Molecular Characterization of Clinical Response and Relapse in Patients with IDH1-Mutant Newly Diagnosed Acute Myeloid Leukemia Treated with Ivosidenib and Azacitidine

Background: Acute myeloid leukemia (AML), a hematologic malignancy characterized by clonal expansion of abnormal myeloid progenitors, is a disease exhibiting a dynamic mutational landscape over time. Somatic mutations in isocitrate dehydrogenase 1 (IDH1) are reported in 6–10% of patients (pts) with AML. Ivosidenib (IVO) is an oral, potent, targeted inhibitor of mutant IDH1 (mIDH1) and is FDA-approved for the treatment of mIDH1 relapsed/refractory (R/R) AML and newly diagnosed (ND) AML in adults ≥ 75 years (yrs) of age or with comorbidities precluding intensive induction chemotherapy (IC). In an ongoing phase 1b study (NCT02677922), 23 pts (11 male; median age 76 yrs [range 61–88]) with mIDH1 ND AML received IVO 500 mg daily and subcutaneous azacitidine (AZA) 75 mg/m² on Days 1–7 in 28‑day cycles. As of 19Feb2019, median number of treatment cycles was 15 (range 1–30); 10 pts remained on treatment. Overall response rate (complete remission [CR] + CRi [incomplete neutrophil recovery] + CRp [incomplete platelet recovery] + morphologic leukemia-free state [MLFS]) was 78% (18/23): including CR in 61% (14/23) and CR with partial hematologic recovery (CRh) in 9% (2/23). mIDH1 clearance assessed in bone marrow mononuclear cells (BMMCs) by BEAMing digital PCR (detection limit 0.02–0.04%) was observed in 11/16 pts (69%) with CR/CRh, including 10/14 (71%) with CR.

Aim: Characterize clonal evolution and resistance in pts with mIDH1 ND AML treated on study with IVO + AZA.

Methods: The secondary efficacy endpoint of CRh was sponsor derived and defined as CR with absolute neutrophil count > 0.5 X 10⁹/L and platelets > 50 X 10⁹/L. Bulk DNA sequencing (DNA-seq, 1400-gene ACE Extended Cancer Panel, 2% variant allele detection limit ) was performed on BMMCs and/or peripheral blood mononuclear cells (PBMCs). Single-cell (sc) targeted DNA-seq was performed on PBMCs using a microfluidic platform (Tapestri^®) with a 20-gene AML panel capable of detecting rare subclones down to 0.1%.

Results: To identify mechanisms of acquired resistance, longitudinal bulk DNA-seq was analyzed for 22/23 pts, including 5 pts with available samples at relapse or disease progression (3 CR and 1 MLFS with morphological relapse; 1 CRh with disease progression). Mutations not detected at baseline but emerging during therapy were categorized into canonical biological pathways (Table). Emerging mutations were observed in 9/22 (41%) pts, including 4 with multiple mutations. 3/22 (14%) pts had emerging IDH2 mutations, with concurrent rise in plasma 2-hydroxyglutarate (2-HG) levels. Within the relapse/progression cases, emerging mutations were observed in 4/5 pts, including 3 where the emerging mutation appeared to be the predominant mutation at relapse/progression (2 CR pts with IDH2 mutations, and 1 CRh pt with a TET2 mutation). To date, from the bulk DNA-seq analysis, no emergence of an IDH1 second-site or receptor tyrosine kinase pathway (FLT3, KRAS, NRAS, PTPN11) mutation has been observed. To further evaluate clonal evolution of clinical response and disease progression, scDNA-seq was performed, with data available for 15 pts (10 CR, 2 CRh, 1 MLFS, and 2 stable disease), including end-of-study time points for 5 relapse/progression pts. In the 2 relapsed pts with an emerging IDH2 mutation observed by bulk DNA-seq, 1 had a minor IDH2 clone present at baseline that expanded independently from IDH1 during therapy (Fig). In a separate case, a subclonal baseline PTPN11 clone evolved to gain both RUNX1 and IDH2 mutations, becoming the predominant clone at relapse. In 2 other cases, scDNA-seq data showed that non-IDH1 clones were selected from baseline clones ancestral (TP53 n = 1) to or emerged separate from mIDH1 (TET2 n = 1). Clonal architecture and evolution from additional pts will be presented.

Conclusion: IVO + AZA combination treatment in IC-ineligible ND AML led to deep and durable molecular remissions. Although the dataset is small, IDH2 clones appeared to expand or emerge separate from the IDH1 clone, with no observation of an IDH1 second-site mutation to date. Understanding patterns of emerging mutations/pathways at relapse will allow for comparison with mIDH1 R/R AML and ND AML pts treated with IVO monotherapy. These results underline the importance of mutational testing, particularly at progression to determine optimal salvage therapy. Potential combination or sequential therapies should be evaluated prospectively in future clinical trials.