Artificial intelligence predicts treatment response in epithelial ovarian cancer

  • Lu H & al.
  • Nat Commun
  • 15 Feb 2019

  • curated by Dawn O'Shea
  • UK Medical News
Access to the full content of this site is available only to registered healthcare professionals. Access to the full content of this site is available only to registered healthcare professionals.

Researchers at Imperial College London and the University of Melbourne have developed a novel radiomics-based prognostic signature which reliably predicts survival and treatment response in epithelial ovarian cancer (EOC).

The software program, known as TexLab 2.0, was developed based on 657 features relating to the shape and size, intensity, texture and wavelet decompositions of 364 pre-operative contrast-enhanced CT scans. A non-invasive summary-statistic of the primary ovarian tumour was created, which the developers named ‘Radiomic Prognostic Vector’ (RPV).

RPV was used to assess 294 primary patients with EOC with fresh frozen tissue treated at Hammersmith Hospital (HH) between 2004 and 2015, as well as 70 patients with EOC from The Cancer Genome Atlas (TCGA) project.

The patient groups stratified by RPV had distinct overall survival (OS) differences. OS differences were confirmed in the TCGA (P=.000105) and HH (P=.0274) data sets. The addition of RPV improved clinically available prognostic methods (stage, age and post-operative residual disease) in all data sets.

High RPV was also associated with chemotherapy resistance and poor surgical outcomes, suggesting that RPV can be used as a potential biomarker to predict treatment response.

The authors suggest RPV and its associated analysis platform could be exploited to guide personalised therapy of EOC and is potentially transferrable to other cancer types.

The research is published in Nature Communications.

Please confirm your acceptance

To gain full access to GPnotebook please confirm:

By submitting here you confirm that you have accepted Terms of Use and Privacy Policy of GPnotebook.

Submit