Cerebrospinal fluid challenges for the diagnosis of herpes simplex infection in the central nervous system

Abstract Herpes simplex virus (HSV) is a cause of a severe disease of the central nervous system (CNS) in humans. The demonstration of specific antibodies in the cerebrospinal fluid (CSF) may contribute to the retrospective neurological diagnosis. However, the commercial immunological tests for HSV infection are for use in serum samples. Objective: The aim of the present study was to adapt a commercial kit anti-HSV IgG used for serum samples to be performed with a CSF sample. Methods: Forty CSF specimens from 38 patients with suspected CNS HSV infection were serially diluted for detecting anti-HSV IgG by enzyme immunoassay (EIA). The same samples were also analyzed with the polymerase chain reaction (PCR). Results: The sensitivity of EIA test for HSV was 5% (dilution 1:40) and 65% (dilution 1:2) in CSF, and HSV DNA PCR was 15%. The combined analysis of EIA (dilution 1:2) and PCR increased the sensitivity up to 72.5%. The inflammatory CSF was associated with positive HSV PCR. Conclusions: We demonstrated the importance to adapt serological anti-HSV IgG EIA test for CSF assays to increase the accuracy of the analysis, considering the low concentration of specific antibodies in CSF.

The diagnosis of HSV infection in the CNS is based on the early detection of viral DNA with the polymerase chain reaction (PCR) in the cerebrospinal fluid (CSF). The sensitivity of this method is greater over the first week after symptoms onset or before treatment 5 . From seven to 10 days after that, the diagnosis can be determined by the presence of antibodies against the viral antigen 5,6 . The presence of IgM indicates recent primary infection being rarely detected in the CSF, whereas the specific IgG demonstration supports the retrospective diagnosis of the herpetic infection in the CNS and may persist for years 7 . The demonstration of the intrathecal synthesis of specific antibodies is an immunological method of choice for retrospective analysis. However, such analysis requires serum and paired CSF, which are rarely available.
Commercial serological tests have been used to investigate viral agents in the routine of CSF analysis. This practice may lead to a decrease in the accuracy of the test due to the lower concentration of CSF immunoglobulin compared to serum if there is no previous adaptation and validation of the method. This procedure should consider the modifications of already validated bioanalytical methods, as well as matrix change 8 . The aim of this study was to adapt a commercial kit anti-HSV IgG used for serum samples to be performed with CSF samples, increasing the accuracy of the laboratory diagnosis of HSV infection in the CNS. In addition, the study evaluates the combined use of immunological and molecular tests to support the diagnosis of HSV infection in the CNS.

METHODS
A panel of 40 CSF samples from 38 patients with a suspected diagnosis of HSV infection in the CNS was evaluated in this study. Two of them had samples analyzed at two different times. All had negative CSF tests for bacterial, fungal or syphilis infections. Routine CSF examination included total and differential cell count, protein and glucose concentration, stain and culture to bacteria and fungi.
First, a pool of six reactive CSF samples for anti-HSV-1/2 IgG SYM (Symbiosys Diagnóstica, Brazil) was serially diluted (1:2 to 1:160) in two stages to obtain a better dilution, using a commercial kit HSV-1/2 IgG EIA (DiaSorin, Italy). In the next step, repeatability assay with twenty replicates on the same day and reproducibility assay with triplicates over the course of 20 days were performed with the pool of CSF.
We performed nested real-time PCR by Sybr Green system for the detection of HSV DNA in all 40 clinical samples, according to the protocol described by Kawada et al. 9 .

Statistical analysis
The repeatability and reproducibility assays were evaluated by the mean, standard deviation (SD), and coefficient of variation (CV) using the GraphPad Prism statistical program. The association between EIA and PCR for HSV and CSF findings (cell count and protein) was performed with the nonparametric Mann Whitney test of the SPSS program. The existence of association was considered for p-value<0.05. In the comparison between the results of the EIA test and PCR, the agreement was analyzed with Kappa (K) value, being K≤0.20 without agreement and K>0.80 optimal agreement. This analysis was performed by SAS System software (SAS Institute, Inc., North California, USA). The present study was approved according to the Ethical Review Board of HUCFF/UFRJ.
Considering that the best optical density (OD) values are between 0.5-1.5 for the EIA test, the 1:40 dilution was chosen as the mean OD value of 1.115 of the reactive CSF pool ( Table 2). In the repeatability and reproducibility assays for the 1:40 dilution, we found 4.5 and 10.2% of CV, respectively. Both values are acceptable according to ANVISA's 8 determination that considers a CV up to 15%. Repeatability and reproducibility assays were also performed for a minimum empirical dilution of 1:2 in an attempt to increase the sensitivity of the test by detecting lower concentrations of anti-HSV IgG in the CSF. The repeatability and reproducibility assays for this dilution were 1.8 and 13.6%, respectively.
Of the 40 CSF samples analyzed, 5% of the samples were reactive for HSV IgG EIA at 1:40 dilution, 65% were reactive at 1:2 dilution, and 15% of the samples amplified the viral DNA by PCR. Three of the samples that amplified in PCR were also reactive in EIA at 1:2 dilution. None of the samples that amplified in PCR were reactive in EIA at 1:40 dilution ( Table 3).
The combined analysis of immunological tests and molecular biology were positive in 20% (EIA at dilution 1:40 and PCR) and 72.5% (EIA at dilution 1:2 and PCR) of the CSF samples. Increased cell count and protein values in CSF were associated with positive PCR in 40 samples (Tables 4 and 5). A trend of association of reactive EIA at dilution 1:40 (p=0.18) was observed with high levels of protein in the CSF. However, this tendency cannot be considered, because only a few samples were reactive (2/40). Kappa (K) index showed no agreement between reactive anti-HSV IgG by EIA and positive PCR for HSV (Table 3).

DISCUSSION
In the present study, in order to increase the diagnosis accuracy in cases of CNS infection by HSV with negative PCR,    we adapted a serological assay for the detection of anti-HSV IgG for CSF analysis, since the commercial kits usually used are for the evaluation of serum samples 12 . Specific EIA tests for serum are used in the laboratory routine for CSF analyses, not rarely without previous adaptation. Otherwise, some criteria are important for the validation of a diagnostic method: its results must corroborate the suspected diagnosis, be evaluated repeatedly and reproducible in the same and in different laboratories 8,12 . Nonetheless, the validation of methods for CSF analysis may present some limitations in its applicability, sensitivity (65%) for the detection of lower concentrations of antibodies found in the 40 CSF samples of the 38 patients with suspected HSV infection in their CNS. There was no agreement between PCR with EIA at various dilutions, since immunological tests became reactive around the tenth day after the onset of the first symptoms, whereas the PCR was positive at the onset of the disease, before treatment 16 .
The results of the molecular biology test on HSV infection in the CNS can be influenced by the duration of symptoms and the use of anti-viral drugs 17 . Although PCR represents gold-standard for diagnosis (95% sensitivity and 100% specificity), its sensitivity decreases after the acute phase of the disease, due to reduced virus replication activity 18 . At this time, the immune system begins to produce specific antibodies against the viral agent, which may persist for years 6,7 . The concentration of HSV antibodies in CSF can increase up to four times, reaching sensitivity at 85% detection after one month of disease onset 14 .
This study demonstrates the importance of a previously adaptation of commercial serological tests for use in CSF samples to obtain higher accuracy. The adaptation of dilution for CSF was a good alternative for retrospective screening of specific anti-HSV IgG antibodies in the CSF. It may contribute to the CNS HSV diagnosis in suspected cases when PCR is already negative. The combined analysis of immunological tests and PCR showed a greater comprehensiveness in the HSV infection in the CNS, considering the different stages of the disease. Viral DNA detection was predominant in inflammatory CSF samples consistent with the acute and active phase of the disease. Such findings confirm that the signs of acute inflammatory CSF represents a marker for HSV activity. such as noble sample obtained with invasive methods, scarce material and few standardized tests for use as reference or gold standard. For the development of a new laboratory test, there are some accepted methodologies. One of them consists of comparing the results of a laboratory test to a clinical diagnosis, based on the symptoms and/or physical exams of the patients that make up the analysis 8,12 . These are situations in which an imperfect pattern is accepted, which, although not considered ideal by the consensus of experts, may still provide results that are closer to the diagnostic reality 8,12 .

ACKNOWLEDGMENTS
The clinical diagnosis suspicion of HSV infection in the CNS together with CSF findings (cell counts, protein and glucose dosage, exclusion of other infections) was the reference in this analysis. The immunological tests and PCR were used to support the laboratorial diagnosis. This type of approach poses limitations in the case of retrospective evaluations, but it is more comprehensive because it allows analyzing distinct phases of the disease (multiple parallel tests). Not all patients were in the acute phase, or untreated, to consider PCR as the gold standard, although this was done in all of them. In fact, the reactive anti-HSV IgG by EIA is not a definition criteria for acute infection in the CNS, but may be helpful to support the retrospective diagnosis.
Most of the patients (77.5%) presented a CSF examination compatible with viral infection (pleocytosis with lymphocyte predominance and/or hiperproteinorrachia and normal glucose). Pleocytosis (>4 cells/mm 3 ) and hiperproteinorrachia (>40 mg/dL) were found in 65 and 50% of the cases, respectively. Ward et al. 13 found pleocytosis in 75% (12/16) of the CSF samples analyzed for patients with HSV infection in the CNS, whereas hiperproteinorrachia was detected in 71.4% (10/14) of them. In accordance with our results, Whitley reported that 50% of the individuals with herpetic encephalitis showed a normal concentration of protein in the first week of the infection 14 . Riera-Mestre et al. described an increase in leucocytes count in 77% of the CSF samples of 35 patients with encephalitis by HSV, and the protein concentration remained >40 mg/dL in 94.3% of the cases 15 .
In relation to the comparison between the EIA tests in the present study, the dilution 1:40 of CSF samples showed excellent results in the repeatability and reproducibility assays. There was no agreement between 1:2 and 1:40 dilutions of EIA. The lower dilution (1:2) presented good laboratory performance (repeatability and reproducibility), besides a higher