Glynn RJ， MacFadyen JG， Ridker PM.
    Center for Cardiovascular Disease Prevention， Division of Preventive Medicine， Division of Cardiovascular Medicine，   Brigham and Women’s Hospital， Harvard Medical School， Boston， MA 02215， USA. rglynn@rics.bwh.harvard.edu
Abstract
    BACKGROUND: The JUPITER (Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin) trial suggests that increased high-sensitivity C-reactive protein (hsCRP) concentrations may be useful in decisions about the initiation of statin therapy for primary prevention of vascular disease. Although studies of specific populations have suggested that hsCRP is a reliable longitudinal marker， it is unclear how strongly hsCRP tracks in individuals after a single increased concentration.
    METHODS: We evaluated tracking of hsCRP in 8901 individuals randomized to placebo in the JUPITER trial. These individuals had screening LDL cholesterol concentrations <130 mg/dL (<3.37 mmol/L) and hsCRP concentrations > or =2 mg/L， with subsequent hsCRP measurements made before randomization; at 13 weeks; 1， 2， 3， and 4 years later; and at trial termination. Longitudinal trends and associations were evaluated nonparametrically with box plots and Spearman correlations. After data transformation to achieve normality， repeated-measures regression models estimated the intraclass correlation of hsCRP， with and without controlling for known demographic， lifestyle， and medical determinants of hsCRP concentration. For comparison， we evaluated tracking of systolic and diastolic blood pressure; total， LDL， and HDL cholesterol; and fasting triglycerides.
    RESULTS: The median hsCRP concentration in these untreated individuals showed modest regression to the mean over time， declining from 3.8 mg/L at randomization to 3.4 mg/L at 4 years. Tracking correlations for hsCRP over time were comparable to those for blood pressure and LDL cholesterol， but lower than those for HDL， fasting triglycerides， and total cholesterol. The intraclass correlation for repeated hsCRP measurements was 0.54 (95% CI， 0.53-0.55) without covariate adjustment and 0.50 (95% CI， 0.49-0.51) after adjustment for demographic， lifestyle， and comorbidity determinants.
CONCLUSIONS: Concentrations of hsCRP show strong tracking， even after selection of individuals with initially high values. Without statin therapy， increased concentrations of hsCRP generally remain high over time.