Tuberculosis (TB) remains a public health problem, especially among young people. The diagnosis of TB in children is difficult to confirm as the bacterium is only cultured in a small percentage of cases. Chest radiography and presence of lymphadenopathy are important diagnostic features. This study assessed the value of ultrasound (US) for detecting mediastinal lymphadenopathy in 32 children with positive tuberculin skin tests. US identified lymphadenopathy in 90.5% of children with radiographic abnormalities suggestive of TB and in 66.7% of children with normal chest radiographs. US findings agreed with computed tomography findings in most cases, showing US is useful for detecting enlarged lymph nodes when chest radiographs are normal.

1. Introduction
Tuberculosis (TB) has re-emerged as a serious public
health problem in developed countries, particularly
among young adults and children. The diagnosis of TB
in children is often difficult to confirm, because Myco-bacterium
TB is cultured only in a small percentage of
cases [1, 2]. Whereas the diagnosis of active TB in adults
is mainly bacteriological, in children it is usually epide-miological
and indirect. In the absence of a positive
culture, the strongest evidence for TB in a child is recent
exposure to an adult with active disease [3]. Indirect
diagnostic techniques, such as the tuberculin skin test,
chest radiography and physical examination offer sup-portive
information [4].
Central to the clinical diagnosis of childhood TB is
the chest radiograph and the presence of lymphade-nopathy
with or without parenchymal involvement is the
single most important diagnostic feature [4]. The nodal
enlargement typically involves the hilar and paratracheal
nodes, with bilateral hilar lymphadenopathy identified
in about 25% of cases. Different studies have docu-mented
right-sided predominance of lymphadenopathy
and parenchymal changes [5]. Both frontal and lateral
views are necessary to evaluate lymphadenopathy. Of
interest is the fact that enlarged lymph nodes may be
Joaquim Bosch-Marcet
Xavier Serres-Cre´ixams
Amalia Zuasnabar-Cotro
Xavier Codina-Puig
Margarita Catala` -Puigbo´
Jose´ L. Simon-Riazuelo
Comparison of ultrasound with plain
radiography and CT for the detection
of mediastinal lymphadenopathy in children
with tuberculosis
Received: 21 February 2004
Revised: 1 April 2004
Accepted: 11 May 2004
Published online: 9 September 2004
Springer-Verlag 2004
Abstract Background: Lymphade-nopathy,
with or without parenchy-mal
abnormality, is the radiological
hallmark of primary tuberculosis
(TB) in children. However, lymph
node enlargement may pass unde-tected
on plain chest radiographs.
Ultrasonography provides comple-mentary
information to that ob-tained
by radiographs. Objective: To
assess the clinical value of US for the
detection of mediastinal lymphade-nopathy
in children with a positive
intradermal tuberculin test. Materi-als
and methods: Thirty-two children
with a mean age of 6 years and a
positive Mantoux test underwent
chest radiography (frontal and lat-eral)
and US (suprasternal and left
parasternal access routes). Chest CT
was performed at the discretion of
the attending physician in six cases.
Results: Eleven children had clinical
symptoms and 90% a recent contact
with a person with active TB. In
90.5% of children with chest radio-graphic
images compatible with TB,
coincident findings in the mediasti-nal
US study were found. By com-parison,
66.7% of those with normal
chest radiography had evidence of
mediastinal lymphadenopathy on
the US scan. In all cases but one, US
and CT findings agreed. Conclu-sions:
Mediastinal US is useful for
the detection of enlarged lymph
nodes in children with a positive
tuberculin reaction and normal chest
radiography.
Keywords Mediastinum Æ
TB Æ Lymphadenopathy Æ
Radiography Æ Ultrasound Æ
CT Æ Children
Pediatr Radiol (2004) 34: 895–900
DOI 10.1007/s00247-004-1251-3 ORIGINAL ARTICLE
J. Bosch-Marcet ()
A. Zuasnabar-Cotro Æ M. Catala` -Puigbo´
J. L. Simon-Riazuelo
Department of Paediatrics,
Avda. Francesc Ribas s/n,
Hospital General de Granollers, 08400
Granollers, Barcelona, Spain
E-mail: 8543jbm@telefonica.net
Tel.: +34-93-8425039
Fax: +34-93-8425036
X. Serres-Cre´ixams
Department of Diagnostic Imaging,
Hospital General de Granollers,
Barcelona, Spain
X. Codina-Puig
Emergency Department,
Hospital General de Granollers,
Barcelona, Spain

2. detected by ultrafast CT in 60% of children with
tuberculous infection and normal findings on chest
radiography [6]. However, this technique would not be
available in most cases and the cost is very high. We
report the clinical value of US to detect mediastinal
lymph node involvement in children with a positive
intradermal tuberculin skin test.
Materials and methods
A retrospective review of the medical records of 32 chil-dren,
17 boys and 15 girls, with a mean age of 6 years
(range 4 months to 17 years), who had a positive intra-dermal
tuberculin skin test was made. These patients had
been referred to our Department of Paediatrics for work-up
studies and eventual treatment between 1994 and 2000.
None of the patients had been exposed to BCG vaccina-tion.
All patients underwent a thorough history (including
exposure tracing), physical examination, frontal and lat-eral
chest radiographs, and sonographic study of the
mediastinum. The radiographic findings considered rep-resentative
of TB included nonspecific localized infil-trates,
hilar adenitis, localized hyperaeration, atelectasis,
segmental lesions, cavitation, calcification, and localized
pleural effusion.CTof the chest was performed in selected
patients at the discretion of the physicians in charge.
Ultrasonography of the mediastinum was performed
with high-resolution equipment (Logiq 700, General
Electric) using a 5-MHz convex probe. The presence of
one or more masses with an ovoid or round shape and
hypoechoic appearance in the anterior or middle medi-astinum
was recorded. The anterior mediastinum in-cluded
the prevascular region, occupied by the thymus
gland and the middle mediastinum, the right paratrac-heal,
supra-aortic, aortopulmonary, and subcarinal re-gions.
On US the normal thymus has a bilobulated
appearance and homogeneous echotexture with some
echogenic strands. It is hypoechoic relative to the thy-roid
gland and has a smooth, well-defined margin due to
its fibrous capsule. It is a soft organ that does not
compress neighbouring vascular structures, a charac-teristic
that can help the radiologist to differentiate it
from mediastinal masses. The normal thymus can vary
considerably in position, extension, size and configura-tion.
In small children, the organ can extend from the
cervical region to the diaphragm. During respiration and
particularly when the child is crying, the thymus can be
above the manubrium and simulate a cervical mass.
The mediastinum was accessed via the suprasternal
and left parasternal approaches [7]. When using the
suprasternal approach, the patient was placed in a
supine decubitus position with a cushion under the
back and the neck slightly extended. The transducer
was placed above the manubrium and titled caudally.
To obtain an oblique sagittal view, the probe was
placed laterally to encounter the space between the
trachea and the sternocleidomastoid muscle. For the
left parasternal approach, the patient was placed in a
left lateral decubitus position to move the mediastinum
downwards and increase the size of the anatomic
acoustic window. Five standard sonographic slices
were used to visualize the complete anterior and mid-dle
regions of the mediastinum. Three sonographic
slices were obtained with the suprasternal approach
(oblique coronal, coronal, and oblique parasagittal)
and two with the left parasternal approach (axial and
parasagittal views). The oblique coronal view through
the suprasternal approach was used to visualize the
paratracheal region and to study the aortopulmonary
region; the coronal view was useful for visualizing the
vessels, particularly the SVC; the oblique parasagittal
view visualized the aortopulmonary region. The axial
and parasagittal views through the left parasternal
approach were used to study the subcarinal and pre-vascular
regions.
In all the cases, the number and size (long axis) of
lymph nodes were determined. The following groups
were established arbitrarily: no adenopathy or lymph
nodes 10 mm in diameter (negative, group 0); a single
lymph node 10 mm (positive +, group 1); a single
lymph node 15 mm (positive ++, group 2); a
single lymph node 20 mm (positive +++, group 3);
more than one lymph node 15 mm (positive++++,
group 4). In the case of clearly matted nodes, the size of
the whole mass was considered. When possible, the size
of each of its components was measured.
For each patient, the results of chest radiography, US
of the mediastinum, and chest CT were compared.
Results
Of the 32 patients who had US studies of the mediasti-num,
90% had recent contact with a person with con-firmed
pulmonary TB. Only 11 (34.4%) children had
clinical manifestations such as fatigue, low-grade fever,
mild cough, weight loss, night sweats, chills, and failure
to thrive. The remaining 21 children were asymptomatic,
but with a positive tuberculin skin test. Pulmonary
radiographic findings were suggestive of TB in 21 chil-dren,
negative in nine, and uncertain in two. With regard
to US of the mediastinum, there were five children in
group 0, 15 in group 1, two in group 2, four in group 3,
and six in group 4. CT of the chest was performed in six
children. Details of findings of chest radiography,
mediastinal US and chest CT are shown in Table 1 and
in Figs. 1, 2.
In the group of nine children with normal findings on
chest radiography, US of the mediastinum confirmed
lymphadenopathy in six cases (66.7%) and was negative
in the remaining three. One of these three patients had a
896

3. Table 1 Reults of chest roentgenograms, mediastinal ultrasonography, and chest CT scans in 32 children with positive intradermal
tuberculin skin test
Case Sex and age Chest X-ray Mediastinal lymph nodes
by ultrasonography
normal chest CT scan and in the other two, CT exami-nation
was not performed. All patients but two with
compatible radiological findings of TB had visible
mediastinal lymph nodes on US. Therefore, 90.5% (19
out of 21) of patients with pathologic images in the chest
radiographs, had visible mediastinal lymphadenopathy
in the ultrasonographic study. In the two patients with
doubtful radiological images, ultrasonography con-firmed
the diagnosis of tuberculous lymphadenopathy in
two.
Mantoux (mm) Chest CT scan
1 Female, 7 years Normal Present (+) 20 ND
2 Male, 2 years Compatible TB Present (++++) 16 ND
3 Male, 2 years Compatible TB Present (+) 20 ND
4 Male, 7 months Normal Present (++) 10 ND
5 Female, 14 years Compatible TB Present (+) 10 Confirmatory
6 Female, 16 months Compatible TB Present (+) 10 ND
7 Female, 11 years Compatible TB Present (++++) 16 Confirmatory
8 Female, 3 years Compatible TB Present (+) 10 ND
9 Male, 16 years Compatible TB Present (++) 10 ND
10 Male, 18 months Compatible TB Present (++++) 10 ND
11 Male, 13 years Doubtful Present (+++) 10 Normal
12 Male, 8 years Normal Present (++++) 10 ND
13 Female, 2 years Compatible TB Present (+) 10 ND
14 Male, 22 months Compatible TB Present (+++) 28 ND
15 Female, 4 years Compatible TB Present (++++) 14 ND
16 Female, 12 years Compatible TB Present (+++) 10 Confirmatory
17 Female, 11 years Doubtful Present (+) 10 ND
18 Male, 4 months Compatible TB Present (++++) 10 ND
19 Male, 2 years Compatible TB Present (+) 10 ND
20 Female, 17 years Compatible TB Present (+) 22 ND
21 Female, 14 months Normal Present (+++) 16 ND
22 Male, 14 months Compatible TB Present (+) 14 ND
23 Female, 14 months Compatible TB Present (+) 10 ND
24 Male, 7 months Compatible TB Absent 12 ND
25 Male, 3 years Compatible TB Present (+) 9 ND
26 Male, 14 years Normal Present (+) 20 Confirmatory
27 Male, 2 years Normal Present (+) 10 ND
28 Female, 15 years Compatible TB Present (+) 10 ND
29 Female, 13 months Normal Absent 10 ND
30 Female, 15 months Normal Absent 10 ND
31 Male, 12 years Normal Absent 10 Normal
32 Male, 5 years Compatible TB Absent 14 ND
ND not done
Fig. 1 Results of chest radiog-raphy
897

4. The chest CT examination, which was carried out in
six patients, confirmed the results of US in four. In one
patient with normal radiographic findings and absence
of mediastinal adenopathy, the CT scan was also nega-tive,
whereas in the other patient with uncertain radio-graphic
findings and mediastinal lymphadenopathy in
the ultrasound examination, the CT scan was negative.
This patient, however, was given antituberculous treat-ment
and his clinical symptoms resolved and radiologi-cal
images cleared. In five of six (83.3%) patients a
concordance between results of mediastinal ultrasonog-raphy
and CT examination was observed.
The case of a patient with lymphadenopathy in the
right paratracheal region using the suprasternal ap-proach
is shown in Fig. 3. In this case, results of US
were confirmed by CT (Fig. 3). In the case of a 14-year-old
patients with active TB involving the left upper lobe,
the suprasternal approach revealed a lymph node,
1.8 cm in diameter, in the aortopulmonary region
(Fig. 4). A lymphadenopathy in the subcarinal space
was detected in a patient with normal chest roentgeno-gram
using the left parasternal approach (Fig. 5).
Discussion
Most TB infections in children and adolescents are
asymptomatic when the tuberculin skin test is positive.
In the present series, only 34.4% of patients had non-specific
symptoms, such as fever, cough, weight loss, and
failure-to-thrive pattern in young infants. All patients
were referred for evaluation because of the tuberculin
skin test and in 90% of them, a recent exposure to an
adult with active disease was present. It should be noted
that there were two patients aged between 15 and
17 years of age. Despite the fact that these patients were
adolescents, they were referred for evaluation to our
department because 18 years is the upper age limit as-signed
to pediatrics by our health care system. On the
other hand, the fact that 32 patients with a tentative
diagnosis of TB had been referred for work-up studies
during the study period indicates that although signifi-cant
progress has been made in the control of TB in
developed countries, this communicable disease has not
yet been eradicated. Furthermore, children with primary
tuberculous infection are the reservoir from which future
cases will emerge.
One of the major practical problems in diagnosing
TB in children is that isolation of Mycobacterium TB
from gastric aspirates or sputum is difficult [8]. Sputum
for acid-fast stain and culture is rarely available from
infants and children. Optimal collection of gastric aspi-rates
requires hospitalization to sample the swallowed
secretions that accumulate overnight. However, the
Fig. 2 Results of mediastinal US
Fig. 3 Right paratracheal lymphadenopathy. a Suprasternal,
oblique coronal US section. The echogenic line originated in the
right upper lobe is displaced by the mass. b Axial CT section
confirming the US findings. IA innominate artery, LBV left
brachiocephalic vein, AO aorta, TR trachea, RUL right upper
lobe, LN lymph node
898

5. sensitivity of acid-fast stain for gastric contents is usu-ally
below 10%. The low yield of positive cultures from
gastric aspirates is a result of the small number of
organisms in primary TB in childhood and possible
inadequate techniques for collection of gastric washings.
Therefore, the diagnosis is frequently based solely on
899
detecting typical radiographic abnormalities in a child
with a reactive tuberculin skin test and with history of
contact of an infectious case. The Mantoux method is
helpful in supporting the diagnosis. Although a reaction
of ‡10 mm induration is the usual cut-point for defining
a significant reaction, a reaction of ‡5 mm is considered
significant for symptomatic children and for recent
contacts with infectious cases [9]. However, a negative
Fig. 4a, b 14-year-old patient with active TB. Suprasternal oblique
parasagittal (a) and suprasternal oblique coronal (b) US sections
showing a lymph node 1.8 cm in diameter in the aortopulmonary
region involving the left upper lobe. IA innominate artery, LBV left
brachiocephalic vein, RUL right upper lobe, TR trachea, AO aorta,
LN lymph node, LPA left pulmonary artery, LB left bronchus, LC
left carotid artery; LS left subclavian artery, LB left bronchus, RPA
right pulmonary artery, LA left atrium
Fig. 5 Positive US with negative radiograph. a Normal frontal
chest radiograph. b Left parasternal axial US section in the same
patient shows lymphadenopathy in the subcarinal space. TH
thymus gland, AO aorta, RPA right pulmonary artery, LC left
carotid artery, LS left subclavian artery, LA left atrium, LB left
bronchus, LBV left brachiocephalic vein, LPA left pulmonary
artery, LN lymph node

6. reaction in a child who has signs and symptoms com-patible
with TB does not rule out the diagnosis. In the
present series, indurations ranged between 9 and 28 mm.
Lymphadenopathy, with or without parenchymal
abnormality, is the radiological hallmark of primary TB
in children [10]. Children less than 3 years of age show a
higher prevalence of lymphadenopathy and a lower
prevalence of parenchymal abnormalities compared with
children 4–15 years [5]. In early childhood, lymphade-nopathy
as the sole radiological manifestation of disease
was seen in 49% of cases versus 9% in late childhood
and adolescence according to data reported by Leung
et al. [11]. Bronchi in infants are of smaller calibre and
more easily compressed by enlarging hilar lymph nodes.
As the hilar lymph nodes enlarge, bronchial obstruction
may occur and signs of air trapping may develop. Al-though
hilar lymphadenopathy may be the only sug-gestive
finding of TB in the chest radiographs, in the
present study, 66.7% of patients with chest radiographs
considered unrevealing showed mediastinal lymphade-nopathy
in the ultrasound examination. Lymph nodes
can sometimes be difficult to visualize on frontal plain
radiographs. Occasionally, lymphadenopathy is visible
only on the lateral film [12]. Apical-lordotic views may
aid in visualizing lesions obscured by the heart. When no
lymphadenopathy is present on the standard radio-graphic
examination of the chest, special imaging tech-niques
such as CT may be of particular value [13]. It has
been shown that CT scan may reveal mediastinal aden-opathies
which are not evident on the chest radiograph
[6]. Ultrafast CT scanning, however, is costly, not
available in many institutions, includes radiation, and
may require the use of sedation in young children. In
contrast, US is much less expensive, the use of sedatives
or contrast medium is not necessary, and can be easily
obtained both in the hospital and in primary care set-tings.
Although subcarinal adenopathy has recently been
reported to be the most common site of lymphadenop-athy
in children with TB [14], we have detected small
adenopathies more frequently in the paratracheal region
and aortopulmonary window because of a better echo-graphic
access. In the subcarinal region, we have docu-mented
large lymph nodes due to limitations in the
echographic access and artifacts (e.g., the oesophagus).
As far as we are aware, no previous study regarding
the usefulness of mediastinal ultrasonography for the
diagnosis of lymphadenopathy in TB in children has
been published. For this reason, the present results
cannot be compared to those reported by others.
Conclusions
In the present series of 32 patients with positive tuber-culin
skin test, 90.5% of those with chest radiographic
images compatible with TB had coincident findings in
the mediastinal ultrasonographic study. On the other
hand, 66.7% of those with normal chest radiography
had evidence of mediastinal lymphadenopathy on US.
In all cases but one, US and CT findings agreed. In view
of the usefulness of US of the mediastinum for the
diagnosis of lymphadenopathy in children with TB, this
non-invasive method could also be of value in the con-trol
and follow-up of children receiving antituberculous
chemotherapy.
Acknowledgement We thank Marta Pulido, MD, for editing the
manuscript and for editorial assistance.
References
1. Vallejo JG, Ong LT, Starke JR (1994)
Clinical features, diagnosis, and treat-ment
of TB in infants. Pediatrics 94:1–7
2. American Thoracic Society (1990)
Diagnostic standards and classification
of TB. Am Rev Respir Dis 142:725–735
3. Hilman BC (1993) Pulmonary TB and
tuberculous infection in infants, chil-dren,
and adolescents. In: Hilman BC
(ed) Pediatric respiratory disease: diag-nosis
and treatment. Saunders, Phila-delphia,
pp 311–319
4. Snider DE, Rieder HL, Combs D et al
(1988) TB in children. Pediatr Infect Dis
7:271–278
5. Burroughs M, Beitel A, Kawamura A
et al (1999) Clinical presentation of TB
in culture-positive children. Pediatr In-fect
Dis 18:440–446
6. Delacourt C, Mani TM, Bonnerot V
et al (1993) Computed tomography with
normal chest radiograph in tuberculous
infection. Arch Dis Child 69:430–432
7. Lucaya J, Strife J (2001) Pediatric chest
imaging: chest imaging in infants and
children. Springer, Berlin Heidelberg
New York, pp 1–25
8. Starke JR (1988) Modern approach to
the diagnosis and treatment of TB in
children. Pediatr Clin N Am 35:441–464
9. American Thoracic Society and the
Centers for Disease Control and Pre-vention
(2000) Diagnostic standards
and classification of TB in adults and
children. Am J Respir Crit Care Med
161:1376–1395
10. Omlor GJ (2001) Pulmonary lymph-adenopathy.
Pediatr Infect Dis 20:437–
438
11. Leung AN, Mu¨ ller NL, Pineda PR et al
(1992) Primary TB in childhood:
radiographic manifestations. Radiology
182:87–91
12. Smuts NA, Beyers N, Gie RP et al
(1994) Value of the lateral chest radio-graph
in TB in children. Pediatr Radiol
24:478–480
13. Vallejo JG, Starke JR (1996) Intratho-racic
TB in children. Semin Respir Dis
11:184–195
14. Andronikou S, Joseph E, Lucas S et al
(2004) CT scanning for the detection of
tuberculous mediastinal and hilar
lymphadenopathy in children. Pediatr
Radiol 34:232–236
900