Agreement between high-sensitivity C-reactive protein and C-reactive protein assays

Abstract Background High sensitivity C-reactive protein (hs-CRP) is a biomarker used for risk prediction for cardiovascular disease (CVD) by assessing low concentrations of inflammatory markers. Measurements of regular CRP assays have become very sensitive too with a detection limit of 0.03mg/dL, while also being more available and cheaper. The aim of this study is to compare the association between CRP and hs-CRP. Methods This study compared CRP and hs-CRP serum concentrations and data acquired by medical chart review of 590 patients from 11/2017 to 10/2018 of our cardiology outpatient clinic who were divided into hs-CRP and CRP risk groups for CVD events: low<0.1 mg/dL, average 0.1–0.3mg/dL, high>0.3mg/dL. The agreement of classification in hs-CRP and CRP risk groups was assessed by kappa statistic, with Kappa coefficient of <0.20, 0.21–0.40, 0.41–0.60, 0.61–0.80, 0.81–0.99 interpreted as slight, fair, moderate, substantial and almost perfect agreement, respectively. Bland-Altman (BA) analysis was used to assess agreement between hs-CRP and CRP by plotting the mean of the two measurements on the x-axis against the mean difference of CRP - hs-CRP on the y-axis. Results Out of all 590 patients, 37.7% were in low risk, 33.9% in average risk and 28.5% in high risk hs-CRP group. Some group changes occurred after reclassification of the patients according to CRP measurements. Eight percent of patients reclassified into a higher risk group, 0.7% into a lower risk group, while 91.4% remained in the same risk group as determined by hs-CRP (kappa: 0.87; p<0.001) (Tab. 1). Important to note, there was a 100% agreement between the high-risk CRP and hs-CRP group patient classification. BA-plot displayed a fixed bias with an average difference between the two laboratory tests for CRP and hs-CRP of 0.02 mg/dL±0.09SD with only sporadic outliers (Fig. 1). The upper limit of agreement was 0.12 and lower limit of agreement was −0.07. In the lower range of CRP values, measurements were tightly clustered around the average difference. Greater variability could be observed at higher serum level of the inflammatory biomarker in the BA-plot with a bias to higher CRP concentrations than hs-CRP concentrations at values greater than 0.5mg/dL. This proportional bias, which was further demonstrated by linear regression analysis, does not affect the risk predicting qualities of hs-CRP or CRP for CVD because the cut-off values for risk groups (0.1mg/dL for low risk, 0.3mg/dL for high risk) are all below this threshold. Conclusion A close agreement between measurements of hs-CRP and CRP assays was identified, therefore regular CRP assays could replace hs-CRP for cardiac risk assessment. Benefits for clinical implementation are: First, CRP assessment is routinely available in most laboratories compared to hs-CRP. Second, CRP is less costly than hs-CRP, since no further laboratory acquisitions are necessary, which is especially relevant in regions where cost efficiency is of importance. Funding Acknowledgement Type of funding sources: None. Table 1. Kappa statisticFigure 1. Bland-Altman plot