5636 - Identification of Prognostic and Predictive Factors for Durvalumab Efficacy by Modeling of Tumor Response and Overall Survival (OS) in Patients with Non–Small-Cell Lung Cancer (NSCLC)

Background

Durvalumab, a human anti–PD-L1 mAb, is currently approved for treatment of patients with Stage III unresectable NSCLC. The objectives of this analysis were to identify prognostic and predictive factors for tumor growth and shrinkage, as well as for OS in NSCLC patients treated with durvalumab.

Methods

Longitudinal tumor size (TS) and OS data obtained from NSCLC patients in Study 1108 (all comers) and ATLANTIC (Stage III and above) who received durvalumab were analyzed using a nonlinear mixed effect model that describes the growth and regression of sensitive and insensitive tumor cells, as well as delay in immune response leading to tumor killing. A linked OS-dropout model was developed by relating model-predicted tumor changes to OS and dropout probability over time. Potential prognostic and predictive factors were evaluated in a multivariate covariate analysis using the models.

Results

The longitudinal TS and OS data from NSCLC patients in both studies are generally well described by the models. Liver metastasis, neutrophil-to-lymphocyte ratio (NLR), EGFR mutation, and durvalumab clearance (CL) are identified as prognostic factors for tumor growth, and tumor cell PD-L1 expression (TC) and baseline tumor size as predictive factors for tumor killing (p < 0.01). The significant factors for OS after accounting for the tumor size changes included TC and immune cell PD-L1 expression (IC), NLR, lactate dehydrogenase, as well as CL (p < 0.01). Among all factors tested, NLR is the most influential factor on the predicted 1-year survival rates (∼60% vs. 30% with NLR below and above the median [4.56]). Positive PD-L1 expression (TC or IC ≥ 25%) is predicted to result in ∼10-20% increase in one-year survival rates. Increasing the cutoff value is not predicted to result in substantially greater improvement in the survival rate.

Conclusions

The modeling results provided quantitative assessments of the impact of various prognostic and predictive factors, as well as biomarker cutoff values on the efficacy of durvalumab in NSCLC patients, and can be used to inform patient selection criteria in future monotherapy or combination studies.

Clinical trial identification

NCT01693562.

Legal entity responsible for the study

MedImmune, LLC.

Funding

MedImmune.

Editorial Acknowledgement

Disclosure

Y. Zheng, R. Narwal, C. Jin, P. Baverel, A. Gupta, B.W. Higgs, L. Roskos: Employee: MedImmune; Stock and/or stock interests or options: AstraZeneca. P. Mukhopadhyay: Employee, Stock and/or stock interests or options: AstraZeneca.