Novel Model Integrating TME Composition with TLS Gene Signatures Predicts Meaningful Survival Differences in Checkpoint Inhibitor-Treated Patients
Cellworks Group Inc., a leader in Personalized Therapy Decision Support and Best-in-Class PTRS, today announced results from a new study showing that the combination of tumor microenvironment (TME) composition and tertiary lymphoid structure (TLS) dynamics is a key predictor of how individual patients with non-small cell lung cancer (NSCLC) respond to immunotherapy. The study introduces a novel prediction model that integrates TME cell composition with a 34-gene TLS score. Together, these measures enabled the Cellworks Platform to accurately predict patient-level survival outcomes and reveal meaningful differences among NSCLC patients treated with checkpoint inhibitors.
Results from the study were showcased in a poster presentation titled, "Cellular Heterogeneity and Tertiary Lymphoid Structure Dynamics Predict Overall Survival in Immune Checkpoint Therapy-Treated NSCLC Patients," as part of the IASLC 2025 World Conference on Lung Cancer (WCLC25) hosted by the International Association for the Study of Lung Cancer held from September 6-9, 2025 in Barcelona, Spain.
"Despite the promise of immune checkpoint inhibitors (ICIs), only a subset of NSCLC patients benefit from these therapies," said Charu Aggarwal, MD, MPH, FASCO, Leslye M. Heisler Professor of Lung Cancer Excellence in the Perelman School of Medicine at the University of Pennsylvania, and co-author on the study. "This study provides new insights into how the structural organization of tertiary lymphoid structures and the immune cell composition of the tumor microenvironment jointly determine immunotherapy response. By moving beyond single biomarkers, this approach holds promise as a more comprehensive way to guide personalized treatment decisions in NSCLC."
"Our findings highlight that both cellular heterogeneity and TLS dynamics play critical roles in determining whether patients respond to checkpoint inhibitor therapy," said James Wingrove, PhD, Chief Development Officer at Cellworks and presenting author of the study. "By integrating these factors, we created a personalized model that can offer oncologists new insight into which patients are most likely to benefit from checkpoint inhibitors. This study underscores how computational modeling of the tumor microenvironment can advance personalized decision support in NSCLC."
"What makes this work exciting is the ability to connect molecular signals within the tumor microenvironment to real patient outcomes," said Michael Castro, MD, Chief Medical Officer at Cellworks. "By biosimulating how both cellular heterogeneity and TLS dynamics shape immunotherapy response, we move closer to a future where treatment selection is not just based on broad population markers, but on each patient's unique tumor biology. This level of personalization has the potential to identify which patients benefit from combination chemotherapy and immunotherapy while also sparing other patients from receiving chemotherapy when it is unlikely to benefit."
Key Findings
- Strong Predictive Value The integrated TLS TME model demonstrated high predictive significance for overall survival in both the training (HR=0.36, C-Index=0.768) and validation cohorts (HR=0.66, C-Index=0.628).
- Clear Survival Differences Patients predicted to have a high benefit from immune checkpoint inhibition showed a significant increase in overall survival, living a median of 30.8 months versus 12 months for patients predicted to have low benefit.
- Immune Balance Matters Survival benefit was tied to immune balance: pro-inflammatory environments enhanced TLS benefit, while suppressive immune cells (like neutrophils and M2-like macrophages) reduced or reversed it.
- New Insights for Personalization The integration of TLS dynamics with TME immune and stromal cell composition provided independent yet complementary insights, showing how structural organization and cellular balance cooperatively determine immunotherapy efficacy and can guide personalized treatment decisions.
Study Design
Cellworks developed and cross-validated an algorithm that deconvolutes bulk transcriptomic data to estimate cell proportions and cell-type-specific gene expression within the TME. This approach was enhanced with a TLS Score based on 34 genes representing cellular interplay and maturity, derived from bulk RNA-sequencing data of tumor samples. By integrating TLS dynamics with immune and stromal cell populations, the model captured complementary and independent contributions, providing collective insight into how tumor structure and cellular composition determine ICI efficacy. A Cox proportional hazards model was trained on advanced NSCLC patients treated with ICIs (n=63, SU2C-MARK cohort) and validated in an independent cohort of 66 patients from the same study. The locked model confirmed predictive performance at the individual patient level, underscoring its potential clinical utility.
The Cellworks Platform
The Cellworks Platform performs computational biosimulation of protein-protein interactions, enabling in silico modeling of tumor behavior using genomic data derived from next-generation sequencing (NGS). This allows for the evaluation of how personalized treatment strategies interact with the patient's unique tumor network. Multi-omic data from an individual patient or cohort is used as input to the in silico Cellworks Computational Biology Model (CBM) to generate a personalized or cohort-specific disease model.
The CBM is a highly curated mechanistic network of 6,000+ human genes, 30,000+ molecular species and over 600,000 molecular interactions. This model along with associated drug models are used to biosimulate the impact of specific compounds or combinations of drugs on the patient or cohort and produce therapy response predictions, which are statistically modeled to produce a qualitative therapy response score for a specific therapy. The Cellworks CBM has been tested and applied against various clinical datasets with results provided in more than 125 presentations and publications with global collaborators.
About Cellworks Group
Cellworks Group, Inc. is dedicated to improving patient outcomes by harnessing the power of computational science to deliver Personalized Therapy Decision Support and Best-in-Class PTRS solutions. The Cellworks Platform predicts patient-specific therapy response for oncology and other serious diseases using a breakthrough Computational Biology Model (CBM) and biosimulation technology. Cellworks is backed by Artiman Ventures, Bering Capital, Sequoia Capital, UnitedHealth Group and Agilent Ventures. Headquartered in South San Francisco, the company also operates a CLIA-certified computational lab in Franklin, Tennessee. Learn more at www.cellworks.life.
All trademarks and registered trademarks in this document are the properties of their respective owners.
View source version on businesswire.com: https://www.businesswire.com/news/home/20250908354231/en/
Contacts:
Media Contact:
Barbara Reichert, Reichert Communications, LLC
barbara@reichertcom.com
650-548-1002
