False positive at colon capsule endoscopy or false negative at conventional colonoscopy?

Dear Sir, We read the interesting paper by Eliakim et al. [1], reporting for the first time the accuracy with regard to colorectal cancer polyps of the second generation of the PillCam Colon capsule endoscopy (CCE) (Given Imaging, Yoqneam, Israel). The authors showed a sensitivity and specificity of 89 % and 76 % for polyps measuring 6 mm or more, and 88 % and 89 % for polyps of 10 mm or more. Although the sensitivity appears to be substantially improved compared to that of the first-generation CCE [2] [3] [4] [5] [6], specificity remains disappointingly low. Considering the low prevalence of significant polyps in a screening setting, this low specificity would result in unnecessary post-CCE colonoscopy in persons with false-positive findings, representing a waste of economic and medical resources. It would also appear quite singular for an endoscopic procedure – with its direct visualization of the gastrointestinal mucosa – to be characterized by suboptimal specificity, when upper and lower conventional endoscopies are usually regarded as having a virtually 100 % specificity.

There are two possible explanations for the apparently low specificity of the second-generation CCE: suboptimal sensitivity of the reference standard colonoscopy, or an imperfect polyp-matching algorithm. In the study by Eliakim et al., conventional colonoscopy was regarded as the gold standard, and patients with positive findings at CCE but negative findings at conventional colonoscopy were counted as false-positive at CCE. The study design provided that a repeat ”unblinded” colonoscopy was offered to all patients with a polyp 10 mm in size or more that was detected at CCE but missed at the first, blinded colonoscopy. It would be expected that detection at repeat colonoscopy of such CCE-detected polyps larger than 10 mm would lead to reclassification of these findings as true-positive at CCE, improving the CCE specificity. Quite surprisingly, however, the authors still classified them as false-positives, rejecting the assumption of an enhanced reference standard, as widely suggested by CT colonography studies [7]. Secondly, polyp matching was apparently based on polyp size. Although the second-generation colon capsule has a system for estimating the size of polyps, this system has never been proven to be accurate. Neither can polyp measurement at conventional colonoscopy be regarded as accurate. Thus, both CCE and colonoscopy might give inaccurate assessments of polyp size, so that it could happen that the two techniques actually describe the same polyp, but the inaccuracy of the measurements prevents adequate matching, resulting in untrue ”false-positive” CCE results.

Analysis of the study by Eliakim et al. suggests that both of these limiting factors may have been present. CCE diagnosed polyps 10 mm or larger that were not reported or were reported differently at conventional colonoscopy in 10 patients [1] [8]. In three of these 10 patients, colonoscopy identified findings that were described differently: in two patients CCE-detected polyps were considered to be smaller (6 – 9 mm) at colonoscopy, whilst in the remaining case the CCE-detected polyp was described at conventional colonoscopy as a hemangioma. In the remaining seven patients, a second colonoscopy was offered. Five patients underwent a second colonoscopy. The CCE finding was confirmed in one; in one patient the CCE finding was confirmed although colonoscopy indicated an 8-mm polyp; while in three patients the CCE findings were not confirmed. The remaining two patients were invited to a second colonoscopy that had not already been performed. Therefore, in five out of eight patients CCE was able to correctly diagnose findings that were missed at the initial colonoscopy or only resulted in a different size estimate as compared to colonoscopy, preventing adequate matching. If these five patients were considered as true-positive rather than false-positive cases, the ”real” sensitivity and specificity of the second generation of colon capsule endoscopy for polyps larger than 10 mm would be 92 % and 94 %, respectively.

In summary, although no definitive conclusions can be made, it seems that this new CCE has substantially better accuracy values than the first version. These improvements, especially in relation to specificity, might be even higher after a careful re-evaluation of conventional colonoscopy and the polyp-matching algorithm. In fact, some of the cases described as false positives at CCE should be better classified as false negatives at conventional colonoscopy ([Fig. 1]).

Fig. 1 A ”false-positive” CCE finding in our experience. The patient is a man who was included in a clinical trial and underwent second-generation colon capsule endoscopy for hematochezia. a The colon capsule detected a 14-mm pedunculated polyp in the left colon. b Conventional colonoscopy confirmed a pedunculated sigmoid polyp, although it indicated a size of 11 mm. c Distal to the first finding, the colon capsule detected another 7-mm polyp that was not confirmed at conventional colonoscopy and was therefore considered as a ”false-positive” capsule endoscopy finding. However, although the finding was formally classified as a false positive on capsule endoscopy, it would have been better classified as a false negative on conventional colonoscopy.

Moreover, matching between the two techniques on the basis of polyp size alone should be performed with caution as it potentially may underestimate the accuracy of CCE, giving rise to a need for new CCE-dedicated matching algorithms.

Competing interests: Guido Costamagna has served as a speaker for Given Imaging Ltd. and has received research funding from Given Imaging Ltd. The other authors certify that they have no commercial associations (e. g., consultancies, stock ownership, equity interest, patent/licensing arrangements, etc.) that might pose a conflict of interest in connection with this letter.

References

• 1 Eliakim R, Yassin K, Niv Y. et al . Prospective multicenter performance evaluation of the second-generation colon capsule compared with colonoscopy.  Endoscopy. 2009;  41 1026-1031
• 2 Eliakim R, Fireman Z, Gralnek I M. et al . Evaluation of the PillCam Colon capsule in the detection of colonic pathology: results of the first multicenter, prospective, comparative study.  Endoscopy. 2006;  38 963-970
• 3 Schoofs N, Deviere J, Van Gossum A. PillCam colon capsule endoscopy compared with colonoscopy for colorectal tumor diagnosis: a prospective pilot study.  Endoscopy. 2006;  38 971-977
• 4 Sieg A, Friedrich K, Sieg U. Is PillCam COLON capsule endoscopy ready for colorectal cancer screening? A prospective feasibility study in a community gastroenterology practice.  Am J Gastroenterol. 2009;  104 848-854
• 5 Van Gossum A, Munoz-Navas M, Fernandez-Urien I. et al . Capsule endoscopy versus colonoscopy for the detection of polyps and cancer.  N Engl J Med. 2009;  361 264-270
• 6 Gay G, Delvaux M, Frederic M. et al . Could the colonic capsule PillCam Colon be clinically useful for selecting patients who deserve a complete colonoscopy?: results of clinical comparison with colonoscopy in the perspective of colorectal cancer screening.  Am J Gastroenterol. 2009;  Nov 3 [Epub ahead of print] PMID: 19888198
• 7 Johnson C D, Chen M H, Toledano A Y. et al . Accuracy of CT colonography for detection of large adenomas and cancers.  N Engl J Med. 2008;  359 1207-1217
• 8 Eliakim R. Capsule endoscopy: from top to bottom. New perspectives. Proceedings of the United European Gastroenterology Federation 2009 Congress. London, 21 – 25 November 2009.