The development of the Xpert MTB/RIF assay (Cepheid, Sunnyvale, United States of America (United States)) was a significant step forward in improving the diagnosis of TB and the detection of rifampicin resistance globally. However, Xpert MTB/RIF sensitivity is suboptimal, particularly among people (including children) with smear-negative TB and people (including children) living with HIV. The Xpert MTB/RIF Ultra (Cepheid, Sunnyvale, United States), hereafter referred to as Xpert Ultra, was developed by Cepheid as the next-generation assay to overcome these limitations. Xpert Ultra has a lower limit of detection than the Xpert MTB/RIF, with an additional "trace" semi-quantitative category. It uses the same GeneXpert® platform as the Xpert MTB/RIF. Xpert MTB/RIF is strongly recommended as the initial diagnostic test for TB and rifampicin-resistance detection in sputum (including induced and spontaneously expectorated sputum), gastric aspirate, NPA and stool specimens rather than smear microscopy/culture and phenotypic DST in children aged below 15 years with signs and symptoms of PTB. The Xpert Ultra assay is recommended in the same population as the initial diagnostic test for TB and detection of rifampicin resistance in sputum or NPA, rather than smear microscopy/culture and phenotypic DST (16).
The 2020 rapid diagnostics guidelines and 2021 update listed the evaluation of the diagnostic accuracy of Xpert Ultra in gastric aspirate or stool specimens for PTB in children as a research priority. Therefore, a systematic review and meta-analysis were conducted to evaluate the use of the Xpert Ultra assay in two paediatric specimens, namely gastric aspirate and stool, for which insufficient data were available at the time of the GDG meeting on molecular assays in December 2019 (18).
PICO question: In children aged below 10 years of age with signs and symptoms of pulmonary TB seeking care at health care facilities, should Xpert Ultra in gastric aspirate or stool be used to diagnose pulmonary TB and rifampicin resistance, as compared with a microbiological/composite reference standard?
Evidence: In preparation for the GDG meeting on the management of TB in children and adolescents, an update of the review on the diagnostic accuracy of Xpert MTB/RIF and Xpert Ultra for TB disease in children (18) was performed. A systematic search of the literature was carried out in January 2021. This update focused specifically on the diagnostic accuracy of Xpert Ultra in gastric aspirate or lavage specimens and stool specimens for the diagnosis of PTB and rifampicin resistance in children below 10 years of age. Nine studies that evaluated the diagnostic accuracy of Xpert Ultra in gastric aspirate and/or stool specimens in children were identified. For the meta-analysis, six studies (653 participants) provided data for gastric specimens (19-24) and six studies (1278 participants) for stool specimens (20, 23-27). The review found that for gastric aspirate, Xpert Ultra sensitivity was 64% in children 0-9 years, against a microbiological reference standard and specificity was 95%. For stool, Xpert Ultra sensitivity was 53% in children 0-9 years, against the microbiological reference standard and specificity was 98%. Sensitivity estimates against a composite reference standard were lower for both specimen types. There were no studies that evaluated the diagnostic accuracy of Xpert Ultra for detection of rifampicin resistance using gastric aspirate or stool specimens.
Of the nine studies, eight (89%) reported the proportion of Xpert Ultra positive results that were trace results. In these eight studies, of the total Xpert Ultra positive results, the proportion (expressed as a percentage) of Ultra trace results ranged from 0% to 66% (median 52%) in studies evaluating gastric specimens and from 0% to 84% (median 52%) in studies evaluating stool specimens.
GDG considerations: The GDG members discussed that all TB tests, including the microbiological reference standard (usually culture or Xpert Ultra on respiratory samples) have suboptimal sensitivity in children, due to the paucibacillary nature of TB disease in this age group. Therefore, the panel highlighted that a positive test accurately determines a case of TB disease, but a negative test does not exclude TB disease. As no studies assessed the diagnostic accuracy for Xpert Ultra for the detection of rifampicin resistance in children, conclusions related to the use of Xpert Ultra for the diagnosis of rifampicin resistance were based on data in adults, evaluated for the 2020 rapid diagnostic guidelines. The GDG members agreed that the desirable effects from the test in both specimens are moderate, but large for the detection of rifampicin resistance, because of the large impact on the choice of effective treatment when rifampicin resistance is detected. The panel noted that false-negative results should be interpreted within the context of the clinical picture, and clinicians should not solely use a negative test to rule out disease. The GDG judged that the balance of effects favours the intervention (considering both the moderate desirable effects and small undesirable effects), based on moderate certainty evidence for both gastric aspirate and stool samples.
The GDG judged that more time and training-related resources would be required to collect gastric aspirate specimens, and that the procedure would also often require hospital admission. Stool processing may result in some additional time requirement to prepare the sample, depending on the processing method. The panel decided upon comparing the cost of doing the tests on the sample types versus not doing the test. The panel then concluded that the costs for both specimen types vary.
While there were no studies on the cost-effectiveness of gastric aspirate, a cost-effectiveness study compared stool testing against clinical diagnosis at PHCs in Uganda, assuming that no respiratory samples would be collected without referral to the hospital or without invasive sampling (see web annex 4). Compared to that standard of care, stool testing was more effective, but also more costly. The GDG members discussed that an Xpert Ultra test on stool samples becomes cost-effective even at a TB prevalence level of 3% at a zero discount rate (the practice of discounting future health effects). Implementation aspects such as the importance of obtaining microbiological confirmation and the detection of rifampicin resistance would further favour cost-effectiveness of stool testing. The panel therefore concluded that the evidence probably favours Xpert Ultra testing on stool samples.
In terms of acceptability, although the gastric aspirate sample collection was found to be an invasive procedure (that could be uncomfortable to children), the panel felt that the procedure is probably acceptable to children, caregivers and health care workers (HCW), considering the role of this procedure to obtain bacteriological confirmation at higher level facilities. The GDG noted the importance of non-invasiveness of stool samples for bacteriological testing for the diagnosis of TB in children and agreed that this sample type was acceptable to all relevant stakeholders, including HCWs, laboratory technicians, parents and children.
The panel agreed that, despite the requirements of training and skills, using Xpert Ultra on gastric aspirate is probably feasible to implement, especially at higher levels of the health care system. The majority of the panel felt that Xpert Ultra on stool samples is feasible to implement at all levels of the health care system (see web annex 3).