Jun. 20 at 12:50 PM
$ACTU Actuate reports biomarker, machine learning data from Phase 2 elraglusib trial
Actuate Therapeutics announced biomarker data from a recent poster presentation at the American Society of Clinical Oncology annual meeting from the Phase 2 trial of elraglusib in combination with gemcitabine/nab-paclitaxel in first-line metastatic pancreatic adenocarcinoma. The study demonstrated the use of machine learning and statistical models to predict overall survival based on pre-dose plasma biomarkers.
The poster described an analysis of 40 cytokines, chemokines, soluble cell receptors, and growth factors from plasma samples obtained prior to treatment from all patients enrolled in Actuate-1801 Part 3B. The analysis identified 7 biomarkers as uniquely significant predictors of favorable survival in the elraglusib-treated cohort, including one, CXCL2, with an inverse survival trend compared to the GnP control arm.
This indicates that in patients not treated with elraglusib, high CXCL2 was unfavorable, but this trend is reversed with elraglusib treatment, highlighting the potential of elraglusib to favorably affect the immune microenvironment. Univariate analysis revealed strong predictive performance with CXCL2 emerging as a consistently reliable biomarker for survival across multiple cross-validation analyses.
Elevated levels of CXCL2 and TRAIL were associated with improved OS, while lower levels of CCL3, IL-1alpha, IL-18, TGF-beta, and TRAIL R3 were similarly linked to better survival. These signatures were combined into multivariate machine learning models that accurately predicted patients who would survive for greater than one year if treated with elraglusib and GnP.
The company plans to test the identified biomarkers prospectively in future trials. Additional efforts will focus on optimizing sequential univariate combinations for patient stratification, refining multivariate machine learning models for predictive accuracy, and comparing these approaches head-to-head.
