|
|
|||||||||||||
 |
 |
Literature review > Issue 9 > Review on Laeyendecker et al. |
|||||||||||
|
|
The paper by Laeyendecker et al.
has been designed to test two objectives: The study used stored sera from 744 subjects, aged 15 to 19 years old, from a previous randomized control trial in the Rakai district, Uganda, from 1994 to 1998. This study compared the ELISA against a well established 'gold standard' of Western Blot (WB) analysis which was previously verified against culture-proven genital herpes infections. These comparisons allowed the sensitivity and specificity to be calculated for the ELISA and a receiver operating curve was determined. Previous evaluation of this ELISA in California, USA showed that it was 100% sensitive and greater than 95% specific [1]. In this earlier study of this ELISA, low positive results (between index values 1.1 and 3.0) were rare and these often represented false positive results. Another study testing this ELISA on HIV positive men in the California, USA found low positive results in 7% [2]. This study showed that in the Ugandan setting, using the manufacturer's index cut-off, the ELISA had a sensitivity of 99% and a specificity of 52%. In addition, 18.1% of the samples were found to have a low positive result. The HIV status of patients with low positive results is not given in the paper. It can be inferred from the data presented in this paper that the low positive findings are likely to have arisen from the HIV positive patients, who made up a third of this sample (n = 248). In those patients positive for HSV-2 by WB, there appears to be no difference in the median index value for the ELISA between HIV negative and positive patients. The authors then infer that the HIV status does not affect the performance of the test. The second half of the paper argues that the cut-off of the ELISA needs to be adjusted in this study population in order to minimize the number of false positive results. It is suggested that the "best index cut-off value to optimize the assay performance in this population was 3.4, with a sensitivity of 84.9% and a specificity of 84.2%. Such a cut-off is calculated as having the probability of a false-positive result of 8%. This is proposed to be acceptable cut-off for epidemiological research. The authors also propose that a cut-off of 2.2 would be acceptable for those working in a resource poor setting with no second confirmatory WB test. Such a cut-off would have a sensitivity of 95% and specificity of 74.4% generating a false-positive rate of 11%. The authors then state "the possibility exists that there are antibodies in African sera that react to the gG2 protein on which the HSV-2 ELISA is based, either for genetic reasons or from reaction to other endemic infections in the region." This paper further proposes that despite previous data demonstrating that the gG2 epitope is highly conserved, that this may not be the case in Uganda or other developing countries. If the ELISA low positive results are caused by HIV infection (leading to false positive diagnoses) then this challenges the authors' claims that HIV does not affect the assay. The authors may be correct that genetics of the patient or variations in the gG2 epitope may affect the result but the most likely cause that must be excluded is that one third of the patients in the study were HIV positive. The fact that 7% of HIV positive males in the USA had a low positive ELISA result, which was rare in the original study, suggests that HIV is related to this phenomenon. It is biologically plausible that HIV could generate false positive results, since it is known to produce immune dysfunction. Certainly if HSV-2 Western blot-positive samples are used, then the ELISA is not affected by HIV status. This is, however, the reverse of standard clinical practice, where the WB is used as a second confirmatory test. The pre-selection of HSV-2 positive samples to test the assay generates false reassurance about the accuracy of the test and ignores the low positive ELISA results described earlier. The authors propose that the ELISA is used alone in a resource poor setting with an elevated cut-off to accommodate the low positive results (possibly produced by HIV co-infection) in a population with a HIV prevalence of 33%. Without WB, it remains possible that erroneous results will occur. It would have been interesting if this paper had analysed the low positive ELISA results further by HIV status. It may be that, in HIV negative patients in Uganda, the ELISA performs as well as the original study. If this were the case, then suggesting the use of the ELISA test alone in HIV negative patients, without a second confirmatory WB test, would be more valuable. This design of this study does not allow
it to answer both of its objectives: Secondly, the evidence that the ELISA performance is not affected by HIV serostatus is only tested on those samples pre-selected to be HSV-2 positive by WB analysis. This sub-analysis ignores the diagnostic problem caused by low positive ELISA results described above and therefore is unable to state conclusively that the assay is not affected by HIV status. References: 1. Prince HE, Ernst CE, Hogrefe WR. Evaluation of an enzyme immunoassay for measuring herpes simplex virus (HSV) type 1- and HSV type 2-specific IgG antibodies. J Clin Lab Anal 2000;14:14-16. 2. Safrin S, Arvin A, Mills J, Ashley R. Comparison of the Western immunoblot and a glycoprotein G enzyme immunoassay for detection of serum antibodies to herpes simplex virus type 2 in patients with AIDS. J Clin Microbiol 1992;30:1312-14. |
||||||||||||
|
about SDI | newsletters | grants | publications | literature reviews WHO
Home -
WHO
Search - TDR Home - SDI Home -
SDI Contact us
|
|||||||||||||
.gif){kind=small}