ZOE D. DANIIL, FRANCES C. GILCHRIST, ANDREW G. NICHOLSON, DAVID M. HANSELL, JESSICA HARRIS, THOMAS V. COLBY, and ROLAND M. du BOIS

Transcription

1 A Histologic Pattern of Nonspecific Interstitial Pneumonia Is Associated with a Better Prognosis Than Usual Interstitial Pneumonia in Patients with Cryptogenic Fibrosing Alveolitis ZOE D. DANIIL, FRANCES C. GILCHRIST, ANDREW G. NICHOLSON, DAVID M. HANSELL, JESSICA HARRIS, THOMAS V. COLBY, and ROLAND M. du BOIS Interstitial Lung Disease Unit and Departments of Histopathology and Radiology, Royal Brompton Hospital, London, United Kingdom; and Department of Pathology, Mayo Clinic, Scottsdale, Arizona This study aimed to investigate whether there was a difference in outcome related to histologic pattern in cryptogenic fibrosing alveolitis (CFA) and to see whether there were correlations between clinical and radiologic findings and histology. One hundred thirteen lung biopsies from consecutive patients taken for the diagnosis of diffuse lung disease were reviewed and reclassified using the Katzenstein and Myers criteria for interstitial pneumonias. Patients lacking full investigational data at presentation and those with conditions predisposing to lung fibrosis were excluded, leaving 15 patients diagnosed with nonspecific interstitial pneumonia (NSIP) and 15 with usual interstitial pneumonia (UIP). Clinical and radiologic findings at presentation and serial lung function information and survival status in November 1998 were compared for the two groups. Survival was found to be significantly greater in the NSIP group compared with the UIP group (p 0.001). This could not be explained by differences in treatment. Patients with UIP showed a progressive deterioration in lung function whereas those with NSIP remained stable. CT scans of patients with UIP showed more fibrosis than those of patients with NSIP (p 0.011). A histologic diagnosis of NSIP is associated with a better prognosis than UIP. This subclassification of CFA is clinically useful. Daniil ZD, Gilchrist FC, Nicholson AG, Hansell DM, Harris J, Colby TV, du Bois RM. A histologic pattern of nonspecific interstitial pneumonia is associated with a better prognosis than usual interstitial pneumonia in patients with cryptogenic fibrosing alveolitis. AM J RESPIR CRIT CARE MED 1999;160: (Received in original form March 1, 1999 and in revised form April 12, 1999) Correspondence and requests for reprints should be addressed to Dr. R. M. du Bois, Royal Brompton Hospital, Sydney Street, London SW3 6NP, UK. r.dubois@rbh.nthames.nhs.uk Am J Respir Crit Care Med Vol 160. pp , 1999 Internet address: Cryptogenic fibrosing alveolitis (CFA) is a chronically progressive interstitial lung disease characterized by poor prognosis and a limited response to treatment. The survival at 5 yr is approximately 50% and an objective response to corticosteroids is observed in only 20% of patients (1 5). Most patients diagnosed as having CFA, also termed idiopathic pulmonary fibrosis (IPF), show the features which fulfill the histopathologic criteria for usual interstitial pneumonia (UIP), the most common histologic type of idiopathic interstitial pneumonia (6, 7). Chronic interstitial pneumonias were originally classified by Liebow and Carrington into five groups: UIP, desquamative interstitial pneumonia (DIP), bronchiolitis obliterans with interstitial pneumonia (BIP), giant cell interstitial pneumonia (GIP), and lymphoid interstitial pneumonia (LIP) (8). Cases of BIP usually fall into the category now termed bronchiolitis obliterans organizing pneumonia (BOOP), and GIP is regarded as a pneumoconiosis associated with hard metal exposure. Furthermore, acute interstitial pneumonia (AIP) (9), nonspecific interstitial pneumonia (NSIP) (10), and respiratory bronchiolitis interstitial lung disease (RBILD) (11) have all been recently recognized. Their current interrelationships have been recently reviewed and a revised classification has been proposed by Katzenstein and Myers (6). The term NSIP has been proposed for cases that do not fit the histopathologic criteria for the other categories described above (10). NSIP is characterized by an interstitial inflammatory cell infiltrate with or without fibrosis and lacking the specific histologic features that characterize other forms of interstitial pneumonia. NSIP is most likely to be confused with UIP, but can be distinguished in most cases as NSIP does not show the temporal heterogeneity that is the cardinal feature of UIP. In NSIP the inflammation and fibrosis appear to be of a similar age, whether recent and active, or old and relatively quiescent. In UIP both processes are present but appear to be of different ages with a mixture of active inflammation and fibroblastic proliferation believed to reflect ongoing injury, and established fibrosis thought to reflect older injury and repair. Because of the wide range of survival in CFA we hypothesized that there may be differences in pathologic pattern between long-term survivors and those who succumb early. Lung biopsies taken for investigation of suspected CFA were reclassified using the Katzenstein and Myers criteria for idiopathic interstitial pneumonias (6). The results were correlated with

2 900 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL clinical, radiologic, and survival data with two aims: first, to confirm the findings of Katzenstein and Fiorelli (10), Bjoraker and coworkers (12), and Nagai and coworkers (13) that survival of patients diagnosed as having CFA or IPF is determined by the precise histopathologic subtype; and second, to relate the pathology to radiologic, in particular high-resolution computed tomography (HRCT), and clinical findings to determine whether there are distinguishing aspects of these histologic variants. METHODS Subject Selection Lung biopsies from the 113 consecutive patients who presented to the Royal Brompton Hospital between March 1990 and December 1995 for diagnosis of diffuse lung disease were reviewed. Open or thoracoscopic lung biopsies from all patients were independently reviewed by two pathologists (A.G.N. and T.V.C.) who were unaware of any clinical or physiologic findings. Histologic classification was based on previously published criteria for idiopathic interstitial pneumonias (6, 10). Cases in which there were different independent diagnoses were reevaluated by the two pathologists and a consensus diagnosis was agreed. Those with fibrosing alveolitis associated with connective tissue disease or environmental exposures known to cause fibrosing lung disease were excluded together with those categorized as having histopathologic diagnoses other than UIP or NSIP. This left 30 patients, 15 with the histologic features of NSIP and 15 with those of UIP. Clinical Information and Bronchoalveolar Lavage (BAL) Clinical data were obtained from patient medical records. Sex, age at presentation, onset of breathlessness, and smoking history were recorded. Subjects were classified as smokers if they had smoked at least one cigarette a day for more than 1 yr. Clinical histories were reviewed and those with any underlying medical condition or potential causes of pulmonary abnormalities (e.g., connective tissue disease, exposure to organic or inorganic dust or toxic fumes, and history of specific drug intake) were excluded. One patient in the UIP group had minor exposure to asbestos and hard metal and one patient in the NSIP group had been working as a thatcher (possible exposure to fungal spores) but fungal precipitins were negative. These two patients were included in the study group. Antinuclear antibodies (ANA), rheumatoid factor, and 99m Tc-DTPA clearance at presentation were recorded if available. ANA were measured by immunofluorescence. Samples were categorized subjectively into those negative for ANA and those weakly positive, moderately positive, and strongly positive. For the purposes of this study any degree of positivity was defined as positive. Rheumatoid factor titers were measured using the rapid particle agglutination test. A titer of greater than 1 in 160 was defined as a positive result. BAL had been performed for diagnostic or prognostic purposes in eight subjects in each group. HRCT Scans HRCT scans taken within 4 mo of biopsy were scored semiquantitatively by two observers. A global assessment was made using a fourpoint scale (0 no evidence of the particular pattern, through 3 pattern present throughout the abnormal lung) for the extent of the following HRCT patterns: honeycombing, interstitial thickening (fine reticular pattern), and traction bronchiectasis. A total fibrosis score (range, 0 to 9) was derived for each HRCT by adding the scores for each pattern. In addition, the HRCT scans were scored for the degree of lower zone predominance (0 no identifiable zonal distribution, through 3 lower zone predominant disease) and the subpleural predilection of the abnormalities (0 no obvious peripheral distribution, through 3 predominantly subpleural disease). A composite score to reflect disease distribution was derived for each case by totaling these two values (range, 0 to 6). The observers were also asked to categorize each HRCT scan as either typical of the published HRCT description of CFA or atypical for a diagnosis of CFA. Finally, HRCT scans were categorized into those which showed a uniform pattern of changes within 85% or more of the abnormal lung and those which showed a variable pattern. Pulmonary Function Tests In all cases lung function measurements were performed at presentation. Eleven patients in each group had carried out further lung function tests during the follow-up period. Tests included FEV 1, FVC, TLC, carbon monoxide diffusing capacity (DL CO ), and carbon monoxide diffusing capacity adjusted for alveolar volume (KCO). In some patients partial pressure of oxygen (Pa O2 ) in arterial blood was also measured. Results were expressed as percentages of values predicted from the subject s age, sex, and height (14). Throughout the study period, lung volumes were measured using an Ohio water-seal spirometer (Ohio Instruments, Atlanta, GA). TLC was measured in a wholebody plethysmograph (Fenyvens and Gut, Basel, Switzerland). Measures of gas transfer (DL CO, KCO) were made by the single-breath technique using a P.K. Morgan respirometer (P.K. Morgan, Chatham, Kent, UK). Criteria used to define significant changes in pulmonary function tests were identical to those used in previous studies (14, 15): improvement was defined as a rise of more than 15%, stability as a change of less than or equal to 15%, and deterioration as a decline of more than 15% from baseline values. Treatment Treatments used were identified from the medical records. Any effect of treatment on disease was assessed by comparison of lung function test results before and after diagnosis. Outcome Survival status in November 1998 was established either from hospital clinical records or from general practitioners records. Data Analysis Differences between subjects in the two groups (NSIP and UIP) were examined by the Mann-Whitney U test or chi-square test (for discrete variables). When patient numbers were small the chi-square test with Yates correction was used. Comparison of the frequency of improvement and deterioration in pulmonary function tests with treatment between the two histologically defined groups was made by the chisquare test. Comparison of the changes in pulmonary function tests during the follow-up period within each group was determined using the Wilcoxon matched pairs test. Survival was analyzed for those with UIP and NSIP and for those with typical and atypical appearances on HRCT. Survival curves were compared by the log rank test and Kaplan-Meier survival curves were plotted (16). RESULTS Pathologic Features In the 15 patients classified as having UIP there was independent agreement on diagnosis by both pathologists in all 15 cases. In the NSIP group, there was independent agreement in 10 cases. On reevaluation a consensus agreement on a diagnosis of NSIP was reached in five cases. Clinical Features All 30 patients had the clinical, radiologic, and physiologic criteria for the diagnosis of CFA. The NSIP group included seven men and eight women with a median age of 43 yr (range, 31 to 66 yr). The UIP group included 12 men and three women with a median age of 56 yr (range, 36 to 68 yr). Their clinical features are summarized in Table 1. No significant differences were found in age, sex, and smoking habits between the two groups. Functional indices of disease severity were similar in the two groups (Figure 1). The median duration of symptoms from onset to the first presentation to the unit were similar in the two groups: 18 mo (range, 7 to 84) in the NSIP group and 18 mo (range, 3 to 84) in the UIP group. Gradually increasing dyspnea and dry cough were the most common presenting symptom in both groups (Table 1). Chest pain was reported in two patients in the NSIP group and one in the UIP group. One patient in each group

3 Daniil, Gilchrist, Nicholson, et al.: Comparison of Histologic Subtypes in CFA 901 TABLE 1 DEMOGRAPHIC CHARACTERISTICS, CLINICAL FEATURES, AND INDICES OF LUNG FUNCTION AT PRESENTATION FOR ALL PATIENTS NSIP (n 15) UIP (n 15) p Value Age at presentation,* yr 43 (31 66) 56 (36 68) 0.21 Male 7 (47%) 12 (80%) 0.06 Ethnic origin 11 Caucasian 14 Caucasian 2 Indian 1 Indian 2 Afro-Caribbean Smoking, ever/never 9/6 12/ Presenting complaints Dyspnea Cough 9 9 Duration of symptoms before presentation, mo* 18.0 (7 84) 18.0 (3 84) 0.71 Crackles, n (%) 12 (80) 14 (93.3) 1.00 Clubbing, n (%) 6 (40) 14 (93.3) 0.01 FVC FEV TLC DL CO KCO PO 2, kpa* Definition of abbreviations: NSIP nonspecific interstitial pneumonia; UIP usual interstitial pneumonia. * Mean with standard deviation. Mean percentage of predicted value with standard deviation. complained of malaise. Wheezing was described by one patient in the NSIP group, and one patient in the UIP group had weight loss. One patient in the NSIP group had a history of possible viral pneumonia 7 mo before the open lung biopsy. Crackles were found in 12 patients (80%) in the NSIP group and in 14 patients (93%) in the UIP group. Digital clubbing occurred significantly more frequently in the UIP group in comparison with patients diagnosed as having NSIP (14/15 versus 6/15, p 0.01). Lung Function Pulmonary function tests at first presentation, including FEV 1, FVC, TLC, DL CO, KCO, and PO 2 were compared between the two groups. No significant difference was found in any of these indices (Table 1 and Figure 1). HRCT HRCT scans were available on 15 of 15 patients in the NSIP group and 12 of 15 in the UIP group. Results are shown in Table 2. The median time between HRCT and biopsy was 50 d in TABLE 2 RESULTS FOR SCORING OF HRCT SCANS FOR PATIENTS IN THE NSIP AND UIP GROUPS NSIP UIP p Value Number Median time between CT scan and biopsy, days (range) 50 (107 0) 56 (105 3) Fibrosis score,* median (range) 3.0 (0 7.0) 5.0 (2 7.0) Distribution score, median (range) 4 (0 6) 5 (1 6) Overal assessment Typical of CFA Atypical of CFA 13 4 Consistency of pattern Consistent Inconsistent 3 2 * Derived from the scores for honeycombing, traction bronchiectasis, and interstitial thickening. Derived from the scores for the degree of lower zone predominance and the predilection for a subpleural distribution. the NSIP group with a range of 107 to 0 d and 56 d in the UIP group with a range of 105 d before to 3 d after the biopsy. The fibrosis scores for the HRCTs in the two patient groups were compared. The median score in the NSIP group was found to be 3 (range, 0 to 7) compared with 5 (range, 2 to 7) in the UIP group (p 0.011). The HRCT scans were regarded as typical of CFA in two of 15 of the NSIP group, compared with eight of 12 in the UIP group (p 0.005). Only four HRCT scans in the UIP group were felt to be atypical for CFA compared with 13 in the NSIP group. Comparison of the scores reflecting distribution of the abnormalities seen on the HRCT scans failed to show a significant difference. In both groups the majority of scans were found to have consistency of pattern with only three in the NSIP group and two in the UIP group lacking homogeneity of the pattern of abnormal lung. DTPA Scan, ANA, and Rheumatoid Factor DTPA clearance was available in nine patients in the NSIP group and 10 in the UIP group. Clearance was abnormal in 10 of 10 patients with UIP and in seven of nine patients in the NSIP group (p 0.1). ANA was positive in five of the nine patients with NSIP in whom it had been checked compared with four of 13 in the UIP group (p 0.2). Rheumatoid factor was positive in two of 15 NSIP patients but negative in all the 13 UIP subjects in whom it had been checked (p 0.1). BAL BAL was performed at the first presentation in eight patients in each of the two groups. No significant differences were found on analysis of the lavage fluid in the proportions of lym- TABLE 3 BAL FLUID ANALYSIS IN PATIENTS WITH NSIP COMPARED WITH THOSE WITH UIP NSIP* UIP* p Value Figure 1. FVC and DL CO values at presentation in patients with UIP and NSIP. Horizontal bars indicate mean values in each group. Subjects 8 8 Macrophages, % Lymphocytes, % Neutrophils, % Eosinophils, % * Results are expressed as mean SD.

4 902 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL TABLE 4 EFFECTS OF TREATMENT ON LUNG FUNCTION IN PATIENTS WITH NSIP AND UIP NSIP (n 12) UIP (n 13) Improvement* 3 1 Stable 4 2 Deterioration 5 10 * Improvement a rise in FVC and/or DL CO of more than 15%. Unchanged no change or change of 15%. Deterioration a decline in FVC and/or DL CO of more than 15%. This category includes those who died during the first year of treatment (two in NSIP group and five in UIP group). phocytes, neutrophils, and macrophages in the two groups as a whole (Table 3). The numbers of patients in each group whose lavage fluid contained abnormal proportions of the different cell types were also comparable. Four patients in the NSIP group and five in the UIP group had above the normal range of eosinophils, one patient in each group had an abnormally high proportion of lymphocytes, and seven of eight in the UIP group and five of eight in the NSIP group had above the normal range of neutrophils in their lavage fluid. Effect of Treatment Of the 30 patients, 25 received treatment during the first year after biopsy, 12 in the NSIP group and 13 in the UIP group. Of the five patients untreated, two diagnosed as having UIP died shortly after presentation having been treated previously with both prednisolone and cyclophosphamide; two patients in the NSIP group had normal lung function and were not treated; one patient in the NSIP group was lost to follow-up immediately after lung biopsy. Seven patients, two in the NSIP group and five in the UIP group who were given treatment died within the first year after biopsy. Three of these failed to carry out further lung function tests after the diagnosis had been made. These are included in the assessment of response to treatment which is shown in Table 4. In the NSIP group, one patient received prednisolone only (60 mg daily for 6 wk, reducing over the next 9 wk to 15 mg TABLE 5 CHANGES IN LUNG FUNCTION BETWEEN PRESENTATION AND FINAL LUNG FUNCTION MEASUREMENT IN NSIP AND UIP PATIENTS NSIP (n 11) UIP (n 11) p Value Time from biopsy to final lung function assessment Median, mo Range, mo FVC change, mean SEM ml Percentage % % DL CO change, mean SEM mmol/min/kpa Percentage % % Dead 4 (29%) 13 (93%) Alive 10 (71%) 1 (7%) Total daily) and had an improvement in lung function tests. Eight patients received both prednisolone (20 mg on alternate days) and azathioprine (100 or 150 mg daily). Two died (one 4 mo and the other 6 mo after the initiation of the treatment), one patient improved, three remained stable, and the final two patients suffered a significant deterioration in lung function tests. Two patients in the NSIP group received prednisolone (20 mg on alternate days in combination with cyclophosphamide (150 mg and 100 mg daily). One deteriorated and the other remained stable. One patient in the NSIP group received cyclosporin (5 mg/kg/d) in combination with prednisolone (10 mg on alternate days) and improved. In the UIP group three patients received corticosteroids only (60 mg daily for 6 wk and reduced to 20 mg on alternate days). Two died 4 mo after the treatment had been started and the other showed significant deterioration in lung function. Three patients received both corticosteroids (two prednisolone 20 mg on alternate day and one prednisolone 20 mg daily) and azathioprine (150 mg daily). One improved, one remained stable, and one deteriorated. Six patients received corticosteroids (prednisolone 20 mg on alternate day in three Figure 2. Changes in DL CO between presentation and final lung function measurement in both groups.

5 Daniil, Gilchrist, Nicholson, et al.: Comparison of Histologic Subtypes in CFA 903 Figure 3. Changes in FVC between presentation and final lung function measurement in both groups. cases and 30 mg daily in two cases) in combination with cyclophosphamide (median dose 125 mg daily). Two died, one 4 mo and one 6 mo after the initiation of therapy, three deteriorated, and one remained stable. One patient in the UIP group was treated with -interferon and showed a decline in lung function. Overall seven of 12 patients (58%) with NSIP who were treated improved or remained stable on treatment compared with only three patients of 13 (23%) with UIP (p 0.03). Lung Function Changes during Follow-up to Last Visit or Death Eleven patients (73%) in each group performed lung function tests during follow-up after the diagnosis had been established by lung biopsy. The median duration of follow-up from the first to the final presentation to the unit was 14 mo (range, 3 to 72.5) in the NSIP group and 23.5 mo (range, 1.5 to 59) in the UIP group (Table 5). No significant changes were observed in either FVC where there was a mean improvement of % (mean SEM) or in DL CO (mean improvement of %) in the NSIP group as a whole. In contrast, patients diagnosed as having UIP showed a significant decline (p 0.03) in DL CO with a mean deterioration of % and in FVC (p 0.02) where there was a mean decline of % between presentation and final lung function measurements. Figures 2 and 3 show the changes seen in individual subjects. Three patients in the NSIP group showed a significant improvement in both FVC and DL CO during the follow-up period whereas DL CO improved significantly in a fourth with FVC remaining stable. Both indices were stable in four patients. Three patients suffered significant deterioration during the follow-up period, two in DL CO only and the third in both parameters. In the UIP group two patients improved significantly, two had stable lung function, and seven had a significant decline in pulmonary function tests, three in DL CO only and four in both DL CO and FVC. Taken as a group the patients with UIP had a significant loss in both FVC (p 0.03) and DL CO (p 0.05) compared with patients in the NSIP group. Survival status in November 1998 was ascertained for 14 patients in each group. In the NSIP group four patients (29%) died during the study period compared with 13 (93%) in the UIP group. This difference is significant (p 0.001). Survival from presentation was also significantly greater in the NSIP group compared with those diagnosed as having UIP (p 0.005) (Figure 4). Figure 5 compares survival in subjects according to the global assessment of their HRCT appearances at presentation. Survival and HRCT data were available on 10 subjects whose CT appearances were judged to be typical of CFA and in 14 whose CTs were felt to be atypical of a diagnosis of CFA. This shows that those with typical appearances have reduced survival compared with those with atypical appearances. The findings of this study are summarized diagramatically in Figure 6. This compares changes in FVC and DL CO and in survival during the study period in the two groups. DISCUSSION This study explored the clinical indices and outcome in a group of 30 patients who had been diagnosed as having CFA (IPF). The diagnosis had been made on the basis of clinical, radiographic, physiologic, and BAL features, coupled with a biopsy that was described at the time as being consistent with CFA. On review of the biopsy material, it was possible to subdivide this group of 30 individuals into those with an NSIP pattern and those with a UIP pattern (6, 10). This study showed Figure 4. Survival from presentation in both groups.

6 904 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL Figure 5. Comparison of survival in those with HRCT scans at presentation with the typical appearances of CFA (n 10) and those with scans atypical for CFA (n 14). that the response to therapy and survival was significantly better if a histopathologic pattern of NSIP was present as part of the CFA syndrome. The only clinical feature that showed a difference between the two groups was digital clubbing which was more frequent in the UIP group. However, we did find different patterns of abnormality on HRCT that increased the likelihood of predicting the pathologic subset preoperatively. Many studies have confirmed that the mean survival from time of diagnosis for CFA is 5 yr at most, but studies have also been consistent in demonstrating a tail on the survival curve (1 5, 17). It has long been suspected that this tail might represent a different disease subset or subsets. This study now provides evidence that would support this concept. Clinical outcome, radiology, and treatment response have all been quite different in the subset of individuals with CFA who have the NSIP pattern. The improved survival in patients with NSIP compared with those with UIP agrees with the findings of Katzenstein and Fiorelli in 1994 who assessed 64 cases of NSIP, and found Figure 6. Visual summary of the findings of this study. The first two pairs of columns comparing mean changes in FVC and DL CO during the study period in patients with NSIP and UIP. Positive values on the y-axis indicate improvement and negative values deterioration. The figure also provides a visual representation of survival in the two groups. The final two columns compare the percentage of patients who died in the two groups. In this case the y-axis indicates the proportion of patients in each group who died with a maximal value of 0% showing no deaths occurred to a minimum of 100% indicating the entire group died. an improved response to treatment and better prognosis when compared with UIP (10). These findings were supported by Bjoraker and coworkers in 1998 who studied 104 patients diagnosed as having IPF. These were reclassified on histological grounds into those with features of UIP, NSIP, and a group termed others. The 63 patients identified as having the UIP pattern showed reduced survival compared with both the 14 patients categorized into the NSIP group and the 20 who made up others (12). The Kyoto group studied 31 patients diagnosed as having NSIP and found their survival to be intermediate between those with UIP and those diagnosed with BOOP (13). The current study confirms the difference in survival determined by the histologic pattern and in addition has looked at radiologic and clinical correlations with histologic pattern showing that the difference in histology is reflected by different patterns on CT. We were surprised to find that, of our 30 patients with CFA, there were as many individuals with the NSIP as the UIP pattern. This suggests possible bias in selection. However, this is not the case. All patients who had surgical biopsies between the years 1991 and 1995 had their biopsies reviewed. From these, all patients with the UIP or NSIP pattern were selected. Those with clear etiologic factors or associated diseases such as systemic sclerosis, rheumatoid arthritis, or hypersensitivity pneumonitis were excluded. From the remaining patient group, only those who had full staging investigations at presentation were included. However, it remains surprising that there were so few patients with what we would regard as the more typical pattern of pathology, i.e., UIP. The reason for this is the use of HRCT. Since its introduction, we and others have reported that a coarse peripheral reticular pattern of disease on HRCT taken in the context of classic clinical features and investigation results predicts with a high degree of confidence, a diagnosis of fibrosing alveolitis (18 22). Thus, the selection of individuals for surgical biopsy has been restricted to those with some unusual features in their clinical, and particularly radiographic, workup. We suspect that classic cases of CFA with a UIP pattern are less likely than those with an NSIP pattern to have unusual features. HRCT was not absolute in being able to differentiate between these two pathologic processes. However, those with UIP tended to show more honeycombing and other signs of

7 Daniil, Gilchrist, Nicholson, et al.: Comparison of Histologic Subtypes in CFA 905 established fibrosis. In more subjective assessment, observers judged that only one of the HRCT scans from patients in the UIP group would have been regarded as atypical for CFA compared with eight in the NSIP group. Survival in those with typical appearances was found to be reduced compared with the other groups. This does therefore support the concept that HRCT can be of some help in distinguishing these groups, particularly in the more established honeycomb phase of the disease. This again is helpful in reducing the number of individuals who will require surgical confirmation of the diagnosis and will be particularly useful for individuals who have comorbid conditions making operative procedures more hazardous. This study has not been able to determine which treatment option is better for the different pathologic process. There are two reasons for this: first, the numbers in each group are too small and second, the spread of treatment options was very similar in the two groups. Answers to this question will require a prospective double-blind control study, stratified on the basis of pathologic subset to provide unequivocal rather than the current anecdotal evidence of treatment efficacy (10, 23, 24). With the exception of the HRCT pattern, no investigational index was helpful in differentiating the two histopathologic subsets. In particular, BAL showed a similar pattern of cellular return in the two groups. This is at variance with reports from the Kyoto group (13). Their studies have shown that BAL results in NSIP are similar to those seen in BOOP where there is an excess of lymphocytes. Our study did not confirm this. The reasons for this are unclear but could reflect ethnic or other differences in the population of patients studied. In summary, our study has shown that a diagnosis of CFA with an NSIP pattern carries a significantly better prognosis than a UIP pattern. This relates to response to treatment as well as survival. The only clinical feature that was different between the groups was the prevalence of digital clubbing. The only investigational difference was the pattern of HRCT abnormality. We conclude, therefore, that the pattern of histopathology is an important determinant in this disease and that we now have both noninvasive and histopathologic means of differentiating the two. Whether the NSIP variant requires change in the nomenclature from CFA must await further debate. Acknowledgment : The authors are grateful to Maria Bernardo for all her help in the preparation of this paper. References 1. Stack, B. H. R., Y. F. J. Choo-Kang, and B. E. Heard The prognosis of cryptogenic fibrosing alveolitis. Thorax 17: Turner-Warwick, M., B. Burrows, and A. Johnson Cryptogenic fibrosing alveolitis: clinical features and their influence on survival. Thorax 35: Carrington, C. B., E. A. Gaensler, R. E. Coutu, M. X. Fitzgerald, and R. G. Gupta Natural history and untreated course of usual and desquamative interstitial pneumonia. N. Engl. J. Med. 298: Tukiainen, P., E. Taskinen, P. Holsti, O. Korhola, and M. Valle Prognosis of cryptogenic fibrosing alveolitis. Thorax 38: Turner-Warwick, M., B. Burrows, and A. Johnson Cryptogenic fibrosing alveolitis: clinical features: response to corticosteroid treatment and its effect on survival. Thorax 35: Katzenstein, A. A., and J. L. Myers Idiopathic pulmonary fibrosis: clinical relevance of pathologic classification. Am. J. Respir. Crit. Care Med. 157: Katzenstein, A. A Idiopathic interstitial pneumonia: classification and diagnosis. In IAP Monograph Series: Current Topics in Lung Pathology. Williams and Wilkins, Baltimore, MD Liebow, A. A., and C. B. Carrington The interstitial pneumonias. In M. Simon, E. J. Potchen, and E. Lemay, editors. Frontiers in Pulmonary Radiology. Grune and Stratton, New York Katzenstein, A. A., J. L. Myers, and M. T. Mazur Acute interstitial pneumonia: a clinicopathologic, ultrastructural, and cell kinetic study. Am. J. Surg. Pathol. 10: Katzenstein, A. A., and R. F. Fiorelli Nonspecific interstitial pneumonia/fibrosis: histologic features and clinical significance. Am. J. Surg. Pathol. 18: Myers, J. L., C. F. Veal, Jr., M. S. Shin, and A. A. Katzenstein Respiratory bronchiolitis causing interstitial lung disease: a clinicopathologic study of six cases. Am. Rev. Respir. Dis. 135: Bjoraker, J. A., J. H. Ryu, M. K. Edwin, J. L. Myers, H. D. Tazelaar, D. R. Schroeder, and K. P. Offord Prognostic significance of histological subsets in idiopathic pulmonary fibrosis. Am. J. Respir. Crit. Care Med. 157: Nagai, S., M. Kitaichi, H. Itoh, T. Izumi, and T.V. Colby Idiopathic nonspecific interstitial pneumonia/fibrosis: comparison with idiopathic pulmonary fibrosis and BOOP. Eur. Respir. J. 12: Quanjer, P. H Standardised lung function testing. Clin. Respir. Physiol. 199(Suppl. 5): Chinet, T., D. Dusser, S. Labrune, M. A. Collignon, J. Chretien, and G. J. Huchon Lung function declines in patients with pulmonary sarcoidosis and increased respiratory epithelial permeability to 99mTc DTPA. Am. Rev. Respir. Dis. 141: Peto, R., M. C. Pike, P. Armitage, N. E. Breslow, D. R. Cox, S. V. Howard, N. Mantel, K. MacPherson, J. Peto, and P. G. Smith Design and analysis of randomised clinical trials requiring prolonged observation of each patient: analysis and examples. Br. J. Cancer 35: Agusti, C., A. Xaubet, J. Roca, A. Agusti, and R. Rodriguez-Roisin Interstitial pulmonary fibrosis with and without associated collagen vascular disease: results of a two year follow-up. Thorax 47: Tung, K. T., A. U. Wells, M. B. Rubens, J. M. E. Kirk, R. M. Du Bois, and D. M. Hansell Accuracy of the typical computed tomographic appearances of fibrosing alveolitis. Thorax 48: Wells, A. U., D. M. Hansell, M. B. Rubens, P. Cullinan, C. M. Black, and R. M. du Bois The predictive value of appearances on thin-section computed tomography in fibrosing alveolitis. Am. Rev. Respir. Dis. 148: Wells, A. U., M. B. Rubens, R. M. du Bois, and D. M. Hansell Serial CT in fibrosing alveolitis: prognostic significance of the initial pattern. Am. Respir. J. 161: Lee, J. S., J. Im, J. M. Ahn, Y. M. Kim, and M. C. Han Fibrosing alveolitis: prognostic implication of ground-glass attenuation at highresolution CT. Radiology 184: Terriff, B. A., S. Y. Kwan, M. M. Chan-Yeung, and N. L. Muller Fibrosing alveolitis: chest radiography and CT as predictors of clinical and functional impairment at follow-up in 26 patients. Radiology 184: Raghu, G., W. J. Depaso, K. Cain, S. P. Hammar, C. E. Wetzel, and D. F. Dreis Azathioprine combination with prednisone in the treatment of idiopathic pulmonary fibrosis: a prospective double-blind, randomized, placebo-controlled clinical trial. Am. Rev. Respir. Dis. 144: Johnson, M. A., S. Kwan, N. J. Snell, A. J. Nunn, J. H. Darbyshire, and M. Turner-Warwick Randomised controlled trial comparing prednisolone alone with cyclophosphamide and low dose prednisolone in combination in cryptogenic fibrosing alveolitis. Thorax 44: