Brush cytology has previously been described as a feasible method for accurately diagnosing colorectal cancer.
Purpose: This study was designed: 1) to determine the sensitivity and specificity of brush cytology for the diagnosis of rectal cancer; 2) to prospectively assess the extent of interobserver variability with this technique; 3) to prospectively examine the cost impact of the addition of brush cytology as a routine method of confirming the diagnosis of rectal cancer.
Patients and methods: Three hundred fifty-seven patients who attended a rectal clinica and who were found to have a lesion between January 1990 and March 1996 were assessed. Each patient underwent rigid proctoscopy, followed by brush cytology and tissue biopsy. Results were compared with the final histologic diagnosis in each patient. The brushings from the last 92 consecutive patients in this series were independently examined by four cytologists and a pathologist to determine the rate of interobserver variability.
Results: Rectal adenocarcinoma was confirmed from surgically resected specimens in 303 patients. Brush cytology accurately diagnosed 278 of them. Of the remaining 25 patients, two had brushings that were insufficient for diagnosis. There was one false-positive case. Forceps biopsy correctly identified cancer in 260 patients, with no false-positive interpretations. Brush cytology accurately identified 53 of 54 adenomas as being benign, and forceps biopsy correctly identified all as benign. Sensitivity of brush cytology in this series was 92 percent, with a specificity of 92 percent. Interobserver agreement was 84 percent. Actual costs incurred with this method was an additional $17.00 per patient.
Conclusions: Brush cytology can accurately diagnose rectal cancer in a high proportion of patients. Interobserver variation is low and compares favorably with other forms of cytologic interpretation. The additional cost remains a concern but can be kept within acceptable proportion.