To the Editors:
In patients latently infected with Mycobacterium tuberculosis, immunosuppression significantly augments the risk of progression to active tuberculosis (TB) and TB is still one of the most frequent opportunistic infections worldwide. Prevention of TB in HIV-positive patients using the tuberculin skin testing (TST) followed by targeted preventive treatment is an effective strategy, but has thus far shown limited clinical success, in part due to the lack of a reliable test for latent TB infection (LTBI) 1. Interferon (IFN)-γ release assays (IGRAs) have shown great potential in the improvement of LTBI diagnosis, but the tests perform suboptimally in immunocompromised populations. We and others have previously shown that HIV-positive patients have high rates of indeterminate QuantiFERON®-TB Gold In-Tube test (QFT-IT) results and that this, at least in part, is due to low CD4 T-cell count 2–4.
We and others have also shown that IFN-γ-inducible protein 10 (IP-10/CXCL-10) is an alternative biomarker may improve immunodiagnosis of M. tuberculosis infection. IP-10 is induced specifically and in large quantities upon in vitro stimulation of whole blood with M. tuberculosis-specific antigens. We have developed an IP-10 based test 5 and shown that the IP-10 test performs comparably to the QFT-IT in adult patients with intact immune function 6.
We compared the sensitivity of an IP-10 based test with the QFT-IT for diagnosing M. tuberculosis infection in HIV-negative and HIV-positive patients with culture-confirmed active pulmonary tuberculosis (PTB). 300 patients newly diagnosed with PTB were recruited prospectively through the National Tuberculosis and Leprosy Programme (Mwanza, Tanzania). Blood was drawn for IGRA and HIV testing, and CD4 cell counts. Sputum was collected for microscopy and M. tuberculosis culture. Patients >15 yrs of age with a positive sputum culture result were enrolled. All HIV-positive patients were newly diagnosed in the study context. A detailed description of recruitment, testing and sample handling has been published previously 3. After QFT-IT testing, supernatants were frozen at -80°C. 1 yr later, samples were thawed and analysed for IP-10 using xMAP technology on the Luminex platform (Luminex Corp., Austin, TX, USA) as described previously 5. Algorithms for interpretation of IP-10 test results have been developed previously using receiver operating characteristic analysis 5; however, the mitogen cut-off was chosen arbitrarily (see online supplement 1). One IU corresponds to 50 pg IFN-γ. In the literature, there is a lack of consensus on how to present IGRA sensitivity and specificity: including or excluding indeterminate results. We suggest the designations “sensitivity” for calculation after exclusion of indeterminate results, and “positivity rate” for calculation amongst all possible test outcomes. High rates of indeterminate results (i.e. low positivity rate) mainly affect usability and cost-effectiveness, and may impair the feasibility of implementing the test in severely immunocompromised individuals. Positivity rate was calculated among all possible tests outcomes (i.e. including indeterminate results), whereas sensitivity was calculated after exclusion of indeterminate test results. All comparisons were done using nonparametric paired tests. A p-value ≤0.05 was considered significant. Permission to conduct the study was granted by the ethics committee of the National Institute for Medical Research (Mwanza, Tanzania) and was approved by The Danish National Committee on Biomedical Research Ethics (Copenhagen, Denmark; reference 2005-7041-57).
HIV-negative (n = 92) and HIV-positive (n = 65) patients were included in the study. HIV-positive patients were older, had lower CD4 cell counts and were more likely to be female (p<0.05 for all, see online supplement 2). The IP-10 test had comparable positivity rate and sensitivity to the QFT-IT in both HIV-positive and HIV-negative patients. Positivity rate was lower and indeterminate rate higher in HIV-positive patients for both tests, while there was no significant difference in sensitivity for either test (table 1).
Impaired performance of the QFT-IT in HIV-positive patients with a low CD4 cell count has been well documented and we have previously demonstrated that QFT-IT sensitivity decreases with decreasing CD4 cell count in HIV-positive patients, which is mediated by an increasing proportion of indeterminate results. Interestingly, for the IP-10 test, we found no such trend with CD4 cell count for rate of positive (p = 0.44) or for indeterminate (p = 0.55) results (fig. 1). Another recent study in HIV-infected patients with active TB reported that, while QFT-IT yielded indeterminate results when CD4 cell count was <200 cells·μL−1, an IP-10-based test yielded indeterminate results only when CD4 cell count was <50 cells·μL−1 4. Thus, it seems that IP-10 is not, or at least is much less, affected by CD4 cell count than is the QFT-IT. Although IFN-γ is directly involved in inducing IP-10 production, IP-10 is produced primarily by the monocyte and might be induced by CD4 T-cell- and IFN-γ-independent pathways (e.g. induction by other cytokines) 7. We found no effect of monocyte count on IP-10 test performance in either HIV-positive or -negative patients (data not shown). Another possible explanation for the finding is the larger amounts of biomarker produced, which may render an IP-10-based test less sensitive to the effect of immune suppression. Antigen-induced IP-10 levels were 21-fold higher than IFN-γ (interquartile range 7–48; p<0.001). Lastly, the mitogen cut-off was rather low, giving less indeterminate IP-10 test results and, thereby, possibly obscuring CD4 dependency. However, increasing the mitogen cut-off did not change these findings (data not shown). We found no further connection between clinical characteristics (listed in online supplement 2) and indeterminate or false negative results by either test.
Agreement between the tests was 0.73 (κ = 0.67) for HIV-negative and 0.78 (κ = 0.60) for HIV-positive patients (online supplement 3). Combining the tests significantly increased sensitivity to 96% in HIV-negative and 88% in HIV-positive patients (table 1). We and others have found a similar improvement in sensitivity among patients from low-endemic setting without a compromise in specificity 8, 9. Thus, combining tests appears to be a feasible way to improve performance of the antigen-specific assays. We found no connection between clinical characteristics (listed in online supplement 2) and discordant test results.
It is important to remember that the diagnostic sensitivity obtained in a population with active and severe TB may be lower than the sensitivity obtained among otherwise healthy individuals with LTBI. Also, a limitation of the present study is the lack of a control group, which prevents estimation of specificity. None of the tests discriminates between active TB and LTBI and, given the prevalence of LTBI in the background population, we could not identify a valid control group in this setting. Inclusion of a control group from another geographic region would not have been a valid approach due to potential influences of genetic background, comorbidity, nutritional status, etc. However, a very high specificity of both QFT-IT and the IP-10 test has been convincingly demonstrated in healthy, unexposed individuals from low-endemic regions. Unfortunately, TST was not performed in this study population.
In conclusion, the IP-10 test performs with equal sensitivity to the QFT-IT in a TB/HIV-endemic setting. Combining the tests significantly improves sensitivity, even in HIV-positive patients. The IP-10 test seems less affected by a low CD4 cell count than the QFT-IT. Further approaches for improvement of TB diagnosis are needed, especially in TB/HIV-endemic settings, and the IP-10 biomarker represents a promising example.
Acknowledgments
The authors would like to thank all colleagues contributing to this project, including the staff at the National Institute for Medical Research (NIMR) Mwanza Centre, Sekou Toure Regional Hospital, Buzuruga Health Centre, Bugando Medical Centre and Butimba Health Centre (all Mwanza, Tanzania), especially laboratory technician O. Kaswamila (NIMR Mwanza Centre). Equally, the authors thank the staff at the Laboratory for Infectious Diseases at University of Copenhagen, Hvidovre Hospital (Hvidovre, Denmark) and at the AIDS Laboratory at University of Copenhagen, Rigshospitalet (Copenhagen, Denmark) for their collaboration, as well as R. Wallis (Pfizer, New London, CT, USA) for his helpful comments and critique.
Footnotes
This article has supplementary material available at www.erj.ersjournals.com
Support Statement
Data from this study were presented, in part, at the 2nd Global Symposium on IGRAs, (Dubrovnik, Croatia; 2009), the European Respiratory Society 19th Annual Congress, (Vienna, Austria; 2009) and the 17th Conference on Retroviruses and Opportunistic Infections (CROI) (New Orleans, LA, USA; 2010).
Statement of Interest
Statements of interest for M.G. Aabye, M. Ruhwald and P. Ravn, and the study itself can be found at www.erj.ersjournals.com/site/misc/statements.xhtml
- ©ERS 2010