Intrathoracic (pulmonary) tuberculosis (PTB) in children is largely paucibacillary. In young children who cannot expectorate, specimen collection requires gastric aspiration, sputum induction (with nasopharyngeal suctioning) or nasopharyngeal aspiration. Respiratory specimens typically contain low concentrations of Mycobacterium tuberculosis bacilli, resulting in low sensitivity of currently available molecular tests and culture.(1)
In children, increasing the number and variety(2–4) of specimens improves the overall diagnostic yield (detection of M. tuberculosis). Specimen collection over consecutive days may have a higher cumulative yield than same-day collection.(3,5) However, it is less practical and more costly if hospitalization is required. In young children, pooling multiple specimens after collection, to obtain a higher volume specimen, has been suggested as an approach to improve the bacteriologic yield.(6)
In this study of children with suspected PTB, we compared the diagnostic yield and culture contamination of multiple respiratory specimen types pooled for microbiologic testing, with the yield from individual respiratory specimens for an individual child.
MATERIALS AND METHODS
This analysis was part of a prospective diagnostic cohort study enrolling children with suspected PTB in Cape Town, South Africa. Eligibility criteria and enrollment investigations have been previously described.(7) In brief, children <13 years of age presenting to 2 public referral hospitals, with history and symptoms of suspected PTB, were consecutively enrolled May 2014 to March 2017.(7) We excluded children who had received >1 dose of antituberculosis (TB) therapy before the first day of respiratory specimen collection, those with a convincing alternative clinical diagnosis, children with only extrathoracic TB or those who lived remotely. The pooling strategy was assessed in children who could not expectorate sputum, typically children <5 years of age.
Investigations included HIV testing, tuberculin skin test (Mantoux, 2 Tuberculin Units PPD RT-23, Statens Serum Institute, Copenhagen) and a chest radiograph (anteroposterior and lateral), evaluated by 2 independent experts. During the study, there was a global stock-out of tuberculin skin test, leading to a number of children not having the test. Interferon-gamma release assays were not used.
The attending clinicians were responsible for treatment decisions, with all study-related results made available to them. International consensus clinical case definitions were used to classify participants as “confirmed TB”, “unconfirmed TB”, and “unlikely TB”.8 Categories were assigned retrospectively at the 2-month follow-up, after assessment of treatment response and review of culture results.
Specimen Collection and Laboratory Methods
The study protocol required the collection of 1 specimen of 3 different types, on each of 2 days (Figure 1, Supplemental Digital Content 1, http://links.lww.com/INF/D362). The standard schedule was an early morning gastric aspirate (GA), a nasopharyngeal aspirate (NPA) and induced sputum (IS). The same order of collection was followed on both days, although specimens for pooling were collected with no time lag between them. On day 1, a minimum of 2 hours between GA, NPA and IS were observed (Document 1_Specimen collection, Supplemental Digital Content 2, http://links.lww.com/INF/D363). In November 2015, NPA collection was stopped, because of low yield of M. tuberculosis for NPA (Table 1, Supplemental Digital Content 3, http://links.lww.com/INF/D364) which did not justify the cost of individual testing. NPA was however still collected for the pooled specimen.
Specimens were processed at the National Health Laboratory Service, Tygerberg Hospital following standard protocols. The reconstituted pellets from specimens collected on the second day were combined (pooled) into 1 centrifuge tube and vortex-mixed (Figure 1, Supplemental Digital Content 1, http://links.lww.com/INF/D362). The individual concentrated day 1 specimens and the pooled respiratory specimen were subjected to fluorescent Auramine-O smear microscopy, Xpert MTB/RIF (Xpert: Cepheid, Sunnyvale, CA) and liquid Mycobacteria Growth Indicator Tube (Becton Dickinson, Sparks, MD) culture.
GenoType MTBDRplus line probe assay (Hain Lifescience, Nehren, Germany) was performed on positive cultures for mycobacterial identification.
The paired binary diagnostic outcomes of the pooled specimens by culture or Xpert (day 2) for each child and individual specimens (GA/IS/NPA culture or Xpert) (day 1) for each child were compared for marginal homogeneity using McNemar’s test. The Benjamini-Hochberg method was used to correct for multiple testing (m = 34) of the pooled specimens using a false discovery rate of 10%, which corresponded to a P ≤ 0.04. Diagnostic yield was defined as number of children (not specimens) positive for M. tuberculosis by Xpert or culture. We compared (1) the diagnostic yield of pooled versus each individual day 1 specimen type separately (pooled vs. GA, pooled vs. NPA, pooled vs. IS); (2) the proportion of contaminated cultures and invalid/error Xpert results for pooled versus individual specimens separately); (3) the total diagnostic yield by Xpert and culture (either positive) of pooled specimens versus the combined yield from day 1 individual specimens (yield by Xpert or culture from any individual specimen). For (1) and (2), the pooled specimen had to contain the specimen type to which it was compared (per-protocol approach). For (3), both per-protocol (the number and type of specimens in the pooled specimen being the same as the number and type of day 1 individual specimens) and pragmatic approach (including all participants) were used.
Stellenbosch University Health Research Ethics Committee (N11/09/282) and local health authorities approved the study.
In 304 enrolled children (Figure 2, Supplemental Digital Content 4, http://links.lww.com/INF/D365), the median age was 15.1 (interquartile range 9.6–27.2) months (Table 2, Supplemental Digital Content 5, http://links.lww.com/INF/D366). Fifty-one of 304 (16.8%) children had confirmed TB: 44 were confirmed by study specimens and 7 on other specimens. Therefore, the total diagnostic yield for the pooling study was 44 children with confirmed TB. Anti-TB treatment was initiated in 51 (100%) children with confirmed TB, 77/97 (79.4%) with unconfirmed TB and 6/156 (3.8%) with unlikely TB.
Pooled Specimens Versus Individual Specimen Types
When comparing pooled specimens to each individual specimen type (Table 1, Supplemental Digital Content 3, http://links.lww.com/INF/D364 for overall specimen results and Fig. 3, Supplemental Digital Content 6, for specimen flow; http://links.lww.com/INF/D367), the proportion difference in diagnostic yield for pooled versus single IS and single NPA was significantly higher for pooled specimens, by culture alone, Xpert alone and culture and Xpert combined (any test positive), but not for pooled versus single GA.
The proportion difference in contaminated cultures was not significant for pooled versus any individual specimen type (Table 1).
The Diagnostic Yield of Pooled Specimens Versus the Overall Yield From Individual Specimens
For the per-protocol analysis, 234/304 (77.0%) children who were included had culture on all specimens, while 223 (73.4%) had Xpert on all specimens. For both the per-protocol and pragmatic analyses, the proportion difference in diagnostic yield, culture contamination and Xpert error rate of pooled specimens versus all individual specimens combined was not significant (Table 1).
Of the 134 children treated for TB, 16 (11.9%) started treatment before (median 1.5 days) collection of specimens for pooling. Only 1 had a pooled specimen with negative culture and an individual positive GA culture, while 2 cases had culture-positive pooled specimens with negative individual specimens.
We found that the overall diagnostic yield from pooled specimens (pooled GA, NPA, and IS) was not different from that of a single GA specimen but was significantly higher than the yield from single IS or single NPA specimen. Pooled specimens and individual GA also had similar incremental yield and proportions of culture contamination, suggesting that the GA contained within the pooled specimen was probably the main contributor to both the diagnostic yield and the contamination rate of pooled specimens.
Pooling may result in high specimen volume and also allows for sampling of the respiratory tract at different time points. Early studies demonstrated that shedding of M. tuberculosis bacilli into respiratory secretions may be intermittent,9 and pooling may increase the chance of collecting respiratory secretions which contain higher concentrations of bacilli. GA may be viewed as a naturally “pooled” specimen, as it consists of multiple expectoration and swallowing cycles, collected within the stomach. In contrast, both IS and NPA reflect a single instance of sampling the respiratory tract.
We elected to pool (in the laboratory) specimens of different types collected consecutively in a standard manner, to avoid prolonged periods of fasting, while allowing for collection on a single day. A same-day collection strategy avoids longer hospital stay and potential loss to follow-up associated with consecutive-day collection in ambulatory settings.
Although we may have compromised some of the diagnostic potential of pooled specimens by collecting specimens consecutively for pooling with no time interval between specimens, our focus was to reduce cost of laboratory testing and improve feasibility, and long waiting times would not be practicable in routine clinical settings. A small proportion of children started anti-TB therapy before collection of pooled specimens. However, our data suggest that this did not impact negatively on M. tuberculosis detection.
As expected in pediatric TB, collecting multiple specimens from children optimizes M. tuberculosis detection (Figure 4, Supplemental Digital Content 7, http://links.lww.com/INF/D368). The yield from multiple specimens on a single day was substantial (38/44 = 86.4%) (Table 1). Previous studies have also shown that “front-loading” specimen collection on 1 day was comparable with consecutive-day collection.3,10 However, pooling specimens did not have the expected increase in diagnostic yield. Given the good performance of GA, pooling 2 GA could be considered in future studies.
The study team wishes to acknowledge the study participants and their families; the staff at the participating health facilities and at the Desmond Tutu TB Centre, for their dedication and assistance. Specific acknowledgement is given to Cornelia Rautenbach, Sven Friedrich and Kim Hoek for technical assistance, and to Rory Dunbar for data management.
1. Detjen AK, DiNardo AR, Leyden J, et al. Xpert MTB/RIF assay for the diagnosis
of pulmonary tuberculosis
in children: a systematic review and meta-analysis. Lancet Respir Med. 2015;3:451–461.
2. Rachow A, Clowes P, Saathoff E, et al. Increased and expedited case detection by Xpert MTB/RIF assay in childhood tuberculosis: a prospective cohort study. Clin Infect Dis. 2012;54:1388–1396.
3. Al-Aghbari N, Al-Sonboli N, Yassin MA, et al. Multiple sampling in one day to optimize smear microscopy in children with tuberculosis in Yemen. PLoS One. 2009;4:e5140.
4. Marcy O, Ung V, Goyet S, et al. Performance of Xpert MTB/RIF and Alternative Specimen Collection Methods for the Diagnosis
of Tuberculosis in HIV-Infected Children. Clin Infect Dis. 2016;62:1161–1168.
5. Zar HJ, Workman L, Isaacs W, et al. Rapid molecular diagnosis
of pulmonary tuberculosis
in children using nasopharyngeal specimens. Clin Infect Dis. 2012;55:1088–1095.
6. Cuevas LE, Petrucci R, Swaminathan S. Tuberculosis diagnostics for children in high-burden countries: what is available and what is needed. Paediatr Int Child Health. 2012;32(Suppl 2):S30–S37.
7. Walters E, van der Zalm MM, Palmer M, et al. Xpert MTB/RIF on stool is useful for the rapid diagnosis
of tuberculosis in young children with severe pulmonary disease. Pediatr Infect Dis J. 2017;36:837–843.
8. Graham SM, Cuevas LE, Jean-Philippe P, et al. Clinical case definitions for classification of intrathoracic tuberculosis in children: an update. Clin Infect Dis. 2015;61(Suppl 3):S179–S187.
9. Hippke E. Neue versuche über die bedeutung der tropfcheninfektion für die ausbreitung der lungenschwindsucht. Zeitschr f Hyg u Infectionskr 1921. 93:122–146.
10. Hatherill M, Hawkridge T, Zar HJ, et al. Induced sputum or gastric lavage for community-based diagnosis
of childhood pulmonary tuberculosis
? Arch Dis Child. 2009;94:195–201.