Laboratory evaluation of the new Access® cytomegalovirus immunoglobulin IgM and IgG assays

Results Total CV is lower than 10% and 12% for Access CMV IgG and CMV IgM. The IgG method is linear ( R 2 = 0.999) with recoveries between 85% and 108%. Correlation between Access and VIDAS CMV IgG is highly significant ( P < 0.001). Observed agreement was 97.4% for CMV IgG and 93.7% for CMV IgM. Relative sensitivity and specificity was 97.2% and 100% for IgG and 100% and 97.4% for IgM. Kinetics of the antibody response measured with Access methods is significantly higher ( P < 0.02) when compared with VIDAS and probably easier to interpret. Prevalence of a recent infection was 0.85% and CMV IgM non-specific rate was 2.9%. Conclusion Good sensitivity and specificity and pronounced anti-CMV antibody response make the Access CMV IgG and IgM tests suitable for screening prenatal CMV infections. Keywords Cytomegalovirus CMV IgG CMV IgM Prevalence Sensitivity Specificity Serial measurements 1 Background Human cytomegalovirus is the most common cause of congenital infection occurring with an incidence between 1% and 2% of all live births. 1 CMV infection is endemic. 2 In Europe, 45% of pregnant women are sero-positive at the beginning of pregnancy. 3 The in utero transmission of CMV can take place during primary maternal infection or non-primary infection of sero-positive mothers, but the transmission rate to the fetus is much higher for non-immune mothers. 4 Naturally acquired immunity results in a 69% reduction in the risk of congenital CMV infection in future pregnancies. 5 Therefore prenatal screening for CMV infection is important for correct follow-up and counseling of the expectant mother. In CMV-sero-negative pregnant women, precautionary hygienic measures can be suggested. In pregnant women with a primary CMV infection, closer follow-up of the fetus by ultrasound or invasive investigation may be considered for further investigation. 6 2 Objectives The present study evaluates the technical and diagnostic performance of the Beckman Coulter Access CMV IgG and IgM assays. The comparative devices used were the bioMérieux VIDAS ® CMV IgG and IgM assays. VIDAS ® CMV IgG avidity test was used for dating of the infection and for analyzing discrepant results. 3 Study design 3.1 Samples Five different groups of samples were collected from prenatal screening and from frozen selected samples. Group 1 consists of 346 fresh random routine serum samples from pregnant women. Group 1 is used to evaluate the correlation and concordance between the DxI and VIDAS CMV IgG results and for estimating relative sensitivity and specificity for the Access CMV IgG method. Group 2 consists of 491 specimens, composed out of three subgroups (a–c). Group 2 is used for determining global agreement between the CMV IgM methods. Group 2a is 382 routine samples with negative VIDAS CMV IgM results. Group 2a is selected to evaluate specificity of the Access CMV IgM test. Group 2b is composed out of 43 frozen serum samples from pregnant women with a proven primary infection (positive IgG, a grey-zone avidity index with a median of 0.46, a minimum of 0.29 and maximum of 0.59 and a proven serum conversion). Group 2b is selected to evaluate the sensitivity of the Access CMV IgM test. Group 2c consists of 66 frozen samples from women with a proven old infection. Criteria were: positive IgG with intermediate to high avidity (median avidity index = 0.8, SD = 0.07, minimum = 0.63, and maximum = 0.94), positive IgM or prior clinical information. Group 2c is selected to study the sensitivity for detecting residual CMV IgM. In Group 3 serial measurements on consecutive serum samples from seven (IgG) and eleven women (IgM) showing evidence of a recent primary infection were collected for comparing the kinetics of the anti-CMV antibodies response to the Access and VIDAS IgG and IgM methods. A total of 50 and 72 tests were performed for IgG and IgM respectively. Maximum time of the follow-up period was 191 days. In Group 4, 55 samples from patients infected with other viruses or concerned with abnormal autoimmune diseases were selected to study cross-reactivity and estimate analytical specificity of the Access CMV methods. In Group 5, 3992 samples from antenatal screening were measured with the Access CMV IgG and IgM methods. Positive CMV IgM results were further analyzed with VIDAS CVM IgM and IgG avidity tests. This survey was performed for estimating the Access CMV IgM specificity and prevalence of a recent infection in a routine situation. 3.2 Methods The Beckman Coulter Access CMV IgG and CMV IgM assays are paramagnetic particle chemiluminescent assays for semi-quantitative and qualitative determination of IgG and IgM antibodies to cytomegalovirus in human serum and plasma. The paramagnetic particles are coated with CMV AD169 purified viral lysate that serves to capture antibodies to CMV. The assays were performed on Unicel ® DxI 800 Access Immunoassay system. Reportable range of Access CMV IgG is 0–400 AU/mL, depending on the highest value of the calibrator. Sera with IgG above 400 AU/mL were diluted with wash buffer II. CMV IgG results ≥15 AU/mL are positive and results between 11 and 15 AU/mL are equivocal. Access CMV IgM results are calculated as signal to cut-off ratio (S/CO) value. A signal to cut-off ratio higher than 1 is considered to be reactive and a ratio of 0.8–1 is equivocal. The Biomérieux VIDAS CMV IgG and IGM methods are quantitative and qualitative enzyme linked fluorescent immunoassays, using solid phase receptacles coated with CMV antigen (strain AD169), derived from cell cultures. These methods have been used in the laboratory since more than 10 years and therefore considered as reference in this study. Samples with VIDAS CMV IgG ≥ 6 AU/mL are positive and results between 4 and 6 AU/mL are equivocal. The VIDAS CMV IgG method is linear up to 400 AU/mL. A VIDAS CMV IgM index >0.9 is positive. Indexes between 0.7 and 0.9 are equivocal. VIDAS IgG avidity test is used for discriminating between recent (younger than 3 months) and old infections. CMV IgG antibodies with avidities of ≤0.2 were considered to be representative for recent infections. IgG antibodies with avidities of ≥0.8 were considered to be indicative for infections dating back more than 3 months. 3.3 Performance evaluation Precision of the Access IgG and IgM tests was determined with the CLSI EP5-A2 protocol using the manufacturer's positive and negative controls. During a minimum of 20 days, two runs per day with two replicates per run were analyzed of each control level. Linearity of the anti-IgG CMV assay was evaluated using three high positive samples with expected levels of 583, 549 and 360 AU/mL. Samples were diluted 1/2 to 1/64. Two replicates of each dilution were measured. Expected and observed values were compared. Relative sensitivity and specificity of Access IgG were studied using fresh random samples. Sensitivity is estimated as the proportion of confirmed VIDAS positives correctly identified by the Access method and specificity is the proportion of VIDAS negatives correctly identified by the Access test. Equivocal samples were excluded from calculations. Relative Sensitivity and specificity of Access IgM were studied using sera stored at −20 °C, previously analyzed with VIDAS. Storage did not exceed 1 year, warranting stability of the samples. Kinetics of CMV IgG and IgM antibodies is different. IgG antibodies steadily increase with time and does not start to return to their initial level during the period of study. To summarize the IgG serial data, the % maximum difference with the first observation ((maximum difference/first value) × 100), and a time weighted value (=area under the curve divided by the length of the time interval) are estimated for each subject. When the first IgG value was zero it was replaced by the limit of blank value. 7 CMV IgM antibodies start from a baseline value, rise to a peak, and then may return to baseline. To summarize IgM data a time weighted value for each woman is calculated. 3.4 Statistical analysis The CMV IgG methods were compared with non-parametric Passing Bablok regression and Spearman correlation. Observed agreement, expected agreement by chance, 8 a chi-square value and a prevalence-adjusted bias-adjusted Kappa statistic PABAK, 9 calculated as [( k × p ) − 1]/( k − 1), where k is the number of categories and p is the proportion matching's, were estimated to study method agreement. PABAK is based on Cohen's kappa and measures the agreement beyond what would be expected by chance. 9 When PABAK is one, there is perfect agreement. When PABAK is zero, there is no agreement better than chance. 10 Summary statistics 11 describing serial responses are analyzed with the non-parametric Mann–Whitney test. Statistical significance is set at P < 0.05. ROC curve analysis was performed for determining lowest IgM value giving a specificity of 100% for a recent infection and IgM value giving the best combination of sensitivity and specificity. 4 Results 4.1 Precision Within-run CVs of both Access IgG and IgM are lower than 6% ( Table 1 ). Total CVs are smaller than 10% for IgG and 12% for IgM and are significantly smaller than the claim of the manufacturer and in compliance with specifications. 4.2 Linearity Percentage recovery of Access CMV IgG was good, ranging from 85.5% to 108%, with a mean value of 98% (95%CI: ±2.8%). Regression curve ( Fig. 1 ) obtained by plotting serial dilutions was linear ( R 2 = 0.999), confirming reliable results are obtained, diluting over range samples. 4.3 Comparison between CMV IgG methods Although substantial scatter ( Fig. 2 ), correlation with the VIDAS IgG assay is acceptable R s = 0.87. Access results are more than 10 times higher, indicating standardization may be desirable. Observed agreement in the CMV IgG group is 97.4% (95%CI: 95.1–98.8%) ( Table 2 ). Nine samples were discordant and negative with Access. Four (median = 7.5 AU/mL, SD = 1.7 AU/mL) and five (median = 4.1 AU/mL, SD = 0.53 AU/mL) were positive and equivocal with VIDAS. Two of the four positive VIDAS samples were confirmed with the AxSYM CMV IgG method. Relative sensitivity and specificity of the Access CMV IgG method were 97.2% (95%CI: 93.0–99.2%) and 100% (95%CI: 98.5–100%) respectively. 4.4 Comparison between CMV IgM methods Agreement with the VIDAS method, except in the residual IgM group ( P = 0.04), is highly significant ( P < 0.001) ( Table 2 ). Relative sensitivity of the Access CMV IgM method is 100%. Relative specificity is 97.4% (95%CI: 95.2–98.7%). Four Access results were equivocal (median = 0.88 (S/CO); SD = 0.05) and six were positive (median value = 1.19 (S/CO), SD = 0.22) in the negative VIDAS group. Agreement in the residual CMV IgM group is 68.2%(95%CI: 55.5–79.1%) and moderate according to PABAK ( Table 2 ). Twenty-one observations in this group were discordant. Overall agreement between IgM methods is 93.7% (95%CI: 91.2–95.7%). 4.5 Serial measurements Fig. 3 shows individual IgG responses of seven women with a recent infection (avidity < 0.8). Table 3 summarizes the results. IgG antibody responses to both methods are significantly different ( P = 0.0175). Although in sample 5, Access IgG appeared more slowly (still negative after 41 days; Access = 5 AU/mL and VIDAS = 10 AU/mL), IgG response is much more pronounced with Access. Time weighted average CMV IgM values are significantly different between methods ( P < 0.001) ( Fig. 4 ). Maximum Access IgM values can rise to around 20 (S/CO), whereas VIDAS maximum is around two. Nevertheless, in only two results, different interpretation is obtained. In sample 2, VIDAS (=0.33 index) returned faster to negative than Access (=2.15 (S/CO)). In sample 6, VIDAS was negative at the first observation (VIDAS = 0.55 index and Access = 1 (S/CO)). As with IgG, Access IgM response is much more pronounced. 4.6 ROC analysis The ability of the Access CMV IgM test to discriminate primo-infections from negative cases is evaluated using ROC curve analysis. 12 Area under the curve is one, which means separation between negatives and positives is perfect ( Fig. 5 ). An IgM value = 1.2 gives the best combination of sensitivity (100%) and specificity (99.5%). IgM = 1.62 (S/CO) is the lowest value giving a relative specificity of 100%. At this level relative sensitivity is 95.4% (95%CI: 84.2–99.4). This means, IgM = 1 (S/CO) is chosen to favor sensitivity. 4.7 Interferences When tested with predefined 10 sera positive for Rubella IgM, 10 sera positive for Toxo IgM and 10 sera positive for Rheumatoid factor, one serum positive for Hepatitis A IgM, 4 sera positive for Parvovirus, 13 sera positive for EBV and 7 sera with monoclonal IgM proteins, Access CMV IgM achieved a analytical specificity of 87.3%, whereas VIDAS CMV IgM specificity was 89.0%. Relative specificity of the Access CMV IgG method compared with VIDAS CMV IgG negative results was 100%. 4.8 Survey of prenatal routine samples: prevalence of recent infection and % non-specific CMV IgM Three groups ( Table 4 ) having significantly different IgM values ( P < 0.0001) were distinguished in the 187 positive Access IgM samples ( Table 4 ). Of these samples, 116 (2.9%) were considered having remaining IgM from a non-primary infection (negative VIDAS, high avidity). Prevalence of a recent infection was 0.9% and in accordance with recent estimates, indicating between 0.1% and 2% of pregnancies being infected. 13 5 Discussion Precision of the new Access methods and linearity of the Access CMV IgG assay are good and comparable with published results, demonstrating reliable performance. 14–17 There is no gold standard for CMV antibodies. Therefore, Access methods were compared with methods used in the laboratory for several years. Hence only relative agreements can be reported. Agreement with comparative methods was good. Sensitivity was identical and specificity of Access CMV IgM was slightly lower. A possible explanation for the observed inter-assay variability is the use of non-standardized native viral antigens (AD169 strain), purified from cell culture-derived lysates. 18–20 Further standardization of CMV serology may be realized using specific combinations of native and recombinant antigens (such as IE1). 21 For both CMV IgG and IgM, similar good agreements were reported between VIDAS and Access. 15–17 For more difficult samples such as HIV patients analytical specificity of the Access CMV methods is reported as 100% (IgG) 15 and 95.5% (IgM). 22 Observed relative specificity of Access IgM compared with VIDAS is 97.4 and claimed specificity of the VIDAS test is 98.8%. 23 Therefore, a realistic estimate of Access CMV IgM specificity should be around 96% (≈97.4 × 98.8%). Consequently additional testing, taking a second sample, performing IgG and eventually IgG avidity tests, may help defining clinical status of the patient. 24,25 In a prospective survey of 3992 pregnant women, 0.85% of the women had a recent CMV infection. Median IgM in this group is 6.7 (S/CO). Three % of these women had CMV IgM not linked with a primary infection. Interpretation of the clinical status of these samples was not easy and additional testing (IgG, VIDAS IgM, and IgG avidity) was done to conclude. A finding that may help interpreting infection status is absolute Access CMV IgM antibody level. Highest IgM antibodies are observed in women with a recent infection, intermediate levels in old infections and lowest positivity with confirmed non-specific CMV IgM antibodies. Further we observed 80% of women having positive IgM antibodies, have values <1.64 (S/CO). A ROC analysis, done with data from Group 2, showed 100% specificity for a primo-infection is achieved at IgM values > 1.62 (S/CO). Therefore status of women with CMV IgM values < 1.64 may be equivocal. At the beginning of a primo-infection, IgM titers can be <1.64 (S/CO). We observed Access CMV IgM antibody kinetics is very pronounced and therefore, for pregnant women with a first Access CMV IgM result between 1 and 1.64 (S/CO), a follow-up sample after a few weeks will provide more than likely an answer. Accepting a specificity of 97%, a sensitivity of 100% and a prevalence of 0.85% for a recent infection (this study), the positive predictive value of the Access IgM method is 22% and the negative predictive value is 100%. Thus, in prenatal screening using Access CMV IgM one out of five positive samples will be a recent infection. The Access CMV IgM and IgG kits showed good performance and meet requirements for routine screening for presence of CMV antibodies. Despite good performance of CMV laboratory methods, interpretation of CMV serology remains complex and challenging. Conflict of interest statement Ronald Bailly and Guido Vranken work for Analis S.A., a company that has the distribution rights for Beckman Coulter immunoanalyser systems and reagents in Belgium. Acknowledgement We thank Beckman Coulter and Fabrice Bouniort for supplying the Access reagents. References 1 H.L. Brown M.P. Abernathy Cytomegalovirus infection Semin Perinatol 22 1998 260 266 2 E.S. Mocarski T. Schenk R.F. Pass Cytomegalovirus D.M. Knipe P.M. Howley Fields virology 5th ed. 2007 Lippincott Williams and Wilkins Philadelphia, PA 2701 2772 3 A.Z. Azam Y. Vial C.L. Fawer J. Zufferey P. Hohlfeld Prenatal diagnosis of congenital cytomegalovirus infection Obstet Gynecol 97 2001 443 448 4 S. Stagno R.J. Whitley Herpesvirus infections of pregnancy. Part I. Cytomegalovirus and Epstein-barr virus infections N Engl J Med 313 1985 1270 1274 5 K.B. Fowler S. Stagno R.F. Pass Interval between births and risk of congenital cytomegalovirus infection Clin Infect Dis 38 2004 1035 1037 6 T. Lazzarotto The best practices for screening, monitoring, and diagnosis of Cytomegalovirus disease. Part II. Clinical microbiology Newsletter 32 2010 9 15 7 D.A. Armbruster T. Pry Limit of blank, limit of detection and limit of quantitation Clin Biochem Rev 29 Suppl 1 2008 S49 S52 8 D.J. Sheskin Handbook of parametric and nonparametric statistical procedures 4th ed. 2007 Chapman and Hall CRC, Boca Raton 9 T. Byrt J. Bishop J.B. Carlin Bias, prevalence and kappa J Clin Epidemiol 46 1993 423 429 10 J.R. Landis G.G. Koch The measurement of observer agreement for categorical data Biometrics 33 1977 159 174 11 J.N. Matthews D.G. Altman M.J. Campbell P. Royston Analysis of serial measurements in medical research BMJ 300 1990 230 235 12 M.H. Zweig G. Campbell Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine Clin Chem 39 1993 561 577 13 M.A.B. van der Sande S. Kaye D.J.C. Miles P. Waight D.J. Jeffries O.O. Ojuola Risk factors for and clinical outcome of congenital cytomegalovirus infection in a Peri-Urban West-African Birth Cohort PLoS ONE 2 6 2007 e492 10.1371/journal.pone.0000492 14 Beckman Coulter. Access Immunoassay Systems. CMV IgG. Fullerton, USA 2008; WB A62000B, 13 pp. 15 N. Groleau B. Rousse V. Potelle M. Bullenger J. Poropane R. Falcou-Briatte Analytical performance of the new Access ® CMV IgG assay Clin Chem Lab Med 47 Special Suppl. 2009 S252 16 L. Grangeot-Keros C. Vauloup-Fellous N. Gaidot M. Rawlins N. Groleau J. Ritchie Multicenter evaluation of the CMV IgM assay on the family of Access ® immunoassay systems Clin Microbiol Infect 16 Suppl. 2 2010 S326 17 L. Grangeot-Keros N. Gaidot M. Rawlins O. Flecheux J.C. Ritchie C. Darte Multi-center evaluation of the CMV IgG assay on the family of Access ® immunoassay systems from Beckman Coulter Clin Chem Lab Med 47 Special Suppl. 2009 S249 18 E. Mendelson Y. Aboudy Z. Smetana M. Tepperberg Z. Grossman E. Mendelson Reprod Toxicol 21 2006 350 382 19 A. Daiminger U. Bäder M. Eggers T. Lazzarotto G. Enders Evaluation of two novel enzyme immunoassays using recombinant antigens to detect cytomegalovirus-specific immunoglobulin M in sera from pregnant women J Clin Virol 13 1999 161 171 20 B. Genser M. Truschnig-Wilders D. Stünzner M.P. Landini G. Halwachs-Baumann Evaluation of five commercial enzyme immunoassays for the detection of human cytomegalovirus-specific IgM antibodies in the absence of a commercially available gold standard Clin Chem Lab Med 39 January 2001 62 70 21 C. Busse A. Strubel P. Schnitzler Combination of native and recombinant cytomegalovirus antigens in a new ELISA for detection of CMV-specific antibodies J Clin Virol 43 October 2008 137 141 22 N. Groleau B. Rousse V. Potelle M. Bullenger J. Poropane C. Chandelier Performance of the Access ® CMV IgM assay. Diagnostics in virology Clin Microbiol Inf 16 Suppl. 2 2010 S297 23 BioMérieux. VIDAS ® CMV IgM. Marcy-L’Etoile, France; 2008. 06637 I, 5 pp. 24 K. Lagrou M. Bodeus M. Van Ranst P. Goubau Evaluation of the new architect cytomegalovirus immunoglobulin M (IgM), IgG, and IgG avidity assays J Clin Microbiol 47 2009 1695 1699 25 M.G. Revello G. Gerna Diagnosis and management of human cytomegalovirus infection in the mother, fetus, and newborn infant Clin Microbiol Rev 15 2002 680 715