Send to

Choose Destination
J Gastrointest Surg. 2009 Jul;13(7):1299-305. doi: 10.1007/s11605-009-0892-9. Epub 2009 Apr 15.

A simple scoring system based on clinical features to predict locally advanced rectal cancers.

Author information

Department of Colorectal Surgery, Cancer Hospital, Fudan University, 270 Dong An Road, Shanghai, 200032, People's Republic of China.



The purpose of this study was to identify clinical risk factors and establish a prediction scoring system for locally advanced rectal cancer.


Retrospective univariate and multivariate logistic analyses were conducted for 413 curable rectal cancer patients. Clinical factors found to be significantly related with tumor stages were incorporated into a scoring system to predict locally advanced stages, which was validated in an independent cohort of 279 rectal cancer patients.


In the training set, tumor size, differentiation, and serum carcinoembryonic antigen (CEA) level (P < 0.01) were significant predictors of locally advanced rectal cancer in both univariate and multivariate analyses, which were incorporated into a proposed scoring system to predict locally advanced stages. The area under the receiver operating characteristic curve (AUROC) of this scoring system was 0.751 and the prediction accuracy was 78.2%. Patients were categorized into three subsets according to the total score. The low-risk group (score 0) had a smaller chance (18.2%) to have locally advanced rectal cancer, compared to mean 49.2% for the intermediate-risk group (score 1) and mean 83.0% for the high-risk group (score of 2-4; P < 0.05). In the validation set, the AUROC of the scoring system was 0.756 and the prediction accuracy was 75.3%.


Tumor size more than 2 cm, poor differentiation, and elevated serum CEA level are high-risk factors of locally advanced rectal cancer. A simple scoring system based on these three factors may be valuable to predict locally advanced rectal cancer.

[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Springer
Loading ...
Support Center