A new multi-institutional study published in the Journal for ImmunoTherapy of Cancer suggests that an artificial intelligence-based imaging biomarker could help oncologists predict which patients with Non-Small Cell Lung Cancer (NSCLC) are likely to benefit from immunotherapy.

The research, conducted by Picture Health, analysed more than 1,300 CT scans from 682 NSCLC patients treated across six medical centres. The study evaluated a novel AI-driven biomarker known as Quantitative Vessel Tortuosity (QVT) score, which measures the structure and complexity of blood vessels surrounding tumours.

Using routine CT scans, the QVT score analyses over 900 measurements related to tumour blood vessel patterns. These measurements help identify abnormalities in the vascular networks that tumours create to sustain their growth. Researchers found that the biomarker could predict patient survival outcomes and identify early signs of treatment response, potentially earlier than traditional imaging methods.

Immunotherapy has significantly improved treatment options for lung cancer, but not all patients respond equally. Existing tests, such as PD-L1 testing, focus mainly on immune-related tumour markers. However, these tests often overlook the role of tumour blood vessel networks, which can influence how effectively immune cells and treatments reach cancer cells.

The study found that the QVT score recorded at the start of treatment independently predicted patient survival. In addition, early decreases in QVT scores during therapy indicated normalisation of tumour blood vessels and were detected before conventional methods showed measurable tumour shrinkage.

Researchers believe that the technology could help clinicians personalise treatment strategies and make earlier decisions about therapy effectiveness. By analysing routine CT scans with AI, oncologists may gain deeper insights into tumour biology without requiring additional tests.

The findings also come at a time when pharmaceutical companies are increasingly exploring therapies that combine immunotherapy with treatments targeting tumour blood vessels. As a biomarker in clinical trials, QVT may help identify patients who are most likely to benefit from specific treatment approaches.

Although the current research focuses on lung cancer, abnormal tumour blood vessel patterns are common across several cancer types. Experts suggest that AI-based imaging tools like QVT could eventually support treatment decision-making across multiple cancers.