Appropriate statistical descriptions for evaluating the predictive role of epithelial-to-mesenchymal transition in patients with pancreatic cancer.

I read with interest the recent article by Yamada et al. entitled ''Epithelial-to-mesenchymal transition predicts prognosis of pancreatic cancer.'' Although the authors reported that epithelial-to-mesenchymal transition is predictive for both survival and response to adjuvant therapy in patients with pancreatic cancer, in the article there are several statistical inaccuracies that could limit the impact of these findings.