Accurate assessment of kidney function is fundamental to the care of all patients, but current methods to estimate GFR have error rates of 35-60% in important populations as well as variable accuracy across race, ethnicity and geography, leading to errors in critical decisions, drug dosing, and determination of prognosis. Current GFR estimates include a coefficient for Black vs non-Black which may restrict access to care. As a result, there are critical knowledge gaps in evaluation and management of GFR in health and disease. Our goal is to develop a valid and robust GFR estimate optimized for an individual person that meets the critical unmet need for a confirmatory test. The confirmatory test will use a panel of endogenous filtration markers to estimate GFR (peGFR) from a single blood sample using an equation that does not require serum creatinine or demographic characteristics. Our research team has extensive experience in biomarker evaluation, GFR estimation, epidemiology, laboratory science and metabolomics. Our preliminary data provide proof of concept that a panel of novel metabolites that does not include serum creatinine or demographics can eliminate bias and greatly improve precision of GFR estimates in patients with heart failure and reduced muscle mass; and that novel techniques can be used to produce a high-accuracy and parsimonious prediction model.
Aim 1 Using global metabolomic discovery (~1000+ metabolites) on five cohorts (N=2583), we will select candidate metabolites based on maximal joint association with mGFR as well as biological and physiological assessment of their properties as filtration markers. Markers that show acceptable analytical properties in initial testing will be incorporated into a liquid chromatography tandem mass spectrometer (LC-MS/MS) multiplex assay.
Aim 2 : We propose to use a spectrum of novel approaches, such as marginalizing predictions to down- weight outlier metabolites and kernel nearest neighbor weighted average predictions, to maximize precision as well as robustness across multiple diverse populations and to compare these approaches to a benchmark model developed using superlearner ensemble modeling techniques. Development and external validation in ~3036 and ~3465 participants across 8 and 12 studies, respectively.
Aim 3 : We propose to evaluate the impact of panel eGFR in patients with heart and liver failure (N=1796) on clinical decisions and outcomes. The expected outcome is development of panel eGFR that can be used as a confirmatory test and ultimately incorporated into clinical practice guidelines. The proposal is highly innovative as it is comprehensive spanning discovery, validation, assessment of clinical utility in populations not previously been studied and use of novel statistical methods to compute individualized GFR estimates. The proposal is significant because it will enable generalizable and accurate individualized GFR estimates in health and disease, particularly diseases with unacceptably GFR estimation high error rates and diverse racial, ethnic and geographic groups.

Public Health Relevance

Kidney function is usually estimated from markers that are measured in the blood, but current estimates are less accurate for people with chronic illness, such as those with heart or liver failure, or diverse racial, ethnic and geographic groups. The current project will fill this important gap in knowledge by developing a simple method to estimate kidney function from markers that can be easily measured from a single blood draw and that is more accurate than current estimates for all people across the continuum of health and disease. More accurate kidney function estimates will enable better clinical decision making for patients, more rigorous research designs, and better estimates of the public health burden of chronic kidney disease in the United States.

National Institute of Health (NIH)
National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)
Research Project (R01)
Project #
Application #
Study Section
Kidney, Nutrition, Obesity and Diabetes Study Section (KNOD)
Program Officer
Chan, Kevin E
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Tufts University
United States
Zip Code