Dual Inhibition of PD-1 and VEGF-A Overcomes Resistance and Enhances T-Cell Infiltration in Non-Small-Cell Lung Cancer
Background: The influx of resistance to single-target immune checkpoint inhibitors and the metabolic data bottleneck in the tumor microenvironment of non-small-cell lung cancer (NSCLC) R&D.
Existing new drug development and clinical R&D for NSCLC have relied on static standard guidelines that inhibit only the PD-1 or PD-L1 single pathway, exposing a critical blind spot: the failure to maintain effective drug concentrations due to vascular endothelial growth factor (VEGF) release and physical extracellular matrix barriers, which are simultaneously induced within the tumor microenvironment (TME). The aberrant structural disruption of tumor tissue and abnormal tumor vasculature limit the physical infiltration of immune effector cells, and inter-species differences and microenvironmental feedback loops that suppress the baseline of anti-cancer immune responses are key data barriers that existing computational omics architectures cannot control. By failing to integrate the dynamic changes in gene expression flux occurring in downstream control pathways after drug administration into a multidimensional matrix, and instead relying solely on single diagnostic markers, it has been difficult to predict early tumor progression due to angiogenesis and metabolic homeostasis disruption, thereby compromising the ability to ensure patient prophylactic concentrations and long-term engraftment.
Discovery: Activation of PD-1/VEGF bispecific antibody modality and demonstration of single-cell resolution tumor-infiltrating lymphocyte tensor synchronization.
The bispecific antibody Ivonescimab (AK112), co-developed by Summit Therapeutics and Akeso, is based on the structural cooperativity that occurs when binding to PD-1, and maximizes the free energy of binding to the VEGF-A molecular target through in silico calculation of the differential equation dynamics model. This platform computationally completely eliminates batch effects on single-cell RNA sequencing and spatial transcriptomics matrices, and demonstrates the synchronization of the spatial distribution of tumor-infiltrating lymphocytes (TILs) before and after drug administration using a multidimensional tensor model. To elucidate the mechanism that disruptively surpasses existing single-target immune checkpoint therapies, the topological variation curves of downstream transcriptomic networks were monitored, and it was demonstrated that vascular normalization induced by VEGF inhibition induces the upregulation of T cell activation markers, thereby proving the molecular biological integrity. This served as the theoretical basis for achieving disruptive prolongation of progression-free survival (PFS) (hazard ratio 0.51, see latest clinical references such as DOI: 10.1016/S0140-6736(24)01026-X) compared to Keytruda (Pembrolizumab) in the Phase 3 HARMONi and HARMONi-2 studies.
Establishment of a model for coordinating tumor vascular-immune interaction pathways and reversibly stratifying metabolic homeostasis.
A computational systems biology framework built on a multidimensional omics matrix establishes a precision stratification model that analyzes patient-specific NSCLC molecular phenotypes and classifies them into high-resolution vascular hyperformation and immune suppression groups. The variability of oxygen partial pressure and lactic acid concentration distribution in the tumor microenvironment is replicated in real-time as a digital twin, and the rate-limiting step constant up/down clamping technique is used to simulate the optimal effective homeostasis backbone that inhibits abnormal angiogenesis while reversibly maintaining immune cell activity within the tumor. In particular, by precisely controlling the kinetics of cell membrane receptor binding, the efficiency of down-regulating the VEGF downstream PI3K-Akt and MAPK signaling pathways induced in drug-resistant cell populations is predicted, and the feedback loop that induces immune suppression is blocked, thereby establishing a computational solution that reversibly preserves tumor killing function even in malignant microenvironmental stress conditions.
Prospects: Establishment of a programmable tumor governance standard and activation of a next-generation IND digital governance.
This computational biology architecture will be the central backbone for completely resetting pharmaceutical R&D governance from a static post-hoc evaluation system to a programmable predictive infrastructure based on multidimensional data tensors. By linking the genetic gradient correction coefficients at the high-throughput screening (HTS) stage in multinational biotech pipelines, it provides a computational moat that eliminates batch effects and inter-batch variations that inevitably occur in multi-institutional clinical trials. Ultimately, this platform is expected to be established as a digital core asset that meets the standards for digital healthcare-based companion diagnostics (CDx), ensures the integrity of high-risk patient population screening models, and disruptively shortens the timelines for clinical trial application (IND) and cGMP commercial launch approval reviews by the US FDA and global regulatory agencies.
Despite recent advances in immunotherapy and targeted therapy, most patients with non-small-cell lung cancer (NSCLC) continue to have disease progression and develop resistance to current treatments. Consequently, there is an urgent need for more effective treatment approaches that provide durable benefit for a greater number of patients, especially for those whose tumours do not harbour actionable genomic alterations. One promising strategy is to simultaneously target multiple mechanisms of immune evasion within the tumour microenvironment.
The PD-1/VEGF-A bispecific in silico validation and clinical demonstration in this study goes beyond theoretical exploration of tumor immune mechanisms and is directly applied to actual global finished drug production hubs and next-generation precision personalized cancer diagnosis and treatment business lines.
First, by instantly scanning the dynamics of angiogenesis and the kinetics of interaction between immune evasion receptors in the tumor microenvironment using a Python-based dynamic system solver algorithm, the temporal noise of existing pathology diagnoses is eliminated at the source, and the unique medical technology moat is maintained by pre-identifying the effective drug response barriers of each patient.
At the same time, by linking to the open-source TCGA database, which aggregates patient genomic multidimensional omics matrices, a companion diagnostic (CDx) panel interface is realized that virtually simulates genetic confounding variables that distort drug metabolism pathways during clinical trial design and real-time reverse-calculates the effective docking concentration of PD-1 and VEGF receptors.
Furthermore, when multinational companies conduct large-scale approval clinical trials for next-generation bispecific cancer immunotherapy drugs, by linking indicators such as tumor vascular normalization index in the tumor as correction coefficients, batch-to-batch efficacy evaluation variations are eliminated, and the backbone infrastructure that maximizes the probability of obtaining global regulatory approval for clinical trial applications and cGMP commercial launch approvals is provided.