Making an equal system equitable: Proposing a sex-adjusted MELDNa score for liver transplantation allocation

Background & Aims: Liver allocation is determined by the model for end-stage liver disease (MELD), a scoring system based on four laboratory measurements. During the MELD era, sex disparities in liver transplant have increased and there are no modifications to MELD based on sex. We use data from electronic health records (EHRs) to describe sex differences in MELD labs and propose a sex adjustment. Methods: We extracted lab values for creatinine, International Normalized Ratio of prothrombin rate, bilirubin, and sodium from EHRs at Vanderbilt University Medical Center (VUMC) and the All of Us Research Project to determine sex differences in lab traits. We calculated MELDNa scores within liver transplant recipients, non-transplanted liver disease cases, and non-liver disease controls separately. To account for sex differences in lab traits in MELDNa scoring, we created a sex-adjusted MELDNa map which outputs adjusted female scores mapped to male scores of equal liver disease severity. Using waitlist data from the Liver Simulated Allocation Modeling, we conducted simulations to determine if the sex-adjusted scores reduced sex disparities. Results: All component MELDNa lab values and calculated MELDNa scores yielded significant sex differences within VUMC (n=623,931) and All of Us (n=56,715) resulting in MELDNa scoring that disadvantaged females who, despite greater decompensation traits, had lower MELDNa scores. In simulations, the sex-adjusted MELDNa score modestly increased female transplantation rate and decreased overall death. Conclusions: Our results demonstrate pervasive sex differences in all labs used in MELDNa scoring and highlight the need and utility of a sex-adjustment to the MELDNa protocol.

Statistical methods versus machine learning techniques for donor-recipient matching in liver transplantation

Donor-Recipient (D-R) matching is one of the main challenges to be fulfilled nowadays. Due to the increasing number of recipients and the small amount of donors in liver transplantation, the allocation method is crucial. In this paper, to establish a fair comparison, the United Network for Organ Sharing database was used with 4 different end-points (3 months, and 1, 2 and 5 years), with a total of 39, 189 D-R pairs and 28 donor and recipient variables. Modelling techniques were divided into two groups: 1) classical statistical methods, including Logistic Regression (LR) and Naïve Bayes (NB), and 2) standard machine learning techniques, including Multilayer Perceptron (MLP), Random Forest (RF), Gradient Boosting (GB) or Support Vector Machines (SVM), among others. The methods were compared with standard scores, MELD, SOFT and BAR. For the 5-years end-point, LR (AUC = 0.654) outperformed several machine learning techniques, such as MLP (AUC = 0.599), GB (AUC = 0.600), SVM (AUC = 0.624) or RF (AUC = 0.644), among others. Moreover, LR also outperformed standard scores. The same pattern was reproduced for the others 3 end-points. Complex machine learning methods were not able to improve the performance of liver allocation, probably due to the implicit limitations associated to the collection process of the database.