Predicting post-kidney transplant dementia/Alzheimer's Disease risk in older patients

PROJECT SUMMARY/ABSTRACT Kidney transplantation (KT) is increasing for older adults (≥50) with ESRD. In 2021, older adults received roughly 60% of all KTs and are at increased risk of dementia/Alzheimer’s disease (AD). KT recipients who develop dementia/AD post-transplant have a 2.4-fold increased risk of mortality and a 1.5-fold increased risk of graft loss. Of older KT recipients who are diagnosed with dementia/AD, 88.6% die within 10 years. These deaths may be due to inability to perform self-care, inadequate nutrition, or medication non-adherence. Despite these risks, predicting who will develop post KT dementia/AD is not part of pre-KT evaluation. Furthermore, factors routinely measured at pre-KT evaluation (age, sex, comorbidities, etc.) have only moderate predictive power for post-KT dementia/AD. Predicting post-KT dementia/AD risk can help identify older candidates who would benefit from interventions such as cognitive prehabilitation or post-KT surveillance. Predicting post-KT dementia/AD risk at transplant evaluation provides enough time to intervene prior to KT. To design a geriatric-specific model that can predict post-KT dementia/AD risk utilizing machine learning, we will leverage an ongoing NIA-funded R01 prospective longitudinal cohort study of frailty among older KT candidates to accomplish the following aims: (1) To identify dementia/AD cases and possible subtypes among KT recipients and quantify the cumulative incidence of AD/dementia in KT recipients in this ongoing cohort study; (2) To identify clinical, geriatric, and ESRD-specific risk factors that are associated with post-KT dementia/AD; and (3) To design a model with the aid of machine learning that successfully predicts the risk of post-KT dementia/AD in older patients undergoing KT evaluation. Our group’s expertise in frailty and dementia/AD and access to the ongoing Frailty Assessment in Renal Disease (FAIR) cohort, along with Dr. Long’s training interests in machine learning and regression, provide a unique opportunity to build prediction models that could identify older candidates at highest risk of post-KT dementia/AD. We hypothesize that a risk prediction tool that incorporates traditional clinical, geriatric, and ESRD-specific risk factors that are commonly measured at KT evaluation, will improve post-KT dementia/AD risk prediction. If the proposed aims are achieved, we will improve our ability to identify older patients at increased risk of developing post-KT dementia/AD, who will need additional interventions to improve post-KT outcomes.