Chee J, Principe N, Chin M, Cook A, Gray E, Nowak A
Funding: WA DOH Future Health and Research Innovation Fund
Lay Synopsis: Checkpoint inhibitor immunotherapy can produce long-lasting benefits in some cancer patients, but many people don’t respond or develop serious side effects, so we need better ways to predict outcomes before treatment. We will use bulk and single-cell T cell receptor sequencing plus machine learning (e.g., DeepTCR, immuneML) to identify immune “fingerprints” that predict response and immune-related adverse events across cancers such as mesothelioma, lung cancer, and melanoma.
Scientific Synopsis: Immunotherapy with checkpoint inhibitors can lead to durable responses in some cancer patients and is being tested across multiple cancer types. However, only a minority of patients benefit. Because these treatments are expensive and can cause serious side effects, it is important to predict which patients are most likely to respond or develop side effects.
Checkpoint inhibitors work by strengthening the patient’s immune response, particularly T cells, to recognise and destroy tumour cells. Each person has a unique “fingerprint” of T cell antigen receptors (T cell receptors, TCRs) that shapes how their immune system responds to cancer.
We believe differences in these receptors help explain why immunotherapy works exceptionally well for some patients but not others. Using bulk and single-cell TCR sequencing, we can measure millions of TCRs and track how individual clonotypes expand, persist, or disappear during treatment. We will then apply machine learning and specificity-discovery tools, including DeepTCR and immuneML, alongside motif and clustering approaches such as GLIPH2 and tcrdist3 to identify TCR patterns linked to treatment response or resistance, and integrate these findings with single-cell transcriptomic state.
We will study TCR repertoires in patients with mesothelioma, lung cancer, and melanoma receiving different combination immunotherapy regimens, and test whether these ML-derived immune signatures predict clinical outcomes such as response and immune related adverse events. The results will support development of robust biomarkers to improve patient selection and guide immunotherapy decisions across cancers.