Machine Learning Prediction of Liver Allograft Utilization From Deceased Organ Donors Using the National Donor Management Goals Registry.

Early prediction of whether a liver allograft will be utilized for transplantation may allow better resource deployment during donor management and improve organ allocation. The national donor management goals (DMG) registry contains critical care data collected during donor management. We developed a machine learning model to predict transplantation of a liver graft based on data from the DMG registry. METHODS: Several machine learning classifiers were trained to predict transplantation of a liver graft. We utilized 127 variables available in the DMG dataset. We included data from potential deceased organ donors between April 2012 and January 2019. The outcome was defined as liver recovery for transplantation in the operating room. The prediction was made based on data available 12-18 h after the time of authorization for transplantation. The data were randomly separated into training (60%), validation (20%), and test sets (20%). We compared the performance of our models to the Liver Discard Risk Index. RESULTS: Of 13 629 donors in the dataset, 9255 (68%) livers were recovered and transplanted, 1519 recovered but used for research or discarded, 2855 were not recovered. The optimized gradient boosting machine classifier achieved an area under the curve of the receiver operator characteristic of 0.84 on the test set, outperforming all other classifiers. CONCLUSIONS: This model predicts successful liver recovery for transplantation in the operating room, using data available early during donor management. It performs favorably when compared to existing models. It may provide real-time decision support during organ donor management and transplant logistics.

Critical care and ventilatory management of deceased organ donors impact lung use and recipient graft survival.

Current risk-adjusted models for donor lung use and lung graft survival do not include donor critical care data. We sought to identify modifiable donor physiologic and mechanical ventilation parameters that predict donor lung use and lung graft survival. This is a prospective observational study of donors after brain death (DBDs) managed by 19 Organ Procurement Organizations from 2016 to 2019. Demographics, mechanical ventilation parameters, and critical care data were recorded at standardized time points during donor management. The lungs were transplanted from 1811 (30%) of 6052 DBDs. Achieving ≥7 critical care endpoints was a positive predictor of donor lung use. After controlling for recipient factors, donor blood pH positively predicted lung graft survival (OR 1.48 per 0.1 unit increase in pH) and the administration of dopamine during donor management negatively predicted lung graft survival (OR 0.19). Tidal volumes ≤8 ml/kg predicted body weight (OR 0.65), and higher positive end-expiratory pressures (OR 0.91 per cm H2 O) predicted decreased donor lung use without affecting lung graft survival. A randomized clinical trial is needed to inform optimal ventilator management strategies in DBDs.

Impact of Deceased Donor Management on Donor Heart Use and Recipient Graft Survival.

BACKGROUND: Current risk-adjusted models used to predict donor heart use and cardiac graft survival from organ donors after brain death (DBDs) do not include bedside critical care data. We sought to identify novel independent predictors of heart use and graft survival to better understand the relationship between donor management and transplantation outcomes. STUDY DESIGN: We conducted a prospective observational study of DBDs managed from 2008 to 2013 by 10 organ procurement organizations. Demographic data, critical care parameters, and treatments were recorded at 3 standardized time points during donor management. The primary outcomes measures were donor heart use and cardiac graft survival. RESULTS: From 3,433 DBDs, 1,134 hearts (33%) were transplanted and 969 cardiac grafts (85%) survived after 684 ± 392 days of follow-up. After multivariable analysis, independent positive predictors of heart use included standard criteria donor status (odds ratio [OR] 3.93), male sex (OR 1.68), ejection fraction > 50% (OR 1.64), and partial pressure of oxygen to fraction of inspired oxygen ratio > 300 (OR 1.31). Independent negative predictors of heart use included donor age (OR 0.94), BMI > 30 kg/m2 (OR 0.78), serum creatinine (OR 0.83), and use of thyroid hormone (OR 0.78). As for graft survival, after controlling for known recipient risk factors, thyroid hormone dose was the only independent predictor (OR 1.04 per µg/h). CONCLUSIONS: Modifiable critical care parameters and treatments predict donor heart use and cardiac graft survival. The discordant relationship between thyroid hormone and donor heart use (negative predictor) vs cardiac graft survival (positive predictor) warrants additional investigation.

Vasopressor selection during critical care management of brain dead organ donors and the effects on kidney graft function.

BACKGROUND: Delayed graft function (DGF), the need for dialysis in the first week following kidney transplant, affects approximately one quarter of deceased-donor kidney transplant recipients. Donor demographics, donor serum creatinine, and graft cold ischemia time are associated with DGF. However, there is no consensus on the optimal management of hemodynamic instability in organ donors after brain death (DBDs). Our objective was to determine the relationship between vasopressor selection during donor management and the development of DGF. METHODS: Prospective observational data, including demographic and critical care parameters, were collected for all DBDs managed by 17 organ procurement organizations from nine Organ Procurement and Transplantation Network Regions between 2012 and 2018. Recipient outcome data were linked with donor data through donor identification numbers. Donor critical care parameters, including type of vasopressor and doses, were recorded at three standardized time points during donor management. The analysis included only donors who received at least one vasopressor at all three time points. Vasopressor doses were converted to norepinephrine equivalent doses and analyzed as continuous variables. Univariate analyses were conducted to determine the association between donor variables and DGF. Results were adjusted for known predictors of DGF using binary logistic regression. RESULTS: Complete data were available for 5,554 kidney transplant recipients and 2,985 DBDs. On univariate analysis, donor serum creatinine, donor age, donor subtype, kidney donor profile index, graft cold ischemia time, phenylephrine dose, and dopamine dose were associated with DGF. After multivariable analysis, increased donor serum creatinine, donor age, kidney donor profile index, graft cold ischemia time, and phenylephrine dose remained independent predictors of DGF. CONCLUSION: Higher doses of phenylephrine were an independent predictor of DGF. With the exception of phenylephrine, the selection and dose of vasopressor during donor management did not predict the development of DGF. LEVEL OF EVIDENCE: Prognostic study, Level III.

Deceased organ donor factors influencing pancreatic graft transplantation and survival.

Criteria for organ acceptance in brain-dead organ donors remain inconsistent, especially when concerning pancreatic transplants. We sought to examine donor-specific predictors of pancreatic graft use and survival to better guide the selection and management of potential donors. A prospective observational study of all donors from ten organ procurement organizations was conducted from March 2012 to January 2015. Critical care endpoints were collected at 4 standardized time points. Data associated with pancreatic transplantation and graft survival rates were first determined using univariate analyses, and then logistic regression was used to identify independent predictors of these two outcomes. From 1819 donors, 238 (13.1%) pancreata were transplanted, and at a mean follow-up of 192 days, 218 (91.6%) grafts had survived. After regression analysis, donor age (OR = 0.89), HgbA1C (OR = 0.07), and achieving the donor management goal (DMG) for ejection fraction at allocation of ≥50% (OR = 3.29) remained as independent predictors of pancreatic utilization. On regression analysis, graft survival was independently predicted by lower donor age (OR = 0.93) and achieving the DMGs for mean arterial pressure (60-110 mm Hg) and glucose (≤180 mg/dL) at separate time points. These results may help guide the management and selection of potential pancreatic donors after brain death.

A porcine model to study the effect of brain death on kidney genomic responses.

INTRODUCTION: A majority of transplanted organs come from donors after brain death (BD). Renal grafts from these donors have higher delayed graft function and lower long-term survival rates compared to living donors. We designed a novel porcine BD model to better delineate the incompletely understood inflammatory response to BD, hypothesizing that adhesion molecule pathways would be upregulated in BD. METHODS: Animals were anesthetized and instrumented with monitors and a balloon catheter, then randomized to control and BD groups. BD was induced by inflating the balloon catheter and animals were maintained for 6 hours. RNA was extracted from kidneys, and gene expression pattern was determined. RESULTS: In total, 902 gene pairs were differently expressed between groups. Eleven selected pathways were upregulated after BD, including cell adhesion molecules. CONCLUSIONS: These results should be confirmed in human organ donors. Treatment strategies should target involved pathways and lessen the negative effects of BD on transplantable organs.

Active Donor Management During the Hospital Phase of Care Is Associated with More Organs Transplanted per Donor.

BACKGROUND: Meeting donor management goals when caring for potential organ donors has been associated with more organs transplanted per donor (OTPD). Concern persists, however, as to whether this indicates that younger/healthier donors are more likely to meet donor management goals or whether active management affects outcomes. STUDY DESIGN: A prospective observational study of all standard criteria donors was conducted by 10 organ procurement organizations across United Network for Organ Sharing Regions 4, 5, and 6. Donor management goals representing normal critical care end points were measured at 2 time points: when a catastrophic brain injury was recognized and a referral was made to the organ procurement organization by the DH; and after brain death was declared and authorization for organ donation was obtained. Donor management goals Bundle "met" was defined as achieving any 7 of 9 end points. A positive Bundle status change was defined as not meeting the Bundle at referral and subsequently achieving it at authorization. The primary outcomes measure was having ≥4 OTPD. RESULTS: Data were collected for 1,398 standard criteria donors. Of the 1,166 (83%) who did not meet the Bundle at referral, only 254 (22%) had a positive Bundle status change. On adjusted analysis, positive Bundle status change increased the odds of achieving ≥4 OTPD significantly (odds ratio 2.04; 95% CI 1.49 to 2.81; p < 0.001). CONCLUSIONS: A positive donor management goal Bundle status change during donor hospital management is associated with a 2-fold increase in achieving ≥4 OTPD. Active critical care management of the potential organ donor, as evidenced by improvement in routinely measured critical care end points can be a means by which to substantially increase the number of organs available for transplantation.

The impact of blood transfusions in deceased organ donors on the outcomes of 1,884 renal grafts from United Network for Organ Sharing Region 5.

BACKGROUND: Historically, strategies to reduce acute rejection and improve graft survival in kidney transplant recipients included blood transfusions (BTs) before transplantation. While advents in recipient immunosuppression strategies have replaced this practice, the impact of BTs in the organ donor on recipient graft outcomes has not been evaluated. We hypothesize that BTs in organ donors after neurologic determination of death (DNDDs) translate into improved recipient renal graft outcomes, as measured by a decrease in delayed graft function (DGF). METHODS: Donor demographics, critical care end points, the use of BTs, and graft outcome data were prospectively collected on DNDDs from March 2012 to October 2013 in the United Network for Organ Sharing Region 5 Donor Management Database. Propensity analysis determined each DNDD's probability of receiving packed red blood cells based on demographic and critical care data as well as provider bias. The primary outcome measure was the rate of DGF (dialysis in the first week after transplantation) in different donor BT groups as follows: no BT, any BT, 1 to 5, 6 to 10, or greater than 10 packed red blood cell units. Regression models determined the relationship between donor BTs and recipient DGF after accounting for known predictors of DGF as well as the propensity to receive a BT. RESULTS: Data were complete for 1,884 renal grafts from 1,006 DNDDs; 52% received any BT, 32% received 1 to 5 U, 11% received 6 to 10, and 9% received greater than 10 U of blood. Grafts from transfused donors had a lower rate of DGF compared with those of the nontransfused donors (26% vs. 34%, p < 0.001). After adjusting for known confounders, grafts from donors with any BT had a lower odds of DGF (odds ratio, 0.76; p = 0.030), and this effect was greatest in those with greater than 10 U transfused. CONCLUSION: Any BT in a DNDD was associated with a 23% decrease in the odds of recipients developing DGF, and this effect was more pronounced as the number of BTs increased. LEVEL OF EVIDENCE: Therapeutic study, level III; epidemiologic/prognostic study, level II.

The impact of meeting donor management goals on the number of organs transplanted per expanded criteria donor: a prospective study from the UNOS Region 5 Donor Management Goals Workgroup.

IMPORTANCE: The shortage of organs available for transplant has led to the use of expanded criteria donors (ECDs) to extend the donor pool. These donors are older and have more comorbidities and efforts to optimize the quality of their organs are needed. OBJECTIVE: To determine the impact of meeting a standardized set of critical care end points, or donor management goals (DMGs), on the number of organs transplanted per donor in ECDs. DESIGN, SETTING, AND PARTICIPANTS: Prospective interventional study from February 2010 to July 2013 of all ECDs managed by the 8 organ procurement organizations in the southwestern United States (United Network for Organ Sharing Region 5). INTERVENTIONS: Implementation of 9 DMGs as a checklist to guide the management of every ECD. The DMGs represented normal cardiovascular, pulmonary, renal, and endocrine end points. Meeting the DMG bundle was defined a priori as achieving any 7 of the 9 end points and was recorded at the time of referral to the organ procurement organization, at the time of authorization for donation, 12 to 18 hours later, and prior to organ recovery. MAIN OUTCOMES AND MEASURES: The primary outcome measure was 3 or more organs transplanted per donor and binary logistic regression was used to identify independent predictors with P < .05. RESULTS: There were 671 ECDs with a mean (SD) number of 2.1 (1.3) organs transplanted per donor. Ten percent of the ECDs had met the DMG bundle at referral, 15% at the time of authorization, 33% at 12 to 18 hours, and 45% prior to recovery. Forty-three percent had 3 or more organs transplanted per donor. Independent predictors of 3 or more organs transplanted per donor were older age (odds ratio [OR] = 0.95 per year [95% CI, 0.93-0.97]), increased creatinine level (OR = 0.73 per mg/dL [95% CI, 0.63-0.85]), DMGs met prior to organ recovery (OR = 1.90 [95% CI, 1.35-2.68]), and a change in the number of DMGs achieved from referral to organ recovery (OR = 1.11 per additional DMG [95% CI, 1.00-1.23]). CONCLUSIONS AND RELEVANCE: Meeting DMGs prior to organ recovery with ECDs is associated with achieving 3 or more organs transplanted per donor. An increase in the number of critical care end points achieved throughout the care of a potential donor by both donor hospital and organ procurement organization is also associated with an increase in organ yield.

Determining optimal threshold for glucose control in organ donors after neurologic determination of death: a United Network for Organ Sharing Region 5 Donor Management Goals Workgroup prospective analysis.

BACKGROUND: The appropriate level of glucose control in organ donors after neurologic determination of death (DNDD) remains uncertain. We hypothesized that a glucose target of 180 mg/dL would be appropriate for optimizing organ transplantation rates and outcomes. METHODS: Demographic, critical care, organ transplantation, and graft outcome data were prospectively collected on all DNDDs in United Network for Organ Sharing (UNOS) Region 5 from 2010 to 2012. Glucose levels were assessed at four time points in the organ donation process. The primary outcome measure was having four or more organs transplanted per donor (OTPD). Univariate analyses were conducted to determine the relationship between glucose levels and OTPD, organ transplantation rates, and graft function. Multivariate analyses were performed to determine independent predictors of four or more OTPDs. Glucose levels were analyzed at the following cutoff points: 150 or less, 180, and 200 mg/dL. Results with a p < 0.05 are listed. RESULTS: A total of 1,611 DNDDs had a mean (SD) age of 38 (17) years and 3.4 (1.7) OTPDs. Forty-one percent had four or more OTPDs. Glucose levels of 150 mg/dL or less were not associated with differences in organ use. Levels of 180 mg/dL or less were associated with more OTPDs (3.5 vs. 3.2), a higher rate of four or more OTPDs (42% vs. 34%), and more heart (34% vs. 28%), pancreas (18% vs. 11%), and kidney (85% vs. 81%) use. Levels of 200 mg/dL or less revealed similar results. However, only a level of 180 mg/dL or less was an independent predictor of four or more OTPDs (odds ratio, 1.4). All three levels were associated with higher kidney graft survival after a mean (SD) of 10 (6.0) months of follow-up (97% vs. 95%). CONCLUSION: Hyperglycemia is common in DNDDs and is associated with lower organ transplantation rates and worse graft outcomes. Targeting a glucose level of 180 mg/dL or less seems to preserve outcomes and is consistent with general critical care guidelines. LEVEL OF EVIDENCE: Therapeutic study, level II.