35IN - How can cancer risk and genetic prediction models help oncologists in the clinics?

Date 29 September 2012
Event ESMO Congress 2012
Session How can medical oncologists deal with the new wave of genetic information about their patients?
Topics Familial Cancer
Personalised Medicine
Presenter Gareth Evans
Authors G.R. Evans
  • St Mary's Hospital, M139WL - Manchester/UK

Abstract

There are two forms main outputs of risk prediction models. These are the likelihood of a patient harbouring a mutation in a cancer predisposition gene and the likelihood of developing specific cancers. Whilst there are a number of these models for different cancer types, the most well developed area is that of breast cancer. An oncologist treating a breast cancer patient may be influenced in discussing treatment options by future risks of contralateral breast cancer and ovarian cancer. Entry into treatment trials such as those of targeted treatments with PARP inhibitors may also be influenced by genetic testing for BRCA1 and BRCA2. Finding a mutation in BRCA1/2 may also influence treatment choices including risk reducing surgery. Outputs of genetic risk for BRCA1/2 will usually drive the offer of mutation screening if the likelihood exceeds 10% although this is currently 20% in the UK. Models include computer based: BRCAPRO, BOADICEA, Tyrer-Cuzick and paper scoring systems such as the Manchester system. There have been numerous validations of the models which show that they all perform reasonably well. The breast cancer risk element is most used for unaffected women and in the context of family history have only really been validated through one study where the Tyrer-Cuzick model was the best performer. Incorporation of genetic testing information from Single Nucleotide Polymorphisms (SNPs) and mammographic density is likely to improve the precision of these models in the near future.

Disclosure

The author has declared no conflicts of interest.