NEW YORK (GenomeWeb) – While Cepheid's GeneXpert MTB/RIF assay has the potential to provide rapid confirmation of tuberculosis infection in children younger than 16, its sensitivity is "suboptimum" and it does not effectively rule out TB, according to a meta-analysis published today.
The analysis appeared in The Lancet, and its publication coincides with World Tuberculosis Day. Cepheid also marked the day by commemorating shipment of the 10 millionth GeneXpert MTB/RIF test cartridge.
In the meta-analysis, an international group of researchers aimed to estimate how accurately MTB/RIF diagnosed pulmonary TB and rifampin resistance in children, including ones who were co-infected with HIV. They also assessed how this related to age, TB smear test results, and inpatient versus outpatient status.
Diagnosing pediatric pulmonary TB is different than diagnosing infection in adults because children are often unable to expectorate sputum. The meta-analysis also considered pediatric studies that used induced sputum and gastric lavage samples.
Fifteen studies met the standards to be included in the analysis. In total, these studies examined 4,768 respiratory specimens from 3,640 children suspected of having TB.
The analysis reported on the pooled sensitivities and specificities of GeneXpert MTB/RIF compared to two other commonly used diagnostic methods: microscopic examination and bacterial culture. In a few cases, the authors also considered a clinical diagnosis of TB – culture-negative cases in which a provider started anti-tuberculosis treatment without knowledge of GeneXpert test results.
The limit of detection of GeneXpert is estimated to be 131 colony-forming units per milliliter, while LOD is between 10 and 100 CFU per mL for culture, and 10,000 CFU per mL for microscopy. Children are more likely than adults to have TB with low bacterial levels, or paucibacillary TB, and the yield of culture in childhood TB can range between 20 and 70 percent. Not surprisingly, then, GeneXpert was found to be between 36 and 44 percent more sensitive than microscopy.
Overall, 420 of the 3,640 children suspected of having TB, or 12 percent, had TB infection confirmed by culture, while 406, or 11 percent, were positive by GeneXpert. In a subset of the studies, 44 percent of culture-negative children were given empirical antituberculosis treatment after clinical diagnosis of TB, and two percent of that group wound up being GeneXpert positive.
The authors concluded that the specificity of GeneXpert was consistently high, but found that sensitivity "varied broadly across studies and specimen types."
More specifically, in smear-positive cases, which presumably have higher content of bacteria, the pooled sensitivity was lowest for children four years old and younger using gastric lavage as a specimen, at 82 percent.
Using induced or expectorated sputum from this age group increased sensitivity to 94 percent. Sensitivity for children between five and 15 years old using sputum samples was 96 percent, while sensitivity on samples from HIV-positive and HIV-negative children was 97 and 94 percent, respectively.
However, in smear-negative cases, focusing only on sputum sample types, the pooled sensitivity was poor.
In children age four and under, as well as HIV-negative children, pooled sensitivity of the MTB/RIF assay was 44 percent compared to culture. For HIV-positive children, sensitivity was 60 percent, while for children age five to 15, it was 66 percent.
A prior meta-analysis had concluded GeneXpert's sensitivity on sputum from smear-negative children was 55 percent. Those results informed the WHO guidelines for pediatric TB published in 2013, which concluded, "Children suspected of having pulmonary TB but who have had a single negative result by Xpert MTB/RIF should undergo further diagnostic testing, and a child for whom there is a high clinical suspicion of TB should be treated even if an Xpert MTB/RIF result is negative or if the test is not available."
An accompanying comment in The Lancet summarized the limitations of the current meta-analysis. It noted that culture is "an imperfect reference standard" for children, and extrapulmonary TB — which is more common among children and has been previously shown to be more difficult to detect using GeneXpert — was not considered. The study also compared a single GeneXpert test result to multiple culture results per patient in some cases.
A majority of studies in the analysis were done on inpatients, who likely have a more severe illness and higher bacterial load, while GeneXpert is frequently being adopted in outpatient clinics as well. "As more data emerges from primary care … overall estimated sensitivity of Xpert will probably diminish," the comment concluded.
The analysis and comment both emphasized a need to develop techniques to use other types of samples to diagnose pediatric TB, such as urine or blood. Ongoing efforts in this domain include a collaboration between Harvard University and Akonni Biosystems to develop a microarray-based drug-resistant TB test and adapt Akonni's TruTip sample prep technology to purify cell-free TB DNA from urine for pediatric diagnostics, as previously reported by GenomeWeb.
Cepheid, meanwhile, is developing additional GeneXpert cartridges for TB that will have improved sensitivity and scan for more resistance mutations. The firm's TB Ultra test is expected to launch in 2016 and will have a multi-copy TB target to boost sensitivity by about 10- to 15-fold.