A novel artificial intelligence-assisted "vascular healing" diagnosis for prediction of future clinical relapse in patients with ulcerative colitis: a prospective cohort study (with video).

BACKGROUND AND AIMS: Image-enhanced endoscopy has attracted attention as a method for detecting inflammation and predicting outcomes in patients with ulcerative colitis (UC); however, the procedure requires specialist endoscopists. Artificial intelligence (AI)-assisted image-enhanced endoscopy may help nonexperts provide objective accurate predictions with the use of optical imaging. We aimed to develop a novel AI-based system using 8853 images from 167 patients with UC to diagnose "vascular-healing" and establish the role of AI-based vascular-healing for predicting the outcomes of patients with UC. METHODS: This open-label prospective cohort study analyzed data for 104 patients with UC in clinical remission. Endoscopists performed colonoscopy using the AI system, which identified the target mucosa as AI-based vascular-active or vascular-healing. Mayo endoscopic subscore (MES), AI outputs, and histologic assessment were recorded for 6 colorectal segments from each patient. Patients were followed up for 12 months. Clinical relapse was defined as a partial Mayo score >2 RESULTS: The clinical relapse rate was significantly higher in the AI-based vascular-active group (23.9% [16/67]) compared with the AI-based vascular-healing group (3.0% [1/33)]; P = .01). In a subanalysis predicting clinical relapse in patients with MES ≤1, the area under the receiver operating characteristic curve for the combination of complete endoscopic remission and vascular healing (0.70) was increased compared with that for complete endoscopic remission alone (0.65). CONCLUSIONS: AI-based vascular-healing diagnosis system may potentially be used to provide more confidence to physicians to accurately identify patients in remission of UC who would likely relapse rather than remain stable.

Impact of computer-aided characterization for diagnosis of colorectal lesions, including sessile serrated lesions: Multireader, multicase study.

OBJECTIVES: Computer-aided characterization (CADx) may be used to implement optical biopsy strategies into colonoscopy practice; however, its impact on endoscopic diagnosis remains unknown. We aimed to evaluate the additional diagnostic value of CADx when used by endoscopists for assessing colorectal polyps. METHODS: This was a single-center, multicase, multireader, image-reading study using randomly extracted images of pathologically confirmed polyps resected between July 2021 and January 2022. Approved CADx that could predict two-tier classification (neoplastic or nonneoplastic) by analyzing narrow-band images of the polyps was used to obtain a CADx diagnosis. Participating endoscopists determined if the polyps were neoplastic or not and noted their confidence level using a computer-based, image-reading test. The test was conducted twice with a 4-week interval: the first test was conducted without CADx prediction and the second test with CADx prediction. Diagnostic performances for neoplasms were calculated using the pathological diagnosis as reference and performances with and without CADx prediction were compared. RESULTS: Five hundred polyps were randomly extracted from 385 patients and diagnosed by 14 endoscopists (including seven experts). The sensitivity for neoplasia was significantly improved by referring to CADx (89.4% vs. 95.6%). CADx also had incremental effects on the negative predictive value (69.3% vs. 84.3%), overall accuracy (87.2% vs. 91.8%), and high-confidence diagnosis rate (77.4% vs. 85.8%). However, there was no significant difference in specificity (80.1% vs. 78.9%). CONCLUSIONS: Computer-aided characterization has added diagnostic value for differentiating colorectal neoplasms and may improve the high-confidence diagnosis rate.

Whole slide image-based prediction of lymph node metastasis in T1 colorectal cancer using unsupervised artificial intelligence.

OBJECTIVES: Lymph node metastasis (LNM) prediction for T1 colorectal cancer (CRC) is critical for determining the need for surgery after endoscopic resection because LNM occurs in 10%. We aimed to develop a novel artificial intelligence (AI) system using whole slide images (WSIs) to predict LNM. METHODS: We conducted a retrospective single center study. To train and test the AI model, we included LNM status-confirmed T1 and T2 CRC between April 2001 and October 2021. These lesions were divided into two cohorts: training (T1 and T2) and testing (T1). WSIs were cropped into small patches and clustered by unsupervised K-means. The percentage of patches belonging to each cluster was calculated from each WSI. Each cluster's percentage, sex, and tumor location were extracted and learned using the random forest algorithm. We calculated the areas under the receiver operating characteristic curves (AUCs) to identify the LNM and the rate of over-surgery of the AI model and the guidelines. RESULTS: The training cohort contained 217 T1 and 268 T2 CRCs, while 100 T1 cases (LNM-positivity 15%) were the test cohort. The AUC of the AI system for the test cohort was 0.74 (95% confidence interval [CI] 0.58-0.86), and 0.52 (95% CI 0.50-0.55) using the guidelines criteria (P = 0.0028). This AI model could reduce the 21% of over-surgery compared to the guidelines. CONCLUSION: We developed a pathologist-independent predictive model for LNM in T1 CRC using WSI for determination of the need for surgery after endoscopic resection. TRIAL REGISTRATION: UMIN Clinical Trials Registry (UMIN000046992, https://center6.umin.ac.jp/cgi-open-bin/ctr/ctr_view.cgi?recptno=R000053590).

Impact of the clinical use of artificial intelligence-assisted neoplasia detection for colonoscopy: a large-scale prospective, propensity score-matched study (with video).

BACKGROUND AND AIMS: Recently, the use of computer-aided detection (CADe) for colonoscopy has been investigated to improve the adenoma detection rate (ADR). We aimed to assess the efficacy of a regulatory-approved CADe in a large-scale study with high numbers of patients and endoscopists. METHODS: This was a propensity score-matched prospective study that took place at a university hospital between July 2020 and December 2020. We recruited patients aged ≥20 years who were scheduled for colonoscopy. Patients with polyposis, inflammatory bowel disease, or incomplete colonoscopy were excluded. We used a regulatory-approved CADe system and conducted a propensity score matching-based comparison of the ADR between patients examined with and without CADe as the primary outcome. RESULTS: During the study period, 2261 patients underwent colonoscopy with the CADe system or routine colonoscopy, and 172 patients were excluded in accordance with the exclusion criteria. Thirty endoscopists (9 nonexperts and 21 experts) were involved in this study. Propensity score matching was conducted using 5 factors, resulting in 1836 patients included in the analysis (918 patients in each group). The ADR was significantly higher in the CADe group than in the control group (26.4% vs 19.9%, respectively; relative risk, 1.32; 95% confidence interval, 1.12-1.57); however, there was no significant increase in the advanced neoplasia detection rate (3.7% vs 2.9%, respectively). CONCLUSIONS: The use of the CADe system for colonoscopy significantly increased the ADR in a large-scale prospective study including 30 endoscopists (Clinical trial registration number: UMIN000040677.).

Combined endocytoscopy with pit pattern diagnosis in ulcerative colitis-associated neoplasia: Pilot study.

OBJECTIVES: Ulcerative colitis-associated neoplasias (UCAN) are often flat with an indistinct boundary from surrounding tissues, which makes differentiating UCAN from non-neoplasias difficult. Pit pattern (PIT) has been reported as one of the most effective indicators to identify UCAN. However, regenerated mucosa is also often diagnosed as a neoplastic PIT. Endocytoscopy (EC) allows visualization of cell nuclei. The aim of this retrospective study was to demonstrate the diagnostic ability of combined EC irregularly-formed nuclei with PIT (EC-IN-PIT) diagnosis to identify UCAN. METHODS: This study involved patients with ulcerative colitis whose lesions were observed by EC. Each lesion was diagnosed by two independent expert endoscopists, using two types of diagnostic strategies: PIT alone and EC-IN-PIT. We evaluated and compared the diagnostic abilities of PIT alone and EC-IN-PIT. We also examined the difference in the diagnostic abilities of an EC-IN-PIT diagnosis according to endoscopic inflammation severity. RESULTS: We analyzed 103 lesions from 62 patients; 23 lesions were UCAN and 80 were non-neoplastic. EC-IN-PIT diagnosis had a significantly higher specificity and accuracy compared with PIT alone: 84% versus 58% (P < 0.001), and 88% versus 67% (P < 0.01), respectively. The specificity and accuracy were significantly higher for Mayo endoscopic score (MES) 0-1 than MES 2-3: 93% versus 68% (P < 0.001) and 95% versus 74% (P < 0.001), respectively. CONCLUSIONS: Our novel EC-IN-PIT strategy had a better diagnostic ability than PIT alone to predict UCAN from suspected and initially detected lesions using conventional colonoscopy. UMIN clinical trial (UMIN000040698).

Beyond complete endoscopic healing: Goblet appearance using an endocytoscope to predict future sustained clinical remission in ulcerative colitis.

OBJECTIVES: Complete endoscopic healing, defined as Mayo endoscopic score (MES) = 0, is an optimal target in the treatment of ulcerative colitis (UC). However, some patients with MES = 0 show clinical relapse within 12 months. Histologic goblet mucin depletion has emerged as a predictor of clinical relapse in patients with MES = 0. We observed goblet depletion in vivo using an endocytoscope, and analyzed the association between goblet appearance and future prognosis in UC patients. METHODS: In this retrospective cohort study, all enrolled UC patients had MES = 0 and confirmed clinical remission between October 2016 and March 2020. We classified the patients into two groups according to the goblet appearance status: preserved-goblet and depleted-goblet groups. We followed the patients until March 2021 and evaluated the difference in cumulative clinical relapse rates between the two groups. RESULTS: We identified 125 patients with MES = 0 as the study subjects. Five patients were subsequently excluded. Thus, we analyzed the data for 120 patients, of whom 39 were classified as the preserved-goblet group and 81 as the depleted-goblet group. The patients were followed-up for a median of 549 days. During follow-up, the depleted-goblet group had a significantly higher cumulative clinical relapse rate than the preserved-goblet group (19% [15/81] vs. 5% [2/39], respectively; P = 0.02). CONCLUSIONS: Observing goblet appearance in vivo allowed us to better predict the future prognosis of UC patients with MES = 0. This approach may assist clinicians with onsite decision-making regarding treatment interventions without a biopsy.

Current status and future perspective on artificial intelligence for lower endoscopy.

The global incidence and mortality rate of colorectal cancer remains high. Colonoscopy is regarded as the gold standard examination for detecting and eradicating neoplastic lesions. However, there are some uncertainties in colonoscopy practice that are related to limitations in human performance. First, approximately one-fourth of colorectal neoplasms are missed on a single colonoscopy. Second, it is still difficult for non-experts to perform adequately regarding optical biopsy. Third, recording of some quality indicators (e.g. cecal intubation, bowel preparation, and withdrawal speed) which are related to adenoma detection rate, is sometimes incomplete. With recent improvements in machine learning techniques and advances in computer performance, artificial intelligence-assisted computer-aided diagnosis is being increasingly utilized by endoscopists. In particular, the emergence of deep-learning, data-driven machine learning techniques have made the development of computer-aided systems easier than that of conventional machine learning techniques, the former currently being considered the standard artificial intelligence engine of computer-aided diagnosis by colonoscopy. To date, computer-aided detection systems seem to have improved the rate of detection of neoplasms. Additionally, computer-aided characterization systems may have the potential to improve diagnostic accuracy in real-time clinical practice. Furthermore, some artificial intelligence-assisted systems that aim to improve the quality of colonoscopy have been reported. The implementation of computer-aided system clinical practice may provide additional benefits such as helping in educational poorly performing endoscopists and supporting real-time clinical decision-making. In this review, we have focused on computer-aided diagnosis during colonoscopy reported by gastroenterologists and discussed its status, limitations, and future prospects.

Artificial Intelligence-assisted System Improves Endoscopic Identification of Colorectal Neoplasms.

BACKGROUND & AIMS: Precise optical diagnosis of colorectal polyps could improve the cost-effectiveness of colonoscopy and reduce polypectomy-related complications. However, it is difficult for community-based non-experts to obtain sufficient diagnostic performance. Artificial intelligence-based systems have been developed to analyze endoscopic images; they identify neoplasms with high accuracy and low interobserver variation. We performed a multi-center study to determine the diagnostic accuracy of EndoBRAIN, an artificial intelligence-based system that analyzes cell nuclei, crypt structure, and microvessels in endoscopic images, in identification of colon neoplasms. METHODS: The EndoBRAIN system was initially trained using 69,142 endocytoscopic images, taken at 520-fold magnification, from patients with colorectal polyps who underwent endoscopy at 5 academic centers in Japan from October 2017 through March 2018. We performed a retrospective comparative analysis of the diagnostic performance of EndoBRAIN vs that of 30 endoscopists (20 trainees and 10 experts); the endoscopists assessed images from 100 cases produced via white-light microscopy, endocytoscopy with methylene blue staining, and endocytoscopy with narrow-band imaging. EndoBRAIN was used to assess endocytoscopic, but not white-light, images. The primary outcome was the accuracy of EndoBrain in distinguishing neoplasms from non-neoplasms, compared with that of endoscopists, using findings from pathology analysis as the reference standard. RESULTS: In analysis of stained endocytoscopic images, EndoBRAIN identified colon lesions with 96.9% sensitivity (95% CI, 95.8%-97.8%), 100% specificity (95% CI, 99.6%-100%), 98% accuracy (95% CI, 97.3%-98.6%), a 100% positive-predictive value (95% CI, 99.8%-100%), and a 94.6% negative-predictive (95% CI, 92.7%-96.1%); these values were all significantly greater than those of the endoscopy trainees and experts. In analysis of narrow-band images, EndoBRAIN distinguished neoplastic from non-neoplastic lesions with 96.9% sensitivity (95% CI, 95.8-97.8), 94.3% specificity (95% CI, 92.3-95.9), 96.0% accuracy (95% CI, 95.1-96.8), a 96.9% positive-predictive value, (95% CI, 95.8-97.8), and a 94.3% negative-predictive value (95% CI, 92.3-95.9); these values were all significantly higher than those of the endoscopy trainees, sensitivity and negative-predictive value were significantly higher but the other values are comparable to those of the experts. CONCLUSIONS: EndoBRAIN accurately differentiated neoplastic from non-neoplastic lesions in stained endocytoscopic images and endocytoscopic narrow-band images, when pathology findings were used as the standard. This technology has been authorized for clinical use by the Japanese regulatory agency and should be used in endoscopic evaluation of small polyps more widespread clinical settings. UMIN clinical trial no: UMIN000028843.

Treatment policy for colonic laterally spreading tumors based on each clinicopathologic feature of 4 subtypes: actual status of pseudo-depressed type.

BACKGROUND AND AIMS: Laterally spreading tumors (LSTs) are originally classified into 4 subtypes. Pseudo-depressed nongranular types (LSTs-NG-PD) are gaining attention because of their high malignancy potential. Previous studies discussed the classification of nongranular (LST-NG) and granular types (LST-G); however, the actual condition or indication for endoscopic treatment of LSTs-NG-PD remains unclear. We aimed to compare the submucosal invasion pattern of LSTs-NG-PD with the other 3 subtypes. METHODS: A total of 22,987 colonic neoplasms including 2822 LSTs were resected endoscopically or surgically at Showa University Northern Yokohama Hospital. In these LSTs, 322 (11.4%) were submucosal invasive carcinomas. We retrospectively evaluated the clinicopathologic features of LSTs divided into 4 subtypes. In 267 LSTs resected en bloc, their submucosal invasion site was further evaluated. RESULTS: The frequency of LSTs in all colonic neoplasms was significantly higher in women (14.9%) than in men (11.0%). Rates of submucosal invasive carcinoma were .8% in the granular homogenous type (LSTs-G-H), 15.2% in the granular nodular mixed type (LSTs-G-M), 8.0% in the nongranular flat elevated type (LSTs-NG-F), and 42.5% in LSTs-NG-PD. Tumor size was associated with submucosal invasion rate in LSTs-NG-F and LSTs-NG-PD (P < .001). The multifocal invasion rate of LSTs-NG-PD (46.9%) was significantly higher than that of LSTs-G-M (7.9%) or LSTs-NG-F (11.8%). In LSTs-NG-PD, the invasion was significantly deeper (≥1000 µm) if observed in 1 site. CONCLUSIONS: For LSTs-G-M and LSTs-NG-F that may have invaded the submucosa, en bloc resection could be considered. Considering that LSTs-NG-PD had a higher submucosal invasion rate, more multifocal invasive nature, and deeper invasion tendency, regardless if invasion was only observed in 1 site, than LSTs-NG-F, we should endoscopically distinguish LSTs-NG-PD from LSTs-NG-F and strictly adopt en bloc resection by endoscopic submucosal dissection or surgery for LSTs-NG-PD. (Clinical trial registration number: UMIN 000020261.).

Endocytoscopy for the differential diagnosis of colorectal low-grade adenoma: a novel possibility for the "resect and discard" strategy.

BACKGROUND AND AIMS: Endocytoscopy, a next-generation endoscopic system, facilitates observation at a maximum magnification of ×520. To our knowledge, no study has reported high-precision diagnosis of colorectal low-grade adenoma, endoscopically. We aimed to reveal which endocytoscopic findings may be used as indicators of low-grade adenoma and to assess whether a "resect and discard" strategy using endocytoscopy is feasible. METHODS: Lesions diagnosable with endocytoscopy were examined retrospectively between May 2005 and July 2017. A normal pit-like structure in endocytoscopic images was considered a normal pit (NP) sign and used as an indicator of low-grade adenoma. The primary outcome was the diagnostic accuracy of the NP sign for low-grade adenoma. We evaluated agreement rates between endocytoscopic and pathologic diagnosis for surveillance colonoscopy interval recommendation (SCIR) and performed a validation study to verify the agreement rates. RESULTS: For 748 lesions in 573 cases diagnosed as colorectal adenoma using endocytoscopy, the results were as follows: sensitivity of the NP sign for low-grade adenoma, 85.0%; specificity, 90.7%; positive predictive value, 96.6%; negative predictive value, 66.1%; accuracy, 86.4%; and positive likelihood ratio, 9.2 (P < .001). The agreement rate between endocytoscopic and pathologic diagnosis for SCIR was 94.4% (95% confidence interval [CI], 92.2%-96.1%; P < .001) under United States guidelines and 96.3% (95% CI, 94.5%-97.7%; P < .001) under European Union guidelines. All inter- and intraobserver agreement rates for expert and nonexpert endoscopists had κ values ≥0.8 except one nonexpert pair. CONCLUSIONS: Endocytoscopy is an effective modality in determining the differential diagnosis of colorectal low-grade adenoma. (University Hospital Medical Information Network Clinical Trials database registration number: UMIN000018623.).

Left-sided location is a risk factor for lymph node metastasis of T1 colorectal cancer: a single-center retrospective study.

PURPOSE: Although some studies have reported differences in clinicopathological features between left- and right-sided advanced colorectal cancer (CRC), there are few reports regarding early-stage disease. In this study, we aimed to compare the clinicopathological features of left- and right-sided T1 CRC. METHODS: Subjects were 1142 cases with T1 CRC undergoing surgical or endoscopic resection between 2001 and 2018 at Showa University Northern Yokohama Hospital. Of these, 776 cases were left-sided (descending colon to rectum) and 366 cases were right-sided (cecum to transverse colon). We compared clinical (patients age, sex, tumor size, morphology, initial treatment) and pathological features (invasion depth, histological grade, lymphatic invasion, vascular invasion, tumor budding) including lymph node metastasis (LNM). RESULTS: Left-sided T1 CRC showed significantly higher rates of LNM (left-sided 12.0% vs. right-sided 5.4%, P < 0.05) and lymphatic invasion (left-sided 32.7% vs. right-sided 23.2%, P < 0.05). Especially, the sigmoid colon and rectum showed higher rates of LNM (12.4% and 12.1%, respectively) than other locations. Patients with left-sided T1 CRC were younger than those with right-sided T1 CRC (64.9 years ±11.5 years vs. 68.7 ± 11.6 years, P < 0.05), as well as significantly lower rates of poorly differentiated carcinoma/mucinous carcinoma than right-sided T1 CRC (11.6% vs. 16.1%, P < 0.05). CONCLUSION: Left-sided T1 CRC, especially in the sigmoid colon and rectum, exhibited higher rates of LNM than right-sided T1 CRC, followed by higher rates of lymphatic invasion. These results suggest that tumor location should be considered in decisions regarding additional surgery after endoscopic resection. TRIAL REGISTRATION: This study was registered with the University Hospital Medical Network Clinical Trials Registry ( UMIN 000032733 ).

Endocytoscopic intramucosal capillary network changes and crypt architecture abnormalities can predict relapse in patients with an ulcerative colitis Mayo endoscopic score of 1.

OBJECTIVES: Recent studies have suggested the necessity of therapeutic intervention for patients with ulcerative colitis at high risk of clinical relapse with a Mayo endoscopic score (MES) of 1. The aim of this retrospective cohort study was to demonstrate the impact of intramucosal capillary network changes and crypt architecture abnormalities to stratify the risk of relapse in patients with an MES of 1. METHODS: All included patients had an MES of ≤1 and confirmed sustained clinical remission between October 2016 and April 2019. We classified patients with an MES of 1 as "intramucosal capillary/crypt (ICC)-active" or "ICC-inactive" using endocytoscopic evaluation. We followed patients until October 2019 or until relapse; the main outcome measure was the difference in clinical relapse-free rates between ICC-active and ICC-inactive patients with an MES of 1. RESULTS: We included 224 patients and analyzed data for 218 (82 ICC-active and 54 ICC-active with an MES of 1 and 82 with an MES of 0). During follow-up, among the patients with an MES of 1, 30.5% (95% confidence interval 20.8-41.6; 25/82) of the patients relapsed in the ICC-active group and 5.6% (95% confidence interval 1.2-15.4; 3/54) of the patients relapsed in the ICC-inactive group. The ICC-inactive group had a significantly higher clinical relapse-free rate compared with the ICC-active group (P < 0.01). CONCLUSIONS: In vivo intramucosal capillary network and crypt architecture patterns stratified the risk of clinical relapse in patients with an MES of 1 (UMIN 000032580; UMIN 000036359).

Fully automated diagnostic system with artificial intelligence using endocytoscopy to identify the presence of histologic inflammation associated with ulcerative colitis (with video).

BACKGROUND AND AIMS: In the treatment of ulcerative colitis (UC), an incremental benefit of achieving histologic healing beyond that of endoscopic mucosal healing has been suggested; persistent histologic inflammation increases the risk of exacerbation and dysplasia. However, identification of persistent histologic inflammation is extremely difficult using conventional endoscopy. Furthermore, the reproducibility of endoscopic disease activity is poor. We developed and evaluated a computer-aided diagnosis (CAD) system to predict persistent histologic inflammation using endocytoscopy (EC; 520-fold ultra-magnifying endoscope). METHODS: We evaluated the accuracy of the CAD system using test image sets. First, we retrospectively reviewed the data of 187 patients with UC from whom biopsy samples were obtained after endocytoscopic observation. EC images and biopsy samples of each patient were collected from 6 colorectal segments: cecum, ascending colon, transverse colon, descending colon, sigmoid colon, and rectum. All EC images were tagged with reference to the biopsy sample's histologic activity. For validation samples, 525 validation sets of 525 independent segments were collected from 100 patients, and 12,900 EC images from the remaining 87 patients were used for machine learning to construct CAD. The primary outcome measure was the diagnostic ability of CAD to predict persistent histologic inflammation. Its reproducibility for all test images was also assessed. RESULTS: CAD provided diagnostic sensitivity, specificity, and accuracy as follows: 74% (95% confidence interval, 65%-81%), 97% (95% confidence interval, 95%-99%), and 91% (95% confidence interval, 83%-95%), respectively. Its reproducibility was perfect (κ = 1). CONCLUSIONS: Our CAD system potentially allows fully automated identification of persistent histologic inflammation associated with UC.

Real-Time Use of Artificial Intelligence in Identification of Diminutive Polyps During Colonoscopy: A Prospective Study.

Background: Computer-aided diagnosis (CAD) for colonoscopy may help endoscopists distinguish neoplastic polyps (adenomas) requiring resection from nonneoplastic polyps not requiring resection, potentially reducing cost. Objective: To evaluate the performance of real-time CAD with endocytoscopes (×520 ultramagnifying colonoscopes providing microvascular and cellular visualization of colorectal polyps after application of the narrow-band imaging [NBI] and methylene blue staining modes, respectively). Design: Single-group, open-label, prospective study. (UMIN [University hospital Medical Information Network] Clinical Trial Registry: UMIN000027360). Setting: University hospital. Participants: 791 consecutive patients undergoing colonoscopy and 23 endoscopists. Intervention: Real-time use of CAD during colonoscopy. Measurements: CAD-predicted pathology (neoplastic or nonneoplastic) of detected diminutive polyps (≤5 mm) on the basis of real-time outputs compared with pathologic diagnosis of the resected specimen (gold standard). The primary end point was whether CAD with the stained mode produced a negative predictive value (NPV) of 90% or greater for identifying diminutive rectosigmoid adenomas, the threshold required to "diagnose-and-leave" nonneoplastic polyps. Best- and worst-case scenarios assumed that polyps lacking either CAD diagnosis or pathology were true- or false-positive or true- or false-negative, respectively. Results: Overall, 466 diminutive (including 250 rectosigmoid) polyps from 325 patients were assessed by CAD, with a pathologic prediction rate of 98.1% (457 of 466). The NPVs of CAD for diminutive rectosigmoid adenomas were 96.4% (95% CI, 91.8% to 98.8%) (best-case scenario) and 93.7% (CI, 88.3% to 97.1%) (worst-case scenario) with stained mode and 96.5% (CI, 92.1% to 98.9%) (best-case scenario) and 95.2% (CI, 90.3% to 98.0%) (worst-case scenario) with NBI. Limitation: Two thirds of the colonoscopies were conducted by experts who had each experienced more than 200 endocytoscopies; 186 polyps not assessed by CAD were excluded. Conclusion: Real-time CAD can achieve the performance level required for a diagnose-and-leave strategy for diminutive, nonneoplastic rectosigmoid polyps. Primary Funding Source: Japan Society for the Promotion of Science.

A novel ability of endocytoscopy to diagnose histological grade of differentiation in T1 colorectal carcinomas.

BACKGROUND AND STUDY AIMS: Endocytoscopic images closely resemble histopathology. We assessed whether endocytoscopy could be used to determine T1 colorectal cancer histological grade. PATIENTS AND METHODS: Endocytoscopic images of 161 lesions were divided into three types: tubular gland lumens, unclear gland lumens, and fused gland formations on endocytoscopy (FGFE). We retrospectively compared endocytoscopic findings with histological grade in the resected specimen superficial layer, and examined the incidence of risk factors for lymph node metastasis. RESULTS: Of the 118 eligible lesions, the sensitivity, specificity, accuracy, negative predictive value, and positive likelihood ratio of tubular or unclear gland lumens to identify well-differentiated adenocarcinomas were 91.0 %, 93.1 %, 91.5 %, 77.1 %, and 13.20, respectively. To identify moderately differentiated adenocarcinomas for FGFE, these values were 93.1 %, 91.0 %, 91.5 %, 97.6 %, and 10.36, respectively. In the 35 lesions with FGFE, the rates of massive invasion, lymphovascular infiltration, and tumor budding were 97.1 %, 60.0 %, and 37.1 %, respectively. CONCLUSIONS: Endocytoscopy could be used to diagnose T1 colorectal cancer histological grade, and FGFE was a marker for recommending surgery.

Artificial intelligence may help in predicting the need for additional surgery after endoscopic resection of T1 colorectal cancer.

BACKGROUND AND STUDY AIMS: Decisions concerning additional surgery after endoscopic resection of T1 colorectal cancer (CRC) are difficult because preoperative prediction of lymph node metastasis (LNM) is problematic. We investigated whether artificial intelligence can predict LNM presence, thus minimizing the need for additional surgery. PATIENTS AND METHODS: Data on 690 consecutive patients with T1 CRCs that were surgically resected in 2001 - 2016 were retrospectively analyzed. We divided patients into two groups according to date: data from 590 patients were used for machine learning for the artificial intelligence model, and the remaining 100 patients were included for model validation. The artificial intelligence model analyzed 45 clinicopathological factors and then predicted positivity or negativity for LNM. Operative specimens were used as the gold standard for the presence of LNM. The artificial intelligence model was validated by calculating the sensitivity, specificity, and accuracy for predicting LNM, and comparing these data with those of the American, European, and Japanese guidelines. RESULTS: Sensitivity was 100 % (95 % confidence interval [CI] 72 % to 100 %) in all models. Specificity of the artificial intelligence model and the American, European, and Japanese guidelines was 66 % (95 %CI 56 % to 76 %), 44 % (95 %CI 34 % to 55 %), 0 % (95 %CI 0 % to 3 %), and 0 % (95 %CI 0 % to 3 %), respectively; and accuracy was 69 % (95 %CI 59 % to 78 %), 49 % (95 %CI 39 % to 59 %), 9 % (95 %CI 4 % to 16 %), and 9 % (95 %CI 4 % - 16 %), respectively. The rates of unnecessary additional surgery attributable to misdiagnosing LNM-negative patients as having LNM were: 77 % (95 %CI 62 % to 89 %) for the artificial intelligence model, and 85 % (95 %CI 73 % to 93 %; P < 0.001), 91 % (95 %CI 84 % to 96 %; P < 0.001), and 91 % (95 %CI 84 % to 96 %; P < 0.001) for the American, European, and Japanese guidelines, respectively. CONCLUSIONS: Compared with current guidelines, artificial intelligence significantly reduced unnecessary additional surgery after endoscopic resection of T1 CRC without missing LNM positivity.

Risk factors of recurrence in T1 colorectal cancers treated by endoscopic resection alone or surgical resection with lymph node dissection.

PURPOSE: The recurrence of T1 colorectal cancers is relatively rare, and the prognostic factors still remain obscure. This study aimed to clarify the risk factors for recurrence in patients with T1 colorectal cancers treated by endoscopic resection (ER) alone or surgical resection (SR) with lymph node dissection, respectively. METHODS: We reviewed 930 patients with resected T1 colorectal cancers (mean follow-up, 52.3 months). Patients were divided into two groups: those who underwent ER alone (298 cases), and those who underwent initial or additional SR with lymph node dissection (632 cases). Group differences in recurrence-free survival were evaluated using the Kaplan-Meier method and log-rank test. Associations between recurrence and clinicopathological features were evaluated in Cox regression analyses; hazard ratios (HRs) were calculated for the total population and each group. RESULTS: Recurrence occurred in four cases (1.34%) in the ER group and six cases (0.95%) in the SR group (p = 0.32). Endoscopic resection, rectal location, and poor or mucinous (Por/Muc) differentiation were prognostic factors for recurrence in the total population. Por/Muc differentiation was prognostic factor in both groups. Female sex, depressed-type morphology, and lymphatic invasion were also prognostic factors in the ER group, but not in the SR group. CONCLUSIONS: Endoscopic resection, rectal location, and Por/Muc differentiation are prognostic factors in the total population. For patients who undergo ER alone, female sex, depressed-type morphology, and lymphatic invasion are also risk factors for recurrence. For such patients, regional en-bloc surgery with lymph node dissection could reduce the risk of recurrence.

Classification of nuclear morphology in endocytoscopy of colorectal neoplasms.

BACKGROUND AND AIMS: We investigated endocytoscopy (EC) findings that were considered risk factors for colorectal neoplasms and determined whether they could be used as new indices to identify carcinomas with massive submucosal invasion (SM-m) or worse outcomes. METHODS: We performed a multivariate analysis of 8 factors on EC images to determine whether they were associated with SM-m or worse. Based on the results, we divided the EC3a category of the EC classification into low grade or high grade and investigated the diagnostic accuracy of this subclassification. In addition, we compared the diagnostic ability of EC for SM-m with that of other modalities (narrow-band imaging and pit pattern). RESULTS: The multivariate analysis indicated that unclear glandular lumens (ULs), high degree of nuclear enlargement (HNE), and multilayered nuclei (MNs) were the most useful factors for the diagnosis of SM-m or worse. The odds ratios for these factors were 12.47, 12.29, and 10.48, respectively (P < .001). The sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and positive likelihood ratio for the diagnostic accuracy of the EC3a subclassification were 88.9%, 91.3%, 75.0%, 96.6%, 90.8%, and 10.2, respectively (P < .001). The sensitivity, negative predictive value, and accuracy of EC were significantly higher than those of narrow-band imaging and pit pattern. CONCLUSIONS: From the EC findings, the presence of ULs, HNE, and MNs are important risk factors for SM-m or worse outcomes. Furthermore, the EC3a subclassification taking these findings into consideration could be effective for the diagnosis of SM-m or worse. (Clinical trial registration number: UMIN 000014906.).

Management and risk factor of stenosis after endoscopic submucosal dissection for colorectal neoplasms.

BACKGROUND AND AIM: Although endoscopic submucosal dissection (ESD) enables en bloc removal of large colorectal neoplasms, the incidence of stenosis after ESD and its risk factors have not been well described. This study aimed to determine the risk factors of stenosis and verify the surveillance and treatment of stenosis. METHODS: This retrospective study included 822 patients, with a total of 912 consecutive colorectal lesions, who underwent ESD from September 2003 to May 2015. The main outcome measures were incidence of stenosis and its relationship with the clinicopathologic factors in surveillance. RESULTS: Surveillance endoscopy was performed 6 months after ESD. Four of the 822 patients (0.49%) developed stenosis and required unanticipated endoscopy. The other 908 cases in 818 patients showed no symptoms or only slight abdominal discomfort (that was controlled with medication) and did not require any dilation or steroid therapies. Post-ESD stenosis was observed in 11.1% (2/18) of patients with circumferential resection between ≥90% and <100% and in 50% (2/4) of patients with circumferential resection of 100%. Among the 50 cases with a circumferential mucosal defect ≥75%, a circumferential mucosal defect ≥90% was a significant risk factor (P = .005). Four patients with stenosis were treated successfully by endoscopic dilation. CONCLUSIONS: Circumferential mucosal defect of more than 90% is a significant risk factor for stenosis after colorectal ESD. Surveillance endoscopy 6 months after ESD is recommended to assess for development of stenosis. Defects smaller than 90% do not require close endoscopic follow-up or prophylactic measures for prevention of post-ESD stenosis. (UMIN clinical trial registration number: UMIN000015754.).

Accuracy of diagnosing invasive colorectal cancer using computer-aided endocytoscopy.

Background and study aims Invasive cancer carries the risk of metastasis, and therefore, the ability to distinguish between invasive cancerous lesions and less-aggressive lesions is important. We evaluated a computer-aided diagnosis system that uses ultra-high (approximately × 400) magnification endocytoscopy (EC-CAD). Patients and methods We generated an image database from a consecutive series of 5843 endocytoscopy images of 375 lesions. For construction of a diagnostic algorithm, 5543 endocytoscopy images from 238 lesions were randomly extracted from the database for machine learning. We applied the obtained algorithm to 200 endocytoscopy images and calculated test characteristics for the diagnosis of invasive cancer. We defined a high-confidence diagnosis as having a ≥ 90 % probability of being correct. Results Of the 200 test images, 188 (94.0 %) were assessable with the EC-CAD system. Sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) were 89.4 %, 98.9 %, 94.1 %, 98.8 %, and 90.1 %, respectively. High-confidence diagnosis had a sensitivity, specificity, accuracy, PPV, and NPV of 98.1 %, 100 %, 99.3 %, 100 %, and 98.8 %, respectively. Conclusion: EC-CAD may be a useful tool in diagnosing invasive colorectal cancer.