Systematic review with network meta-analysis: endoscopic techniques for dysplasia surveillance in inflammatory bowel disease
Funding information
None.
Summary
Background
International guidelines recommend dysplasia surveillance in IBD.
Aim
To compare endoscopic techniques for dysplasia surveillance
Methods
We searched MEDLINE, Embase, CENTRAL for randomised trials through May 2019. We estimated odds ratios (ORs) for binary and mean differences (MDs) for continuous outcomes, using frequentist random-effects network meta-analysis. We assessed study risk of bias and appraised evidence certainty using GRADE.
Results
Eighteen trials (2638 participants) were included. Standard definition white-light endoscopy (OR 0.44, 95% CI 0.26-0.73; high certainty) and i-SCAN (OR 0.47, 95% CI 0.25-0.90; moderate certainty) had lower odds of detecting neoplasia than chromoendoscopy. Fujinon intelligent colour enhancement (FICE), standard definition white-light endoscopy and i-SCAN had lower odds for this outcome than full spectrum high definition white-light endoscopy (ORs 0.02 to 0.15; low certainty). Standard definition white-light endoscopy had lower odds of detecting nonpolypoid neoplasia than full spectrum high definition white-light endoscopy, narrow band imaging, chromoendoscopy and high definition white-light endoscopy (ORs 0.01-0.14; moderate certainty). Full spectrum high definition white-light endoscopy ranked as the best technique for both outcomes (moderate certainty). Standard definition white-light endoscopy had lower odds of detecting neoplasia by target biopsy (OR 0.27, 95% CI 0.08-0.91) and had shorter procedure time (MD −14.81 minutes, 95% CI −25.03, −4.06) than chromoendoscopy (moderate certainty).
Conclusions
Chromoendoscopy, high definition white-light endoscopy, narrow band imaging, autofluorescence, FICE and full spectrum high definition white-light endoscopy may be comparable for dysplasia surveillance. Standard definition white-light endoscopy and i-SCAN probably provide lower yields for neoplasia identification. Full spectrum high definition white-light endoscopy may represent the first-line approach.
1 INTRODUCTION
People with ulcerative colitis and Crohn's disease experience a two-fold higher risk of colorectal cancer compared to the general population.1, 2 The risk of death associated with this malignancy is around 1.5 times greater in people with IBD than in the general population.3
Colorectal cancer develops through a multistep process, where low- and high-grade dysplasia represent an intermediary stage that progresses to cancer.4-6 International guidelines recommend endoscopic surveillance in people with IBD to identify and eradicate colonic lesions at an early non-invasive stage, with the aim of reducing colorectal cancer incidence and mortality.7-14 The target population for surveillance programs includes patients with left-sided or extensive ulcerative colitis,7-11 and those with Crohn's disease involving at least one third or more than one segment of colon.7-9, 12 The first screening colonoscopy is recommended to occur at 8-10 years after disease onset.7-10 Several endoscopic techniques for dysplasia surveillance have been evaluated for IBD in randomised trials, including standard definition and high definition white-light endoscopy, chromoendoscopy, narrow band imaging (Olympus, Tokyo, Japan), i-SCAN (Pentax, Tokyo, Japan) and autofluorescence (Olympus, Tokyo, Japan). Clinical guidelines recommend chromoendoscopy with target biopsies for surveillance programs in this population.8-11, 13, 14 White-light endoscopy with random biopsies is considered only appropriate if chromoendoscopy is not available, whereas other endoscopic techniques have not been considered superior to chromoendoscopy.7, 9, 11, 13
Guideline recommendations supporting the use of chromoendoscopy are based on evidence from observational studies and randomised trials that have methodological limitations due to selection, performance, detection, attrition or reporting bias. In a recent systematic review of randomised trials, we found that chromoendoscopy identifies more patients with dysplasia only when compared to standard definition white-light endoscopy, with no evidence of a difference in the comparison with high definition white-light endoscopy, narrow band imaging and i-SCAN.15
Standard pairwise meta-analysis only allows the direct comparison of two endoscopic techniques that have been evaluated head-to-head in randomised trials. In a setting in which there are severall options for endoscopic dysplasia surveillance, a network meta-analysis can facilitate comparisons of all endoscopic procedures simultaneously within a single framework and rank available techniques according to efficacy and safety. We aimed to provide evidence regarding the comparative efficacy of all endoscopic approaches for dysplasia surveillance in IBD using network meta-analysis of randomised trials.
2 MATERIALS AND METHODS
This systematic review was reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Network Meta-Analyses (PRISMA-NMA) guidelines.16
2.1 Data sources and searches
We searched for randomised trials in MEDLINE (1996 to May 2019), Embase (1996 to May 2019) and the Cochrane Central Register of Controlled Trials (CENTRAL; from 1996 to May 2019). We did not limit searching by publication date or language. An information specialist with expertise in systematic reviews of randomised trials designed the search strategy (Table S1). We also searched relevant trials from reference lists of all identified trials, guidelines and reviews on the topic.
2.2 Study selection
Two reviewers (AI and MR) independently screened the retrieved search records by title and abstract. Potentially eligible citations were reviewed by the same two reviewers in full text. Disagreements were resolved through consensus and discussion with a third reviewer (SCP).
We included randomised and quasi-randomised controlled trials comparing any endoscopic technique (standard definition white-light endoscopy, high definition white-light endoscopy, chromoendoscopy, narrow band imaging, i-SCAN, autofluorescence, Fujinon intelligent colour enhancement [FICE] and full-spectrum endoscopy) for dysplasia surveillance in adults with ulcerative colitis and Crohn's disease. We included any study enrolling people with IBD or with concurrent primary sclerosing cholangitis, a condition at high risk of colorectal cancer that requires dysplasia surveillance following the diagnosis of biliary disease.10 We excluded trials evaluating colorectal cancer surveillance in the general population or hereditary polyposis syndromes.
2.3 Data extraction and quality assessment
Two reviewers (AI and MR) independently extracted data on characteristics of study, population, interventions and outcomes from included randomised trials using an electronic database. In case of randomised crossover trials, we extracted data regarding only the first phase of the study to avoid a carry-over effect. Any disagreements in data extraction were resolved through consensus and discussion with a third reviewer (SCP).
2.3.1 Outcomes
The key outcomes in this review were number of participants with one or more neoplastic lesions, number of participants with any type of lesion (including neoplastic and non-neoplastic lesions), number of neoplastic lesions detected by target biopsy, procedural time and any adverse events associated with the diagnostic procedure. According to the Vienna classification, we considered as neoplastic lesions those showing low-grade dysplasia, high-grade dysplasia or invasive neoplasia at histological examination.17 Non-neoplastic lesions included endoscopic findings with no evidence of dysplasia or invasive neoplasia at histology. In the analysis of number of neoplastic lesions detected by target biopsy, we included only trials in which both target and random biopsy specimens were taken in all intervention arms and we calculated this outcome as the number of lesions detected by target biopsy divided by the total number of neoplastic lesions identified.
Additional outcomes included all-cause mortality, colorectal cancer-related mortality, interval colorectal cancer and health-related quality of life.
We also evaluated the number of morphological and histological subtypes of neoplastic lesions identified using different endoscopic techniques. According to the SCENIC consensus, we considered the morphological classification of neoplasia in polypoid (including sessile and pedunculated lesions) and nonpolypoid (including elevated, flat and depressed lesions).13 We applied the histological classification of neoplastic lesions in low-grade dysplasia, high-grade dysplasia or invasive neoplasia, according to the Vienna classification.17 We calculated the number of nonpolypoid neoplastic lesions, low-grade dysplastic lesions and high-grade dysplastic lesions as the number of each morphological or histological subtype of neoplastic lesions divided by the total number of neoplastic lesions identified.
If published outcome data were not reported or provided in sufficient detail in included trials, an author (AI) contacted the investigators up to three times by email to request any relevant additional information.
2.3.2 Risk of bias assessment
Two independent reviewers (AI and MR) assessed the risks of bias using the Cochrane tool.18 Disagreements were resolved through consensus and discussion with a third reviewer (SCP). We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach for network meta-analysis to critically appraise the certainty of evidence for all evaluated outcomes.19 For this purpose, we used the Confidence in Network Meta-Analysis (CINeMA) web application.20
2.4 Statistical analysis
First, we performed random-effects pairwise meta-analysis. We assessed heterogeneity between trials in pairwise meta-analysis using the Chi-squared test and I2 statistics. Then, we used a frequentist framework, random-effects network meta-analysis to compare all endoscopic techniques for each pre-specified outcome.21, 22 We calculated treatment estimates for binary outcomes as odds ratios (ORs) and continuous outcomes as mean differences (MDs), together with their 95% CI.
We estimated the extent of heterogeneity in each network analysis using the restricted maximum likelihood method to generate a common heterogeneity variance (tau [τ]), which was compared with an empirical distribution of heterogeneity variances.23 To explore for network inconsistency, we used a loop-specific approach, which compared the estimated effects derived from direct and indirect evidence in all triangular and quadratic loops in a network.24 We used the design-by-treatment interaction approach to check the assumption of consistency in the entire analytical network.25 We ranked endoscopic techniques to generate a hierarchy for each endpoint. The relative ranking probability of each endoscopic procedure being among the best technique was obtained using surface under the cumulative ranking (SUCRA) curves and displayed using rankograms. We planned to explore publication bias with funnel plots in which 10 or more studies were available.26
We performed pairwise and network meta-analyses in Stata (StataCorp LP), using the network command27 and self-programmed Stata routines.28
3 RESULTS
3.1 Study characteristics
The search retrieved 565 citations (Figure 1). Twenty-one trials involving 2889 participants with IBD were eligible for inclusion in the review.29-49 We could not extract data from three crossover trials31, 32, 41 because results at the end of the first phase of the study were not reported. Thus, 18 trials involving 2638 participants were included in our analysis.29, 30, 33-40, 42-49 Table S2 shows the study design, interventions and population characteristics in the 21 trials included in the review.
The number of participants enrolled in each trial ranged between 2032 and 305.29 In nine30, 31, 34, 35, 39-42, 45 of 21 included trials, a total of 125 participants were excluded after randomisation for insufficient bowel preparation, active colonic inflammation, short disease duration or lack of compliance with the study protocol. Fourteen trials30-32, 34, 37, 39-41, 43, 44, 46-49 included only participants with ulcerative colitis, six trials29, 33, 35, 38, 42, 45 enrolled participants with both ulcerative colitis and Crohn's disease, and one trial included participants with ulcerative colitis, Crohn's disease and unclassified colitis.36
Eight endoscopic techniques for dysplasia surveillance were evaluated: chromoendoscopy (13 trials), high definition white-light endoscopy (9 trials), narrow band imaging (8 trials), standard definition white-light endoscopy (6 trials), autofluorescence (3 trials), i-SCAN (2 trials), full spectrum high definition white-light endoscopy (Endochoice, Alpharetta, USA) (1 trial) and Fujinon intelligent colour enhancement FICE (Fujifilm, Tokyo, Japan) (1 trial). Full spectrum high definition white-light endoscopy has additional lateral camera lenses, providing a 330 degree field of view as opposed to the 170 degree field of view of conventional forward-viewing endoscopes.50 We received additional unpublished information from the authors of two trials.34, 48
3.2 Risk of bias
Random sequence generation was adequate in nine trials and unclear in the remaining 12 (Figure S1). The risk of bias for allocation concealment was low in one trial and high or unclear in the other trials. Masking of participants, investigators and outcome assessment was adequate in 14 trials and unclear in the remaining seven. Ten trials were characterised by completeness of outcome reporting, while 11 were at high or unclear risk of bias for this domain. The risk of selective reporting bias was low in 13 and high or unclear in eight.
3.3 Network consistency
Figure 2 and Figure S2 show the networks of individual endoscopic techniques for all outcomes. Included trials were deemed sufficiently comparable by population, intervention and study characteristics (Table S2) to be summarised using network meta-analysis. Pairwise and network meta-analysis estimates were similar in magnitude (Tables 1-3; Tables S3-S6) and there was evidence of no loop-specific inconsistency between direct and indirect evidence (Figure S3). The design-by-treatment interaction approach did not identify global inconsistency in any outcome network (Table S7).
| Indirect comparison | ||||||||
|---|---|---|---|---|---|---|---|---|
| Direct comparison | SD white-light endoscopy | 5.51 (0.27-110.50) | 0.54 (0.27-1.08) | 0.93 (0.41-2.10) | 0.47 (0.23-0.99) | 0.56 (0.24-1.33) | 0.14 (0.02-0.86) | 0.44 (0.26-0.73) |
| — | FICE | 0.10 (0.00-1.94) | 0.17 (0.01-3.46) | 0.09 (0.00-1.73) | 0.10 (0.00-2.12) | 0.02 (0.00-0.77) | 0.08 (0.00-1.52) | |
| — | — | HD white-light endoscopy | 1.73 (1.01-2.96) | 0.89 (0.48-1.63) | 1.05 (0.47-2.33) | 0.25 (0.05-1.40) | 0.82 (0.50-1.32) | |
| — | — | 1.84 (1.06-3.21) | i-SCAN | 0.51 (0.24-1.10) | 0.61 (0.24-1.52) | 0.15 (0.02-0.88) | 0.47 (0.25-0.90) | |
| — | — | 1.06 (0.42-2.71) | — | NBI | 1.18 (0.50-2.79) | 0.29 (0.05-1.76) | 0.92 (0.54-1.56) | |
| — | — | 0.28 (0.05-1.53) | — | — | Autofluorescence | 0.24 (0.04-1.60) | 0.78 (0.39-1.57) | |
| — | — | 0.25 (0.05-1.40) | — | — | — | Full spectrum HD white-light endoscopy | 3.22 (0.55-18.98) | |
| 0.44 (0.26-0.73) | 0.08 (0.00-1.52) | 0.85 (0.39-1.84) | 0.58 (0.25-1.35) | 0.95 (0.45-1.99) | 0.60 (0.28-1.28) | — | Chromoendoscopy | |
Note
- Values are given as OR (95% confidence interval). The grid should be read from left to right. The lower part of the grid reports estimates for direct comparison and risk estimate is for the column-defining intervention compared to the row-defining intervention. An OR < 1 indicates the column technique is associated with a lower odds of identifying participants with one or more neoplastic lesions than the row technique. The upper part of the grid reports estimates for indirect comparison and the risk estimate is for the row-defining intervention compared to the column-defining intervention. An OR < 1 indicates the row technique is associated with a lower odds of identifying participants with one or more neoplastic lesions than the column technique. There are 18 trials involving 2543 participants in this network.
- Abbreviations: FICE, Fujinon intelligent colour enhancement; HD, high definition; NBI, narrow band imaging; SD, standard definition.
| Indirect comparison | ||||||||
|---|---|---|---|---|---|---|---|---|
| Direct comparison | SD white-light endoscopy | 0.58 (0.01-51.75) | 0.14 (0.02-0.93) | 0.12 (0.01-1.26) | 0.09 (0.02-0.46) | 0.22 (0.04-1.23) | 0.01 (0.00-0.34) | 0.13 (0.04-0.49) |
| — | FICE | 0.24 (0.00-22.27) | 0.20 (0.00-23.34) | 0.15 (0.00-12.57) | 0.38 (0.00-32.42) | 0.01 (0.00-3.52) | 0.23 (0.00-17.06) | |
| — | — | HD white-light endoscopy | 0.84 (0.14-5.24) | 0.61 (0.13-2.80) | 1.56 (0.30-8.06) | 0.05 (0.00-1.49) | 0.95 (0.24-3.75) | |
| — | — | 1.00 (0.15-6.53) | i-SCAN | 0.73 (0.09-6.20) | 1.85 (0.20-16.89) | 0.06 (0.00-2.82) | 1.13 (0.16-8.15) | |
| — | — | 0.80 (0.08-8.47) | — | NBI | 2.54 (0.55-11.68) | 0.09 (0.00-3.36) | 1.55 (0.53-4.50) | |
| — | — | 0.22 (0.01-5.28) | — | — | Autofluorescence | 0.03 (0.00-1.40) | 0.61 (0.20-1.89) | |
| — | — | 0.05 (0.00-1.49) | — | — | — | Full spectrum HD white-light endoscopy | 18.04 (0.49-668.36) | |
| 0.13 (0.03-0.51) | 0.23 (0.00-17.06) | 1.56 (0.25-9.75) | 1.56 (0.18-13.11) | 1.65 (0.52-5.20) | 0.46 (0.14-1.52) | — | Chromoendoscopy | |
Note
- Values are given as OR (95% confidence interval). The grid should be read from left to right. The lower part of the grid reports estimates for direct comparison and risk estimate is for the column-defining intervention compared to the row-defining intervention. An OR < 1 indicates the column technique is associated with a lower odds of identifying nonpolypoid neoplastic lesions than the row technique. The upper part of the grid reports estimates for indirect comparison and the risk estimate is for the row-defining intervention compared to the column-defining intervention. An OR < 1 indicates the row technique is associated with a lower odds of identifying nonpolypoid neoplastic lesions than the column technique. There are 10 trials involving 1336 participants in this network.
- Abbreviations: FICE, Fujinon intelligent colour enhancement; HD, high definition; NBI, narrow band imaging; SD, standard definition.
| Indirect comparison | |||||
|---|---|---|---|---|---|
| Direct comparison | SD white-light endoscopy | 0.60 (0.01-40.43) | 2.47 (0.01-545.12) | 3.59 (0.13-99.65) | 0.27 (0.08-0.91) |
| — | HD white-light endoscopy | 4.09 (0.14-120.69) | 5.95 (0.04-956.76) | 0.45 (0.01-25.16) | |
| — | 4.09 (0.14-120.69) | NBI | 1.45 (0.00-651.29) | 0.11 (0.00-21.11) | |
| — | — | — | Autofluorescence | 0.08 (0.00-1.67) | |
| 0.25 (0.06-0.99) | 0.45 (0.01-25.16) | — | 0.08 (0.00-1.67) | Chromoendoscopy | |
Note
- Values are given as OR (95% confidence interval). The grid should be read from left to right. The lower part of the grid reports estimates for direct comparison and risk estimate is for the column-defining intervention compared to the row-defining intervention. An OR < 1 indicates the column technique is associated with a lower odds of identifying neoplastic lesions by target biopsy than the row technique. The upper part of the grid reports estimates for indirect comparison and the risk estimate is for the row-defining intervention compared to the column-defining intervention. An OR < 1 indicates the row technique is associated with a lower odds of identifying neoplastic lesions by target biopsy than the column technique. There are six trials involving 1048 participants in this network.
- Abbreviations: HD, high definition; NBI, narrow band imaging; SD, standard definition.
3.4 Outcomes
Endoscopic technique effects in pairwise and network meta-analyses for the considered outcomes are reported in Tables 1-3 and Tables S3-S6. GRADE assessments for the comparison of each endoscopic technique with chromoendoscopy, the recommended first-line approach for dysplasia surveillance in IBD, on number of participants with one or more neoplastic lesions, number of participants with any type of lesion, number of nonpolypoid neoplastic lesions, number of low-grade dysplastic lesions and number of high-grade dysplastic lesions are provided in Table 4.
| Outcome and intervention | Confidence in evidence | Reasons for downgrading confidence in evidence | Network intervention estimate vs chromoendoscopya |
|---|---|---|---|
| Number of participants with one or more neoplastic lesions | |||
| SD white-light endoscopy |
High ●●●● |
No downgrades in confidence | 0.44 (0.26-0.73)b |
| FICE |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 0.08 (0.00-1.52) |
| HD white-light endoscopy |
Moderate ●●●○ |
Downgrade 1 level for imprecision (−1) | 0.82 (0.50-1.32) |
| i-SCAN |
Moderate ●●●○ |
Downgrade 1 level for inconsistency (−1) | 0.47 (0.25-0.90)b |
| NBI |
Low ●●○○ |
Downgrade 2 levels for study limitations (−1) and imprecision (−1) | 0.92 (0.54-1.56) |
| Autofluorescence |
Moderate ●●●○ |
Downgrade 1 level for imprecision (−1) | 0.78 (0.39-1.57) |
| FSHD white-light endoscopy |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 3.22 (0.55-18.98) |
| Number of participants with any type of lesion | |||
| HD white-light endoscopy |
Moderate ●●●○ |
Downgrade 1 level for imprecision (−1) | 0.69 (0.44-1.10) |
| i-SCAN |
Moderate ●●●○ |
Downgrade 1 level for inconsistency (−1) | 0.46 (0.22-0.95)b |
| NBI |
Low ●●○○ |
Downgrade 2 levels for study limitations (−1) and imprecision (−1) | 0.93 (0.56-1.54) |
| Autofluorescence |
High ●●●● |
No downgrades in confidence | 0.42 (0.24-0.73)b |
| Number of nonpolypoid neoplastic lesions | |||
| SD white-light endoscopy |
Moderate ●●●○ |
Downgrade 1 level for inconsistency (−1) | 0.13 (0.04-0.49)b |
| FICE |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 0.23 (0.00-17.06) |
| HD white-light endoscopy |
Moderate ●●●○ |
Downgrade 1 level for imprecision (−1) | 0.95 (0.24-3.75) |
| i-SCAN |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 1.13 (0.16-8.15) |
| NBI |
Low ●●○○ |
Downgrade 2 levels for study limitations (−1) and imprecision (−1) | 1.55 (0.53-4.50) |
| Autofluorescence |
Moderate ●●●○ |
Downgrade 1 level for imprecision (−1) | 0.61 (0.20-1.89) |
| FSHD white-light endoscopy |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 18.04 (0.49-668.36) |
| Number of low-grade dysplastic lesions | |||
| SD white-light endoscopy |
Moderate ●●●○ |
Downgrade 1 level for imprecision (−1) | 1.30 (0.38-4.40) |
| FICE |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 0.07 (0.00-8.55) |
| HD white-light endoscopy |
Very Low ●○○○ |
Downgrade 3 levels for study limitations (−1) and imprecision (−2) | 0.91 (0.17-4.86) |
| i-SCAN |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 1.96 (0.16-23.84) |
| NBI |
Very Low ●○○○ |
Downgrade 3 levels for study limitations (−1) and imprecision (−2) | 1.22 (0.21-6.91) |
| Autofluorescence |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 1.44 (0.13-16.11) |
| FSHD white-light endoscopy |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 0.19 (0.01-6.96) |
| Number of high-grade dysplastic lesions | |||
| SD white-light endoscopy |
Moderate ●●●○ |
Downgrade 1 level for imprecision (−1) | 0.72 (0.19-2.74) |
| FICE |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 15.00 (0.12-1923.71) |
| HD white-light endoscopy |
Very Low ●○○○ |
Downgrade 3 levels for study limitations (−1) and imprecision (−2) | 1.56 (0.24-9.92) |
| i-SCAN |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 1.90 (0.10-35.39) |
| NBI |
Very Low ●○○○ |
Downgrade 2 levels for study limitations (−1) and imprecision (−2) | 0.84 (0.11-6.14) |
| Autofluorescence |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 1.12 (0.09-13.85) |
| FSHD white-light endoscopy |
Low ●●○○ |
Downgrade 2 levels for imprecision (−2) | 4.67 (0.11-197.25) |
- Abbreviations: FICE, Fujinon intelligent colour enhancement; FSHD, full spectrum high definition; HD, high definition; NBI; narrow band imaging; SD, standard definition.
- a Values expressed as odds ratio (95% confidence interval).
- b Statistically significant.
3.4.1 Number of participants with one or more neoplastic lesions
Eighteen trials involving 2543 participants evaluated the effect of different endoscopic techniques on the number of participants who were identified to have one or more neoplastic lesions (Figure 2A). Standard definition white-light endoscopy (OR 0.44, CI 0.26-0.73; high certainty evidence) and i-SCAN (OR 0.47, CI 0.25-0.90; moderate certainty evidence) probably had lower odds of detecting participants with neoplasia compared to chromoendoscopy (Tables 1 and 4). Differently from the estimated CI, the 95% prediction interval for the comparison between i-SCAN and chromoendoscopy included the null value (ie ‘1’) (Figure S4A). A 95% prediction interval takes into consideration the full uncertainty around the intervention estimate, providing a range of values that includes, with 95% confidence, effect estimates from future studies. Standard definition white-light endoscopy (OR 0.14, CI 0.02-0.86; low certainty evidence), i-SCAN (OR 0.15, CI 0.02-0.88; low certainty evidence) and FICE (OR 0.02, CI 0.00-0.77; low certainty evidence) may have lower odds of identifying participants with neoplasia compared to full spectrum high definition white-light endoscopy (Table 1). The 95% prediction interval for these comparisons included the null value (Figure S4A). Standard definition white-light endoscopy may have also lower odds of detecting participants with neoplasia compared to narrow band imaging (OR 0.44, CI 0.23-0.99; low certainty evidence) (Table 1), with the 95% prediction interval including the null value (Figure S4A). High definition white-light endoscopy may have higher odds of identifying participants with neoplasia compared to i-SCAN (OR 1.73, CI 1.01-2.96; low certainty evidence) (Table 1), with the 95% prediction interval including the null value (Figure S4A).
Full spectrum high definition white-light endoscopy ranked as the best endoscopic technique to detect participants with one or more neoplastic lesions (Figure 3). There was moderate confidence in hierarchical ranking for study limitations.
There was no evidence of publication bias for this outcome (Figure S5A).
3.4.2 Number of participants with any type of lesion
Seven trials involving 1246 participants evaluated the effect of five endoscopic techniques (chromoendoscopy, high definition white-light endoscopy, narrow band imaging, autofluorescence and i-SCAN) on the number of participants who were identified to have one or more colonic lesions of any type (including neoplastic and non-neoplastic lesions) (Figure 2B). Autofluorescence (OR 0.42, CI 0.24-0.73; high certainty evidence) and i-SCAN (OR 0.46, CI 0.22-0.95; moderate certainty evidence) probably had lower odds of detecting participants with any type of lesion compared to chromoendoscopy (Table 4 and Table S3). The 95% prediction interval for the comparison between i-SCAN and chromoendoscopy included the null value (Figure S4B). Narrow band imaging probably had higher odds of detecting participants with any type of lesion compared to autofluorescence (OR 2.20, CI 1.06-4.56; moderate certainty evidence) (Table S3), with the 95% prediction interval including the null value (Figure S4B).
Chromoendoscopy ranked as the best endoscopic technique to detect participants with any type of lesion (Figure S6A), with moderate confidence in hierarchical ranking for study limitations.
We could not assess publication bias for this outcome due to the paucity of trials.
3.4.3 Number of nonpolypoid neoplastic lesions
The network for number of nonpolypoid neoplastic lesions included 10 trials involving 1336 participants (Figure 2C). Standard definition white-light endoscopy probably had lower odds of detecting nonpolypoid neoplastic lesions compared to high definition white-light endoscopy (OR 0.14, CI 0.02-0.93; moderate certainty evidence), chromoendoscopy (OR 0.13, CI 0.04-0.49; moderate certainty evidence), narrow band imaging (OR 0.09, CI 0.02-0.46; moderate certainty evidence) and full spectrum high definition white-light endoscopy (OR 0.01, CI 0.00-0.34; moderate certainty evidence) (Tables 2 and 4). The 95% prediction interval for these comparisons included the null value (Figure S4C). There was no evidence of detectable differences in the odds of detecting nonpolypoid neoplasia in any other comparison between endoscopic techniques (moderate to low certainty evidence) (Tables 2 and 4).
Full spectrum high definition white-light endoscopy ranked as the best endoscopic technique for this outcome, with moderate confidence in hierarchical ranking for imprecision (Figure 3).
There was no evidence of publication bias for this outcome (Figure S5B).
3.4.4 Number of low-grade and high-grade dysplastic lesions
The networks for number of low-grade and high-grade dysplastic lesions included 14 trials involving 1736 participants (Figure 2D,E). There was no evidence of detectable differences between interventions in the odds of detecting low-grade (moderate to very low certainty evidence) (Table S4) and high-grade (moderate to very low certainty evidence) (Table S5) dysplastic lesions. There was limited confidence in hierarchical endoscopic technique rankings for number of low-grade and high-grade dysplastic lesions for study limitations and imprecision (Figure 3).
There was no evidence of publication bias for these outcomes (Figure S5C,D).
3.4.5 Number of neoplastic lesions detected by target biopsy and procedural time
Six trials involving 1048 participants evaluated the effect of five endoscopic techniques (chromoendoscopy, high definition white-light endoscopy, narrow band imaging, standard definition white-light endoscopy and autofluorescence) on number of neoplastic lesions detected by target biopsy (Figure S2A). Standard definition white-light endoscopy probably had lower odds of detecting neoplastic lesions by target biopsy compared to chromoendoscopy (OR 0.27, CI 0.08-0.91; moderate certainty evidence) (Table 3), with the 95% prediction interval including the null value (Figure S4F). There was no evidence of detectable differences in any other comparison between interventions (low to very low certainty evidence) (Table 3). Chromoendoscopy ranked as the best endoscopic technique to detect neoplastic lesions by target biopsy (Figure S6B), with limited confidence in hierarchical ranking for study limitations and imprecision.
The network for procedural time included five trials (675 participants) evaluating the effect of four endoscopic techniques (chromoendoscopy, high definition white-light endoscopy, standard definition white-light endoscopy and full spectrum high definition white-light endoscopy) (Figure S2B). Standard definition white-light endoscopy was probably associated with a shorter procedural time than chromoendoscopy (MD −14.81 minutes, CI −25.03 to −4.06 minutes; moderate certainty evidence) (Table S6), with the 95% prediction interval including the null value (ie ‘0’) (Figure S4G). There was no evidence of detectable differences in procedural time in any other comparison between endoscopic techniques (low to very low certainty evidence) (Table S6). Standard definition white-light endoscopy ranked as the best endoscopic technique for procedural time (Figure S6C). There was low confidence in hierarchical ranking for study limitations and imprecision.
We could not assess publication bias for number of neoplastic lesions detected by target biopsy and procedural time due to the paucity of trials (<10 studies).
3.4.6 Adverse events
Eight trials34-37, 39, 42, 47, 48 involving 1050 participants evaluated adverse events associated with the different endoscopic techniques. We could not perform a network meta-analysis for this outcome because six of these trials did not report any endoscopic procedure-related side effects. The remaining two studies42, 48 described a total of 28 participants with adverse events: 24 complained of abdominal pain or discomfort with high definition white-light endoscopy (14 participants), full spectrum high definition white-light endoscopy (nine participants) or autofluorescence (one participant); three had intraprocedural mild bleeding with chromoendoscopy (two participants) or autofluorescence (one participant); and one experienced bowel perforation with chromoendoscopy.
3.4.7 Other outcomes
No trial provided data on health-related quality of life, all-cause mortality, colorectal cancer-related mortality and interval colorectal cancer.
4 DISCUSSION
This network meta-analysis found that standard definition white-light endoscopy and i-SCAN probably had lower odds of identifying participants with neoplastic lesions compared to chromoendoscopy. There was no evidence of detectable differences in the odds of identifying participants with neoplasia between chromoendoscopy and any other endoscopic technique, although chromoendoscopy had higher odds of detecting participants with any type of colonic lesion, including non-neoplastic ones, compared to autofluorescence and i-SCAN. Standard definition white-light endoscopy had lower odds of detecting nonpolypoid neoplastic lesions than full spectrum high definition white-light endoscopy, narrow band imaging, chromoendoscopy and high definition white-light endoscopy. There was no evidence of detectable differences in the odds of identifying nonpolypoid neoplasia in any other comparison between endoscopic techniques. Full spectrum high definition white-light endoscopy ranked as the best technique for identifying both participants with neoplasia and nonpolypoid lesions. This technique showed higher odds of detecting participants with neoplastic lesions than standard definition white-light endoscopy, i-SCAN and FICE. Full spectrum high definition white-light endoscopy also had higher odds of identifying nonpolypoid neoplasia compared with all other techniques, although these comparisons did not reach statistical significance and were mainly derived from indirect comparisons. There was no evidence of differences between endoscopic techniques in identifying low-grade or high-grade dysplastic lesions. Standard definition white-light endoscopy probably had lower odds of detecting neoplastic lesions by target biopsy and shorter procedural time compared to chromoendoscopy. Based on our findings, chromoendoscopy, high definition white-light endoscopy, narrow band imaging, autofluorescence, FICE and full spectrum high definition white-light endoscopy may be alternative endoscopic techniques for dysplasia surveillance in people with IBD, whereas standard definition white-light endoscopy and i-SCAN showed lower efficacy than chromoendoscopy, the recommended first-line approach in this setting. However, the discrepancy in our results between the estimated CI and the 95% prediction interval for the comparison between i-SCAN and chromoendoscopy indicates that i-SCAN efficacy requires further investigation. Additionally, considering the high prevalence of nonpolypoid neoplasia in people with IBD (23%-67%),51, 52 standard definition white-light endoscopy should not be used for neoplasia surveillance in this population. Finally, our results suggest that full spectrum high definition white-light endoscopy may represent the first-line endoscopic approach for dysplasia surveillance in IBD.
Two recently published systematic reviews with network meta-analysis compared the efficacy of different endoscopic techniques for dysplasia surveillance in people with IBD.53, 54 Imperatore et al53 found a higher likelihood of identifying dysplasia with chromoendoscopy compared to white-light endoscopy (OR 2.12, CI 1.18-5.23), with no significant differences in any other comparison between endoscopic techniques. The authors analysed standard definition and high definition white-light endoscopy in the same intervention arm, although these techniques require distinct endoscopes which provide images at different resolution.55 Our network meta-analysis confirmed that chromoendoscopy probably had higher odds of detecting participants with neoplastic lesions when compared with standard definition white-light endoscopy. Additionally, the authors included both randomised (11 trials) and observational studies in their review and performed a network meta-analysis for the single outcome of dysplasia detection, without providing network plots, endoscopic technique rankings or assessment of loop-specific and global network consistency. Bessissow et al54 did not identify any statistically significant difference in the detection of dysplasia in any comparison between endoscopic techniques, with chromoendoscopy ranking as the best technique for dysplasia identification. These findings differ from our analysis because we found higher odds of identifying participants with neoplasia with chromoendoscopy compared to standard definition white-light endoscopy and i-SCAN, with full spectrum high definition white-light endoscopy ranking as the best endoscopic technique for this outcome. However, the authors of this network meta-analysis included a low number of randomised trials (eight studies), which investigated only four endoscopic techniques (ie chromoendoscopy, narrow band imaging, high definition white-light endoscopy and standard definition white-light endoscopy) for dysplasia surveillance in IBD. They also included in their analysis data from one crossover trial31 that did not provide the results for the first phase of the study and performed a network meta-analysis for the single outcome of dysplasia detection, without providing the assessment of global network consistency. Finally, both previous network meta-analyses53, 54 did not report the efficacy of different endoscopic techniques on identifying morphological and histological subtypes of dysplastic lesions, detecting neoplasia by target biopsy, procedural time and adverse events.
We compared our findings with recommendations from key international guidelines, which indicate chromoendoscopy with target biopsy as the first-line approach for dysplasia surveillance in people with IBD.8-11, 13, 14 White-light endoscopy with random biopsies is considered appropriate if chromoendoscopy is not available, while other endoscopic techniques are not recommended.7, 9, 11, 13 Our findings support the superiority of chromoendoscopy over standard definition white-light endoscopy and i-SCAN in detecting people with neoplasia. Standard definition white-light endoscopy also showed lower odds of detecting nonpolypoid neoplastic lesions compared with chromoendoscopy, high definition white-light endoscopy, narrow band imaging and full spectrum high definition white-light endoscopy. In contrary, we found similar efficacy in identifying people with neoplasia and detecting nonpolypoid neoplastic lesions when comparing chromoendoscopy with high definition white-light endoscopy, narrow band imaging, full spectrum high definition white-light endoscopy, FICE or autofluorescence. Thus, considering the prevalence of nonpolypoid dysplasia in this population (23%-67%),51, 52 white-light endoscopy with standard definition endoscopes should no more be proposed for dysplasia surveillance in IBD, while high definition white-light endoscopy, narrow band imaging, full spectrum high definition white-light endoscopy, FICE and autofluorescence may represent reasonable alternatives to chromoendoscopy. Finally, based on our results, we contend that more evidence is needed to clearly assess the efficacy of i-SCAN for dysplasia surveillance in IBD and evaluate the possible role of full spectrum high definition white-light endoscopy as the first-line endoscopic approach in this setting.
This systematic review with network meta-analysis compares eight different endoscopic techniques for dysplasia surveillance in IBD, with a comprehensive evaluation of efficacy and harms of these interventions. We assessed the totality of evidence, including a larger number of randomised trials (18 studies) in our analysis than existing network meta-analyses,53, 54 and limited analyses to randomised trials without incorporating cohort studies as in one53 of the two previous reviews.
The study had limitations that should be considered in the interpretation of the results. First, eight out of 18 randomised trials included in our analysis had unclear/high risk of bias for most domains. The study limitations related to these trials and the imprecision in effect estimates, due to the amount of data available for analyses, were the main reasons for downgrading the evidence of certainty to low or very low for most endoscopic techniques. Second, we estimated very low odds ratio values and wide CI in most comparisons between FICE and other endoscopic techniques. This technique has been investigated in a single trial with a crossover study design, in which no neoplastic lesions were identified using FICE in the first phase of the trial.35 This may have reduced precision in comparative estimates for FICE on assessed outcomes. Third, we could not perform subgroup analysis by IBD subtype (ulcerative colitis or Crohn's colitis) because 11 out of 18 trials included in our analyses enrolled only ulcerative colitis participants, while the remaining seven studies enrolled a mixed ulcerative and Crohn's colitis population and did not provide outcome data by disease subtype. Fourth, we could not analyse the diagnostic accuracy of high definition white-light endoscopy compared to dye-based and virtual chromoendoscopy on optical discrimination between neoplastic and non-neoplastic lesions. Only one36 of the randomised trials included in our network meta-analysis assessed this endpoint, using histology as the reference standard. The authors found similar sensitivity and specificity to predict histology of colonic lesions for high definition white-light endoscopy, chromoendoscopy and i-SCAN. Fifth, we could not evaluate the effect of dysplasia surveillance on colorectal cancer-related mortality, as no selected trial included a follow-up period to capture this outcome. There is a need for high-quality trials investigating the impact of surveillance programs in people with IBD on colorectal incidence and mortality, since unequivocal evidence for the benefit of surveillance endoscopy on these relevant clinical outcomes is still lacking.10, 13 Finally, adverse events, all-cause mortality, interval cancer and health-related quality of life could not be analysed in the present network meta-analysis because few or no trials reported data on these outcomes. Gulati et al35 analysed participants’ experience in their study investigating chromoendoscopy versus FICE. However, they did not provide the results for the first phase of this randomised crossover trial.
The results of this network meta-analysis indicate that chromoendoscopy, high definition white-light endoscopy, narrow band imaging, autofluorescence, FICE and full spectrum high definition white-light endoscopy may be appropriate endoscopic techniques for dysplasia surveillance programs in people with IBD. On opposite, standard definition white-light endoscopy should not be proposed in this setting. Thus, if standard definition white-light endoscopy is the only available technique, patients with IBD should be referred to specialised endoscopic centres for dysplasia surveillance.
The efficacy of i-SCAN in dysplasia surveillance should be further investigated in randomised trials, based on the discrepancy in our results between the estimated CI and the 95% prediction interval for the comparison between this technique and chromoendoscopy. Full spectrum high definition white-light endoscopy may represent the best endoscopic technique to identify both participants with neoplasia and nonpolypoid lesions. However, currently only few endoscopic centres have full spectrum high definition white-light endoscopy available. Additionally, evidence supporting this technique mainly derives from indirect comparisons. Thus, there is a need for high-quality multi-centre randomised trials comparing full spectrum high definition white-light endoscopy with chromoendoscopy, which is currently recommended as the first-line endoscopic approach, to clearly assess if this technique may play a predominant role in dysplasia surveillance of IBD in the near future.
5 CONCLUSIONS
Chromoendoscopy, high definition white-light endoscopy, narrow band imaging, autofluorescence, FICE and full spectrum high definition white-light endoscopy may be reasonably comparative techniques for dysplasia surveillance in IBD, while standard definition white-light endoscopy and i-SCAN are probably less effective. Full spectrum high definition white-light endoscopy may represent the first-line approach.
ACKNOWLEDGEMENTS
We thank Ruth Mitchell, Trial Search Coordinator, for development of search strategies, and Jasper L. A. Vleugels, MD and Paulo Freire, MD, for providing additional information.
Declaration of personal interests: None.
AUTHORSHIP
Guarantor of the article: Giovanni F. M. Strippoli.
Author contributions: Andrea Iannone, Suetonia C. Palmer and Giovanni F. M. Strippoli were involved in study concept and design; Andrea Iannone and Marinella Ruospo were involved in acquisition of data; Andrea Iannone, Suetonia C. Palmer and Marinella Ruospo were involved in statistical analysis and interpretation of data; Andrea Iannone, Suetonia C. Palmer (native English speaker), Marinella Ruospo and Giovanni F. M. Strippoli were involved in drafting the manuscript; Mariabeatrice Principi, Michele Barone, Alfredo Di Leo and Giovanni F. M. Strippoli were involved in critical revision of the manuscript for important intellectual content; Suetonia C. Palmer and Giovanni F. M. Strippoli were involved in study supervision. Andrea Iannone, Marinella Ruospo, Suetonia C. Palmer, Mariabeatrice Principi, Michele Barone, Alfredo Di Leo and Giovanni F. M. Strippoli approved the final version of the manuscript.
