Faculty Bibliography

The performance of a CRC screening blood test was validated in a prospective, multicenter, observational study (PREEMPT CRC). The composition of the clinical study population can impact performance measures, potentially affecting the generalizability of the observed outcomes. We conducted a prespecified post-stratification adjustment analysis in which PREEMPT CRC performance values were adjusted to US Census age and sex distribution. The PREEMPT CRC evaluable cohort had a higher proportion of younger individuals and females than the census population. Compared to observed values, census adjustment demonstrated nominally higher CRC sensitivity (81.1% [95% confidence interval or CI, 71.3-88.1%] vs 79.2% [95% CI, 68.4-86.9%]) and advanced precancerous lesion sensitivity (13.7% [95% CI, 12.4-15.0%] vs 12.5% [95% CI, 11.3-13.8%]), with lower advanced colorectal neoplasia specificity (90.4% [95% CI, 90.0-90.7%) vs 91.5% [95% CI, 91.2-91.9%]). Negative and positive predictive values were consistent across age groups, highlighting consistent clinical interpretability of test results regardless of patient age.

BACKGROUND:Emerging blood tests may improve colorectal cancer (CRC) screening uptake and outcomes but are less sensitive for advanced precancerous lesions than some currently recommended tests. We examine whether these tests meet expectations for U.S. Preventive Services Task Force (USPSTF) recommendation. METHODS:A decision-analytic model that informed USPSTF was replicated and used to estimate the lifetime benefits (averted CRC cases & deaths, life-years gained [LYG]), burdens (required screening tests & colonoscopies), and harms (colonoscopy-related complications) for annual, biennial or triennial blood testing through age 45-75 years vs a benchmark of recommended and contemporary stool-based strategies, with colonoscopy screening as the reference. Base-case analyses assumed 100% adherence. Sensitivity analyses evaluated more realistic scenarios. RESULTS:Among benchmark strategies, colonoscopy screening had the most benefit, with an estimated 30 CRC deaths averted, 356 LYG, 4270 colonoscopies required and 15 complications per 1000 adults; stool-based strategies resulted in 81-88% of LYG for colonoscopy, 6829-19,476 screening tests, 1523-1880 colonoscopies, and 9-10 complications. By comparison, annual blood testing resulted in 85-87% of LYG for colonoscopy and an intermediate number of screenings, colonoscopies and complications. Biennial and triennial blood testing provided 57-72% of LYG for colonoscopy but resulted in net population benefit under plausible scenarios for increased utilization vs existing strategies. CONCLUSIONS:The estimated benefits, burdens and harms of annual blood testing are within the range of current CRC screening strategies. Biennial and triennial testing should also be considered for recommendation given potential for increased utilization and net population benefit.

BACKGROUND:Colorectal cancer (CRC) is the fourth most common and second deadliest cancer in the US. Screening is effective at reducing CRC incidence and mortality, but rates of screening remain suboptimal. There are no sensitive machine learning models for accurately identifying individuals at risk for colorectal cancer or precancerous polyps. OBJECTIVES/OBJECTIVE:The aim of our study was to develop and validate a novel machine learning model that uses multimodal Electronic Health Record (EHR) data, including the most recent complete blood count (CBC), basic metabolic panel (BMP), ICD codes, and medications, to estimate a patient's likelihood of having CRC or an advanced precancerous lesion. METHODS:We developed ColAI, an L1-regularized logistic regression model trained on 1-year trailing EHR data, to predict CRC or advanced adenoma at screening colonoscopy. Labs are treated as continuous variables, while ICD codes and medications are represented as binary indicators of presence. ColAI was trained using 87,825 screening colonoscopies and validated using 21,957 independent colonoscopies between August 1, 2020, and March 31, 2024, from the NYU Langone Health system. RESULTS:ColAI achieved an AUROC of 0.93 for CRC and 0.98 for CRC or advanced adenoma. Performance remained consistent across different hospitals and time periods within NYU Langone, demonstrating strong generalizability. Performance also remained consistent between first and follow-up colonoscopies, decreasing concern for selection bias. CONCLUSIONS:ColAI accurately identifies patients at elevated risk for CRC using only routine EHR data. It has the potential to enhance targeted outreach to high-risk, unscreened individuals and improve early cancer detection at the population level.

INTRODUCTION/BACKGROUND:Primary sclerosing cholangitis (PSC) is a chronic cholestatic liver disease frequently associated with inflammatory bowel disease (IBD). While both conditions independently increase cancer risk, the comparative burden of cancer in patients with coexisting IBD and PSC (IBD-PSC), isolated IBD, and isolated PSC remains inadequately defined. METHODS:We conducted a retrospective cohort study using the TriNetX nationwide electronic health records database. Patients with IBD-PSC were compared to individuals with isolated IBD and isolated PSC. Propensity score matching (PSM) was employed to balance key baseline characteristics across groups. Adjusted hazard ratios (aHRs) with 95% confidence intervals (CIs) were calculated for intestinal and extraintestinal malignancies across three pairwise comparisons: IBD-PSC vs isolated IBD, IBD-PSC vs isolated PSC, and isolated PSC vs isolated IBD. RESULTS:After matching, 4,187 patients were included in each of the IBD-PSC and isolated IBD groups. IBD-PSC was associated with increased risks of colorectal cancer (aHR 4.01, 95% CI: 2.79-5.75, p < 0.001), cholangiocarcinoma (aHR 27.54, 95% CI: 15.05-50.38, p < 0.001), liver cancer (aHR 13.41, 95% CI: 7.42-24.26, p < 0.001), pancreatic cancer (aHR 2.37, 95% CI: 1.18-4.76, p = 0.013), and gallbladder cancer (aHR 36.26, 95% CI: 4.94-266.23, p < 0.001). Compared to isolated PSC, IBD-PSC had higher risks of colorectal (aHR 5.72, 95% CI: 3.17-10.31, p < 0.001) and gallbladder cancer (aHR 4.14, 95% CI: 1.69-10.14, p = 0.001). CONCLUSION/CONCLUSIONS:IBD-PSC is associated with substantially elevated risks of both intestinal and extraintestinal malignancies compared to isolated IBD or PSC. These findings highlight the synergistic oncogenic potential of coexisting IBD and PSC and underscore the need for tailored surveillance and early detection strategies in this high-risk population.

INTRODUCTION/BACKGROUND:Colorectal cancer (CRC) screening rates among patients receiving care at multiple federally qualified health care centers (FQHCs) in New York city are low. Proactive outreach through mailed fecal immunochemical tests (FIT), reminders and navigation are evidence based interventions to improve CRC screening rates but remain untested in this study population. OBJECTIVE:To evaluate the effectiveness, implementation, and cost-effectiveness of a multilevel proactive outreach strategy to improve CRC screening rates among underserved adults in Brooklyn, New York. METHODS:This is a randomized controlled trial across five FQHCs serving predominantly Black and low-income populations. Adults aged 45-75 who are overdue for CRC screening are randomized to usual care or a multi-level proactive intervention. The intervention includes mailed education and FIT kits, patient navigation, and support for colonoscopy scheduling and follow-up. The primary outcome is CRC screening completion (FIT or colonoscopy) within six months. Secondary outcomes include colonoscopy follow-up after a positive FIT, implementation barriers and facilitators, and cost-effectiveness. RESULTS:A total of 1379 participants have been enrolled through May 2025. DISCUSSION/CONCLUSIONS:This trial addresses a critical gap in CRC prevention by testing a scalable, multilevel outreach model tailored to underserved populations. Findings will inform future strategies to enhance screening rates while reducing screening disparities through sustainable FQHC-based programs.

This study evaluated the effectiveness of large language models (LLMs) and vision-language models (VLMs) in gastroenterology. We used board-style multiple-choice questions to assess the performance of both proprietary and open-source LLMs and VLMs-including GPT, Claude, Gemini, Mistral, Llama, Mixtral, Phi, and Qwen, across different interfaces, computing environments, and levels of compression (quantization). Among the proprietary models, o1-preview (82.0%) and Claude3.5-Sonnet (74.0%) had the highest accuracy, outperforming the top open-source models: Llama3.3-70b (65.7%) and Qwen-2.5-72b (61.0%). Among the small quantized open-source models, the 8-bit Llama 3.2-11b (51.7%) and 6-bit Phi3-14b (48.7%) performed the best, with scores comparable to their full-precision counterparts. Notably, VLM accuracy on image-containing questions improved (~10%) when given human-generated captions, remained unchanged with original images, and declined with LLM-generated captions. Further research is warranted to evaluate model capabilities in real-world clinical decision-making scenarios.

INTRODUCTION/BACKGROUND:Colorectal cancer (CRC) is the fourth most common cancer in the USA and second leading cause of cancer deaths. While screening rates have increased and mortality rates have declined, disparities persist. This study investigates the screening rates and mortality correlation over 25 years. METHODS:We analyzed trends in age-adjusted CRC screening and mortality rates (AAMRs) for adults aged ≥ 50 using BRFSS and CDC WONDER databases, respectively. Correlation analysis between CRC screening rates and AAMRs and projected AAMRs at 100% screening rates were calculated using Jamovi and R software. RESULTS:CRC screening rates increased from 41.5% in 1999 to 76.3% in 2023. Non-Hispanic Whites recorded the highest rates (80.1%), while American Indians or Alaskan Natives (AI/AN) had a low screening rate of 48.65% in 2023. Non-insured individuals had a screening rate of 33.02%, while insured recorded 78.13% in 2023. AAMRs of CRC declined significantly over time, from 69.3% to 40.7% per 100,000 (1999-2024). AAMRs demonstrated a strong inverse correlation (- 0.885) with screening rates. Correlation analysis revealed stronger associations between screening and mortality for NH Whites and African Americans (AA) populations (- 0.824 and - 1.19, respectively). The projected AAMR at 100% screening was 18.91 (95% CI 17.92-19.91), versus 40.4 at 76.3% in 2023. CONCLUSION/CONCLUSIONS:CRC screening increased over the past 25 years, achieving 76.3% in 2023, correlating with decrease in AAMRs. Disparities persist across races and different socioeconomic groups. At 100% screening rates, projected AAMR is 18.919. Equity-focused interventions are needed to further increase CRC screening rates.

BACKGROUND:Infliximab (IFX) is commonly used in the management of acute severe ulcerative colitis (ASUC), yet up to 30% of individuals still require colectomy within 1 year. Clinical data characterizing these patients, however, are limited. AIMS/OBJECTIVE:We aimed to determine risk factors for colectomy among patients with ASUC who received in-hospital IFX treatment. METHODS:We performed a retrospective analysis of patients with ASUC who were treated with at least one dose of IFX while admitted between 2014 and 2022. Cox proportional hazards (PH) models were used to assess demographic, clinical, and laboratory risk factors for colectomy within 30 days and 1 year of IFX initiation. RESULTS:Overall, 36/170 (21.2%) patients underwent colectomy within 1 year of IFX initiation, with 22 (12.9%) individuals requiring colectomy within 30 days. On univariable analysis, concomitant Clostridioides difficile infection during admission, a ≤50% decrease in C-reactive protein (CRP) and experiencing 3 or more bowel movements per day within 48 hours after an initial IFX dose were significantly associated with 1-year colectomy. On multivariable Cox PH analysis, C. difficile infection during admission (aHR=2.92, 95% CI: 1.12-7.58) and a higher CRP/albumin ratio on admission (aHR=1.13, 95% CI: 1.01-1.27) were associated with increased colectomy risk within 1 year of IFX initiation. CONCLUSIONS:C. difficile infection and a higher CRP/albumin ratio on admission are associated with decreased time to colectomy within 1 year of IFX among patients presenting with ASUC. These factors may aid in early risk stratification to minimize delays in JAK-inhibitor initiation or surgical referral.

Primary Obesity Surgery Endoluminal 2 (POSE-2) is a minimally invasive endoscopic bariatric procedure. We conducted a systematic review and meta-analysis of one randomized controlled trial (RCT) and three observational studies (N = 210) to evaluate its efficacy, safety, and metabolic outcomes (HbA1c, glucose, cholesterol, triglycerides, LDL, and liver enzymes). Pooled percent total body weight loss (%TBWL) was 13.23% (I² = 87%) at 3 months, 16.22% (I² = 76%) at 6 months, and 16.17% (I² = 0%) at 12 months, showing high heterogeneity early but consistency by 12 months. Percent excess weight loss (%EWL) at 12 months was 56.95% (I² = 0%). HbA1c improved at 6 months (SMD = -0.67, p = 0.036, I² = 0%), exceeding the minimal clinically important difference (-0.5%), and cholesterol decreased significantly (SMD = -0.25, p = 0.013, I² = 0%). Fasting glucose and liver enzymes showed nonsignificant favorable trends with high heterogeneity for ALT/AST (I² > 90%). Adverse events were infrequent (2.5-5%), mostly mild, with rare perforations or bleeding. POSE-2 demonstrates promising weight loss and selected metabolic improvements in HbA1c and cholesterol with a favorable short-term safety profile, though evidence is limited to four studies with small sample size, short follow-up, and variable heterogeneity, highlighting the need for larger, longer RCTs.