Comparison of inflammatory cell counts in asthma: induced sputum ...

Clinical and Experimental Allergy, 1997, Volume 27, pages 769-779

Comparison of inflammatory cell counts in asthma: inducedsputum vs bronchoalveolar lavage and bronchial biopsies

D. C. GROOTENDORST, J. K. SONT, L. N. A. WILLEMS,J. C. KLUIN-NELEMANS*, J. H. J. M. VAN KRIEKENf,M. VESELIC-CHARVATt atid P. J. STERK

Departments of Pulmonology, *Haematology, Laboratory of Haematomorphology and Immunotyping and"^Pathology. Leiden University Medical Centre. The Netherlands

Summary

Background Induced sputum potentially allows monitoring of airway inflammation inpatients with asthma in a non-invasive way. However, the relationship between the cellularcontent in sputum and airway tissue has not been fully clarified.Objective We compared the cellular compositions of hypertonic saline-induced sputum,bronchoalveolar lavage fluid (BAL) and bronchial biopsies in 18 clinically stable patientswith mild to moderate atopic asthma (baseline FEV,: range 61-114%pred, PC20 methacho-line: 0.04-4.7 mg/mL). They were treated with inhaled short-acting bronchodilators ondemand, with (n = 8) or without (n = 10) regular inhaled steroids.Methods Each patient underwent sputum induction and fiberoptic bronchoscopy onseparate days in random order. Differential cell counts of induced sputum, bronchoalveolarlavage and bronchial wash were determined on May-Grunwald-Giemsa stained cytospins.Flow cytometry was performed on sputum and BAL samples. Immunohistochemicaltechniques were used to stain inflammatory cells in 6 fxxn cryostat sections of bronchialbiopsies.Results Sputum cell differentials were not different between the patients with and withoutinhaled steroids, and showed a median value of 19.4% squamous cells, with 1.0%eosinophils, 3.3% lymphocytes, 28.7% neutrophils, 49.4% macrophages and 6.9% cylindricepithelial cells (in percentage non-squamous cells). The percentage eosinophils in sputumwas significantly correlated with their percentage in bronchial wash {R^ = 0.52, P = 0.03)and in BAL (/?s = 0.55, P = 0.02), whilst there was a trend towards such a correlationbetween the number of eosinophils/mL sputum and the number of EG2"'" eosinophils/mm^lamina propria in bronchial biopsies (7?, = 0.44, P = 0.07). In addition, the percentage ofCD4"*" lymphocytes correlated between sputum and BAL {R^ — Q.55, f = 0.03).Conclusion We conclude that the eosinophil counts in hypertonic saline-induced sputumfrom patients with asthma are related to those in bronchial wash and BAL and, to a lesserextent, with the counts in bronchial biopsies. This suggests that induced sputum can be usedto monitor the presence and severity of airway inflammation in asthma.

Keywords: airways, bronchoscopy, eosinophils, flow cytometry, lymphocytes

Clinical and Experimental Allergy, Vol. 27, pp. 769-779. Submitted 28 June 1996; revised4 November 1996; accepted 14 November 1996.

Correspondence: Professor P. J. Sterk, Lung Function Laboratory, C2-P, IntroductionDepartment of Pulmonology, Leiden University Medical Centre, PO Box r r . i i - r - i . 1.1 ri--9600, NL-23OORC Leiden, The Netherlands. ^^e charactenstic feature m the pathology of asthma isSupported by a grant from the Commission of the European Communities airway inflammation [1], which has extensively been inves-(ECSC grant 7280-03-056). tigated in bronchial biopsies and bronchoalveolar lavage

© 1997 Blackwell Science Ltd 769

770 D. C. Grootendorst et al.

(BAL). Bronchoscopy, however, is an invasive procedure,and is not easily applicable in large scale diagnostic andfollow-up studies. Analysis of sputum might be analternative, non-invasive method to obtain airway secre-tions, which can potentially be used to monitor airwayinflammation in asthma.

Studies of bronchial biopsies from asthmatic subjectshave demonstrated increased infiltration of macrophages,T lymphocytes, eosinophils and mast cells into the laminapropria when compared with healthy subjects [2-5]. Inaddition, increased numbers of activated lymphocytes andeosinophils have been found in the lamina propria [6]. Theinflammatory state of the airways in biopsies and BALappears to be dependent on the treatment level of thedisease. In asthmatics using inhaled steroids fewer T lym-phocytes, eosinophils, and mast cells have been found inbronchial biopsy specimens as compared with asthmaticsusing inhaled short-acting |S2-agonists only [4]. And eventhough treatment of asthmatic subjects with inhaled steroidsdoes not affect cell differentials in BAL [7,8], it does reducethe absolute number of recovered eosinophils and theexpression of the T lymphocyte activation markers CD25and HLA-DR in BAL [8].

Similar changes have been observed in the cellularcontent of sputum, either spontaneously expectorated orinduced by inhalation of hypertonic saline aerosols, inasthma. Induced sputum from asthmatics is characterizedby an increased percentage of eosinophils and mast cells,and by a decreased percentage of macrophages when com-pared to healthy subjects [9-11]. Exacerbations seem to beassociated with even higher percentages of eosinophils insome, but not all patients, together with an increase in mastcells and a decrease in the percentages of macrophages ininduced sputum [12,13]. These findings have been extendedby using intervention studies. Fahy et al. have shown thateosinophils and neutrophils are elevated in induced sputum24 h after allergen challenge [14], whereas treatment ofasthmatic subjects with oral corticosteroids decreased thepercentage of eosinophils [15].

At present, it is unknown whether cellular inflammationin induced sputum is fully representative for the ongoinginflammatory process in airway tissue. The first studies onthe comparison between the cellular content of sputum withthat in bronchial wash (BW) and BAL in patients withasthma show variable correlations in eosinophil numbersbetween induced sputum and BW [16-18], and betweeninduced sputum and BAL [18-20]. One of the studies alsoshowed a correlation between induced sputum and BW orBAL for neutrophils [18]. With regard to the relationshipbetween induced sputum and bronchial biopsies, it hasrecently been reported that there is a trend towards a corre-lation between the two materials for eosinophil counts only[20].

The aim of this study was to further examine whetherthere is a relationship between inflammatory cell counts inhypertonic saline-induced sputum on the one hand, andthose in BW, BAL and bronchial biopsies on the other, ina relatively large group of asthmatic subjects. To that end,we have conducted a study in clinically stable atopicasthmatic subjects, covering a wide range in lung functionand airway hyperresponsiveness. In these patients, we com-pared the inflammatory cell counts between sputum inducedby inhalation of hypertonic saline, and BW, BAL andbronchial biopsies as obtained by fibreoptic bronchoscopy.

Methods

Subjects

Eighteen non-smoking, atopic asthmatic subjects partici-pated in this study. Ten subjects (seven male and threefemale, 18-44 years) used inhaled short-acting /32-agonistson demand only, and eight subjects (four male and fourfemale, 19-50 years) were treated with regular inhaledcorticosteroids and inhaled short- and/or long-acting 182-agonists on demand. All subjects had an FEV] >60% pre-dicted [21], which was reversible in all subjects by >12%predicted [21] or by reaching the normal range (>80% pred)after administration of 400 /xg salbutamol per metered doseinhaler (MDI) connected to an aerosol chamber. All subjectswere hyperresponsive to inhaled methacholine as shown bya provocative concentration to cause a 20% fall in FEVi(PC20FEV1) of 8 mg/mL or less [22]. Atopy was indicatedby a positive skin-prick test (>3 mm weal) to one or morecommon airbome allergen extracts (Soluprick, ALK, Den-mark). They were clinically stable without a recent historyof an upper respiratory tract infection. During the study,FEV| had to be stable (within 10% range of predictedvalue). The subject characteristics are shown in Table 1.The study was approved by the Medical Ethics Committeeat the Leiden University Medical Centre, The Netherlands.

Design

All subjects attended the laboratory on four occasions. Atthe first visit, clinical characteristics were documented by aquestionnaire, and a skin-prick test and spirometry werecarried out. Bronchial hyperresponsiveness to methacholinewas measured approximately 14 days later (second visit).On visit 3 (at least 48 h later) and visit 4 (within 4-12 days)either sputum was induced or fibreoptic bronchoscopy wasperformed. The sequence of these procedures was rando-mized in order to preclude any bias due to potential carry-over effects. Immediately prior to bronchoscopy, reversi-bility of lung function to 400/^g of inhaled salbutamol wastested. Short-acting bronchodilators were withheld for at

1997 Blackwell Science Ltd, Clinical and Experimental Allergy, 27, 769-779

Comparison of inflammatory cell counts in asthma 771

Table 1. Subject characteristics

Subject nr

123456789

10

Mean (SD)

1112131415161718

mean (SD)

Sex

MMMFMMFMMF

MFFMMFMF

Age (years)

20372224232324264418

26,1 (8,1)

5019363022212521

28,0(10,5)

FEV| (%pred)

113,078,362,061,295,088,084.183,085,0

107,0

85,7 (16,7)

113,065,084,092,076,183,993,0

114,0

90,1 (16,9)

% reversibility*

7,022,028,016,58,0

16,97,87,07,0

15,0

13,5 (7,4)

7,021,013,09,07,4

33,06,06,0

13,8 (9,9)

PC20 METHt

1,140,700,100,421,280,710,0420,300,560,25

0,37 (L6):l:

0,550,304,670,760,410,0960,100,36

0,41 (l,8)t

Medication

S pmS pmS pmS pmS pmS pmS pmS pmS pmS pm

BUD 1600, S pmBDP 800, S pmBDP 200, S pmBUD 400, T pmBDP 400, S pmBDP 900, S pm

BDP 800, Sm 100, S pmBUD 1600, Tpm

* improvement in FEV, in %pred 15min after inhalation of 400;tg salbutamol per MDI connected to an aerosol chamber,t provocative concentration of methacholine causing a 20% fall in FEVj,i geometric mean PC20 (SD in doubling doses), S, salbutamol; T, terbutaline; Sm, salmeterol; BUD, budesonide; BDP, beclomethasonedipropionate daily dose in ^g.

least 8 h, and long-acting bronchodilators were withheld for24 h before each procedure, whilst the prescribed doses ofinhaled corticosteroids were continued.

Spirometry

Flow-volume curves were recorded using a dry rolling sealspirometer (Morgan Spiroflow, UK) and displayed on a XY-recorder (Kipp, The Netherlands) with a timing device toprovide a spike to determine FEVj, Maximal flow-volumecurves were performed according to European standards[21],

Inhalation challenge test

Methacholine provocation tests were performed accordingto the standardized 2min tidal breathing method [22], Theresponse to methacholine was measured by FEV, at 90safter inhalation of each dose. The provocative concentrationcausing 20% fall in FEV, (PC20FEV1) was calculated bylinear interpolation of the last two points of the logconcentration-response curves, Salbutamol (200/tg) was

administered by MDI connected to an aerosol chamber,after completion of the challenges.

Sputum induction

Sputum was induced by inhalation of hypertonic (4.5%)saline aerosols [23]. Sodium chloride aerosols were gene-rated at room temperature by a DeVilbiss Ultraneb 99ultrasonic nebulizer at maximal output setting (2.5 mL/min). The generated aerosols of this nebulizer have aparticle size with a mass median aerodynamic diameter of4.5 ^m [24], The aerosols were administered to the subjectthrough a 100 cm long tube with an internal diameter of22 mm, and inhaled by the mouth through a two-way valvebox (no,2700; Hans Rudolph, Kansas City, MO, USA) withthe nose clipped.

Immediately before inhalation of the aerosols, baselineFEV] was recorded, and 200/xg salbutamol were admini-stered per MDI connected to an aerosol chamber, for safetyreasons. The subjects then inhaled hypertonic saline aero-sols in periods of 5 min up to a maximum of 50 min or untilenough sputum was obtained (approximately 5 g). Every

© 1997 Blackwell Science Ltd, Clinical and Experimental Allergy, 27, 769-779


5 min, subjects were asked to rinse their mouth and throatwith water and to expectorate sputum into a clean plasticcontainer by coughing. FEV i was monitored directly beforeand after sputum induction, and whenever the subject feltuncomfortable. When FEVi fell by more than 20% frombaseline during or after sputum induction, a final dose of200 fx,g salbutamol was administered.

Sputum processing

Sputum samples were processed according to the method ofFahy et al. [11,23]. Differential cell counts of leucocytes,cylindric epithelial cells and squamous cells were performedby a qualified cytologist on coded, May-Grlinwald-Giemsastained cytospins. A sputum sample was considered ade-quate when the percentage of squamous cells was lowerthan 80%. To correct for the variable salivary contami-nation, differential leucocyte and cylindric epithelial cellcounts were expressed as percentages of 500 nucleated cellsexcluding squamous cells. Cell differentials of metachro-matic cells (mast cells and basophils) were made by count-ing 5000 nucleated cells on slides stained with 1 % toluidineblue (w/v) in methanol/aqua dest (v/v: 30/70, pH6.6).Metachromatic cells were expressed as the number of posi-tively stained cells per 1000 (non-squamous) cells. Further-more, absolute cell numbers per mL sputum were calculatedby multiplying the cell percentage with the total (non-squamous) cell number in sputum, divided by the volumeof the sputum sample.

Processing of biopsies and cell quantification

The biopsies were embedded in OCT medium (Miles Inc.Diagnostics Division, Elkhart USA), and snap-frozen inisopentane cooled by iced CO2 [27]. Samples were storedat —70 °C pending further processing. Six /xm sections wereair-dried for 1 h, and fixed in a mixture of equal partsacetone:methanol for 2 min. Immunohistochemical stainingwas performed using an indirect immunoperoxidase techni-que with the following monoclonal antibodies against:eosinophil cationic protein (EGl), secreted form of eosino-phil cationic protein (EG2) (both Pharmacia, Sweden), mastcell tryptase (AAl, kindly donated by Dr R. Djukanovic,Southampton University Hospital, Southampton, UK),neutrophil elastase (Dako, The Netherlands), CD45 (CLA),CD3 (UCHT-1), CD4 (leu 3a) and CD8 (leu 2a) (all BectonDickinson, Mountain View, CA, USA). Phosphate bufferedsaline was used as a negative control of immunostaining.

For each antibody and each localization, the number ofpositively stained cells were counted in a 115 ^m deep,randomly selected zone along the length of the epithelialbasement membrane by means of a Video Interactive Dis-play Analysis System (VIDAS II, Kontron Electronik GmbH,Munich, Germany) [6,27]. Slides were coded and countedtwice by one observer in a blinded fashion. Light micro-scopy analysis was performed using a Zeiss microscope at amagnification of x200. The mean number of positive cells/mm'̂ lamina propria, calculated on pooled data of the threelocalizations, is presented. Counted areas covered at least

Bronchoscopy

Fibreoptic bronchoscopy and BAL were carried out accord-ing to a standardized protocol based on recent recommen-dations [25-27], and was performed using a Pentaxbronchoscope (outer diameter 6 mm; Pentax Optical Co.,Japan) [27]. The bronchoscope was introduced through thenose or mouth with the patient in supine position andwedged into a subsegmental bronchus of the lingula.Lavage was performed with three 50 mL aliquots of physio-logical saline pre-warmed to 37 °C. Each aliquot was imme-diately recovered using a negative pressure <20 mmHg. Thefirst recovered tube of fluid (maximal 20 mL) was con-sidered as BW. After completion of lavage, a total of threemucosal biopsies were taken from the main carina, themiddle lobe carina and a segmental carina of the rightlower lobe using cup forceps (Olympus FB-21C, Tokyo,Japan). Throughout the procedure 100% oxygen was deliv-ered via a nasal cannula at a rate of 4 L/min whilst HbO2saturation was monitored continuously using an oximeter(N-180, Nellcor, Hay ward, CA, USA) with the probe placedon a finger.

Processing of lavage fluid

After recovery of the lavage, fluids were kept on ice pendingfurther processing. BW and BAL fluids were filtrated andcentrifugated at 1500r.p.m. (4°C) for 10 min. The super-natant was aspirated and stored at —20°C for later analysis.The cell pellet was resuspended in PBS/BSA 1% (pH7.8).The cell suspension was used for determination of the cellconcentration using a haemacytometer (Tamson, Zoeter-meer. The Netherlands), cytospin preparation (Nordic cyto-centrifuge, Tilburg, The Netherlands) and determination ofcell markers by flow cytometry.

Differential cell counts of BW and BAL of both inflam-matory and metachromatic cells were performed in the sameway as in sputum. However, squamous epithelial cells werenot counted since the percentage of these cells in lavagefluid was negligible.

Cell preparation for flow cytometry

For detection of the different T lymphocyte subpopulations(CD3, CD4 and CD8 positive cells) both sputum and BAL



were analysed by flow cytometry. Monoclonal antibodieslabelled with fluorescein (FITC), phycoerythrin (PE) orperidinin-chlorophyll-protein-conjugate (PerCP) (all obtainedfrom Becton Dickinson, USA) were used to study the fol-lowing combinations of determinants: CD45 (FITC) andCD 14 (PE) (positive control to set the lymphocyte gate),cells only (control for determination of autofluorescence)and CD4 (FITC) and CD8 (PE). In addition, mouse IgGl(FITC) and IgG2 (PE) were used as a control for aspecificbinding. As a third fluorescent parameter, CD3 (PerCP) wasused in sputum samples and LDS-751 in methanol con-taining 1% paraformaldehyde in BAL samples, CD3(PerCP) was added to the cell suspensions at the sametime as the previously mentioned combinations ofmonoclonal antibodies.

In order to be able to set lymphocyte gates, heparinizedblood was used. As a control for sputum, blood of eachindividual subject was lysed with NH4CI for 20 min on ice,centrifugated (10 min 1500r,p.m.), the pellet resuspended inlysis buffer and centrifugated again, followed by sputolysintreatment as sputum. As control for BAL, lymphocytes wereisolated from blood with a standard ficoU procedure.

A maximum of 0,5 x 10^ cells in PBS/BSA 1% (pH7.8)were incubated with saturating concentrations of the abovementioned combinations of monoclonal antibodies (to afinal volume of 100 fiL) for 30 min in the dark. All stepswere performed at 4°C. Suspensions were washed with4mL of PBS/BSA 1% (pH7.4) (lOmin 1500r,p.m,), Thecell pellet of sputum and control blood samples were thenresuspended in 0,5 mL PBS/BSA 1% (pH7.8), and the cellpellet of BAL and control blood samples in 0,5 mL metha-nol containing LDS-751 and 1% paraformaldehyde andincubated 30 min at 4°C in the dark. Cell suspensionswere stored in the dark at 4°C pending flow cytometricanalysis,

Elow cytometry

Flow cytometric analyses of the labelled sputum, BAL andperipheral blood (lymphocyte) samples were performed ona FACScan (Becton Dickinson, USA), Subsequent gene-ration of graphics and analysis of lymphocyte subsets wereperformed on a Hewlett-Packard 340 computer.

For each combination of monoclonal antibodies, 10000cells of BAL and blood, and 50 000 cells of sputum wereanalysed. Both for sputum and BAL, a gate on lymphocytesand a gate on CD3''' cells were combined in order to be ableto analyse lymphocytes. Gates were set on blood samplesand the same gate specifications were used for sputum orBAL, CD3 expression within these gates was used for deter-mination of CD4 and CD8 positive lymphocytes. Results ofthe analyses for BAL are presented as percentages of vitallymphocytes within the gated lymphocyte compartment.

Since CD3 positive lymphocytes in sputum were morespread over the forward/sideward scatter plot than in BAL,CD4 and CD8 positive lymphocytes were expressed aspercentages of the number of gated CD3 positive lympho-cytes. CD4 and CD8 percentages in sputum and BAL werecorrected for non-specific antibody binding. T lymphocytesubsets were also evaluated by calculating their CD4/CD8ratio.

Statistics

Before analysis, PC20 values for methacholine were loga-rithmically transformed. Since the cell counts in bronchialbiopsies were not statistically different between the threelocalizations (ANOVA P > O . O 6 ) , cell counts from the threebiopsy localizations were pooled and the mean was calcu-lated. Part of the cell counts showed a skewed distributionwhile the rest of the data were normally distributed. There-fore, all data on cell counts are presented as median (range).For presentation in figures, eosinophil percentages andnumbers were logarithmically transformed because of non-normal distribution. However, since eosinophil countsincluded some zero values, one was added to all valuesbefore logtransfonnation, Between-group differences weredetected using the non-parametric Mann-Whitney t/-test.Non-parametric statistical tests were employed to test forany correlation (Spearman's rank correlation coefficient,Rs). Differential sputum cell counts (expressed as percen-tage non-squamous cells) were used in the comparison withdifferential cell counts in BW and BAL, while absolute cellnumbers in sputum (expressed as cells per mL) were used inthe comparison of inflammatory cell counts between inducedsputum and bronchial biopsies, A probability value of lessthan 0.05 was considered significant. All analyses wereperformed using the Statistical Package of the SocialSciences (SPSS-PC -t-).

Results

All subjects completed the study. Sputum was inducedsuccessfully in all subjects. All sputum samples could beanalysed for differential cell counts since the highest per-centage of squamous epithelial cells was 69.7. In onesubject (No. 18) total cell count was not noted and absolutecell counts could therefore not be calculated. Bronchoscopywas well tolerated by all the subjects. In one subject (No. 6),the BW was contaminated with erythrocytes and this samplewas not included in the analysis. Flow cytometric data ofsputum were not obtained in three cases (No. 9, 10 and 18)because their sputum did not contain enough cells to allowanalysis with flow cytometry.

Differential cell counts in induced sputum, BW and BALare shown in Table 2, absolute cell counts in sputum in



Da•a

3•ac

•a-a

<DaC3

IIo3

T3C

•a'3ili

iti

fg3

ti

§•8

a

a

00,0

(N c s >r)o rJ cN

o r ^̂tN -H 00

00,0,

O -H ,_^r ) -H 00

S S- 2-O 00 CN

ON OO t ^r^ OO OSI I I

VO in ON

^ ^ ON ^ ^Ci w wr-i P-. 00ON Tf MD• * ( ^ 00

i77

i n CN enI I I

(N -̂ i- q

ON

ICN

00 ^ .^ r-H

^-i CN CNI I I

CN CN 0C5

ci 2- S00 CN OO^ - ^ 00

°? P odCN O t ^— CN I

I I O

CN r-; rni n i n CN

Da oa

uI =— O

r8-2

I 2 Ic

c «.3 S•a c2 uE -=

. _ GO ^—V

^ O O

> £ £

Table 3, flow cytometric data in Table 4 and cell counts inbronchial biopsies in Table 5.

The subjects using 182-agonists only and steroid-usingsubjects were pooled in the analysis, because they were notsignificantly different with respect to age, FEVj, PC20FEV,to methacholine, and reversibility in FEVj following salbu-tamol. Furthermore, no significant differences were foundbetween the two subgroups concerning differential cellcounts in sputum, BW and BAL, except for CD8 positivelymphocytes in BAL, as determined by flow cytometry.BAL fiuid from subjects using 182-agonists only containedless CDS"*" lymphocytes as compared with subjects on inhaledsteroids {P = 0.045). Considering the absolute cell counts insputum, induced sputum from subjects using /32-agonistsonly contained more macrophages/mL sputum, as comparedwith induced sputum from subjects on inhaled steroids(P = 0.04). In bronchial biopsies there was a significantdifference between the two subgroups in the number of mastcells (P = 0.04), with higher numbers of mast cells insubjects using j32-agonists only.

Sputum vs bronchial wash

For eosinophils there was a significant correlation betweentheir percentage in sputum and in B W (R^ = 0.52, P = 0.03)(Fig. la). Furthermore, there was a trend towards such acorrelation for lymphocytes (i?s = 0.46, P = 0.07). Therewere no correlations between induced sputum and bronchialwash for neutrophils, macrophages or cylindric epithelialcells percentages.

Sputum vs bronchoalveolar lavage

There was a significant correlation for the percentage ofeosinophils between sputum and BAL {R^ = 0.55, P = 0.02)(Fig. lb). Furthermore, CD4+ lymphocytes, as determinedby flow cytometry, correlated positively between inducedsputum and BAL {R^ = 0.55, P = 0.03) (Fig. 2). In addition,there was a trend towards a correlation between sputum andBAL for CD8+ lymphocytes (/?s = 0.48, P = 0.08) and forthe CD4/CD8 ratio {R^-Q.Al, / ' = 0.09). The percentageneutrophils, macrophages or cylindric epithelial cells insputum did not correlate with those in BAL.

Sputum vs bronchial biopsies

There was a trend towards a correlation between the numberof eosinophils/mL sputum and the number of EG2"'' eosino-phils/mm^ lamina propria in bronchial biopsies {R^ = 0.44,P = 0.07) (Fig. lc). Lymphocyte, neutrophil and mast cellnumbers in induced sputum did not correlate with cellnumbers in broncbial biopsies.



Table 3. Absolute cell numbers x 10'*/mL of sputum*

Squamous cells

14.1 (1.8-65.3)

Lymphocytes

2.5(0.5-16.6)

Neutrophils

19.6 (3.8-202.2)

Alveolarmacrophages

37.9(6.1-181.6)

Eosinophils

1.6(0-13.6)

Cylindricepithelial cells

5.4(1.4-63.9)

Mast cells

0.02(0-1.3)

* Values are expressed as median (range).

Discussion

Tbis study bas sbown tbat for tbe percentage of eosinopbilstbere is a correlation between bypertonic saline-inducedsputum on tbe one band and BW and BAL on tbe otber inpatients witb astbma. In addition, tbere was a trend towardssucb a correlation between tbe absolute numbers of eosino-pbils per mL sputum and per mm^ lamina propria in bron-cbial biopsy specimens. Finally, we bave sbown a correlationfor 004"*" lympbocytes between sputum and BAL. Tbesefindings indicate tbat eosinopbilic and lympbocytic inflam-mation in bypertonic saline-induced sputum modestly reflectssucb inflammation in tbe intrapulmonary airways. Tbis sug-gests tbat induced sputum, is a non-invasive alternative forbroncboscopy in studying intraluminal airway inflammationin patients witb astbma.

Tbe inflammatory cell differentials in sputum, BW andBAL, and tbe flow cytometric data in sputum and BALpresented in tbis study seem to be in accordance witb tbosein otber studies [7,15,17,28-30]. Comparison of tbe inflam-matory infiltrate in broncbial biopsies between our andprevious studies, bowever, reveals tbat we bave found rela-tively bigb inflammatory cell numbers in tbe lamina propria[4,5,31,32]. A plausible explanation for tbis discrepancymigbt be our usage of 6 /xm cryostat sections [27] instead of1 ixm plastic embedded sections [5,31], wbicb allows morecell layers to be counted.

Tbis is one of tbe first studies in atopic astbmatic subjectson tbe comparison of cell differentials between induced

sputum and broncbial biopsies. We bave demonstrated atrend towards a relationsbip between eosinopbils numbers insputum and £02"*" cells in broncbial biopsies in a group ofastbmatic subjects irrespective of tbe treatment level. Tbisfinding is in close agreement witb a recent study byMaestrelli et al. sbowing a non-significant trend towards acorrelation between induced sputum and broncbial biopsiesfor eosinopbil counts in a group of astbmatics consisting ofatopic, intrinsic and isocyanate allergic subjects [20]. Hence,induced sputum appears only to be a weak reflection ofeosinopbilic inflammation in tbe broncbial walls of patientswitb astbma.

Cell differentials obtained from sputum, BW and BALbave been compared by several groups [16-20]. Tbe corre-lations for tbe percentage of eosinopbils between sputumand BW, and between sputum and BAL sbown bere, areconfirming and extending earlier publisbed results [16,18,20].We bave demonstrated tbat tbe correlations for eosinopbilsexist in astbmatics covering a wide range in disease severity.In addition, we bave found a correlation between sputumand BAL for CD4~'' lympbocytes, and weak correlations forCD8'*' lympbocytes and tbe CD4/CD8 ratio, and betweensputum and BW for lympbocytes. Tbese relationsbips migbtreacb significance wben tbe group would bave been enlarged.Our findings are not confirming tbose by Faby et al. [17],wbo observed a closer resemblance between sputum andBW as compared to sputum and BAL in 10 astbmatic sub-jects. Particularly wben looking at CD4''" lympbocytes, as

Table 4. Percentage CD3-, CD4- and CD8-positive lymphocytes in sputum and BAL asmeasured by flow cytometry*

Material CD3 CD4 CD8 CD4/CD8 ratio

SputumBAL

398 (89-895)t8L5 (65.0-95.0)

57.0 (5.3-74.1)49.0 (28.0-61.0)

23.2 (5.7-67.4)32.1 (18.4-57.0)

1.7 (0.4-3.7)1.5 (0.5-3.2)

* Values are expressed as median (range) of gated lymphocytes.t Number of gated CD3 positive lymphocytes used to quantitate CD4 and CD8 positivelymphocytes. CD4 and CD8 positive lymphocytes in sputum are expressed as apercentage of the number of CD3 positive lymphocytes.



+o

D.Oc

+00QU•a

sU

E

+COQU

QO

ft

2 i•Si

U U

•aoc

Cl.

o

I

ooQO

QU

O

eno

o

o

o

00

2

T3

CDc

aoU)o111

10

(a)

10 1 0 0

%Eosinophils in sputum

10

mc

en

(b)

A

A

c

N

E

Q)U

+CMCDID

10OO

10O

1 0

1

1 10

%Eosinophils in sputum

(c)

A A A

100

1 10

^ #Eosinophils in sputum

Fig. 1. Rank correlation between the percentage of eosinophils insputum and BW (a), between the percentage of eosinophils insputum and BAL (b), and between the number of eosinophils/mLof sputum and the number of EG2 '*' cells/mm^ lamina propria inbronchial biopsies (c) in our group of asthmatic subjects. (A)asthmatic subjects using inhaled short-acting /32-agonists ondemand only; (A) asthmatic subjects on regular inhaled steroids.Since eosinophils in sputum, BW, BAL and bronchial biopsiesshowed a skewed distribution, cell percentages and numbers arepresented on logarithmic scale, (a) /?s = 0.52; ^ = 0.03; (b)R, = 0.55; P = 0.02; (c) R, = 0.44; P = 0.07.



CD 70

c

tes

o0a

4-

QU

60

50

40

30

2O

10

O 10 20 30 40 50 60 70 80

%CD4'' lymphocytes in sputum

Fig. 2. Rank correlation between the percentage of vital CD4'''lymphocytes in sputum and BAL (R^ = 0.55, P = 0.03). (A) asth-matic subjects using inhaled short-acting j32-agonists on demandonly; (A) asthmatic subjects on regular inhaled steroids.

measured by flow cytometry, our data might be indicative ofthe reverse. This might be remarkable, since BAL fluid isconsidered to represent both the airway and alveolar com-partments. However, we believe that the anatomical inter-pretation of these relationships has to await clarification ofthe sites of origin of inflammatory cells in induced sputum,which at present is still unknown.

Since this study was designed to compare three methodsto investigate airways inflammation, we tried to eliminateall sources of potential errors involved in subject selection,study design, the method of sputum induction and process-ing, the methodology of bronchoscopy, cell counting pro-cedures and the statistical analysis. We have selectedasthmatic subjects following strict inclusion criteria in orderto exclude any other diseases than asthma. All participatingsubjects were atopie, non-smoking adults with hyperrespon-siveness to inhaled methacholine and with reversible air-ways obstruction. This group of subjects had a wide range inlung function, airway hyperresponsiveness and anti-asthmamedication usage and, thus, yielded a representative cross-section of an asthmatic population. We believe this widerange of asthma severity was needed in our study, becausewe aimed to investigate the relationship between inflamma-tory cell differentials in sputum, lavage fluid and bronchialbiopsy specimens in patients in whom monitoring of sputumcells can potentially improve their clinical management.

Induction of sputum by inhalation of aerosolized hyper-tonic saline is a frequently used method to obtain intrapul-monary epithelial lining fluid [9-20,23,28,33]. The used'whole sample' method has been shown to provide repeat-able measurements of cellular, as well as soluble markers inboth mild and severe asthmatic patients [23]. And even

though the whole sample method results in variable salivarycontamination, it enabled us to examine all cells in theobtained sample of airway lining fluid, and allowed storingof sputum supematants. Differential cell counts made oncytospin preparations of sputum were performed accordingto recommendations for differentiation of BAL, i.e. count-ing 500 inflammatory cells [34]. However, we includedcolumnar epithelial cells in the cell differentials, since theamount of ciliated epithelial cells in BAL is inverselyrelated to methacholine hyperresponsiveness and therebypossibly with disease severity [35,36]. Differential cell countswere expressed as percentage of non-squamous cells, inorder to be able to compare cell differentials betweensubjects with variable salivary contamination. The samestandardization of inflammatory cell counts was used forBW and BAL cytospins, which justifies the comparison ofcell differentials between sputum, BW and BAL.

Finally, it is usual to express inflammatory cells in inducedsputum as percentage of total (non-squamous) cells. How-ever, for comparison of inflammatory cell counts in sputumwith those in bronchial biopsies, we chose to expressinflammatory cells in induced sputum as number of cells/mL sputum as opposed to percentage in sputum [20], since itcan be argued that it is not appropriate to correlate relativecell differentials in sputum with absolute cell counts inbronchial biopsies. Earlier studies on the comparison ofinflammatory cells in BAL with those in bronchial biopsieshave also used this method to analyse correlations [37].

How can the present findings be interpreted? Transepi-thelial migration of inflammatory cells from the airway tissueinto the lumen is a highly speciflc process, requiring multiplepathways of cellular adherence and activation [38,39]. There-fore, in analogy with the comparison of inflammatory cellsbetween BAL and biopsies [1,30,37,39], it is not unexpectedthat the leucocyte counts in sputum do not fully representthose in bronchial biopsies. In this view, it is quiteremarkable that we did find a relationship between thesetwo compartment for eosinophils. Apparently, transen-dothelial cell trafflc from blood into the airway tissue isoften accompanied with transepithelial migration in asthma.

What are the clinical implications of this study? Assum-ing that inflammatory cell counts in bronchial biopsies canbe considered as the present gold standard of measuringairway pathology in asthma [1], our flndings indicate thatinduced sputum only modestly reflects mucosal inflamma-tion in bronchial biopsies. However, induced sputum doesseem to be a suitable, non-invasive technique providing arough estimate of intraluminal inflammation in asthma. Thissuggests that cellular markers in induced sputum can beused in monitoring patients with asthma. However, suchapproach needs to be established in follow-up studies on thecomparison of induced sputum with bronchial biopsies,BAL or BW in prospective, long-term trials, in well-deflned



subgroups of patients with varying disease severity anddifferent treatment levels.

Acknowledgements

The authors would like to thank the Commission of theEuropean Communities for their financial support; RatkoDjukanovic, MD, PhD for providing us with the AAl mono-clonal antibody; Christine Evertse, Nizet Dijkstra, Maartenvan der Keur and Arie van der Marel for their technicalassistance.

References

1 Djukanovic R, Roche WR, Wilson JW et al. Mucosal inflam-mation in asthma. Am Rev Respir Dis 1990; 42:434-57.

2 Jeffery PK, Wardlaw AJ, Nelson FC, Collins JV, Kay AB.Bronchial biopsies in asthma. An ultrastructural, quantitativestudy and correlation with hyperreactivity. Am Rev Respir Dis1989; 140:1745-53.

3 Poston RN, Chanez P, Lacoste JY, Lee TH, Bousquet J.Immunohistochemical characterization of the cellular infiltra-tion in asthmatic bronchi. Am Rev Respir Dis 1992; 145:918-21.

4 OUerenshaw SL, Woolcock AJ. Characteristics of the inflam-mation in biopsies from large airways of subjects with asthmaand subjects with chronic airflow limitation. Am Rev RespirDis 1992; 145:922-27.

5 Laitinen LA, Laitinen A, Haahtela T. Airway mucosal inflam-mation even in patients with newly diagnosed asthma. Am RevRespir Dis 1993; 147:697-704.

6 Azzawi M, Bradley B, Jeffery PK et al. Identification ofactivated T lymphocytes and eosinophils in bronchial biopsiesin stable atopic asthma. Am Rev Respir Dis 1990; 142:1407-13.

7 Adelroth E, Rosenhall L, Johansson SA, Linden M, Venge P.Inflammatory cells and eosinophilic activity in asthmaticsinvestigated by bronchoalveolar lavage. The effects of anti-asthmatic treatment with budesonide or terbutaline. Am RevRespir Dis 1990; 142:91-9.

8 Wilson JW, Djukanovic R, Howarth PH, Holgate ST. Inhaledbeclomethasone dipropionate downregulates airway lympho-cyte activation in atopic asthma. Am J Respir Crit Care Med1994; 149:86-90.

9 Pin I, Gibson PG, Kolendowicz R et al. Use of induced sputumcell counts to investigate airway inflammation in asthma.Thorax 1992; 47:25-9.

10 Iredale MJ, Wanklyn SAR, Phillips IP, Krausz T, Ind PW.Non-invasive assessment of bronchial inflammation in asthma:no correlation between eosinophilia of induced sputum andbronchial responsiveness to inhaled hypertonic saline. ClinExp Allergy 1994; 24:940-5.

11 Fahy JV, Liu J, Wong H, Boushey HA. Cellular and biochemi-cal analysis of induced sputum from asthmatic and fromhealthy subjects. Am Rev Respir Dis 1993; 147:1126-31.

12 Gibson PG, Girgis-Gabardo A, Morris MM et al. Cellular

characteristics of sputum from patients with asthma andchronic bronchitis. Thorax 1989; 44:693-9.

13 Turner MO, Hussack P, Sears MR, Dolovich J, Hargreave FE.Exacerbations of asthma without sputum eosinophilia. Thorax1995; 50:1057-61.

14 Fahy JV, Liu J, Wong H, Boushey HA. Analysis of cellular andbiochemical constituents of induced sputum after allergenchallenge: A method for studying allergic inflammation. JAllergy Clin Immunol 1994; 93:1031-9.

15 Claman DM, Boushey HA, Liu J, Wong H, Fahy JV. Analysisof induced sputum to examine the effects of prednisone onairway inflammation in asthmatic subjects. J Allergy ClinImmunol 1994; 94:861-9.

16 Spanevello A, Migliori GB, Giudici MR et al. Sputum inducedand bronchoalveolar lavage in asthmatics: preliminary resultson their relationship. Eur Respir J 1994; 7 (Suppl. 18):3 s.

17 Fahy JV, Wong HH, Liu J, Boushey HA. Comparison ofsamples selected by sputum induction and bronchoscopyfrom asthmatic and healthy subjects. Am J Respir Crit CareMed 1995; 152:53-8.

18 Keatings VM, Evans DJ, O'Connor BJ, Barnes PJ. Cellularprofiles in asthmatic airways: a comparison of induced sputum,bronchial washings, and bronchoalveolar lavage fluid.. Thorax1997; 52:372-4.

19 Kidney J, Pizzichini E, Adelroth E et al. Comparison ofsputum, bronchoalveolar lavage and blood inflammatory cellsin asthma. Am J Respir Crit Care Med 1995; 151:A384.

20 Maestrelli P, Saetta M, Stefano AD et al. Comparison ofleukocyte counts in sputum, bronchial biopsies, and broncho-alveolar lavage. Am J Respir Crit Care Med 1995; 152:1926-31.

21 Quanjer PhH, Tammeling GJ, Cotes Journal of Ecology et al.Lung volumes and forced ventilatory flows. Eur Respir J 1993;6 (Suppl. 16):5-40.

22 Sterk PJ, Fabbd LM, Quanjer PhH et al. Airway responsive-ness. Standardized challenge testing with pharmacological,physical and sensitizing stimuli in adults. Eur Respir J 1993;6 (Suppl. 16):53-83.

23 In'tVeen JCCM, de Gouw HWFM, Smits HH et al. Repeat-ability of cellular and soluble markers of inflammation ininduced sputum of patients with asthma. Eur Respir J, 1996;9:2441-7.

24 Leigh TR, Nazir T, Wiggins J, Ganderton D, Collins JV.Performance characterisation of the DeVilbiss Ultraneb 99ultrasonic nebuliser. Thorax 1990; 45:P817.

25 Djukanovic R, Wilson JW, Lai CKW, Holgate ST, HowarthPH. The safety aspects of fiberoptic bronchoscopy, broncho-alveolar lavage, and endobronchial biopsy in asthma. Am RevRespir Dis 1991; 143:772-7.

26 Workshop summary and guidelines. Investigative use of bron-choscopy, lavage and bronchial biopsies in asthma and otherairways diseases. Eur Respir J 1992; 5:115-21.

27 Sont JK, van Krieken JHJM, Evertse CE et al. Relationshipbetween the inflammatory infiltrate in bronchial biopsy speci-mens and clinical severity of asthma in patients treated withinhaled steroids. Thorax 1996; 51:496-502.

28 Kidney JC, Wong AG, Efthimiadis A et al. Elevated B cells in


Comparison of infiammatory cell counts in asthma 779

sputum of asthmatics. Close correlation with eosinophils. Am JRespir Crit Care Med 1996; 153:540-4.

29 Walker C, Bode E, Boer L, Hansel TT, Blaser K, Virchow JC.Allergic and nonallergic asthmatics have distinct patterns ofT-cell activation and cytokine production in peripheral bloodand bronchoalveolar lavage. Am Rev Respir Dis 1992;146:109-15.

30 Lacoste JY, Bousquet J, Chanez P et al. Eosinophilic andneutrophilic inflammation in asthma, chronic bronchitis, andchronic obstructive pulmonary disease. J Allergy Clin Immunol1993; 92:537-48.

31 Djukanovic R, Wilson JW, Britten KM, Wilson SJ, Walls AF,Roche WR, Howarth PH, Holgate ST. Effect of an inhaledcorticosteroid on airway inflammation and symptoms inasthma. Am Rev Respir Dis 1992; 145:669-74.

32 Tdgg CJ, Manolitsas ND, Wang J et al. Placebo-controlledimmunopathologic study of four months of inhaled cortico-steroids in asthma. Am J Respir Crit Care Med 1994; 150:17-22.

33 Pizzichini E, Pizzichini MMM, Efthimiadis A et al. Indices ofairway inflammation in induced sputum: reproducibility andvalidity of cell and fluid-phase measurements. Am J Respir CritCare Med 1996; 154:308-17.

34 American Thoracic Society. Clinical role of bronchoalveolarlavage in adults with pulmonary disease. Am Rev Respir Dis1991; 142:481-6.

35 Wardlaw AJ, Dunnette S, Gleich GJ, Collins JV, Kay AB.Eosinophils and mast cells in bronchoalveolar lavage in sub-jects with mild asthma. Relationship to bronchial hyperreac-tivity. Am Rev Respir Dis 1988; 137:62-9.

36 Beasley R, Roche WR, Roberts JA, Holgate ST. Cellular eventsin the bronchi in mild asthma and after bronchial provocation.Am Rev Respir Dis 1989; 139:806-17.

37 Foresi A, Bertorelli G, Pesci A, Chetta A, Olivieri D. Inflam-matory markers in bronchoalveolar lavage and in bronchialbiopsy in asthma during remission. Chest 1990; 98:528-35.

38 Hansel TT, Walker C. The migration of eosinophils into thesputum of asthmatics: the role of adhesion molecules. Clin ExpAllergy 1992; 22:345-56.

39 Kroegel C, Liu MC, Hubbard WC, Lichtenstein LM, BochnerBS. Blood and bronchoalveolar eosinophils in allergic subjectsafter segmental antigen challenge: Surface phenotype, densityheterogeneity, and prostanoid production. J Allergy ClinImmunol 1994; 93:725-34.


Comparison of inflammatory cell counts in asthma: induced sputum ...

Documents