INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary Langerhans cell histiocytosis (PLCH), also referred to as pulmonary histiocytosis X, is a disorder of unknown etiology characterized by the presence in the lung of granulomas containing large numbers of Langerhans cells (LC) that invade and destroy distal bronchioles (1). Typically, lesions of apparently different age can be found in a single biopsy specimen. Early lesions have a granulomatous appearance. They are mainly composed of LC accumulated in large numbers and associated with inflammatory cells, predominantly lymphocytes. These florid granulomas are often centered by a small cavity corresponding to the residual lumen of the preexisting bronchiole. In more advanced stages, the granulomas are clearly cavitary. They are still inflammatory, although fibrotic changes begin to occur in the periphery. As the lesions evolve, they are less and less cellular, whereas fibrotic changes are prominent, and late stages are characterized by cystic lesions surrounded by fibrous rings of variable thickness. Given the destructive nature of PLCH granulomas, when florid lesions rich in LC are present in lung tissue, the disease is usually considered as "histopathologically" active, independently of the clinical status of the patient. This is of critical importance, as the presence of florid lesions is compatible with a progressive disease.

Although the diagnosis of PLCH and the activity of the pathological process can be established with certainty only by examination of pathologic material, high-resolution computed tomography (HRCT) has proved to be of considerable value in the diagnosis and follow-up of the disease (5). HRCT allows the identification of characteristic cystic lesions in essentially all patients, including patients whose chest radiographs were interpreted as being normal or showing interstitial abnormalities. These cysts are usually thin walled and are frequently associated with small nodular lesions, sometimes cavitary. When the characteristic nodular and cystic lesions are present together, they are highly suggestive of PLCH, especially when they predominate in the upper to middle lung fields. Little is known, however, concerning the histopathological counterpart of the characteristic lesions shown by HRCT in the course of the disease. In this respect, nodules shown by HRCT are usually supposed to represent florid granulomatous lesions, but the precise relationship between these entities remains to be established. Likewise, it remains to be determined whether thin-wall cysts observed by HRCT, which are supposed to be cicatricial cavitary lesions, are actually fibrous cysts, or still inflammatory cavitary granulomas. Studies comparing both HRCT and histopathological findings could be helpful in assessing the relevance of HRCT as a noninvasive test to estimate the histopathological activity of PLCH.

In this context, we compared quantitative data obtained by HRCT and data obtained by histopathological examination of corresponding lung tissue specimens in patients with biopsy-proven PLCH. Only pulmonary samples from patients evaluated by HRCT within a range of 1 mo of the biopsy were included in the study. The nature and extent of both HRCT and histopathological lesions were compared to determine their respective counterpart. In particular, cystic lesions visible on CT scan were analyzed to determine whether they could represent still inflammatory lesions.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

Tissue biopsies obtained by thoracotomy or videothoracic surgery in the course of diagnostic evaluation (n = 11), or at the time of lung transplantation (n = 2), were obtained from 13 patients (10 men; mean age, 36 ± 11 yr; all but one current smokers). All but two patients had isolated lung involvement at the time of thoracic surgery and were evaluated by HRCT within a range of 1 mo of the biopsy, or 1 mo prior to lung transplantation. Clinical data are summarized in Tables 1 and 2. Final diagnosis was established by standard histopathological criteria, including electron microscopy and/or immunohistochemistry.

Study Design

Histopathological analysis. Lung tissue was fixed, processed by routine techniques, and sections were stained with hematoxylin-eosin. All samples were large enough (about 2 × 1 cm) to perform the histopathological analysis accurately. For each patient, three sections taken at different levels were evaluated to determine the distribution and extent of the characteristic lesions encountered in PLCH: florid granulomas, inflammatory cavitary lesions (cavitary granulomas), and late-stage cavitary lesions (fibrous cysts). The lesions were first evaluated according to a 0 to ++ scoring system. They were then enumerated, and their relative frequency was expressed as the number of lesions per square micrometer of lung tissue. Analysis of sections was performed simultaneously by consensus by three histopathologists (P.S., M.K., and O.G.).

HRCT analysis. HRCT examination consisted of a series of 1- or 1.5-mm-thick sections separated by 10-mm gaps. Scanning was performed from the apex of the lungs to the costophrenic angles on a CE 10000 unit (Thomson CGR, Buc, France) or a Toshiba X-Press unit (Tokyo, Japan). CT scans were interpreted simultaneously by consensus by two chest radiologists (M.G. and P.G.). For each patient, the CT scan was evaluated to determine the distribution and extent of the characteristic abnormalities observed in the course of PLCH (micronodules, nodules, thick- and thin-wall cysts) using a 0 to ++ scoring system. A quantitative analysis of the predominant lesion was then performed on the whole lobe corresponding to the site of the biopsy using a 5 × 5-mm grid according to Miller and coworkers (11) and Lucidarme and coworkers (12). The numbers of squares containing the lesions or overlying lung tissue were counted separately. A main lesion score (CT score) was calculated as the ratio of the number of squares containing the lesion to the total number of squares counted. The two observers easily established a consensus about the quantification of the lesions. Thus, given the small number of patients included in the study, the assessment of the interobserver variability did not appear to be necessary. For the two transplanted patients, the CT score was calculated as the average of the different slices of the whole lung.

Comparison between histopathological and HRCT findings. To better evaluate the precise relationships between HRCT and histopathological abnormalities, the nature and extent of the predominant lesions scored on CT scans were compared with the nature and density of lesions evaluated on biopsy samples. Correlations were sought between nodular abnormalities and florid granulomatous lesions, respectively, as well as between cystic abnormalities and cavitary lesions (both cavitary granulomas and fibrous cysts). Statistical analysis was performed using linear regression to test for correlations.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Histopathological Findings

Typical lesions of PLCH were found in all samples examined. They were focal and separated from each other by intervals of apparently normal lung parenchyma. They essentially consisted of florid granulomas and cavitary lesions, which could be either cavitary granulomas or fibrous cysts. Florid granulomas, which were early lesions, were irregular in size and poorly demarcated. They were sometimes compact, but often presented a small central cavity (Figure 1A). These lesions were mainly composed of typical LC associated with macrophages and inflammatory cells, particularly lymphocytes and eosinophils (Figure 1B). Cavitary granulomas, which represented more advanced lesions, were characterized by a prominent central cavity, few fibrotic changes, but numerous LC and inflammatory cells in their walls (Figures 1C and 1D). Fibrous cysts of variable size, representing late-stage lesions, were demarcated by a fibrous ring of variable thickness, containing no LC and few or no inflammatory cells (Figure 2). They produced a paracicatricial enlargement of the adjacent alveolar structures, contributing to the cystic appearance of the surrounding tissue. Florid and/or cavitary granulomas were present in almost all patients (12 of 13 patients), whereas fibrous cysts were observed less constantly (7 of 13 patients).

View larger version (150K): [in this window] [in a new window]   View larger version (146K): [in this window] [in a new window]   View larger version (115K): [in this window] [in a new window]   View larger version (145K): [in this window] [in a new window]   Figure 1.   (A) Typical florid granulomatous lesion of PLCH presenting a small central cavity. (B) High-power view of the region framed in A. The granuloma is mainly composed of Langerhans cells and lymphocytes, associated with neutrophils and eosinophils. The Langerhans cells are recognizable by their pale folded nucleus and weakly stained cytoplasm. (C ) Large cavitary granuloma corresponding to a more advanced lesion and characterized by a prominent central cavity. (D) High-power view of the region framed in C. The wall of the granuloma is still composed of numerous Langerhans cells associated with inflammatory cells. Hematoxylin and eosin; original magnification: A, ×80; B and D, ×250; C, ×31.

View larger version (95K): [in this window] [in a new window]   Figure 2.   Cicatricial cystic lesions (cl) surrounded by a fibrous ring. Note the paracicatricial enlargement (e) of adjacent alveolar structures. Hematoxylin and eosin; original magnification: ×31.

HRCT Findings

Typical abnormalities usually observed in the course of PLCH were found on CT scan. They consisted of micronodules (less than 5 mm in diameter), nodules (more than 5 mm in diameter), thick-wall cysts (more than 1 mm thickness), and thin-wall cysts (1 mm thickness or less). Poorly defined groundglass centrilobular nodules and nonnodular areas of groundglass opacities were not observed. Nodules and thick-wall cysts were rarely found, whereas micronodules or thin-wall cysts were the most frequent abnormalities present in all patients (Table 3 and Figure 3). In a given patient, micronodules or thin-wall cysts were either unique or largely predominant and, thus, to test for correlations, patients were divided into two groups according to their CT scan pattern: patients with a predominant nodular pattern (n = 7) and patients with a predominant cystic pattern (n = 6).

View larger version (99K): [in this window] [in a new window]   View larger version (109K): [in this window] [in a new window]   Figure 3.   (A) HRCT of the lung of patient 6 showing well-marginated nodules and micronodules. (B) HRCT of the lung of patient 5 showing thin-walled, joined, and confluent cysts, and a pneumothorax.

Correlations between Histopathological and HRCT Findings

Analysis of histopathological lesions observed in both groups of patients according to their CT scan pattern is summarized in Table 4. All patients showing a predominant nodular pattern were found to present florid granulomas in their lung tissue, whereas cavitary granulomas and fibrous cysts were infrequent (two of seven patients in both cases). Florid granulomas were numerous and well developed, either isolated or more often confluent, ranging from 0.7 to 11.7 mm in their largest dimension (mean, 2.6 ± 1.9 mm). Interestingly, the extent of nodular lesions observed on CT scan was strongly correlated with the density of granulomatous lesions present in corresponding lung tissue samples (Figure 4), reinforcing the idea that nodules observed on HRCT corresponded to areas of inflammatory granulomas.

View larger version (22K): [in this window] [in a new window]   Figure 4.   Correlations between histopathological and HRCT findings. The CT score calculated for nodular abnormalities observed by HRCT is strongly correlated with the density of granulomatous lesions present in corresponding lung tissue from patients with a nodular pattern (A) and from all patients (B).

All patients showing a predominant thin-wall cyst pattern were found to present cavitary lesions in lung tissue, which were not only fibrous cysts, but also cavitary granulomas. It should be stressed that only two of six patients presented cavitary lesions that were fibrous cysts alone, whereas the other patients presented cavitary granulomas associated with cysts (three of six patients) or, even, cavitary granulomas alone (one of six patients). Interestingly, the density of cavitary lesions, including both cavitary granulomas and fibrous cysts, correlated well with the CT score (Figure 5), strongly suggesting that thin-wall cysts observed on HRCT could correspond indifferently to cavitary granulomas or cystic lesions. Most important, although micronodules were virtually absent on CT scans of patients with a cystic pattern, small isolated florid granulomas (< 1 mm in their largest dimension; mean, 0.7 ± 0.2 mm) were observed in lung tissue of four of six patients, and, on the whole, inflammatory lesions (florid and/or cavitary granulomas) were present in almost all patients (five of six patients).

View larger version (22K): [in this window] [in a new window]   Figure 5.   Correlations between histopathological and HRCT findings. The CT score for thin-walled cystic abnormalities is strongly correlated with the density of cavitary lesions, including both cavitary granulomas and fibrous cysts. (A) Patients with a cystic pattern; (B) all patients.

Taken together, these data demonstrate that in the course of PLCH, a nodular CT scan pattern reflects a histopathologically active disease, and that a cystic pattern, although usually presumed to reflect an end-stage disease, is frequently associated with the presence of still inflammatory lesions.

Correlations between Histopathological and HRCT Findings and Clinical Presentation

Although in larger series most patients with PLCH have been reported to resolve spontaneously (13), in the present study only 3 of 13 patients recovered or improved. As shown in Table 2, they were all patients with a nodular CT scan pattern, the other patients of this group presenting no evolution (two of seven patients), or deterioration of their status with appearance of cystic lesions accompanying a nodular regression (two of seven patients). In our study, all patients with a cystic CT scan pattern had a severe disease, and two of them required lung transplantation. The follow-up characteristics of the patients of this group showed that one of them remained stable, whereas two patients progressively developed a respiratory insufficiency of increasing severity over a 5-yr period. This respiratory insufficiency was associated with a severe deterioration of lung parenchyma, one patient requiring lung transplantation. The presence of still inflammatory lesions in the lung of most of these patients (five of six patients) may help explain the poor evolution of their disease. No correlation was observed, however, between the presence of cystic lesions and/or cavitary granulomas and the duration of the disease.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of the present study, comparing both HRCT and histopathological findings in the course of PLCH, demonstrate that HRCT is relevant only with a relative accuracy for the assessment of the granulomatous reaction in the course of the disease. A nodular pattern clearly reflects a histopathologically active process, but caution is necessary with patients presenting predominant cystic abnormalities, because still cellular lesions are frequently present in their lung tissue.

The histopathological presence of LC granulomas in lung tissue is the reference marker of disease activity (1). However, since pulmonary biopsy is not always required in clinical practice and, most importantly, cannot be repeated, noninvasive tools are clearly needed to improve the follow-up of the disease. This is necessary to better estimate the natural history and potential evolution of the disease in a given patient, and to evaluate, if necessary, the efficacy of drugs. Gallium 67 scintigraphy does not show significant uptake in PLCH, and, in addition, smoking habits, which are generally important in adult patients with PLCH, are known to increase gallium uptake in the lung (14). Some years ago, the identification of CD1a⁺ LC in bronchoalveolar lavage (BAL) has been considered a promising diagnostic tool (15). It should be stressed, however, that more than two of three patients with PLCH have less than 5% LC in their lavage fluid, a level similar to that of normal smoking individuals or patients with other diffuse lung disorders (4). Thus, because of its low sensitivity, BAL is not suitable for the assessment of disease evolutivity.

HR-CT is a noninvasive procedure that is very useful in the etiological diagnosis of interstitial lung disease (9, 10), particularly PLCH, in the course of which two of three patients present both nodular and cystic abnormalities (5, 8). In several cases, some lesions have been shown to develop a cavitation, and subsequently to evolve from cavitation to cysts (5). In a study of 21 patients including a CT scan follow-up (8), it has been shown that CT scan was initially characterized by a predominant nodular pattern followed by a cystic pattern, less frequently than by a relative normalization. The hypothesis that HRCT could follow the evolution of histopathological abnormalities was raised, but the correspondence between the latter and CT findings has never been investigated.

Our results confirm the idea that nodular lesions visible on CT scans correspond to granulomatous lesions in lung tissue, and are suggestive of an active process. The CT score strongly correlated with the granuloma density in patients with a nodular pattern. In these patients, the granulomas always corresponded to florid inflammatory lesions, demonstrating that the presence of predominant nodules on HRCT reflects a histopathologically active disease. Our results also demonstrate that thin-wall cysts on HRCT frequently correspond to cavitary but still inflammatory granulomas, suggesting that in patients presenting with a cystic scan pattern, the persistence of an active, albeit limited granulomatous process cannot be excluded. It should be noted, however, that in these patients, the relative importance of granulomatous lesions could not be predicted by HRCT, as thin-wall cysts observed on scan could indifferently correspond to fibrotic cysts or cavitary granulomas on lung histology. In addition, florid granulomas were found in lung tissue from most patients with a cystic pattern (four of six patients). The small numbers of these granulomas, as well as their small size (< 1 mm), explain why they were not visible on HRCT. Thus, although in patients with a nodular pattern the histopathological activity of PLCH can be easily predicted by HRCT, this is not the case in patients with a cystic pattern because of the possible coexistence of both fibrotic and still cellular lesions in their lung tissue.

It has been previously reported that a nodular pattern is observed early in the course of the disease, whereas a predominant thin-wall cyst pattern is observed later during the evolution of the process, the majority of patients presenting with both nodular and cystic lesions (8). In the present study, patients presented a main CT scan pattern clearly nodular or cystic, and no patient had simultaneously significant numbers of both nodular and cystic lesions. This is probably due, at least in part, to the fact that biopsies are now performed in patients in whom diagnosis is difficult to establish, or at the time of pulmonary transplantation in patients with severe disease. Another possible reason is suggested by CT follow-up studies of patients with PLCH (8, 16). These studies demonstrated that the presence of nodular lesions is often transient and the relapse of nodules is very rare. Thus, the probability of seeing the suggestive combination of both nodules and cysts in a given patient seems relatively weak.

Finally, in our study, most patients with PLCH had inflammatory granulomas in their lung tissue, even at a very advanced stage of the disease (five of six patients with a cystic CT scan pattern). As previously discussed, patients in this study with a cystic pattern presented a particularly poor evolution of their disease. The persistence of cellular lesions in these patients can help explain their progressive respiratory insufficiency.

In conclusion, the present study demonstrates that HRCT, which has been previously shown to be of considerable value in the diagnosis and follow-up of pulmonary LCG, can be used as a noninvasive test to grossly evaluate the histopathological activity of the disease. A nodular pattern clearly reflects a histopathologically active disease, but caution must be used in evaluating patients with a cystic pattern, since HRCT does not permit us to differentiate between fibrous cysts and cavitary granulomas. Thus, although it is usually considered to correspond to an end-stage disease, a cystic pattern does not exclude a still active pathological process. This is of clinical relevance, and patients presenting with a cystic CT scan pattern should benefit from long-term follow-up. Furthermore, because these patients are susceptible to progression to severe respiratory insufficiency, they should also be considered for treatment as soon as an effective therapy for LCH is available.