Back to Annual Meeting Program
Absolute Preoperative Carcinoembryonic Antigen Serum Level is Not a Reliable Predictor of Recurrence or Survival in Colon Cancer
*Ramzi Amri, *Liliana G Bordeianou, *Patricia Sylla, David L Berger Massachusetts General Hospital/Harvard Medical School, Boston, MA
Objective: Serum carcinoembryonic antigen (CEA) is used as a tumor marker in colon cancer. Several reports suggest that stratified preoperative CEA levels can be of prognostic value, potentially implying that absolute preoperative CEA levels are of prognostic value. This analysis is designed to test this hypothesis. Design: Retrospective review of a prospectively maintained, IRB-approved data repository. Setting Tertiary care center with high patient volume. Patients 723 patients operated for colonic adenocarcinoma at Massachusetts General Hospital between 2006 and 2011 were included. 57.9% (419) had a preoperative CEA drawn and were included in the analysis. Main Outcome Measures: Preoperative CEA was first correlated with the presence of metastatic disease on presentation, survival and long-term recurrence. Subsequently, the correlation coefficients were recalculated in a partial model, correcting for the presence of metastatic disease on presentation. Results: Preoperative CEA value has a significant, moderately strong correlation with postoperative metastatic status (Point-Biserial rpb=0.354; p<0.001), is weakly correlated with survival (rpb=-0.186; p=<0.001) and does not significantly prove to correlate with long-term recurrence (rpb=0.044; p=0.370). When using a partial correlation model to correct for the confounding effect of cases with perioperative metastatic disease, CEA becomes a nonsignificant weak correlate of survival (r=-0.081 p=0.1) and no significant correlation for the development of metastasis (r=0.005 p=0.917) can be found. Conclusion Though CEA values are a predictor for metastatic disease on presentation, the absolute value of CEA is not a reliable predictor of recurrence or survival in patients presenting without metastasis; nor does CEA strongly predict the prognosis of those who do present with metastasis. This further challenges the confounding effect of using the absolute value of a tumor marker as a prognostic tool.
Back to Annual Meeting Program
|
|