Topazian, Mark D

ABSTRACT Background and Aims It remains unclear whether different factors influence 30‐ and 90‐day readmission rates. We identified predictors of 90‐day mortality in a global cohort of hospitalised patients with cirrhosis. Methods Variables associated with 30‐ and 90‐day post‐discharge readmission, liver transplantation (LT), and mortality were compared in prospectively enrolled, non‐electively hospitalised adult patients with cirrhosis. Results We enrolled 4208 patients from 125 centres across 37 countries. Alcohol‐related liver disease was the most common aetiology of cirrhosis (40%), followed by hepatitis B virus infection (21%); 47% were readmitted, 8% received liver transplantation (LT), and 23% died within 90 days post‐discharge. In multivariable analysis, country income was significantly associated with both 30‐ and 90‐day readmission rates, the receipt of LT, and mortality within 30 and 90 days post‐discharge (all p ‐values < 0.001). Independent predictors of higher rates of 30‐ and 90‐day readmissions, respectively, were disease severity ( p < 0.001, p = 0.003), prior ascites ( p = 0.025, p = 0.016), prior hospitalisation ( p < 0.001), hyponatremia ( p < 0.001, p = 0.008), and in‐hospital vasopressor use ( p = 0.002, p = 0.011). Prior variceal bleeding independently predicted higher 90‐day readmission rates, whereas age, disease severity, mechanical ventilation, and in‐hospital vasopressor use (all, p < 0.001) increased odds of 30‐ and 90‐day post‐discharge death. Additionally, admission ( p = 0.011) and nosocomial ( p = 0.022) infections increased the odds of 90‐day post‐discharge death. Conclusion In a global cohort of hospitalised patients with cirrhosis, patients in high‐income countries exhibited the highest rates of 30‐ and 90‐day readmission, alongside lower mortality rates and a higher incidence of LT. Variables that predict 30‐day readmission also predicted 90‐day readmission. Understanding these disparities is essential for achieving health equity.

INTRODUCTION: Nosocomial infections (NIs) in cirrhosis are associated with high mortality but could be preventable. Logistic regression (LR) models have failed to identify high-risk patients. We aimed to develop machine learning (ML) models to predict NI. METHODS: The CLEARED consortium consists of prospectively enrolled cirrhosis inpatients from >120 centers. Using day-of-admission clinical data, 3 ML approaches (random forest [RF], extreme gradient boosting, and neural networks [NNs]) were used to predict NI. Data were split 80:20 for training and testing stratified by the outcome. Models were compared using area under the receiver operating characteristic curve (AUC). RESULTS: In total, 8,263 patients (55.90 ± 13.34 years; 64.1% men) from 127 centers in 37 countries were included. NI developed in 869 (10.5%), a median of 6 (4–11) days of postadmission. Major NIs were respiratory (29.6%) and urinary tract infection (15.7%), and spontaneous bacterial peritonitis (13.5%). NIs occurred more frequently in patients from low/low-middle income countries and those with severe liver disease, alcohol etiology, and admission infections. NIs were associated with inpatient mortality (31.9% vs 8.1%, P < 0.001) and liver transplantation (4.8% vs 1.9%, P < 0.001). Although the RF model (AUC 0.69) showed good calibration (Brier score 0.09), outperforming extreme gradient boosting, neural network, and LR models (AUC 0.66 for all; LR comparison P = 0.043), no model achieved AUC ≥0.80 for clinical utility. At 10% predicted probability threshold, the RF model demonstrated only 75.4% sensitivity, 52.9% specificity, and 15.9% positive predictive value (PPV). DISCUSSION: NIs cannot be accurately predicted from day-of-admission data using ML models, even in a large, prospective, global cirrhosis cohort. Every hospitalized patient with cirrhosis should receive protocolized infection control measures.

PURPOSE Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related deaths worldwide. The higher incidence and earlier onset of HCC in sub-Saharan Africa suggest a potential role for a genetic predisposition. This study evaluated the association of TERT -rs2242652-(A), a variant linked to lower HCC risk, with HCC in populations from Ghana, Nigeria, and Cameroon, compared with a population from the United States. METHODS The study included 537 patients with HCC: United States (n = 348), Ghana (n = 79), Nigeria (n = 43), and Cameroon (n = 67). The control group had 2,872 cancer-free individuals: United States (n = 2,399), Ghana (n = 323), Nigeria (n = 85), and Cameroon (n = 65). Whole-exome sequencing was conducted using germline DNA, and data for TERT -rs2242652 were analyzed. Odds ratios (ORs) and 95% CIs were calculated using unconditional logistic regression. Chi-square tests assessed protective allele frequencies. RESULTS In the US cohort, TERT -rs2242652 was significantly associated with lower HCC risk (OR, 0.75 [95% CI, 0.58 to 0.96]; P = .02), with an allele frequency of 18.8% (15.4% in HCC cases v 19.3% in controls). In the combined sub-Saharan African population, no significant association was observed, but there was a trend toward decreased HCC risk (OR, 0.80 [95% CI, 0.53 to 1.22]; P = .29), with an allele frequency of 12.2% (10.6% in HCC cases v 12.9% in controls). Separate analyses of Ghanaian, Nigerian, and Cameroonian populations showed similar nonsignificant trends. The protective allele frequency in the combined African populations was significantly lower than in the US cohort ( P < .0001). CONCLUSION In sub-Saharan African populations, there was a lower frequency of the HCC protective allele TERT -rs2242652 compared with European Americans. These findings underscore the importance of multiethnic genetic studies in understanding population differences in HCC risk and developing prevention strategies.

Abstract Background: Discovery of methylated DNA markers (MDM) of esophageal squamous cell carcinoma (ESCC) has sparked interest in assessing these markers in tissue. We evaluated MDMs in ESCC from three geographically and ethnically distinct populations, and explored the feasibility of assaying MDMs from DNA obtained by swallowed balloon devices. Methods: MDMs were assayed in ESCC and normal tissues obtained from the populations of United States, Iran, and China, and from exfoliative cytology specimens obtained by balloons in a Chinese population. Areas under the receiver operating curve (AUC) of MDMs discriminating ESCC from normal tissues were calculated. Random forest prediction models were built, trained on U.S. cases and controls, and calibrated to U.S.-only controls (model 1) and three-country controls (model 2). Statistical tests were used to assess the relationship between dysplasia and MDM levels in balloons. Results: Extracted DNA from 333 ESCC and 322 normal tissues was analyzed, in addition to archival DNA from 98 balloons. For ESCC, model 1 validated in Iranian and Chinese tissues with AUCs of 0.90 and 0.87, and model 2 yielded AUCs of 0.99, 0.96, and 0.94 in tissues from the United States, Iran, and China, respectively. In Chinese balloons, MDMs showed a statistically significant trend of increasing levels with increasing grades of dysplasia (P &lt; 0.004). Conclusions: MDMs accurately discriminate ESCC from normal esophagus in tissues obtained from high- and low-incidence countries. Preliminary data suggest that levels of MDMs assayed in DNA from swallowed balloon devices increase with dysplasia grade. Larger studies are needed to validate these results. Impact: MDMs coupled with minimally invasive collection methods have the potential for worldwide application in ESCC screening.