Value of ELISA Using Antigen 60 for the Diagnosis of Tuberculosis in Children

Value of ELISA Using Antigen 60 for the Diagnosis of Tuberculosis in Children* Christophe Delacourt, M.D.; ]ovana Gobin, M.D.; Jean-Louis Gaillard, M.D.; jacques de Blic, M.D.; Michel Veron, M.D. ; and Pierre Scheinmann, M.D. We evaluated the possible value of enzyme-linked immunosorbent assay (ELISA) using antigen 60 (A60) for improved diagnosis of tuberculosis in children. Results obtained in 31 children with active tuberculosis and in 16 patients with tuberculous infection without disease were compared with the results of 198 control subjects with no mycobacterial disease. In control children, anti-A60 IgG increased with age and the optical density (OD) in ELISA assays rose from .079± .053 (OD ± SD) in children younger than 5 years old to 0.146±0.082 ODin children older than 5 years. In control subjects younger than 2 years old, IgG OD values were significantly higher in BCG-vaccinated children than in nonvaccinated children. At a chosen specificity of 98 percent, a positive serodiagnosis was observed in 68 percent of children with clinically active

tuberculosis. In these children with active disease, smears were positive in only 26 percent of cases and mycobacterial cultures yielded the organism in 45 percent of cases. None of the infected children without disease had high lgG OD values. IgM measurements were also evaluated. Mean values from control and diseased children overlapped, leading to a low sensitivity ( 19 percent) in children with clinically tuberculosis active. We conclude that anti-A60 IgG measurement is a rapid and low-cost technique that enhances the diagnosis of clinically active tuberculosis in children and may distinguish recent infection without disease from infection with disease. (Chest 1993; 104:393-98)

Tuberculosis remains a major health problem in developed countries, and its incidence increased in the 1980s both in the United States 1 and in Europe. The incidence of tuberculosis in Paris (France) and its surrounding region was 33/100,000 in 1989. 2 The influx of high-risk individuals and the prevalence of human immunodeficiency virus infection may largely account for this. 3 The growing number of new adult cases naturally represents a source of infection for children. The diagnosis of tuberculosis in children can be difficult and is usually indirect and based on epidemiologic considerations. 3 A history of recent exposure

shown its usefulness in adults, using either crude bacillary antigens, 7 tuberculin purified protein derivatives,H or purified mycobacterial antigens. 9 - 11 In particular, the use of antigen 60 (A60) provided high diagnostic values. 12 However, there has been no adequate evaluation of this approach for children. We therefore assessed the usefulness of ELISA using A60 in the diagnosis of tuberculosis in children and the diagnostic values of both IgG and lgM measurements.

For editorial comment see page 329 to a subject with the active disease, the results of tuberculin skin tests, chest radiographs, and physical examinations are often the only supports for the diagnosis. Fiberoptic bronchoscopy may be also helpful for the initial diagnosis. 4 The recovery of tubercle bacilli, which would establish the diagnosis with certainty, is difficult in children. 5 Children rarely produce sputum and patients with minimal pulmonary disease rarely have positive smears. Other laboratory analyses are thus required to improve the diagnosis of tuberculosis in children. Enzyme-linked immunosorbent assay (ELISA) serodiagnosis of tuberculosis is a potentially valuable technique .6 Many studies have *From the Service de Pneumologie et Allergologie de l"enfant , Departement de Pediatrie (Drs. Delacourt, de Blic, and Scheinmann), and the Service de Bacteriologie (Drs. Gohin, Gaillard, and Veron), Hopital Necker-Enfants Malades, Paris, France. Manuscript received Mav 13; revision accepted Decem her 8 Reprint requests: Dr: Delacourt, Service du Pr Scheinmann, Ho11ital tks Enfant Malatks, 149 Rue tk Sevres, 75015 lllris, France

A60 =antigen 60 of BCG; EUSA =enzyme-linked immunosorbent assay; OD=optical density

METHODS

Population Sennn samples were ~"llected from 245 children under inw-stigation for pulmonary and allergic diseases in our pediatric department. Thirty-one of the children had tuherculous diseast• (20 hoys and 11 girls) with a mean age of5.7±5.3 (SD) years (range, 0.7 to 16.5 years), and 16 had tuherculous infection without disease (6 hoys and 10 girls) with a mean age of 4.8± 4.2 years (range, 0.1 to 16.4 years). The remaining 198 suhjects had rw my~"hacterial disease and were c:.1msidered as control suhjects (122 hoys and 76 girls). The mean aJ?;e of c:.1mtrol subjects was 3.8±4.2 years (ranJ!:e, 0.05 to 17 years). Tuherculosis was suspected on history of close contact with aetive tuherculosis, clinical and radiographic findings, and the results of tuberculin skin tests. Diagnosis was stated according to the re~"m­ mendations of the American Thoracic Societ):"' AmonJ!: tht• 16 children with tuherculosis infection hut no diseast•, II had ht>en recently and massively exposed to active pulmonary disease and had an increase of at least 10 mm in their intradermal skin test. The five remaining infected children had no known t1mtaet with pulmonary disease , hut had positive intradermal skin tests su~·rior to 15 mm in the absence of BCG vaccination. Twenty-nine of the 31 children diagnosed as having clinically active tuberculosis had ahnormal chest radiographs. The other two children presented with an isolated peripheral lymph node and an isolated involvenwnt of middle ear. Mycobacterium tuberculosis was isolated in cultures from 14 of the 31 diseased children. Eight of thest' 14 childn•n had CHEST I 104 I 2 I AUGUST, 1993

393

Table l-Characteriatics of Children With Tuberculoris Clauified Into Three Groupa: Children With Tuberculoua Infection, But No Disease; Children With Tuberculoua Disease But Negative Cultures for M tuberculosis; Children With Bacteriologicallg Confirmed Tuberculoua Disease (Positive Culture) Tuberculous Infection No Disease (n = 16)

Tuberculous Disease Negative Culture (n = 17)

Tuberculous Disease Positive Culture (n = 14)

Total (n=47)

6110

1314

7n

26121

10 6 0

9 6 2

5 6 3

24 (51%) 18 (38%) 5 (11%)

9 5 1 0

9 5 I 0 2

3 2 7

21 12 9 1 4

Sex, M/F BCG vaccination (n) Positive Negative Unknown Ethnic origin (n) European North African Black African South American Asian

positive smears. The sex, BCG vaccination history, and ethnic origin of the subjects in each group are listed in Table 1. All blood samples were collected prior to treatment. Among the children with no mycobacterial disease and no known contact with active tuberculosis, the diagnoses were as follows : acute pulmonary infection (bacterial or viral) (n = 78), asthma or other allergic diseases (n =54), chronic brqnchopulmonary disease (n = 36), extrapulmonary acute infection (n = 13), congenital malformation (n = 9), and miscellaneous disorders (n = 8). One hundred forty-seven (74 percent) of the children were ethnically European and 51 (26 percent) came from developing countries. One hundred fifty-seven control subjects (mean age 4.i ± 4.3 years) had previously been vaccinated with BCG and 41 (mean age 0.6±0.6 year) had not . All the nonvaccinated control subjects were younger than 2 years of age, except 1 child aged 3.5 years. An intradermal tuberculin skin test was performed in 112 of ~he 157 control vaccinated children after their serum was collected . Technique for Antibody Determination Serology was analyzed by ELISA. Diagnostic kits were supplied by Anda Biologicals, Strasbourg, France . Antigen 60 was extracted from the cytoplasm of Mycobacterium bovis BCG and purified a(:cording to Cocito and Van linden." The antigens were coated on microtiter wells. Samples of sera diluted 1:100 in a dilution buffer (included in the kit) were added to the wells and incubated for I h at 37"C. All samples were pretreated with a preparation of antihuman lgG sheep antibodies (Absorbant RF, Behring, Germany) that eliminates rheumatoid factors. The plates were washed in buffer and then incubated with peroxidase anti-human IgG or peroxidase anti-human lgM for 30 min at 37"C. After rewashing, 0.1 ml of tetramethyl-benzidin (TMB) was added to the wells and the plates were incubated for 10 min at room temperature in darkness. The reaction was stopped with 0.1 ml 0.5 N H,SO,. The optical density (00) was read in a microtiter reader (1itertek Multiscan) at 450 nm . The following controls were included on every plate: negative control , low positive control (OD =0.300), high positive control (OD = 0.600), and two sera from patients with documented tuberculosis and 00 values superior to 0.400. Data Analysis First, the data from the control population were analyzed to define cut-off antibody 00 values chosen to correspond to better than 98 percent specificity (the fraction of children without the disease and having a negative test result). Second, the results from children with tuberculosis were assessed . Sensitivity (the fraction of children with tuberculosis who gave a PoSitive test result) and positive predictive value (the fraction of children giving a positive test result who had the disease) were calculated by standard

394

(45%) (26%) (19%) (2~)

(8%)

methods" from the defined cut-off points. All values are reported as mean± SO . Antibody 00 values were compared between two groups by the Mann-Whitney U test. A p
Antibody L£vels in Control Children Influence of BCG Vaccination: Since BCG vaccination is obligatory in France before children go to day nursery, most infants are vaccinated before they are 2 years old. This was reflected in our population. We therefore only compared vaccinated and nonvaccinated children younger than 2 years old. The mean age was 0.5 ± 0.4 year in nonvaccinated children (n = 40) and 0.8±0.5 year in vaccinated children (n=58). lgG and lgM OD values in nonvaccinated children were 0.054±0.019 (range, 0.019to0.138) and 0.167 ±0.118 (0.027 to 0.574) respectively, significantly lower than in the vaccinated group: 0.075 ± 0.050 (0.023 to 0.280) (p<0.03) and 0.277 ± 0.228 (0.054 to 1.059) (p<0.003), respectively. Influence of Age on Antibody OD Values: The distribution of individual values showed that most of the higher OD values corresponded to children older than 5 years. We therefore divided the 157 control vaccinated children into 2 age groups: <5 years old (n = 96) and >5 years old (n = 61), and we compared the results between these 2 groups. The mean IgG OD value was significantly higher in the group of older children than in the group of younger children: 0 . 146 ± 0.082 and 0.079 ± 0.053, respectively (p=0.0001). In contrast, lgM OD values in each age group were widely scattered and the mean values of the two groups were similar: 0.323 ± 0.178 in children older than 5 years old and 0.305 ± 0.234 in children younger than 5 years old. Absence of Influence of Ethnic Origin: The mean ages of control children originating from Europe (4. 0 ± 4. 3 years) and those originating from developing countries (3.3 ± 3.8 years) were not significantly different. The mean lgG OD value was similar in both Diagnosis of Active Tuberculosis in Children (Delacourt et at)

groups: 0.094 ± 0.068 and 0.098 ± 0.071, respectively. Mean lgM values were also not significantly different, 0.273 ± 0.186 and 0.307 ± 0.253, respectively. Relationship Between Intradermal Skin Test Results and Antibody OD Values: Intradermal skin tests were performed on 112 control vaccinated children. Fiftyfive gave positive results (induration of 5 mm or more in diameter) and 57 gave negative results. The mean IgG and lgM OD values for children with positive intradermal skin tests were not significantly different from those for children with negative tests: IgG OD values were 0.107 ± 0.071 and 0.102 ± 0.070, respectively; lgM OD values were 0 .365 ± 0.236 and 0.311 ± 0.217, respectively. The Underlying Disease Had No Effect on Antibody OD Values: In each of the different age groups, mean lgG and IgM OD values did not significantly differ between the different categories of diagnosis.

younger than 5 years old; and vaccinated children older than 5 years old. If we require a specificity over 98 percent for this test, the cut-off OD values for the 3 groups would be 0.150, 0.225, and 0.310, respectively, giving specificities of 100 percent, 98 percent, and 98 percent, respectively. The overall specificity for IgG determination would be 0.985. The wide scattering of lgM values made classification into groups difficult. We therefore defined only 2 groups: nonvaccinated children younger than 2 years old and others. Cut-offOD values of0.500 and 0.960, respectively, give an overall specificity of0.980. None of the four control children with lgM OD values over the cut-off points had elevated IgG OD values. Results in Children With Tuberculosis Children with tuberculous disease were significantly older than control subjects. Direct comparison between mean antibody 0 D values was therefore uninformative. The mean age of infected children without disease was similar to that of the control subjects. The proportion of BCG-vaccinated children was significantly higher in control subjects than in tuberculous children. The antibody OD values for each child with tuberculosis were compared with the cut-off values for the appropriate control group: 10

Definition of Cut-Off Points The influence of BCG vaccination in young infants on the antibody OD values and the variation of OD values with increasing age must be taken into account in the definition of cut-off points. The IgG values of our control population fell into 3 groups: nonvaccinated infants younger than 2 years old; vaccinated children

8

C

0

(!)

C)

8

c

I I I I •

•

0.8

•

0.6

•

0.2 0 0

• •• 0

0

•

I

I

di:n

w

0 0

r--------------, BCG- and< 2 yrs

'• • •

•

•

0.4

0

[J

•

1.0 0

A

I I I

1.2

-

D

BCG+ and < 5 yrs

• 0

•

•

'

8

£

Q)

CD

0

BCG ... and > 5 yrs

FIGURE 1. Individual (single circles) IgG OD values in control children (column A), in children with tuberculous infection but no disease (column B), in children with tuberculous disease but negative culture (column C), and in children with tuberculous disease and positive culture for M tuberculosis (column D). Data points are displayed according to the cut-olf points defined in the control population: BCG nonvaccinated and <2 years old, BCG vaccinated and <5 years old, and BCG vaccinated and ~5 years old. Open circles represent children under the de fined cut-olf points and closed circles represent the children above these points.

CHEST I 104 I 2 I AUGUST, 1993

395

A

B C D

I I I

1.6

-c -

0

•

1.4

1.2

•

1.0

•

0.8

0.6

•

•

0.4 0.2

0 0

9

0 0

0

• •

t

J

0

BCG- and < 2 yrs

0

0

8

cib

f 0 0

0

0

•

0

0

~

:,

<8

0

(I)

Vaccinated children

FIGURE 2. Individual (single circles) lgM OD values in control children and in patients. Columns A, B, C , D as in Figure I . Data points are displayed according to the cut-off points defined in the control population: BCG nonvaccinated and <2 years old and BCG vaccinated children. Open circles and closed circles are as in Figure 1.

children had not been vaccinated and were younger than 2 years old, 15 children were BCG vaccinated and younger than 5 years old, and 13 children were BCG vaccinated and older than 5 years old. For the 4 nonvaccinated children suffering from tuberculosis and older than 2 years old and for the 5 children with an unknown BCG status, we used the cut-off defined for vaccinated children of same group of age. IgG Measurements (Fig 1): None of the 16 children with tuberculous infection but no disease had a positive serodiagnosis (mean value, 0.084 ± 0.069 OD). In contrast, 11 of the 17 children with tuberculous disease but negative cultures (mean value, 0.308 ± 0.219 OD) and 10 of the 14 children with positive cultures (mean value, 0 .440±0.350 OD) had IgG OD values above the cut-off points. In particular, four of the six children with negative smears but positive cultures had high IgG OD values. The sensitivities in children with tuberculous disease were 0 .647 (negative culture) and 0.714 (positive culture). The overall sensitivity was thus 0 .677 for active tuberculosis and the positive predictive value was 0 .875. lgM Measurements (Fig 2): None of the infected children without disease had high lgM OD values (mean value, 0.406±0.201 OD). Only one child with active tuberculosis but negative culture (mean value, 0.318 ± 0.337 OD) and five children with bacteriologically confirmed active t uberculosis (mean value , 0.593 ± 0 .386) had an lgM OD value above the cut-off 396

points. The overall sensitivity was therefore 0.194 in active tuberculosis, and the positive predictive value was 0.667. Five of these six children with high lgM OD values also had elevated lgG OD values. DISCUSSION

The diagnosis of tuberculosis in children can be difficult and is delayed in many cases. Bacterial proof of the infection, and especially positive smears, are rarely obtained. New methods are thus required for improved and more rapid diagnosis of the disease in children. Previous reports demonstrated that serodiagnosis of tuberculosis may be useful in adults, 6 • 11 • 12 especially when using A60. 10 However, the value of this approach for diagnosis in children has been poorly investigated. Our results show first that antibody OD values in control children are influenced by a prior BCG vaccination and by age . Second, measurements of anti A60-IgG OD values may be helpful for the diagnosis of active tuberculosis in children and may separate diseased children from infected children without disease. Third, lgM measurements in children with tuberculosis provide no useful results because of the great degree of overlap with control patients.

Interpretation of Antibody OD Values in Control Children No data concerning the results of ELISA serodiagnosis were previously available in children without Diagnosis of Active Tuberculosis in Children (Delacourt eta/)

tuberculosis. Both BCG vaccination and age appear to inAuence the antibody OD values in our control population. In contrast, neither the ethnic origin nor the underlying disease affected the ELISA reactions. The increase of anti A60-IgG and IgM OD values after BCG vaccination is not surprising since A60, purified from the cytoplasm of M bovis BCG, induces the formation of anti-A60 antibodies when injected subcutaneously into mice.l 6 Furthermore, anti-A60 IgG OD values increase in healthy adult subjects three weeks after revaccination. 17 lgG OD values also increase significantly with age . Our results are thus consistent with the higher mean anti A60 IgG OD value (0.230 ± 0.130) obtained in control adults by Charpin et al. 10 Since A60 is not strictly species specific, but is characteristic of microorganisms, including the genera Corynebacterium , Mycobacterium, and Nocardia, 14 it is most likely that repeated contacts with environmental mycobacteria may lead to the observed increase in antibody OD values with age. 17 This explanation of inapparent infections caused by environmental mycobacteria and leading to crossreactivity with the antigens of tubercle bacilli obliges us to establish cut-off lines for serologic tests according to local control values. 12 Finally, the absence of significant correlation between antibody levels and tuberculin skin test positivity in the control group is consistent with previous data. 8 It underlines that these two tests do not reAect the same defense mechanism against tuberculosis.

Usefulness of lgG Measurements for the Diagnosis of Active Tuberculosis in Children The diagnosis of tuberculosis in adults is largely reliant on bacteriology. The recovery of tubercle bacilli from children, however, is difficult and diagnosis is therefore mainly indirect. 3 Children rarely produce sputum, and gastric aspirates yield the organism in only 30 to 40 percent of cases of active pulmonary tuberculosis. 18 Tubercle bacilli had been recovered in 14 children (45 percent) in our population but smears were positive in only 8 cases (26 percent). Thus, in 23 of the 31 children with tuberculous disease, bacteriology did not help initial diagnosis. Furthermore, interpretation of tuberculin skin test may be particularly difficult in BCG-vaccinated patients since there are no satisfactory criteria of true conversion in this population. 19 The ELISA is a rapid and inexpensive procedure that greatly improves the diagnosis of active tuberculosis in children. Twenty-one of the 31 children with active disease had IgG OD values above the cutoff point. In particular, 15 of the 23 children with negative smears (65 percent) had a positive serodiagnosis. The ELISA serodiagnosis with A60 is thus one of the most sensitive, high-specificity laboratory tests available for the diagnosis of active tuberculosis in

children. Indeed, most previous studies of ELISA serodiagnosis in children have been disappointin~ and thus, highlight our results. Rosen2<1 used mycobacterial sonicates, but sensitivity (20.7 percent) and specificity (39.5 percent) were poor, essentially due to crossreactivity with BCG. More promising was the study of Barrera et al 21 using purified protein derivati\'t' and showing a specificity of 98 percent and a sensitivity of 51 percent in bacteriologically confirmed cases. The best results were obtained by Aide et al 22 using antigen 5. They found a specificity of 100 percent and a sensitivity of 85.7 percent in bacteriologically confirmed tuberculous children, but the low number of control subjects (n = 19) limits the interpretation of this study. Furthermore, our study adds that serologic study with A60 can distinguish recent infection without disease from infection with disease and contributes to the resolution of one of the major problems in diagnosing tuberculosis in children. Indeed, chest radiographic findings can he subtle, and their interpretation determines whether the child receives preventive or curative therapy. Our results show that high IgG OD values are obtained only in diseased children and not in infected children without disease . Thus, in front of doubtful radiographic interpretation, high IgG OD values invite us to propose a multiple drug therapy. In this study, we did not include children with diseases such as cervical adenitis caused by environmental mycobacteria. Thus, the possible distinction between tuberculosis and diseases due to other mycobacteria by anti-A60 antibodies remains unknown.

lgM Measurements Do Not Help Diagnosis in Children Most previous studies have shown that lgM measurements using antigens other than A60 do not correlate with the presence of tuberculosis. Kardjito et al,7 using crude bacillary antigens, concluded that IgM measurements were not helpful in the diagnosis of tuberculosis. Radin et al~ came to similar conclusions using a tuberculin purified protein derivative . In both studies, patients with tuberculosis overlapped the control groups, and very high IgM titers were detected in some control subjects. In contrast, Charpin et al, 10 using A60, found that IgM measurements were more discriminating than IgG measurements in adults. Our data from children do not confirm the findings of Charpin et al but are consistent with the former studies. Indeed, the lgM OD values in the control group were widely divergent and IgM measurements in children with tuberculosis had a very poor diagnostic sensitivity. Technical factors could not satisfactorily explain this difference . In particular, rheumatoid factors capable of causing false-positive results were eliminated by our technique. We must therefore speculate that true differences exist between children and adults regarding anti-A60 IgM levels. Thus, only CHEST I 104 I 2 I AUGUST, 1993

397

6 of the 31 children with active disease had lgM OD values above the cut-off points. Of these six children, five also had high lgG OD values. This combination ofboth high IgG and IgM values was not found in any of the control patients. The only value of lgM measurements would be therefore to confirm a diagnosis, based on high IgG OD values. In conclusion, ELISA serodiagnosis of tuberculosis using A60 is useful in children, especially in cases of active disease with negative smears. Its low cost ($25) and rapidity greatly enhance its value. The discriminatory power of the test is essentially dependent on IgG determinations, and IgM measurements are of little value. Newer promising methods, such as the specific detection of M tuberculosis DNA by polymerase chain reaction, would also greatly improve the diagnosis of tuberculosis in children, but the higher cost and the technical requirements of this method keep intact the advantages of the serodiagnosis. REFERENCES

1 Nemir RL, Krasinski K. Tuberculosis in children and adolescents in the 1980s. Pediatr Infect Dis J 1988; 7:375-79 2 Quenum B, Hubert B, Grossoet J. La tuberculose en France de 1970 a 1989. BEH 1990; 3:10-1 3 Starke JR. Modern approach of the diagnosis and treatment of tuberculosis in children. Pediatr Clin North Am 1988; 35:44164 4 De Blic J, Azevedo I, Burren CP, Le Bourgeois M, Scheinmann P. The value of fiheroptic bronchoscopy in childhood pulmonary tuberculosis. Chest 1991 ; 100:688-92 5 Inselman LS, Kendig EL. Tuberculosis. In: Chernick V, Kendig EL, eds. Disorders of the respiratory tract in children. 5th ed. Philadelphia: WB Saunders Co, 1990; 730-69 6 Daniel TM, Debanne SM . The serodiagnosis of tuberculosis and other mycobacterial diseases by enzyme-linked immunosorhent assay. Am Rev Respir Dis 1987; 135:1137-51 7 Kardjito T, Handoyo I. Diagnosis of active tuberculosis by immunological methods: I. The effect of tuberculin reactivity and previous BCG vaccination on the antibody levels determined by ELISA. Tuhercle 1982; 63:269-74 8 Radin RC, Zeiss CR, Phair JP. Antibodies to purified protein derivative in different immunoglobulin classes in the diagnosis of tuberculosis in man. Int Arch Allergy Appl Immunol 1983;

398

70:25-9 9 Chan SL, Reggiardo Z, Daniel TM, Girling DJ, Mitchison DA. Serodiagnosis of tuberculosis using an ELISA with antigen 5 and a hemagglutination assay with gly<.'Oiipid antigens. Am Rev Respir Dis 1990; 142:385-90 10 Charpin D, Herbault H, Gevaudan MJ, Saadjian M, DeMicco P, Arnaud A, et al. Value of ELISA using A60 antigen in the diagnosis of active pulmonary tuberculosis. Am Rev Respir Dis 1990; 142:380-84 ll Bothamley GH, Rudd R, Festenstein F, Ivanyi J. Clinical value of the measurement of Mycobacterium tuberculosis specific antibody in pulmonary tuberculosis. Thorax 1992; 47:270-75 12 Cocito CG. Properties of the my<.~>bacterial antigen (.'()mplex A60 and its application to the diagnosis and prognosis of tuberculosis. Chest 1991; 100:1687-93 13 American Thoracic Society. Diagnostic standards and classification oftuherculosis. Am Rev Respir Dis 1990; 142:725-35 14 Cocito C, Vanlinden F. Preparation and properties of antigen 60 from Mycobacterium bovis BCG. Clin Exp Immunol 1986; 66:262-72 15 li>man K. Sensitivity, specificity and predictive value of diagnostic tests. Bull Int Union Tuberc 1981; 56:18-28 16 Cocito C, Baelden MC, Benoit C. Immunological properties of antigen 60 of BCG: induction of humoral and cellular immune reactions. Scand J lmmunol1987; 25:579-85 17 Maes R. Incidence of inapparent active mycobacterial infections in France detected by an IgG serological test based on antigen 60. Med Microbiol Immunol1989; 178:315-21 18 Lin(.'()ln EM, Harris LC, Bovornkitti S, Carretero RW Endobronchial tuberculosis in children: a study of 156 patients. Am Rev Tuherc 1958; 77:39-61 19 De March-Ayuela P. Choosing an appropriate criterion fi>r true or false conversion in serial tuherculin testing. Am Rev Respir Dis 1990; 141:815-20 20 Rosen EU. The diagnostic value of an enzyme-linked immune sorhent assay using adsorhed mycobacterial sonicates in children. Tubercle 1990; 71:127-30 21 Barrera L, Miceli I, Ritacco V, Torrea G, Broglia B, Botta R, et al. Detection of circulating antibodies to purified protein derivative hy enzyme-linked immunosorhent assay: its potential for the rapid diagnosis of tuberculosis. Pediatr Infect Dis J 1989; 8:763-67 22 Aide SLM, Pinas(.'() HM, Pelosi FR , Budani HF, Palma-Beltran OH, Gonzalez-Montaner LJ. Evaluation of an enzyme-linked immunosorbent assay (ELISA) using an IgG antibody to Mycobacterium tuberculosis antigen 5 in the diagnosis of active tuberculosis in children. Am Rev Respir Dis 1989; 139:748-51

Diagnosis of Active Tuberculosis in Children (Delacourt eta/)