LONDON — Clinical symptoms are of little value in the selection of appropriate patients for colonoscopy for the purpose of early detection of colorectal cancer, despite common beliefs to the contrary.
These were the disappointing results of a large Australian retrospective analysis presented at the 13th World Congress of Gastroenterology meeting.
The early warning signs for colorectal cancer are well known—recurrent abdominal pain, rectal bleeding or pain or excretion of mucus, or radical alteration in bowel habits, among other symptoms. But most of these symptoms have little predictive value and, when considered in isolation, these symptoms may lead to thousands of unnecessary colonoscopies at great cost, according to Dr. Peter Katelaris, one of the lead researchers of the CRISP study (Colorectal Research in Symptom Prediction) performed at the Concord Repatriation General Hospital in Sydney.
“Most symptoms are not predictive of bowel cancer and are a poor guide to the best use of colonoscopy,” Dr. Katelaris said. “Basic screening tests have much higher predictive value of co-lorectal cancer than [do] patient symptoms. Perhaps it is time for a reappraisal,” said Dr. Katelaris, clinical associate professor in the department of gastroenterology at the University of Sydney.
The CRISP analysis of 5,577 patients compared their self-reported symptoms on a presenting questionnaire and their medical histories against their diagnosis based on colonoscopy. A total of 159 patients (3%) were confirmed to have cancer. Yet a similar population of the same median age would generally have had a 2% rate of colon cancer. “We are talking about an absolute increase of 1%, which is not very useful for interventions on this scale. We're wasting a lot of colonoscopy resources on this,” Dr. Katelaris said.
Only one predictive factor stood out: patient age. Those aged 70 years and older showed an 8.6% increased likelihood of a cancer diagnosis upon colonoscopy. A history of previously diagnosed rectal polyps or having undergone colonoscopy in the preceding 10 years also indicated increased risk of a colon cancer diagnosis. Heavy smoking also slightly increased the likelihood of a cancer diagnosis.
But the most common triggers for colonoscopy referral—abdominal pain, rectal bleeding, and related bowel irregularities—showed almost no correlation with histologic findings, unless symptoms had persisted for months. In women, these symptoms had low predictive value for a cancer diagnosis.
Dr. Katelaris noted that 20% of those patients diagnosed with cancer upon colonoscopy in this cohort showed no symptoms whatsoever.
“Colonoscopy to detect cancer need not be done for many bowel symptoms [that] are currently considered to be indications,” they said.