J Clin Pathol 1997;50:981-984

Papers

Pulmonary alveolar proteinosis: diagnosis usingroutinely processed smears of bronchoalveolarlavage fluid

Tetuo Mikami, Yoshiko Yamamoto, Masaru Yokoyama, Isao Okayasu

AbstractAims-For the diagnosis of pulmonaryalveolar proteinosis from bronchoalveolarlavage specimens it is normally necessaryto make an ultrastructural examination.However, this is thought to be impracticalfor bronchoalveolar lavage specimens thathave been routinely fixed in ethanol. In thepresent study, bronchoalveolar lavage cy-tology smears on slide glasses were exam-ined directly ultrastructurally to make adiagnosis ofpulmonary alveolar proteino-sis.Methods-Bronchoalveolar lavage smearsfrom three pulmonary alveolar proteino-sis patients were stained with Papanico-laou and periodic acid-Schiff (PAS) foridentification of amorphous globularstructures. Subsequently, they were re-fixed with glutaraldehyde and osmiumtetroxide, and embedded in epoxy resin.Ultrathin sections were cut and examinedultrastructurally.Results-Papanicolaou stained specimensfrom pulmonary alveolar proteinosis pa-tients contained scattered amorphous orgranular globules, 20-50 gm in diameter,which were PAS positive. Ultrastructuralexamination ofthe globules revealed mul-tilamellated structures, characteristic ofpulmonary alveolar proteinosis, in allcases.Conclusions-In general, it is thought thatthe morphological diagnosis of pulmo-nary alveolar proteinosis from bronchoal-veolar lavage specimens requires bothcytological and ultrastructural examin-ation. However, the amorphous globulesevident on cytology smears proved to con-tain multilamellated structures so thatthey can themselves be used as diagnosticevidence.( Clin Pathol 1997;50:981-984)

Keywords: pulmonary alveolar proteinosis; bronchoal-veolar lavage; multilamellated structure

Pulmonary alveolar proteinosis is a rare diseasefirst described by Rosen et al in 1958.1 It is

characterised by deposition within the airspaces of a granular extracellular materialcomposed ofproteins and lipids that is periodicacid-Schiff (PAS) positive and diastaseresistant.' 3 Its diagnosis is based mainly onhistological findings obtained with transbron-chial lung biopsy or open lung biopsy speci-mens. However, transbronchial lung biopsyspecimens are often so small that they do notprovide specific evidence. Because the charac-teristic intra-alveolar material is localisedmainly in the pulmonary hilar region,2 openlung biopsy may also fail to ensure diagnosis.Bronchoalveolar lavage can sample a muchwider area and is safer for the patient thantransbronchial lung biopsy and/or open lungbiopsy.Pulmonary alveolar proteinosis cytology

findings from bronchoalveolar lavage speci-mens have been reported previously,"7 butresults of Papanicolaou staining are not consid-ered sufficient for a definitive diagnosis and it isgenerally thought necessary to confirm thepresence of characteristic multilamellatedstructures under the electron microscope.4'6 7However, it is not practical to process allroutine bronchoalveolar lavage specimens forultrastructural examination. Therefore, in thepresent study we directly compared the ultra-structure of cytology materials prepared ini-tially for routine Papanicolaou smears. Ourpurpose was to prove that the globules seenunder the light microscope have characteristicelectron microscopic features, and thus toassess the possibility of pulmonary alveolarproteinosis diagnosis from routinely processedbronchoalveolar lavage smears alone.

MethodsThree cases of pulmonary alveolar proteinosis(including idiopathic and secondary types)were collected by reviewing the pathologyreports of Kitasato University Hospital and theTokyo Medical and Dental University Hospi-tal. In all three cases, transbronchial lungbiopsy was performed to prove the presence ofPAS positive and diastase digestion resistantintra-alveolar material. Clinical information onthe cases is summarised in table 1.

Department ofPathology, School ofMedicine, KitasatoUniversity,Sagamihara,Kanagawa, JapanT MikamiI Okayasu

Department ofPathology, TokyoMedical and DentalUniversity Hospital,Tokyo, JapanY Yamamoto

Cytology Section,Department ofPathology, KitasatoUniversity HospitalM Yokoyama

Correspondence to:Dr Mikami, Departnent ofPathology, School ofMedicine, KitasatoUniversity, 1-15-1, Kitasato,Sagamihara, Kanagawa, 228,Japan.

Accepted for publication10 October 1997

981

on July 23, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.50.12.981 on 1 D

ecember 1997. D

ownloaded from

Mikami, Yamamoto, Yokoyama, Okayasu

Table 1 Clinical details of the three patients

Case nol Age Final diagnosticsex (years) Symptomsl signs Background disease procedure

1/F 47 Dyspnoea Chronic myelogenous Transbronchial lungleukaemia biopsy

2/F 50 Fever, coughing None (idiopathic) Transbronchial lungbiopsy

3/M 33 Abnormal shadow None (idiopathic) Transbronchial lungon chest x ray biopsy

Smears of bronchoalveolar lavage fluid sedi-ment from the patients were fixed routinelywith 95% ethanol and subjected to Papanico-laou staining. After examination under the lightmicroscope, coverslips were removed by rins-ing in xylene. Following decoloration byrehydration, PAS staining was carried out. Theslides were examined microscopically and usedfor electron microscopy as described below.The smears were fixed with 2.5% glutaralde-

hyde for 30 minutes, rehydrated with gradedconcentrations of ethanol, and postfixed with1% osmium tetroxide for one hour. The smearswere then dehydrated in graded ethanols up to100% and embedded by inverting a gelatincapsule of fresh epoxy resin over each slide.The epoxy resin was polymerised by overnightincubation at 37°C and a further incubation at600C for 24-36 hours. The epoxy resin blockswere removed from the slides by heating andultrathin sections were cut. These were stainedwith 3.5% uranyl acetate and Reynolds' leadcitrate and examined with a transmission elec-tron microscope (H-600; Hitachi, Tokyo,Japan).To serve as controls, Papanicolaou stained

routine cytology smears from bronchoalveolarlavage fluid samples of 31 patients without pul-

A'.

II.

C

monary alveolar proteinosis (initially diag-nosed as "no malignant cells") were reviewed.These control patients consisted of20 men and11 women aged between 39 and 82 years old.The clinical diagnoses before bronchoalveolarlavage were as follows: 15 cases of suspectedlung tumour, three cases of suspected pulmo-nary haemorrhage, three cases of pneumonia,two cases of atelectasis, two cases of bron-chiectasis, two cases of suspected tuberculosis,one case of interstitial pneumonia, one case ofatypical mycobacterial infection, one case ofsuspected eosinophilic pneumonia, and onecase of consolidation in chest radiography.

ResultsCYTOLOGY FINDINGSThe cytological features of the bronchoalveolarlavage fluid were similar in all three cases. Thesmears contained globules with amorphous orgranular structures, between 20 and 50 ,um indiameter, stained green. Some globules werestained green but had an orange zone in thecentre. They were found as solitary elements inthe smears, with approximately 5-10 per slide.Finely granular or amorphous material wasapparent in the background along with scat-tered neutrophils and lymphocytes. The glob-ules and the amorphous material were PASpositive (fig 1). Focal clusters of alveolar mac-rophages with foamy or vacuolated cytoplasmwere noted.

In the 31 control bronchoalveolar lavagespecimens, these globules were not detected.Occasionally, degenerated squamous cells ofthe same size were observed, at 0-2 per slide,but these could be differentiated by differences

B **)

D

Figure 1 (A) A typical globule detected in a Papanicolaou stained bronchoalveolar lavage smear of case 2; (B) the sameglobule stained with periodic acid-Schiff (PAS). (C) Another globule stained with Papanicolaou found in case 3; (D) thesame globule stained with PAS; note the PAS positive amorphous material in the background.

982

AP7654ib ..

AL., Akk...

it`7

on July 23, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.50.12.981 on 1 D

ecember 1997. D

ownloaded from

Pulmonary alveolar proteinosis diagnosis from routine smears

Figure 2 Ultrastructuralfindings for a globule found in a routine cytology bronchoalveolarlavage smear. The arrow indicates a type B multilamellated structure forming regularlyrepeated concentric lamellar layers. Its periodicity is in the range of 4-5 nm. The arrowheadindicates a type D multilamellated structure consisting of electron dense double layersalternating with wider electron lucent layers. The double layers measured 8-10 nm in widthand the wider layers were 20-30 nm. (Original magnification, x60 000; bar = 200 nm.)

in thickness, the presence of a degeneratednucleus, and the lack of granular structures.

ULTRASTRUCTURAL FINDINGSSeven globules detected by cytological examin-ation could be examined ultrastructurally.They were all found to consist of collections ofround or oval shaped electron lucent struc-tures, approximately 200-600 nm in diameter.Although their finer structure was almostdestroyed, parts were made up of multilamel-lated structures consisting of electron densedouble layers alternating with wider electronlucent layers. In other areas, concentric multi-lamellated structures with a periodicity in the4-5 nm range were found (fig 2). These twodifferent multilamellated structures were com-patible with the type D and type B multilamel-lated structures described by Takemura et al,8respectively. Type D multilamellated structureswere detected in cases 1 and 3 while type Bstructures were detected in cases 1 and 2.

DiscussionIn bronchoalveolar lavage specimens, amor-phous or granular globules stained orange orgreen by Papanicolaou staining have beendescribed to be characteristic of pulmonary

alveolar proteinosis.4 Although ultrastructuralfindings have been described in several papers,bronchoalveolar lavage sediments fixed withglutaraldehyde or bronchoalveolar lavage cellblocks were used.9 Therefore, the globulesseen in Papanicolaou stained bronchoalveolarlavage smears have hitherto not been linkeddirectly with ultrastructural findings. Further-more, because in most hospitals cytologymaterials are fixed routinely with ethanol assmears, ultrastructural examination for diagno-sis is not practical.

In this study, however, globules themselvescould be examined ultrastructurally, and al-though most of the ultrastructural featureswere destroyed as a result of poor alcohol fixa-tion, it was confirmed that the globulescomprised collections of multilamellated struc-tures. Takemura et al examined the ultrastruc-tural findings of bronchoalveolar lavage sedi-ments from pulmonary alveolar proteinosispatients that had first been fixed with glutaral-dehyde and divided them into four types ofmultilamellated structures, from A to D.8According to their description, type A (themajor component) consisted of concentric tri-laminar structures that comprised two electrondense layers and a central lucent layer (5.7-7.5 nm in overall width) alternating with wider(25-30 nm) electron lucent layers. Type Bwere formed by concentric lamellae with a5.0-5.3 nm periodicity. Type C were com-posed of wavy, electron dense lamellae with a4.0-4.5 nm periodicity. Type D were conglom-erated masses of intricately arranged double ortriple electron dense layers (7.5-13.5 nm wide)alternating with wider (30-40 nm) electronlucent layers. Comparison with their detailedfindings for the four types allowed identifica-tion of types B and D in the globules of thebronchoalveolar lavage fluid smears in thisstudy. In other reports using bronchoalveolarlavage sediments or cell blocks, the structureswere not categorised into the four types, butfrom the photographs we can identify the typeA multilamellated structure4-7 9 and the type Dstructure.9

In this study, the typical type A multilamel-lated structure, which is normally the majorcomponent, was not detected. It is possible thatit is destroyed by routine alcohol fixation andthat the electron lucent round or oval struc-tures might be equivalent to type A structures.However, because their inner structures weredestroyed, there is no definitive proof. On theother hand, the type B structure was well pre-served and the type D multilamellated struc-ture was partly preserved, despite the alcoholfixation. Although the routine alcohol fixationhad modified the ultrastructure, it was con-firmed that the globules seen in Papanicolaoustained smears comprise multilamellated struc-tures at the electron microscope level.The important differential diagnosis of

pulmonary alveolar proteinosis is from Pneu-mocystis carinii pneumonia. In P carinii pneu-monia the bronchoalveolar lavage smear con-tains characteristic findings described as a"foamy mass" or "foamy exudate".'0 " Wethink that the globules of pulmonary alveolar

983

on July 23, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.50.12.981 on 1 D

ecember 1997. D

ownloaded from

Mikamii, Yamamoto, Yokoyamiia, Okayasu

proteinosis can be differentiated from this foamymass by Papanicolaou staining. When the differ-entiation is difficult, PAS staining and Grocottmethenamine silver staining are useful.

In conclusion, confirmation that the specificPAS positive globular structures found in rou-

tinely processed bronchoalveolar lavage fluidsmears comprise multilamellated structuresindicates that they themselves can be useddiagnostically as evidence for pulmonary alveo-lar proteinosis.

We are grateful to the members of the Electron MicroscopeLaboratory Center of Kitasato University, School of Medicinefor preparing electron micrographs.

1 Rosen SH, Castleman B, Liebow AA. Pulmonary alveolarproteinosis. N EnglJ3 Med 1958;258: 1123-42.

2 Dale DH, Hammer SP. Pulmonary pathology. 2nd edn.Berlin: Springer Verlag, 1994:745-51.

3 Hasleton PS. Spencer's pathology of the lung. 5th edn. NewYork: McGraw-Hill, 1996:773-5.

4 Cardillo MR. Pulmonary alveolar proteinosis-a cytomor-phological histochemical and ultrastructural study of onecase. Arch Anat Cytol Path 1989;37:259-61.

5 Cordonnier C, Fleury-Feith J, Escudier E, Atassi K,Bernaudin JF. Secondary alveolar proteinosis is a reversiblecause of respiratory failure in leukemic patients. Am _7Respir Crit Care Med 1994;149:788-94.

6 Mermolja M, Rott T, Deeljak A. Cytology of bronchoalveo-lar lavage in some rare pulmonary disorders: pulmonaryalveolar proteinosis and amiodarone pulmonary toxicity.Cytopathology 1994;5:9-16.

7 Burkhalter A, Silverman JF, Hopkins MB, Geisinger AK.Bronchoalveolar lavage cytology in pulmonary alveolarproteinosis. Am Clin Pathol 1996;106:504-10.

8 Takemura T, Fukuda Y, Harrison M, Ferrans VJ. Ultra-structural, histochemical, and freeze-fracture evaluation ofmultilamellated structures in human pulmonary alveolarproteinosis. Anm jAnat 1987;179:258-68.

9 Hook GE, Gilmore LB, Talley FA. Multilamellatedstructures from the lungs of patients with pulmonaryalveolar proteinosis. Lab Invest 1984;50:711-25.

10 Greaves TS, Strigle SM. The recognition of Pneumocystiscarinii in routine Papanicolaou-stained smears. Acta Cytol1985;29:714-20.

11 Ghali VS, Garcia RL, Skolom J. Fluorescence of Pneumo-cystis carinii in Papanicolaou smears. Hum Pathol 1984;15:907-9.

984

on July 23, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.50.12.981 on 1 D

ecember 1997. D

ownloaded from