Serology as an Adjunct to Polymerase Chain Reaction Assays for Surveillance of Acute Respiratory Virus Infections

Diagnostic testing for viral infections has evolved during the past decade. Documenting a seroconversion, or significant increase in antibody titer between paired acute-phase and convalescent-phase serum specimens, is a well-proven method for detecting acute viral infection but can be challenging, because blood sample collection is invasive, a second blood sample is required, and late collection of the acute-phase serum sample complicates interpretation. In contrast, polymerase chain reaction (PCR) assays are rapid and sensitive when performed on respiratory specimens collected early in the illness but can lack specificity because of amplicon contamination or presence of virus not etiologically linked to the illness. PCR assays have largely replaced serologic examination and culture for detecting viral pathogens in respiratory disease surveillance [1]. The added value of serologic examination in disease surveillance has not been fully assessed. We compared serologic examination and PCR results among patients enrolled in a pneumonia etiology study in rural Thailand from September 2003 through August 2005 [2]. Patients who were hospitalized and met a broad case definition for respiratory disease were enrolled. We compared patients who had specimens (nasopharyngeal swab and serum) tested for adenovirus, human metapneumovirus (HMPV), influenza viruses A and B, parainfluenza viruses (PIVs) 1–3, and respiratory syncytial virus (RSV), as described elsewhere [2]. Conventional (first 18 months) and real-time (last 6 months) reverse-transcription PCR assays were used [3]. Influenza human serologic examination was conducted using hemagglutination inhibition test [4]. Adenovirus; PIVs 1, 2, and 3; HMPV; and RSV immunoglobulin G antibodies were tested using enzyme immunoassay [5–8]. A positive result was defined as a positive PCR test result and/or a ≥4-fold increase in antibody titer of the convalescent serum sample. We analyzed PIV 1 and 3 together, because the presence of cross-reactive epitopes on these related viruses complicates serologic discrimination. We compared proportions using McNemar’s χ2 statistic. For all viruses except influenza, PCR detected more positive results than did serologic testing, but the difference was statistically significant only for adenovirus (adenovirus, 2.4% vs 1.2%; HMPV, 1.3% vs 0.8%; PIV 1 and 3, 6.8% vs 6.6%; PIV 2, 0.7% vs 0.6%; RSV, 7.2% vs 6.4%; Table 1). The comparable frequencies for influenza A were 7.1% and 11.1%, respectively, and for influenza B were 2.1% and 3.7%, respectively. Among patients who had positive PCR results, negative serologic test results for that virus were more common among children aged <5 years (Table). Use of combined assays increased detection for all pathogens. We did not ascertain date of illness onset but did find that, among patients positive for any pathogen, the mean number of days between admission and specimen collection was significantly lower for those who were positive by PCR, compared with those who were positive by serologic testing but negative by PCR (1.28 vs 1.77 days; P = .003), suggesting that prompt collection of respiratory samples is important for PCR. Numbers were too small to compare assay results by illness severity. Viral Respiratory Pathogens Detected by Polymerase Chain Reaction and Serologic Testing September 2003–August 2005, by Age Group, Thailand Abbreviations: neg, negative; PCR, polymerase chain reaction; PIV, parainfluenza virus; pos, positive; seroneg, seronegative; seropos, seropositive. Denominators differ because serology results that were indeterminant were excluded in the analysis. P < .05. Only 16 months of complete testing for influenza viruses, September 2003–December 2004. Viral Respiratory Pathogens Detected by Polymerase Chain Reaction and Serologic Testing September 2003–August 2005, by Age Group, Thailand Abbreviations: neg, negative; PCR, polymerase chain reaction; PIV, parainfluenza virus; pos, positive; seroneg, seronegative; seropos, seropositive. Denominators differ because serology results that were indeterminant were excluded in the analysis. P < .05. Only 16 months of complete testing for influenza viruses, September 2003–December 2004. Lower seropositivity for most viruses may be related to late collection of initial serum samples after patients were already hospitalized with clinical pneumonia and acute-phase antibodies may have already increased. Higher PCR positivity could also be attributable to high assay sensitivity or to persistent nasopharyngeal shedding that is not related to acute infection, as has been shown to occur with adenovirus, particularly in young children [9]. These hypotheses do not explain the difference observed for influenza viruses, which may be related to the different seroassay used or to less optimized influenza conventional PCR used at that time. In addition, influenza viruses undergo extensive genetic variations; thus, the PCR results should not be generalized to other PCR assays. When used together, PCR and serologic assays may provide a more complete picture of the burden of some viral respiratory pathogens. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.