AstraZeneca, Daiichi Sankyo Use AI to Understand Phase III NSCLC Failure for Dato-DXd

AstraZeneca and Daiichi Sankyo have introduced an innovative AI-powered biomarker, dubbed the TROP2-QCS (quantitative continuous scoring), to predict clinical outcomes for non-small cell lung cancer (NSCLC) patients being treated with their investigational antibody-drug conjugate, datopotamab deruxtecan (Dato-DXd). This computational biomarker focuses on the normalized membrane ratio (NMR) of the TROP2 protein, which is instrumental in determining patients who would benefit significantly from Dato-DXd.

Through an exploratory analysis of the Phase III TROPION-Lung01 study, the companies found that patients who tested positive for TROP2-QCS experienced a 43% reduction in the risk of disease progression or death when treated with Dato-DXd compared to those receiving docetaxel. This marked a statistically significant difference. The median progression-free survival stood at 6.9 months for the Dato-DXd group, compared to 4.1 months for those on docetaxel.

In contrast, the primary analysis of the overall study population, which did not account for TROP2-QCS, revealed that Dato-DXd only managed to reduce the risk of disease progression or death by 25% when compared to docetaxel.

Ken Takeshita, global head of R&D at Daiichi Sankyo, emphasized that the findings support the potential of TROP2-QCS as a predictive biomarker for Dato-DXd. He highlighted that this new AI biomarker could help explain why certain NSCLC patients respond better to treatments.

The unveiling of TROP2-QCS took place at the International Association for the Study of Lung Cancer’s 2024 World Conference on Lung Cancer. AstraZeneca announced its collaboration with Roche Tissue Diagnostics to further develop and commercialize this AI biomarker.

Unlike most biomarkers that rely on the simple expression levels of specific proteins in cancer cells, TROP2-QCS evaluates the amount of TROP2 expressed on tumor cell membranes relative to its expression in the cytoplasm. Patients are classified as TROP2-QCS-positive if at least 75% of their tumor cells have an NMR below a predefined value, indicating higher TROP2 expression in the cytoplasm.

This biomarker was developed using AstraZeneca's fully automated computational pathology platform, which leverages AI to produce biomarker data beyond the mere presence or absence of a protein. This platform also analyzes signal intensity and its localization within the cell.

AstraZeneca and Daiichi Sankyo utilized QCS to shed light on the underwhelming results of their TROPION-Lung01 study, which in May 2024 revealed that Dato-DXd did not significantly improve overall survival rates for NSCLC patients. At that time, they identified a clinically meaningful improvement in overall survival for patients with non-squamous disease. According to the latest QCS analysis, a greater proportion of patients with non-squamous NSCLC were TROP2-QCS-positive, standing at 66% compared to 44% for those with squamous disease.

The data from the TROPION-Lung01 study has prompted AstraZeneca and Daiichi Sankyo to file for regulatory approval of Dato-DXd for previously treated non-squamous NSCLC patients. The FDA is expected to make a decision in the fourth quarter of this year.