Introduction Response to neoadjuvant therapy in oesophago-gastric (OG) cancer can be measured using a range of radiological and histological measures. However, there is no agreed definition of what constitutes a beneficial response. An accurate definition is critical for standardisation of outcome reporting and will be necessary if predicting response can be used to provide personalised treatment. This study re-defines response to neoadjuvant therapy in OG cancer.

Method A literature search was performed to identify potential measures for accurately defining response the neoadjuvant therapy. All patients undergoing surgical resection for OG cancer at this Unit between January 2010 and December 2014 were identified from a database.

Part a) Patients meeting the threshold for offering neoadjuvant therapy (≥T3 and/or ≥N1) were identified and divided according to whether they were TRG responders, non-responders or had surgery only. These groups were then compared with respect to the apparent T- and N-down-staging (defined as a reduction in stage between pre-operative staging and post-operative pathological staging) in order to assess the validity of down-staging for use in defining response.

Part b) Patients undergoing neoadjuvant therapy were identified and potentially valid measures of response were then analysed with respect to overall survival in order to determine and validate a definition of response.

Results The literature review identified methods using histopathological regression as the most promising measures of response with the Becker and Mandard scoring systems validated for use in OG cancer. T and N down-staging offer the potential to identify partial responders. 587 patients were identified from the database.

Part a) In 419 patients above the neoadjuvant therapy threshold, the rate of apparent T down-staging was 62.8% in TRG responders, 15.7% in non-responders and 21.4% in the surgery only patients. The rate of apparent N down-staging was 35.9% in responders, 13.0% in non-responders and 12.9% in the surgery only patients.

Part b) In 376 patients undergoing neoadjuvant therapy, patients with a Mandard TRG score of 1–3 had longer survival than those with TRG 4–5, mean survival 49.5 months vs. 35.7 months respectively (P = 0.001, Log Rank Mantel-Cox).

Conclusion Histopathological regression scores correlate well with survival and represent the most accurate measure of response to neoadjuvant therapy in OG cancer. Survival in TRG1–3 patients (Mandard Score) is better than in TRG 4–5 patients. Apparent T and N down-staging were no higher in the TRG non-responder group than the surgery only group suggesting this represents clinical over-estimation of stage rather than a true therapy effect and is not valid for use in the definition of response to therapy.

Disclosure of interest None Declared.