Amy Shui’s research while affiliated with University of California, San Francisco and other places

Background Frailty is strongly associated with mortality after liver transplantation. However, national guidelines discourage its use as a sole reason to decline a patient for liver transplantation, as some frail patients have acceptable outcomes. We aimed to develop a composite index, the Liver Transplant Comorbidity Index (LTCI), integrating frailty and other comorbidities, as a risk factor for longer-term (3-year) post-transplant mortality. Methods This 8-center prospective Functional Assessment in Liver Transplantation (FrAILT) Study included adult recipients of a primary deceased donor liver transplant from 2012-2022. Frailty was measured using the Liver Frailty Index (LFI ≥4.5=frail). Other candidate variables included demographics, laboratories, and comorbidities. Cox proportional hazards regression with best subset selection was used to identify risk factors of 3-year post-transplant death. The final model was selected based on Aikaike Information Criterion and clinical pragmatism. Results Of 1,472 liver transplant recipients. 290 (20%) were frail. Three-year post-transplant mortality was higher in frail versus non-frail patients (13 vs. 8%; p =0.03). The final LTCI included 5 variables: frailty, coronary artery disease, hepatocellular carcinoma, renal dysfunction, and diabetes. Three-year post-transplant mortality in low-, moderate-, and high-risk LTCI groups was 93%, 87%, and 80% respectively. In multivariable analysis, after adjusting for donor factors (age, DCD), both moderate- (HR 2.23; 95% CI 1.46-3.40; p <0.001) and high-risk (HR 2.78; 95% CI 1.67-4.64; p <0.001) status were associated with 3-year post-transplant mortality. Conclusion The LTCI, comprising 5 pre-transplant clinical parameters, effectively identifies patients at increased risk of post-transplant mortality. By integrating frailty in the context of other co-morbidities, the LTCI can help providers better weigh the relative transplant risks and benefits and standardize selection of transplant candidates.

Background Alcohol use and hepatitis C virus (HCV) often coexist and are associated with cardiovascular disease. One of the underlying drivers is dyslipidemia. We assessed lipid and lipoprotein levels and the relationship between alcohol use and atherogenic lipid profiles, specifically small dense low‐density lipoprotein cholesterol (sdLDL‐C), in Latinos with and without HCV. Methods From June 1, 2002, to January 1, 2016, 150 Latino adults underwent demographic, clinical, metabolic, lipid/lipoprotein, and genetic evaluations. Linear regression (adjusted for age, sex, and recent alcohol use) assessed factors associated with sdLDL‐C. Results Participant characteristics were as follows: median age 44 years, 64% male, 39% HCV+, and alcohol use in the last 12 months was 19% heavy and 47% moderate. Ancestries were as follows: 52% European, 40% Native American (NA), and 4.3% African. 29% had non‐CC PNPLA3, 89% non‐CC TM6SF2, and 73% non‐CC IL‐28b genotypes. High‐density lipoprotein (HDL) cholesterol, HDL‐3, apolipoprotein A‐1, and lipoprotein‐associated phospholipase A2 levels differed by alcohol use groups (p < 0.05). On multivariable analysis, female sex (est. –6.08, p < 0.001), HCV+ status (est. –8.49, p < 0.001), and heavy alcohol use (vs. none) (est. –4.32, p = 0.03) were associated with lower, while NA ancestry (est. 0.92; p = 0.01) and adipose tissue insulin resistance (est. 3.30, p < 0.001) were associated with higher sdLDL‐C levels. The positive association between NA ancestry and sdLDL‐C was dampened by the presence of a non‐CC IL28b genotype (interaction est. −1.95, p = 0.01). Conclusions In this Latino cohort, ancestry and metabolic dysfunction, independent of alcohol use and HCV, were associated with atherogenic risk. In addition to HCV treatment in this population, cardiometabolic health should be optimized.

Background Cannabis use is increasing in the United States, including among liver transplant candidates. Although the anesthesia literature suggests an association between cannabis use and increased post-operative pain, the impact of cannabis use on post-liver transplant (LT) opioid use remains unknown. This study investigates changes in cannabis use at a transplant center over time, as well as the impact of cannabis use on post-LT opioid use, healthcare utilization, and mortality. Methods We included 4,236 patients evaluated for LT at our institution between January 2013 and July 2023. Our primary risk factor was cannabis use, defined as urine toxicology (UTox) positive for cannabis within 90 days of LT evaluation. Our primary outcome was post-LT opioid use, including oral morphine equivalents (OME) received during the LT hospitalization and discharge opioid prescriptions. We used multivariable logistic and quantile regression to compare post-LT opioid use, healthcare utilization outcomes, and mortality between cannabis users and nonusers. Results Cannabis use was associated with higher OME use in the 48 hours post-LT ( p =0.04). There were no statistically significant differences between groups in 72-hour ( p =0.07) or 7-day cumulative OME ( p =0.33), opioid prescriptions on discharge ( p =0.25), hospital length of stay (LOS, p =0.69), intensive care unit LOS ( p =0.94), 90-day readmission ( p =0.66), or 90-day mortality ( p =0.96). Conclusions While cannabis use pre-LT was associated with significantly higher opioid use in the immediate postoperative period, this did not translate to differences in opioid use beyond 48 hours post-LT, or short-term healthcare utilization or clinical outcomes. These findings should help set provider expectations for immediate post-LT pain control. Our findings support the growing body of literature that fails to identify an association between pre-LT cannabis use and post-LT outcomes.

Background & Aims Patients with hepatocellular carcinoma (HCC) meeting UNOS-downstaging (DS) criteria have excellent post-liver transplantation (LT) outcomes. Studies on HCC beyond UNOS-DS criteria (“All-comers” (AC)) have been limited by small sample size and short follow-up time, prompting this analysis. Approach & Results 326 patients meeting UNOS-DS and 190 meeting AC criteria from 9 LT centers across 5 UNOS regions were enrolled from 2015 to 2023 and prospectively followed. Competing risk analysis and Kaplan-Meier method were used to evaluate downstaging and LT outcomes, and Fine-and-Gray and Cox models were used to identify predictors of outcomes. AC and UNOS-DS had similar median alpha-fetoprotein (15 vs. 12 ng/mL; p =0.08), MELD (9 vs. 9; p =0.52), and Child-Pugh (A vs. A; p =0.30). At 2 years from first local regional therapy, 82% of UNOS-DS and 66% of AC were successfully downstaged ( p <0.001). In AC, DS rates were 72% for tumor number plus diameter of largest lesion<10, 51% for sum 10-12, and 39% for sum>12 ( p =0.01). Y-90 achieved higher DS success than transarterial chemoembolization (TACE) in AC (74% vs. 65%; p <0.001). 48% of UNOS-DS and 40% of AC underwent LT ( p =0.10). 5-year post-LT survival was similar between UNOS-DS and AC (74% vs. 72%; p =0.77), although 5-year post-LT recurrence was higher in AC (30% vs. 14%; p =0.02). Conclusions Despite higher HCC recurrence and lower ITT survival in AC, post-LT survival was comparable between UNOS-DS and AC. Y-90 attained higher DS success than TACE in AC. LT after downstaging is feasible in AC, though defining an upper limit in tumor burden may be necessary.

Pediatric liver transplant outcomes exhibit disparities, necessitating identification of modifiable risk factors to develop targeted interventions. We characterized associations between household material economic hardship (e.g., financial barriers to housing or food) and pediatric liver transplant outcomes. We recruited pediatric liver transplant recipients <18 years at the time of transplant across 8 U.S. centers. Our primary exposure was >/=1 household material economic hardship (i.e., food insecurity, housing instability, transportation challenges, or utility concerns), measured using the Accountable Healthcare Communities screening tool. Outcomes included 90-day and 1-year (1) total inpatient-bed days, and (2) episodes of T-cell mediated rejection (TCMR). Of the 77 participants (36% female), 34% reported household material economic hardship. Such hardship was associated with increased total inpatient bed-days within 90 days (ratio estimate: 1.45, 95%CI: 1.08, 1.96); the association persisted after adjusting for health literacy, insurance, and transplant center (ratio estimate: 1.37, 95%CI: 1.02, 1.84). Household material economic hardship was associated with total inpatient bed-days within 1-year post-transplant (ratio estimate: 3.2, 95%CI: 1.1, 10.1); associations diminished in multivariable analyses (ratio estimate: 2.2, 95%CI: 0.7, 6.9). Household material economic hardship was associated with increased risk of TCMR within 1 year of transplant (RR: 2.1, 95%CI: 1.1, 4.2); the association diminished in propensity-score matched analyses (RR: 1.4, 95%CI: 0.9, 2.3). Our findings highlight the adverse influence of household material economic hardship on pediatric liver transplant outcomes within the first year. Targeted social risk assistance and adjustment strategies offer actionable avenues to mitigate these challenges and enhance outcomes in pediatric liver transplant recipients.

Background Alcohol use disorder (AUD) treatment can help improve clinical outcomes among patients with alcohol‐associated cirrhosis but is underutilized. Among socioeconomically disadvantaged patients with alcohol‐associated cirrhosis, we examined rates of lifetime and past 12‐month AUD treatment utilization and associated demographic and clinical characteristics. Methods Racial/ethnically diverse patients with alcohol‐associated cirrhosis who had at least one hepatology clinic visit in the prior 6 months were recruited from three Northern California medical centers serving veterans and safety‐net populations. Participants self‐reported their AUD treatment utilization, liver disease quality of life (LDQoL), history and current symptoms of anxiety and depression, and problematic drinking as measured by the Alcohol Use Disorders Identification Test (AUDIT). Clinical measures including liver disease severity were captured from medical records. Results Among 196 participants, the majority were male (88%) with a mean age of 62 years. Two‐thirds of participants (67%) reported ever utilizing AUD treatment and 32% reported utilizing AUD treatment in the past 12 months. Compared with those who did not utilize AUD treatment, participants who utilized lifetime or past 12‐month AUD treatment were younger, had lower LDQoL scores, and had higher scores on current symptoms of anxiety, depression, and problematic drinking. In multivariable analyses, the odds of ever utilizing pharmacological treatment alone or both behavioral and pharmacological treatment (vs. none) were lower with older age or higher LDQoL, and higher among those with a history of anxiety/depressive disorder. For past 12‐month treatment utilization, odds were lower with older age, and higher among those with current clinically significant anxiety/depression or problematic drinking. Conclusions Patients with alcohol‐associated cirrhosis who were younger or had anxiety/depression and problematic drinking were more likely to utilize AUD treatment. To improve AUD treatment utilization, targeted outreach to patients less likely to receive care and the provision of integrated ALD and AUD treatment is warranted.

Importance Conclusions vary substantially among studies examining associations between area-based social determinants of health (SDOH) and pediatric health disparities based on the selected patient population and SDOH index. Most national studies use zip codes, which encompass a wide distribution of communities, limiting the generalizability of findings. Objectives To characterize the distributions of composite SDOH indices for pediatric surgical patients within a national sample of academic children’s hospitals and to assess SDOH index precision in classifying patients at similar levels of disadvantage. Design, Setting, and Participants This multicenter retrospective cohort study included patients younger than 18 years who underwent surgery from January 1, 2016, to December 31, 2021, at 8 American College of Surgeons National Surgical Quality Improvement Program Pediatric children’s hospitals. Data were analyzed November 15, 2023, to September 25, 2024. Exposures Exposures included the Social Vulnerability Index (SVI), Area Deprivation Index (ADI), and Child Opportunity Index (COI), which are composite scores of SDOH factors within a geographic area. A standardized, high-fidelity process was developed to link individual patients to SDOH indices at the US census tract and census block group level. Main Outcomes and Measures The primary outcome was composite SDOH index distribution, which was assessed using interrater reliability scores. Substantial agreement was defined as a Cohen κ statistic higher than 0.60. Results Of 55 865 included patients, 54.6% were male; 34.8% were infants and toddlers (0-3 years of age), 39.0% were school age (4-12 years), and 26.2% were adolescents (13-17 years). A total of 3468 patients (6.2%) could not be matched to either an SVI, ADI, or COI. Patients with missing geocodes were more likely to be Hispanic (20.1% vs 17.1%; P < .001) and have Medicaid insurance (48.1% vs 44.6%; P < .001) compared with patients with addresses that could be geocoded. With all institutions grouped, SDOH indices showed only minor variations. However, within each institution and among institutions, indices varied considerably, especially the ADI. Indices had low-to-fair interrater reliability within institutions (κ range, 0.15-0.33), indicating that each index classified individual patients differently according to disadvantage. Conclusions and Relevance In this multicenter retrospective cohort study of 55 865 pediatric surgical patients, 3 different composite measures of SDOH classified disadvantage for individual patients differently. The SDOH index components should be understood and carefully considered prior to inclusion of a composite measure in the analysis of children’s surgical outcomes.

Background Anorectal malformations (ARMs) have an incidence of up to 1 in 4000 live births and can require immediate neonatal surgery due to associated intestinal blockage. Due to limited surgical access, Ugandan children present late and undergo three separate staged operations: (1) initial colostomy formation; (2) repair of the ARM (called anoplasty); and (3) colostomy closure. Three operations result in long treatment duration, potential complications with each procedure, delays in care, and stigmata associated with colostomies. By offering primary repair for ARMs in a resource-limited setting, we expect to: reduce healthcare expenditure by families, length of treatment, length of hospital stay, frequency of hospital visits, and social rejection. Materials and methods A pragmatic clinical trial was performed examining the effectiveness of primary repair (prospective arm) and comparing it with the three-stage repairs (retrospective arm). Results Of the 241 patients included for analysis—157 patients had a three-stage repair, whereas 84 patients had one- or two-stage repair. The median [IQR age at the last surgery (days) was 730.0 (365.0, 1460.0) vs 180.0 (90.0, 285.0)] in three-stage and one- or two-stage repairs, respectively. There was no difference in postoperative complications compared to patients who had three-stage repair. Patients who had a two-stage repair had less time with colostomy than those with three-stage repair. Non-inferiority analysis demonstrated that the primary repair approach was non-inferior to the three-stage approach. Conclusions Primary repair for ARM is effective in low-income settings. It allows for less time with colostomy with no difference in post-operative complications. The decision on approach for treatment depends on the surgeon’s experience and clinical judgment.