IFN indicates interferon; TB, tuberculosis.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Kang YA, Lee HW, Yoon HI, et al. Discrepancy Between the Tuberculin Skin Test and the Whole-Blood Interferon γ Assay for the Diagnosis of Latent Tuberculosis Infection in an Intermediate Tuberculosis-Burden Country. JAMA. 2005;293(22):2756–2761. doi:10.1001/jama.293.22.2756
Context A recently developed whole-blood interferon γ (IFN-γ) assay
based on stimulation with the Mycobacterium tuberculosis–specific antigens early secreted antigenic target 6 and culture
filtrate protein 10 shows promise for the diagnosis of latent tuberculosis
Objective To compare the tuberculin skin test (TST) and the whole-blood IFN-γ
assay in the diagnosis of latent TB infection according to the intensity of
Design and Setting A prospective comparison between the whole-blood IFN-γ assay and
the TST using a 2-TU dose of purified protein derivative RT23 in a population
with intermediate TB burden was conducted sequentially between February 1,
2004, and February 28, 2005, in a Korean tertiary referral hospital.
Participants Of 273 participants, 220 (95.7%) had received BCG vaccine. Participants
were grouped according to their risk of infection: group 1, no identifiable
risk of M tuberculosis infection (n = 99); group
2, recent casual contacts (n = 72); group 3, recent close contacts (n = 48);
group 4, bacteriologically or pathologically confirmed TB patients (n = 54).
Main Outcome Measures Levels of agreement between the TST and the IFN-γ assay and the
likelihood of infection in the various groups.
Results For the TST with a 10-mm induration cutoff, the positive response rate
in group 1 was 51%; group 2, 60%; group 3, 71%, and group 4, 78%. For the
IFN-γ assay, the positive response rate in group 1 was 4%; group 2,
10%; group 3, 44%; and group 4, 81%. The overall agreement between the TST
and the IFN-γ assay in healthy volunteers was κ = 0.16. The odds
of a positive test result per unit increase in exposure across the 4 groups
increased by a factor of 5.31 (95% confidence interval [CI], 3.62-7.79) for
the IFN-γ assay and by a factor of 1.52 (95% CI, 1.20-1.91) for the
TST (P<.001). Using a 15-mm induration cutoff
for the TST did not make a substantial difference to the test results.
Conclusion The IFN-γ assay is a better indicator of the risk of M tuberculosis infection than TST in a BCG-vaccinated population.
Tuberculosis (TB) remains a significant global health problem. Tuberculosis
in humans is the most frequent cause of death from a single infectious agent,
and often causes severe morbidity. Despite the worldwide effort for eradicating
TB, it is responsible for an estimated 8.3 million new infections and 1.8
million deaths each year.1,2 For
effective and efficient treatment of active TB in developing countries, rapid
diagnosis and treatment of patients with sputum smear-positive test results
are the key points in disease control. However, in countries with low or intermediate
rates of TB endemism, the treatment of latent TB infection to prevent progression
to active disease has been an essential component of public health efforts
to eliminate TB.3,4
The tuberculin skin test (TST), which has been used for years for the
diagnosis of latent TB infection, has many limitations, including false-positive
test results in individuals who were vaccinated with BCG and in individuals
who have infections not related to Mycobacterium tuberculosis.5-9 Recently,
a whole-blood interferon γ (IFN-γ) assay based on the M tuberculosis–specific antigens early secreted antigenic target
6 (ESAT-6) and culture filtrate protein 10 (CFP-10)10,11 has
been introduced for the diagnosis of latent TB infection. The whole-blood
IFN-γ assay QuantiFERON-TB Gold (Cellestis Ltd, VIC, Australia) is based
on the elaboration of inflammatory cytokines by T cells previously sensitized
to mycobacterial antigens when they encounter ESAT-6 and CFP-10.12 In
previous studies and when compared with TST, this assay showed high sensitivity
and specificity in patients with active TB and correlated better with exposure
to M tuberculosis.13-17 However,
there has been no study on the usefulness of this assay in an intermediate
TB-burden area. In addition, the utility of the commercialized IFN-γ
assay has yet to be reported.
The aim of this study was to estimate the usefulness of the IFN-γ
assay for the diagnosis of latent TB infection in Korea, where the incidence
of active pulmonary TB is intermediate (92/105 per year) and BCG
vaccination is mandatory.18 We hypothesized
that this assay would correlate better with the risk of infection and the
degree of exposure than the TST and could help to eliminate the limitations
of the TST in this population.
The participants were recruited between February 1, 2004, and February
28, 2005. The protocol was approved by the ethics review committee of the
Seoul National University Hospital (Seoul, Republic of Korea). After providing
written consent, each individual was asked to complete a questionnaire about
his/her possible risk of exposure to M tuberculosis.
All participants were prospectively recruited from the Seoul National University
College of Medicine and from the Seoul National University Hospital.
The participants were classified into 1 of 4 groups that reflected the
risk of infection. Group 1 consisted of healthy medical students without an
identified risk for M tuberculosis exposure. Group
2 consisted of healthy hospital staff with a history of casual contact with
active pulmonary TB patients. Group 3 consisted of individuals who had household
contact with or who had worked in the same rooms as patients with smear-positive
pulmonary TB for longer than 8 hours per day. Group 4 consisted of patients
with active pulmonary TB. Diagnosis was confirmed by culture of M tuberculosis from sputa or by the presence of caseating granuloma
in the lung tissue of patients in whom masslike consolidation was evident
from chest radiographs. Participants in groups 1 through 3 were excluded if
they showed abnormal simple chest radiographs, if they had taken immunosuppressive
drugs during the previous 3 months, or if they had positive test results for
the human immunodeficiency virus. The participants with indeterminate IFN-γ
assay test results were excluded from further analysis.
The sample size was determined by the following factors: the prevalence
of TB infection in Korea, the κ coefficient, α and β errors,
and the “supposed” missing rate. We used a value of 33% for the
TB infection prevalence rate. This is a presumptive value based on the 44.4%
infection rate in 1990 in the sixth nationwide TB prevalence survey of Korea.
We consider the test results from the TST and the IFN-γ assay to be
in agreement when κ coefficients are greater than 0.75. We assumed that
the α error was .05, that the β error was .20, and that the “supposed
missing” rate was 20%. Using these parameters and assumptions, and the
method previously described,19 the minimum
sample size for each group was estimated to be 51.
After collection of blood samples for the IFN-γ assay, the TST
was performed on the volar side of the forearm according to the Mantoux method20 using a 2-TU dose of purified protein derivative
RT23 (Statens Serum Institut, Copenhagen, Denmark), and any induration was
measured in millimeters after between 48 and 72 hours using the ball-point
method.21 The investigator who performed the
TST was blinded to the status of all groups except group 4 (patients with
active pulmonary TB).
The IFN-γ assay was performed in 2 stages according to the manufacturer’s
instructions. First, 1 mL of heparinized whole blood was incubated with aliquots
of antigen-free control and antigens ESAT-6, CFP-10, or phytohemagglutinin
for 16 to 24 hours at 37°C in a carbon dioxide incubator. Then, after
overnight incubation, 200 μL of plasma was removed from each well and
the concentration of IFN-γ was determined using the assay kit according
to the manufacturer’s instructions (Cellestis Ltd). A positive response
value of 0.35 IU/mL of IFN-γ was used as the cutoff.13,16 The
investigator who performed the IFN-γ assay was blinded to the group
status of all participants.
Concordance between test results from the TST and the IFN-γ assay
was assessed using κ coefficients (κ>0.75, excellent agreement; κ<0.4,
poor agreement; and κ between 0.4 and 0.75, fair to good agreement).
Instead of calculating sensitivity and specificity of the TST and the IFN-γ
assay, we measured the correlation of the 2 tests with the risk of latent
TB infection by estimating the odds ratio (OR) and relating the test results
to the likelihood of TB infection.
Given that there is no criterion standard for the diagnosis of latent
TB infection, it is hard to estimate the sensitivities of the TST and the
IFN-γ assay. Although sensitivity could be calculated from patients
with active TB, active pulmonary TB would differ from latent TB infection
in terms of the size of the bacterial inoculum, duration and period of exposure,
and host immune status. In addition, it would be difficult to make an accurate
estimate of the specificity of the 2 tests using Korean populations because
the prevalence of latent TB infection can be as high as 33%. Furthermore,
to adjust for the effects of age, sex, and BCG vaccination status on the TST
and the IFN-γ assay, we performed a logistic regression analysis.
We estimated the increase in odds of a positive test result per unit
increase in exposure by logistic regression. Matched-pairs logistic regression
was used to assess any significant difference in the associations between
the tests. We also calculated an OR adjusted for age, sex, and BCG vaccination
status. Analyses were performed using SPSS version 12.0 (SPSS Inc, Chicago,
Ill) and STATA version 7.0 (STATA Corp, College Station, Tex) software packages.
Group 1 contained 99 participants; group 2, 72; group 3, 48; and group
4, 58. All participants had negative test results for human immunodeficiency
virus. Four patients with active TB (from group 4) were excluded from further
analysis because of indeterminate IFN-γ assay test results. No one in
groups 1 through 3 had an indeterminate result. The demographic characteristics
of the enrolled participants are listed in Table
When a 10-mm induration cutoff was used for the TST, 50 participants
(51%) had positive test results in group 1, whereas only 4 participants (4%)
had positive test results for the IFN-γ assay (κ = 0.08). In group
2, 43 participants (60%) had positive test results with the TST, but only
7 participants (10%) had positive test results with the IFN-γ assay
(κ = 0.14). In group 3, 34 participants (71%) had positive
test results and an induration of 10 mm or more and 21 participants (44%)
had a positive test result with the IFN-γ assay (κ = 0.17).
The overall agreement between the TST and IFN-γ assay in healthy volunteers
(groups 1 through 3) was κ = 0.16. In group 4, 42 participants
(78%) had positive test results with the TST and 44 participants (81%) had
positive test results with the IFN-γ assay (κ = 0.43; Figure). When using a 15-mm induration cutoff,
the correlation between a positive test result on the TST and IFN-γ
assay was κ = 0.13 in group 1, κ = 0.25 in
group 2, κ = 0.25 in group 3, and κ = 0.40
in group 4.
The odds of a positive test result for each increase in risk across
the 4 groups increased by a factor of 5.31 for the IFN-γ assay (95%
confidence interval [CI], 3.62-7.79; P<.001),
by a factor of 1.52 for the TST with a 10-mm cutoff (95% CI, 1.20-1.91; P<.001), and by a factor of 1.74 for the TST with a
15-mm cutoff (95% CI, 1.39-2.18; P<.001). However,
the IFN-γ assay correlated significantly better with the increased risk
of infection across the groups compared with the TST using a 10-mm cutoff
or a 15-mm cutoff (both P<.001). We estimated
the relationship between the TST test results and BCG vaccination status for
groups 1 through 3 using the χ2 test. However, no significant
relationship was found (P = .89 for TST
with a 10-mm cutoff and P = .88 for a 15-mm
cutoff). After adjustment for the effects of age, sex, and BCG vaccination
status, the OR was 1.68 (95% CI, 1.24-2.26) for the TST with a 10-mm cutoff,
1.82 (95% CI, 1.38-2.41) for the TST with a 15-mm cutoff, and 4.23 (95% CI,
2.79-6.41) for the IFN-γ assay (Table 2 and Table 3).
Our study showed poor correlation between the TST and the IFN-γ
assay among healthy volunteers in Korea. In addition, the correlations in
each subgroup of the participants according to the risk of infection were
poor. This poor correlation between the TST and the IFN-γ assay is different
from prior test results. In industrialized countries with low TB endemism,
the reported agreement was good between the TST and the IFN-γ assay based
on the M tuberculosis–specific antigen in TB
contacts (94% concordance in Denmark [κ = 0.87] and 89% in
the United Kingdom [κ = 0.72]).15,16
Given that the proportions of positive test results with the TST (58%
for a 10-mm cutoff and 37% for a 15-mm cutoff) and the IFN-γ assay (15%)
in participants were much different, this discrepancy might be explained by
the false-positive TST results and false-negative IFN-γ assay results.
In Korea, BCG vaccination is given at birth and again at age 12 or 13
years if the child proves to be a TST nonresponder. Most of the participants
in group 1 and 2 had BCG vaccination scars. Therefore, the positive test result
for the TST in 51% of participants in group 1 (least possibility of latent
TB infection) might be explained by the confounding effects of previous BCG
the predicted prevalence of TB infection (33%) in the Korean population.23 Non-TB mycobacteria infection could be another explanation
for the observed discrepancy in this study because non-TB mycobacteria infections
also give positive test results for the TST.24,25 Given
that non-TB mycobacteria infection usually causes the TST false-positive result
with indurations of 5 to 14 mm,24,25 the
analysis using a positive result with the 15-mm cutoff for the TST could validate
this possibility. The fact that the discrepancy between the TST and the IFN-γ
assay was not corrected with a 15-mm induration cutoff for TST weakens the
probability of cross-reactivity between non-TB mycobacteria and M tuberculosis infections.
However, the possibility of underestimation of latent TB infection by
IFN-γ assay could not be excluded. The TB infection rate in 20- to 29-year-old
Koreans was 59% in 1995 and the expected prevalence of latent TB infection
in all Koreans was about 33% in 2004,23,26 so
the TB infection rates of 4% and 10% in groups 1 and 2 measured by the IFN-γ
assay might be too low. A previous large evaluation comparing enzyme-linked
immunospot assays and skin tests for the diagnosis of M
tuberculosis infections in Gambia, a region with a high TB prevalence,
also raised concern about the low sensitivity of assays based on M tuberculosis–specific antigens.27
The possible underestimation of latent TB infection by IFN-γ assay
in our study could be understood in terms of the antigenicity of ESAT-6 and
CFP-10. Although these antigens are specific for M tuberculosis, they do not represent the whole spectrum of antigenicity of M tuberculosis.28,29 In
addition, the different response rates to the immunodominant peptide ESAT-6
between the United Kingdom and Indian patients with TB suggests that the response
rate may be affected by human leukocyte antigens.30 Another
possible explanation for the potential underestimation of latent TB infection
by the IFN-γ assay may be found in the high response rate (81%) in patients
with clinically active TB to this assay in contrast to the lower response
rates among participants with close TB contact without active TB. Memory T
cells are less likely to release IFN-γ following a short period of exposure
to antigens in an ex vivo assay.31,32 Moreover,
activated lymphocytes and effector memory cells producing IFN-γ persist
for a limited time in circulation once the antigen is cleared.33,34 Thus,
the IFN-γ assay might reflect recent rather than remote TB infections.
The measurement of correlation with the risk of infection could be a
more sensitive method to compare the utility of the TST and the IFN-γ
assay than a detailed determination of the specificity and sensitivity of
tests considering that there is no criterion standard for the diagnosis of
latent TB infection, especially in endemic regions, because of high levels
of environmental exposure to mycobacteria and M tuberculosis. In our study, the IFN-γ assay test results were more closely
associated with the risk of infection than the TST (P<.001),
and this was not significantly affected by BCG vaccination status, age, and
sex. Therefore, the IFN-γ assay could be more helpful than the TST for
the detection of latent TB infection in Koreans despite possible underestimation.
However, we should consider the relatively higher costs, practical inconvenience,
and the presence of indeterminate test results of the IFN-γ assay. The
tentative cost of the IFN-γ assay to process a sample from 1 patient
would be between $20 and $30 US while the cost of the TST is as low as $1
US in Korea. In addition, samples of at least 22 patients should be analyzed
per run not to waste wells. Because the IFN-γ assay is based on an enzyme-linked
immunosorbent assay method that involves the use of a standard curve, minimum
sample costs will only be achieved when all wells are used. Moreover, the
presence of indeterminate test results of the IFN-γ assay, although
rare, should be acknowledged.
In conclusion, the IFN-γ assay based on the ESAT-6 and CFP-10 M tuberculosis–specific antigens is a useful method
for detecting latent TB infections and might help to eliminate the limitations
of the TST in BCG-vaccinated populations in intermediate TB-burden countries.
However, the negative IFN-γ assay test results should be cautiously
appreciated because of its possible low sensitivity in the diagnosis of latent
Corresponding Author: Jae-Joon Yim, MD,
Department of Internal Medicine, Seoul National University College of Medicine,
28 Yongon-Dong, Chongno-Gu, Seoul, 110-744, Republic of Korea (firstname.lastname@example.org).
Author Contributions: Dr Yim had full access
to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Yim.
Acquisition of data: Kang, Lee, Yoon, Cho.
Analysis and interpretation of data: Kang,
Han, Shim, Yim.
Drafting of the manuscript: Kang, Lee.
Critical revision of the manuscript for important
intellectual content: Yoon, Cho, Han, Shim, Yim.
Statistical analysis: Kang.
Administrative, technical, or material support:
Lee, Yoon, Cho, Han, Yim.
Study supervision: Han, Shim, Yim.
Financial Disclosures: None reported.
Funding/Support: This work was supported by
grant 05-2004-005-0 from the Seoul National University College of Medicine
Research Fund (Seoul, Republic of Korea). The purified protein derivative
assay kits were donated by Korean Institute of Tuberculosis (Seoul, Republic
of Korea) and the QuantiFERON-TB Gold assay kits were donated by Woongbee
Meditech (Seoul, Republic of Korea).
Role of the Sponsor: The Seoul National University
College of Medicine Research Fund had no involvement in the design and conduct
of the study; in the collection, management, analysis, and interpretation
of the data; or in the preparation, review or approval of the manuscript.
Acknowledgment: We thank the numerous individuals
who volunteered to participate in this study. We also acknowledge the insightful
contributions of Seung-Sik Hwang, MD, Young Whan Kim, MD, Chul-Gyu Yoo, MD,
Sang Min Lee, MD, Sang Won Um, MD, Chang-Hoon Lee, MD, and Eun Kyoung Lee,