|Home | About | Journals | Submit | Contact Us | Français|
Human rhinoviruses (HRV) cause acute upper respiratory illness. The frequency of HRV-associated illnesses appears greater when PCR assays are used to detect rhinoviruses. The present study performed PCR-based detection of HRV upon entry of subjects into respiratory syncytial virus and parainfluenza type 3 vaccine trials when subjects were symptom-free and upon subsequent development of clinical symptoms of respiratory illness during the trial. The background of HRV PCR positivity in symptom-free individuals (30/139 [22%]) was only slightly lower than in those with respiratory illness (28/77 [36%]). For subjects with multiple samples, it was estimated that HRV was detectable by PCR for approximately 100 days before, during, and after clinical symptoms were documented. PCR is a remarkably more sensitive method of detecting HRV than is tissue culture. The presence of HRV RNA may not always reflect an association with infectious virus production. The limited association of HRV RNA with illness suggests caution in assigning causality of HRV PCR positivity to clinical symptoms of respiratory illness.
Over 100 serotypes of human rhinoviruses (HRV), members of the family Picornaviridae, have been identified through community surveillance of respiratory illness (19). HRV has traditionally been identified by growth in tissue culture, as characterized by cytopathic effect. HRV are differentiated from enteroviruses by the capacity to grow at 33°C and acid lability. HRV are responsible for a substantial proportion of upper respiratory illnesses (19). Because infectious HRV are difficult to recover from clinical samples, it has been postulated that HRV may be responsible for other respiratory illnesses currently of unknown etiology (1, 8, 18, 24). The fact that less than 1 50% tissue culture infectious dose of HRV caused an experimental human rhinovirus infection (5) suggests that tissue culture may not efficiently detect HRV. Recent reports suggest that PCR detection extends the scope of HRV illness to include lower respiratory tract illness (10) and establishes a strong association between rhinovirus and exacerbations of asthma (12, 26). HRV has been shown to replicate in cells of both the upper (14, 27) and lower (21) respiratory tracts and can grow at the higher temperature of the lower respiratory tract (22).
However, more limited studies have detected HRV by PCR in 12 to 20% of asymptomatic children and in a relatively high percentage of adenoid and tonsillar tissues, suggesting that identification of HRV by PCR may not necessarily confirm it as the etiologic agent (16, 20, 23,25). Identification of rhinovirus infection by PCR may not be accompanied by a serologic response, an unusual occurrence with most symptomatic respiratory viral infections (4).
Clinical studies evaluating the safety of live, attenuated strains of respiratory syncytial virus (RSV) (17, 29) and parainfluenza virus type 3 (PIV3) (3) have recently been carried out with adults and children as young as 4 weeks of age. Nasal wash samples were collected from symptom-free subjects at the start of the trials and from those with respiratory tract illnesses during the trials. Samples were cultured for vaccine virus and for adventitious (wild-type) viral agents to determine the etiology of intercurrent illness. In some samples, a common wild-type respiratory virus (RSV, PIV1, PIV2, PIV3, influenza A or B virus, enterovirus, HRV, or adenovirus) was identified as a probable etiologic agent; however, no viruses were detected in more than three quarters of illness samples. Although most respiratory illnesses are caused by viruses (6), some of these illnesses may be caused by other pathogens, such as Streptococcus pyogenes, Mycoplasma pneumoniae, Streptococcus pneumoniae, nonencapsulated Haemophilus influenzae, Chlamydia pneumoniae, and Moraxella catarrhalis.
The same nasal wash samples were then tested for HRV by both tissue culture and real-time quantitative reverse transcription-PCR (QPCR) to examine alternative methods for detecting HRV, to better evaluate the role of HRV in upper respiratory tract illnesses, and to expedite the assignment of causality of adverse events reported during the course of clinical vaccine evaluations. This data set formed the basis for an analysis of the clinical relevance of QPCR in the detection of HRV illness.
(This work was presented in part at the VIII International Symposium on Respiratory Viral Infections, Kohala Coast, HI, March 2006.)
Nasal wash samples for adventitious viral detection were obtained from subjects enrolled in live RSV and PIV3 clinical studies on the day of vaccination (day 0) and on subsequent days on which the subjects had symptoms of respiratory illness, per clinical study protocol (19). These samples were routinely evaluated in the clinical investigators' laboratories for the presence of common respiratory viral agents, specifically, RSV, PIV1, PIV2, PIV3, influenza A and B viruses, enterovirus, and adenovirus; 203 of these nasal wash samples were also analyzed at ViroMed Laboratories, Minneapolis, MN, by tissue culture and by QPCR for HRV detection. The data were from four clinical studies that included 35 infants 1 to 3 months old, 48 young children 6 to 36 months old, and 33 adults. The data were analyzed with respect to the reliability of QPCR versus culture to detect HRV infection and the correlation of positive QPCR and culture results with respiratory illness reported during the trials. The studies were carried out during the spring, summer, and fall, thus encompassing the seasonal peak of rhinovirus illness (2, 28).
HRV was isolated and identified in clinical samples by standard virus culture techniques using human lung cells (either MRC-5 or WI-38) inoculated with 200 μl of nasal wash sample and incubated in a roller drum in a 33 to 35°C humidified CO2 incubator for up to 14 days. All specimens were either snap-frozen and stored at −80°C for less than 2 weeks or sent directly at 4°C for inoculation within 24 h. HRV cytopathic effect in human lung cells typically appeared as rounded and refractile cells distributed among the cell monolayer. To differentiate HRV from enterovirus, an aliquot of the infected culture was tested for viability at pH 3; only enteroviruses are stable at this pH.
At the same time the samples were cultured, RNA was extracted from 200 μl of each nasal wash sample and 1/10 of the extracted sample was used for each QPCR. The PCR primers and probe used in the TaqMan assay (Applied Biosystems, Foster City, CA) were derived from conserved sequences within the 5′ untranslated region of sequenced HRV genomes. The forward primer sequence corresponded to nucleotides 414 to 432 of HRV-89, as previously described (11). The reverse primer sequence corresponded to the reverse complement of nucleotides 461 to 481 of HRV-89 (25). The double-labeled fluorescent probe sequence corresponded to nucleotides 438 to 459 of HRV-89. (11) The reactions were run in a GeneAmp 5700 sequence detection system (AME Bioscience, Toroed, Norway) for real-time detection of fluorescence signal changes. The threshold cycle number was determined and compared to positive and negative controls. The level of HRV present is reported in PFU equivalents per milliliter, based on the standard curve generated from a positive control (HRV) RNA serially diluted from 10,000 to10 PFU/ml. The amount of HRV present is based on the copy number of rhinovirus genomes present in the positive control standards rather than on the infectivity of these standards.
Specificity of the picornavirus QPCR TaqMan detection assay for HRV was determined by spiking the reaction mixtures with high titers of closely related members of the Enterovirus genus of the Picornaviridae family, including 12 strains of coxsackie A virus, 6 strains of coxsackie B virus, 3 strains of poliovirus, and 30 strains of echovirus, as well as the unrelated viruses RSV subgroup A, adenovirus 5, coronaviruses OC43 and 229E, mumps virus, measles virus, PIV3, and influenza A and B viruses. All but three of the enteroviruses evaluated (specifically coxsackievirus A18, coxsackievirus A20, and coxsackievirus B4) yielded negative results. Of 101 available HRV prototype strains tested in this assay, 90 strains had detectable virus nucleic acid. The following HRV strains were not detected: HRV-5, HRV-19, HRV-26, HRV-42, HRV-45, HRV-79, HRV-84, HRV-86, HRV-87, HRV-98, and HRV-99. These assay results indicate a high level of specificity for HRV.
The comparisons of methods of HRV detection included the following: (i) frequency of HRV detection by tissue culture and QPCR in subjects with and without clinical symptoms (samples collected within 7 days from patients with symptoms were considered to represent the same illness); (ii) frequency of HRV detection by tissue culture and QPCR in adults, children, and infants; (iii) frequency of detection of HRV by QPCR versus tissue culture (three quantitative PCR cutoffs are presented: all positive, those positive with a cutoff of ≥1 PFU equivalent, and those with >150 PFU equivalents); (iv) quantity of HRV by QPCR in positive versus negative tissue cultures; (v) quantity of HRV by QPCR in subjects with and without symptoms; and (vi) predictive value of a positive rhinovirus sample upon entry into the trial for subsequent illness in the next 6 days.
Individuals with multiple samples of which one or more were positive for rhinovirus were used to describe the best-fit curve, with the highest detected titer used as the zero point and quantitative values plotted temporally from zero either before or after the highest PCR result. We used restricted cubic splines, as described by Harrell (9), to model this relationship and to better approximate the true biological relation between days from peak and virus shedding. Restricted cubic splines allow continuous data to fit within the regression model without the assumption of a linear relation.
A total of 216 “day 0” and “illness visit” nasal wash samples collected from subjects were tested for rhinovirus by both tissue culture and QPCR. Individuals had to be well on day 0 to permit entry into the trial and administration of one of the experimental vaccines. Of the total samples, 7 were positive by tissue culture and 58 were positive by QPCR (P < 0.01) (Fig. (Fig.1).1). The percentage of QPCR-positive samples was lower when the higher cutoff values of >1 or ≥150 PFU equivalents were used (Fig. (Fig.1).1). All tissue culture-positive samples were QPCR positive, with values of ≥150 PFU equivalents by PCR.
To reduce the risk of multiple sampling of the same illness, the illness samples were analyzed only when separated by ≥7 days. A significantly higher frequency of positive QPCR samples was obtained from subjects with symptoms (28/77 [37%]) than from subjects without symptoms (30/139 [22%]) (Table (Table11).
The proportions of subjects who had HRV-positive samples differed significantly by age (P = 0.01 [for trend]) (Fig. (Fig.2).2). Surprisingly, QPCR-positive samples were most common in early infancy, with 18% of samples from adults, 33% of samples from children 6 to 36 months of age, and 54% of samples from infants being positive (Fig. (Fig.2).2). QPCR-positive samples were detected in subjects without symptoms in all age ranges. The quantities of rhinovirus copies by QPCR were not different between those with and without symptoms (Fig. (Fig.3).3). Furthermore, those with a QPCR-positive sample on day 0 did not have any greater frequency of illness in the next 6 days than those with negative QPCR at day 0.
With the highest detected value used as the set point in those with multiple samples, it was possible to calculate the median duration of QPCR detection to be as long as 30 days before and after the highest recorded QPCR value (Fig. (Fig.4).4). Visually, the points create a sharp curve, suggesting that QPCR detection is not chronic.
Other adventitious viruses were recovered by culture in 16 of 77 illness visit samples (21%). These included wild type PIV 3 (six samples), adenovirus (three samples), enterovirus (three samples), wild-type RSV (two samples), influenza A virus (one samples), and nontyped PIV (one sample). QPCR was not performed for viruses other than rhinovirus.
The ability to detect HRV in nasal wash samples was increased many times when a validated QPCR assay was used. However, in our study there was frequent detection of rhinovirus by QPCR in symptom-free individuals. This result raises a number of questions.
Was the QPCR accurately detecting HRV RNA? The assay was well validated in terms of specificity and sensitivity against reference strains. The curve generated by progressive dilution of HRV standards was linear. Nevertheless, the relative quantitation of each HRV strain may be limited by the use of a limited number of primers for all HRV serotypes. Designing specific primers for all rhinovirus serotypes seems a daunting task.
Can a QPCR cutoff that increases the correlation with illness be identified? As all positive tissue culture results were >150 copies, this could be considered a cutoff value for a positive QPCR result. However, the overlap of the quantity of virus present between individuals with and without symptoms was almost complete, and setting such a high cutoff raises the issue of how to interpret weakly positive samples. There is logic to a cutoff of 1 PFU equivalent, but the early literature suggested that human infectivity was a more sensitive measure than was tissue culture infectivity (5). The wide overlap in QPCR values between well and sick individuals is also disconcerting. Any proposed assay needs a definitive correlation of QPCR level with clinical disease. We suggest that future studies must include tissue culture and QPCR determinations from healthy individuals in the population(s) being monitored in order to more accurately assess the relationship between HRV infection and clinical illness.
Are there unique aspects to the conduct of this trial that might have altered the findings or limited its generalizability? The studies were done outside of the winter respiratory viral season and were done with otherwise healthy individuals without a history of asthma. Each of these factors makes the study less general but does not change the validity of the comparisons between subjects with and without symptoms.
Do these observations give insight into the pathogenesis of HRV infection? The prolonged period of time that HRV can be detected is consistent with the high frequency of recovery of rhinovirus from tonsil and adenoid tissues (23, 25). At least two other reports have documented rhinovirus recovery in well children for up to 2 weeks after a respiratory illness (15, 28). A longitudinal study has shown an association of illness with rhinovirus recovery (28). However, in that study only 62% of the illnesses ascribed to rhinovirus had a direct temporal association of virus identification with the onset of illness, and 20% of the rhinoviruses detected had no association with illness. Other studies have shown 4% of adults and 12 to 18% of well children with rhinovirus (16,20). Certainly, from our study and those conducted by others, HRV are ubiquitous and occur in all ages, including early infancy, and appear to persist in the respiratory tract with or without illness for relatively long periods of time. Interpretation of our observations is confounded by the possibility of multiple strains in a single volunteer, affecting the duration of apparent shedding, and the possibility that detection of rhinovirus in asymptomatic individuals may be from a previous symptomatic infection. However, to be permitted entry into the trial, the volunteers could not have had a respiratory illness in the week before vaccine administration.
There is no question that rhinoviruses cause respiratory illness. Our data support that fact, with a statistically significantly higher identification of rhinovirus in subjects with illness than in those without. Experimental infections (5, 7) and the effectiveness of specific antivirals (11) all make unequivocal arguments that HRV are causes of respiratory illness. Additionally, rhinovirus has been reported to be a cause of illness in immunologically compromised individuals (13). However, given that HRV was detected in greater than 20% of well individuals, further research is necessary to convincingly show the extent to which the impact of HRV should be broadened beyond the traditional association with upper respiratory tract infections to having a major role in the triggering of asthma or as a causative agent in serious lower respiratory tract disease.
These studies were supported by Wyeth Vaccines Research and a grant from the NIH (NO1 AI015444). Studies at Vanderbilt were done on the GCRC with support from grant M01RR00095.
The performance of the PCR assays by ViroMed Laboratories under contract from Wyeth Vaccines Research is gratefully acknowledged.
Published ahead of print on 2 May 2007.