INT J TUBERC LUNG DIS 17(6):787793

2013 The Union

http://dx.doi.org /10.5588/ijtld.12.0892

Can tuberculous pleural effusions be diagnosed by pleural

fluid analysis alone?

S. A. Sahn,* J. T. Huggins,* M. E. San Jos, J. M. lvarez-Dobao, L. Valds

* Division of Pulmonary, Critical Care, Allergy and Sleep Medicine, Medical University of South Carolina, Charleston,
South Carolina, USA; Servico de Analisis Clinicos and Servicio de Neumologia, Complexo Hospitalario Clinico,
Universitario de Santiago, Santiago de Compostela, A Corua, Spain

SUMMARY

O B J E C T I V E : To assess whether pleural fluid analysis pH, cholesterol, triglycerides, adenosine deaminase
(PFA) can confidently diagnose tuberculous pleural effu- (ADA), and total percentages of lymphocytes, neutrophils
sion (TPE). and macrophages when TPEs were compared to all
M E T H O D S : PFA of 548 TPEs was performed between other groups. Of the TPEs, 99.1% were exudates. Pleural
January 1991 and December 2011. The control group fluid protein 5.0 g /dl, lymphocytes > 80% and ADA
consisted of patients with malignant PE (MPE), compli- > 45 U/l were diagnostic of TPE, with a specificity of
cated parapneumonic/empyema (infectious) PE (IPE), 100%, a sensitivity of 34.9% and an area under the
miscellaneous PE (MisPE) and transudative PE (TrPE). curve of 0.975.
R E S U LT S : The PFA of 548 histologically or culture- C O N C L U S I O N : PFA alone was diagnostic in one third
positive consecutive cases of TPE was compared with of the TPE cases, with a high probability in nearly
that of 158 consecutive cases of MPE, 113 cases of IPE, 60%.
37 cases of MisPE and 115 cases of TrPE. Statistically sig- K E Y W O R D S : adenosine deaminase; ADA; total pro-
nificant differences were noted in pleural fluid glucose, tein; lymphocytosis

IN SOME geographical areas, tuberculous pleural of the PF. In TPE, the clinician may not have access to
effusion (TPE) is the most common cause of pleural ef- the diagnostic procedures mentioned above, and PF
fusion (PE).1 PEs are the most common manifestation culture is often negative.
of extra-pulmonary tuberculosis (TB) in Galicia, The aim of this study was to assess whether it is
Spain,2 although globally it is considered to be the possible to diagnose TPE based on pleural fluid anal-
second most common, exceeded only by lymphatic ysis (PFA) in isolation.
involvement.3
A definitive diagnosis of TPE requires a positive
MATERIAL AND METHODS
culture from pleural fluid (PF) or pleural tissue with
histological presence of pleural granulomas.4 How- All patients with documented TPE consecutively di-
ever, in some cases it is difficult to establish a diagno- agnosed between 1 January 1991 and 31 December
sis. TPE results from a delayed hypersensitivity reac- 2011 were included in the study. The diagnosis of
tion to mycobacterial antigens that have entered the TPE was substantiated by previously established cri-
pleural space.5 It is likely that tuberculous antigens teria.7,8 The study was performed in a university gen-
enter the pleural space following the leakage or rup- eral hospital in Galicia in the northwest of Spain,
ture of a subpleural focus of disease.6 At times this where the incidence of TB in 1996 (the first year for
equates to a small number of mycobacteria that enter which reliable data are available) was 72.3 per
the pleural cavity; for this reason, the yield of acid-fast 100 000 population9 and 28/100 000 in 2010,2 with
bacilli (AFB) stain and culture may be low. Pleural bi- an incidence of TPE of 4.8/100 000 in 2009.10
opsy is not readily available in most hospitals, and even The TPE cases were divided into four periods ac-
when pleural biopsies are performed the mycobacterial cording to the year of diagnosis (period I: 1991
yield does not exceed 80%, and the tuberculin skin test 1996; period II: 19972001; period III: 20022006
can be negative in up to one third of patients.7 and period IV: 20072011) to determine whether
Clinicians therefore need to explore other diag- there were significant differences between the study
nostic options, such as analysing the characteristics periods.

Correspondence to: Steven A Sahn, Division of Pulmonary, Critical Care, Allergy and Sleep Medicine, Medical University of
South Carolina, 96 Jonathan Lucas Street, CSB 812, Charleston, SC 29425, USA. Tel: (+1) 843 792 3167. Fax: (+1) 843
792 0732. e-mail: sahnsa@musc.edu
Article submitted 27 November 2012. Final version accepted 24 January 2013.
788 The International Journal of Tuberculosis and Lung Disease

The PEs from the database for the years 2007 used to identify the threshold values of the analytical
2011 were used as controls and were classified as ma- parameters. This value corresponds to the point in
lignant PE (MPE), infectious PE (empyema and para- the ROC curve farthest from the diagonal line, and
pneumonic effusions, IPE), miscellaneous PE (MisPE) the statistical programme calculated a bootstrapped
and transudative PE (TrPE). A PE was classified as tu- 95% confidence interval for both the Youden index
berculous if Ziehl-Neelsen stain or Lwenstein-Jensen and its corresponding threshold. In the case of sev-
(LJ) culture was positive in PF or on biopsy, or if eral combined parameters (ADA + % lymphocytes
granulomas were identified on biopsy; malignant or ADA + % lymphocytes + protein), the cut-offs
only when confirmed by positive cytology in PE or were chosen to maximise specificity.
pleural biopsy; parapneumonic in the case of bacte- Statistical analyses were performed using MedCalc
rial pneumonia, lung abscess or bronchiectasis, or if 12.2.1.0 software (MedCalc, Ostend, Belgium).
PF culture was positive; and empyema if the fluid was
purulent. Other diagnoses were based on clinical, ra- Ethics approval
diographic and PF findings.7,8 All patients provided signed informed consent before
The data collection process was determined before any procedure was performed (chest computed to-
beginning the study. PF and peripheral blood samples mography scan, thoracocentesis or pleural biopsy).
were obtained on the patients initial visit after fasting, The protocol was evaluated and approved by the
and a closed pleural biopsy was performed using a Clinical Research Ethics Committee of Galicia (Reg-
Cope or Abrams needle (Grupo Taper S A, Alcobnedas, istry 2012/215).
Madrid, Spain). PF samples were sent to the biochem-
istry, cytology and microbiology laboratories for cul-
RESULTS
ture in liquid and LJ media. PFA, which included total
protein (g /dl), lactate dehydrogenase (LDH; IU/l), The flow charts for the patients with TPE and the
glucose (mg /dl), pH, cholesterol (mg /dl), triglycerides control group are shown in Figure 1. Of the 548 pa-
(mg /dl), adenosine deaminase (ADA; U/l) and total tients with TPE, 60% were male. PF AFB stain and
nucleated and differential cell counts, was performed culture were positive in 6% and 36% of patients, re-
for all patients. Given its low prevalence in our spectively. Table 1 summarises the sensitivity of each
community, human immunodeficiency virus (HIV) of the criteria used for a definitive diagnosis of TBE.
serology was only examined if there were risk factors Of the 548 patients, 393 (71.7%) were aged <40 years,
or clinical suspicion.11 with 187 (85.4%), 89 (64.5%), 73 (70.9%) and 44
All biochemical measurements were performed us- (50%) diagnosed in periods I, II, III and IV,
ing standard methodology. The analytical systems respectively.
underwent technological updates during the study, but A complete PFA was performed on all 548 pa-
before any modifications were made to the technique tients. The median and 5th and 95th percentiles of all
of a biological parameter or the analysers used, the parameters are summarised in Table 2. Significant
reproducibility of the results was analysed to minimise differences were found only in age, PF:LDH and PF/S
any changes that may have occurred in the historical (serum):LDH upper limits of normal (ULN) ratios
controls. The laboratory manager was the same throughout the study (Table 2). Only 38 patients
throughout the study (ESJ). (6.9%) had a PF protein concentration < 3.0 g /dl, 72
ADA activity (U/l 37C) was determined colori- (13.1%) had a PF LDH concentration < 210 IU/l (2/3
metrically by the Galanti and Giusti method.12 Activ- of the ULN in plasma), 17 (3.1%) had a PF/S:protein
ity was considered to be increased if values were ratio < 0.5, and 95 (17.3%) had a PF/S:LDH ratio
45 U/l. Neutrophilic and lymphocytic effusions were < 0.82. Only five (0.9%) patients met the biochemical
defined as effusions with a neutrophil/lymphocyte criteria for a transudate.14 Twenty-one of the 548
count >50% of the total nucleated cell count. Only patients (3.8%) had total nucleated cell counts of
the first PF chemistry panel was used for statistical < 300/l and in 54 (9.9%) they were > 6000/l. The
analysis in patients who had had more than one thora- vast majority of the patients (96.7%, 530/548) had
cocentesis. Age and sex were recorded. PF lymphocytes 50% and for 59.1% (324/548)
they were 80%. Only 2.9% (16/548) of the TPE
Statistical analysis patients had pleural fluid pH < 7.30. Glucose values
Normality of distribution was checked using the < 40 mg /dl were observed in 6% of the patients
DAgostino-Pearson test. Quantitative variables across (33/548), of whom only three also had pH < 7.30
the different time periods were compared using the and 20.6% (113/548) had glucose levels < 60 mg /dl.
Kruskal-Wallis test. 2 analysis was used to compare Although 19 patients (3.5%) had cholesterol values
frequencies of categorical covariates. The receiver op- in the transudate range (< 55 mg /dl), only two met
erating characteristic (ROC) curve analyses, as well the traditional criteria (PF/S:protein ratio, PF:LDH
as their comparisons, were performed using the and PF/S:LDH ULN ratio).14 Only one patient had
method of DeLong et al.13 The Youden index was PF triglyceride levels > 110 mg /dl (340 mg /dl). The
 Diagnosis of TB pleural effusion 789

Figure 1 Patient flow chart of patients with TPE (A) and the control group (B).
TPE = tuberculous pleural effusion.

vast majority of the patients (97.6%, 535/548) had were lower among TBE patients than in the three ex-
ADA levels 45 U/l. udate groups and higher in the transudates. Table 4
The control group consisted of 423 cases of non- shows the diagnostic yield among the TPE cases for
tuberculous PE: MPE (n = 158), IPE (n = 113), all parameters studied. The three largest areas under
MisPE (n = 37) and TrPE (n = 115). The origin of the curve (AUCs) of the individual parameters were
the MPEs and the aetiology of the MisPEs are shown obtained with ADA, lymphocytes and glucose con-
in Table 3. The transudates were secondary to heart centration (0.979, 0.898 and 0.822, respectively).
failure (n = 104), chronic liver disease (n = 10), and The combinations with the highest yields were PF
hypoalbuminaemia (n = 1). lymphocytes 80% and PF ADA 45 U/L (sensitiv-
The TPE patients had significantly higher lympho- ity 58.4%, specificity 99.5%, positive likelihood ratio
cyte, cholesterol and triglyceride levels, and signifi- [PLR] 111.9, negative likelihood ratio [NLR] 0.42
cantly lower glucose levels than the control group. and AUC 0.974). With a PF protein 5.0 g /dl, PF
There were also differences in pH between all the groups lymphocytes 80% and PF ADA 45 U/L, the sensi-
(significantly higher among TBE patients than among tivity was 34.9%, specificity 100%, PLR > 150, NLR
IPE cases and lower in the remainder). Triglycerides 0.65 and AUC 0.975. Figure 2 shows the AUC ROC
of PF ADA and the two parameter combinations used;
no significant differences were observed (Table 4).
Table 1 Sensitivity of each of the methods used for a
definitive diagnosis of tuberculous pleurisy
Methods n (%)
DISCUSSION
ZN stain of pleural fluid 33/548 (6) In the largest series of TPE cases published to date,
Culture of pleural fluid in L J medium 199/548 (36.3) our results confirm that PFA can accurately diagnose
ZN stain of biopsy tissue 126/517 (24.3)
Culture of biopsy tissue in L J medium 273/517 (52.8)
up to one third of TPE cases, based on an ADA
Observation of caseating granulomas 401/517 (77.6) 45 U/l, lymphocytes 80% and total protein
ZN stain of sputum 62/104 (59.6) 5.0 g /dl, and that it has a high diagnostic probabil-
Culture of sputum in L J medium 104/104 (100)
ity in 60% of cases, based on ADA and percentage
ZN = Ziehl-Neelsen; L J = Lwenstein-Jensen. of lymphocytes.
 790 The International Journal of Tuberculosis and Lung Disease

An increase in the mean age of the TPE cases was

TPE = tuberculous pleural effusion; MPE = malignant pleural effusion; IPE = infectious pleural effusion; MisPE = miscellaneous pleural effusion; TrPE = transudative pleural effusion; PF = pleural fluid; S = serum; LDH = lactate dehy-
P value*
<0.001
0.028
0.009
<0.001
<0.001
<0.001
<0.001

<0.001
<0.001
<0.001
<0.001
0.024
<0.001
<0.001
observed during the study, from 34 years in period I
to 43 years in period IV (P < 0.001). It has been sug-
TrPE (n =115)

gested that at older age at onset, the development of

2.53 (1.254.21)
0.42 (0.210.65)

7.42 (7.107.60)
TPE is a result of reactivation, while it could be a pri-

0.38 (0.20.78)
120 (63.5250.4)

570 (1702894)
120 (82.5225)
percentiles)

mary form of the disease in younger individuals.15 It

(5th95th

80 (54.591)

30 7.561.7)
37 (1379.2)
Median

15 (259.6)

18 (537.4)
14 (521.1)
42 (1080)
is possible that in our region there was a changing
frequency of primary TB7 (period I) compared to a
higher number of cases involving reactivation of dis-
ease (period IV).
We compared TPE cases diagnosed over a period
P value*
0.734
0.418
0.027
0.281
0.276
<0.001
<0.001

0.025
0.001
<0.001
<0.001
<0.001
<0.001
<0.001
of 21 years with a control group of all cases of non-
tuberculous PE in our database from the last 5 years.
Although the possibility of bias cannot be excluded,
MisPE ( n = 37)

by comparing the different PF parameters obtained

298 (115.41122.5)
0.64 (0.440.85)

0.93 (0.363.51)

7.40 (7.117.58)

2350 (66410 510)

102 (56.2172.7)

33.5 (9.72780)

in different time periods, this appears to be low; min-

61 (28.588.3)

26 (10.770.3)
4.2 (2.75.5)

87 (28.8151)
percentiles)

22 (9.360.8)
(5th95th

imal differences were found in the analytical markers

Median

25 (272.5)
26 (856.6)

for TPE in the different time periods studied, except

for the PF:LDH and PF/S:LDH ULN ratios. Further-
more, these two parameters do not follow a trend
over time: there were lower levels in period III than
in the other periods. Therefore, the inclusion of all
P value*
0.127
0.003
0.408
<0.001
<0.001
<0.001
<0.001

<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
0.269

TPE cases in a single group gives the study greater

power to detect differences in the analytical markers
compared to the non-TPE cases.
IPE (n = 113)

PFA of TPE has well defined characteristics, and

384.0 (120.28685.7)
1.20 (0.3827.14)
0.66 (0.480.86)

7.40 (6.807.59)

78.0 (45.0125.6)
35.0 (14.5115.5)

our results are in agreement with those published

4.4 (2.725.50)

100.0 (9.5246.1)

26.0 (12.097.4)
3300 (28632 967)

15.0 (3.0 44.7)

21.0 (3.082.6)
47.0 (4.090.0)
68 (39.088.9)
percentiles)

previously.4,7,1622 Of the TPEs, 99.1% were exu-

(5th95th
Median

dates, 97.6% had pH > 7.30, 80.2% had glucose

> 60 mg /dl, 52% had total protein 5.0 g /dl, 90%
had nucleated cell count < 6000/l, 96.3% had lym-
phocytes > 50%, 96.5% had cholesterol 55 mg /dl
Descriptive statistics of the studied parameters for each pleural effusion group

and 97.6% had ADA 45 U/l. The value of each in-

dividual test alone in the diagnosis of TPE is low, ex-
P value*
<0.001
<0.001
0.627
<0.001
<0.001
<0.001
<0.001

0.170
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001

drogenase; IU = international units; ULN = upper limits of normal; ADA = adenosine deaminase.

cept for ADA, lymphocyte percentage and glucose

concentration (Table 4). The two most significant
MPE (n = 158)

1.1 (0.395 4.45)

0.67 (0.480.83)

7.40 (7.207.60)
348 (128.61446)

101 (29.2182.4)

85 (41.6139.4)
4.5 (2.925.7)

1715 (3167704)

Table 3 Origin of malignant PE (left) and aetiology of

30 (10137.3)
percentiles)
73 (50.687)

55 (1283.8)
(5th95th

miscellaneous PE (right)
14 (160.4)
18.5 (455)
Median

23 (11 43)

Malignant PE n Miscellaneous PE n
Lung 85 Traumatic haemothorax 10
Lymphoma 21 Viral pleuropericarditis 8
Breast 14 Systemic lupus
Mesothelioma 6 erythematosus 4
1.52 (0.47 4.73)
0.73 (0.550.89)

7.35 (7.317.40)

Colon 5 Pancreatitis 3
486 (1521512)

2678 (5388200)

Pancreas 5 Post-abdominal surgery 3

5.0 (2.86.5)
percentiles)
(5th95th
(n = 548)

93 (61130)

109 (50192)

Stomach 4 Pulmonary embolism 2

Median

27 (1574)

72 (3896)

82 (5495)

32 (1364)
9 (236)
6 (020)
TPE

Ovary 3 Subphrenic abscess 2

Kidney 1 Sarcoidosis 2
Prostate 1 Porphyria 1
Oesophagus 1 Rheumatoid arthritis 1
* TBE are the comparison group.

Sarcoma 1 Chylothorax of
Bladder 1 unknown origin 1
Chronic lymphocytic
PF triglycerides, mg /dl
PF cholesterol, mg /dl

leukemia 1
cell count, cells/l

PF macrophages, %
PF/S LDH ULN ratio

PF lymphocytes, %
PF total nucleated

Malignant fibrous
PF neutrophils, %
PF/S protein ratio

PF glucose, g /dl
PF protein, g /dl

histiocytoma 1
Liver 1
PF LDH, IU/l

PF ADA, U/l
Age, years
Parameter

Non-associated 7
Table 2

Total 158 37
PF pH

PE = pleural effusion.
 Diagnosis of TB pleural effusion 791

Table 4 Diagnostic yield for tuberculous pleural effusion for all parameters studied (individual and combinations)

Sensitivity Specificity AUC

Parameter Cut-off % % PLR NLR Accuracy ROC P value*
PF protein, g /dl >4.68 67.7 73.3 2.53 0.44 70.3 0.751 <0.0001
PF/S protein ratio >0.67 78.7 63.5 2.15 0.34 70.9 0.753 <0.0001
PF LDH, IU/l >313 77.7 60.7 1.98 0.37 70.5 0.697 <0.0001
PF/S LDH ULN ratio* >0.99 77.6 61.0 1.99 0.37 70.0 0.698 <0.0001
PF glucose, g /dl 90 88.5 72.8 3.26 0.16 81.9 0.822 <0.0001
PF pH 7399 96.5 63.5 2.65 0.06 84.3 0.721 <0.0001
PF total nucleated cell
count, cells/l >1780 66.4 56.0 1.51 0.60 62.3 0.594 <0.0001
PF lymphocytes, % >64 89.1 76.4 3.77 0.14 83.7 0.898 <0.0001
PF neutrophils, % 11 61.9 65.5 1.79 0.58 63.5 0.677 <0.0001
PF macrophages, % 11 84.5 76.1 3.54 0.20 81.3 0.867 <0.0001
PF cholesterol, mg /dl >74 82.3 53.0 1.75 0.33 69.6 0.696 <0.0001
PF triglycerides, mg /dl >21 81.6 32.9 1.22 0.56 73.6 0.536 <0.0001
PF ADA, U/ l 44 97.6 93.1 14.4 0.03 95.8 0.979
PF lymphocytes, % 80 58.4 99.5 111.9 0.42 75.6 0.974 0.1378
and PF ADA, U/ l 45
PF protein, g /dl + 5 34.9 100 >150 0.65 61.9 0.975 0.1479
PF lymphocytes, % 80
and PF ADA,U/ l 45

* With respect to PF ADA.

PLR = positive likelihood ratio; NLR = negative likelihood ratio; AUC = area under the curve; ROC = receiver operating characteristics; PF = pleural fluid; S =
serum; LDH = lactate dehydrogenase; ULN = upper limits of normal; ADA = adenosine deaminase.

parameter combinations were PF lymphocytes and The diagnosis of TPE using PFA has not to date
PF ADA. PF lymphocytes 80% and PF ADA been evaluated in patients aged >40 years. At this age,
45 U/l have a specificity of 99.5% and a sensitivity the higher incidence of malignant effusions or systemic
of 58.4% for the diagnosis of TPE. Only two non- diseases that can involve an increased ADA, as with
TPE cases met both criteria (one lymphoma MPE and other non-tuberculous PE cases,26 can make the diag-
one IPE of long duration). If a third parameter was nosis problematic. An ADA lower than the cut-off
included (PF protein 5.0 g /dl), the specificity for a point does not exclude TPE, as occurred in 13 cases
diagnosis of TPE was 100%, which occurred in in our series, but a PE with repeatedly low ADA values
34.9% of the cases. would be unlikely.27 In our series ADA was the most
A PF ADA 45 U/l is highly suggestive of TPE.23 important parameter for diagnosing TPE, due to its
The vast majority (97.6%) of the TPE cases had levels high individual diagnostic yield, and when included in
45 U/l, as did 29 non-TPE cases (4 malignant [MPE], the two parameter combinations with higher yields.
22 infectious [IPE], 2 systemic lupus erythematosus and More than half (59%) of the patients had PF lym-
1 haemothorax [MiPE]), with a specificity of 93.1%. phocytes 80%. The differential diagnosis of PE
In young patients and in regions with a high preva- with this lymphocyte percentage is well established.28
lence of TB, the presence of a pleural exudate with Some of these diseases are easily excluded after a de-
high ADA concentration and PF lymphocytes 80% tailed history and PFA. Only 60% of patients with a
is highly suggestive of TPE, to the extent that pleural non-lymphoma metastatic PE would have PF lym-
biopsy may be superfluous.24,25 phocytes > 50% and the level would rarely reach

Figure 2 Receiver operating characteristics of PF ADA, PF lymph + PF ADA and PF protein + PF lymph + PF ADA. PF ADA = pleural
fluid adenosine deaminase; PF lymph = pleural fluid lymphocytes (%); AUC = area under the curve.
792 The International Journal of Tuberculosis and Lung Disease

80%;29 in our series, this was 10.2% (14/137) in the References

non-lymphoma MPE and 14.3% (3/21) in the lym- 1 Valds L, lvarez D, Valle J M, Pose A, San Jos E. The etiology
phomas. Therefore, in the case of PE with PF lym- of pleural effusions in an area with high incidence of tubercu-
phocytes 80% (59% in our series), TPE would losis. Chest 1996; 109: 158162.
have to be considered. However, neutrophils can 2 Informe da tuberculose en Galicia. Caractersticas dos casos de
tuberculose de Galicia no ano 2010. Evolucin do perodo 1996
be the predominant cell early on in the course of 2010. Versin ampliada 29.11.2011. Galicia, Spain: Xunta de
TPE.30,31 On combining the PF lymphocyte percent- Galicia, 2011. http://www.sergas.es/gal/DocumentacionTecnica/
age with the PF ADA, IPE that also had high ADA docs/SaudePublica/Tuberculose/Informe%20TB%202010_
levels are excluded, as they are typically predomi- ampliado_galego.pdf Accessed March 2013. [Galician]
nantly neutrophils. 3 Baumann M H, Nolan R, Petrini M, et al. Pleural tuberculosis
in the United States. Incidence and drug resistance. Chest 2007;
The mean PF protein concentration was 4.9 g /dl;
131: 11251132.
52% of the TPE cases had values 5.0 g /dl (24.7% 4 Sahn S A. State of the art: the pleura. Am Rev Respir Dis 1989;
in the other exudates). Although it has been reported 138: 188234.
previously that protein levels > 5.0 g /dl are sugges- 5 Leibowitz S, Kennedy L, Lessof M H. The tuberculin reaction
tive of TPE,7,21,30 this parameter has never been thor- in the pleural cavity and its suppression by antilymphocyte
oughly investigated. In other high-prevalence exu- serum. Br J Exp Pathol 1973; 54: 152162.
6 Stead W W, Eichenholz A, Stauss H K. Operative and patho-
dates, such as IPE or MPE, the PF protein range is logic findings in twenty-four patients with syndrome of idio-
wider (1.58 g /dl) and generally lower.3234 This pa- pathic pleurisy with effusion, presumably tuberculosis. Am
rameter would tend to exclude the few MPEs with a Rev Tuberc 1955; 71: 473502.
high PF ADA and lymphocyte percentage. 7 Valds L, lvarez D, San Jos E, et al. Tuberculous pleurisy. A
The main limitation of this study is that it was per- study of 254 cases. Arch Intern Med 1998; 158: 20172021.
8 Villena-Garrido V, Ferrer-Sancho J, Hernndez-Blasco L, et al.
formed in an area with a medium prevalence of TB
Diagnosis and treatment of pleural effusion. Arch Bronconeumol
where the predictive value of ADA is high;7 for this 2006; 42: 349372. [Spanish]
reason, these results cannot be extrapolated to low- 9 Cruz-Ferro E, Fernndez-Nogueira E. Epidemiology of tuber-
incidence regions, as in these cases the predictive culosis in Galicia, Spain, 19962005. Int J Tuberc Lung Dis
value of ADA will be low. There were no cases of TPE 2007; 11: 10731079.
with HIV infection in our series. This is partly due to 10 Valds L, Ferreiro L, Cruz-Ferro E, et al. Recent epidemiological
trends in tuberculous pleural effusion in Galicia, Spain. Eur J
the low incidence of HIV in patients with TPE in Intern Med 2012; 23: 727732.
Galicia (2.7%; 50/1835 patients diagnosed over a 11 Informe VIH-SIDA en Galicia. Diagnsticos de infeccin polo
10-year period),10 and also because there is a desig- VIH: 20042009. Casos de Sida: 19842009. Galicia, Spain:
nated clinic for HIV patients. Finally, PFA might offer Xunta de Galicia, 2010. http://www.sergas.es/Publicaciones/
a higher yield if other parameters were determined, DetallePublicacion.aspx?IdPaxina=40008&IDCatalogo=2012
Accessed March 2013. [Galician]
such as interferon-gamma,22 ADA isoform ADA-2,24
12 Giusti G. Adenosine deaminase. In: Bergmeyer H U, ed. Meth-
or interleukin-12 p40,35 which were not included as ods of enzymate analysis. New York, NY, USA: Academic
they are not routinely performed in clinical practice. Press, 1974: pp 10921099.
In summary, the results of this TPE series, the larg- 13 DeLong E R, DeLong D M, Clarke-Pearson D L. Comparing
est published to date, confirm that PFA can confi- the areas under two or more correlated receiver operating
characteristic curves: a nonparametric approach. Biometrics
dently diagnose one third of TPEs, and with a high
1988; 44: 837845.
probability in almost 60% of cases. It is well known 14 Joseph J, Badrinath P, Basran G S, Sahn S A. Is the pleural fluid
that several biochemical parameters are individually transudate or exudate? A revisit of the diagnostic criteria.
elevated in TPE; however, this is the first study to Thorax 2001; 56: 867870.
show the simultaneous use of PF protein 5.0 g /dl, 15 Moudgil H, Sridhar G, Leitch A G. Reactivation disease: the
lymphocyte percentage 80 and ADA 45 U/l in the commonest form of tuberculous pleural effusion in Edinburgh,
19801991. Respir Med 1994; 88: 301304.
diagnosis of TPE. These data are important for our 16 Gopi A, Madhavan S M, Sharma S K, Sahn S A. Diagnosis and
understanding of TPE given: 1) the lack of diagnostic treatment of tuberculous pleural effusion in 2006. Chest 2007;
confirmation by currently accepted methods, even tak- 131: 880889.
ing into account mycobacterial yield of pleural biopsy 17 Sharma S K, Mohan A. Extrapulmonary tuberculosis. Indian J
of 80%; 2) the lack of universal availability to per- Med Res 2004; 120: 316353.
18 Antonangelo L, Vargas F S, Seiscento M, Bombarda S, Teixera L,
form pleural biopsies; and 3) the excessive delay in
Sales R K. [Clinical and laboratory parameters in the differen-
obtaining results of PF culture, which can take up to tial diagnosis of pleural effusion secondary to tuberculosis or
4 weeks. In these circumstances, PFA can provide a cancer]. Clinics (San Paulo) 2007; 62: 585590. [Portuguese]
diagnosis in a significant number of cases. However, 19 Zaric B, Kuruc V, Milovancev A, et al. Differential diagnosis of
it cannot be refuted that in certain patient populations, tuberculous and malignant pleural effusions: what is the role
such as those with HIV disease, and in areas with a of adenosine deaminase? Lung 2008; 186: 233240.
20 Porcel J M. Tuberculous pleural effusion. Lung 2009; 187:
high incidence of drug-resistant TB, TB isolation will 263270.
be crucial to allow for drug susceptibility testing. 21 Light R W. Tuberculous pleural effusions. In: Light R W, ed.
Pleural diseases. 3rd ed. Baltimore, MD, USA: Williams &
Conflict of interest: none declared. Wilkins, 1995: pp 154166.
 Diagnosis of TB pleural effusion 793

22 Valds L, San Jos E, lvarez D, et al. Diagnosis of tuberculous 29 Yam L T. Diagnostic significance of lymphocytes in pleural
pleurisy using the biologic parameters adenosine deaminase, effusions. Ann Intern Med 1967; 66: 972982.
lysozyme, and interferon-. Chest 1993; 103: 458465. 30 Antony V B, Repine J E, Harada R N, Good J T Jr, Sahn S A.
23 Liang Q L, Shi H Z, Wang K, Qin S M, Qin X J. Diagnostic Inflammatory response in experimental tuberculosis pleurisy.
accuracy of adenosine deaminase in tuberculous pleurisy: a Acta Cytol 1983; 27: 355361.
meta-analysis. Respir Med 2008; 102: 744754. 31 Antony V B, Sahn S A, Antony A C, Repine J E. Bacillus
24 Valds L, Alvarez D, San Jos E, et al. Value of adenosine Calmette-Gurin-stimulated neutrophils release chemotaxins
deaminase in the diagnosis of tuberculous pleural effusions in for monocytes in rabbit pleural spaces and in vitro. J Clin
young patients in a region of high prevalence of tuberculosis. Invest 1985; 76: 15141521.
Thorax 1995; 50: 600603. 32 Light R W, MacGregor M I, Luchsinger P C, Ball W C. Pleural
25 Valds L, San Jos M E, Pose A, et al. Diagnosing tuberculous effusions: the diagnostic separation of transudates and exudates.
pleural effusion using clinical data and pleural fluid analysis. A Ann Intern Med 1972; 77: 507513.
study of patients less than 40 years old in an area with a high 33 Chernow B, Sahn S A. Carcinomatous involvement of the
incidence of tuberculosis. Respir Med 2010; 104: 12111217. pleura: analysis of 96 patients. Am J Med 1977; 63: 695
26 Valds L, Pose A, San Jos E, Martnez Vzquez J M. Tubercu- 702.
lous pleural effusions. Eur J Intern Med 2003; 14: 7788. 34 Sahn S A, Good J T Jr. Pleural fluid pH in malignant effusions.
27 Ferrer J S, Muoz X G, Orriols R M, Light R W, Morell F B. Diagnostic, prognostic and therapeutic implications. Ann Intern
Evolution of idiopathic pleural effusion. A prospective, long- Med 1988; 108: 345349.
term follow-up study. Chest 1996; 109: 15081513. 35 Valds L, San Jos E, lvarez-Dobao J M, et al. Diagnostic
28 Sahn S A. The value of pleural fluid analysis. Am J Med Sci value of interleukin-12 p40 in tuberculous pleural effusions.
2008; 335: 715. Eur Respir J 2009; 33: 816820.
 Diagnosis of TB pleural effusion i

RSUM

O B J E C T I F : Evaluer dans quelle mesure une analyse du concernant le glucose, le pH, le cholestrol, les triglyc-
liquide pleural (PFA) peut conduire avec scurit au di- rides, ladnosine daminase (ADA) et les pourcentages
agnostic des panchements pleuraux tuberculeux (TPE). totaux de lymphocytes, de neutrophiles et de macro-
M T H O D E S : On a tudi entre janvier 1991 et dcembre phages du liquide pleural lors de la comparaison des
2011 la PFA de 548 cas de TBPE. Le groupe contrle TBPE tous les autres groupes. Les TPE taient de
tait constitu de patients dont les panchements nature exsudative dans 99,1% des cas. Un taux de pro-
pleuraux taient de nature maligne (MPE), infectieuse tine du liquide pleural 5,0 g /dl, un pourcentage de
(pneumonie ou dempyme, IPE), varie (MisPE) ou des lymphocytes > 80% et un ADA > 45 U/l ont permis
transsudats (TrPE). le diagnostic de TPE avec une spcificit de 100%,
R S U LTAT S : On a compar la PFA de 548 cas conscu- une sensibilit de 34,9% et une aire sous la courbe de
tifs de TPE confirms par lexamen histologique ou la 0,975.
culture par comparaison avec 158 cas conscutifs de C O N C L U S I O N S : La PFA a permis elle seule le di-
MPE, 113 cas dIPE, 37 cas de MisPE et 115 cas de TrPE. agnostic dun tiers des cas de TPE avec une probabilit
On a not une diffrence statistiquement significative leve dans presque 60% des cas.

RESUMEN

O B J E T I V O : Valorar si el anlisis del lquido pleural DPTB con los otros grupos se encontraron diferencias
permite diagnosticar con seguridad los derrames pleu- significativas con la glucosa, el pH, la ADA y los porcen-
rales tuberculosos (DPTB). tajes de linfocitos, neutrfilos y macrfagos. El 99,1%
M T O D O S : Se estudiaron 548 DPTB diagnosticados de los DPTB fueron exudados. Unas protenas en lquido
entre enero de 1991 y diciembre de 2011. El grupo pleural > 5,0 g /dl, un porcentaje de linfocitos > 80% y
control lo componan pacientes con derrames pleurales una ADA > 45 U/l fue diagnstico de TBPEs con una
malignos (MPE), infecciosos (IPE), miscelneos (MisPE) especificidad del 100%, una sensibilidad del 34,9% y
y trasudados (TrPE). una AUC de 0,975.
R E S U LTA D O S : El PFA de 548 casos consecutivos de C O N C L U S I O N E S : El anlisis del lquido pleural fue
TBPEs con histologa o cultivo positivo se compar con diagnstico en la tercera parte de los DPTB, mientras que
158 casos consecutivos de MPE, 113 con IPE, 37 con en un 60% sugera una alta probabilidad diagnstica.
MisPE y 115 casos de TrPE. Cuando se compararon los