BRIEF REPORT • CID 2007:44 (1 January) • 69
B R I E F R E P O R T
Low Sensitivity of a Whole-Blood
Interferon-g Release Assay for
Detection of Active Tuberculosis
Puneet K. Dewan,1Jennifer Grinsdale,2and L. Masae Kawamura2
1International Research and Programs Branch, Division of Tuberculosis
Elimination, National Center for HIV, STD, and TB Prevention, Centers for
Disease Control and Prevention, Atlanta, Georgia; and
Department of Public Health, San Francisco, California
(See the editorial commentary by Pai and Menzies on pages 74–7)
The sensitivity of an interferon-g assay (Quantiferon-TB
Gold; Cellestis) was evaluated for the detection of tubercu-
losis among 242 persons with suspected tuberculosis in San
Francisco, California. Thirty-seven subjects had culture-
confirmed tuberculosis. Excluding 1 indeterminate result, 23
(64%; 95% confidence interval, 48%–78%) of 36 subjects had
positive results using the QuantiFERON-TB Gold assay. The
should not be used alone to exclude active tuberculosis.
In December 2004, the US Food and Drug Administration
approved the Quantiferon-TB Gold (QFT-G; Cellestis), a new
assay for the detection of Mycobacterium tuberculosis infection.
The QFT-G assay measures the amount of IFN-g released after
blood is incubated with synthetic antigens (early secretory an-
tigen 6 and culture filtrate protein 10) that simulate proteins
present in M. tuberculosis. Although the QFT-G assay is re-
ported to detect M. tuberculosis infection with high specificity,
it does not distinguish between active tuberculosis and latent
tuberculosis infection [1–7]. Furthermore, the sensitivity of
QFT-G for detection of active tuberculosis has not been well
defined among individuals with suspected tuberculosis in the
The San Francisco Department of Public Health (San Fran-
cisco, CA) has used the QFT-G assay as a component of routine
Received 19 July 2006; accepted 23 August 2006; electronically published 28 November
Reprints or correspondence: Dr. Puneet K. Dewan, International Research and Programs
Branch, Div. of Tuberculosis Elimination, National Center for HIV, STD, and TB Prevention,
Centers for Disease Control and Prevention, 1600 Clifton Rd., MS E10, Atlanta, GA 30333
Clinical Infectious Diseases 2007;44:69–73
This article is in the public domain, and no copyright is claimed.
evaluation for patients with suspected tuberculosis since Feb-
ruary 2005. We sought to evaluate the sensitivity and predictive
value of QFT-G for the detection of active tuberculosis among
persons with suspected tuberculosis at a public health clinic.
San Francisco Department of Public Health Tu-
berculosis Clinic guidelines require QFT-G testing of patients
who are suspected of having active tuberculosis (defined as
American Thoracic Society tuberculosis class 5), in conjunction
with routine clinical, microbiologic, and radiographic exami-
as a patient with clinical or radiographic evidence consistent
with active tuberculosis for whom laboratory confirmation of
the final disease classification has not been received. Because
specimens require same- and subsequent-day processing, the
QFT-G assay is not used if patients are evaluated at the San
Francisco Department of Public Health Tuberculosis Clinic af-
ter 3 p.m., on Fridays or weekends, or if phlebotomy is un-
successful or declined. The assay is performed in accordance
with the manufacturer’s instructions: specimens are processed
?8 h after phlebotomy, and incubation time was standardized
at 16–18 h . A tuberculin skin test (TST) is not required for
patients with suspected tuberculosis, but results are often
In this retrospective evaluation, we reviewed clinical records
for all consecutive patients from San Francisco who were re-
ported to have suspected tuberculosis during the period from
1 March through 31 December 2005. We excluded patientswho
transferred out of the jurisdiction or were treated for 114 days
with antituberculosis drugs before QFT-G testing. We collected
information about TST results obtained within 2 weeks of the
QFT-G test date that were reported to San Francisco Depart-
ment of Public Health by the referring provider or that were
obtained at the tuberculosis clinic using standard methods .
TST indurations ?5 mm in diameter were classified as positive
results. We defined “high clinical suspicion” as suspected tu-
berculosis in a patient who was receiving empirical antituber-
culosis medication before the availability of diagnostic micro-
biologic test results. Cases were classified by final diagnosis as
either a verified case of tuberculosis or as negative for tuber-
Ninety-five percent CIs for sensitivity were calculated using
the Wilson score method . Proportions were compared
using the x2test or, if there were ?5 observations, Fisher’s test.
In the evaluation period, the San Francisco De-
partment of Public Health noted 522 consecutive patients with
suspected tuberculosis; 41 (8%) were excluded as transfers out,
by guest on May 21, 2011
70 • CID 2007:44 (1 January) • BRIEF REPORT
San Francisco, California, March–December 2005.
Characteristics of patients evaluated for tuberculosis,
No. (%) of patients
(n p 242)
Not tested by
(n p 200)
In United States !5 years
Injection drug use
HIV infection status
History of prior TB treatment
Treatment for LTBI
Treatment for active TB
Level of clinical suspicion for TB
16 (8) .15a
tained after receipt of 114 days of antituberculosis treatment were classified
as not tested. P values were calculated using the Mantel-Haenszel x2test or,
if there were ?5 observations, with Fisher’s exact test. LTBI, latent tuber-
culosis infection; TB, tuberculosis.
aP value by x2test for difference in trend.
bPersons classified as class B during immigration health evaluation.
cCalculated in comparison with the HIV infection status “not positive.”
Patients with QuantiFERON-TB Gold (QFT-G; Cellestis) results ob-
and 39 (7%) had not yet received a final diagnosis as of 15
March 2006. Of the remaining 442 subjects with suspected
tuberculosis, 242 (55%) were tested using the QFT-G assay
(table 1). Persons who underwent QFT-G testing did not sig-
nificantly differ from those who did not with regard to distri-
bution of age, HIV infection status, high clinical suspicion of
tuberculosis, or final diagnosis of tuberculosis. Patients who
underwent QFT-G testing were more likely to be foreign born
or Asian and were less likely to declare non-Hispanic black or
Hispanic white ethnicity/race.
Among the 242 persons with suspected tuberculosis, 45
(19%) received a diagnosis of tuberculosis, including 37 (82%)
who had culture-confirmed disease (table 2). Among these 45
patients with tuberculosis, QFT-G results were positive for 25
(55%), negative for 17 (38%), and indeterminate for 3 (7%).
Excluding indeterminate results, we calculated an overall QFT-
G sensitivity for this population of 60% (95% CI, 44%–73%)
and a negative predictive value of 86% (95% CI, 79%–91%).
False-negative QFT-G results were distributed equallythrough-
out the 46-week evaluation period (data not shown). We ob-
served similar results among the 37 patients with culture-
confirmed tuberculosis (sensitivity, 64% [95% CI, 48%–78%];
negative predictive value, 89% [95% CI, 83%–94%] andamong
persons with no history of prior tuberculosis infection (in-
cluding those with latent tuberculosis infection and active dis-
ease) or tuberculosis treatment (table 2). Only 3 HIV-infected
patients received a diagnosis of tuberculosis, precluding mean-
ingful evaluation of this subgroup.
We compared the clinical characteristics of patients with ver-
ified tuberculosis and true-positive QFT-G results (25 patients)
with those who had false-negative QFT-G results (17 patients)
(table 3). We observed no significant differences with regard
to age distribution, proportion of patients without chronic
medical conditions, HIV infection status, or duration of treat-
ment of active tuberculosis at the time of testing. Patients with
false-negative QFT-G results were more likely to have extra-
pulmonary tuberculosis (35% vs. 4%;
observe cases of multiple-site or miliary tuberculosis in this
cohort of patients who underwent QFT-G testing.
TST results were available for 24 (53%) of the 45 patients
with verified tuberculosis who were tested using the QFT-G.
Of these 24 patients, the QFT-G and TST results were both
positive for 12 (50%) and both negative for 1 (4%). Nine
patients (38%) had negative QFT-G and positive TST results,
and 2 (8%) had indeterminate QFT-G and negativeTSTresults.
No patient with verified tuberculosis had negative TST and
positive QFT-G results. Eighteen (75%) of the 24 patients’ TST
results were verified by the San Francisco Department ofPublic
Health, and 6 (25%) results were reported in induration (in
mm) by community providers; between these groups, the pro-
). We did not
P ! .05
portion of discordant TST and QFT-G results did not signif-
icantly differ (data not shown).
The QFT-G IFN-g assay was demonstrated to
have 64% sensitivity and 89% negative predictive value for
culture-confirmed cases of active tuberculosis among patients
with suspected tuberculosis in San Francisco. Excluding in-
determinate results, we observed negative QFT-G results in 17
by guest on May 21, 2011
BRIEF REPORT • CID 2007:44 (1 January) • 71
QuantiFERON-TB Gold (QFT-G; Cellestis) results among 242 persons with suspected active tuberculosis, by patient population
Patient population, QFT-G result
No. (%) of subjects
(95% CI) With TB (%) Without TB (%)
Patients with high clinical suspicion of TBa
Patients with low clinical suspicion of TB
Patients with culture-confirmed TB
Patients with no known prior TB or treatmentb
NPV, negative predictive value; PPV, positive predictive value; TB, tuberculosis.
aPatients with abnormal chest radiograph findings and/or clinical symptoms of active tuberculosis whostartedreceivingantituberculosistreatmentimmediately,
regardless of the availability of microbiological examination results.
bPatients had no known prior positive tuberculin skin test result or IFN-g release assay result, no prior treatment history for active tuberculosis or latent
tuberculosis infection, and no current use of antituberculosis medications for 15 days. Prior tuberculosis infection was defined as either (1) documented history
of a tuberculin skin test induration 110 mm at least 1 year prior to notification as having suspected tuberculosis, or (2) prior self-reported history of active
tuberculosis disease at any time in the past.
Patient population subgroups are not mutually exclusive. Persons with indeterminate QFT-G results are excluded from calculations of test accuracy.
(40%) of 42 patients with tuberculosis, including 13 (36%) of
36 patients with culture-confirmed tuberculosis and 10 (40%)
of 25 patients with no prior history of tuberculosis infection,
disease, or treatment. Our results reinforce the recent recom-
mendation that negative results should not be used alone to
exclude active tuberculosis and that results should be inter-
preted in conjunction with other clinical and diagnostic find-
ings . The combination of low sensitivity for active tuber-
culosis and the intrinsic inability of the test to distinguish
between latent infection and active disease suggests that the
QFT-G assay has a limited role in the evaluation of patients
with suspected tuberculosis.
The sensitivity of the QFT-G assay for the detection of active
tuberculosis cases observed in San Francisco was slightly lower
by guest on May 21, 2011
72 • CID 2007:44 (1 January) • BRIEF REPORT
Table 3.Characteristics of patients who had verified cases of tuberculosis diagnosed.
Patients with positive
(n p 25)
Patients with negative
(n p 17)
Age, median years (range)
Prior history active TB treatment
Chronic medical conditions
Chronic obstructive lung disease
Lung cancer (untreated)
HIV infection status
Smear positive for acid-fast bacilli, n/N (%)
Culture positive for Mycobacterium tuberculosis, n/N (%)
Culture positive for M. tuberculosis, n/N (%)
Current treatment for TB
14 (56) 10 (59)
QFT-G, QuantiFERON-TB Gold (Cellestis); TB, tuberculosis.
aOne patient had both diabetes mellitus and alcoholism.
bPatient was not taking oral corticosteroids at time of testing.
dThere was 1 case of culture-confirmed tuberculous lymphadenitis.
eCulture-confirmed diagnosis in case of tuberculous lymphadenitis (1 patient), colitis (1 patient), and peritonitis (1 patient),
as well as pathological (culture-negative) diagnosis of tuberculous lymphadenitis (2 patients) and osteomyelitis (1 patient).
Data are no. (%) of patients, unless otherwise indicated. Patients with indeterminate QFT-G results are excluded.
but similar to that observed in referral hospitals . Lee et
al.  reported a sensitivity of 70% among 87 patients who
received a diagnosis of tuberculosis; the subjects in that study
differed from our population in that 55 patients (63%) in that
study had culture-confirmed tuberculosis, and 29 (33%) were
classified as immunocompromised. Among patients with cul-
ture-confirmed tuberculosis, Kang et al.  reported 81% sen-
sitivity among 54 patients, and Mori et al.  reported 89%
sensitivity among 118 patients. Lee et al.  suggested that
using lower IFN-g responsevaluesforclassificationofapositive
QFT-G result may increase the assay’s sensitivity with minimal
trade-off in specificity and that this should be considered when
testing patients with a high pretest probability of active tuber-
culosis, such as close contacts or patients with suspected tu-
berculosis. This approach, however, has not been validated.
Among patients with extrapulmonary tuberculosis, we ob-
served that 6 (86%) of 7 had negative QFT-G results, including
3 (75%) of 4 with culture-confirmed cases. Poor IFN-g re-
sponse to culture filtrate protein 10 has been previously re-
ported among patients with extrapulmonary tuberculosis .
Our retrospective evaluation had several limitations. Our re-
sults may have been affected by the fact that 8 (47%) of 17
patients with tuberculosis diagnoses who had false-negative
QFT-G results had received 1–14 days of antituberculosistreat-
ment. Our evaluation lacked the power to assess the possible
effects of treatment [15, 16]. However, similar QFT-G sensi-
tivity was observed among the subset of patients with no prior
history of antituberculosis treatment or M. tuberculosis infec-
tion; thus, this limitation has likely not affected our conclu-
sions. Because the TST was only performed for a subset of
by guest on May 21, 2011
BRIEF REPORT • CID 2007:44 (1 January) • 73
patients and by multiple health care providers, our evaluation
should not be interpreted as a comparative study with the TST.
The limitations of the TST in the diagnosis of active tuber-
culosis are well established, and clinical recommendationssug-
gest that the TST should not be used alone to exclude active
Our results are specific for the QFT-G assay and should not
be assumed to apply to other IFN-g assays. Newer commercial
assays, such as the T-SPOT.TB (Oxford-Immunotec) and the
QuantiFERON-TB Gold In Tube (Cellestis) assays, have many
differences and may perform with higher sensitivity [5, 18].
Nevertheless, the QFT-G assay is currently the only IFN-gassay
commercially available in the United States, and providers
should be aware of the limitations we have observed. Similarly,
these findings should not be extrapolated to assume that the
QFT-G or other IFN-g assays lack adequate sensitivity for the
detection of latent tuberculosis infection. Prospective studies
are critically needed to characterize the future risk of tuber-
culosis among persons with different responses to IFN-g in
these commercial assays.
In conclusion, we observed a low sensitivity for the QFT-G
assay for the detection of active tuberculosis among persons
with suspected tuberculosis at a public health clinic in San
Francisco. As with the TST, providers should be aware of the
limitations of this assay and should not use negative QFT-G
based clinical trials should be conducted fornewerIFN-gassays
before any adjustments are made in currentdiagnosticpractices
We acknowledge Sally Liska and Ernest H. Wong (Public Health Lab-
oratory of the San Francisco Department of Public Health), for their in-
valuable contributions to IFN-g assay implementation in San Francisco;
Robert Fallstead, for data entry assistance; and the frontline staff of the
Tuberculosis Control Section, San Francisco Department of Public Health.
We also thank Dr. Andrew Vernon (Centers for Disease Control and Pre-
vention [CDC]; Atlanta, GA), for his review of this manuscript, and Ann
Lanner (CDC), for editorial assistance.
Potential conflicts of interest.
All authors: no conflicts.
1. Pai M, Riley LW, Colford JM Jr. Interferon-gamma assays in the im-
munodiagnosis of tuberculosis: a systematic review. Lancet Infect Dis
2. Ferrara G, Losi M, Meacci M, et al. Routine hospital use of a new
commercial whole blood interferon-g assay for the diagnosis of tu-
berculosis infection. Am J Respir Crit Care Med 2005;172:631–5.
3. Kang YA, Lee HW, Yoon HI, et al. Discrepancy between the tuberculin
skin test and the whole-blood interferon gamma assay for thediagnosis
of latent tuberculosis infection in an intermediate tuberculosis-burden
country. JAMA 2005;293:2756–61.
4. Mori T, Sakatani M, YamagishiF,et al.Specificdetectionoftuberculosis
infection: an interferon-gamma-based assay using new antigens. Am J
Respir Crit Care Med 2004;170:59–64.
5. Lee JY, Choi HJ, Park IN, et al. Comparison of two commercial in-
terferon gamma assays for diagnosing Mycobacterium tuberculosis in-
fection. Eur Respir J 2006;28:24–30.
6. Tsiouris SJ, Coetzee D, Toro PL, Austin J, SteinZ, El-SadrW.Sensitivity
analysis and potential uses of a novel gamma interferon release assay
for diagnosis of tuberculosis. J Clin Microbiol 2006;44:2844–50.
7. Ravn P, Munk ME, Andersen AB, et al. Prospective evaluation of a
whole-blood test using Mycobacterium tuberculosis-specific antigens
ESAT-6 and CFP-10 for diagnosis of active tuberculosis. Clin Diagn
Lab Immunol 2005;12:491–6.
8. Diagnostic standards and classification of tuberculosis in adults and
children. Am J Respir Crit Care Med 2000;161:1376–95.
9. Cellestis International. Quantiferon-TB Gold [package insert]. 2005.
Available at: http://www.cellestis.com. Accessed 10 February 2006.
10. Centers for Disease Control and Prevention. Case definitions for in-
fectious conditions under public health surveillance. MMWRRecomm
Rep 1997;46(RR-10):1–55. Available at: http://www.cdc.gov/mmwr/
preview/mmwrhtml/00047449.htm. Accessed 20 August 2006.
11. Newcombe RG. Two-sided confidence intervals for the single propor-
tion: comparison of seven methods. Stat Med 1998;17:857–72.
12. Mazurek GH, Jereb J, Lobue P, Iademarco MF, Metchock B, Vernon
A. Guidelines for using the QuantiFERON-TB Gold test for detecting
Mycobacterium tuberculosis infection, United States. MMWR Recomm
13. Pai M, Kalantri S, Dheda K. New tools and emerging technologies for
the diagnosis of tuberculosis: Part I. Latent tuberculosis. Expert Rev
Mol Diagn 2006;6:413–22.
14. Hussain R, Kaleem A, Shahid F, et al. Cytokine profiles using whole-
15. Al-Attiyah R, Mustafa AS, Abal AT, Madi NM, Andersen P. Restoration
of mycobacterial antigen-induced proliferation and interferon-gamma
responses in peripheral blood mononuclear cells of tuberculosis pa-
tients upon effective chemotherapy. FEMS Immunol Med Microbiol
16. Carrara S, Vincenti D, Petrosillo N, Amicosante M, Girardi E, Goletti
D. Use of a T cell–based assay for monitoring efficacy of antituber-
culosis therapy. Clin Infect Dis 2004;38:754–6.
17. Targeted tuberculin testing and treatment of latent tuberculosis infec-
tion. Am J Respir Crit Care Med 2000;161:S221–47.
18. Mahomed H, Hughes EJ, Hawkridge T, et al. Comparison of mantoux
skin test with three generations of a whole blood IFN-gamma assay
for tuberculosis infection. Int J Tuberc Lung Dis 2006;10:310–6.
19. Pai M, O’Brien R. Tuberculosis diagnostics trials: do they lack meth-
odological rigor? Expert Rev Mol Diagn 2006;6:509–14.
by guest on May 21, 2011