Polyp recurrence is common after endoscopic mucosal resection (EMR) of non-pedunculated colonic polyps ≥ 20 mm. Two models haven been published for polyp recurrence prediction: Sydney EMR recurrence tool (SERT) and the size, morphology, colonic site, and access to target (SMSA) score. None of these models have been evaluated in a real-world United States (U.S.) cohort. We aimed to evaluate the external validity of these two models and develop a new model. Retrospective cohort study of patients with non-pedunculated polyps ≥ 20 mm that underwent EMR between 1/1/2012 and 6/30/2020. Univariate and multivariate analysis were performed to identify predictors of polyp recurrence to build a new model. Receiver Operating Characteristic (ROC) curves for the new model, SERT and a modified version of SMSA were derived and compared. A total of 461 polyps from 461 unique patients were included for analysis. The average polyp size was 29.1 ± 12.4 mm. Recurrence rate at first or second surveillance colonoscopy was 29.0

Introduction: Identifying dysplasia in the setting of chronic colitis requires manual review of unstructured pathology reports, which may vary in terminology or description of dysplasia. Natural Language Processing (NLP) technologies are used to extract data from free text format in the electronic medical record (EMR). We aimed to develop and validate an NLP based algorithm to identify presence of dysplasia in the setting of chronic colitis from pathology reports within an integrated EMR system. Methods: We developed an unsupervised, rule-based regular expression NLP algorithm to identify “dysplasia” with “chronic colitis” and their corresponding location alongside a list of negation terms within pathology reports derived from the EMR (EPIC) at a large quaternary care medical center in Houston, Texas. The algorithm’s performance was evaluated in comparison to authors NSL, SB, and MK's interpretation of the contents of the same pathology reports. A portion of the pathology reports were reviewed by multiple authors to ensure adequate intra-observer agreement. The algorithm's performance was calculated as accuracy, sensitivity, precision and F- measure. Results: We queried 9508 pathology reports and identified 480 patients with chronic colitis, of whom 48 had dysplasia on colonic biopsies. The NLP algorithm identified dysplasia with 97.5% accuracy, 89.5% sensitivity, 86% precision and an F-measure of 93.7% when compared with manual review. The NLP algorithm was able to identify the location of dysplasia with 93% accuracy, 87.9% precision and an F-measure of 78%. Conclusion: Unsupervised NLP approach identified the presence and location of dysplasia in the setting of chronic colitis with high degree of accuracy from pathology reports. We expect our algorithm’s performance to improve with the utilization of training sets. Application of this algorithm has the potential to improve patient identification to enhance research and clinical care across large EMRs (Table 1). Table 1. - Performance characteristics of the NLP algorithm for detection of colonic dysplasia and dysplasia location in pathology reports Measure Accuracy Sensitivity Specificity Precision F-Measure Dysplasia 97.5 89.5 98.4 86.0 93.7 Location 93.1 64.5 98.3 87.9 77.9

Background: Clinical guidelines reserve endoscopic surveillance after a gastric intestinal metaplasia (GIM) diagnosis for high-risk patients. However, it is unclear how closely guidelines are followed in clinical practice. We examined the effectiveness of a standardized protocol for the management of GIM among gastroenterologists at a US hospital. Methods: This was a preintervention and postintervention study, which included developing a protocol and education of gastroenterologists on GIM management. For the preintervention study, 50 patients with GIM were randomly selected from a histopathology database at the Houston VA Hospital between January 2016 and December 2019. For the postintervention study, we assessed change in GIM management in a cohort of 50 patients with GIM between April 2020 and January 2021 and surveyed 10 gastroenterologists. The durability of the intervention was assessed in a cohort of 50 GIM patients diagnosed between April 2021 and July 2021. Results: In the preintervention cohort, GIM location was specified (antrum and corpus separated) in 11 patients (22%), and Helicobacter pylori testing was recommended in 11 of 26 patients (42%) without previous testing. Gastric mapping biopsies were recommended in 14% and surveillance endoscopy in 2%. In the postintervention cohort, gastric biopsy location was specified in 45 patients (90%, P <0.001) and H. pylori testing was recommended in 26 of 27 patients without prior testing (96%, P <0.001). Because gastric biopsy location was known in 90% of patients ( P <0.001), gastric mapping was not necessary, and surveillance endoscopy was recommended in 42% ( P <0.001). One year after the intervention, all metrics remained elevated compared with the preintervention cohort. Conclusions: GIM management guidelines are not consistently followed. A protocol for GIM management and education of gastroenterologists increased adherence to H. pylori testing and GIM surveillance recommendations.

Background Endoscopic mucosal resection (EMR) is an effective method for removing non-pedunculated polyps >= 20 mm. We aimed to examine changes in EMR techniques over a 9-year period and evaluate frequency of histologic-confirmed recurrence. Methods We identified patients who underwent EMR of non-pedunculated polyps >= 20 mm at a safety net and the Veteran's Affairs (VA) hospital in Houston, Texas between 2012 and 2020. Odds ratios (ORs) and 95% confidence intervals (CI) for associations with recurrence risk were estimated using multivariable logistic regression. Results 461 unique patients were included. The histologic-confirmed recurrence was 29.0% at 15.6 months median follow up (IQR 12.3 - 17.4). Polyps removed between 2018 and 2020 had a 0.43 decreased odds of recurrence vs. polyps removed between 2012 and 2014. The use of viscous lifting agents increased over time (from 0 to 54%), and the use of saline was associated with increased risk of recurrence (OR 2.28 [CI 1.33 - 3.31]). Conclusions Histologic-confirmed recurrence after EMR for non-pedunculated polyps >= 20 mm decreased over the seven year-period. Saline was associated with a higher risk of recurrence and the use of more viscous agents increased over time.

Introduction: Ascertaining disease location in inflammatory bowel disease requires manual review of unstructured pathology reports, which may vary in style and terminology. Natural Language Processing (NLP) technologies are used to extract data from free text format in the electronic medical record (EMR). We aimed to develop and validate an unsupervised NLP based algorithm to identify presence of ileal and/or colonic inflammation as well as differentiate acute from chronic inflammation from pathology reports within an integrated EMR system. Methods: We developed an unsupervised, rule-based regular expression NLP algorithm to identify keywords corresponding to the findings of acute or chronic ‘ileitis’, ‘colitis’, ‘crypt architectural distortion’ and ‘granulomas’ alongside a list of negation terms within pathology reports. The algorithm’s performance was evaluated in comparison to authors NSL, SB, and MK's interpretation of the contents of the same pathology reports. A portion of the pathology reports were reviewed by multiple authors to ensure adequate intra-observer agreement. The algorithm’s performance was calculated as accuracy, sensitivity, precision, and F-measure. Results: We queried 9508 pathology reports spanning a 36-month period and identified 649 reports with findings of acute or chronic inflammation on colonoscopy. The NLP algorithm demonstrated high accuracy in detecting acute colitis (93.5%), chronic colitis (80.2%), acute ileitis (96.4%), chronic ileitis (86.4%) and the presence of granulomas (98.7%) compared to manual review of pathology reports. Detailed performance across the variables studies is in Table 1. Conclusion: Unsupervised NLP approach identified the location and chronicity of inflammation from biopsies with high degree of accuracy. We expect our algorithm’s performance to improve further with the utilization of training sets with expert input. Application of this algorithm has the potential to improve patient identification to enhance research and clinical care across large EMRs. Table 1. - Performance characteristics of the NLP algorithm for detecting acute and chronic colitis, acute and chronic ileitis, and granulomas from pathology reports Variable (N) Accuracy (%) Sensitivity (%) Specificity (%) Precision (%) F-Measure Colitis Acute (n=31) 93.5 12.9 99.3 57.1 22.8 Chronic (n=411) 80.2 79.6 84.1 96.7 81.8 Ileitis Acute (n=19) 96.4 15.8 99.8 75 27.3 Chronic (n=137) 86.4 52.6 100 100 68.9 Granulomas (n=51) 98.7 90.2 98.1 85.2 94

A 45-year-old woman with common variable immunodeficiency and interstitial lung disease on weekly intravenous immune globulin therapy and mycophenolate mofetil presents to the gastroenterology clinic following a year of worsening diarrhea. She reports 5 to 6 nonbloody bowel movements a day with 10-lb weight loss. On colonoscopy multiple 8- to 10-mm mucosal nodules were noted in the rectum (Figure A) surrounded by patchy areas of flattened carpet-like erythematous mucosal lesions (Figure B). Biopsies of the rectal nodules and mucosa are detailed at ×200 magnification pathology slides in Figures C and D through a hematoxylin-eosin and von Kossa stain, respectively.

Introduction: Colorectal cancer is the third cause of cancer-related mortality in the United States. Non-pedunculated polyps ≥20 mm have increased risk of recurrence; hence, surveillance colonoscopy is recommended at 6 months. Unfortunately, patient follow up after resection of these high-risk polyps is not perfect. We aim to study the characteristics of patients who were lost in follow up after resection of non-pedunculated polyps ≥20 mm at a safety-net and Veteran’s Affairs (VA) hospitals, with the purpose of developing qualitive improvement interventions targeting these patients. Methods: Retrospective study of patients who undergone endoscopic mucosal resection of non-pedunculated polyps larger ≥20 mm at a safety-net and VA hospital in Houston, Texas from 2012 to 2019. Patients who had two stage procedures, non-neoplastic histology or evidence of invasive cancer were excluded. Patients were divided by competition of at least 1 surveillance colonoscopy. Demographic data, polyp and resection characteristics were analyzed. Distance to hospital and income was calculated based on zip code. Univariate and multivariate analyzed were performed. Results: A total of 522 non-pedunculated polyps were included from 468 unique patients. 328 lesions completed at least one surveillance colonoscopy and 194 did not follow up. Patients who did not follow up were more likely to be younger male, uninsured, and had longer driving time and distance to the hospital. Ethnicity, need of English interpreter and average income was comparable between groups. Polyps larger in size, in the right colon, removed by piecemeal resection and with high grade dysplasia were more likely to have at least 1 surveillance colonoscopy (Table). After multivariable analysis, male gender (OR 2.88 [1.52-5.47] P=< 0.001) was significantly associated with increased risk of no follow up. Conclusion: Younger males without insurance and with longer driving time and distance to the hospital are more likely to not have surveillance colonoscopy after resection of high-risk colonic polyps. Understanding the patient population who is lost in follow up is key for the implementation of targeted quality improvement interventions.Table 1.: Patient characteristics by type of follow up after index colonoscopy.

INTRODUCTION: Nonalcoholic Fatty Liver Disease (NAFLD) can culminate in liver cirrhosis with the potential complication of portal hypertension. Among patients with cirrhosis, prognosis of their condition has been demonstrated through the use of the Model for End-Stage Liver Disease (MELD)-Na. Other methods of assessing a patient with cirrhosis’s prognosis is through measuring their hepatic venous pressure gradient (HVPG). The HVPG has been shown in previous studies to correlate with a patient’s MELD-Na score. The aim of this study was to determine the relationship between non-invasive measures of fibrosis with invasive HVPG measurements in a NAFLD-cirrhotic population. The MELD-Na scores were then calculated to assess their correlation with these noninvasive and invasive measures of clinical prognosis. METHODS: A retrospective study was done on a cohort of NAFLD-cirrhosis patients seen at Liver Associates of Texas in Houston, TX from 1/18 – 5/20. Patients’ portal hypertension was assessed using the HVPG calculated from transjugular pressure measurements. Fibrosis measurements were measured through liver stiffness readings through transient elastography. A multivariate linear regression was performed using fibrosis on elastography as the outcome variable to assess factors independently associated with fibrosis progression: gender, ethnicity, HPVG measurements, AST, ALT, and MELD-Na. Then a linear regression was performed between the MELD-Na and kilopascal (kPa) score to assess the correlation between these two values. RESULTS: 99 patients were included in the study, 38 had transjugular liver biopsies, and 49 had fibrosis measurements. 69% were females. 43% were Caucasian, 5% African American, 20% Hispanic, 4% Asian, and 27% were other. 29.05 (± 17.81) and 10.20 (±4.3) were the mean kPa and MELD-Na scores respectively. No significant finding was seen between HVPG and kPa measurements. Elastography and MELD-Na score were significantly correlated (P = 0.03) as well as elastography with AST (P = 0.034). CONCLUSION: This study demonstrated a significant correlation between elastography and the MELD-Na score in patients with NAFLD-cirrhosis. AST was also identified as a possible surrogate marker of fibrosis while HVPG was not. This gives more credence to the idea that NAFLD patients with cirrhosis could be better prognosticated through noninvasive studies such as the use of elastography or serum AST levels rather than more invasive methods such as HVPG measurements.Table 1.: Linear Regression on Selected Variables Compared to Fibrosis Measured on Elastography

INTRODUCTION: Hereditary hemochromatosis is an autosomal recessive genetic disorder characterized by an iron overload that affects multiple organs. The lifetime incidence of cirrhosis is estimated to be 10% among patients with hereditary hemochromatosis. The aim of this study is to determine host factors that could impact the natural history of fibrosis in patients with hereditary hemochromatosis. METHODS: A cross-sectional study was conducted at Liver Associates of Texas Hepatology clinics in Houston, Texas on a sample of patients initially diagnosed with hereditary hemochromatosis between January 2018 and June 2020. The diagnosis was made by a combination of genetic testing, lab measurements, imaging, and/or liver biopsy. Fibrosis stage was determined either by transient elastography or liver biopsy. A multivariate linear regression analysis was performed using fibrosis stage as the outcome variable. Initial serum ferritin levels, liver enzymes, age, body mass index (BMI), alcohol consumption, diabetes, and presence of fatty liver disease were assessed as factors independently associated with fibrosis stage at the time of diagnosis. All statistical tests were two-sided, and a value of P < 0.05 was considered to indicate statistical significance. RESULTS: Primary analysis included 16 patients with fibrosis measurements. The median age was 62 years old (± 16 years). 5 (31%) patients were male. The median initial ferritin level was 426.5 ng/mL (± 1221 ng/mL). 5 (31%) patients had diabetes mellitus, 6 (38%) had fatty liver disease, and the median BMI was 27 kg/m2 (± 10 kg/m2). A multivariate linear regression analysis revealed that higher initial ferritin levels and advanced age were associated with more advanced fibrosis stage (β = 0.0006, P = 0.041, CI: 0.000032–0.0012) and (β = 0.1112, P = 0.006, CI: 4.01–0.18), respectively. There was no correlation with advanced fibrosis in patients with diabetes, fatty liver disease, or high BMI. CONCLUSION: Advanced age and higher levels of initial ferritin are associated with the progression of fibrosis in hereditary hemochromatosis. This supports the use of initial ferritin level and older age as indirect markers of fibrosis in patients diagnosed with hereditary hemochromatosis. Further studies will be needed to determine other predictors of fibrosis progression and the underlying mechanisms that may impact the natural history of hemochromatosis.Figure 1.: Predicted Effect of Initial Ferritin on Fibrosis Level.Table 1.: Multivariate Linear Regression Analysis

Introduction: Endoscopic mucosal resection (EMR) technique is the most common approach for non-pedunculated colonic polyps ≥20 mm. EMR is usually performed by both general gastroenterologists and advance endoscopists. EMR performed by advanced endoscopists have been associated with decreased risk of incomplete polyp resection, but little is known about the recurrence rate of large non-pedunculated polyps performed by general gastroenterologists. Thus, we aim to investigate recurrence rate and EMR characteristics for non-pedunculated polyps ≥20 mm performed by general gastroenterologists. Methods: Retrospective study of patients who undergone EMR of non-pedunculated polyps ≥20 mm at Ben Taub and Veteran’s Affairs (VA) hospital in Houston, Texas from 2012 to 2019. Patients who had two stage procedures, non-neoplastic histology or evidence of invasive cancer were excluded. Patients were divided by type of endoscopist who performed index colonoscopy. Demographic data, polyp and resection characteristics were analyzed. Univariate and multivariate analyzed were performed. Results: A total of 522 non-pedunculated polyps were included from 470 unique patients. 271 polyps had EMR performed by general gastroenterologists and 199 polyps by advance endoscopists. 177 lesions in the general gastroenterologist group completed at least 1 surveillance colonoscopy, 49 (27.7%) had histologic recurrence. 150 lesions in the advance endoscopists group completed at least 1 surveillance colonoscopy, 46 (30.7%) had histologic recurrence. The average polyp size of the cohort was 28.6 mm. Polyps from male patients (85.6% vs 74.4%, P = 0.002) at the VA hospital (82.7% vs 55.3%, P ≤ 0.001) were more likely to be resected by general gastroenterologists. Other demographic characteristics were comparable among groups. There were no significant differences between groups in polyp size, histological recurrence, delayed post-polypectomy bleeding and/or post procedural pain prompting abdominal imaging. Advance endoscopists were more likely to use clips for prophylactic defect closure and less likely to have intraprocedural bleeding (Table 1). Conclusion: General gastroenterologists had a similar recurrence rate after EMR of non-pedunculated polyps ≥20 mm compared with advanced endoscopists. Advance endoscopists had less intraprocedural bleeding but delayed post-polypectomy bleeding rate was comparable.Table 1.: Univariable and multivariable logistic regression analysis of polyp resection characteristics

INTRODUCTION: Non-alcoholic Fatty Liver Disease (NALFD) is regarded to be the result of the metabolic syndrome and has become an increasing concern with the rise in obesity and diabetes especially in the primary care setting. A subset of NAFLD patients with non-alcoholic steatohepatitis (NASH), can progress to advanced fibrosis, cirrhosis, and hepatocellular carcinoma. NAFLD is often under-diagnosed due to its non-specific symptoms, allowing its silent progression. Our aim is to determine which clinical parameters in a well-defined NAFLD population in a primary care environment that could be of utility to detect patients with NAFLD who are at high risk for progressive disease. This will enable early detection, monitoring, and aggressive management in this subset of NAFLD patients. METHODS: A cohort study was performed at Liver Associates of Texas Hepatology for all patients diagnosed with NAFLD from 9/2018-5/2020. Patients with history of significant alcohol use, HCV/HBV infections, and other chronic liver disease were excluded. Steatosis and fibrosis were assessed using transient elastography (Fibroscan). Demographic data, body mass index (BMI), AST, ALT, and platelet values were recorded. A multivariate linear regression model was used to determine the correlations of age, BMI, ethnicity, ALT, AST, and platelet counts on fibrosis. All statistical tests were two sided and a value of < 0.05 was considered to indicate statistical significance. RESULTS: 154 patients were identified with NAFLD. Patients were 54.96 ± 13.42 years old, 34% male, 42% white, 6% African American, 14% Hispanic, 6% Asian/pacific Islander, and 32% unspecified. Mean BMI and liver stiffness measurements were 33.4 kg/m2 and 11.95 kPa respectively. Using a multivariate linear regression model, BMI, AST, ALT, and platelet counts were determined to significantly correlate with liver stiffness and fibrosis stage. As Table 1 shows: liver stiffness positively correlated with higher BMI (P < 0.001), increasing AST (P < 0.001), lower levels of platelets (P < 0.001), and decreasing ALT (P = 0.003). CONCLUSION: In NASH patients without other confounding liver-related diseases, we demonstrate that BMI, AST, ALT, and platelet counts are significant predictors of fibrosis. In patients with higher BMI, elevated AST, lower ALT, or decreased platelet counts, evaluation and aggressive management of patients with NASH should be considered. Other factors need to be explored in future studies.TABLE 1.: Multivariate Regression Analysis Summary for Liver Stiffness (in kPa)

INTRODUCTION: Non-alcoholic fatty liver disease (NAFLD) is one of the major chronic liver diseases impacting the world today. NAFLD can silently progress to fibrosis and lead to cirrhosis and hepatocellular carcinoma. Several non-invasive models have been developed using blood biomarkers to assess progression of fibrosis in patients with NAFLD. The aim of this study was to compare four of the most commonly used biomarker models in patients diagnosed with NAFLD to identify which of these models could best assess liver fibrosis in diabetic versus nondiabetic patients diagnosed with NAFLD. METHODS: This retrospective cohort study was performed at the Liver Associates of Texas Hepatology outpatient clinics in Houston, Texas from 9/2018 to 5/2020. Patients with other liver pathologies such as Hepatitis B, C, alcoholic liver disease, or hepatocellular cancer were excluded. Patients’ fibrosis levels were measured through two different modalities: lab serology and transient elastography through Fibroscan®. Using serology results, aspartate aminotransferase to platelet ratio index (APRI), fibrosis-4 index (FIB-4), BARD, and NAFLD fibrosis score (NFS) were calculated and compared to kilopascals (kPa) measurements on elastography. A multivariate linear regression was used to compare the four models and assess their correlation with the kPa measurements in patients with and without diabetes. RESULTS: One hundred seventy patients were identified for this study. The mean age of patients was 55 years, 112 (66%) were female, and 79 (46%) had diabetes. APRI, BARD, and NFS scores significantly correlated with fibrosis level for all patients studied. When isolating diabetic patients, NFS was the only score that significantly correlated with advanced fibrosis level (68% sensitivity, 80% specificity). The NFS was not significantly correlated with advanced fibrosis in the non-diabetic group (18% sensitivity, 98% specificity). APRI and FIB-4 scores were the only scores that were significantly correlated with advanced fibrosis in non-diabetic patients. CONCLUSION: In NAFLD patients without other liver pathologies, APRI, BARD, FIB-4, and NFS models were all identified to be the indirect measures of fibrosis. The NFS demonstrated greater sensitivity of advanced fibrosis in diabetic patients than in nondiabetic patients. APRI score was more sensitive in nondiabetic patients compared to diabetics. This study emphasizes the relative strengths of the different fibrosis serology models in particular populations and warrants further review.Table 1.: Multivariate Regression of Indirect Fibrosis ModelsTable 2.: Epidemiological Tests Comparing APRI & NFS Scores for Diabetic & Nondiabetic Patients