Leveraging a novel health records platform to predict the development of cardiovascular disease following kidney transplantation

PROJECT SUMMARY
Cardiovascular disease (CVD) is the leading cause of death among kidney transplant (KT) recipients with a
functioning allograft. KT patients face a 3- to 5-fold higher risk of CVD morbidity and mortality than the general
population, and within three years of kidney transplantation, 11% of these patients will have had a myocardial
infarction. Evidence suggests that this increased risk is driven by multiple intersecting pathways contributing to
CVD, including the metabolic side-effects of immunosuppression medications, a history of chronic kidney
disease and volume overload, current allograft function, chronic and acute inflammation, and socioeconomic
factors such as housing and income. Despite this, a KT-specific CVD-risk prediction model incorporating known
risk factors has not been developed. Existing datasets lack the ability to capture granular CVD events, fully
characterize contributions of longitudinal biomarkers, or incorporate traditional, transplant-specific, and
socioeconomic factors in their risk estimation. Furthermore, current studies predict disparate composite CVD
outcomes confusing the interpretation of predicted risk and highlighting the lack of a standard CVD outcome to
assess burden in this population. Finally, beyond potential risk miscalculation, existing models remain largely
unused in the clinical setting as they require manual input of data into an online calculator. To address this, we
have leveraged a unique health records platform within our institution to identify a cohort of KT patients and
retrospectively capture their highly granular longitudinal data to assess CVD risk. We have successfully used
this platform to build risk prediction models for two other patient populations and embedded clinical tools into the
health record for use in real time. Thus, my proposed research strategy is to 1) quantify the cumulative incidence
of CVD events in our KT population and define the optimal compositive outcome to assess meaningful risk, 2)
identify and characterize risk factors associated with CVD after KT accounting for time-varying disease states,
longitudinal biomarker trajectories, and socioeconomic factors, and 3) implement and pilot-test an individualized
CVD-risk prediction tool embedded in our health record. The proposed work will generate a comprehensive and
transportable risk-prediction tool specific to the KT population with implications for dissemination across multiple
institutions. Our findings will allow patients and providers to engage in shared decision-making and identify
targets of intervention that will ultimately improve outcomes in this unique population. This work will be
immediately applicable to KT patients burdened with excessive CVD risk and their physicians who must optimize
the balance between maintaining allograft health and minimizing cardiovascular disease.

Grant Number: 5F30HL168842-03
NIH Institute/Center: NIH
Principal Investigator: Mary Bowring