INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The high morbidity of patients with cystic fibrosis (CF), chronic obstructive pulmonary disease (COPD), and bronchiectasis is associated with recurrent and persistent lower respiratory tract infections due to various microorganisms. Among the bacteria most frequently isolated, nonencapsulated Haemophilus influenzae has been recognized as an important pulmonary pathogen. H. influenzae, although a commensal of the upper airway in healthy individuals, causes acute upper and lower respiratory tract infections, and persists in the lower respiratory tract of CF and COPD patients (1). Because H. influenzae is associated with the occurrence of exacerbations in CF and COPD patients (2, 5), persistent H. influenzae in the lower respiratory tract of these patients may contribute to the high pulmonary morbidity, in addition to the contribution of other frequently occurring pathogens. It has been suggested that H. influenzae hides in the lower respiratory tract during persistence, thereby evading host immune defenses and the effect of antibiotic treatment (3, 4, 6). The microbial flora (5), and the pathology of the airways of CF patients in autopsied cases (7) have been studied, but no data about the localization of H. influenzae are available.

Examination of lungs from autopsied COPD patients revealed epithelial and subepithelial localization of H. influenzae during respiratory tract infections (8). In vitro experiments using cell lines and organ cultures showed that H. influenzae penetrated into the subepithelial layer by passing between epithelial cells (9, 10). Since in adenoid tissue of young children H. influenzae was detected intracellularly in macrophage-like cells localized in the subepithelial layer (11), these cells may contribute to the persistence of H. influenzae. A serious problem encountered in studies using autopsied lungs is the bacterial overgrowth occurring during the postmortem interval prior to autopsy, particularly in CF patients with multiple colonizing or infecting microorganisms. Generally, the period of time between the explanation of the lungs from lung transplant recipients and microbiological examination is shorter for explanted lungs than for lungs obtained by autopsy. Therefore, we determined the presence and localization of H. influenzae in explanted lungs from patients undergoing lung transplantation because of end-stage pulmonary disease due to various disorders consisting of CF, COPD, bronchiectasis, and noninfectious pulmonary diseases. Sputum and lung specimens from all patients were analyzed using selective culture methods and nonculture detection methods. The presence of H. influenzae in tissue sections from the explanted lungs was studied by immunoperoxidase (IP) staining, and by polymerase chain reaction (PCR) (12) followed by DNA hybridization and DNA sequence analysis of the PCR products. The results of this study show that H. influenzae was more frequently detected in tissue sections of lung explants from patients with CF and COPD than from patients with bronchiectasis or noninfectious pulmonary diseases. H. influenzae was diffusely present in the respiratory epithelium and subepithelial layers of the lungs of patients with end-stage pulmonary disease, especially patients with CF and COPD.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

In total, 49 patients (20 females and 29 males) with end-stage pulmonary disease were included in the study. These patients comprised 16 CF patients, 16 COPD patients (five patients with chronic bronchitis and 11 patients with emphysema according to the criteria of the American Thoracic Society [13]), five bronchiectasis patients, and 12 patients with noninfectious pulmonary diseases consisting of pulmonary hypertension (n = 9), idiopathic pulmonary fibrosis (n = 2), and Langerhans cell histiocytosis (n = 1). The diagnosis of CF was based on a sweat chloride concentration of > 70 mmol/L, COPD was defined according to the American Thoracic Society criteria (13), and bronchiectasis was defined by the presence of purulent secretions and caliber changes of the airways as evidenced by high-resolution computed tomography (HRCT) scanning. In contrast to the infectious pulmonary diseases CF, COPD, and bronchiectasis, the other pulmonary diseases were regarded as noninfectious. Characteristics of the patients are summarized in Table 1. The age distribution of the patients ranged from 12 to 62 yr (median age, 40 yr). Nearly all patients with CF and COPD had a FEV₁ of less than 30% of the predicted value (14). After institutional approval, patients were accepted for lung transplantation at the University Hospital Groningen between November 1990 and December 1995. Patients were accepted for lung transplantation if they had a life expectancy of less than 12 to 18 mo and if their quality of life was severely impaired by respiratory disease, as assessed by a stepwise selection procedure including an extensive review of the patients' medical history and pretransplant performance (15). A bilateral lung transplantation was performed in 16 CF patients, 15 COPD patients, five bronchiectasis patients, and three patients with noninfectious pulmonary diseases including pulmonary hypertension (n = 2), and idiopathic pulmonary fibrosis (n = 1). Unilateral lung transplantation was performed in patients with pulmonary hypertension (n = 7), COPD (n = 1), idiopathic pulmonary fibrosis (n = 1), and Langerhans cell histiocytosis (n = 1). Perioperative antibiotic prophylaxis consisted of intravenous treatment with ceftazidime (initial dose 2 g, thereafter 1 g every 3 h) started 1/2 h before surgery. Explanted lungs, collected in sterile containers and stored at 4° C within 30 after explantation, were transported to the Department of Pathology for examination within 4 h after collection.

Culturing and IP Staining of Expectorated Sputum Specimens and Specimens from the Explanted Lungs

Six to 12 mo preoperatively, at least two expectorated sputum specimens, including one specimen obtained just prior to lung transplantation, were cultured from all patients. Within 4 h after lung explantation, specimens were obtained from the main bronchus and peripherally from the explanted lungs by aspiration with a sterile syringe or taking a swab if no aspirate could be obtained. Expectorated sputum specimens and lung specimens were cultured by standard culture techniques (16). Specimens from CF patients were also cultured on selective media appropriate to recover Staphylococcus aureus, H. influenzae, Pseudomonas aeruginosa, and Burkholderia cepacia (5, 17). Lung specimens were also cultured in brain heart infusion broth (Oxoid, Basingstoke, Hampshire, UK) supplemented with saponin (0.1% wt/vol) (Sigma Chemical Co., St. Louis, MO) and hemin (5 mg/L) (Sigma) for the recovery of intracellularly localized bacteria (9). The bacterial isolates were identified by standard procedures (18). H. influenzae was identified as growth-dependent for hemin and nicotinamide-adenine dinucleotide (NAD) using a disk test with X and V factors (19). IP staining using monoclonal antibody (MoAb) 8BD9 for the detection of H. influenzae in sputum specimens and specimens from the explanted lungs was performed as described previously (20, 21).

Major Outer Membrane Protein (MOMP) and Randomly Amplified Polymorphic DNA (RAPD) Analysis of H. influenzae

H. influenzae isolates obtained as single colonies from primary culture media were characterized phenotypically by MOMP and genotypically by RAPD analysis to distinguish distinct strains from MOMP variant strains (4). Distinct H. influenzae strains differ in both their MOMP and RAPD patterns, whereas MOMP variants differ in MOMP pattern but not in RAPD pattern (4).

Preparation of Lung Tissue Sections

Cryostat sections were routinely obtained from the main bronchus, the first to fifth bronchus generation, and the peripheral site of the explanted lungs within 6 h after explantation. These specimens were immediately snap frozen in liquid isopentane and stored at 70° C. From the same sites of the lung as used for the preparation of cryostat sections, tissue specimens were fixed in 4% (vol/vol) formalin for at least 24 h. After fixation, these specimens were routinely embedded in paraffin. Subsequently, 20 serial sections of 3 µm thickness were cut from each specimen, yielding a total of approximately 80 sections per explant from each patient. Sections were deparaffinized in xylol and rehydrated with ethanol-water mixtures. Endogenous cellular peroxidase activity was reduced by incubation with 0.3% (vol/vol) H₂O₂ in phosphate-buffered saline (PBS) (pH 7.2) for 30 min.

Detection of H. influenzae in Lung Tissue Sections by IP Staining

IP staining using MoAb 8BD9 was applied on deparaffinized tissue sections as described for sputum specimens (20). In each staining run, a slide with a smear from a pure culture of H. influenzae was used as a positive control. Slides with a smear of H. influenzae that were not exposed to MoAb 8BD9 were included as negative controls. The slides were counterstained with hematoxylin. Microscopic examination of the IP-stained slides at ×1,000 magnification with oil immersion was performed blindly with respect to the culture results for H. influenzae from the explanted lungs. In an IP-positive tissue section, rods characteristic for H. influenzae were seen in at least one microscopic field.

Detection of H. influenzae in Tissue Sections by PCR

From each explant, one tissue section with a thickness of 25 µm was used for PCR amplification of the MOMP P6 gene of H. influenzae. Tissue sections were placed in 200 ml of digestion buffer containing 50 mM Tris-HCl, pH 8.5, 1 mM EDTA, 0.5% (vol/vol) Tween 20, and 200 µg/ml proteinase K at 55° C overnight to release total DNA (23). After subsequent heating at 100° C for 10 min and cooling of the specimens to 18° C, digested material was extracted with 250 µl chloroform added to the reaction mixture followed by centrifugation (10,000 × g for 5 min). A 5-µl volume of the supernatant was used to perform the PCR with the primer set designed to amplify the MOMP P6 gene of H. influenzae as described previously (12), using 35 cycles of amplification (1 min at 95° C, 1 min at 55° C, and 2 min at 72° C) in a programmable DNA thermal cycler (Perkin-Elmer Cetus, Gouda, the Netherlands). The products were analyzed by electrophoresis in 2% agarose gels with a 123-bp DNA ladder (Gibco; Bethesda Research Laboratory [BRL], Grand Island, NY), as a molecular weight marker. The presence of components inhibiting the PCR was determined by spiking the tissue sections with H. influenzae DNA. The presence of eukaryotic DNA was examined by amplifying the -globulin gene as previously described (12, 24).

Southern blot hybridization of the PCR products was performed as previously reported (12). Briefly, DNA was transferred onto nylon filters (Zetaprobe; Bio-Rad Laboratories, Veenendaal, The Netherlands) under alkaline conditions (0.4 M NaOH), the filters were hybridized with a digoxigenin (DIG)-labeled oligonucleotide probe (HI-VI) derived from the MOMP P6 gene of H. influenzae (12) using the DIG DNA labeling kit (Boehringer Mannheim, Mannheim, Germany), followed by washing, blocking, and immunochemically staining of the bound probe using the DIG detection kit (Boehringer Mannheim).

The specificity of the PCR products, obtained after amplification of a H. influenzae culture-positive tissue section and a H. influenzae culture-negative but IP-positive section, was further established by DNA sequence analysis of the PCR products. The sequence reaction was performed with the Dye Terminator Cycle Sequencing Kit (Perkin-Elmer Cetus 402117) using primers derived from MOMP P6 gene, as described previously (12). Sequences were analyzed by the automatic sequencer ABI 373 (Perkin-Elmer Cetus) using computer programs which included the program package Sequence Navigator (Applied Biosystems, Perkin Elmer, San Jose, CA) and PC/GENE (IntelliGenetics, Inc., 1991, Mountain View, CA).

Indirect Double Immunocytochemical Staining of Tissue Sections

An indirect double immunocytochemical staining procedure was performed to identify whether H. influenzae was associated with macrophages. From two CF patients with IP-positive lung tissue sections for H. influenzae, consecutive paraffin-embedded tissue sections were first incubated with MoAb 8BD9 (IgG2a) diluted in PBS to which 0.5% (wt/vol) bovine serum albumin (BSA) and 15% normal rabbit serum (Gibco BRL, Life Technologies LTD, Paisley, Scotland) were added (20). The slides were washed three times in PBS for 5 min and endogenous peroxidase activity was reduced by incubation with 0.3% (vol/vol) H₂O₂ in PBS for 30 min. Binding of MoAb 8BD9 was detected with peroxidase-conjugated goat anti-mouse IgG2a antibodies (Dakopatts) diluted in PBS. For the identification of macrophages, the same slides were subsequently incubated with anti-CD68 antigen (M814; Dako A/S, Glostrup, Denmark) diluted in PBS-BSA (25). The slides were washed three times in PBS for 5 min. The binding of the CD68-specific antibodies was detected with biotin-labeled goat anti-mouse IgG1 (Southern Biotechnology Associates, Birmingham, AL) followed by a biotin-streptavidin alkaline phosphatase labeling system and staining by Fast Blue BB (Sigma Chemical Co.) as chromogen in conjunction with the substrate naphthol AS-MX (Sigma), yielding a blue reaction product. Subsequently, the peroxidase-labeled MoAb 8BD9 was detected by 3-amino-9-ethylcarbazole and H₂O₂ resulting in a brown-staining of H. influenzae, as described previously (20). Microscopic examination of the stained slides was performed as described above for IP staining.

Statistics

Statistical analysis of the presence of H. influenzae in the various patient groups was performed by chi-square analysis with Bonferroni correction for multiple comparisons. p Values of < 0.05 were considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cultures of Expectorated Sputum Specimens and Specimens from Explanted Lungs

Six to 12 mo prior to lung transplantation, H. influenzae and S. aureus were mainly recovered from sputum specimens of a limited number of patients, particularly CF and COPD patients (Table 2). P. aeruginosa was isolated from nearly all CF and bronchiectasis patients. Other Pseudomonas spp. were rarely isolated. Cultures from specimens of the explanted lungs yielded H. influenzae in only one CF patient and one COPD patient. Prior to lung transplantation sputum specimens of both patients were also culture-positive for H. influenzae. Pseudomonas spp., including 13 isolates of P. aeruginosa, were the most frequently isolated pathogens from the explanted lungs. The culture results of the expectorated sputum specimens obtained from 49 patients just before surgery were similar to the results of the specimens cultured from the main bronchus and/or peripherally from the explanted lungs. The results of specimens from the explanted lungs cultured in saponin broth and on standard and selective culture media were identical, suggesting that bacteria present in these lung specimens were mainly localized extracellularly.

IP Staining of Expectorated Sputum Specimens and Specimens from Explanted Lungs

H. influenzae was identified by IP staining in expectorated sputum specimens obtained 6 to 12 mo prior to lung transplantation from 17 of 43 (48%) patients, and just prior to lung transplantation from 24 of 49 (49%) patients. The IP staining results of the paired expectorated sputum specimens collected from 43 of 49 patients were the same except for three patients. IP staining results obtained from the expectorated sputum specimens just prior to lung transplantation and from the specimens of the explanted lungs corresponded completely. H. influenzae could not be detected by IP staining in specimens from the explanted lungs in 4 of 11 (36%) patients with H. influenzae culture-positive results of their sputum specimens 6 to 12 mo prior to lung transplantation.

IP Staining of Tissue Sections

H. influenzae was identified by IP staining in tissue sections from two H. influenzae culture-positive patients (one CF patient and one COPD patient) and in 22 of 47 (47%) H. influenzae culture-negative patients. IP staining results of the tissue sections corresponded with those of the specimens from the explanted lungs for each individual patient. Control slides without exposure to H. influenzae-specific MoAb 8BD9 were always IP-negative, indicating MoAb 8BD9 specificity.

H. influenzae was present in 748 of 1,280 (58%) tissue sections obtained from 16 CF patients, in 600 of 1,280 (47%) tissue sections from 16 COPD patients, in 133 of 400 (33%) tissue sections from five bronchiectasis patients, and in 282 of 960 (29%) tissue sections from 12 patients with noninfectious pulmonary diseases. H. influenzae was more frequently detected in tissue sections from patients with CF and COPD compared with patients with bronchiectasis (² = 77.5, p < 0.0001, and ² = 23.0, p < 0.0001, respectively) and noninfectious pulmonary diseases (² = 186.5, p < 0.0001, and ² = 70.4, p < 0.0001, respectively). No difference in the number of IP-positive lung tissue sections was observed between patients with CF and COPD (² = 0.04, p = 0.85).

PCR of H. influenzae DNA in Tissue Sections and DNA Sequence Analysis of PCR Products

PCR amplification with the H. influenzae MOMP P6 gene specific primer set of DNA extracted from tissue sections of both patients who were H. influenzae culture-positive and 14 of 22 IP-positive but H. influenzae culture-negative patients resulted in clearly visible bands of the correct molecular weight as detected by gel electrophoresis (Figure 1, Table 3). These 16 PCR bands hybridized with the H. influenzae MOMP P6 gene derived probe HI-VI. PCR products from tissue sections of the remaining eight IP-positive but H. influenzae culture-negative patients were only observed after hybridization of the gel with the MOMP P6 specific gene probe. The PCR and hybridization results were negative for all patients with IP-negative tissue sections (Figure 1, Table ).

View larger version (77K): [in this window] [in a new window]   Figure 1.   Polymerase chain reaction (A) and hybridization with a H. influenzae-specific probe (B) of lung tissue sections obtained from a H. influenzae culture-positive patient (C+), immunoperoxidase (IP)-negative, and H. influenzae culture-negative patient (C), and H. influenzae culture-negative patients (lanes 1-8). A 123 base-pairs marker is indicated at left of figure (M). IP staining results of sections in the same area of the lung are shown below the lanes.

DNA sequence analysis of PCR products obtained from the lung tissue sections of the CF patient who was H. influenzae culture-positive and IP-positive and a CF patient who was H. influenzae culture-negative but IP-positive revealed a 100% sequence homology with the published nucleotide sequences of MOMP P6 of H. influenzae (26). These findings indicate that cross reactions of the primer set with Haemophilus parainfluenzae or other closely related Haemophilus species are very unlikely.

Distribution of H. influenzae in Tissue Sections from IP-positive Patients

The distribution of H. influenzae in tissue sections obtained from the main bronchus, first to fifth bronchus generation, and peripheral sites of the lungs of patients with CF, COPD, bronchiectasis, and noninfectious pulmonary diseases as determined by IP staining, is shown in Table 4. H. influenzae was detected by IP staining in the majority (> 83%) of tissue sections examined, irrespective of the patient group. Although major differences were found between the numbers of H. influenzae-positive tissue sections in the various patients groups, H. influenzae was detected in tissue sections at various levels of the lungs in all patient groups. This result indicates a generalized distribution of H. influenzae in lungs of patients with end-stage pulmonary diseases. In tissue sections of the first bronchus generation, H. influenzae was detected in the (mucus) layer overlying the surface epithelium of the bronchus and in association with the bronchial epithelial cells. Some sections showed the presence of stained bacterial clusters associated with the ciliated epithelium of the bronchus. In other sections H. influenzae had locally invaded the bronchial epithelium and was detected in association with the epithelial cells and inflammatory cells infiltrating the bronchial mucosa. H. influenzae was also diffusely spread in the epithelial and submucosal area of the bronchi. Visually damaged epithelium, cartilage, and the fibroinflammatory tissue of dilatated bronchi were occasionally infiltrated by IP-positive bacteria. H. influenzae was further detected in association with the epithelial cells and in the submucosa of intralobular bronchioles. In sections of parenchymal lung tissue and visceral pleura, H. influenzae was found in small numbers in the alveolar epithelium, in the interstitium, in the surrounding alveoli, and in the visceral pleura. Local clustering of H. influenzae was observed in the corresponding areas of serial tissue sections obtained from different sites of the lungs.

In order to assess the relationship between H. influenzae and tissue macrophages, tissue sections from two CF patients including one patient who was H. influenzae culture-positive and IP-positive and one patient who was H. influenzae culture-negative but IP-positive, were analyzed by indirect double immunocytochemical staining. H. influenzae was present in these sections at the same sites as in a subsequent IP-stained section of the same tissue specimen. H. influenzae was detected in close association with alveolar and interstitial macrophages of the lungs. It was not possible to determine by electron microscopy whether bacteria were localized intracellularly or extracellularly attached to the macrophage cells of the lungs.

MOMP and RAPD Analysis of H. influenzae Isolates from Sputum and Lung Specimens

To determine the presence and distribution of distinct H. influenzae strains or MOMP variants in the respiratory tract, isolates were obtained from different sites of the explanted lungs from the only H. influenzae culture-positive CF patient. In total, 152 single H. influenzae colonies were analyzed for their MOMP and RAPD pattern. One H. influenzae isolate was obtained from an expectorated sputum specimen. Specimens from the first bronchus generation (58 colonies), second bronchus generation (45 colonies), third bronchus generation (24 colonies), and peripheral sites (24 colonies) of the lungs were obtained. In total, three H. influenzae MOMP P2 variants with an identical RAPD pattern were identified. One of these MOMP variants was present in the sputum specimen, as well as in cultures from the first, second, and third bronchus generation of the left lung. The second MOMP variant was isolated only from the first bronchus generation of the left lung. The third MOMP variant was isolated from a peripheral site of the left lung and from the first bronchus generation of the right lung. These results show that MOMP variants of H. influenzae were present in different areas of both lungs, although they were not restricted to one lung or a part of the lung.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The presence and distribution of H. influenzae was determined in explanted lungs from lung transplantation recipients with end-stage pulmonary disease. Bacterial overgrowth and extensive tissue deterioration frequently occurring in autopsied material was prevented by the procedure applied after explantation of the lungs. Compared with lung biopsy material, examination of explanted lungs provides the advantage that various levels in the same lung can be analyzed. However, lung transplantation recipients have reached an end-stage of their pulmonary disease. Therefore, our findings may represent features predominantly present during the end-stage of various pulmonary diseases.

The culture results of expectorated sputum specimens collected just before transplantation and specimens from the explanted lungs were similar, suggesting that the bacterial flora of expectorated sputum specimens is representative for the bacteria present in the lower airways. Despite extensive perioperative antibiotic therapy, culture results of specimens obtained from the explanted lungs revealed frequently Pseudomonas spp., similar to results of sputum specimens cultured before lung transplantation. The correlation between culture results of sputum and lung specimens for CF patients has been previously described (27). Other microorganisms such as H. influenzae were less often recovered from specimens collected just prior to lung transplantation or from the explanted lungs (Table ), probably owing to perioperative antibiotic treatment (28). H. influenzae in the respiratory tracts of patients with end-stage pulmonary disease was detected more frequently by IP staining than by culture. This is in accordance with earlier data in which IP staining was shown to be a valuable addition to culture (21). Including the two H. influenzae culture-positive patients, IP-positive results were obtained from specimens of the explanted lungs in almost half of the patients. The large discrepancy between culture and IP staining results may be due to the presence of Pseudomonas spp. which exert an inhibitory effect on the growth of H. influenzae (29). This assumption is supported by the finding that discrepancies between IP staining and culture results occurred more frequently among CF than COPD patients in whom P. aeruginosa was isolated rarely from respiratory tract specimens.

Using IP staining, H. influenzae was detected in tissue sections of the explanted lungs from the same patients with IP-positive lung specimens (Figure 1). IP staining was previously shown to be a suitable technique for the detection of H. influenzae in sections from various human tissues (22). Although IP staining using MoAb 8BD9 has a high sensitivity and specificity for the detection of H. influenzae in clinical specimens (21), cross-reactivity of MoAb 8BD9 with the subspecies H. aegyptius, H. haemolyticus, and a minority of H. parainfluenzae isolates has been demonstrated (20). It is unlikely that H. aegyptius and H. haemolyticus were present in the lower respiratory tract, because these species are rarely isolated from patients with respiratory tract infections (19). H. parainfluenzae was not cultured from lung specimens of any patient during the study. The IP staining results of tissue sections were confirmed by PCR using primers specific for the gene encoding MOMP P6 of H. influenzae and/or DNA hybridization with P6 gene as probe (12). This PCR was shown to be also positive for H. haemolyticus and H. aegyptius, but these species were not cultured from the explanted lungs. In addition, sequence analysis of PCR products from a H. influenzae culture-positive and a culture-negative but IP-positive lung tissue section showed a 100% homology with the published nucleotide sequences of the MOMP P6 gene of H. influenzae (26). Therefore, the results of IP staining, PCR with DNA hybridization, and sequence analysis of the PCR products all indicate the presence of H. influenzae in lung specimens of culture-negative patients with end-stage pulmonary disease.

H. influenzae was much more frequently detected by IP staining in tissue sections from patients with CF and COPD compared with patients with bronchiectasis and noninfectious pulmonary diseases. The frequency of H. influenzae detection in CF and COPD patients was similar. The low frequency of H. influenzae present in bronchiectasis patients was not expected, but the significance of this finding is not clear due to the small number of patients (n = 5) analyzed. The presence of H. influenzae in patients with noninfectious pulmonary diseases indicates that H. influenzae also penetrated the airway epithelium of these patients with end-stage pulmonary disease. H. influenzae was shown to be able to penetrate between epithelial cells in vitro (9, 10), and to enter the mucosa of healthy children without symptoms (11). These observations strongly suggest that H. influenzae is an invasive pathogen, persisting in patients with severe pulmonary disease, and that H. influenzae spreads in lungs of CF and COPD patients. If H. influenzae was present in the lungs of patients with end-stage pulmonary disease, this bacterium was diffusely present in the bronchi, the large and small bronchioles, the bronchioloectatic areas, and the damaged epithelium of the airways (Table ). In addition to the epithelial and subepithelial localization of H. influenzae in autopsied COPD lungs reported previously (8), H. influenzae was widely distributed in the lungs of patients with end-stage pulmonary disease. The wide distribution of H. influenzae in the lung also differs from that of P. aeruginosa, which is predominantly present endobronchially and endobronchiolarly in the small airways of CF patients with limited invasion of the lung parenchyma (30). H. influenzae was mainly localized extracellularly either alone or in bacterial clusters. For the H. influenzae culture-positive CF patient, the wide distribution of H. influenzae in the lung was demonstrated by molecular analysis of the strains. MOMP variants of one distinct H. influenzae strain were isolated from various sites of the explanted lung, and two MOMP variants were present in both lungs at the same time. These observations suggest diffuse spreading of H. influenzae into the lung after establishment in the bronchi and bronchioli, particularly of CF and COPD patients. During persistence, MOMP variants appear which also spread throughout the lung and from one lung to the other. The diffuse extracellular presence of H. influenzae in the explanted lungs may serve as a reservoir for this bacterium during persistence (8, 31), at least for patients with end-stage pulmonary disease. It has been suggested that cell wall- deficient bacteria induced by -lactam antibiotic treatment are responsible for the persistence of H. influenzae in the respiratory tract (32).

The close relationship between H. influenzae and alveolar and interstitial macrophages of the lung may reflect activity of the natural immune defense mechanism against bacterial intruders. It is not known whether this association provides a mechanism by which H. influenzae is cleared from the lungs, but pulmonary macrophage phagocytosis of H. influenzae has been shown to be rather inactive in CF patients (33). Additionally, most bacteria were observed extracellularly in the lung, and macrophages densely packed with H. influenzae as found in human adenoid tissue of children (11) were not detected. Alternatively, the tissue macrophages ineffective in killing H. influenzae may be a reservoir for H. influenzae during persistence, because the bacteria cannot be recognized by antibodies and complement-mediated defense mechanisms.

We conclude that H. influenzae is often present in the lungs of patients with end-stage pulmonary disease, especially patients with CF and COPD. H. influenzae is diffusely present in the lungs at all levels and occurs mostly extracellularly in patients with end-stage pulmonary disease. Further prospective studies are needed to assess the effect of H. influenzae infection on the progression of pulmonary damage and to determine whether an early start of antibiotic treatment during an exacerbation may postpone persistent H. influenzae in the lower respiratory tract of patients with end-stage pulmonary disease.