Chromoendoscopy Detects More Adenomas than Colonoscopy Using Intensive Inspection without Dye Spraying

Most colorectal cancers are believed to arise from adenomatous polyps (1, 2) and there is convincing evidence that colonoscopy with removal of colorectal adenomas reduces the risk of colorectal cancer (1, 3). Although colonoscopy is considered the gold standard for detecting adenomatous polyps, studies have documented colonoscopy miss rates of 6% to 27% for adenomas (4, 5). Studies using back-to-back colonoscopy (4), comparing colonoscopy to computed tomography colonography (6), and matching large colorectal cancer registries with endoscopic databases (7) have shown that adenomas, as well as cancers, can be missed during conventional colonoscopic exams. There continue to be cases of colorectal cancer diagnosed in individuals who had recent colonoscopic examinations in which no neoplastic lesions were seen (8), suggesting that missed lesions may be clinically important.

Chromoendoscopy, performed by spraying dye on the colorectal mucosa during colonoscopy, has been reported to improve detection of flat dysplasia in patients with ulcerative colitis (9) and detection of flat adenomas in screening and high-risk populations (10–17). However, previous randomized trials that have examined the sensitivity of chromoendoscopy for adenoma detection in average and moderate-risk subjects reached different conclusions about the clinical utility of spraying dye (11, 12, 17). Critics of chromoendoscopy have argued that the technique is messy and time-consuming and have proposed that the prolonged inspection time required to perform chromoendoscopy, and not the actual dye spraying, is the reason for the higher sensitivity for detecting polyps.

We conducted a randomized multicenter study to ascertain if chromoendoscopy was superior to intensive conventional endoscopy for detecting adenomas missed by standard colonoscopy in patients with a prior history of colorectal neoplasia. Our study was designed to control for the potential effect of procedure time on adenoma detection rate.

Study subjects were enrolled at five collaborating study centers associated with the Great Lakes New England Clinical Epidemiology and Validation Center of the Early Detection Research Network [University of Michigan, Dana-Farber/Brigham and Women's Hospital, M.D. Anderson Cancer Center, University of Toronto, and the Tel Aviv Sourasky (Ichilov) Medical Center]. Subjects were recruited from among patients scheduled to undergo surveillance colonoscopy. Eligible subjects were those with a prior history of colorectal cancer or ≥3 colorectal adenoma(s). Individuals under 18 years of age, with poor performance status, receiving active treatment for cancer, or using anticoagulant medications were ineligible for the study. This study was approved by the institutional review board or ethics board at each institution.

All subjects underwent back-to-back colonoscopy exams, with a conventional colonoscopy followed immediately by a second endoscopy with either chromoendoscopy or intensive colonoscopy. Subjects were randomized after the cecum was reached during the second exam and randomization was done in block sizes of two, stratified by study site. The endoscopist, study coordinator, and endoscopy nurse were not made aware which randomization arm had been assigned until the cecum was reached during the second colonoscopy and the randomization envelope was opened.

Subjects provided informed consent and completed demographic and medical history questionnaires before colonoscopy. Subjects took a standardized preparation on the day before colonoscopy [magnesium citrate (12 oz) followed by either large volume (4 liter) polyethylene glycol colonic lavage, 1.5 oz of oral phosphosoda followed by 24 oz water (2 doses), or Visicol tablet preparation].

The first exam for all subjects was a standard colonoscopy with removal of all visualized polyps. On completion of the first colonoscopy, subjects were considered eligible to undergo the second exam if all of the following criteria were satisfied: the preparation was considered excellent; the first standard colonoscopy was completed in less than 30 minutes; the endoscopist considered the exam to be technically easy; and the endoscopist, study coordinator, and endoscopy nurse all agreed that the subject was comfortable and clinically able to immediately undergo a second procedure. Study participants were contacted 24 to 72 hours after their procedures to determine if they had experienced any adverse effects.

All 8 endoscopists participating in this study underwent training in chromoendoscopy technique and recognition of polyp morphology. Standard nonmagnifying Olympus-160 or Pentax-160 colonoscopes (3 and 2 study sites, respectively) were used for all study procedures.

During each colonoscopy, a study coordinator recorded data from the endoscopic procedures, including the duration of each aspect of each procedure (time from endoscope insertion to visualization of cecal landmarks, time withdrawing from cecum to anal verge, and time spent performing polypectomy). The endoscopist assessed the location and size of each polyp as measured by placing an open standardized biopsy forceps (Bard 00823 C diameter 9.6-mm inner dimension) adjacent to the polyp. Endoscopists were instructed to classify polyp morphology as polypoid or flat, with flat polyps defined as having height less than half of the diameter of the lesion (10-12). All polyps were numbered and photographed before they were fully removed with standardized biopsy forceps or snares, according to standard clinical practice.

For subjects randomized to chromoendoscopy as their second exam, the entire colon was sprayed during withdrawal of the colonoscope with 0.2% indigo carmine solution with a standardized (Olympus pw-5v-1) spraying catheter and the mucosa was inspected in 10-cm segments. Each 20 mL of indigo carmine solution contained 1 mL of simethicone as an antifoaming agent. An average of 100 mL of solution was used per patient.

Subjects randomized to intensive inspection received a thorough examination of the colon without indigocarmine dye. Endoscopists were instructed to spend at least 20 minutes visualizing the colonic mucosa during withdrawal from the cecum, exclusive of time spent performing polypectomy.

All polyps seen on withdrawal of the endoscope during each of the two colonoscopic examinations were removed. Tissue removed from each visualized lesion was placed in its own jar, fixed in 10% buffered formalin, and processed by the routine pathology procedures at the local institutions. The pathologic diagnosis was made by the pathologist at each of the five collaborating institutions per routine practice. Based on the histopathologic diagnosis, lesions were categorized as adenomatous polyps, hyperplastic polyps, normal tissue, or “other.”

The primary objective of this study was to compare the adenoma detection rates of chromoendoscopy and intensive inspection colonoscopy without dye spraying performed after a standard colonoscopic examination. The study was designed as a multicenter randomized trial with 50 subjects. Polyp and biopsy counts were analyzed by means of generalized linear models (SAS PROC GENMOD, SAS Institute), which assume that the number of lesions of any particular type identified in a given patient follows a Poisson distribution, with different means in each of the two study groups. Linear mixed models (SAS PROC MIXED) were used to compare the size of lesions between the treatment groups. Predictors in both patient-level and polyp-level models included clinical site, age, sex, race, smoking status, total number of previous colonoscopies, and number of months since most recent colonoscopy. Descriptive statistics are reported as mean ± SD.

A total of 50 subjects completed the study. All subjects had a history of at least one prior colonoscopy with adenomatous polyps and/or colorectal cancer. After the first routine colonoscopy was completed and the cecum was reached during the second exam, 27 subjects were randomized to undergo chromoendoscopy and 23 subjects were randomized to intensive inspection colonoscopy without dye spraying. The discrepancy in the numbers of subjects in the two study arms was the result of the blocked randomization by study center. The baseline characteristics of subjects are shown in Table 1. There were no statistically significant differences between the two arms with respect to baseline subject characteristics.

Before randomization, all subjects underwent an initial conventional colonoscopy. The average procedure time (from insertion of colonoscope to removal, minus time spent in removal of polyps) was 20.8 minutes. Twenty-five of 50 (50%) subjects had polyps on the first exam. Of the 78 lesions biopsied, 40 (51%) were adenomatous polyps, 21(27%) were hyperplastic polyps, 15 (19%) were normal tissue, 1 (1%) was a fragment of an adenoma, and 1 was patchy active colitis.

The characteristics of the first colonoscopy procedures are presented in Table 2 by randomization arm. There was no statistically significant difference in procedure time between subjects subsequently randomized to intensive inspection versus chromoendoscopy for their second exam, or in the number of subjects who had one or more polyps on the first exam (14 versus 11, respectively). Although there were no significant differences by arm in the total number of adenomas detected during the first standard colonoscopy exams (21 found in subjects in intensive inspection arm versus 19 in chromoendoscopy arm), more of the subjects with adenomas on the first standard exam were randomized to undergo intensive inspection for their second exam (11 versus 6 assigned to chromoendoscopy, P = 0.08).

The characteristics of polyps found on the first exam are presented in Table 3 and the polyp locations are shown in Fig. 1. There was no statistically significant variation by arm in the sizes or location of polyps removed in the first colonoscopy nor were there significant differences in distributions of adenomas, hyperplastic polyps, and normal biopsies. Most adenomas found on the first exams were removed from the right colon proximal to the splenic flexure. Fourteen of 40 (35%) of the adenomas found on first exams were considered flat, with 12 of these found in the intensive inspection arm (P = 0.002). There was no relationship between procedure time of the first colonoscopy and number of adenomas detected (Fig. 2).

Twenty-three subjects were randomized to intensive inspection without dye spraying for their second colonoscopy, with a mean procedure time of 27.3 ± 6.2 minutes (range 19-42 minutes).

During the exams with intensive inspection, lesions were biopsied from 12 of 23 (52%) subjects. Eight subjects had polyps and four had adenomas (Table 2). Of the total of 12 lesions biopsied, 5 (41.5%) were adenomatous polyps, 5 (41.5%) were hyperplastic polyps, and 2 (17%) were normal tissue (Table 3).

The location of adenomatous polyps found on second colonoscopy with intensive inspection is shown in Fig. 1. Three of the 5 (60%) adenomas were considered flat. There was no association between procedure time of the intensive inspection colonoscopy and the numbers of polyps and adenomas identified (Fig. 2). Of 14 subjects with adenomas discovered at either the first routine or second intensive colonoscopy, 3 (21%) subjects had adenomas found only on the second colonoscopy. Adenomas removed during the intensive inspection colonoscopy were not significantly smaller than those obtained during the first standard colonoscopy (mean size 3.20 ± 0.84 mm versus 3.57 ± 3.25 mm, respectively; Table 3).

Twenty-seven subjects were randomized to chromoendoscopy, with an average procedure time of 36.9 ± 14.5 minutes (range 15-86 minutes). During the chromoendoscopy examination, lesions were biopsied from 19 of 27 (70%) subjects. Seventeen subjects had polyps and 12 had adenomas (Table 2). Of the total of 57 lesions biopsied, 19 (32%) were adenomatous polyps, 16 (27%) were hyperplastic polyps, 2 were mucosal polyps, 1 was patchy active colitis, and 22 (37%) were normal tissue (Table 3).

Of the 19 adenomas found during chromoendoscopy exams, 16 (84%) were removed from the right colon, 1 (5%) from the left colon, and 2 (11%) from the rectum (Fig. 1). Seven (37%) of the adenomas were considered flat and six of the flat adenomas were located in the right colon. There was no association between the duration of the chromoendoscopy procedure and the numbers of polyps and adenomas identified during the exam (Fig. 2).

Of 13 subjects randomized to chromoendoscopy who had adenomas discovered at either colonoscopy, 7 (54%) had adenomas found only during the second exam. Adenomas obtained during chromoendoscopy were significantly smaller than those obtained during the first examination (mean size 2.66 ± 0.97 mm versus 4.68 ± 3.59 mm, respectively; Table 3).

Chromoendoscopy took significantly longer than intensive inspection, with an average procedure time of 36.9 ± 14.5 versus 27.3 ± 6.2 minutes, respectively (P < 0.01). Subjects randomized to chromoendoscopy had more biopsies on their second exams (2.4 ± 2.3 biopsies per subject compared with 0.7 ± 0.8 for intensive inspection), and chromoendoscopy detected more hyperplastic polyps (16 versus 4 in intensive inspection) and more adenomas (19 versus 4 in intensive inspection; Tables 2 and 3). Although chromoendoscopy exams yielded 22 biopsies that were normal tissues, the percentages of biopsies that were normal tissues were similar in the standard colonoscopy, intensive inspection, and chromoendoscopy examinations. There were no adverse events reports reported for any of the 50 subjects.

Twelve of 27 (44%) subjects in the chromoendoscopy arm and 4 of 23 (17%) in the intensive inspection arm had additional adenomas found during the second colonoscopy (Table 2). Overall, 24 of 64 (38%) adenomas were found on second exams. The adenomas detected on the second exams were significantly smaller than those removed during the first colonoscopy; however, the size did not significantly differ between the intensive inspection and chromoendoscopy arms (Table 3). None of the polyps detected during the second colonoscopies had features of high-grade dysplasia.

In multivariate analysis controlling for procedure time and study center, the use of chromoendoscopy was significantly associated with a greater likelihood of finding one or more additional adenomas on the second exam than intensive inspection (P = 0.04).

We designed this randomized trial of back-to-back colonoscopies to determine whether chromoendoscopy is better than intensive inspection without dye spraying for detecting small adenomatous lesions that might be missed during a routine colonoscopy. We found that chromoendoscopy doubled the adenoma yield after a standard colonoscopy and detected significantly more adenomas than intensive inspection exams done without using dye. Chromoendoscopy identified additional adenomas in 44% of subjects and changed management for 26% of subjects (n = 7) who would have been misclassified as “adenoma-free” after the first standard colonoscopy. Although chromoendoscopy exams lasted nearly 10 minutes longer than exams using intensive inspection without dye spraying, after controlling for procedure time through study design and multivariate analysis, our data support that the increase in adenoma detection seen with chromoendoscopy is independent of inspection time.

Overall, 38% of adenomas found in our subjects were detected on the second exams, suggesting that a single conventional colonoscopy may miss ∼1 of every 3 adenomas present. The conventional exam missed half of the total adenomas in subjects who underwent chromoendoscopy, a much higher miss rate than the 26% to 27% for adenomas 1 to 5 mm previously reported in studies of tandem exams using conventional colonoscopy (4, 5). Most of the adenomas found in our second exams were small (<5 mm) and none met definitions for advanced adenomas based on size or histology. Even so, 75% (18 of 24) of the missed adenomas were located in the right colon and 42% (10 of 24) had a flat morphology—characteristics that may be associated with a more aggressive natural history (18-20).

Several recent European studies have also reported that chromoendoscopy detects more, albeit diminutive, colorectal adenomas (11, 12, 17). Although our findings are similar, it is important to note that our study is the first multicenter North American trial to examine the utility of chromoendoscopy for adenoma detection. Furthermore, our study design was unique in its use of randomized tandem colonoscopies to compare chromoendoscopy to a time-intensive conventional colonoscopy control. Our findings support that dye spraying improves adenoma yield and that results using a standard chromoendoscopy technique are generalizable.

We acknowledge that our study has several limitations. The chromoendoscopy exams lasted, on average, 10 minutes longer than the intensive inspection exams. Recent reports have shown an association between inspection time and adenoma detection rates (21), and in designing our study we included the intensive inspection arm to control for time, considering that a 20-minute inspection would greatly exceed the threshold of 6 minutes recommended by expert opinion (22). In examining our data, we found no association between procedure time and number of adenomas detected and after controlling for time in the multivariate analysis the effect of chromoendoscopy remained statistically significant. This supports our conclusion that the dye spraying, and not longer inspection time, is responsible for the higher sensitivity of chromoendoscopy for adenoma detection.

This was a small study and, despite blinded randomization, there were differences among subjects by randomization arm. However, none of these differences were statistically significant and they are unlikely to completely explain the increased adenoma yield with chromoendoscopy. Although endoscopists could not be truly blinded to procedure type, they were not aware of which randomization arm had been assigned until after the first colonoscopy was completed, and there were no differences in procedure characteristics of the first colonoscopy by randomization arm (procedure time, number of biopsies) to suggest differential bias in adenoma detection.

We did all of the exams using standard colonoscopes, rather than high-definition or magnification colonoscopes, because we believed this technique would be more exportable to other clinical practice settings. However, there are data showing that high-definition/magnification endoscopes increase sensitivity of chromoendoscopy; thus, our results may underestimate the effect of chromoendoscopy. We recognize that because all study subjects had prior history of colorectal cancer or adenomas, the adenoma yield of chromoendoscopy would probably be lower in an average-risk population.

Our results show that chromoendoscopy improves detection of colorectal adenomas missed by conventional colonoscopy independent of inspection time. Our findings from this North American multicenter study are consistent with reports from randomized trials conducted in European centers with expertise in chromoendoscopy (11, 12, 17). However, all but one of those studies (11) concluded that because most of the adenomas detected by chromoendoscopy were small (<5 mm), there was insufficient evidence to support the routine use of chromoendoscopy in the clinical setting. Still, studies indicate that flat adenomas, which can be difficult to see during conventional white light colonoscopy, are 10 times more likely than polypoid lesions to contain invasive carcinoma (20), and 7% to 15% of small adenomas (5-10 mm) show advanced histology (18, 19). Although it is believed that cancers that arise in the interval between colonoscopic exams may be the result of missed lesions, we currently have no way of knowing if any of the additional lesions detected by chromoendoscopy would be clinically significant.

Recent reports have recommended that chromoendoscopy should be used for routine screening for flat neoplasia in patients with ulcerative colitis (9, 23–25). As there are limited data regarding the natural history of small flat adenomas, the utility of chromoendoscopy or other new endoscopic modalities for colorectal cancer screening in average and moderate-risk individuals will ultimately depend on the biological significance of small flat lesions missed by conventional colonoscopy.

No potential conflicts of interest were disclosed.