Clinical and Radiomics prediction of complete response in Rectal cancer

  • Peter Mbanu

Student thesis: Doctor of Medicine


Abstract Purpose: About 15% of patients post neoadjuvant chemoradiotherapy in rectal cancer achieve clinical complete response (cCR) and could avoid or defer surgery by entering a watch and wait surveillance treatment plan. Patients that achieve cCR have overall better outcomes. Prediction of complete response before treatment is essential for neoadjuvant treatment selection. Method: Using the UK-based research OnCoRe (The Rectal Cancer Oncological Complete Response Database) database, we performed a propensity-score matched (1:1) case-control study of 322 patients (161 patients with cCR and 161 without cCR) who received neoadjuvant chemoradiotherapy. We collected pre-treatment MR images, demographics, clinical and blood parameters and radiotherapy-related characteristics. We segmented the gross tumour volume on the T2W MR Images and extracted 1430 stable radiomics features per patient. We wanted to compare the predictive power of clinical parameters and the radiomics variable in predicting complete response. Results: Using Logistic regression analysis, the PCA-derived combined model (radiomics plus clinical variables) gave a ROC AUC of 0.76 in the training set and 0.68 in the validation set. The clinical-only model achieved an AUC of 0.73 and 0.62 in the training and validation set. The radiomics-only model had an AUC of 0.68 and 0.66 in the training and validation sets. Various clinical variables were associated with cCR. A nomogram using only routinely acquired clinical variables was developed with a resulting ROC AUC of 0.75. Conclusion: The predictive abilities of clinical variables for cCR are better than radiomics variables. Combining clinical and radiomics variables improves predictability. However, their predictive characteristics remain modest. The Nomogram of the clinical variable produced will need to be enhanced before prospective validation and clinical use.
Date of Award1 Aug 2023
Original languageEnglish
Awarding Institution
  • The University of Manchester
SupervisorEliana Vasquez Osorio (Supervisor), Mark Saunders (Supervisor), Andrew Renehan (Supervisor), Rohit Kochhar (Supervisor) & Marcel Van Herk (Supervisor)

Cite this