Main

Respiratory viruses are an extremely common cause of childhood morbidity-it is estimated that an average child will experience four to five respiratory illnesses in each of the first 2 y of life(1). Thirty percent of healthy infants acquire lower respiratory tract illnesses severe enough to merit medical attention in the first year of life, of which 60% are bronchiolitis(2). Nasopharyngeal and throat swabs have often been used to establish epidemiologic patterns of respiratory infection(2,3) because collecting blood samples in infancy, whether for purposes of diagnosis or for epidemiologic data collection, is both technically and ethically difficult. As a result the true seroprevalence of viruses in infant populations may be unknown or based on outdated data. In 1939 the use of filter paper was suggested for collection, storage, and transport of blood(4), and since 1963, dried blood on filter paper has been in use for widespread neonatal screening programs to detect phenylketonuria(5). In Northern Ireland it is also used to screen for neonatal hypothyroidism, homocystinuria, and cystic fibrosis. Filter paper minimizes the need for invasive tests, enhances acceptability, and reduces the cumbersome transport difficulties associated with whole-blood samples. Its use has been extended to encompass a wide range of purposes(6–19), including the identification of a variety of viruses(20–24).

This study set out to explore a new multiantigen viral immunofluorescence assay as a potential epidemiologic and diagnostic tool to establish the seroprevalence of respiratory viruses in a pediatric population in Northern Ireland and to evaluate the use of filter paper for blood sample collection.

METHODS

Subjects. The study population consisted of a subsample of mothers from a larger study (n = 386) looking at factors influencing the prevalence and severity of RSV infection in infancy. The subsample included 256 mothers delivering babies in the Royal Maternity Hospital, Belfast (January 9, 1995 to June 12, 1995) who completed an initial and follow-up questionnaire as part of the larger study and permitted collection of both cord blood at birth and a follow-up heel-prick blood sample taken from their infant at 7 mo. A pediatrician (J.K.N.) prospectively enrolled mothers in the immediate perinatal period. Ethical approval was granted by the hospital ethical committee, and written informed consent was obtained before completion of the maternal questionnaire and cord blood analysis. Seven months was chosen for follow-up because the majority were seen by health visitors and follow-up coincided with the 7-mo child surveillance check. Mothers missed by health visitor follow-up were seen at one of three follow-up clinics or at a home visit.

Collection of cord blood samples. Two to three milliliters of cord blood was collected at birth, stored at 4°C until centrifugation and the sera were stored frozen at -70°C until analysis.

Collection of 7-mo heel-prick blood samples. Health visitors performed the majority of heel-prick blood tests at 7 months. Drops of blood fully saturated up to three circles (15-mm diameter) on a filter paper card in routine use for neonatal screening (Summit Printing Systems, Lisburn, Northern Ireland). Blood was allowed to dry at room temperature, packaged in a polythene resealable bag, and posted to the Regional Virology Laboratory where it was frozen at -20°C.

Elution of blood from filter paper. Filter paper cards were defrosted. Three-millimeter-diameter discs were punched from the center of the blood-soaked circles, using a ticket punch (Maun Industries, Mansefield, England). Twenty discs were added to 200 µL of PBS with 0.05% Tween and 1% skimmed milk in a Sarstedt tube (Sarstedt Ltd, Leicester, England). This gave a previously determined serum dilution of 1/10. Each tube was agitated for 30 min at room temperature before centrifugation at 1300 rpm for 5 min. This method has been described previously(22). The eluate was withdrawn and frozen at -70°C.

Immunofluorescence. All cord sera and heel-prick sera were assayed for measles, RSV, IFA, ADV, PF1, PF2, and PF3 specific IgG by a standard indirect immunofluorescence technique. These viruses were chosen because all (except measles) were demonstrated in or grown from other inpatient clinical specimens, reflecting the circulating respiratory viruses in Northern Ireland during the winter period under study; influenza B virus was not included as it was not recovered during the study. The analysis of measles IgG was included as a comparative control to demonstrate diminution of maternally derived antibody between cord sera at birth and heel-prick sera at 7 months.

Infected Cell Preparation. Laboratory isolates of measles and ADV were grown in Vero E6 cells, RSV and PF3 were grown in HEp-2 cells, and IFA, PF1, and PF2 were grown in primary rhesus monkey kidney cells. The cell monolayers were grown in 75-cm2 tissue culture flasks (Bibby Sterilin Ltd, Stone, England). They were infected with stock virus preparations and harvested when approximately 80% of the cells showed a cytopathic effect. The cells were removed with glass beads and centrifuged in a plastic universal bottle (Bibby Sterilin Ltd) at 1500 rpm in a Mistral bench-top centrifuge. They were resuspended in 500 µL of PBS, aliquoted in 100 µL amounts, and stored at -20°C.

Slide preparation. An aliquot of each antigen was thawed. Indirect immunofluorescence with in-house MAb on doubling dilutions of the cell suspensions was used to find the working concentration of the respective antigens; this was chosen as the dilution preceding the one with a noticeable fall-off in fluorescence staining. A 1-µL volume of each antigen at its working dilution was added to each well of a 10-well, multiwell slide, in a predetermined pattern. IFA, measles, and ADV were added at 1, 3, and 5 o'clock positions. PF1, PF2, and PF3 were added at 11, 9, and 7 o'clock, and RSV was added to the center of the well. Each antigen was air-dried before the next was applied, and the completed slide then fixed in acetone for 10 min at 4°C.

Virus IgG assay. Four microliters of cord sera was diluted 1/50 in PBS and 8 µL of heel-prick sera was diluted 1/5 in PBS (final dilution 1/50). Each pair was added to adjacent multiantigen wells, incubated at 37°C for 30 min, washed in PBS for 30 min, and counterstained with goat anti-human IgG FITC conjugate (Sigma Chemical Co, Poole, England). They were washed for 30 min in PBS, mounted with glass cover slips using glycerol saline, and read using a Zeiss IV epifluorescence microscope. Positive and negative control sera and conjugate controls were included in each run. All sera were ranked for strength of antibody reactivity in ascending order of reactivity as -, ±, +, ++, and +++. The presence of virus-specific IgG in heel-prick sera of either equal or greater strength to that seen in the corresponding cord sera was taken to indicate de novo antibody synthesis and childhood infection with that virus.

RESULTS

The larger study population comprised 386 mother and baby pairs of whom 357 (92.5%) were contacted at follow-up. Of 325 (84%) who agreed to continue to participate in the study, 69 (21%) refused the heel prick. Thirty-two (8.3%) refused any follow-up, and 29 (7.5%) could not be traced. Of the 256 mother and baby pairs who permitted both heel prick and questionnaire follow-up, 234 paired cord sera and heel-prick sera were successfully tested. Eleven (4%) of the heel-prick sera had insufficient blood for analysis, and 11 (4%) of the samples were not analyzed either because there was insufficient eluate or because of investigator error or spillage during analysis.

Sixteen of the cord sera were strongly reactive for one of the viral antibodies under assay and were ranked +++; this included nine sera for RSV, two for measles, and one each for the remaining viruses. In the corresponding heel-prick sera the respective antibody became undetectable for 14 of the 16 with the remaining two having a low level of antibody still detectable (rank ±). These were against RSV and PF3.

A total of 100 (43%) children had evidence of respiratory virus infection during the study (Table 1). Seventy-five children had IgG specific for one respiratory virus only. A further 25 had IgG specific for two respiratory viruses (Table 2). The staining pattern of fluorescence was almost always solely against the single virus. Occasional minimal reactivity against a second virus in the panel was also present.

Table 1 Seroprevalence of six respiratory viruses and measles in the infant cohort of 234 children at 6-12 mo of age
Full size table
Table 2 Seroprevalence of two respiratory virus infections in the infant cohort of 234 children at 6-12 mo of age
Full size table

ADV was the most frequently recorded infection (53/234) whereas PF1 and PF2 were the least common (2/234 for each virus). All but one of the cord sera were positive for measles IgG. None of the heel-prick sera had detectable measles IgG.

DISCUSSION

This study introduced a multiantigen fluorescence immunoassay format with paired heel-prick and cord sera to examine exposure to common respiratory viruses in infancy. The filter paper collection of blood samples and multiantigen assay format have implications for easy, widespread viral serodiagnosis in both seroepidemiology studies and the diagnosis of pediatric viral illnesses. A filter paper method of blood collection is a safe and more acceptable option than venipuncture for follow-up serology(22). Of the parents in our study agreeing to questionnaire follow-up, only 21% refused the heel prick.

The format of the immunoassay is designed to maximize the number of tests available from the infected cell preparations by using a minimal number of cells per well. Storage at -20°C with a subsequent single thawing is minimally damaging to the cells and allows the various antigens to be prepared in advance of making the slides without having to synchronize the respective cell cultures. Having multiple antigens (in this case, seven) on each well minimizes the volume of serum and reagents used and the manual labor of the test. It also controls for anticellular staining of each serum by having multiple cell lines assayed concurrently, eliminating a problem encountered with immunofluorescence assays and increasing specificity.

Fall-off in maternally transmitted IgG antibody was controlled for by two qualitative approaches. First, measles was included in the panel of viruses tested because antibody would be expected to be present in the entire cohort of cord sera and would be unlikely to be stimulated by natural infection during the study. All except one of the cord sera, but none of the heel-prick sera, were positive for measles IgG, indicating the likelihood that the latter did not contain residual maternal IgG.

Second, by ranking the cord sera for all viruses, we were able to confirm that only two of the 16 cord sera antibodies ranked +++ had detectable heel-prick sera antibody and that was of the lowest rank (±). From these two parameters we are confident that we were not detecting maternal antibody in the heel-prick sera. It is likely that our criteria for accepting that a childhood infection had occurred, with the viruses under investigation, actually underestimates the true number of infections that have taken place.

In our region we detect ADV in approximately 10% of all feces from children younger than the age of 5, and it is likely that the ADV infections in this study are inclusive of more than the respiratory types. Also detection of group-specific antibody reactivity by parainfluenza-infected cells could have resulted in inaccurate conclusions. However, in practice this did not happen, and we recorded no instance in which there was evidence of infection by more than a single parainfluenza virus type.

The strength of immunofluorescence in the heel-prick sera was noticeably stronger for ADV and, as we suggest, probably indicates infection by more than one serotype. This would result in our detection of a secondary rather than a primary immune response and higher antibody levels. The opposite is true for the respiratory pathogens, especially those with an epidemic transmission pattern, such as RSV. These viruses would be expected to result in a primary infection with the generation of a low level of antibody, which would quickly fall to low or undetectable levels. It is recognized that in a pediatric population (1 mo to 14 y), antibody assay alone may detect only 29% of respiratory viral infections and in infants antigen detection has been shown to be the most accurate method of microbial diagnosis(25). Understandably we recorded a different pattern of reactivity in the cord sera, in which RSV was the strongest antibody detected, indicating repeated maternal exposure to this virus.

Thus, a limitation of this study is the overestimation of the prevalence of ADV as a respiratory pathogen and the likely underestimation of other viruses. However, the profile of exposure to the different respiratory viruses (Table 1), excluding ADV for the reasons mentioned, reflects that seen in our laboratory for confirmed infection using culture and direct antigen detection methods and is comparable with other studies, especially with regard to the relative frequency of RSV infection in the first year(1,2,25–28). We believe this supports the view that our findings are a valid determination of exposure to respiratory viruses in a newborn population during the winter months in Northern Ireland. However, accurate direct comparison with other studies is limited by differing study design, inclusion criteria (different age groups and restriction to analysis of lower respiratory tract infections), and the spectrum of viruses analyzed.

Identifying the exact virus for each illness will permit a more accurate understanding of virus-specific signs and symptoms, prognosis, and complication rates, and would be helpful for infection control purposes. As respiratory viral vaccines become feasible(29,30) it is only by establishing the prevalence and severity of specific viruses in our community that we can progress to consider the risks and benefits of prevention strategies.

This study has combined the advantages of filter paper blood collection with a multiantigen screening approach and has introduced a method that could facilitate future seroepidemiology studies of infectious diseases.