Introduction

Antituberculosis drug resistance is a major public health problem that threatens progress made in tuberculosis (TB) care and control worldwide. Globally, 3.7 % (2.1 %–5.2 %) of new cases and 20 % (13 %–26 %) of previously treated cases are estimated to have multidrug-resistant (MDR) TB [1]. The World Health Organization (WHO) estimates that there are 650,000 prevalent cases of MDR-TB globally [2], and since children (<15 years of age) constitute up to 20 % of the TB caseload in high-burden settings [3, 4], the number of children with drug-resistant (DR) TB is undoubtedly high. Data regarding this vulnerable population, however, are lacking; a recent systematic review of children with MDR-TB was able to include only eight studies from five countries [5••]. Children serve as a “sentinel” of TB transmission in the community, and drug resistance in this group mirrors the situation in the adult population in the region.

Major obstacles to understanding the epidemiology of pediatric TB in general and DR-TB in particular include the difficulty of confirming the diagnosis (needing multiple specimens other than sputum and a laboratory capable of performing culture), a higher proportion of smear- and culture-negative and extrapulmonary TB in young children, and the low priority given to this group by public health programs. However, the occurrence of DR-TB among children has been documented by several groups [6, 7•, 8••, 9–11]. In the Western Cape, repeat surveys among children, done in 1997–1998, 2001–2002, and again in 2005–2006, showed that resistance to isoniazid (INH) or rifampicin (RIF) increased from 6.9 % to 12.9 % to 15.1 % and multidrug resistance from 2.3 % to 5.6 % to 6.7 % [12, 13]. Drug resistance among children has been documented in both pulmonary and extrapulmonary disease [14].

When children have MDR-TB, it is usually “primary resistance”—that is, they are infected with strains transmitted from adults with MDR-TB—rather than secondary resistance acquired as a result of suboptimal therapy or nonadherence [13]. The concordance between the Mycobacterium tuberculosis strain infecting the child and the adult index case varies from 45 % to 80 % in different studies, suggesting that children are exposed to TB both within and outside the household [15].

Diagnosis

The diagnosis of pediatric MDR-TB is often delayed due to reliance on the adult case definition and the need for bacteriologic confirmation [16]. Systematic approaches to the diagnosis of children with suspected drug resistance and consensus case definitions have been proposed recently [17, 18]. A diagnosis of TB in children can be made on clinical and radiological grounds in the majority of cases, when bacteriological confirmation is not possible. Depending on the age of the child, site of disease, and available facilities, attempts can be made to obtain sputum, gastric aspirates, induced sputum, biologic fluid samples, nasopharyngeal aspirates, lymph node aspiration biopsy, or tissue biopsy [19–23]. With extensive sampling, the proportion of children with a confirmed diagnosis can be >50 % [24]. Invasive methods, such as bronchoalveolar lavage, bronchoscopic biopsy, or open lung biopsy may sometimes be required [25].

Diagnostic Assays

Culture can be performed using solid media, such as the egg-based Lowenstein–Jensen or the agar-based Middlebrook medium, where the cultures are examined after 3–4 weeks, instead of 4–6 weeks using the classic method. Liquid media systems such as the radioactive (Bactec) or nonradioactive (MGIT), allow detection of growth in 8–14 days. Table 1 shows the TB diagnostic tests in use recently endorsed by the WHO [26].

Table 1 Tuberculosis (TB) diagnostic tests in use, recently endorsed by WHO and in later stages of development
Full size table

Tuberculin skin testing, using purified protein derivative and chest radiography, is used as an adjunct to smear microscopy (and culture, if available); however, the former has poor sensitivity and specificity for active TB, and the latter is often not available at the point of primary patient care [26].

In a large, multicountry study in adults, Boehme et al. evaluated an automated tuberculosis assay (Xpert MTB/RIF) for the presence of Mycobacterium tuberculosis (MTB) and resistance to RIF. With a single test, this assay identified 98 % of patients with smear-positive and culture-positive TB (including more than 70 % of patients with smear-negative and culture-positive disease) and correctly identified 98 % of bacteria that were resistant to RIF [27•]. It has several advantages over conventional nucleic acid amplification tests, which have been licensed for nearly 20 years: simple to perform with minimal training, not prone to cross-contamination, requires minimal biosafety facilities, and has a high sensitivity in smear-negative TB (the last factor being particularly relevant in patients with HIV infection) [27•].The Xpert MTB/RIF assay has demonstrated sensitivity of 50 %–70 % in specimens like gastric aspirates and induced sputum [28••, 29].

Molecular line probe assays focused on rapid detection of RIF resistance alone or in combination with INH resistance are now widely used; examples are the INNO-LiPARif.TB kit (Innogenetics, Zwijndrecht, Belgium) [30], labeled for use on M. tuberculosis isolates grown on solid culture, and the Genotype MTBDR and Genotype MTBDRplus assays (Hain Lifescience, Germany) [31], labeled for use on isolates from solid and liquid culture, as well as directly on smear-positive pulmonary specimens. Both assays are complete, polymerase chain reaction (PCR)-based, hybridization assays simultaneously detecting M. tuberculosis complex and specific mutations in the rpoB gene conferring RIF resistance. The Genotype MTBDRplus assay also simultaneously detects specific mutations in the katG gene conferring high-level INH resistance, as well as those in the inhA conferring low-level resistance.

The molecular basis of resistance to INH and RIF (and some other drugs) is now understood (Table 2) [32]. Resistance to INH is due to mutations at one of two main sites, in either the katG or the inhA gene [33, 34]. Resistance to RIF is nearly always due to point mutations in the rpo gene in the beta subunit of DNA-dependent RNA polymerase [35]. These mutations are not directly connected, and so separate mutations are required for organisms to change from a drug-susceptible isolate to MDR-TB. However, genetic probes that detect drug resistance to RIF with >95 % accuracy is very suggestive of MDR-TB; <10 % of RIF resistance is monoresistant, and so RIF resistance is a marker for MDR-TB in >90 % of cases [36]. Whenever RIF and/or INH resistance is determined by a rapid molecular test, the results should be confirmed by phenotypic testing.

Table 2 [32] Genetic sites for drug resistance in tuberculosis
Full size table

There must be recognition, however, that there will be a group of children who need treatment for MDR-TB in whom bacteriological confirmation is either pending or not possible. The category of “probable” MDR-TB will allow providers to initiate timely care within programmatic guidelines in order to decrease the morbidity and mortality of MDR-TB in children, while at the same time ensuring that any potential therapeutic “chaos” does not ensue.

Children with signs and symptoms of active TB disease who, in addition, have the following risk factors should be considered as having “probable” MDR-TB and started on MDR-TB treatment, even in the absence of bacteriological confirmation [8••]:

  1. 1.

    Close contact with a known case of MDR-TB;

  2. 2.

    Close contact with a person who died whilst on TB treatment;

  3. 3.

    Close contact with a person who failed TB treatment;

  4. 4.

    Failure of a first -line regimen;

  5. 5.

    Previous treatment with second-line medications.

Treatment

The basic principles of treatment regimen design for children are the same as for adults with MDR-TB [37]. One major difference for children is that their treatment is often empiric and based on the drug susceptibility pattern of the source case, if available, or on past history of treatment. Depending on country guidelines, the regimen used is either individually constructed or a standardized one, such as the Category IV regimen recommended by WHO [18].

The basic principles are the following:

  • Use any first-line medication to which susceptibility is documented or likely (high-dose INH could be included routinely, unless high-level INH resistance or Kat-G mutation is documented).

  • Use of at least four second-line drugs to which the strain is likely to be sensitive; one of these agents should be an injectable, one should be a fluoroquinolone, and PZA should be continued.

  • All doses should be given using DOT (directly observed therapy) to ensure that patients adhere to treatment.

  • Treatment duration should be for 18–24 months, at least 12 months after the last positive culture/smear with minimal disease or 18 months with extensive (lung cavities or widespread parenchymal involvement) disease.

Table 3 shows the five groups of drugs recommended by WHO for use in treating DR-TB in children [5••]. The pharmacokinetics and toxicity of drugs in children differ considerably from those in adults. Almost every aspect of pharmacokinetics (absorption, distribution, metabolism, excretion) is subject to age-related change. Young children often require a higher mg/kg bodyweight dosage of a drug to achieve the same pharmacokinetic exposure as in adults. Current dosing recommendations are based on adult mg/kg doses [18].

Table 3 Drugs used to treat tuberculosis in children (5)
Full size table

For children, amikacin is usually given in preference to kanamycin, since it has a lower minimum inhibitory concentration and the available ampoule sizes are smaller, preventing wastage. Capreomycin is usually reserved for the treatment of extensively DR (XDR) TB. The fluoroquinolones have a central role in the management of MDR-TB in children. Resistance to early generation fluoroquinolones (ofloxacin) may not necessarily imply resistance to later generations (moxifloxacin or levofloxacin) [38]. Few studies have assessed the pharmacokinetics of fluoroquinolones in children; the available data are largely from studies in older children with cystic fibrosis [39].

The second-line drugs are rarely produced in pediatric formulations or appropriate tablet sizes, necessitating breaking, splitting, crushing, or grinding. Hence, dosing may be inaccurate, and subtherapeutic or toxic levels are possible. The taste of the medications is often unpalatable.

Adherence to treatment is a critical factor in the management of MDR-TB, and adverse events associated with second-line drugs could have a severe impact on adherence [40]. In general, children tolerate drugs better than do adults, and most side effects are mild and manageable with counseling and symptomatic drugs. The published information on treatment outcomes for children with MDR-TB suggests that when appropriately treated, outcomes are as good if not better than in adults [41•].

MDR-TB and HIV Coinfection

In settings with a high burden of TB and HIV, up to 40 % of children with MDR-TB are also HIV infected [42]. However, there are few reports of DR-TB/HIV cotreatment in pediatric patients [43–45]. The combination of MDR-TB and HIV can have serious psychological effects. Both conditions are stigmatized and are perceived to carry poor prognosis. The second- and third-line TB regimens demonstrate their own distinct cumulative toxicities with concomitant antiretroviral administration; the nephrotoxicity associated with tenofovir may be compounded by the antituberculous aminoglycosides, and the peripheral neurotoxicity induced by stavudine and didanosine and psychiatric disturbances associated with efavirenz may be exacerbated by the antituberculous agent cycloserine. Additionally, the pill burden and gastrointestinal distress associated with drug-susceptible TB regimens are even greater with MDR-TB and XDR-TB regimens [46, 47].

Studies have demonstrated that, even in a setting of high HIV prevalence, it is possible to achieve favorable outcomes among children treated for MDR-TB using early empiric treatment delivered through a comprehensive community-based program [16, 41•, 48–50]. Four pediatric XDR-TB patients with HIV coinfection were successfully cured with cotreatment in South Africa [43]. Another study in South Africa examined outcomes in 111 children with MDR-TB, including 43 children with HIV coinfection, most of whom initiated ART prior to or during MDR-TB treatment. In that report, 82 % of patients achieved favorable outcomes, and 5 of the 13 deaths occurred before confirmation of MDR-TB and initiation of appropriate treatment [45].

Supportive Care

In addition to TB drugs, guidelines recommend that children with TB should be given pyridoxine if they are HIV infected, malnourished, or breast fed or are being given terizidone, cycloserine, or high-dose INH [51, 52]. Most experts put all children being treated for DR-TB on multivitamin supplements. Nutritional and metabolic requirements should be assessed, because these children are commonly malnourished, and supplements should be provided when necessary [44, 45]. Physiotherapy and occupational therapy may be of benefit for children with respiratory and musculo-skeletal deficit. Social workers should assess home circumstances and support the caregiver to look after a child who may have complex medical needs and must take multiple medications.

New TB Drugs

There are six novel drugs in four new classes in clinical trials, including TMC207 (Bedaquiline), OPC-67683 (Delamanid), PA824, SQ109, and Oxazolidinones (PNU-100480 and AZD5847) [53].

Table 4 shows the overview of anti-TB drugs in the clinical pipeline [54]. These agents are anticipated to shorten and improve the treatment of drug-resistant, and possibly drug-susceptible, tuberculosis—used either separately or in novel combinations. A recent study from South Africa evaluated several novel combinations in an early bactericidal activity study, which measures decline in sputum colony counts per day among patients with sputum smear-positive pulmonary TB, and got encouraging results [55•].

Table 4 Overview of anti-TB drugs in the clinical pipeline [54]
Full size table

Conclusions and Future Directions

MDR-TB in children is often an underrecognized and neglected problem. Although accurate prevalence or incidence data are not available, wherever surveillance has been done, the rates have been found similar to those for adults in the region. Diagnosis should be presumptive, based upon a number of clinical and epidemiologic factors, in situations where bacteriologic confirmation is not available. While principles of treatment are similar to those for adults, lack of pediatric formulations and paucity of information on pharmacokinetics of second-line drugs in children make treatment challenging. Outcomes are good when appropriate therapy is initiated, even in the presence of HIV coinfection. Research is urgently required to establish optimal dosing schedules of second-line drugs, investigate shorter, more patient-friendly, fully oral regimens for treatment and prevention, and initiate dose-finding and safety studies of newer anti-TB molecules (e.g., Bedaquiline, PA 824, and Delamanid).