Annals of Tropical Paediatrics (2000) 20, 223226

A retrospective review of paediatric lymph node

tuberculosis in Port Moresby, Papua New Guinea
P. McMASTER, N. EZEILO & J. D. VINCE
Division of Paediatrics, Clinical Sciences Department, Port Moresby General Hospital, Papua New
Guinea
(Accepted March 2000)
Summary The study aimed to examine the perception that the relapse rate following standard short-
course chemotherapy in children with lymph node tuberculosis is greater than in similarly treated children
with pulmonary tuberculosis (TB). The treatment records, clinical data and results of investigations of 427
children treated for lymph node TB between 1989 and 1996 were analysed. The results and role of
investigations are discussed. The relapse rate was compared with that of 892 children treated for pulmonary
TB during the same period. The documented relapse rate for lymph node TB was 2.8% compared with
0.6% in pulmonary TB. This highly signicant difference led to a prospective study of outcome of lymph
node TB treatment in Port Moresby.
Introduction
The short-course chemotherapy recom-
mended by the World Health Organization
(WHO) is endorsed by many authorities in-
cluding the British Thoracic Society, the
American Thoracic Society, the Centers for
Disease Control (CDC) in Atlanta, the Amer-
ican Academy of Pediatrics and the Canadian
Pediatric Society.
14
However, scientic data
on which to base this recommendation are not
readily available. A literature survey using
Medline and Paperchase produced little infor-
mation. The only publication found dealing
exclusively with paediatric lymph node TB
was a study in which streptomycin was used
5
and this is not currently recommended by
WHO for uncomplicated lymph node TB.
1
A
standard clinical text on TB states that 5% of
adults with lymph node TB are estimated to
have palpable glands at the end of treatment,
but that these rarely cause further trouble.
6
One report from India suggests that 9
months therapy is insufcient for lymph node
disease. Sixty of 82 (74.3%) adults treated
with rifampicin, isoniazid and ethambutol for
9 months showed complete disappearance of
the lymph nodes, but this increased to 70 of
82 (87.2%) when the treatment was continued
for 1 year.
7
Another study suggests that there
is no further evidence of relapse during follow-
up of those with residual nodes.
8
Other litera-
ture relates to lymph node TB in the context
of HIV infection, but this is not the issue
discussed in this study.
9
In a British Thoracic Society study, three of
51 adults who completed 6 months of treat-
ment with currently recommended drugs
(pyrazinamide (Z), isoniazid (H) and ri-
fampicin (R), Z2H6R6) were judged to have
Reprint requests to: Dr Paddy McMaster, Department
of Infectious Diseases, The New Childrens Hospital,
PO Box 3515, Parramatta, NSW 2124, Australia.
Fax: 161 29 845 3421; e-mail: Paddy McMaster@
compuserve.com
ISSN 0272-4936 print/ISSN 1465-3281/online/00/030223-04 2000 The Liverpool School of Tropical Medicine
Carfax Publishing
224 P. McMaster et al.
relapsed, although none of the biopsies was
positive for TB.
2
In the paediatric TB clinic at Port Moresby
General Hospital (PMGH), about 700 chil-
dren less than 12 years of age are commenced
on TB treatment each year. A substantial pro-
portion of these have nodal TB. In Biddulphs
previous study of 639 children enrolled at the
same clinic, 110 (17%) were classied as hav-
ing peripheral lymph node TB.
10
In Port Moresby, the current short-course
regimen for children with lymph node TB or
sputum/gastric aspirate-negative, pulmonary
TB is daily pyrazinamide (Z), isoniazid (H)
and rifampicin (R) for 2 months, followed by
twice-weekly isoniazid (H) and rifampicin (R)
for 4 months (2HRZ/4H2R2). WHO recom-
mends 2HRZ/4H3R3 but states that 2HRZ/
4H2R2 is acceptable.
1
Senior medical staff
working in the clinic had the impression that
the relapse rate following this regimen was
higher in children with nodal TB than in those
with pulmonary TB, and the study was under-
taken to examine this perception.
Methods
A retrospective study was conducted in the
paediatric TB clinic and ward of Port Moresby
General Hospital, Papua New Guinea. Data
were collected from the paediatric TB regis-
ters. On discharge from the ward or on start-
ing treatment in clinic, the following
information is entered in the register: name,
address, age, diagnosis, weight, BCG immu-
nization status, chest X-ray (CXR) ndings,
ne needle aspiration biopsy (FNAB)/sputum/
gastric aspirate, Mantoux or other investiga-
tion results, date of starting treatment, family
screening and if previously commenced on
treatment and whether it was completed. At
each review (usually monthly) a record is kept
of compliance, weight and clinical status. The
Mantoux was performed using 10 IU human
tuberculin and read as positive if the indura-
tion was of more than 10 mm diameter, irre-
spective of BCG history. A CXR was not
performed on every child but only if deemed
necessary by the paediatrician seeing the pa-
tient. FNAB was performed using a 25 French
gauge needle without sedation, anaesthetic or
analgesia.
Generally, ne needle aspiration biopsy
(FNAB) was done only when resistant organ-
isms were suspected or to rule out malignancy
when the diagnosis was in doubt. Whilst
FNAB was easily organized, results were often
difcult to obtain. When the FNAB was in-
conclusive and there were worrying features
such as lymphadenopathy in multiple sites, a
biopsy of the lymph node was obtained. This
was much more difcult to organize as there
was little theatre time for anything as elective
as this. The presence of granulomata or
caseating necrosis was taken as positive for
TB. Fite Faraco (modied Ziehl Neelsen
stain) was used for AFB. The specimens have
been cultured on liquid broth medium (Bactec
12 B) since 1994, and prior to this on Lowen-
steinJensen medium. Identication of my-
cobacterium and antimycobacterial drug
sensitivity testing were routine during only
part of the study period.
The study period was from 1 August 1989
to 31 December 1996. The children enrolled
in the study are those recorded as having only
lymph node TB. Those with a diagnosis of
lymph node and other forms of TB were ex-
cluded. However, those whose chest X-rays
were consistent with TB but with a primary
diagnosis of lymph node TB were included.
Relapse is dened as a patient requiring a
second course of TB chemotherapy for en-
largement of lymph nodes associated with
other symptoms of TB (such as pyrexia, anor-
exia, etc.) after having completed a 6-month
course, with apparent resolution of symptoms
and signs of active TB. The comparison group
was children with pulmonary TB enrolled over
the same period. Children with extra-pul-
monary TB in sites other than lymph nodes
alone were excluded from both groups. The
data were analysed on Epi Info version 6.02.
Results
Of the 427 children who started TB treat-
ment, our records conrm completion of
Lymph node tuberculosis in PNG 225
treatment in only 207 (48%). There is an
equal sex ratio for presentation of lymph node
TB and for completion of treatment. Eleven
children with lymph node TB were recorded
as having had no BCG (3.4% of those for
whom BCG immunization status was
recorded).
Fine needle aspiration biopsy (FNAB) was
performed in 122 children, although only 65
results were recorded. Of those, 59 (90%)
were positivedened as the presence of ep-
ithelioid cells or granulomata suggestive of tu-
berculosis on histology or positive culture.
Eighteen excision biopsies were performed
and of the 15 results available, 13 were posi-
tive (by the same histological or microbiologi-
cal criteria). AFBs in FNAB or biopsy were
detected far less frequently with only ve posi-
tive results from all specimens. All specimens
were cultured for TB but no results are avail-
able.
Fifteen (3.5%) of 427 children diagnosed as
having lymph node TB were restarted on
treatment. Of the 207 who completed treat-
ment, ve (2.4%) relapsed. A lymph node
biopsy taken from one of them on restarting
treatment was histologically positive for TB.
Fifteen of 220 (6.8%) children with incom-
plete records restarted treatment.
During the same period, 892 cases of pul-
monary TB (with no record of associated per-
ipheral lymphadenopathy or any other
extra-pulmonary site) were documented as
having completed treatment by the same cri-
teria as the TB lymph node patients. Of these,
ve relapsed, giving a relapse rate of 0.56%.
This is highly signicantly different from the
incidence of relapse in lymph node TB (v
2
test, p 50.0001).
Discussion
The results of this study lend support to the
clinical impression that there is a higher re-
lapse rate in children with lymph node TB
than in those with pulmonary TB. The
signicance of this retrospective study is obvi-
ously limited by large omissions in record-
keeping. However, it does reveal a number of
important issues in the management of lymph
node TB in children.
There are issues concerning the diagnosis of
lymph node TB. Our diagnoses were primarily
clinical. Lymphadenopathy in children, partic-
ularly in developing countries, is common. As
a general rule, TB was suspected when a cervi-
cal lymph node was of more than 1 cm diam-
eter, an axillary node more than 1.5 cm and an
inguinal node more than 2 cm, and when
there was no response to a course of antibi-
otics. Almost all diagnoses were made by staff
at consultant level with considerable experi-
ence of TB.
The very high positive diagnostic rate of the
FNAB and excision biopsies supports the ac-
curacy of clinical diagnoses. There was no
record of any patient with clinically diagnosed
lymph node TB subsequently being shown to
have malignant disease. No problems associ-
ated with FNAB such as tracking and scarring
were recorded.
The pick-up rate of acid- and alcohol-fast
bacilli (AAFB) staining was low. The micro-
biology department at PMGH has very rarely
identied atypical mycobacteria (personal
communication).
Although Mantoux testing can be very use-
ful in diagnosing TB, interpretation of results
in developing countries is fraught with
difculty. Tuberculin strengths and prepara-
tions may vary, and potency is reduced with
storage, particularly if the preparation is not
kept cool. Severely malnourished children are
likely to have negative Mantoux tests despite
heavy infection. Interpretation is made
difcult by a high BCG vaccination rate in
infancy and in situations where there is a very
high rate of primary infection among the
young population. Nevertheless, in a well-
nourished child with lymphadenopathy, a
positive Mantoux test is useful supporting evi-
dence.
It is recognized that lymph nodes occasion-
ally enlarge after treatment has been com-
pleted. It is thought that this represents an
immunological reaction to tubercular protein
rather than relapse. Even during therapy,
lymph nodes may increase in size, become
226 P. McMaster et al.
uctuant and suppurate. This makes the
denition of relapse somewhat difcult but,
for the purposes of the study, relapse was
diagnosed when enlargement of glands was
associated with other symptoms of TB
(pyrexia, anorexia, etc.) after cessation of
treatment and apparent resolution of symp-
toms and signs of active TB. Lymph nodes
persisting but not enlarging after treatment
were not regarded as indicating relapse.
Nevertheless, the question remains as to
whether patients who still have lymph nodes
after 6 months of treatment have an increased
relapse rate compared with those whose gland
enlargement has resolved.
The prevalence of HIV infection in children
during the study period was very low: only 31
PNG children had been diagnosed as HIV-
positive after 18 months of age by the end of
the study period. None of these was in the
study population. There is no doubt, however,
that with the rapidly escalating epidemic of
sexually transmitted HIV infection in PNG,
the diagnosis of TB lymphadenopathy is likely
to become even more complex than it is at
present.
The present study raises the possibility that
the recommended standard 6 months short-
course anti-TB chemotherapy may be in-
sufcient in childhood lymph node TB. A
prospective study with improved follow-up is
needed before rm conclusions can be drawn.
In a prospective study, diagnosis before start-
ing treatment and of possible relapse should
be conrmed histologically or microbiologi-
cally. A prospective study may also help ident-
ify causal factors in those who relapse. Such a
prospective study is currently in progress at
PMGH.
References
1 World Health Organization. Treatment of Tu-
berculosis. Guidelines for National Programmes,
2nd edn. Geneva: WHO, 1997.
2 Campbell IA, Ormerod LP, Friend JAR, Jenkins
PA, Prescott RJ. Six-months versus nine-months
chemotherapy for tuberculosis of lymph nodes:
nal results. Respir Med 1993; 87:621

3.
3 American Thoracic Society. Treatment of tubercu-
losis and tuberculosis infection in adults and chil-
dren. Am J Respir Crit Care Med 1994;
149:1359

74.
4 Canadian Paediatric Society, Infectious Diseases
and Immunization Committee. Short course ther-
apy for tuberculosis in infants and children. Can
Med Assoc J 1994; 150:1233

9.
5 Jawahar MS, Sivasubramanian S, Vijayan VK, et al.
Short course chemotherapy for tuberculous
lymphadenitis in children. Br Med J 1990;
301:359

62.
6 Crofton J, Horne N, Miller F. Clinical Tubercu-
losis. London: Macmillan, 1992; 125.
7 Behra D, Jindal SK. Short course chemotherapy for
lymphnode tuberculosis. Indian J Chest Dis Allied
Sci 1990; 32:73

4.
8 Bedi RS. Short course chemotherapy for
lymphnode tuberculosis. Indian J Chest Dis Allied
Sci 1991; 33:171

2.
9 Bem C. Human immunodeciency virus-positive
tuberculous lymphadenitis in Central Africa: clini-
cal presentation of 157 cases. Int J Tuberc Lung
Dis 1997; 1:215

9.
10 Biddulph J. Short course chemotherapy for child-
hood tuberculosis. Pediatr Infect Dis J 1990;
9:794

801.