DOI: 10.1097/MCG.0b013e31819fcd0a
,
,
PMID: 19525860
Issn Print: 0192-0790
Publication Date: 2009/11/01
Facebook
Twitter
LinkedIn
Email
 |
Checking for direct PDF access through Ovid | |
Excerpt
The question of how many biopsies to take from a suspected colorectal cancer found at colonoscopy is a common clinical dilemma. An inconclusive set of index biopsies may result in a further colonoscopic examination, with its inherent risks and inconvenience. Maximizing the sensitivity of colonoscopy in diagnosing cancer is therefore desirable and hence the optimal number of biopsies needs to be known. To date this variable has only been examined in 1 prospective study by Marshall et al1 in 1993. Currently there are no guidelines, possibly because of the lack of evidence, recommending the number of biopsies that should be taken at index colonoscopy. To allow for the most efficient diagnostic strategy at colonoscopy the optimal number of biopsies needs to be known. We attempted to answer this question with a retrospective study. All cases of suspected cancer, as recorded on the endoscopy report, during a 26 months period were identified. For these patients all colonoscopy reports, corresponding histologic data and subsequent surgical specimens were collected. Two sets of patients were identified, first patients with suspected carcinoma confirmed on initial biopsies. Second, patients with suspected carcinoma not confirmed on initial biopsies but subsequently diagnosed histologically from biopsies taken either at a second colonoscopy or surgical resection. The number of biopsies taken during colonoscopy for each group was statistically analyzed using a Mann-Whitney test and the sensitivity for making a correct diagnosis of cancer was calculated for individual biopsy number. Two hundred and seventeen patients with a complete data set were analyzed, of whom 198 patients were identified with a histologic diagnosis of cancer, 107 male and 91 female. On the 198 case of cancer, 182 patients were correctly diagnosed at index colonoscopy and 16 required either a repeat colonoscopy (10) or surgery (6). The mean number of biopsies in the 16 patients with cancer “missed” at index colonoscopy was statistically lower than in confirmed cases [4.250 (95% confidence interval: 3.391-5.109) compared with 5.729 (confidence interval: 5.455-6.004), P=0.0020].
The sensitivity of diagnosis of a suspected cancer seen at colonoscopy positively correlated with the number of biopsies taken. A significant increase in sensitivity is seen at 6 biopsies. Our data show a 98% sensitivity for 6 or more biopsies in confirming cancer at index colonoscopy. The sensitivity for each number of biopsies were, 2 (80%), 3 (86%), 4 (86%), 5 (88%), 6 (98%), 7 (100%), 8 (94%), and greater than 9 (100%). A lower number of biopsies taken at index colonoscopy correlated with a higher chance of further investigation being necessary to make the cancer diagnosis.
Currently there are no guidelines and few data available to suggest the optimal number of biopsies necessary. A study in 1993, by Marshall et al,1 demonstrated that 4, 6, and 10 biopsies detected cancer, when present, in 68.3%, 78.3%, and 78.3% of cases, respectively. The findings that the sensitivity could be increased with cytology lead to the suggestion that this should be routine. Our data found a higher sensitivity, 80%, than Marshall et al's data even for the lowest number of biopsies taken. This discrepancy may be accounted for by improvements in endoscope technology and colonoscopy training. Although we did not aim to study the use of cytology, the sensitivities from our study suggest than it is not routinely necessary as standard if 6 or more biopsies are taken.
Only 1 study has attempted to identify the optimal number of biopsies in the diagnosis of esophageal cancer. Lal et al2 prospectively investigated 48 patients with esophageal cancer and found that for 2, 4, or 6 biopsies, the sensitivities were 95.8%, 97.9%, and 100%, respectively.
