By Carol A. Kemper, MD, FACP

The Problem of Diagnosing TB in Children

Starke JR and Committee on Infectious Diseases. Interferon-gamma release assays for diagnosis of tuberculosis infection and disease in children. Pediatr 2014;134:1763-1773.

From 2008-2010, nearly one-third (31%) of children in the United States with active tuberculosis (TB) were foreign-born. Among those who were U.S.-born, 86% had at least one foreign-born parent, and three-fourths had some international connection, through family or travel.

Detection of latent or active tuberculosis in children presents problems different from that of adults. Both the tuberculin skin test (TST) and the interferon-gamma release assays (IGRA) have limitations although less is known about the use of IGRAs in children. Many of the studies of the comparative sensitivity of theses assays were conducted outside of the United States where the risks of latent infection and underlying health conditions affect the interpretation/sensitivity of the assays. Part of the problem is the only real way to test the accuracy of the IGRA assays is to observe untreated children with clear exposure which doesn’t happen, at least not in the United States. And, in children with active TB disease, microbiologic diagnosis is often lacking much of the treatment is pre-emptive or empiric.

In examining various publications and meta-analysis of the use of IGRAs in children, the strongest and most consistent finding is that IGRAs work better in kids with latent TB (LTBI) than in those with active TB especially in settings of low risk and higher rates of BCG vaccination.

Observations of pooled specificity for the Quantiferon TB Gold test (QTF) compared with TST in unvaccinated children are reported to be 95% vs 93%, respectively. In contrast, in BCG-vaccinated children, the pooled specificity for the two tests was 89% vs 49%. Of 207 children in New York City with a positive TST, only 23% had a positive QTF, although a positive QTF more strongly correlated with exposure. Nonetheless, because the risks of active TB infection in children are so much greater, many providers will treat a positive TST in a child with exposure irrespective of the QTF test results. Among 55 children in Montreal with household exposure and a positive TST, 52 received isoniazid a negative QTF result altered the treatment course in only 3 cases. Thus, because the risk of progression to active TB outweighs the risk of treatment in most cases, the decision to treat a child with exposure and a positive TST and a negative QTF is based on the clinical assessment of exposure risk and age of the child.

Sensitivity of the IGRAs in children compared with TST is more difficult to determine and is essentially derived from studies of children with active TB. Reported sensitivity results for all TB disease in children range from 57-96% for QTF, 40-100% for the TSPOT-TB test, and 43-100% for TST. All three tests were less sensitive in children with active TB than in those with LTBI. At least two meta-analyses suggest that IGRAs are even less sensitive in children with higher TB burden and should not be used to rule-out the possibility of TB disease.

Indeterminate IGRA test results in children occur for a number of reasons, including operator error (the tubes may be shaken too hard), or patients may have underlying medical conditions, such as malaria, hepatitis, malnutrition, or parasitic infections, that render the response to the control well insufficient (a negative control renders the test “uninterpretable” or indeterminate). Studies have documented a higher frequency of indeterminate IGRA test results in children compared with adults. Two studies conducted in children in Croatia and in Greece found that indeterminate test results were more frequent in younger children (those younger than 5 years of age) (8.1%) compared with older children (2.7%), but the tests behaved otherwise similarly.

Immunocompromised children: In general, both TST and IGRA perform poorly in children with HIV-infection or underlying immunosuppressive disease, similar to that observed in adults. Three systematic reviews suggest that TSPOT-TB test may be more sensitive than the QTF in children with HIV (72% vs 61%), but neither performed better than TST. Earlier studies suggested that the IGRAs had diminished sensitivity in younger children (those younger than 5 years of age), but the data have been inconsistent and muddied by the fact that very young children are more likely to quickly progress to active TB with larger burdens of infection. Of 397 South African children younger than 3 years of age suspected of having active TB, the sensitivity of the QTF was 38% compared with 35% for traditional TST testing. Better results may be observed in healthier children with lower rates of malnutrition, but the small numbers of such children in developing countries make such observations difficult.

In summary:

• A positive IGRA in a child should be viewed as consistent with TB infection, but, similar to that in adults, does not correlate with disease activity;

• Children should be treated for household exposure if either a TST or IGRA is positive, especially if they are younger than the age of 5.

• Children younger than 5 years of age are more likely to have indeterminate test results than adults, and repeating the test is not likely to be of value. Such children with household exposure should receive empiric treatment.

• And finally, a negative IGRA should NEVER be used to rule-out active TB in a child. The sensitivity of the assay in children with active TB is too low.

How Best to Treat Latent TB?

Stagg HR, et al. Treatment of latent tuberculosis infection: A meta-analysis. Ann Intern Med 2014;161(6):419-428.

The most widely accepted treatment of latent TB infection (LTBI) is 6 to 9 months of isoniazid. Nine months of INH is recommended in the United States although recent data suggest that a combination of weekly INH and rifapentine for 12 weeks may be more effective and less toxic than the standard approach but this approach requires directly observed therapy on a weekly basis for three months. In order to examine the most effective and least toxic regimens in current use for the treatment of LTBI, a meta-analysis was performed, using 1516 published articles and 53 different studies. Data on 15 different treatment regimens were available, although direct comparative treatment data were available for only a few. Only 45 studies contained data on progression to TB, and 25 contained data on hepatotoxicity. Twelve studies examined only household contacts of patients with pulmonary TB; only one was performed in health care workers; and 5 studies included children. Only 8 studies contained a non-treatment control arm.

Fifteen LTBI regimens were compared through a series of complex models (using MTC modeling), including various durations of INH; rifampin as a single agent for 3-4 months; combinations of INH and rifampin for 3-4 months; combination INH and rifapentine weekly for 3 months; and combinations containing PZA for various time periods. At least 40 different comparisons were made. Combinations of INH-rifabutin (utilizing different doses of rifabutin) ranked highest in efficacy, although the confidence intervals were quite wide. The INH-rifampin-PZA regimens ranked third in efficacy. And the use of either rifampin for 3-4 months or INH-rifampin for 3-4 months also ranked high. Regimens of INH for 12 months or longer also ranked high.

On the other side of the analysis were the results of hepatotoxicity (HT), for which the data were more limited. Data were available for only 20 comparisons, but nonetheless proved interesting. Lower rates of HT were observed for rifampin alone or for the combination of INH-rifapentine both of which were significantly lower than that for INH for either 6 or 9 months (or for INH for 12-72 months). Shorter courses of INH-rifampin had less toxicity than INH for 12-72 months. Regimens containing PZA had the highest rates of HT, even when compared with INH-rifapentine.

A total of 5 deaths were reported for all of the publications, at least 4 of which occurred in patients receiving INH for 12 months or longer. The degree to which the LTBI regimen contributed to the death was not always apparent.

This meta-analysis suggests that LTBI regimens containing a rifamycin are generally more effective and less hepatotoxic than those using INH for 9 months or longer. Given the long half-life of rifapentine, weekly observed administrations of combination INH-RPT for three months may turn out to be the most effective and least toxic regimen. Studies of non-directly observed weekly INH-RFP are in progress. It is, however, important to recognize that increasing rates of INH and rifampin may alter the relative benefits observed in these studies. In our area, baseline resistance to INH and rifampin are running about 4 and 17%, respectively.

Transmission Risk in Smear-negative TB

O’Shea MO, et al. Time-to-detection in culture predicts risk of Mycobacterium tuberculosis transmission: A cohort study. Clin Infect Dis 2014;59(2):177-185.

Sputum acid-fast smears are a fairly quick and widely used method for screening patients with possible active TB and have been used as a marker of infectivity for more than a hundred years. But the technique has some distinct shortcomings. Smears are able to detect approximately 5,000 to 10,000 organisms per mL sputa so a lot has to do with the specimen provided and the degree of TB involvement. This number of colony-forming units generally corresponds to more severe TB involvement at a minimum, a very large infiltrate, multi-lobar involvement, or cavitary disease. This limits its utility in other cases of pulmonary TB. It is also time intensive and subject to the availability and skill of the microbiology technician. And in patients on treatment, it cannot distinguish between viable organisms in sputa or non-viable organisms. In contrast, the newer automated liquid culture systems are much more sensitive, and able to detect 10-100 cfu/mL.

Further, increasing data suggest that sputum smears may not be the best marker of contagious risk smear-negative cases of TB may, in fact, transmit infection to others. In this study conducted in Birmingham, England, in 2010, risk factors for transmission were examined for 184 cases of pulmonary TB, including time to culture detection and chest radiographs, graded for disease severity. (Grading of chest radiographs were as follows: 0 for normal, 1 for adenopathy, 2 focal area of consolidation, 3 bilateral focal consolidation, 4 unilateral cavity, 5 bilateral cavity, and 6 miliary disease). Sputum samples were cultured for up to 70 days (10 weeks) using the BACTEC-MGIT-960 system. Time to culture detection was grouped into shorter (< 9 days) and longer (> 9 days) durations. Organisms from both index cases and subsequent cases were analyzed using newer molecular techniques, and clusters of microbiologically confirmed cases were identified.

Including only those patients with confirmed positive sputum cultures and excluding those with HIV infection, a total of 111 index cases were evaluated. More than half the index cases were smear-positive (51.5%), and the median time to culture-detection was 13 days. These cases resulted in 1246 contacts, of which 825 completed screening. Of these, 148 (17.9%) were diagnosed with latent TB and 17 (2.1%) were diagnosed with active TB giving an overall transmission rate of 0.2.

The risk of transmission strongly correlated with time to culture-detection (TTD) and was more than double for patients with short TTD compared with long TTD (0.32 vs 0.15). Smear-negative index cases resulted in 37/148 (25%) of all latent infections and 3/17 (17.3%) of new active cases. A strong inverse collection was also observed between TTD and the higher grades of chest radiographs every 1+ point increase in severity grade correlated with a 3.2 day-decrease in TTD (p < .0001). Only 2 of the 66 smear-negative cases had TTD < 9 days compared with 35/70 smear-positive cases.

Sputum smears may not be the best marker of infectivity. This study indicates that 25% of smear-negative cases resulted in transmission of infection to others. Only 7 smear-negative cases had TTD of < 9 days but smear-negative cases resulted in a total of 40 new cases of latent and active TB. Grading of chest radiographs and time to sputum culture positivity appear to be better predictors of contagiousness.