The Practical Value of Computed Tomography to Evaluate Response to Neoadjuvant Chemotherapy in Oesophageal Adenocarcinoma

Background The current standard of care for locally advanced curable oesophageal adenocarcinoma (OAC) is surgery with peri-operative chemotherapy. Computed tomography (CT) is widely used for staging and restaging OAC after neoadjuvant chemotherapy (NAC). The restaging scan needs to provide accurate information for surgical planning, including changes in the dimensions of the primary tumour, and to exclude metastatic disease. The value of CT in evaluating the primary tumour has not been clearly established, and there are no accepted reporting guidelines for restaging scans. This project aimed to determine if CT imaging to restage the primary tumour after NAC provides prognostically relevant information. Methods This was a retrospective single-centre study of patients receiving treatment for OAC. Patients were eligible if they had a diagnosis of OAC treated with curative-intent oesophagectomy following NAC, with available baseline and post-NAC CT scans. Tumour dimensions were measured on both scans (short axial diameter, long axial diameter, length, location compared to gastroesophageal junction/carina). The study outcomes were histopathologic tumour regression measured by the Mandard grade, and overall survival. Receiver operating characteristic (ROC) curves were fitted to generate cut-offs, to cohort the survival analysis. The timing of the interval scan was assessed, given its importance in other cancer pathways. Results One-hundred-and-eighty-seven patients were analysed. All tumour measurements were significantly different after NAC (all P < 0.0001). Only tumour length was inversely associated with Mandard grade (Mandard 1–2 17.72% ± 19.06, Mandard 3–5 5.45% ± 19.2, P = 0.0001). Patients defined as non-responders by ROC-selected tumour measurement cut-offs were not associated with poorer survival, despite poorer survival linked higher Mandard. However, the pattern of regression was prognostic; patients with a regression towards the GOJ had a poorer survival (towards GOJ median survival 28 months; away from GOJ, 38.5 months, P = 0.045). The timing of restaging scans within the treatment pathway was not linked to differences in response. Conclusion Although CT-derived tumour measurements change after neoadjuvant treatment and reduced length correlate with histopathologic Mandard regression, these were not prognostic in the current dataset. This may be due to the limitation of the technique in distinguishing fibrosis from viable cancer. However, when tumour length was analysed in the context of anatomical landmarks it demonstrated prognostic value, with implications for the planning of resection margins.