Objective: To identify by routine pathology which Dukes B colorectal cancer patients may benefit from chemotherapy.
Method: Retrospective study of the five year survival of colorectal cancer patients for whom colorectal pathology minimum datasets had been collected between 1997 and 2000 in the Yorkshire region of the UK. The study population consisted of 1625 Dukes B and 480 Dukes C patients who possessed one positive node treated between 1997 and 2000. The predictive ability of the Petersen prognostic model was investigated and survival of Dukes B patients with potentially high risk pathological features was compared to that of Dukes C patients with one positive node.
Results: Only 23.3% of patients had all the pathological variables required for the application of Petersen's index reported. The index offered a statistically significant survival difference of 24.3% and 30.3% between high and low risk colon (p<0.01) and rectal cancer patients (p<0.01). The size of these effects was smaller than predicted by the original model. Survival of Dukes B patients with any of the high risk pathological factors or low nodal yields was lower than that of Dukes C patients who possessed one positive node.
Conclusion: Petersen's index discriminated between high and low risk Dukes B colorectal tumours, but inadequate pathological reporting diminished its ability to identify all high risk patients. The survival of patients with any high risk feature was lower than the threshold for adjuvant therapy of one lymph node positive Dukes C colorectal cancer. Chemotherapy may benefit patients with such features. Improving the quality of pathological reporting is vital if high risk patients are to be reliably identified.