Xpert Ultra, Pediatric Pulmonary Tuberculosis and Stool: Forward Progress Raises New Questions

In this issue of Clinical Infectious Diseases, Kabir et al present novel data from 454 Bangladeshi children with clinical suspicion of pulmonary tuberculosis (TB) who underwent stool testing with Xpert MTB/RIF Ultra, the newest generation of this assay with a limit of detection of approximately 16 colony-forming units (CFU)/mL (compared to ~131 CFU/mL on the previous version) [1, 2]. Making a microbiologic diagnosis of pediatric TB is difficult given the paucibacillary character of disease and the logistical and technical challenges of getting adequate samples. Thus, the work by Kabir et al represents an important advance in our understanding of diagnostic potential from an easily collected nonpulmonary specimen. At the same time, the observations in this study raise some notable questions that must be addressed before the use of Xpert MTB/RIF Ultra on stool samples can be widely adopted for diagnosis of pediatric TB—most notably the significance of positive results at the very lowest limit of detection, so called “trace-call.”

Multiple adult and pediatric studies using Xpert Ultra on pulmonary specimens have demonstrated increased sensitivity relative to the previous generation of Xpert MTB/RIF, making Ultra particularly useful in patients with paucibacillary disease, such as persons with human immunodeficiency virus (HIV) and children [3, 4]. Similar gains in sensitivity have been noted for extrapulmonary specimens including adult stool specimens [5]. While a recent systematic review demonstrated 67% sensitivity of the Xpert MTB/RIF in pediatric stool samples against a microbiologic reference standard, the work presented by Kabir et al is the first to utilize Xpert Ultra on stool from pediatric patients [6].

In emerging Xpert Ultra literature, increased sensitivity—particularly with inclusion of trace-call positives—comes at the cost of a small decrease in specificity. However, this trade-off can either be mitigated by repeat testing (as is recommended for trace-call–positive pulmonary samples from adults without HIV infection), or can be justified in higher-risk populations such as persons infected with HIV, children, or those with suspected extrapulmonary disease including meningitis [7]. Given the potential for rapidly progressive and severe disease in children, an emphasis on sensitivity at the expense of specificity is a important characteristic of decision making in pediatric tuberculosis. In this context, current World Health Organization recommendations state that trace-call positives be considered confirmation of active disease in children [7].

The findings presented by Kabir et al of numerous children in whom the only microbiologic evidence of TB is a trace-call result in stool appear to push this calculus to an extent that the trade-offs between increased sensitivity and decreased specificity may need to be reexamined. Furthermore, all considerations of sensitivity and specificity of testing of extrapulmonary samples are hampered by significant inherent weaknesses of the reference standard, which in this study is microbiologically confirmed TB on induced sputum specimens using culture, Xpert, and Xpert Ultra—tests which themselves have variable and imperfect sensitivity for confirming pediatric TB disease [8].

Of 454 patients with clinical suspected TB presented by Kabir et al, 72 were microbiologically confirmed. However, of these, 60% (43/72) were “confirmed” only through the Ultra result on stool testing, 98% (42/43) of which were trace-call. To treat the stool trace-calls as positive in this context is to operate under the assumption that a significant proportion of these results are true positives, and that the sensitivity of Ultra in stool exceeds that of induced sputum. However, there are 2 key observations in this study that do not support this assumption. First, such high sensitivity, if true, should be reflected in a rate of Ultra positivity (including trace-call) among sputum-confirmed patients higher than the 59% (17/29) observed. Second, Kabir et al describe 6 children who were trace-call positive in stool with no additional microbiologic confirmation and not treated for tuberculosis disease. Despite the initial clinical suspicion of pulmonary TB that led to entry in to this study, none of these patients developed “active TB” over a 6-month follow-up period.

If some or many of the stool trace-call positives without additional microbiologic confirmation are then not true positives, it will be important to answer the question as to what these results do represent. The most common explanation for “false-positive” trace-call results is the presence of nonviable TB, particularly among patients who have been recently treated or self-recovered from incipient TB [7]. Laboratory contamination is also a proposed possibility. However, neither explanation appears to be particularly likely in the pediatric population described by Kabir et al: The authors do not note cases of previously treated TB among the study population, and laboratory methods and quality control appear robust. Could these tests represent intestinal exposure to environmental mycobacteria such as Mycobacterium bovis? It would be informative in future studies to quantify rates of trace-call–positive stools among healthy patients from a population similar to the one included in the Kabir et al study, as well as from patients from low-TB-incidence populations. Also useful would be correlation of stool trace-call results with immunologic markers of TB infection such as the interferon-γ release assay (IGRA).

Such unresolved questions notwithstanding, this study represents an important step toward incorporation of stool Xpert MTB/RIF Ultra into available diagnostic tools for TB. In clinical environments where individualized clinical decision making is possible, adjunct data such as clinical and radiologic characteristics, local epidemiology/TB incidence, exposure history, and IGRA test results may be utilized to enhance the predictive value of a stool result, including trace-call positives. Furthermore and most importantly, Kabir et al demonstrate that more than one-third of patients with microbiologically confirmed TB on induced sputum are positive above the trace-call threshold in stool. Thus, if these results are replicated in future studies and in practice, stool testing is likely to represent a very practical first step in the diagnosis of pediatric TB disease. A positive result above trace-call could confirm TB, demonstrate drug susceptibility, and preclude the need for more invasive sampling. A negative or (trace-call positive) would need to be followed by more extensive testing including sputum collection when available.

The transition in microbiologic diagnostics from culture-based to molecular-based techniques frequently yields new questions related to the implications of positive results at the lowest thresholds, and application of the Xpert TB/RIF Ultra in children using stool samples appears to be no exception. Nevertheless, such advances in technology and work such as presented in this study are essential to progress toward global TB elimination.

Note

Potential conflicts of interest. The author: No reported conflicts of interest. The author has submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.

References