Incidence Of Interval Colorectal Cancer Research Articles

An Accurate and Automated Method for Adenoma Detection Rate and Report Card Generation Utilizing Common Electronic Health Records.

To develop an automated method for Adenoma Detection Rate (ADR) calculation and report card generation using common electronic health records (EHRs). ADR is the most widely accepted colonoscopy quality indicator and is inversely associated with interval colorectal cancer incidence and mortality. However, ADR is difficult to efficiently measure and disseminate, due to need for data integration from distinct electronic databases. We migrated data from an endoscopy reporting software (Endosoft) to Epic Reporting Servers where it was combined with anatomic pathology data (Beaker Lab Information System, EPIC Systems). A natural language processing expression was developed to search Beaker pathology reports for accurate identification of adenomatous polyps. A blinded physician manually validated a final cohort of 200 random procedures. ADR report cards were automatically generated utilizing the Crystal Reports feature within EPIC. Validation of the natural language processing algorithm for ADR showed a sensitivity, specificity, and accuracy of 100%. ADR was automatically calculated for 12 endoscopists over a calendar year. Two thousand two hundred seventy-six screening colonoscopies were performed with 775 procedures having a least one adenoma detected, for a total ADR of 34%. Report cards were successfully generated within the EPIC EHR and distributed to endoscopists by secure e-mail. We describe an accurate, automated and scalable process for ADR calculation and reporting utilizing commonly adopted EHRs and data integration methods. By integrating the process of ADR collection and streamlining dissemination of reports, this methodology is poised to enhance colonoscopy quality care across health care networks that use it.

Using fecal immunochemical test values below conventional cut-off to individualize colorectal cancer screening.

Background and study aims Of the participants in the Danish screening program, 89.9 % to 92.5 % have fecal immunochemical test (FIT) values < 10 μg/g feces (equivalent to 50 ng hemoglobin/mL buffer). This study aimed to investigate the risk of interval colorectal cancer (CRC) in this group before the next biennial screening round. Patients and methods This cohort study included all citizens from the region of Southern Denmark who participated in the Danish bowel screening program from 2014 trough 2016 and had a FIT value < 10 μg/g feces. Individuals receiving a CRC diagnosis were identified through the national CRC registry, with a follow up of 2 years corresponding to the current screening interval. We also examined the 3-year CRC incidence. Hazard ratios (HRs) were estimated using univariate and multivariate Cox proportional hazard regression models. Results Data from 185,654 citizens presenting with a FIT value < 10 μg/g feces were eligible for analysis. Overall, interval CRC incidence was 0.07 % within 2 years with HRs of 4.16 (95 % confidence interval [CI] 2.67;6.48) and 5.8 (95 % CI 3.34;10.05) for FIT values of 4 to 6.9 μg/g feces and 7 to 9.9 μg/g feces, respectively, compared to those having a FIT value below the limit of quantification of 4 μg/g feces. After 3 years, the overall CRC incidence increased to 0.14 %; however, this was not significant. Conclusions This study demonstrates a positive correlation between FIT value and risk of interval cancer even for very low values. It further suggests that an increase in the screening interval could be reasonable in the low FIT categories.

Racism Is a Modifiable Risk Factor: Relationships Among Race, Ethnicity, and Colorectal Cancer Outcomes

Racism Is a Modifiable Risk Factor: Relationships Among Race, Ethnicity, and Colorectal Cancer Outcomes

Significant decrease in interval colorectal cancer incidence after implementing immunochemical testing in a multiple-round guaiac-based screening programme.

We aimed to evaluate the effects of switching to faecal immunochemical testing (FIT) on the cumulative 2-year incidence rate of interval cancers, interval cancer rate and test sensitivity within a mature population-based colorectal cancer screening programme consisting of six rounds of biennial guaiac faecal occult blood testing (gFOBT). The FIT results were compared with those of gFOBT used in each of the previous two rounds. For the three rounds analysed, 279,041 tests were performed by 156,186 individuals. Logistic regression analysis was used to determine interval cancer risk factors (Poisson regression) and to compare the sensitivity of FIT to gFOBT. There were 612 cases of screen-detected cancers and 209 cases of interval cancers. The sex- and age-adjusted cumulative 2-year incidence rates of interval cancers were 55.7 (95% CI, 45.3-68.5), 42.4 (95% CI, 32.6-55.2) and 15.8 (95% CI, 10.9-22.8) per 100,000 person-years after the last two rounds of gFOBT and FIT, respectively. The FIT/gFOBT incidence rate ratio was 0.38 [95% CI, 0.27-0.54] (P < 0.001). Sex- and age-adjusted sensitivity was significantly higher with FIT than with gFOBT (OR = 6.70 [95% CI, 4.48-10.01], P < 0.0001). This population-based study revealed a dramatic decrease in the cumulative incidence rates of interval cancers after switching from gFOBT to FIT. These data provide an additional incentive for countries still using gFOBT to switch to FIT.

Effect of dynamic position changes on adenoma detection rate during colonoscope withdrawal: systematic review and meta-analysis.

Background and study aims The adenoma detection rate (ADR) is an important quality metric of colonoscopy. Higher ADR correlates with lower incidence of interval colorectal cancer. ADR is variable between endoscopists and depends upon the withdrawal technique amongst other factors. Dynamic position change (lateral rotation of patients with a view to keep the portion of the colon being inspected at a higher level) helps with luminal distension during the withdrawal phase. However, impact of this on ADR is not known in a pooled sample. We performed a systematic review and meta-analysis to study the impact of dynamic position changes during withdrawal phase of colonoscopy on ADR Methods A comprehensive search of MEDLINE, EMBASE, Google Scholar, and the Cochrane Database was conducted from each database’s inception to search for studies comparing dynamic position changes during colonoscope withdrawal with static left lateral position (control). The primary outcome of interest was ADR. Other studied outcomes were polyp detection rate (PDR) and withdrawal time. Outcomes were reported as pooled odds ratio (OR) with 95 % confidence intervals (CI) with statistical significance ( P < 0.05). RevMan 5.3 software was used for statistical analysis. Results Six studies were included in our analysis with 2860 patients. Of these, dynamic position change was implemented in 1177 patients while 1183 patients served as the controls. ADR was significantly higher in the dynamic position change group with pooled OR 1.36 (95 % CI, 1.15–1.61; P < 0.01). There was low heterogeneity in inclusion studies (I 2 = 0 %). PDR was numerically higher in position change group (53.4 % vs 49.6 %) but not statistically significant ( P = 0.16). Mean withdrawal time did not significantly change with dynamic position change (12.43 min vs 11.46 min, P = 0.27). Conclusion Position change during the withdrawal phase of colonoscopy can increase the ADR compared to static left lateral position. This is an easy and practical technique that can be implemented to improve ADR.

Incidence of interval colorectal cancer: A tertiary hospital report

AbstractColorectal cancer (CRC) remains the leading cause of death in Taiwan despite mandatory CRC screening programs for early detection. The development of CRC between periods of follow‐up remains a Taiwanese health and economic burden. This study aimed to investigate the relationship between consequent progression of polyps and interval CRC and identify potential risk factors. This retrospective chart review study reviewed 1671 CRC patients between 1 January 2002 and 31 December 2018, in Kaohsiung Chang Gung Memorial Hospital for interval CRC analysis. Interval CRC is defined as CRC diagnosed within 5 years after a negative colonoscopy. We further analyzed potential risk factors associated with interval CRC such as sex, age, underlying disease, and previous pathologic type of polyps by univariate and multivariate analyses. A total of 44 patients developed interval CRC in this study (2.63%). Ten patients were from the proctologist division and five from the gastroenterology department. Mean diagnostic interval was 919.45 ± 547.78 days in male and 897.87 ± 586.14 days in female patients. Multivariate analysis showed that right‐side colon polyp demonstrated independent risk factors for interval CRC development in our observation (P &lt; .001; hazard ratio (HR), 3.897; 95% confidence interval (CI), 1.934‐7.780). In conclusion, interval CRC prevalence in a cohort is 2.63%. Location of polyps over the right colon is relevant to interval CRC occurrence.

Interval colorectal cancers after negative faecal immunochemical test in a 13-year screening programme.

To assess faecal immunochemical test sensitivity for cancer in a very large population-based cohort followed up for six rounds with biennial faecal immunochemical test repetition. This study is based on interval colorectal cancers diagnosed in a cohort of subjects aged 50-69 undergoing repeated faecal immunochemical test screening (six rounds) from 2002 to 2015. Test sensitivity was calculated using both the Proportional Interval Cancer Rate and the Interval Cancer Proportion method. Among 441,647 faecal immunochemical tests (123,347 individuals), 150 interval colorectal cancers were detected after a negative faecal immunochemical test. Interval colorectal cancer incidence rate was 1.87 per 10,000 person-years (95%CI: 1.60-2.20), and it was higher during the second interval year (rate ratio: 1.78; 95%CI: 1.28-2.47), for proximal locations (rate ratio: 3.00; 95%CI: 1.92-4.68), and among 60-71 year old subjects (rate ratio: 2.37; 95%CI: 1.61-3.50). The Proportional Interval Cancer Rate was 13.1%, with an overall faecal immunochemical test sensitivity of 86.9% (95%CI: 84.7-89.0). Sensitivity was lowest at the first round (81.5%; 95%CI: 75.6-86.2), and increased to 91.9% (95%CI: 83.9-96.5) for subsequent rounds. Applying the Interval Cancer Proportion method, sensitivity was 83.9% (95%CI: 81.3-86.2), and it was highest at the first round (89.0%; 95%CI: 85.7-91.6), ranging between 73% and 83.1% at subsequent rounds. A faecal immunochemical test sensitivity for cancer higher than 80% resulted in a low incidence of interval colorectal cancers, representing an accurate estimate of one of the major limits of screening programmes. Due to intrinsic biases, the Proportional Interval Cancer Rate and the Interval Cancer Proportion methods generated different trends in faecal immunochemical test sensitivity by screening round.

Retrograde inspection vs standard forward view for the detection of colorectal adenomas during colonoscopy: A back-to-back randomized clinical trial

BACKGROUNDThe adenoma detection rate (ADR) is inversely associated with the incidence of interval colorectal cancer and serves as a benchmark quality criterion during screening colonoscopy. However, adenoma miss rates reach up to 26% and studies have shown that a second inspection of the right colon in retroflected view (RFV) can increase ADR.AIMTo assess whether inspection of the whole colon in RFV compared to standard forward view (SFV) can increase ADR.METHODSPatients presenting for screening or surveillance colonoscopy were invited to participate in this randomized controlled trial and randomized into two arms. In RFV arm colonoscopy was initially performed with SFV, followed by a second inspection of the whole colon in RFV. In the SFV arm first withdrawal was performed with SFV, followed by a second inspection of the whole colon again with SFV. Number, size and morphology of polyps found during first and second inspection in each colonic segment were recorded and all polyps were removed and sent for histopathology in separate containers.RESULTSTwo hundred and five patients were randomly assigned to the RFV (n = 101) and SFV (n = 104) arm. In the RFV arm, both polyp detection rate (PDR) and ADR were increased under second inspection in RFV (PDR 1st SFV: 39.8%, PDR 2nd RFV: 46.6%; ADR 1st SFV: 35.2%, ADR 2nd RFV: 42%). Likewise, in the SFV arm, PDR and ADR were increased under second inspection (PDR 1st SFV: 37.5%, PDR 2nd SFV: 46.6%; ADR 1st SFV: 34.1%, ADR 2nd SFV: 44.3%) with no significant differences in ADR and PDR between the SFV and RFV arm. Mean number of adenomas per patient (APP) was increased in the RFV and SFV (APP RFV arm: 1st SFV: 1.71; 2nd RFV: 2.38; APP SFV arm: 1st SFV: 1.83, 2nd SFV:2.2). The majority of adenomas additionally found during second inspection in RFV or in SFV were located in the transverse and left-sided colon and were > 5 mm in size.CONCLUSIONSecond inspection of the whole colon leads to increased adenoma detection with no differences between SFV and RFV. Hence, increased detection is most likely a feature of the second inspection itself but not of the inspection mode.

Proportional incidence of interval colorectal cancer in a large population-based faecal immunochemical test screening programme

BackgroundThe European guidelines for quality assurance in colorectal cancer (CRC) screening recommend that interval cancer rate be expressed as a proportion of background incidence rate. AimTo determine the crude and adjusted proportional incidence of interval CRC in an Italian regional two-yearly faecal immunochemical test (FIT) screening programme. MethodsThe programme (year of implementation, 2005) is targeted at over 1,000,000 people aged 50–69 years. The test is a one-sample OC-Sensor (Eiken Chemical Co., Tokyo, Japan). The study covered one-third of the regional area. Excerpts of 434,295 eligible negative FIT records dated 2005–2012 from 193,193 subjects were retrieved from the regional CRC screening data warehouse. By 31 December 2013, the cohort accumulated 198,302 man-years and 235,370 woman-years. Interval CRCs were identified by record-linkage with the local population-based cancer registry. Their number was divided by the expected number, estimated with age-period-cohort models, to obtain the proportional incidence. ResultsThe proportional incidence of interval CRC for men and women was, respectively, 0.06 (95% confidence interval (CI), 0.04–0.09) and 0.17 (95% CI, 0.13–0.23) in the first interval year, and 0.21 (95% CI, 0.16–0.26) and 0.28 (95% CI, 0.22–0.36) in the second year. ConclusionsThe results were acceptable and in line with previous studies.

Deep learning for colorectal polyp detection: time for clinical implementation?

Colonoscopy is one of the most valuable and frequently done clinical procedures worldwide, offering protection against colorectal cancer by detection and removal of neoplastic lesions. 1 Winawer SJ Zauber AG Ho MN et al. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med. 1993; 329: 1977-1981 Crossref PubMed Scopus (3910) Google Scholar Unfortunately, colonoscopy is imperfect and highly operator-dependent, leading to variable protection against colorectal cancer. Global research efforts have focused on improving the quality of colonoscopy. Effect of a deep-learning computer-aided detection system on adenoma detection during colonoscopy (CADe-DB trial): a double-blind randomised studyPolyps initially missed by the endoscopist had characteristics that are sometimes difficult for skilled endoscopists to recognise. Such polyps could be detected using a high-performance CADe system during colonoscopy. The effect of CADe during colonoscopy on the incidence of interval colorectal cancer should be investigated. Full-Text PDF

Effect of a deep-learning computer-aided detection system on adenoma detection during colonoscopy (CADe-DB trial): a double-blind randomised study

Colonoscopy with computer-aided detection (CADe) has been shown in non-blinded trials to improve detection of colon polyps and adenomas by providing visual alarms during the procedure. We aimed to assess the effectiveness of a CADe system that avoids potential operational bias. We did a double-blind randomised trial at the endoscopy centre in Caotang branch hospital of Sichuan Provincial People's Hospital in China. We enrolled consecutive patients (aged 18-75 years) presenting for diagnostic and screening colonoscopy. We excluded patients with a history of inflammatory bowel disease, colorectal cancer, or colorectal surgery or who had a contraindication for biopsy; we also excluded patients who had previously had an unsuccessful colonoscopy and who had a high suspicion for polyposis syndromes, inflammatory bowel disease, and colorectal cancer. We allocated patients (1:1) to colonoscopy with either the CADe system or a sham system. Randomisation was by computer-generated random number allocation. Patients and the endoscopist were unaware of the random assignment. To achieve masking, the output of the system was shown on a second monitor that was only visible to an observer who was responsible for reporting the alerts. The primary outcome was the adenoma detection rate (ADR), which is the proportion of individuals having a complete colonoscopy, from caecum to rectum, who had one or more adenomas detected. The primary analysis was per protocol. We also analysed characteristics of polyps and adenomas missed initially by endoscopists but detected by the CADe system. This trial is complete and is registered with http://www.chictr.org.cn, ChiCTR1800017675. Between Sept 3, 2018, and Jan 11, 2019, 1046 patients were enrolled to the study, of whom 36 were excluded before randomisation, 508 were allocated colonoscopy with polyp detection using the CADe system, and 502 were allocated colonoscopy with the sham system. After further excluding patients who met exclusion criteria, 484 patients in the CADe group and 478 in the sham group were included in analyses. The ADR was significantly greater in the CADe group than in the sham group, with 165 (34%) of 484 patients allocated to the CADe system having one or more adenomas detected versus 132 (28%) of 478 allocated to the sham system (odds ratio 1·36, 95% CI 1·03-1·79; p=0·030). No complications were reported among all colonoscopy procedures. Polyps initially missed by the endoscopist but identified by the CADe system were generally small in size, isochromatic, flat in shape, had an unclear boundary, were partly behind colon folds, and were on the edge of the visual field. Polyps initially missed by the endoscopist had characteristics that are sometimes difficult for skilled endoscopists to recognise. Such polyps could be detected using a high-performance CADe system during colonoscopy. The effect of CADe during colonoscopy on the incidence of interval colorectal cancer should be investigated. None.

Incidence of Interval Colorectal Cancer After Negative Results From First-Round Fecal Immunochemical Screening Tests, by Cutoff Value and Participant Sex and Age

We evaluated the incidence of interval cancers between the first and second rounds of colorectal cancer (CRC) screening with the FOB-Gold fecal immunochemical test (FIT), and the effects of different cutoff values and patient sex and age. We collected data from participants in a population-based CRC screening program in the Netherlands who had a negative result from a first-round of FIT screening. We calculated the cumulative incidence of interval cancer after a negative result from a FIT and the sensitivity of the FIT for detection of CRC at a low (15 μg Hb/g feces) and high (47 μg Hb/g feces) cutoff value. Among the 485,112 participants with a negative result from a FIT, 544 interval cancers were detected; 126 were in the 111,800 participants with negative results from a FIT with the low cutoff value and 418 were in the 373,312 FIT participants with negative results from a FIT with the high cutoff value. The mean age of participants tested with the low cutoff value was 72.0 years and the mean age of participants tested the high cutoff value was 66.7 years. The age-adjusted 2-year cumulative incidence of interval cancer after a negative result from a FIT were 9.5 per 10,000 persons at the low cutoff value vs 13.8 per 10,000 persons at the high cutoff value (P < .005). The age-adjusted sensitivity of the FIT for CRC were 90.5% for the low cutoff value vs 82.9% for the high cutoff (P < .0001). The FIT identified men with CRC with 87.4% sensitivity and women with CRC with 82.6% sensitivity (P < .001). In an analysis of data from a FIT population-based screening program in the Netherlands, we found that incidence of interval CRC after a negative result from a FIT to be low. Although the sensitivity of detection of CRC decreased with a higher FIT cutoff value, it remained above 80%.

Challenges Facing the Detection of Colonic Polyps: What Can Deep Learning Do?

Colorectal cancer (CRC) is one of the most common causes of cancer mortality in the world. The incidence is related to increases with age and western dietary habits. Early detection through screening by colonoscopy has been proven to effectively reduce disease-related mortality. Currently, it is generally accepted that most colorectal cancers originate from adenomas. This is known as the “adenoma–carcinoma sequence”, and several studies have shown that early detection and removal of adenomas can effectively prevent the development of colorectal cancer. The other two pathways for CRC development are the Lynch syndrome pathway and the sessile serrated pathway. The adenoma detection rate is an established indicator of a colonoscopy’s quality. A 1% increase in the adenoma detection rate has been associated with a 3% decrease in interval CRC incidence. However, several factors may affect the adenoma detection rate during a colonoscopy, and techniques to address these factors have been thoroughly discussed in the literature. Interestingly, despite the use of these techniques in colonoscopy training programs and the introduction of quality measures in colonoscopy, the adenoma detection rate varies widely. Considering these limitations, initiatives that use deep learning, particularly convolutional neural networks (CNNs), to detect cancerous lesions and colonic polyps have been introduced. The CNN architecture seems to offer several advantages in this field, including polyp classification, detection, and segmentation, polyp tracking, and an increase in the rate of accurate diagnosis. Given the challenges in the detection of colon cancer affecting the ascending (proximal) colon, which is more common in women aged over 65 years old and is responsible for the higher mortality of these patients, one of the questions that remains to be answered is whether CNNs can help to maximize the CRC detection rate in proximal versus distal colon in relation to a gender distribution. This review discusses the current challenges facing CRC screening and training programs, quality measures in colonoscopy, and the role of CNNs in increasing the detection rate of colonic polyps and early cancerous lesions.

Colonoscopic withdrawal time and adenoma detection in the right colon

Shorter colonoscopic withdrawal time (CWT) has been associated with lower adenoma detection rate (ADR), which can increase the risk of interval colorectal cancer (ICC) that commonly arises in the right colon (RC). Therefore, a better ADR in the RC could decrease the incidence of ICC. We analyzed the relationship between CWT and ADR in the RC and entire colon.We retrospectively reviewed the patients who had undergone screening colonoscopy at Chungnam National University Hospital between March 2015 and February 2016. We enrolled 5370 patients in whom colonoscopies were performed by 7 gastroenterologists. We categorized patients into 4 groups in the RC and 6 groups in the entire colon by CWT. Multivariable analysis was used for detection of adenoma in the RC and entire colon.In the RC, the odds ratio (OR) of CWT longer than 3 minutes was 3.70, compared to CWT of <2 minutes [3.06–4.85, 95% confidence interval (CI), P < .001]. In the entire colon, the OR of CWT between 9 to 10 minutes and longer than 10 minutes was 3.34 [2.61–4.27, 95% CI, P < .001] and 3.49 [2.80–4.33, 95% CI, P < .001] compared to CWT of <6 minutes.Based on our result, we suggest that the optimum CWT in the RC should exceed 3 minutes, and considering the “ceiling effect,” the optimum CWT in the entire colon should exceed 9 minutes.

Incidence of faecal occult blood test interval cancers in population-based colorectal cancer screening: a systematic review and meta-analysis.

Faecal immunochemical tests (FITs) are replacing guaiac faecal occult blood tests (gFOBTs) for colorectal cancer (CRC) screening. Incidence of interval colorectal cancer (iCRC) following a negative stool test result is not yet known. We aimed to compare incidence of iCRC following a negative FIT or gFOBT. We searched Ovid Medline, Embase, Cochrane Library, Science Citation Index, PubMed and Google Scholar from inception to 12 December 2017 for citations related to CRC screening based on stool tests. We included studies on FIT or gFOBT iCRC in average-risk screening populations. Main outcome was pooled incidence rate of iCRCs per 100 000 person-years (p-y). Pooled incidence rates were obtained by fitting random-effect Poisson regression models. We identified 7 426 records and included 29 studies. Meta-analyses comprised data of 6 987 825 subjects with a negative test result, in whom 11 932 screen-detected CRCs and 5 548 gFOBT or FIT iCRCs were documented. Median faecal haemoglobin (Hb) positivity cut-off used was 20 (range 10-200) µg Hb/g faeces in the 17 studies that provided FIT results. Pooled incidence rates of iCRC following FIT and gFOBT were 20 (95% CI 14 to 29; I2=99%) and 34 (95% CI 20 to 57; I2=99%) per 100 000 p-y, respectively. Pooled incidence rate ratio of FIT versus gFOBT iCRC was 0.58 (95% CI 0.32 to 1.07; I2=99%) and 0.36 (95% CI 0.17 to 0.75; I2=10%) in sensitivity analysis. For every FIT iCRC, 2.6 screen-detected CRCs were found (ratio 1:2.6); for gFOBT, the ratio between iCRC and screen-detected CRC was 1:1.2. Age below 60 years and the third screening round were significantly associated with a lower iCRC rate. A negative gFOBT result is associated with a higher iCRC incidence than a negative FIT. This supports the use of FIT over gFOBT as CRC screening tool.

Su1729 - 5-Year Interval Colorectal Cancer Incidence Trends under age 50. Results of Population-Based Study

Su1729 - 5-Year Interval Colorectal Cancer Incidence Trends under age 50. Results of Population-Based Study

Efficacy of Narrow-Band Imaging (NBI), Especially the Second Generation NBI for Adenoma Detection Rate: Systematic Review and Meta-Analysis

Introduction: Incidence of interval colorectal cancer (CRC), CRC diagnosed between the time of screening colonoscopy and the scheduled time of surveillance colonoscopy, was reported to be inversely correlated with higher adenoma detection rate (ADR). Image enhanced endoscopy, such as narrow band imaging (NBI) was expected to improve ADR. However, there has been no evidence of significant efficacy of the 1st generation NBI for ADR compared to conventional white light. The 2nd generation NBI was invented with fixing relatively dark image of the 1st generation NBI. The 2nd generation NBI was reported to provide at least two-fold brighter image. The aim of this study was to determine the efficacy of NBI, especially the 2nd generation NBI for ADR. Methods: We performed a comprehensive literature review following Preferred Reporting Items for Systematic reviews and Meta-Analyses from 1966 through July 2016 using appropriate search terms in MEDLINE, EMBASE, and Cochrane library.Inclusion criteria required full literatures published in English and randomized controlled trial (RCT). Heterogeneity was evaluated by I2, pooled estimates were calculated by fixed effects model (FEM) (Mantel-Haenszel method) or random effects model (REM) (DerSimonian and Laird method). The quality of each study was evaluated by randomization sequence and allocation concealment. Publicaton bias or small study effect was checked by funnel plot with Egger's test. Results: Regarding the 1st and 2nd NBI, 21 RCTs including 7442 patients were selected. The 2nd generation NBI was used in three studies. Assuming that relative risk (RR) more than 1 meant significant efficacy of improving ADR compared to conventional white light. Pooled RR [95%CI] of NBI (both 1st and 2nd generation) was 1.03 [0.98-1.09] (I2=0%) (FEM). Pooled RR [95%CI] of the 1st generation NBI and the 2nd generation NBI was 1.20 [0.96-1.08] (I2=0%) (FEM) and 1.12 [0.94-1.33] (I2=61%) (REM), respectively. The validity of allocation sequence and allocation concealment were confirmed in more than half studies. There was no significant publication bias or small study effect. Conclusion: Our study did not show any evidence of significant efficacy of NBI (Both 1st and 2nd generation) for ADR. The 2nd generation NBI improved ADR compared to the 1st generation NBI, but it was not statistically significant. The 2nd generation NBI did not show any significant efficacy for ADR compared to conventional white light, either.

Adenoma Detection Through Deep Learning: 2017 Presidential Poster Award

Introduction: For every percentage increase in adenoma detection rate (ADR), risk of interval colorectal cancer is reduced by 3%. Adenoma prevalence is estimated to be 50%. Ideally, ADR should reflect adenoma prevalence, yet ADR varies between 5 and 55% among colonoscopist. We hypothesize that computer-assisted polyp identification could help bring low ADRs closer to true prevalence and reduce interval colorectal cancer incidence. Methods: Utilizing our colonoscopy quality database (Qulaoscopy, Docbot, Inc,), we applied the VGG-16 model as starting point (see [1]), which is a convolutional neural network (CNN) with 16 convolutional layers that was trained on natural images as part of the ImageNet challenge. We then “fine-tuned” this CNN on a set of 9000 screening colonoscopy images originating from all locations in the colorectum (and half of which contain polyps) to classify the presence or absence of polyps in images. The RGB images are scaled to a size of 224x224 pixels for this purpose and the CNN is trained using ten-fold crossvalidation, i.e. we repeat the training and testing ten times with different parts of the data for training/testing the model. Results: On the task of classifying polyp versus non-polyp containing images we obtain an accuracy of 96% and AUC (area under the ROC curve) of 0.99, which is an almost perfect score (the maximum attainable AUC is 1). At a processing rate of 170 images per second, the algorithm was easily applied to live video. Conclusion: Our deep learning methology achieved extremely high accuracy for polyp detection in images and is proven applicable to live video. We predict that application of this technology during colonoscopy will assist all colonoscopist to achieve ADRs that approach true adenoma prevalence.Figure: AUC (area under the ROC curve) for Polyp Detection.

Covering the Cover

Covering the Cover

Racial and Ethnic Disparities in Interval Colorectal Cancer Incidence: A Population-Based Cohort Study.

Interval colorectal cancer (CRC) accounts for 3% to 8% of all cases of CRC in the United States. Data on interval CRC by race/ethnicity are scant. To examine whether risk for interval CRC among Medicare patients differs by race/ethnicity and whether this potential variation is accounted for by differences in the quality of colonoscopy, as measured by physicians' polyp detection rate (PDR). Population-based cohort study. Medicare program. Patients aged 66 to 75 years who received colonoscopy between 2002 and 2011 and were followed through 2013. Kaplan-Meier curves and adjusted Cox models were used to estimate cumulative probabilities and hazard ratios (HRs) of interval CRC, defined as a CRC diagnosis 6 to 59 months after colonoscopy. There were 2735 cases of interval CRC identified over 235146 person-years of follow-up. A higher proportion of black persons (52.8%) than white persons (46.2%) received colonoscopy from physicians with a lower PDR. This rate was significantly associated with interval CRC risk. The probability of interval CRC by the end of follow-up was 7.1% in black persons and 5.8% in white persons. Compared with white persons, black persons had significantly higher risk for interval CRC (HR, 1.31 [95% CI, 1.13 to 1.51]); the disparity was more pronounced for cancer of the rectum (HR, 1.70 [CI, 1.25 to 2.31]) and distal colon (HR, 1.45 [CI, 1.00 to 2.11]) than for cancer of the proximal colon (HR, 1.17 [CI, 0.96 to 1.42]). Adjustment for PDR did not alter HRs by race/ethnicity, but differences between black persons and white persons were greater among physicians with higher PDRs. Colonoscopy and polypectomy were identified by using billing codes. Among elderly Medicare enrollees, the risk for interval CRC was higher in black persons than in white persons; the difference was more pronounced for cancer of the distal colon and rectum and for physicians with higher PDRs. American Cancer Society.