Background: Perioperative chemotherapy improves outcomes of surgery for upper GI cancer. Some patients have rapid recurrence. We hypothesized that any type of response to chemotherapy would predict better disease-free survival (DFS).
Methods: From May 2007 to Sep 2009, 43 patients with operable adenocarcinoma of the esophagus, stomach or gastroesophageal junction went on a multicenter phase II trial. 3 cycles of docetaxel/cisplatin/5FU were given pre and post surgery. We compared DFS between responders and nonresponders after 3 cycles. Clinical response was defined as improvement of ≥ 2 points on a dysphagia score (0: normal swallowing, to 4: total obstruction), metabolic response as ≥ 35% reduction of maximum SUV by PET scan. Lack of nodal involvement or significant histologic regression (less than 50% viable tumor) was considered pathologic response. Log-rank test was used for univariate analysis, Cox regression model for multivariate analysis. All p values are double sided, median follow-up from surgery is 808 days.
Results: Clinical response data for 29/40 patients: 26 responses (90%). Median DFS was 405 days for clinical responders vs. 210 days for nonresponders (p= 0.018). Metabolic response data for 28/40 patients : 22 responses (70%). Median DFS was 276 days in metabolic nonresponders, and has not been reached in responders (p = 0.009). 14 patients had no nodal involvement at resection (N0) and had longer DFS (median not reached) than the 26 with N+, who had median DFS of 562 days (p = 0.045). A nonsignificant trend (p = 0.192) was seen in favor of the 20 with significant tumor regression (DFS 565 days) compared to the 20se without it (median DFS not reached). All results were supported by multivariate analysis.
Conclusions: These preliminary results indicate that clinical, metabolic and pathologic responses may be individually predictive of longer DFS in patients receiving perioperative chemotherapy for upper GI adenocarcinoma. This requires confirmation in larger datasets. The optimal management of nonresponders is unclear at present. Acknowledgements: Biostatistics Unit (McGill University Health Centre) and Cedars Cancer Institute (Henry R Shibata Fellowship to Dr Alcindor).