Dysplasia detection is similar between chromoendoscopy and high-definition white-light colonoscopy in inflammatory bowel disease patients: a US-matched case-control study

Abstract

Introduction

Inflammatory bowel disease (IBD) patients are at greater risk of developing colorectal cancer (CRC). Detection of precursor dysplasia is important for cancer prevention. Recent guidelines recommend dye chromoendoscopy (DCE) as the preferred method for dysplasia detection. Universal adoption of DCE is time-consuming and may limit endoscopy access. The benefit of universal application of the guidelines is unclear. We compared high-definition white-light colonoscopy (HD-WLC) with DCE for dysplasia detection in IBD patients.

Methods

We conducted a retrospective case-control study of adult IBD patients undergoing dysplasia surveillance between September 1, 2015, and February 1, 2020. DCE cases were matched to HD-WLC in a 1:1 ratio for gender, IBD diagnosis, and age. DCE patients were considered high risk for colorectal cancer by the referring provider.

Results

A total of 187 subjects were enrolled. Majority were males, were Caucasian, and had longstanding IBD (primarily ulcerative colitis). Baseline characteristics were similar between the two groups, except for history of surgery, duration of IBD, and history of dysplasia. There was no significant difference in dysplasia detection between DCE and HD-WLC (10.2% vs 6.7%, p = 0.39). More polyps were detected in the DCE arm compared with the HD-WLC group (1.35 vs 0.80, p = 0.018), but adenoma detection rate was not statistically different between the two groups (10.2% vs 9.0%, p = 0.31). Mean withdrawal time was longer in the DCE group (24.6 min vs 15.4, p < 0.001).

Conclusions

There were no differences in dysplasia detection using DCE compared with HD-WLC, although withdrawal times were longer with DCE.

Introduction

Patients with IBD are at a substantially greater risk of developing colorectal cancer compared with the general population. The pooled standard incidence ratio for CRC in patients with inflammatory bowel disease is 1.7 (95% CI, 2.9–17.8) based on a meta-analysis by Lutgens et al. in 2013 [1]. There is a reported increased risk of 2–5-fold in patients with ulcerative colitis (UC) [2] and 3-fold among patients with ileocolonic Crohn’s disease [3]. Risk factors for cancer development in this cohort include the duration and extent of the disease, extent of active inflammation, family history of sporadic CRC, anatomic abnormalities such as pseudopolyps, and presence of primary sclerosing cholangitis [4,5]. Colorectal cancer in IBD is predominantly derived from dysplasia as part of the dysplasia–adenocarcinoma sequence [6]. As such, accurate detection of precursor dysplasia is paramount for early CRC detection.

There has been significant debate about the optimal strategy for dysplasia detection in IBD patients. Detection has traditionally relied on both examination of the mucosa using standard-definition colonoscopy with targeted biopsies of visible lesions and random biopsies to identify invisible dysplasia [7–10]. With the advent of better optic colonoscopes and endoscopic accessories, this technique has fallen out of favor. More recently, Watanabe et al. demonstrated that the strategy of targeted biopsies was more cost-effective with similar rates of neoplasia detection compared with random biopsies [11].

Dye chromoendoscopy (DCE) involves the topical application of dyes (indigo carmine or methylene blue) to improve identification and delineation of surface abnormalities during endoscopy. A study by Kiesslich et al. in 2003 was one of the first to examine its role in an IBD cohort which included 263 UC patients [12]. That study noted a 3.2-fold increase in the number of intraepithelial neoplasias compared with standard-definition white-light colonoscopy (SD-WLC) [12]. Since then, high-definition endoscopes (with improved optics) have been incorporated into endoscopy practice, resulting in improved characterization of lesions compared with standard-definition colonoscopy [13]. The SCENIC international consensus statement in 2015 suggested DCE over high-definition white-light colonoscopy (HD-WLC) for surveillance of dysplasia in IBD (conditional recommendation; low-quality evidence) based on a single observational study [14]. Additionally, a study of 75 IBD patients by Picco et al. showed that dysplasia detection was increased with DCE by more than 2-fold (21% vs 9%, respectively) as compared with HD-WLC [15].

Since then, several other studies have been published evaluating the efficacy of DCE for detection of dysplasia in IBD patients. In a systematic review published in 2017 by Iannone et al. [16], 10 randomized trials were identified since the SCENIC guidelines, comparing DCE with other endoscopic techniques. Three of these included DCE versus HD-WLC and found no significant differences in identifying dysplasia between the two groups [17–19].

Furthermore, the practicality of universally adopting DCE in the real-world practice setting has not been fully evaluated [20]. Implementation of DCE into practice requires a learning period as well as additional endoscopy time and resources, and, as a result, has a potential impact on endoscopic access. We aimed to compare the use and evaluate the outcomes of HD-WLC versus DCE for dysplasia and polyp detection in IBD patients at increased risk for CRC.

Materials and methods

Study design and participant characteristics

We conducted a retrospective case-control study of adult IBD patients undergoing dysplasia surveillance by IBD-specialized gastroenterologists at a single US tertiary care academic center from September 1, 2015, to February 1, 2020. Exclusion criteria included age less than 18 years, patients presenting for any other indication aside from colorectal cancer surveillance, history of total abdominal colectomy, or patients who had incomplete procedures. Subjects were categorized into two groups: [1] those undergoing DCE for dysplasia surveillance and [2] controls undergoing surveillance with HD-WLC. Cases undergoing DCE were referred due to being perceived as high risk for dysplasia by the requesting provider. Reasons for referral included (a) history of low- or high-grade dysplasia on previous biopsy specimens, (b) history of indefinite for dysplasia on previous biopsies, (c) previous diffuse pseudopolyps, and (d) advanced adenoma and/or sessile (e) serrated adenoma with longstanding IBD. Cases were matched to corresponding controls (i.e., IBD patients undergoing HD-WLC) in a 1:1 ratio for gender, IBD diagnosis, and age within 3 years. All colonoscopies were performed by IBD specialists with training in DCE, using HD colonoscopes (CF-HQ190L or PCF-H190DL, Olympus co., Tokyo, Japan). Methylene blue was used in 95.7% of cases with DCE (with the rest of cases used indigo carmine). Biopsy specimens were reviewed by board-certified gastrointestinal pathologists at the institution. Pathologic diagnoses were categorized into the following designations: negative for dysplasia, indefinite for dysplasia, low-grade dysplasia, or high-grade dysplasia. Baseline characteristics including IBD diagnosis, years of IBD diagnosis, family history of colon cancer, associated primary sclerosing cholangitis (PSC), history of colon resection, known history of dysplasia, quality of bowel preparation, type of DCE contrast used, time of the procedure (morning or afternoon), extent of disease activity, polyp characteristics, and mean withdrawal time were recorded.

Institutional review board approval was obtained prior to the initiation of the study.

Outcomes

The primary outcome was dysplasia detection rates in the two groups. Dysplasia detection rate was defined as the number of cases with either low-grade or high-grade dysplasia found divided by the number of colonoscopies. Secondary outcomes included polyp detection and mean withdrawal times.

Statistics

Univariate analysis using the chi-square test was used for independent variables, and Kruskal–Wallis and Wilcoxon rank sum tests were used to compare groups. Statistical tests for significance were two-sided, and a significance level p value ≤ 0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 25.0 (Armonk, NY; IBM Corp).

Results

One hundred eighty-seven subjects were enrolled (n = 98 DCE, n = 89 WLC), with a mean age of 54.6 years (IQR 52.5–56.7). Baseline characteristics were generally similar when comparing cases with controls (Table 1) except for history of small-bowel/colon surgery (which was seen in a higher proportion in the HD-WLC group), duration of IBD (higher proportion in the DCE group), and history of dysplasia (higher proportion in DCE). The majority were males (68.4%) and Caucasian (95.6%). Most of the subjects in our study had UC (68.4%), and 114 (60.9%) had longstanding IBD (defined as more than 8 years prior to the time of colonoscopy of interest). Eleven (5.9%) subjects had a family history of colorectal cancer, and 17 (9.1%) carried a concomitant diagnosis of PSC. Thirty six of the 187 (19.2%) subjects had a history of dysplasia (low-grade or high-grade), and 17 (9.1%) had a history of indefinite for dysplasia on previous biopsies. In the DCE group, 36.9% of cases were referred for a history of dysplasia on previous colonoscopy, followed by 35.2% of cases being referred for a history of indefinite for dysplasia on previous colon biopsy. There was a higher proportion of subjects in the DCE group with longer-standing disease (68.4% vs 52.8%; p = 0.02), while the HD-WLC group had an increased proportion with a history of colonic surgery (23.9% vs 11.2%; p = 0.03). There was a difference in the extent of active disease at the time of the procedures between the two groups, with the control HD-WLC group having significantly more extensive/proximal colonic disease (27% vs 10.2%; p = 0.003), while the DCE group had a higher proportion of subjects in remission (61.2% vs 41.6%; p = 0.007). A higher proportion of cases in the DCE group had no active disease compared with HD-WLC (61.2% vs 41.6%), and more cases in the HD-WLC group had proximal disease (27% vs 10.2%). There was also a significant difference in the classes of medications for maintenance therapy between the two groups. Subjects who were on immunomodulators (as primary therapy) were seen in higher proportions in the DCE group in comparison with HD-WLC (21.4% vs 10.1%). The rate of anti-TNF agent use for maintenance was higher in the control group (32.6% vs 18.4%).

Table 1.

Baseline characteristics of IBD subjects undergoing CRC surveillance with dye chromoendoscopy or high-definition white-light colonoscopy

	Dye chromoendoscopy (n = 98)	HD-WLC (n = 89)	p value
Age at time of procedure, years	55.4 (52.3–58.4)	53.7 (50.7–56.7)	0.41
Male gender, n (%)	68 (69.4%)	60 (67.4%)	0.77
Ethnicity, n (%)
Caucasian	92 (93.8%)	84 (95.4%)	0.42
African American	1 (1.0%)	1 (3.4%)
Asian	3 (3.1%)	1 (1.1%)
Hispanic	2 (2.0%)	2 (2.2%)
Other	0 (0%)	1 (1.1%)
Ulcerative colitis (%)	69 (70.4%)	59 (66.3%)	0.55
Crohn’s disease (%)	29 (29.6%)	30 (34.1%)	0.51
Duration of IBD > 8 years (%)	67 (68.4%)	47 (52.8%)	0.02
PSC	12 (12.2%)	5 (5.7%)	0.09
History of small-bowel/colon surgery	11 (11.2%)	21 (23.9%)	0.03
Family history of colorectal cancer	6 (6.1%)	5 (5.7%)	0.74
History of dysplasia			< 0.001
No	49 (50%)	69 (77.5%)
Indefinite	12 (12.2%)	5 (5.6%)
Yes	31 (31.6%)	7 (7.9%)
Unknown	6 (6.1%)	8 (8.9%)
Extent of active disease			0.035
None	60 (61.2%)	37 (41.6%)
Distal (rectum to splenic flexure)	19 (19.4%)	15 (16.9%)
Proximal (beyond splenic flexure)	10 (10.2%)	24 (27.0%)
Terminal ileum only	5 (5.1%)	4 (4.5%)
Maintenance therapy			< 0.001
None	14 (14.3%)	10 (11.2%)
Mesalamines	35 (35.7%)	30 (33.7%)
Immunomodulators	21 (21.4%)	9 (10.1%)
Anti-TNF agents	18 (18.4%)	29 (32.6%)
Vedolizumab	3 (3.1%)	7 (7.9%)
Ustekinumab	2 (2.0%)	1 (1.1%)
Tofacitinib	0 (0%)	2 (2.2%)

Sixteen of the 187 subjects (8.6%) had dysplasia. A majority of the cases with dysplasia (62.5%) had longstanding IBD. None of the subjects with dysplasia also had PSC. One of the 16 cases (6.2%) had a family history of colorectal cancer. In the DCE group,10 of the 98 subjects (10.2%) had dysplasia detected—3 of whom did not have a prior history of dysplasia, 1 with a history of indefinite dysplasia, 5 with a prior history of dysplasia, and 1 with an unknown history of dysplasia. In comparison, 6 of the 89 subjects (6.7%) in the control group (HD-WLC) had dysplasia detected—4 of whom had no prior history of dysplasia and 2 with a prior history of dysplasia. There was no significant difference in dysplasia detection when comparing DCE cases with HD-WLC controls (p = 0.39) on univariate analysis (Table 2). On multivariate analysis (which adjusted for age, gender, duration and type of IBD, and history of dysplasia), dysplasia detection with DCE was not statistically different when compared with HD-WLC (OR 0.91; 95% CI (0.15–5.67); p = 0.92) (Table 3).

Table 2.

Univariate analysis of dysplasia detection with dye chromoendoscopy vs high-definition white-light colonoscopy

	DCE (n = 98)	HD-WLC (n = 89)	p value
No dysplasia detected, n (%)	81 (82.7%)	78 (87.6%)	0.07
No previous history	46	68
History of indefinite dysplasia	10	3
History of dysplasia	20	4
Unknown	5	3
Indefinite for dysplasia	7 (7.1%)	1 (1.1%)	0.07
No previous history	0	1
History of indefinite dysplasia	1	0
History of dysplasia	6	1
Total dysplasia detected, n (%)	10 (10.2%)	6 (6.7%)	0.38
No previous history	3	4
History of indefinite dysplasia	1	0
History of dysplasia	5	2
Unknown	1	0
Unavailable data	0	4

Table 3.

Multivariate analysis of dysplasia detection with dye chromoendoscopy vs high-definition white-light colonoscopy

Variable	OR (95% CI)	p value
Dysplasia with DCE (compared with WLC)	0.91 (0.15–5.67)	0.92
Longstanding disease (duration > 8 years)	1.69 (0.45–6.44)	0.44
Age	1.08 (1.02–1.14)	0.01
Female gender	2.19 (0.63–7.57)	0.22
Crohn’s disease (compared with UC)	0.26 (0.03–2.18)	0.21
History of dysplasia vs no dysplasia	2.84 (0.56–14.42)	0.05
History of indefinite for dysplasia vs no dysplasia	0.69 (0.07–7.11)	0.35

The mean number of polyps detected overall was 1.09 (95% CI 0.78–1.39). More polyps were detected in the DCE cohort compared with the HD-WLC cohort per colonoscopy (1.35 vs 0.80, p = 0.018). In the DCE cohort, 25 of the 42 polyps (59.5%) were considered precancerous (defined as tubular adenoma, sessile serrated lesion, or dysplastic), while 20 of the 27 polyps (74%) were precancerous in the HD-WLC group (p = ?). There was no statistically significant difference in adenoma detection rate, with one or more adenomas found during 10 of the cases in the DCE group, compared with 8 in the HD-WLC group (10.2% vs 9.0%, p = 0.31). The single case of adenocarcinoma discovered on polyp pathology was in the DCE group.

Mean withdrawal time was significantly longer in the DCE group, by an average of 8 additional minutes (24.6 min vs 15.4 min, p < 0.001).

Discussion

Our study demonstrated no difference in dysplasia detection rates between DCE and HD-WLC. This finding is particularly noteworthy considering that IBD disease duration (an established risk factor for dysplasia and CRC in IBD) and mean withdrawal time (a primary determinant of colonic polyp detection) were both longer in our DCE cohort (Table 4). To our knowledge, this is the largest US-based study to have evaluated the clinical application of guidelines for colon cancer screening in IBD patients comparing the two modalities for dysplasia detection in a tertiary care IBD population.

Table 4.

Univariate analysis of polyp detection with dye chromoendoscopy vs high-definition white-light colonoscopy

	DCE	WLC	p value
Polyps detected, n (95% CI)	1.35 (0.86–1.85)	0.80 (0.44–1.15)	0.018
Normal colon mucosa	8	1
Hyperplastic	13	4
Inflammatory	4	3
Tubular adenoma	10	8
Sessile serrated adenoma	9	5
Dysplasia (low-grade or high-grade)	6	7
Adenocarcinoma	1	0

These findings indicate that there is insufficient evidence to support the general implementation of DCE for surveillance of colorectal cancer in IBD patients, as recommended by the SCENIC international consensus statement in 2015 [14]. While multiple studies have been published since the SCENIC guidelines, literature pertaining to optimal surveillance methods to detect and manage dysplasia in the IBD population remains sparse, with limited longitudinal follow-up. Our findings support the notion that HD-WLC is adequate for surveillance in the general IBD population and is just as effective in dysplasia detection as DCE, even when considering the fact that a significant proportion of this study population had additional risk factors for colorectal cancer (including having persistent active inflammation, personal history of dysplasia, family history of sporadic CRC, and/or concomitant PSC).

As previously indicated, mean withdrawal time was significantly longer in the DCE group compared with the control arm, by an average of 9.2 min (Table 5). This is similar to the results noted in the meta-analysis published by Subramanian et al. [21]. The additional time required for DCE is important because it has significant impacts on resource utilization, including access to colonoscopy at the performing facility, anesthesia, and other staffing costs. Additionally, while there were no adverse events reported in our DCE cohort, the use of methylene blue has been associated with methemoglobinemia in patients with glucose-6-phosphate-dehydrogenase deficiency and serotonin syndrome in patients on long-term serotonin reuptake inhibitor therapy [22].

Table 5.

Unadjusted univariate analysis of mean withdrawal time with dye chromoendoscopy vs high-definition white-light colonoscopy

	DCE	WLC	p value
Mean withdrawal time, min (95% CI)	24.6 (22.4–26.8)	15.4 (13.4–17.4)	< 0.001

Our study showed an increase in polyp detection with patients undergoing DCE when compared with those undergoing HD-WLC. It is unclear if this is the result of the longer withdrawal times. However, adenoma detection rate was similar between our two study cohorts. This finding supports the assertion that the use of DCE did not alter plan of treatment in this cohort since the lesions were not clinically significant (i.e., precancerous or cancerous). These findings were consistent with the results of the systematic review conducted by Iannone et al. [16].

The referred cohort had only a limited number of patients on 5’ASA medication. This is likely explained by a referral bias in that sicker patients with prolonged duration of disease are more likely to have been referred and may have progressed from 5’ASA medications to escalated therapy.

The data on potential chemopreventive benefits of 5’ASA medications for colorectal cancer in IBD patients is mixed. Results have been confounded by study designs. A large observational study by Terdiman et al. [23] did not show any benefit. However, a more recent systemic review with a meta-analysis of 26 observational studies of over 15,000 patients reported a chemopreventive effect at a maintenance dosage of ≥ 1.2 g/day [23,24].

Optimal management of dysplasia found on surveillance colonoscopy in patients with IBD is multifaceted and should consider several factors including extensive discussion between patient and provider. Grade of dysplasia, polypoid or flat dysplasia, presence or absence of associated inflammation which may make pathology difficult to interpret, multifocal or unifocal areas of dysplasia, and risk for surgery of the patient are some of the factors that should be considered. Ultimately, the decision to continue intensive surveillance, colectomy, or endoscopic therapy should be made after extensive discussion of the above factors in a manner consistent with shared decision-making.

As with all retrospective studies, our study may suffer from bias due to missing data. Although it included a cohort of IBD patients undergoing DCE similar to previous studies, the small dataset is attributed to the limited use of DCE. However, our cohort of IBD patients undergoing DCE was similar in sample size to previous studies. Our study is also limited by being from a single-center specialist IBD clinical practice and may not be generalizable or representative of the general IBD population. However, several of the patients included in our cohort were referred from outside practitioners. In addition, as previously indicated, many of the referred patients were deemed “high risk” for CRC (implying that they should have demonstrated an increased rate of dysplasia and/or cancer).

Future studies may benefit from investigation of DCE use under particular circumstances (i.e., high-risk subpopulations within IBD patients). Examples would include patients with simmering persistent inflammation, personal history of dysplasia, family history of sporadic CRC, and presence of primary sclerosing cholangitis [4,5]. Larger prospective, multicenter trials are needed to further evaluate this comparison and its clinical application with a particular focus on resource utilization, clinical utility, and cost benefit. In addition, longer-term, longitudinal follow-up may be required with serial DCE examinations.

Conclusions

We found no difference in dysplasia detection in IBD patients using DCE when compared with HD-WLC despite greater polyp detection rates and increased withdrawal times.

To our knowledge, this is the largest real-world experience in the USA comparing DCE with HD-WLC for surveillance of dysplasia and/or colorectal cancer in patients with IBD. Our findings need to be confirmed by large-scale prospective studies to determine best endoscopic practices for dysplasia and colorectal cancer surveillance in patients with IBD.

Study highlights.

1. What is current knowledge

   • Accurate detection of dysplasia is crucial for early colorectal cancer detection in patients with inflammatory bowel disease.
   • Guidelines have suggested dye chromoendoscopy as the preferred method for dysplasia detection; however, its benefit in a real-world setting with challenges in implementation remains unknown.
2. What is new here

   • Our study shows that there is no difference in dysplasia detection between chromoendoscopy and high-definition white-light colonoscopy.
   • These findings support the growing evidence that implementing universal chromoendoscopy for dysplasia surveillance is costlier and time-consuming with little benefit in everyday IBD clinical care.