Detection of colorectal adenomas with texture and color enhancement imaging: Multicenter observational study

Abstract

Objectives

We aimed to evaluate the efficacy of texture and color enhancement imaging (TXI), which allows the acquisition of brighter images with enhanced color and surface structure in colorectal polyp detection compared to white light imaging.

Methods

Patients who underwent colonoscopy with repeated ascending colon observation using TXI and white light imaging between August 2020 and January 2021 were identified in three institutions. The outcomes included the mean number of adenomas detected per procedure (MAP), adenoma detection rate (ADR), and ascending colonic adenoma miss rate (Ac‐AMR). Logistic regression was used to determine the effects of the variables on the outcomes.

Results

We included 1043 lesions from 470 patients in the analysis. The MAP, ADR, flat polyp detection rate, and Ac‐AMR in TXI and white light imaging were 1.5% (95% confidence interval 1.3–1.6%) vs. 1.0% (0.9–1.1%), 58.2% (51.7–64.6%) vs. 46.8% (40.2–53.4%), 66.2% (59.8–72.2%) vs. 49.8% (43.2–56.4%), and 17.9% (12.1–25.2%) vs. 28.2% (20.0–37.6%), respectively. TXI, age, withdrawal time, and endoscopy type were identified as significant factors affecting the MAP and the ADR using multivariate regression analysis.

Conclusions

Our study indicates that TXI improve the detection of colorectal neoplastic lesions. However, prospective randomized trials are required to confirm these findings.

INTRODUCTION

Early detection and resection of a colorectal adenoma, a precancerous lesion, are crucial for preventing the development of colorectal cancer (CRC). Colonoscopy is a reliable method for detecting, diagnosing, and resecting adenomas and early‐stage cancers. However, it has been reported that conventional endoscopic systems that use white light imaging (WLI) miss approximately 25% of colonic lesions. ¹ , ² , ³ Lesions may be missed because of technical factors and factors related to the lesion itself, such as flat and depressed morphologies that are difficult to detect. Therefore, improvements in colonoscopy procedures and various colonoscopy modalities are considered to overcome these limitations. These include improvements in bowel preparation, longer withdrawal time, repeated examination of the right colon, cap‐assisted colonoscopy, and image‐enhanced endoscopy (IEE). Despite its significant advantages in detection, dye‐based IEE is not widely used because it is time‐consuming and cumbersome. Equipment‐based IEE includes digital and optical–digital methods that involve image enhancement through signal processing and an image‐processing algorithm. Several reports have described its usefulness in clinical practice. ⁴ , ⁵

Narrowband imaging (NBI) is a widely used technique. Although NBI has been acknowledged for its usefulness in the qualitative diagnosis of colorectal polyps, its effectiveness in detecting colorectal lesions is difficult to demonstrate. ⁶ , ⁷ , ⁸ Linked color imaging (LCI) is a relatively new IEE method that has shown promising results in detecting colorectal lesions. ⁹ , ¹⁰ However, only a few reports on IEE technologies improve the detection of lesions in the right colon, particularly sessile serrated lesions (SSLs).

The new EVIS X1 colonoscopy system (Olympus Medical Systems, Tokyo, Japan) features the novel “texture and color enhancement imaging (TXI)” mode designed to enhance three image factors in WLI (texture, brightness, and color). The system has two settings for TXI: mode 1, with color enhancement, and mode 2, without color enhancement (Fig. 1). TXI improves the visibility of lesions by using image processing to correct the dark portions of WLI and highlight structural and color changes in concavities and convexities. ¹¹ , ¹² , ¹³ However, the magnitude of the additional effect of detecting colorectal lesions in clinical practice remains unclear. In this study, we compared colorectal polyp detection in TXI mode with that in WLI mode in a clinical setting.

Figure 1.

A case photographed in each observation mode. (a) WLI: a slight reddish flat type lesion is present, but its border is partially difficult to recognize. (b) TXI mode 1: the addition of color and texture enhancement improves lesion visibility. (c) TXI mode 2: no color enhancement, but better lesion visibility than WLI due to texture enhancement. TXI, texture and color enhancement imaging; WLI, white light imaging.

METHODS

Study design

This retrospective cohort study was conducted at three Japanese institutions. The study was approved by the involved institutional review boards (IRB Approval No. 2020‐313). This study was conducted in accordance with the Japanese Ethical Guidelines for Medical and Health Research involving Human Subjects.

We reviewed the patients' medical records between August 2020 and January 2021. The Solemio ENDO system (Olympus Medical Systems) and the NEXUS system (Fujifilm Medical, Tokyo, Japan) were used to obtain the medical records and pathological diagnosis results of the lesions.

Study population

The inclusion criteria were as follows: (i) patients aged ≥20 years; (ii) patients who underwent colonoscopy using the new endoscopy system; (iii) indications for colonoscopy were CRC screening (including a positive fecal occult blood test result), posttreatment surveillance, a workup for lower gastrointestinal symptoms (including melena, hematochezia, and other symptoms), polyp follow‐up, and pretreatment workup; (iv) all lesions detected during colonoscopy were diagnosed histopathologically, excluding those that did not clinically require resection; (v) repeated ascending colon observation, which was recommended by the Japanese Colonoscopy Screening and Surveillance Guidelines; ¹⁴ and (vi) the second observation of the ascending colon was performed using the TXI or WLI mode, which was not used in the first observation of the entire colorectum. Figure 2 shows the procedure and group allocation. Patients with a history of colorectal resection involving the ascending colon and inadequate bowel preparation (Aronchick scale) ¹⁵ and those judged by the investigators to be inappropriate for analysis were excluded.

Figure 2.

Schematic illustration of the study. Initially, the colonoscope was inserted into the cecum, to observe the cecum and ascending colon with TXI or WLI as the first observation in TXI and WLI group, respectively. Colonoscope was re‐inserted into the cecum, then the cecum and ascending colon were re‐observed as the second observation using TXI or WLI, which was not used for the first observation. The observation mode was re‐switched to TXI or WLI, which was used for the first observation. Colonoscope was then withdrawn and the transverse colon to the rectum was observed. TXI, texture and color enhancement imaging; WLI, white light imaging.

Colonoscopy

The new EVIS X1 colonoscopy system was used in combination with a video system center (CV‐1500; Olympus Medical Systems), new colonoscope (CF‐EZ1500DI; Olympus Medical Systems), or conventional endoscope (CF‐HQ290ECI, CF‐HQ290I, CF‐HQ290ZI, PCF‐H290TI, PCF‐H290ZI, or GIF‐H290T; Olympus Medical Systems).

Patients and outcomes

Patient data and procedural characteristics were recorded. The outcomes of interest were the mean number of adenomas detected per procedure (MAP), adenoma detection rate (ADR), polyp detection rate (PDR), flat polyp detection rate (FDR), ascending colonic adenoma miss rate (Ac‐AMR), ascending colonic polyp miss rate (Ac‐PMR), and the mean number of polyps detected per procedure (MPP). ADR, PDR, and FDR were defined as the proportion of patients with at least one adenoma or cancer, one polyp, or one non‐polypoid polyp detected during the first observation of the entire colorectum (per‐patient analysis). In this study, we defined lesions of the Paris classification type 0‐IIa, 0‐IIb, 0‐IIc, 0‐IIa + 0‐IIc, or 0‐Is + 0‐IIa as non‐polypoid polyps. ¹⁶ , ¹⁷ In the per‐polyp analysis, we calculated MAP and MPP by dividing the total number of adenomas and cancers or the total number of polyps detected during the first observation of the entire colorectum by the total number of colonoscopies. Ac‐AMR and Ac‐PMR were calculated as the ratio of adenomas and cancers or polyps detected in the second observation to the total number of adenomas and cancers or polyps detected during the two observations in the ascending colon, respectively. Intraprocedural and postprocedural complications were compared as the safety outcomes.

Our study's histopathological diagnosis was based on the Japanese classification of colon and rectal cancers. ¹⁸ Therefore, intramucosal carcinoma corresponding to high‐grade dysplasia and mucosal high‐grade neoplasia by the World Health Organization classification was defined as cancer.

Statistical analysis

No procedures were performed to compensate for missing data. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA). Descriptive analyses were performed using means and standard deviations for continuous variables and percentages for categorical variables. Point estimation values and 95% confidence intervals (CIs) for MAP, ADR, PDR, FDR, Ac‐AMR, Ac‐PMR, and MPP were calculated. The distributions used to construct the CIs for individual estimates were as follows: a Poisson distribution was used for MAP and MPP. F‐distribution (Clopper–Pearson exact CI) was used for ADR, PDR, FDR, Ac‐AMR, and Ac‐PMR. The CIs for the estimates were obtained using a two‐sided 95% CI.

Bias might have existed in the patient background factors between the groups because this was an observational study. To ascertain the magnitude of the effect of TXI mode after adjusting for these background factors, we performed univariate and multivariate regression analyses using a model with MAP, ADR, and Ac‐AMR as objective variables. The explanatory variables included in the univariate analysis were as follows: TXI mode (used or not), risk factors for CRC or polyps (age, smoking history [with or without], drinking history [with or without], family history of CRC [with or without]), sex, total withdrawal time (min), bowel preparation (excellent or other), and endoscope used (new or conventional). Explanatory variables included for multivariate analysis were TXI mode (used or not), risk factors for CRC or polyps (age, smoking history [with or without], drinking history [with or without], family history of CRC [with or without]); in addition, sex, total withdrawal time (min), bowel preparation (excellent or other), and endoscope used (new or conventional endoscope) were added if P < 0.20 in univariate regression analysis. Poisson regression was performed for MAP and logistic regression for ADR and Ac‐AMR. The two‐sided statistical significance level was set at 5%.

RESULTS

Patient and procedural characteristics

We included 1043 lesions from 470 patients in the analysis. Patient characteristics are shown in Table 1. The two groups were comparable in terms of their demographic characteristics. Colonoscopies were performed by 35 endoscopists. The number of screening colonoscopies was smaller 58 (24.5%) in the TXI group than in the WLI group 84 (36.1%), and more colonoscopies were performed for pretreatment workups in the TXI group 44 (18.6%) than in the WLI group 21 (9.0%). Colonoscopy procedural characteristics were similar in both groups (Table 2).

Table 1.

Patient characteristics

Characteristics of patients	TXI, N (%)	WLI, N (%)
	237 (100)	233 (100)
Age (mean, SD)	64.2 (12.1)	63.7 (12.8)
Sex
Male	154 (65.0)	147 (63.1)
Female	83 (35.0)	86 (36.9)
Drinking history
Yes	73 (30.8)	70 (30.0)
No	163 (68.8)	162 (69.5)
Unknown	1 (0.4)	1 (0.4)
Smoking history
Yes	103 (43.5)	108 (46.4)
No	134 (56.5)	124 (53.2)
Unknown	0 (0.0)	1 (0.4)
Family history of CRC
Yes	40 (16.9)	30 (12.9)
No	194 (81.9)	191 (82.0)
Unknown	3 (1.3)	12 (5.2)
Indications for colonoscopy
Screening, including FIT positives	58 (24.5)	84 (36.1)
Posttreatment surveillance	126 (53.2)	116 (49.8)
Abdominal symptoms	3 (1.3)	2 (0.9)
Polyp follow‐up	6 (2.5)	10 (4.3)
Pretreatment workup	44 (18.6)	21 (9.0)

CRC, colorectal cancer; FIT, fecal immunochemical test; polyp follow‐up, colonoscopy for patients with residual polyps that did not need to be removed; Posttreatment surveillance, colonoscopy for patients who had one or more neoplastic polyps that were completely removed during a baseline colonoscopy; Pretreatment workup, colonoscopy as a workup before treating colorectal polyps; SD, standard deviation; TXI, texture and color enhancement imaging; WLI, white light imaging.

Table 2.

Procedural characteristics

Characteristics	TXI	WLI
Total number of procedure, N (%)	237 (100)	233 (100)
Aronchick scale, N (%)
Excellent	133 (56.1)	117 (50.2)
Good	93 (39.2)	103 (44.2)
Fair	9 (3.8)	10 (4.3)
Poor	2 (0.8)	1 (0.4)
Other	0 (0.0)	0 (0.0)
Unknown	0 (0.0)	2 (0.9)
Cecal intubation, N (%)
Yes	237 (100)	232 (99.6)
No	0 (0.0)	1 (0.4)
Withdrawal time
Number of procedure, N (%)	235 (100)	230 (100)
Mean, SD (min)	8.6 (3.1)	9.0 (3.5)

SD, standard deviation; TXI, texture and color enhancement imaging; WLI, white light imaging.

Endoscopic findings

The MAP, ADR, PDR, FDR, and MPP were higher in the TXI group than in the WLI group. Ac‐PMR was lower in the TXI group than in the WLI group (Table 3). Details of the lesion characteristics included in the analysis are summarized in Table S1.

Table 3.

Comparison of the detected lesions between TXI and WLI

	TXI		WLI
	Point estimate	95% CI	Point estimate	95% CI
MAP	1.5 (352/237)	1.3–1.6	1.0 (233/233)	0.9–1.1
ADR, %	58.2 (138/237)	51.7–64.6	46.8 (109/233)	40.2–53.4
PDR, %	80.2 (190/237)	74.5–85.0	63.9 (149/233)	57.4–70.1
FDR, %	66.2 (157/237)	59.8–72.2	49.8 (116/233)	43.2–56.4
Ac‐PMR, %	18.6 (38/204)	13.5–24.7	31.7 (53/167)	24.8–39.4
Ac‐AMR, %	17.9 (26/145)	12.1–25.2	28.2 (31/110)	20.0–37.6
MPP	2.3 (555/237)	2.2–2.5	1.6 (384/233)	1.5–1.8

Ac‐AMR, ascending colonic adenoma miss rate; Ac‐PMR, ascending colonic polyp miss rate; ADR, adenoma detection rate; CI, confidence interval; FDR, flat polyp detection rate; MAP, mean number of adenomas detected per procedure; MPP, mean number of polyps detected per procedure; PDR, polyp detection rate; TXI, texture and color enhancement imaging; WLI, white light imaging.

Subgroup analysis revealed that TXI detected more neoplastic lesions than WLI in lesions located in the ascending colon, non‐polypoid morphology, lesion size of 6–9 mm, and histopathologically diagnosed adenomas (0.430 [95% CI 0.354–0.523%] vs. 0.270 [0.211–0.346%], 1.055 [0.932–1.194%] vs. 0.687 [0.588–0.802%], 0.321 [0.256–0.402%] vs. 0.112 [0.076–0.164%], and 1.405 [1.262–1.564%] vs. 0.961 [0.843–1.096%], respectively; Table 4). In the subgroup analysis of missed lesions in the ascending colon, we found that fewer non‐polypoid lesions were missed in the TXI group (16.6%; 95% CI 10.1–24.8%) than in the WLI group (30.6%; 21.0–41.5%; Table 5). Ac‐PMR was lower in TXI than in WLI, even in 0‐IIa, 0‐IIb, and 0‐IIc lesions, according to the Paris classification. There was no significant difference in the Ac‐AMR.

Table 4.

Subgroup analysis of the mean number of adenomas detected per procedure

	TXI		WLI
	Point estimate	95% CI	Point estimate	95% CI
Lesion location
Cecum	0.072 (17/237)	0.045–0.115	0.069 (16/233)	0.042–0.112
Ascending colon	0.430 (102/237)	0.354–0.523	0.270 (63/233)	0.211–0.346
Transverse colon	0.485 (115/237)	0.404–0.583	0.292 (68/233)	0.230–0.370
Descending colon	0.122 (29/237)	0.085–0.176	0.069 (16/233)	0.042–0.112
Sigmoid colon	0.300 (71/237)	0.237–0.378	0.206 (48/233)	0.155–0.273
Rectum	0.076 (18/237)	0.048–0.121	0.094 (22/233)	0.062–0.143
Morphologic type
Polypoid	0.376 (89/237)	0.305–0.462	0.279 (65/233)	0.219–0.356
Non‐polypoid	1.055 (250/237)	0.932–1.194	0.687 (160/233)	0.588–0.802
Others	0.013 (3/237)	0.004–0.039	0/233	–
Size, mm
≤5	0.924 (219/237)	0.809–1.055	0.773 (180/233)	0.668–0.894
6–9	0.321 (76/237)	0.256–0.402	0.112 (26/233)	0.076–0.164
≥10	0.160 (38/237)	0.117–0.220	0.082 (19/233)	0.052–0.128
Histopathology
Adenoma	1.409 (334/237)	1.266–1.569	0.979 (228/233)	0.859–1.114
Cancer	0.076 (18/237)	0.048–0.121	0.021 (5/233)	0.009–0.052

CI, confidence interval; TXI, texture and color enhancement imaging; WLI, white light imaging.

Table 5.

Subgroup analysis of missed lesions in the ascending colon

Subgroup	Ac‐AMR				Ac‐PMR
	TXI		WLI		TXI		WLI
	Point estimate, %	95% CI	Point estimate, %	95% CI	Point estimate, %	95% CI	Point estimate, %	95% CI
Size, mm
≤5	21.5 (23/107)	14.1–30.5	30.9 (30/97)	21.9–41.1	21.5 (32/149)	15.2–28.9	33.8 (45/133)	25.9–42.5
6–9	9.1 (2/22)	1.1–29.2	14.3 (1/7)	0.4–57.9	12.9 (4/31)	3.6–29.8	27.8 (5/18)	9.7–53.5
≥10	7.1 (1/14)	0.2–33.9	0 (0/6)	0.0–45.9	9.5 (2/21)	1.2–30.4	20.0 (3/15)	4.3–48.1
Morphologic type
Polypoid	23.5 (8/34)	10.7–41.2	20.0 (5/25)	6.8–40.7	26.3 (10/38)	13.4–43.1	24.1 (7/29)	10.3–43.5
Non‐polypoid	16.5 (18/109)	10.1–24.8	30.6 (26/85)	21.0–41.5	17.2 (28/163)	11.7–23.9	33.3 (46/138)	25.5–41.9
Histopathology
Adenoma	19.0 (26/137)	12.8–26.6	27.8 (30/108)	19.6–37.2	19.0 (26/137)	12.8–26.6	27.8 (30/108)	19.6–37.2
SSL	–	–	–	–	7.7 (1/13)	0.2–36.0	45.5 (5/11)	16.7–76.6
Cancer	0 (0/8)	0.0–36.9	50.0 (1/2)	1.3–98.7	0 (0/8)	0.0–36.9	50.0 (1/2)	1.3–98.7

Ac‐AMR, ascending colonic adenoma miss rate; Ac‐PMR, ascending colonic polyp miss rate; CI, confidence interval; SSL, sessile serrated lesion; TXI, texture and color enhancement imaging; WLI, white light imaging.

Factors associated with the MAP, ADR, and Ac‐AMR

Significant differences were observed in the multivariate regression model with MAP as the objective variable (odds ratio in lesion units) for the use of TXI mode (1.4; 95% CI 1.2–1.6%; P < 0.001), older age (1.0; 1.0–1.0%, P < 0.001), smoking history (1.3; 1.1–1.5%, P = 0.010), longer withdrawal time (1.1; 1.0–1.1%, P < 0.001), and the use of new endoscope (1.5; 1.2–1.9%, P < 0.001). Significant differences were observed in the multivariate regression model with ADR as the objective variable (odds ratio in case units) for the use of the TXI mode (1.5; 1.0–2.3%, P = 0.044), older age (1.0; 1.0–1.1%, P < 0.001), longer withdrawal time (1.1; 1.0–1.1%, P = 0.040), and the use of new endoscope (1.7; 1.1–2.7%, P = 0.025).

A significant difference was observed in the univariate regression model with Ac‐AMR (odds ratio of lesion units) as the objective variable for sex and withdrawal time. None of the variables showed significant differences in the multivariate regression model (Table 6).

Table 6.

Adjustment for confounders of TXI mode add‐on effect

Objective variable	Explanatory variable	Univariate regression			Multivariate regression
		Odds ratio	95% CI	P value	Odds ratio	95% CI	P value
MAP	TXI mode	1.5	1.3–1.8	<0.001	1.4	1.2–1.6	<0.001
	Older age	1.0	1.0–1.0	<0.001	1.0	1.0–1.0	<0.001
	Smoking history	1.3	1.1–1.6	<0.001	1.3	1.1–1.5	0.01
	Drinking history	1.1	0.9–1.3	0.556	1.0	0.8–1.2	0.628
	Family history of CRC	1.0	0.8–1.2	0.715	1.0	0.8–1.3	0.941
	Male sex	1.5	1.3–1.8	<0.001	1.2	1.0–1.5	0.094
	Withdrawal time (min)	1.0	1.0–1.1	<0.001	1.1	1.0–1.1	<0.001
	Excellent bowel preparation	1.2	1.0–1.4	0.026	1.1	0.9–1.3	0.355
	New endoscope	1.7	1.4–2.1	<0.001	1.5	1.2–1.9	<0.001
ADR	TXI mode	1.6	1.1–2.3	0.013	1.5	1.0–2.3	0.044
	Older age	1.0	1.0–1.1	<0.001	1.0	1.0–1.1	<0.001
	Smoking history	1.5	1.0–2.2	0.028	1.5	0.9–2.2	0.09
	Drinking history	1.4	1.0–2.1	0.076	1.3	0.8–2.1	0.217
	Family history of CRC	0.7	0.4–1.1	0.106	0.7	0.4–1.2	0.19
	Male sex	1.4	1.0–2.0	0.09	0.9	0.6–1.4	0.64
	Withdrawal time (min)	1.0	1.0–1.1	0.097	1.1	1.0–1.1	0.04
	Excellent bowel preparation	1.5	1.0–2.1	0.032	1.4	0.9–2.1	0.155
	New endoscope	2.0	1.4–2.9	<0.001	1.7	1.1–2.7	0.025
Ac‐AMR	TXI mode	0.6	0.3–1.0	0.053	0.7	0.3–1.3	0.231
	Older age	1.0	1.0–1.0	0.988	1.0	1.0–1.0	0.853
	Smoking history	0.7	0.4–1.2	0.212	0.7	0.4–1.5	0.42
	Drinking history	1.0	0.5–1.9	0.916	0.9	0.4–2.0	0.878
	Family history of CRC	0.7	0.3–1.7	0.427	0.7	0.3–1.7	0.435
	Male sex	0.5	0.3–1.0	0.049	0.7	0.3–1.3	0.236
	Withdrawal time (min)	1.1	1.0–1.2	0.025	1.1	1.0–1.2	0.089
	Excellent bowel preparation	0.7	0.4–1.2	0.207	–	–	–
	New endoscope	0.6	0.3–1.2	0.149	0.6	0.3–1.4	0.244

Ac‐AMR, ascending colonic adenoma miss rate; ADR, adenoma detection rate; CI, confidence interval; CRC, colorectal cancer; MAP, mean number of adenomas detected per procedure; TXI, texture and color enhancement imaging.

Complications

No intraprocedural bleeding was observed. Bleeding occurred after endoscopic submucosal dissection in one patient (0.6%) in the WLI group. During or after the procedure, there were no perforations, cardiopulmonary complications, anaphylaxis, or other complications (Table S2).

DISCUSSION

To the best of our knowledge, this is the first clinical study to directly compare the detection of colorectal polyps between the newly available TXI and WLI colonoscopies. Our study showed that the TXI mode improved the detection of colorectal polyps and neoplastic lesions, especially non‐polypoid lesions. We also found that TXI tended to miss fewer SSLs in the ascending colon than WLI did.

Studies that used upper gastrointestinal endoscopy images reported that TXI improved lesion visibility in squamous cell carcinomas of the pharynx and esophagus, neoplastic lesions of the stomach, and atrophy of the gastric mucosa. ¹⁹ , ²⁰ However, no studies have compared the detection of lesions using TXI and WLI. Tamai et al. ¹³ reported that TXI improved the visualization of colorectal lesions compared with WLI and NBI using images of hyperplastic polyps, SSLs, adenomas, cancers, and submucosal tumors. Nishizawa et al. ¹² reported that TXI provided higher visibility for serrated colorectal polyps, including SSLs, than WLI. In our study, TXI detected more lesions, including non‐polypoid lesions, than WLI, which is consistent with the findings of previous reports.

A large fraction of precancerous colorectal lesions are endoscopically subtle. Therefore, superior lesion recognition skills are required for detection. ²¹ Several IEE technologies have been developed and studied to improve polyp detection during colonoscopy. NBI uses two specialized lights to display superficial mucosal capillaries and their patterns. Because hemoglobin appears brown in NBI images and the color difference between blood vessels and the surrounding mucosa is more prominent than in WLI, it is expected that NBI will facilitate the detection of colorectal lesions. ²² , ²³ Five meta‐analyses and systematic reviews on the detection of colorectal lesions using WLI and NBI have been reported. In three of these studies, NBI did not improve the ADR compared with WLI. ⁶ , ²⁴ , ²⁵ It was shown that the detection of flat adenomas was improved in only one meta‐analysis. ⁷ In the most recent meta‐analysis that included data from randomized controlled trials using second‐generation NBI, NBI had a higher ADR than WLI (50.2% vs. 44.4%, P = 0.02) when bowel preparation was optimized. The stool appears brick‐red with NBI; even a thin film of stool and mucus can significantly impair mucosal visualization. ADR improved in the NBI of the second generation. However, the improvement in ADR was not seen in the NBI of the first generation. ⁸ LCI images are lighter and closer to WLI images than to NBI images. In early trials, LCI was reported to be superior to WLI for polyp detection. ⁹ , ²⁶ Reports on ADR improvements were inconsistent, showing both significant and non‐significant improvements in LCI compared with WLI. ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ TXI has a bright image similar to that of LCI, which is more suitable for lesion detection than NBI. In our study population, TXI had a higher ADR than WLI did. Factors that improve lesion detection with TXI include enhanced red contrast, which facilitates the identification of slight color differences between lesions and the normal mucosa. TXI selectively enhances brightness in the dark areas of an endoscopic image, making it easy to identify lesions that are distant from the endoscope. ¹¹ , ¹³ , ²⁰ , ³¹ Moreover, TXI with WLI‐like images is preferable, particularly for inexperienced endoscopists. ¹⁹ However, it improves the detection of colorectal polyps only in non‐polypoid lesions. One possible explanation is that nonpolypoid lesions are more likely to be missed by WLI. Thus, there is a significant opportunity to improve detection.

Recently, sessile serrated polyps have been considered important in terms of their histology and molecular abnormalities. The frequency of CRCs involving the serrated pathway is estimated to be 5–10% of all CRCs. ³² , ³³ Serrated polyps, especially SSLs, are more difficult to detect than conventional adenomas. The SSL has a flat or sessile morphology and is often located in the proximal colon. In addition, the mucus caps and indistinct borders of these lesions make it difficult to detect them endoscopically, so they are often not completely removed or easily overlooked. ³⁴ SSLs are mainly responsible for the inability of colonoscopy to prevent proximal colonic and interval cancers, which occur during the interval between complete colonoscopy and surveillance. ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰

However, the benefits of IEE technologies for SSL detection remain unclear. In 2016, Parikh et al. ⁴¹ reported that in their meta‐analysis to evaluate the utility of IEE in the diagnosis of SSL, they did not find a study that assessed the detection of SSL. More recently, studies comparing the detection of SSL for NBI and LCI with WLI have been reported. However, improved detection of SSL has not been fully verified. ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ Our results demonstrated that TXI tends to miss fewer SSLs than WLI. However, the 95% CI range was very wide and overlapped between the two groups owing to the limited sample size.

This study has several limitations. First, there may have been a selection bias because this observational study retrospectively selected patients. Colonoscopy was associated with a higher rate of pretreatment workup in the TXI group (18.6%) than in the WLI group (9.0%), and patients in the TXI group may have had more adenomas. Second, because the mode used was not blinded, there may have been a diagnostic bias that resulted in careful observation with TXI, which is expected to detect more polyps. Third, the choice of endoscope or observation mode was left to the endoscopist's discretion, which may have affected patient selection. Therefore, we performed a multivariate analysis to minimize confounding factors. Fourth, the study was conducted only in tertiary centers in Japan and may not reflect the environment in community hospitals. Fifth, the number of patients included in the analysis was insufficient, and the sample size may have been insufficient for some outcome analyses. For example, Ac‐PMR and Ac‐AMR showed similar trends; however, only a higher number of polyps in Ac‐PMR showed statistical differences. Sixth, hyperplastic polyps that are <5 mm in diameter in the rectosigmoid colon are not resected in routine practice. If endoscopists misdiagnosed neoplastic lesions as hyperplastic polyps, this might have affected the study results. However, it has been reported that the accuracy of endoscopic diagnosis is extremely high, ⁴⁶ and it is presumed that the influence of misdiagnosis is small. Finally, the indication for colonoscopy for the patients included in the analyses was associated with lower screening rates and more patients at a higher risk of detecting colorectal pathology. Thus, the results may not be generalizable to the general colonoscopy population.

Our study indicates that TXI improves the detection of colorectal neoplastic lesions, especially nonpolypoid lesions. However, prospective randomized trials are required to confirm these findings.

CONFLICT OF INTEREST

Author Y.S. has received lecture fees from Olympus Corporation. This study was supported by Olympus Corporation as a joint research organization for the three institutions. The other authors declare no conflict of interest for this article.

FUNDING INFORMATION

None.

Supporting information

Table S1 Characteristics of the lesion included in the analysis.

Table S2 Safety evaluation.