This study was approved by the Institutional Review Boards (IRB) of the CDC and Emory University (CDC IRB study protocol #4630). Written informed consent for participation in the study was obtained from parents or guardians for the children in this study. We enrolled children between the ages of 6 and 60 months who were having blood drawn for other purposes at four Emory University outpatient pediatric clinics. The enrollment questionnaire collected information on age, sex, attendance in daycare (institutional or home-based), race/ethnicity, and type of health insurance as a marker for socioeconomic status (SES). We did not have information on the timing of CMV infection for any children since CMV is typically an asymptomatic infection. We did not have information on whether any of the 13 seropositive children may have been born with CMV, but this would have low probability since the congenital CMV infection rate is approximately 0.65% (1 per 154 infants) [2].

There was a one-time collection of 3 bodily fluids: whole blood, urine, and saliva. Whole blood was collected through venipuncture into EDTA Vacuatiners (Becton-Dickinson, New Jersey, USA) with a portion processed for antibody testing. CMV IgG and IgM status were determined by VIDAS. End-point titer was performed by immunofluorescense assay (IFA) (MBL-Bion, Des Plaines, IL) because the IFA requires far less specimen volume than the VIDAS. IgG avidity was measured with 3 tests (Euroimmun, Vidia, and VIDAS) because of reports of low concordance among commercial CMV avidity tests [10]. Interpretation of the Euroimmun and Vidia tests followed manufacturer’s instructions. Interpretation of the VIDAS was adjusted to provide improved sensitivity [11–13]. Final avidity determination was based on agreement of two or more tests results. Saliva was collected with the Oracol foam swab (Malvern Medical Developments Ltd, Worcester, England). Urine was collected by having children urinate into a specimen cup with their mothers helping as needed. Total DNA was extracted from whole blood, urine, and saliva using the MagNAPure with DNA extraction kits (Roche Diagnostics) followed by Taqman-based PCR [14, 15]. Viral copy numbers were normalized to the input volume for each specimen so that viral loads were comparable across the 3 bodily fluids. Statistical significance for various associations was calculated using Fisher’s exact test (two-tailed, 95% confidence interval). Differences in CMV viral load by specimen type were compared using a two-sided t-test with unequal variances.

Total enrollment was 48 children; 54% were male and 46% were female; 18% were 6–11 months old, 74% were 1–2 years old, and 8% were 3–4 years old. Blood was collected from all 48 enrolled children, saliva was collected from 46/48 (96%) children, and urine was collected from only 26/48 (54%) children because many of them were unable to produce urine during the 30 minute clinic visit. VIDAS determinations for IgG and IgM were the same as those by IFA. CMV IgG was detected in 13/48 (27%) of the children (Table 1) all of whom were over 12 months old, eliminating the possible interference of maternal antibody. Their racial/ethnic makeup and SES are shown on Table 1. Three of the IgG-positive children were CMV IgM-positive. Among IgM positive sera, two were low IgG avidity and one was low-intermediate IgG avidity (Table 1), indicating that all three represented primary infection. Two out of 3 IgM-positive sera had virus in bodily fluids that might suggest a correlation between the presence of IgM and the presence of shedding. Avidity results for the 3 commercial tests were in complete agreement for 10/13 sera and had imperfect agreement for the following 3 borderline sera: H1 had two intermediate values and one high value; D6 had two low values and one intermediate value; D8 had two intermediate values and one low value. Results of the Euroimmun test were in the majority result for all sera and are shown in Table 1. Ten of the 13 antibody-positive children (69%) had CMV DNA in one or more bodily fluid. Shedding in urine is not shown on Table 1 because of the low number of specimens collected. The prevalence of CMV DNA was 62% (8/13) for saliva and 46% (6/13) for blood (Table 1). Geometric mean viral loads were 3727 copies/ml for urine, 4449 copies/ml for saliva, and 172 copies/ml for blood. The difference in mean viral loads was statistically significant for saliva and blood (p = 0.01). There were too few urine samples to include them in analysis. The 3 positive urine samples had the following viral loads: (Table 1 child numbers) D4: 27,756 copies/ml, D6: 667 copies/ml, and D7: 2,798 copies/ml. None of the 35 CMV IgG-negative children had CMV DNA in any bodily fluid. Enrollment in daycare showed a significant association with CMV viral shedding (p = 0.037). Low IgG antibody titer, defined as endpoint titer 1:20–1:80 (range 1:20 – 1:1280), was associated with the absence of shedding (p = 0.014). Age, race/ethnicity, and socioeconomic status showed no association with CMV antibody positivity or shedding; however, the small sample size had limited ability to show associations.Table 1

CMV antibody titers, IgG avidity status, and viral loads for the CMV seropositive children according to daycare attendance