CT‐based radiomic features to predict pathological response in rectal cancer: A retrospective cohort study

• Published in: Journal of medical imaging and radiation oncology 2020-06, Vol.64 (3), p.444-449
• Main Authors: Yuan, Zhigang, Frazer, Marissa, Zhang, Geoffrey G, Latifi, Kujtim, Moros, Eduardo G, Feygelman, Vladimir, Felder, Seth, Sanchez, Julian, Dessureault, Sophie, Imanirad, Iman, Kim, Richard D, Harrison, Louis B, Hoffe, Sarah E, Frakes, Jessica M
• Format: Article
• Language: English
• Subjects: Adult; Aged; Aged, 80 and over; Biomarkers; Biomarkers, Tumor - analysis; Cancer; Chemoradiotherapy; Cohort analysis; Colorectal cancer; Computed tomography; Female; Florida; Humans; Image contrast; Machine Learning; Male; Medical imaging; Middle Aged; neoadjuvant chemoradiation therapy; Neoadjuvant Therapy; Neoplasm Grading; Neoplasm Staging; pathologic response; Prediction models; Predictive Value of Tests; Radiomics; rectal cancer; Rectal Neoplasms - diagnostic imaging; Rectal Neoplasms - pathology; Rectal Neoplasms - therapy; Regression analysis; Retrospective Studies; Tomography, X-Ray Computed; Toxicity; Training
• ISSN: 1754-9477; 1754-9485
• DOI: 10.1111/1754-9485.13044

Introduction Innovative biomarkers to predict treatment response in rectal cancer would be helpful in optimizing personalized treatment approaches. In this study, we aimed to develop and validate a CT‐based radiomic imaging biomarker to predict pathological response. Methods We used two independent cohorts of rectal cancer patients to develop and validate a CT‐based radiomic imaging biomarker predictive of treatment response. A total of 91 rectal cancer cases treated from 2009 to 2018 were assessed for the tumour regression grade (TRG) (0 = pathological complete response, pCR; 1 = moderate response; 2 = partial response; 3 = poor response). Exploratory analysis was performed by combining pre‐treatment non‐contrast CT images and patterns of TRG. The models built from the training cohort were further assessed using the independent validation cohort. Results The patterns of pathological response in training and validation groups were TRG 0 (n = 14, 23.3%; n = 6, 19.4%), 1 (n = 31, 51.7%; n = 15, 48.4%), 2 (n = 12, 20.0%; n = 7, 22.6%) and 3 (n = 3, 5.0%; n = 3, 9.7%), respectively. Separate predictive models were built and analysed from CT features for pathological response. For pathological response prediction, the model including 8 radiomic features by random forest method resulted in 83.9% accuracy in predicting TRG 0 vs TRG 1–3 in validation. Conclusion The pre‐treatment CT‐based radiomic signatures were developed and validated in two independent cohorts. This imaging biomarker provided a promising way to predict pCR and select patients for non‐operative management.