Safety, Efficacy and Biomarker Analysis of a Phase 1b/2 Study of Onvansertib (ONV), a Polo-like Kinase 1 (PLK1) Inhibitor, in Combination with Low-Dose Cytarabine (LDAC) or Decitabine (DEC) in Patients (pts) with Relapsed/Refractory Acute Myeloid Leukemia (R/R AML)

Background: PLK1 is a serine/threonine kinase and master regulator of G2/M cell-cycle progression. PLK1 is overexpressed in numerous cancer types including AML and inhibition induces mitotic arrest and cell death. ONV (PCM-075) is an orally active, highly selective PLK1 inhibitor with significant activity in preclinical AML models as a single agent and in combination with cytarabine. Here we provide an updated analysis of the ongoing Phase 1b dose-escalation trial (NCT03303339), evaluating the safety, efficacy and correlative analyses of ONV in combination with either LDAC or DEC in R/R AML.

Methods: R/R AML pts are treated with ONV for 5 days in combination with LDAC (20 mg/m² SC qd x 10d) or DEC (20 mg/m² IV qd x 5d) within a 21 to 28-day cycle. Dose escalation is in 50% increments, and dose limiting toxicities (DLTs) are evaluated at the end of cycle 1. Efficacy is evaluated using modified International Working Group 2003 criteria. We evaluated PLK1 inhibition and gene expression changes with treatment. Based on preclinical studies, PLK1 activity can be monitored in vivo through changes in phosphorylation of its direct substrate: the translationally controlled tumor protein (TCTP). Blood samples are collected on day 1 at pre-dose and 3h post-dose (corresponding to ONV ≈ C_max), pTCTP and TCTP are assessed by capillary Western-Blot in isolated peripheral mononuclear cells and biomarker positivity is defined as ≥ 35% decrease in pTCTP/TCTP at 3h versus pre-dose. In addition, blood samples are collected on day 1 at pre-dose, 3h and 24h post-dose for gene expression analysis (RNA-seq).

Results: As of 22Jul2019, 37 pts were treated (20 DEC, 17 LDAC) with the ONV dose escalated in 3-subject cohorts from 12, 18, 27, 40, 60 to an ongoing 90mg/m² cohort. No DLTs or drug-related deaths were reported. Adverse events (AEs, all grades) reported in ≥10% pts (n=37) included anemia (11 pts), fatigue (11), thrombocytopenia (10), febrile neutropenia (9), neutropenia (9), dyspnea (8) and nausea (8). Grade 3 and 4 AEs possibly related to ONV were hematological AEs (12), elevated bilirubin (1) and stomatitis (1). Preliminary efficacy for the 30 evaluable patients in the Phase 1b showed 4 pts achieving objective responses: Complete response (CR,n=2) or CR with incomplete count recovery (CRi,n=1) in the DEC arm (27 and 40mg/m²) and CR with incomplete platelet count recovery (CRp,n=1) in the LDAC arm (40mg/m²). Among responders, 3 pts remain on treatment with ongoing responses for ≥100 days. Fourteen (45%) of the 31 evaluable pts in the Phase 1b were biomarker positive (achieved a ≥ 35% reduction in pTCTP upon treatment). pTCTP inhibition was not correlated with ONV dose level, individual pts pharmacokinetics (AUC_0-24), or % circulating blasts. However, response to treatment was significantly associated with biomarker positivity with all 4 responding pts being in the biomarker positive group (n=11). In contrast, no responses were observed in the biomarker negative group (n=15) (p=0.02). Whole transcriptomic comparison between biomarker negative (n=11) and positive pts (n=11) indicated a significant enrichment for increased expression of oxidative phosphorylation (OXPHOS) hallmark genes within the biomarker positive pts (p = 0.009). Interestingly, OXPHOS gene expression was downregulated at 3h and 24h after treatment in the biomarker positive but not the biomarker negative group. Finally, to determine whether the underlying tumor biology that is classified as biomarker positive is associated with pt outcome, a predictive gene expression signature for biomarker positive pts was developed and subsequently tested as a prognostic for R/R pts from the BeatAML dataset; this predictive signature identified patients with decreased overall survival (p=0.0005).

Conclusion: In the Phase 1b, the combination of ONV with either LDAC or DEC demonstrated safety, tolerability and preliminary efficacy. Biomarker positivity was observed in a subset of pts, possibly related to dependency of the tumor on PLK1 activity, and was associated with objective response. In addition, gene expression data suggest that biomarker positive pts have a worse prognosis and an underlying OXPHOS-dependency of the tumor. The OXPHOS-dependent AML population has been shown to be resistant to chemotherapy (Farge et al., Cancer Discov. 2017 Jul 1;7(7):716-35 ), therefore this hard-to treat population could potentially benefit from the addition of ONV.