15 Of note, IDHi-based therapies retained efficacy in the R/R setting, with an ORR of 50% by 2012 MPN-BP criteria and a median OS of 1.1 years. The most durable survival outcomes were seen in the 10 patients who received allo-HCT. Although our cohort noted CALR mutation status as affecting OS in multivariate analysis, other reports have shown no impact of driver mutations in MPN-AP/BP outcomes.16, 17, 18 In addition, our cohort only had 3 patients who were IDH1/2m with a CALR mutation. We also noted a discrepancy in response rate when using ELN 2017 and 2012 MPN-BP, which is in line with previous reports that have demonstrated that AML-based response criteria may not adequately capture the benefit of therapy in MPN-AP/BP.^3,19,20 Although IDH1/2m patients with MPN-AP/BP appear to benefit from the therapeutic advances in our cohort, it is necessary to prospectively investigate these strategies, with efforts to study the combination of JAK2 inhibition and IDH inhibition underway (ClinicalTrials.gov: identifier: NCT04281498 and NCT06291987).

Conflict-of-interest disclosure: R.M.S. reports consultancy fees and honoraria from Gilead Sciences, Inc, Servier, Rigel, and Kura Oncology; and other fees including for steering committee from Servier. V.G. reports consultancy and advisory board fees from Novartis, Incyte, Bristol Myers Squibb (BMS)-Celgene, AbbVie, and GlaxoSmithKline (GSK); and research grants from AbbVie and Novartis. E.C.C. reports consultancy fees from AbbVie and Rigel. S.G.I. reports honoraria from Medical Loquix; and membership on an entity’s board of directors or advisory committees in MorphoSys. R.K.R. reports research funding from Incyte Corporation, Stemline Therapeutics, Constellation/MorphoSys, Zentalis, and Ryvu; and consultancy fees from Incyte Corporation, Celgene/BMS, Blueprint, AbbVie, CTI BioPharma, Stemline Therapeutics Galecto, PharmaEssentia, Jubilant, Constellation/MorphoSys, Sierra Oncology/GSK, Protagonist, Cogent, Sumitomo Dainippon, Kartos, Servier, Zentalis, Karyopharm, Disc Medicine, Jazz Pharmaceuticals, Novartis, and Promedior. G.S.G.M. reports advisory board fees from BeiGene, Gilead Sciences/Kite, Pfizer, Autolus, Syndax, and BMS; research funding from BeiGene, Gilead Sciences/Kite, LOXO/Lilly, Zentalis, Schrodinger, and Merck; consultancy fees from Amgen; and speakers bureau fees from Amgen, Rigel, and Stemline. T.B. reports membership on an entity’s board of directors or advisory committees on and speakers bureau fees from Novartis, Geron Corporation, and Gilead. Y.A. reports consultancy fees from Kite/Gilead, Pfizer, Servier, Astellas, Rigel, BMS/Celgene, and Geron; and research funding from ALX Oncology, Biosight, Curis, Novartis, Biomea Fusion, and AbbVie. J.S.G. reports consultancy fees from AbbVie, Genentech, and Servier; research funding from AbbVie, Genentech, Taiho, and Newave; and membership on an entity’s board of directors or advisory committees in Genentech. K.M.P. reports consultancy fees from AbbVie Incyte, PharmaEssentia, Protagonist Therapeutics, and Sierra Oncology; membership on an entity’s board of directors or advisory committees at AbbVie, Merck, GSK, PharmaEssentia, and Protagonist Therapeutics; and research funding from AbbVie, Blueprint Medicines, BMS, Imago, Kura Oncology, and Merck. O.O. reports research funding from AbbVie (Inst), Agios (Inst), Aprea AB (Inst), Astex Pharmaceuticals (Inst), AstraZeneca (Inst), BMS (Inst), Celgene (Inst), CTI BioPharma Corp (Inst), Daiichi Sankyo (Inst), Incyte (Inst), Janssen Oncology (Inst), and Kartos Therapeutics (Inst); honoraria from AbbVie, Blueprint Medicines, BMS/Celgene (Inst), Celgene, CTI BioPharma, Impact Biomedicines, Kymera, Novartis, Servier, Taiho Pharmaceutical, and Treadwell Therapeutics. A.A.P. reports research funding from Pfizer, Kronos Bio, Sumitomo, Incyte, and Servier; honoraria from AbbVie, BMS, and Sobi. The remaining authors declare no competing financial interests.