The impact of linked color imaging on adenoma detection rate in colonoscopy: a systematic review and meta-analysis
Article information
Abstract
Background/Aims
Colorectal cancer prevention relies on surveillance colonoscopy, with the adenoma detection rate as a key factor in examination quality. Linked color imaging (LCI) enhances lesion contrast and improves the examination performance. This systematic review and meta-analysis aimed to evaluate the effect of LCI on adenoma detection rate in adults who underwent colonoscopy.
Methods
We searched the Medline, PubMed, BIREME, LILACS, and Scientific Electronic Library Online databases for randomized controlled trials comparing the use of LCI versus white light (WL), published up to March 2023. The outcomes included lesion characteristics, number of adenomas per patient, and the additional polyp detection rate.
Results
Sixteen studies were included in the analysis, which showed that LCI was more accurate than WL in detecting adenomas, with an increased number of adenomas detected per patient. Although LCI performed well in terms of lesion size, morphology, and location, the subgroup analyses did not reveal any statistically significant differences between LCI and WL. The addition of LCI did not result in significant improvements in the detection of serrated lesions, and there were no differences in the withdrawal time between groups.
Conclusions
LCI has been shown to be effective in detecting colonic lesions, improving the number of adenomas detected per patient and improving polyp detection rate without negatively affecting other quality criteria in colonoscopy.
INTRODUCTION
Colorectal cancer (CRC) is the third most common malignancy worldwide and the second leading cause of cancer-related death.1 The disease often progresses from adenomatous polyps, and colonoscopic detection of these polyps has proven to be the most effective method to reduce CRC risk and mortality.2 The adenoma detection rate (ADR), defined as the proportion of colonoscopies detecting at least one adenomatous lesion, inversely impacts the risk of CRC.3 An increase of approximately 1% in ADR leads to a 3% reduction in CRC incidence between surveillance intervals.4 Despite this, conventional colonoscopy has a notable limitation, with an undetected polyp rate of up to 20%.5
To improve the detection of premalignant colonic lesions, new technologies have been implemented. Image enhancement techniques have become popular due to their ease of use and lack of special preparations or additional devices. While narrow-band imaging (NBI) and blue laser imaging (BLI) initially showed good results in differentiating premalignant from benign lesions by analyzing polyp surface patterns,6 their effectiveness in increasing the polyp detection rate (PDR) compared to conventional white light (WL) colonoscopy has been controversial.7,8
In 2014, Fujifilm Co. Ltd. developed linked color imaging (LCI), a technology that provides superior illumination of the colorectal lumen compared to other virtual chromoendoscopy techniques.9,10 LCI enhances the differentiation of blue, red, and green color spectra, highlighting color changes in the colonic mucosa and increasing hemoglobin contrast, thereby making intestinal lesions appear more reddish than the surrounding mucosal surface.9
LCI is used with Fujifilm's Laser Endoscopic System (LASEREO) and Light-Emitting Diode Endoscopic System (ELUXEO), which differ in their light sources.7,11 LASEREO employs laser light generating two wavelengths (410±10 nm and 450±10 nm),12 while ELUXEO uses LEDs emitting four wavelengths (blue-violet, blue, green, and red).13 The 410-nm violet light of LCI penetrates superficially, highlighting neoplastic lesions as red and adjacent mucosa as violet due to higher blood vessel concentration in neoplastic areas.10
LCI also accurately distinguishes between adenomatous lesions and inflammatory changes, which both appear reddish under WL. Under LCI, inflammatory changes appear purple, while neoplastic lesions remain red.10,11 This study evaluated the impact of LCI on ADR and its benefits by reviewing randomized controlled trials (RCTs) comparing LCI with WL in adults undergoing colonoscopy.
This study evaluated the impact of LCI on ADR and the benefits associated with this technology by reviewing randomized controlled trials (RCTs) that compared LCI with WL in adults undergoing colonoscopy.
METHODS
Following Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines,14 this systematic review and meta-analysis addressed the question: “Does virtual chromoendoscopy with LCI technology increase ADR in colonoscopy?”. The review protocol was registered with the International Prospective Register of Systematic Reviews (CRD42023438359).
Search strategy
An experienced librarian developed and an investigator (BHFZ) reviewed the search strategy. We searched MEDLINE, PubMed, Latin American and Caribbean Center on Health Sciences Information (BIREME), Latin American and Caribbean Health Sciences Literature (LILACS), and Scientific Electronic Library Online (SciELO) databases for articles published up to March 2023 using Medical Subject Headings (MeSH), Descriptors in Health Sciences (DeCS), and natural language search terms: colonoscopy, chromoendoscopy, quality colonoscopy, Linked Color Imaging, polyp detection, and adenoma detection (Table 1).
Search strategy
Selection criteria
Two independent reviewers (BHFZ and JCA) compiled search results using the PICO approach: Patients (≥18 years undergoing colonoscopy), Interventions (use of LCI), Comparators (use of WL), and Outcomes (ADR, detection rates of serrated lesions, colonic lesions ≤5 mm and >5 mm, lesions in the right and left colon, flat and non-flat lesions, number of adenomas per patient, additional PDR, and withdrawal time). They independently screened titles, abstracts, and full-text articles based on the relationship between LCI and PDR compared to WL. Eligible studies were RCTs in English reporting results of LCI versus WL in adults undergoing colonoscopy. Case reports, reviews, retrospective studies, and animal experiments were excluded.
Study definitions
The ADR was defined as the proportion of colonoscopies during which at least one adenoma was detected, considering all colonoscopies. Serrated lesion detection rate was defined as the proportion of colonoscopies during which at least one sessile serrated adenoma/polyp was detected considering all colonoscopies. The number of adenomas per patient was calculated as the total number of adenomas divided by the total number of patients. The number of flat lesions was defined as the total number of lesions with a diameter of at least twice the height of the lesion in each study. In many studies, flat lesions were defined as IIa, IIb, and IIc lesions according to the Paris classification.15 Non-flat lesions included sessile, pedunculated, and subpedunculated lesions classified as Is, Ip, and Isp, respectively, according to the Paris classification.15 Some studies7,16 only differentiated lesions morphologically into polypoid and nonpolypoid. Additional PDR was defined as the number of polyps detected during the second examination divided by the total number of procedures.
By definition, the right colon accommodates segments of the cecum and ascending colon, whereas the left colon anatomically corresponds to the descending and sigmoid colon. Withdrawal time was defined as the amount of time spent viewing the colonoscope withdrawn from the cecum.
Data extraction and study outcomes
The selected articles were assessed for ADR and serrated lesion detection rates. The benefit of LCI in detection rates was evaluated by considering lesion size (≤5 mm or >5 mm), location (right or left colon), and morphology, and individualizing the results for the detection rate of flat and non-flat lesions. Additionally, the number of adenomas per patient and additional PDR were counted when the study design allowed for such an assessment.
For crossover study designs, a duplicate assessment of the colonic mucosa was performed with both WL and LCI. For proper statistical comparison, only the number of lesions found in the initial examination was considered and included. Additionally, withdrawal times with and without this chromoendoscopy technology were evaluated. For the triple-arm study, data were extracted from the LCI and WL groups.
Data extracted from the studies included the first author, year of publication, study period, type of study, study design, endoscopic system, number of participants, baseline demographics (age and sex), indications for colonoscopy, total number of study participants, and number of colonoscopists.
Two independent reviewers (BHFZ and JCA) tabulated the extracted data on a Microsoft Excel spreadsheet (Microsoft), and any discrepancies in data collection were resolved through discussion and consensus between reviewers.
Bias assessment
Using the Cochrane tool for assessing risk of bias in RCTs,17 the highlighted domains included the randomization process (D1), deviations from the intended interventions (D2), missing outcome data (D3), measurement of outcomes (D4), and selection of reported results (D5). The bias assessment of the primary outcome revealed considerations in domain D2 stemming from the inability to blind the performing colonoscopist who inevitably became aware of the ongoing intervention. Some studies also encountered issues with pre-registration of proposed outcomes and problems with randomization.
Statistical analysis
Meta-analyses were performed to determine outcomes of interest. The combined results for continuous measures are presented as the mean difference, and aggregated results for dichotomous outcome measures are presented as the risk ratio (RR) between the LCI and WL methods. All analyses were performed using the meta package (ver. 6.5-0) in the R programming language. Random-effects models were used for all analyses, and this choice was based on their plausible assumptions in medicine and the extent of heterogeneity among studies. The results of the meta-analysis are shown as forest plots. Statistical significance was set at p<0.05.
1) Subgroup analysis
Post hoc subgroup analyses were conducted when common trends in interventions considered clinically relevant emerged. These analyses included lesion size (≤5 mm and >5 mm), morphology (flat and non-flat), and location (right and left).
2) Publication bias assessment
Funnel plots were inspected to assess publication bias in the meta-analyses and the findings were confirmed using Egger’s regression test. If publication bias was identified, robustness was tested using the trim-and-fill method.
The leave-one-out method was used to explore heterogeneity for the primary outcome, systematically removing one study at a time, and ultimately assessing the effect size and new heterogeneity encountered. Heterogeneity among studies was quantified using I2 statistics.
RESULTS
Study selection
The study selection process is illustrated in Figure 1. The database search yielded 279 studies. After title and abstract screening, 24 studies were retrieved for full-text review, 16 of which met the eligibility criteria and were included in this review.7,16,18-31
Flowchart of study selection. RCT, randomized controlled trial.
Characteristics of the studies
The characteristics of the included studies are summarized in Table 2.7,16,18-31 The RCTs were published between 2017 and 2023 and were conducted in Japan, China, Brazil, Hungary, Thailand, and Italy. Two international RCTs were conducted in multiple countries,30,31 whose objectives were to analyze the effectiveness of LCI in the detection of polyps compared with examinations performed with WL alone.
Characteristics of the included studies
Most studies used the LASEREO system,7,16,18,20-22,24,25,27-29 whereas three studies used ELUXEO as the prevailing system,19,23,26 with LED as the endoscopic light source. These studies did not clearly explain or distinguish between the two lighting systems, often raising questions about whether this could be a point of divergence in the findings. In 2023, Okada et al.13 compared these two systems using the BLI mode to assess image quality for determining the Japan NBI expert team classification and found a concordance rate of 92.5%. The weighted κ‐statistic was 0.99. Similar studies have not been conducted on the use of LCI. Extrapolating from these results, we conclude that there was no significant difference or interference in the outcomes between the laser and LED systems as light sources.
In three studies,13,18,19 results were extracted solely from the analysis of the right colon, and in another study,22 analysis was extended to the splenic flexure, allowing for assessment of the benefit of incorporating LCI to detect lesions that might go unnoticed under WL. Min et al.,21 Dos Santos et al.,27 and Suzuki et al.,31 evaluated the results of LCI and WL, individualizing each segment of the colon, while the others included an analysis of the entire colon from the cecum to the rectum.
Seven studies were multicenter,16,18,20,21,29-31 while the remaining studies were conducted in a single center. Most studies included patients with multiple indications for colonoscopy, surveillance, and screening to investigate gastrointestinal symptoms and positive fecal occult blood tests. The studies exhibited notable similarities regarding the sex and mean age of participants. Only two studies evaluated the detection rate of serrated lesions,16,22 considered by some to be more difficult to visualize because of their similarity to the adjacent mucosa.32
The included studies exhibited differences in design, with some employing a parallel design,7,16,20,23-31 whereas others used a crossover design.19,21,22 Yoshida et al.,18 conducted an initial examination with WL, followed by additional assessments with either WL or LCI, combined with an extended 30-second duration for the second evaluation. Conversely, Hasegawa et al.,25 adopted a design in which the second observation was standardized using WL. Houwen et al.,30 included only adult patients with Lynch syndrome undergoing surveillance colonoscopy.
Adenoma detection rate
The primary outcome of this study was the comparative evaluation of ADR using LCI and WL. Our systematic review included 15 studies7,16,18-21,23-31 involving observations of 10,558 patients (5,200 in the LCI group and 5,358 in the WL group). The total number of recorded events was 4,751. The meta-analysis revealed that ADR using LCI was 1.20 (95% confidence interval [CI], 1.13–1.28) times higher than that using WL (Fig. 2). The heterogeneity among the studies was moderate (I2=44%, p=0.03).
Quantitative analysis of linked color imaging (LCI) versus white light (WL) for the primary outcome: adenoma detection rate. RR, risk ratio; CI, confidence interval.
Table 3 shows the analysis of the influence of each study on heterogeneity using the leave-one-out method.7,16,18-21,23-31 When the study by Miyaguchi et al.,20 was omitted, a substantial decrease in heterogeneity was observed, reaching values close to 24%. However, there were no alterations in the primary effect. In a sensitivity analysis using the random-effects model, the results appeared consistently in favor of LCI (RR, 1.20; 95% CI, 1.13–1.28), and the true effect was likely close to this (Fig. 3).
Leave-one-out analysis
Sensitivity analysis of linked color imaging (LCI) versus white light (WL) for adenoma detection rate. RR, risk ratio; CI, confidence interval.
A funnel plot was constructed to investigate publication bias, with the results presented (Fig. 4); Egger’s linear regression test was used to formally assess the presence of asymmetry in this plot. The results of this analysis indicated no effect or significant asymmetry in the funnel plot (beta=0.4851, p=0.5067).
Funnel plot to investigate publication bias for adenoma detection rate.
Serrated lesions
Among the studies that comparatively assessed total colonoscopy with and without LCI, the serrated lesion detection rate was reported in 11 studies.7,16,18-20,22,24,25,27,30,31 In the analysis of the literature, we observed favorable outcomes for LCI, with low heterogeneity (I2=0%, p=0.51) but with no statistical significance (RR, 1.11; 95% CI, 0.86–1.44) (Fig. 5).
Quantitative analysis of linked color imaging (LCI) versus white light (WL) for serrated lesions. SD, standard deviation; MD, mean difference; CI, confidence interval.
Number of adenomas per patient
Among the selected studies, the number of adenomas per patient was evaluated in nine RCTs,16,21,23-26,29-31 including observations from 7,845 patients (3,837 in the LCI group and 4,008 in the WL group), with a greater number of lesions detected in the groups in which LCI was used. In the analysis of the average number of adenomas per patient, the overall average of all studies was 0.26 (95% CI, 0.16–0.37) more adenomas per patient in the LCI group and heterogeneity was low (I2=28%, p=0.20) (Fig. 6).
Quantitative analysis of linked color imaging (LCI) versus white light (WL) for mean number of adenomas per patient. SD, standard deviation; MD, mean difference; CI, confidence interval.
Lesion morphology
Analyses were conducted to differentiate between flat and non-flat lesions. These analyses were feasible in ten studies,7,18-21,24,27,28,30,31 yielding results that favored LCI in the detection rate of flat lesions, but with no statistical significance. For non-flat lesions, specifically sessile, pedunculated, and subpedunculated lesions, the results did not favor LCI. Depressive lesions were excluded from the analysis. It is noteworthy that, in this analysis, the results were not isolated solely for adenomas, converging towards the macroscopic morphological characterization of different colonic lesions, primarily based on the Paris classification.15 LCI demonstrated a tendency towards superior detection rates for flat lesions compared with non-flat lesions. In a comparative exploratory subgroup analysis, no statistically significant results were observed (RR, 1.02; 95% CI, 0.97–1.09; χ21, 2.62; p=0.11) (Fig. 7).
Comparative exploratory subgroup analysis for morphology comparing flat and non-flat lesions. LCI, linked color imaging; WL, white light; RR, risk ratio; CI, confidence interval.
Lesion size
We subdivided observations using a 5-mm size threshold for colonic lesions. Results for size analyses were identified in seven studies,7,15,19-21,24,30 demonstrating that, overall, LCI exhibited a trend towards the detection of lesions ≤5 mm, but with no statistical significance. For larger lesions, the results did not favor LCI. In a comparative exploratory subgroup analysis, no statistically significant results were observed (RR, 1.10; 95% CI, 0.95–1.28; χ21=2.82; p=0.09) (Fig. 8).
Comparative exploratory subgroup analysis for size comparing lesions ≤5 mm and >5 mm. LCI, linked color imaging; WL, white light; RR, risk ratio; CI, confidence interval.
Lesion location
In eight studies,7,20,21,24,27,28,30,31 it was possible to assess ADR in the right and left colons comparatively. Proportionally higher values were observed in the right than in the left colon (11% vs. 2%). However, these differences were not statistically significant (RR, 1.06; 95% CI, 0.94–1.20; χ21=0.38; p=0.54) (Fig. 9).
Comparative exploratory subgroup analysis for location comparing lesions in right and left colon. LCI, linked color imaging; WL, white light; RR, risk ratio; CI, confidence interval.
Additional polyp detection rate
Three studies19,21,22 used a crossover design and evaluated the impact of LCI on lesion detection in patients who had already undergone an initial WL examination. For comparison, PDR with WL was also evaluated after the initial comparison with LCI alone. The additional PDR between studies was 60% higher for the group in which a second observation was performed with LCI than with WL (RR, 1.99; 95% CI, 1.30–3.06). Heterogeneity was low (I2=0%, p=0.53) (Fig. 10).
Additional polyp detection rate. LCI, linked color imaging; WL, white light; RR, risk ratio; CI, confidence interval.
Withdrawal time
Withdrawal time was defined as the removal time after cecal intubation. This period was evaluated by comparing the use of LCI and WL. The LCI group presented a median time 0.14 minutes longer than the WL group, but the difference was not statistically significant between the methods (RR, 0.15; 95% CI, −0.03 to 0.34). Heterogeneity was high (I2=86%, p=0.10) (Fig. 11).
Withdrawal time. LCI, linked color imaging; WL, white light; SD, standard deviation; MD, mean difference; CI, confidence interval.
DISCUSSION
LCI is an innovative and promising technological resource that leads to better CRC screening results, and a growing number of studies are aimed at evaluating and quantifying its benefits. Pre- and post-processing programs produce higher-quality, clearer, and sharper images with a unique color pattern that allows neoplastic lesions to appear prominently with a more reddish color than the rest of the adjacent mucosa. By increasing the staining contrast and accentuating certain wavelength spectra, these lesions become more detectable by the examiner.9,10 Current studies corroborate that the indiscriminate application of LCI increases the detection of polyps and colonic lesions across all histologies and sizes. Our meta-analysis consolidates the growing body of research aimed at assessing and quantifying the benefits of LCI, as evidenced by earlier meta-analyses by Shinozaki et al.,10 Wang et al.,33 and Sun et al.,34 which comprehensively summarized these advantages.
In the evaluation of our primary outcome, although we found results consistent with those previously reported, which primarily demonstrated an increase in ADR with the addition of LCI compared with WL, our subgroup analysis for the assessment of secondary outcomes revealed results that slightly diverged from those previously reported. Overall, although the results favored LCI, we did not find any statistically significant differences in lesion location, morphology, size, or histological type.
A comparative sub-analysis of lesion location demonstrated that LCI enabled greater detection of lesions in the right colon. However, no statistically significant differences were observed, consistent with the results reported by Wang et al.,33 but differs from those demonstrated by Sun et al.,34 who stratified the analyses by comparing lesions detected in proximal and distal colons.
In addition, we evaluated the effect of LCI on the detection of lesions based on lesion size and morphology. Although the addition of this feature improved the results, there was no statistical significance in the detection of flat lesions, with preferentially lateral growth, or lesions ≤5 mm. Both types are considered more challenging to visualize using WL alone. Wang et al.,33 using a cutoff point of 10 mm demonstrated that the use of LCI provided even more pronounced and favorable results for the detection of these lesions. In contrast to our findings, Sun et al.,34 reported statistically significant favorable results for LCI in the detection of flat lesions. A possible explanation for our discrepant results could be the more comprehensive inclusion of data in our review, which was approximately twice the number of studies included in previous reviews. To minimize publication bias, we attempted to include as many studies as possible, adhering to pre-established selection criteria.
In addition to the global ADR, we also individually evaluated the detection rate of serrated lesions, which are often difficult to assess because they present surface patterns similar to those of the adjacent colonic mucosa,32 accounting for 15% to 30% of the causes of CRC.35-38 The results indicate a potential advantage with the incorporation of virtual chromoendoscopy technology; however, in our analysis, the observed differences did not reach statistical significance, consistent with the findings reported by Wang et al.,33 In contrast to the aforementioned meta-analyses, our study encompassed a broader range of investigations evaluating the number of serrated lesions. Our comprehensive approach involved the consideration of crossover study designs, exemplified in the study by Paggi et al.,19 wherein duplicate assessments of colonic mucosa with both WL and LCI necessitated the inclusion of only lesions detected in the initial examination for proper statistical comparison. The inclusion of diverse studies beyond those specifically designed to detect serrated lesions may explain the disparities observed in our results. Our results are consistent with those reported by Wang et al.,33 Another possible explanation for the results found, in addition to the inclusion of a greater number of studies, is that serrated lesions are generally covered by a mucus cap, making it difficult to visualize them regardless of the technology used. To optimize these strategies, it is suggested that the mucus cap be removed in advance.39
In these crossover studies,19,21,22 we observed significant benefits when LCI was used as an additional tool in patients who had already undergone prior evaluation with WL. This technology can overcome some of the deficiencies inherent in the conventional method, which is associated with a higher rate of unnoticed lesions. Instead of evaluating the adenoma miss rate, we assessed the additional PDR when incorporating LCI and observed that a second evaluation with the addition of this feature significantly increased PDR. Additionally, in the analysis of the number of adenomas per patient, we extracted data from more studies than those included in previously published reviews.
We also evaluated withdrawal time as an additional outcome. The results showed that despite a slight increase in the time to remove the colonoscope from the cecum when using LCI, there was no statistical significance in these data. Furthermore, considering the higher ADR with LCI, the longer procedure duration could be attributed to the additional time involved in the removal of these lesions.
The main limitations of our study are inherent in the heterogeneity of the included studies with different designs, populations, multiple indications for colonoscopy, and primary outcomes. Altogether, these limitations made it difficult to homogenize the data for adequate meta-analysis and statistical analysis. Nevertheless, this more comprehensive approach minimized possible publication bias and provided more robustness to the outcomes evaluated.
It was concluded that LCI, an increasingly accessible and available technology, should be seen as a valuable tool in the arsenal of CRC prevention, mainly by reducing the rate of unnoticed lesions associated with interval CRCs,10 without negatively affecting other quality indicators in colonoscopy. However, our study stands out from those previously published by demonstrating that, in subgroup analyses, despite a favorable trend towards the use of LCI, no statistically significant differences were observed compared with WL.
Notes
Ethical Statements
Not applicable.
Conflicts of Interest
Luiz Ronaldo Alberti is a member of editorial board of Clinical Endoscopy. The other authors have no potential conflicts of interest.
Funding
None.
Author Contributions
Conceptualization: BHFZ, LRA, FGR, JCA; Data curation: BHFZ, JCA; Formal analysis: NCJ; Funding acquisition: BHFZ; Investigation: BHFZ, OMN, GCOC, JCA; Methodology: BHFZ, MMFD, FRG; Project administration: BHFZ; Resources: BHFZ, FGR; Software: NCJ; Supervision: BHFZ, CEOS, LRA, FGR, JCA; Validation: CEOS, LRA, FGR, JCA; Visualization: all authors; Writing–original draft: BHFZ, MMFD, FRG, OMN, GCOC; Writing–review & editing: all authors.