Serologic Diagnosis of Tuberculosis

Using a Simple Commercial

Multiantigen Assay*
Mark D. Perkins, MD; Marcus B. Conde, MD; Marneili Martins, MD;
Afranio L. Kritski, MD, PhD

Setting: Seven primary health clinics and a pulmonary disease specialty clinic in Rio de Janeiro
City, Brazil.
Objective: To evaluate a commercial immunochromatographic test kit (ICT Tuberculosis;
AMRAD-ICT; Sidney, Australia) employing five recombinant Mycobacterium tuberculosis pro-
teins for the detection of pulmonary tuberculosis (TB).
Design: Serology test results were compared with duplicate sputum microscopy and culture in 277
patients with symptomatic pulmonary disease (243 with pulmonary TB and 34 with nontubercu-
lous disease). An additional 110 healthy control subjects were also tested.
Results: The serology test was simple and rapid to perform and detected 64.2% of smear-positive
and 46.3% of smear-negative TB patients overall. HIV co-infection was present in 15.3% of TB
patients, and serology was much less sensitive (overall 27.6%) in this small group, as was
microscopy (13.8%). Specificity of the serology test was 100% in healthy control subjects and
85.2% in the small number of control patients with pulmonary disease, including those with prior
TB. Combined with microscopy, serology detected 72.8% of TB patients.
Conclusion: Depending on the population studied, multiantigen serologic tests for TB may be as
sensitive as microscopy, but detect a different and overlapping subset of patients. The use of
multiple antigens in this kit increased test sensitivity without significant loss of specificity. Bacille
Calmette-Guérin vaccination and tuberculin sensitivity did not affect serology results. Estimating
specificity in clinical use will require testing a much larger cohort of symptomatic patients with
nontuberculous disease. The TB diagnostic performance of this group of antigens in HIV
co-infected individuals was poor. (CHEST 2003; 123:107–112)

Key words: developing countries; diagnosis; serology; tuberculosis

Abbreviations: AFB ⫽ acid-fast bacilli; BCG ⫽ bacille Calmette-Guérin; CI ⫽ confidence interval; PPD ⫽ purified
protein derivative; TB ⫽ tuberculosis

T hecost-effective
treatment of tuberculosis (TB) is a highly
medical intervention. Selection of
patients for treatment, however, remains compli-
cated by the lack of specificity of clinical findings and
the poor performance characteristics of diagnostic
methods available in the majority of developing-
*From the Department of Medicine (Dr. Perkins), Duke Uni-
versity Medical Center, Durham, NC; and Tuberculosis Re-
search Unit (Drs. Conde, Martins, and Kritski), Chest Service,
Clementino Fraga Filho Hospital, Federal University of Rio de
Janeiro, Rio de Janeiro, Brazil.
The study received partial financial support from AMRAD ICT;
CNPq-processo (35 04 77/95. 7), NIH Fogarty AIDS Interna-
tional Training Grant (5 D43 TW00018), and Fogarty Interna-
tional Center (TBITRP 5 D43 TW00018).
Manuscript received November 28, 2000; revision accepted July
12, 2002.
Correspondence to: Mark D. Perkins, MD, World Health Orga-
nization, Tropical Diseases Research (TDR), 20 Appia Ave,
CH-1211 Geneva 27, Switzerland; e-mail: perkinsm@who.int
world laboratories. Many attempts have been made
to develop a serologic TB test, challenges to which
include the need to discriminate active from latent
infection, to avoid cross-reactivity to bacille
Calmette-Guérin (BCG) or nontuberculous myco-
bacteria, and to perform consistently in genetically
and immunologically diverse populations.
Early studies using partially purified antigens
showed that antimycobacterial antibody was readily
detected in tuberculosis patients, but test specificity
was poor.1,2 Monoclonal antibodies and highly puri-
fied native or recombinant antigens have improved
specificity, but at a cost of decreased sensitivity.
Recently, the degree to which the humoral response
to TB is heterotypic has received greater recognition,
stimulating the development of serology tests using
multiple antigens.3
Though sensitivity, specificity, and cost are the

www.chestjournal.org CHEST / 123 / 1 / JANUARY, 2003 107

Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21987/ on 04/20/2017

 characteristics most frequently considered in TB
diagnostic tests, robustness and speed are also criti-
cal to field effectiveness. Delays of even 1 to 2 days
result in loss of patients to follow-up, increased cost
from return visits, and diminished physician control
and test confidence.4,5 A diagnostic tool for TB that
could be performed rapidly in the field would be of
great benefit to tuberculosis control programs, espe-
cially in developing countries, even if test perfor-
mance were less than ideal. In this report, we
evaluate a rapid immunochromatographic test kit to
detect antibody directed against any one of five
recombinant antigens in TB patients in Rio de
Janeiro City, Brazil. Rio de Janeiro, the fourth-
largest city in Latin America, is an evolving mega-city
with an expanding epidemic of TB. There were
10,210 cases of TB registered in the city in 1997, and
a incidence rate based on passive case finding of 120
per 100,000 population citywide.6
Materials and Methods
Patient Enrollment
Sera were collected between July 1996 and December 1996
from three populations of patients not receiving antituberculous
therapy: (1) patients with newly diagnosed TB at seven Rio de
Janeiro City primary health clinics; (2) patients with pulmonary
disease of indeterminate cause presenting for evaluation by
induced sputum and/or BAL at Clementino Fraga Filho Hospital,
Federal University of Rio de Janeiro; and (3) healthy subjects,
including asymptomatic household TB contacts and purified
protein derivative (PPD)-positive and PPD-negative health-care
workers at the study site. The control group was thus composed
of both asymptomatic volunteers and patients with pulmonary
disease in whom Mycobacterium tuberculosis was not detected
on mycobacterial culture or microscopic examination of Ziehl-
Neelsen–stained sputum smear. BCG vaccination history was not
recorded, but the age of the patients and the history of universal
BCG vaccination of infants in Brazil predict that between 50%
and 75% would have received the vaccine.
Exclusions
Symptomatic patients for whom no smear or culture data were
available or for whom relevant clinical information was lacking
were excluded from analysis. A small number of samples that did
not yield test results for technical reasons were also excluded.
Clinical and Laboratory Evaluation
All case patients were interviewed and underwent physical and
laboratory evaluation including a chest radiograph. Two sputum
samples from each patient were submitted for microscopic
examination for acid-fast bacilli (AFB) and culture on Löwen-
stein-Jensen media following standard alkaline decontamination.
Smear and culture results were considered positive if either of
the two duplicate samples were positive. HIV testing was per-
formed whenever possible. Asymptomatic control patients under-
went no laboratory evaluation other than tuberculin skin testing.
108
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21987/ on 04/20/2017
Serology
From case patients, serum for testing was drawn at the time of
diagnosis in all but one patient, who was tested 6 months after
initial presentation. Testing was done using a commercial kit
(ICT Tuberculosis; AMRAD-ICT; Sydney, Australia) consisting
of a folding cardboard device containing a nitrocellulose strip
onto which the 38-kd antigen and four other recombinant
antigens were fixed in discrete lines. The construction of the
device and its use has been described elsewhere in detail.7
Briefly, 30 ␮L of serum was placed at one end of the nitrocel-
lulose strip and allowed to migrate across four bands containing
the five fixed antigens. After the serum had reached a marked
limit line, migration was stopped with buffer, and captured
antibody detected using an immunogold goat anti-human IgG
conjugate. Testing required 5 to 15 min, and ease of use was such
that multiple tests could be run concurrently. Interpretation was
performed as suggested by the manufacturer, and a visible line at
any of the four antigen-binding areas was considered positive.
Each positive band was scored for location (A through D) and
intensity of staining (1⫹ to 4⫹). A single investigator read all
strips. Serum was frozen, not fresh, and went through at least one
freeze-thaw cycle prior to use in this test.
Results
Adequate clinical and laboratory information was
available to evaluate 393 of the 494 patients from
whom serum was collected. In six subjects (1.5%),
kits did not yield readable results, either because of
operator error or because of viscous serum that
failed to migrate on the nitrocellulose strip. The
remaining 387 tests formed the basis for our evalu-
ation.
Patients were classified into four groups for anal-
ysis, depending on the results of clinical and labora-
tory findings: (1) patients with smear-positive TB,
(2) patients with smear-negative TB with or without
positive mycobacterial culture, (3) patients with
symptomatic pulmonary disease and no evidence of
TB, and (4) healthy individuals who were hospital
workers or household contacts of pulmonary TB
patients.
Case patients had a mean age of 36.0 years,
whereas healthy control subjects tended to be
younger (29.8 years) and patients with non-TB pul-
monary disease older (55.4 years). The proportion of
men (n ⫽ 158) and women (n ⫽ 85) among the case
patients reflected that routinely reported by the
Brazilian National Tuberculosis Program. HIV test
results were available from 189 of 243 case patients,
of which 29 were positive (15.3%), and from 28 of
the 34 control subjects with nontuberculous pulmo-
nary disease, of which none were positive. HIV
testing was not performed on healthy control sub-
jects.
The results of the ICT Tuberculosis test kit are
shown in Table 1. Overall, 55.1% of TB patients
were test positive with detectable antibodies to one
Clinical Investigations
 Table 1—TB Test Result in Each Study Group and Antibody Detection by Test Antigen Line

% Positive by Individual Antigen Line

Positive Test Result to Any Antigen,
Patient Group Tested No. (%; 95% CI) Line A Line B Line C Line D

All TB patients (n ⫽ 243) 134 (55.1; 48.8–61.4)

Smear-positive TB (n ⫽ 120) 77 (64.2*; 55.6–72.8) 42.5 10.8 20.0 27.5
Smear-negative TB (n ⫽ 123) 57 (46.3*; 37.5–55.1) 28.5 9.8 16.3 13.8
All non-TB patients (n ⫽ 144) 5 (3.5; 0.5–6.5)
Non-TB pulmonary disease (n ⫽ 34)† 5 (14.8; 2.9–26.7) 5.9 2.9 5.9 8.8
Healthy control subjects (n ⫽ 110)‡ 00
*For test-positive proportion in smear-positive and smear-negative patients (p ⫽ 0.005).
†Including prior TB.
‡Including PPD-positive and PPD-negative subjects.

or more of the five recombinant Mycobacterium
tuberculosis antigens in the test. Smear-positive TB
patients were more frequently test positive (64.2%,
p ⫽ 0.005), but nearly half of the smear-negative
patients (46.3%) were also detected by the kit,
including 50% of the 38 smear-negative, culture-
negative patients with diagnoses made on clinical
and radiographic grounds. The duration of symp-
toms ranged from 1 to 58 weeks (average 10.4 weeks)
in 49 of 243 TB patients for whom firm dates were
available. Duration of symptoms in patients with
positive serology results was not significantly longer
than in those with negative results (11.5 weeks vs 8.4
weeks, p ⫽ 0.28).
Band A in the kit used for this study contained the
38 kd antigen. The identity of the other four antigens
is proprietary. As shown in Table 1, the most com-
monly detected antibody in case patients was that
directed against the 38-kd antigen, but antibody to
the other test antigens was present in roughly 10 to
30% of patients. The addition of antigens beyond the
38-kd antigen increased sensitivity of the assay 51%
in smear-positive patients and 63% in smear-
negative patients, without markedly diminishing
specificity. There were an insufficient number of
false-positive test results to evaluate the impact of
multiple antigens on specificity.
The serology test result was positive in 5 of the 144
sera from control patients, all of whom were from
the subset with symptomatic pulmonary disease.
Three of these patients had a history of prior TB.
Two additional symptomatic control patients with no
history of prior TB were also seropositive using the
test kit. One was a patient with bilateral interstitial
lung disease, and the other was a patient with COPD
and lung cancer of unknown type who died within a
few months of presentation. No autopsies were
performed. The test kit results from all five of these
sera classified as false-positive showed either 3⫹ or
stronger scoring on a single band or had multiple
positive bands. Of 113 control subjects who had skin
tests performed, 23 subjects (20.3%) had positive
tuberculin reactions, none of whom had positive
serology test results.
The results of smear, culture, and serology tests
for tuberculosis case patients, divided by HIV-
infection status, are shown on Table 2. All patients
submitted two or more sputum specimens for exam-
ination; AFB microscopy and culture were each
considered positive if any one of the respective
examinations were positive. Laboratory results from
54 TB patients for whom HIV results were unavail-
able were not statistically different from those from
HIV-negative TB patients (p ⬎ 0.5 for all tests), and
these results were pooled for analysis into a single
HIV-uninfected group. Among the 214 HIV-
uninfected TB patients, microscopy was positive in
54.2%, culture in 79.0%, and serology in 58.4%.

Table 2—Results of Duplicate Microscopy (AFB) and Culture and of TB Kit in All TB Patients,
Divided by HIV-Infection Status*

HIV Status TB, No. Positive AFB† Positive Culture† Positive Serology Positive AFB or Serology

Infected 29 13.8 (1.2–26.4) 69.0 (52.2–85.8) 27.6 (11.3–43.9) 37.9 (20.2–55.6)

Uninfected‡ 214 54.2 (47.5–60.9) 79.0 (73.5–84.5) 58.4 (51.8–65.0) 77.1 (71.5–82.7)
Overall 243 49.4 (43.1–55.7) 77.8 (72.6–83.0) 55.1 (48.8–61.4) 72.4 (66.8–78.0)
*Values shown as percentage of test-positive patients (95% CIs).
†Smear and culture considered positive if either of the two samples were positive.
‡Includes 54 patients without HIV test result. Microbiology and serology results from untested group was not statistically different from those of
uninfected (p ⬎ 0.5 for all tests).

www.chestjournal.org CHEST / 123 / 1 / JANUARY, 2003 109

Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21987/ on 04/20/2017

 Among the 29 HIV-infected TB patients, microscopy
was positive in 13.8%, culture in 69.0%, and serology
in 27.6%, but confidence intervals (CIs) were wide.
CD4 counts were not available on enough of these
HIV-infected patients to evaluate any association
between the degree of immunosuppression and the
likelihood of false-negative serology results. Combin-
ing acid-fast sputum smear microscopy with serology
yielded sensitivity of 77.1% in HIV-uninfected TB
patients, which was equivalent to culture (p ⫽ 0.80).
Combined sensitivity of microscopy plus serology
was 37.9% in HIV-coinfected patients and 72.4%
overall.
Discussion
The need for improved TB diagnostics has long
been recognized. Local diagnostic needs may vary
markedly, and it is unlikely that a single diagnostic
test be ideal for all situations. However, an accurate,
simple, rapid, point-of-care assay would be widely
useful and could be expected to improve case hold-
ing, empower health-care workers at the peripheral
level, and decrease diagnostic confusion and delay.
Recognizing these needs, serologic approaches to
TB diagnosis have been tried on and off since the
seroagglutination work of Arloing8 in 1898. Early TB
serology tests used PPD or other crude antigens and
showed relatively poor specificity, as many of the
dominant antibody responses are to shared anti-
gens.9 –12
The availability of monoclonal antibodies, recom-
binant antigens, and methods for the accurate appli-
cation of controlled amounts of immunologic re-
agents to nitrocellulose or other solid matrix have
enhanced the specificity and reproducibility of such
tests.13 The best studied M tuberculosis diagnostic
antigen is the 38-kd antigen, and its availability as a
recombinant expressed in Escherichia coli led to the
development of tests that could easily distinguish
between healthy PPD-positive controls and tubercu-
lous patients.14 –16 Development of immunochro-
matographic strips has put these tests into a form
appropriate for disease endemic countries, though
the change in format from enzyme-linked immu-
nosorbent assay has sometimes been accompanied
by a loss of favorable operating characteristics.17 A
number of simple serologic assays using highly puri-
fied or recombinant proteins are now being mar-
keted in the developing world, many based on
detection of antibody to 38 kd, but limited data exist
to support their use.
Variations in specific antibody responses to myco-
bacterial antigens in different human populations
may be linked to human leukocyte antigen-DR
110
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21987/ on 04/20/2017
phenotype, and tests designed to detect responses to
a single antigen could show important geographic
variability and limited sensitivity.18 –21 To overcome
these problems, multiantigen tests have been devel-
oped.22–24 Loss of specificity is additive, however, so
each of the antigens used must be very highly
specific.
The test kit used in this study employed five
different protein antigens to detect an IgG response
to M tuberculosis. Each individual antigen showing a
specificity of at least 98% (differences not statistically
significant), resulting in an overall specificity of
96.5% in the study group. However, this result is
relatively arbitrary, as antibody was detected in none
of the healthy control subjects, and simply increasing
their number would increase the reported specificity
of the test. Of the 34 patients with nontuberculous
pulmonary disease included here, 5 patients (14.7%)
had a positive serology test result, but because of the
small size of this group, CIs were wide (2.9 to 26.7).
Symptomatic, culture-negative patients in endemic
areas may often have positive results of nucleic acid
amplification assays25 or serology tests for TB, as was
found in studies of a previous version of the ICT
Tuberculosis kit.7,26 Determining the true specificity
of nonculture-based tests in symptomatic patients
will require evaluation of patients from nonendemic
sites and prolonged and careful follow-up of study
patients in disease endemic countries.
The performance characteristics of diagnostic tests
for TB vary widely with the population studied. The
poor performance of microscopy in HIV-coinfected
patients in this study, for example, was due in part to
an enrollment bias with the purposeful inclusion of
smear-negative HIV-infected patients with pulmo-
nary disease undiagnosed at initial evaluation. Dif-
ferences in patient populations studied may also
account for the higher sensitivity found in previous
trials of a serology test using the 38-kd antigen alone
in a similar format to that employed here. Two
studies of a single antigen ICT Tuberculosis kit
carried out in China demonstrated 70 to 90% sensi-
tivity in smear-negative and smear-positive patients,
respectively.7,26 Greater duration or severity of ill-
ness, which have been correlated with likelihood of
positive serologic test for TB, may be one explana-
tion. In the current study, duration of illness did not
correlate with test results in the group from which
such data were available. Comparison with other
published studies of the ICT Tuberculosis test is
confounded by limitations in the available data. For
example, a study in California performed with the
same multiantigen kit used in the present trial found
only 20% sensitivity in a small number (n ⫽ 59) of
culture-positive TB cases.27 The lack of supporting
Clinical Investigations
 clinical and microbiologic data, including an absence
of smear microscopy results, make that study diffi-
cult to interpret.
There were a number of limitations to the current
study. The first is that the serology testing was done
retrospectively, using stored frozen sera that had
passed through at least one freeze-thaw cycle. The
use of fresh serum may increase sensitivity. A second
limitation to the study is the lack of inclusion of
patients with known nontuberculous mycobacterial
infections, as cross-reactive antibodies to antigens
from other mycobacterial species could result in loss
of specificity as mentioned by others.28 Thirdly,
patients from a single geographic region were in-
cluded, limiting the certainty that these data apply to
other regions. Lastly, the study enrolled a limited
number of HIV-coinfected patients and patients
with other comorbidities, decreasing its resemblance
to most field-use conditions. The most pertinent
group, persons with nontuberculous pulmonary dis-
ease, represented a minority of our study subjects,
though among individuals seeking care for respira-
tory symptoms in primary care clinics they generally
greatly outnumber TB patients.
Conclusion
The principal findings of the current study are
that serologic testing for TB with a simple mul-
tiantigen format is feasible in field settings with
minimal technical requirements for the user. The
use of multiple antigens allowed detection of a
significant number of additional patients over
those detected by the 38-kd band alone without a
loss of specificity, but did not result in an overall
sensitivity markedly greater than that reported in
previous studies. In HIV-uninfected patients,
combined use of serology and sputum microscopy
was as sensitive as culture, thus representing an
opportunity to greatly shorten the time to diagno-
sis in a substantial subset of patients. As has been
found in other serologic studies, test performance
in HIV co-infected patients in this study was poor,
albeit superior to smear microscopy.
Specificity was very high in healthy individuals and
was not compromised by tuberculin sensitivity or
childhood BCG vaccination. The positive predictive
value in true use situations, however, was not calcu-
lable due to the artificial composition of the study
population. Larger trials with more representative
populations are needed before the appropriate role
of TB serology can be ascribed.
ACKNOWLEDGMENT: We thank Dr. Draurio Barreira, Rio
de Janeiro City DST/AIDS Program Coordinator; Dr. Guida
Vasconcelos, Rio de Janeiro City Tuberculosis Program Coordi-
www.chestjournal.org
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21987/ on 04/20/2017
nator; Dr. Oscar Berro, Director of Rio de Janeiro State Refer-
ence Laboratory-LACEN; and the health-care workers from the
seven Primary Health Services in Rio de Janeiro City.
References
1 Daniel T. Antibody and antigen detection for the immunodi-
agnosis of tuberculosis: why not? What more is needed?
Where do we stand today? J Infect Dis 1988; 158:678 – 680
2 Grange JM, Lazlo A. Serodiagnostic tests for tuberculosis: a
need for assessment of their operational predictive accuracy
and acceptability. Bull World Health Organ 1990; 68:571–576
3 Genaro M. Tests to replace AFB microscopy in the diagnosis
of smear positive pulmonary diagnosis. WHO Tuberculosis
Diagnostics Workshop. Geneva, Switzerland: World Health
Organization, 1997
4 Foulds J, O’Brien R. New tools for the diagnosis of tubercu-
losis: the perspective of developing countries. Int J Tuberc
Lung Dis 1998; 2:778 –783
5 Squire B. Tests to replace AFB microscopy in the diagnosis of
smear positive pulmonary tuberculosis: meeting report.
WHO Workshop on Tuberculosis Diagnostics, November,
1997. Available at: www.who.int/tdr/publications/publications/
workshop.htm. Accessed December 11, 2002.
6 Cavalcante SC, Pacheco AG, Lauria L, et al. Epidemiologia
da tuberculose no municı́pio do Rio de Janeiro: revisão dos
casos notificados de 1995 a 1997. Boletim de Pneumologia
Sanitária 1998; 6:81–92
7 Cole RA, Lu HM, Shi YZ, et al. Clinical evaluation of a rapid
immunochromatographic assay based on the 38-kd antigen of
Mycobacterium tuberculosis on patients with pulmonary tu-
berculosis in China. Tuber Lung Dis 1996; 77:363–368
8 Arloing S. Agglutination du bacille de la tuberculose vraie.
Comptes Rendues de l’Academie de Sciences 1898; 126:
1398 –1400
9 Daniel TM, Ellner JJ. Immunology of tuberculosis. In: Reich-
man LB, Hershfield ES, eds. Tuberculosis: a comprehensive
international approach. New York, NY: Marcel Dekker, 1993;
75–101
10 Grange JM, Gibson J, Batty A. The specificity of the humoral
immune response soluble mycobacterial antigens in tubercu-
losis. Tubercle 1980; 61:153–156
11 Straus E, Wu N. Radioimmunoassay of the tuberculoprotein
derived from Mycobacterium tuberculosis. Proc Natl Acad
Sci U S A 1980; 77:4301– 4304
12 Straus E, Wu N, Quraishi MAH, et al. Clinical applications of
the radioimmunoassay of secretory tuberculoprotein. Proc
Natl Acad Sci U S A 1981; 78:3214 –3217
13 Hewitt J, Coates ARM, Mitchison DA, et al. The use of
murine monoclonal antibodies without the purification of
antigen in the serodiagnosis of tuberculosis. J Immunol
Methods 1982; 55:205–211
14 Harboe M, Wiker HG. The 38-kd protein of Mycobacterium
tuberculosis: a review. J Infect Dis 1992; 166:874 – 884
15 Anderson AB, Hansen EB. Structure and mapping of anti-
genic domains of protein b, a 38,000-molecular weight pro-
tein of Mycobacterium tuberculosis. Infect Immun 1989;
57:2481–2488
16 Singh M, Andersen AM, McCarthy JEG, et al. The Mycobac-
terium tuberculosis 38-kd antigen: overproduction in Esche-
richia coli, purification and characterization. Gene 1992;
117:53– 60
17 McDonough JA, Sada ED, Sippola AA, et al. Microplate and
dot immunoassays for the serodiagnosis of tuberculosis. J Lab
Clin Med 1992; 120:318 –322
CHEST / 123 / 1 / JANUARY, 2003 111
 18 Lee B, Hefta SA, Brennan PJ. Characterization of the major
membrane protein of virulent Mycobacterium tuberculosis.
Infect Immun 1992; 69:2006 –2074
19 Bothamley GH, Schreuder GMT. Human leukocyte antigen,
tuberculosis, and Mycobacterium tuberculosis-specific anti-
body [letter]. J Infect Dis 1992; 165:598
20 Bothamley GH. Serologic diagnosis of tuberculosis. Eur
Respir J Suppl 1995; 20:676s– 688s
21 Lyashchenko K, Colangeli R, Houde M, et al. Heterogeneous
antibody responses in tuberculosis. Infect Immun 1998;
66:3936 –3940
22 Hoeppner V, Mayers I. The antibody response to primary
tuberculous infection in man. Chest 1989; 95:171S
23 Verbon A, Weverling GJ, Kuijper S, et al. Evaluation of
different tests for the serodiagnosis of tuberculosis and the
use of likelihood ratios in serology. Am Rev Respir Dis 1993;
148:378 –384
112
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21987/ on 04/20/2017
24 Bothamley GH, Rudd R, Festenstein F, et al. Clinical
value of the measurement of Mycobacterium tuberculosis
specific antibody in pulmonary tuberculosis. Thorax 1992;
47:270 –275
25 Roth A, Schaberg T, Mauch H. Molecular diagnosis of
tuberculosis: current clinical validity and future perspectives.
Eur Respir J 1997; 10:1877–1891
26 Zhou AT, Ma WL, Zhang PY, et al. Detection of pulmonary and
extrapulmonary tuberculosis patients with the 38-kd antigen
from Mycobacterium tuberculosis in a rapid membrane-based
assay. Clin Diagn Lab Immunol 1996; 3:337–341
27 Mathur ML, LoBue PA, Catanzaro A. Evaluation of a
serologic test for the diagnosis of tuberculosis. Int J Tuberc
Lung Dis 1999; 3:732–735
28 Freeman R, Magee J, Barratt A, et al. Rapid immunochromato-
graphic assay for diagnostics of tuberculosis: antibodies detected
may not be specific. J Clin Microbiol 1999; 37:2111–2112
Clinical Investigations