Nonpolypoid Colorectal Neoplasia in the United States: Title and subTitle BreakThe Parachute Is Open

People's minds are like parachutes—they only function when they are open.—Thomas Robert Dewar (1864-1930)

The history of modern medicine is riddled with stories of discovery, skepticism, rejection, and, sometimes, redemption and acceptance. For instance, the discovery of Helicobacter pylori by Marshall¹ is a fairly recent example of initial rejection by the mainstream scientific community and ultimate redemption (as recognized by the Nobel Prize).

Similarly, reports from Japan in the 1980s and 1990s suggested that nonpolypoid (ie, flat or depressed) colorectal neoplasms (NP-CRNs) were common and ominous.^{2 - 3} Depressed lesions commonly harbored in situ or submucosal carcinoma. However, some western investigators were skeptical because such lesions were not commonly found in North American patient populations, and the rates of high-grade dysplasia in small flat polyps were low.^{4 - 5} In 1992, a prospective study from Nebraska found that although flat adenomas could be found with careful examination, none contained high-grade dysplasia.⁶ Subsequent studies from Sweden,⁷ Canada,⁸ the United States,⁹ and the United Kingdom¹⁰ reported that with careful examination technique, flat and depressed lesions were indeed found in the West. Even though there was agreement that flat lesions are difficult to detect and might easily be missed with routine colonoscopy, the importance of such findings in the West remained uncertain but of concern.

In this issue of JAMA, Soetikno et al¹¹ provide new information about rates of NP-CRN in 1819 predominantly male veterans who received colonoscopy for screening, surveillance, or symptoms. Using special techniques (chromoendoscopy with indigo carmine spray) to highlight neoplastic lesions, the overall prevalence of NP-CRN was 9.35%, accounting for 15% of neoplastic lesions. The likelihood that NP-CRNs harbor serious pathology (in situ or submucosal carcinoma) was more than 5 times higher than the rate in polypoid lesions after adjusting for polyp size. These data are consistent with reports in Asian populations and suggest that these lesions are important and may differ biologically from polypoid neoplasia.⁹ What are the implications in the West?

First, an important issue is whether important colon pathology is being missed with colonoscopy. Over the past decade, several interventional studies have performed baseline colonoscopy and repeated examinations within 2 to 5 years to detect interval neoplasia.¹² These studies found that 0.3% to 0.9% of patients (in 1.7-2.8 per 1000 patient-years of follow-up) have interval cancers within only a few years of colonoscopy and polypectomy.¹² Explanations could include new de novo, rapidly growing tumors in some cases, but either missed lesions or incompletely removed lesions probably account for most of the interval cancers. Incomplete removal may be especially important for NP-CRNs, which may have indistinct borders that are difficult to discern with white-light endoscopy. Chromoendoscopy with special stains¹¹ or the use of new technologies such as narrow band imaging or fluorescence endoscopy¹³ may help enhance the visualization and demarcation of flat lesions at colonoscopy. Further work is needed to determine the optimal imaging technique. Methods of removal, as described by Soetikno et al¹¹ using endoscopic mucosal resection, may need to be used more commonly to ensure complete removal. In the study by Soetikno et al, the rate of NP-CRN was highest in the cohort undergoing surveillance colonoscopy, raising the possibility that some lesions may have been incompletely removed or missed at the baseline examination.

The elephant in the endoscopy suite is missed lesions. There has been increasing recognition that colonoscopy, even by experienced and well-trained endoscopists, may fail to detect important colon pathology. Computed tomography colonography studies have shown that optical colonoscopy misses 2% to 12% of polypoid lesions larger than 10 mm.^{14 - 16} It is possible, if not likely, that additional NP-CRNs may be missed by both studies so that these studies underestimate the actual colonoscopic miss rate. These missed lesions may represent the most common explanation for interval cancers.

Second, these findings may have implications for colorectal cancer screening. The finding that NP-CRNs have high rates of serious pathology suggests that effective screening programs will need to accurately identify patients who harbor these lesions. Colonoscopy is increasingly used as a primary screening test and remains the secondary test for all other screening programs. High-quality colonoscopy remains the key to any effective screening program. Recent work has highlighted quality indicators of colonoscopy¹⁷ and provided a reporting tool (Colonoscopy Reporting and Data System) to promote quality improvement in the performance of colonoscopy.¹⁸ For example, a recent study demonstrated that adenoma detection rate is improved when endoscopists spend more time during endoscope withdrawal.¹⁹ Accurate detection and complete removal of NP-CRNs at colonoscopy will be important elements of every screening program. The Colonoscopy Reporting and Data System calls for documentation of polyp morphology, which will enable endoscopists to monitor rates of NP-CRN in their endoscopic experience. The current study of NP-CRNs highlights the challenges facing colonoscopists and the importance of a high-quality examination.

Imaging tests such as computed tomography colonography are unlikely to detect flat lesions, which do not protrude into the lumen of the colon. The failure to detect clinically important NP-CRNs may represent an important limitation of computed tomography colonography as a screening test and will require further study.

A third consideration is the implication for colon surveillance. Soetikno et al¹¹ report on partial follow-up of 580 patients for whom surveillance colonoscopy was recommended within 3 years. Among 393 patients (68%) who had surveillance colonoscopy, the proportion with advanced neoplasia was low (3.3%) and included 1 rectal cancer. Data from the study are insufficient to determine whether patients with NP-CRNs at baseline need more intensive surveillance. Other longitudinal studies, such as VA Cooperative Study 380,²⁰ reported higher rates of advanced neoplasia within 3 years for those patients with advanced neoplasia at baseline (10.8%) and in patients with tubular adenomas smaller than 10 mm (4.5%). Patients with NP-CRNs were not distinguished in these surveillance studies. If the biological fingerprint of NP-CRNs differs from that of polypoid lesions, then prognosis might also differ. Longitudinal follow-up of patients with NP-CRNs is needed. At this time, there is no basis for modifying the current surveillance recommendations for patients with NP-CRNs.²¹

It is now clear that both Asian and Western populations may develop NP-CRNs. These lesions may be biologically distinct from polypoid lesions and appear to be more likely to harbor malignant features. Detection and complete removal at colonoscopy may be challenging. The current study emphasizes the importance of quality in the performance of colonoscopy. Future reports of colonoscopy performance should include descriptors of polyp morphology and methods of polyp removal consistent with the Colonoscopy Reporting and Data System.¹⁸ Several areas of further investigation are needed. The optimal methods for enhancing colonoscopic imaging of NP-CRNs are uncertain. Chromoendoscopy with indigo carmine seems to work, but other methods that might be technically easier might also be equally effective, such as narrow band imaging or fluorescence endoscopy, which do not require time for spraying the mucosa. Moreover, additional studies are needed to determine whether imaging modalities such as computed tomography colonography will be able to detect NP-CRNs. Finally, longitudinal studies are needed to determine whether patients with NP-CRNs require more intensive colonoscopic surveillance compared with patients with polypoid lesions of similar size and histology.