Published ahead of print on December 3, 2004, doi:10.1164/rccm.200403-417OC
© 2005 American Thoracic Society doi: 10.1164/rccm.200403-417OC
Pulmonary Manifestations of Primary Sjögren's SyndromeA Clinical, Radiologic, and Pathologic StudyDepartment of Respiratory Medicine and Laboratory of Anatomic Pathology, Kyoto University Hospital; Central Clinic in Kyoto, Kyoto; Department of Medical Physics and Radiology, Osaka University Graduate School of Medicine, Osaka; Department of Radiology, Tenri Hospital, Nara, Japan; Department of Histopathology, Royal Brompton Hospital, London, United Kingdom; and Division of Pulmonary and Critical Care Medicine, Asan Medical Center, University of Ulsan, College of Medicine, Seoul, Korea Correspondence and requests for reprints should be addressed to Sonoko Nagai, M.D., Ph.D., Department of Respiratory Medicine, Kyoto University Hospital, 54 Shogoin-kawaharacho, Sakyo, Kyoto, Japan. E-mail: nagai{at}kuhp.kyoto-u.ac.jp
Rationale: Clinicopathologic pulmonary manifestations associated with primary Sjögren's syndrome have yet to be reviewed in a large series since the recognition of nonspecific interstitial pneumonia (NSIP) as a distinct histologic pattern. Objectives: To determine clinical presentations, high-resolution computed tomographic (HRCT) and histologic findings of the lung disease associated with primary Sjögren's syndrome in the light of NSIP, and to analyze prognosis of the disease. Methods: On the basis of 33 cases (31 surgical lung biopsies and 2 autopsies) collected consecutively from multiple centers, we have retrospectively evaluated clinical, radiologic, and pathologic manifestations of the disease. Prognostic factors were identified by univariate and multivariate analysis. Measurements and main results: We found that NSIP was the most frequently seen histologic pattern (20 of 33 cases [61%], 19 fibrosing and 1 cellular). Bronchiolar diseases and amyloid and malignant lymphoma were seen less frequently. HRCT-pathologic correlation resulted in a 94% positive predictive value of CT-NSIP pattern for pathologic diagnosis of NSIP, whereas the diagnostic value of HRCT was low (15%) with an HRCT pattern other than NSIP, data that may influence the decision to biopsy. The 5-year survival rate was 84% overall and 83% in patients with NSIP. Multivariate analysis on all patients showed that low PaO2 (p = 0.02) and presence of microscopic honeycombing (p = 0.04) were independently associated with survival. Patients with NSIP showed lower vital capacity (mean ± SD: 68.5 ± 16.6%pred) than patients without NSIP (92.5 ± 18.6%pred; p < 0.001). Conclusion: Among a diversity of pulmonary lesions in primary Sjögren's syndrome, NSIP was the commonest histologic pattern and had a favorable prognosis.
Key Words: bronchiolitis • honeycombing • nonspecific interstitial pneumonia • surgical lung biopsy • survival Sjögren's syndrome (SjS) is a chronic inflammatory autoimmune disease that is characterized typically by lymphocytic infiltration of the exocrine glands resulting in the sicca syndrome (1). SjS can occur alone as primary SjS (pSjS) or in association with other autoimmune diseases, such as secondary SjS (2). Various extraglandular organ systems, such as the lung, kidney, and small vasculature, are sometimes affected in pSjS (3). A broad range of computed tomographic (CT) (4–6) and pathologic (4, 7–9) findings have been reported as pulmonary manifestations of pSjS, but there are limited data on clinical manifestation and prognosis on the basis of pathologic diagnosis and radiologic findings in patients with clinically overt symptoms (4, 7–9). Furthermore, nonspecific interstitial pneumonia (NSIP) in patients, associated with or without collagen-vascular diseases, has been recognized as a distinct histologic pattern of interstitial pneumonia in the last decade (10–20) and has been shown to have different clinicopathologic features from idiopathic pulmonary fibrosis/usual interstitial pneumonia (UIP) (11, 13–18, 20). We have therefore reviewed retrospectively a series of patients with pSjS and associated pulmonary disease to determine the following: (1) the clinical features at initial examination, (2) high-resolution CT (HRCT) and histologic findings in the light of a histologic pattern of NSIP, and (3) prognostic analysis. Some of these results have been reported in the form of abstract (21).
Study Subjects and Pathologic Analysis Forty-two patients, who had presented with lung disease possibly associated with SjS and who had undergone pathologic investigation at 14 hospitals in Japan and one hospital in Korea between January 1992 and May 2002, were consecutively identified from records of each hospital. At the eighth meeting of the Kyoto Respiratory Disease Symposium held in August 2002, these patients' records were retrospectively reviewed. A total of 33 cases that fulfilled the revised diagnostic criteria for pSjS published by the American-European Consensus Group were enrolled in the study (22). Patients were followed up through April 2003 or until death before April 2003. After an initial review by 10 pathologists who were experienced in lung pathology (see ACKNOWLEDGMENT), two of the pathologists (M.K. and A.G.N.) further reviewed all pathologic specimens from 33 cases (31 surgical lung biopsies [SLB] and 2 autopsies) using the histologic patterns described in the International Consensus Classification of the Idiopathic Interstitial Pneumonias (19). NSIP was subdivided into cellular NSIP and fibrosing NSIP. Other categories used were as follows: chronic bronchiolitis, amyloidosis, lymphoma, and atelectatic fibrosis; the last term indicates that fibrosis was accompanied by acinar collapse. No radiographic findings and clinical data were made available to the pathologists until the final diagnoses were made. Final diagnoses were reached by the consensus of M.K. and A.G.N. in a simultaneous joint review, with consideration of comments from the other eight pathologists. A major histologic pattern was assigned for each patient, which was the dominant pattern believed to be the most important for diagnosis. Minor findings were defined as other histologic patterns seen in addition to the major pattern.
Clinical and Radiologic Analysis For the radiologic analysis, we used chest radiographs and HRCT taken at the initial presentation and during the follow-up period. Each film was reviewed by two experienced chest radiologists (T.J. and S.N.) who were blinded to pathologic findings, and final diagnoses were obtained by consensus. Radiologists reviewed the findings of HRCT examinations taken before therapeutic interventions. According to the classification of CT patterns (19), they categorized the findings into CT-UIP, CT-NSIP, CT–organizing pneumonia, or CT–lymphoid interstitial pneumonia (CT-LIP) patterns. In addition, CT-bronchiolitis and CT-cysts were described.
Statistical Analysis Details on the pathologic, clinical/radiologic, and statistical analyses are provided in the online supplement.
Selection of Patients and Their Pathologic Findings Of 42 patients, 33 (27 women and 6 men) were diagnosed as having pSjS and selected for the study. The age of the patients at the time of the respiratory presentation was 55.9 ± 11.0 years (mean ± SD). Among the 33 patients, three patients had no respiratory symptoms when their respiratory disease was identified: two were found by incidental chest radiograph at health checks, and one at chest screening for pSjS. Nine patients were not diagnosed as having pSjS and were excluded from the study. The average number of specimens was 2.1 ± 0.6 (mean ± SD) in 31 biopsied cases. In two autopsied cases, specimens from all lobes were evaluated. All pathologists reviewed all slices available. All the specimens were of sufficient size for pathologic evaluation, except one, which was interpreted as honeycomb lung and no pattern of interstitial pneumonia was assigned.
Table 1 shows the major histologic patterns and the number of cases in each category. The most frequent major finding was a pattern of NSIP, which was diagnosed in 20 patients (61%). One patient showed cellular NSIP and 19 showed fibrosing NSIP (further pathologic data are provided in Figures E1 and E2 in the online supplement). Four patients showed diffuse bronchiolitis as the major pattern. Two of these cases showed subepithelial cellular infiltration of lymphoid cells around the entire circumference of the bronchiole, causing luminal stenosis (see Figure E3), and one case showed a follicular bronchiolitis. The fourth case showed a major pattern of chronic bronchiolitis with additional minor involvement of the adjacent alveolar walls, akin to cellular NSIP. Four patients showed non-Hodgkin lymphomas of mucosa-associated lymphoid tissue (MALT) origin. Two patients showed deposition of amyloid in the lung parenchyma as the major pattern. Two cases with interstitial fibrosis did not fit into either of the previously described pattern and were categorized as having atelectatic fibrosis (see Figure E4). There was good interobserver agreement in major diagnosis between the two pathologists (M.K. and A.G.N.) (
A variety of minor histologic patterns were seen in addition to the major histologic pattern in many cases. Four cases showed bronchiolitis, two in association with NSIP, making eight cases overall with bronchiolitis as a major or minor finding. Among four cases with primary pulmonary lymphoma, amyloid was observed in three cases as a minor finding. Thus, a total of five cases of major and minor findings showed amyloid. (Details of pathologic findings in each case are shown in Table E1.) In terms of individual histologic parameters, honeycombing with relatively homogenous cysts was observed in 13 cases as a minor finding, 12 of which were classified as fibrosing NSIP. Therefore, microscopic honeycombing was observed in 60% of cases with a major diagnosis of NSIP. Scanty fibroblastic foci were found in only six cases, in insufficient numbers for a diagnosis of UIP with preferential classification as fibrosing NSIP. Lymphoid hyperplasia, organizing pneumonia, and pleuritis were also noted as minor features in two cases each of NSIP. Four of the 31 patients who underwent biopsy (three patients diagnosed as having fibrosing NSIP and one as having cellular bronchiolitis) had received corticosteroid therapy with some remission before pathologic investigation. However, further deterioration of lung diseases after tapering of steroid therapy prompted SLB. Both patients who underwent autopsy (fibrosing NSIP and atelectatic fibrosis) also had been treated with steroid therapy. Further details of this section are provided in the online supplement.
Clinical Findings
The frequency of symptoms experienced by patients during the follow-up period that were possibly related to the disease was as follows: sicca symptoms in 31 patients (94%), cough in 24 (73%), dyspnea in 21 (64%), arthralgia in eight (24%), fever in seven (21%), sputa in six (18%), and Raynaud's symptom in six patients (18%). None of the pathologic categories were associated with specific symptoms. Pulmonary function tests were performed on a mean of 23 days before lung biopsy (range 2–99 days). Nineteen (58%) of the 33 patients showed restrictive changes of VC%predicted of less than 80%, whereas three patients (9%) had obstructive changes of FEV1/FVC of less than 70%. Bronchoalveolar lavage was performed in 28 patients. Cell differentials of bronchoalveolar lavage fluid were abnormal in all patients. Lymphocyte counts were elevated in 18 (64%), neutrophils in 19 (68%), and eosinophils in 21 (75%) patients. We then assessed the 33 patients in relation to groups on the basis of histologic findings and compared the mean or median of clinical indices: (1) 20 patients with a major diagnosis of NSIP versus 13 patients with other diagnoses, (2) four patients with a major diagnosis of bronchiolitis versus 29 with other diagnoses, (3) four patients with a major diagnosis of malignant lymphoma versus 29 others, (4) eight patients with major or minor bronchiolar findings versus 25 with other patterns, and (5) 14 patients with microscopic honeycombing versus 19 with other patterns. VC%predicted was lower in patients with NSIP (mean ± SD: 68.5 ± 16.6%) than in patients with other patterns (92.5 ± 18.6%; p < 0.001). FEV1/FVC (%) was lower in patients with major or minor bronchiolar finding (median 74.8%, range 57.9–87.8) than in patients without it (median 83.1%, range 71.7–91.0; p < 0.05). Further details of correlation between clinical and pathologic findings are shown in Table E2.
Radiologic Findings HRCTs were obtained for 31 patients. HRCTs taken on the closest day before pathologic investigation (median 27 days, range 2–127 days) were analyzed. A summary of HRCT findings is shown in Table 3. Fourteen patients (45%) were assessed as having CT-NSIP pattern, and four (13%) as having CT-UIP pattern. Three (10%) patients showed CT–organizing pneumonia pattern as the initial CT finding and responded well to corticosteroid therapy. However, those three patients all experienced relapses of lung diseases with development of a CT-NSIP pattern at the time of SLB. Therefore, 17 patients (55%) had a CT-NSIP pattern at the time of biopsy. Radiologic honeycombing was seen in four cases, and all of them were diagnosed as CT-UIP pattern, two of which were NSIP pathologically. Another patient (3%) showed CT-LIP pattern and three patients (10%) showed CT-cysts. Nine cases (29%) showed CT-bronchiolitis combined with (n = 5) or without (n = 4) other patterns. The presence of CT-bronchiolitis did not affect residual volume/total lung capacity (RV/TLC) in pulmonary function tests (p = 0.48). In two cases, it was difficult to categorize their CT findings into previously defined patterns. In two cases, multiple nodules were seen bilaterally, in combination with CT-LIP pattern in one case and with CT-cysts in another. HRCT findings of each case are shown in Table E1.
The HRCT-pathologic correlation is summarized in Table 3. In the four cases with CT-UIP pattern, a diagnosis of fibrosing NSIP with honeycombing was given pathologically in two cases, atelectatic fibrosis with bronchiolectasis in one and only honeycomb change because of insufficient material in the other case. The positive predictive value of CT-NSIP pattern against pathologic major diagnosis of NSIP was 94% (16 of 17 cases), and the negative predictive value of CT-NSIP pattern was 77% (10 of 13 cases). However, the major pathologic diagnosis was predictable only in 2 (15%) of the 13 cases with CT diagnosis other than CT-NSIP pattern.
Treatment and Survival Analysis The Kaplan-Meier survival curve for all patients is shown in Figure 1. The 5-year survival rate for all patients with pSjS was 84%. The hazard ratio, 95% confidence interval, and p value for each variable in the univariate analysis using Cox's proportional hazard model for the risk of death in all 33 patients are shown in Table 4. In all 33 patients with pSjS, a univariate analysis of 21 factors revealed that age at the time of respiratory presentation, RV/TLC, and PaO2 under room air condition influenced mortality (p = 0.04, p = 0.003, and p = 0.01, respectively; Table 4). Age, sex, smoking history, VC%predicted, RV/TLC, PaO2, and presence of microscopic honeycombing were included in the multivariate analysis. By multivariate analysis, PaO2 (p = 0.02, hazard ratio 1.266, 95% confidence interval 1.036–1.546) and absence of microscopic honeycombing (p = 0.03, hazard ratio 0.031, 95% confidence interval 0.001–0.779) were found to be independently associated with survival. The Kaplan-Meier survival curve of the subgroup of patients with pathologic NSIP is shown in Figure 2. The 5-year survival rate of patients with NSIP was 83%.
In this retrospective review of the clinical, radiologic, and pathologic manifestations of lung diseases associated with pSjS, we found the following: (1) pathologically, NSIP was present in the majority of the cases (61%) and no case with LIP was included in our series; (2) a diversity of other pathologies, such as bronchiolitis, malignant lymphoma, amyloid, and atelectatic fibrosis, were seen both as major and minor patterns; (3) CT-NSIP pattern had a high positive predictive value for pathologic NSIP pattern (94%), but HRCT patterns other than CT-NSIP rarely helped identify the correct diagnoses (15%); (4) the 5-year survival rate for all patients was 84% and that for patients with NSIP was 83%; and (5) the survival of all patients was related to PaO2 and absence of microscopic honeycombing. In terms of pathology, it is notable that no cases were diagnosed as LIP pattern, although pSjS has previously been associated with this pattern (19, 24). There are several possible reasons for this finding. First, many have likely been reclassified as NSIP when criteria from the American Thoracic Society/European Respiratory Society classification are applied (19). Previous classifications would only have had options of UIP or LIP and these biopsies would have been closer to the latter. However, given that 19 of 20 cases were fibrotic in nature and the current American Thoracic Society/European Respiratory Society classification concentrates more on the intensity of the interstitial inflammatory cell infiltrate, reclassification as NSIP is appropriate. We did consider whether corticosteroid therapy dampened the intensity of inflammation but consider this unlikely because all cases were biopsied in deteriorating phases of disease after stopping therapy for more than 1 year and initial HRCTs had not shown a CT-LIP pattern. We also considered whether this cohort comprised patients presenting late in the course of their disease and therefore with a greater extent of fibrosis, leading to classification as fibrosing NSIP, but this is not possible to substantiate from our data. Finally, some cases that are now classified as lymphoma may also have been previously termed as LIP, given that LIP was once viewed as a preneoplastic condition in relation to primary pulmonary lymphomas. Furthermore, one case that showed CT-LIP pattern was malignant lymphoma on pathology. Whether this particular case represents malignant transformation of LIP is uncertain, but this finding is now believed to be exceptionally rare based on histologic and molecular data, and many cases of LIP believed to develop into lymphoma were, in fact, malignant from the outset (19, 24, 25). It is possible that previously reported cases with "radiologic LIP pattern" could have also included lymphomas (19, 26). We therefore conclude that LIP, as currently defined, is rare even among patients with pSjS and that NSIP is the most common histologic pattern of interstitial pneumonia in pSjS. This is not so surprising because it is the most prevalent pattern of interstitial pneumonia seen in patients with systemic sclerosis (68–78%) (27–29) and polymyositis/dermatomyositis (36–82%) (30–32); in addition, it is also increasingly being described in association with rheumatoid arthritis and systemic lupus erythematosus (10, 33, 34). Yamadori and colleagues (9) also reported NSIP in association with pSjS, although the total number of cases was small. In addition to NSIP as the major category, follicular/cellular bronchiolitis and amyloidosis were observed in several cases, both as major and minor features. Some of these phenomena have also been described in a report that investigated 40 lung diseases associated with rheumatoid arthritis (35) but are comparatively rare in other collagen-vascular diseases. Our data show that they are also not infrequently seen in patients with pSjS and clinical pulmonary involvement. Furthermore, the presence of combined pathologic findings is believed to suggest the presence of background collagen-vascular disease (33, 36), also supported by these data. Of these findings, clinical studies suggest that the small airways involvement may play a particularly important role in pulmonary presentation of patients with pSjS (4–6, 33–37). One of the most important issues in managing lung diseases associated with pSjS is whether to obtain an SLB for pathologic diagnosis. In the current series, CT-NSIP pattern in HRCT had a high positive predictive value for the presence of NSIP in pathology. However, cases without CT-NSIP pattern correlated poorly with the histologic diagnosis. Accordingly, performance of SLB should be considered in particular for cases without CT-NSIP pattern, when pathologic diagnosis is required. This is in contrast to idiopathic disease, in that a CT-NSIP pattern is not an accurate predictor of histologic NSIP in this setting and cases are more frequently biopsied (20). However, further studies on larger cohorts may be required to assess how minor CT and histologic features (e.g., bronchiolitis pattern) impact on mortality and morbidity in patients with a major pattern of NSIP. Cases other than CT-NSIP pattern included those with CT patterns suggestive of LIP or cysts. In this series, five cases with CT-cysts, CT-LIP pattern, or CT-bronchiolitis with and without solid nodules were diagnosed as malignant lymphoma and/or amyloid. The cases have certain similarities to those reported as a combination of amyloid and multiple cysts (38, 39) or LIP (40) in pSjS, and are a group in which biopsy may be warranted to diagnose or exclude lymphoma. Assessment of bronchoalveolar lavage fluid and/or bronchoscopic biopsies can lead to diagnosis of lymphoma using immunohistochemistry and molecular analyses (41), but minor B-cell clones caused by chronic autoantigenic stimulation may impact on specificity of molecular techniques (41). Of note, the mechanism for cyst formation in such cases remains unknown, although a check-valve mechanism via peribronchiolar inflammation and/or amyloid deposition has been proposed (42). For prognosis, our data show that mortality is associated with decreased baseline PaO2 and presence of microscopic honeycombing on multivariate analysis. In studies of idiopathic interstitial pneumonias, several studies report associations with mortality and baseline physiologic tests (18, 43, 44), pathologic findings (13, 14, 16–18), and more recently, serial pulmonary function tests (45, 47). Some of these studies also show an association between mortality and baseline PaO2 (43, 45, 46). In terms of histologic parameters, there are few studies on correlation with mortality but those in patients with idiopathic pulmonary fibrosis/UIP show an association with the extent of fibroblastic foci (48, 49) rather than the extent of fibrosis/honeycomb change. Furthermore, fibroblastic foci were sufficiently rare in this cohort that no cases were classified as UIP. This apparent association between prognosis and extent of fibrosis, not seen in idiopathic pulmonary fibrosis/UIP or between cellular and fibrosing cases of NSIP in systemic sclerosis (27) may reflect a different biology in the progression to fibrotic disease in association with pSjS. We also found that a high value of RV/TLC was associated with mortality on univariate analysis, but this association disappeared on multivariate analysis. In idiopathic pulmonary fibrosis, low VC%predicted has been related to poor prognosis, but RV/TLC does not seem to be much affected because both RV and TLC would be reduced as the VC decreases (43). In restrictive conditions, the RV, VC, and TLC are all reduced, but the RV/TLC may be elevated, normal, or reduced depending on the possibility of concomitant obstruction, especially in cigarette smokers. However, the RV/TLC was not related to the smoking history in our series (data not shown). Other factors, such as higher age and poorer general condition, which cause difficulty in examination of RV/TLC, may also have affected the multivariate analysis. We recognize that this study is a retrospective review and criteria for performing SLB differed in each hospital. As such, we cannot exclude that some patients with typical HRCT findings of UIP pattern may not have underwent biopsy, leading to an overestimation of the frequency of NSIP, as mentioned in previous studies on systemic sclerosis (28). As such, our data may not necessarily reflect the general population. Also, the number of the cases is insufficient to compare with confidence clinical courses according to the various patterns of disease. Finally, therapeutic intervention also differed between institutions, which may have complicated the evaluation of outcome. Nevertheless, we believe that the presented data remain valid despite these shortcomings.
The authors thank those investigators who supported the symposium in clinical work, radiologic review, pathologic review, and clinical evaluation and also thank all the members of the symposium. Clinical work: Takashi Mouri (Iwate), Hiroshi Mukae (Nagasaki), Hideo Kobayashi (Saitama), Kazuyoshi Kurashima (Saitama), Yasuo Matsuzawa (Chiba), Takashi Ogura (Cardiovascular and Respiratory Center, Kanagawa), Futoshi Kotajima (Tokyo), Hidefumi Shimizu (Rosai Hospital, Tokyo), Yoshihiro Nambu (Kanazawa), Kingo Chida (Hamamatsu), Hiroyuki Taniguchi (Tosei General Hospital, Aichi), Tatsuya Kanamori (Okayama), Yoshio Taguchi (Tenri), Riyo Fujiyama (Kobe), Jiro Fujita (Kagawa), Masakazu Aitani (NTT West Osaka Hospital, Osaka), and Kunio Hamada (Kyoto). Radiologic evaluation: Masashi Takahashi (Shiga). Pathologic evaluation: Yoshinori Kawabata (Saitama), Tamiko Takemura (Tokyo), Yuh Fukuda (Tokyo), Yoichiro Kobashi (Tenri), Yoshitetsu Yuba (Tenri), Osamu Matsubara (Saitama), Ichiro Yamadori (Okayama), and Angela Chong Pek Yoon (Singapore). Clinical evaluation: Yoshikazu Inoue (Osaka) and Shigeki Makino (Osaka). This manuscript was reviewed by an experienced scientific editor whose first language is English.
Supported by the Otsuka Pharmaceutical Co., Ltd. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Conflict of Interest Statement: I.I. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.N. received $18,000 in 2002 from Otsuka Pharmaceutical Co., Ltd., in research grants for holding the Eighth Kyoto Respiratory Disease Symposium; M.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; A.G.N. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; T.J. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.N. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; D.S.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; T.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; T.I. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; M.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form March 26, 2004; accepted in final form November 24, 2004
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
= 0.63). 
