Objectives: RSV persistence was investigated by testing respiratory secretions from subjects with COPD during illness and at regular intervals over 1 yr.

Methods: Nasal and sputum samples from subjects with COPD were tested by one-tube nested RT-PCR for RSV every 2 mo and during respiratory illnesses for 1 yr. Subjects positive for RSV were evaluated weekly until negative in two consecutive samples. Nasal secretions and serum were tested for RSV antibody. A rise of fourfold or greater was defined as evidence of RSV infection.

Results: A total of 112 patients were enrolled and the illnesses of 92 patients were evaluated. RSV was detected by RT-PCR in 6/92 (6.5%) illness nasal samples versus 0/685 routine nasal samples and in 5/69 (7.2%) illness sputum samples versus 3 /315 (0.9%) routine. Four additional RSV infections were identified by serum antibody responses. Of the RSV infections 86% were associated with serum or nasal antibody responses and 73% had symptoms of acute respiratory illness.

In our laboratory, we have developed a sensitive and specific real-time, one-tube nested RT-PCR for the diagnosis of RSV in adults (19, 20). Because the assay is performed without opening the reaction tube, PCR contamination has been markedly reduced while the sensitivity of a nested assay is retained. Therefore, we sought to explore the question of RSV persistence in patients with COPD by testing upper- and lower-respiratory-tract secretions at regular intervals over 1 yr in a cohort of subjects with COPD.

Overall, 685 routine evaluations were performed, yielding 685 nasal samples and 315 sputum samples for RT-PCR analysis. RSV RNA was detected in 0 of the 685 nasal and in 3 (0.9%) of the 315 sputum samples (Table 2). During the 12-mo follow-up period, 134 respiratory illnesses were reported, of which 92 (69%) were evaluated during the acute phase with the collection of 92 nasal and 69 sputum samples. Ill subjects were evaluated an average of 2.9 ± 1.6 d after onset of symptoms, and six nasal (6.5%) and five sputum (7.2%) samples were RT-PCR positive. The RSV detection rate in acute illness samples was significant greater than in routinely obtained nasal (p < 0.0001) and sputum samples (p = 0.006). Overall, 30% of the 384 sputum samples were judged to be adequate, 44% adequate but contaminated, and 26% inadequate.

From RT-PCR and serology, 14 subjects showed evidence of RSV infections with 1 subject showing evidence of two infections (Table 3) Seven infections were RT-PCR positive and associated with a greater-than-fourfold rise in serum or nasal antibody, three were RT-PCR positive only, and four were identified by serologic response only. Of the 15 RSV infections, 11 were identified during illness visits (subjects 1–10), including both infections in subject 1. The latter subject had two RT-PCR positive illnesses approximately 2 mo apart, the second one associated with seroconversion. Of the four RSV-infected subjects with clinically unrecognized infection, three were identified by a positive sputum RT-PCR collected on routine visits (subjects 12–14). Of these three, one had no symptoms, one complained of increased cough at the time of the visit, and the other complained of feeling unwell with increased dyspnea for 1 wk preceding the visit. However, neither of the latter two subjects perceived a “new respiratory illness” warranting an illness evaluation at the time of the positive RT-PCR sample collection. One of these subjects (subject 12) demonstrated a serologic response at the time of the positive RT-PCR.

Quantitative RT-PCR was performed on all positive nasal and sputum samples and expressed as pfu/ml of sample (Table 4). The mean peak viral load by quantitative RT-PCR was 38,620 ± 64,750 pfu (range, 2–156,000) in illness samples and 82 ± 140 (range, 0.01–250) in routine samples. Due to the small numbers and the wide range of values, this difference was not statistically significant. The viral load of RSV RNA in nasal samples was lower than in sputum in almost all instances. In 9 of 11 nasal and sputum pairs collected simultaneously, the sputum viral load was higher than the nasal (p = 0.008, Mann-Whitney test). In addition, sputum samples remained positive longer than nasal samples (13.8 ± 7.6 vs. 10.1 ± 7.3 d) although this difference was not significant.

All RT-PCR positive samples were collected between October and May, with clear clustering in January (Figure 1), and all were group B by strain-specific RT-PCR, including both illness samples from subject 1. Sequences of the 190 nucleotide amplified F gene segment from these two samples were both consistent with the B1 genotype, although they differed by 2 nucleotides at positions 713 A/G and 916 A/G.

The subjects who were identified as RSV positive by RT-PCR or serology did not differ significantly from subjects who did not have RSV in age, sex, smoking status, pulmonary function, or steroid use (Table 1). The clinical characteristics of the 11 RSV illnesses are outlined in Table 5. Most subjects were moderately ill with complaints of increased cough (100%), sputum (45%), and dyspnea (73%). Only 1 of 11 had symptoms limited to the upper respiratory tract. At least one office visit was made by 73 % of the subjects and one subject was hospitalized for 3 d. On average, subjects were unable to perform their activities of daily living for 9 d. Evaluated by standards set forth by Anthonsien and colleagues for acute exacerbation of COPD, 91% of those with an identified illness met exacerbation criteria (25).

In our study, the identification of RSV RNA in respiratory samples from patients with COPD was most often associated an acute exacerbation and a measurable serum and mucosal immune response. However, clearly there is also a spectrum of disease ranging from asymptomatic infection or mild coryza to severe illness requiring hospitalization. In only three cases did we detect RSV RNA in respiratory secretions collected during routine visits.

The subjects with no symptoms and a robust immune response undoubtedly had asymptomatic infection. For the three subjects in whom viral RNA was identified at low copy number from a single specimen without a demonstrable immune response (subject 1, illness 1, and subjects 13 and 14), it must be questioned whether true “infection” has occurred. Possible explanations for these episodes include laboratory contamination of the RT-PCR assay, “aborted infection,” or low-grade persistence of RSV. Although all samples with low copy number or those found during routine visits were confirmed by repeat testing, contamination of the original sample cannot be excluded with absolute certainty. “Aborted infection” seems the most plausible explanation since the organism was cleared without resulting in symptoms or an immune response. It is conceivable that some pathogen encounters are successfully eliminated in the early phases by the host innate immune response or preexisting mucosal antibody. The fact that RSV RNA was only identified in sputum and not nasal samples in two of these three cases argues somewhat against this possibility. Low-grade persistence seems an unlikely alternative explanation given the clear winter temporal clustering of the RSV positive cases and that no subject had the same RSV RNA identified on more than one occasion.

One subject in our study had the same genotype of RSV identified 2 mo apart. Prolonged shedding of RSV has been described in a number of immunosuppressive conditions such as corticosteroid use and HIV infection (33). Because our subject used inhaled and oral corticosteroids on a chronic basis, prolonged infection was possible. However, the identification of two nucleotide differences in a conserved portion of the F gene in the two specimens most likely represents two different strains of RSV B1. Thus, reinfection, which has been well documented in children, seems the most likely explanation for this finding although chronic infection with the development of quasi-species cannot be ruled out without more complete sequencing data (34).

Our results are clearly different from those of previous reports that describe a high percentage of patients with stable COPD with detectable RSV RNA at low copy number. Possible explanations for this discrepancy include: different study populations, variable sensitivities of RT-PCR methodology, and assay contamination. Our study has several limitations that could have potentially affected the results. Because spirometry was not available for all patients, it is possible that the study population contained some subjects who did not have COPD and thus were not at risk for persistent RSV. Because 97% of the subjects were under the care of a pulmonary specialist we feel that this was unlikely. Another limitation of our study is that we did not have freshly induced sputum specimens from all subjects, raising concerns that viral RNA may have degraded during specimen transport. Although this is a valid concern, we have found that sputum samples may be frozen and thawed multiple times with no significant diminution in viral load. In addition, all nasal specimens were collected during clinic visits and processed immediately. The failure to identify RSV in significant numbers of subjects with stable COPD seems not to be due to an insensitive RT-PCR assay because the limit of detection of our nested assay is one RNA copy. The use of highly sensitive RT-PCR methodology requires rigorous procedures to avoid contamination and false-positive results. Thus, an alternative explanation for the disparate results of this study compared with other reports is that high rates of RSV identification in subjects with stable COPD are due to RT-PCR contamination.