I’ve been talking about trying to model human brain tumours using human rather than mouse cells but injecting those human cells into a mouse recipient. We were inspired by the idea that you could take normal neural stem cells that we published a few years ago, you could take normal neural stem cells from a mouse, introduce the genetic events that lead to medulloblastoma and use that to drive the disease. We wondered if we could take normal neural stem cells from a person and do the same experiment but there are ethical issues associated with taking normal neural stem cells from a person so we chose instead to use an induced pluripotent stem cell, make that into a neural stem cell and then use that to model medulloblastoma.

What have you found so far?

That we can generate human-based models for medulloblastoma through this methodology; that we can use those models to weed through the vast number of genes that have been identified as mutant to identify which ones are functionally important.

We’ve also been able to generate mouse models for some of the more therapy-resistant human medulloblastoma subtypes and we can then use the material we’ve generated to do screens.

One of the fundamental problems in paediatric cancer is that the wealth of genomics has led to taking relatively rare cancers and subdividing them into even rarer populations. So there are 90 medulloblastomas in the UK every year and you start subdividing those into twenty groups and you think about powering a clinical trial enough to get enough patients to do something meaningful and it becomes a challenge. But funding aside, just getting enough patients with a very defined subtype… so having mouse models that recapitulate that is a start to get a therapy that we can then test in that small number of patients.