Single radial hemolysis as a cost effective determinant of Rubella antibody status

(1)

0095-1137/79/01-0115/05$02.00/0

Single-Radial Hemolysis

as a

Cost-Effective

Determinant of

Rubella

Antibody Status

JOHN M. FORGERIII* AND ROBERT F. GILFILLAN

Virology Laboratory, State Laboratory Institute,Massachusetts DepartmentofPublic Health, Boston,

Massachusetts 02130

Received forpublication6October 1978

Single-radial hemolysis was examined for sensitivity, reliability, and cost for

determination ofrubella antibody levels in the general population. Results ob-tained with single-radial hemolysis plates made in this laboratory, and those under development by a commercial manufacturer, were compared with those

obtained by the hemagglutination inhibition method normally used for rubella antibody determinations. The results suggest that single-radial hemolysis is comparabletohemagglutination inhibition withrespect tosensitivity and

relia-bility; it is superiortohemagglutination inhibition intermsofeaseofmanipulation

andeconomyofuse.

Single-radial hemolysis (SRH) isasimple and inexpensive method for the determination in serumof thepresenceandamount ofantibody to specific viral antigens. First described by Weileretal. (9), thistechnique consistsof sus-pending in agarose erythrocytes (RBC) coated with the antigens of interest. The antiserum undertestisthenaddedto acircularwell in the solidifiedagarose.Afterasuitableperiod of time forantibodydiffusion, complement (C')isadded; thecombination withantibody andcomplement causes lysis of the antigen-sensitizedcells. The resultis acircular areaofhemolysis around the well, contrasting sharply with the surrountding intactRBC and varyingin sizeaccordingtothe amountofantibodypresent in theserumsample added (3,5, 7).

Limited experiments have shown that SRH can be used to detect and measure antibody againstavariety of viralantigens including ru-bella (2). The technique has been used exten-sively inseroepidemiological studies of influenza (7). It has not, however, received widespread attentionas adiagnostic method for large-scale applicationtorubella.

Thepurposeof thisstudywas,first,to assess thesensitivity, accuracy,andreliability of SRH plates prepared in this laboratory (with com-mercially available components)toplates under development and prepared by a commercial manufacturer. Second, thisstudywas designed tocompare the costeffectiveness of both labo-ratory andcommercially preparedplates to the standardized hemagglutination inhibition (HI) test(6)for:(i) thelaboratory diagnosisof rubella infection and(ii)the determinationof immunity in candidatesfor vaccine. The present

commu-nication describes results obtained from this study andpresentsevidencetoshow that SRH compares favorably with the HItest in identi-fying recent rubella infections and candidates for vaccination.

MATERIALS AND METHODS

SRHprocedure.The SRH plateswere made

es-sentially according to the methods of Griliner and

Strannegard (2). Lyophilized rubella antigen (Flow

Laboratories, Bethesda,Md.) wasreconstituted to 1

mlwithdistilledwater; thehemagglutinationtiterwas 1:256byastandardtest(6). Thisantigenwasusedto

sensitize sheep RBC as follows: a 10% suspension

(packedRBC volume[milliliters]/milliliterofantigen)

ofcellswasmade in reconstitutedantigenand allowed

to stand for 10minat roomtemperature. After

cen-trifugation at6,000xgfor5min, the sensitized cells

wereresuspendedat00Cin5 x 10' M sodium,

5,5'-diethyl barbiturate buffer, pH 7.3, containing 0.14 M

NaCl, 5 x 10-4 M MgCl2, and 5 x 10' M CaCl2.

Suspendedcellswerecentrifugedasabove and

resus-pended at a concentration of 10% (vol/vol) in the

barbiturate buffer(00C).Agarose (L'Industrie Biologic

Francaise, S.A.) was dissolvedat a concentration of

1.5%(wt/vol) byboilingin phosphate-buffered saline,

pH 7.2,towhich 0.1%(wt/vol) sodium azide had been

addedas apreservative. The molten agarose was then

distributed into 5-ml glass tubes (2.7 ml/tube) and

cooled for450Cinawaterbath.A0.3-ml amount of

the 10% RBCsuspensionwasadded quickly to each

tube of agarose, and the contents of the tube were

mixed vigorously and poured into empty diffusion

plates(Hyland Laboratories, Costa Mesa, Calif;

cata-logno. 085-710) to cool. After solidification, wells 2

mmin diameter were punchedinthe RBC/agarose,

and0.02ml of thetestorcontrolsera(heat inactivated

at560Cfor30min)wasaddedtoeachwell. Theplates

were then allowed tostandfor 18 to24h at40C.At

the end ofthistime, 2.0 ml ofguinea pig serum (a

115

on February 7, 2020 by guest

http://jcm.asm.org/

(2)

116 FORGER AND GILFILLAN

sourceof C') diluted 1:10 in barbiturate buffer (above) was pipetted over the surface of the gel.

Lyoph-ilized/reconstitutedguinea pig sera used in this study

hadcomplementactivities of 280±3050% hemolytic

complement units/ml. Theplatesthen stoodfor3 to

6h at37°C; after removal of theexcessC', diameters

of thehemolytic zonesweremeasured tothenearest

0.5mm.

Commerciallyprepared SRH plates. For

com-parative studies, rubella hemagglutinin-sensitized

sheep RBC SRH plates were obtained through the

courtesy of Norman Finter, Weilcome Laboratories,

Kent,England,along with supplies ofstrong andweak

positive andnegative control sera. Control and test

seraweretestedasdescribedabove.

HI. HI tests were carried out by a standardized

method, using fresh human type 0 Rh(-) cells as

indicator (6).Testsera were notheat inactivated. All

test sera had been submitted to the laboratory for

immunitytests orfor serodiagnosis ofrubellainfection.

RBC. Sheep,chicken,andchick RBCwereobtained

fromlaboratory animals maintainedattheState

Lab-oratoryInstitute. Humantype0Rh(-)cells (6)were

obtained from the Red Cross Laboratories, Boston.

Cellswerecollectedaseptically and stored in Alsever

solution (6)at4°C.Sheepcells remained usable for6

weeks after collection; chicken, chick, andhuman 0

cells couldbe storedfor3weeks.

Serumsamples for SRH.Inexperiments designed

todetermine thesensitivity andaccuracyof theSRH

method,serumsampleswereselected,grouped

accord-ingtocategories, andtestedunblinded (Table 1; Fig.

1).Afterexploration and developmentof the SRH test

presentedhere, sampleswereselected and tested blind

by code(Tables2and 3).

RESULTS

Theoptimal values ofseveral variableswere

examined with a view toward balancing

maxi-mumeconomy ofreagentswith sensitivityand reliability.We studiedfinal RBC concentration in thegel,the RBC/antigen ratioduring sensi-tization, the antigen concentration during sen-sitization, and theminimumamountof

comple-mentneeded foraninterpretable result. Concentration of sheep RBC. RBC are

commonly usedat aconcentrationof 1% in the gel. Since some economy can be realized by using fewer cells, weexamined final concentra-tions of 0.83, 0.66, 0.50, 0.33, and 0.1%. Using triplicate determinations with several different antisera, wefoundnoeffect on the diameter of the hemolytic zonesas afunction of RBC

con-centration. The clarity of the zones, however, changed noticeably; asthe redbackground de-creased, it became increasingly difficult to dis-cern hemolysis. Below 0.33% the contrast was too small to be suitable for routine diagnostic

use. Thus,final RBC concentrations of greater than0.5%(generally 1%)wereusedfor the

pres-entstudy.

Ratio of RBC to reconstituted antigen. The RBC/antigen ratio, i.e., packed RBC vol-ume(in milliliters) per milliliter of reconstituted antigen, was examined next. The ratios tested

were0.1/0.9(as cited in Materials and Methods), 0.2/0.8, 0.3/0.7, 0.4/0.6, and 0.5/0.5. Three dif-ferentantisera (one each ofhigh, intermediate, and low titer) weretested intriplicatefor each ratio. We observed that variation in the ratio had no effect on the measured hemolytic zone diameter. However, the opacity of the zone de-creased asthe amount ofantigenper RBC de-creased. The observable difference between hemolyzed and intact cells on the plates was unacceptable for routineuse asratios of0.4/0.6 and 0.5/0.5. However, a ratio of 0.3/0.7 was suitable andrepresentedaconsiderable savings inantigen, since0.7ml ofantigenyielded0.3ml ofusable sensitizedcells, whereasatthe 0.1/0.9 ratio (prominent in the literature and cited in Materials and Methods) a volume of 0.9 ml of antigenyieldedonly0.1ml ofusablecells.

Antigen concentration. Inasimilar experi-ment,wevariedtheantigen concentration while maintainingthesamefinalRBCpercentage(by volume) during sensitization. The antigen was

used at three different levels: (i) normal lx

reconstitution of the antigen (cf.Materials and Methods); (ii) a 10-fold dilution of the normal reconstitution; (iii)a10-fold concentration of the usual amount (achieved by hydratingthe anti-gen asusual, centrifugingthereconstituted

an-tigenat 105,000xgfor45min, andredissolving the precipitateinsupernatant fluid at 1/10 the starting volume).

Forty sera with HI titers of 1:8 were run at each antigen dilution. Diameters oflysis were

measured, converted into area, and compared for eachantigenconcentration.

Themean area ofthehemolyticzone didnot

vary significantly asafunction ofantigenlevel used during sensitization. Qualitatively, how-ever, thecontrast on the O.lx plates waspoor,

zones ofclearing were muchmore pronounced

atlx, andinthelOxplatesthezonesof hemol-ysis were extremely sharp and free ofunlysed

cells.

Concentration of complement.

Experi-ments werealso carriedout totestthe effect of

varying the concentration ofcomplement used. Thefollowingdilutions ofguinea pigserumwere

used:undiluted, 1:5, 1:10, 1:20, and1:40.In trip-licatetests on twodifferent antisera, represent-inghighand lowtiters,therespective diameters of the hemolytic zones did not change as a

function ofC'concentration. Theclarityof these zones, however, decreased with increasing C' dilution so that 1/10 was the highest dilution

J. CLIN. MICROBIOL.

http://jcm.asm.org/

(3)

which could be successfully employed for rou-tineuse.

Comparison of Massachusetts

Depart-ment of Public Health Virus Laboratory (VL) and commercial rubella SRH plates. Having determined optimal values for several variablesimportantin the preparation of SRH plates, it was then decided that a comparison between platesprepared in our laboratory and those availablecommerciallywould be in order. Ifthiscomparison proved favorable, considera-ble economic benefit wouldresult.

For the comparison, serum samples were se-lectedatrandom fromspecimens submitted to theStateLaboratory Institute Virus Laboratory fordeterminationofrubellaantibody status: 287 sera weretested on VL plates, and 279 of these sera were tested on commercial (Wellcome

[W])

plates.

Initially, thesamples were grouped by HI titer andrun onSRH plates. Theresulting data were convertedto areasandthe mean and standard errorof themeanwerecalculatedfor n samples in each group as shown in Table 1. These data are also plotted in Fig. 1, whichillustrates the linear relationships between the SRHareas and the logarithm of the HI titer;this linearity was seenindata from bothsetsof plates. The equa-tions for these relaequa-tionships assume the form: SRH (area) =KlogHI titer +K', where K is theslope of the line and K'is the y intercept.

Using the method of leastsquares, the respec-tive equations of the linesin Fig. 1 were deter-mined, along with the correlationcoefficient (r). Itisevident thatthe VL and W plates generate very similar equations; in both cases the lines are excellent representations of the respective data used toderive them (r=0.97 and 0.98 for the VL and Wplates,respectively).

Sensitivity

ofSRH versus HI. The equa-tions derived above suggest a general

relation-TABLE 1. ComparisonofhemolysisonVL and

commerciallyprepared(WR SRHplates

W VL

HI titer

na Area(mm2)b n Area(mm2)

<8 59 1.9±1.1 44 2.1±1.2

8 7 24.0±6.7 40 29.4±2.3

16 31 31.1±3.1 27 47.3±4.5 32 64 46.5+2.0 60 64.6 ± 2.7 64 54 61.9+2.1 58 83.0+2.7 128 44 69.8+2.4 39 84.8±1.4 256 13 87.1+4.1 12 101.8±4.3 512 7 85.0+6.8 7 91.7±2.7

aNumberofsamples testedateachrange of

anti-body level.

bMean hemolytic area + standard error of the mean.

- 80

E

E 70

cn

i>

60

0

w I 50

0

< 40

w

4r

20

10

0

o =VL

Area=16.7 log[HI]+2.3 r= 0.97

Area= 15log [HIJ-3.6 r=0.98

0 l 6

log [HI]

FIG. 1. Relationship ofthe SRHarea to the HI

titerforVL

and

Wplates.

ship between SRH andHI inthecaseofrubella antiserum titers.They do not,

however,

reveal anysystematicdifferencesinthe sensitivities of therespectivemethods. We

thought

it desirable to assesswhat(if any) proportionofserachosen at random (including acute,

convalescent,

and negative sera) and tested blind wouldgive neg-ative results whenscreenedbySRHversusthose whichwerenegativebyHI.

Table 2 gives the results on 293 specimens testedbyHIandSRH(bothVLandW).

Two specimens found positive by HI

con-tainednorubellaantibodymeasurable

by

SRH. Both sera negative by SRH had HI antibody titers of 1:16 and were acute specimens from diagnostic pairs. Differences between HI and SRH reactivitiesmaybe duetofalse-positiveHI testsor to lackof

immunoglobulin

M

antibody

activityinSRH (2,8). The sera werenot frac-tionated and tested for rubella-specific immu-noglobulinMantibody (6).Resultsfrom VL and Wplateswereessentiallythesame.

We also compared the relative efficiency of SRH and HI in determining recentrubella in-fections. Paired (acute/convalescent) serafrom

12 cases clinically diagnosed as rubella were

testedbyHI and onWplates (Table 3).In all

VOL. 9,1979

http://jcm.asm.org/

(4)

118 FORGER AND GILFILLAN

TABLE 2. Detection of antibody by SRH and HI Method Positive/totala % Positive

HI 252/293 86.0

SRH 250/293 85.3

aTotal number with detectable antibody/total tested.

instancesafourfold increase inantibody against rubellawas seenby HI. Byuseofthe W equation from Fig. 1, we determined theHI equivalents of the SRH reading foreach serum(Table 3).In all 12 pairs, fourfold or greater antibody titer rises werealso foundby SRH.

Specificity.TheHItestissubjectto mislead-ing results caused by nonspecific agglutinins whichmaybepresent inthe sample (6). When chickcellsareusedasindicators, theagglutinins areroutinely removed by adsorption of thesera with chickencells; if trypsinized humantype0 Rh(-) cells are used, prior absorption is not normally required (8). We tested85unadsorbed serumsamples containing chick cell agglutinins (at titers of1:8 orgreater)byHIwith human0 Rh(-) cells and by SRHonVLplates. By both methods, 63 rubella-positive samples were found, and22negatives.

DISCUSSION

Aneffortwasmadetostudy the SRHtestin orderto define those circumstances permitting maximum reliability and sensitivity with mini-mumexpenditure ofresources.Themost expen-sivecomponents of thetest arebloodcells, an-tigen, and C'. Itwasfound thatvaryingamounts and ratios of these reagents did not markedly affect the final values observed but did affect the ease ofmeasurement. At present, the cost per test for the components can be as low as

$0.16(12 sera/plate) foraneasily interpretable testcomparedto$1.15 foranHItest. All com-ponentsfor theHI test arebased onbidprices fromcommercialsources.

Somepreliminarytestsinourlaboratory sug-gest that SRHplates (both VL and W) havea shelf life of about2weeks.Thus, sufficientplates canbeprepared and standardized in 1dayand stored toanticipate short-term needs.

Themajor advantageof thistechniqueisthat only onewellintheplateandonesimple

mea-surement are required to titrate the antibody concentration of aserum. Once the plates are

made,thetechnician needonlyadd thesample, incubate, add C', and measure the hemolytic zone. Thus,many more testscanbecompleted thanbythemorecumbersomeconventional di-lutiontechniquesforantibodytitration.

Platesmade in this laboratory compared fa-vorably with plates under development by a

commercial source. For alargenumber of tests the effort in making the plates from starting materials isverysmallascomparedtothe effort of standard HI determinations. Thus, savings arerealized from two sources: (i) theinherent efficiency of the SRH techniqueversusthe HI procedure; (ii)theeconomyof using homemade plates as opposed to commercially produced ones.Thedisadvantage of SRHascomparedto HI is that SRH requires 24 h for completion, whereas theHI test can becompletedin 1 work-ingday.

Twoserawith HI titers of<1:8produced low levels ofhemolysis by SRH (Table 1).Whether reactivity in gel was a response to antibody missed byHI oris in factanSRHfalse-positive reactioncannotbedetermined from thepresent experiments. Serum samples with HI titers of <1:8 maycontain antibody inthe rangeof1:2 to 1:7 andnotbedetected by HI. It is not known whether antibodytorubellameasured bySRH isthesame asHIorneutralizing antibody. Klin-geborn and Dinter (4) carriedoutexperiments withequid herpesvirus which demonstrated that antibodymediating SRHwasprobably that re-sponsible for neutralization.

Two sera(Table 2)werefoundtocontain low levels ofantibody by HI (1:8 and 1:16) but not by SRH. Differences betweenHI andSRH

reac-TABLE 3. Diagnosis of recent rubella infection by

SRH

Case AcUte/COnVa- Reciprocal SRH area HI

equiva-no. recimen HItiter (mm2) lenta

nrUnsPeCimen

1 a <8 0 <8

c 128 88.0 300

2 a <16 0 <8

c 256 71.5 120

3 a <8 0 <8

c 256 96.8 500

4 a <8 0 <8

c 64 49.7 40

5 a <8 0 <8

c 64 56.5 50

6 a <8 0 <8

c 128 43.2 30

7 a <16 0 <8

c 512 106 1,000

8 a <8 0 <8

c 128 79.5 220

9 a <8 0 <8

c 32 43.2 30

10 a <8 0 <8

c 64 71.5 120

11 a <16 0 <8

c 128 79.5 220

12 a <8 0 <8

c 256 88 300

Determined

graphically,

using theWrelationship

ofFig. 1.

J. CLIN. MICROBIOL.

http://jcm.asm.org/

(5)

tivities may be duetononspecificfalse-positive HI reactionsortoimmunoglobulin M antibody (8). Rubellaimmunoglobulin M antibody tests were notcarried out on these sera.As a diagnos-tictool, however,thisis not aparticular disad-vantagesincepairedsera andseroconversionor fourfold rises in antibody are required.

In thedevelopment ofnew methodologyfor diagnostic investigation, it is necessary tohave asimple and reliablemeansofrelating thenew technique to more established procedures in widespreaduse. The SRH technique is readily relatedtoconventional HI titrationby asimple linear function. Thus, results determined by SRH can bereadily converted tothe more fa-miliar HI format ifsodesired.

ACKNOWLEDGMENTS

Wegratefully acknowledgetechnical assistance from Albert FoleyandMichael Broder and assistance with the presenta-tionand themanuscriptby GeorgeGrady.We thank Michael Oxman and Morton Madoff forsuggestions,encouragement, and support inundertakingthe presentstudy.

This work wassupported in partbycontract 4512-0179, MassachusettsDepartmentof PublicHealth,National Influ-enzaImmunizationProgram.

LITERATURE CITED

1. Farrohi, K., F.K.Farrohi,G. R. Noble, H. S. Kaye, and A. P.Kendal.1977.Evaluation of the single radial

hemolysis test for measuring hemagglutinin- and neur-aminidase-specific antibodies to H3N2 influenza strains and antibodies to influenza B. J. Clin. Microbiol. 5:

353-360.

2. Grillner, L, and 0. Strannegard. 1976. Evaluation of the hemolysis-in-gel test for the screening of rubella immunity and the demonstration of recent infection. J. Clin.Microbiol. 3:86-90.

3. Hiramoto, R. N., J. R. McGhee, D. C. Hurst, and N. M.Hamlin. 1971. A study of the single radialhemolysis ingel System-I. Factors affecting the model. Immuno-chemistry 8:355-365.

4. Klingeborn, B.,and Z. Dinter. 1978.Measurementof neutralizing antibody to equid herpesvirus 1 bysingle radialhemolysis. J. Clin. Microbiol. 7:495-496. 5. McGhee, J. R., D. C.Hurst, N. M. Hamlin, and R. N.

Hiramoto.1971.Astudyof thesingleradialhemolysis in gel System-II.Application of the model to mouse 19S hemolytic antibodies. Immunochemistry 8:367-373. 6. Palmer, D.F.,J. J.Cavallaro, and K. H. Herrmann.

1977.Aprocedural guide to theperformance of rubella hemagglutination-inhibition tests. U.S. Department of Health, Education and Welfare, Center for Disease Control, Atlanta.

7. Schild, G. C., M. S.Pereira, and P.Chakraverty. 1975. Single radial haemolysis: a new method for the assay of antibody to influenza haemagglutinin. Bull. W.H.O. 52: 43-50.

8. Strannegard, O., L. Grillner, and L. Lindberg. 1975. Hemolysis-in-gel test for the demonstration of antibod-ies to rubella virus. J. Clin.Microbiol.64:491-494. 9. Weiler,E., E. W.Melletz, and E.Brueninger-Peck.

1965.Facilitation of immune hemolysis by an interac-tionbetween red cell-sensitizing antibody and globulin allotype antibody. Proc. Natl. Acad. Sci. U.S.A. 54: 1310-1317.

http://jcm.asm.org/