Accuracy of Xpert Mtb/Rif Ultra for the Diagnosis of Pulmonary Tuberculosis in Children

Abstract

We determined the accuracy of the new Xpert MTB/RIF Ultra assay for diagnosis of pulmonary tuberculosis in children, using banked-induced sputum. Sensitivity and specificity were 75.3% and 96.9%, respectively, compared with a culture result from the same specimen. Of those children with microbiologically confirmed tuberculosis (positive Xpert, Ultra or culture), Xpert detected 63.2%, Ultra 73.7% and culture 82.9%.

Microbiologic confirmation of pulmonary tuberculosis (PTB) in children remains important, as clinical and radiologic diagnosis lack specificity.¹ Xpert MTB/RIF (Xpert, Cepheid, Sunnyvale, California), a cartridge-based nucleic acid amplification assay, permits rapid detection of Mycobacterium tuberculosis complex and identification of rifampin resistance.² However, sensitivity for the diagnosis of pediatric PTB is limited; 62% for sputum samples in a recent meta-analysis.³ Xpert MTB/RIF Ultra (Ultra, Cepheid, Sunnyvale, California), a new tuberculosis (TB) diagnostic on the GeneXpert platform, has a substantially lower limit of detection and improved sensitivity than Xpert in studies in adults.⁴ Here we report the first data on the accuracy of Ultra testing of sputum for diagnosis of PTB in children.

MATERIAL AND METHODS

Children (<15 years of age) presenting to Red Cross Children’s Hospital, Cape Town, South Africa, with suspected PTB between April 2012 and September 2016 were considered for enrolment. Suspected PTB was based on cough of any duration and one of: household contact with an infectious TB source case within the preceding 3 months, loss of weight or failure to gain weight in the preceding 3 months, a positive tuberculin skin test to purified protein derivative (2Tuberculin Units (TU), PPD RT23; Staten Serum Institute, Denmark, Copenhagen; positive defined as ≥10 mm in HIV-uninfected children and ≥5 mm in HIV-infected children) or a chest radiograph suggestive of PTB. Children were excluded if they received more than 72 hours of TB treatment, if they were not resident in Cape Town or if informed consent was not obtained. For this analysis, children were excluded if 2 induced sputum samples could not be obtained. Written informed consent for enrolment in the study was obtained from a parent or legal guardian. The Research Ethics Committee of the Faculty of Health Sciences, University of Cape Town, approved the study.

TB therapy was initiated at the discretion of the treating doctor (resident or specialist). Follow-up visits were done at 1, 3 and 6 months for children on TB therapy and at 1 and 3 months for those not treated. Response to treatment was assessed at follow-up by recording symptoms, signs, weight and repeating the chest radiograph at completion of TB treatment.

Two consecutive induced sputum specimens were obtained as previously described.² After standard decontamination with sodium hydroxide, centrifuged sputum deposits were resuspended in 1.5 mL of phosphate buffer. For both sputum samples, a drop of sediment was used for fluorescent acid-fast smear microscopy, and 0.5 mL of resuspended pellet for automated liquid culture (BACTEC MGIT; Becton Dickinson, Cockeysville, MD). Cultures were incubated for up to 6 weeks. Positive cultures were identified by acid-fast staining followed by MTBDRplus testing (Hain Lifescience, Nehren, Germany). For the first induced sputum, and for a subset of the second induced sputums, a further aliquot of 0.5 mL was used for Xpert testing, as per manufacturer’s recommendations. For the second induced sputum, an aliquot of 0.5 mL was frozen at −80°C for Ultra testing in September 2016. Frozen aliquots from the second sample were subsequently thawed in batches for Ultra testing according to manufacturer’s recommendations. Staff performing Xpert and Ultra test results were blinded to clinical and other microbiologic results.

TB diagnostic categorization was based on clinical and microbiologic investigations, in line with consensus definitions, but excluding results of Xpert and Ultra tests to avoid inclusion bias⁵: “confirmed TB” (any induced sputum culture positive for M. tuberculosis), “unlikely TB” (culture negative, no TB treatment given and documented resolution of symptoms and signs at 3-month follow-up visit) or “unconfirmed TB” [all other children, including (1) children clinically diagnosed who were placed on TB treatment and (2) children with negative culture for M. tuberculosis, who were not placed on TB treatment and who did not have documented resolution of symptoms and signs at follow-up (either due to loss to follow-up or due to persistent symptoms or signs at follow-up)]. We excluded children who had a positive culture from an extrapulmonary sample only.

RESULTS

Four hundred fifty-three eligible children produced 2 sputum samples over the study period and were included (Figure 1). Fifty Ultra tests did not yield a valid result; 37/50 of the invalid results occurred in 2 batches processed on a single day. Invalid results were not related to a specific batch of test cartridges or to an individual operator. Of the remaining 403 children, 36 were excluded: 17 with contaminated TB cultures, 6 with no TB culture and 13 with a positive TB culture result from an extrapulmonary sample only.

The median (interquartile range) age of the remaining 367 children was 33.0 (15.2–74.0) months, 186 (50.7%) were males, 71 (19.4%) were HIV infected, 31 (8.5%) had previously been treated for TB and 186/322 (57.8%) were tuberculin skin test positive. A total of 83 (22.6%) had a positive TB culture (from any sputum sample) and so were classified as confirmed TB, while 187 (51.0%) were classified as unconfirmed TB and 97 (26.4%) as unlikely TB. Overall 83/250 (33.2%) of those treated for TB were culture positive. The proportions of children with a positive culture from the first and second sputum samples were similar (20.9% and 19.9%; P = 0.796); 7 children had culture confirmation from the second sample only.

All 367 children had interpretable results from both TB culture and Ultra on the second sputum specimen; 73 (19.9%) had a positive TB culture from this sample, while 64 (17.4%) had a positive Ultra (P = 0.083). Ultra semiquantitative results were 14 (21.9%) trace, 27 (41.2%) very low, 6 (9.4%) low, 13 (20.3%) medium and 4 (6.2%) high positive. When comparing Ultra with a reference standard of culture performed “on the same sample,” sensitivity and specificity were 55/73 [75.3%; 95% confidence interval (CI): 63.9–84.7] and 285/294 (96.9%; 95% CI: 94.3–98.6), respectively. When comparing Ultra with a reference standard of culture performed on “any” sputum sample (83 children had a positive TB culture on any sample), sensitivity and specificity were 56/83 (67.5%; 95% CI: 56.3–77.4) and 276/284 (97.2%; 95% CI: 94.5–98.8), respectively. All the children who had a positive Ultra but negative TB cultures were classified as unconfirmed TB, and so specificity of Ultra was 100% if a reference standard of confirmed plus unconfirmed TB was used. Sensitivity was similar in HIV-infected and HIV-uninfected children (66.7% and 67.7%; P = 0.935). Specificity of Ultra was similar in children who had versus had not been previously treated for TB 23/24 (95.8%; 95% CI: 78.9–99.9) versus 249/256 (97.3%; 95% CI: 94.4–98.9), P = 0.687.

When comparing Xpert with Ultra results, 306/367 (83.4%) children had a valid Xpert result from the first sample, of these 48 (15.7%) were Xpert positive and 73 (23.9%) were culture positive (on any sputum sample). In this subgroup, the sensitivity and specificity of Xpert (using culture of any sample as the reference standard) were 47/73 (64.4%; 95% CI: 52.3–75.3) and 232/233 (99.6%; 95% CI: 97.6–100) and the sensitivity and specificity of Ultra were 48/73 (65.8%; 95% CI: 53.7–76.5) and 225/233 (96.6%; 95% CI: 93.3–98.5), respectively. Among those children with a positive Xpert, Ultra or TB culture (n = 76), Xpert detected 48/76 [63.2% (51.3–73.9)], Ultra 56/76 [73.7% (62.3–83.1)] and culture 63/76 [82.9% (72.5–90.6)], P = 0.117, for comparison of Xpert and Ultra.

In addition, a subgroup of 112 children had all 3 tests (TB culture, Xpert and Ultra) performed on the second sputum sample. Among this group (using culture from the second sample only as the reference standard), the sensitivities of Xpert and Ultra were identical; 17/22 (77.3%; 95% CI: 54.6–92.2). Specificities were 90/90 (100%; 95% CI: 96.0–100) and 86/90 (95.6%; 95% CI: 89.0–98.8), respectively.

DISCUSSION

In this cohort of South African children hospitalized with suspected PTB, Ultra detected 75.3% of cases positive by culture on the same sputum sample. Assuming that Xpert or Ultra positive, clinically confirmed but culture-negative samples were true positives, the sensitivity of Xpert and Ultra were 63.2% and 73.7%, respectively. False-positive Ultra results have been described, particularly in adults previously treated for TB. This is less likely to be a problem in young children, who infrequently have had a previous TB treatment episode. In this study, we did not detect any positive Ultra tests in children who were not diagnosed with TB by culture or on clinical grounds.

Limitations of this study include the relatively small sample size and retrospective Ultra testing. We cannot exclude the possibility that prolonged storage may have affected the likelihood of a positive Ultra result; however, there was no difference in positive rates by duration of storage (data not shown). Ultra was done on the second sputum sample, while Xpert was done on the first. It is possible that the quality of the first (tested with Ultra) and second (tested with Xpert) sputum samples were different; however, the proportion of first and second samples with a positive culture result was similar.

Because Ultra detected three-quarters of culture-confirmed cases, Ultra should not be used as a replacement test for culture in children. Larger studies are required to accurately determine the incremental benefit of Ultra over Xpert in this patient population.

FIGURE 1.: Study profile.

ACKNOWLEDGMENTS

The authors thank the contributions of the National Health Laboratory Service diagnostic microbiology laboratory at Groote Schuur Hospital, the study, laboratory and clinical staff at Red Cross Children’s Hospital and the Division of Medical Microbiology, and the children and their caregivers.