Prognostic Value of Bronchoalveolar Lavage Lymphocyte Count in Recently Diagnosed Pulmonary Sarcoidosis

Prognostic Value of Bronchoalveolar Lavage Lymphocyte _Count in Recently Diagnosed Pulmonary Sarcoidosis* Michel Uwiolette, M.D., F.C.C.P..; }ocques lA Forge, M.D.; Sonia 7Bnnina, M.Sc.; and Louis-Philippe Boulet, M.D., F.C.C.P.. We prospectively looked at the prognostic value of broncboalveolar lavage (BAL) lymphocyte count in 98 patients with recently diagnosecl (<4 months) untreated sarcoidosis. 1bese 50 men and 48 women (mean age, 37.4) were followed up £or a period of 6 to 60 months (mean, 25.6), and were clinically evaluated every three to six months with repeated chest roentgenograms and pulmonary function tests.

'IWenty-Cour patients required steroid treatment during the study period. The proportion of treated patients was not signi&cantly higher in the group presenting a BAL lymphocyte count i!::30 penient at diagnosis than in the group with fewer lymphocytes (31.9 and 17.7 penient of total group

a granulomatous disease of unknown Sarcoidosis, etiology, is characterized by an increased number of T-lymphocytes in the lung, more specifically, the helper subset. 1-7 These activated T-lymphocytes may modulate the histologic changes seen in this disease. s. 10 Immunoglobulin-producing cells are also increased. 11 Alveolar macrophages are activated, secrete angiotensin convertase, and show an enhanced ability to present antigens. 12 The demonstration of such active inflammation raised the question of a possible role of these cells and their mediators in the lung functional impairment seen in this disease. Since the clinical course is variable and the outcome unpredictable, steroid treatment is usually given when significant morphologic and/or functional alterations are present. The availability of an accurate prognostic marker would be useful to identify the patients at risk of deterioration of their lung function and to allow possible early treatment. Keogh and colleagues 13 showed that subjects with high-intensity lymphocytic alveolitis had a worse prognosis than those with low-intensity lymphocytic alveolitis. However, their study included a small number of subjects who were followed up for a rather short period of time. Other studies looked at the value of lymphocytic alveolitis in predicting the outcome of sarcoidosis1'"13 and presented somewhat controversial data. This may. in part, be explained by the heterogeneity of studied populations, which included patients with long-standing or treated sarcoidosis, or racial differences. In this report, we present a prode Recherche, Centre de Pneumologie de l'H6pital et l'Universit6 Laval, Sainte-Foy (Qu6bec), Canada. Manuscript received J~ ~9i revision accepted December 10 llqlr!nt requeata: Dr: LIJVIOleffe, Cenm de Pneumologie, HopUal

*From the Uni~

-

Ltioal, Salnte-lby, Quebec, Canada GlV 4G5

respectively, p=0.10). No signi&cant change in TLC, FRC, FVC, FEV1 or DI.co was found at follow-up between the groups with or without an initial high lymphocyte count. In the treated group, BAL lymphocyte percent weakly correlaled with the improvement of FEV1 and FVC while on steroid treatment (mean duration: 3.5 months): r=0.41, p=0.031 and r=0.36, p=0.05 respectively; no correlation was found with lung volumes and Dco. We conclude that BAL lymphocyte count at the time of diagnosis is not a help£ul predictor of lung £unction deterioration in recently diagnosed sarcoidosis and is not very use£ul in predicting (Chat 1991; 100:380-84) response to treatment.

spective study of 98 white patients with newly diag: nosed sarcoidosis who were followed up for 6 to 60 months. We looked at the prognostic value of bronchoalveolar lavage (BAL) lymphocyte count obtained at the time of diagnosis of sarcoidosis and its value in the therapeutic decision. METHODS

Subjects All consecutive patients with recently diagnosed pulmonary sarcoidosis referred to the Laval Hospital Sarcoidosis Clinic during a period of 72 months from September 1980 to September 1986, were enrolled in this prospective study. We considered sarcoidosis recent ifsymptoms dated less than four months or, in asymptomatic subjects, if the first abnormal chest roentgenogram had been documented within the same period. The diagnosis of sarcoidosis was based on clinical presentation and chest x-ray abnormalities. Histologic confirmation was made by transbronchial biopsies and by other techniques (mediastinoscopy, lymph node biopsy, or salivary gland biopsy) when judged necessary. The diagnosis was also supported by excluding other entities and by the follow-up evaluation. Eoaluaflon

As part of their initial evaluation, subjects had posteroanterior and lateral lung roentgenograms, pulmonary function tests, and a BAL. Chest roentgenograms were interpreted according to the criteria of Scadding"' by a radiologist unaware of the patients' clinical status: stage 0, normal; stage 1, hi1ar adenopathies; stage 2, hi1ar adenopathies and pulmonary infiltrate; stage 3, pulmonary infiltrate; and stage 4, pulmonary fibrosis. Pulmonary function parameters tested were total lung capacity (TLC), forced vital capacity (FVC), forced expiratory volume in one second (FEV J, functional residual capacity (FRC) (plethysmography), and diffusing capacity with the single-breath carbon monoxide method (Dco). These parameters were measured as previously described"" and expressed as a percentage of predicted values.

BAL and l.aooge Fluid ITI1ceuing BAL was performed during ftexible bronchoscopy with the patient PIOgnosllc Value ol BAL Lymphocyte Count In San:oldo8ls (Laviolette et al)

under local anesthesia as previously described.• A segment or a subsegment of the right middle lobe or another lobe was lavaged. When there was an asymmetric distribution of lung infiltrates, lavage Buid was obtained from the region with the predominant infiltration. Five 20-ml aliquots (n =57) or four boluses of SO ml of 0.9 percent saline solution (n=41) were instilled and withdrawn by gentle aspiration with a syringe. The total cell count was determined on a bemocytometer. The cell differential was obtained by counting at least 300 cells on both Wrigbt-Giemsa and nonspeci6c esterasestained cytocentrifuge preparations. Recovery of less than 35 percent of the instilled volume was considered an inadequate return, and these results were discarded. A percentage of BAL lymphocytes 2:30 percent was considered to respresent higbintensity alveolitis; a lymphocyte count <30 percent, low-intensity alveolitis.

Snuly Population and lbllow-up One hundred nineteen subjects with newly diagnosed sarcoidosis 12 were seen at the clinic during the study period. ~excluded subjects who had a follow-up of less than six months, as well as nine subjects who had a BAL Buid recovery of less than 35 percent. This report is based on the remaining 98 subjects. During their followup, the patients were reevaluated at three- to six-month intervals. The reevaluations included a questionnaire, physical examination, pulmonary function tests, and a lung roentgenogram. The decision to treat (start prednisone therapy) was based on the following criteria: (1) impairment of organs other than the lung; (2) TLC and FVC less than 85 percent of the predicted values; (3) hypoxemia; and (4) persistent symptoms.

Stati8tical Analysis The proportions of treated subjects during follow-up were compared within categories: histologic confirmation of diagnosis, chest x-ray stage, and the lymphocyte count observed at the time of diagnosis (<30 percent or 2:30 percent). Statistical differences were assessed by x• tests. Age-adjusted comparison of proportions were made by logistic regression .., Means of the pulmonary function measures at the end of follow-up were compared for the two groups determined by the BAL lymphocyte count. In patients who were treated, only the pretreatment pulmonary function values were used for analysis. Adjusted means were estimated by covariance analysis and tested with an F statistic.• Residuals were examined to verify the normality assumption. In the treated group, the predictive value of BAL lymphocyte count for the responsiveness to steroid treatment was evaluated by correlating the variation of pulmonary functions before and during treatment to the initial BAL lymphocyte percent (Spearman's correlation coefficient). RESULTS

The 98 subjects included in the current analysis included 50 men and 48 women aged 17 to 69 years (mean, 37.4 years). All were white. Sixty had never smoked, 15 had stopped smoking for more than ten months (mean±SD, 59.7±45.7months), and 15were smokers (mean usage, 17.7±8.9 pack-years). The patients were followed up for a mean period of 25.6 months (range, 6 to 60 months). Histologic confirmation (presence of sarcoid granuloma on biopsy) was obtained in 48 subjects (49 percent). Transbronchial biopsies were performed in 76 subjects, and noncaseating granuloma was seen in 38 patients (50 percent). Mediastinoscopy, peripheral lymph node biopsy, and salivary gland biopsy showed histologic findings com-

Table 1-Pcdrnortatv Fundion and BAL Bwameten at the 7iru tfDiagnoeU

TLC•

FRC• FVC•

FEV,• DLc:o•

1btal number of cells (x 10'/ml) Macrophages, % Lymphocytes, % Neutrophils, % Eosinophils, %

No.

Mean

Range

'31 '31 98 98 96 98

95.3 102.2 90.6 86.3 88.7 18.7

69-128 56-159 48-122 42-125 49-155 0.7-66.4

98 98 98 98

68.5 29.7 1.5 0.3

12.0-'37.5 2.0-85.5 0.0-14.0 0.0-4.0

*Expressed as a percentage of predicted values.

patible with sarcoidosis in 7, l, and 2 patients, respectively. Three subjects had a normal lung roentgenogram (stage 0), 39 had hilar adenopathies (stage 1), 41 had both adenopathies and pulmonary infiltrates (stage 2), and 15 had pulmonary infiltrates alone (stage

3). Tuble 1 summarizes the results of pulmonary function testing and the BAL parameters of the studied subjects at time of diagnosis. The mean values of the pulmonary function tests for the group were normal. The BAL cell determinations showed an increased number ofcells (mean, 18.7 x 10'/ml) and an increased lymphocyte count (mean: 29. 7 percent) compared to normal volunteers studied in our laboratory over the same period (5.8 x 10'/ml and 9.6 percent, respectively).• Very few neutrophils and eosinophils were found. No correlation was found between BAL lymphocyte count and roentgenographic staging. A total of 24 patients (IO men and 14 women) fulfilled the treatment criteria chosen for this study: 18 at the first visit and two before and four after the first six-month follow-up. Reasons for treating were as follows: decreased lung volumes in 21 patients, decreased lung volumes associated with persistent symptoms in one, decreased lung volumes and myocardial involvement in one, and low Dco and significant hypoxemia in one. Seven patients received prednisone every other day starting with 40 mg for three months, and 15 were placed on a daily regimen beginning with 50 mg per day; steroids were then progressively tapered. In two patients, steroids were postponed because of concomitant medical conditions (severe diabetes and cardiac disease). These two patients were, however, included in the treated group. The treated patients were older (mean, 41.1 years) than the nontreated patients (mean: 36.1 yr), but this difference was not statistically significant (p=0.12). The two groups had similar lavage volume recovery (nontreated, 56.2 percent; treated, 60.5 percent [p = 0.15]; lavage total number of cells (18.8 X IOt/ml and 18.5X 10'/ml [p=0.94]; lavage lymphocyte perCHEST I 100 I 2 I AUGUST. 1991

381

Total No.of Subjects Biopsy Positive Negative Chest x-ray stage 0 1

2 3 BAL lymphocyte count <~ :2:~

100

Treated Subjects, ~ofTotal

p

80

48 50

9.5.0 24.0

3 39 41 15

0.0 12.8 29.3 46.7

0.04

20

51 47

17.7 31.9

0.10

0

0.91

60

40

centages (27.8 percent and 35.3 percent [p=0.09]), and lymphocyte counts (99 X 103/ml [recovered Javage ftuid] and 72X 103/ml, respectively [p=0.15]). The other cell counts were also similar. The treated patients were equally distributed in the positive or negative (including those without biopsy) biopsy groups; however, the proportion of treated patients increased from lung roentgenogram stage 0 to 3 (p = 0.04) ('lable 2). The proportion of treated patients was higher among the patients with a BAL lymphocyte count ::!::30 percent at diagnosis compared to those with lower lavage lymphocyte percent (31. 9 percent and 17. 7 percent, respectively). However, this difference did not reach statistical significance (p=0.10). Most of the treated patients (18 of 24) required treatment at the first visit. From this subgroup, seven patients had a BAL lymphocyte count <30 percent and 11 had a count ::!::30 percent (61 percent). On the other hand, 36 subjects out of the 80 not requiring treatment at the first visit (45 percent) had a lymphocyte count ;;::30 percent; the difference in the mean proportion of high-intensity alveolitis between these two groups was not significant (p=0.087). To evaluate the prognostic value of the initial BAL lymphocyte count, we excluded the 20 patients who were treated before six months of follow-up. Of the remaining patients, the 44 subjects with a highintensity alveolitis (c::30 percent lymphocytes) at diagnosis were compared with the 34 with a lower BAL lymphocyte count (<30 percent). Pulmonary function test results obtained at their last follow-up visit or before treatment were used as the evaluation criteria. In both groups, the mean duration of follow-up was 23.1 months. To ensure the comparability of the mean pulmonary function measurements in the two groups, all the analyses were adjusted for the duration of follow-up and for the pulmonary function values at the time of diagnosis. Pulmonary function values were not different for subjects with high- or low-intensity alve-

-

A

120

Table !-Proporlion of'lrealed .ftdienls According to Their c~"' 11ae 7irrae ofDiagnoeia

n.c

FRC

n.c

FRC

FVC

FEVI

DLCO B

120 100 80 60

40

20 0

FVC

FEVI

DLCO

FIGURE 1. Mean pulmonary function measured at the end offollowup for patients who bad a follow-up of at least six months without treabnent. A, All patients. 8, Patients with biopsy-proven sarcoidosis. Subjects were grouped according to the BAL lymphocyte count at diagnosis (<30 percent, solid bars; :2:30 percent, hatched bars). The means (SEM) are adjusted for the pulmonary function values obtained at diagnosis and the follow-up duration. No significant difference could be seen in these parameters between the subgroups of patients.

olitis at diagnosis (Fig l, A). Similar results were obtained when only subjects with biopsy-proven sarcoidosis were analyzed (Fig 1, B). Further analysis grouping subjects into three categories according to their percentage lymphocyte count (<30, 30 to 39, ::!::40 percent) instead of two showed no trend in any of these parameters (data not shown). We also evaluated whether the responsiveness to steroid treatment could be predicted by the percentage of BAL lymphocytes. The difference in the pulmonary functions before and during steroid treatment (mean duration of treatment, 3.5 months) were correlated to BAL lymphocyte counts. Weak but significant correlations were found between the lymphocyte count and AFEV1 (r=0.41, p=0.031) and AFVC (r=0.36, p=O.o.5) (Fig 2). The ATLC, AFRC, ADco, and ADcoNA values did not correlate with BAL PIOgnoallc Vlllue of BAL Lymphocyte Count In San:oidoals (l..avioletl8 et al)

PERCENT OF PREDICTED

AFEVI

40 0

30

0

0 0

20

0

0 0

10

0

0

0

0

0

0 0

0

0

0

'

0

r= 0.41 p=0.03

0 0

-10.+---.....-.......--.-.---..--.......--.-.---.....-----.-.-----. 20 80 100 0 40 60

AFVC

40 30

0

0 0

20

0

0

10 0

0

"

0

0

0

0 0

0

0

00

0

0

0

r=0.36 p =0.05

-10.+---.....-.......--.-.---..------.-..---.....-----.-.-----. 100 20 0 60 80 40 BAL LYMPHOCYTES(%) FIGURE 2. Correlation between . pretreatment BAL lymphocyte percentages and the variation ofFEV, and FVC (AFEV, and AFVC) with treatment in the 22 patients who received steroids. The improvements in both pulmonary function parameters showed a weak but significant correlation with lymphocyte count.

lymphocyte counts (r=0.33, r=0.18, r=0.04, and r=0.30, respectively). DISCUSSION

In this prospective study, we show that BAL lymphocyte count at diagnosis is not a valuable prognostic factor in patients with newly diagnosed sarcoidosis. Patients who needed treatment had a BAL lymphocyte percentage similar to that of those who did not; therefore, BAL lymphocyte count was not useful in the clinical decision to treat. In patients who required treatment, the BAL lymphocyte count weakly predicted response. In comparison with other reports, 14-113 this study has a large number of untreated patients, all with newly diagnosed sarcoidosis and a long follow-up. Our obser-

vations do not include patients with long-standing sarcoidosis and previous physiologic changes or treatment. Moreover, all our subjects were white. Therefore, the study population was more homogeneous than in previous reports. Although histologic confirmation was obtained in only 48 of 98 patients, in the others, the diagnosis of sarcoidosis was strongly supported by the typical clinical presentation and elimination of other diseases by the long follow-up (mean, 25.6 months). We believe it is, therefore, quite unlikely that entities other than sarcoidosis could have been included in the 50 patients without histologic confirmation. Furthermore, subanalysis showed similar results for the nonhistologically proven sarcoidosis compared to the whole group (Fig 1). We did not measure the number of BAL T-lymphocytes but used the total number of lymphocytes for the statistical analysis. Since in sarcoidosis there is an important increase in BAL T-lymphocyte count (.... 90 percent of the total lymphocyte population),2 we used 30 percent as an arbitrary cutoff between high- and low-intensity alveolitis. We believe that this value is similar to that proposed by Keogh et al. 13 We did not obtain a lung gallium scan to further define the highintensity alveolitis group as proposed by Keogh et al. 13 They found that a patient with a T-lymphocyte count >28 percent rarely (13 percent) had a negative lung gallium scan, but that BAL lymphocyte count alone could not reliably predict functional deterioration. Our study confirms that this criterion cannot identify patients whose condition will deteriorate. Although we included only subjects with recently diagnosed sarcoidosis, interestingly, most patients who required treatment did so at the first visit (18 of 24). Treated patients had a higher incidence of pulmonary infiltrates than nontreated patients, but not a higher incidence of high-intensity lymphocytic alveolltis. Recently, Ward et al29 showed that a high BAL lymphocyte count was found in subjects with acute-onset sarcoidosis, such as that in erythema nodosum and uveitis, entities which have a good prognosis. Therefore, their data, like ours, suggest that a high BAL lymphocyte percentage is not related to lung function impairment. Our data show that pretreatment BAL lymphocyte count has some predictive value for the therapeutic response to steroid. Hollinger et al15 studied a group of patients of different races, different duration of disease, and, in some cases, previous steroid treatment. They found a significant correlation between the BAL lymphocyte count and FVC, but not with the other parameters measured. Contrary to Hollinger et al, 15 we do not believe that the BAL lymphocyte count can be clinically useful; the correlation between lymphocyte percentage and pulmonary function presents too much variability and is too weak to be applied CHEST/ 100 I 2 I AUGUST, 1991

383

to any specific patient. The presence of a high BAL lymphocyte count at diagnosis was not associated with the need to treat, either at diagnosis or at follow-up. This is similar to what was found in extrinsic allergic alveolitis where also no correlation between the presence of lymphocytic alveolitis and deterioration of lung function was found. 30 In asymptomatic farmers, large numbers of lymphocytes have been found in repeated lavages without evidence of lung function abnormalities or deterioration. 31 •32 In sarcoidosis, inftammatory cells could be present in the lung without altering lung function, inducing fibrosis and irreversible morphologic changes, and ultimately, predicting an unfavorable outcome. REFERENCES 1 Thomas PD, Hunningbake GW. Current concepts of the pathogenesis of sarcoidosis. Am Rev Respir Dis 1987; 135:747-60 2 Hunningbake GW, Crystal RG. Pulmonary sarcoidosis: a disorder mediated by excess helper T-lymphocyte activity at sites of disease activity. N Engl J Med 1981; 305:429-34 3 Danel C, Dewar A, Corrin B, Turnel"Warwick M, Chretien J. Ultrastructural changes in bronchoalveolar lavage cells in sarcoidosis and comparison with the tissue granuloma. Am J Pathol 1983; 112:7-17 4 Ceuppens JL, Lacquet LM, Marii!n G, Demedts M, Van Den Eeckhout A, Stevens E. Alveolar T-cell subsets in pulmonary sarcoidosis: correlation with disease activity and effect of steroid treatment. Am Rev Respir Dis 1984; 129:563-68 5 Costabel U, Bross KJ, Riihle KH, Inhr GW, Matthys H. Ia-like antigens on T-cells and their sub-populations in pulmonary sarcoidosis and in hypersensitivity pneumonitis: analysis of bronchoalveolar and blood lymphocytes. Am Rev Respir Dis 1985; 131:337-42 6 Rossi GA, Sacco 0, Cosulich E, Risso A, Balbi B, Ravazzoni C. Helper T-lymphocytes in pulmonary sarcoidosis: functional analysis of a lung T-cell subpopulation in patients with active disease. Am Rev Respir Dis 1986; 133:1086-90 7 Saltini C, Spwzem JR, Lee JJ, Pinkston P, Crystal RG. Spontaneous release of interleukin-2 by lung T lymphocytes in active pulmonary sarcoidosis is primarily from the Leu+ DR+ T-cell subset. J Clio Invest 1986; 77:1962-70 8 Hunnlnghalre GW, Gadek JE, Young RC, ICawanami 0, Ferrans VJ, Crystal RG. Maintenance of granuloma formation in pulmonary sarcoidosis by T lymphocytes within the lung. N Engl J Med 1980; 302:594-98 9 Robinson BWS, McLemore TL, Crystal RG. Gamma interferon is spontaneously released by lung T-lymphocytes and alveolar macrophages in patients with pulmonary sarcoidosis. J Clio Invest 1985; 75:1488-95 10 Pinkston P, Bitterman PB, Crystal RG. Spontaneous release of interleukin-2 by lung T-lymphocytes in active pulmonary sarcoidosis. N Engl J Med 1983; 308:793-80 11 Hunningbake GW, Crystal RG. Mechanisms of hypergammaglobulinemia in pulmonary sarcoidosis: site of increased antibody production and role ofT-lymphocytes. J Clio Invest 1981;

67:86-92 12 Hunningbake GW. Release of interleukin-I by alveolar macrophages of patients with active pulmonary sarcoidosis. Am Rev Respir Dis 1984; 129:569-72 13 Keogh BA, Hnnninghalce GW, Line BR, Crystal RG. The alveolitis of pulmonary sarcoidosis: evaluation of natural history

384

and alveolitis-dependent changes in lung function. Am Rev Respir Dis 1983; 128:256-65 14 Chretien J, Venet A, Danel C, Israel-Biet D, Snadron D, Arnoux A. Bronchoalveolar lavage in sarcoidosis. Respiration 1985;

48:222-30 15 Hollinger WM, Staton GW Jr, Fajman WA, Gilman MJ, Pine JR, Check IJ. Prediction of therapeutic response in steroidtreated pulmonary sarcoidosis. Am Rev Respir Dis 1985; 132:6569 16 Baughman RP, Fernandez M, Bosken CH, Mantil F, Hurtubise P. Comparison of gallium-67 scanning, bronchoalveolar lavage, and serum angiotensin-converting enzyme levels in pulmonary sarcoidosis: predicting response to therapy. Am Rev Respir Dis 1984; 129:676-81 17 Costabel U, Bross KJ, Guzman J, Nilles A, Riihle KH, Matthys H. Predictive value ofbronchoalveolar lavage T cell subsets for the course of pulmonary sarcoidosis. Ann NY Acad Sci 1986; 465:418-26 18 Israel-Biel D, Venet A, Chretien J. Persistent high alveolar lymphocytosis as a predictive criterion of chronic pulmonary sarcoidosis. Ann NY Acad Sci 1986; 465:395-406 19 Buchalter S, App W, Jackson L, Chandler D, Jackson R, Fulmer J. Bronchoalveolar lavage cell analysis in sarcoidosis: a comparison of lymphocyte counts and clinical course. Ann NY Acad Sci 1986; 465:678-84 20 Greening AP, Nunn P, Dobson N, Rudolf M, Rees ADM. Pulmonary sarcoidosis: alterations in bronchoalveolar lymphocytes and T cell subsets. Thorax 1985; 40:278-83 21 Turnel"Warwick M, McAllister W, Lawrence R, Britten A, Haslam PL. Corticosteroid treatment in pulmonary sarcoidosis: do serial lavage lymphocyte counts, serum angiotensin converting enzyme measurements, and gallium-67 scans help management? Thorax 1986; 41:903-13 22 Bjemer L, Bosenhall L, Angstrom T, Hallgren R. Predictive value of bronchoalveolar lavage cell in sarcoidosis. Thorax 1988; 43:284-88 23 Foley NM, Coral AP, Tung IC, Hudspith BN, James DG, Johnson NMcl. Bronchoalveolar lavage cell counts as a predictor of short term outcome in pulmonary sarcoidosis. Thorax 1989; 44:732-

38 24 Scadding JG. Prognosis of intrathoracic sarcoidosis in England. Br Med J 1961; 2:1165-72 25 Cormier Y, Belanger J, 'Iludif A, Leblanc P, Laviolette M. llelationships between radiographic change, pulmonary function, and bronchoalveolar lavage fluid lymphocytes in farmers lung disease. Thorax 1986; 41:28-33 26 Laviolette M. Lymphocyte fluctuation in bronchoalveolar lavage of normal volunteers. Thorax 1985; 40:651-56 27 Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research: principles and quantitative methods. Toronto: Lifetime Learning Publications, 1982 28 Kleinbaum DG, Kupper LL. Applied regression analysis and other multivariate methods. Boston: Duxbury Press, 1978 29 Ward IC, O'Connor C, Odium C, Fitzgerald MX. Prognostic value of bronchoalveolar lavage in sarcoidosis: the critical influence of disease presentation. Thorax 1989; 44:6-12 30 Cormier Y, Belanger J, Leblanc P, Laviolette M. Bronchoalveolar lavage in farmer's lung disease: diagnostic and physiological significance. Br J Ind Med 1986; 43:401-5 31 Gariepy L, Cormier Y, Laviolette M, 'Iludif A. Predictive value ofbronchoalveolar lavage cells and serum precipitins in asymptomatic dairy farmers. Am Rev Respir Dis 1989; 140:1386-89 32 Moore VL, Pederson GM, Hauser WC, Fink JN. A study of lung lavage materials in patients with hypersensitivity pneumonitis: in vitro response to mitogen and antigen in pigeon breeder's disease. J Allergy Clio Immunol 1980; 65:365-70 0

PIOgno8llc Value of BAL Lymphocyte Count In San:oldoela (Lavloletl8 et el)