Temporal multiomic profiling reveals intrinsic resistance mechanisms to ERK inhibition in relapsed Acute Myeloid Leukemia

Abstract Background: The RAS/MAPK signaling pathway is hyperactivated in acute myeloid leukemia (AML). RAS/MAPK signaling converges on the extracellular signal-regulated kinase (ERK) which in turn promotes the transcription of immediate-early genes (IEGs) that mediate blast proliferation and survival. Temuterkib is an orally bioavailable small molecule inhibitor of ERK1/2. In preclinical studies, temuterkib downregulated ERK targets in AML cell lines and caused growth arrest (Weisberg et al., 2020), but itseffects in human leukemia in vivo have not been defined. We therefore performed multiomic studies in primary AML blasts from patients treated with temuterkib to characterize the acute and adaptive responses to ERK inhibition and identify potential mechanisms of intrinsic resistance. Methods We performed a phase 1 dose escalation trial of temuterkib in patients with R/R AML using continuous daily dosing (NCT04081259). We evaluated bone marrow blast response and pre/post-treatment genetics using DNA sequencing. Blood samples were collected pre-dose; 1, 4, 8, and 24 hours after first dose, and weekly thereafter. Blasts were purified for bulk and single-cell RNA-seq with CITE-seq targeting 330 surface and intracellular proteins and 28 phospho-proteins to assess transcriptional and signaling changes over time. Results We enrolled 17 patients over 3 dose levels (3 at 200mg, 3 at 100mg with concurrent azole, 11 at 400mg). There were no dose-limiting toxicities. Ten patients had TP53 mutations, 7 had AML-MR mutations and 9 had RAS pathway mutations (FLT3, NRAS, KRAS, CBL, PTPN11). Among 15 patients evaluable for response, 7 had progressive disease (PD) and 8 had stable disease (SD) per IWG criteria. Median time on study was 32 days (range 15-121) with disease progression as the primary reason for discontinuation. Among patients treated with 200mg and 400mg, temuterkib reached peak concentrations by 4 hours and showed on-target inhibition, reflected by decreased ERK target gene expression. We evaluated baseline DNA and RNA-sequencing to define pre-treatment differences between patients. Baseline genotype (TP53, AML-MR, RAS mutations) did not correlate with measures of clinical activity, including PD vs. SD or time on study. We compared RAS mutant (n=9) vs. WT (n=8) leukemias and identified 300 differentially expressed genes (>1.5 fold change, padj <0.05). Sixty-two genes with greater expression in RAS mutant leukemias showed enrichment for peptide biosynthetic processes (GO:0043043) and TORC1 signaling (GO:0038242). We performed serial DNA and RNA-seq to assess genetic and transcriptional responses to ERK inhibition despite lack of clinical response. Paired pre/post treatment NGS showed no subclonal expansion or contraction, including of RAS mutations. Due to high inter-patient heterogeneity at baseline, we analyzed longitudinal changes in global transcription in each patient separately. ERK dependent IEGs (EGR1, JUN, and FOS) and negative feedback regulators (DUSP1, DUSP6, and SPRY1) were among the most downregulated transcripts at 4 and 8 hours across all patients. However, in RAS mutant leukemias, this effect was dampened with a lower magnitude of EGR1 (mean log2fc -0.9 vs. -2.2; p=0.07) and DUSP6 decrease (mean log2fc -0.5 vs. -1.2; p=0.04) at 4 hours. MYC expression decreased by 20% at 4 hours post dose, returned to baseline by 8 hours, and increased to 54% greater than baseline by day 15. This rebound was independent of RAS mutation status. To compare the acute and adaptive response to ERK inhibition, we performed longitudinal single-cell RNA-seq with CITE-seq on samples from 2 patients treated at 400mg. In a patient with NRAS and TP53 mutations, we found an acute increase in JAK-STAT signaling proteins (pSTAT5A, pSTAT5B) at 4 and 8 hours, followed by a sustained increase above baseline at day 15. In another patient with FLT3-TKD/ITD and RUNX1 mutations we observed an acute increase in MTOR signaling proteins (AKT1, pMTOR, pRPS6KA1). These findings indicate that compensatory activation of alternative signaling pathways occurs rapidly in response to ERK inhibition.Conclusions: Temuterkib produced acute on-target ERK suppression without objective clinical responses in R/R AML regardless of RAS mutation status. RAS mutations did not confer increased sensitivity to ERK inhibition and may instead attenuate downstream suppression. Rapid induction of JAK-STAT and MTOR signaling may underlie resistance to ERK inhibition in AML.