Improved assessment of donor liver steatosis using Banff consensus recommendations and deep learning algorithms.

BACKGROUND & AIMS: The Banff Liver Working Group recently published consensus recommendations for steatosis assessment in donor liver biopsy, but few studies reported their use and no automated deep-learning algorithms based on the proposed criteria have been developed so far. We evaluated Banff recommendations on a large monocentric series of donor liver needle biopsies by comparing pathologists' scores with those generated by convolutional neural networks (CNNs) we specifically developed for automated steatosis assessment. METHODS: We retrospectively retrieved 292 allograft liver needle biopsies collected between January 2016 and January 2020 and performed steatosis assessment using a former intra-institution method (pre-Banff method) and the newly introduced Banff recommendations. Scores provided by pathologists and CNN models were then compared, and the degree of agreement was measured with the intraclass correlation coefficient (ICC). RESULTS: Regarding the pre-Banff method, poor agreement was observed between the pathologist and CNN models for small droplet macrovesicular steatosis (ICC: 0.38), large droplet macrovesicular steatosis (ICC: 0.08), and the final combined score (ICC: 0.16) evaluation, but none of these reached statistically significance. Interestingly, significantly improved agreement was observed using the Banff approach: ICC was 0.93 for the low-power score (p <0.001), 0.89 for the high-power score (p <0.001), and 0.93 for the final score (p <0.001). Comparing the pre-Banff method with the Banff approach on the same biopsy, pathologist and CNN model assessment showed a mean (±SD) percentage of discrepancy of 26.89 (±22.16) and 1.20 (±5.58), respectively. CONCLUSIONS: Our findings support the use of Banff recommendations in daily practice and highlight the need for a granular analysis of their effect on liver transplantation outcomes. IMPACT AND IMPLICATIONS: We developed and validated the first automated deep-learning algorithms for standardized steatosis assessment based on the Banff Liver Working Group consensus recommendations. Our algorithm provides an unbiased automated evaluation of steatosis, which will lay the groundwork for granular analysis of steatosis's short- and long-term effects on organ viability, enabling the identification of clinically relevant steatosis cut-offs for donor organ acceptance. Implementing our algorithm in daily clinical practice will allow for a more efficient and safe allocation of donor organs, improving the post-transplant outcomes of patients.

Tailoring allocation policies and improving access to paediatric liver transplantation over a 16-year period.

BACKGROUND & AIMS: Mortality on the paediatric liver transplantation (pLT) waiting list (WL) is still an issue. We analysed the Italian pLT WL to evaluate the intention-to-treat (ITT) success rate and to identify factors influencing success. METHODS: All children (<18 years) listed for pLT in Italy between 2002-2018 were included (Era 1 [2002-2007]: centre-based allocation; Era 2 [2008-2014]: national allocation; Era 3 [2015-2018]: national allocation+mandatory-split policy). RESULTS: A total of 1,424 patients (median age: 2.0 [IQR 1.0-9.0] years; median weight: 12.0 kg [IQR 7-27]) were listed for pLT. Median WL time was 2 days (IQR 1-5) for Status 1 and 44 days (IQR 15-120) for non-Status 1 patients; 1,302 children (91.4%) were transplanted (67.3% with split grafts), while 50 children (3.5%) dropped off the WL (2.5% death, 1.0% clinical deterioration). Predictive factors for receiving LT included Status 1 (hazard ratio [HR] 1.66, p = 0.001), Status 1B (HR 1.96, p = 0.016), Status 2A (HR 2.15, p = 0.024) and each 1-point increase in PELD/MELD score. Children with recipient's weight >25 kg, blood group O or awaiting pLT combined with other organs had less chance of being transplanted. ITT patient survival rates were 90.5% at 1 year and 87.5% at 5 years, remaining stable across eras. Risk factors for ITT survival were re-transplantation (HR 5.83, p <0.001), Status 1 (HR 2.28, p = 0.006), Status 1B (HR 2.90, p = 0.014), Status 2A (HR 9.12, p <0.001), recipient weight <6 kg (HR 4.53, p <0.001) and low-volume activity (HR 4.38, p = 0.001). CONCLUSIONS: In Italy, continuous adaption of paediatric organ allocation policies via the introduction of national allocation, paediatric prioritisation rules and a mandatory-split policy have helped maximise the use of donors for paediatric candidates and to minimise WL mortality without compromising outcomes. IMPACT AND IMPLICATIONS: Globally, paediatric liver transplant candidates still suffer from high mortality. Over recent decades, the continuous adaption of organ allocation policies in Italy has led to excellent outcomes for children awaiting liver transplantation. The mortality rate of paediatric liver transplant candidates has been minimised to almost zero, mainly using grafts from deceased donors. Paediatric prioritisation rules, national organ exchange organisation and a mandatory-split liver policy have resulted in a unique allocation model for paediatric liver transplant candidates and represent a landmark for the paediatric transplant community.

Viability assessment of livers donated after circulatory determination of death during normothermic regional perfusion.

BACKGROUND: Abdominal normothermic regional perfusion (A-NRP) allows in-situ reperfusion and recovery of abdominal organs metabolism in donors after circulatory death (DCD). Besides improving liver transplantation outcomes, liver injury and function can be assessed during A-NRP. METHODS: To refine liver viability assessment during A-NRP, prospectively collected data of controlled DCD donors managed at our Institution between October 2019 and May 2022 were retrospectively analyzed. Baseline characteristics, procedural variables and A-NRP parameters of donors whose liver was successfully transplanted were compared to those of donors whose liver was discarded. RESULTS: Twenty-seven donors were included and in 20 (74%) the liver was accepted (positive outcome). No differences between study groups were observed concerning baseline characteristics and warm ischemia times (WIT). Initial lactate levels were positively correlated with functional WIT (r2 = 0.4, p = 0.04), whereas transaminase levels were not. Blood flow during A-NRP was comparable, whereas oxygen consumption (VO2 ) was significantly higher in the positive outcome group after 1 h. Time courses of lactate, AST and ALT were significantly different between study groups (p < 0.001). Donors whose liver was accepted showed faster lactate clearance, a difference which was amplified by normalizing lactate clearance to oxygen delivery (DO2 ) and VO2 . Lactate clearance was correlated to transaminase levels and DO2 -normalized lactate clearance was the parameter best discriminating between study groups. CONCLUSIONS: DO2 -normalized lactate clearance may represent an element of liver viability assessment during A-NRP.

Viability assessment and transplantation of fatty liver grafts using end-ischemic normothermic machine perfusion.

End-ischemic viability testing by normothermic machine perfusion (NMP) represents an effective strategy to recover liver grafts having initially been discarded for liver transplantation (LT). However, its results in the setting of significant (≥30%) macrovesicular steatosis (MaS) have not been specifically assessed. Prospectively maintained databases at two high-volume LT centers in Northern Italy were searched to identify cases of end-ischemic NMP performed to test the viability of livers with MaS ≥ 30% in the period from January 2019 to January 2022. A total of 14 cases were retrieved, representing 57.9% of NMP and 5.7% of all machine perfusion procedures. Of those patients, 10 (71%) received transplants. Two patients developed primary nonfunction (PNF) and required urgent re-LT, and both were characterized by incomplete or suboptimal lactate clearance during NMP. PNF cases were also characterized by higher perfusate transaminases, lower hepatic artery and portal vein flows at 2 h, and a lack of glucose metabolism in one case. The remaining eight patients showed good liver function (Liver Graft Assessment Following Transplantation risk score, -1.9 [risk, 13.6%]; Early Allograft Failure Simplified Estimation score, -3.7 [risk, 2.6%]) and had a favorable postoperative course. Overall, NMP allowed successful transplantation of 57% of livers with moderate-to-severe MaS. Our findings suggest that prolonged observation (≥6 h) might be required for steatotic livers and that stable lactate clearance is a fundamental prerequisite for their use.

Clinical assessment of liver metabolism during hypothermic oxygenated machine perfusion using microdialysis.

BACKGROUND: While growing evidence supports the use of hypothermic oxygenated machine perfusion (HOPE) in liver transplantation, its effects on liver metabolism are still incompletely understood. METHODS: To assess liver metabolism during HOPE using microdialysis (MD), we conducted an open-label, observational pilot study on 10 consecutive grafts treated with dual-HOPE (D-HOPE). Microdialysate and perfusate levels of glucose, lactate, pyruvate, glutamate, and flavin mononucleotide (FMN) were measured during back table preparation and D-HOPE and correlated to graft function and patient outcome. RESULTS: Median (IQR) MD and D-HOPE time was 228 (210, 245) and 116 (103, 143) min. Three grafts developed early allograft dysfunction (EAD), with one requiring retransplantation. During D-HOPE, MD glucose and lactate levels increased (ANOVA = 9.88 [p = 0.01] and 3.71 [p = 0.08]). Their 2nd-hour levels were higher in EAD group and positively correlated with L-GrAFT score. 2nd-hour MD glucose and lactate were also positively correlated with cold ischemia time, macrovesicular steatosis, weight gain during D-HOPE, and perfusate FMN. These correlations were not apparent when perfusate levels were considered. In contrast, MD FMN levels invariably dropped steeply after D-HOPE start, whereas perfusate FMN was higher in dysfunctioning grafts. CONCLUSION: MD glucose and lactate during D-HOPE are markers of hepatocellular injury and could represent additional elements of the viability assessment.

Heterogeneous indications and the need for viability assessment: An international survey on the use of machine perfusion in liver transplantation.

Although machine perfusion (MP) is being increasingly adopted in liver transplantation, indications, timing, and modality are debated. To investigate current indications for MP a web-based Google Forms survey was launched in January 2021 and addressed to 127 experts in the field, identified among first and corresponding Authors of MP literature in the last 10 years. The survey presented 10 real-life cases of donor-recipient matching, asking whether the liver would be accepted (Q1), whether MP would be used in that particular setting (Q2) and, if so, by which MP modality (Q3) and at what timing during preservation (Q4). Respondents could also comment on each case. The agreement was evaluated using Krippendorff's alpha coefficient. Answers from 39 (30.1%) participants disclosed significant heterogeneity in graft acceptance, MP indications, technique, and timing. Agreement between respondents was generally poor (Q1, α = 0.11; Q2, α = 0.14; Q3, α = 0.12, Q4, α = 0.11). Overall, respondents preferred hypothermic MP and an end-ischemic approach in 56.3% and 81.1% of cases, respectively. A total of 18 (46.2%) participants considered only one MP approach, whereas 17 (43.6%) and 3 (7.7%) considered using alternatively 2 or 3 different techniques. Of 38 comments, 17 (44.7%) were about the use of MP for graft viability assessment before implantation. This survey shows considerable variability in MP indications, emphasizing the need to identify scenarios of optimal utilization for each technique. Viability assessment emerges as a fundamental need of transplant professionals when considering the use of MP.

Impact of MELD 30-allocation policy on liver transplant outcomes in Italy.

BACKGROUND & AIMS: In Italy, since August 2014, liver transplant (LT) candidates with model for end-stage liver disease (MELD) scores ≥30 receive national allocation priority. This multicenter cohort study aims to evaluate time on the waiting list, dropout rate, and graft survival before and after introducing the macro-area sharing policy. METHODS: A total of 4,238 patients registered from 2010 to 2018 were enrolled and categorized into an ERA-1 Group (n = 2,013; before August 2014) and an ERA-2 Group (n = 2,225; during and after August 2014). A Cox proportional hazards model was used to estimate the hazard ratio (HR) of receiving a LT or death between the two eras. The Fine-Gray model was used to estimate the HR for dropout from the waiting list and graft loss, considering death as a competing risk event. A Fine-Gray model was also used to estimate risk factors of graft loss. RESULTS: Patients with MELD ≥30 had a lower median time on the waiting list (4 vs.12 days, p <0.001) and a higher probability of being transplanted (HR 2.27; 95% CI 1.78-2.90; p = 0.001) in ERA-2 compared to ERA-1. The subgroup analysis on 3,515 LTs confirmed ERA-2 (odds ratio 0.56; 95% CI 0.46-0.68; p = 0.001) as a protective factor for better graft survival rate. The protective variables for lower dropouts on the waiting list were: ERA-2, high-volume centers, no competition centers, male recipients, and hepatocellular carcinoma. The protective variables for graft loss were high-volume center and ERA-2, while MELD ≥30 remained related to a higher risk of graft loss. CONCLUSIONS: The national MELD ≥30 priority allocation was associated with improved patient outcomes, although MELD ≥30 was associated with a higher risk of graft loss. Transplant center volumes and competition among centers may have a role in recipient prioritization and outcomes. CLINICAL TRIAL NUMBER: NCT04530240 LAY SUMMARY: Italy introduced a new policy in 2014 to give national allocation priority to patients with a model for end-stage liver disease (MELD) score ≥30 (i.e. very sick patients). This policy has led to more liver transplants, fewer dropouts, and shorter waiting times for patients with MELD ≥30. However, a higher risk of graft loss still burdens these cases. Transplant center volumes and competition among centers may have a role in recipient prioritization and outcomes.

Fully automated quantitative assessment of hepatic steatosis in liver transplants.

BACKGROUND: The presence of macro- and microvesicular steatosis is one of the major risk factors for liver transplantation. An accurate assessment of the steatosis percentage is crucial for determining liver graft transplantability, which is currently based on the pathologists' visual evaluations on liver histology specimens. METHOD: The aim of this study was to develop and validate a fully automated algorithm, called HEPASS (HEPatic Adaptive Steatosis Segmentation), for both micro- and macro-steatosis detection in digital liver histological images. The proposed method employs a hybrid deep learning framework, combining the accuracy of an adaptive threshold with the semantic segmentation of a deep convolutional neural network. Starting from all white regions, the HEPASS algorithm was able to detect lipid droplets and classify them into micro- or macrosteatosis. RESULTS: The proposed method was developed and tested on 385 hematoxylin and eosin (H&E) stained images coming from 77 liver donors. Automated results were compared with manual annotations and nine state-of-the-art techniques designed for steatosis segmentation. In the TEST set, the algorithm was characterized by 97.27% accuracy in steatosis quantification (average error 1.07%, maximum average error 5.62%) and outperformed all the compared methods. CONCLUSIONS: To the best of our knowledge, the proposed algorithm is the first fully automated algorithm for the assessment of both micro- and macrosteatosis in H&E stained liver tissue images. Being very fast (average computational time 0.72 s), this algorithm paves the way for automated, quantitative and real-time liver graft assessments.

Noninvasive assessment of coronary artery bypass graft patency by multislice computed tomography.

BACKGROUND: Follow-up of coronary artery bypass after cardiac surgery is routinely performed by means of X-ray coronary angiography. However, this is an invasive procedure, expensive and includes ionizing radiation exposure, hospitalization and a small risk of complications. Multislice computed tomography is a noninvasive diagnostic tool that permits the visualization of the cardiac structures, including the coronary arteries. The purpose of our study was to compare multislice computed tomography with conventional angiography for the evaluation of graft patency following cardiac surgery. METHODS: Forty-seven asymptomatic patients (44 men and 3 women, mean age 67 +/- 7 years) who had undergone coronary bypass surgery at least 10 years previously, were retrospectively investigated by means of ECG-gated multislice computed tomography, within 6 months of coronary angiography. RESULTS: Overall, 116 out of the possible 127 (91.4%) grafts were assessable at computed tomography, including 87 saphenous vein grafts, 26 left internal mammary artery, 2 right internal mammary artery, and 1 gastroepiploic artery. Coronary angiography showed that 79 of 116 grafts (68.1%) were patent and that 37 (31.9%) were occluded. All grafts which were patent and occluded at coronary angiography were correctly identified at multislice computed tomography, with a sensitivity and specificity of 100%. CONCLUSIONS: Multislice computed tomography with retrospective gating permits an accurate and noninvasive evaluation of coronary artery bypass patency, and could replace conventional angiography for the follow-up of asymptomatic, stable patients.