Main

Although architectural immaturity and deficient surfactant production may be central in initiating RDS(1, 2), there is much evidence to suggest that long-term outcome may depend on the extent of the inflammatory response to ventilation(3). Lung lavage studies have demonstrated excess production within the first days of life of leukocyte proteases(4, 5) and macrophage cytokines such as TNF-α and IL-1(68). Both theα-chemokine (polymorphonuclear neutrophil-recruiting) IL-8(9, 10) and the β-chemokine(macrophage-recruiting) macrophage inflammatory protein-1α(MIP-1α)(7) have been demonstrated in high concentrations in bronchoalveolar lavage fluid. Conventional analysis of cellular aspirates has confirmed high level polymorphonuclear neutrophil recruitment(11), whereas immunohistochemistry has also shown equivalent numbers of macrophages and monocytes, much higher than have been identified by routine staining(7).

Despite evidence for the central role of macrophages in inducing fibrosis(12, 13), their contribution to the development of bronchopulmonary dysplasia is less clearly established than for polymorphonuclear neutrophils(11). However, there is clear evidence that excess macrophage activation is highly damaging within the pulmonary microenvironment, and that specific antagonism of macrophage cytokines or chemokines may be strikingly effective in reducing acute and chronic lung damage in animal models(1416).

We have therefore attempted to define the inflammatory response at the whole organ level, using carefully selected postmortem specimens, by determining immunohistochemically the distribution of macrophages, neutrophils, and TNF-α-immunoreactive cells. In addition we have used chargebased histochemistry specific for sulfated GAGs, which are known to be degraded by both TNF-α and IL-1(17) and which are extensively disrupted in intestinal inflammation(18). The physiologic importance of GAGs includes restriction of albumin and ion flux(1921), inhibition of vascular thrombosis(17, 22, 23), maintenance of tissue turgor and interstitial compliance(22, 24), inhibition of fibrosis in fetal animals(25), and control of cellular proliferation and differentiation(22, 26, 27). Congenital deficiency of GAGs on intestinal epithelium is associated with massive leakage of albumin and electrolytes, as well as impaired absorption(28). Similar disruption of sulfated GAGs in RDS would thus be likely to contribute to much of the characteristic pathophysiologic derangement.

METHODS

Use of whole-lobe postmortem specimens. The local inflammatory response was studied in formalin-fixed whole lung-lobe sections, obtained at postmortem examination from infants who had died during the course of acute RDS. These were selected from our bank of autopsy specimens by several criteria: 1) obtained from preterm infant with clinical history of acute RDS; 2) histologic changes typical of RDS (all had been reported by a neonatal pathologist as showing no inflammation); 3) no clinical or microbiologic evidence of pulmonary or systemic infection; and 4) postmortem examination performed within 48 h of death.

Many of the infants had died from acute events, such as pneumothorax or intraventricular hemorrhage, whereas none had received surfactant or steroid therapy. Five specimens were from infants stillborn at 24-28-wk gestation and 30 from preterm infants dying in the first week of life with acute RDS. In addition eight archival specimens, from infants who died with RDS in the early 1970s, were studied. These infants had never received ventilatory support, although all had been treated with oxygen. Details of all infants are given in Table 1.

Table 1 Clinical details and significant pathologic findings (in addition to RDS) of infants undergoing whole-lung postmortem examination

Immunohistochemical analysis. Serial sections were dewaxed and immunostained, using MAb recognizing CD68+ macrophages (PG-M1, Dako Corp., High Wycombe, UK), L1+ macrophages (MAC-387, Dako), HLA-DR(CR3/43, Dako), polymorphonuclear neutrophils (neutrophil elastase, Dako), and TNF-α (52-B-83, a gift from Celltech Ltd., Slough, UK). Staining was developed using alkaline phosphatase/anti-alkaline phosphatase immunohistochemistry as previously described(7, 18). Immunoreactive cell density was estimated by point counting with a Lennox graticule(29) over pulmonary lamina propria(excluding points overlying epithelium or airways).

Staining for sulfated GAGs. Sulfated GAGs were visualized with a 5-nm gold-conjugated poly-L-lysine probe (Biocell Research Laboratories, Cardiff, UK) applied at pH 1.2(17). After developing with a silver enhancer (Biocell) for approximately 15 min, the slides were counterstained in Mayer's hemalum and mounted in Apathy's medium. Semiquantitative assessment of the density of anionic residues within the pulmonary interstitium was made using a Lennox graticule(29), counting at least five high power fields, and scoring as positive graticule points overlying silver-stained tissue(18). Double staining after immunohistochemical analysis was performed as previously described(18).

Statistical analysis. Mean tissue densities of immunoreactive cells and sulfated GAGs were compared using the unmatched t test. Regression analysis was additionally used to compare the density of sulfated GAGs with inflammatory cells.

RESULTS

Inflammatory cell distribution. Macrophages and neutrophils were present within the pulmonary interstitium of the stillborn preterm infants at low density (<5 cells/mm2) and TNF-α immunoreactivity was low (9.75 cells/mm2 (Figs. 1 and 2). All liveborn infants who had developed RDS showed interstitial inflammation, including those infants who died between 5 and 12 h of age. Study of groups based on age at death showed a sequential increase in the mean density of all inflammatory cell types. This reached a maximum by 48-72 h, when the interstitial density of CD68+ macrophages was 81/mm2, neutrophils 51/mm2, and TNF-α+ cells 183/mm2 (Fig. 2). TNF-α immunoreactivity was not confined to cells and could be detected throughout the interstitium in many cases (Fig. 1), in contrast of tissue from stillborn infants or inflammatory controls (active Crohn's disease) stained at the same time with 52-B-83. The distribution of all inflammatory cell types appeared uniform within pulmonary interstitium, although macrophages but not neutrophils were frequently aggregated around airspaces and in hyaline membranes (Fig. 1). There was no evidence of focal perivascular accumulation. Again it was notable that CD68+ cells were highly variable in appearance and thus underestimated by conventional staining techniques. Almost all of the macrophages expressed the L1 antigen, as MAC-387+ staining of serial sections appeared identical.

Figure 1
figure 1

Immunohistochemical and histochemical analysis of postmortem whole lung lobe specimens. (A) Single CD68+ macrophage (red) in a 28-wk stillbirth. Similar appearances were seen in sections stained for TNF-α, neutrophil elastase, and MAC-387 (L1 antigen). (B) Interstitial infiltration of CD68+ cells in a ventilated 27-wk gestation infant at 23 h of age. (C) Multiple TNF-α+ cells, with significant noncellular immunoreactivity (not seen in stillbirths or infants dying in first 24 h), in 25-wk infant at 72 h.(D) Lung from 26-wk stillbirth double-stained for sulfated GAGs(black), showing a dense interstitial network, and CD68+ macrophages(red). (E) Double staining for neutrophil elastase (red) and sulfated GAGs (black) in 27-wk infant at 37 h. Note almost complete attenuation of interstitial and vascular GAGs. (F) Double-staining for MAC-387 (L1 antigen-red) and sulfated GAGs in 25-wk infant at 55 h, showing dense macrophage infiltration, and absent interstitial GAG staining, with marked GAG staining within the hyaline membrane.

Figure 2
figure 2

Immunostaining and histochemistry for sulfated GAGs in whole-lung postmortem specimens, showing mean interstitial density (with SE) in groups of infants dying at different ages. (A) Density of CD68+ macrophages (filled bars) and neutrophils (open bars) in five stillborn infants, five who died before 12 h, six between 12 and 23 h, six between 24 and 47 h, four between 48 and 72 h, six between 3 and 7 d, and eight between 7 and 14 d. (B). Temporal increase in mucosal TNF-α+ cell density (filled bars) and decrease in interstitial sulfated GAGs (open bars) in the same groups of infants as in (A).

Distribution of sulfated GAGs. Within the pulmonary interstitium from stillborn infants there was a dense meshwork of sulfated GAGs (mean density 697.5 U/mm2) (Figs. 1 and 2), confirmed by chondroitinase ABC digestion to be predominantly composed of chondroitin and dermatan sulfate (not shown). Vascular endothelium, perivascular connective tissue, and subepithelial basement membrane stained strongly in all specimens, confirmed by heparanase II digestion as predominantly heparan sulfate. In addition, cartilage around major airways stained intensely.

In all infants with RDS, there was significant attenuation of pulmonary interstitial GAGs, with almost 50% loss in infants dying at 12-24 h, and almost complete attenuation beyond 48-72 h (Figs. 1 and 2). Similar degradation of vascular endothelial and subepithelial basement membrane GAGs was also present in all specimens (Fig. 1). Significant negative correlation was found between the density of interstitial GAGs and of CD68 cells (r =-0.782, p < 0.00001), neutrophils (r = -0.722,p < 0.00001) and TNF-α+ cells (r = -0.733,p < 0.00001). In all specimens from 48 h there was striking and increasingly dense accumulation of stainable GAGs within the hyaline membranes, although this phenomenon was not detected in infants dying before this time. In addition, 6/18 infants with sufficient pulmonary cartilage for assessment showed abnormal staining, characterized by unstained lacunae around swollen chondrocytes.

Archival specimens from unventilated infants. Similar features of neutrophil and macrophage infiltration, TNF-α immunoreactivity, and disruption of sulfated GAGs were seen in all of these archival specimens. However, in contrast to the modern specimens from infants who had received positive pressure ventilation, bullous distortion of the immediate subpleural region with focal fibrotic change was also noted in 4/8 specimens.

DISCUSSION

We have studied by immunohistochemistry the distribution and density of macrophages expressing the surface marker CD68 and MAC-387 (L1 antigen), elastase-immmunoreactive neutrophils, and TNF-α-immunoreactive cells within the pulmonary interstitium of infants who died from acute RDS. The inflammatory changes were demonstrated in all studied infants and were much more extensive than could be identified, even retrospectively, by routine staining with hematoxylin and eosin. The interstitial density of CD68+ macrophages was at least 15-fold, and neutrophils 10-fold, higher in infants who died at 2-3 d age than in stillborn infants of equivalent gestation. The density of TNF-α+ cells increased similarly and was associated with marked noncellular immunoreactivity that was not seen in concomitantly stained inflammatory controls (inflammatory bowel disease specimens). This phenomenon was repeatedly seen, and may represent binding of secreted TNF-α to tissue receptors, although we did not confirm this by direct study of TNF-receptor expression.

The clinical status, physical size, and immaturity of preterm infants place severe constraints on the investigation of pulmonary inflammation in RDS. Bronchoscopic biopsy is likely to be unjustifiably hazardous, and thus the only direct insight into mucosal events may come from such autopsy studies. The major question underlying such work is the relevance of findings, in a group that is likely to be self-selected for severity, to the main population of infants with RDS. We have thus attempted, as far as possible, to use specimens obtained from infants with a classical history of RDS who died from acute events such as intraventricular hemorrhage. We also excluded infants who had clinical or histologic evidence of infection. None of these infants had received surfactant, and it is possible that different inflammatory patterns may be seen in surfactant-treated infants. However, our preliminary studies have not shown evidence of this. Whether surviving infants would have less severe interstitial inflammation is also unknown, although immunohistochemistry of lavaged cells shows an increase of bronchoalveolar macrophage and neutrophil density with a very similar temporal pattern(7). It was also noticeable that macrophages in particular tended to accumulate around alveoli, suggestive of an epithelial chemotactic effect.

Use of a specific cationic probe demonstrated extensive disruption of sulfated GAGs on vascular endothelium, basement membranes, and throughout the interstitium. There was a reciprocal relationship between the mucosal density of sulfated GAGs and of macrophages, neutrophils, and TNF-α cells. Thus similar changes are seen within pulmonary GAGs in acute RDS to those found in relation to inflammatory cells and their products in both in vitro studies of endothelial monolayers(17) and in inflammatory bowel diseases(18). We have also found similar changes within the lungs of five infants who died from pulmonary infection, suggesting that this phenomenon is not specific for RDS. The degradation of sulfated GAGs may be mediated locally by the macrophage-derived cytokines TNF-α and IL-1(17) or by neutrophil or macrophage production of MMP such as stromelysin (MMP-3)(30, 31). In cultured fetal intestinal explants, where T cell activation induces macrophage-mediated tissue destruction(32), we have shown similar disruption of matrix GAGs which was inhibited in a dose-dependent fashion by dexamethasone or the protease inhibitor α2-macroglobulin(33). The dynamic balance between production of MMP and their tissue inhibitors may thus determine the extent of matrix degradation(31). This may be directly amenable to therapeutic intervention, as has now been shown in experimental allergic encephalomyelitis(34).

Among the proinflammatory effects of macrophage and neutrophil activation, enzymatic or cytokine-mediated disruption of sulfated GAGs is likely to be of particular resonance within the pulmonary microenvironment(26). Loss of endothelial and epithelial heparan sulfate(28) will contribute to the increased permeability of pulmonary exchange vessels(35) and massive alveolar albumin leak characteristic of RDS(36), which is even greater in those who develop chronic lung disease(37). Inhibition of inflammatory GAG disruption(33) may thus explain the reduction of albumin leakage by dexamethasone therapy(38).

In addition to increased albumin leakage (which may inactivate surfactant)(36) and microthrombosis from loss of vascular GAGs, breakdown of the interstitial gel matrix is likely to affect pulmonary compliance. It was notable that several infants also showed abnormality of peribronchial cartilage, which represents another potential mechanism for alteration of lung function. The time course of interstitial GAG breakdown appears consistent with the decrement in pulmonary compliance and ventilation parameters seen in RDS over the first 48 h of life, and might explain in part the beneficial effects of fluid restriction.

However, in addition to sulfated GAG breakdown, increased fibroblast production of unsulfated and sulfated GAGs may be stimulated by pulmonary TNF and IL-1 production(39). Such duality of effect has been seen in a primate model of RDS(40). Thus the differential control of fibroblast function, leading to production of either normal extracellular matrix components or of collagen, is also likely to be of signal importance in determining overall outcome(1216). In this context it is interesting to note the focal subpleural changes in some of the archival infants who had never been artificially ventilated. This region of the lung is likely to have been subjected to the greatest shear stress, which has now been shown to induce fibroblast proliferation and activation in vitro(41).

Necrotic epithelial cells are thought to provide the initial framework for hyaline membranes(42). Our finding that GAG degradation products contribute to hyaline membrane structure is consistent with early experimental work in which exogenously applied GAGs interacted with plasma protein to induce hyaline membrane formation in vivo(43). This had suggested that the material destined to form hyaline membranes was present in fluid form and precipitated within the airway(43). It is notable that intense hyaline membrane staining for surfactant-associated proteins also occurs only after 48 h(44), and it is possible that the negative charge contributed by complexed GAGs may contribute to this sequestration. Surfactant may also be inactivated directly by leakage of albumin and other proteins(36).

These findings support our contention that routine staining techniques significantly underestimate the inflammatory response, in particular the macrophage component, in neonatal RDS. They show that the density of TNF-α immunoreactivity, and the extent of matrix degradation, may be at least comparable to that found in better documented inflammatory conditions. However, such changes occur within 2 d of birth. In addition to controlled attenuation of inflammatory infiltration by blockade of specific adhesion molecules(45) or targeting of individual chemokines or cytokines, inhibition of MMP activity by administration of a therapeutic tissue inhibitor of metalloproteinases may represent a third potential new approach to the prevention of inflammatory lung destruction in RDS.