Early detection of colorectal cancer

Abstract

This thesis addresses the early detection of colorectal cancer from several aspects. An audit of the diagnostic process was conducted in 227/235 (97%) patients with recentlydiagnosed colorectal cancer in the County of Västmanland, Sweden in 1998-99. The median work-up-time was 42 (QR 12-110) days for colon cancer and 23 (QR 0-49) days for rectal cancer. A family history was documented on the first visit in 2 /179 (1%) cases. For patients with right-sided cancer and a positive F-Hb, the median work-up time was 53 days while it was 448 days for patients with a negative result (p<0.01). The lack of ascertainment of family history as a means of risk assessment and the benefit of F-Hb in patients attending with symptoms could be questioned.

The same cohort of patients was asked for a date on which the first symptoms appeared with an accuracy of two weeks. Emergency cases were excluded and the elective population was analysed separately. In this group, the median symptom duration was 20 weeks; 19 weeks for Dukes A and B and 21 weeks for Dukes C and D (p=0.8). Symptom duration did not influence overall survival either. Other public health strategies than a reduction in symptom duration have to be found.

The risk of developing colorectal cancer is highly affected by genetic susceptibility. A family history was obtained from 400/411 (97%) eligible colorectal cancer patients in the County of Västmanland in 1998-2001. Five (1.2%) cases of HNPCC were identified. At least one first-degree relative with colorectal cancer was found in 47 (11.4%) patients. Every ninth patient with colorectal cancer thus represents a highly or intermediately increased risk of the disease among relatives. The low frequency makes the establishment of surveillance programs feasible.

Mutations of the APC gene initiate the adeno-carcinoma sequence. As such, they are theoretically optimal markers for the early detection of colorectal cancer. Stool samples were collected from 28 patients with Dukes B cancer, 18 with adenomas of > 1 cm and 28 controls. A Digital Protein Truncation Test was applied and 17 / 28 (61%) cancer patients and 9 / 18 adenoma patients were identified. All the controls were negative. This suggests a new approach to screening for colorectal cancer.

It is uncertain whether proximal cancer could also be detected using faecal DNA analysis. The BAT 26 tract is a marker of microsatellite instability, a characteristic trait of proximal cancer. Stool samples were collected from 46 patients with proximal cancer, 19 with proximal adenoma and 69 controls. Among cancer patients, 18 were found to have BAT 26 alterations in their tumours and 17 of them could be detected in stools. All the controls and adenoma patients were negative. The high specificity strengthens the prospect of DNA-based screening.