Comparative Effectiveness of Glasdegib or Venetoclax in Combination with Low-Dose Cytarabine Using Simulated Treatment Comparisons

Background: Older patients with acute myeloid leukemia (AML) who are unfit for intensive chemotherapy have a poor prognosis, further encumbered by the lack of well-tolerated standard therapies. Combinations of glasdegib with low-dose cytarabine (GLAS+LDAC) and venetoclax with LDAC (VEN+LDAC) are non-intensive chemotherapy (NIC) options recommended by the National Cancer Comprehensive Network Guidelines. As these regimens have not been compared in a head-to-head trial, indirect treatment comparisons (ITC) are of clinical relevance to assess their comparative effectiveness. Hazard ratios (HR) for overall survival (OS) estimate relative survival valued equally throughout follow-up, unlike landmark measures of OS, which may overvalue or undervalue survival at a single time point. This simulated treatment comparison (STC) study compared OS HRs between GLAS+LDAC and VEN+LDAC among older patients with AML unfit for IC.

Methods: This analysis utilized individual AML patient data (IPD) from the BRIGHT AML 1003 GLAS+LDAC trial (N=116) and published summary data extracted from the Phase III VEN+LDAC trial (N=211). STC was used following Decision Support Unit Guidelines from the National Institute for Health and Care Excellence. Parametric time-to-event regression models for OS were generated using the GLAS+LDAC IPD; optimal exponential models are presented based on fit statistics. OS HRs were estimated after adjusting for baseline patient characteristics of the GLAS+LDAC trial to match the VEN+LDAC trial. Full and backwards stepwise models (retention p-value <0.2) were constructed considering all mutually available patient covariates including age, sex, de novo vs secondary AML, bone marrow blast >50%, Eastern Cooperative Oncology Group performance status, and cytogenetic risk. OS HRs were indirectly compared by ITC (unadjusted for covariates) and STC (adjusted) with 95% confidence intervals (CIs). ITC and STC used post-hoc analysis results for VEN+LDAC; a sensitivity analysis of pre-specified analysis results for VEN+LDAC was conducted.

Results: Overlapping Kaplan Meier (KM) curves for GLAS+LDAC and VEN+LDAC (Figure 1A) suggest that survival probability was similar throughout follow-up. Comparisons of crude survival rates, however, do not account for differences in patient characteristics between studies. Patients in the GLAS+LDAC trial were more likely to be male, have secondary AML, and had worse cytogenetic risk profiles than the VEN+LDAC trial.

Unadjusted ITC estimated that GLAS+LDAC numerically (but not statistically) favored over VEN+LDAC, with a 34% (HR: 0.66; 95%CI: 0.38, 1.15) reduction in mortality (Figure 1B). Similarly, following STC with full covariate adjustment, the estimated comparative advantage of GLAS+LDAC over VEN+LDAC widened to 44% (HR: 0.56; 95%CI: 0.24, 1.31). Results of the sensitivity analyses were consistent with these findings.

Conclusion: This study indirectly compared OS HRs of GLAS+LDAC and VEN+LDAC in older patients with AML unfit for intensive chemotherapy. In the absence of a head-to-head trial, a well-conducted STC provides the best adjusted estimate of comparative effectiveness between treatments. Unadjusted ITC and adjusted STC revealed that the OS HR numerically favored GLAS+LDAC over VEN+LDAC, albeit not significantly. These findings suggest that the decision between these two recommended NIC regimens may not be based solely on differences in survival outcomes. Rather, the safety profile, the burden of administration, and patient preference will need to be factored into treatment decisions.

Figure 1. KM and ITC/STC Results