Immunotherapy and Targeted Therapy: New Data from ASCO 2026

Immunotherapy has transformed cancer treatment, yet a substantial proportion of patients do not achieve a clinical response. This necessitates the exploration of alternative strategies, including novel immuno-oncology targets and combination regimens, to overcome resistance to standard therapies.^1,2,3

What the studies did

A multimodal graph neural network system, Mining Immunotherapy Drug tArgetS (MIDAS), was developed for immuno-oncology target discovery. MIDAS integrates gene interactions, multi-omic patient profiles, immune cell biology, antigen processing, disease associations, and phenotypic consequences of genetic perturbations.¹ The system demonstrated generalizability to time-sliced data, outperforming existing baselines like OpenTargets, and accurately ranked approved targets above those in clinical development.¹ MIDAS also identified immunotherapy-response-associated genes in unseen patients, indicating its ability to capture determinants of immunotherapy response.¹ Interpretability analyses revealed MIDAS's reliance on autoimmunity, regulatory networks, and immuno-oncology pathways.¹ Functional perturbation of oncostatin M-oncostatin M receptor signaling, a target proposed by MIDAS, in TRACERx melanoma-patient-derived explants, resulted in reduced dysfunctional CD8+ T cells, which are associated with immunotherapy response, and decreased CCL4 levels.¹ Oncostatin M and oncostatin M receptor expression correlated with altered T cell and macrophage profiles in bulk transcriptomic data from patient samples, consistent with a role in modulating the tumor microenvironment towards immunosuppressive, tumor-promoting phenotypes.¹

In advanced gastric or gastroesophageal junction (GEJ) adenocarcinoma, a post-hoc analysis of the RATIONALE-305 study investigated tislelizumab plus chemotherapy versus placebo plus chemotherapy as first-line treatment.² This analysis included patients with or without peritoneal metastases.²

For patients with EGFR-mutated advanced non-small cell lung cancer (NSCLC) who progressed on first-line osimertinib, the ORCHARD study evaluated the combination of osimertinib plus selumetinib in those with BRAF alterations.³

Key Findings

The MIDAS framework presents a machine learning approach for analyzing multimodal data to discover immuno-oncology targets.¹

In the RATIONALE-305 post-hoc analysis, tislelizumab plus chemotherapy demonstrated a median overall survival (OS) of 15.0 months (95% CI, 13.0-16.9) compared to 12.9 months (95% CI, 11.1-14.3) for placebo plus chemotherapy (HR, 0.77; 95% CI, 0.65-0.92; p=0.004).² The objective response rate (ORR) was 46.3% (95% CI, 41.6-51.0) with tislelizumab plus chemotherapy versus 35.0% (95% CI, 30.5-39.7) with placebo plus chemotherapy.² The median progression-free survival (PFS) was 6.9 months (95% CI, 6.1-7.9) versus 5.6 months (95% CI, 5.4-6.0) (HR, 0.70; 95% CI, 0.59-0.83; p<0.001).²

The ORCHARD study explored osimertinib plus selumetinib in EGFR-mutated advanced NSCLC patients with BRAF alterations after progression on first-line osimertinib.³ Specific efficacy data for this combination in this subgroup were not detailed in the provided abstract, but the study design addresses a critical unmet need in resistance mechanisms to EGFR TKIs.³

Limitations & Next Steps

The MIDAS study is a computational framework validated with patient-derived explants; further clinical validation in human trials is required to confirm the efficacy of targeting oncostatin M-oncostatin M receptor signaling.¹ The RATIONALE-305 data on gastric/GEJ adenocarcinoma is a post-hoc analysis, which may be subject to selection bias and confounding, and prospective studies are needed to confirm these findings in specific subgroups, such as those with peritoneal metastases.² The ORCHARD study abstract did not provide specific outcome data, limiting the ability to assess the clinical utility of osimertinib plus selumetinib in the described patient population.³