C-reactive protein levels in patients on maintenance hemodialysis: reliability and reflection on the utility of single measurements.

BACKGROUND: Single C-reactive protein (CRP) values have been associated with death and cardiovascular disease in dialysis patients. We prospectively obtained multiple CRP values in stable patients, hypothesizing that values would remain stable in the absence of disease and that a single CRP value would be a reliable marker of risk.
METHODS: Four CRP values per week for three consecutive weeks were obtained in 10 clinically stable patients receiving conventional HD. Using prespecified cutoffs of 2.2 and 4.4 mg/l, the frequency of risk misclassification relative to the lowest CRP value obtained was determined. Within and between patient variability was also calculated.
RESULTS: The median age was 54 years, and the average duration of dialysis was 41 months. Nine out of ten patients had at least one abnormal CRP value (>2.2 mg/l), six had all values elevated, and seven had an abnormal median CRP. The overall coefficient of reliability was 0.63 (95% CI 0.42-0.87). The misclassification rate varied with cutoff, and ranged from 0-83% and 0-58% using upper limit of normal (ULN) and twice ULN, respectively. The within patient variability was 0.37 for the entire cohort, and 0.33 when three patients with intercurrent acute inflammation were excluded.
CONCLUSIONS: CRP exhibits short term variability in HD patients, resulting in a risk of misclassification depending on sampling time and chosen cutoff point. Single CRP values must be interpreted with caution, and multiple measurements, or use of other biomarkers, should be considered.