Published ahead of print on January 18, 2007, doi:10.1164/rccm.200609-1322OC
© 2007 American Thoracic Society doi: 10.1164/rccm.200609-1322OC
Herpes Simplex Virus Lung Infection in Patients Undergoing Prolonged Mechanical Ventilation1 Service de Réanimation Médicale, 2 Service de Virologie, and 3 Service d'Anatomo-Pathologie, Groupe Hospitalier Pitié-Salpêtrière, Assistance Publique-Hôpitaux de Paris, Université Pierre et Marie Curie, Paris, France Correspondence and requests for reprints should be addressed to C.-E. Luyt, M.D., Ph.D., Service de Réanimation Médicale, Institut de Cardiologie, Groupe Hospitalier Pitié-Salpêtrière, Université de Paris Pierre et Marie Curie, Assistance Publique-Hôpitaux de Paris, 47, boulevard de l'Hôpital, 75651 Paris Cedex 13, France. E-mail: charles-edouard.luyt{at}psl.aphp.fr
Rationale: It is not known whether the isolation of herpes simplex virus (HSV) from lower respiratory tract samples of nonimmunocompromised ventilated patients corresponds to bronchial contamination from the mouth and/or throat, local tracheobronchial excretion of HSV, or true HSV lung involvement (bronchopneumonitis) with its own morbidity/mortality. Objectives: This prospective, single-center, observational study was conducted to define the frequency, risk factors, and relevance of HSV bronchopneumonitis. Methods: All consecutive nonimmunocompromised patients receiving mechanical ventilation for 5 days or more were evaluated. Bronchoalveolar lavage, oropharyngeal swabs, and bronchial biopsies (presence of macroscopic bronchial lesions) were obtained for all who deteriorated clinically with suspected lung infection. HSV bronchopneumonitis was defined as this deterioration, associated with HSV in bronchoalveolar lavage and HSV-specific nuclear inclusions in cells recovered during lavage or bronchial biopsies. Measurements and Main Results: HSV bronchopneumonitis was diagnosed in 42 (21%) of the 201 patients who deteriorated clinically, with a mean mechanical ventilation duration before diagnosis of 14 ± 6 days. Risk factors associated with HSV bronchopneumonitis were oral–labial lesions, HSV in the throat, and macroscopic bronchial lesions seen during bronchoscopy. Patients with HSV bronchopneumonitis were comparable to those without at admission, but their courses were more complicated, with longer duration of mechanical ventilation and intensive care unit stays. Conclusions: HSV bronchopneumonitis is common in nonimmunocompromised patients with prolonged mechanical ventilation, is associated with HSV reactivation or infection of the mouth and/or throat, and seems to be associated with poorer outcome.
Key Words: Herpesviridae • bronchopneumonia • intensive care unit • immunocompetence • ventilator-associated pneumonia
Herpes simplex virus (HSV) can be isolated from the saliva of 1 to 5% of normal subjects. In the intensive care unit (ICU), the frequency of its detection in the throat reached 41% in patients after surgery (1). It can be isolated from the tracheobronchial secretions of a high percentage of patients, and from 30 to 71% of those with acute respiratory distress syndrome (ARDS) (2–5). According to a large prospective study, HSV was recovered from the upper and lower respiratory tract of 22 and 16% of ICU patients, respectively (6). Those findings were confirmed in another study (5). In those studies, carriers with HSV in either the throat or respiratory tract had poorer outcomes than patients without HSV. However, the mere demonstration of a virus in tracheobronchial secretions does not permit any conclusion to be drawn as to its causative role in lung involvement. It is not known whether HSV isolation from lower respiratory tract samples of nonimmunocompromised ventilated patients corresponds to viral contamination of the lower respiratory tract from the mouth and/or throat, a local tracheobronchial excretion of the virus resulting from its reactivation without parenchymal involvement, or real HSV bronchopneumonitis (BPn; corresponding to HSV infection of the lung parenchyma). Therefore, the exact role of HSV in ICU patients—that is, whether it is a marker of disease severity or a true pathogen with its own morbidity/mortality— remains unclear (5–7). To date, HSV BPn has been described mostly in patients taking immunosuppressants (8), patients with ARDS (2), or burn victims (9).
We conducted a prospective study to determine the frequency, risk factors, and relevance of HSV BPn in nonimmunocompromised patients receiving prolonged ( Preliminary results of the present study were previously reported in an abstract (10).
Patient Selection Between October 2004 and January 2006, all consecutive patients ventilated for 5 days or more in our medical ICU were eligible and monitored prospectively. Patients were not included if they were (1) pregnant, (2) neutropenic (white blood cell counts < 1 x 109/L or neutrophils < 0.5 x 109/L), (3) had concomitant acquired immunodeficiency syndrome (stage 3 according to the Centers for the Disease Control and Prevention 1993 classification), or (4) had received immunosuppressants or long-term corticosteroid therapy, defined as prednisone or an equivalent at or exceeding 0.5 mg/kg per day for more than 1 month.
Study Design The sampling area was selected on the basis of the infiltrate location on the chest radiograph or the segment with purulent secretions visualized during bronchoscopy (12, 13). The presence of any tracheal or bronchial lesions was recorded by the endoscopist. For patients with bronchial lesions, for example, superficial ulcerations or erosions, and without contraindications, bronchial biopsies were taken, placed in formalin, and sent to the pathology department. An endotracheal aspirate and oropharyngeal swabs were obtained from all patients immediately before bronchoscopy and sent to the pathology department. If the attending physician noted an oral–labial lesion, suspected of being of HSV origin, vesicles were vigorously swabbed and samples were placed in virus transport medium and sent to the virology department. Blood was also drawn for HSV serology testing. For patients exhibiting more than one clinical sign of deterioration during his/her stay, the diagnostic work-up, described above, was repeated.
Procedures Slides prepared with endotracheal aspirate, BAL fluid, and bronchial biopsy sections were stained with Perls, Papanicolaou, May-Grünwald-Giemsa, and Grocott stains, and carefully examined for pathogens and HSV-specific nuclear inclusions (15). All slides were reviewed separately by two experienced pathologists (F.C. and M.-H.A.-L.), who were blinded to the clinical status of the patients and their virology results. For additional details on viral and pathological procedures, including virus load determination in BAL fluid, and on baseline assessment and data collection, see the online supplement.
Definitions The study was approved by the Ethics Committee of the Société de Réanimation de Langue Française (Paris, France), which waived the need for written consent, because the protocol did not impact on patient management and complied with standard care in our unit (Service de Réanimation Médicale, Groupe Hospitalier Pitié-Salpêtrière). However, all patients and/or next of kin were informed. In accordance with French law, the Commission Nationale de l'Informatique et des Libertés approved this project for the use of computerized medical data and protection of patient confidentiality.
Statistical Analysis To confirm the effect of HSV BPn on outcomes, a nested case–control analysis was also performed. Matching baseline (Day 5 of MV) criteria for the control subjects, scored "yes" or "no," were age ± 5 years, Simplified Acute Physiology Score (SAPS) II ± 5, McCabe and Jackson comorbidity score, postcardiac surgery reason to pursue MV, and MV duration at least equal to that of HSV BPn onset for the paired case. Analyses were performed with StatView 5.0 (SAS Institute, Inc., Cary, NC) and SPSS 11.5 (SPSS, Inc., Chicago, IL) software. Significance was defined as p < 0.05. Additional details on statistical analysis are available in the online supplement.
Among the 784 patients admitted to our ICU during the study period, 376 (48%) received MV and 238 underwent MV for 5 days or more. Thirty-four were not included, leaving 204 eligible patients (Figure 1). Among them, 201 (99%) experienced at least one episode of clinical deterioration and were evaluated for HSV infection. All these patients underwent bronchoscopy with BAL, oropharyngeal swabbing, and HSV serology testing. Macroscopic bronchial lesions were seen in 109 patients, who underwent bronchial biopsies without complications.
Oropharyngeal swabs yielded HSV from 109 patients, and PCR or culture of their BAL fluids detected HSV in 98. HSV BPn was diagnosed in 38 of those patients and 4 others with HSV-negative oropharyngeal swabs. Thus, 42 patients developed HSV BPn, after a mean of 14 ± 6 (range, 5–37) days of MV. Immunofluorescence labeling of HSV-positive cultures, regardless of their origin (i.e., oropharyngeal swab, oral–labial lesion swab, and/or BAL fluid), always identified HSV-1. HSV BPn was diagnosed on the basis of histologic examination of BAL fluids from 30 patients, BAL and bronchial biopsies from 5 patients, and bronchial biopsies alone for the remaining 7 patients. For all patients diagnosed with HSV BPn, including the seven in whom HSV-specific inclusions were seen only in bronchial biopsies, cytologic examination of the BAL fluid showed squamous cell metaplasia. Cytologic examination of tracheal aspirates detected nuclear inclusions suggestive of HSV infection in only 19 (45%) of the 42 patients with HSV BPn. At the time of diagnosis, 19 (45%) of the 42 patients had combined HSV BPn–bacterial VAP, whereas 23 (55%) had only HSV BPn. All patients with HSV BPn showed signs of clinical deterioration on D1 and then slowly improved. No significant differences were demonstrated in the clinical outcomes of HSV BPn between patients with and without associated bacterial VAP on D1, but the radiologic score and white blood cell count were higher and the PaO2/FIO2 ratio was lower for those with bacterial VAP (Figure 2).
Among the 42 patients with HSV BPn, 19 were given acyclovir. The decision to use it was left to the discretion of the treating physicians, and was based mainly on the importance of the oral–labial lesions. Clinical courses and outcomes did not differ significantly between acyclovir-treated and untreated patients, even though in-hospital mortality was slightly lower for the former (respectively, 37 vs. 57%) (see the online supplement).
Risk Factors for HSV BPn
The percentages of patients developing ARDS did not differ significantly between groups, but patients with HSV BPn tended to have a higher ARDS rate (Table 2). For all patients with HSV BPn and ARDS, the latter preceded HSV BPn onset or was present the day of the diagnosis. Moreover, these patients had significantly more frequent HSV oral–labial lesions, and HSV-positive throat and lower respiratory tract samples, than did patients without HSV BPn. In patients with HSV BPn and oral–labial vesicles (23 of 42), these lesions were detected on the same day as, or 1 to 5 days before, HSV BPn diagnosis. Macroscopic bronchial lesions too were seen significantly more frequently in patients with HSV BPn and they were limited to mucosal erythema and/or a few superficial ulcerations; no patient had bronchial mucosal vesicles.
Factors potentially associated with HSV BPn retained by multivariable analyses were oral–labial lesions, HSV-positive throat-swab cultures, and macroscopic bronchial lesions (Table 3).
HSV BPn and Outcome Patients with HSV BPn had longer durations of MV and ICU stays, and higher total numbers of VAP episodes while receiving MV (Table 4). The adjusted proportional hazards ratio for MV duration in patients with HSV BPn versus those without was 1.52 (95% confidence interval [CI], 0.94–2.46; p = 0.09) after adjustment for age, sex, McCabe and Jackson classification, postcardiac surgery reason to pursue MV, admission SAPS II, and baseline (Day 5 of MV) SOFA (Sepsis-related Organ Failure Assessment) score. However, the in-hospital mortality rates were comparable for the two groups.
As shown in Table 5, it was possible to pair every patient with HSV BPn with a well-matched control subject. Although patients with HSV BPn had longer durations of MV and ICU stays, the numbers of bacterial VAP episodes per patient and in-hospital mortality rates did not differ significantly between the two groups, but these values were slightly higher for patients with HSV BPn (p = 0.09 and 0.3, respectively).
Virus Load for the Diagnosis of HSV BPn Among the 42 patients with HSV BPn, BAL viral cultures were positive for 37 and negative for 5; for these latter, virus was detected exclusively by quantitative PCR. Box plots (Figure 3) illustrate the normalized virus loads for patients with and without HSV BPn: patients with HSV BPn had a higher normalized median virus load than those without (p < 0.0001). The ability to differentiate between patients with and without HSV BPn, on the basis of virus load, was assessed by ROC curve analysis (Figure 4). The area under the ROC curve was 0.89 (95% CI, 0.84–0.93) and significantly larger than that obtained when only clinical variables (oral–labial lesions and macroscopic bronchial lesions) were used to predict HSV BPn (area under the ROC curve, 0.73; 95% CI, 0.66–0.79; p = 0.004). To predict HSV BPn, a virus load cutoff value of 8 x 104 copies/106 cells had 81% sensitivity (95% CI, 69–90%) and 83% specificity (95% CI, 71–91%).
The results of this prospective, observational study showed that HSV was detected in the upper and lower respiratory tract of 54 and 64% of our patients, respectively. Moreover, HSV BPn was common in nonimmunocompromised (as defined herein) patients ventilated for 5 days or more, thus affecting 21% of the population evaluated. Potential risk factors for HSV BPn (i.e., those retained in multivariable analysis as being independently associated with it), were oral–labial lesions, HSV in the throat, and macroscopic bronchial lesions seen during fiberoptic bronchoscopy. Pertinently, patients with HSV BPn had poorer outcomes than those without: longer total durations of MV and ICU stays, and more bacterial VAP episodes per patient. These findings were confirmed by the nested case–control study: patients with HSV BPn underwent MV longer and stayed longer in the ICU than did matched patients. Although we detected HSV in BAL fluids from 64% our patients, including some without HSV BPn, the use of virus load normalized to the total number of cells recovered was a strong predictor of HSV BPn. Thus, the high sensitivity and specificity of this approach could allow its use as a surrogate for cytologic examination to diagnose HSV BPn. In accordance with previous studies (1, 5–7, 16, 17), HSV was detected in the throat and/or lower respiratory tract of a high percentage of our patients. Lower respiratory tract shedding was even higher in our study, probably because of the specific population that was included, which consisted of a subset of severely ill ICU patients, who required prolonged (at least 5 d) MV and had multiple organ/system dysfunction or failure. The authors of only a few studies were able to differentiate between asymptomatic carriage and/or local HSV shedding and true infection (lung parenchymal involvement), as defined by HSV detection and specific herpetic nuclear inclusions in the tracheal secretions (2, 4). Using a combination of three criteria to diagnose HSV BPn, we were able to demonstrate that, in 21% of our patients, lower respiratory tract HSV shedding was associated with histologically proven lung involvement. Risk factors for HSV reactivation and/or infection of the respiratory tract of nonimmunocompromised ICU patients remain poorly studied. In accordance with the few previously published investigations, postoperative status, underlying disease severity, HSV oral–labial lesions, macroscopic bronchial lesions, and HSV reactivation in the throat were identified as potential risk factors (1, 4–6). According to our multivariable analysis, the only three factors that remained significantly associated with HSV BPn were oral–labial lesions, positive throat swab cultures, and macroscopic bronchial lesions. The precise disease mechanism involved is not clearly understood. We hypothesized that, in most patients, HSV BPn was initially due to viral reactivation in the throat (possibly secondary to critical illness and local microtrauma caused by endotracheal and gastric tubes, and oropharyngeal cavity suctioning), followed by contamination, colonization, and infection of the bronchial tree and the lungs (descending infection). This hypothesis is supported by previously reported findings: on the basis of an autopsy series, Nash concluded that the anatomic distribution of HSV involvement in the tracheobronchial tree and the lungs suggests that aspiration or contiguous spread from the upper respiratory tract was the most likely dissemination pathway (18). Moreover, Bruynseels and coworkers showed that, for 72% of their patients with lower respiratory tract HSV-positive specimens, HSV was detected in the throat on the same day as, or before, detection in the lower respiratory tract (6). Our observations were similar: in our 23 patients with oral–labial lesions and HSV BPn, the lesions were detected before or the same day as HSV BPn diagnosis. Moreover, we found that oral–labial lesions and HSV detection in the throat were independent risk factors for HSV BPn. These data suggest that viral reactivation or infection in the oropharynx reaches the lower respiratory tract by aspiration (6). The presence of macroscopic bronchial lesions, possibly due to local microtrauma and/or preexisting acute lung injury with distal squamous cell metaplasia, might also have paved the way for distal infection. However, because we were unable to detect the virus in the throats of four patients with HSV BPn, another mechanism, that is, local distal reactivation and infection or hematogenous spread, cannot be categorically excluded. Several mechanisms could even differ from one patient to another. Regardless of the mechanism of lung involvement, HSV BPn probably resulted from reactivation of latent infection, because all patients had HSV-positive serology (IgG without IgM) at the time of diagnosis. Outcomes of ICU patients with HSV reactivation and/or infection seems to be worse than those of patients without, but the pathogenic role of HSV—that is, benign colonizer activated in proportion to the severity of the underlying illness or infectious agent with a true attributable morbidity and/or mortality—remains unclear (1, 5–7). Our findings are in accordance with those previously reported and extend them: using strict criteria to define HSV BPn, infected patients had poorer clinical outcomes, with prolonged MV duration and longer ICU stays, and more bacterial VAP episodes per patient. These results were confirmed when patients with HSV BPn were carefully matched to control subjects, who had the same disease severity at admission and received MV for at least a duration equal to that of HSV BPn onset for the paired case. Even though these observations argue in favor of a true pathogenic role of HSV in nonimmunocompromised, severely ill ICU patients, any conclusion regarding this issue seems premature and only a randomized controlled trial comparing a specific antiviral agent, for example, acyclovir, with placebo could attempt to answer this question. To date, only one study has shown that prophylactic acyclovir effectively prevented HSV infection in patients with ARDS, but their outcomes were not improved (3). Unfortunately, that study lacked sufficient power to be able to demonstrate a benefit associated with prophylaxis, because only 45 patients were included. Treating patients with documented HSV BPn or with high virus load might achieve better outcomes, but that remains to be determined. Giving acyclovir to ICU patients, especially when multiple organ failures are present, could lead to several potentially severe adverse effects, including renal dysfunction and neurotoxicity (19). Our study has several limitations. First, because our specialized referral ICU cares for patients with severe disease, receiving prolonged MV, and many postcardiac surgery patients with shock or other organ failures, it is difficult to extrapolate our results to other ICUs with different patient mixes. Second, bronchoscopy was done only when clinical deterioration was observed and therefore we might have missed some patients with only mild symptoms of HSV BPn. However, among the 204 eligible patients, 3 experienced no clinical deterioration while undergoing MV, did not undergo fiberoptic bronchoscopy, and were excluded. For all other patients, extreme vigilance was maintained throughout the entire time they required MV, and fiberoptic bronchoscopy with BAL was performed as soon as even minor clinical deterioration was noted. Third, we looked at a particular subgroup of mechanically ventilated patients, so we might have missed some patients with early-onset HSV infections, that is, those occurring before Day 5 of MV. However, authors of other studies reported that HSV colonization occurred after several days in the ICU (2, 4, 6). Furthermore, most of our patients from whom samples were collected before 7 days of MV had no detectable HSV in their lower respiratory tracts. Fourth, HSV BPn frequency could have been underestimated because only a small area of the lung surface was rinsed during BAL. However, bronchoscopy and BAL were oriented by chest radiography and bronchial lesions. Fifth, HSV BPn diagnosis was based on histologic examination of two different types of specimens, that is, BAL fluid and bronchial biopsies, but not on true lung biopsy. BAL is, nonetheless, a technique that enables the recovery of both cellular and noncellular components from the epithelial surface of the lower respiratory tract and alveoli (20). Sixth, although both BAL fluid and bronchial biopsies were positive for 10 patients, only one specimen was positive for others (BAL alone for 25, and bronchial biopsies alone for 7). Therefore, we might have inadvertently assimilated two different types of HSV lung involvement, one corresponding to proximal lesions of the tracheobronchial tree and the other to more distal lung involvement. In summary, HSV BPn appears to be a common disease in patients receiving prolonged MV. Risk factors for this disease are oral–labial lesions, HSV in the throat, and macroscopic bronchial lesions. HSV BPn seems to have a real impact on outcome, but only an interventional study will be able to determine its true impact in an ICU setting.
Supported in part by an unrestricted grant from the Société de Réanimation de Langue Française (SRLF). This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200609-1322OC on January 18, 2007 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form September 15, 2006; accepted in final form January 17, 2007
Related articles in AJRCCM:
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
5 d) mechanical ventilation (MV). HSV BPn was defined as clinical deterioration leading to suspicion of lower respiratory tract infection, associated with HSV in the lower respiratory tract and specific herpetic nuclear inclusions in cells collected during bronchoalveolar lavage (BAL) or in bronchial biopsies. 
