Advances in endoscopic dysplasia detection in inflammatory bowel disease

Abstract

BACKGROUND

Dysplasia surveillance in inflammatory bowel disease (IBD) has evolved significantly with the adoption of advanced endoscopic technologies.

AIM

To synthesize evidence on image-enhanced endoscopy techniques, biopsy protocols, and surveillance practices optimizing dysplasia detection in IBD.

METHODS

A scoping review was conducted following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. A comprehensive search of PubMed and EMBASE from inception to June 2025 identified studies reporting on endoscopic dysplasia detection or characterization in IBD. Forty-five studies were included for qualitative synthesis, covering dye-based chromoendoscopy (DCE), virtual chromoendoscopy (VCE), confocal laser endomicroscopy, artificial intelligence-based tools, and panoramic endoscopy.

RESULTS

DCE consistently demonstrated high dysplasia detection rates, especially when indigo carmine was used, and enabled accurate pit-pattern-based lesion characterization. High definition (HD) DCE may offer procedural benefits over HD white light endoscopy (WLE), though superiority in dysplasia detection remains inconsistent across studies. In comparative studies, DCE outperformed or matched white-light approaches, with higher dysplasia yield in selected trials (e.g., 9.7% vs 1.9%, P = 0.004), while HD WLE with segmental re-inspection was non-inferior to DCE in expert settings. Virtual chromoendoscopy modalities such as i-SCAN and narrow band imaging showed comparable performance to DCE in several trials, with artificial intelligence-assisted computer-aided detection systems demonstrating equivalent sensitivity but lower specificity. Studies comparing biopsy protocols revealed that targeted biopsies under image-enhanced endoscopy, particularly DCE, were generally superior or equivalent to random biopsies, with random sampling offering marginal benefit in select high-risk subgroups. Multimodal imaging and panoramic endoscopy further improved dysplasia yield in challenging cases. Cost-effectiveness analyses favored DCE over WLE, and long-term surveillance data confirmed declining colorectal cancer rates with high-quality endoscopic programs. However, real-world practice audits revealed substantial variation in surveillance quality and guideline adherence.

CONCLUSION

Image-enhanced targeted surveillance - particularly using DCE or validated virtual platforms - has improved dysplasia detection in IBD and may allow for a reduction in random biopsies. Despite technological advancements, major quality gaps and interobserver variability persist in clinical practice. Standardized training, quality benchmarks, and cost-effective implementation of advanced endoscopic techniques are needed to optimize colorectal cancer prevention in IBD.

Key Words: Colitis-associated neoplasia; Dye-based chromoendoscopy; Virtual chromoendoscopy; Narrow band imaging; White light endoscopy; High-definition endoscopy; Dysplasia; Biopsy

Core Tip: Dye-based chromoendoscopy (DCE), particularly with indigo carmine, provides high dysplasia detection rates and accurate lesion characterization in inflammatory bowel disease. While high-definition DCE may offer procedural advantages over high-definition white light endoscopy, detection superiority remains inconsistent. Virtual chromoendoscopy shows comparable dysplasia detection to DCE in several trials, with added efficiency and reduced procedure time. Targeted biopsies under image-enhanced endoscopy outperform or match random biopsies, with limited added value from random sampling except in select high-risk patients. Multimodal imaging and panoramic endoscopy enhance dysplasia yield. Despite strong evidence, real-world surveillance quality varies widely, highlighting the need for standardization.

INTRODUCTION

Colorectal cancer (CRC) remains a major cause of morbidity and mortality in patients with inflammatory bowel disease (IBD), particularly those with longstanding ulcerative colitis (UC) or colonic Crohn’s disease. Dysplasia is the most established precursor to CRC in this population, and its timely detection is a key goal of surveillance colonoscopy. Advances in endoscopic imaging have enabled earlier identification and characterization of neoplastic lesions, potentially reducing the need for colectomy and improving patient outcomes. However, the optimal strategy for dysplasia surveillance - balancing accuracy, efficiency, cost, and accessibility - remains an evolving challenge in both clinical practice and guideline development[1]. Over the past two decades, several image-enhanced endoscopy (IEE) techniques - including dye-based chromoendoscopy (DCE), virtual chromoendoscopy (VCE), confocal laser endomicroscopy (CLE), and, more recently, artificial intelligence (AI)-assisted platforms - have been evaluated for dysplasia detection in IBD. While DCE is endorsed by major societies as the preferred surveillance tool, its real-world adoption has been inconsistent[2]. Furthermore, the comparative utility of different biopsy strategies (random vs targeted), the role of newer modalities like panoramic endoscopy, and practical issues such as cost-effectiveness and interobserver variability remain incompletely understood[3]. Previous studies have yielded conflicting results, with some showing marginal or no benefit of advanced imaging over high-definition (HD) white light endoscopy (WLE), particularly outside of expert centers.

This scoping review aims to synthesize and map the breadth of evidence surrounding endoscopic dysplasia detection and characterization in IBD. By comparing technologies, biopsy protocols, characterization tools, and implementation practices, the review seeks to clarify areas of consensus, highlight gaps in current knowledge, and inform best practices for surveillance. The findings can guide future research priorities and support the development of standardized, high-quality, and cost-effective surveillance strategies for CRC prevention in IBD. In addition to surveillance colonoscopy, pharmacologic prevention strategies targeting inflammation-driven carcinogenesis - such as mesalazine, and agents modulating the cyclic guanosine monophosphate adenosine monophosphate synthase-stimulator of interferon genes (STING) stimulator of interferon genes pathway (e.g., STING agonists like 5,6-dimethylxanthenone-4-acetic acid or inhibitors like H-151) - may complement endoscopic surveillance to mitigate IBD-associated CRC risk[4].

MATERIALS AND METHODS

This scoping review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews 2020 guidelines. A comprehensive literature search of PubMed and EMBASE was performed from inception (1966) to June 2025 using the following search strategy: (“Inflammatory Bowel Diseases” OR “Ulcerative Colitis” OR “Crohn Disease”) AND (“Dysplasia” OR “Colorectal Neoplasms”) AND (“Chromoendoscopy” OR “Narrow Band Imaging” OR “Confocal Microscopy” OR “Virtual Chromoendoscopy” OR “Virtual chromoendoscopy”). Eligible studies included original articles that reported on endoscopic dysplasia detection and characterization. Only English-language articles were included. Case reports, conference abstracts, animal studies, preclinical, and articles unrelated to dysplasia detection/characterization were excluded. A total of 1453 records were identified (PubMed: 337; EMBASE: 1116), and after removing duplicates, 1254 records underwent title and abstract screening. Of these, 1142 were excluded, and 112 full-text articles were reviewed. Ultimately, 45 studies met the inclusion criteria (Figure 1). Two independent reviewers (Pal P and Kata P) screened all titles, abstracts, and full-text articles for eligibility. In cases of uncertainty or disagreement, a third reviewer was consulted to reach a consensus. Data were extracted on study design, patient population, detection/characterization technique, and outcomes.

Figure 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 checklist.

RESULTS

IEE for dysplasia detection in IBD

Various advanced endoscopic techniques have significantly improved the detection of dysplasia in patients with IBD (Figure 2 and Table 1). DCE using indigo carmine has been shown to have a high dysplasia detection rate of 24% in real-life settings and 10.5% in community practice, with targeted biopsies yielding substantially more dysplasia than random ones[5,6]. CLE, when combined with DCE, demonstrated a modest dysplasia detection rate (9.8%) but limited sensitivity and practicality in daily surveillance due to equipment challenges[7]. Narrow band imaging (NBI), especially with magnification, allows visualization of vascular and mucosal patterns, with tortuous patterns being predictive of dysplasia and achieving 80% sensitivity and 84% specificity[8]. NBI-based assessment of mucosal vascular patterns further correlates with epithelial proliferation indices like Ki-67, making it a valuable non-invasive marker for identifying high-risk mucosa[9]. Autofluorescence imaging (AFI), though limited in detecting visible dysplasia (only 37.5% appeared purple), showed promise in quantifying dysplastic risk using green/red fluorescence intensity ratios[10]. These findings support a shift from random biopsies to targeted, image-enhanced techniques tailored to lesion morphology and mucosal risk.

Figure 2 Representative images of a Japanese Narrow band imaging Experts Team 2A/B lesion in the descending colon observed under dye-based chromoendoscopy. A: Methylene blue chromoendoscopy reveals a regular surface pattern with slight vascular irregularity, consistent with low-grade dysplasia features; B: Crystal violet staining enhances the pit pattern, demonstrating a type IV-VI Kudo pattern suggestive of possible high-grade dysplasia.

Table 1 Summary of studies on individual techniques of dysplasia detection in inflammatory bowel disease.

Ref.	Endoscopy method	Key findings
Rubín de Célix et al[6], 2021	Dye-based Chromoendoscopy (Indigo carmine)	24% dysplasia detection rate; high yield in flat and right-sided lesions; lesions > 5 mm and age > 60 were risk factors
Klepp et al[5], 2018	Dye-based Chromoendoscopy (Indigo carmine)	10.5% dysplasia detection rate; 20.8% yield with targeted vs 3.5% with random biopsies; high NPV (97%)
Wanders et al[7], 2016	Chromoendoscopy + confocal laser endomicroscopy (iCLE)	9.8% dysplasia detection rate; low sensitivity (42.9%); frequent equipment failures limited routine use
Matsumoto et al[8], 2007	Narrow band imaging + magnification	Tortuous pattern correlated with dysplasia; 80% sensitivity; high specificity (84.2%)
Guo et al[9], 2021	Narrow band imaging + mucosal vascular pattern analysis	MVP correlates with histological inflammation and Ki-67 index; useful for predicting mucosal proliferation
Yoshioka et al[10], 2016	AFI + NBI + CE	AFI green/red ratio predicted dysplasia; only 37.5% neoplastic lesions visible as purple; AFI quantitation promising

NPV: Negative predictive value; iCLE: Integrated confocal laser endomicroscopy; MVP: Mucosal vascular pattern; AFI: Autofluorescence imaging; CE: Chromoendoscopy; NBI: Narrow band imaging.

Indigo carmine as a substitute for methylene blue in DCE

Several studies have demonstrated that indigo carmine is an effective alternative to methylene blue for DCE in dysplasia surveillance among patients with UC. In a prospective evaluation, Hurlstone et al[11] showed that HD DCE using indigo carmine significantly increased intraepithelial neoplasia detection compared to conventional colonoscopy (69 lesions vs 24 lesions, P < 0.0001), especially for flat lesions. It also enabled accurate in vivo pit-pattern-based differentiation of neoplastic from non-neoplastic mucosa (93% sensitivity, 88% specificity)[11]. Similarly, Rutter et al[12] demonstrated that pancolonic indigo carmine dye spraying enabled the detection of additional dysplastic lesions not seen with standard-definition (SD) WLE. Targeted biopsies after dye spraying detected seven additional dysplastic lesions in five patients that were missed by random biopsies (P = 0.02). More recently, González-Bernardo et al[13] used indigo carmine in a randomized controlled trial (RCT) comparing DCE with VCE and found comparable dysplasia detection rates, supporting the dye’s efficacy in contemporary surveillance protocols. Collectively, these studies affirm indigo carmine's utility as a safe, effective, and resource-friendly dye for enhancing dysplasia detection in IBD surveillance.

Panoramic endoscopy enhances dysplasia detection in IBD

Full-spectrum endoscopy (FUSE), which provides a 330° panoramic field of view using three integrated cameras, has shown significant promise in improving dysplasia detection during IBD surveillance colonoscopy. In a prospective crossover study, Leong et al[14] compared FUSE to standard forward-viewing colonoscopy in 52 IBD patients. FUSE significantly reduced the dysplasia miss rate per lesion (25% vs 71.4%, P = 0.0001) and per subject (25% vs 75%, P = 0.046), and detected more dysplastic lesions overall (mean 0.37 vs 0.13, P = 0.044) despite a longer withdrawal time. These findings underscore FUSE’s potential to detect flat or hidden lesions behind mucosal folds, making it a valuable adjunct to DCE, particularly in high-risk IBD surveillance cohorts.

Comparison of endoscopic techniques for dysplasia detection

DCE and HD WLE: (1) Comparative performance of HD DCE and HD WLE: Several studies have assessed whether HD DCE improves dysplasia detection in IBD over HD WLE (Table 2). Yang et al[15] conducted a multicenter RCT in 210 UC patients and found no significant difference in dysplasia detection between HD DCE (3.9%) and HD WLE (5.6%), although HD DCE required fewer biopsies. Similarly, Alexandersson et al[16] reported no overall difference in detection rates but noted higher dysplasia detection with DCE in the primary sclerosing cholangitis (PSC) subgroup. In a large real-world cohort study, Coelho-Prabhu et al[17] found a higher dysplasia yield with DCE (33% vs 12%) that lost statistical significance after multivariate adjustment. Carballal et al[2] demonstrated that targeted biopsies under DCE identified 91% of dysplasia, suggesting a minimal role for random biopsies. Together, these studies underscore that while HD DCE may offer procedural efficiencies and improved lesion characterization, its superiority over HD WLE in dysplasia detection remains unproven in most settings; and (2) SD WLE vs DCE: Multiple studies have investigated whether chromoendoscopy offers a diagnostic advantage over SD WLE for dysplasia surveillance in IBD. In an early RCT, Kiesslich et al[18] demonstrated a significantly higher dysplasia detection rate with CE compared to SD WLE (32 lesions vs 10 lesions, P < 0.005), particularly for flat and endoscopically subtle lesions. Marion et al[19] reinforced these findings in a prospective study, where CE identified additional dysplastic lesions missed by SD WLE, advocating its routine use. However, in a large multicenter retrospective study, Mooiweer et al[20] found no significant difference in neoplasia detection between CE and SD WLE plus random biopsies (11% vs 10%, P = 0.80), raising questions about CE’s real-world benefit. Similarly, Marion et al[21] reported no added yield of CE in patients with prior negative SD colonoscopies. Contrarily, Freire et al[22] found that CE improved the neoplasia detection rate (28.2% vs 11.6%, P = 0.01) in UC, and Wan et al[23], in a multicenter RCT with long-term follow-up, showed that CE with targeted biopsies significantly outperformed WLE with targeted biopsies (9.7% vs 1.9%, P = 0.004). More recently, Te Groen M et al[24] in the multicenter HELIOS trial demonstrated that HD WLE with segmental re-inspection was non-inferior to HD DCE for neoplasia detection (10.3% vs 13.1%), challenging the presumed superiority of chromoendoscopy in expert settings. Collectively, while controlled studies often support CE’s superiority, its effectiveness in everyday practice remains mixed, influenced by endoscopist experience, lesion characteristics, and procedural standardization.

Table 2 Comparison of conventional/high definition colonoscopy and dye chromoendoscopy for dysplasia detection in inflammatory bowel disease.

Ref.	Techniques compared	Study design	Key findings
Alexandersson et al[16], 2020	HD WLE vs HD DCE	RCT, 305 IBD patients	HD DCE detected dysplasia in 14% vs 6% with HD WLE (P = 0.020); superior for macroscopic dysplasia and lesions per 10-minute withdrawal
Carballal et al[2], 2018	HD WLE vs HD DCE (real-life)	Prospective real-world cohort	Dysplasia detection: 11.5% with CE vs 2.6% with WLE; CE significantly improved detection, mostly by targeted biopsies
Coelho-Prabhu et al[17], 2021	HD WLE vs HD DCE	Prospective observational	No significant difference in dysplasia detection; CE associated with longer procedure time
Yang et al[15], 2019	HD WLE (random) vs HD DCE (targeted)	Multicenter RCT	No significant difference in CAD detection (3.9% CE vs 5.6% WLE); CE reduced the number of biopsies
Kiesslich et al[18], 2003	SD WLE vs CE (Indigo carmine)	RCT	CE detected 32 lesions vs 10 lesions with WLE (P < 0.005); more flat/invisible dysplasia
Marion et al[19], 2008	SD WLE vs DCE	Prospective single-center	CE detected more dysplasia and additional lesions missed by WLE
Mooiweer et al[20], 2015	SD WLE + random biopsies vs DCE + targeted	Multicenter retrospective	No significant difference in dysplasia detection (11% vs 10%, P = 0.80)
Marion et al[21], 2016	SD WLE vs DCE	Prospective follow-up	CE did not increase detection in patients with prior negative colonoscopy
Freire et al[22], 2014	SD WLE vs DCE	RCT in UC patients	CE had higher dysplasia detection (28.2% vs 11.6%, P = 0.01)
Wan et al[23], 2021	SD WLE (targeted) vs DCE (targeted)	Multicenter RCT, long-term follow-up	CE superior for long-term surveillance (9.7% vs 1.9%, P = 0.004); detected more non-polypoid lesions
Te Groen M et al[24], 2025	HD WLE with segmental re-inspection vs HD DCE	Multicenter RCT (HELIOS)	HD WLE with re-inspection was non-inferior to HD DCE (10.3% vs 13.1%) for dysplasia detection

HD: High definition; WLE: White light endoscopy; DCE: Dye-based chromoendoscopy; SD: Standard definition; CE: Chromoendoscopy; RCT: Randomized controlled trial; IBD: Inflammatory bowel disease; UC: Ulcerative colitis; CAD: Colitis-associated dysplasia.

VCE vs WLE: Several studies have evaluated whether VCE, including technologies such as NBI and i-SCAN, enhances dysplasia detection in IBD compared to WLE (Table 3). In the multicenter VIRTUOSO trial, Kandiah et al[25] demonstrated that VCE using i-SCAN optical enhancement was non-inferior to HD WLE for neoplasia detection, while also allowing for better lesion characterization and a reduction in unnecessary random biopsies. However, earlier studies have shown mixed results. In an RCT, van den Broek et al[26] reported no significant difference in dysplasia detection between NBI and HD WLE (81% vs 69%, P = 0.727), with limited added value in real-time histologic prediction. Similarly, Dekker et al[27] observed that NBI offered poor sensitivity and specificity in a prospective surveillance cohort, failing to outperform SD WLE. Leifeld et al[28], in an RCT among UC, also found no significant difference in dysplasia detection between NBI and WLE, and noted that NBI poorly correlated with histological inflammation. Collectively, while newer VCE platforms like i-SCAN show promise, especially in HD formats, traditional NBI has not consistently demonstrated superiority over WLE, and its role in IBD surveillance remains adjunctive rather than primary.

Table 3 Comparative studies of virtual chromoendoscopy, white light endoscopy, and artificial intelligence-assisted detection for dysplasia surveillance in inflammatory bowel disease.

Ref.	Techniques compared	Study design	Key findings
Kandiah et al[25], 2021	VCE (i-SCAN OE) vs HD WLE	Multicenter RCT (VIRTUOSO trial)	VCE non-inferior to HD WLE for neoplasia detection; improved dysplasia characterization and fewer random biopsies
van den Broek et al[26], 2011	NBI vs HD WLE	Randomized crossover trial	No significant difference in neoplasia detection (NBI 81% vs HDE 69%, P = 0.727); NBI did not improve real-time differentiation
Dekker et al[27], 2007	NBI vs SD WLE	Prospective observational	NBI had limited sensitivity and specificity for dysplasia; not superior to WLE in UC surveillance
Leifeld et al[28], 2015	NBI vs SD WLE	RCT in UC patients	No significant difference in neoplasia detection; NBI had poor correlation with histology
López-Serrano et al[29], 2025	VCE with iSCAN vs CADe	Prospective, cross-sectional, non-inferiority diagnostic test comparison	Similar dysplasia detection (15.4% vs 13.5%); equal sensitivity (90%); VCE had better specificity and diagnostic accuracy

VCE: Virtual chromoendoscopy; i-SCAN OE: I-SCAN optical enhancement; HD: High definition; SD: Standard definition; WLE: White light endoscopy; NBI: Narrow band imaging; RCT: Randomized controlled trial; UC: Ulcerative colitis; CADe: Computer-aided detection; IBD: Inflammatory bowel disease.

VCE vs AI-assisted CADe in dysplasia surveillance: López-Serrano et al[29] compared VCE with iSCAN to an AI-based computer-aided detection (CADe) system for dysplasia detection in UC. Both identified dysplasia in approximately 15% of patients with equal sensitivity (90%). VCE had better specificity and accuracy, while CADe showed more false positives, highlighting the need for refinement before routine use.

DCE vs VCE: Multiple RCTs have compared DCE with VCE modalities such as NBI, i-SCAN, and AFI in the surveillance of IBD-related dysplasia (Table 4). In a large multicenter RCT, Bisschops et al[30] found no significant difference in neoplasia detection between HD DCE and NBI (21.2% vs 21.5%), though NBI was faster. Similarly, Pellisé et al[31] showed comparable dysplasia detection between CE and NBI but noted a higher miss rate with NBI, albeit not statistically significant. González-Bernardo et al[13] and Jans et al[32] both demonstrated equivalent dysplasia detection rates between CE and i-SCAN, though CE slightly improved lesion characterization. Efthymiou et al[33] also found similar detection yields between CE and NBI, with NBI offering faster procedures and fewer false positives. The FIND-UC trial, Vleugels et al[34], compared CE with AFI, reporting higher specificity and fewer false positives with CE, despite no difference in detection rates. Overall, while VCE offers practical advantages, especially in reducing procedure time, DCE remains the benchmark, particularly for nuanced lesion characterization in high-risk IBD surveillance.

Table 4 Studies comparing dye-based and virtual chromoendoscopy techniques for dysplasia detection in inflammatory bowel disease.

Ref.	Techniques compared	Study design	Key findings
Bisschops et al[30], 2018	HD CE vs NBI	Multicenter RCT	No significant difference in dysplasia detection (21.2% vs 21.5%); NBI had shorter procedure time
Pellisé et al[31], 2011	HR CE vs HR NBI	Randomized crossover study	NBI is less time-consuming, but CE had a lower miss rate for intraepithelial neoplasia; not statistically significant
González-Bernardo et al[13], 2021	CE vs i-SCAN (VCE)	RCT	No significant difference in dysplasia detection rate; CE required more procedure time
Jans et al[32], 2024	CE vs i-SCAN OE	RCT	Both techniques are comparable in dysplasia detection; CE offered slightly better lesion characterization
Efthymiou et al[33], 2013	CE vs NBI	RCT	Dysplasia detection rates were similar; NBI had fewer false positives and was faster
Vleugels et al[34], 2018	CE vs AFI	RCT (FIND-UC trial)	CE had higher specificity and fewer false positives than AFI; no difference in dysplasia detection

CE: Chromoendoscopy; NBI: Narrow band imaging; VCE: Virtual chromoendoscopy; i-SCAN OE: I-SCAN optical enhancement; AFI: Autofluorescence imaging; HD: High definition; HR: High resolution; RCT: Randomized controlled trial; IBD: Inflammatory bowel disease.

Comparison of multiple endoscopic modalities for dysplasia surveillance in IBD: Several studies have assessed the comparative performance of more than two endoscopic techniques - typically involving combinations of WLE, DCE, VCE, and CLE - to determine optimal dysplasia surveillance strategies in IBD (Table 5). In a prospective cohort study, Hlavaty et al[35] compared SD WLE, DCE, and CLE, concluding that targeted biopsies were superior to random ones, with DCE improving intraepithelial neoplasia detection. However, CLE was limited by the poor evaluability of polypoid lesions, offering little added clinical value[35]. Iacucci et al[36], in a multicenter RCT, compared HD WLE, HD DCE, and VCE (iSCAN), and found no significant difference in dysplasia detection among the three modalities. Notably, VCE was the most time-efficient and yielded the best lesion visibility. Gasia et al[3] examined HD WLE, HD DCE, and HD CLE in a real-world setting and found that CLE, when combined with targeted biopsies, yielded the highest dysplasia detection rate, although it was more resource-intensive. Across these studies, while DCE and CLE may enhance dysplasia yield under specific circumstances, VCE offers practical efficiency, and random biopsies continue to show limited incremental benefit.

Table 5 Comparative studies of multiple advanced endoscopic modalities for dysplasia detection in inflammatory bowel disease.

Ref.	Techniques compared	Study design	Key findings
Hlavaty et al[35], 2011	SD WLE vs DCE vs CLE	Prospective cohort	Targeted biopsies are superior to random; DCE increased IEN detection; CLE did not add clinical benefit due to low evaluability of polypoid lesions
Iacucci et al[36], 2018	HD WLE vs HD DCE vs VCE (iSCAN)	Multicenter RCT	No significant difference in neoplasia detection; VCE had the best lesion visibility and was the most time-efficient
Gasia et al[3], 2016	HD WLE vs HD DCE vs HD CLE	Prospective observational	CLE combined with targeted biopsies provided the highest dysplasia detection rate, though more resource-intensive; random biopsies added little value

SD: Standard definition; HD: High definition; WLE: White light endoscopy; DCE: Dye-based chromoendoscopy; CLE: Confocal laser endomicroscopy; VCE: Virtual chromoendoscopy; IEN: Intraepithelial neoplasia; RCT: Randomized controlled trial; IBD: Inflammatory bowel disease.

Comparison of biopsy protocols in dysplasia surveillance

The role of random vs targeted biopsy strategies for dysplasia detection in IBD remains debated (Table 6). In a pivotal multicenter retrospective study, Mooiweer et al[20] reported no significant difference in neoplasia detection between random biopsies with WLE and targeted biopsies using DCE, questioning the utility of random sampling. However, several other studies have highlighted complementary roles. Moussata et al[37] found that 20% of dysplastic lesions were detected only through random biopsies, especially in high-risk groups such as patients with PSC, previous neoplasia, or tubular colons. Wan et al[23] demonstrated that DCE with targeted biopsies was superior to both random and targeted WLE approaches in dysplasia yield. Similarly, Carballal et al[2] showed that DCE with targeted biopsies identified significantly more dysplasia than WLE with random biopsies. Other studies, including those by Gasia et al[3], Hlavaty et al[35], and Günther et al[38], consistently reported that random biopsies add limited incremental value over targeted strategies, particularly with HD imaging. Marion et al[19] also supported a targeted approach with DCE over random WLE sampling. Leifeld et al[28] found no added benefit of random biopsies in NBI surveillance. Hu et al[39] echoed these findings in a real-world cohort, where random biopsies demonstrated low diagnostic yield. A Japanese multicenter study by Watanabe et al[40] demonstrated that step biopsy was not superior to targeted biopsy, with the latter offering a more efficient, less invasive approach. Collectively, these findings support a shift toward image-enhanced, targeted surveillance, especially when performed by experienced endoscopists using HD platforms.

Table 6 Comparative studies of targeted vs random biopsy strategies in inflammatory bowel disease dysplasia surveillance.

Ref.	Biopsy protocol compared	Study design	Key findings
Mooiweer et al[20], 2015	Random biopsies (WLE) vs Targeted (DCE)	Retrospective multicenter	No significant difference in neoplasia detection (11% vs 10%, P = 0.80); questions the routine benefit of CE
Moussata et al[37], 2018	Targeted + Random biopsies (CE)	Prospective multicenter cohort	20% of neoplastic sites detected only by random biopsies; the highest yield in PSC, prior neoplasia, and tubular colon
Wan et al[23], 2021	Targeted (CE) vs Random (WLE) vs Targeted (WLE)	Multicenter RCT with long-term follow-up	CE with targeted biopsies was superior to WLE targeted (9.7% vs 1.9%, P = 0.004); random WLE was also better than WLE targeted alone
Carballal et al[2], 2018	Targeted (CE) vs Random (WLE)	Real-life prospective cohort	CE + targeted biopsies detected significantly more dysplasia than WLE with random biopsies
Gasia et al[3], 2016	Random vs Targeted (WLE, DCE, CLE)	Prospective observational	Random biopsies added minimal yield; targeted approach is more efficient
Günther et al[38], 2011	Targeted vs Random (WLE)	Retrospective analysis	Targeted biopsies are sufficient; random biopsies rarely add additional findings
Hlavaty et al[35], 2011	Targeted vs Random (CLE, DCE, WLE)	Prospective cohort	Targeted biopsies increased yield; CLE is not superior to DCE/WLE for flat lesions
Marion et al[19], 2008	Targeted (CE) vs Random (WLE)	Prospective single center	Targeted CE biopsies identified more dysplasia than random WLE biopsies
Leifeld et al[28], 2015	Targeted (NBI) vs Random (WLE)	RCT	No significant advantage of random biopsies in NBI-guided exams
Hu et al[39], 2021	Random vs Targeted (WLE, CE, VCE)	Retrospective cohort	Low diagnostic yield from random biopsies supports image-enhanced targeted approaches
Watanabe et al[40], 2011	Step biopsy vs Targeted biopsy	Japanese multicenter study	Target biopsy is non-inferior to step biopsy in neoplasia detection; fewer biopsies and reduced invasiveness

WLE: White light endoscopy; CE: Chromoendoscopy; DCE: Dye-based chromoendoscopy; VCE: Virtual chromoendoscopy; CLE: Confocal laser endomicroscopy; PSC: Primary sclerosing cholangitis; RCT: Randomized controlled trial; IBD: Inflammatory bowel disease.

Endoscopic techniques for dysplasia characterization

Accurate endoscopic characterization of dysplasia is crucial in IBD surveillance (Table 7). Bisschops et al[41] demonstrated that DCE was more sensitive than NBI for pit-pattern-based dysplasia detection, although both showed comparable negative predictive values. Carballal et al[2] further validated the diagnostic utility of pit pattern, with non-polypoid morphology and proximal lesions linked to higher dysplasia risk. Van den Broek et al[26] found that AFI had the highest sensitivity and negative predictive value among WLE, NBI, and AFI, with combined use enhancing lesion characterization. Vleugels et al[34] confirmed in a pre-specified analysis that DCE outperformed tri-modal imaging (AFI, NBI, WLE) marginally in sensitivity, but both modalities had high negative predictive values. These findings support a multimodal, morphology-guided strategy for optimal dysplasia characterization in IBD.

Table 7 Endoscopic techniques for dysplasia characterization in inflammatory bowel disease.

Ref.	Study design	Techniques compared	Key findings
Bisschops et al[30], 2018	Prospective comparative study	HD CE vs HD NBI	CE had higher sensitivity (88%) than NBI (60%) for Kudo pit pattern-based dysplasia diagnosis; NPV was similar (89%)
Carballal et al[2], 2018	Large prospective cohort study	WLE with targeted biopsies using the pit pattern	Non-polypoid morphology, proximal location, and type III-V pit pattern were associated with dysplasia; supports the pit pattern as a diagnostic adjunct
van den Broek et al[26], 2011	Prospective study	AFI, NBI, WLE	AFI had the highest sensitivity and NPV for dysplasia characterization; the combined use improved detection
Vleugels et al[34], 2018	Pre-specified analysis of FIND-UC RCT	Trimodal (AFI, NBI, WLE) vs CE	CE had slightly better sensitivity (82% vs 77%) for dysplasia; both had high NPV (> 94%); AFI alone had the best sensitivity (92%)

HD: High definition; CE: Chromoendoscopy; NBI: Narrow band imaging; WLE: White light endoscopy; AFI: Autofluorescence imaging; PDD: Photodynamic diagnosis; NPV: Negative predictive value; RCT: Randomized controlled trial; IBD: Inflammatory bowel disease.

Cost-effectiveness of surveillance strategies in UC

Two studies have evaluated the cost-effectiveness of dysplasia surveillance strategies in UC. Konijeti et al[42] used a decision-analytic Markov model to compare DCE, WLE, and no surveillance, finding DCE to be both more effective and less costly than WLE. DCE became cost-effective compared to no surveillance when performed at intervals ≥ 7 years, with an incremental cost-effectiveness ratio of $77176 per quality-adjusted life year. CE dominated WLE across nearly all sensitivity analyses. Kisiel et al[43] assessed the combination of stool DNA testing and CE, reporting that this strategy improved the detection of advanced neoplasia and might be cost-effective in high-risk UC patients, although further validation is needed. Together, these studies highlight the economic value of CE-based surveillance and the potential role of non-invasive adjuncts in optimizing cost and clinical outcomes.

Interobserver agreement for dysplasia detection and characterization

Accurate visual differentiation of dysplastic from non-dysplastic lesions during IBD surveillance remains challenging, with considerable variability among endoscopists. In a large real-life cohort, Carballal et al[2] observed that despite high-resolution chromoendoscopy with pit pattern analysis, variability in lesion interpretation contributed to inconsistent dysplasia detection across different centers. This issue was further underscored by Wanders et al[7], who systematically assessed interobserver agreement using standardized endoscopic images. Their study demonstrated only fair interobserver agreement (κ = 0.24) among 17 endoscopists, including both experts and nonexperts, in distinguishing dysplastic from non-dysplastic lesions. Experts had slightly higher specificity (61% vs 47%) and confidence, but accuracy remained suboptimal overall. These findings highlight the inherent subjectivity in endoscopic dysplasia assessment and underscore the importance of histopathological confirmation for any suspicious lesion, regardless of the endoscopist’s experience.

Real-world practice patterns and quality gaps in dysplasia surveillance

Despite guideline recommendations, significant variability persists in the implementation and quality of dysplasia surveillance in IBD (Table 8). In a national French study nested within the CESAME cohort, Vienne et al[44] found that only 54% of eligible patients with longstanding extensive colitis underwent surveillance colonoscopy, with wide inter-center variation and lower uptake in Crohn’s colitis. Chromoendoscopy was used in just 30% of cases, with random biopsies predominating. Similarly, a United States-based survey by Lewis et al[45] revealed inconsistent adherence among high-volume IBD providers, with only 20% routinely using DCE and 58% continuing random biopsies despite doubts about their utility. Te Groen M et al[46], in a multicenter Dutch audit, showed that only 10.6% of colonoscopies adhered to quality standards, citing lapses in withdrawal time, lesion reinspection, and documentation. These findings underscore a persistent gap between recommendations and real-world practice, highlighting the need for structured quality improvement initiatives in IBD surveillance programs.

Table 8 Practice pattern and quality of care in dysplasia surveillance.

Ref.	Study type	Sample size	Key findings
Vienne et al[44], 2011	Survey (France, CESAME cohort)	583 patients	Only 54% of eligible IBD patients received surveillance colonoscopy; large inter-center variability (27%-70%) and low uptake in Crohn’s colitis. Chromoendoscopy was used in 30%, and random biopsies in 71%
Lewis et al[45], 2022	Survey of high-volume IBD providers (United States)	55 providers	Wide variation in practice: 20% regularly used DCE, 27% VCE, 58% random biopsies; random biopsy use is inversely related to DCE use; less than half believed random biopsies increased detection with HD-WLE
Te Groen M et al[46], 2024	Multicenter retrospective audit (Netherlands)	644 colonoscopies from 391 IBD patients	Major guideline deviations: Inadequate withdrawal time (52%), lack of inspection of prior dysplasia sites (68%), and suboptimal documentation; only 10.6% received surveillance consistent with guideline-defined high-quality colonoscopy

IBD: Inflammatory bowel disease; CESAME: Cohorte Epidémiologique Sur les Maladies de l’Appareil Digestif; DCE: Dye-based chromoendoscopy; VCE: Virtual chromoendoscopy; HD-WLE: High definition white light endoscopy.

Long-term outcomes of dysplasia surveillance

Choi et al[1] reported 40 years of surveillance experience in UC from St. Mark’s Hospital, involving over 1300 patients. The study showed a steady decline in advanced CRC and colectomy rates over time, with improved dysplasia detection, particularly after the adoption of chromoendoscopy. Despite increased detection of low-grade dysplasia, progression to CRC remained stable, supporting the effectiveness of long-term surveillance in reducing CRC-related morbidity.

DISCUSSION

IEE techniques, including DCE, VCE, NBI, and AFI, have significantly improved the visualization and detection of dysplasia in IBD. These tools support a targeted biopsy strategy, enhancing diagnostic yield and allowing for more precise lesion characterization compared to standard approaches. While some studies show a modest benefit of HD DCE over HD WLE, others report comparable detection rates, with DCE offering procedural advantages like fewer biopsies. Studies comparing indigo carmine with methylene blue suggest that indigo carmine is an effective, resource-efficient alternative for DCE.

The evidence remains mixed, especially outside expert centers, suggesting DCE’s advantage may depend on operator expertise and patient risk factors. Earlier studies favored DCE over SD WLE in detecting flat and subtle lesions. However, more recent findings, especially from real-world practice, show inconsistent benefits[2]. Factors like endoscopist experience and inspection time play significant roles in influencing outcomes. VCE techniques like i-SCAN and NBI provide practical alternatives to DCE. While newer VCE platforms show comparable dysplasia detection and improved efficiency, older NBI studies reveal limitations in real-time accuracy. VCE remains a suitable adjunct, particularly when DCE is not feasible. Multiple trials demonstrate comparable dysplasia detection rates between DCE and VCE, with VCE offering shorter procedure times and fewer false positives[13,32]. However, DCE remains superior for nuanced lesion characterization, particularly in high-risk populations. Studies comparing multiple modalities indicate that while CLE and DCE can enhance dysplasia yield, VCE provides a more time-efficient option[35]. CLE is resource-intensive and often limited in evaluability, while DCE remains the most balanced in performance and practicality. CLE is resource-intensive and often limited in evaluability, while DCE remains the most balanced in performance and practicality[3].

In addition to surveillance colonoscopy, pharmacologic prevention strategies targeting inflammation-driven carcinogenesis - such as mesalazine, STING agonists, and emerging immunomodulators - may offer complementary benefits and warrant integration in future protocols, Papadakos et al[4]. Collectively, these strategies underscore a pragmatic, combined approach - pairing optimized image-enhanced surveillance with targeted chemopreventive mechanisms along the cyclic guanosine monophosphate-adenosine monophosphate synthase-STING axis.

Targeted biopsies under IEE, particularly DCE, consistently outperform or match random biopsy protocols. Random biopsies may still offer incremental yield in high-risk subgroups but add little value in standard practice, especially with HD imaging[3,35,38]. Pit pattern analysis, mucosal vascular patterns, and fluorescence-based techniques improve dysplasia characterization[2]. Combining modalities like CE, NBI, and AFI enhances diagnostic accuracy and helps distinguish sporadic from colitis-associated neoplasia[34]. DCE is both clinically effective and cost-saving compared to WLE in most models. Integration of stool DNA testing or extended surveillance intervals can further improve cost-efficiency, especially in high-risk patients[43]. Interobserver variability remains a major limitation in IBD surveillance. Despite HD imaging, accuracy and agreement in dysplasia interpretation are suboptimal, reinforcing the need for histologic confirmation and possibly AI-based standardization[7].

While promising, AI-based dysplasia detection tools remain limited by hardware costs, the need for annotated training data, and integration challenges with existing hospital infrastructure. Similarly, CLE adoption is restricted by equipment availability, cost, and operator learning curves, limiting its use to expert centers. Despite encouraging performance metrics, real-world implementation of AI systems requires (1) Adequate on-premise or embedded compute (processing unit/accelerator) to minimize latency for real-time CADe/computer-aided diagnostic; (2) Robust pipelines for large, representative, and continuously updated annotated datasets; (3) Interoperability with existing processors, endoscopy reporting systems, and hospital information technology; and (4) Endoscopist training, credentialing, and quality assurance to avoid over-reliance or alert fatigue. Cost and reimbursement remain center-specific and may limit adoption outside tertiary units. Likewise, CLE - although valuable in expert hands - is constrained by capital and per-use costs, dye/consumable needs, added procedure time, and a steep learning curve, restricting scalability in routine surveillance. These practical considerations are essential when translating trial efficacy into program-level effectiveness, particularly in resource-constrained settings. When selecting advanced technologies for an IBD surveillance program, centers should match capability with context - prioritizing platforms that are maintainable, staff-trainable, reimbursable, and compatible with existing workflows to ensure sustained quality rather than isolated technical excellence.

Patient experience meaningfully influences surveillance adherence. Frequent colonoscopies, burdensome bowel preparation, procedure-related anxiety, and time away from work/family are commonly cited barriers. Some patients may prefer shorter examinations (e.g., VCE) or non-invasive adjuncts such as stool DNA testing, where appropriate. Embedding patient-reported outcome measures within surveillance quality frameworks can align modality choice and intervals with patient preferences without compromising neoplasia detection. The future directions include: (1) Validation of combined pharmacologic-endoscopic surveillance strategies; (2) Evaluation of the cost-effectiveness of AI and panoramic endoscopy in real-world settings; (3) Standardization of patient-reported outcome measures for acceptability of advanced surveillance; and (4) Defining optimal intervals based on lesion type, risk factors, and surveillance modality.

Surveillance practices for dysplasia in IBD vary widely across centers and providers. Use of advanced imaging and adherence to biopsy protocols remain inconsistent, emphasizing the need for quality assurance initiatives and training[46]. Patient perspectives must also inform surveillance strategies. Acceptability of frequent colonoscopies, bowel preparation burden, and preference for non-invasive options are key considerations for ensuring adherence and optimizing outcomes. Longitudinal data show that sustained surveillance efforts, especially with chromoendoscopy, reduce CRC and colectomy rates. Enhanced lesion detection without increased progression risk supports the long-term utility of endoscopic surveillance in IBD[1]. Key evidence gaps include: (1) Head-to-head DCE vs modern VCE in high-risk subgroups (e.g., PSC); (2) The incremental value of random biopsies when HD IEE quality metrics are optimized; (3) Real-world cost-effectiveness and equity of AI and panoramic systems across income settings; (4) Standardized training and competency benchmarks to reduce interobserver variability; and (5) Implementation strategies that integrate chemoprevention with image-enhanced surveillance.

CONCLUSION

In conclusion, dye-based and VCE have significantly advanced dysplasia surveillance in IBD, with targeted biopsies largely replacing random sampling in most clinical contexts. While DCE remains the gold standard, virtual platforms offer comparable detection with greater efficiency. Emerging tools such as AI-assisted systems and panoramic endoscopy show promise but require further validation. Despite robust evidence, wide variability in practice and surveillance quality persists, underscoring the urgent need for standardized protocols, training, and quality assurance measures to ensure optimal CRC prevention in IBD. Standardized protocols, cost-effectiveness evaluation, and training are key to optimizing CRC prevention in IBD through endoscopic surveillance.