Thorax 1985;40:561-570

Editorial

Pneumocystis carinii pneumoniaThe first description of pneumocysts is credited toChagas' in 1909, who mistook them for trypano-somes in guineapig lungs. Carini reported similarstructures in rat lungs infected with Trypanosomacruzi. Subsequently Dr and Madame Delanoe inParis found pneumocysts in sewer rats withouttrypanosomiasis and proposed the namePneumocystis carinii in honour of the BrazilianCarini.' Although Chagas probably reported thefirst human case of pneumocystosis, its importancewas not recognised until the second world war, whenoutbreaks of pneumonia occurred throughoutEurope in malnourished children in overcrowdedorphanages.23 Histologically, the lungs of suchpatients showed considerable mononuclear cellinterstitial infiltration and foamy vacuolated alveo-lar exudates. The condition was named interstitialplasma cell pneumonia. A definite association withP carinii infection was made in 1952.4 Subsequently,pneumocystis pneumonia was diagnosed in infantswith congenital immunodeficiency states. Epidemicpneumocystis pneumonia has virtually disappearedfrom Europe, but is still recognised in parts of theworld where malnutrition and poverty prevail.' 2 5

Until recently pneumocystis pneumonia largelyaffected patients receiving immunosuppressivetreatment and corticosteroids for cancer, organtransplantation, and other conditions. Reports from1981 onwards, however, detailed the associationbetween pneumocystis pneumonia, other oppor-tunistic infections, and Kaposi's sarcoma in previ-ously well homosexual men and drug abusers.6 Theacquired immune deficiency syndrome (AIDS) hascaused intense interest in the medical and lay press.Recently, 169 cases of AIDS have been recorded inthe United Kingdom, 762 in Europe, and 8495 inthe United States (Communicable Disease Surveil-lance Center, 1985, unpublished), As pneumocystispneumonia is by far the commonest opportunisticinfection (85%) in AIDS, clinicians and micro-

Address for correspondence: Dr JT Macfarlane, City Hospital,Nottingham NG5 1PB.(Reprints will not be available.)

biologists will need to become familiar with thediagnosis and treatment of pneumocystosis.

The organism

The taxonomy of P carinii continues to causedebate. At times it has been considered both as atrypanosome and as a fungus, largely owing to itsability to take up silver stains. It is now generallyheld to be a sporozoan with a life cycle similar tothat of Toxoplasma gondii. This conclusion is sup-ported by its susceptibility to antiparasitic agents.Although there are minor antigenic differences bet-ween strains isolated from various animal sources,they would all appear to belong to a single speciessince the similarities are greater than the differ-ences.The life cycle remains incompletely understood.

Nevertheless detailed morphological studies ofinfected human and animal lungs by light and elec-tron microscopy have increased our knowledge inrecent years.78 The parasite exists in bothtrophozoite and cystic forms within the lung. Thetrophozoites are pleomorphic and 1-4 ,um in diame-ter, staining well with Giemsa. In contrast, the cystsare 5-7 gm in diameter, possess a thick wall, andstain best with methenamine silver stain, whichmakes them readily identifiable in tissues, althoughthey are often outnumbered by the trophozoiteforms at certain stages of the infection. The thickwalled cysts contain up to eight smaller bodies orsporozoites. These are thought to emerge anddevelop into trophozoites, which may or may notundergo binary fission before maturing and develop-ing into cysts to repeat the life cycle. Electron mic-roscopic studies have shown the cyst to undergomaturation with the liberation of trophozoitesbefore collapse.9 The trophozoite possesses a nuc-leus but few cytoplasmic organelles, although tubu-lar extensions of the surface cytoplasm, or filopodia,can be visualised.

Koch's postulates have yet to be fulfilled in rela-tion to pneumocystis pneumonia. Isolation of theparasite from the human lung, first reported by Piferand her colleagues,'" is not routinely possible. Vari-ous cell lines permit isolation of the parasite in tissue

561

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

562

culture. These include Vero, CEL, WI-38, MRC 5,and Chang liver cells.'0-'2 The most successful so farappears to be the use of the Vero cell line with min-imal essential medium and 2% fetal bovine serum.'3The ratio of organisms to cells appears to be critical,as do the conditions of incubation. An atmosphereof 5% oxygen and 10% carbon dioxide favoursgrowth, which is often maximal within a few days.'4There is a trend towards a greater predominance ofcystic forms over the trophozoites with more pro-longed incubation. More recently P carinii has beencultivated in cell lines derived from lung.'5The ready isolation of the parasite from the

human lung would undoubtedly facilitate diagnosisand allow cultivation of large numbers of cysts andtrophozoites. This in turn would lead to a greaterunderstanding of the parasite, its antigenic deter-minants, and factors governing its various stages ofgrowth.

Pathogenesis

P carinii has been found to exist saprophytically inthe lungs of a wide range of animal species, includ-ing man.'6 There is evidence to suggest that it isfrequently acquired in early childhood, where itusually causes inapparent infections, coinciding withloss of maternally acquired antibody.'7 18 Infectionpresumably occurs by inhalation, although in view ofthe reports of congenital infection transplacentalinfection cannot be excluded.'9 It probably remainsin a latent state for many years and undergoes reac-tivation in response to immune suppression,although case clustering has raised the possibility ofhospital acquired infection.20 The main predisposingconditions include lymphatic leukaemia,2' malignantlymphoma, and organ transplantation.2224 Morerecently pneumocystis pneumonia has proved themajor infectious complication among those withAIDS.25 Other, less frequent predisposing causesare collagen vascular disease,26 solid tumours, otherhaematological conditions, and primaryimmunodeficiency states,27 and it occasionallyoccurs in immunocompetent neonates.28 In all thesecircumstances, except in starvation and AIDS, Pcarinii infection is uncommon unless corticosteroidsor cytotoxic agents have been used.26 Of the twoclasses of agents, corticosteroids are the major pre-disposing factor and this is reflected in the variousanimal models of pneumocystis pneumonia.8Among the cytotoxic and immunosuppressive drugsazathioprine, methotrexate, and the vinca alkaloidsare the most frequently identified.26 There is littledoubt that the frequency of pneumocystispneumonia is proportional to the degree ofimmunosuppression.

Models of infection

The principle animal model has been the rat treatedwith corticosteroids. This has allowed the study ofchanges in the lung and growth characteristics of Pcarinii.8 30 Rats treated with cortisone for eightweeks show progressive increase in the intensity ofthe P carinii infection. This is enhanced by a lowprotein diet. Control rats not treated with cortisoneshow only a low level of infection throughout. Thepneumonia occurs by reactivation of latent infectionand there is no experimental animal that can reliablybe infected with exogenous P carinii. Rats treatedwith cortisone for only four weeks show a gradualdecrease in the intensity of infection after steroidsare withdrawn, although organisms were still evi-dent up to 21 weeks later.

Light microscopy shows that in mild infectionorganisms lie in small numbers along alveolar walls.As infection increases, more alveoli become filledwith clumps of organisms, alveolar lining cellshypertrophy, and there is a mild mononuclear cellinterstitial infiltration. Foamy eosinophilic materialappears in the alveolar spaces, which eventuallybecome completely filled. When the steroids arestopped at four weeks, however, there is a dramaticincrease in the interstitial cellular infiltration andalveolar macrophages. Interstitial fibrosis mayappear subsequently.8

Electron microscopy shows that the P cariniitrophozoites lie in tight apposition to type Ipneumocytes and are covered by the liquid alveolarlining layer.8 3" Increased permeability of the alveolarcapillary membrane (as shown by leakage of horse-radish peroxide) is the first change seen in alveolarstructure after infection.32 Later the type I pneumo-cytes degenerate, denuding the alveolar wall, andthe trophozoites come into direct contact with thebasement membrane. Some are found below theepithelium or in the interstitium. Extrapulmonaryspread has been seen rarely in man,33 most com-monly in lymph nodes and spleen. The foamyeosinophilic material in the alveoli probably repres-ents degenerative membranes of the organism andalveolar cells, surfactant, and protein exudate. Theevolution of these alveolar changes explains whydyspnoea is the earliest manifestation ofpneumocystis pneumonia.

Healthy rats usually have no serum antibodies toP carinii when young, but these appear with age.Corticosteroid treatment and protein malnutritionstimulate heavy infection but depress antibody pro-duction. When corticosteroids are stopped highantibody levels develop. Specific antibody can bedetected in lavage fluid from infected rats and alsocoating the organism.34 Alveolar macrophages will

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

not ingest trophozoites unless antipneumocystisserum is present, suggesting that both cellular andhumoral forms of immunity are important.35 Indirectfluorescent antibody staining, with specific rabbitantisera to rat P carinii, has proved the most sensi-tive method of detecting organisms in tissue sectionsand alveolar fluid of cortisone treated rats.36

Clinical features

The clinical features of pneumocystis pneumoniahave been well described in infants with epidemicintersitial plasma cell pneumonia, children andadults having immunosuppressive treatment, andthose with AIDS. Constant features include symp-toms of a febrile illness with dyspnoea and dry coughfrom progressive alveolar filling. Those who developpneumocystis pneumonia are, however, also at riskof various other pulmonary diseases that maymodify the clinical presentation. Gadjusek3 reportedthat epidemic pneumocystis pneumonia started in-sidiously over seven to 14 days with non-specificmalaise and a gradually increasing respiratory rate.Coryza, fever, cough, and weight loss were unusualinitially. At the height of the illness there was con-siderable dyspnoea, cyanosis, intercostal recession,flaring of the nasal alae, and sometimes mediastinalemphysema and pneumothorax. Characteristically,few signs were heard on chest examination. Treat-ment was with rest and oxygen and over a quarter ofpatients died. Survivors recovered slowly over fourto six weeks. Children with congenital immuno-deficiency diseases may present in a similar way.

Occasionally, apparently normal individualsdevelop pneumocystis pneumonia and may presentwith a non-specific pneumonia. In one study Pcarinii infection was implicated in 18% of appar-ently normal children admitted to hospital becauseof pneumonia. The clinical features were similar tothose in children with other types of pneumonia.28By contrast, patients who are on immunosuppres-

sive treatment usually present with an abrupt onsetof illness with dyspnoea in over 90%, high fever intwo thirds, and dry cough in half. Sputum produc-tion, haemoptysis, and chest pain occur in less than8%. Lung crepitations are detected in only a third ofcases; other chest signs are unusual. Symptoms maydevelop acutely37 but the average duration of symp-toms is about two weeks. Symptoms may coincidewith a reduction or cessation of corticosteroidtreatment.38 This may reflect a loss of anti-inflammatory effect-as in cortisone treated rats,where the cessation of corticosteroids is associatedwith a considerable increase in the lung's

30inflammatory response.There appear to be several differences between

563

pneumocystis pneumonia in AIDS patients and thecondition affecting other immunosuppressedadults.3739-1 One likely explanation for this may bethat, unlike AIDS, immunosuppression in non-AIDS patient may often be modified both duringand after pneumocystis pneumonia. The attack rateof 35-60% is high in AIDS. Symptoms appear todevelop much more insiduously, over four weeks orso, although increased awareness of the conditionmay result in earlier presentation and diagnosis.39Fevers, dyspnoea, dry cough, weight loss, diarrhoea,and malaise are common. Chest signs are sparse orabsent. Pneumocystis pneumonia has, however,been diagnosed in both symptom free patients withabnormal chest radiographs39 and patients with mildsymptoms who have normal radiographs.25 42 43Patients with AIDS and pneumocystis pneumoniaimprove more slowly than others and the mortalityrate is higher.37

Laboratory investigations

There are no haematological and biochemical teststhat are helpful in diagnosing pneumocystispneumonia. Abnormalities often result from theunderlying disease or as a result of its treatment.Arterial blood gas estimations usually show severehypoxia; arterial oxygen tensions (Pao2) of 8 kPa(60 mm Hg) or less are common.3942 Hypercapniaand respiratory acidosis occur at a late stage. Seriallung function testing in patients with AIDS andpneumocystis pneumonia shows progressive reduc-tion in total lung capacity and vital capacity and arise in the alveolar-arterial partial pressure gradientfor oxygen. A reduction in diffusing capacity forcarbon monoxide (TLCO) to less than 70% of thepredicted value is reported in nearly all cases.254244

Radiographic findings

The typical radiographic finding is diffuse, bilateralalveolar and interstitial shadowing, beginning in theperihilar region and later coalescing and spreadinginto a butterfly pattern with relative sparing of lungapices and bases. Initially the shadowing is reti-culonodular but progresses over three to five daysinto diffuse, fluffy shadows and finally densehomogeneous consolidation. Air bronchograms arepresent in over half the cases.384546 This pattern isnot unique to pneumocystis pneumonia and adiagnosis cannot be made on radiographic appear-ance alone. Many atypical features have been des-cribed.46 Nevertheless, the presence of focal lobar orsegmental shadows, lung cavitation, large pleuraleffusions, and hilar lymphadenopathy are againstthe diagnosis of pneumocystis pneumonia, although

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

564

they may be associated with the underlying diseaseprocess or a concurrent infection. Radiographicshadowing may lag behind symptoms. Normal chestradiographs have been reported in 10-20% ofpatients with AIDS at the time of diagnosis of

2541243 erdopneumocystis pneumonia. Although the radiographs subsequently became abnormal, suchpatients have a good prognosis.4"

Gallium 67 lung scans invariably have diffuseabnormalities in patients with pneumocystispneumonia even in the presence of a normal chestradiograph.43 In one series, gallium scans were posi-tive in 96 of 98 cases of pneumocystis pneumonia,including 10 with normal chest radiographs.25Although there is a very high sensitivity forpneumocystis pneumonia, scans are positive in awide variety of other active lung conditions. Forinstance, 47% of patients with AIDS who had acutediffuse lung shadowing not caused by pneumocystispneumonia had a positive gallium scan.25 Thespecificity of the scan for diagnosing pneumocystispneumonia can be increased by using a graded scor-ing system.43

DiaosisThe clinical, laboratory, and radiographic patternsare not characteristic and the diagnosis can only bemade with confidence by demonstrating the organ-ism in the lung or bronchoalveolar fluid.

DIAGNOSITIC PROCEDURESThere are several different ways of sampling lungtissue.48 Most have been used successfully to diag-nose pneumocystis pneumonia. Percutaneousneedle aspiration is popular in some centres for chil-dren; a diagnostic rate of 60-95% for pneumocystispneumonia is reported in experienced hands.4950 Itis quick, well tolerated, and a relatively safe proce-dure in children, pneumothorax being the main risk.In one series 17% of patients required a chest drainfor pneumothorax.49For adults the initial procedure of choice is

fibreoptic bronchoscopy.4251-4 This is widely avail-able and safe. At bronchoscopy bronchial and trans-bronchial biopsy specimens, peripheral brush biopsyspecimens, bronchial aspirates, and bronchoalveolarlavage fluid can be obtained. In a recent series ofpatients with AIDS who had lung shadowingpneumocystis pneumonia was diagnosed by bron-choscopy in 368 of 373 (95%) episodes.25 The diag-nosis was made after both touch preparation andfixation of transbronchial lung tissue in 95% ofcases, lavage in 79%, bronchial washings in 55%42and brush biopsies in 39% of cases. No importantcomplications were recorded but a repeat broncho-

scopy was sometimes necessary for a diagnosis. Otherinfections such as cytomegalovirus and Mycobac-terium avium intracellulare can also be diagnosed bythis technique. Kaposi's sarcoma, however, is notreadily identified by bronchoscopy.25 P carinii infec-tion has also been diagnosed by bronchoalveolarlavage in AIDS patients with symptoms who hadnormal chest radiographs.

If pneumocystis pneumonia is suspected in apatient with AIDS who has normal chest radio-graph, it is recommended that invasive tests shouldbe considered if any of the following are abnormal:gallium scan, TLco, or alveolar-arterial oxygengradient.25 We do not yet know whether any of thesetests will be a useful screening procedure forpatients with AIDS. Precautions to be taken when apatient with AIDS is being investigated by bron-choscopy have been reviewed elsewhere.2554 Thesuccess rate for diagnosing pneumocystis pneumoniaat bronchoscopy is probably marginally lower inthose who do not have AIDS,38 perhaps because of alower pathogen burden in the lungs. Bronchoalveo-lar lavage, however, has been used successfully inimmunosuppressed adults to diagnose pneumocystispneumonia.40 55 56

The diagnostic problem is somewhat different inpatients without AIDS because of the diversity ofinfective and non-infective pulmonary disease thatmay be present, and open lung biopsy is favoured bygroups who have had extensive experience of diag-nosing pneumocystis pneumonia. In patients with-out AIDS who have acute lung shadowing3338 s pcarinii is the commonest pathogen identified at openbiopsy. Open lung biopsy is successful in giving aspecific diagnosis (usually of infection) in 50-80%,48and is also the best way of identifying the othercauses of acute lung shadowing.57-60 The operativemortality rate is up to 1% and minor complicationsoccur in a proportion of cases.48

LABORATORY DIAGNOSIS OF P carinjiINFECTIONAt present laboratory diagnosis depends on theidentification of the parasite within tissue or lavagematerial. Isolation of P cannii in tissue cultureremains a research procedure. Serological testing inmost instances provides only supportive evidence,except in selective centres offering antigen detec-tion.

Histologically the typical lung changes onhaematoxylin and eosin staining include foamyintra-alveolar exudate; mild interstitial infiltrates oflymphocytes, plasma cells, and histiocytes; and anincrease in alveolar macrophages, but an absence ofneutrophils.3 Various atypical features have beenreported, including lack of alveolar exudate and the

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

presence of interstitial fibrosis and of granulomas.6'The parasite can be found in both cystic andtrophozoite forms. The cystic form is most readilyvisualised with silver stains such as the Gomorimethenamine variety and the Gram-Weigert stain.Cysts have also been visualised by Gram staining.62The trophozoites frequently outnumber the cysticforms and may be easily overlooked. Staining ofbiopsy touch preparations, ground transbronchialbiopsy specimens, and centrifuged lavage depositcan provide a rapid diagnosis of pneumocystispneumonia.The serodiagnosis of pneumocystis pneumonia

has been widely explored owing to the difficulty inisolating the pathogen and the hazards of obtainingbiopsy material. Various tests have been in use,although critical standardisation of the techniqueshas yet to be established. For example, the source ofthe antigen has varied widely: some centres haveused naturally infected animal sources while othershave relied on human tissue.'0 63 6 The complementfixation test has been useful in the diagnosis ofepidemic pneumocystis pneumonia,65 but of littlehelp in the immunocompromised patient in whomantibody responses are considerably impaired.Immunofluorescence tests have gained popularityand have used antisera raised in immunised animalsas well as convalescent serum from both infectedpatients and animals.6667 Using indirect immuno-fluorescent antibody (IFA) testing with a cyst sus-

pension as antigen, Pifer found some 71% of chil-dren with pneumocystis pneumonia to be seroposi-tive at the time of diagnosis, with a titre of 1/16 or

more. More than 75% of normal children are, how-ever, seropositive by 4 years of age.'763 An enzymelinked immunospecific assay (ELISA) has also beenused and has shown varying titres among differentgroups.68 This emphasises the fact that the mere pre-sence of detectable antibody cannot be equated withthe presence of active pneumocystis pneumonia. Afourfold rise in titre is, however, usually diagnosticwhile titres above 1:32 are certainly suspicious andappear to be associated with pneumocystispneumonia. In the United Kingdomimmunofluorescence testing is carried out only inreference centres, a human lung source being usedas the antigen. The anticipated increase in patientswith AIDS means that there will be considerableopportunities for the critical evaluation of serologi-cal tests in the diagnosis of pneumocystispneumonia.Another approach to the serodiagnosis of

pneumocystis pneumonia is antigen detection. Thishad considerable appeal since a host response toinfection is not required. Pifer et all7 used counter-current immunoelectrophoresis (CIE) to detect P

565

carinii antigen, with rabbit antiserum against Pcarinii isolated in tissue culture from human andmurine lung. They found that 95% of patients withcancer who had pneumocystis pneumonia were anti-genaemic, compared with 15% of cancer patientswithout pneumonitis and none of the normal chil-dren tested. Thus positive tests for antigen do notalways indicate active disease but may reflect anti-gen mobilisation for other reasons. Nevertheless,clearance of antigenaemia does correlate well withclinical improvement." Fractionated P carinii anti-gens may improve specificity.69 In the future moresensitive antigen detection systems, such asradioimmunoassay and ELISA,70 are likely to makethe diagnosis of pneumocystis pneumonia simple,rapid, and reliable through the examination ofserum, lavage material, or possibly urine.

Management

GENERAL MEASURESThe commonest complication of pneumocystispneumonia is respiratory failure. This may requireoxygen treatment and in some instances assistedventilation. The need for assisted ventilation isassociated with a poor prognosis. Of 102 patientswith AIDS who required assisted ventilation, only14% survived.25 In patients having immunosuppres-sive treatment the aim should be to improve theimmunocompetence of the individual by reducingimmunosuppressive treatment or control of theunderlying disease. The fact that at present theimmunocompetence of patients with AIDS cannotbe improved may explain the higher mortality ratein this group. Some clinicians favour the introduc-tion, or an increased dosage, of corticosteroids toreduce the inflammatory response to pneumocystispneumonia; but the fact that the condition is fre-quently associated with other opportunistic infec-tions is probably the biggest argument against re-introduction of corticosteroids in such patients.Anecdotal reports suggest that lung lavage may bebeneficial in removing some of the alveolar exudateand improving oxygenation (analogous to the man-agement of alveolar proteinosis).38

SPECIFIC TREATMENTThe initial observation in 1958 that pentamidinewas effective in the treatment of pneumocystispneumonia in infants with the epidemic form of thedisease was a major breakthrough." Pentamidine isan aromatic diamidino compound synthesised some50 years ago and used in the treatment of Africantrypanosomiasis and resistant forms of leish-maniasis. Its efficacy in pneumocystis pneumoniawas shown in a series of some 212 patients, in whom

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

566

the mortality fell from 50% in the untreated groupto 3.5% in those receiving pentamidine isethion-ate.72 The mode of action is not clear. One sugges-tion is that it interferes with folate metabolism,73although folinic acid does not reduce its efficacy inexperimental animal models.74The standard dosage regimen of pentamidine

isethionate is 4 mg/kg a day for 14 days by theintramuscular route." Unfortunately toxicity iscommon, nephrotoxicity being recognised in almosta quarter of recipients,26 probably as a result of drugconcentration within the kidney.76 The list of otherreported adverse reactions is impressive andincludes hepatotoxicity with hypoglycaemia andoccasionally liver necrosis, weakness, seizures,hypocalcaemia, nausea, vomiting, sweating,tachycardia, thrombocytopenia, and sterile absces-ses at the injection site. Occasional deaths have beenrecorded.75Animal studies in 1966 indicated that antifolate

agents such as pyrimethamine and sulphonamideswere effective in pneumocystis pneumonia.74 Thesubsequent availability of trimethoprim (TMP) ledHughes and his colleagues to conduct a successfulstudy of the combination of trimethoprim and sul-phamethoxazole (TMP-SMX) in the treatment ofpneumocystis pneumonia in the rat model.77 TMP isconcentrated in lung tissue and is less toxic to themarrow than pyrimethamine. Subsequent carefullyexecuted trials showed the effectiveness of TMP-SMX in human pneumocystis pneumonia.78Response rates are similar to those with pen-tamidine79 but adverse reactions are largely avoided.TMP-SMX is known to inhibit nucleic acid synthesisthrough inhibition of the folic acid pathway. Thesulphonamide inhibits the conversion of para-aminobenzoic acid to dihydrofolate, which in turn isblocked by TMP by its action on the enzyme dihyd-rofolate reductase. Neither pentamidine nor TMP-SMX is microbicidal-they are thought to act largelyon the trophozoite form. The daily dose of TMP-SMX associated with optimal response rates consistsof TMP 20 mg/kg plus SMX 100 mg/kg in three orfour divided doses for a total of 14 days,79 althoughin patients with AIDS complicated by pneumocystispneumonia more prolonged treatment is oftennecessary.47 Intravenous SMX-TMP is a usefulalternative.808' The dose of TMP is 10-20 mg/kg/day. The higher dose is often recommended initially,being reduced to 10-15 mg/kg a day when there isclinical improvement.Although TMP-SMX is better tolerated than pen-

tamidine it is not entirely without side effects. Themost common complications are various rashes,which may occasionally be manifest as the Stevens-Johnson syndrome. Adverse reactions to high dose

TMP-SMX are much more frequent in patients withAIDS.82 This is somewhat surprising owing to theoverall anergy of this population and possibly arguesagainst hypersensitivity as the basis for the reaction.The use of TMP-SMX in renal failure may result

in toxic blood levels. There is rather limited infor-mation available concerning dose modification.8384Drug assay would appear desirable in these circums-tances since it has been suggested that the therapeu-tic serum concentrations are 3-S5 mg/l for TMP and100-150 mg/l for SMX two hours after oraladministration.7

RESPONSE TO TREATMENTThe best guide to successful treatment is clinicalimprovement of the patient. Resolution is indicatedby a fall in fever within one to five days and clearingof the chest radiograph within four to 10 days.Patients with AIDS take longer to improve.39Repeat lung function measurements44 and galliumscans43 do not appear to be helpful in monitoringresponse. Similarly, repeat lung biopsy or bron-choalveolar lavage not infrequently shows persis-tence of P carinii, even after two to three weeks ofapparently effective treatment, particularly inpatients with AIDS.254' 43 Clearance of P carinii andthe inflammatory response appears faster in thosewithout AIDS.4' In patients showing a poorresponse repeat bronchoscopy or lung biopsy mayshow a concurrent opportunistic infection. In sur-vivors lung function abnormalities return to normalover a period of many weeks,44 although fibrosis hasbeen reported.38

In patients failing to respond to TMP-SMX byabout the fifth day of treatment a change to pen-tamidine should be considered, particularly if thereare signs of deterioration. Combined treatment withpentamidine and TMP-SMX has been considered inthe hope of achieving either a synergistic effect or,by dose reduction, fewer side effects. In the cor-tisone treated rat model, however, this combinationhas not been shown to be more effective.85

CHEMOPROPHYLAXISChemoprophylaxis was first used in the control ofepidemic pneumocystis pneumonia86 and subse-quently in children with acute lymphoblasticleukaemia and other solid tumours undergoingchemotherapy.8'88 Most experience has been gainedat St Jude's Hospital, Memphis, where it has beenshown that no cases of pneumocystis pneumoniahave occurred since 1977 in patients with acutelymphoblastic leukaemia who have received adequ-ate chemoprophylaxis with doses of TMP of 150mg/m2/day and SMX of 750 mg/m2/day in twodivided doses.89

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

Second attacks of pneumocystis pneumonia arewell recognised and have occurred in up to 15% ofleukaemic patients after treatment with pen-

tamidine.90 Prophylactic TMP-SMX is prescribedfor at least one year after recovery from pneumocys-tis pneumonia unless there is improvement in theimmunological state of the patient. Chemopro-phylaxis of pneumocystis pneumonia in patientswith AIDS may present problems owing to the highincidence of side effects with TMP-SMX. Alterna-tive agents are needed and preliminary experienceusing pyrimethamine and sulfadoxine (Fansidar) hasproved encouraging in those with AIDS9' and inepidemic pneumocystis pneumonia.86 Confirmationof efficacy and the optimal dosage have yet to beestablished.

MANAGEMENT OF THE IMMUNOSUPPRESSEDPATIENT WITH LUNG SHADOWSIn practice the immunosuppressed patient presentsnot with obvious pneumocystis pneumonia butrather with an acute progressive lung infiltrate ofunknown cause. Possible diagnoses include suchdiverse conditions as pulmonary oedema, pulmo-nary haemorrhage, the underlying disease (forexample tumour or vasculitis) directly affecting thelung, reaction to the drug treatment or radiotherapyand various bacterial, fungal, viral and protozoalinfections. In those with AIDS other infections arecommon and include in decreasing order of fre-quency: cytomegalovirus, Mycobacterium aviumintracellulare, M tuberculosis, Legionella pneumo-phila, and cryptococcus, although Kaposi's sarcoma

must also be considered.25 This emphasises theimportance of making a correct diagnosis of acutelung shadowing.When the clinician is faced with an immunosup-

pressed patient with acute lung shadowing, thecommoner respiratory pathogens should be soughtimmediately by routine methods. Most would thengive broad spectrum antibiotics while awaiting theresults. Failure to improve, or deterioration, makesurgent consideration of invasive techniques necessary.Fibreoptic bronchoscopy is the next procedure ofchoice, although transtracheal aspiration may behelpful.38 Bronchoalveolar lavage may be performedwithout transbronchial biopsy in patients withuncorrectable and serious bleeding disorders. If theresults are unhelpful and further information isneeded, open lung biopsy should be considered as amatter of urgency. Useful clinical algorithms aregiven elsewhere.2592 Some studies show that open

lung biopsy results in a change or confirmation ofthe preoperative treatment in half to two thirds ofsuch cases.57 960 Open biopsy offers a high chance ofobtaining a definite diagnosis, which in turn allows a

567

more confident management plan and will preventunnecessary and potentially toxic, multiple, empiri-cal treatments. It can be argued, however, that theresults of open lung biopsy do not alter the overallprognosis in such patients and empirical treatment ispreferable and as safe.579293 The choice of diagnosticprocedure will depend to a large extent on theexpertise and facilities that are available locally,both in performing the various invasive techniquesand also in handling the specimens in the laboratory.Close liason is essential between the clinician,surgeon, microbiologist, and cytohistopathologist.An urgent "out of hours" invasive technique is oflittle use if the specimens are not dealt with immedi-ately in the laboratory by an experienced operator.

The future

In Britain, most clinicians' experience ofpneumocystis pneumonia is very limited. It seemslikely, however, that doctors other than oncologistsand transplantation specialists will need to be awareof the presentation, diagnosis, and treatment of thisinfection. Clinicians will hope that further researchwill establish reliable serological methods for dia-gnosis by antigen and antibody detection, find prac-tical ways of culturing human P cainii, and clarifythe best method of diagnosing pneumocystispneumonia at the earliest stage in susceptible indi-viduals.

JOHN T MACFARLANE

ROGER G FINCH

City Hospital, Nottingham

References

1 Dutz W. Pneumocystis carinii. Pathol Ann 1970;5:309-41.

2 Young LS. Pneumocystis carinii pneumonia. NewYork: Marcel Dekker, 1984:1-6.

3 Gajdusek DS. Pneumocystis carinii-etiological agentof interstitial plasma cell pneumonia of premature andyoung infants. Pediatics 1957; 19:543-65.

4 Vanek J, Jirovec 0. Parasitare pneumonie. Inters-titielle plasma zell pneumonie der Fru geburten verur-sach durch Pneumocystis carinii. Bakt Orig1952; 158:120-7.

5 Hughes WT, Price RA, Sisko F, et al. Protein caloriemalnutrition-a host determinant for Pneumocystiscarinii pneumonitis.Am J Dis Child 1974; 128:44-52.

6 Gottlieb MS, Schroff R, Shanker HM. Pneumocystiscarinii pneumonia and mucosal candidiasis in previ-ously healthy homosexual men: evidence of a newacquired cellular immunodeficiency. N Engl J Med1981;301: 1425-30.

7 Murphy MJ Jr, Pifer LL, Hughes WT. Pneumocystiscarinii in vitro: a study by scanning electron micro-scopy. Am J Pathol 1977;86:387-402.

8 Walzer PD. Experimental models of Pneumocystis

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

568

carinii infections. In: Young LS, ed. Pneumocystiscarinii pneumonia. New York: Marcel Dekker, 1984:7-76.

9 Hasleton PS, Curry A, Rankin EM. Pneumocystiscarini pneumonia: a light microscopical and ultrastruc-tural study. J Clin pathol 1981;34:1138-46.

10 Pifer LL, Hughes WT, Murphy MJ. Propagation ofPneumocystis carinii in vitro. Pediatr Res 1977;11:305-16.

11 Bartlett MS, et al. Cultivation of Pneumocystis cariniiwith WI-38 cells. J Clin Microbiol 1979; 10:796-9.

12 Latorre CR, Sulzer AJ, Norman LG. Serial propaga-tion of Pneumocystis cannii in cell line cultures. ApplEnviron Microbiol 1977;33:1204-6.

13 Pifer LL, Woods D, Hughes WT. Propagation ofPneumocystis carnii in Vero cell culture. InfectImmun1978;20:66-8.

14 Bartkett MS, Eichholtz R, Miller J, Smith JW. Activityof antimicrobial agents against Pneumocystis carinii inculture. In: Abstracts ofthe Interscience Conference onAntimicrobial Agents and Chemotherapy, Las Vegas,Nevada. Washington, DC: American Society for Mic-robiology, 1983:173.

15 Cushion MT, et al. Cultivation of Pneumocystis cariniiin lung-derived cell lines. J Infec Dis 1984; 149:644.

16 Pifer LL. Pneumocystis carinii: a misunderstoodopportunist. Eur J Clin Microbiol 1984; 3: 169-73.

17 Pifer LL, Hughes WT, Stagno S, Woods D.Pneumocystis carinii infection: evidence for high pre-valence in normal and immunosuppressed children.Pediatrics 1978;61:35-41.

18 Meuwissen JHE, Tauber I, Leeuwenberg AD. Para-sitologic and serologic observations of infection withpneumocystis in humans. J Infec Dis 1977; 136:43-50.

19 Bazaz GR, Manfredi OL, Howard RG, Claps AA.Pneumocystis carinii pneumonia in three full-term sibl-ings. J Pediatr 1970;76:767-9.

20 Singer C, Armstrong D, Rosen PP, Schottenfield D.Pneumocystis carinii pneumonia: a cluster of elevencases. Ann Intern Med 1975;82:772-7.

21 Hughes W, Price RA, Kim H, Coburn TP, Grigby D,Feldman S. Pneumocystis carinii pneumonitis in chil-dren with malignancies. J Pediatr 1973;82:404-15.

22 LeClair RA. Transplantation pneumonia, associatedwith Pneumocystis carinii, among recipients of cardiactransplants. Am Rev Resp Dis 1969; 100:874-5.

23 Hardy AM, Wajszyczuk CP, Suffredinin AF, HakalaTR, Mo M. Pneumocystis carinii pneumonia in renaltransplant recipients treated with cyclosporin andsteroids. J Infect Dis 1984; 149:143-7.

24 Kuller J, First MR, I'Achiardi R, Munda R.Pneumocystis carinii pneumonia in renal transplantrecipients. Am J Nephrol 1982;2:213-5.

25 Murray JF, Felton CP, Garay SM, etal. Pulmonarycomplications of the acquired immunodeficiency syn-drome. N Engl J Med 1984;310: 1682-8.

26 Walzer PD, Perl DP, Krogstad DJ, Rawson PG,Schultz MG. Pneumocystis carinii pneumonia in theUnited States. Ann Intern Med 1974;80:83-93.

27 Walzer PD, Schultz MG, Western KA, Robbins JB.Pneumocystis carinii pneumonia and primary immunedeficiency diseases of infancy and childhood. J Pediatr1973;82:416-22.

28 Stagno S, Brasfield DM, Brown MB, et al. Infantpneumonitis associated with cytomegalovirus,chlamydia, pneumocystis and ureaplasma: a prospec-

tive study. Pediatrics 1981;68:322-9.29 Hughes WT, Feldman S, Aur RJ, Verzosa MS, Hustu

HO, Simone JV. Intensity of immunosuppressivetherapy and the incidence of Pneumocystis cariniipneumonitis. Cancer 1975;36: 2004-9.

30 Walzer PD, Powell RD, Yoneda K, Rutledge ME,Milder JE. Growth characteristics and pathogenesis ofexperimental Pneumocystis carinii pneumonia. InfectImmun 1980;27:928-37.

31 Yoshida Y, Matsumoto Y, Yamada M, et al.Pneumocystis carinii: electron microscopic investiga-tions on the interaction of trophozoite and alveolarlining cell. Zbl Bakt Hyg A 1984;256: 390-9.

32 Yoneda K, Walzer PD. Experimental Pneumocystiscarinii pneumonia in the rat: mechanism of alveolarinjury. Br J Exper Pathol 1981;62:339-46.

33 Walzwer PD. Pneumocystis carinii infection: a review.South Med J 1977; 70:1130-7.

34 Walzer PD, Rutledge ME. Humoral immunity inexperimental Pneumocystis carinii infection. J Lab ClinMed 1981;97:821-33.

35 Masur H. Interactions between Pneumocystis cariniiand phagocytic cells. In: Young LS, ed. Pneumocystiscarinii pneumonia. New York: Marcel Dekker,1984:77-95.

36 Milder JE, Walzer PD, Coonrod JD, Rutledge ME.Comparison of histological and immunological techni-ques for the detection of Pneumocystis carinii in ratbronchial lavage fluid. J Clin Microbiol 1980; 11: 409-17.

37 Sterling RP, Bradley BB, Khalil KG, Kerman RH,Conklin RH. Comparison of biopsy provenPneumocystis carinii pneumonia in acquired immunedeficiency syndrome patients and renal allograft reci-pients. Ann Thorac Surg 1984; 38:494-9.

38 Young LS. Clinical aspects of pneumocystis in man. In:Young LS, ed. Pneumocystis carinii pneumonia. NewYork: Marcel Dekker, 1984:139-74.

39 Engelberg LA, Lerner CW, Tapper ML. Clinical fea-tures of Pneumocystis carinii in the acquired immunedeficiency syndrome. Am Rev Respir Dis 1984;130:689-94.

40 Kovacs JA, Hiemenz J, Macher AM, et al. A compari-son of Pneumocystis carinii pneumonia in patients withand without acquired immunodeficiency syndrome.Clin Res 1983;31:691.

41 Shelhamer JH, Ognibene FP, Macher AM, et al. Per-sistence of Pneumocystis carinii in lung tissue ofacquired immunodeficiency syndrome patients treatedfor pneumocystis pneumonia. Am Rev Respir Dis1984; 130:1161-5.

42 Coleman DL, Dodek PM, Luce JM, et al. Diagnosticutility of fibreoptic bronchoscopy in patients withPneumocystis carinii pneumonia and the acquiredimmune deficiency syndrome. Am Rev Respir Dis1983; 128:795-9.

43 Coleman DL, Hattner RS, Luce JM, et al. Correlationbetween gallium lung scans and fiberoptic broncho-scopy in patients with suspected Pneumocystis cariniipneumonia and the acquired immune deficiency syn-drome. Am Rev Respir Dis 1984; 130:1166-9.

44 Coleman DL, Dodek PM, Golden JA, et al. Correla-tion between serial pulmonary function tests andfiberoptic bronchoscopy in patients with Pneumocystiscarinii pneumonia and the acquired immune deficiencysyndrome. Am Rev Respir Dis 1984; 129:491-3.

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

45 Dee P, Winn W, McKee K. Pneumocystis carinii infec-tion of the lung: radiologic and pathologic correlation.Am J Radiol 1979; 132:741-6.

46 Doppmann JL, Geelhoed GW. Atypical radiographicfeatures in Pneumocystis carinii pneumonia. In: Rob-bins JB et al, eds. Symposium on Pneumocystis cariniiinfection. Bethesda: National Cancer Institute Mono-graph No 43, 1976:89-96.

47 Rosen MJ et al. Prognosis of Pneumocystis cariniipneumonia in the Acquired Immunodeficiency Syn-drome. Ann Intern Med 1984; 101:276.

48 Macfarlane JT. Lung biopsy. Br MedJ 1985;290:97-8.49 Chaudhary S, Hughes WT, Feldman S, et al. Per-

cutaneous transthoracic needle aspiration of the lung.Am J Dis Child 1977; 131:902-7.

50 Hughes WT. Pneumocystis carinii pneumonia. N EnglJ Med 1977;297:1381-3.

51 Ognibene FP, Shelhamer J, Gill V, et al. The diagnosisof Pneumocystis carinii pneumonia in patients with theacquired immunodeficiency syndrome using subseg-mental bronchoalveolar lavage. Am Respir Dis1984; 129:929-32.

52 Stover DE, White DA, Romano PA, Gellene RA.Diagnosis of pulmonary disease in acquired immunedeficiency syndrome. Am Rev Respir Dis 1984;130:659-62.

53 Swinburn CR, Pozniak AL, Sutherland S, et al. Earlyexperience and difficulties with bronchoalveolar lavageand transbronchial biopsy in the diagnosis of AIDSassociated pneumonia in Britain. Thorax 1985;40:166-70.

54 Pozniak AL, Swinburn CR, Mc I Johnson N. Manage-ment of AIDS pneumonia. Br J Dis Chest1985;79: 105-14.

55 Hopkin JM, Turney JH, Young JA, Adu D, Michael J.Rapid diagnosis of obscure pneumonia in immunosup-pressed renal patients by cytology of alveolar lavagefluid. Lancet 1983;ii:299-301.

56 Kelley J, Landis JN, Davis GS, et al. Diagnosis ofpneumonia due to pneumocystis by subsegmental pul-monary lavage via the fiberoptic bronchoscope. Chest1978;74:24-8.

57 Imoke E, Dudgeon DL, Colombani P, et al. Open lungbiopsy in the immunocompromised pediatric patient. JPediatr Surg 1983; 18:816-21.

58 Jaffe JP, Maki DG. Lung biopsy in immunocomprom-ised patients. Cancer 1981;48: 1144-53.

59 Mason WH, Siegel SE, Tucker BL. Diagnostic openlung biopsy for diffuse pulmonary disease inimmunocompromised pediatric patients. Am J PedHematlOncol 1982;4: 355-9.

60 Prober CG, Whyte H, Smith CR. Open lung biopsy inimmunocompromised children with pulmonary infiltra-tion. Am J Dis Child 1984; 138:60-3.

61 Weber WR, Askin FB, Dehner LP. Lung biopsy inPneumocystis carinii pneumonia. A histopathologicstudy of typical and atypical features.Am J Clin Pathol1977;67: 11-9.

62 Felegie TP, et al. Recognition of Pneumocystis cariniiby Gram stain in impression smears of lung tissue. JClin Microbiol 1984;20: 1190-1.

63 Meuwissen JHE Th, Tauber I, Leeuwenberg ADEM,Beckers PAJ, Sieben M. Parasitologic and serologicobservations of infection with pneumocystis in humans.J Infect Dis 1979;136:43-9.

64 Meyers JD, Pifer LL, Sale GE, Thomas ED. The value

569

of Pneumocystis carinii antibody and antigen detectionfor diagnosis of Pneumocystis carinii pneumonia aftermarrow transplantation. Am Rev Respir Dis 1979;120:1283-7.

65 Barta K. Complement fixation test for pneumocystosis.Ann Intern Med 1969;70:235.

66 Lim SK, Eveland WC, Porter RJ. Development andevaluation of a direct fluorescent antibody method forthe diagnosis of Pneumocystis carinii infection inexperimental animals. Appl Microbiol 1973;26:666-71.

67 Norman L, Kagan IG. A preliminary report of an indi-rect fluorescent antibody test for detecting antibodiesto cysts of Pneumocystis carinii in human sera. Am JClin Pathol 1972;58:170-6.

68 Pifer L, Niell H, Morrison B, Counce J. Pneumocystiscarinii antigenemia in patients with malignancy,acquired immunodeficiency syndrome, Kaposi's sar-coma or renal allograft. In: Proceedings of the Ameri-can Society for Clinical Oncology. San Diego, Califor-nia, 1983:49 (abstract).

69 Maddison SE, Hayes GV, Ivey MH, et al. Fractiona-tion of Pneumocystis carinii antigens used in anenzyme-linked immunosorbent assay for antibodiesand in the production of antiserum for detectingPneumocystis carinii antigenemia. J Clin Microbiol1982; 15: 1029-35.

70 Maddison SE, Hayes GV, Slemenda SB, et al. Detec-tion of specific antibody by enzyme-linkedimmunosorbent assay and antigenemia by counter-immunoelectrophoresis in humans infected withPneumocystis carinii. J Clin Microbiol 1982; 15: 1036-43.

71 Ivady G, Paldy L. A new form of treatment of intersti-tial plasma-cell pneumonia in premature infants withpentavalent antimony and aromatic diamidines. MschrKinderheilk 1958; 106:10-4.

72 Ivady G, Paldy L, Koltay M, Toth G, Kovacs Z.Pneumocystis carinii pneumonia. Lancet 1967;i: 616-7.

73 Robbins JB. Pneumocystis carinii pneumonitis-areview. Pediatr Res 1967; 1: 131-58.

74 Frenkel JK, Good JT, Schultz JA. Latent pneumocystisinfection in rats relapse and chemotherapy. Lab Invest1966; 15: 1559-77.

75 Western KA, Perera DR, Schultz MG. Pentamidineisethionate in the treatment of Pneumocystis cariniipneumonia. Ann Intern Med 1970;73:695-702.

76 Waalkes TP, Denham C, Devita VT. Pentamidine:clinical pharmacologic correlations in man and mice.Clin Pharmacol Ther 1970;11:505-12.

77 Hughes WT, McNabb PC, Makres TD, Feldman S.Efficacy of trimethoprim and Sulfamethoxazole in theprevention and treatment of Pneumocystis cariniipneumonitis. Antimicrob Agents Chemother 1974;5:289-93.

78 Hughes WT, Feldman S, Sanyal SK. Treatment ofPneumocystis carinii pneumonitis with trimethoprim-sulfamethoxazole. Can Med Assoc J 1975;112 suppl:47-50.

79 Hughes WT, Feldman S, Chaudhary SC, Ossi MJ, CoxF, Sanyal SK. Comparison of pentamidine isethionateand trimethoprim-sulfamethoxazole in the treatmentof Pneumocystis carinii pneumonia. J Pediatr1978;92:285-91.

80 Sattler FR, Remington JS. Intravenous trimethoprim-sulfamethoxazole therapy for Pneumocystis carinii

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from

570

pneumonia. Am J Med 1981;70: 1215-21.81 Winston DJ, Lau WK, Gale RP, Young LS.

Trimethoprim-sulfamethoxazole for treatment ofPneumocystis carinii pneumonia. Ann Intern Med1980;92: 762-9.

82 Gordin FM, Simon GL, Wofsy CB, Ills J. Adversereactions to trimethoprim-sulfamethoxazole in patientwith the acquired immunodeficiency syndrome. AnnIntern Med 1984;100:495-9.

83 Bourgault AM, Van Scoy RE, Brewer NS, RosenowEC III. Trimethoprim with sulfamethoxazole fortreatment of infection with Pneumocystis carinii inrenal insufficiency. Chest 1978;74:91-2.

84 Reider J, Schwarz DE, Fernex M, et al. Pharmacokine-tics of the antibacterial combination sulfamethoxazoleplus trimethoprim in patients with normal or impairedkidney function. Antibiotic Chemother 1974; 18:148-98.

85 Kluge RM, Spaulding DM, Spain AJ. Combination ofpentamidine and trimethoprim-sulfamethoxazole inthe therapy of Pneumocystis carinii pneumonia in rats.Antimicrob Agents Chemother 1978; 13: 975-8.

86 Post C, Fakouhi T, Dutz W, Bandarizadeh B, KoboutE. Prophylaxis of epidemic infantile pneumocystis witha 20:1 sulfadoxime + pyrimethamine combination.

Curr Ther Res 1971; 13:273-9.87 Harris RE, McCallister JS, Allen SA, Barton AS,

Baehner RL. Prevention of Pneumocystispneumonia-use of continuous sulfamethoxazole-trimethoprim therapy. Am J Dis Cild 1980;134:35-8.

88 Hughes WT, Kuhn S, Chaudhary S, etal. Successfulchemoprophylaxis for P carinii pneumonitis. N Engl JMed 1977;279: 1419-26.

89 Hughes WT. Five year absence of Pneumocystis cariniipneumonitis in a pediatric oncology centre. J Infect Dis1984; 150:305-6.

90 Hughes WT, Johnson WW. Recurrent Pneumocystiscarinii pneumonia following apparent recovery. JPediatr 1971;79:755-9.

91 Gottlieb MS, Knight S, Mitsuyasu R, Weisman J, RothM, Young LS. Prophylaxis of Pneumocystis cariniiinfection in AIDS with pyrimethamine-sulfadoxine.Lancet 1984;ii:398-9.

92 Wardman AG, Cooke NJ. Pulmonary infiltrates inadult acute leukaemia: empirical treatment or lungbiopsy? Thorax 1984;39:647-50.

93 Hiatt JR, Gong H, Mulder DG, Ramming KP. Thevalue of open lung biopsy in the immunocompromisedpatient. Surgery 1982;92:285-91.

on 4 May 2018 by guest. P

rotected by copyright.http://thorax.bm

j.com/

Thorax: first published as 10.1136/thx.40.8.561 on 1 A

ugust 1985. Dow

nloaded from