Ahn, Joseph

INTRODUCTION: Cirrhosis is associated with cardiac dysfunction and distinct electrocardiogram (ECG) abnormalities. This study aimed to develop a proof-of-concept deep learning-based artificial intelligence (AI) model that could detect cirrhosis-related signals on ECG and generate an AI-Cirrhosis-ECG (ACE) score that would correlate with disease severity. METHODS: A review of Mayo Clinic's electronic health records identified 5,212 patients with advanced cirrhosis ≥18 years who underwent liver transplantation at the 3 Mayo Clinic transplant centers between 1988 and 2019. The patients were matched by age and sex in a 1:4 ratio to controls without liver disease and then divided into training, validation, and test sets using a 70%-10%-20% split. The primary outcome was the performance of the model in distinguishing patients with cirrhosis from controls using their ECGs. In addition, the association between the ACE score and the severity of patients' liver disease was assessed. RESULTS: The model's area under the curve in the test set was 0.908 with 84.9% sensitivity and 83.2% specificity, and this performance remained consistent after additional matching for medical comorbidities. Significant elevations in the ACE scores were seen with increasing model for end-stage liver disease-sodium score. Longitudinal trends in the ACE scores before and after liver transplantation mirrored the progression and resolution of liver disease. DISCUSSION: The ACE score, a deep learning model, can accurately discriminate ECGs from patients with and without cirrhosis. This novel relationship between AI-enabled ECG analysis and cirrhosis holds promise as the basis for future low-cost tools and applications in the care of patients with liver disease.

Physiologic aging leads to attrition of telomeres and replicative senescence. An acceleration of this process has been hypothesized in the progression of chronic liver disease. We sought to examine the association of telomere length (TL) with liver disease and its impact on mortality risk. A cohort of 7,072 adults with leukocyte TL measurements from the National Health and Nutrition Examination Survey 1999‐2002 with mortality follow‐up through 2015 was analyzed. Liver disease was defined by aminotransferase levels and classified into etiology‐based and advanced fibrosis categories. Multivariable‐adjusted linear regression models estimated effect sizes, with 95% confidence intervals (CIs), of the presence of liver disease on TL. Cox regression models evaluated associations between TL and all‐cause mortality risk using adjusted hazard ratios (HRs). The cohort was representative of the US population with mean age 46.1 years and mean TL 5.79 kilobase pairs. No overall association between TL and liver disease was found; however, there was a significant negative association of TL and advanced liver fibrosis in individuals aged 65 and above. The liver disease cohort (HR 1.22, 95% CI 0.99‐1.51) and those with metabolic syndrome (HR 1.26, 95% CI 0.96‐1.67) had increased mortality risk with shorter TL. The relationship between TL and all‐cause mortality was stronger in women (HR 1.51, 95% CI 1.02‐2.23) and in non‐Hispanic Whites (HR 1.37, 95% CI 1.02‐1.84). Conclusion: Shortened leukocyte TL is independently associated with advanced liver disease at older ages, and with a higher risk of all‐cause mortality in those with liver disease. These associations reaffirm the need to better understand the role of telomeres in the progression of liver disease.

Prognosis of hepatocellular carcinoma (HCC) could be affected by lack of or delayed therapy. We aimed to characterize the prevalence, correlates, and clinical impact of therapeutic underuse and delay in patients with HCC. Patients with HCC diagnosed between 2010 and 2017 were analyzed from the United States National Cancer Database. Logistic regression analysis identified factors associated with no and delayed (>90 days after diagnosis) HCC treatment. Cox proportional hazards regression with landmark analysis assessed the association between therapeutic delay and overall survival (OS), accounting for immortal time bias. Of 116,299 patients with HCC, 24.2% received no treatment and 18.4% of treated patients had delayed treatment. Older age, Black, Hispanic, lower socioeconomic status, earlier year of diagnosis, treatment at nonacademic centers, Northeast region, increased medical comorbidity, worse liver dysfunction, and higher tumor burden were associated with no treatment. Among treated patients, younger age, Hispanic, Black, treatment at academic centers, West region, earlier tumor stage, and receipt of noncurative treatment were associated with treatment delays. In multivariable Cox regression with a landmark of 150 days, patients with and without treatment delays had similar OS (adjusted hazard ratio [aHR], 1.01; 95% confidence interval [CI], 0.98‐1.04) with a median survival of 33.7 vs. 32.1 months, respectively. However, therapeutic delay was associated with worse OS in patients who had tumor, nodes, and metastases (TNM) stage 1 (aHR, 1.06; 95% CI, 1.01‐1.11) or received curative treatment (aHR, 1.12; 95% CI, 1.05‐1.18). Conclusion: One‐fourth of patients with HCC receive no therapy and one‐fifth of treated patients experience treatment delays. Both were associated with demographic, socioeconomic, and clinical characteristics of patients as well as facility type and region. The association between therapeutic delay and survival was stage and treatment dependent.

Introduction Despite extensive studies on the relationship between diabetes mellitus (DM) and depression, the associations of depressive symptom severity with prevalence, awareness, treatment, and control of diabetes remain unclear. We aimed to investigate changes in these outcomes of diabetes as depressive symptoms aggravate. Research design and methods We conducted a cross-sectional analysis of 14 328 participants in the 2011–2016 National Health and Nutrition Examination Survey. Participants were classified into depressive symptom groups of none, mild, moderate, moderately severe, and severe depending on their Patient Health Questionnaire-9 scores. Multivariate logistic regression analyses were conducted in three models adjusted for expanding confounders to evaluate the associations between severity of depressive symptoms and prevalence, awareness, treatment, and control of DM. Results As depressive symptom severity worsened, both prevalence and awareness of DM increased regardless of models (p value for trend <0.01 in all models for prevalence and awareness; adjusted OR (aOR) 2.14, 95% CI 1.29 to 3.56 for prevalence in the severe group, model 1; aOR 2.43, 95% CI 1.27 to 4.64 for awareness in the moderately severe group, model 1). Notwithstanding higher awareness of diabetes in the moderately severe and severe groups (84.5% and 86.2%, respectively, vs 71.3% in the none group), these groups were treated less (aOR 0.25, 95% CI 0.11 to 0.55 in the severe group, model 3) or inadequately controlled (aOR 0.51, 95% CI 0.27 to 0.98 in the moderately severe group, model 3). Conclusions The gap between patients’ higher awareness and lower treatment rate or control of diabetes among individuals with severe depressive symptoms highlights the unmet needs for postdiagnostic multidisciplinary care for patients with comorbid depression and DM.

Background and Aims Patients with cirrhosis on the liver transplant (LT) waiting list may die or be removed because of complications of portal hypertension (PH) or infections. von Willebrand factor antigen (vWF‐Ag) and C‐reactive protein (CRP) are simple, broadly available markers of these processes. Approach and Results We determined whether addition of vWF‐Ag and CRP to the Model for End‐Stage Liver Disease‐Sodium (MELD‐Na) score improves risk stratification of patients awaiting LT. CRP and vWF‐Ag at LT listing were assessed in two independent cohorts (Medical University of Vienna [exploration cohort] and Mayo Clinic Rochester [validation cohort]). Clinical characteristics, MELD‐Na, and mortality on the waiting list were recorded. Prediction of 3‐month waiting list mortality was assessed by receiver operating characteristics curve (ROC‐AUC). In order to explore potential mechanisms underlying the prognostic utility of vWF‐Ag and CRP in this setting, we evaluated their association with PH, bacterial translocation, systemic inflammation, and circulatory dysfunction. In the exploration cohort (n = 269) vWF‐Ag and CRP both improved the predictive value of MELD‐Na for 3‐month waitlist mortality and showed the highest predictive value when combined (AUC: MELD‐Na, 0.764; MELD‐Na + CRP, 0.790; MELD‐Na + vWF, 0.803; MELD‐Na + CRP + vWF‐Ag, 0.824). Results were confirmed in an independent validation cohort (n = 129; AUC: MELD‐Na, 0.677; MELD‐Na + CRP + vWF‐Ag, 0.882). vWF‐Ag was independently associated with PH and inflammatory biomarkers, whereas CRP closely, and MELD independently, correlated with biomarkers of bacterial translocation/inflammation. Conclusions The addition of vWF‐Ag and CRP—reflecting central pathophysiological mechanisms of PH, bacterial translocation, and inflammation, that are all drivers of mortality on the waiting list for LT—to the MELD‐Na score improves prediction of waitlist mortality. Using the vWFAg‐CRP‐MELD‐Na model for prioritizing organ allocation may improve prediction of waitlist mortality and decrease waitlist mortality.

Modern medical care produces large volumes of multimodal patient data, which many clinicians struggle to process and synthesize into actionable knowledge. In recent years, artificial intelligence (AI) has emerged as an effective tool in this regard. The field of hepatology is no exception, with a growing number of studies published that apply AI techniques to the diagnosis and treatment of liver diseases. These have included machine‐learning algorithms (such as regression models, Bayesian networks, and support vector machines) to predict disease progression, the presence of complications, and mortality; deep‐learning algorithms to enable rapid, automated interpretation of radiologic and pathologic images; and natural‐language processing to extract clinically meaningful concepts from vast quantities of unstructured data in electronic health records. This review article will provide a comprehensive overview of hepatology‐focused AI research, discuss some of the barriers to clinical implementation and adoption, and suggest future directions for the field.

Abstract Background and Aims: For hepatocellular carcinoma (HCC) with vascular invasion, systemic treatment is recommended but transarterial radioembolization (TARE) is often used in clinical practice. The aim of this study was to investigate the current usage of TARE vs. systemic therapy as a first-line treatment for HCC with vascular invasion in the United States (US). Method: We performed a retrospective cohort study using the National Cancer Database (NCDB), which represents more than 70 percent of newly diagnosed cancer cases nationwide. All HCC patients diagnosed between 2010 and 2015 with the tumor stage of T3BN0M0 according to the 7th edition of the American Joint Committee on Cancer TNM staging system (tumor involving a major branch of a large vein of the liver without lymph nodes or extrahepatic metastasis) and received TARE or systemic treatment as a first-line treatment were included. Multivariable logistic regression was used to estimate factors associated with the utilization of TARE vs. systemic therapy. Results: A total of 1979 patients who received systemic treatment and 524 patients who received TARE were included. The proportion of patients receiving TARE increased from 15.7% in 2010 to 31.4% in 2015 (P&lt;0.01). The median tumor size of the largest lesion was comparable (7.0cm for TARE vs. 7.1 cm for systemic treatment, P=0.43). Factors independently associated with increased likelihood of receiving TARE over systemic treatment were: advanced age (Odds Ratio [OR]: 1.02, 95% confidence interval [CI]: 1.01 - 1.03, p=0.005), insurance status: private (OR: 3.06, 95% CI: 1.61 - 5.65, p&lt;0.001), Medicaid/Medicare (OR: 2.91, 95% CI: 1.52 - 5.31, p&lt;0.001), others (OR: 4.85, 95% CI: 1.90 - 11.62, p&lt;0.001) vs. no insurance, and receiving care in the Midwest vs. Northeast region of the US (OR: 1.89, 95% CI: 1.38 - 2.53, p&lt;0.001). Patients were significantly less likely to receive TARE if receiving treatment at community cancer program (OR: 0.24, 95% CI: 0.098 - 0.49, p&lt;0.001) or comprehensive community cancer program (OR: 0.62, 95% CI: 0.48 - 0.82, p&lt;0.001) vs. academic cancer program. Larger tumor size (OR per 1cm: 0.97, 95% CI: 0.95 - 1.00, p=0.03) and higher Charlson comorbidity index of ≥3 vs. 0 (OR: 0.53, 95% CI: 0.38 - 0.73, p&lt;0.001) were inversely associated with receiving TARE. Recent HCC diagnosis year was associated with an increased likelihood of receiving TARE (OR: 1.23, 95% CI:1.16 - 1.32, p&lt;0.001). Conclusion: The proportion of patients receiving TARE as a first-line treatment for HCC with macrovascular invasion has been increasing in the US. Age, comorbidities, insurance status, tumor size, and the type and location of cancer centers are associated with the usage of TARE vs. systemic treatment in HCC patients with vascular invasion. Citation Format: Joseph C. Ahn, Michael Luu, Walid Ayoub, Alexander Kuo, Andrew Hendifar, Jun Gong, Kambiz Kosari, Shelly Lu, Nicholas Nissen, Ju Dong Yang. Transarterial radioembolization versus systemic treatment for the management of hepatocellular carcinoma with vascular invasion: Analysis of the US National Cancer Database [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2350.