|Home | About | Journals | Submit | Contact Us | Français|
Few data exist regarding respiratory virus quantitation in lower respiratory samples and detection in serum from hematopoietic cell transplantation (HCT) recipients with respiratory virus-associated pneumonia.
We retrospectively identified HCT recipients with respiratory syncytial virus (RSV), parainfluenza (PIV), influenza, metapneumovirus (MPV), and coronavirus (CoV) detected in BAL samples, and tested stored BAL and/or serum samples using quantitative PCR.
In 85 BAL samples from 82 patients, median viral load in copies/ml for RSV (n=35) was 2.6×106, PIV (n=35) 4.9×107, influenza (n=9) 6.8×105, MPV (n=7) 3.9×107, and CoV (n=4) 1.8×105. Quantitative viral load was not associated with mechanical ventilation or death. Viral RNA was detected in serum of 6 of 66 patients: 4 of 41 with RSV pneumonia, 1 with influenza B, and 1 with MPV/influenza A/CoV co-infection (influenza A and MPV RNA detected). RSV detection in serum was associated with high viral load in BAL (p=0.05). Detection of viral RNA in serum was significantly associated with death (adjusted RR 1.8, p=0.02).
Quantitative PCR detects high viral load in BAL samples from HCT recipients with respiratory virus pneumonia. Viral RNA is also detectable in serum of patients with RSV, influenza, and MPV pneumonia, and may correlate with disease severity.
Respiratory virus infections are associated with high morbidity and mortality after hematopoietic cell transplantation (HCT). The most common viruses causing progression from upper to lower respiratory tract disease in HCT recipients are respiratory syncytial virus (RSV), parainfluenza virus (PIV), influenza, and human metapneumovirus (MPV), with mortality rates up to 25–45% within 30 days after progression to pneumonia [1–10]. Recent studies suggest that PIV and RSV, in particular, are associated with the complication of fixed airflow obstruction following HCT, further contributing to transplant-related mortality [11, 12].
Higher viral loads of cytomegalovirus (CMV) and herpes simplex virus in bronchoalveolar lavage (BAL) specimens from immunosuppressed patients have been associated with clinical outcomes such as severe respiratory illness and death [13, 14]. Quantitative polymerase chain reaction (PCR) analysis of BAL samples from predominantly immunosuppressed patients including lung transplant recipients showed that high MPV viral load was associated with severe pneumonia and complications requiring prolonged hospitalization . Similar studies are lacking for HCT recipients.
In immunocompetent hosts, respiratory virus replication is generally limited to the respiratory epithelium. However, for viruses such as avian influenza/H5N1, seasonal influenza, and SARS-coronavirus (CoV), detection of viral RNA by PCR and isolation (of influenza [16–21]) from plasma/serum samples has been described; these viruses may cause disseminated infection with replication outside the respiratory tract [16–27]. Notably, among individuals with avian influenza/H5N1 infection, viral RNA in blood has been detected only in fatal cases and is associated with higher pharyngeal viral loads, indicating potential viral dissemination associated with poor prognosis [16, 17, 27]. Detection of other viruses, such as RSV and rhinovirus, from serum, whole blood, and peripheral blood mononuclear cell (PBMC) samples by PCR and culture has rarely been described in neonates and children, although there is no consistent correlation between virus detection and disease severity [28–32]. Serum DNA viral load in immunocompromised patients with disseminated adenovirus or CMV infection reflects disease activity and can be used to predict severity and monitor response to antiviral treatment [33–39]. The presence of respiratory virus RNA in serum has not been systemically evaluated among HCT recipients.
We retrospectively identified a cohort of 104 HCT recipients with 108 episodes of respiratory virus-associated pneumonia within one year after HCT. Four patients had distinct pneumonia episodes with two viruses, separated by at least one month. All patients underwent HCT at the Fred Hutchinson Cancer Research Center (FHCRC) between 1993–2007 and provided written informed consent allowing use of stored specimens and medical records. The study was approved by the FHCRC Institutional Review Board.
All patients eligible for analysis had radiographic and clinical evidence of lower respiratory tract disease confirmed by respiratory virus detection in BAL. BAL samples were obtained from adults (≥18 years) by washing with 90–150 mL of sterile, isotonic saline via bronchoscopy, with lesser volumes used for children. Pneumonia was virologically confirmed by testing BAL samples with direct fluorescent antibody (DFA), culture, shell vial, or qualitative RT-PCR for RSV, PIV 1–4, influenza A/B, adenoviruses, and rhinoviruses, and by qualitative RT-PCR alone for MPV and the non-SARS human CoVs (OC43, 229E, HKU1, and NL63). A multiplexed qualitative PCR panel has been available for testing of clinical samples at our center since May 2006. BAL samples were also submitted for routine bacterial, fungal, and acid-fast bacilli cultures, and detection of CMV and herpes simplex virus.
All patients had serum and/or plasma samples prospectively collected weekly for laboratory monitoring during the first 100 days and at varying intervals up to one year after HCT. Although both sera and plasma were tested, these will be referred to collectively as serum samples throughout this manuscript. BAL and residual serum samples were stored frozen at −20°C or −70°C. For this analysis, 1–3 serum samples were obtained that most closely corresponded to date of bronchoscopy. We included patients with either stored BAL or serum samples; concomitant stored BAL and serum samples were available from a subset of patients. Stored BAL samples that had previously tested positive for RSV, PIV, influenza A/B, MPV, or CoV were tested by quantitative RT-PCR for the previously detected virus. Stored serum samples were also tested by quantitative RT-PCR for the virus that had been previously detected in BAL.
Total nucleic acids were isolated from 200 μL of BAL specimens and quantitative RT-PCR assays were performed using 10 μL of specimen/reaction [40–43]. Each quantitative assay was linear from 10–108 viral copies/reaction, with a 95% limit of detection of 10 copies/reaction or 1000 copies/mL [40–43]. A different nucleic acid extraction method (QIAamp Viral RNA Mini Kit, Qiagen, Inc.) was used for serum, which allowed for processing a larger volume (50 μL), providing sensitivity of 200 copies/mL when using the cut-off of 10 copies/reaction. Serum samples were tested by quantitative RT-PCR using the same primers and method described for BAL specimens [40–43]. All PCR methods were performed according to College of American Pathologist standards and the laboratories passed proficiency testing in viral diagnostics.
The Wilcoxon rank-sum test was used to compare quantitative viral load in BAL samples among patients grouped by clinical characteristics. Cox proportional hazards regression was used to analyze overall survival at one and 6 months by quantitative viral load levels in the first BAL. The model was adjusted for potential confounders, including age, stem cell source (peripheral blood stem cells vs. bone marrow and cord blood), presence of lymphopenia (≤100 lymphocytes/μl ≤7 days before BAL), mechanical ventilation at or within 30 days after BAL, and presence/absence of co-pathogens. Co-pathogens were defined as pathogenic bacteria, fungus, or opportunistic viruses from the same BAL or concomitant lung biopsy sample. For these evaluations, each patient was represented once using the first viral load available from the first pneumonia episode.
Characteristics of patients with and without detection of viral RNA in serum were compared using the Wilcoxon rank-sum test for nonparametric continuous data and Fisher’s exact test for categorical variables. Poisson regression with robust standard error estimates was used to calculate the prevalence rate ratio (RR) for clinical outcomes among HCT recipients with and without detection of viral RNA . The models were adjusted for stem cell source, presence of lymphopenia, and day of BAL after transplant [4–6]. Age, donor type (HLA-identical sibling vs. alternate donors), gender, conditioning regimen (nonmyeloablative vs. myeloablative), CMV risk group (donor and recipient seronegative vs. other), underlying disease risk, presence of grades 2–4 acute graft-versus-host disease (GVHD) at or any time before diagnosis of pneumonia, presence or absence of co-pathogens, and transplant year were also considered as potential confounders. These confounders were included in the multivariable model if they altered the adjusted RR for the outcomes of interest by ≥10%. For RSV pneumonia, the model was limited to one confounder.
Two-sided P-values <0.05 were considered statistically significant. No adjustments were made for multiple comparisons.
Of 104 HCT recipients, 87 had 100 stored BAL samples available for testing (Figure 1). Thirty-eight patients had 45 stored BAL samples without sera available, and 17 patients had 37 stored sera without BALs. Forty-nine patients had 55 concomitant BAL and 95 concomitant serum samples. Sera had been collected a median of one day after BAL (interquartile range, 3 days before to 4 days after).
Study cohort characteristics, grouped by type of samples available, are shown in Table 1. Most patients were lymphopenic in the week before pneumonia diagnosis. The groups differed with respect to proportion of pneumonia episodes accompanied by co-pathogens. Of 27 patients with pulmonary co-pathogens, 11 had more than one type. Aspergillus was most common, with 9 infections confirmed as Aspergillus fumigatus. Twelve patients had CMV pneumonia, and 10 had co-existing bacterial pathogens (5 Pseudomonas aeruginosa, 3 Streptococus pneumoniae, 2 Staphylococcus aureus, 3 coliforms). Other co-pathogens included 2 Candida species, 1 Mycobacterium fortuitum, and 1 rhizopus species.
Five BAL samples from five HCT recipients that had previously tested positive were negative upon retesting (2 RSV, 2 influenza A, one MPV); that is, viral RNA was below the limit of detection by quantitative RT-PCR. These samples were collected in 1990, 1999, and 2006 (3 samples) and had previously been positive by culture (RSV ×2), DFA (influenza A ×2, RSV), and RT-PCR (MPV). We analyzed quantitative viral load versus sample collection date and found no correlation between viral load and time over the study period to suggest that sample degradation may have consistently contributed to lack of RNA amplification (Pearson coefficient 0.08). Further analyses were performed using the 95 amplifiable BAL samples from 82 HCT recipients.
Three patients had two separate pneumonia episodes, and nine patients had more than one BAL sample per infection. Using the BAL with maximum quantitative viral load for each respiratory virus per pneumonia episode (85 BAL samples), median (range) respiratory virus copy number for each was: RSV (n=35) 2.6×106 (1.5×103 – 1.0×109) copies/ml, PIV (n=35) 4.9×107 (2.7×103 – 1.1×109) copies/ml, influenza (n=9) 6.8×105 (7.4×103 – 8.3×108) copies/ml, MPV (n=7) 3.9×107 (2.9×104 – 2.8×108) copies/ml, and CoV (n=4) 1.8×105 (2.5×103 – 2.0×107) copies/ml (Figure 2. A.).
For 77 patients, we examined the association of quantitative viral load in BAL samples from first episodes of pneumonia (33 RSV, 30 PIV, five MPV, eight influenza, and two CoV) with the presence of lymphopenia, co-pathogens, and need for mechanical ventilation. Three patients with multiple respiratory viruses co-detected in BAL samples, and two patients for whom the first positive BAL was not available, were excluded. No statistically significant difference was found for median quantitative viral load between patients with and without lymphopenia (≤100 lymphocytes/μl ≤7 days before BAL) or co-pathogens; there was a trend toward higher viral load in patients who required mechanical ventilation (p=0.06; Figure 3). There was no association between quantitative viral load measured in first BAL and overall survival at one or 6 months (data not shown).
One hundred thirty-two serum samples obtained from 66 HCT recipients near the date of diagnostic BAL were tested by quantitative RT-PCR, corresponding to 68 episodes of pneumonia: 41 RSV, 17 PIV, 5 influenza, three MPV, one PIV2/CoV co-infection, and one MPV/influenza A/CoV co-infection (Table 2). Forty-nine patients had concomitant BAL (n=55) and serum (n=95) samples. Respiratory viral RNA was detected in sera from six patients: four (10%) with RSV pneumonia, one with influenza B, and the patient with MPV/influenza A/CoV co-infection (influenza and MPV RNA detected). Median serum RNA detection in copies/ml for RSV was 5.3×102 (range 3.0×102 – 1.2×104) copies/ml, 3.3×102 copies/ml for influenza B, and 7.9×102 and 3.7×102 copies/ml for MPV and influenza A, respectively. The six positive serum samples were collected a median of one day after closest concomitant positive BAL sample (range, 11 days before to six days after), similar to the 126 negative serum samples (median one day after BAL, range 16 days before to 19 days after; p=0.78).
Figure 2. B. provides quantitative viral loads for BAL samples corresponding to negative and positive serum samples. For two patients with RSV RNA detected in serum, corresponding BAL samples with maximum viral load had 1.7×108 – 1.0×109 copies/ml RNA detected. One patient with detection of influenza B RNA in serum had 6.7×104 copies/ml in BAL, and the patient with MPV and influenza A detected in serum had 7.8×107 and 7.4×103 copies/ml, respectively, in BAL. No viral RNA was detected in serum of 18 patients with PIV pneumonia, even among patients with highest BAL viral loads.
The association of quantitative RSV viral load in BAL samples and detection of RSV RNA in serum was analyzed for 23 HCT recipients with RSV pneumonia and quantitative viral load from concurrent BAL and serum samples. A higher median maximum RSV BAL viral load was present in two patients with detection of RSV RNA in serum (5.9 ×108 copies/ml) compared with patients without (3.2×106, copies/ml); p=0.05.
All six patients with viral RNA detected in serum underwent allogeneic bone marrow transplantation with pneumonia and serum RNA detection in the first 120 days after HCT; additional characteristics are in Table 3. Serum viral RNA was detected in three patients after receiving at least a week of antiviral therapy, two with aerosolized ribavirin for RSV (patients #3, #4) and one with oseltamivir (patient #6). Positive serum samples were collected within 1–12 days of positive BAL. Five of the six patients died within one week of a positive serum or BAL sample and viral pneumonia was regarded as the final diagnosis and cause of death. Autopsies were performed for patients #5 and # 6; both had diffuse alveolar damage and negative viral cultures; focal bronchiolitis obliterans organizing pneumonia was also reported for patient #5. No patients with viral RNA detected in serum had bacterial, fungal, or viral co-pathogens in the BAL.
Characteristics from Table 1 were similar between patients with and without viral RNA detection in serum (data not shown). When the entire population was analyzed, there was no difference in timing of pneumonia diagnosis after HCT. However, for the subset of 40 patients with RSV pneumonia after HCT, the 4 patients with RSV RNA detected in serum were diagnosed with pneumonia earlier after transplant than patients without RSV RNA in serum (median 10 days [range 4–27] versus 48 days [range 8–237], p=0.02).
For first episodes of respiratory virus-associated pneumonia, detection of viral RNA in serum was assessed as a risk factor for receipt of mechanical ventilation and death within 30 days after first positive BAL. This analysis was restricted to the 59 patients who underwent allogeneic transplantation. In univariate analysis, patients with detection of viral RNA in serum had increased risk of mechanical ventilation (RR 2.4, p=0.02) and death (RR 2.0, p=0.005) within 30 days of first positive BAL; the association with death persisted in an adjusted model (RR for death 1.8, p=0.02; Table 4).
We have presented the first description of quantitative viral load in BAL and detection of viral RNA in serum samples among HCT recipients with respiratory virus pneumonia. We found high viral loads in BAL samples for all virus types, comparable to another study in which MPV was identified by quantitative RT-PCR in BAL and bronchial wash samples from predominantly immunocompromised patients . We also detected viral RNA in serum among a subset of HCT recipients early after transplantation with respiratory virus pneumonia caused by RSV, influenza, and MPV, but not among patients with PIV or CoV. Patients with detection of viral RNA in serum had an increased risk of death, suggesting that RNA detection may correlate with disease severity and poor outcome. Among patients with RSV, patients with RNA detected in serum were diagnosed earlier after transplant than patients without.
It is possible that detection of respiratory viral RNA in serum of these patients may indicate systemic viral dissemination associated with poor prognosis. We did not have access to other specimens to evaluate for viral presence or replication in extrapulmonary sites. Pathogenic viral dissemination is just one possible mechanism to explain our findings. Alternatives include physical release of intact viral particles into the circulation resulting from high viral load in the respiratory tract leading to epithelial cell death, or detection of virus or viral RNA from antigen-presenting cells, including pulmonary macrophages and dendritic cells, which gain direct access to the bloodstream during severe infection. It may be that detection of viral RNA in serum is typical in respiratory virus infections among HCT recipients, and that use of sensitive RT-PCR assays enabled us to detect this occurrence. Further studies are necessary to investigate whether patients with URI alone, or with URI that progresses to lower tract disease, may also have respiratory virus RNA detected in serum.
Implications of viral RNA detection in serum likely vary for different viruses. Studies of avian influenza suggest that detection of viral RNA in serum/plasma is a marker for disease severity and poor outcome [16, 17, 27]. Indeed, in avian influenza, a viremic phase may contribute greatly to pathogenesis. Although we do not have definitive evidence of detection of replication-competent virus, we did find that detection of RNA in serum was associated with increased risk of death. Among patients with RSV, high levels of RNA were detected in concomitant BAL samples at the time of RNA detection in serum, suggesting that the amount of infecting virus may influence the likelihood of serum RNA detection. This may indicate that we are detecting RNA that has spilled over due to pulmonary tissue damage. However, for the two cases with influenza detected in serum, the corresponding BAL viral load was considerably lower, suggesting detection of a true viremic phase. Clinically, complications of influenza include myocarditis and encephalopathy, although whether these represent direct viral invasion or an aberrant host immune response is not clear [45–48]. These are important questions for investigation, since RT-PCR testing of serum for respiratory viruses may offer an adjunctive method for diagnosis of severe viral respiratory pneumonia, and may have important implications for therapy and monitoring of certain patients. Further study is needed to determine if viruses such as influenza and RSV have a true viremic phase associated with detection of viral RNA in serum; if so, this may indicate that aerosolized therapy may have limitations for some HCT patients with viral pneumonia. Treatment with systemic antivirals may be beneficial for these patients.
The major strength of this study is that it provides the first quantitative analysis of respiratory virus RNA in BAL samples and, to our knowledge, represents the largest study to evaluate respiratory viral load in BAL and serum samples from HCT recipients. Our stored repository of BAL and serum samples, prospectively collected during the study period, provided a valuable opportunity to evaluate a large quantity of specimens. Importantly, all patients received a standardized diagnostic work-up for pneumonia, including bronchoscopy at the first clinical or radiographic indication of lower respiratory disease. Although detection of viral RNA alone does not ensure that replicating virus is present, nucleic acid detection in serum has been associated with transmission and/or disease severity for both RNA and DNA viruses [24, 27, 35, 38, 49, 50]. PCR testing is often used for prompt, sensitive diagnosis of pneumonia in patients with possible lower respiratory tract disease, particularly in the immunocompromised population. Furthermore, many transplant centers are moving toward molecular-based laboratory techniques for diagnosis of respiratory viruses so that PCR testing of serum may soon be widely available. Our infrequent sampling of serum may have even underestimated the frequency of detection of circulating RNA. More frequent and regular sampling, as well as cell-based assays performed at the time of pneumonia, may increase the frequency of detection of respiratory viral RNA since others have reported respiratory viral RNA detection in whole blood or PBMC samples but not in sera [28, 30–32].
The retrospective nature of this study is limiting. Varying collection volumes for BAL samples may have influenced quantitative viral results, although we would not expect BAL dilution factors of 1–2 fold to cause major differences in viral load. In a previous study of quantitative testing in nasal wash samples in which collection volumes were recorded, uncorrected viral loads were compared with viral loads corrected by sample volume and differences in samples were within the range of assay reproducibility (<1/2 log10 viral load) . In this study, we assumed one copy of viral RNA equals one virus particle for all viruses studied, i.e., that we are detecting genomic RNA. We do not know the proportion of viral genomes that is infectious. If not all genomic copies in a specimen are infectious in vivo, quantifying viral RNA by PCR may overestimate the number of infectious particles. In addition, a small number of BAL samples tested negative by RT-PCR. At times, these samples were stored at −20°C, a temperature not optimal for long-term RNA storage. Thus, it is not surprising that degradation may have occurred and viral RNA was undetectable in a few samples. We did not see a trend of decreasing quantitative viral load over time to suggest that duration of storage impacted RNA amplification. Because we identified only six patients with viral RNA detected in concomitant serum samples, our ability to perform more rigorous multivariable analysis of outcomes was limited by sample size.
In conclusion, our analysis found that respiratory viruses may be detected at high virus copy numbers in BAL samples and that detection of viral RNA in serum may be more frequent than previously appreciated among HCT recipients with virologically-confirmed respiratory virus pneumonia early after transplantation. This study provides evidence that detection of respiratory virus RNA in the bloodstream of severely immunocompromised patients may be associated with poor outcome. Circulating respiratory viral RNA may provide some understanding of viral pathogenesis, especially for RSV infection in which RSV RNA was documented in serum of 10% of patients with pneumonia. Larger studies are needed to validate these findings and determine whether detection of respiratory virus RNA in serum may be associated with disease severity in HCT recipients and other severely immunocompromised populations.
Financial support: This work was supported by NIH grants CA 18029, HL081595, K23HL091059, L40 AI071572, K24HL093294, and K24AI 071113. A.P.C. also received support from the MedImmune Pediatric Fellowship Grant Award and the Pediatric Infectious Diseases Society Fellowship Award funded by MedImmune.
We thank Amalia Magaret for providing statistical expertise. We also thank Craig Silva, Sanam Hussein, Peter Choe, Nido Nguyen, and George Counts for database services; and Anne Cent, Bruce Ulness, Nancy Wright, Terry Stevens-Ayers, Kristen White, Tera Matson, Sam Chatterton-Kirchmeier, Jason Daza, and Vikram Nayani for specimen processing, testing, and laboratory expertise.
Author contributions: A.P.C. performed the research, collected data, analyzed data, and wrote the manuscript; J.W.C. contributed to the study design and analysis plan, and critically reviewed the manuscript; J.K. performed PCR testing and critically reviewed the manuscript; J.A.E. contributed to the analysis plan and critically reviewed the manuscript; A.W. contributed to the analysis plan and critically reviewed the manuscript; K.A.G. provided statistical expertise and critically reviewed the manuscript; L.C. critically reviewed the manuscript and provided resources for the study; and M.B. designed and performed the research, collected data, analyzed data, and critically reviewed the manuscript.
Presented in part: Infectious Diseases Society of America Annual Meeting, San Diego, California, 6 October 2007
Potential conflicts of interest: A.P.C. received research support from MedImmune. J.A.E. received research support from Sanofi Pasteur, MedImmune, ADMA Biologics, Adamas Pharmaceuticals, and Novartis. M.B. received research support from ADMA Biologics, Adamas Pharmaceuticals, Roche Pharmaceuticals, and is a consultant for Roche Pharmaceuticals and Novartis. J.W.C., A.W., J.K., K.A.G., and L.C. do not have any potential conflict of interests.