Decitabine and Venetoclax for IDH1/2-mutated Acute Myeloid Leukemia

Isocitrate dehydrogenase (IDH) 1/2 mutations occur in approximately 20% of older patients with acute myeloid leukemia (AML). Ivosidenib, an IDH1 inhibitor, is approved for the treatment of relapsed/refractory (R/R) IDH1^mut AML and newly diagnosed (ND) patients ineligible for standard therapy, and enasidenib, an IDH2 inhibitor in IDH2^mut R/R AML. In ND IDH1^mut AML, ivosidenib offers complete remission (CR) plus CR with partial hematologic recovery (CRh) rate of 42% and median overall survival (OS) of 12.6 months, and in the R/R setting, CR/CRh of 30% and median OS of 8.8 months.¹ Comparably, enasidenib in ND IDH2^mut AML patients showed a CR/CR with incomplete hematologic recovery (CRi) rate of 21% and median OS of 11.3 months, and in R/R patients, a CR/CRi of 26% and median OS of 8.8 months.²

Venetoclax with hypomethylating agent (HMA) is now standard for elderly patients ineligible for intensive therapy. Preclinical studies have shown that IDH1/2^mut primary AML cells are highly sensitive to BCL-2 inhibition which was confirmed by clinical studies showing durable responses in IDH1/2^mut patients receiving venetoclax monotherapy. In the recently published randomized placebo-controlled trial, Venetoclax and azacitidine showed a CR/CRi rate of 75% of patients with ND IDH1/2^mutAML with a median OS of 24.5 months.³ Herein, we describe the outcomes of patients with treatment naïve and previously treated IDH1/2^mut AML who were treated on a prospective phase 2 trial of 10-day decitabine with venetoclax (DEC10-VEN, NCT03404193).⁴

Older patients with ND patients (>60 years) and adult patients >18 years with secondary AML (sAML) from antecedent hematological disorder (AHD), with or without prior therapy for AHD, and R/R AML, with ECOG performance status ≤3 without prior BCL-2 inhibitor exposure were enrolled. In line with our prior report and due to differences in outcomes, we classified treatment-naïve AML as ND AML or sAML without prior therapy for AHD, and the previously treated group as patients with prior treatment for AHD or R/R AML. Patients received 10-day decitabine at 20 mg/m²/day for induction followed by 5-days for consolidation. Venetoclax dose was 400 mg PO daily or equivalent (with azole co-administration). Full protocol of the study have been published previously.⁴ Reduction in venetoclax duration to <21 days per cycle was allowed to mitigate any myelosuppression. Addition of IDH1/2 inhibitors was allowed. Responses, overall survival (OS), and relapse-free survival (RFS) were graded per ELN2017 guidelines for AML. Measurable residual disease (MRD) was assessed on bone marrow specimens using multiparametric flow cytometry (FCM) validated to a sensitivity level of 0.01–0.1%. IDH1/2^mut testing was performed using a next-generation sequencing panel with an analytical sensitivity of 5% mutant reads in a background of wild-type reads; or separately using Sanger sequencing during follow-up (variant allele frequency sensitivity 20%).

Between January 20, 2018 and March 19, 2020, we treated 10 and 25 patients with IDH1^mut and IDH2^mut AML, respectively (Supplement Table 1). No patients with IDH1^mut received concurrent ivosidenib. Among patients with treatment naive IDH1^mut AML (n=7), the median age was 69 years (range, 65–77), and three (43%) patients had adverse risk AML per ELN 2017. ND IDH1^mut AML cohort was enriched with DNMT3A (57%), NPM1, SRSF2 (43% each) mutations. (Fig. 1f) The overall response rate (ORR) was 100% (n=7/7) with MRD negativity by FCM (MRD^neg) achieved in 43% patients (n=3/7). (Fig. 1a) The 60-day mortality was 0%. After a median follow-up of 23.2 months three patients have relapsed (43%). The 1-year OS was 71% and the 1-year RFS was 84% (Fig. 1b and c).

Fig. 1.

a.Outcomes in treatment-naïve (TN) and previously treated (PT) patients with IDH1 and IDH2 mutant acute myeloid leukemia (AML), b-e. Kaplan-Meier plots showing b. overall survival (OS), c. relapse-free survival (RFS), in IDH1 mutant AML, and d. OS, e. RFS, in IDH2 mutant AML f. mutational landscape of the whole cohort.

Abbreviations: All results expressed as % (No.). ORR = overall response rate = CR+CRi+MLFS, CR = complete remission CRi = CR with incomplete hematologic recovery, MLFS = morphologic leukemia-free state; aplasia was defined as in evaluable bone marrow sample due to cellularity <10%; MRD = measurable residual disease, FCM = flow cytometry; PCR = polymerase chain reaction; NGS = next generation sequencing. VAF = variant allele frequency, ELN = European LeukemiaNet 2017, CG = cytogenetic, ND = newly diagnosed, R/R = relapsed or refractory, sAML = secondary AML, SCT = stem-cell transplantation, fav = favorable, int = intermediate, adv = adverse, Neg = negative, Pos = positive N/A = not available

Three patients with previously treated IDH1^mut AML had a median age of 77 years (range, 75–79) and had received a median of one prior therapy for AHD or AML (range, 1–2). This cohort was enriched with DNMT3A, K/NRAS, NPM1, PHF6, TP53 (33% each). (Fig. 1f) Only one patient (33%) attained CR MRD^neg; one patient (33%) was inevaluable and one patient (33%) was refractory. The 60-day mortality was 0%. After a median follow-up of 14.7 months, the 1-year OS was 66% and the 1-year RFS was 100% (Fig. 1b and c).

Among patients with IDH2^mut AML, 14 patients had treatment-naive AML and 11 patients had previously treated AML. One treatment-naïve and one previously treated patient continued to receive decitabine and venetoclax with the addition of enasidenib from cycle 15 and 6, respectively. The median age in the treatment-naïve group was 72 years (range, 65–80) and 36% patients (n=5) had adverse risk AML. ND IDH2^mut AML cohort was enriched with SRSF2 (57%), K/NRAS (29%), DNMT3A (28%). (Fig 1f) The ORR was 86% (n=12/14) and MRD^neg was achieved in 57% patients (n=8/14). (Fig. 1a) One patient (7%) had aplasia, one patient (7%) had refractory disease, and four patients (29%) relapsed. The 60-day mortality was 7% (n=1/14) which was related to bacteremic sepsis. At a median follow up of 19.9 months, the 1-year OS was 79%, the median OS was 29.6 months (Fig. 1d) and 1-year RFS was 74% (Fig 1e).

Among 11 patients with previously treated IDH2^mut AML, the median age was 62 years (range, 34–75), six patients (55%) had ELN adverse risk AML and had received a median of three prior therapies (range, 1–5). This cohort was enriched with DNMT3A (36%) and TP53 (27%). (Fig. 1f) The ORR was 82% (n=9/11) and MRD^neg rate was 54% (n=6/11). One patient was inevaluable (9%), one patient was refractory (9%) and four patients (36%) relapsed. The 60-day mortality was 9% (n=1) which was related to traumatic intracranial hemorrhage. At a median follow up of 22.1 months the 1-year OS was 59% and median OS was 14.7 month (Fig. 1d). The 1-year RFS was 80% and median RFS was 15.6 months (Fig. 1e).

The most frequent new mutations emerged on NGS testing in 11 patients at the time of relapse included SRSF2 (n=5/10), DNMT3A (n=3), WT1 (n=3), and K/NRAS (n=3, Fig. 1f). DEC10-VEN appeared effective in clearing IDH1/2^mut clones in majority of patients and three patients (3/10) relapsed with undetectable IDH1/2^mut. Co-mutations in tumor suppressor genes (TP53, WT1, PHF6) were more frequent at baseline or at relapse in patients with refractory disease or relapse compared to those with durable responses without relapse till data cut-off for this report (9/14 vs 4/18, p=.02).

After a median follow-up of 20.2 months in all patients, 9 patients (26%) continue therapy and 21 patients (60%) are alive. The reasons for treatment discontinuation in 26 patients included relapse (n=8, 23%), SCT (n=6, 17%), death (n=4, 11%), refractory disease (n=3, 9%), patient choice (n=2, 6%), toxicity (n=2, 6%) and new lung cancer diagnosis (n=1, 3%). Causes of death in 14 patients included infections (n=4, 11%), death in hospice (n=4, 11%), unknown reason (n=3, 9%), intracranial hemorrhage (n=2, 6%), and cardiac arrest (n=1, 3%).

With necessary caution for cross-trial comparisons and differences in patient populations, CR/CRi rates with DEC10-VEN in both treatment-naive and previously treated IDH1/2^mut AML were numerically higher compared to IDH1/2 inhibitor monotherapy ^{1, 2} and were comparable with available data on HMA+IDH inhibitors or HMA+venetoclax. In IDH1^mut treatment-naive patients ineligible for intensive chemotherapy, CR/CRi rate of 86% with DEC10-VEN was modestly higher compared to CR/CRi of 70% with ivosidenib and azacitidine.⁵ Similarly, in treatment-naive patients with IDH2^mut AML,CR/CRi of 70% with DEC10-VEN was comparable to CR/CRi of 63% with enasidenib and azacitidine.⁶ These responses rates of 70–86% with DEC10-VEN were comparable to CR/CRi rate of 75% with azacitidine and venetoclax for IDH1/2^mut AML.³

Although all AML patients with IDH1/2^mut and/or NPM1^mut treated with frontline venetoclax combination have been reported to have excellent outcomes in terms of sustained remission and survival benefit,³ durable remissions are possibly unique to IDH2/NPM1 co-mutated patients.⁷ In our cohort, outcomes in treatment naïve IDH1^mut AML appeared favorable compared to IDH2^mut AML, which may be due to the limited sample size, enrichment for NPM1 mutations in IDH1^mut cohort (43% vs 21%) and previously treated cohort having a median of one prior line of therapy in IDH1^mut patients compared to three lines in IDH2^mut cohort. Of note, the triplet combination of DEC10-VEN+enasidenib maintained and prolonged the duration of remission in two patients who experienced disease progression on DAC10 + VEN combination (for 11 and 8 months, respectively). The value and optimal regimen of venetoclax and IDHi doublets and triplets in IDH1/2^mut AML need further clarification through larger datasets. One prospective trial evaluating ivosidenib with venetoclax +/− HMA triplet combination is underway (NCT03471260).

Although our study is limited by the lack of apriori post-hoc analysis and a small sample size, DEC10-VEN appears to be a possibly efficacious frontline option with high activity for older patients with IDH1/2^mut ND AML, and additionally may be an effective strategy for R/R IDH1/2^mut AML. Early referral for high-risk patients and post-SCT maintenance with IDH inhibitors, as well as ongoing evaluation of this and other clinical trials incorporating “triplet” therapy, may offer even enhanced future outcomes.