Secondary Logo

Journal Logo

Original Studies

The Likelihood of an Indeterminate Test Result from a Whole-Blood Interferon-γ Release Assay for the Diagnosis of Mycobacterium tuberculosis Infection in Children Correlates With Age and Immune Status

Haustein, Thomas MD*†‡; Ridout, Deborah A. MSc§; Hartley, John C. FRCPath*†‡; Thaker, Urvashi MSc*†‡; Shingadia, Delane FRCPCH, MPH; Klein, Nigel J. PhD; Novelli, Vas FRCP; Dixon, Garth L. J. PhD*†‡

Author Information
The Pediatric Infectious Disease Journal: August 2009 - Volume 28 - Issue 8 - p 669-673
doi: 10.1097/INF.0b013e3181a16394
  • Free


Assessing patients for infection with Mycobacterium tuberculosis is often complex. In children, specific challenges are posed by the diverse and nonspecific nature of signs and symptoms, the difficulty in obtaining bacteriologic specimens, the poor yield of mycobacterial culture in active disease, and a high risk of rapid progression to disease from latent infection.1 The tuberculin skin test (TST) has been widely used for greater than a century to detect latent and active tuberculosis (TB), but has important limitations.2,3 It is therefore not surprising that the interferon-γ release assays (IGRAs) introduced in recent years have been met with considerable interest.2 IGRAs are performed ex vivo and detect interferon-γ (IFNγ) production by peripheral blood mononuclear cells in response to stimulation with specific antigens including ESAT-6, CFP-10, and TB7.7. These antigens are produced by M. tuberculosis, but are absent in BCG and most nontuberculous mycobacteria, which should make IGRAs more specific than TST.2 IGRAs also incorporate a positive control where peripheral blood mononuclear cells are nonspecifically stimulated with the mitogen phytohemagglutinin (PHA). A test result is reported as indeterminate rather than negative if there is no significant response to both M. tuberculosis antigens and PHA.2 Potentially, this could reduce the number of false negatives and improve the negative predictive value of the test. However, for individual patients, indeterminate results are of limited use, because they do not provide a diagnosis nor guide clinical management.

Two test formats are commercially available. Whole-blood IGRAs (QuantiFERON-TB, Cellestis Ltd, Victoria, Australia) measure concentrations of IFNγ in plasma by ELISA, whereas ELISpot assays (T-SPOT.TB, Oxford Immunotec Ltd, Oxford, UK) detect individual IFNγ-secreting cells. Of the 2 formats, ELISpot assays may be more sensitive, while whole-blood IGRAs are technically more convenient and less costly.2

UK guidelines issued by the National Institute for Health and Clinical Excellence recommend that IGRA testing be considered for the diagnosis of latent TB in adults and children with a positive TST, and in those “in whom Mantoux testing may be less reliable.”4 In the United States, CDC guidelines endorse the use of QuantiFERON-TB Gold (QFT-G) in the diagnosis of latent and active infection with M. tuberculosis instead of TST, but highlight the lack of data regarding its performance in children.5

Young age or impaired immunity have previously been associated with higher rates of indeterminate IGRA results, in particular when using the whole-blood format,6–9 but the likelihood of indeterminate results in different pediatric age groups and in immunocompromised children is currently unknown.

Here, we report our experience with the In-Tube format of QuantiFERON-TB Gold In-Tube assay (QFT-IT) in pediatric patients of all ages and with a high prevalence of immunocompromising conditions.


Test Procedure

QFT-IT was used as a routine diagnostic test at Great Ormond Street Hospital if either active or latent TB was suspected, and was performed and interpreted according to manufacturer's instructions.10 Briefly, instructions were given to fill the blood collection tubes in the correct order, and to ensure thorough mixing of the blood with the contents of the tubes. After receipt in the laboratory, samples were mixed again and incubated as soon as possible. Samples with a documented time to incubation of >16 hours were excluded from analysis. After incubation, the plasma was harvested and stored at 4°C. IFNγ-levels were measured in batches by ELISA within a few days of sample collection. For a test result to be determinate, the IFNγ concentration in the Mitogen tube had to be at least 0.5 IU/mL higher than in the Nil tube, or the concentration in the TB antigen tube had to be at least 0.35 IU/mL higher than in the Nil tube with the difference in concentration between Nil and TB antigen tube being at least 25% of the Nil value, and the background IFNγ-level in the Nil tube had to be ≤8.0 IU/mL.

Clinical information, including immunization history and TST results, was obtained contemporaneously from the clinicians requesting the test, or retrospectively from patient records. TSTs were usually applied and read by a member of the Infectious Diseases team. Mantoux test was done using 2 TU/0.1 mL tuberculin PPD RT 23 SSI (Statens Serum Institut, Copenhagen, Denmark). According to UK national guidelines, an induration ≥6 mm (≥15 mm if BCG-vaccinated) after 48 to 72 hours is considered positive.11


Children were deemed immunocompromised if one or more of the following conditions were documented at the time of testing: (1) acute hematologic malignancies not yet in remission, and other malignancies up to 6 months after completion of chemotherapy; (2) known primary immunodeficiencies, including severe combined immunodeficiency, hypogammaglobulinemia, hyper-IgM syndrome, ataxia telangiectasia, and congenital aplastic anemia; (3) hematopoietic stem cell or solid organ transplant on immunosuppressive therapy; (4) inflammatory disorders on systemic immunosuppressive treatment, including steroids, azathioprine, methotrexate, and cyclophosphamide; (5) asplenia; and (6) known HIV infection with severe suppression according to the 1994 CDC definitions.12 All other children were classified as not immunocompromised. A case of culture-proven TB was defined as a child with signs or symptoms compatible with TB and isolation of M. tuberculosis from a clinical sample. TB was considered probable if clinical and radiologic features were consistent with TB and were deemed unlikely to be due to another disease process, and if the child responded appropriately to antituberculous therapy. BCG status was determined based on the documentation of a corresponding history or scar in the patient notes.

Statistical Analysis

The effects of age, sex, and immune status on the probability of obtaining an indeterminate QFT-IT result, along with the 2-way interactions between these factors, were analyzed by logistic regression analysis. Fractional polynomials were used to investigate the relationship between age and outcome. Agreement between QFT-IT and TST results was assessed using a κ statistic. Calculations were performed on Stata Release 9 (StataCorp LP, College Station, TX).

Ethical Approval

The project was registered as a clinical evaluation of a commercially available test with the Research and Development Office at the Institute of Child Health and Great Ormond Street Hospital, and was considered exempt from ethics review.


Between January 2006 and February 2008, we received 269 samples from children aged <18 years for QFT-IT testing. One hundred seventy-nine samples (76%) were referred from within our hospital, and 58 specimens (24%) were requested from other local hospitals in consultation with our infectious disease team. Thirty-two samples (12%) were excluded from analysis; 16 were invalid because of technical errors during sampling or processing (incomplete set of tubes, >16 hours delay in transit, incubation at wrong temperature), and 16 were repeats on patients with a previous technically valid test. For 24 specimens, incubation within 16 hours could not be ascertained. However, inclusion or exclusion of these specimens did not affect our results, and we retained them in our analysis. Thus, in total 237 tests were analyzed from 237 patients.

Characteristics of the study population are summarized in Table 1. QFT-IT was positive in 41 children (17%), negative in 113 (48%), and indeterminate in 83 (35%). The test was indeterminate in 82 children because of an insufficient response to PHA, and in 1 child because of a high background IFNγ-level. Among 81 children whose samples were incubated within 2 hours of venipuncture (median age: 7.4 years, interquartile range: 2.7–12.8 years, 24 [29.6%] immunocompromised), 33 (40.7%) had an indeterminate result. Of note, 9 of the 41 children with a positive test (22%) had a failed mitogen control. Individual IFNγ-responses to PHA are displayed in Figure 1.

Characteristics of the Study Population
IFNγ-levels in response to stimulation with PHA (mitogen control) after subtraction of the background value (Nil tube) by age. Concentrations >15 IU/mL are censored (black triangles), as QFT-IT is not designed to reliably quantify higher levels. Negative values are displayed as zero. The threshold for a positive mitogen control is represented by the horizontal lines at 0.5 IU/mL. A, children without known immunocompromise, (B) children with immunocompromise.

On univariate analysis, immunocompromise and age ≤5 years were significantly associated with an indeterminate QFT-IT result. Multivariable analysis showed that younger age and immunocompromise were independently associated with a higher probability of an indeterminate QFT-IT (Table 2). There was no evidence of a significant interaction between any of the factors.

Interactions Between Age, Immune Status, Sex, and Likelihood of an Indeterminate QFT-IT Result

To explore more precisely the relationship of age and risk of indeterminate result without categorization into arbitrary age brackets (Table 3), we performed logistic regression analysis using fractional polynomials. A linear trend was found to be the best fit to the data with OR for every year of age 0.94, 95% CI: 0.89 to 0.99, P = 0.03. Analyzing the 178 children without known immunodeficiencies only, OR per year of age was 0.87, 95% CI: 0.81 to 0.94, P < 0.001, corresponding to a 13% reduction of the risk of an indeterminate test per year (Fig. 2). There was no evidence of such a relationship in the immunocompromised group (P = 0.87).

Frequencies of Indeterminate QFT-IT Results by Age Group and Immune Status
Estimated probability of indeterminate results with 95% confidence intervals as a function of age based on a logistic model. A, children without known immunocompromise, (B) children with immunocompromise.

TST results were documented in 161 children, of whom 30 (19%) were considered immunocompromised. TST and QFT-IT were both positive in 30 children. The combination of a negative TST with a negative or indeterminate QFT-IT was seen in 114 patients. Nine individuals had a positive TST with a negative or indeterminate QFT-IT, while 8 had a negative TST with a positive QFT-IT (89% concordant results, κ = 0.71, 95% CI: 0.55–0.87, P < 0.001).

Culture-proven TB occurred in 16 patients (median age: 9 years, interquartile range: 7.6–12.8 years), none of whom were considered immunocompromised. QFT-IT was positive in 12 children, negative in 1, and indeterminate in 3, yielding a sensitivity of 75.0% (95% CI: 47.4–91.7). TST was positive in 10 children (9/12 children with positive QFT-IT, and the child with the negative QFT-IT). Among the 3 patients with indeterminate QFT-IT, 2 had TST performed, which was negative. Thus, the sensitivity of TST in proven TB was 67.7% (95% CI: 38.7–87.0). When including a further 11 children with probable TB, sensitivities of QFT-IT and TST were 77.8% (95% CI: 57.3–90.1) and 72.0% (95% CI: 50.4–87.1), respectively.


We have shown that the risk of an indeterminate QFT-IT result in apparently immunocompetent children at our tertiary care institution is linearly correlated with age, and 25% on average. Conditions associated with impaired immunity increase this risk to 66%, independently of age.

Immunosuppression,7 cancer chemotherapy,6 or HIV-infection with CD4 lymphocyte count <100/μL,2,8 have previously been found to be significantly associated with indeterminate QFT results, but specific data for immunocompromised children have been lacking. High rates of indeterminate results in children have previously been observed (7/22 children ≤5 years,6 17/101 children aged 0.4–17.9 years13), but the correlation between age and probability of an indeterminate test result has not been well characterized. We found a striking inverse relationship between age and likelihood of an indeterminate test in apparently nonimmunocompromised children. The reasons for this finding are not immediately apparent. The magnitude of the IFNγ-response to PHA can vary widely between subjects. Statistically, however, responses are lowest in newborns and increase during early childhood, reaching adult levels as early as around the age of 3 years.14,15 Although important differences in methodology make direct comparison with previous studies difficult, results presented here are consistent with this observation. Nevertheless, these data do not fully explain why the proportion of children exceeding the relatively low threshold of the QFT-IT mitogen control should continue to increase beyond early childhood.

In contrast with our results, studies from Cambodia,16 rural India,17 and Australia9 have found very low proportions of indeterminate results in children screened for infection with M. tuberculosis (9/204 children ≤5 years using QFT-G; 0/105 children aged 1–12 years, and 3/96 children aged 0.5–19 years using QFT-IT, respectively). These differences merit further investigation. Environment, underlying or intercurrent diseases, and genetic features, could all play a role. For example, differing prevalence patterns of atopy or intercurrent viral infections, both of which have been shown to affect T-cell cytokine responses,18,19 could impact on the performance of QFT-IT.

All indeterminate QFT-IT results in this study except one were due to an inadequate response to PHA, rather than high background IFNγ-levels in the nil control. Although QFT-IT and QFT-G are not comparable in all respects,20 this finding is in agreement with published experience in adult hospitalized patients using QFT-G.7 A previous study in children found that 12 of 17 indeterminate QFT-G results were caused by a high negative control response, based on a lower cut-off of 1.0 IU/mL. Even when applying the cut-off value of >8.0 IU/mL recommended in the package insert, the test failed in a third of children.13,21 The difference between these observations and our results could again reflect the influence of environmental or individual-related factors on test performance.

Furthermore, in our cohort, nearly a quarter of children with a significant response to TB-antigen (including some with culture-proven TB) failed to produce IFNγ responses to mitogen above the prespecified level, suggesting that the control indices may not be optimized for use in our patient group. The proportion of indeterminate results with QFT-IT might be reduced by lowering the threshold for the mitogen control or by adjusting the concentration of PHA. However, even changing the positivity threshold of the mitogen control arbitrarily from 0.5 IU/mL to 0.3 IU/mL would only have reduced the number of indeterminate results by 12 (14%). More importantly, this reduction might come at the expense of an increase in the rate of false-negative results.

The sensitivity of IGRAs could be affected by the delay between sampling and incubation. In healthy adult volunteers, a decline in the magnitude of the IFNγ-response to PHA was seen with a resting time as short as 2 hours, but was still detectable by ELISA if samples were incubated after 24 hours.22 In a recent study involving immunosuppressed adults that found very low rates of indeterminate QFT-IT results,23 all samples were incubated within 2 hours, suggesting that time to incubation might be more critical in certain patient groups. However, in our setting the rate of indeterminate results was not lower if only specimens processed within 2 hours were considered, and it seems unlikely that faster processing would have changed our results significantly.

We are confident that the rate of indeterminate results observed in our setting is not due to errors in the testing process, as manufacturer's instructions were strictly followed. More importantly, there were significant and biologically plausible differences between subgroups and good concordance with TST.24 The sensitivity of QFT-IT in children with active TB was also in line with previously reported data.2,25

The limitations of our study are that analysis and part of the data collection were retrospective. Furthermore, our definition of immunocompromise was broad, but the predominant mechanism was iatrogenic immunosuppression. Further studies are necessary to determine the performance of these tests in particular subgroups of patients, including doses, combinations, and duration of different immune suppressive therapies. Screening of young patients with inflammatory disorders on immune suppressive therapy before prescribing TNF inhibitors for example appears to be particularly problematic. It is likely that alternative immunologic tests may be appropriate in these patients.

Given our data, how should TST and IGRAs be used in young or immunocompromised children? Unfortunately, there is no clear answer to this question at present. In view of the high risk of morbidity and mortality from TB in this patient group and the difficulty of obtaining a microbiologic or histologic tissue diagnosis, many clinicians will feel that maximizing sensitivity is paramount. Thus, in our practice both diagnostic routes are often pursued in parallel, an approach that has also been advocated for adult immunocompromised patients.26 The interpretation of discordant results remains difficult and, for now, relies on sound clinical judgment.

We certainly do not dispute that IGRAs can yield valuable diagnostic information in children, but our results highlight some problematic issues. The combination of young age and primary and acquired immune deficiency could explain why QFT-IT may not provide a determinate test result in a substantial proportion of children in tertiary care settings. Further modification of this test may be required to account for age specific responses to both nonspecific and antigen specific T-cell stimuli.


The authors thank all laboratory staff at the Department of Microbiology, Virology, and Infection Control, Great Ormond Street Hospital, for processing the samples, Paul Lock for database creation, and the clinical teams who provided us with samples and clinical information, in particular Jemma Lawley, TB nurse specialist. The authors also thank to Haitham El Bashir and his colleagues from the TB clinic at North Middlesex Hospital NHS Trust.


1. Marais BJ, Gie RP, Schaaf HS, et al. Childhood pulmonary tuberculosis: old wisdom and new challenges. Am J Respir Crit Care Med. 2006;173:1078–1090.
2. Menzies D, Pai M, Comstock G. Meta-analysis–new tests for the diagnosis of latent tuberculosis infection: areas of uncertainty and recommendations for research. Ann Intern Med. 2007;146:340–354.
3. Shingadia D, Novelli V. Diagnosis and treatment of tuberculosis in children. Lancet Infect Dis. 2003;3:624–632.
4. National Collaborating Centre for Chronic Conditions. Tuberculosis: Clinical Diagnosis and Management of Tuberculosis, and Measures for its Prevention and Control. London, UK: Royal College of Physicians; 2006.
5. Mazurek GH, Jereb J, Lobue P, et al. Guidelines for using the QuantiFERON-TB Gold test for detecting Mycobacterium tuberculosis infection, United States. MMWR Recomm Rep. 2005;54:49–55.
6. Ferrara G, Losi M, D'Amico R, et al. Use in routine clinical practice of two commercial blood tests for diagnosis of infection with Mycobacterium tuberculosis: a prospective study. Lancet. 2006;367:1328–1334.
7. Ferrara G, Losi M, Meacci M, et al. Routine hospital use of a new commercial whole blood interferon-gamma assay for the diagnosis of tuberculosis infection. Am J Respir Crit Care Med. 2005;172:631–635.
8. Luetkemeyer AF, Charlebois ED, Flores LL, et al. Comparison of an interferon-gamma release assay with tuberculin skin testing in HIV-infected individuals. Am J Respir Crit Care Med. 2007;175:737–742.
9. Connell TG, Ritz N, Paxton GA, et al. A three-way comparison of tuberculin skin testing, QuantiFERON-TB gold and T-SPOT. TB in children. PLoS ONE. 2008;3:e2624.
10. Cellestis Ltd. QuantiFERON-TB Gold (In-Tube Method) Package Insert. 2007. Available at: Accessed February 11, 2009.
11. Salisbury D, Ramsay M, Noakes K. Immunization Against Infectious Disease. 3rd ed. Norwich, Norfolk, UK: The Stationery Office; 2006.
12. Caldwell MB, Oxtoby MJ, Simonds RJ, et al. 1994 Revised classification system for human immunodeficiency virus infection in children less than 13 years of age. MMWR Recomm Rep. 1994;43:1–10.
13. Connell TG, Curtis N, Ranganathan SC, et al. Performance of a whole blood interferon gamma assay for detecting latent infection with Mycobacterium tuberculosis in children. Thorax. 2006;61:616–620.
14. Frenkel L, Bryson YJ. Ontogeny of phytohemagglutinin-induced gamma interferon by leukocytes of healthy infants and children: evidence for decreased production in infants younger than 2 months of age. J Pediatr. 1987;111:97–100.
15. Miyawaki T, Seki H, Taga K, et al. Dissociated production of interleukin-2 and immune (gamma) interferon by phytohaemagglutinin stimulated lymphocytes in healthy infants. Clin Exp Immunol. 1985;59:505–511.
16. Okada K, Mao TE, Mori T, et al. Performance of an interferon-gamma release assay for diagnosing latent tuberculosis infection in children. Epidemiol Infect. 2008;136:1179–1187.
17. Dogra S, Narang P, Mendiratta DK, et al. Comparison of a whole blood interferon-γ assay with tuberculin skin testing for the detection of tuberculosis infection in hospitalized children in rural India. J Infect. 2007;54:267–276.
18. Tang ML, Coleman J, Kemp AS. Interleukin-4 and interferon-gamma production in atopic and non-atopic children with asthma. Clin Exp Allergy. 1995;25:515–521.
19. Copenhaver CC, Gern JE, Li Z, et al. Cytokine response patterns, exposure to viruses, and respiratory infections in the first year of life. Am J Respir Crit Care Med. 2004;170:175–180.
20. Mahomed H, Hughes EJ, Hawkridge T, et al. Comparison of mantoux skin test with three generations of a whole blood IFN-γ assay for tuberculosis infection. Int J Tuberc Lung Dis. 2006;10:310–316.
21. Radford AJ, Rothel JS, Sberna G. Whole blood IFN-gamma assay for detecting TB in children. Thorax. 2006;61:919–920, author reply 920–921.
22. Doherty TM, Demissie A, Menzies D, et al. Effect of sample handling on analysis of cytokine responses to Mycobacterium tuberculosis in clinical samples using ELISA, ELISPOT and quantitative PCR. J Immunol Methods. 2005;298:129–141.
23. Matulis G, Juni P, Villiger PM, et al. Detection of latent tuberculosis in immunosuppressed patients with autoimmune diseases: performance of a Mycobacterium tuberculosis antigen-specific interferon gamma assay. Ann Rheum Dis. 2008;67:84–90.
24. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174.
25. Detjen AK, Keil T, Roll S, et al. Interferon-gamma release assays improve the diagnosis of tuberculosis and nontuberculous mycobacterial disease in children in a country with a low incidence of tuberculosis. Clin Infect Dis. 2007;45:322–328.
26. Lalvani A, Millington KA. Screening for tuberculosis infection prior to initiation of anti-TNF therapy. Autoimmun Rev. 2008;8:147–152.

tuberculosis; IGRA; children; immunocompromised; quantiferon

© 2009 Lippincott Williams & Wilkins, Inc.