The dengue ELISA in this study was performed by the same skilled persons, working in two laboratories, one of which is a virological reference laboratory in an industrialized country (EMC) and another in a resource poor setting (CR), using the same operating procedures and commercially available test assays. The conditions in this study are a realistic reflection of routine diagnostics with commercially available assays in endemic settings.
The diagnostic classification applied in this study is based on the different time course of antibody concentrations after primary and secondary dengue virus infections In the acute stage of secondary infections, IgM may be detected in all patients, but after convalescence, IgM concentrations may be lower or even undetectable[16,17] In this study the second sample was taken more than three weeks after onset of fever. An increase of IgG without detectable IgM is thus compatible with secondary dengue but it may also indicate another flavivirus infection.
Cross reactivity with other flavivirusses is a potential source of error. Test kits are usually validated against a set of well defined patients and healthy individuals. In endemic regions such as Vietnam however, repeated dengue and other flavi-virus infections, such as JEB, are common. Other studies have attempted to enhance the distinction with other flavivirusses by applying higher cut off values, thereby losing sensitivity[2,5] This study did not address cross reactivity with JEB virus but cases with clinically evident encephalitis were excluded. Although the clinical pattern of JEB is different compared to dengue virus infections we cannot completely rule out that some of the reactivity measured in serum and on filter paper was due to the serological cross-reactivity between both viruses
Inter-laboratory disagreement on serological results is not uncommon. The interpretation of serological tests is often difficult and random and systematic errors may occur in the serological confirmation of dengue in endemic as well as in non-endemic areas [18-20] This is partly due to differences between the used test kits but also to other technical aspects, among which the optical density (OD) ELISA plate readers Variability in OD values is usually accounted for by making reference to a control However, the relation between "real" extinction and measured OD values is not always linear, especially not at the extremes of its measurement range. The cut off value, the distinction between positive and negative, is ideally set at a value where most OD ELISA plate readers show linear characteristics. Nonetheless, in this study the inter-laboratory variability also occurred in the range around the cut off index value, 1. Although this was mainly a random error, with almost equal variability within both laboratories, with a small systematic compound (the intercept of the regression lines), it led to differences in classification of positive and negative in the two laboratories. Other potential sources of variability and error, related to laboratory procedures, were not uncovered in this study.
Differences in storage time did not explain the variability in this study, neither for serum nor for filter papers. Other studies have indicated that filter papers should be stored in a refrigerator but that this does not completely prevent a decline of antibody concentrations[9,10] It was also shown that in particular the anti-dengue IgM in filter paper blood spots from secondary cases decreased significantly on storage In this study filter papers were not stored in a refrigerator but in an air-conditioned room. This may be an advantage because it avoids temperature differences and thus condensation of moisture on the papers. In addition, it simulates the reality of using filter papers, usually under conditions where electrical power is not available. We paid special attention to keeping the filter papers dry and found no significant decay of antibody contents.
Filter paper index values tended to be lower than the corresponding serum values. Several studies have shown that the antibody titers measured in serum eluted from filter paper blood were 1 or 2 dilution factors lower compared to serum, especially for IgM antibodies[10,12,22].The ratio of serum and filter paper concentrations is sometimes used as conversion factor but in this study it was mainly a random variation so that we could not calculate a conversion factor to standardize the filter paper results. A problem with collection of blood on filter paper is that the amount of reconstituted serum is variable. It depends on the subject's hematocrite and on several technical aspects of reconstitution There are approaches to get around this variability. For example, in neonatal screening for PKU and hypothyroidism the test characteristics are set to a high sensitivity with the possibility of confirmation by a blood test of all positive cases Such a two stage screening is probably not suitable for the diagnosis of acute infectious diseases such as dengue.
Public health specialists are especially interested in early detection of outbreaks, notably of complicated dengue, which is mainly secondary dengue. This study shows that especially the value of dengue virus IgM detection on t0 and t3 with filter papers may lead to diagnostic misclassification and is thus of limited value for clinical confirmation. But, the results of measuring IgG in serum and filter papers eluates showed a rather good agreement with almost equal misclassification in both directions (Fig 1, panel C) Thus, with sufficient sample sizes and while taking cross reactivity with other flavivirusses, notably JEB virus, into account, dengue virus sero-prevalence studies, based on the detection of IgG antibodies eluted from filter papers, are a useful tool for epidemiologists.