Research Spotlight

Nitski, O., Azhie, A., Qazi-Arisar, F.A., Wang, X., Ma, S., Lilly, L., Watt, K.D., Levitsky, J., Asrani, S.K., Lee, D.S., Rubin, B.B., Bhat, M. and Wang, B. (2021). “Long-term mortality risk stratification of liver transplant recipients: real-time application of deep learning algorithms on longitudinal data.” Lancet Digit Health 3(5): e295-e305.

Full text of this article.

BACKGROUND: Survival of liver transplant recipients beyond 1 year since transplantation is compromised by an increased risk of cancer, cardiovascular events, infection, and graft failure. Few clinical tools are available to identify patients at risk of these complications, which would flag them for screening tests and potentially life-saving interventions. In this retrospective analysis, we aimed to assess the ability of deep learning algorithms of longitudinal data from two prospective cohorts to predict complications resulting in death after liver transplantation over multiple timeframes, compared with logistic regression models. METHODS: In this machine learning analysis, model development was done on a set of 42 146 liver transplant recipients (mean age 48·6 years [SD 17·3]; 17 196 [40·8%] women) from the Scientific Registry of Transplant Recipients (SRTR) in the USA. Transferability of the model was further evaluated by fine-tuning on a dataset from the University Health Network (UHN) in Canada (n=3269; mean age 52·5 years [11·1]; 1079 [33·0%] women). The primary outcome was cause of death, as recorded in the databases, due to cardiovascular causes, infection, graft failure, or cancer, within 1 year and 5 years of each follow-up examination after transplantation. We compared the performance of four deep learning models against logistic regression, assessing performance using the area under the receiver operating characteristic curve (AUROC). FINDINGS: In both datasets, deep learning models outperformed logistic regression, with the Transformer model achieving the highest AUROCs in both datasets (p<0·0001). The AUROC for the Transformer model across all outcomes in the SRTR dataset was 0·804 (99% CI 0·795-0·854) for 1-year predictions and 0·733 (0·729-0·769) for 5-year predictions. In the UHN dataset, the AUROC for the top-performing deep learning model was 0·807 (0·795-0·842) for 1-year predictions and 0·722 (0·705-0·764) for 5-year predictions. AUROCs ranged from 0·695 (0·680-0·713) for prediction of death from infection within 5 years to 0·859 (0·847-0·871) for prediction of death by graft failure within 1 year. INTERPRETATION: Deep learning algorithms can incorporate longitudinal information to continuously predict long-term outcomes after liver transplantation, outperforming logistic regression models. Physicians could use these algorithms at routine follow-up visits to identify liver transplant recipients at risk for adverse outcomes and prevent these complications by modifying management based on ranked features. FUNDING: Canadian Donation and Transplant Research Program, CIFAR AI Chairs Program.

Bhakta, D., Patel, M., Ma, T.W., Boutté, J., Sarmast, N. and Asrani, S.K. (2021). “Model for End Stage Liver Disease-Lactate score and prediction of inpatient mortality in critically ill patients with cirrhosis.” Liver Transpl

Full text of this article.

The burden of decompensated liver disease is high with a significant proportion of cirrhosis patients requiring inpatient hospitalization. Patients with cirrhosis have high inpatient mortality in the intensive care unit (ICU) setting, especially as compared to other chronic conditions. Objective models that predict short-term mortality at the time of ICU admission are needed for assessing response to therapy, transplant candidacy as well as futility of care.

Asrani, S.K. and Nadim, M.K. (2021). “Chronic Kidney Disease after Simultaneous Liver Kidney Transplantation: Refining Patient Selection.” Liver Transpl.

Full text of this article.

The number of simultaneous liver kidney transplantations (SLKT) performed in the United States have steadily increased over the last 15 years: approximately 1 out of 10 liver transplantations is a dual organ transplant.(1) Given the disparity between availability of donor organs and recipients awaiting transplant and the increasing numbers of patients on the waitlist with MELD ≥ 40 with the majority having acute kidney injury (AKI), the number of patients who will qualify for SLKT will most likely continue to increase.(2) Over the years, the transplant community has refined SLKT listing criteria in order to ensure appropriate utilization of kidney organs given the increasing numbers of SLKT.(2, 3) In 2017, new criteria were formalized to set up a “sustained AKI” and “chronic kidney disease” (CKD) criteria for listing for SLKT.

Asrani, S.K., Mellinger, J., Arab, J.P. and Shah, V.H. (2021). “Reducing the Global Burden of Alcohol-Associated Liver Disease: A Blueprint for Action.” Hepatology 73(5): 2039-2050.

Full text of this article.

Alcohol-associated liver disease (ALD) is a major driver of global liver related morbidity and mortality. There are 2.4 billion drinkers (950 million heavy drinkers) and the lifetime prevalence of any alcohol use disorder (AUD) is 5.1%-8.6%. In 2017, global prevalence of alcohol-associated compensated and decompensated cirrhosis was 23.6 million and 2.5 million, respectively. Combined, alcohol-associated cirrhosis and liver cancer account for 1% of all deaths worldwide with this burden expected to increase. Solutions for this growing epidemic must be multi-faceted and focused on both population and patient-level interventions. Reductions in ALD-related morbidity and mortality require solutions that focus on early identification and intervention, reducing alcohol consumption at the population level (taxation, reduced availability and restricted promotion), and solutions tailored to local socioeconomic realities (unrecorded alcohol consumption, focused youth education). Simple screening tools and algorithms can be applied at the population level to identify alcohol misuse, diagnose ALD using non-invasive serum and imaging markers, and risk-stratify higher-risk ALD/AUD patients. Novel methods of healthcare delivery and platforms are needed (telehealth, outreach, use of non-healthcare providers, partnerships between primary and specialty care/tertiary hospitals) to proactively mitigate the global burden of ALD. An integrated approach that combines medical and AUD treatment is needed at the individual level to have the highest impact. Future needs include (1) improving quality of ALD data and standardizing care, (2) supporting innovative healthcare delivery platforms that can treat both ALD and AUD, (3) stronger and concerted advocacy by professional hepatology organizations, and (4) advancing implementation of digital interventions.

Askar, M. and Hanna, R. (2021). “A Step Toward Comprehensive Transplant Solutions for Aplastic Anemia.” Transplantation 105(5): 955-957.

Full text of this article.

In their article published in this issue, Park et al6 investigate the timely and important question of which alternative donor source is more advantageous for transplantation in SAA patients. One of the remarkable findings of this study is the comparable outcomes among SAA patients who received transplants from HLA-matched (8/8) donors, HLA-mismatched (7/8) unrelated donors, and haploidentical donors using total body irradiation (TBI) based regimen in terms of 3-y overall survival (OS), failure-free survival (FFS), cumulative incidence of graft-failure, transplant-related mortality (TRM), and graft versus host disease (GVHD). [No abstract; excerpt from article].