Glioblastoma is the most common and most aggressive adult brain cancer and has resisted decades of therapeutic effort, in part because it offers the immune system few distinctive targets. The LKS Faculty of Medicine at HKU, through its Clinical Neuroscience Consortium established jointly with Queen Mary Hospital, worked with the Hong Kong Genome Institute to look for those targets at the level of transcript isoforms rather than genes.
The team applied long-read single-cell sequencing to data from more than 210,000 cells drawn from 27 glioblastoma patients, producing what is described as the most comprehensive map of isoform diversity in the disease to date. Crucially the map spans tumour, immune and stromal cells, so transcripts can be attributed to a cell population rather than to bulk tissue in which tumour and microenvironment signals are mixed together.
Thousands of previously unannotated isoforms emerged, including many found only in tumour cells and absent from healthy tissue. The team predicted that a subset of these bind strongly to MHC class I molecules, which is the requirement for a peptide to be presented to cytotoxic T cells and therefore to serve as a neoantigen. The findings were published in Nature Communications.
The release is explicit that this does not change current treatment options, and prediction of MHC binding is not the same as demonstrated immunogenicity. What the work provides is a candidate list for personalised cancer vaccines in a tumour type where the shortage of candidates has been a central obstacle. It also illustrates how much tumour-specific antigen may be hiding in alternative splicing rather than in mutation.