Molecular patterns of response and treatment failure after frontline venetoclax combinations in older patients with AML

Abstract

The BCL-2 inhibitor venetoclax combined with hypomethylating agents or low-dose cytarabine represents an important new therapy for older or unfit patients with acute myeloid leukemia (AML). We analyzed 81 patients receiving these venetoclax-based combinations to identify molecular correlates of durable remission, response followed by relapse (adaptive resistance), or refractory disease (primary resistance). High response rates and durable remissions were typically associated with NPM1 or IDH2 mutations, with prolonged molecular remissions prevalent for NPM1 mutations. Primary and adaptive resistance to venetoclax-based combinations was most commonly characterized by acquisition or enrichment of clones activating signaling pathways such as FLT3 or RAS or biallelically perturbing TP53. Single-cell studies highlighted the polyclonal nature of intratumoral resistance mechanisms in some cases. Among cases that were primary refractory, we identified heterogeneous and sometimes divergent interval changes in leukemic clones within a single cycle of therapy, highlighting the dynamic and rapid occurrence of therapeutic selection in AML. In functional studies, FLT3 internal tandem duplication gain or TP53 loss conferred cross-resistance to both venetoclax and cytotoxic-based therapies. Collectively, we highlight molecular determinants of outcome with clinical relevance to patients with AML receiving venetoclax-based combination therapies.

Graphical abstract

Figure 1.

Categorization of outcomes and genomic annotation of the study population. (A) Kaplan-Meier plots of OS for patients with AML treated with venetoclax in combination with either DNMTi’s or LDAC. n = 81, P = .52 (log-rank) for comparison between different combinations. (B) Kaplan-Meier plots for relapse-free survival for patients with AML achieving response (CR, n = 35; CRi, n = 17; and MLFS, n = 7) after treatment with either venetoclax in combination with DNMTi’s or LDAC. n = 59, P = .93 (log-rank) for comparison between different combinations. (C) Categorization of patients into those achieving durable remission (group A, n = 18), remission then relapse with adaptive resistance (group B, n = 25), primary refractory disease (group C, n = 20), or into an uncategorized group when the criteria were not met for any of the other categories (n = 18). The presence of adverse cytogenetic risk, complex karyotype, del(17p), indicated mutations, study ID number, best response (CR, CRi, MLFS, resistant disease [RD], or nonevaluable [NE]), cytotoxic therapy received (AZA, azacitidine; DEC, decitabine; or LDAC), and prior exposure to DNMTi are shown for each case. Shown to the right of the plot are the number of cases achieving (dark gray) or not achieving (light gray) CR or CRi.

Figure 2.

Characterization of AML cases with durable remission. (A) Genomic landscape of AML cases with durable remission. Molecular profile for 18 patients with ongoing remission for >12 months (date cutoff, 28 February 2019). The presence of adverse cytogenetic risk, complex karyotype, del(17p), indicated mutations, study ID number, relapse-free survival (in months), and cytotoxic therapy received (AZA, azacitidine; DEC, decitabine; or LDAC) are shown for each case. The bar graphs (right side of the plot) summarize the number of cases with persistent mutations or reduced/cleared mutations or where MRD was not assessed in remission. (B) Quantitative MRD profile of NPM1 mutation in 4 AML cases as assessed by RT-qPCR. MRD levels are expressed as percent relative to baseline transcript level. Positive detection is shown as solid-filled black circles, whereas negative detection is shown as open circles, indicating the assay sensitivity relative to amplification of the ABL gene. (C) Quantitative MRD profile of IDH2 mutation in 7 AML cases as assessed by Droplet Digital PCR (n = 5) and next-generation sequencing (NGS) panel (n = 2, both had high VAF). One additional case (#044) tested negative in remission by Sanger sequencing only and is not included in this figure. Morphologic relapse is indicated by an “R”. Undetectable IDH2mut level is indicated by “Neg”. (D) Kaplan-Meier plots of OS of patients with NPM1 or IDH2 mutant AML treated with venetoclax in combination with either DNMTi or LDAC compared with patients WT for both NPM1 and IDH2. Four patients had concurrent NPM1 and IDH2 mutations and were included twice. One patient was excluded due to unknown molecular status. (E) Kaplan-Meier plots of OS of patients with IDH1 mutant AML treated with venetoclax in combination with either DNMTi or LDAC compared with patients WT for IDH1. One patient was excluded due to unknown molecular status. (F) Volcano plot of differential gene expression associated with NPM1 mutation in responders: NPM1mut (n = 4) vs NPM1 WT (n = 6). No genes were significantly differentially expressed (DE) with FDR <5%. HOX genes are highlighted in green and apoptosis pathway genes in purple. Gene set enrichment analysis confirmed activation of the NPM1mut signature (using correlation adjusted mean rank gene set analysis [CAMERA], top 3 NPM1mut gene sets, FDR range from 5.17 × 10⁻²¹ to 2.51 × 10⁻²⁷). (G) Volcano plot of differential gene expression associated with NPM1 mutation in TCGA dataset: NPM1mut (n = 35) vs NPM1 WT (n = 115). Gene sets are colored as in panel F, with differentially expressed genes in blue and red. Gene set enrichment analysis confirmed activation of the NPM1mut signature (using CAMERA, top 3 NPM1mut gene sets, FDR range 1.9 × 10⁻³⁴ to 1.1 × 10⁻³⁷).

Figure 3.

Characterization of AML cases with adaptive resistance. (A) Genomic landscape of AML cases with adaptive resistance. Molecular mutations in 25 patients with relapsed AML after an initial response. The presence of adverse cytogenetic risk, complex karyotype, del(17p), indicated mutations, study ID number, relapse-free survival, and cytotoxic therapy received (AZA, azacitidine; DEC, decitabine or LDAC, low-dose cytarabine) are shown for each case. The bar graphs (right side of the plot) summarize the number of cases with persistent, expanded, acquired, or reduced/cleared mutations at relapse. (B-F) Dynamic changes in clonal architecture from diagnosis to relapse in 5 cases with FLT3-ITD. The VAF for each mutation is shown, along with the bone marrow blast count at the corresponding time point. The time elapsed from remission to treatment failure is shown in days. (G-H) Single-cell analysis of clonal architecture at screening and relapse. Mission Bio Tapestri clonograms showing the relative mutation composition (%) of samples at indicated time points for cases #041 (G) and #392 (H). In case #041 (G), the NPM1 and IDH1 parental clone included 2 PTPN11mut subclones, which were extinguished, and a FLT3-ITD subclone, which expanded at relapse. In case #392 (H), the proportion of WT and SRSF2mut-only cells have decreased and been replaced by 4 new clones activating the FLT3 or NRAS kinase pathways.

Figure 4.

Impact of TP53 defects on venetoclax combinations. (A-D) Dynamic changes in clonal architecture from diagnosis to relapse in 4 illustrative cases with TP53 mutations. The VAF of each mutation is shown, along with the bone marrow blast count at the corresponding time point. The time elapsed from remission to treatment failure is shown in days. (E) Interval changes in chromosome 17 and TP53 at diagnosis and relapse. Patterns of clonal evolution of biallelic TP53 abnormalities in 5 AML cases at relapse. Changes in TP53 VAF % were quantitated by targeted NGS. Changes in chromosome 17 were assessed using standard karyotypic techniques. (F) Competitive growth assay comparing the survival of TP53 WT (gray) and TP53 CRISPR/Cas9–deleted (red) in MV4;11 cells during exposure to vehicle, venetoclax 100 nM, cytarabine (ara-C) 500 nM, decitabine 1 µM, venetoclax 100 nM plus low-dose cytarabine (LDAC) 100 nM, or venetoclax 100 nM plus decitabine (DEC) 1 µM in culture over a period of 10 days. The proportion of each cell genotype was indicated by a fluorescent reporter and enumerated by flow cytometry. The total cell viability (%) is shown above each bar. (G-I) Viability of RN2 cells (TP53 WT, p53^WT) or p53^R172H/Δ murine AML cell lines after exposure to venetoclax (G), venetoclax plus azacitidine (1 μM) (H), or venetoclax plus cytarabine (20 nM) (I) for 48 hours. The tables indicate the 50% inhibitory concentration values (nM). Errors represent the standard deviation of 3 technical replicates.

Figure 5.

Characterization of AML cases with primary resistance. (A) Genomic landscape of AML cases with primary resistance. Molecular profile of 20 AMLs that were refractory to venetoclax-based combinations. The presence of adverse cytogenetic risk, complex karyotype, del(17p), indicated mutations, study ID number, and cytotoxic therapy received (AZA, azacitidine; DEC, decitabine or LDAC, low-dose cytarabine) are shown for each case. Mutations detected by RNA-seq are represented by a dark gray box. For case #353, the KIT-ITD increased in frequency from 0.9% to 27% VAF from screening to after cycle 1 assessment. (B) TP53 abnormalities at baseline in refractory AML cases. TP53 VAFs % were quantitated by targeted NGS. (C) Baseline VAF of TP53 mutations according to patients with late relapse (>12 months), early relapse (<12 months), or primary refractory to treatment. AMLs with baseline and acquired del(17p) are indicated by orange and red, respectively. Each patient in a group is represented by a symbol. Some AML cases had >1 TP53 mutation: late relapse (#182 [inverted triangle], #335 [open circle] and #330 [upright triangle]); early relapse (#056 [square]); and refractory disease (#053 [hexagon], #176 [black circle] and #212 [upright triangle]). (D) Single-cell analysis of clonal architecture at screening and after treatment. Mission Bio Tapestri clonogram of case #064 showing the relative mutation composition (%) of samples at patient screening or at the time of refractory disease. In this case, within the EZH2mut clone, 5 parallel RASmut subclones are shown to emerge, with concurrent suppression of EZH2mut-only cells.