Fibreoptic bronchoscopy in the diagnosis of pulmonary disease in the immunocompromised host in northern Alberta

Pulmonary Division, Department of Medicine, University of Alberta, Edmonton, Alberta Correspondence and reprints: Dr Jennifer A Crocket, 2E4.39 Walter C MacKenzie Health Sciences Centre, University of Alberta, Edmonton, Alberta T6G 2B7. Telephone 403-492-6151, fax 403-492-6384 Received for publication October 5, 1994. Accepted June 16, 1995 JA CROCKET, MR CHAPUT, DC LIEN. Fi bre op tic bron cho scopy in the di ag no sis of pul mo nary dis ease in the im mu no com pro mised host in north ern Al berta. Can J In fect Dis 1995;6(6):286290.

B OTH IN FEC TIOUS AND NON IN FEC TIOUS ETI OLO GIES MAY cause pul mo nary dis ease in the im mu no com pro mised host, and di ag no sis can be dif fi cult.Clini cal and ra dio graphic fea tures are of ten non spe cific, and fi bre op tic bron cho scopy is fre quently used as a pri mary means of evalua tion.Pre vi ous stud ies have as sessed the util ity of bron cho scopy in the di agno sis of pul mo nary dis ease in the im mu no com pro mised host with var ied re sults de pend ing on the tech niques em ployed, popu la tion stud ied and di ag nos tic cri te ria used (1)(2)(3)(4)(5)(6)(7)(8)(9)(10).The ob jec tive of this study was to de ter mine the di ag nos tic util ity of bron cho scopy in a popu la tion of im mu no com pro mised hosts in north ern Al berta by com par ing bron choal veo lar lavage (BAL) with pro tected double-sheathed brushes, cy tology brushes and trans bron chial lung bi opsy (TBB).

PA TIENTS AND METH ODS
Study popu la tion: The study popu la tion was se lected ret rospec tively from a re view of all pa tients who un der went broncho scopy with BAL at the Uni ver sity of Al berta Hos pi tals, Ed mon ton, Al berta be tween August 1991 and August 1992.From a to tal of 143 pa tients, 86 im mu no com pro mised pa tients who un der went a to tal of 101 bron cho scopy pro ce dures were in cluded.Fifty-seven pa tients of the 143 were ex cluded: 55 pa tients were not im mu no com pro mised hosts, and in two immu no com pro mised pa tients in com plete speci mens were obtained.
The study popu la tion com prised solid or gan trans plan tation (n=21), bone mar row trans plan ta tion (n=9), hu man immu no de fi ciency vi rus (HIV) in fec tion (n=28), im mu no sup pres sive drug ther apy (n=10), he ma to logi cal malig nancy (n=16) and lung can cer (n=2) (Fig ure 1).Im mu nosup pres sive drug ther apy in cluded che mo thera peu tic agents, im mu no sup pres sive drugs in clud ing cy clo spor ine, cy clophos pha mide and azathio prine, or cor ti cos ter oid ther apy taken for at least three months, the low est dose be ing 10 mg of predni sone on al ter nate days.Of the 86 pa tients, 68 were male and 18 fe male with a mean age of 45 years (range 18 to 84).Bron cho scopy: Bron cho scopy was un der taken to evalu ate an ab nor mal chest ra dio graph and/or pul mo nary symp toms of dysp nea, cough or ab nor mal spu tum pro duc tion with increased vol ume or pu ru lent col ora tion on 98 oc ca sions, and for lung trans plant sur veil lance on three oc ca sions.The patients re ceived stan dard pre medi ca tion of in tra mus cu lar or intra ve nous at ro pine, and in tra mus cu lar or in tra ve nous nar cot ics or di aze pam.The oro pha rynx and up per air way were an es thes tized with lido caine so lu tion be fore in tro duc tion of the bron cho scope.Pro tected sheath brushes (1 mm mi crobi ol ogy brushes) were taken be fore BAL from the area of ra diographic ab nor mal ity.All pa tients un der went BAL from the right mid dle lobe or lin gula in the cases of dif fuse lung dis ease.Where a lo cal ized ra dio graphic in fil trate was pres ent, the bron cho scope was wedged in a bron chus in that area.A to tal of 150 mL of nor mal sa line in five ali quots of 30 mL was instilled with as pi ra tion through the side chan nel of the bron choscope af ter each ali quot.All five re cov ered ali quots were then com bined.Fol low ing BAL, cy tol ogy brushes and TBB were obtained from the area of ra dio graphic ab nor mal ity.BAL speci mens: BAL ef flu ent was di vided and sub mit ted for cul tures for bac te ria in 99 cases, fungi in 101, myco plasma in 95, chla my dia in 88, my co bac te rium in 100, vi ruses in 87 and cy tomega lovi rus (CMV) in 87.CMV early an ti gen was de tected by the shell vial tech nique (11)(12)(13)(14).BAL bac te rial cul tures were not quan ti fied for the num bers of bac te rial colo nies be cause this is not stan dard prac tice in the authors' labo ra tory.Protected double-sheathed brushes were sent for aero bic and anaero bic cul tures in 89 pa tients and for le gionella cul ture in 99.These brushes were trans ferred to Todd Hew lett and Cary Blair me dia with ster ile tech nique in the bron cho scopy suite, and then vor texed and in ocu lated on to blood agar, choco late agar, and an aero bic and Mac Con key me dia in the mi cro bi ology labo ra tory.Sheathed brush cul tures were quan ti fied with a sig nifi cant posi tive cul ture de ter mined to be greater than 10 6 colo nies/L.The speci men for le gionella was rou tinely cul tured on blood agar and buff ered charcoal-yeast ex tract me dia with and with out an ti bi ot ics.A di rect fluo res cent an ti body test was done if spe cifi cally re quested by the phy si cian.
Cy to logi cal ex ami na tion was per formed on BAL ef flu ent in 100 cases and on brush speci mens in 92.Slides were routinely stained with Pa pani co laou stain, and a Go mo ri's methenamine-silver stain was used to iden tify Pneu mo cys tis car inii and fungi.Iron stains to de tect he mo sid erin laden macro phages were not rou tinely done.TBB for his tol ogy was done in 30 pa tients.These tis sues were proc essed in the hospi tal his tol ogy labo ra tory, stained with he ma tox ylin and eo sin and in ter preted by a pul mo nary pa tholo gist.

RE SULTS
A posi tive re sult was ob tained in 94% of pro ce dures, but the re sults were de ter mined to be clini cally rele vant by the authors in only 57%.Clini cal rele vance was based on a change of ther apy, in sti tu tion of new ther apy or con fir ma tion of pres ent ther apy fol low ing bron cho scopy.Al ter na tively, results were con sid ered clini cally rele vant when a posi tive cul -

Fig ure 1) Dis tri bu tion of the pa tient popu la tion (n=86). HIV Hu man immu no de fi ciency vi rus
CAN J IN FECT DIS VOL 6 NO 6 NO VEM BER/DE CEM BER 1995 ture of the same or gan ism from other sites in clud ing blood and urine, or posi tive his tol ogy from bi opsy or autopsy specimens was pres ent.Posi tive re sults that were ex cluded were those thought to rep re sent oro pha ryn geal con tami na tion, airway colo ni za tion, CMV an ti gen ic ity with out evi dence of pneumo ni tis in the HIV group, and posi tive fun gal cul tures for Can dida al bi cans in the ab sence of sys temic or in va sive disease.Di ag nos tic yield was high est in pa tients on im mu no suppres sive drug ther apy (80%) and low est in bone mar row trans plant pa tients (27%).Di ag nos tic yield in the solid or gan trans plan ta tion group was 69%, in HIV pa tients 62%, in he mato logi cal ma lig nancy 35% and in the lung can cer group 50% (Fig ure 2).Of the clini cally rele vant re sults, an in fec tious eti ology was found in 85%, non in fec tious eti ol ogy in 14% and com bined eti ol ogy in 1%.
The di ag nos tic yield of BAL cul tures (bac te ria, fungi, mycoplasma, chla my dia, my co bac te rium and vi ruses) was 38%, of BAL cy tol ogy was 19%, of brush cy tol ogy was 13%, of sheathed brush cul tures (bac te ria and le gionella) was 10% and of TBB was 27% (Fig ure 3).No cases of le gionella were diag nosed.Posi tive BAL cy tol ogy re sults com prised ma lig nancy in two pa tients, CMV in fec tion in four pa tients, P car inii in fection in 12 pa tients and fun gus in one pa tient.Cy tol ogy brushes were di ag nos tic for ma lig nancy in two pa tients, for CMV in two pa tients and for P car inii in eight pa tients.TBB was posi tive in eight pa tients: CMV in three pa tients, lung trans plant re jec tion in two pa tients, fi bro sis in one pa tient, a plasma cell in fil trate in one pa tient and tal co sis in one pa tient.
Ta ble 1 out lines the fre quency of vari ous di ag no ses within the six pa tient groups.Bac te ria were the most com mon patho gens, fol lowed by vi ruses and P car inii.BAL bac te rial cultures were not quan ti fied and the false posi tive rate was high.Of 122 posi tive cul tures, 24 were judged to be clini cally relevant.Clini cally rele vant bac te rial iso lates were Pseu do mo nas ae rugi nosa (three cases), En te ro cocci (three), Kleb siella pneu mo niae (two), Ser ra tia marc es cens (two), Hae mo phi lus in flu en zae (three), Mo rax ella ca tar rhalis (one), Strep to coc cus pneu mo niae (six), Staphy lo coc cus au reus (one), Neis se ria men in gi ti dis (one), coagulase-negative staphy lo cocci (one) and mixed an aer obes (one).Of 15 posi tive CMV re sults, eight were de ter mined to be clini cally sig nifi cant, and of 35 posi tive fun gal re sults, five were judged to be of clini cal rele vance.
Dif fer en ti at ing fun gal colo ni za tion from tis sue in va sion and in the case of C al bi cans, oro pha ryn geal con tami na tion, remains a dif fi cult clini cal prob lem.The pres ence of sys temic or in va sive dis ease with C al bi cans is best con firmed by co ex isting posi tive cul tures from the other sites such as blood and urine, or by posi tive his tol ogy from bi opsy or autopsy specimens.Of the five fun gal re sults deemed to be clini cally relevant, two rep re sented as per gil lus in fec tion in lung trans plant pa tients, both of whom were treated with sys temic an ti fun gal Fig ure 3 ther apy and one of whom had autopsy evi dence of in va sive as per gil lus.A third sig nifi cant re sult was blas to my ces in a neu tro penic pa tient on chronic cor ti cos ter oid ther apy with posi tive blood cul tures and autopsy evi dence of dis semi nated fun gal in fec tion.The re main ing two clini cally rele vant re sults were C al bi cans.One pa tient had con cur rent posi tive blood cul tures and in a sec ond pa tient the re sult was con sid ered rele vant be cause the pa tient was treated for a pro longed time with oral an ti fun gal ther apy by the at tend ing phy si cian.The re main ing 30 posi tive fun gal re sults were C al bi cans and thought to rep re sent oro pha ryn geal con tami na tion or air way colo ni za tion.

TA BLE 1 Dis tri bu tion of clini cally rele vant posi tive re sults*
The com pli ca tion rate of bron cho scopy was 4%.Se vere hy poxe mia oc curred in two in stances (one pa tient re covered spon ta ne ously while the sec ond re quired me chani cal ven ti la tion).Two pa tients had pneu motho ra ces.

DIS CUS SION
Pul mo nary dis ease is a com mon clini cal prob lem in the immu no com pro mised host and fi bre op tic bron cho scopy is advo cated as a pri mary means of evalua tion al low ing di rect sam pling from the lung with low mor bid ity.In this study, the over all di ag nos tic yield of bron cho scopy was 57%.This was de ter mined by di vid ing the number of posi tive pro ce dures with clini cally rele vant re sults, which num bered 58, by the total number of pro ce dures done, of which there were 101.The di ag nos tic yield of BAL alone was 50%, with sheathed brushes con trib ut ing ad di tional in for ma tion in three cases, brush cy tology in one case and TBB in seven cases.A re view of other reports in the lit era ture in di cates a di ag nos tic yield for BAL in simi lar pa tient groups rang ing from 39 to 87% (1-10).This vari abil ity may in part be re lated to dif fer ences in pa tient popu la tions, as sug gested by Pi sani and Wright (2).The yield of BAL tends to be higher in the HIV in fected pa tient popu la tions (8,9) as op posed to pa tient popu la tions with a high pre ponder ance of he ma to logi cal ma lig nancy (2,3).Our study confirms this trend with di ag nos tic yield be ing low est in the bone mar row trans plant pa tients (27%) and in pa tients with he mato logi cal ma lig nancy (35%).The ma jor ity of pa tients in these groups had re ceived em piri cal an ti bac te rial ther apy be fore bron cho scopy, pos si bly de creas ing the di ag nos tic yield.In ad di tion, the in ci dence of other di ag no ses, in clud ing pul monary hem or rhage and chemoradiation-induced lung in jury, may be higher in these pa tient groups.Idio pathic pneu mo nia syn drome is an other di ag nos tic con sid era tion in the bone marrow trans plant group (15).
In this se ries of pa tients, pro tected sheathed brushes added lit tle ad di tional in for ma tion to BAL re sults.Bac te ria were thought to rep re sent sig nifi cant patho gens in 13 cases where sheathed brush cul tures were ei ther nega tive or of in suf fi cient growth to de fine sig nifi cance.Topi cal lido caine so lu tion was used be fore the in tro duc tion of the bron cho scope for pa tient com fort.Al though lido caine is known to have an ti bac te rial prop er ties, there is no good in for ma tion on its ef fect on mi crobio logi cal speci mens ob tained by bron cho scopy.How ever, the po ten tial to re duce the re cov ery of mi cro or gan isms in quan ti ta tive cul tures must be noted.Kahn and Jones (16) re viewed the di ag no sis of bacte rial res pi ra tory in fec tion us ing quan ti ta tive BAL cul tures in a group of pre domi nantly im mu no com pro mised pa tients.Protected brush cul tures were done con cur rently in a sub group of these pa tients.Their find ings in di cated that quan ti ta tive BAL cul tures are both sen si tive and spe cific with a rea son able corre la tion (r=0.78,P<0.001) with pro tected brush cul tures.They con cluded that quan ti ta tive bac te rial cul ture of the BAL specimen (in com bi na tion with analy sis of BAL cell dif fer en tial) can be used to di ag nose bac te rial res pi ra tory tract in fec tions.In our study sheathed brush cul tures added lit tle in for ma tion to BAL bac te rial cul tures and the ques tion of util ity arises.It may be more cost ef fec tive to quan tify BAL cul tures and forego the ex pense of sheathed brush cul tures.This is sue could not be di rectly ad dressed in this study be cause the mi cro bi ol ogy labo ra tory at our in sti tu tion did not rou tinely quan tify BAL speci mens dur ing this study pe riod.
Le gionella cul tures from the pro tected sheath brushes were done in 99 pa tients with no posi tive cul tures.Le gionella is an un com mon patho gen in north ern Al berta and we rec ommend that cul tures be done only in spe cific pa tients where there is a high de gree of clini cal sus pi cion.
Brush speci mens for cy to logi cal ex ami na tion were done in 92 pa tients and yielded a clini cally rele vant posi tive re sult on 12 oc ca sions.This con trib uted ad di tional di ag nos tic in for mation in only one pa tient and there fore the role of rou tine cy tology brush ing may need to be re viewed.
In this study, 57% of re sults ob tained by bron cho scopy were clini cally rele vant.We de fined clini cal rele vance as a result that caused a change in ther apy, in sti tu tion of new therapy or con fir ma tion of pres ent ther apy.De fin ing clini cal rele vance was a sub jec tive pro cess, but per haps best reflected the ac tual use of the re sults.Al though this is a methodo logi cal limi ta tion, it may be the best means of ana lyz ing clini cal im por tance.Cul tures that were not clini cally rele vant could be con sid ered as false posi tives.In as sess ing whether a re sult is of clini cal rele vance, proba bly the best guide line to of fer from the re view of our data is that in di vid ual re sults must be in ter preted in light of the clini cal situa tion.Cases in which a re sult is not of clini cal rele vance are most likely to oc cur where there is a high prob abil ity that the or gan ism rep re sents contami na tion from the oro pha ryn geal air way, air way colo ni zation or, in the case of pre vi ous CMV ex po sure, per sis tence of the or gan ism with out ac tive in fec tion.In ad di tion to clini cal cor re la tion, quan ti fi ca tion of BAL cul tures could po ten tially assist the cli ni cian in dis tin guish ing false posi tives from clini cally rele vant re sults.
This study has ob vi ous limi ta tions in that it is a ret ro spective analy sis of a het ero ge ne ous popu la tion of im mu no compro mised pa tients.The false nega tive rate for bron cho scopy could not be de ter mined be cause a de fini tive di ag no sis was not avail able in all pa tients re viewed.The study does, however, pro vide use ful in for ma tion about the di ag nos tic yield of bron cho scopy and the util ity of cer tain pro ce dures in clud ing BAL, pro tected double-sheathed brushes and rou tine cy tol ogy brushes.
We con clude that bron cho scopy is a valu able tool in our CAN J IN FECT DIS VOL 6 NO 6 NO VEM BER/DE CEM BER 1995 cen tre for the evalua tion of pul mo nary dis ease in the im mu nocom pro mised host.The com pli ca tion rate is low com pared with open lung bi opsy (17), and our over all di ag nos tic yield of 57% is simi lar to other re ports in the lit era ture (1-10).