ENDOANGEL improves detection of missed colorectal adenomas in second colonoscopy: A retrospective study

Abstract

The ENDOANGEL (EN) computer-assisted detection technique has emerged as a promising tool for enhancing the detection rate of colorectal adenomas during colonoscopies. However, its efficacy in identifying missed adenomas during subsequent colonoscopies remains unclear. Thus, we herein aimed to compare the adenoma miss rate (AMR) between EN-assisted and standard colonoscopies. Data from patients who underwent a second colonoscopy (EN-assisted or standard) within 6 months between September 2022 and May 2023 were analyzed. The EN-assisted group exhibited a significantly higher AMR (24.3% vs 11.9%, P = .005) than the standard group. After adjusting for potential confounders, multivariable analysis revealed that the EN-assisted group had a better ability to detect missed adenomas than the standard group (odds ratio = 2.89; 95% confidence interval = 1.14–7.80, P = .029). These findings suggest that EN-assisted colonoscopy represents a valuable advancement in improving AMR compared with standard colonoscopy. The integration of EN-assisted colonoscopy into routine clinical practice may offer significant benefits to patients requiring hospital resection of lesions following adenoma detection during their first colonoscopy.

1. Introduction

Colorectal cancer (CRC) poses a significant global health burden, ranking as the third most prevalent malignant tumor worldwide and the second leading cause of cancer-related deaths.^[1] The classical adenoma-carcinoma sequence is the predominant pathway in CRC development, typically unfolding over a 10 to 15 years.^[2] This sequence involves the progressive evolution of colorectal adenomas from low-grade dysplasia to high-grade dysplasia and eventually invasive carcinoma.^[3] Detecting colorectal adenomas through effective colonoscopy is critical for reducing CRC incidence and mortality rates.^[4] Despite its high sensitivity, colonoscopy does not completely prevent future CRC,^[5–9] largely due to the failure to detect precancerous lesions,^[8,10] leading to the concept of adenoma miss rate (AMR) as a measure of post-colonoscopy CRC risk.^[11] Previous studies comparing consecutive colonoscopies have reported miss rates for colorectal lesions ranging from 20% to 40%.^[12–14]

The application of computer-aided detection systems based on artificial intelligence (AI) in colonoscopy has shown promise in improving adenoma detection rates and reducing AMR,^[15,16] potentially decreasing the risk of interval CRC. One such system, ENDOANGEL (EN), is a novel AI-based detection system developed for the diagnosis and detection of gastrointestinal diseases.^[17–19] EN significantly enhances the detection of colorectal adenomas by providing real-time monitoring synchronized with colonoscopy.^[20] Additionally, short-term secondary colonoscopies (within 1 year) have been shown to detect an additional approximately 15% of missed colorectal adenomas.^[21,22] However, it remains unclear whether EN-assisted colonoscopy can further reduce AMR compared with standard colonoscopy during a second colonoscopy. Therefore, this study aims to investigate whether the use of EN-assisted colonoscopy can improve AMR compared with standard colonoscopy.

2. Patients and methods

2.1. Patients

Patients who underwent a second colonoscopy for polypectomy or endoscopic mucosal resection within 6 months after their first colonoscopy at Yijishan Hospital of Wannan Medical College between September 2022 and May 2023 were retrospectively analyzed. The first colonoscopy was performed as either a surveillance or screening colonoscopy using standard techniques. The second colonoscopy utilized either EN-assisted colonoscopy (EN group) or standard colonoscopy (ST group). If a diminutive lesion was identified during the first colonoscopy, it was removed with the patients’ informed consent. However, if patients had multiple lesions requiring polypectomy or endoscopic mucosal resection, a second colonoscopy was performed after admission. The exclusion criteria comprised age under 18 years, presence of polyposis syndrome, inflammatory bowel disease, CRC, history of colorectal resection, known or suspected intestinal obstruction or perforation, and a total score on the Boston Bowel Preparation Scale (BBPS) of <6 points or any segment with a score of <2 points. The BBPS score was used to evaluate the quality of bowel preparation during withdrawal.^[23] Two experienced senior endoscopists, each having performed over 2500 EN-assisted colonoscopies, conducted the colonoscopies using an Olympus Optical colonoscope (GIF-H290, Tokyo, Japan).^[20] The bowel was prepared using polyethylene glycol-electrolyte split-dose, following the provided instructions. All patients received sedation with propofol or midazolam and fentanyl.^[11] Adenomas were classified as either tubular or (tubulo) villous lesions based on histological examination, with sessile serrated lesions included in the adenoma group. Basic demographic information, abdominal or pelvic surgery history, insertion time, withdrawal time, and BBPS score were recorded. Ethics approval and informed consent were waived by the Ethics Institution of Yijishan Hospital of Wannan Medical College in compliance with national legislation and institutional requirements.

2.2. Confounding variables and study outcome

This study considered several confounding variables, including sex (female/male), age, BBPS score, history of abdominal or pelvic surgery (no/yes), and withdrawal time. The study’s primary outcome was AMR, defined as the number of adenomas detected during the second colonoscopy divided by the total number of adenomas detected during both the first and second colonoscopies.^[11]

2.3. Statistical analysis

Continuous variables were reported as means ± standard deviation or as a median with interquartile range (25th–75th percentiles). They were compared using Student t test or the Mann–Whitney U test. Categorical variables were presented as frequencies and percentages and analyzed using the chi-square test or Fisher exact test. Multivariable logistic regression was used to assess the performance of EN-assisted colonoscopy compared with standard colonoscopy in detecting patients with missed adenomas. Statistical analysis was conducted using SPSS 26.0 (SPSS Inc, Chicago, IL) and R version 4.0.2 (The R Foundation for Statistical Computing, Vienna, Austria), with a significance level set at P < .05.

3. Results

3.1. Baseline characteristics

A total of 118 eligible patients were retrospectively included. Table 1 and Table S1, Supplemental Digital Content, http://links.lww.com/MD/N210 present the distribution of patient characteristics and baseline colonoscopy characteristics, classified based on the methods used during the second colonoscopy, respectively. Except for the BBPS score and withdrawal time, no significant differences were noted in all variables between the 2 groups.

Table 1.

Characteristics of patients received a second standard or ENDOANGEL-assisted (EN) colonoscopy.

Characteristics	Standard (n = 62)	EN (n = 54)	P value
Sex, n (%)			.150
Female	37 (59.7)	25 (46.3)
Male	25 (40.3)	29 (53.7)
Age, mean (SD)	55.8 (±12.9)	57.0 (±12.3)	.597
Withdrawal time (min), median (IQR)	6.7 (6.4–7.2)	6.3 (6.2–6.7)	<.001
Insertion time (min), median (IQR)	4.1 (3.3–4.9)	4.0 (3.3–5.0)	.744
BBPS score, median (IQR)	9 (8–9)	8 (7–9)	.014
History of abdominal surgery, n (%)			.447
No	35 (56.5)	35 (64.8)
Yes	27 (43.5)	19 (35.2)
Adenoma miss rate, n (%)	18/152 (11.9)	35/144 (24.3)	.005
Patients with missed adenoma rate, n (%)	15/62 (24.2)	23/54 (42.6)	.035

BBPS = Boston Bowel Preparation Scale.

3.2. Characteristics of missed adenomas

Out of the 118 patients, 38 were identified to have missed adenomas during the second colonoscopy, resulting in a patients with missed adenoma rate of 32.2%. This corresponded to a total of 53 colorectal adenomas that were missed, resulting in an AMR of 17.9%. Upon analysis, the EN group had a significantly higher AMR (P = .005) and missed adenoma rate than the ST group (P = .035). The majority of the missed adenomas were located in the right colon, had a size of ≤5 mm, and were predominantly tubular adenomas (Table 2).

Table 2.

Characteristics of missed adenomas.

Characteristics	Stndard	ENDOANGEL
Total missed adenomas	18	35
Location
Right	13	25
Left	5	10
Size
≤5 mm	13	28
6–9 mm	5	7
≥10 mm	1	0
Pathology
Tubular	13	31
Tubulovillous	4	2
Sessile serrated lesions	1	2

3.3. ENDOANGEL-assisted colonoscopy and detection of missed adenomas

The univariate analysis revealed a significant association between patients who received EN-assisted colonoscopy and a higher likelihood of detecting missed adenomas than those who underwent standard colonoscopy (odds ratio = 2.32, 95% confidence interval = 1.06–5.22, P = .037) (Table 3). The multivariable analysis showed that the EN group also exhibited better performance in terms of detecting missed adenomas than the ST group (odds ratio = 2.89, 95% confidence interval = 1.14–7.80, P = .029).

Table 3.

Multivariate analysis of the influence of ENDOANGEL-assisted colonoscopy versus standard colonoscopy on the detection of patient with missed adenoma.

Characteristics	Univariate analysis		Multivariate analysis
	OR (95% CI)	P value	OR (95% CI)	P value
Sex		.089		.034
Female	Ref		Ref
Male	1.98 (0.91–4.40)		3.24 (1.14–10.21)
Age (year)	1.02 (0.99–1.06)	.154	1.02 (0.98–1.06)	.280
History of abdominal surgery		.978		.240
No	Ref		Ref
Yes	0.99 (0.44–2.18)		1.89 (0.67–5.68)
BBPS score	0.63 (0.38–1.02)	.060	0.59 (0.34–1.00)	.054
Withdrawal time (min)	1.42 (0.65–3.07)	.367	3.08 (1.20–8.35)	.022
Methods		.037		.029
Standard	Ref		Ref
EN	2.32 (1.06–5.22)		2.89 (1.14–7.80)

CI = confidence interval, EN = ENDOANGEL, OR = odds ratio.

4. Discussion

This study represents a pioneering analysis of the impact of EN-assisted colonoscopy on AMR. EN-assisted procedures yielded higher patients with missed adenoma rate and AMR than standard colonoscopy. Furthermore, multivariate analysis underscored the significant enhancement in detecting patients with missed adenomas through EN-assisted colonoscopy during subsequent procedures.

Tandem colonoscopy serves as a benchmark for evaluating AMR.^[24] Findings from a randomized controlled trial revealed that AI-driven computer-aided detection systems markedly mitigate AMR and the miss rate of sessile serrated lesions, compared with solitary high-definition white light colonoscopy.^[25] Additional tandem colonoscopy investigations corroborated these findings by demonstrating a substantial decrease in AMR.^[26] Meta-analyses of tandem colonoscopy studies reported an average AMR of 26%.^[27] Nonetheless, owing to low patient compliance and the considerable economic burden associated with tandem colonoscopy, these procedures are predominantly confined to research settings and seldom performed in clinical practice.

Adenomas grow <1 mm annually.^[28] Consequently, conducting a follow-up colonoscopy within a short time frame is justified to identify missed colorectal adenomas. Dong et al observed an AMR of 15.8% among patients undergoing a second colonoscopy within 1 year of their first procedure.^[21] Similarly, Lee et al reported an AMR of 13.4% in patients undergoing a second colonoscopy within 6 months of their first.^[22] Our findings align with these studies. Notably, in the EN group, the majority of missed adenomas (80%) measured ≤5 mm, indicating minimal impact on patient prognosis. Consequently, the clinical effectiveness of EN may be less pronounced than anticipated. These preliminary findings necessitate validation through prospective research endeavors.

The capability of EN to detect lesions in real-time, monitor withdrawal speed and time, mitigate human error, and enhance the likelihood of identifying diminutive adenomas that might be missed with visual inspection alone has been underscored.^[29] Moreover, EN helps standardize the detection process, minimizing variances in the cognitive and technical skills of endoscopists, thus improving reproducibility, fidelity, and uniformity in lesion detection.^[20,30,31] EN significantly enhances the detection of colorectal adenomas by facilitating real-time monitoring synchronized with colonoscopy.^[20] Our findings suggest that EN-assisted secondary colonoscopy effectively improves AMR and identify more patients with missed adenomas than standard colonoscopy in patients necessitating hospital resection of lesions after adenoma detection during their first colonoscopy, potentially mitigating the risk of interval CRC.

Nevertheless, our study has several limitations. First, being a retrospective single-center study, it inherently carries the risk of selection bias, particularly since it only included patients undergoing a second colonoscopy within 6 months. Second, certain patient characteristics, such as alcohol and smoking habits, were not available, which could have influenced the outcomes. Third, our results were based on the second colonoscopy within 6 months after the first procedure, possibly overestimating the miss rate due to newly developed lesions. Lastly, our findings necessitate validation through future large-scale multicenter clinical trials.

In conclusion, EN-assisted colonoscopy presents promising prospects for improving AMR and enhancing the detection of patients with missed adenomas compared with standard colonoscopy. We advocate for the integration of EN-assisted colonoscopy into contemporary clinical practice, especially for patients requiring hospital resection of lesions after adenoma detection during their first colonoscopy.

Author contributions

Conceptualization: Chiyi He.

Data curation: Yundong Wang.

Formal analysis: Yundong Wang.

Funding acquisition: Yundong Wang.

Methodology: Yundong Wang, Chiyi He.

Validation: Chiyi He.

Validation: Chiyi He.

Writing – original draft: Yundong Wang.

Writing – review & editing: Chiyi He.