Superiority of Serum Cystatin C Over Creatinine in Prediction of Long-Term Prognosis at Discharge From ICU

Abstract

Objectives: 

Renal outcomes after critical illness are seldom assessed despite strong correlation between chronic kidney disease and survival. Outside hospital, renal dysfunction is more strongly associated with mortality when assessed by serum cystatin C than by creatinine. The relationship between creatinine and longer term mortality might be particularly weak in survivors of critical illness.

Design: 

Retrospective observational cohort study.

Patients: 

In 3,077 adult ICU survivors, we compared ICU discharge cystatin C and creatinine and their association with 1-year mortality. Exclusions were death within 72 hours of ICU discharge, ICU stay less than 24 hours, and end-stage renal disease.

Interventions: 

None.

Measurements and Main Results: 

During ICU admission, serum cystatin C and creatinine diverged, so that by ICU discharge, almost twice as many patients had glomerular filtration rate less than 60 mL/min/1.73 m² when estimated from cystatin C compared with glomerular filtration rate estimated from creatinine, 44% versus 26%. In 743 patients without acute kidney injury, where ICU discharge renal function should reflect ongoing baseline, discharge glomerular filtration rate estimated from creatinine consistently overestimated follow-up glomerular filtration rate estimated from creatinine, whereas ICU discharge glomerular filtration rate estimated from cystatin C well matched follow-up chronic kidney disease status. By 1 year, 535 (17.4%) had died. In survival analysis adjusted for age, sex, and comorbidity, cystatin C was near-linearly associated with increased mortality, hazard ratio equals to 1.78 (95% CI, 1.46–2.18), 75th versus 25th centile. Conversely, creatinine demonstrated a J-shaped relationship with mortality, so that in the majority of patients, there was no significant association with survival, hazard ratio equals to 1.03 (0.87–1.2), 75th versus 25th centile. After adjustment for both creatinine and cystatin C levels, higher discharge creatinine was then associated with lower long-term mortality.

Conclusions: 

In contrast to creatinine, cystatin C consistently associated with long-term mortality, identifying patients at both high and low risk, and better correlated with follow-up renal function. Conversely, lower creatinine relative to cystatin C appeared to confer adverse prognosis, confounding creatinine interpretation in isolation. Cystatin C warrants further investigation as a more meaningful measure of renal function after critical illness.