We agree with their comment on the urgent need to develop novel prognostic markers for colon cancer. The current prognostification of colon cancer is predominantly based on the TNM staging classification. However, colon cancer should be regarded as a heterogeneous and multi-pathway disease, and this observation is substantiated by the fact that tumors with similar histopathological characteristics may have different clinical outcomes and responsiveness to therapy [2]. This explains why numerous studies are now being directed towards developing powerful prognostic factors for colon cancer.
In the April issue of the Journal of Surgical Oncology [3], we addressed the prognostic value of the preoperative serum carcinoembryonic antigen (CEA) level in patients with colon cancer. Although whether or not the preoperative CEA level is an independent prognostic variable for colon cancer patients remains controversial, CEA has been the most widely accepted and frequently used tumor marker in the field of colorectal cancer. As we mentioned in our study, the preoperative CEA level was a reliable predictor of recurrence and survival for colon cancer patients and it seemed to have a stage-specific prognostic role.
In addition, we also concur with the comment of Dr. Chen and Dr. Zhao that individualized therapies tailored to a patient's genetic composition and test results may be of great value for treating colon cancer. Various chemosensitivity tests have been studied for this approach, but they have not yet been adopted on a worldwide clinical basis . Only long-term, randomized trials in the future can justify individualized neoadjuvant or adjuvant treatment according to the tumor characteristics of a particular patient. Therefore, developing novel prognostic markers for patient-tailored treatment is essential in the current clinical setting. New research should facilitate evaluating the stage-specific incidence of different markers and their prognostic role. Such studies need to be comprehensive and validated in independent settings, and they should be controlled for multiple variables.