Evaluation of an indirect hemagglutination test for Legionella pneumophila serogroups 1 to 4

0095-1137/81/061040-06$02.00/0

Evaluation of an Indirect

Hemagglutination

Test

for

Legionella

pneumophila

Serogroups

1

to

4

CHERYL A. YONKE,' HARLAN E. STIEFEL,2 DAVID L.WILSON,3 AND BERTTINA B.

WENTWORTH2*

DepartmentofMicrobiologyand PublicHealth,MichiganStateUniversity,EastLansing,

Michigan

48824'; Bureauof Disease Control andLaboratoryServices, MichiganDepartment ofPublicHealth,Lansing,

Michigan 489092; and DifcoLaboratories, Detroit,Michigan482013

Received 10 December1980/Accepted 24February1981

Paralleltestingof895 seraby indirecthemagglutination andindirect fluores-cent-antibodytechniquesshowed 97.3%agreement.Althoughthe indirect hemag-glutination technique usually showed more

cross-reactivity

among serogroups

than the indirect fluorescent-antibody technique with Formalin-fixed antigens

anda conjugate which detected primarily

immunoglobulin

G

antibodies,

heter-ologous serogroup reactionswere

significantly

lower than

homologous

serogroup

titers and theetiological serogroup couldbe easily defined.Theindirect hemag-glutinationtechnique showed nocross-reactivity witha crudeextract of Esche-richiacoli013:K92:H4. Since the indirecthemagglutinationtechniquewasshown

to detect both immunoglobulin M and immunoglobulin G antibodies and was found to berapid,simple,andinexpensive,itappearstobeanexcellentalternative

tothe indirect

fluorescent-antibody

testfor

serodiagnosis

oflegionellosis.

The indirect fluorescent-antibody (IFA) test

forlegionellosis (10, 15) hasbecome an

impor-tantdiagnostictool in the 3yearssinceits

intro-duction.

Performing

thetest on aroutinebasis,

however, is becoming an increasingly complex

task; with the recent addition of the agents TATLOCK (13), WIGA (1), and NY 23 (4) to

the genus Legionella(proposeddesignationsL. micdadei [9], L. bozemanii, and L. dumoffii

[2], respectively), proper serodiagnosis of

Le-gionnaires disease involvesthe useof nine

sep-arate antigens. The use ofpolyvalent antigens

(7, 15) can simplify testing somewhat, buteven

this doesnotdiscountsome inherent disadvan-tages of thetechnique, i.e., it istechnically dif-ficult and expensive to perform. As a practical

alternative, Edson and associates (5) have

intro-ducedanindirecthemagglutination (IHA) tech-nique for legionellosis. The test is rapid, inex-pensive, and simple to perform and shows a 94.9% agreement with IFA serogroup 1 results. The IHA test has recently been expanded to includeL.pneumophila serogroups 2, 3, and 4, and thisstudywas madetoevaluate the

sensi-tivityandspecificityof themethod.

MATERIALS AND METHODS Sera forparallel testing ofIHAand IFA

tech-niques. Serawerereceived at theMichigan

Depart-ment ofPublic Health (Lansing, Mich.) for routine

IFAlegionellosistestingbetween MarchandOctober

1979.Ailserareceivedduringthisperiodwereincluded

inthestudy exceptforthose in whichthequantityof serum wasinsufficient for testing by bothtechniques. The sampletestedincludedatotalof895sera.There were single serum specimens from 483 patients and

pairedorserialserafrom 188patients.

Special sera selected for comparison of IHA andIFAtechniques. In additiontothe above-men-tioned, unselected sera, sera from 55 patients with

IFA-confirmed (fourfold or greater rise in titer) or presumptive (titer of -1:128) legionellosis were

se-lected for comparative studies. Single serum

speci-mens were available from 8patients, and paired or

serialsera wereavailablefrom47patients. Thesecases

represent 41% ofthe 134 cases oflegionellosis

diag-nosedby serology betweenOctober 1977 andAugust

1980. The onlyadditional criterionforinclusioninthe studywassufficient serum volumefortesting.

Sera for cross absorption andblocking fluid studies.The 13 sera withserogroup 1titers usedfor thesestudies(seeTable 6) were fromIFA-confirmed orpresumptivecases.Of the12serawithEscherichia colititers,4 werefromLegionnaires diseasepatients,

and 8werefrom patients without legionellosis.

Sera for sucrose density gradient fractiona-tion. Twelve sera selected from the 55 patients

de-scribed above were used. An additional three sera whichshowed fourfoldorgreater rises or titers of.-1:

128by IHA and were IFA negative werefractionated foranalysis.

Bacterial strains.Bacterial strains were grownin Rouxbottlesonenrichedchocolateagar (GCmedium base with heated sheep blood and vitamin supple-ments). The L. pneumophila Flint 1, Flint 2, and Detroit1strains weremaintained as stock culturesat the Michigan Department of Public Health. The L. 1040

on February 7, 2020 by guest

http://jcm.asm.org/

IHA FOR L. PNEUMOPHILA 1041 pneumophila Knoxville 1, Togus 1, Bloomington 2,

and Los Angeles 1 strains and E. coli 013:K92:H4

were obtained from the Centers for Disease Control

(Atlanta, Ga.).

IFAantigenpreparation. Bacterialgrowthwas

washed fromthesurfaceof the Rouxbottle after72h of incubation with 0.4% formalinized phosphate-buffered saline, filtered through a fine-mesh wire screentoremoveagardebris,andthenrefrigeratedat 4°Cfor1week.Thesuspensionwasthen washed three

timesinformalinizedphosphate-buffered saline, and

the opacity was standardized to 300 international

opacityunitsperml by comparisonwithWorld Health

Organization opacity standards. Suspensions of for-malinizedwholeorganisms,adjustedto2international opacityunitsperml,servedastheantigen.

IHAantigenpreparation. AllIHAantigenswere

suppliedasexperimental antigens byDifco

Laborato-ries(Detroit, Mich.).Themethodofpreparationwas

similar to that of Edson et al. (5) for other IHA

antigens. The antigens consisted ofturkey

erythro-cytesstabilizedbyglutaraldehydetreatmentand sen-sitized withboiled,sonicatedantigenfromagar-grown

isolatesbythebis-diazotized benzidinetechnique.

An-tigenswerepreparedfromthefollowingstrains: sero-group1 antigenas amixture ofFlint 1,Flint2, and

Detroit1;serogroup2fromTogus1;serogroup3 from

Bloomington2; andserogroup4fromLosAngeles1. Thepolyvalent antigenwasmadefromamixtureof

Knoxville1,Togus1, Bloomington 2,and LosAngeles 1.

IFA technique. A standard microtiter technique

with0.005-ml amounts ofantigenon12-well, acetone-resistantglassmicroscope slideswasemployed.

Con-jugatesweredilutedinphosphate-bufferedsalinewith 2% Tween 80. Rabbitanti-humanglobulin conjugate

waspurchasedfromBeckmanInstruments,Inc.

(Ful-lerton,Calif.).Anti-rabbitglobulin conjugate (Cappel Laboratories, Downington, Pa.)wasusedto determine titers on hyperimmune rabbit antisera. Anti-human

immunoglobulinM (IgM) conjugatewassupplied by

Burroughs Wellcome Co. (Research Triangle Park, N.C.).

IRAtechnique. TheIHA techniquewassimilar to

that ofEdson etal.(5). Unless otherwisenoted,sera werescreenedataninitial1:16dilutionagainst

poly-valent sensitized cellsand unsensitized controlcells.

Anyserumshowingareactionwith the testcellsbut notwiththecontrolcellswassubsequentlyscreened

withthe four monovalentantigens (serogroups1 to4)

andcorrespondingcontrolcells.Seraweretitrated to endpointinserial,twofold dilutions withappropriate

monovalentantigens.

Anyserumshowingagglutination ofboth

antigen-sensitizedand unsensitized controlcellswastreatedin the followingmanner. A 0.02-ml amountofpacked,

unsensitized (control) cells was added to0.12 mlof serum. After overnight incubation at 4°C, the cell suspension wascentrifuged at3,000rpmfor 15min, the absorbedserum wasremoved,and the IHA test

wasrepeated. Sera withhightitersofanti-cell anti-bodyrequiredtwoorthree successiveabsorptionsfor completeremoval of theseagglutinins.

Crossabsorption.Forabsorption,0.1mlofserum

was added to approximately 3.7 x 1010 organisms

which had been washed three times in phosphate-buffered saline to remove Formalin. The absorption mixture was incubated overnight at 4°C and centri-fuged at 3,000 rpm for 15 min, and the absorbed serum wasremoved.

Preparation of hyperimmune sera. Rabbits weighing >2 kg each were immunized with an initial intramuscular injection of1ml of antigen mixed with

anequal amount of Freund complete adjuvant, fol-lowed by five intravenous boosters of1ml of antigen

atdays 22, 24, 26, 28, and 30. Rabbits were exsangui-nated 9 days after the last booster.

Comparisonof immune sera preparedagainst IFA and IHA antigens. Rabbits were divided into four groups of two rabbits each and immunized as described above. Groups1and 3received injections of

fornalinizedIFAantigens at 16international opacity units per ml. Groups 2 and 4 received injections of IHA antigen(boiled and sonicated) at 16 international opacity units per ml. Groups1and 2 wereinoculated with the Flint1 strain of L.pneumophila. Groups 3 and4 were immunized with the Detroit 1strain of serogroup 1.Sera from two rabbits in each immuni-zation group were pooled, and IFA and IHA titers were thendetermined on each of the pools.

Blocking fluid preparation. Blocking fluids of the Flint 1 strain and E. coli 013:K92:H4 were pre-pared in the manner described by Wilkinson et al. (14). Sera were initially diluted in IHA diluent or

phosphate-buffered saline as a control and in both blocking fluids. Subsequent dilutions were made in diluent or phosphate-buffered saline, and IHA or IFA titers were determined.

Sucrose density gradients.Sera were fraction-atedon 10 to 50% sucrose density gradients (12) by centrifugation at 33,500 rpm for 17 h in a Spinco ultracentrifuge with anSW50.1 rotor. Ten0.5-mi frac-tions werecollected for analysis.

lmmunodiffusion.The presence of IgM or IgG or

both in each of the samples was confirmed with radial immunodiffusion kits (Meloy Laboratories, Spring-field, Va.). These kits employ the Fahey and Mc-Kelvey (6) techniqueto measureimmunoglobulins.

RESULTS

Parallel testing of 895 sera by the IHA

and IFA

techniques.

Results of

parallel

testing

of 895 sera

by

theIHAandIFA

techniques

are

shown inTable 1. There were 796nonreactive

TABLE 1. Resultsofparallel testingbythe IHAand IFAtechniqueson 895 sera

No.ofserainserogroup: Results oftesting

1 2 3 4

IHAa

andIFA- 796 879 885 858

IHA+bandIFA+ 46 1 5 15

IHA-and IFA+ 14 0 2 2

IHA+andIFA- 39 15 3 20

a_ Titerof<1:16.

b+,Titer of

.1:16.

VOL.13,1981

on February 7, 2020 by guest

http://jcm.asm.org/

sera and 46 sera which reacted in both

tech-niques inserogroup 1. Therewere14sera

reac-tiveonly in the IFA technique whichwere

non-reactiveinthe IHAtechnique ata1:16dilution. There were 39 sera reactive only in the IHA

technique that failedto react in the IFA tech-niqueata1:16 dilution. Discrepanciesbetween

IFA and IHA reactivities withserogroups2, 3,

and 4 were due largely to the cross-reactive nature of the serogroup 1 antibodyand not to

serawithspecific serogroup2, 3,or4 antibody.

Agreement (percentage ofserapositive byboth

techniques plus percentage ofseranegative by

bothtechniques) was94.1, 98.3, 99.4, and 97.5%

forserogroups1,2, 3,and4, respectively. There was anoverall agreementforserogroups 1 to4

of 97.3%.

Among the 188 paired or serial sera, there

were nine instances of fourfold or greater

in-creases in titer and four discrepancies byboth

theIHAandIFAtechniques. Onepatientwith

afourfoldrisein IFA titer showedatwofoldrise

byIHA,anotherpatientshowed thereverse(i.e., afourfoldrisebyIHA andatwofoldrisebythe

IFAtechnique). Another patientwithagreater

than fourfoldincreasebythe IFAtechniquehad

a high stable titer (1:2,048) by IHA. Finally, serumfromapatientwith successive IFAtiters

of1:256and 1:128 inserogroup 1wasIHA

neg-ative.

Some of the diverse cross-reactivity among serogroups of the human antibody response

notedby Wilkinsonetal. (15) in the IFA tech-nique was observed with the IHA technique.

However, with Formalin-fixed antigens, the IFA technique didnotappeartobesocross-reactive

inthisstudy

(Table

2).

When cross-reactivity wasseen,itwasfound

that theetiologicalserogroup could bestbe de-finedbycrossabsorption. Absorption ofa serum

with a serogroup-specific antigen removed the

cross-reacting antibody aswell as the

homolo-gous antibody. Absorption witha heterologous

serogroupantigen removedonlythe cross-react-TABLE 2. Diverse serogroup reactivity of the human

antibody response Convalescentserum titer

Sero-group an- Serum1 Serum2 Serum3 tigen

IFA IHA IFA IHA IFA IHA

1 64a 512 64 16 128 2,048

2 __b _c - 256 - 32

3 - - 512 512 -

-4 - 32 16 128 8 128

aReciprocalofhighest

dilution

showing reactivity. bIFA titer of <1:8.

cIHA titer of <1:16.

ing antibody but didnotdecrease the

serogroup-specifictiter. Tencasesofserogroup 1 andone case of serogroup 4 were identified by cross

absorption. Two examplesareshown inTable3.

Patient 1appears tohaveserogroup 1

legionel-losis,and patient 2appearstohaveLegionnaires

disease ofserogroup4etiology.

Comparison of IHA and IFA techniques

onselected serafrom 55 patients. Of the 55

patients with IFA-confirmedorpresumptive

le-gionellosis, 54 were identified as serogroup 1,

and 1 was identified as serogroup 3. The IFA

technique detected fourfold or greater rises in titer in 42 patients,41of whomwereinserogroup

1 and 1 of whom was in serogroup 3. Ofthese

patients,37(36 inserogroup 1and1inserogroup

3) also showed significant rises in titer by the IHA technique (Table 4). In three of the

re-maining five patients, the IHA technique showed a rise from <1:16 to 1:16, and in two cases the IHA technique was nonreactive. Of

13 patients with presumptive legionellosis, 7

showed IHA titers of -1:128, 1 showed a rise

from <1:16to1:16, 2showed titers between 1:16

and 1:128, and 3 were nonreactive in the IHA

technique (Table 4).

TABLE 3. Crossabsorption to define etiology Serum IHAtiter of

Patient Serum serogroup:

1 2 3 4

1 Unabsorbed 256a - - 64

Absorbedwith Flint 1 _b - -

-(serogroup 1)

Absorbed with LosAn- 256 - -

-geles1(serogroup4)

2 Unabsorbed 256 - - 2,048

Absorbed with Flint1 - - - 2,048 (serogroup1)

Absorbed with LosAn- - - -

-geles1(serogroup 4)

Reciprocal of highest dilutionshowingreactivity. bTiter of<1:16.

TABLE 4. Comparison of IFAandIHAresults from

55patients with IFA-confirmed or presumptive legionellosis

IHA(no.of patients) IFA Pre-

~~Change

in Sain

IA Con-

Pr-titer

from

Stto-Nea

firmedA _tivebP negative' tivec

t 6 ters,

Confirmed 37 O 3 O 2

Presumptive 0 7 1 2 3

a Fourfold or greater rise in titer.

bTiter of -1:128. Titer of <1:16.

dReciprocalofhighest dilutionshowingreactivity. e Between 1:16 and 1:128.

on February 7, 2020 by guest

http://jcm.asm.org/

IHA FOR L. PNEUMOPHILA 1043

Itwasconsidered that differencesbetween the

resultsfrom the IHA and IFA techniquesmight

beexplained by: (i) measurement oftwo

differ-entantigen-antibodysystems,(üi)reactionswith

across-reactingantigen(s),or (iii)differencesin

immunoglobulin specificity. Toexplore each of

these possibilities, the studies discussed below

wereconducted.

Comparison of immune sera prepared

against

IFA and IHA antigens. Results of

IFA and IHA testing of hyperimmune rabbit

sera that were prepared with IFA and IHA

antigens areshowninTable5. Iftwodifferent

antigen-antibody

systems are involved, titers

withhomologousantigens might beexpected to

behigher than those with

heterologous

antigens.

However, therewere no significant differences

intiters withIFAandIHAantigens, regardless

of theantigen usedtopreparethe antisera.

Blocking fluid andcross

absorption

stud-ies. Wilkinson and associates reported that a

crude extract of E. coli 013:K92:H4 can block

97% ofIFA-positive reactions withavariety of

gram-negativebacterial species and block6% of

L.pneumophila titers (14). ThisstrainofE.coli

wasobtained from H. W.Wilkinson, and

block-ing fluids of it and the Flint 1strain were

pre-pared.

Twelveserawith

anti-serogroup

1 IHAtiters

were tested by comparing titrations done with

blocking

fluid as diluent with titrations done with

regular

test diluent (Table 6). Dilution in

E. coli

blocking

fluid hadno

significant

effecton

anyofthe 12sera

tested,

whereasFlint1

block-ingfluid causedafourfoldorgreaterdecrease in

all of theseratested.

Blocking fluid studies by theIFA technique

showed similar results. E. coli

blocking

fluid had

noeffectonthe 13seratested.Dilution in Flint

1blocking fluid reducedserogroup 1titers

by

at

leastfourfoldin100% of thecasestested.Twelve

serawith initial IFAE.colititerswere

similarly

tested

(Table 6).

Flint1

blocking

fluid causedno

TABLE 5. Comparisonofantiserapreparedagainst

IFA and IHAantigens

Titer'by:

tiommunia

Immunizing antigen

tiongroup IHA IFA

1 Flint1,IFA 4,096b 2,048c

2 Flint1,IHA 4,096 2,048c 3 Detroit1,IFA 4,096 2,048d

4 Detroit1, IHA 8,192 4,096d

a

Results

shown are theaverage valuesoftriplicate determinations.

bReciprocal ofhighestdilutionshowingreactivity. cFlint1servedasthetestantigen.

dDetroit1servedasthetestantigen.

TABLE 6. Results of blocking fluid and cross-absorptionstudies

Test

IHA IHA IHA IHA IFA IFA IFA IFA IFA IFA IFA IFA

Initialserum titer to: Serogroup1

Serogroup1

Serogroup1

Serogroup1

Flint1

Flint1

Flint1

Flint1

E.coli E.coli E. coli E.coli

BF'orAbs.b used Flint1BF E.coli BF Flint1 Abs.

E.coli Abs. Flint1BF E.coli BF Flint1Abs. E. coli Abs. Flint1BF E.coli BF Flint1Abs. E.coli Abs.

Decrease in titer' 12/12

0/12 11/11 0/11 13/13 0/13 9/9 0/9 0/12 12/12 1/8

8/8

aBF, Blocking fluid used for initial

dilution

of

se-rum.

bAbs., Whole-cellantigen usedtoabsorb sera.

CFourfoldorgreater decrease in titer.

significant decreases in titer in any of the 12

sera,whereas E. coliblocking fluid significantly

decreased titersinallof thesera. There was one

patientwho showed a concurrent fourfold rise in

titer to both E. coli and the Flint 1 strain in

whichblocking was specific,i.e.,Flint 1 blocking

fluid reduced only the Flint 1 titer, and E. coli

blocking fluid reduced only the E. coli titer.

Titersin another patientwho showed a fourfold

rise in serogroup 1 titer to 1:128 by the IFA

techniqueand a maximumIHAtiter of 1:16 were

blocked by Flint1 blocking fluidbut not by E.

coli

blocking

fluid.

Cross absorptionwasfelttobe a more

conclu-sivetestfor common antigenthan was the use

ofcrudeblocking fluidextract.Absorption of11

serathat hadserogroup 1 IHAtiters with

whole-cell E. coli antigen caused no significant

de-creasein any serogroup 1titers, whereas

absorp-tionwiththewhole-cell Flint 1 strain caused a

fourfoldor greaterdecrease inall titers (Table

6). Similarly, in the IFAtechnique, absorption

withwhole-cell E. coli did not change Flint 1

titers in nine sera, whereasabsorptionwith the

Flint 1 strain significantly decreased all of the

titers. Crossabsorptionstudiesofeightserawith

E.coliIFA titers

proved specific

for

all

butone

serum.Thisserumhad aninitialE. colititerof 1:32, andabsorptionwith E. colidecreasedthe

titer to <1:8. Absorption with the whole-cell

Flint 1 strain decreased the titer fourfoldto 1:8.

Fractionation of sera on sucrose

gra-dients to determine the

immunoglobulin

class reactive in theIRAand IFAtests. A

total of 15 serawerefractionated

by

thesucrose

density gradient

technique,

and the fractions

wereanalyzed byboththe IFA and IHA

meth-ods (Table 7).Threeserafromcases

diagnosed

VOL.13,1981

on February 7, 2020 by guest

http://jcm.asm.org/

1044 YONKE ET AL.

TABLE 7. Immunoglobulin reactivity of15sera fractionatedon sucrosedensitygradients

No. of sera withreactivityin: Technique

IgMfractions IgG fractions

IFA O 12

IHA 13 il

TABLE 8. ComparisonofIHA and IFA resultson 10sucrosedensity gradientfractions

Immunodiffu-Fraction IHA IFAa IFA-Mb sion'

no.

IgM IgG

1 8d _e - _ _

2 32 - 2 +

-3 512 - 32 +

-4 128 - 8 +

-5 32 4 - - +

6 32 16 - - +

7 4 32 - - +

8 - 2 - - +

9 - _

-10 - -

-aAnti-humanglobulin

conjugate.

bAnti-humanIgM

conjugate.

'Presence (+) or absence (-) ofIgM and IgG in

eachfraction.

dReciprocalofhighestdilutionshowingreactivity.

e_,

Noreaction with

undiluted

sample.

by the IHAtechnique butnotbythe IFA

tech-nique showed reactivity in only the IgM

frac-tions. Tenserashowed

reactivity

intheIgMand

IgGfractionsbythe IHAtestbut showed

reac-tivityinonlythe

IgG

fractionsbythe IFAtest.

Another serum showed reactivity in the IgG

fractions by both techniques. The remaining serumshowedreactivityintheIgGfractionsby

the IFAtestbutwasnonreactive inallfractions

bythe IHAtest.Thisserumhadaninitial IFA titerof1:128(serogroup 1) andwasnonreactive by the IHA test.Itappearedthatthe standard

antiglobulin conjugate used in the IFA

tech-nique did not adequately detect IgM. To test

this hypothesis, the fractions of 10 sera which

showed the presence of IgMby theIHAtestbut

notby the IFAtestwerereexamined by the IFA

test with an IgM-specific conjugate. All sera

thenshowedreactivity in the IgM fractions (Ta-ble8).

DISCUSSION

Parallel testing of 895 sera showed that the

IHA and IFA testsyielded essentially the same resultsfor serogroups1to4.There was approx-imately 97% overall agreement between the two techniques,anddiscrepancies were observed in both directions. IHA-positive andIFA-negative

discrepancieswereassociatedwiththeinability

to detect IgM with the antiglobulin conjugate

used in thisstudy.Nagingtonandassociates(11)

reported that 3 of 22 Legionnaires disease pa-tientsshowed arisein IgM antibody withouta

concurrent rise in IgG antibody. Such cases would go undetected with a conjugate of this type. Polyvalent immunoglobulin conjugates

should be usedformaximumsensitivity(11, 14). Thisrequirement posesnoproblem intheIHA technique, asit wasconclusively showntodetect

IgMandIgG antibodies.

The problem ofIFA-positive and IHA-nega-tive sera is yettobe resolved. Onesuch serum

showedreactivityinonlytheIgGfractions. Con-tinued fractionation of sera ofthis naturemay provide an explanation for the differences be-tweenthe two techniques.

Correlation of clinical data with laboratory

dataprovides furtherinsightintothesensitivity

andspecificityof thetwotechniques.Twocases

of Legionnaires disease defined by the IHA which wereundiagnosed by the IFAtechnique hadclinical symptoms consistent with

legionel-losis and responded promptly to erythromycin

therapy. Among six patients diagnosed by the

IFA test but not by IHA, three had clinical

histories consistent with Legionnaires disease

andrespondedtotreatmentwitherythromycin;

one patient had symptoms consistent with the disease but responded well to treatment with

gentamicin and cephalothin, two antibiotics

which are of low efficacy in the treatment of legionellosis. Clinical histories of the two re-maining patients were not consistent with the

diagnosis made by IFA serology. Clinical data

were notavailable onthe remaining

discrepan-ciesdescribedinthis study.

Even when cross absorption was used in ad-ditiontothecruderblocking technique, the IHA test showed no significant cross-reactivity with E. coli. The IFA test showed evidence of cross-reactivity with E. coli in only one instance, but it should be noted that this serum had a low initial titer. Studies with 12 common gram-neg-ative bacterialantigens,including Haemophilus, showed nocross-reactivity in the IFA test with Legionella antigens (B. Wentworth, personal communication). Cherry and

associates

(3) found that of 400 bacterial strains tested, one strain ofPseudomonas fluorescens is antigeni-cally related to L.pneumophila serogroup 1.

Although the IHA test showed more cross-reactivity among serogroups than did the IFA

testwith

Formalin-fixed

antigens, heterologous reactivity is usuallysignificantly lower than ho-mologous serogroup reactivity. The IHA test adequately defined the etiological serogroup in

on February 7, 2020 by guest

http://jcm.asm.org/

most cases. If therewasanyquestionas tothe

etiological

serogroup, crossabsorption could be

usedtodefine it.

Microagglutination andmicro-enzyme-linked

immunosorbentassayshave alsobeendescribed

for the detection of Legionella antibodies (8).

However, microagglutination is less sensitive

thantheIFAtest,and the

micro-enzyme-linked

immunosorbent assaytechnique, although

sen-sitive, ismoreexpensive and difficulttoperform

than theIHA

technique.

The IHAtechnique is rapid,

simple,

inexpen-sive, and well suited to

large-scale

studies and

routineuse.Itprovidesasuitable alternativeto

the useof the morecomplicated and expensive

IFAtechnique.

LITERATURE CrIED

1.Bozeman, F.N.,J. W.Humphries,and J. M. Camp-bell. 1968. A new group of rickettsia-like agents re-covered from guineapigs.ActaVirol. (Engl.Ed.) 12: 87-93.

2. Brenner,D.J.,A. G.Steigerwalt,G. W.Gorman,R. E.Weaver, J. C.Feeley, C. Patton,B. M. Thoma-son,andK.R.Lewallen-Sasseville.1980.Legionella bozemaniispeciesnovaandLegionelladumofflispecies nova:classification of two additionalspeciesof

Legion-elaassociated with humanpneumonia. Curr. Micro-biol. 4:115-120.

3. Cherry, W.B.,B.Pittman,P. P.Harris,G. A.Hebert, B.M.Thomason, L.Thacker, and R. E. Weaver. 1978. Detection ofLegionnaires disease bacteria by directinmunofluorescent staining.J.Clin.Microbiol. 8:329-338.

4. Cordes, LG.,H. W.Wilkinson,G. W.Gorman,B. J. Fikes,and D. W. Fraser.1979.Atypical Legionella-like organisms: fastidious water associated bacteria pathogenictoman.Lancetii:927-930.

5. Edson,D.C.,H. E.Stiefel,B. B.Wentworth, andD. L Wilson. 1979. Prevalence of antibodiesto

Legion-naires disease: aseroepidemiologic survey of Michigan residentsusing thehemagglutinationtest. Ann. Intern. Med. 90:691-693.

6.Fahey,J. L, and E. M. McKelvey. 1965. Quantitative determination of serumiemmunoglobulinsin antibody agarplates. J. Immunol. 94:84-90.

7.Fallon, R. J., and W. H. Abraham. 1978. Detecting Legionnaires disease. Lancet ii:1318.

8. Farshy,C. E., G. C. Klein, and J. C. Feeley. 1978. Detection of antibodiesto Legionnairesdisease organ-ismby microagglutination and micro-enzyme-linked im-munosorbent assay tests. J.Clin. Microbiol. 7:327-331. 9. Hebert,G. A., A. G. Steigerwalt, and D. J. Brenner. 1980. Legionella micdadeispecies nova: classification ofa third species of Legionella associated with human pneumonia.Curr. Microbiol.3:255-257.

10. MeDade,J. E., C. C. Shepard,D. W. Fraser, T. F. Tsai, M. A.Redus,W. R.Dowdle, and the Labo-ratoryInvestigation Team. 1977. Legionnaires dis-ease: isolation of a bacteriumanddemonstrationof its role in other respiratory disease. N.Engl. J. Med. 297: 1197-1203.

11. Nagington,J.,and T.G.Wreghitt. 1979. The antibody response in Legionnaires disease. J. Hyg. 83:377-381. 12. Palmer, D. F., J. J. Calavero, K. Herrman, J. A.

Stewart, and K. W. Walls. 1974. Serodiagnosis of: toxoplasmosis, rubella, cytomegalic inclusion disease, herpessimplex, p. 101-110. Immunology Series no. 5. U.S. Department of Health, Education and Welfare, Atlanta, Ga.

13.Tatlock, H. 1944. Arickettsia-like organism recovered from guinea pigs. Proc. Soc. Exp. Biol. Med. 57:95-99. 14. Wilkinson,H. W., C. E.Farshy, B. J. Fikes, D. D.

Cruce,and L P.Yealy. 1979. Measure of immuno-globulin G-, M-, andA-specifictiters against Legionella pneumophila and inhibition of titersagainst nonspe-cific, gram-negative bacterialantigens in the indirect immunofluorescence test forlegionellosis.J.Clin. Mi-crobiol.10:685-689.

15. Wilkinson,H.W.,B. J.Fikes,and D. D.Cruce.1979. Indirect immunofluorescence test forserodiagnosis of Legionnairesdisease: evidence for serogroupdiversity ofLegionnairesdiseasebacterial antigens and for mul-tiple specificity of humanantibodies.J. Clin. Microbiol. 9:379-383.

13,

on February 7, 2020 by guest

http://jcm.asm.org/