Key points

Introduction

Colonoscopy has three main roles in the area of colorectal neoplasia. The first is to screen for the disease, preventing it by the detection and removal of potentially premalignant lesions, or providing presymptomatic diagnosis of cancers at an early stage. The second is to diagnose the disease by the investigation of symptoms. The third is to prevent metachronous cancer by the surveillance of patients who already have had a colorectal neoplasm. The efficacy of colonoscopy in fulfilling these roles depends on an accurate examination of the entire colorectal mucosa. This is the rub. In an era of “pay for performance” and a time of increased demand for screening and surveillance colonoscopy, quality has become an important issue. Indices of quality of colonoscopy are now more pragmatic, focused more on what is found than how complete the examination or how satisfied the patient. In this article, the miss rates of colonoscopy are considered and the impact of miss rates on the incidence of colorectal cancer is examined.

Biology

Before discussing colonoscopy and its ability to detect colorectal neoplasia, it is important to consider the molecular mechanisms by which colorectal neoplasia develops. There are three main ones. Chromosomal instability is the most common, giving rise to 60% of colon cancers and 90% of rectal cancers. The accumulation of genetic changes (mutations, hypomethylation, and loss of heterozygosity) is reflected in the mucosa by a progression from small adenomas to larger adenomas to severely dysplastic adenomas to cancer. CpG island methylator phenotype (CIMP) is the second most common, a widespread methylation of CpG promoter islands that causes loss of expression of many genes. When combined with an initial BRAF mutation, such hypermethylation produces advanced serrated polyps (sessile serrated adenomas/polyps [SSA/Ps]) that become severely dysplastic and ultimately malignant. Approximately 18% of colon cancers and 4% of rectal cancers are CIMP-high. The third most common mechanism is loss of DNA mismatch repair. This causes mutations throughout the DNA due to single base or loop mismatches at DNA microsatellites, a phenotype known as microsatellite instability (MSI). High levels of MSI can lead to cancer by causing carcinogenic mutations in a variety of genes. Most mutator cancers arise from adenomas, but serrated polyps are also effects of the mechanism. Approximately 18% of colon and 3% of rectal cancers are mutator cancers. The most common cause of MSI colon cancers is sporadic hypermethylation of MLH1 , a DNA mismatch repair gene. Some MSI cancers, however, are due to a germline mutation in 1 of 4 DNA mismatch repair genes, known as Lynch syndrome. Colonoscopy intervenes in the mucosal manifestation of these mechanisms but does nothing to the underlying risk. Patients diagnosed with the hereditary versions of these mechanisms (chromosomal instability = familial adenomatous polyposis, CIMP+ = serrated polyposis, and mutator = Lynch syndrome) need specialized care that takes into account the instability of the colorectal epithelium with a subsequent high risk of colorectal cancer, the risk of extracolonic cancer, and the status of the family. This is best accomplished in the context of a registry.

Aims of colonoscopy

The aims of colonoscopy, as far as colorectal cancer is concerned, are to prevent it or, if it is already there, at least to diagnose it early. The aim is not necessarily to remove all polyps regardless of size. The number and types of polyp in the colorectal mucosa are, however, a reflection of the mechanisms active at a molecular level. This is why it is important to be able accurately to document the cumulative numbers of adenomas and serrated polyps. Just as every colorectal cancer has a unique molecular fingerprint, so every colon has a unique blend of molecular changes that either encourage or discourage neoplasia. Recognizing the degree of instability of the colorectal epithelium is a key to designing an appropriate endoscopic surveillance program. Therefore, the aim of colonoscopy is to find every polyp and to remove and biopsy most of the polyps—not because all polyps turn into a cancer (they do not) but because a risk status can then be assigned to the colon that determines the recommended surveillance interval.

Colonoscopy and cancer prevention

The National Polyp Study caused a big stir in 1993 when it published its initial results showing a considerable reduction in the death rate from colorectal cancer in patients who had their adenomas removed. Compared to three different control groups, the reduction in colorectal cancer incidence rate was 76%, 88%, and 90%. Recently, a longer follow-up of the patients confirmed a reduction in colorectal cancer mortality of 53%. These data are not analyzed by site of the cancers, however. This was reported by Baxter and colleagues in 2009, in a much different kind of study, looking at the integrated results of population-based colonoscopy; a sort of warts and all approach but one that is likely closer to reality than a prospective, randomized polypectomy study. They included 10,292 patients who underwent colonoscopy and compared them to 51,460 controls. Colonoscopy was associated with fewer deaths from left-sided colorectal cancer (odds ratio [OR] 0.33) but not with any reduction in deaths from right-sided colon cancer (OR 0.99). These findings have been reproduced by others, raising concerns about the performance of colonoscopy and the problem with finding and removing premalignant lesions in the right colon.

Miss rates

Polyp miss rates are the best way of determining the accuracy of colonoscopy in detecting neoplasia. Most miss rates are determined by tandem colonoscopy, and a selection of recent series is shown in Table 1 . The absolute miss rates vary but so do study designs. The summary is useful, however, in showing an average adenoma miss rate of rate of 22%, a range from 12% to 26%, and some of the factors influencing miss rates. The most influential factor is size. Small polyps are harder to find. Shape makes a difference in that flat or sessile lesions are also a problem. The number of polyps was significant in two studies, showing that if there are two polyps, there is a strong possibility of a third or a fourth. The influence of polyp location was not consistent, with left-sided polyps easier to find in one study but harder in others. Finally, none of the new techniques seemed to make a difference in detection rates—not narrow band imaging, not the third eye retroscope, not high-definition imaging.

Adenoma detection rates

Adenoma detection rates (ADRs) are subtly different from miss rates. ADR is usually defined as the proportion of patients with at least one histologically proved adenoma whereas miss rates are defined as the proportion of the total number of adenomas missed. Thus, a patient who has four adenomas represents a tick in the detection rate column if only one is found and removed. The same scenario is an adenoma miss rate of 75%. This demonstrates the fallacy of overall ADR and one of the problems of using it as a quality indicator. More detailed ADRs have been reported to show the average number of adenomas per patient screened and the number of patients with multiple adenomas or serrated polyps. These rates are not generally used, however, as a summary quality indicator. Advanced ADRs are also reported to show the incidence of high-risk lesions that may be expected during screening colonoscopy. Some recent studies are summarized in Table 2 to give an indication of the sort of detection rates that are reported. Both total adenoma detection and detection of advanced adenomas are considerably higher in men than women. Detection rates in patients on surveillance are also higher than those in average risk screening.

Kaminski and colleagues showed that the risk of interval cancer in patients on a colonoscopy screening or surveillance program is related to ADR. This makes sense and confirms ADR as a significant quality measure, a surrogate for the ability of endoscopists to protect patients from cancer.