Mediastinal Granulomatous

Lymphadenitis in a Population
at Risk for HIV and Tuberculosis
Daniel R. De Wet, M.D.,1 Colleen A. Wright, M.D., Ph.D., F.I.A.C.,1,2
Pawel T. Schubert, M.D., M.I.A.C.,1 Coenraad F. N. Koegelenberg, M.D., Ph.D.,3
Mercia Louw, M.D.,1 and Andreas H. Diacon, M.D., Ph.D.3*

Background: Granulomatous inflammation on transbronchial areas of tuberculosis. Diagn. Cytopathol. 2015;43:696–700.

needle aspirates from mediastinal lymph nodes is an infre- V
C 2015 Wiley Periodicals, Inc.

quent yet important finding. We determined associations

between cytomorphological features and underlying aetiology
in an area of high prevalence of HIV-infection and
tuberculosis.
Methods: We identified cases with granulomatous inflammation
on mediastinal aspirates from January 2003 to July 2010. Cyto-
morphological features were evaluated and graded according to
a simple and reproducible system including the presence, quality
(discrete or vague), and number (≤5 or more) of granulomas as
well as the presence of necrosis, lymphocytes, multinucleated
giant cells, and neutrophils.
Results: In 81 patients (36 male, 9 HIV-positive) the final
diagnosis was tuberculosis in 37 (46%), sarcoidosis in 40
(49%), fibrosing mediastinitis in 1 (1%), and unknown in 3
(4%). The presence of necrosis (P < 0.001) and neutrophils
(P 5 0.05) was associated with tuberculosis and numerous
discrete granulomas were associated with sarcoidosis
(P 5 0.03). All HIV-positive patients were diagnosed with
tuberculosis.
Conclusion: Granulomatous disease identified on TBNA from
mediastinal lymph nodes is mostly associated with sarcoidosis
and tuberculosis. Ancillary investigations for sarcoidosis are
appropriate if numerous discrete granulomas are found. Tuber-
culosis must be excluded if necrosis and neutrophils are present
and in HIV-positive individuals, particularly in high-burden
1
Faculty of Medicine and Health Sciences, Department of Pathology,
Stellenbosch University, National Health Laboratory Service, Tygerberg
Hospital, Cape Town, South Africa
2
National Health Laboratory Service, Port Elizabeth, South Africa
3
Faculty of Medicine and Health Sciences, Division of Pulmonology,
Department of Medicine, Stellenbosch University Cape Town, South
Africa
*Correspondence to: Dr Andreas Diacon, MD, PhD, Faculty of Medi-
cine and Health Sciences, Stellenbosch University, Tygerberg 7505,
South Africa. E-mail: ahd@sun.ac.za
Received 13 November 2014; Revised 23 February 2015; Accepted
30 March 2015
DOI: 10.1002/dc.23280
Published online 4 May 2015 in Wiley Online Library
(wileyonlinelibrary.com).
Key Words: transbronchial needle aspiration; rapid on-site
evaluation; granulomatous inflammation; tuberculosis; sarcoido-
sis; HIV-infection
Introduction
Transbronchial needle aspiration (TBNA) is an estab-
lished, minimally invasive, effective, and cost-saving
bronchoscopic procedure used in the diagnosis of pri-
mary tumours; to obtain material from mediastinal and
hilar nodes for staging as well as the diagnosis of reac-
tive lymphadenopathies.1–5 Although generally used in
the diagnosis and staging of lung carcinoma, the diagno-
sis of non-neoplastic conditions such as granulomatous
inflammation has significant implications for clinical
management of the patient. As a distinctive pattern of
chronic inflammation, granulomatous inflammation is
encountered in a limited number of infectious (most
notably tuberculosis) and noninfectious conditions includ-
ing sarcoidosis.
The Western Cape Province of South Africa is regarded
as a very high burden tuberculosis region with the inci-
dence of tuberculosis in 2012 reported as 935/100,000 as
compared to the national incidence rate of 823/100,000. 6
Half of all new tuberculosis cases are found to be HIV-
coinfected and such immunocompromised subjects can
present with intrathoracic lymphadenopathy and no
parenchymal infiltration. Sputum examination is usually
negative and more invasive methods may be necessary to
make an early diagnosis.1,7 This more invasive approach
is warranted for early and optimal treatment in an era of
multidrug resistance.7

696 Diagnostic Cytopathology, Vol. 43, No 9 V

C 2015 WILEY PERIODICALS, INC.

Table I. Cytomorphological Grading System
Feature
Discrete granulomas 0 ≤5
Vague granulomas 0 ≤5
Multinucleated Absent Present
giant cells
Necrosis 0 11
Absent <33% of slide
Lymphocytes 0 11
Absent <33% of slide
Neutrophils 0 11
Absent <33% of slide
Sarcoidosis is a multisystem disorder of unknown etiol-
ogy characterized by noncaseating epithelioid cell granu-
lomas. There is no single gold standard diagnostic test for
sarcoidosis and diagnostic criteria include identification of
granulomatous inflammation with noncaseating granulo-
mas usually accompanied by multinucleated giant cells.2
Less common causes of granulomatous lymphadenopathy
include infections by mycobacteria other than Mycobacte-
rium tuberculosis (such as Mycobacterium avium-intracel-
lulare), fungal organisms such as Histoplasma and
Cryptococcus, foreign body reactions to metal dusts such
as Berylliosis or organic antigens and even malignancies
such as Hodgkin Lymphoma (HL), and carcinomas (par-
ticularly with squamous differentiation).7,8
It is essential that cytopathologists are able to recog-
nize granulomatous inflammation and recommend appro-
priate ancillary tests to confirm the most likely aetiology.
This is especially important during TBNA of mediastinal
lymph nodes where rapid on-site evaluation (ROSE) of
aspirates allows triaging of aspirated material for stains
for mycobacteria and addititonal microbiological culture
or PCR to confirm tuberculosis, or to advise clinicians to
collect bronchoalveolar lavage and transbronchial or
mucosal biopsies to confirm sarcoidosis. This retrospec-
tive study was aimed to determine the relative frequency
and spectrum of underlying pathology of granulomatous
disease on TBNA in an area of high prevalence of
tuberculosis.
Materials and Methods
Study Design, Patients, and Procedures
We conducted a retrospective review of all patients diag-
nosed with granulomatous inflammation on TBNA from 1
January 2003 to 31 July 2010 in Tygerberg Academic
Hospital, Cape Town, South Africa. Cases were identified
on the computerized laboratory information management
system. Search parameters included the procedure (bron-
choscopy), site (respiratory system), and diagnosis (granu-
lomatous inflammation, cytological picture suggestive of
mycobacterial infection, epithelioid granulomata with
Diagnostic Cytopathology DOI 10.1002/dc
MEDIASTINAL GRANULOMATOUS LYMPHADENITIS
Grading
>5
>5
21 31
33-66% of slide >66% of slide
21 31
33-66% of slide >66% of slide
21 31
33-66% of slide >66% of slide
caseous necrosis present and consistent with mycobacte-
rial infection). Cases excluded from the study were non-
TBNA procedures, malignancies, and cases without avail-
able cytological slides or sufficient clinical information
and those without appropriate consent.
Bronchoscopy and TBNA
Flexible bronchoscopy was done under local anaesthesia
and TBNA of mediastinal and hilar lymph node stations
with 21G needles without endobronchial ultrasound. ROSE
was done in real time by a cytopathologist in the bronchos-
copy theatre. Two slides were produced per pass, one spray
fixed (Fencott, Sangene Products, Cape Town, South
Africa) for an alcohol-based modified Papanicolaou (Pap)
stain and one air-dried for a Wright-Giemsa-based rapid
stain (Rapidiff, Clinical Sciences Diagnostic, Southdale,
South Africa). Additional samples for ancillary investiga-
tions were requested as deemed necessary based on the pre-
liminary on-site diagnosis.
Cytopathology
All cytological slides were retrieved and retrospectively
reviewed by DDW, ML, and CAW who were blinded to
any diagnosis issued beforehand. Specific cytological fea-
tures were evaluated and graded according to a simple
and reproducible system based on the presence, quality,
and quantity of granulomas (Table I). A discrete granu-
loma was defined as a cluster of epithelioid histiocytes
with or without multinucleated giant cells (Fig. 1) and a
vague granuloma as loose epithelioid histiocytes with or
without multinucleated giant cells (Fig. 2). We also
scored the presence of multinucleated giant cells and the
presence and quantity of necrosis (Fig. 3), lymphocytes
and neutrophils. Ziehl-Neelsen stains on cytological slides
were reviewed for the presence of organisms in the
laboratory.
Statistics and Ethics
A final case diagnosis was established in interdisciplinary
collaboration between pathologists and clinicians by
reviewing microbiological culture and PCR results from
Diagnostic Cytopathology, Vol. 43, No 9 697
Diagnostic Cytopathology DOI 10.1002/dc
DE WET ET AL.
Fig. 1. Discrete granuloma, papanicolaou stain (3100). [Color figure can
be viewed in the online issue, which is available at wileyonlinelibrary.
com.]
Fig. 2. Vague granuloma, Papanicolaou stain (3100). [Color figure can
be viewed in the online issue, which is available at wileyonlinelibrary.
com.]
the TBNA and any additional cytological and histological
information available from the procedure. This was com-
plemented with the patients’ medical history, current case
notes, and where available, follow-up information col-
lected after the procedure. For statitistical evaluation
semiquantitiative scales were dichotomized for necrosis
into absent (scores 0) or present (scores 11 to 31) and
for the cell counts into absent (scores 0 and 11) or pres-
ent (scores 21 and 31). Comparisons were made with v2
using Fisher exact testing for small numbers. Ethics
approval for this study (N10/08/251) was obtained from
the Stellenbosch University Health Research Ethics
Committee.
Results
Patients and Diagnosis
Eighty-one cases of TBNA with granulomatous inflamma-
tion were identified within the study period. Of these
698 Diagnostic Cytopathology, Vol. 43, No 9
Fig. 3. Amorphous necrosis, Ziehl-Neelsen stain (3200). [Color figure
can be viewed in the online issue, which is available at wileyonlinelibrary.
com.]
patients 36 were male (44%). Thirteen were aged 14
years or younger (16%). Nine of 43 patients with known
HIV-status were HIV-positive (21%). The indication for
TBNA was for diagnostic workup of enlarged mediastinal
or hilar lymph nodes suspected to represent sarcoidosis,
tuberculosis, or lymphoma in 64 patients (79%), sus-
pected carcinoma in 9 (11%), unknown systemic disease
in 5 (6%) and 3 had suspected tuberculosis not respond-
ing to treatment (4%). Eventually, tuberculosis was diag-
nosed in 37 patients (46%) of whom 12 were younger
than 14 years of age (32%). Tuberculosis was confirmed
by a positive ZN and/or microbiological culture on a
TBNA sample in 25 cases (68%). In 4 cases (11%) the
diagnosis was made microbiologically from other speci-
mens (gastric washing: 1; histology: 1; sputum: 2), and 8
cases (21%) were diagnosed clinically including confir-
mation by appropriate response to treatment. Sarcoidosis
was diagnosed in 40 cases (49%) of whom all were older
than 14 years of age. A specific diagnosis could not be
made in 3 patients (4%), all older than 14 years of age,
while in one paediatric case (1%) a diagnosis of fibrosing
mediastinitis was made. All 9 HIV-positive patients were
diagnosed with tuberculosis.
Cytopathology
Discrete and vague granulomas were present in all 81
cases (Table II). Lymphocytes were found in almost half
of all cases, necrosis in a quarter and neutrophils were
infrequently seen. Numerous discrete granulomas were
more frequently seen in sarcoidosis (P 5 0.03), while
necrosis (P < 0.001) and neutrophils (P 5 0.05) were
associated with tuberculosis. Vague granulomas, multi-
nucleated giant cells and lymphocytes did not help to dis-
criminate between tuberculosis and sarcoidosis. In the
subgroup of cases in which the diagnosis of tuberculosis
could be confirmed bacteriologically from the TBNA

Table II. Cytomorphological Findings
All N 5 81 Tuberculosis n 5 37
Discrete granulomas 81 (100%) 37 (100%)
>5 granulomas 24 (30%) 7 (19%)
Vague granulomas 81 (100%) 37 (100%)
>5 granulomas 5 (6%) 2 (5%)
Multinucleated giant cells presenta 5 (6%) 2 (5%)
Necrosis presenta 27 (22%) 25 (68%)
Lymphocytes presenta 40 (49%) 17 (46%)
Neutrophils presenta 5 (6%) 4 (11%)
a
Present 5 21 or 31, see Table I. P 5 between tuberculosis and sarcoidosis (n 5 77).
sample (n 5 25) necrosis was present in 19 (76%), the
Ziehl-Neelsen stain was positive in 17 (68%) and the cul-
ture in 18 (72%).
Discussion
The diagnostic yield of TBNA varies from 20 to 90% in
reported series and is related to the size and location of
the lesion as well as the experience of the operator, 4,5
with results improving over time with practice and expe-
rience.3 ROSE performed in real time by a cytopatholo-
gist has proven to be accurate during TBNA and
increases the value of the procedure.3,4,9 In some cases it
may shorten the procedure and lower the overall cost as
it reduces the number of passes required and allows for
termination of the procedure on diagnosis. In an environ-
ment with high rates of tuberculosis and HIV-
coninfection sputum-negative tuberculous lymphadenitis
contributes considerably to the diagnostic burden. It is
thus essential that cytologists (and cyto-technologists) are
able to identify granulomatous inflammation and to triage
the material for the necessity of specific additional inves-
tigations. A typical, discrete granuloma is defined as a
cluster of epithelioid histiocytes with or without the pres-
ence of multinucleated giant cells. Although typical dis-
crete granulomas are regularly found, vague granulomas
should also be recognised as a loose cluster of epithelioid
histiocytes with or without multinucleated giant cells. The
presence of granulomatous inflammation indicates a spe-
cific differential diagnosis, which is well established.7,8
The main cause specific to our patient population (West-
ern Cape Province, South Africa) is tuberculosis which
can cause intrathoracic lymph node pathology similar to
lymphoma or sarcoidosis in the absence of lung disease.
Those principles, however, might not be generalizable to
settings with lower TB prevalence such as North America
or most parts of Europe.
The presence of necrosis has been confirmed as a stat-
istically significant positive finding associated with tuber-
culosis (P < 0.001). The absence of necrosis however,
does not exclude the possibility of tuberculosis as 24% of
ZN and/or culture positive tuberculosis cases did not
Diagnostic Cytopathology DOI 10.1002/dc
MEDIASTINAL GRANULOMATOUS LYMPHADENITIS
Sarcoidosis n 5 40 Fibrosis n 5 1 Unknown n 5 3 P
40 (100%) 1 (100%) 3 (100%)
16 (40%) 0 (0%) 1 (33%) 0.03
40 (100%) 1 (100%) 3 (100%)
3 (8%) 0 (0%) 0 (%) 0.33
2 (5%) 0 (0%) 1 (33%) 0.38
1 (2.5%) 0 (0%) 1 (33%) <0.001
21 (53%) 0 (0%) 2 (67%) 0.15
0 (0%) 0 (0%) 1 (33%) 0.05
show any necrosis. The presence of neutrophils was infre-
quently found but was associated with tuberculosis
(P 5 0.05) as was HIV-status, as all HIV-positive cases
were eventually diagnosed with tuberculosis. Confirming
the diagnosis of tuberculosis relies on identification of the
organism on Ziehl-Neelsen stained smears or growth of
Mycobacterium species in culture. We were able to iden-
tify organisms in about 56% of culture-positive TBNA
samples, which is higher than the 20% reported in the lit-
erature.10 Identification of organisms may be dependent
on the immune status of the patient, however our series
did not confirm this as four of nine (44%) HIV-positive
tuberculosis cases and seven of 15 (47%) HIV-negative
tuberculosis cases had ZN-positive aspirates.
Sarcoidosis is a multisystem disorder of unknown aeti-
ology and there is no single diagnostic gold standard test.
The diagnosis is one of exclusion and is based on identifi-
cation of granulomatous inflammation with noncaseating
granulomas usually accompanied by multinucleated giant
cells; clinical picture and chest radiographic findings
compatible with sarcoidosis; elevated serum angiotensin
converting enzyme (ACE), lysozyme; and response to
treatment.11,12 The diagnostic accuracy of TBNA cytol-
ogy for sarcoidosis varies between 65 and 76% in differ-
ent studies.2,3,7 Trisolini and Wang suggest that the high
density of granulomas in sarcoid lymph nodes is responsi-
ble for the high percentage of positive granulomas in
cytology samples.2 Our results confirm their suggestion as
40% of our sarcoidosis cases showed >5 well formed
granulomas versus 19% of tuberculosis cases (P 5 0.03).
All agents known to cause a granulomatous response
must be excluded before the diagnosis of sarcoidosis can
be made according to criteria of the World Association
and Other Granulomatous Disease (WASOG).12
After exclusion of all the known entities associated
with granulomatous inflammation, there may be cases in
which a specific diagnosis cannot be confirmed. Our study
included three cases, all adult patients, in which ancillary
investigations could not identify a specific cause for the
lymphadenopathy with granulomatous inflammation. Of
these patients, two had associated medical conditions
Diagnostic Cytopathology, Vol. 43, No 9 699
Diagnostic Cytopathology DOI 10.1002/dc
DE WET ET AL.
including infiltrating duct carcinoma of the breast (40-
year-old woman) and erosive lichen planus on immuno-
suppressive therapy (26-year-old man). The clinical follow-
up was thorough but the usual limitations of retro- spective
studies apply. Granulomatous inflammation can also occur
in infections such as Mycobacterium avium- intracellulare,
Histoplasma, and Cryptococcus, foreign body reactions and
malignancies.
In conclusion this study confirms that granulomatous
disease can reliably be diagnosed on TBNA and that
cytomorphological features such as the presence of necro-
sis and neutrophils are associated with tuberculosis
whereas sarcoidosis more frequently presents with dis-
crete granulomas without a necrotic background. Recogni-
tion of these features during ROSE will help guide the
cytopathologist in requesting ancillary investigations in
order to differentiate specific underlying causes of the
inflammatory reaction pattern. Mycobacterial stains and
cultures should be strongly considered when granuloma-
tous inflammation is found in high burden areas of tuber-
culosis and HIV-infection.
References
1. Cetinkaya E, Yildiz P, Kadakal F, Tekin A, Soysal F, Elibol S
Yilmaz V. Transbronchial needle aspiration in the diagnosis of
intrathoracic lymphadenopathy. Respiration 2002;69:335–338.
2. Smojver-Ježek S, Peroš-Golubičić T, Tekavec-Trkanjec J,
Mažuranić I Alilović M. Transbronchial fine needle aspiration
700 Diagnostic Cytopathology, Vol. 43, No 9
cytology in the diagnosis of mediastinal/hilarsarcoidosis. Cytopa-
thology 2007;18:3–7.
3. Trisolini R, Tinelli C, Cancellieri A, et al. Transbronchial needle
aspiration in sarcoidosis: Yield and predictors of a positive aspirate.
J Thorac Cardiovasc Surg 2008;135:837–842.
4. Baram D, Garcia RB Richman PS. Impact of rapid on-site cytologic
evaluation during transbronchial needle aspiration. Chest 2005;128:
869–875.
5. Herth FJ, Becker HD Ernst A. Ultrasound-guided transbronchial
needle aspiration—An experience in 242 patients. Chest 2003;123:
604–607.
6. http://www.westerncape.gov.za/news/world-tb-day-24-March-2012.
7. Bilaçeroğlu S, G€unel O, Eris N, Cağirici U, Mehta AC. Transbron-
chial needle aspiration in diagnosing intrathoracic tuberculous
lymphadenitis. Chest 2004;126:259–267.
8. Ioachim HL, Medeiros LJ. Sarcoidosis lymphadenopathy. In: Ioa-
chim’s lymph node pathology. Philadelphia: Wolters Kluwer; 2009.
p 203–212.
9. Diacon AH, Schuurmans MM, Theron J, Louw M, Wright CA,
Brundyn K, Bolliger CT. Utility of rapid on-site evaluation of
transbronchial needle aspirates. Respiration 2005;72:182–188.
10. Wright CA, Warren RM, Marais BJ. Fine needle aspiration biopsy:
An undervalued diagnostic modality in paediatric mycobacterial
disease. Int J Tuberc Lung Dis 2009;13:1467–1475.
11. Cetinkaya E, Yildiz P, Altin S, Yilmaz V. Diagnostic value of
transbronchial needle aspiration by wang 22-gauge cytology needle
in intrathoracic lymphadenopathy. Chest 2004;125:527–531.
12. Koo V, Lioe TF, Spence RAJ. Fine needle aspiration cytology
(FNAC) in the diagnosis of granulomatous lymphadenitis. Ulster
Med J 2006;75:59–64.