Use of neural network models to predict liver transplantation waitlist mortality.

Current liver transplantation (LT) organ allocation relies on Model for End-Stage Liver Disease-sodium scores to predict mortality in patients awaiting LT. This study aims to develop neural network (NN) models that more accurately predict LT waitlist mortality. The study evaluates patients listed for LT between February 27, 2002, and June 30, 2021, using the Organ Procurement and Transplantation Network/United Network for Organ Sharing registry. We excluded patients listed with Model for End-Stage Liver Disease (MELD) exception scores and those listed for multiorgan transplant, except for liver-kidney transplant. A subset of data from the waiting list was used to create a mortality prediction model at 90 days after listing with 105,140 patients. A total of 28 variables were selected for model creation. The data were split using random sampling into training, validation, and test data sets in a 60:20:20 ratio. The performance of the model was assessed using area under the receiver operating curve (AUC-ROC) and area under the precision-recall curve (AUC-PR). AUC-ROC for 90-day mortality was 0.936 (95% confidence interval [CI], 0.934-0.937), and AUC-PR was 0.758 (95% CI, 0.754-0.762). The NN 90-day mortality model outperformed MELD-based models for both AUC-ROC and AUC-PR. The 90-day mortality model specifically identified more waitlist deaths with a higher recall (sensitivity) of 0.807 (95% CI, 0.803-0.811) versus 0.413 (95% CI, 0.409-0.418; p < 0.001). The performance metrics were compared by breaking the test data set into multiple patient subsets by ethnicity, gender, region, age, diagnosis group, and year of listing. The NN 90-day mortality model outperformed MELD-based models across all subsets in predicting mortality. In conclusion, organ allocation based on NN modeling has the potential to decrease waitlist mortality and lead to more equitable allocation systems in LT.

Transfusion requirements and alloimmunization to red blood cell antigens in orthotopic liver transplantation.

BACKGROUND AND OBJECTIVES: Orthotopic liver transplantation (OLT) has been associated with high blood transfusion requirements. We evaluated the transfusion needs and frequency of alloimmunization to RBC antigens among OLT recipients pre- and post-transplantation. MATERIALS AND METHODS: We reviewed the medical records of patients who underwent a first OLT between January 2007 and June 2017. Transfusions given only during the perioperative period, defined by 1 week before OLT until 2 weeks following OLT, were included in this study. Records were reviewed in June 2019 for updated antibody testing results. RESULTS: A total of 970 patients underwent OLT during the study period. The median age of patients was 57 years; 608(62.7%) were male. During the perioperative period, transfused patients received an average of 10.7 (±10.7) RBC units, 15.6 (±16.2) thawed plasma units and 4.1 (±4.3) platelet units. At the time of OLT, a total of 101 clinically significant RBC alloantibodies were documented in 58(5.98%) patients. Fifty-three of these antibodies were directed against Rh blood group antigens. Twenty-two (37.9%) patients had more than one alloantibody. Patients with alloimmunization before OLT (N = 58) received perioperatively comparable number of RBCs to non-alloimmunized patients (10.5 ± 10.6 vs. 9.6 ± 10.7; p = 0.52). There was no significant difference in perioperative or intraoperative RBC transfusion between patients with one alloantibody and those with multiple alloantibodies. Only 16 patients (16/737; 2.17%) developed new alloantibodies at a median of 61 days after OLT. The overall alloimmunization rate was 9.8% (72/737), and female patients were more likely to be alloimmunized. CONCLUSION: Blood transfusion requirements in OLT remain high. However, the rate of RBC alloimmunization was not higher than the general patient population.

Outcomes in Living Donor Compared With Deceased Donor Primary Liver Transplantation in Lower Acuity Patients With Model for End-Stage Liver Disease Scores <30.

Although recent studies have reported favorable outcomes in living donor liver transplantation (LDLT), it remains unclear which populations benefit most from LDLT. This study aims to evaluate LDLT outcomes compared with deceased donor LT (DDLT) according to Model for End-Stage Liver Disease (MELD) score categories. Using data from the United Network for Organ Sharing registry, outcomes were compared between 1486 LDLTs; 13,568 donation after brain death (DBD)-DDLTs; and 1171 donation after circulatory death (DCD)-DDLTs between 2009 and 2018. Because LDLT for patients with MELD scores >30 was rare, all patients with scores >30 were excluded to equalize LDLT and DDLT cohorts. Risk factors for 1-year graft loss (GL) were determined separately for LDLT and DDLT. Compared with LDLT, DBD-DDLT had a lower risk of 30-day (adjusted hazard ratio [aHR], 0.60; P < 0.001) and 1-year GL (aHR, 0.57; P < 0.001). The lower risk of GL was more prominent in the mid-MELD score category (score 15-29). Compared with LDLT, DCD-DDLT had a lower risk of 30-day GL but a comparable risk of 1-year GL, regardless of MELD score category. In LDLT, significant ascites was an independent risk for GL in patients with mid-MELD scores (aHR, 1.68; P = 0.02), but not in the lower-MELD score group. The risk of 1-year GL in LDLT patients with ascites who received a left liver was higher than either those who received a right liver or those without ascites who received a left liver. In LDLT, combinations of MELD scores of 15 to 29, moderate/severe ascites, and the use of a left liver are associated with worse outcomes. These findings help calibrate appropriate patient and graft selection in LDLT.

Paradigm Change in Liver Transplantation Practice After the Implementation of the Liver-Kidney Allocation Policy.

The Organ Procurement and Transplantation Network (OPTN)/United Network for Organ Sharing (UNOS) policy regarding kidney allocation for liver transplantation (LT) patients was implemented in August 2017. This study evaluated the effects of the simultaneous liver-kidney transplantation (SLKT) policy on outcomes in LT alone (LTA) patients with kidney dysfunction. We analyzed adult primary LTA patients with kidney dysfunction at listing (estimated glomerular filtration rate [eGFR] less than 30 mL/minute or dialysis requirement) between January 2015 and March 2019 using the OPTN/UNOS registry. Waitlist practice and kidney transplantation (KT) listing after LTA were compared between prepolicy and postpolicy groups. There were 3821 LTA listings with eGFR <30 mL/minute included. The daily number of listings on dialysis was significantly higher in Era 2 (postpolicy group) than Era 1 (prepolicy group) (1.21/day versus 0.95/day; P < 0.001). Of these LTA listings, 90-day LT waitlist mortality, LTA probability, and 1-year post-LTA survival were similar between eras. LTA recipients in Era 2 had a higher probability for KT listing after LTA than those in Era 1 (6.2% versus 3.9%; odds ratio [OR], 3.30; P < 0.001), especially those on dialysis (8.4% versus 2.0%; OR, 4.38; P < 0.001). Under the safety net rule, there was a higher KT probability after LTA (26.7% and 53% at 6 months in Eras 1 and 2, respectively; P = 0.02). After the implementation of the policy, the number of LTA listings among patients on dialysis increased significantly. While their posttransplant survival did not change, KT listing after LTA increased. The safety net rule led to high KT probability and a low waitlist mortality rate in patients who were listed for KT after LTA. These results suggest that the policy successfully achieved the goals of providing appropriate opportunities of KT for LT patients, which did not compromise LTA waitlist or posttransplant outcomes in patients with kidney dysfunction and provided KT opportunities if patients developed kidney failure after LTA.

Effect of Mandatory 6-Month Waiting Period on Waitlist and Transplant Outcomes in Patients With Hepatocellular Carcinoma.

BACKGROUND AND AIMS: Organ Procurement and Transplantation Network/United Network for Organ Sharing (OPTN/UNOS) policy mandates a 6-month waiting period before exception scores are granted to liver transplant candidates with hepatocellular carcinoma (HCC). This study aims to evaluate waitlist and posttransplant outcomes in patients with HCC, before and after implementation of the 6-month waiting rule. APPROACH AND RESULTS: We examined two groups from the UNOS registry: Group 1 (pre-6-month rule) consisted of patients registered as transplant candidates with HCC from January 1, 2013, to October 7, 2015 (n = 4,814); group 2 (post-6-month rule) consisted of patients registered from October 8, 2015, to June 30, 2018 (n = 3,287). As expected, the transplant probability was higher in the first 6 months after listing in group 1 than group 2 at 42.0% versus 6.3% (P < 0.001). However, the 6-month waitlist mortality/dropout rate was lower in group 2 at 1.2% than group 1 at 4.1% (P < 0.001). To assess regional parity of transplant, UNOS regions were categorized into three groups based on Model for End-Stage Liver Disease score at transplant: lower-score (regions 3, 10, and 11), middle-score (1, 2, 6, 8, and 9), and higher-score region groups (4, 5, and 7). Outcomes were compared from the time exception points were given, which we defined as conditional waitlist outcomes. Conditional waitlist mortality/dropout decreased, and transplant probability increased in all region groups, but the benefits of the policy were more pronounced in the higher and middle-score groups, compared with the lower-score group. The decline in waitlist mortality/dropout was only significant in the high Model for End-Stage Liver Disease group (P < 0.001). No effect was observed on posttransplant mortality or percent of patients within Milan criteria on explant. CONCLUSIONS: The HCC policy change was associated with decreased waitlist mortality/dropout and increased transplant probability. The policy helped to decrease but did not eliminate regional disparities in transplant opportunity without an effect on posttransplant outcomes.

Disease-specific waitlist outcomes in liver transplantation - a retrospective study.

This study aimed to evaluate possible discrepancies in waitlist outcomes between liver diseases, including alcohol-related liver disease (ALD), nonalcoholic steatohepatitis (NASH), hepatitis C virus infection (HCV), primary biliary cirrhosis (PBC), and primary sclerosing cholangitis (PSC). Patients registered for liver transplantation from January 11, 2016, to June 30, 2018, were evaluated using OPTN/UNOS registry. Waitlist outcomes were compared between the five-disease groups. Patients were categorized by initial MELD-Na-score (6-20, 21-29, and ≥30) to identify outcome variations. Prognostic impact of transplantation was assessed according to final MELD-Na scores using Cox regression analysis modeling transplantation as a time-dependent covariate. 6053 with ALD, 3814 with NASH, 1558 with HCV, 602 with PBC, and 819 with PSC were eligible. Compared to ALD with comparable MELD-Na-scores, NASH with lower [adjusted hazard ratio (aHR) = 1.30, P = 0.042] and mid-scores (aHR = 1.35, P = 0.008) showed significantly higher risk of 1-year waitlist mortality, and PBC with higher scores showed significantly higher risk of 90-day (aHR = 1.69, P = 0.03) and 1-year waitlist mortality (aHR = 1.69, P = 0.02). Positive prognostic impact of transplantation was not seen until score of 24-27 in ALD, 18-20 in HCV, 15-17 in NASH, and 24-27 in PBC and PSC. There are significant differences in waitlist outcomes among etiologies, which may differ the optimal transplant timing.

Short recipient warm ischemia time improves outcomes in deceased donor liver transplantation.

While adverse effects of prolonged recipient warm ischemia time (rWIT) in liver transplantation (LT) have been well investigated, few studies have focused on possible positive prognostic effects of short rWIT. We aim to investigate if shortening rWIT can further improve outcomes in donation after brain death liver transplant (DBD-LT). Primary DBD-LT between 2000 and 2019 were retrospectively reviewed. Patients were divided according to rWIT (≤30, 31-40, 41-50, and >50 min). The requirement of intraoperative transfusion, early allograft dysfunction (EAD), and graft survival were compared between the rWIT groups. A total of 1,256 patients of DBD-LTs were eligible. rWIT was ≤30min in 203 patients (15.7%), 31-40min in 465 patients (37.3%), 41-50min in 353 patients (28.1%), and >50min in 240 patients (19.1%). There were significant increasing trends of transfusion requirement (P < 0.001) and increased estimated blood loss (EBL, P < 0.001), and higher lactate level (P < 0.001) with prolongation of rWIT. Multivariable logistic regression demonstrated the lowest risk of EAD in the WIT ≤30min group. After risk adjustment, patients with rWIT ≤30 min showed a significantly lower risk of graft loss at 1 and 5-years, compared to other groups. The positive prognostic impact of rWIT ≤30min was more prominent when cold ischemia time exceeded 6 h. In conclusion, shorter rWIT in DBD-LT provided significantly better post-transplant outcomes.

Extreme hyponatremia as a risk factor for early mortality after liver transplantation in the MELD-sodium era.

The impact of hyponatremia on waitlist and post-transplant outcomes following the implementation of MELD-Na-based liver allocation remains unclear. We investigated waitlist and postliver transplant (LT) outcomes in patients with hyponatremia before and after implementing MELD-Na-based allocation. Adult patients registered for a primary LT between 2009 and 2021 were identified in the OPTN/UNOS database. Two eras were defined; pre-MELD-Na and post-MELD-Na. Extreme hyponatremia was defined as a serum sodium concentration ≤120 mEq/l. Ninety-day waitlist outcomes and post-LT survival were compared using Fine-Gray proportional hazard and mixed-effects Cox proportional hazard models. A total of 118 487 patients were eligible (n = 64 940: pre-MELD-Na; n = 53 547: post-MELD-Na). In the pre-MELD-Na era, extreme hyponatremia at listing was associated with an increased risk of 90-day waitlist mortality ([ref: 135-145] HR: 3.80; 95% CI: 2.97-4.87; P < 0.001) and higher transplant probability (HR: 1.67; 95% CI: 1.38-2.01; P < 0.001). In the post-MELD-Na era, patients with extreme hyponatremia had a proportionally lower relative risk of waitlist mortality (HR: 2.27; 95% CI 1.60-3.23; P < 0.001) and proportionally higher transplant probability (HR: 2.12; 95% CI 1.76-2.55; P < 0.001) as patients with normal serum sodium levels (135-145). Extreme hyponatremia was associated with a higher risk of 90, 180, and 365-day post-LT survival compared to patients with normal serum sodium levels. With the introduction of MELD-Na-based allocation, waitlist outcomes have improved in patients with extreme hyponatremia but they continue to have worse short-term post-LT survival.

A Share 21 model in liver transplantation: Impact on waitlist outcomes.

With the introduction of Model for End-Stage Liver Disease-Sodium (MELD-Na)-based allocation, the score at which patients benefit from liver transplantation (LT) has shifted from a score of 15 to 21. This study aimed to evaluate waitlist outcomes in patients with MELD-Na scores <21 and explore the utility of replacing "Share 15" with "Share 21." The study uses data from the Organ Procurement and Transplantation Network/United Network for Organ Sharing registry. All adult patients registered for LT after implementation of the MELD-Na-based allocation were evaluated. Waitlist patients with initial and final scores <21 were eligible. Patients with exception scores were excluded. To explore the potential impact of a Share 21 model, patients with an initial MELD-Na score of 6-14 (Group 1) and those with a score of 15-20 (Group 2) were compared for waitlist outcomes. There were 3686 patients with an initial score of 6-14 (Group 1) and 3282 with a score of 15-20 (Group 2). Group 2, when compared to Group 1, showed comparable risk of mortality (adjusted hazard ratio [aHR] 1.00, P = .97), higher transplant probability (aHR 3.25, P < .001), and lower likelihood of removal from listing because of improvement (aHR 0.74, P = .011). Share 21 may enhance transplant opportunities and increase parity for patients with higher MELD-Na scores without compromising waitlist outcomes.

Variation of Liver Transplant Practice and Outcomes During Public Holidays in the United States: Analysis of United Network for Organ Sharing Registry.

It has been reported that patients hospitalized outside regular working hours have worse outcomes. This study aims to compare outcomes following liver transplantation (LT) performed during public holidays and nonholidays. Methods: We analyzed the United Network for Organ Sharing registry data for 55 200 adult patients who underwent an LT between 2010 and 2019. Patients were grouped according to LT receipt during public holidays ±3 d (n = 7350) and nonholiday periods (n = 47 850). The overall post-LT mortality hazard was analyzed using multivariable Cox regression models. Results: LT recipient characteristics were similar between public holidays and nonholidays. Compared with nonholidays, deceased donors during public holidays had a lower donor risk index (median [interquartile range]: holidays 1.52 [1.29-1.83] versus nonholidays 1.54 [1.31-1.85]; P = 0.001) and shorter cold ischemia time (median [interquartile range]: holidays 5.82 h [4.52-7.22] versus nonholidays 5.91 h [4.62-7.38]; P < 0.001). Propensity score matching 4-to-1 was done to adjust for donor and recipient confounders (n = 33 505); LT receipt during public holidays (n = 6701) was associated with a lower risk of overall mortality (hazard ratio 0.94 [95% confidence interval, 0.86-0.99]; P = 0.046). The number of livers that were not recovered for transplant was higher during public holidays compared with nonholidays (15.4% versus 14.5%, respectively; P = 0.03). Conclusions: Although LT performed during public holidays was associated with improved overall patient survival, liver discard rates were higher during public holidays compared with nonholidays.

Hepatitis E Diagnosis and Management After Liver, Kidney, or Heart Transplant: A Single-Center Experience.

BACKGROUND: Transplant-related hepatitis E virus (HEV) infection is a rarely recognized phenomenon with significant clinical importance given its potential to result in chronic hepatitis posttransplant. METHODS: We retrospectively evaluated HEV diagnosis and treatment after liver, kidney, and heart transplant in a single center. We identified patients diagnosed with HEV by serologic testing and evaluated their treatment regimens. RESULTS: Fifteen transplant recipients (12 liver, 2 kidney, and 1 heart) presented with elevated liver enzymes and were positive for HEV IgM antibody. Liver enzymes normalized in 4 patients after being treated with ribavirin. One of the 4 patients had 2 recurrences with positive HEV RNA results following ribavirin treatment but recovered after 12 months of ribavirin therapy. After treatment with reduction in immunosuppression without antiviral treatment, 6 of 8 patients' liver enzymes normalized. One of these patients died of acute pancreatitis 2 months after testing positive for HEV IgM antibody. CONCLUSIONS: The potential for complications related to active HEV infections in transplant recipients necessitates prompt diagnosis and treatment to prevent irreversible damage. Diagnosis with HEV reverse transcriptase-polymerase chain reaction should follow a positive HEV IgM antibody test. This manuscript provides evidence that ribavirin antiviral therapy and reducing immunosuppression are effective treatments for HEV infections in liver, kidney, and heart transplant recipients, which has not been sufficiently investigated in the population of the United States. Larger multicenter studies are needed to confirm the risks and benefits of using ribavirin antiviral therapy as first-line therapy of HEV posttransplant.

Can Weather Be a Factor in Liver Transplant Waitlist and Posttransplant Outcomes? Analysis of United Network for Organ Sharing Registry.

BACKGROUND: Cold climate is known to affect the frequency and attributable mortality of various illnesses. This study aims to evaluate the effect of climate among regions on liver transplant (LT) outcomes. METHODS: We analyzed data from the United Network for Organ Sharing registry for 98,517 adult patients (aged ≥ 18 years) who were listed for LT between 2010 and 2019. During this period, 51,571 patients underwent single-organ, deceased LT. States were categorized based on their mean winter temperature: warm states (45°F-70°F), intermediate states (30°F-45°F), and cold states (0°F-30°F). Post-LT outcomes at 1 month, 1 year, and 3 years were compared using Cox proportional hazard models. Ninety-day and 1-year waitlist outcomes were compared among climate regions using Fine-Gray hazard regression model. RESULTS: After adjusting risks for recipient and donor characteristics, LT candidates in cold states had a significantly higher waitlist (90-day: subdistribution hazard ratio (HR) 1.46; 1-year: subdistribution HR 1.41; P < .001) and posttransplant mortality (30-day: subdistribution HR 1.23; P = .009, 1-year: subdistribution HR 1.16; P = .001; 3-year: subdistribution HR 1.08; P = .007). LT recipients in cold states had a higher proportion of deaths due to infections than warm states (cold states: 2.3%; intermediate states: 2.1%; and warm states: 1.7%; P < .001). CONCLUSIONS: Potential reasons include weather-related changes in the behavioral and physiological parameters of patients.

Incidence and Risk Factors for Fatal Graft-versus-host Disease After Liver Transplantation.

BACKGROUND: Graft-versus-host disease (GVHD) after liver transplantation (LT) is a rare but serious complication. The aim of this study is to identify risk factors, including immunosuppressive regimens, for mortality due to GVHD (fatal GVHD). METHODS: Using data from the Organ Procurement and Transplantation Network and United Network for Organ Sharing registry, 77 416 adult patients who underwent LT between 2003 and 2018 were assessed. Risk factors for fatal GVHD were analyzed by focusing on induction and maintenance immunosuppression regimens. RESULTS: The incidence of fatal GVHD was 0.2% (121 of 77 416), of whom 105 (87%) died within 180 d and 13 (11%) died between 181 d and 1 y. Median survival after LT was 68.0 (49.5-125.5) d. Recipient age minus donor age >20 y (hazard ratio [HR], 2.57; P < 0.001) and basiliximab induction (HR, 1.69; P = 0.018) were independent risk factors for fatal GVHD. Maintenance therapy with mycophenolate mofetil (MMF) was associated with a decrease in fatal GVHD (HR, 0.51; P = 0.001). In an increased risk cohort of patients with recipient-donor age discrepancy >20 y, MMF use was associated with a 50% decline in fatal GVHD (HR, 0.50; P < 0.001). CONCLUSIONS: Recipient age minus donor age >20 y remains a significant risk factor for fatal GVHD. The risk of fatal GVHD significantly increases in association with basiliximab induction and decreases with MMF maintenance. These associations were pronounced in patients with recipient minus donor age >20 y. These results emphasize the importance of donor age and individualized immunosuppression regimens on the risk of fatal GVHD.

Acute Kidney Injury in a Predominantly African American Cohort of Kidney Transplant Recipients With COVID-19 Infection.

BACKGROUND: Renal involvement in severe or critical acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is frequent. Acute kidney injury (AKI) in African American (AA) kidney transplant recipients (KTRs) with COVID-19 is not well described. We report our experience with a predominantly AA cohort (79%) of KTRs with COVID-19 infections in the Detroit Metropolitan area. METHODS: In this retrospective, single-center study, we identified 39 KTRs who tested positive for SARS-CoV-2 between March 16 and April 25, 2020. Data from electronic medical records were retrieved and compared between KTRs without AKI and KTRs with AKI. RESULTS: One patient was excluded due to delayed graft function. Final analysis of AKI in KTRs with proven COVID-19 was done on 38 patients of which 30 were AA (79%). AKI occurred in 71.1% of COVID-19 KTRs (n = 27), of whom 6 (22.2%) patients required HD. The incidence of AKI in our cohort was 71% (27/38). AKI rate among AA was 76.7% versus 50% in non-AA cohort (P = 0.195). In a univariate logistic regression analysis, AA race was not significantly associated with AKI odds ratio (3.4; CI, 0.68-17.4; P = 0.14). After risk adjustment by race, patients with diabetes showed a significantly higher risk of AKI (adjusted odds ratio, 19.85; CI, 1.65-58.66; P = 0.012). KTRs with AKI had more preexisting renin angiotensin aldosterone system inhibitor use than KTRs without AKI (P = 0.03). CONCLUSIONS: KTRs infected with SARS-CoV-2 have a high incidence of AKI, with associated increased morbidity and mortality. Although no racial differences in mortality were noted in our KTRs with AKI, we await data from registries to help elucidate this difference.