The World Health Organization estimated that 1 million children developed tuberculosis (TB) in 2017, representing ~10% of the global TB burden.1 Estimates of multidrug-resistant TB (MDR-TB; ie, Mycobacterium tuberculosis resistant to at least isoniazid and rifampicin) in children range from 25,000 to 32,000 cases per year.2
Children, especially those <5 years of age, who have drug-resistant TB (DR-TB), usually have transmitted resistance, ie, they have been infected by an organism with established resistance.3 TB in children serves as a form of sentinel surveillance for recently transmitted TB, as young children usually develop disease soon after being infected (>90% in the first 12 months following infection). However, DR-TB in children is rarely reported because of the paucibacillary nature of TB in children and the challenges in obtaining suitable samples for bacteriology and drug susceptibility testing (DST). There are few long-term surveillance studies, which may show trends in drug resistance in communities and children.
This study is a continuation of ongoing anti-TB drug resistance prevalence surveys in children at Tygerberg Hospital, Western Cape province, South Africa.3 We compare the two 2-year periods reported here with results of the previous five 2-year surveillance periods.
All children (<13 years) routinely investigated for TB and who had positive bacteriology (culture or Xpert MTB/RIF) for M. tuberculosis complex during the period 1 March 2013 to 28 February 2017, at Tygerberg Hospital, a provincial referral hospital, Cape Town, South Africa, were included. Tygerberg Hospital serves as a referral center for complicated pediatric TB cases including DR-TB cases; this referral pattern was already established in 1994. In the Western Cape, the all-cases notification rate for TB in 2013 was 601 new cases/100,000 population; 762 cases/100,000 population among children 0–4 years of age, 217 among those 5–9 years of age and 141 among children 10–14 years of age.4 The HIV prevalence among the general population in the Western Cape was 18.9% [95% confidence interval (CI) 16.4%–21.7%] in 2015.5
Management of specimens for this survey period was similar to that reported in the previous survey period.3 Samples were routinely processed using the Mycobacteria Growth Indicator Tube 960 system (Becton Dickinson, Sparks, MD). Confirmation of M. tuberculosis complex and genetic mutations conferring resistance to rifampicin (rpoB gene) and isoniazid (inhA and katG genes) was done on a single isolate from every child with a positive culture using a line-probe assay (LPA; GenoType MTBDRplus; Hain Lifescience, Nehren, Germany). As M. tuberculosis complex includes M. bovis BCG, identification for M. bovis BCG was done if clinically suspected by polymerase chain reaction (PCR) identifying the absence of the Region of Difference 1 and/or by rapid MPB64 commercial test kit, as Danish strain BCG was mainly used (Capilia TB test, Tauns Co. Ltd).3 All patients with only M. bovis BCG or non-tuberculosis mycobacteria detected were excluded from analysis. If rifampicin (RIF) resistance was detected, phenotypic DST to isoniazid (INH), amikacin (AMK; as marker of second-line injectable resistance) and ofloxacin (OFX; as marker of fluoroquinolone resistance) was completed using the indirect proportion method on Middlebrook 7H11 agar.6
Xpert MTB/RIF (Xpert; Cepheid, Sunnyvale, CA) was introduced in the National Health Laboratory Service laboratory at Tygerberg Hospital in August 2013. Initially it was only used on sputum specimens, but over time, a wider range of specimen types were tested by Xpert, including gastric aspirates and fine needle aspirates. Both mycobacterial culture and Xpert were done on concentrated specimens.
For the first 3 survey periods, cultures were done by the same method, but conventional phenotypic DST was undertaken for INH and RIF using the BACTEC™ 460 TB system from March 2003 to July 2008; thereafter LPA method was used. INH was tested at a concentration of 0.1 μg/ml and RIF at 2.0 μg/ml as previously described.7
Once bacteriologic confirmation of M. tuberculosis was obtained, clinical data and laboratory results were reviewed and data extracted using a standard tool.3 Chest radiographs were evaluated by a single reader using a standard approach. If the HIV status was unknown, HIV testing was routinely done. HIV-positive status was defined as HIV PCR positive in children <18 months or >6 weeks after discontinuing breast-feeding, and in all other children if 2 HIV tests, including an HIV ELISA, confirmed HIV infection.
The DST data of the current 4-year period of reporting was divided in two 2-year periods to be able to compare DST data over the total seven 2-year periods from 2003 to 2017. Numbers are presented as absolute with percentages, with 95% CI for new data from the last 2-year period. Chi-square test for trends and linear regression was used to assess the changes in proportions of the different resistance patterns over the time periods. P < 0.05 was considered statistically significant. Statistical analyses were conducted with Stata software, version 14.0 (Stata Statistical Software: Release 14.College Station, TX: StataCorp LP).
The study was approved by Stellenbosch University Health Research Ethics Committee and Tygerberg Hospital (2003/005).
Bacteriologic confirmation of mycobacterial infection was obtained in 710 children over the total 4-year period. Of these, 22 were confirmed as M. bovis BCG and 26 as non-tuberculosis mycobacteria; these were excluded. M. tuberculosis was confirmed in 662 children with median age 34 months (interquartile range 14–79 months). Of these, 587/662 (88.7%) were culture-confirmed; 75 (11.3%) children were positive by Xpert only. Of the 587 children with culture-confirmed TB, 310 (52.8%) children had known contact with infectious TB source cases: 229 of 381 (60.1%) children <5 years of age and 81 of 206 (39.3%) children 5–13 years (odds ratio 2.33, 95% CI 1.64–3.29; P < 0.001). Further, of the 381 children <5 years, 161/240 (67.1%) of those <2 years and 68/141 (48.2%) of those 2–4 years had known source cases (OR 2.18, 95% CI 1.43–3.35; P < 0.001). The demographic data and clinical features of the children with culture-confirmed TB are summarized in Table, Supplemental Digital Content 1, http://links.lww.com/INF/D800.
Cultures for M. tuberculosis were positive in 1150 specimens with a median (interquartile range) of 2 (1–2) positive isolates per child; 146 (24.9%) children had 3 or more culture-positive specimens. Xpert was positive in 333 (56.7%) children with positive cultures, but was not done in all children.
Of the 587 isolates tested by LPA (1 per child), 509 (86.7%) were susceptible to both INH and RIF. Any INH and/or RIF resistance was present in 78 (13.3%) children. INH resistance was present in 65 (11.1%) and RIF resistance in 60 (10.2%). Of all children, 18 (3.1%) had INH-resistant/RIF-susceptible TB; 13 (2.2% of total cases; 21.7% of RIF-resistant cases) had RIF-resistant/INH-susceptible and 47 (8.0%) had MDR-TB. INH resistance was conferred by katG gene mutation in 27/65 (42%), inhA promoter region mutation in 34/65 (52%), by both mutations in two (3%) and in an unknown mutation (phenotypic resistance) in 2 (3%) isolates. Of the 47 MDR-TB cases, 10 (21%) were pre-extensively drug-resistant (pre-XDR; 8 with OFX and 2 AMK resistance) and 7 (15%) were XDR-TB cases. Of 47 MDR-TB cases, 15 (32%) were OFX resistant. One of 13 (8%) children’s isolate with RIF-resistant/INH-susceptible TB also had resistance to OFX.
Of the 13 children with RIF-resistant/INH-susceptible TB by LPA, 6 were confirmed with phenotypic INH DST and, in a further 4, the source cases had confirmed RIF-resistant/INH susceptible TB. In the remaining 3 cases, no phenotypic INH DST was done and no source case was known. Only 2 of the 13 children were HIV-positive and 1 further child was HIV-exposed but uninfected.
Xpert was selectively introduced during the study period and thus not performed in all cases. Four hundred and four (61.0%) of the 662 children had an Xpert positive for M. tuberculosis complex, of which 356/404 (88.1%) showed susceptibility to RIF, 41 (10.1%) showed resistance to RIF and 7 (1.7%) had RIF indeterminate results. Of 329 cases on whom Xpert (done on processed specimens) and LPA (on cultured isolates) for RIF susceptibility were available, RIF DST results were concordant in 324 (98.5%). In the remaining 5 cases, 3 had indeterminate RIF DST results on Xpert but RIF susceptibility on LPA. One case was RIF-susceptible on LPA but RIF-resistant on Xpert (gene sequencing confirmed rpoB L511P mutation) and in 1 case who had 3 Xpert-tests done, the initial 2 showed susceptibility to RIF similar to the LPA DST for RIF and 1 Xpert on the second day of sampling showed RIF resistance, but with very low semi-quantitative reading.
Seventy five (11.3%) of 662 children with bacteriologically confirmed TB had M. tuberculosis identified only by Xpert: of these, 9/75 (12%) had no cultures done and at least 24 (32%) were already on anti-tuberculosis treatment for days to weeks; RIF DST results showed RIF resistance in 3 (4%), indeterminate result in 4 (5%) and RIF susceptibility in 68 (91%) cases. Of the 3 with RIF resistance, 1 had a known RIF-resistant/INH-susceptible source case; the second child’s only bacteriologically positive specimen (Xpert) was obtained 2 months following initiation of first-line treatment (source case had INH and RIF-susceptible TB) with uncertain adherence and clinical deterioration, and the third case was a newborn baby whose mother had confirmed INH and RIF-susceptible TB at birth. The first 2 were started on DR-TB treatment, while the newborn infant, who was clinically healthy, was started on preventive therapy.
HIV testing was done in 648/662 (97.9%) children and was positive in 81/648 (12.5%).
Twenty-one (3.2%) of 662 children died during hospital admission. Of these, 2 had DR-TB (1 pre-XDR; 1 XDR), both with stage 3 TB meningitis and 1 was also HIV-positive. Nineteen had drug-susceptible TB with deaths related to stage 3 TB meningitis in 6, miliary TB (excluding TB meningitis) in 4, abdominal TB (including 1 congenital TB) in 5, extensive pulmonary TB in 3 (including 1 congenital TB and 1 with lymphoma) and septicemia in 2 children; 5 of these children were HIV-positive and 5 had severe malnutrition (marasmus). Median time from admission to death was 8 days (range 0–44).
The DST and HIV results from the current two 2-year surveillance periods are compared with the 5 previous 2-year periods in Table 1 and Figure 1. RIF mono-resistance increased steadily with the test for trend statistically significant (χ2 = 7.050, P = 0.0079, slope = 0.003, z = 2.655), while INH mono-resistance decreased with test for trend statistically significant (χ2 = 6.422, P = 0.0113, slope = −0.0059, z = 2.534). Line graphs showing key trends in TB drug resistance over the 7 two-year periods (March 2003 to February 2017) of surveillance in age groups 0–4 years, the age most likely due to transmission, and 5–13 years are presented in Figures 2 and 3, respectively; complete data of drug resistance for these 2 age groups are supplied (Table, Supplemental Digital Content 2, http://links.lww.com/INF/D801).
In this total 4-year drug resistance surveillance period in children with confirmed TB from the Western Cape, the prevalence of DR-TB remains high. The lower prevalence of MDR-TB seen in the fifth 2-year period compared with the previous 4 periods was not continued;3 MDR-TB prevalence was high at 7.1% and 8.9% in the recent two 2-year periods, respectively, similar to the first 4 study periods.
There has been a significant decreasing trend in INH mono-resistance compared with the previous periods (Table 1). This may be due to fluctuation in INH mono-resistance, but it could also be because LPA DST for INH is missing some INH-resistant cases. LPA INH DST is less sensitive compared with phenotypic INH DST, as it only identifies inhA and katG gene mutations conferring INH resistance;8 phenotypic INH DST is not routinely done in South Africa from August 2008, except in isolates that show resistance to RIF and susceptibility to INH on LPA. In adults in South Africa, INH mono-resistance increased between the 2001–2002 (2.7%; 95% CI 2.2%–3.2%) and 2012–2014 (4.9%; 95% CI 4.1%–5.8%) DR-TB surveys; in the Western Cape, new INH mono-resistance cases were 6.9% (95% CI 5.1%–8.7%) in the 2012–2014 survey—higher compared with INH mono-resistance rates in children in our study in the same period.9 The true prevalence of INH mono-resistance therefore requires further investigation in our setting.
We note with concern a definite steady increase in RIF-resistant/INH-susceptible (RMR) TB cases in this ongoing surveillance in children. Although this is a low proportion of all culture-confirmed cases, it represents nearly a quarter of all RIF-resistant isolates. Previous studies have shown an association of RMR-TB with severe HIV disease, but only 3 of 13 children with RMR-TB in our current study period were HIV-positive.10,11 Furthermore, only 5 children with culture-confirmed RMR-TB in our study had known contact with infectious RMR-TB source cases. Only one of these children had previous exposure to RIF; therefore transmission of RMR strains remains the most likely explanation, which implies that RMR-TB is increasing in adults, as has been previously shown.11 The underlying reason(s) for this increase in RMR-TB still needs further study. Confirming RMR-TB has important implications for the treatment of RMR-TB in children; they are currently managed as MDR-TB, but if INH susceptibility could be phenotypically confirmed, adding INH to their regimen could lead to fewer toxic second-line drugs being added to the regimen. Confirming INH susceptibility in RIF-resistant cases is also important for providing preventive therapy to children in contact with such cases, as INH preventive therapy should still be effective.
Of the MDR-TB cases identified during this period, 32% were resistant to ofloxacin and therefore likely to levofloxacin, with an increasing trend (Table 1). This has important implications for the design of treatment regimens for MDR-TB, which currently rely heavily on a fluoroquinolone. Additionally, 2 randomized, placebo-controlled MDR-TB preventive trials with levofloxacin are ongoing in South Africa (in children <5 years of age—TB-CHAMP)12 and Viet Nam (mainly adults MDR-TB contacts, V-QUIN)13 and high rates of fluoroquinolone resistance will have important implications for using levofloxacin as a single drug regimen in DR-TB prevention.
Xpert was introduced during this study period and proved to be important in rapidly confirming the diagnosis of TB as well as RIF resistance. Xpert detected RIF resistance with high concordance compared with culture-based DST for RIF by LPA method. As shown in some cases in this study, Xpert may also be helpful in confirming TB disease and RIF susceptibility in children with paucibacillary TB already started on anti-tuberculosis treatment, in whom cultures are negative most probably due to critical reduction of live mycobacteria. However, care should be taken in interpreting these results, as Xpert results may remain positive for a long time into treatment, and even after treatment completion.14,15 Obtaining a history of previous TB episodes, and careful clinical evaluation for signs and symptoms of active TB is important to assess the likelihood of false-positive Xpert results in cases where cultures are negative. However, in children, especially young children, recurrent TB is less frequent, and may be less relevant than in adults. Low positive Xpert results, on the other hand, are common in children,16 and have been associated with false-positive detection of RIF resistance by Xpert due to PCR probe delays in some cases.17 The child in this cohort who had 2 RIF-susceptible samples by Xpert (and 2 cultured isolates—also RIF susceptible by LPA) and a single RIF-resistant sample by Xpert is a case in point: the RIF resistance reported on the third sample collected is likely to have been due to the very low semi-quantitative read-out. The same sample was culture positive and RIF-susceptible on LPA. The child was treated for drug-susceptible TB and was clinically well at the end of treatment. The new, more sensitive Xpert Ultra assay which has recently replaced the Xpert MTB/RIF in many settings including South Africa, automatically generates an indeterminate RIF-resistance result for the lowest “trace” positive for M. tuberculosis complex read-outs, and may therefore reduce false RIF-resistance detection.18
The significant decrease in HIV co-infection in culture-confirmed TB cases from a peak of 29% in the 2007–2009 period to 10.6% in the 2015–2017 period is most likely due to high uptake of an effective mother-to-child HIV prevention program in the Western Cape. However, opportunities remain for improved HIV prevention in children. Although the reasons for not identifying HIV-positive children before the TB diagnosis was not documented in this study, this has been studied before in our setting and earlier HIV diagnosis needs ongoing attention.19
Finally, it is interesting to note that known close, mainly household, contact to an infectious TB source case was much higher and age related (39%–67%) in our study than the 10%–30% suggested by a recent study challenging the dogma that most transmission occurs in the household TB.20
A limitation of the study is that not all children had samples tested by all 3 diagnostic modalities (smear, culture and Xpert), which restricts our ability to assess sensitivity of Xpert MTB/RIF. However, we can report that Xpert added at least 8.4% of confirmed diagnoses. Furthermore, in the last 2-year period (2015–2017) when most children had at least 1 Xpert done, in 73% of children the diagnosis could be rapidly confirmed by Xpert, before cultures became positive. We also cannot evaluate the impact of Xpert (or culture) on treatment decisions as the study was not designed for this.
Given the high prevalence of any DR-TB in children, the reduction in use of second-line injectable agents and the increasing use of new and repurposed drugs, there is an urgent need to expand phenotypic and molecular DST. It is insufficient to only have phenotypic DST in cases with RIF resistance, as was the case in this setting. Whole genome sequencing is becoming more readily available and affordable compared with multiple drug phenotypic DST; this could provide DST results much more rapidly and better guide drug regimens and clinical management in DR-TB cases.
In conclusion, resistance to INH and RIF in children with confirmed TB remains high in this setting. RIF-resistant, INH-susceptible cases are slowly but steadily increasing and reasons for this should be investigated. The high rate of OFX resistance among MDR-TB cases is of concern.
1. World Health Organization. Global tuberculosis report 2018. 2018Geneva: WHO;
2. Jenkins HE, Yuen CM. The burden of multidrug-resistant tuberculosis in children. Int J Tuberc Lung Dis. 2018;22:3–6.
3. Schaaf HS, Garcia-Prats AJ, du Preez K, et al. Surveillance provides insight into epidemiology and spectrum of culture-confirmed mycobacterial disease in children. Int J Tuberc Lung Dis. 2016;20:1249–1256.
4. Snow K, Hesseling AC, Naidoo P, et al. Tuberculosis in adolescents and young adults: epidemiology and treatment outcomes in the Western Cape. Int J Tuberc Lung Dis. 2017;21:651–657.
5. National Department of Health. 2015 National antenatal sentinel HIV & syphilis survey report. file:///C:/Users/hss/Downloads/2015%20national%20antenatal%20hiv%20prevalence%20survey_final_23oct17%20(1).pdf
. Accessed July 2, 2019.
6. Schaaf HS, Hesseling AC, Rautenbach C, et al. Trends in childhood drug-resistant tuberculosis in South Africa: a window into the wider epidemic? Int J Tuberc Lung Dis. 2014;18:770–773.
7. Seddon JA, Hesseling AC, Marais BJ, et al. The evolving epidemic of drug-resistant tuberculosis among children in Cape Town, South Africa. Int J Tuberc Lung Dis. 2012;16:928–933.
8. Nathavitharana RR, Cudahy PG, Schumacher SG, et al. Accuracy of line probe assays for the diagnosis of pulmonary and multidrug-resistant tuberculosis: a systematic review and meta-analysis. Eur Respir J. 2017;49:1601075.
9. National Institute for Communicable Diseases. South African Tuberculosis Drug Resistance Survey 2012–14. http://www.nicd.ac.za/assets/files/K-12750%20NICD%20National%20Survey%20Report_Dev_V11-LR.pdf
. Accessed September 21, 2019.
10. Dramowski A, Morsheimer MM, Jordaan AM, et al. Rifampicin-monoresistant Mycobacterium tuberculosis
disease among children in Cape Town, South Africa. Int J Tuberc Lung Dis. 2012;16:76–81.
11. Mukinda FK, Theron D, van der Spuy GD, et al. Rise in rifampicin-monoresistant tuberculosis in Western Cape, South Africa. Int J Tuberc Lung Dis. 2012;16:196–202.
12. Seddon JA, Garcia-Prats AJ, Purchase SE, et al. Levofloxacin versus placebo for the prevention of tuberculosis disease in child contacts of multidrug-resistant tuberculosis: study protocol for a phase III cluster randomised controlled trial (TB-CHAMP). Trials. 2018;19:693.
13. V-QUIN – ANZCTR – Registration. The V-QUIN MDR TRIAL: A randomized controlled trial of six months of daily levofloxacin for the prevention of tuberculosis among household contacts of patients with multi-drug resistant tuberculosis. https://anzctr.org.au/Trial/Registration/TrialReview.aspx?id=369817
. Accessed January 15, 2019.
14. Theron G, Venter R, Calligaro G, et al. Xpert MTB/RIF
results in patients with previous tuberculosis: can we distinguish true from false positive results? Clin Infect Dis. 2016;62:995–1001.
15. Theron G, Venter R, Smith L, et al. False-positive Xpert MTB/RIF
results in retested patients with previous tuberculosis: frequency, profile, and prospective clinical outcomes. J Clin Microbiol. 2018;56: e01696–17.
16. Bates M, O’Grady J, Maeurer M, et al. Assessment of the Xpert MTB/RIF
assay for diagnosis of tuberculosis with gastric lavage aspirates in children in sub-Saharan Africa: a prospective descriptive study. Lancet Infect Dis. 2013;13:36–42.
17. Ocheretina O, Byrt E, Mabou MM, et al. False-positive rifampin resistant results with Xpert MTB/RIF
version 4 assay in clinical samples with a low bacterial load. Diagn Microbiol Infect Dis. 2016;85:53–55.
18. Chakravorty S, Simmons AM, Rowneki M, et al. The new Xpert MTB/RIF
Ultra: improving detection of Mycobacterium tuberculosis
and resistance to rifampin in an assay suitable for point-of-care testing. mBio 2017;8:e00812–e00817.
19. Byamungu LN, du Preez K, Walters E, et al. Timing of HIV diagnosis in children with tuberculosis managed at a referral hospital in Cape Town, South Africa. Int J Tuberc Lung Dis. 2018;22:488–495.
20. Martinez L, Lo NC, Cords O, et al. Paediatric tuberculosis transmission outside the household: challenging historical paradigms to inform future public health strategies. Lancet Respir Med. 2019;7:544–552.