Population based studies show that for rectal cancer the incidence of local recurrence after apparently curative resection is about 20%. Local recurrence after surgery for colon cancer is less common. The liver is the commonest site of distant spread, followed by the lungs; brain and bone metastases are relatively rare. Most recurrences are within 24 months of surgery.

Traditionally surgeons have reviewed their patients at regular intervals after apparently curative resection. Recent surveys, however, have highlighted the lack of consensus among surgeons about the optimal modality and intensity of follow up; surveillance strategies range from a single postoperative visit to lifelong surveillance. Enthusiasts believe that intensive follow up and early intervention will lead to a reduction in the number of deaths from colorectal cancer; others point to the fact that the value of follow up is unproved. With so many tests available and no consensus on their value, it is not surprising that individual clinicians have tended to devise their own protocols.

A meta-analysis in the mid-1990s did little to clarity the situation. The researchers evaluated the results of seven non-randomised studies (covering over 3000 subjects in total) that compared intensive follow up with minimal or no follow up. Clearly several potential biases could and did exist. In the intensive group, investigations included clinical examination, faecal occult blood testing, liver function tests, measurement of the carcinoembyronic antigen, sigmoidoscopy, and either colonoscopy or barium enema examination. Liver ultrasonography was performed in only three studies and even then infrequently. In the intensive group more asymptomatic recurrences were detected, more patients underwent “second look” laparotomy, and more patients had a second potentially curative resection; more metachronous turnouts were also detected and resected. However, although there were fewer deaths in the group receiving intensive follow up, this difference did not reach significance.

Results of randomised clinical trials

Since the meta-analysis, four randomised trials of intensive follow up have been reported. Ohlsson and his colleagues randomised 107 patients to no follow up or to intensive follow up, similar to that described above. No liver imaging was performed routinely. No differences were found in recurrence rates or in overall or cancer specific mortality.

Makela and his associates compared conventional with intensive follow up in 106 patients. In the intensive group flexible sigmoidoscopy was performed every three months, ultrasonography every six months, and colonoscopy and abdominal computed tomography at yearly intervals. Recurrences were detected at an earlier stage (median 10 months v 15 months) in the intensive group. Despite this, no difference in survival was found between the two groups.

Kjeldsen and his colleagues randomised almost 600 patients to either six monthly follow up or to follow up visits at five and 10 years only. Investigations included chest x ray and colonoscopy; no routine liver imaging was performed. Recurrence rates were similar (26%) in both groups, but the recurrences in the intensive group were detected on average nine months earlier, often at an asymptomatic stage. More patients with local recurrence underwent repeat surgery with curative intent. No difference existed, however, in overall survival (68% v 70%) or cancer related survival.

More recently, Schoemaker and his colleagues evaluated the addition of annual chest radiography, colonoscopy, and computed tomography of the liver to a standard follow up based on clinical examination, faecal occult blood testing, liver function tests, and measurement of the carcinoembyronic antigen, with further investigations as clinically indicated. At five years, fewer patients in the intensive group had died, but the result was not significant. At the cost of 505 additional investigations, annual colonoscopy failed to detect any asymptomatic local recurrences; only one asymptomatic metachronous colon tumour was detected. Six hundred and eight additional liver computed tomograms detected only one asymptomatic patient with liver metastases who might have benefited from liver resection.

Carcinoembryonic antigen

Carcinoembryonic antigen concentrations have also been used to predict recurrence. About three quarters of patients with recurrent colorectal cancer have a raised carcinoembryonic antigen concentration before developing symptoms.

An alternative approach therefore would be to monitor this concentration regularly during follow up and, in those patients showing a rising concentration, undertake second look laparotomy. However, although early non-randomised studies suggested that surgery that was prompted by this method resulted in more potentially curative repeat operations for recurrence, more recent studies have failed to show a survival advantage.