Our results present the first prevalence estimates for CMV IgM and low IgG avidity in a nationally representative U.S. population. CMV IgM seroprevalence among women ranged from 2.3 to 4.5% across age groups and did not show significant trends with age. These results are consistent with the understanding that CMV IgM can be produced throughout life after primary CMV infection or as a result of reinfection or reactivation (18
) and suggest that older people may be as likely to have a recurrent episode as younger people are to have a primary infection. No risk factors based on race, ethnicity, age, or household income emerged for CMV IgM seroprevalence in contrast to risk factors such as race/ethnicity and household income for CMV IgG seroprevalence reported by Staras et al. (25
). The lack of identifiable risk factors for CMV IgM may be due to the relatively low number of observations (3.0% prevalence for IgM), and because over 80% of the IgM reactivity in our population sample was high avidity and thus presumably from nonprimary CMV infection, which may be less associated with identifiable risk than primary infection. In addition, a portion of the IgM-positive sera may have been false-positive determinations known to occur with CMV IgM testing (13
Testing all available IgM-positive sera and a sample of IgM-negative sera from the NHANES repository for IgG avidity allowed us to provide prevalence estimates for low IgG avidity ( and ) not previously reported. We applied 3 different cutoffs in the Vidas test for determination of low IgG avidity: an index value of 0.80, shown to reliably identify sera from acute CMV infection in one study (7
); an index value of 0.7, recommended as a cutoff for low avidity by some studies (1
); and an index value of 0.6. The cutoff value of 0.80 showed a strong association between low avidity and young age () but gave an unfeasibly high prevalence for low avidity of 6.6% compared to U.S. incident infection estimates for CMV of 1 to 2% (5
), indicating low specificity for identification of primary CMV infection as shown in other studies (1
). The cutoff value of 0.70 gave a low-avidity prevalence estimate of 2.0%, disproportionately excluding IgM-negative sera compared to the 0.8 test cutoff, and showed a significant association between young age and low avidity (). The index value of 0.60 resulted in a low-IgG-avidity prevalence of 0.8% with no association between low avidity and young age (). The 0.6 test cutoff likely decreased test sensitivity because 3 of the 11 specimens excluded from low avidity compared to the 0.7 cutoff were from young females 12 to 19 years old with strong CMV IgM reactivity suggestive of recent primary infection. Our results and other studies (1
) indicate that the optimal single-point cutoff for the Vidas avidity test is likely 0.7 or possibly midway between 0.6 and 0.7, which was not analyzed. Thus, 0.7 was used as the cutoff for the main results of this study. Use of a two-point cutoff with an indeterminate zone benefits many serology tests but was not applied to the three cutoffs in our study since it would have created too many result categories for analysis. The optimal test cutoff for any serology test depends on whether the test is used primarily for screening, where sensitivity is more important, or confirmation, where specificity is more important. As with nearly all serology tests, adjusting the cutoff involves a tradeoff between sensitivity and specificity of the test.
IgG avidity testing of 170 IgM-positive sera showed that high IgM antibody titers (>2.0 UA/ml) were strongly associated with low avidity and presumably primary CMV infection. This extended work by Prince and Leber (22
), who measured IgG avidity in 64 CMV IgM-positive sera that were primarily low titer or high titer and showed that the high-titer group was 93% low IgG avidity. Our study analyzed about 3 times more IgM-positive sera with titers spanning the test range and showed a steadily increasing proportion of low IgG avidity with increasing IgM antibody level, and the high-antibody group was 79% low IgG avidity.
Strengths of the study include the large national sample size and first population-based estimates for CMV IgM and IgG avidity. Limitations of the study include the cross-sectional design of NHANES, which did not allow documentation of IgG seroconversion to support low-avidity determination. The sample size of 6,000 lacked sufficient power to analyze risk factors for low avidity and primary CMV infection. The avidity analysis would have benefited from testing more of the IgM-negative sera (129/3,384 were tested), but this became evident during analysis after the serum collection had been returned to NHANES. Commercial CMV IgG avidity testing has the general limitation of being relatively new and requiring further standardization among commercial avidity tests recently highlighted by Revello et al. (23
). Borderline avidity values should be interpreted with caution and considered together with CMV IgM determination.
In summary, our findings support the understanding that CMV IgM reactivity can occur throughout life as the result of primary and nonprimary CMV infection. Our prevalence estimates for low IgG avidity suggest that at a given point in time, 14 to 18% of CMV IgM reactivity represents primary CMV infection, consistent with results from a large study in France where 10 to 16% of the IgM-positive sera identified from pregnant women were low avidity (21
). Because about 90% of the low-avidity sera were IgM positive, screening for IgM can greatly enrich the testing pool for low-IgG-avidity sera and should remain part of a testing algorithm. High IgM antibody titer was a strong predictor for low IgG avidity, which may aid the identification of primary infection and risk for congenital CMV infection, especially with the current limited availability of avidity testing. To the best of our knowledge, CMV IgG avidity testing in the United States is available at only two reference laboratories (Focus Diagnostics, Cypress, CA, and ARUP Laboratory, Salt Lake City, UT), and kits are sold through Abbott Laboratories (Abbott Park, IL) but only for use on the Abbott instrument.