Donation-after-circulatory-death (DCD), donation-after-brain-death (DBD), and living-donation (LD) are the three possible options for liver transplantation (LT), each with unique benefits and complication rates. We aimed to compare DCD-, DBD-, and LD-LT-specific graft survival and biliary complications (BC). We collected data on 138 DCD-, 3,027 DBD- and 318 LD-LTs adult recipients from a single center and analyzed patient/graft survival. BC (leak and anastomotic/non-anastomotic stricture (AS/NAS)) were analyzed in a subset of 414 patients. One-/five-year graft survival were 88.6%/70.0% for DCD-LT, 92.6%/79.9% for DBD-LT, and, 91.7%/82.9% for LD-LT. DCD-LTs had a 1.7-/1.3-fold adjusted risk of losing their graft compared to DBD-LT and LD-LT, respectively (p < 0.010/0.403). Bile leaks were present in 10.1% (DCD-LTs), 7.2% (DBD-LTs), and 36.2% (LD-LTs) (ORs, DBD/LD vs. DCD: 0.7/4.2, p = 0.402/<0.001). AS developed in 28.3% DCD-LTs, 18.1% DBD-LTs, and 43.5% LD-LTs (ORs, DBD/LD vs. DCD: 0.5/1.8, p = 0.018/0.006). NAS was present in 15.2% DCD-LTs, 1.4% DBDs-LT, and 4.3% LD-LTs (ORs, DBD/LD vs. DCD: 0.1/0.3, p = 0.001/0.005). LTs w/o BC had better liver graft survival compared to any other groups with BC. DCD-LT and LD-LT had excellent graft survival despite significantly higher BC rates compared to DBD-LT. DCD-LT represents a valid alternative whose importance should increase further with machine/perfusion systems.

Background

Sarcopenia has been identified as a predictive variable for surgical outcomes. We hypothesized that sarcopenia could be a key measure to identify frail patients and potentially predict poorer outcomes among recipients of simultaneous pancreas and kidney (SPK) transplants.

Methods

We estimated sarcopenia by measuring psoas muscle mass index (PMI). PMI was assessed on perioperative computed tomography (CT) scans of SPK recipients.

Results

Of the 141 patients identified between 2010 and 2018, 107 had a CT scan available and were included in the study. The median follow-up was 4 years (range, 0.5-9.1 y). Twenty-three patients had a low PMI, and 84 patients had a normal PMI. Patient characteristics were similar between the 2 groups except for body mass index, which was significantly lower in low PMI group (P < 0.001). Patient and kidney graft survival were not statistically different between groups (P = 0.851 and P = 0.357, respectively). A multivariate Cox regression analysis showed that patients with a low PMI were 6 times more likely to lose their pancreas allograft (hazard ratios, 5.4; 95% confidence intervals, 1.4-20.8; P = 0.015). Three out of 6 patients lost their pancreas graft due to rejection in the low PMI group, compared with 1 out of 9 patients in the normal PMI group. Among low PMI patients who had a follow-up CT scan, 62.5% (5/8) of those with a functional pancreas graft either improved or resolved sarcopenia, whereas 75.0% (3/4) of those who lost their pancreas graft continued to lose muscle mass.

Conclusion

Sarcopenia could represent one of the predictors of pancreas graft failure and should be evaluated and potentially optimized in SPK recipients.

Parathyroid allotransplantation is a burgeoning treatment for severe hypoparathyroidism. Deceased donor parathyroid gland (PTG) procurement can be technically challenging due to lack of normal intraoperative landmarks and exposure constraints in the neck of organ donors. In this study, we assessed standard 4-gland exposure in situ and en bloc surgical techniques for PTG procurement and ex vivo near-infrared autofluorescence (NIRAF) imaging for identification of PTGs during organ recovery.

Methods

Research tissue consent was obtained from organ donors or donor families for PTG procurement. All donors were normocalcemic, brain-dead, solid organ donors between 18 and 65 y of age. PTGs were procured initially using a standard 4-gland exposure technique in situ and subsequently using a novel en bloc resection technique after systemic organ preservation flushing. Parathyroid tissue was stored at 4 °C in the University of Wisconsin solution up to 48 h post-procurement. Fluoptics Fluobeam NIRAF camera and Image J software were utilized for quantification of NIRAF signal.

Results

Thirty-one brain-dead deceased donor PTG procurements were performed by abdominal transplant surgeons. In the initial 8 deceased donors, a mean of 1.75 glands (±1.48 glands SD) per donor were recovered using the 4-gland in situ technique. Implementation of combined en bloc resection with ex vivo NIRAF imaging in 23 consecutive donors yielded a mean of 3.60 glands (±0.4 SD) recovered per donor (P < 0.0001). Quantification of NIRAF integrated density signal demonstrated >1-fold log difference in PTG (2.13 × 10⁵ pixels) versus surrounding anterior neck structures (1.9 × 10⁴ pixels; P < 0.0001). PTGs maintain distinct NIRAF signal from the time of recovery (1.88 × 10⁵ pixels) up to 48 h post-procurement (1.55 × 10⁵ pixels) in organ preservation cold storage (P = 0.34).

Conclusions

The use of an en bloc surgical technique with ex vivo NIRAF imaging significantly enhances the identification and recovery of PTG from deceased donors.

Introduction

Donation after circulatory death (DCD) liver transplantation (LT) makes up well less than 1% of all LTs with a Model for End-Stage Liver Disease (MELD)≥35 in the United States. We hypothesized DCD-LT yields acceptable ischemia-reperfusion and reasonable outcomes for recipients with MELD≥35.

Methods

We analyzed recipients with lab-MELD≥35 at transplant within the UCSF (n=41) and the UNOS (n=375) cohorts using multivariate Cox regression and propensity score matching.

Results

In the UCSF cohort, five-year patient survival was 85% for DCD-LTs and 86% for matched-Donation after Brain Death donors-(DBD) LTs (p=0.843). Multivariate analyses showed that younger donor/recipient age and more recent transplants (2011-2021 versus 1999-2010) were associated with better survival. DCD vs. DBD graft use did not significantly impact survival (HR: 1.2, 95%CI 0.6-2.7). The transaminase peak was approximately doubled, indicating suggesting an increased ischemia-reperfusion hit. DCD-LTs had a median post-LT length of stay of 11 days, and 34% (14/41) were on dialysis at discharge versus 12 days and 22% (9/41) for DBD-LTs. 27% (11/41) DCD-LTs versus 12% (5/41) DBD-LTs developed a biliary complication (p=0.095). UNOS cohort analysis confirmed patient survival predictors, but DCD graft emerged as a risk factor (HR: 1.5, 95%CI 1.3-1.9) with five-year patient survival of 65% versus 75% for DBD-LTs (p=0.016). This difference became non-significant in a sub-analysis focusing on MELD 35-36 recipients. Analysis of MELD≥35 DCD recipients showed that donor age of <30yo independently reduced the risk of graft loss by 30% (HR, 95%CI: 0.7 (0.9-0.5), p=0.019). Retransplant status was associated with a doubled risk of adverse event (HR, 95%CI: 2.1 (1.4-3.3), p=0.001). The rejection rates at 1y were similar between DCD- and DBD-LTs, (9.3% (35/375) versus 1,541 (8.7% (1,541/17,677), respectively).

Discussion

In highly selected recipient/donor pair, DCD transplantation is feasible and can achieve comparable survival to DBD transplantation. Biliary complications occurred at the expected rates. In the absence of selection, DCD-LTs outcomes remain worse than those of DBD-LTs.

Pancreas transplantation improves glycemic control and mortality in patients with diabetes but requires aggressive immunosuppression to control the alloimmune and autoimmune response. Recent developments in "omics' methods have provided gene transcript-based biomarkers for organ transplant rejection. The tissue Common Response Module (tCRM) score is developed to identify the severity of rejection in kidney, heart, liver, and lung transplants. Still, it has not yet been validated in pancreas transplants (PT). We evaluated the tCRM score's relevance in PT and additional markers of acute cellular rejection (ACR) for PT. An analysis on 51 pancreas biopsies with ACR identified 37 genes and 56 genes significantly upregulated in the case of grade 3 and grade 2 ACR respectively (P<0.05). Significant differences were seen with higher grades of rejection among several transcripts. Of the 22 genes differentially expressed in Grade 3 ACR, 18 were also differentially expressed in Grade 2 ACR. The rejection signal was attributable to activated leukocytes' infiltration. Significantly higher tCRM scores were found in grade 3 ACR (p = 0.007) and grade 2 ACR (p = 0.004), compared to normal samples. The tCRM score was able to distinguish treatment-resistant cases from those successfully treated for rejection.

Background: Scoring systems have been proposed to select donation after circulatory death (DCD) donors and recipients for liver transplantation (LT). We hypothesized that complex scoring systems derived in large datasets might not predict outcomes locally. Methods: Based on 1-year DCD-LT graft survival predictors in multivariate logistic regression models, we designed, validated, and compared a simple index using the University of California, San Francisco (UCSF) cohort (n = 136) and a universal-comprehensive (UC)-DCD score using the United Network for Organ Sharing (UNOS) cohort (n = 5,792) to previously published DCD scoring systems. Results: The total warm ischemia time (WIT)-index included donor WIT (dWIT) and hepatectomy time (dHep). The UC-DCD score included dWIT, dHep, recipient on mechanical ventilation, transjugular-intrahepatic-portosystemic-shunt, cause of liver disease, model for end-stage liver disease, body mass index, donor/recipient age, and cold ischemia time. In the UNOS cohort, the UC-score outperformed all previously published scores in predicting DCD-LT graft survival (AUC: 0.635 vs. ≤0.562). In the UCSF cohort, the total WIT index successfully stratified survival and biliary complications, whereas other scores did not. Conclusion: DCD risk scores generated in large cohorts provide general guidance for safe recipient/donor selection, but they must be tailored based on non-/partially-modifiable local circumstances to expand DCD utilization.

Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.