Article Text

Download PDFPDF

801 Risk-benefit ratio in immunotherapy (IO) with or without chemotherapy (CT) according to PD-L1 expression: a model-based approach to inform treatment decisions in metastatic non-small cell lung cancer
  1. Molly Zhao1,
  2. Naveen Mangal1,
  3. Matthew L Zierhut2,
  4. Han Witjes2,
  5. Ting Chen1,
  6. Thao Dang1,
  7. Ana Ruiz1,
  8. Sandhya Girish1 and
  9. Sreeneeranj Kasichayanula1
  1. 1Gilead Sciences, Inc., Foster City, CA, USA
  2. 2Certara USA, Inc., Princeton, NJ, USA
  • Journal for ImmunoTherapy of Cancer (JITC) preprint. The copyright holder for this preprint are the authors/funders, who have granted JITC permission to display the preprint. All rights reserved. No reuse allowed without permission.


Background Anti-programmed death-(ligand)1 (PD-[L]1) therapies, either as monotherapy or in combination, have demonstrated remarkable efficacy in first-line (1L) treatment for metastatic non-small cell lung cancer (mNSCLC). Treatment decisions are often influenced by multiple factors, e.g., PD-L1 expression level or adverse event (AE) profile. Model-based meta-analyses (MBMA) of safety and efficacy can evaluate clinical risk-benefit ratio among FDA-approved IO therapies and inform treatment decisions in mNSCLC patients.

Methods MBMA of median overall survival (mOS) and treatment-related AEs Grade 3–5 (TRAEs Gr3+) were conducted using a curated database containing results from both randomized controlled and single-arm studies (published from 2015 to 2022) with FDA-approved (as of January 2023) IO-based treatments in mNSCLC patients. The analysis dataset comprised mOS and TRAEs Gr3+ results from 49 and 45 studies, and represented 17,918 and 19,238 patients, respectively. The mOS dataset included 34, 28, and 11 studies containing treatment of IO monotherapy, CT, and IO+CT, respectively. In the TRAE analysis, there were 33, 23, and 12 studies containing treatment of IO monotherapy, CT, and IO+CT, respectively. Analyses of endpoints utilized linear models in appropriately transformed domains (i.e., log for mOS, logit for TRAE Gr3+) and data were weighted by standard error of the outcome. After accounting for differential outcomes from specific treatments, potential impact of baseline characteristics (e.g., PD-L1 expression, line of therapy, ECOG PS, sex, histology, and geography/race) on the risk and benefit was evaluated. Models included between-study random effects and all covariates were additive.

Results Regardless of specific IO agent or PD-L1 expression level, fewer patients experienced TRAEs Gr3+ with IO alone (14% [6.2–29]) compared to CT alone (39% [21–61]) or IO+CT (48% [27–70]). Among patients with PD-L1 expression ³50%, mOS was comparable for IO with or without CT (figure 1A,B). In patients with PD-L1 expression <1% (figure 1C,D), therapeutic benefit was higher for IO+CT (16 months [13–19]) than IO alone (12 months [10–15]), justifying the choice of IO+CT in this patient population despite higher TRAEs than IO alone. table 1 summarizes all model parameter estimates, including influential covariates on mOS. No clinically meaningful covariates were identified for TRAEs Gr3+.

Conclusions This MBMA evaluates the risk-benefit ratio and provides quantitative insight into the therapeutic utility of IO agents. Our findings are consistent with treatment guidelines for 1L patients with mNSCLC, highlighting the potential utility of a quantitative model-based approach to strengthen clinical recommendations and to facilitate decision making.

Abstract 801 Figure 1

Model predicated mOS and TRAE Gr3+ for first-line, non-Asian, mNSCLC patient population

Abstract 801 Table 1

Model Parameter Estimates with 95% Confidence Intervals

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.