Evaluation of a sensitive immunoluminometric assay for the determination of C-reactive protein (CRP) in serum and plasma and the establishment of reference ranges for different groups of subjects

A sensitive immunoluminometric assay originally designed to measure C-reactive protein (CRP) in neonates and minimal serum volumes was adapted to measure this protein in a routine method without prior sample dilution. The concentration range covered without prior dilution was 10 micrograms/l to 20 mg/l using a sample volume of 5 microliters serum and a total assay time of less than 2 h. Serum samples were assayed from participants in a community medicine programme (SHIP--Study of Health in Pomerania) of the University of Greifswald, Germany (n = 414), as well as from mother-child pairs at birth (n = 30) and women attending the infertility clinic (n = 36). The validation of the assay was compared with a commercial latex-enhanced turbidimetric immunoassay (Roche Diagnostics--Integra 700) using routine serum samples (n = 60) from hospital patients. Comparison was made with the routine assay used in the SHIP study (Roche Diagnostics--Hitachi 717/Tina Quant). From 414 SHIP samples measured in the immunoluminometric assay, 289 were below the detection level in the turbidimetric (Tina Quant) assay. A significant positive correlation (p < 0.01) between log C-reactive protein concentration with age was found, both in the non-screened (all CRP values) (n = 414, r = 0.222) and selected (CRP < 5.00 mg/l = 90th percentile) (n = 370, r = 0.242) SHIP participants. Women were found to have significantly higher CRP levels than men (women: median age 47 a, median CRP 1.29 mg/l; men: median age 55 a, median CRP 1.00 mg/l--p = 0.016) in the non-selected SHIP participants. The situation was different in the selected group, (median age: men 54 a, women 48 a) where no significant difference in median CRP values between the sexes was seen (men: 0.874 mg/l, women 0.951 mg/l, p = 0.206). The distribution of CRP values in a "Normal Healthy Population" is skewed (mean/median--SHIP: all--2.08; selected--1.49). From the 414 SHIP samples measured in the immunoluminometric assay, 289 were below the detection level (2.5 mg/l) in the turbidimetric (Tina Quant) assay. From the 125 remaining samples the correlation between both methods was acceptable (r = 0.813), the regression line y = a + bx being: CRP (ILMA) = 1.83 + 0.842*CRP (Tina Quant). The Tina Quant assay gave values significantly higher than the ILMA in the range 2.5-25 mg/l CRP (p < 0.001). The total information loss in 289/414 subjects with a CRP < 2.5 mg/l with the Tina Quant assay makes it no longer suitable for epidemiological studies in which CRP is to be studied as a risk factor for cardiovascular events. The comparison between the immunoluminometric assay and the latex-enhanced immunoturbidimetric assay (Roche Integra) was much better. The latter measured down to less than 0.3 mg/l, thus being more suitable for epidemiological studies than the Tina Quant assay from the same producer. The correlation and regression data between the ILMA (x) and the Roche Integra assay (y) were: r = 0.971; CRP (Roche Integra) = 0.635 + 0.984*CRP (ILMA); n = 50.10 sera with CRP levels between 25 and 460 mg/l showed no high-dose hook effect in either assay. The remaining 50 sera were measurable in both assays. The turbidimetric assay gave rise to marginally but significantly higher values than the immunoluminometric assay (p = 0.004). The mothers at birth had a median CRP of 3.64 mg/l (range 1.49-12.6 mg/l), the neonates a median CRP of 34 micrograms/l (range 4-288 micrograms/l). All births were without complications, with gestational periods between 38 and 42 weeks. There was no correlation between maternal and neonatal CRP at birth. Mothers at birth had significantly higher CRP levels than healthy non-pregnant women (p < 0.001). Women attending the infertility clinic had CRP-values similar to age-matched healthy non-pregnant women (median 0.698 mg/l, range 0.05-9.97 mg/l). Interassay coefficients of variation at CRP concentrations of 0.85 and 7.9 mg/l were 8.99 and 7.93%, respectively, for the immunoluminometric