Assessment of Cystatin C Level for Risk Stratification in Adults With Chronic Kidney Disease

IMPORTANCE: Kidney function is usually estimated from serum creatinine level, whereas an alternative glomerular filtration marker (cystatin C level) associates more closely with future risk of cardiovascular disease (CVD) and mortality. OBJECTIVES: To evaluate whether testing concordance between estimated glomerular filtration rates based on cystatin C (eGFRcys) and creatinine (eGFRcr) levels would improve risk stratification for future outcomes and whether estimations differ by age. DESIGN, SETTING, AND PARTICIPANTS: A prospective population-based cohort study (UK Biobank), with participants recruited between 2006-2010 with median follow-up of 11.5 (IQR, 10.8-12.2) years; data were collected until August 31, 2020. Participants had eGFRcr greater than or equal to 45 mL/min/1.73 m(2), albuminuria (albumin <30 mg/g), and no preexisting CVD or kidney failure. EXPOSURES: Chronic kidney disease status was categorized by concordance between eGFRcr and eGFRcys across the threshold for hronic kidney disease (CKD) diagnosis (60 mL/min/1.73 m(2)). MAIN OUTCOMES AND MEASURES: Ten-year probabilities of CVD, mortality, and kidney failure were assessed according to CKD status. Multivariable-adjusted Cox proportional hazards models tested associations between CVD and mortality. Area under the receiving operating curve tested discrimination of eGFRcr and eGFRcys for CVD and mortality. The Net Reclassification Index assessed the usefulness of eGFRcr and eGFRcys for CVD risk stratification. Analyses were stratified by older (age 65-73 years) and younger (age <65 years) age. RESULTS: There were 428 402 participants: median age was 57 (IQR, 50-63) years and 237 173 (55.4%) were women. Among 76 629 older participants, there were 9335 deaths and 5205 CVD events. Among 351 773 younger participants, there were 14 776 deaths and 9328 CVD events. The 10-year probability of kidney failure was less than 0.1%. Regardless of the eGFRcr, the 10-year probabilities of CVD and mortality were low when eGFRcys was greater than or equal to 60 mL/min/1.73 m(2); conversely, with eGFRcys less than 60 mL/min/1.73 m(2), 10-year risks were nearly doubled in older adults and more than doubled in younger adults. Use of eGFRcys better discriminated CVD and mortality risk than eGFRcr. Across a 7.5% 10-year risk threshold for CVD, eGFRcys improved case Net Reclassification Index by 0.7% (95% CI, 0.6%-0.8%) in older people and 0.7% (95% CI, 0.7%-0.8%) in younger people; eGFRcr did not add to CVD risk estimation. CONCLUSIONS AND RELEVANCE: The findings of this study suggest that eGFRcr 45 to 59 mL/min/1.73 m(2) includes a proportion of individuals at low risk and fails to capture a substantial proportion of individuals at high-risk for CVD and mortality. The eGFRcys appears to be more sensitive and specific for CVD and mortality risks in mild CKD.