INTERNATIONAL. JOURNAL. OF LEPROSY^ Volume 64, Number 4

Printed in the U.S.A.(ISSN 0148-9I6X)

Enzyme-Linked Immunosorbent Assay (ELISA) withMycobacterial Crude Antigens for the

Sero-Epidemiological Diagnosis of Active Tuberculosis'Alejandro Escobar-Gutierrez, Maria Eugenia Amezcua-Chavarria,

Sergio Pasten-Sanchez, Ernestina Ramirez-Casanova, Jose Victor Cazares,Guadalupe Granados, Elia Loo-Mendez, and Raul Cicero2

Tuberculosis and leprosy are the majorhuman mycobacterioses, and both remainas important public health problems, inspite of worldwide efforts to control them( 7 . 3"). Consequently, it is necessary to per-form large epidemiological surveys for thedetection of active cases which allow thecontinuous transmission of both diseases,particularly when limited or no interventiontakes place. Also, the determination of theextent of the corresponding endemiaswould provide a solid basis for better sur-veillance programs ( 6 . 34 ). Nevertheless, theroutine diagnosis of tuberculosis and lep-rosy largely relies on highly expensive andtime-consuming clinical and laboratorystudies, quite inadequate for epidemiologi-cal studies. The best choice for this purposewould be nonexpensive laboratory diagnos-tic procedures which can he set up in everypublic health laboratory, especially in de-veloping countries where the magnitude ofboth problems is greater '').

The first methodological approach to theepidemiological diagnosis of tuberculosisand leprosy employed serological tech-niques (I". 2'). Great difficulties with thespecificity of the mycobacterial antigensand the development of the polymerise

' Received for publication on 12 December 1995;accepted for publication in revised 16rm On 29 April1996.

= A. Escobar-Gutierrez, Ph.D.; M. E. Amezcua-Chavarria, M.Sc.; S. Pasten-Sanchez, B.Sc.; J. V.Cazares, Biol.Tech.; G. Granados, B.Sc.; E. Loo-Mendez, M.D., Laboratorio de Investigaciones In-munologicas, Instituto Nacional de Diagnostico y Re-ferencia Epidemiologicos (INDRE), Secretaria deSalmi, Carpio 470, Mexico, DE 11340, Mexico. E.Ramirez-Casanova, M.D.; R. Cicero, M.D., Scrviciode Neumologia, I lospital General de Mexico, Secre-taria de Salud, Dr. Bahnis 14K, Mexico, OF 06726,Mexico.

chain reaction (PCR) techniques loweredthe interest in the diagnosis of mycobacte-rioses, especially tuberculosis, through anti-body-based methods ( 25). However, the use-fulness of the PCR tests for tuberculosiswas not well established until now 31 )and, in any case, this procedure is too ex-pensive and out of reach for developingcountries with limited financial resources.For these reasons, this paper forms a part ofour interest in re-evaluating the usefulnessof serodiagnostic tests for mycobacterialdiseases, identifying their limitations andthe conditions necessary for the correct in-terpretation of the results.

Among serological techniques, the en-zyme-linked immunosorbent assay (ELISA)seems to be the most appropriate for the di-agnosis of mycobacterial diseases (5 ' '"). Inthe case of Mycobacterium tuberculosis, thelack of a species-specific molecule has con-tributed to the development of ELISAs us-ing crude, semicrude and purified mole-cules. Even though there are reports aboutbetter diagnostic performances with puri-fied molecules ( 0 ' 25 ' 33 ), there is no unifiedconsensus concerning which one is the best,and consequently, there are no protocols fortheir pilot or industrial production and stan-dardization of the reagents. Therefore, noneis commercially available. Consequently,tuberculosis serodiagnostic techniques un-der use in most laboratories from develop-ing countries usually employ crude antigenswhose crossreactivity with other mycobac-teria limit the accuracy of the positive pre-dictive results. This situation is especiallystriking in leprosy-endemic areas becausethe large amounts of bacilli spread by activecases easily induce Immoral anti-mycobac-terial responses in households as well as inthe general population (').

417

418^ International Journal of Leprosy^ 1996

In this paper, the diagnostic perfor-mances of ELISAs employing six com-monly used, crude mycobacterial prepara-tions were evaluated in confirmed activecases, with either tuberculosis or leprosy, aswell as in household contacts of tuberculo-sis patients. Assayed antigens were intactwhole bacilli, lipid-free whole bacilli, andcrude, protein-enriched, soluble extractsfrom M. tuberculosis H37Rv and M. bovisBCG strain. The evaluation of sensitivityand specificity of each preparation was ac-complished using sera from well-definedtuberculosis and leprosy patients, householdcontacts of tuberculosis cases, and samplesfrom healthy individuals from a low and ahigh leprosy-endemic area. Cases and con-trols were carefully matched for their ethnicand socioeconomic backgrounds.

MATERIALS AND METHODSCases and controls. One-hundred-six

serum samples from tuberculosis cases andtheir contacts were included in the study.Tuberculosis sera were obtained from 46adult patients with nontreated pulmonarytuberculosis admitted to the Servicio deNeumologia, Hospital General de Mexico,Secretaria de Salud, Mexico City, Mexico.All cases had clinical and radiological evi-dence of pulmonary tuberculosis, con-firmed through positive sputum smears andM. tuberculosis cultures. In addition, 60household contacts from active tuberculosiscases, without any clinical feature suggest-ing tuberculosis, were selected among fa-milial and nonfamilial individuals living to-gether for more than 6 months.

Twenty sera from adult lepromatous lep-rosy patients, all positive for antibodiesagainst M. leprae by the FLA-ABS test,were randomly chosen from our laboratorycollection ( 2."). Patients were diagnosed andclassified according to the standard clinical,histopathological and bacteriological meth-ods at Servicios Coordinados de Salud Pu-blica in Culiacan, state of Sinaloa, Mexico.

As healthy controls, two groups of adultswere selected according to the tuberculosisand leprosy morbidities in the areas wherethey lived. In Mexico, the tuberculosis mor-bidity rate for 1993 was 0.14 per 1000 in-habitants; the leprosy rate for the same yearwas 0.12 per 1000 inhabitants. The firstcontrol group consisted of 20 sera from thestate of Sinaloa, in northwestern Mexico,

where morbidity rates for tuberculosis andleprosy were: 0.265 cases per 1000 inhabi-tants and 0.83 cases per 1000 inhabitants,respectively. The second group was formedby 47 sera obtained from altruistic blooddonors in Mexico City; here, the rate for tu-berculosis reached 0.06 cases per 1000 in-habitants, and the one for leprosy was of0.02 cases per 1000 inhabitants. None ofthem had any history of recent mycobac-terial disease or any other infectious con-dition, but no attempt to assess theirPPD-skin reactivity was undertaken. Bothpatients and controls shared similar socio-economic status and belonged to the Mexi-can mestizo ethnic background. Sera wasaseptically separated from blood and keptfrozen at —70°C until use.

Mycobacterial strains. M. tuberculosisH37Rv was obtained from the collection ofthe Departamento de Micobacterias, IN-DRE, Mexico City, and the M. bovis Danish1331 BCG strain used for vaccination wasdonated by Institute Nacional de Higiene,Secretaria de Salud, Mexico City. In bothcases, the identity and purity of the strainswere confirmed using standard methods andthey were grown in Sauton's medium untila luxurious growth was reached. Bacterialpellets were washed three times with pH7.4, 0.01 M phosphate buffered saline (PBS),resuspended in the same buffer, and main-tained frozen at —70°C.

Intact whole bacilli preparations. Forboth mycobacterial strains, frozen bacilliwere thawed in a water bath at 37°C,weighed and resuspended in PBS pH 7.4.M. tuberculosis cells were killed by irradia-tion at 2.5 Mrad with a `iOCo source. Finalconcentration of this suspension was ad-justed to 10 mg of moist bacilli per ml.

Lipid-free whole bacilli preparations.With both organisms, isopentanol-ex-tracted, formalin-fixed whole cells wereprepared as described by Wayne, et al. ( 3")as follows: 20 mg of moist mycobacteriawere suspended in 1 ml of 10% formalin inPBS, pH 7.4 vol/vol. After three washeswith distilled water, the cells were sus-pended in 200 ml of distilled water, mixedwith 10 ml of isopentanol (isoamyl alcohol,ACS reagent; Sigma Chemical Co., St.Louis, Missouri, U.S.A.) and mixed for 10min in a vortex. Bacilli were separated bycentrifugation at 900 x g x 30 min, washedtwice with 0.02% Tween 80 in distilled wa-

64, 4^Escobar-Gutierrez, et al.: ELISA in TB Seroepidendology^419

ter and resuspended at a concentration of 10mg of moist cells per ml.

Protein-enriched soluble extracts.Sonic extracts of both mycobacterial strainswere prepared with an adaptation of amethod published elsewhere ( 14 ). Briefly,bacterial pellets were suspended in 2 ml ofPBS pH 7.4 and disrupted by sonication onan ice-water bath for 15 min at 2-min inter-vals at 250 W on a Sonifier cell disrupter(Model W-370, Heat Systems; UltrasonicsInc., North Tonawanda, New York, U.S.A.).The sonicated material was examined forthe absence of intact cells by Ziehl-Neelsenstaining and centrifuged at 20,000 x g x 15min in the cold. The supernatant was fil-tered through a 0.22-pm pore membrane fil-ter, and the protein concentration was ad-justed to 2.5 pg per ml.

ELISA procedure. Polystyrene plates(EIA/RIA plate, flat-bottom, high binding;Costar Corporation, Cambridge, Massachu-setts, U.S.A.) were sensitized at 4°C over-night with 100 p1 of the correspondingpreparations described above. After block-ing 90 min at room temperature with 1%bovine serum albumin (BSA grade IV;Sigma) in PBS pH 7.4, the wells were filledin duplicate with test or control sera diluted1:100, 1:500 and 1:1000 in 2.5% normalgoat sera and 0.05% Tween 20 in PBS pH7.4, for the intact or lipid-free bacilli, and inthe same solution used for blocking for theprotein-enriched soluble extract. After dis-carding the contents, the wells were washedthree times and incubated at room tempera-ture with polyclonal goat anti-human im-munoglobulin coupled to horseradish per-oxidase (Sigma) at a 1:1000 dilution. After1 hr of incubation at room temperature, theplates were washed with PBS, pH 7.2, and100 p1 of freshly prepared 0.4 mg/ml o-phenylenediamine with 100 p1 3% hydro-gen peroxide was added as the enzyme sub-strate. The reaction was stopped with 4 Nsulfuric acid and the absorbance (A) wasread at 492 nm using an ELISA reader(Multiskan plus; Labsystems, Helsinki, Fin-land) against a blank without serum.

Cut-off values for each antigen prepara-tion were calculated considering the corre-sponding mean absorbances obtained in thecontrol group plus two standard deviations( 13 ).

Data analysis. For each serum, the meanabsorbance value for duplicate measures

was calculated. When a variation largerthan 10% was found, the results were dis-carded and the test was repeated. An intra-assay reliability test was performed consid-ering variation coefficients of all the dupli-cates included in a given plate, and a <10%variation coefficient was considered opti-mal. In addition, an intra-assay reproduci-bility test was included in each plate withduplicates of known sera (high and lowpositive and negative controls); a <10%variation coefficient was acceptable.

We calculated the means and standarddeviations of the A 4,„ values for the controland study groups. The significance of thevariations in antibody levels among groupswas determined through Student's t test.Comparisons among cases and controlswere made considering one or another con-trol group separately. Specificity, sensitiv-ity, and positive and negative predictivevalues for each antigen were determined us-ing Galen and Gambino's ( 1 ') recom-mended method. After selection of the bestperforming antigen, we determined a coef-ficient of agreement for each of the otherantigens as follows: 100 x number of re-sults in agreement/total number of samplestested. To compare the selected preparationwith other antigens, we used Cohen's kappatest ( 15 ) and a 95% confidence interval wascalculated. The x value was corrected forchance and ranged from —1 to 1, where K =1 corresponds to a perfect agreement, lc =—1 indicates perfect disagreement, and K = 0is an agreement expected by chance alone.

RESULTSSelection of optimal amount of antigen

preparations. Whole intact and wholelipid-free bacilli were only tested at 10mg/ml as has been recommended ( 3 '). Tofind the optimal concentrations of M. tuber-culosis and BCG soluble extract prepara-tions, we performed ELISAs with doublingdilutions of these antigens, starting from 15pg/ml, using randomly chosen sera, 10from the healthy control group and 10 fromthe tuberculosis group. In both cases, 2.5plz/ml was the optimal concentration found.

Comparison of the six antigen prepa-rations. In order to determine the differ-ences between test and control sera witheach antigen, 1:100, 1:500 and 1:1000 dilu-tions of 10 randomly chosen tuberculosisand leprosy sera were tested. Most of the

420^ International Journal of Leprosy^ 1996

TABLE I . Results of ELISA for detection of anti-invcobacterial antibodies in pal-inonary tuberculosis active cases and tuberculosis household contacts using six antigens.

"Fuberculosis active cases^Tuberculosis household contacts

Antigens Cut-offvalue . '

No.positive/

Positivetotal

A492(mean ± S.D.)

No.

Positive/total

PositiveA

2(mean ± S.D.)

M. tubeivulosisWhole intact bacilli 0.614 33/47 70 1.116 ± 0.680 6/60 10 0.220 ± 0.240Whole lipid-free bacilli 0.347 38/47 81 0.812 ± 0.596 1/60 2 0.267 ± 0.085Soluble extract 0.281 31/47 66 0.669 ± 0.513 2/60 3 0.156 ± 0.074

M. bovis, BCG strainWhole intact bacilli 0.464 37/47 79 1.200 ± 0.789 6/60 10 0.350 ± 0.097Whole lipid-free bacilli 0.313 35/47 74 0.939 ± 0.768 0/60 0 0.078 ± 0.037Soluble extract 0.464 43/47 91 1.389 ± 0.842 6/60 10 0.283 ± 0.159

Mean^in controls + 2 standard deviations (S.D.).

sera diluted 1:100 gave important back-ground color, regardless of their origin,antigen employed and number or durationof washings. Contrastingly, consistent re-sults were obtained with sera diluted at1:500 and 1:1000, but some sera from tu-berculosis patients were negative at the1:1000 dilution in spite of a clear positiveresult at a 1:500 dilution. Therefore, onlysera diluted at 1:500 were employed, and

the results for controls, contacts, and pa-tients are shown in Tables 1 and 2 as well asin Figures 1 and 2.

All control sera showed low absorbancereadings with normal distribution curves.Since the analysis of variance showed nodifference between results in healthy con-trols from high- or low-endemic leprosy ortuberculosis areas, they were combined toprovide a single, large control group for

0

3.2 —A 442^3.0 —

2.8 —

2.6 —

2.4 —

2.2 —

2.0 —

1.8 —

1.6 —

1.4 —

1.2 —

1.0-

0.8 —

0.6-

0.4—

0.2 — MO-.

0IIC LL^TB

M. tuberculosisintact bacilli

0

0

COO

0000

000

0000.0^00

0

0^000

000^00

02,0

0

LL^'LB

tuberculosislipid-free bacilli

LI,^TB^

iii

Al tuberculosissoluble extract

TII^

IIC Tit^

!IC

0

0

CO

0

0

00

000

0

COO

000

0

FIG. I. Absorbance values in sent from healthy controls (11C), lepromatous leprosy cases (LL), tuberculosispatients (TB), and household contacts from tuberculosis patients (TI I), using M. tubeindosis crude antigens inELISA (horizontal bars show cut-off values).

0

(s)

00

0

cip

0^ 00c.00

TB

I3CGintact bacilli

00

0

00

0

0

00

0?poo00

000

0000

0

800

&c.

BCClipid-free bacilli

000000

0

0^0000

00^0

LL^TB^TII

BCGsoluble extract

3.2 —

3.0 —

2.8

2.6

2.4

2.2 —

2.0 —

1.8 —

1.6 —

1.4 —

1.2 —

1.0 —

0.8 —

0.6 —

0.4 —

0.2 —

0TII^

IICTII^

IIC^

LL

64, 4^Escobar-Gutierrez, et al.: ELISA in TB Seroepidendology^421

TABLE 2. Results of ELISA for detection of anti-invcobacterial antibodies in leproma-tous lepmsv cases using six antigens.

Antigens Cut-offvalue'

No. positive/total Positive

A(mean ± S.D.)

M. tuberculosisWhole intact bacilli 0.614 9/20 45 0.694 ± 0.486Whole lipid-free bacilli 0.347 15/20 75 0.606 ± 0.381Soluble extract 0.281 11/20 55 0.536 ± 0.516

M. bolls, BCG strainWhole intact bacilli 0.464 7/20 35 0.682 ± 0.811Whole lipid-free bacilli 0.313 12/20 60 0.678 ± 0.720Soluble extract 0.464 3/20 15 0.270 ± 0.366

Mean A 4 ,,, in controls + 2 standard deviations (S.D.).

comparison with the test sera. As deter-mined in the large control group, cut-offvalues were distinct for each preparation(Tables I and 2). Five out of 67 control seragave positive results with one or anotherantigen, and all of them with soluble extractof BCG. In four of them absorbance wasnear to the respective cut-off value, and onlyone control serum was clearly positive to allbut one of the antigens tested. Even thoughwe tried several times, it was impossible to

locate this positive responsive donor to clar-ify the actual individual's clinical status.

Mean A 49 , values for antibodies againstthe six antigens tested in the studied groupsare presented in Tables 1 and 2. As can beseen, significant differences with controlsare consistently present only in the case oftuberculosis patients (p < 0.05). With any ofthe preparations it was possible to detect aspositive all of the tuberculosis sera. Thehighest number of positive results in the tu-

A 49 ,

Fla. 2. Absorbance values in sera from healthy controls (I IC), lepromatous leprosy cases (LL), tuberculosispatients (TB), and household contacts from tuberculosis patients (ill), using BCC; crude antigens in ELISA (ho-rizontal bars show cut-off values).

429^International Journal of Leprosy^ 1996

TABLE 3. Perprmance or six antigens in ELISA Pr the serological detection of anti-mycobacterial antibodies in tuberculosis-active patients.

Antigen Sensitivity (%) Specificity (%) (</e) NPV'' ('4)

M. tuberculosisWhole intact bacilli 74.5 97.0 94.6 84.6Whole lipid-free bacilli 80.8 94. I 90.5 87.7Soluble extract 70.2 94.3 89.2 82.5

M. bobs, BCG strainWhole intact bacilli 78.7 97.0 94.9 86.6Whole lipid-free bacilli 74.5 98.5 97.2 84.6Soluble extract 91.5 92.5 89.5 93.9

Positive predictive value.'' Negative predictive value.

berculosis patients' group was obtainedwith the use of soluble extract of BCG(91%) and, interestingly, this preparationidentified the lowest number of leprosycases. Also, with this preparation the bestvalue for sensitivity was achieved (91.5%)(Table 3). Even in this group, however, sev-eral serum samples had lower absorbancesthan the cut-off value, regardless of the factthat they came from confirmed, active tu-berculosis patients (Figs. I and 2). As alsocan be seen in Table I, most of the serafrom the household contacts gave negativeresults with all of the preparations, and theirmean antibody level for each one was verysimilar to the healthy controls. After obtain-ing these results, the six positive contacts,regardless of the preparation to which theyreacted, were submitted to clinical studies.Three of them showed suggestive chest ra-diographs and, therefore, standard treat-ment was started. The remaining positivecontacts are now under permanent surveil-lance.

Considering the results in the pulmonarytuberculosis group, specificities were up to92% for all antigens tested, but their sensi-tivities showed varying levels (Table 3).Because sensitivity is the parameter ofchoice for our epidemiological studies, theresults with BCG protein-enriched solubleextract were considered as standard for theanalysis of the remaining results. Figure 3shows the contingency tables used for suchcomparisons.

DISCUSSIONInterest in the serodiagnosis of mycobac-

terioses, particularly of tuberculosis, has

been dampened somewhat because most ofM. tuberculosis antigens show crossreativ-ity with other mycobacterial species, bothenvironmental and pathogenic. Neverthe-less, inexpensive and efficient methodssuch as ELISAs remain the best choice forthe epidemiological studies especiallyneeded in developing countries ( 10). In aprevious study, efficiency in the serodiag-nosis for leprosy was explored ( 14 ). There-fore, this work was designed to try severalwhole mycobacterial preparations inELISAs for tuberculosis serodiagnosis ap-plicable to population surveys. Conse-quently, special care was taken in the metic-ulous search for active tuberculosis patientsconfirmed through positive bacilloscopyand M. tuberculosis culture. In order to es-tablish the level of false-positive results,sera from individuals living in the samehouse as a confirmed tuberculosis case,who did not show any symptoms of the dis-ease, were included. Also, a group of well-defined sera from leprosy cases was studiedin order to identify the preparation with thelowest crossreactive reactions. Controlgroups were checked for the absence of anysuggestive history of mycobacterial infec-tion other than BCG vaccination (which iscompulsory in Mexico). According to theage of the individuals included in our con-trol groups, most of them should have beenBCG vaccinated and the low anti-mycobac-terial antibody levels observed in all ofthem are in agreement with other reports

24-2(') of a nonsignificant effect of theBCG vaccination on the number of false-positive results in antibody-based diagnos-tic tests for tuberculosis. It is known that

64, 4^Escobar-Gutierrez, et al.: ELISA in TB Seroepidentiology^423

BCG strain soluble extract

M. tuberculosislipid-free bacilli

^36^2

^

6^3

Agreement = 73.47% K = 0.34 (0.07 - 0.61)

Agreement = 65.31% x = x = 0.21 ( -0.03 - 0.46)

BCG strain soluble extract

BCG strainlipid-free bacilli

32^210^3

BCG strain soluble extract

M. tuberculosis 31^2intact bacilli 11^3 Agreement = 63.26% K = 0.19 (-0.05 - 0.43)

BCG strain soluble extract+^-

BCG strain 30^1intact bacilli 12^4 Agreement = 61.22% x = 0.26 (0.04 - 0.48)

BCG strain soluble extract

M. tuberculosissoluble extract

28^114^4 Agreement = 57.14% x = 0.22 (0.01 - 0.43)

FIG. 3. Contingency tables of results from comparisons of ELISAs using six antigens for the serodiagnosisof tuberculosis cases (nunilms in parentheses after ix - values are 90% confidence intervals).

BCG vaccination provides a very limitedantigenic stimulus compared to that of an ac-tive infection (") and its antibody responsecould interfere with the serodiagnosis onlywhen the vaccination was recently per-formed (32 . 3").

Systematic statistical analysis of the re-sults for each serum duplicate and the rou-tine repetition of all the samples with differ-ences > I (Y/o in absorbance, as practiced inthis work, greatly reduced the variability inreactivity commonly seen in ELISA sys-tems. Nevertheless, it must be stressed thatthe specificity and sensitivity of any ELISA

depends not only on variables related to themethod itself but also on the antigen se-lected and the populations under study. Anexample of this variability in the tuberculo-sis serodiagnosis can be seen in the widedifferences obtained with the use of purifiedprotein derivative (PPD) by different au-thors (1, 11, 23, 24, 27 15 17 41‘ .) As a result of thissituation, laboratory personnel workingwith ELISA for tuberculosis have to dealwith several options when choosing theantigen preparation for routine screening,always taking into account the epidemio-logical background of the population to he

424^ In ternational Journal of Leprosy^ 1996

screened. In this work, whole intact bacilliand soluble extracts were chosen becausethey are frequently used in some laboratories;the inclusion of mycobacteria depleted of ex-ternal lipids (lipid-free whole bacilli) wasincluded because such treatment could elim-inate masked, relevant, antigenic surfacemolecules. Lipid-free bacilli, from either M.tuberculosis or BCG, reacted more frequentlywith the tuberculosis and leprosy sera thandid the respective intact preparations.

The positive reactions of BCG prepara-tions with sera from tuberculosis and lep-rosy cases, as well as those between BCGand M. tuberculosis preparations with lep-rosy sera, are the obligate consequence ofthe strong molecular similarity among my-cobacteria (12,40). Even though BCG and M.tuberculosis crude preparations showed bet-ter reactivity with tuberculosis samples thanwith leprosy sera, such crosspositivitymade it inadequate to use these antigens inareas where both mycobacteria are preva-lent in the population. Thus, crude prepara-tions for seroepidemiological studies of tu-berculosis can be used only in leprosy-freeareas. Under these circumstances, resultspresented here point out that the crudepreparation of choice for tuberculosis sero-diagnosis is the BCG soluble extract, as hasbeen reported by other authors (18. 21, 24).

The occurrence of false-negative resultsin some tuberculosis patients was not unex-pected because this situation is fairly com-mon in the serodiagnosis of the disease,even with methods other than ELISA orwhen more purified antigens are used. Theavailable clinical and laboratory data of theseronegative patients were not sufficient toclarify the origin of such situations and spe-cific investigations in this direction willhave to be done.

It must be stressed that we suggest theuse of crude mycobacterial preparations forepidemiological studies only. Moreover, inpopulation surveys this preparation can beused confidently because it can readily de-tect most of the active tuberculosis casesand even in the close contacts of activecases, false-positive results seem to he notimportant. Studies to confirm this assump-tion are now in progress in our laboratory.On the other hand, for individual diagnosisof pulmonary tuberculoid cases, the use ofserological methods for the routine work

are excluded because results of sputum ex-amination by microscopy, in combinationor not with sputum culture, yield an accu-rate and highly efficient diagnosis of thecondition ( 20 ' 22 ). An attractive possibility,which now is under study in our laboratory,is the employment of serological tests forthe identification of extrapulmonary tuber-culosis cases only where radiography is notuseful and bacilloscopy and/or culture aredifficult to carry out. The preliminary re-sults are encouraging and indicate that anegative result with a BCG protein-en-riched soluble extract ELISA does not ruleout tuberculosis, but a positive result couldbe a highly reliable indicator of the diseasebecause the method seems to be specific formycobacterial infection.

In conclusion, the results of our presentstudy strongly support an acceptable diag-nostic value for BCG soluble extract as anantigen in an ELISA to be used for tubercu-losis epidemiological surveys in leprosy-free areas when purified antigens are notavailable. However, the use of serologicaltests for individual case diagnosis of extra-pulmonary tuberculosis must be regarded asa complementary procedure and the resultsmust he cautiously interpreted, always tak-ing into account compatible clinical data.

SUMMARYIn search for reliable, nonexpensive pro-

cedures for tuberculosis diagnosis suitablefor seroepidemiological studies in leprosy-endemic areas, enzyme-linked immunosor-bent assays (ELISAs) with whole intactbacilli, whole lipid-free bacilli and protein-enriched soluble extracts from the H37RvMycobacterium tuberculosis strain wereevaluated. Sera tested came from 47 active,pulmonary tuberculosis adult cases, 60household contacts of active tuberculosiscases, 20 lepromatous leprosy adult pa-tients, and 67 healthy adult controls ob-tained from low and high leprosy and tuber-culosis endemicity areas. There was no in-fluence of such endemicity levels in thenumber of positive results in control sera.Antibody levels obtained with each of theantigens in ELISAs were significantly dif-ferent in tuberculosis patients and the con-trol groups. Ten percent of tuberculosiscontacts were positive with some of theantigens and three of them showed sugges-

64, 4^Escobar-Gutierrez, et al.: ELISA in TB Scroepidennology^425

tive chest radiographs. The best combina-tion for a high number of positive resultswith tuberculosis sera and low positive re-sults with leprosy sera was the BCG solubleextract (91% and 15%, respectively). Thispreparation also yielded excellent sensitiv-ity and specificity values for tuberculosis(91.5% and 92.5%, respectively). Thesedata suggest that BCG soluble extractELISAs could provide helpful informationto estimate tuberculosis prevalence only inleprosy-free areas, under a situation of un-availability of purified antigens. In pul-monary cases, sputum microscopic exami-nation and culture have higher sensibilitythan serodiagnosis; therefore, the utilizationof BCG soluble extract ELISAs as a diag-nostic aid in individual patients with sus-pected active tuberculosis only can he use-ful in extrapulmonary cases.

RESUMENSe evaluaron procedimientos inmunoenzimaticos

(ELISAs) con bacilos intactos, bacilos deslipidizadosy extractos solubles de Mycobacterium tuberculosis113712v, con el fin de encontrar un metodo confiable ybarato para el diagnOstico de la tuberculosis en areasendemicas de lepra. Los sueros prohados corre-spondieron a 47 adultos con tuberculosis pulmonar ac-tiva, 60 contactos convivientes de los pacientes con tu-berculosis, 20 pacientes con lepra lepromatosa y 67controles sanos de areas de alta y baja endemicidadpara lepra y tuberculosis. El grado de endeinicidad noinfluy6 en los resultados encontrados en los sueroscontrol. Los niveles do anticuerpo con cab uno de losantigenos fueron significativamente diferentes entre.los pacientes con tuberculosis y el grupo control. Die,.porciento de los contactos de los pacientes fueronpositivos con alguno de los antigenos y 3 de ellosmostraron radiogralias de torax sugestivas de enfer-medad. Con el extracto soluble de BCG se logrO unaalta reactividad con los sueros de tuberculosis y unabaja positividad con los sueros de lepra (913 y 15%respectivamente). Estos datos sugieren que en situa-ciones donde no se cucnta con antigenos purificados,los ELISAs con extractos solubles de BM podrianservir para estimar la prevalencia de tuberculosis en lasareas libres dc lepra. Ya que en los casos pulmonares elexamen microscopico del esputo y el cultivo tienenmayor sensibilidad que los metodos serolOgicos, Ia uti-lizacian de los ELISAs con extractos solubles de BCGpara el diagnOstico de tuberculosis, solo podria scr tinten los casos extrapulmonares.

RESUMEA la recherche de tests fiables, bon marche et ap-

propries pour le diagnostics de la tuberculose pour des

etudes sero-epidemiologiques dans des regions ou Ialepre est endemique, nous avons evalue des tests enzy-matiques (ELISA) avec des bacilles entiers intacts, desbacilles entiers sans lipides et des extraits solubles en-richis en proteines provenant de la souche 1137Rv deMycobacterium tuberculosis. Les serums testes prove-naient de 47 cas de tuberculose pulmonaire active, 60contacts domiciliaires de cas actifs de tuherculose, 20cas adultes de lepre lepromateuse et 67 temoins adultesen bonne sante venant de regions a endemicites faibleset elevees pour la lepre et Ia tuberculose. 11 n'y avaitpas d' influence des taux d'endemicite stir le nombre deresultats positifs parmi les serums des temoins. Lestaux d'anticorps obtenus avec chacun des antigenesetaient significativement differents chez les patients tu-berculeux et les grouper de contrule. Dix pourcentsdes contacts de tuberculeux etaient positifs pour cer-tains des antigenes et trois d'entre eux ont montre desradiographies suggestives. La meilleure combinaisonpour un nombre eleve de resultats positifs parmi lesserums de tuberculeux et tin nombre faible de positifsparmi les serums de lepreux etait l'extrait soluble deI3CG (respectivement 91% et 15%). Cette preparationdonnait egalement d'excellentes valeurs de sensibiliteet de specificite pour la tuherculose (respectivement91.5% et 92.5%). Ces donnees suggerent que les ELISAa l'extrait soluble de BCG pourraient dormer des infor-mations utiles pour estimer la prevalence de la tubercu-lose dans des regions ou la lepre n'existe pas, dans dessituations ou les antigenes purifies ne soot pas dis-poniblcs. Dans les cas pulmonaires, Pexamen micro-scopique des expectorations et Ia culture ont tine sensi-bilite plus elevee que le serodiagnostic; c'est pourquoi('utilisation d'ELISA aux extraits solubles de BCGcomme aide au diagnostic chez des patients individuelssuspects de tuherculose active peuvent setilement etreutiles (11111S les cas de tuberculose extra-pulmonaire.

Acknowledgment. The authors wish to expresstheir gratitude to Maria Teresa Garrido and LambertoBlancarte from the Instituto Nacional de DiagnOstico yReferencia EpidemiolOgicos; Maria Elena Juarez fromthe I lospital General de Mexico; Juvencio Ruiz-Puentefrom the Institute Nacional de 1 ligiene, and Enrique.Dominguez from the Instituto Nacional de Investiga-clones Nucleares for their valuable help in the accom-plishment of this project. We are also indebted to EthelGarcia and Aurora del Rio for their excellent criticalreview of this manuscript.

REFERENCESI. ANIEZCUA, NI. E., ESCOBAR-GUTIERREZ, A.,

BARBA-Rumo, J., CAZARES, J. V., MAYEN, E.,CliAvEz-Ntn;a:z, N4., PENA, R. C., Roommiu, R.and PAsTEN, S. Prospective immunological fol-low-up in household contact of Mexican leprosypatients. Int. J. Lepr. 58 (1990) 651-659.

2. AMEZCUA, M. E., ESCOBAR-GUTIERREZ, A.,MAVEN, E. and CAZARES, J. V. Sensitivity andspecificity of the FLA-ABS test for leprosy in

426^ International Journal of Leprosy^ 1996

Mexican population. Int. J. Lepr. 55 (1987)286-292.

3. BALEsTRINo, E. A., DANIEL, T. M., DE LATINI, M.D. S., LATINI, 0. A., MA, Y. and EscocozzA, J. B.Serodiagnosis of pulmonary tuberculosis in Ar-gentina by enzyme-linked immunosorbent assay(ELISA) of IgG antibody to 111 .scobacterium tu-berculosis antigen 5 and tuberculin purified pro-tein derivative (PPD). Bull. WHO 62 (1984)755-761.

4. BARRERA, L., Micui, I., RrrAcco, V., TORREA, G.,BRoGLIA, B., BorrA, R., PEREz-MALDoNADo, C.,FERRER°, N., PistAsco, A., Curnio, 1., CoRNEK),M., PRokorfo, E. and KANT0R, I. Detection of cir-culating antibodies to purified protein derivativeby enzyme-linked immunosorbent assay: its po-tential for the rapid diagnosis of tuberculosis. Pe-diatr. Infect. Dis. J. 8 (1989) 763-767.

5. BENJAMIN, R. G., DEBANNE, S. NI., MA, Y. andDANIEL, T. M. Evaluation of mycobacterial anti-gens in an enzyme-linked immunosorbent assay(ELISA) for the serodiagnosis of tuberculosis. J.Med. Microbiol. 18 (1984) 309-318.

6. BfiARADwAt, V. R and KATOCH, K. An overviewof the current status of serological techniques inthe epidemiology of leprosy. Trop. Med. Parasitol.41 (1990) 359-360.

7. BLoom, B. R. and MURRAY, C. J. L. Tuberculosis:commentary on a re-emergent killer. Science 257(1992) 1055-1063.

8. CENTERS FOR DISEASE CONTROL AND PREVENTION.Diagnosis of tuberculosis by nucleic acid amplifi-cation methods applied to clinical specimens.MMWR 42 (1993) 686.

9. DANiEr., T. M. Rapid diagnosis of tuberculosis:laboratory techniques applicable in developingcountries. Rev. Infect. Dis. 11 Suppl. (1989)S47 I-S478.

10. DANIEL, T. M. and DEBANNE, S. M. The serodiag-nosis of tuberculosis and other mycobacterial dis-eases by enzyme-linked immunosorbent assay.Am. Rev. Respir. Dis. 135 (1987) 1137-1150.

I I. DANIEL, T. M., DEBANNE, S. M. and VAN DERKUYP, F. Enzyme-linked immunosorbent assayusing Mycobacterium tuberculosis antigen 5 andPPD for the serodiagnosis of tuberculosis. Chest88 (1985) 388-392.

12. DANIEL, T. M. and JANICKI, B. W. Mycobacterialantigens: a review of their isolation, chemistry,and immunological properties. Microbiol. Rev. 42(1978) 84-113.

13. DEsAviGNY, A. and VotiER, A. The communica-tion of ELISA data from laboratory to clinician. J.Immunoassay 1 (1980) 105-108.

14. ESCOBAR-GUTIERREZ, A., AMEZCUA, NI. E., PAS-TEN, S., PAI.I.ARES, F., CAZARES, J. V., P01.11)0, R.M., FLoiths, 0., CASTRO, E. and RefimuEz, 0.Comparative assessment of the leprosy antibodyabsorption test, Mycobacterium 'quite extract en-zyme-linked immunosorbent assay, and gelatin

particle agglutination test for serodiagnosis of lep-romatous leprosy. J. Clin. Microbiol. 31 (1993)1329-1333.

15. Floss, J. L. Statistical Methods for Rates and Pm-portions. New York: John Wiley and Sons, 1981,pp. 212-236.

16. FUJIwARA, T., HUNTER, S. W., Cito, S., ASPINALL,G. 0. and BRENNAN, P. J. Chemical synthesis andserology of disaccharides and trisaccharides ofphenolic glycolipid from the leprosy bacillus andpreparation of a disaccharide protein conjugate forserodiagnosis of leprosy. Infect. Immun. 43 (1984)245-252.

17. GALEN, R. S. and GAMBINO, R. S. Beyond Nor-mality: the Predictive Value and Efficiency ofMedical Diagnoses. New York: John Wiley andSons, 1975.

18. GARCIA-ORTIGOZA, E. and GUTIERREZ-VELAZQUEZ,A. DiagnOstico de la tuberculosis pulmonarcrOnica por el metodo de inmunoensayo en-zimatico (ELISA). Rev. Latinoamer. Microbiol. 24(1982) 193-204.

19. GEORGIEV, G. D. and McDomALL, A. C. Thebacteriological examination of slit skin smears inleprosy control programmes using multidrug ther-apy: a plea for radical changes in current opera-tional methodology. Indian J. Lepr. 59 (1987)373-385.

20. GoRDIN, F. and St.uTKIN, G. The validity of acid-fast smears in the diagnosis of pulmonary tuberculo-sis. Arch. Pathol. Lab. Med. 114 (1990) 1025-1027.

21. GRANGE, J. M., GntsoN, J., BATTY, E. and KARD-filo, T. The specificity of the Immoral immune re-sponse to soluble mycobacterial antigens in tuber-culosis. Tubercle 61 (1980) 153-166.

22. GREENBAUM, M., BEYT, B. E. and MURRAY, P. R.The accuracy of diagnosing tuberculosis at a largeteaching hospital. Am. Rev. Respir. Dis. 121(1980) 477-481.

23. Gum-A, A. K., JAMIE, Z., SRIVASTAVA, V. K., TAN-DON, A. and SAXENA, K. C. Antibodies to purifiedtuberculin (PPD) in pulmonary tuberculosis andtheir correlation with PPD skin sensitivity. IndianJ. Med. Res. 78 (1983) 484-488.

24. KALISH, S. B., RABIN, R. C., PII,\IR, J. B., LEVITS,D., ZiAss, C. R. and METZGER, E. Use of an en-zyme-linked immunosorbent assay technique inthe differential diagnosis of active pulmonary tu-berculosis in humans. J. Infect. Dis. 147 (1983)523-530.

25. KAPLAN, M. II., ARMSTRONG, D. and PINSKY, C.Antibody responses to mycobacterial antigens inpatients immunized with BCG for treatment ofneoplastic disease. Cancer 46 (1990) 2195-2202.

26. KARDJITO, T., HANDovo, I. and GRANGE, J. M. Di-agnosis of active tuberculosis by immunologicalmethods. I. The effect of tuberculin reactivity andprevious BCG vaccination on the antibody levelsdetermined by ELISA. Tubercle 63 (1982) 269-274.

64, 4^Escobar-Gutierrez, et al.: ELISA in TB Semepidennology^427

an emphasis On control in developing countries. Rev.Infect. Dis. 11 Suppl. 2 (1989) S339-S346.

35. TAtsboN, A., SAXENA, R. P., SAxlits4A, K. C., JANiff,,Z. and GUPTA, A. K. Diagnostic potentialities ofenzyme-linked immunosorbent assay in tubercu-losis using purified tubercular antigen. Tubercle61 (1980) 87-89.TURNEER, M., VAN VOOREN, J. P., NYABENDA, J.,LEGRos, F., LEcomTE, A., TIIIRIAUX, J., SERRUYS,E. and YERNAULT, J. C. The Immoral immune re-sponse after BCG vaccination in humans: conse-quences for the serodiagnosis of tuberculosis. Eur.Respir. J. 1 (1988) 589-593

37. VILJANEN, NI. K. V., SKALA, J. and TAI.,\, F. En-zyme-linked immunosorbent assay (ELISA) forantibodies to purified protein derivative of tuber-culosis (PPD): IgNI, IgA and IgG-anti PPD anti-bodies in active pulmonary tuberculosis. Eur. J.Respir. Dis. 63 (1982) 257-262.

38. WALLACE, R., DIENA, 13.13., JESSAMINE, A. 0. andGREENBERG, L. Circulating antibody response inBCG-vaccination, tuberculosis infection and sar-coidosis. Can. Med. Assoc. J. 96 (1967) 585-588.

39. WAYNE, L. G., HOLLANDER, D., ANDERSON, B.,SRAN1EK, H. A., VADtiont, C. M. and RorrER, J. 1.Immunoglobulin A (IgA) and IgG serum antibod-ies to mycobacterial antigens in Crohn's diseasepatients and their relatives. J. Clin. NIi CFobiol. 30(1992) 2013-2018.YOUNG, D. B., KAUFMANN, S. H. E., HERMANS, P.W. M. and Thou:, J. E. R. Mycobacterial proteinantigens: a compilation. Mol. Microbiol. 6 (1992)133-145.Znizz, C. R., KALISH, S. 13., ERIII.IUII, K. S.,LEvrrz, D., METZGER, E., RADIN, R. and NAIR, J.P. IgG antibody to purified protein derivative byenzyme-linked immunosorbent assay in the diag-nosis of pulmonary tuberculosis. Am. Rev. Respir.Dis. 130 (1984) 845-848.

27. KIRAN, U., SIIRINIWAS, KUMAR, R. and SHARMA,A. Efficacy of three mycobacterial antigens in theserodiagnosis of tuberculosis. Eur. J. Respir. Dis.66 (1985) 187-195.

28. LEFFoub, M. J. Immune response to mycobacte-ria: diagnostic methods. In: Manual (4' ClinicalLaboratory Immunology. 4th edn. Rose, N. R.,

^

Conway de Macario, E., Fahey, J. L., Friedman,^36.H., Penn, G. NI., eds. Washington, D.C.: AmericanSociety for Microbiology, 1993, pp. 489-497.

29. LoND, A. and Ripo.f., NI. Immunologically baseddiagnostic tests; Immoral antibody methods. In:The Mycobacterias a Sourcebook. Kubica, G. P.and Wayne, L. G., eds. New York: Marcel Dekker,Inc., 1984, pp. 221-248.

30. NOORDEEN, S. K. Leprosy 1962-1992. 4. Epi-demiology and control of leprosy-a review ofprogress over the last 30 years. Trans. R. Soc."Prop. Med. Hyg. 87 (1993) 515-517.

31. Noomium:K, G. T., Koi.k, A. II., 13ifirsi, G.,CA1TY, D., DAL F, J. W., FINE, P. E. M., GODFREY-FAUSSETT, P. G., Cfm, S.-N., SIIINNICK, T., SVEN-SON, S. B., WILSON, S. and VAN Entimov, J. D. A.Sensitivity and specificity of PCR for detection ofMycobacterium tuberculosis: a blind comparisonstudy among seven laboratories. J. Clin. Micro-biol. 32 (1994) 277-284.

32. PEREZ-FERNANDEZ, L., GARCIA-LATORRE, E., SOSADE MARTINEZ, C., LOPEZ DE ARCILA, C. and AR-

^CILA-CASTANO, L. E. Utilidad de Ia tecnica de in-^40.munoensayo enzinuitico ELISA en el diagnOsticode Ia tuberculosis pulmonar en el Milo. Acta. Pedi-atr. Mex. 15 (1994) 55-63.

33. SCIILUGER, N. W. and Rom, W. N. Current ap- 41.proaches to the diagnosis of active pulmonary tu-berculosis. Am. J. Respir. Crit. Care Med. 149(1994) 264-267.

34. STYI31D, K. Overview and epidemiologic assess-ment of the current global tuberculosis situation with