© 2002 American Thoracic Society
Three-Dimensional Characterization of Pathologic Lesions in Pulmonary Langerhans Cell HistiocytosisService d'Anatomie Pathologique, Hôpital Avicenne, Bobigny; Inserm U82 and Inserm U408, Faculté de Médecine Xavier Bichat, Université Paris 7; Service d'Anatomie Biomédicale des Saints Pères, Université Paris 5; and Inserm U552, Hôpital Bichat-Claude Bernard, Paris, France Correspondence and requests for reprints should be addressed to Paul Soler, Ph.D., Inserm U408, BP 416, 75870 Paris cedex 18, France. E-mail: soler{at}bichat.inserm.fr
The characteristic lesions of pulmonary Langerhans cell histiocytosis (LCH) associate destructive granulomas containing large numbers of Langerhans cells and cysts. The lesions are usually considered to develop around small airways, and cysts are thought to result from destruction of the bronchiolar wall by the granulomatous reaction. However, the extent to which the granulomatous reaction is truly bronchocentric remains unknown, and the mode of formation of the cysts has not been defined. By using serial sections, this study aimed to explore further the relationships between pulmonary LCH lesions and distal airways, and the development of cysts. The results demonstrated that the granulomatous process of pulmonary LCH affected exclusively small airways, in an acinar distribution. The lesions extended without interruption along the bronchiolar axis, forming a continuous sheath around distal airways. The granulomatous reaction seemed to progress along the bronchiolar axis over time, extending the abnormalities in both the proximal and distal directions. Cystic lesions resulted from the destruction of the bronchiolar wall and progressive dilatation of the lumen, subsequently circumscribed by fibrous tissue. Because pulmonary LCH lesions affect and progressively destroy distal airways, it may be proper to consider the disease a bronchiolitis rather than an interstitial lung disorder.
Key Words: Langerhans cell histocytosis • 3-D modeling • lesions
Pulmonary Langerhans cell histiocytosis (LCH), also called histiocytosis X, is a disease of unknown etiology characterized by the presence in the lung of destructive granulomatous lesions that in the active stages of the disease contain large numbers of activated Langerhans cells (LCs) (1–7). Typically, the granulomas are poorly delimited focal lesions that are scattered throughout the lung and separated by apparently normal lung parenchyma. Lesions of apparently different age are usually found in the same biopsy. The earliest lesions are composed of cellular infiltrates containing large aggregates of LCs associated with lymphocytes and eosinophils, often adjacent to small airways. In the middle stages of their evolution, the lesions increase in size, LCs decrease in number, and inflammatory cells, especially eosinophils and macrophages, become predominant. Fibrotic changes begin to appear at this stage, both within and at the periphery of the lesions. The process heals by fibrosis, leading to the formation of stellate scars or cystic cavities. Despite extensive evaluation, several important questions remain unanswered concerning the pathology of granulomatous lesions in pulmonary LCH. First, the lesions are commonly described as focal, nodular infiltrates with irregular borders. Although this gives the impression that the granulomas have a spherical form, this cannot be determined from the evaluation of standard pathologic sections, and attempts at three-dimensional reconstruction of these lesions have not been reported. Second, the extent to which the granulomatous reaction is truly bronchocentric remains unknown. It is well established that encroachment of lesions on bronchioles is a typical feature of LCH (2), and small airway involvement is seen in a high percentage of biopsies (1, 3, 5). In a given specimen, however, not all lesions contain identifiable bronchiolar structures, and biopsies without apparent bronchiolar involvement have been described, raising the possibility that lesions can arise de novo in the alveolar parenchyma. Because LCH is a destructive process, however, it is also possible that preexisting bronchioles in such "parenchymal" lesions have been obliterated. Another area of uncertainty is the relationship between lesions of apparently different age, which are typically present in biopsies from patients with pulmonary LCH (3, 5). This finding could result from the episodic initiation of new lesions over a considerable time span. Alternatively, the onset of the pathologic process could be more constrained in time, with existing lesions slowly progressing to involve adjacent tissues. In either case, random sections through the lung would be expected to traverse both recent and established lesions. Finally, central cavitation of some, but not all, LCH granulomas is frequently observed (1–3). It has been suggested that cavities can either represent the remnants of an airway, or arise de novo in cellular lesions (3). Necrosis of LCH granulomas is common in some tissues (e.g., bone and skin), but necrosis has been identified in only a minority of biopsies from patients with pulmonary LCH (2, 3, 5). Thus, the mechanisms responsible for the formation of the cavitary lesions and their subsequent evolution toward cysts have not been defined. To address these questions, we have prepared serial sections of open lung biopsies from patients with pulmonary LCH, and systematically examined these specimens to characterize the distribution, topography, and evolution of the lesions. The results demonstrate that the granulomatous process of pulmonary LCH is bronchocentric. The lesions extend without interruption along the bronchiolar axis, and appear to progress over time to involve both more proximal and distal portions of the airways. Cystic lesions are produced by the destruction of the bronchiolar wall, followed by dilatation of the lumen and subsequent encasement by fibrous tissue.
Study Populations All archival specimens of pulmonary tissue from patients diagnosed as having pulmonary LCH were screened. Specimens that contained only late fibrotic lesions or large coalescent granulomas without intervening normal lung parenchyma were excluded. Twelve biopsies obtained from seven patients with active pulmonary LCH were selected for further study. All patients (two men and five women; mean age, 26 ± 7 years) had clinical symptoms and results of chest radiographs and/or high-resolution computed tomography that were consistent with pulmonary LCH. All patients were current smokers. Lung tissue was obtained at the time of open thoracotomy performed to establish the diagnosis or for treatment of recurrent pneumothoraces. None of the patients was receiving therapy at the time of the biopsy.
Histopathologic Techniques
Immunohistochemical Techniques
Three-Dimensional Reconstruction of a Granuloma
Statistical Analysis
Distribution and Topography of Pulmonary LCH Lesions Pathologic findings typical of pulmonary LCH were observed in all samples examined in this study. As previously described, the lesions were focal and separated by intervals of apparently normal lung parenchyma. All specimens contained both cellular, granulomatous lesions in various stages of evolution and fibrous scars and cysts of variable diameter. Granulomatous lesions were identified in sections from one end of the series, and followed in adjacent sections. The proximal/distal orientation of the sections was determined by following the branching pattern of the airways. The lesions were found to extend over considerable distances. In 32 of the 36 granulomas evaluated, the lesions continued without interruption throughout the sections available for study, and therefore could achieve a length of more than 5 mm. In a given section, the diameter of the lesions was variable, but usually ranged from 2 to 4 mm. Thus, the lesions were elongate structures, and did not have a spherical shape.
Relationship of Granulomas to Bronchovascular Structures
In the example shown in Figure 2, the most proximal sections showed a relatively early granulomatous lesion invading the bronchiole in an eccentric fashion (Figure 2A). In more distal sections, the florid granuloma progressively increased in size, and the entire perimeter of the bronchiole was invaded. As the granulomatous process was destructive, often only remnants of the bronchiolar structures were visible (Figure 2B). In some individual sections (Figure 2C), all evidence of the preexisting bronchiole had been destroyed. Even in these cases, however, remnants of bronchioles or accompanying vascular structures were identified in adjacent sections both proximal and distal to these sites. Using this approach, we found that all 36 lesions studied closely followed the distribution of preexisting bronchiolar structures over their entire length. Focal involvement of a given bronchiole at two distinct sites with an intervening interval of histologically normal airway was never observed.
The progressive destruction of bronchiolar walls by LC granulomas was particularly evident on sections in which LCs were stained by immunohistochemical techniques using anti-CD1a antibodies. In early lesions, LCs invaded bronchiolar walls in an eccentric fashion, accumulating between the epithelial lining and the underlying smooth muscle bundles (Figure 3A). As the lesions advanced, the LCs were present in dramatically increased numbers, penetrating the bronchiolar wall through gaps in the smooth muscle layer and infiltrating into the bronchiolar epithelium (Figure 3B). In subsequent sections, CD1a+ LCs were organized in a granulomatous mass, and only small residual remnants of the bronchiolar structures remained (Figure 3C).
In 32 of the 36 lesions studied, the granuloma persisted in the most proximal sections available. In the remaining four cases, the proximal end of the lesion could be identified, and was always localized to a terminal bronchiole. In several cases, the lesions could be followed distally to the point where respiratory bronchioles gave way to alveolar ducts, indicating that the most peripheral airways could be involved. In this regard, alveolar openings were frequently visible adjacent to the granulomas, especially when the lesions were not too large or destructive, suggesting that respiratory bronchioles were primarily affected. Quantitative assessment demonstrated that respiratory bronchioles were more often involved by the pathologic process than terminal bronchioles (75 and 25%, respectively). Because biopsies were generally taken from the periphery of the lung, however, the overrepresentation of smaller airways in these samples may have contributed to this finding. In 12 of the 36 lesions evaluated through serial sections, 1 or more bifurcations of the bronchovascular tree were encountered in the involved segment. In 9 of 12 cases, both daughter bronchioles continued to be included in the pathologic process. More rarely, only one of the two subdivisions was affected. Apparently intact bronchioles were commonly observed in the vicinity of bronchioles implicated by the granulomatous reaction. We were unable to establish, however, whether the granulomatous reaction extended to the presumed site of communication of these adjacent airways.
Evolution of Histopathologic Changes within Individual Lesions
Lesions appeared to be able to propagate in both the proximal-to-distal and distal-to-proximal directions. In 15 of the 36 lesions evaluated, the granulomas were more cellular and less fibrotic in the proximal sections than in the distal sections. Conversely, in 20 of 36 cases, more extensive fibrotic changes were observed in proximal sections, whereas the lesion was more florid and cellular in the distal sections (p > 0.2, comparing the direction of progression). In only one case (a lesion with mild fibrotic changes) were the histologic abnormalities relatively constant over the sections studied. Taken together, these findings strongly suggest that the lesions progress both over time and over space. It should be noted, however, that the time required for the lesions to propagate along the bronchiolar axis, as well as the time required for granulomas to develop and evolve into end-stage fibrotic lesions, could not be evaluated from our study. As indicated above, the proximal end of four lesions was identified. It is noteworthy that despite the relatively short distance between the normal bronchiole and the involved segment (10 sections; 150 µm), the granulomatous lesion appeared to be well established. Thus, in these cases, the progression of the granulomatous reaction to involve more proximal portions appears to have been arrested.
Development of Cavitary Lesions and Scarring
The frequency with which fibrotic lesions assumed the form of thick-walled cysts, nodules, and stellate scars appeared to depend, at least in part, on the diameter of the airway involved. As discussed above, in 20 of 36 lesions, fibrosis was more advanced in the proximal portion of the involved airway (e.g., where fibrosis was more likely to involve airways of larger caliber). In these cases, thick-walled cysts were relatively common in the most proximal sections (9 of 20 lesions, Table 1). In the lesions where fibrosis was more advanced in the distal portion of the involved airway (e.g., in airways of smaller caliber), thick-walled cysts were encountered less frequently in the most distal sections (3 of 15 lesions, p < 0.05).
When present, the diameter of the central cavities often appeared larger than that of adjacent normal airways. Several mechanisms appeared to contribute to the enlargement of these cavities. First, the destruction of the bronchiolar epithelium and connective tissue framework by the granulomatous reaction appeared to permit the enlargement of the lumen (compare Figures 5A and 5B). At sites just distal to bifurcations, the lumen of the two affected subdivisions could communicate, producing even larger cavities. Similarly, the coalescence of cavitary lesions involving adjacent alveolar ducts led to the formation of large cysts with irregular contours. Traction emphysema of adjacent alveolar ducts and alveoli also contributed to the cystic appearance of end-stage lesions (Figure 5D), as is observed in a variety of other pathologic processes in the lung.
Three-Dimensional Reconstruction of a Granuloma
This study, based on the analysis of serial sections of lung tissue from patients with pulmonary LCH, provides several insights into the nature of the granulomatous reaction. First, LC granulomas were found to be systematically centered on small airways, extending without interruption down to and including alveolar ducts. Second, the granulomatous reaction appears to progress along the bronchiolar axis over time and is capable of extending the abnormalities in both the proximal and distal directions. Third, the cavities present in pulmonary LCH lesions represent the often highly distorted bronchiolar lumen remaining after destruction of the airway walls, and can evolve into thick- or thin-walled cysts when they are subsequently encased in fibrous tissue. These results suggest that pulmonary LCH should be considered a bronchiolitis, not a diffuse or interstitial lung disease. Several previous authors have emphasized the predilection for LCH granulomas to form adjacent to bronchioles (1–3, 5–7). Because the granulomatous reaction destroys the bronchiole, however, not all lesions have appeared to be associated with airways (1). By carefully analyzing serial sections of the same lesion, we found in this study that all the granulomas evaluated had a strictly bronchocentric distribution. Two aspects of our study were important in allowing us to establish this point. First, we carefully sought often subtle signs of the prior existence of bronchioles at the site of the lesions. This included both the detection of remnants of bronchiolar structures within the lesions (e.g., short segments of epithelium lining the central cavity of the granuloma or smooth muscle bundles present in the lesion), as well as the persistence of structures that normally are found adjacent to bronchioles (e.g., lymphatics, arterioles, and alveolar entrance rings). Second, the ability to analyze serial sections of the same lesion was essential. Even where preexisting bronchiolar structures were not recognizable in a given section, their presence in immediately adjacent sections both proximal and distal to these areas allowed us to infer that the lesions were following the bronchovascular axis. Thus, unlike other immune granulomatous disorders, such as sarcoidosis, where specific lesions are spherical in shape and fully delimited (9), the granulomatous process of pulmonary LCH was found to spread along the bronchiolar axis, forming a continuous sheath around distal airways, often infiltrating adjacent alveolar structures.
It has previously been recognized that lesions in different stages of development are frequently encountered in a given histologic section of tissue involved with pulmonary LCH (3, 5–7). Our evaluation of serial sections of single granulomatous lesions may help explain the mechanism responsible for this observation. We found that the apparent stage of development of a given lesion evolved as the lesion was followed in serial sections. This evolution always followed an orderly progression that encompassed part of the spectrum from early invasion of the bronchiolar wall by LCs These observations provide several potentially important insights into the development of pulmonary LCH. First, the existence of lesions of different age in the same biopsy has suggested that new lesions must develop in the lung over an extended period of time. Our findings are more consistent with the alternative hypothesis that the onset of the granulomatous reaction could be relatively constrained in time, but the process subsequently progresses slowly along the bronchiolar axis as it extends to involve adjacent segments of the airway. It should be stressed, however, that evaluation of biopsies taken at a single point in time cannot provide definitive evidence concerning the temporal evolution of the process. Our results also give some insight into the sites at which the granulomatous reaction may develop. If the site of the initial lesions were restricted to more proximal conducting airways (present in small numbers in the open lung biopsies evaluated by us) or in the most distal respiratory bronchioles, preferential extension in the distal or proximal directions, respectively, would be expected. We observed, however, that the proportion of lesions extending proximally and distally was approximately the same, suggesting that lesions may originate frequently in small terminal bronchioles and respiratory bronchioles. The possibility that the lesions can originate at discrete sites in airways of diverse size cannot be excluded by these findings, but the absence of skip lesions weighs against this idea. It is noteworthy that at sites of bifurcations, both daughter bronchioles were sometimes involved in the granulomatous process. The fact that the lesions could involve both daughter bronchioles raises the possibility that entire acinar units could be destroyed en bloc by the granulomatous reaction. The coalescence of lesions involving multiple ramifications of a given airway might explain the appearance of large nodular masses of up to 1 cm in diameter that can be observed on computed tomography scans (10, 11) or histologic sections (3). Because of their divergent orientation, it was usually possible to follow both daughter bronchioles for only short distances, which limited our ability to confirm this possibility. Finally, the reconstruction of LC granulomas firmly supports the idea that the cavities observed in these lesions essentially always originate from the lumen of the involved bronchiole. The lumen of an involved bronchiole could be generally followed, often without interruption, throughout a given lesion, including areas where the preexisting bronchiole had been destroyed. Considerable distortion of the central cavities was observed as the lesions progress, resulting from the enlargement of the lumen after destruction of the bronchiolar wall, the establishment of communications between daughter bronchioles near sites of bifurcation, and the coalescence of adjoining lesions involving smaller respiratory bronchioles and alveolar ducts. The granulomatous reaction may contribute to the enlargement of the cavities in other ways, such as the loss of bronchiolar muscle or expansion by inflammatory infiltrates transiently present within the lumen. In addition, processes observed in the course of other lung diseases may play a role, including traction by adjacent fibrosis, and distension proximal to obstructive lesions. Finally, in lesions where the proximal portion of the airway was obstructed by the granulomatous reaction, increased intralumenal pressure in the distal airway during expiration could also promote dilatation. The pathogenesis of pulmonary LCH remains unknown, although several factors have been identified that may predispose to the disease (6, 7). In this regard, it has been recognized for many years that most adult patients with pulmonary LCH are cigarette smokers, although the reason that smoking predisposes to the disease has remained obscure (1, 2–7, 12). The exquisitely bronchiolar distribution of the lesions demonstrated in the present study supports the idea that smoking-induced alterations in the distal airways are important for the development of pulmonary LCH. In conclusion, pulmonary LCH lesions form continuous sleevelike structures that are centered on distal airways. Thus, it appears to be more accurate to classify the disease as a form of bronchiolitis, rather than as an interstitial lung disorder. The appreciation that the lesions are bronchiolar, not interstitial, and that they progressively destroy the airways in an acinar distribution helps explain many of the unusual clinical and radiologic features of the disease.
The contribution of Dr. Martine Antoine (Hôpital Tenon, Paris) in obtaining pathologic samples and the technical help of Marie Odile Vendeuil (Hôpital Saint Louis, Paris) in processing pathologic samples are gratefully acknowledged. Received in original form January 22, 2002; accepted in final form August 7, 2002
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
2 test. p < 0.05 was considered significant. 
florid cellular granuloma 
