Published ahead of print on February 16, 2006, doi:10.1164/rccm.200511-1809SO
© 2006 American Thoracic Society doi: 10.1164/rccm.200511-1809SO
Childhood Pulmonary TuberculosisOld Wisdom and New ChallengesDesmond Tutu Tuberculosis Centre and Department of Pediatrics and Child Health, Faculty of Health Sciences, Stellenbosch University, Cape Town, South Africa; and Infectious Diseases Section, Department of Pediatrics, Baylor College of Medicine, Houston, Texas Correspondence and requests for reprints should be addressed to B.J. Marais, M.R.C.P. (Paed UK), F.C.P. (Paed SA), M.Med. (Paed), Department of Paediatrics and Child Health, Desmond Tutu Tuberculosis Centre, Faculty of Health Sciences, Stellenbosch University, P.O. Box 19063, Tygerberg 7505, South Africa. E-mail: bjmarais{at}sun.ac.za
Childhood tuberculosis is neglected in endemic areas with resource constraints, as children are considered to develop mild forms of disease and to contribute little to the maintenance of the tuberculosis epidemic. However, children contribute a significant proportion of the disease burden and suffer severe tuberculosis-related morbidity and mortality, particularly in endemic areas. This review provides an overview of well-documented concepts and principles, and demonstrates how this "old wisdom" applies to current and future challenges in the field of childhood tuberculosis; the aim was to articulate some of the most pressing issues, to provide a rational framework for discussion, and to stimulate thought and further scientific study. The prechemotherapy literature that described the natural history of disease in children identified three central concepts: (1) the need for accurate case definitions, (2) the importance of risk stratification, and (3) the diverse spectrum of disease pathology, which necessitates accurate disease classification. The relevance of these concepts and their application to pertinent issues such as the diagnosis of childhood tuberculosis are discussed. The concepts are also linked to the basic principles of antituberculosis treatment, providing a simplified approach to the diagnosis and treatment of childhood tuberculosis that is independent of resource constraints. The main challenges for future research are highlighted and in conclusion it is emphasized that the infrastructure provided by the directly observed therapy, short-course strategy, combined with well-targeted interventions, slightly improved resources, and greatly improved political commitment, may lead to a dramatic reduction in tuberculosis-related morbidity and mortality among children.
Key Words: childhood tuberculosis • classification • diagnosis • treatment
From a tuberculosis control point of view, the treatment of children is not considered a priority in endemic areas as they rarely transmit the disease and contribute little to the maintenance of the tuberculosis epidemic. However, children do contribute a substantial proportion of the global tuberculosis disease burden. Of the estimated 8.3 million new tuberculosis cases diagnosed in 2000, 884,019 (11%) were children (1), and their contribution to the disease burden in endemic areas is estimated to be even higher (2, 3). The United States experienced a resurgence of childhood tuberculosis cases during the 1990s, driven mainly by immigration from endemic areas and the breakdown of basic tuberculosis control practices, such as case reporting, contact tracing and screening, and the use of preventive chemotherapy (4). This resurgence has been brought under control, but outside the United States, children in endemic areas continue to suffer. In a report from a tuberculosis endemic area in Cape Town, South Africa, children less than 13 yr of age contributed 13.7% of the total disease burden, and experienced a tuberculosis incidence rate in excess of 400/100,000 per year (5). The common perception is that children rarely develop severe forms of tuberculosis. However, although this may be the case in nonendemic areas where diligent contact tracing is enforced, an autopsy study conducted in Zambia demonstrated that tuberculosis rivals acute pneumonia as a major cause of death from respiratory disease in children from endemic areas (6). Because of the difficulty of establishing an accurate diagnosis of childhood tuberculosis, the true extent of the tuberculosis-related morbidity and mortality suffered by children in endemic areas is rarely appreciated (6). Despite this huge disease burden, children's access to antituberculosis treatment in most endemic areas remains poor, as tuberculosis control programs focus predominantly on the treatment of sputum smear–positive adults (7). This review provides an overview of well-documented concepts and principles, and demonstrates how the application of this "old wisdom" may assist us to address current and future challenges in the field of childhood tuberculosis. We recognize that some suggestions are at odds with current guidelines, but our aim is to articulate the most pressing issues that hinder effective service delivery to children with tuberculosis, particularly in endemic areas.
The prechemotherapy literature documented the natural history of tuberculosis in children. Unfortunately, clinicians and researchers have limited access to these important studies as they were conducted before 1950 and are not included in modern electronic databanks. Since the discovery of safe and effective antituberculosis treatment, conducting studies on the natural history of disease became unethical and therefore these historic disease descriptions remain invaluable today. The prechemotherapy literature provides a strong body of evidence; multiple studies monitored large cohorts of children for prolonged periods of time and carefully documented the development of disease after primary infection with Mycobacterium tuberculosis. A critical review of the natural history of disease identified three central concepts that are important to consider when addressing current and/or future challenges in the field of childhood tuberculosis: (1) the need for accurate case definitions, (2) the importance of risk stratification, and (3) the diverse spectrum of disease pathology, which necessitates accurate disease classification (8).
Case Definition Occult dissemination frequently occurs during this early proliferative phase before cell-mediated immunity is fully activated (12, 14). Bacteriologic cultures collected at this time may be positive; Wallgren demonstrated in the 1930s that M. tuberculosis is sometimes recovered from recently infected children who are not diseased (11). Therefore, with active contact tracing and aggressive screening that includes the collection of mycobacterial cultures in asymptomatic children, it is not unexpected to find some positive cultures in recently infected children who are not diseased. This illustrates the overlap that exists between recent primary infection and case definitions of disease that rely exclusively on bacteriology. It is important to consider this overlap when case definitions are formulated for research purposes, particularly within the contact setting, although it is less relevant in everyday practice where there is no reason to obtain cultures from completely asymptomatic children. Uncomplicated hilar adenopathy remains the most common disease manifestation in children and is usually regarded as the hallmark of primary tuberculosis (15). However, the prechemotherapy literature documented transient hilar adenopathy in the majority (50–60%) of children after recent primary pulmonary infection, of whom only a few progressed to disease (8, 16). The natural history of disease illustrates that progression to disease is indicated by the onset of persistent, nonremitting symptoms, referred to as the breakpoint of clinical significance (8), whereas the complete absence of symptoms usually indicates good organism containment (8). By convention, asymptomatic hilar adenopathy is currently treated as active disease, although early experience (the U.S. Public Health trials of the 1950s and 1960s) with isoniazid alone demonstrated that one-drug therapy was sufficient in these cases. In terms of pathophysiology, microbiology, and natural history, asymptomatic hilar adenopathy is more indicative of recent primary infection than active disease (8, 13). This indicates that radiologic signs should be interpreted with caution in the absence of clinical data. The entity of so-called asymptomatic tuberculosis, where the case definition rests exclusively on radiographic criteria, is a case in point (12). High-resolution computed tomography is the most sensitive tool available to detect hilar adenopathy (17), as demonstrated by the fact that in children with recent M. tuberculosis infection and a normal chest radiograph, prominent intrathoracic nodes are frequently demonstrated by high-resolution computed tomography (18). Particular caution is required when interpreting the relevance of these radiologic signs in the absence of clinical data. It is important to point out that there is no role for high-resolution computed tomography in the evaluation of asymptomatic, immune-competent children exposed to M. tuberculosis (19). In reality, differences in patient selection may result in the use of different functional case definitions even though the definitions appear similar on paper. In nonendemic areas where active contact tracing is diligently enforced, more children with transient radiologic signs indicative of recent primary infection will be identified, and those with active disease will be diagnosed at an earlier, less advanced stage. Active contact tracing is rarely enforced in endemic areas and children usually present to health care facilities with suspicious symptoms and more advanced disease (20). Unlike asymptomatic contacts in whom visible radiologic signs probably indicate recent primary infection only, radiologic signs in symptomatic children indicate active disease. From a research perspective it is important to be aware of these differences, as inconsistent case definitions may confound the scientific interpretation of results. In everyday practice, distinguishing between the signs and symptoms of recent primary infection and active disease is less relevant in high-risk children (less than 3 yr of age and/or immune compromised) in whom infection frequently progresses to disease, sometimes with rapid disease progression.
Risk Stratification
Immune-competent children at least 3 yr of age are at low risk of progression to disease after primary infection. However, as the vast majority of children in endemic areas become infected after 2 to 3 yr of age, these low-risk children still contribute a significant percentage of the total disease burden (21). In addition, although these children are at low risk to progress to disease, latent infection with M. tuberculosis does pose the risk of future reactivation disease. In nonendemic areas, where transmission rates are low and eradicating the pool of latent infection is an achievable aim, the provision of preventive therapy to these low-risk children is warranted. In endemic areas, where the majority of disease in immune-competent adults results from ongoing transmission and not from reactivation (22, 23), the provision of preventive therapy after exposure and/or infection becomes less relevant. The major diagnostic challenge in this low-risk group is the differentiation between latent infection and active disease (8). Fortunately, active disease is accompanied by persistent, nonremitting symptoms and disease progression is slow, which provides a window of opportunity for symptom-based diagnosis (8).
Disease Diversity Within the Ghon focus, containment is usually successful, but disease progression may result from either poor or "excessive" containment (14, 24). Poor containment and unrestrained organism proliferation may cause progressive parenchymal damage, with ultimate breakdown of the Ghon focus (24). Infants (25, 26) and HIV-positive children (27, 28), who have poor cell-mediated immune responses, are most vulnerable to this type of cavitation. In contrast, immune-competent adolescents seem to mount an "excessive" (damaging) immune response in an attempt to contain the organism (24). The exact immune mechanisms underlying adult-type disease remain uncertain, but it is a striking observation that it emerges only as children enter into puberty (24, 29, 30). It is important to remember that children with adult-type disease are frequently sputum smear–positive and that they do contribute to disease transmission (30), particularly in congregate settings such as schools (31). Complications that arise from affected lymph nodes are most common in children less than 5 yr old, because of exuberant lymph node enlargement and small airway size (13). Extralumenal compression results when the airway is encircled by enlarged lymph nodes and associated inflammatory edema (13, 24). Intralumenal obstruction results from polyps or granulomatous tissue that develops secondary to inflammatory changes in the bronchial wall, or when caseous material is deposited into an airway after lymph node eruption (13, 24). Radiologic signs vary from segmental or lobar hyperinflation with partial obstruction and a check-valve effect (13), to segmental or lobar collapse with total obstruction and resorption of distal air (13). The pathology that results from the aspiration of caseous material is influenced by the dose and virulence of the bacilli aspirated. The pathology may range from transient parenchymal consolidation, resulting from a pure hypersensitivity response to dead bacilli and/or toxic products, to an expansile pneumonic process with progressive caseating pneumonia in the affected segment or lobe (13, 24). Expansile caseating pneumonia frequently leads to parenchymal destruction and cavity formation (24, 32). Thus, cavitary disease in children may result from three distinct pathologic processes: (1) poor containment at the site of organism deposition (very young and/or immune-compromised children); (2) aspiration of live bacilli when a diseased lymph node erupts into an airway, with destructive caseating pneumonia in the distal segment or lobe (children less than 5 yr of age); and (3) adult-type disease (mainly children greater than 10 yr of age) (24). The fact that immune-competent children 5 to 10 yr of age experience the lowest risk to progress to disease after primary infection with M. tuberculosis is an interesting immunologic phenomenon that is poorly understood. A better understanding of age-related differences in the immune response to M. tuberculosis may provide important insight into immune correlates of disease and protection (24). Disseminated (miliary) disease occurs predominantly in very young (immune-immature) and/or immune-compromised children, such as the HIV-infected or severely malnourished (27, 28, 33–35). These children have suboptimal cellular immune responses and demonstrate poor containment of the organism, both within the regional lymph nodes and at the multiple sites of occult dissemination. Tuberculous meningitis (TBM) is the most dangerous complication of disseminated (miliary) disease, occurring in 20 to 30% of cases (21, 33). From a public health perspective the challenge is to develop a pragmatic classification of these diverse disease manifestations with a clear focus on treatment relevance and optimal case management. A pragmatic public health–oriented disease classification is proposed in the treatment discussion, but issues relating to the diagnosis of childhood tuberculosis are discussed first.
The diagnosis of childhood tuberculosis presents a major challenge, as it is complicated by the absence of a practical "gold standard" (36, 37). Bacteriologic confirmation, the accepted gold standard, is of limited use in children because of the paucibacillary nature of their disease and poor bacteriologic yields. Sputum smear microscopy, often the only diagnostic test available in endemic areas, is positive in less than 10 to 15% of children with probable tuberculosis (38, 39). However, the yield is high in children with adult-type disease and sputum smear microscopy has definite diagnostic value in older children (more than 10 yr of age) (30). Culture yields are also low; reported yields in children with probable tuberculosis are less than 30 to 40% (38, 39). However, the bacteriologic yield depends on the specific intrathoracic disease manifestation. A study from South Africa reported a yield of 77% in children with advanced intrathoracic disease, whereas the yield in those with uncomplicated hilar adenopathy was only 35% (odds ratio, 6.3; 95% confidence interval, 3.2–12.8) (40). This indicates the potential value of sensitive bacteriology-based diagnostic approaches, particularly in endemic areas where children frequently present with advanced disease. In addition to poor bacteriologic yields, the collection of bacteriologic specimens is often problematic. Two or three fasting gastric aspirates collected on consecutive days, usually requiring hospital admission, are routinely performed in young children who cannot cough up phlegm. A retrospective study from California compared the bacteriologic yield achieved in gastric aspirates collected from hospitalized and nonhospitalized children (41). Although the yield in hospitalized children was higher (percentage of positive cultures, 48 vs. 37%), this difference was not statistically significant (41), which suggests that hospitalization may not be a prerequisite for the collection of a good gastric aspirate specimen. Bronchoalveolar lavage, using flexible fiberoptic bronchoscopy, has additive value when used in combination with gastric lavage, but this technique is highly specialized and is unavailable in most endemic areas (42). In a study from Peru, midmorning nasopharyngeal aspiration was compared with early morning gastric aspiration; gastric aspiration provided a slightly better yield than nasopharyngeal aspiration (38 vs. 30%), but the results were comparable (43). Nasopharyngeal aspiration is minimally invasive, does not require hospitalization or fasting, and can be performed any time of the day. A study from South Africa demonstrated that a single specimen, using hypertonic saline–induced sputum collection, may provide the same yield as three gastric aspirate specimens (39). However, the overall yield in this study remained poor (15% with one and 20% with three induced sputum specimens) and the technique has not been used outside the hospital setting. Additional studies are awaited to confirm the feasibility and diagnostic value of collecting induced sputum specimens in primary health care settings. Because of the difficulty in achieving bacteriologic confirmation, the diagnosis of childhood tuberculosis in nonendemic areas is usually based on (1) known contact with an adult index case (frequently within the household), (2) a positive tuberculin skin test (TST), and (3) suggestive signs on the chest radiograph. This triad provides a fairly accurate diagnosis in settings where exposure to M. tuberculosis is rare and well documented. However, its diagnostic accuracy is greatly reduced in endemic areas where exposure to M. tuberculosis is common and often undocumented, as exposure frequently occurs outside the household (44, 45). Despite reservations about the specificity of the TST response after bacillus Calmette-Guérin (BCG) vaccination and/or exposure to environmental mycobacteria, a positive TST reaction remains a fairly accurate measure of M. tuberculosis infection in immune-competent children. Current U.S. guidelines recommend the use of three different cutoff points to define a positive TST reaction (46). In endemic areas a positive TST is not uncommon in randomly selected healthy children (47), which limits its diagnostic value. Consequently, the diagnosis of tuberculosis in children from endemic areas depends mainly on clinical features and the subjective interpretation of the chest radiograph (48, 49). However, chest radiography is unavailable in many endemic areas and it has well-known limitations that may result in both under- and overdiagnosis of disease (4, 50, 51). Despite these limitations it provides an accurate diagnosis in the majority of symptomatic children with tuberculosis and the interpretation of the chest radiograph remains the most widely used diagnostic criterion in clinical practice (48, 52). Various clinical scoring systems have been developed. A critical review of these clinical scoring systems concluded that they are limited by a lack of standard symptom definitions and adequate validation (53). Developing standard symptom definitions through consensus of expert opinion is a difficult and subjective exercise; better guidance may be provided by objectively measuring the potential diagnostic value of different symptom definitions. A community-based survey demonstrated that the poorly defined symptoms traditionally associated with tuberculosis (such as a cough greater than 3 wk in duration) are frequently reported in a random selection of healthy children (54). Of 1,397 children without tuberculosis, 253 (26.4%) reported a cough during the preceding 3 mo and 66 (6.9%) reported a cough greater than 3 wk in duration (54). In addition, nearly 50% of children with visible hilar adenopathy on the chest radiograph (diagnosed with tuberculosis) reported no symptoms at all (54). These observations demonstrate the limited diagnostic value of poorly defined symptoms and the need for improved symptom and case definitions. In a follow-on study the use of well-defined symptoms with a persistent, nonremitting character showed greatly improved diagnostic accuracy (55). However, the potential diagnostic value offered by the use of these well-defined symptoms requires further validation in a prospective, community-based study that includes children from all relevant risk groups. It is expected that symptom-based diagnostic approaches would have less value in high-risk children (less than 3 yr of age and/or immune compromised) where disease progression may occur rapidly, emphasizing the need for preventive chemotherapy and early diagnosis of disease in this group (56). Other diagnostic modalities may hold promise, but have not shown convincing results to date (57). Serologic tests are currently unable to diagnose childhood tuberculosis with accuracy (58), and sputum-based polymerase chain reaction (PCR) tests have shown variable results and limited utility (59–62). Good results were reported with the use of a heminested PCR technique in Peru, but the study used uninfected children as the control group and therefore could not evaluate the ability of this novel PCR-based test to differentiate latent infection from active disease (63), which is important, as specific concerns have been raised regarding the specificity of PCR-based tests (61, 62). The diagnostic dilemma is even more pronounced in HIV-infected children. The specificity of symptom-based diagnostic approaches is reduced by the presence of chronic HIV-related symptoms, while the potential window for symptom-based diagnosis is limited by the rapidity with which disease progression may occur. Chest radiograph interpretation is complicated by HIV-related comorbidity and atypical disease presentation (51). These difficulties increase the potential diagnostic value of sensitive bacteriology-based approaches, to identify HIV-infected children with tuberculosis (64). However, as HIV-infected children are in the high-risk group the detection of M. tuberculosis infection is also highly relevant. Disease progression may occur soon (less than 12 mo) after primary or reinfection, or latent infection may be reactivated at a later date because of a decline in immunity. The traditional TST has poor sensitivity to detect M. tuberculosis infection in HIV-infected children; 50% or less of HIV-infected children with bacteriologically confirmed tuberculosis are TST positive, despite using an induration size of at least 5 mm (64, 65). This is a major limitation and a more reliable measure of infection will be valuable to identify HIV-infected children who may benefit from preventive chemotherapy; it may also provide supportive evidence to establish a diagnosis of active tuberculosis. Novel T-cell–based assays (T-SPOT.TB [Oxford Immunotec, Abingdon, UK] and QuantiFERON-TB Gold [Cellestis, Valencia, CA]) may improve the ability to diagnose M. tuberculosis infection, which has particular relevance in HIV-infected or severely malnourished children in whom the traditional TST performs poorly. Compared with the TST the enzyme-linked immunospot (T-SPOT.TB) assay demonstrated improved sensitivity in children, as reflected by better correlation with the degree of exposure after a school tuberculosis outbreak in the United Kingdom (66). It also demonstrated better performance in HIV-infected children treated for probable tuberculosis (67), and demonstrated good performance, independent of the CD4+ T-cell count, in HIV-infected adults (68). In addition to improved sensitivity, the use of M. tuberculosis–specific antigens such as ESAT-6 and CFP-10 provides superior specificity, as these tests are not influenced by BCG vaccination and only rarely by environmental mycobacteria (69). It is important to emphasize that in the absence of symptoms or radiologic signs indicative of disease, qualitative T-cell–based assays, like the TST, fail to make the crucial distinction between latent M. tuberculosis infection and active disease. The application of these new diagnostic tools is a priority for future research: in nonendemic areas mainly to assess its value in contact and immigrant screening, and in endemic areas mainly to assess its ability to detect M. tuberculosis infection in HIV-infected individuals.
We first discuss the basic principles of antituberculosis treatment and then apply it to specific clinical scenarios. Antituberculosis treatment aims to cure the individual patient and to prevent the emergence of drug-resistant organisms within the community by adhering to the following principles: (1) rapidly reduction of the organism load, (2) ensuring effective eradication of dormant and intermittently metabolizing (persistent) bacilli, and (3) achieving this with minimal adverse effects for the child (70–75). Bactericidal drugs ensure rapid reduction of the organism load, which improves clinical symptoms, limits disease progression, terminates transmission, and prevents the emergence of drug resistance. Sterilizing drugs ensure the effective eradication of dormant and intermittently metabolizing (persistent) bacilli, thus preventing disease relapse (70–75). These principles provide the rationale behind the intensive and continuation phases of current antituberculosis treatment regimens (72–76). Each of the first-line drugs makes a specific contribution during different periods of drug action (assuming complete drug susceptibility and the absence of significant immune compromise). Period 1 lasts 2 to 3 d (70, 76), during which time fast-growing extracellular bacilli, comprising the vast majority of the organism load, are killed, mainly by the excellent bactericidal activity of isoniazid (INH) (77, 78). Period 2 lasts 4 to 8 wk. Slower growing extracellular bacilli are killed (70), and the rate of killing is determined more by the physiological state of the bacilli and less by the bactericidal activity of the drug. During this period, the bactericidal activity of rifampin (RMP) is important and pyrazinamide (PZA) contributes by killing extracellular bacilli that persist in acidic areas of inflammation (70, 79). Period 3 lasts 4 to 6 mo. Persistent intracellular bacilli are eradicated mainly by RMP, although INH will continue to offer protection against the development of resistance and may assist with organism eradication, especially in fibrocaseous tissue with poor drug penetration (79). Host immunity plays an important role throughout, but is of particular importance to effect organism eradication and prevent disease relapse, as indicated by the high relapse rate in HIV-infected children (80). In the presence of a high mycobacterial load, any antituberculosis drug used in isolation is vulnerable to resistance from random, naturally occurring mutants (72–76). The number of drug-resistant mutants is proportional to the size of the mycobacterial population. Random resistance to multiple drugs is extremely rare and therefore the use of multiple drugs in combination, during the intensive phase of treatment, drastically reduces the risk of treatment failure despite a high organism load (76). The risk of random drug resistance is virtually eliminated once the intensive phase of treatment is successfully completed and the bulk of the organism load has been eliminated. Good adherence is essential to effect cure and to prevent the emergence of drug resistance, as several cycles of mycobacterial killing (when drugs are taken) and regrowth (when drugs are not taken) favors selection of drug-resistant mutants (81). These mutants may accumulate additional random mutations resulting in the emergence of multiple drug resistance (81). Practical, operational issues are extremely important for effective public health intervention. Operational issues include access to early and accurate diagnosis, the uninterrupted provision of quality-assured drugs and appropriate treatment regimens, as well as the establishment of systems to ensure good treatment adherence. Fixed-dose combinations should be used whenever possible to reduce the risk of drug resistance and to improve simplicity and adherence, but quality assurance is essential to ensure optimal bioavailability of all the constituent drugs (82, 83). With proper implementation, the World Health Organization's directly observed therapy, short-course (DOTS) strategy addresses most of the important operational issues. However, the predominant emphasis of the DOTS strategy on sputum smear–positive disease excludes the vast majority of children. There is a desperate need to improve service delivery to children with tuberculosis, particularly in endemic areas with limited resources (7).
Preventive Chemotherapy In reality, most endemic areas do not have the capacity to follow current World Health Organization guidelines regarding the use of preventive chemotherapy in children, which advise active tracing and screening of all children less than 5 yr old in household contact with a sputum smear–positive adult source case. This results mainly from the huge burden of adult tuberculosis and resource constraints that limit the ability to perform TST and chest X-ray screening tests. Because the TST and chest X-ray are regarded as prerequisite screening tests, screening of exposed children and the provision of preventive chemotherapy are not even attempted in most resource-constrained areas. Access to preventive chemotherapy in these settings may be improved by employing symptom-based screening, although the benefits and risks of such a simplified approach require further evaluation. A study from an endemic area indicated that symptom-based screening may identify those children who require further investigation to exclude active tuberculosis (20), thus allowing asymptomatic household contacts, especially those who are at high risk to progress to disease, immediate access to preventive therapy despite the inability to perform TST and chest X-ray–based screening. Another consideration is that in some endemic areas the majority of disease transmission, particularly in children greater than 2 to 3 yr of age, occurs outside the household (44, 45). In endemic areas, narrowing the focus of contact tracing to those children who are at highest risk to progress to disease after exposure or infection (less than 3 yr of age and/or immune compromised) will decrease the burden placed on already overstretched health care systems, while still ensuring access to preventive chemotherapy for the children who need it most (20, 84). In older (greater than 3 yr of age), immune-competent children the risk of tuberculosis after exposure is low and disease progression is usually indicated by the presence of persistent, slowly progressive symptoms; therefore passive case finding together with adequate diagnostic vigilance seems appropriate in this low-risk group. In nonendemic areas where resources permit and where the risk of future reinfection is low, it seems warranted to extend preventive chemotherapy to low-risk children as well, to eliminate the reservoir of latent infection within the community. INH monotherapy for 6 to 9 mo is the best-studied chemoprophylactic regimen (85–87), and it reduces the tuberculosis risk in exposed children by at least two-thirds; probably by more than 90% with good adherence (87). However, poor adherence is a major concern, particularly in endemic areas (84, 88–92). In real life the effectiveness of a preventive chemotherapy regimen is determined first by its efficacy and second by adherence to the prescribed regimen. Because of documented poor adherence to 6–9 mo of unsupervised INH monotherapy, consideration should be given to alternative preventive strategies with comparable efficacy but with improved adherence. Theoretically the addition of RMP has important advantages; RMP has strong sterilizing activity to eradicate latent bacilli and its addition will shorten the duration of treatment required (76, 78). It will also improve efficacy in settings where INH monoresistance is prevalent (93). The use of a 3-mo INH and RMP regimen for preventive chemotherapy is well established and trials have shown equivalence to 6 to 9 mo of INH alone, although the evidence is not as comprehensive as that for INH monotherapy (92, 94–96). PZA is another important sterilizing drug (76, 78), and in theory the combination of RMP and PZA represents the treatment of choice for latent infection (97). This combination has proven efficacy in animal models (98, 99), but adverse reactions in adults have limited the initial enthusiasm (100). However, these adverse reactions have not been observed in children (101), in whom the three-drug combination of INH, RMP, and PZA is generally well tolerated. Adherence may be improved by shortening the duration of treatment, but consideration may also be given to the provision of supervised preventive therapy. Creative approaches will be required to achieve this, particularly in places where health care services are already overburdened. With curative treatment, intermittent (two or three times weekly) therapy during the continuation phase is as effective as daily therapy to achieve organism eradication, once the organism load has been sufficiently reduced (102, 103). The same principle would apply to the treatment of latent infection, where the organism load is low. Targeting high-risk children for short-course, supervised intermittent preventive therapy seems achievable, but defining optimal preventive therapy regimens remains a fertile and important area for future research. Vaccination with BCG is the most widely used preventive strategy, although its efficacy remains controversial and studies have shown highly variable protection in different settings (104). Many factors may contribute to this variable protection: variations in strain-specific immunogenicity, timing and technique of vaccine administration, genetic factors, the presence or absence of environmental mycobacteria, and the effect of multiple reinfection events as may occur in highly endemic areas (105, 106). It is generally accepted that BCG vaccination offers significant protection against disseminated (miliary) disease in young children (less than 2 yr of age), but that it offers little or no protection against adult-type tuberculosis (106, 107). However, reports have documented significant protection against the development of adult-type tuberculosis when BCG was administered to TST-negative adolescents in locations with a low prevalence of environmental mycobacterial exposure (108, 109). In addition, a report from Turkey indicated that contrary to the prevailing theory, BCG may also protect against M. tuberculosis infection (as registered by a positive enzyme-linked immunospot result) (110). An even more controversial area is the risk versus benefit that BCG provides to HIV-infected children. There is a definite risk for HIV-infected infants to develop severe forms of BCG disease after neonatal BCG vaccination (111), but it remains poorly quantified. As the risk:benefit ratio has not been determined, the World Health Organization still advises BCG vaccination of asymptomatic HIV-exposed infants in tuberculosis endemic areas (112). Establishing the risk:benefit ratio of BCG vaccination in HIV-infected infants and the development of novel vaccines with improved efficacy and safety remain major research challenges.
Curative Treatment From a public health perspective the challenge is to develop a pragmatic classification of childhood tuberculosis that incorporates the diverse spectrum of disease, but focuses primarily on treatment relevance. The main variables that influence the success of chemotherapy identify three groups of children with tuberculosis: (1) those with sputum smear–negative disease, (2) those with sputum smear–positive (often cavitary) disease, and (3) those with disseminated (miliary) disease. Table 2 reflects current treatment guidelines for these three groups; it also includes new regimens to consider on the basis of established treatment principles.
A flow diagram has been developed to guide individual patient classification and management (Figure 1). It is based on answering five simple questions: (1) Is the child exposed to or infected with M. tuberculosis? (2) Does the child have active tuberculosis? (3) If the child is exposed or infected, but does not have active tuberculosis, is preventive chemotherapy indicated? (4) If the child has active tuberculosis, what is the appropriate treatment regimen? (5) Are there any special circumstances such as HIV infection, retreatment, or exposure to a drug-resistant source case to consider? The underlying rationale is universally applicable irrespective of diagnostic or resource constraints; although areas with access to advanced technology may achieve improved levels of diagnostic certainty.
Sputum smear–negative disease is usually paucibacillary and therefore the risk of acquired drug resistance is low. Drug penetration into the anatomic sites involved is good and the success of three drugs (INH, RMP, and PZA) during the 2-mo intensive phase, and of two drugs (INH and RMP) during the 4-mo continuation phase, is well established (113, 114). In the presence of extensive radiographic disease with or without cavitation, and/or suspicion of INH resistance, the use of ethambutol (EMB) in addition to the three drugs during the intensive phase should be contemplated. After completion of the intensive phase, successful organism eradication may be achieved with intermittent (two or three times weekly) therapy during the continuation phase (102, 103). The efficacy of shorter treatment durations for HIV-uninfected immune-competent children with sputum smear–negative disease requires further evaluation, as a 4-mo regimen of INH and RMP may be an acceptable therapy for some adults with sputum smear– and culture-negative tuberculosis (115, 116). Sputum smear–positive disease (often cavitary) implies a high organism load and an increased risk for random drug resistance against individual drugs. Selecting drug-resistant mutants is a particular concern where INH monoresistance is prevalent, as this increases the likelihood of selecting multidrug-resistant (MDR) organisms. The use of four drugs (INH, RMP, PZA, and EMB) during the 2-mo intensive phase should reduce this risk (117). Once the organism load is sufficiently reduced, intermittent (two or three times weekly) therapy with INH and RMP during the 4-mo continuation phase is sufficient to ensure organism eradication (102, 103). However, caution should be exercised when initial treatment response has not been optimal and in HIV-infected patients. The use of long-acting rifamycins together with INH is discouraged (118, 119). Disseminated (miliary) disease is frequently associated with CNS involvement (33, 120). It is therefore essential to consider the cerebrospinal fluid (CSF) penetration of drugs used in the treatment of disseminated (miliary) disease. INH and PZA penetrate the CSF well (116, 121). RMP and streptomycin penetrate the CSF poorly, but may achieve therapeutic levels in the presence of meningeal inflammation (116, 121). The value of streptomycin is limited by poor CSF penetration and intramuscular administration (116, 121). EMB hardly penetrates the CSF, even in the presence of meningeal inflammation, and has no demonstrated efficacy in the treatment of TBM (116, 121). Ethionamide shows good CSF penetration and has been used successfully as a fourth drug in the treatment of TBM (121–123). The fact that RMP penetrates the CSF poorly in the absence of meningeal inflammation reduces its sterilization value and may warrant the inclusion of PZA during the continuation phase, to assist with CNS sterilization (123). Several reports have illustrated the efficacy of short-course regimens in the treatment of TBM, but the risk of CNS relapse is rarely reported. In two of these studies a relapse was documented despite the completion of 6 mo of treatment with INH and RMP with an initial 2 mo of PZA (124, 125). Therefore, it seems prudent to include a fourth drug with good CNS penetration (such as ethionamide) for the treatment of disseminated (miliary) disease, at least during the intensive phase, and to consider PZA for the full 6 mo of treatment to reduce the risk of CNS relapse (123). CNS relapse is rare in the United States, where PZA is routinely discontinued after 2 mo, but the total treatment duration is 9 to 12 mo. Current fixed-dose combination tablets provide 4 to 6 mg of INH per kilogram. This dose may be suboptimal, particularly in settings where the majority of the bacterial population rapidly acetylates INH (126, 127). In addition, the serum level achieved with a similar dose of INH per kilogram is lower in children than in adults, increasing the risk for suboptimal dosing in children (126, 127). The majority of new INH resistance encountered in endemic areas is of an intermediate or low level, which underscores the importance of optimal INH dosing (126). A standard INH dose of 10 mg/kg seems appropriate in children, as even doses up to 20 mg/kg are well tolerated (127, 128); children are less susceptible to the toxic effects of INH than are adults. In general, adverse events are less common in children than in adults. The most severe adverse event is the development of hepatotoxicity, which can be caused by INH, RMP, PZA, or ethionamide. An elevation of liver enzymes (less than five times normal values) is not an indication to stop treatment, but the occurrence of liver tenderness, hepatomegaly, or jaundice should prompt the immediate stopping of all potentially hepatotoxic drugs. Jaundice is often preceded by a period of days or weeks of malaise and nausea. Hepatic reactions usually occur in the first weeks of therapy, but may happen at any time during the treatment period (129). Drug-related hepatic toxicity is usually caused by a single drug, but rarely a combination of drugs, which individually cause no problem, may cause hepatic toxicity (130). Children should be screened for other causes of hepatitis, as in many cases the antituberculosis drugs are not the cause of liver function derangement (131). In South Africa, hepatitis A infection is frequently responsible for non–drug-related liver function derangement in children receiving antituberculosis treatment (131, 132). Potentially hepatotoxic drugs should be reintroduced only after liver functions have normalized. Nonhepatotoxic drugs should be used in the interim and expert opinion should be sought. Ethambutol is usually not advised in children less than 7 yr of age as visual acuity cannot be evaluated; however, its use may be warranted in children with hepatotoxicity, cavitary disease, or resistance to first-line drugs; it seems safe at recommended dosages (116). Ethionamide frequently causes vomiting, but this can usually be overcome by dividing the daily dose and by a slow increase up to the full dose during the first week or two of therapy. Recommended dosages for the various first- and second-line drugs are reflected in Table 3.
Radiographic disease resolution may take many months despite significant symptomatic improvement; persistent radiographic signs are not an indication to change treatment if there is clinical improvement. Paradoxical exacerbation of symptoms or signs may also occur after antituberculosis therapy is initiated. This results from immune reconstitution with increased inflammation (particularly surrounding diseased lymph nodes or tuberculomas) that may follow nutritional rehabilitation (8), and/or antiretroviral therapy (133, 134). The release of bacterial toxins after successful antituberculosis treatment may also contribute. Treatment should be continued unaltered, although the temporary addition of corticosteroids may be considered. Such adjunctive therapy may be helpful in a number of disease manifestations where the host inflammatory response contributes to disease pathology (135), such as CNS involvement (TBM or tuberculoma) (136), severe lymph node compression of the airways, and pericardial effusion (137). There has been insufficient evidence to demonstrate whether steroids are effective in tuberculous pleural effusion (138).
Retreatment
HIV Infection When initiating treatment (curative treatment or RMP-containing preventive therapy) in HIV-infected children already receiving HAART or for whom HAART is contemplated, it should be appreciated that the rifamycins, especially RMP, and some of the nonnucleoside reverse transcriptase inhibitors and/or protease inhibitors may cause significant drug interactions. HIV-infected children may also develop particularly pronounced paradoxical reactions after the institution of HAART, because of immune reconstitution inflammatory syndrome. Recommendations on optimal drug combinations are frequently revised. The most recent recommendations can be obtained from the Centers for Disease Control and Prevention website, http://www.cdc.gov/nchstp/tb/.
Drug Resistance As the majority of INH monoresistance is of a low or intermediate level (143), the use of INH in a high dose (15–20 mg/kg) should still be considered when INH monoresistance is known or suspected (144). If INH monoresistance is expected before treatment is initiated, then the use of high-dose INH with the addition of EMB as a fourth drug will probably suffice. If resistance to INH is discovered after treatment has been started, then the addition of at least two new drugs should be considered. A rational guide for preventive chemotherapy would be to use at least two drugs to which the organism is sensitive, for at least 6 mo. Monoresistance to RMP is rare and it is frequently used as a marker of MDR disease. The term "MDR tuberculosis" implies resistance to both INH and RMP, with or without resistance to other antituberculosis drugs. Treatment can be difficult and should be discussed with an expert. Basic principles are as follows: to treat the child according to the drug susceptibility pattern of the likely source case if an isolate from the child is not available, give preferably three or more drugs to which the isolate is susceptible and/or naive (never add one drug to a failing regimen), use only daily directly observed therapy, and schedule regular follow-up visits to monitor progress and adverse events; treatment should continue for at least 12 mo after the first negative culture. With correct dosing, few adverse events are seen in children, although second-line drugs are generally more toxic.
Reducing the Burden of Childhood Tuberculosis
In our opinion, the following are the 10 most important issues that require additional research.
We aimed to provide a rational framework to stimulate critical thinking, discussion, and further scientific study, as we believe that better-targeted interventions, slightly improved resources (e.g., better availability of chest radiography and child-friendly drug formulations), and greatly improved political commitment may drastically reduce the tuberculosis-related morbidity and mortality suffered by children in endemic areas.
DOI: 10.1164/rccm.200511-1809SO Conflict of Interest Statement: B.J.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. R.P.G. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. H.S.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. N.B. received £500 in 2003 for speaking at a conference where one session was sponsored by GlaxoSmithKline. She also received £111,150 in 2000 and £150,000 per year in 2001–2003 from GlaxoSmithKline Action TB Program as a research grant for developing and maintaining an epidemiologic field site and for doing a study aimed at identifying surrogate markers for response to treatment in patients with TB. P.R.D. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. J.R.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form November 25, 2005; accepted in final form February 13, 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
CONTENTS