 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
,||,¶
* Department of Pediatrics, Infectious Disease Section, Baylor College of Medicine and Texas Children's Hospital, Houston, Texas
Pathology
¶ Medicine, Baylor College of Medicine, Houston, Texas
Department of Pediatrics, University of Texas-Houston Medical School, Houston, Texas
|| Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Hamilton, Montana
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Methods. A retrospective review of a prospective population-based active surveillance and molecular epidemiology project was conducted in private and public pediatric clinics within Houston and Harris County, Texas. The study population consisted of patients who had pediatric TB diagnosed from October 1, 1995, through September 30, 2000. Cases and potential source cases (PSC) were interviewed using a standardized questionnaire. Available Mycobacterium tuberculosis isolates from cases and PSCs were characterized and compared by IS6110 restriction fragment length polymorphism, spoligotyping, and genetic group assignment. Clinical characteristics were described, and molecular characterizations were compared. Data were analyzed by using EpiInfo 6.02b and SAS 8.2.
Results. A total of 220 (92%) of 238 pediatric TB cases were included. Epidemiologic and clinical findings were consistent with previous studies. Molecular profiles from 3 cases did not match the profile of PSC. Four previously unknown PSCs were identified using molecular techniques. Fifty-one (71.8%) of 71 isolates matched at least 1 other Houston Tuberculosis Initiative TB database isolate and were grouped into 33 molecular clusters. Cases were more likely to be clustered when the patients were younger than 5 years, identified a source case, or were US born.
Conclusions. Traditional contact tracing may not always be accurate, and molecular characterization can lead to identification of previously unrecognized source cases. Recent transmission plays a significant role in the transmission of TB to children as evident by the high degree of clustering found in our study population.
Key Words: epidemiology • infectious disease • pediatric • tuberculosis
Abbreviations: TB, tuberculosis • MTB, Mycobacterium tuberculosis • TCR, TB Control Registry • HTI, Houston Tuberculosis Initiative • TST, tuberculin skin test • PSC, potential source case • RFLP, restriction fragment length polymorphism • OR, odds ratio • CI, confidence interval • CSF, cerebrospinal fluid
Tuberculosis (TB) continues to be a significant health problem in the United States. Although the total number of cases has declined 44.2% from 1992 to 2003, the population that is 
24 years of age accounts for 16% of reported cases in the United States. Houston, ranked fourth among US metropolitan areas in total number of TB cases in 2002, mirrors national trends with 63 (13.8%) of 456 cases occurring in patients who are 24 years.
In contrast to adults, the diagnosis of pediatric TB is usually established on clinical and epidemiologic grounds as available diagnostic techniques have a low yield in this population.1–3 In addition, children are more likely to develop disease from a recent infection (ie, <1 year). Thus, pediatric TB often reflects recent TB transmission within a community and consequently provides useful markers for monitoring and directing TB control programs.4
Conventional contact tracing and molecular analysis are 2 methods that are used to monitor Mycobacterium tuberculosis (MTB) transmission. Characterizing and comparing TB isolates through molecular analysis can identify contacts that are missed by traditional techniques and can estimate the proportion of TB that is attributable to recent transmission, otherwise known as clustering.5–12 The aim of this study was to describe the clinical characteristics of pediatric TB in Houston, Texas; to verify source case information for pediatric TB patients using molecular techniques; and to determine the degree of clustering among pediatric TB patients.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Study Population
From October 1995 through September 2000, the City of Houston TB Control Registry (TCR) identified pediatric TB patients and referred them to the Houston Tuberculosis Initiative (HTI) for enrollment in the study. The HTI is an ongoing population-based active TB surveillance and molecular epidemiology project in Harris County, Texas. Patients were excluded from enrollment when they (1) could not be located, (2) refused the interview, (3) had lived in Harris County for <3 months before TB diagnosis, or (4) had incomplete records. The majority of patients were clinically evaluated by 1 of 2 authors (K.C.S. or J.R.S.).
Case Definitions
We defined a case as a patient who was aged 0 to 18 years and identified by TCR during the period of October 1995 through September 2000 and enrolled by the HTI with clinically or culture-confirmed TB. We defined cases as clinically confirmed when 2 of the following 3 criteria were met: (1) the patient's clinical course was consistent with TB, (2) the tuberculin skin test (TST) was positive (>5 mm of induration), or (3) the patient clinically improved with treatment of 
2 anti-TB drugs. We defined a case as culture confirmed when MTB was isolated from the patient's clinical specimen.13 Cases were also classified as pulmonary, extrapulmonary, or pulmonary-extrapulmonary TB on the basis of clinical presentation and/or laboratory results2:
1 of the following specimens: sputum, gastric aspirate, bronchus, bronchial fluid, or lung. A potential source case (PSC) was defined as an adult who had pulmonary TB and was identified by public health and/or HTI interviews as a close contact to a case patient. A close contact was defined as anyone who had close, regular, or prolonged contact with a case patient while he or she was infectious.14
Data Collection
After informed consent was obtained, case patients or their parents were interviewed using a standardized questionnaire. Evaluated risk factors and patient characteristics included demographics, social contacts, economic status, travel history, living and social environment, medical history including TST status, family history, and exposure to TB. The same questionnaire was administered to PSCs. Using an interactive network of laboratory personnel, pediatric TB specialists, TCR workers, infection control officers, and HTI personnel, inpatient and outpatient records were reviewed as well as public health data to create a comprehensive medical history for each case patient and PSC.
Molecular Characterization
MTB isolates from case patients and PSCs were obtained, and a molecular profile for each isolate was created using 3 techniques; IS6110 restriction fragment length polymorphism (RFLP), spoligotyping, and genetic grouping. After extraction of mycobacterial chromosomal DNA from isolates, IS6110 RFLP was performed using an internationally standardized protocol.15 Each IS6110 profile was analyzed using BioImage Whole Band Analysis Program, version 3.2 (Ann Arbor, MI) and archived. Because IS6110 RFLP cannot reliably differentiate isolates with 4 or fewer IS6110 copies, spacer oligonucleotide typing (spoligotyping) was used to analyze MTB isolates further.16–20 A commercially available kit was used for spoligotyping in accordance with the instructions supplied by the manufacturer (Bioscience BV, Maarssen, Netherlands). On the basis of the presence of nucleotide polymorphisms in codon 463 of the katG gene encoding catalase-peroxidase and codon 95 of the gyrA gene encoding the A subunit of DNA gyrase, isolates were analyzed to 1 of 3 major genetic groups.21 When isolates had the same IS6110 profile, spoligotype, and major genetic group, they were considered clonally related and grouped into the same cluster. In addition, we compared profiles to the HTI TB isolate database. Through 2002, the HTI had characterized 4342 isolates representing 3518 patients; 2166 (61.6%) isolates were grouped into 242 clusters, and 1352 (38.4%) were unique.
Data Analysis
Data from questionnaires were transcribed onto a stand-alone computer using EpiInfo 6.02b (Atlanta, GA) and Access (Microsoft, Redmond, WA). Data analysis was performed using SAS 8.2 (SAS Institute Inc, Cary, NC). 
2 analysis with odds ratio (OR) and 95% confidence interval (CI), Fisher's exact test when the value of stratified age data were <5, and t test were used for analysis of appropriate covariates.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The mean age for case patients was 6.8 ± 6.1 years, and the median age was 4 years. The proportion of enrolled male pediatric patients was slightly higher than that of female patients (Table 1). Although there were no statistical differences between the age distribution of male and female patients in this study, patients who were aged 15 to 18 and presented with TB were more likely to be female. The majority of case patients were ethnic minorities, with more than half being Hispanic. In addition, most case patients were born in the United States (n = 185). Additional investigation of the US-born pediatric TB case patients revealed that their parents were foreign born (n = 56), US born (n = 36), or unknown (n = 93). The birth countries for foreign-born case patients were Mexico (n = 19), Latin American countries (n = 9), Vietnam (n = 4), Cameroon (n = 1), Nigeria (n = 1), India (n = 1), and Columbia (n = 1). Eighty-five percent of US-born and foreign-born case patients had lived in Houston for >1 year at the time of diagnosis.
|
For all enrollees, only 154 (70%) cultured cases were submitted for mycobacteriology evaluation, 78 (51%) of which had MTB isolated. The majority of cultures from which MTB was isolated were pulmonary sources, including sputum (n = 25) and gastric aspirate (n = 24); however, extrapulmonary sources were identified as well, including lymph node (n = 6), CSF (n = 2), pleural fluid (n = 2), bone (n = 1), blood (n = 1), peritoneal fluid (n = 1), joint (n = 1), gastric lymph node (n = 1), and omentum biopsy (n = 1). In several patients, MTB was isolated from >1 site, including sputum/gastric aspirate (n = 3), tissue biopsy/lung (n = 2), bronchial wash/sputum (n = 1), CSF/gastric aspirate (n = 4), CSF/gastric aspirate/urine (n = 1), and ear/sputum (n = 1). Of the 30 patients who were MTB culture positive from sputum specimens, 15 (50%) were acid-fast bacilli smear positive. The mean age of enrollees who submitted gastric aspirate–positive MTB cultures was 2.1 ± 4.0 years and for enrollees who submitted sputum-positive MTB cultures was 13.5 ± 6.1 years. Only 7(9%) of the MTB isolates were resistant to at least 1 antimycobacterial drug, and multidrug-resistant TB was identified in 1 pediatric case.
On the basis of clinical and laboratory data, the site of disease for the majority of cases was extrapulmonary (n = 111) followed by pulmonary (n = 81) and pulmonary-extrapulmonary (n = 28). Of the extrapulmonary sites, the intrathoracic lymph nodes were the most commonly involved. Case patients who were aged 5 years (OR: 1.80; 95% CI: 1.03–3.13; P = .04) as well as foreign-born case patients (OR: 3.13; 95% CI: 1.40–6.57; P = .002) were more likely to have pulmonary TB than case patients who were younger than 5 years and US-born cases, respectively.
Molecular Characterization
We obtained MTB isolates from case patients and compared their molecular profiles with profiles of isolates that were obtained from PSCs. Of 33 case patients from whom PSC isolates were also available, the profiles matched in 30 (91%; Fig 1). In 3 cases (patients A, B, and C), the isolates did not match according to our cluster definition. Patient A was an otherwise healthy 3-year-old black, US-born boy who received a diagnosis of miliary TB. Patient A's second cousin was identified as his PSC (PSC A). PSC A was a 41-year-old black man who had diabetes, mental illness, depression, and drug use and received a diagnosis of pulmonary TB 1 month before patient A (Table 2). The molecular profiles for patient A and PSC A did not match but were similar with identical band patterns and genetic group. The spoligotype differed by 2 of 43 spacers (Table 3).
|
|
|
Patient C was an otherwise healthy 1-year-old black, US-born boy who received a diagnosis of pulmonary TB. Patient C's grandfather was identified as his PSC (PSC C). PSC C was an otherwise healthy 35-year-old black man who received a diagnosis of pulmonary TB 2 months before patient C. PSC was born in the United States, reported drug use 20 years before TB diagnosis, and was incarcerated 8 years before diagnosis. The molecular profiles of patient C and PSC C did not match but were similar, differing by only 1 band and belonging to the same genetic group (Table 3).
MTB isolates from case patients with no known source case were evaluated to determine whether PSCs could be identified using molecular techniques (Fig 2). After comparing isolates with the HTI TB isolate database, we identified 4 PSCs. Additional evaluation of these cases (patients D, E, F, and G) revealed that patient D was an otherwise healthy Hispanic 1-year-old girl who had a diagnosis of pulmonary TB. Patient D was born in the United States; however, during the 6 months before her diagnosis, she was exposed to multiple adults who moved from Honduras during the previous 5 years. Molecular characterization identified patient D's aunt as a PSC (PSC D). PSC D was an otherwise healthy 26-year-old Hispanic woman who received a diagnosis of pulmonary TB 
1 year after patient D.
|
Patient F was otherwise healthy 16-year-old Hispanic girl who received a diagnosis of pulmonary TB. Patient F was born in the United States and reported no exposure to jail or long-term facilities. Molecular characterization identified an unrelated adult with no epidemiologic link to patient F as a PSC (PSC F). PSC F received a diagnosis of TB 4 years before patient F, and only limited clinical information was available for this 35-year-old white man.
Patient G was an otherwise healthy 7-year-old Hispanic girl who received a diagnosis of MTB osteomyelitis of her left femoral head. Patient G was born in El Salvador and had visited her brother while he was in jail. Molecular characterization identified an unrelated adult with no known epidemiologic link to patient G as a PSC (PSC G). PSC G was an 82-year-old Asian Pacific man who received a diagnosis of pulmonary TB 2 years after patient G. PSC G was born in China and was enlisted in the military from 1940 to 1950. He moved to the United States 20 years before TB presentation and then to Houston after 10 years.
We received MTB cultures from 73 (93.5%) of the 78 culture-positive cases. We were able to perform molecular characterization on 71 (97.3%) of the 73 MTB isolates. Fifty-one (71.8%) MTB isolates matched at least 1 other isolate within the HTI TB isolate database and were grouped into 33 molecular clusters. Case patients who were younger than 5 years were more likely to be clustered than those who were 5 years of age (OR: 17.98; 95% CI: 3.6–84.6; P < .001). Case patients who identified a source case (OR: 3.73; 95% CI: 1.25–11.19; P = 0.02) were more likely to be clustered than those who did not identify a source case. In addition, case patients who were born in the United States (OR: 6.90; 95% CI: 2.00–23.82; P = <0.001) were more likely to be clustered than foreign-born case patients. Using methods outlined by Sreevatsan et al,20 we divided case isolates into major genetic groups. Most isolates were classified in major genetic group 2 (n = 36), followed by major genetic group 1 (n = 25) and major genetic group 3 (n = 10). On the basis of the molecular profiles and characteristic spoligotype pattern, 20 (28.2%) of the 71 isolates belong to the Beijing family strains.22 Eighteen (90%) of the 20 isolates that were Beijing family strains were clustered.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Although the majority of case patients were TST positive with abnormal chest radiographs, 17 (7.7%) were TST negative and 21 (9.6%) had normal chest radiographs. In addition, 76 (49%) had negative cultures. Our results are consistent with previous reports of TST and culture sensitivity in children and serve as a reminder that pediatric TB may be clinically silent and difficult to diagnose.26 Rates of extrapulmonary disease, however, were higher in our study as compared with national data.27 Possible explanations include an overall trend toward younger age at diagnosis in Houston. Between 1996 and 2000, the mean age of pediatric TB patients in Houston declined, and this may have had an impact on disease presentation. Another explanation is that cases were diagnosed earlier. If, for example, most cases were evaluated as part of a contact investigation rather than because of symptomatic disease, then they may not have had time to progress from hilar adenopathy, which was defined as extrapulmonary to pulmonary disease.
Molecular characterization helped to verify source case information. Although the majority of case isolates matched their source case isolates, 3 did not. Additional evaluation of isolates revealed that in 2 cases (patients A and C), the isolates from case patients and PSCs were similar. It is interesting that for both patients A and C, the PSCs were family members; thus, the nonmatching profiles may have reflected our strict definition matching rather than a misidentified PSC. Also, the slight change in molecular profile between case patient and PSC suggests the possibility of a change in molecular profile with transmission.20 In contrast, the isolate from patient B and from PSC B had truly different molecular profiles with different drug susceptibility patterns; patient B was drug resistant, and the PSC B was drug susceptible. Our results suggest that for certain pediatric patients, therapy that is based on adult source case culture data cannot be assumed, as transmission of multidrug-resistant TB may have occurred through an unknown contact.28,29 In addition, molecular characterization helped to identify 4 previously unrecognized PSCs. Although an epidemiologic link was found in 1 of the 4 cases (patient D), transmission that is based on molecular characterization is controversial and ignores the possibility of additional exposures such as through travel. It is interesting that patient E's cluster included isolates that were obtained from family members, but because they were diagnosed after patient E, they were not identified as a PSC. Thus, molecular characterization did not identify the most likely PSC, but it did highlight the role that asymptomatic adults play in TB transmission to children.
The majority of cases from whom molecular characterization was available were grouped into clusters, suggesting that recent transmission played a significant role in our population. Risk factors for clustering included age <5 years, having an identified source case, and being born in the United States. Similar trends have been found in the adult population, in which higher clustering rates among US-born case patients has been documented.5,11,30–33 Reasons for foreign-born case patients' having lower rates of clustering may be different in adults and children. Foreign-born adults are more likely to have reactivated disease acquired before immigration, reflecting the epidemiology of their country of origin. In contrast, children are more likely to progress from infection to disease and mirror the epidemiology of their source case.
Spoligotyping has been shown to correlate well with major genetic group designations and can help to differentiate between clones. In our study, the majority of MTB isolates belonged to major genetic group 2. Soini et al34 found a similar distribution of the genetic groups among adult TB cases in Houston. We identified one case of Mycobacterium bovis using spoligotyping, whereas Dankner et al35 found that 33.9% of cases were M bovis. A difference in unpasteurized milk consumption, a known risk factor for M bovis infection, may explain the difference in frequency of M bovis. Whether the patient who had a case of M bovis in our study consumed unpasteurized milk is unknown.
Twenty-eight percent of the isolates in our study belonged to the Beijing family based on spoligotype patterns. This strain has been associated with outbreaks of TB and drug resistance in several parts of the world. Soini et al34 found a similar trend among adults in Houston, where 25% were found to have patterns consistent with the Beijing family. These findings support other studies in which an association between age and Beijing strain isolation has not been established.36–38 It is interesting that 90% of the Beijing strains were clustered, suggesting that recent transmission may also play a role among Beijing strains that are isolated from children.
An important limitation to this study is the differences in case presentation. Whether a case was referred to TCR for evaluation because of school policy, a positive TST, as part of contact investigation, or as a result of symptoms consistent with TB may have had an impact on both the severity of disease at presentation and, consequently, the initial work-up (ie, decision to obtain chest x-ray or cultures). In addition, cultures were not obtained from all patients; thus, the degree of clustering may have been underestimated. The decision to collect cultures was made on an individual basis by pediatricians and dependent on whether cultures were available from a source case. Cultures were less likely to be obtained from a patient when a PSC had been identified, as pediatric TB clinicians often rely on the culture and the sensitivity results from the PSC in treatment decisions for the child. Only 2 investigators performed the majority of clinical evaluations, so the degree of variability in case evaluation may have been inherently limited. In addition, most of the data analysis was presented in aggregate. Thus, the ability to make conclusions about risk factors for TB exposure may be limited, as many risk factors for TB exposure are age dependent. Finally, usefulness of DNA fingerprinting in pediatric TB is limited as children are often culture negative. In our study, only 154 (70%) patients had cultures obtained, of which only 78 (51%) had MTB isolated. Thus, MTB was isolated from only 78 (35%) of 220 patients.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
For clinicians, acquiring a culture in a pediatric patient should be a high priority. In addition, identifying adult contacts by all available methods, including molecular characterization as recently recommended by the Centers for Disease Control and Prevention, should be attempted to identify correctly source cases and to determine correct susceptibility profile so that appropriate therapy can be initiated.39 Emphasis on these strategies will give a more complete picture of both pediatric and overall TB transmission in the community.
|
ACKNOWLEDGMENTS |
|---|
We acknowledge the following members of the Houston Tuberculosis Initiative: Natalie Williams-Bouyer, Xi Pan, Conception Cantu, Heather Tooker-Blue, Thanh Tung Bui, and Syed N. Khalil.
|
FOOTNOTES |
|---|
Reprint requests to (E.A.G.) Department of Pathology (209E), Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030-3498. E-mail: egraviss{at}bcm.tmc.edu
No conflict of interest declared.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  B. J. Marais, T. C. Victor, A. C. Hesseling, M. Barnard, A. Jordaan, W. Brittle, H. Reuter, N. Beyers, P. D. van Helden, R. M. Warren, et al. Beijing and Haarlem Genotypes Are Overrepresented among Children with Drug-Resistant Tuberculosis in the Western Cape Province of South Africa J. Clin. Microbiol., October 1, 2006; 44(10): 3539 - 3543. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Tracing TB with Molecular Precision Journal Watch Infectious Diseases, December 9, 2005; 2005(1209): 10 - 10. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2005 American Academy of Pediatrics. All rights reserved. |
||
|
||
