The immune infiltrate, TGFb signalling and APC mutation determine complete response to radiation in rectal cancer


Session type:

Enric Domingo1, Sanjay Rathee2, Andrew Blake2, Leslie Samuel3, Graeme Murray4, David Sebag-Montefiore5, Simon Gollins6, Nicholas West5, Rubina Begum7, Marian Duggan7, Laura White7, Sylvana Hassanieh2, Susan Richman5, Phil Quirke5, James Robineau2, Keara Redmond8, Aikaterini Chatzipli9, Ultan McDermott9, Viktor Koelzer10, Simon Leedham2, Ian Tomlinson11, Philip Dunne8, Francesca Buffa2, Tim Maughan2
1University of Oxford, Oxford, UK, 2University of Oxford, 3NHS Grampian – Aberdeen Royal Infirmary, 4University of Aberdeen, 5University of Leeds, 6North Wales Cancer Treatment Centre, 7University College London (UCL), 8Queen’s University Belfast, 9Wellcome Sanger Institute, 10University of Zurich, 11University of Edinburgh



Neoadjuvant radiotherapy is commonly used to treat rectal cancer but patients have different levels of response and/or toxic effects.


As part of the S:CORT programme, we collected 249 rectal biopsies from two cohorts: Grampian and Aristotle. All patients had been subsequently treated with identical regimen (neoadjuvant radiotherapy and capecitabine). We performed transcriptomic, mutation and copy number profiling and aimed to identify biomarkers associated with pathological complete response (pCR). Key biological determinants were identified by linear regression of pre-defined, hypothesis-driven biomarkers for radiotherapy response, adjusted by the confounders T/N stage. In parallel, a novel RNA signature was derived using a personalised bioinformatical pipeline using a wide range of machine learning approaches. All results were validated in a publicly available transcriptomic cohort of 107 patients treated with radiotherapy and 5-fluorouracil infusion. Further comparison of the biological determinants and the novel RNA signature were performed in the same cohorts and also TCGA by linear regression. Previously published transcriptomic signatures for radiotherapy stratification were assessed.


Grampian and Aristotle cohorts had similar statistical power and showed similar associations of pCR with biological candidates. Accordingly, both cohorts were merged into a single discovery set. Following multivariable stepwise regression the final model was composed of the immune biomarkers cytotoxic lymphocytes and CMS1 for radiosensitivity while the stromal TGFb Fibroblasts and epithelial APC mutations were for radioresistance. The first three variables validated in the transcriptomic validation set. In parallel, a 33-gene signature trained in the discovery cohort by a comprehensive machine learning pipeline showed excellent predictive ability in the validation cohort (0.9 AUC; 88% accuracy). Most genes associated with at least one of the four biological features in 3 cohorts. Our novel signature showed much better predictive ability than other previously published transcriptomic signatures.


The immune, stromal and epithelial components of rectal tumours are important players for prediction of pCR to radiotherapy in rectal cancer.

Impact statement

A transcriptomic biomarker can be used to effectively select patients that are highly likely to achieve pCR allowing organ preservation while modulation of the relevant biological features in the other patients may be tested to improve their poor outcome with current treatment strategies.