Pleural Fluid Biomarkers

Beyond the Light Criteria

José M. Porcel, MD

KEYWORDS
 Pleural effusion  Biomarkers  Natriuretic peptides  Adenosine deaminase  Tumor markers
 Mesothelin  Fibulin-3  C-reactive protein

KEY POINTS
 Accurate diagnosis of the cause of a pleural effusion can be challenging.
 Analysis of soluble biomarkers from effusions may be a useful adjunctive.
 Ideal biomarkers should be both sensitive and specific to the disease state being examined and be
available at the bedside.
 None of the many reported biomarkers for pleural infection have yet been shown to be more effec-
tive than pH in predicting which parapneumonic effusions are complicated.
 The value of pleural fluid tumor markers in diagnosing malignancy remains limited because further
pathologic examination is warranted.

The Light criteria have stood the test of time for Hence, moving beyond the Light criteria using
separating transudates from exudates, a nearly in- effective biomarkers for particular causes may
dispensible first step in the evaluation of patients improve clinical management.
with pleural effusions.1 The rationale behind this A biomarker can be defined as a biological
simple initial discrimination is that, if the patient molecule found in blood, other body fluids, or
has a transudate, the probable cause is heart tissues that is a sign of a normal or abnormal
failure (HF) or, less commonly, cirrhosis, and the process, or of a condition or disease, such as
use of diuretics will solve the problem. In contrast, cancer, infection, or HF.4 Ideal biomarkers are
the finding of an exudate widens the differential easily measured at a reasonable cost (analytical
diagnosis and requires a more precise etiologic validity), must provide information that is not
search, which may include invasive procedures. already available from a routine clinical assess-
Although the Light criteria are extremely sensitive ment (clinical validity), and, eventually, should aid
in identifying exudates, they lack specificity in in decision making (clinical usefulness).4 Although
that about 30% and 20% of HF-associated and there are many emerging pleural fluid biomarkers
cirrhosis-associated pleural effusions are misclas- under study, few fulfill these 3 criteria well enough
sified as exudates.2 Another criticism of the Light to be used clinically. Therefore, this article only
criteria is that, other than grouping pleural effu- discusses those that have consistently shown
sions into transudates and exudates, they do not diagnostic relevance and/or are easily available.
attach a specific label to a disease. Some experts,
including Dr Light, have recommended focusing BIOMARKERS OF HF
research on the identification of specific pleural
disease markers rather than wasting time and HF causes more pleural effusions than any other
resources in transudate-exudate differentiation.3 disease. The finding of a transudate by applying
chestmed.theclinics.com

The author has no potential conflicts of interest.

Pleural Diseases Unit, Department of Internal Medicine, Arnau de Vilanova University Hospital, Biomedical
Research Institute of Lleida, Avda Alcalde Rovira Roure 80, Lleida 25198, Spain
E-mail address: [email protected]

Clin Chest Med 34 (2013) 27–37

http://dx.doi.org/10.1016/j.ccm.2012.11.002
0272-5231/13/$ – see front matter Ó 2013 Elsevier Inc. All rights reserved.
28 Porcel

the Light criteria reinforces its diagnosis, because peptide testing, the current recommendation is to
HF accounts for 80% of all transudates. However, select NT-proBNP, based on existing evidence
HF-related effusions in patients who receive and analytical advantages.7 Several studies have
diuretics or have bloody fluids frequently meet shown a strong correlation between pleural fluid
the Light exudative criteria by a narrow margin.5 and serum levels of NT-proBNP and comparable
In addition, differentiating cardiac from noncardiac operating characteristics at similar or close cutoff
transudates requires the use of specific disease values for both.7,13 As a result, applying the test
biomarkers. to serum alone may be sufficient.14
In addition, more than 80% of HF-associated
effusions that have been misclassified as exudates
Natriuretic Peptides
by the Light criteria exhibit diagnostic levels of
Natriuretic peptides are neurohormones secreted NT-proBNP.5 However, calculating the albumin
by the cardiomyocytes in response to stretch gradient (serum concentration minus pleural fluid
and increased pressure in the chambers of the concentration) seems to be equally effective, and
heart.6 Automated immunometric assays are simpler, for this purpose.2 As expected, noncar-
readily available for brain natriuretic peptide diac transudates, such as hepatic hydrothoraces,
(BNP), the amino-terminal fragment NT-proBNP, have characteristically low pleural fluid concentra-
and the midregional proatrial natriuretic peptide tions of natriuretic peptides.8,11
(MR-proANP). The measurement of BNP or NT-
proBNP can be useful in the evaluation of patients BIOMARKERS OF PLEURAL INFECTION
presenting in the urgent care setting in whom the
clinical diagnosis of HF is uncertain. Patients with The terms pleural infection and parapneumonic
a serum BNP level less than 100 pg/mL or NT- effusions are used interchangeably, although
proBNP less than 300 pg/mL are unlikely to have one-fourth of pleural infection cases occur without
HF, whereas those with respective concentrations a concurrent bacterial pneumonia. The typical
of more than 500 pg/mL and 450 to 1800 pg/mL patient with pleural bacterial infection presents
(cutoffs depending on and increasing with the with symptoms of pneumonia (ie, fever, chest
patient’s age) are likely to have HF.7 Natriuretic pain, dyspnea, cough) along with leukocytosis,
peptides can be detected in pleural fluid using raised serum C-reactive protein (CRP) levels, and
assays originally designed for measuring serum a chest radiograph showing the effusion and
or plasma levels. radiological lung infiltrates. However, patients
Since the first report in 2004,8 several studies may have a more indolent presentation and
have supported the usefulness of pleural fluid several conditions (eg, tuberculosis, connective
NT-proBNP in determining the cardiac origin of tissue diseases, pulmonary embolism, pancreatic
a pleural effusion.9 In a meta-analysis of 10 publi- diseases, or malignancy) can all mimic pleural
cations, totaling 429 cardiac and 691 noncardiac bacterial infection.
effusions, the test had a pooled sensitivity and
specificity of 94%, a likelihood ratio (LR) positive Leukocyte Esterase Reagent Strips
of 15.2, an LR negative of 0.06, a diagnostic
Pleural fluid sampling and analysis are essential to
odds ratio of 246, and a summary receiver oper-
confirm an infection.15 In a resource-limited health
ating characteristic curve of 0.98.10 In general,
care setting or if biochemistries of the aspirated
LRs of more than 10 or less than 0.1 generate large
fluid are not available on an emergency basis,
and usually conclusive shifts from pretest to post-
urine reagent strips applied to pleural fluid may
test probability. Therefore, the finding of a pleural
expedite diagnostic information. One study tested
fluid NT-proBNP greater than 1500 pg/mL, which
commercially available reagent strips for leukocyte
is the most commonly reported diagnostic deci-
esterase in the pleural fluid of 42 patients with
sion threshold,7 argues convincingly for HF. More-
bacterial infections, 15 with tuberculosis, and
over, levels less than this cutoff are similarly
71 with noninfectious causes.16 A positive test
compelling, decreasing the probability of HF.
yielded 42% sensitivity, 100% specificity, and an
Pleural fluid BNP is a less powerful test for deci-
LR positive of 75 for the identification of bacterial
sively determining HF compared with NT-proBNP,
infections in the pleural space.
as was shown in a head-to-head comparison
study (respective areas under the curve [AUC] of
Rapid Pneumococcal Antigen Test
0.90 and 0.96).11 However, the measurement of
pleural fluid MR-proANP has discriminating When used in pleural fluid specimens, another
capacities similar to NT-proBNP.12 If clinicians quick and simple urinary test, the immunochroma-
choose pleural fluid specimens for natriuretic tographic detection of the pneumococcal antigen
 Pleural Fluid Biomarkers 29

(Binax NOW Streptococcus pneumoniae, Inver- 100 mg/L suggested complicated parapneumonic
ness Medical, Scarborough, Maine), allows the effusions (positive LRs of 5.5, 5.7, and 5, respec-
etiologic diagnosis of parapneumonic effusions tively), but the absence of these findings did not
and may help to optimize antimicrobial therapy. change the probability of complicated effusions
One study tested the pneumococcal antigen assay sufficiently (all negative LRs w0.48).18 As ex-
in 140 adult patients with pleural effusions.17 It was pected, a combination of CRP with either pH or
positive in 71% of 34 patients with pneumococcal glucose using an Or rule, wherein the pleural
pneumonia, 7% of 89 nonpneumococcal effu- space would be drained if any test is positive,
sions, and 29% of 17 parapneumonic effusions increased sensitivity for detecting complicated
of unknown microbial cause. Among 28 patients parapneumonic effusions to about 80%. In
with pneumococcal pneumonia for whom both contrast, the combination using an And rule,
pleural fluid and urine antigen assays were avail- wherein a chest tube would be inserted if both
able, the results were concordant in 18 and discor- tests were positive, increased the specificity to
dant in 10 (3 of the 10 had positive tests in pleural 97%.18 Whether pleural fluid CRP may be included
fluid and negative in the urine, whereas, in the re- among the routine biochemical data used to make
maining 7, the opposite occurred).17 Therefore, decisions on the drainage of nonpurulent infec-
ordering a pleural fluid pneumococcal antigen tious collections warrants further study.
assay is a logical option in patients with parapneu-
monic effusions and negative urinary tests in which
the bacterial cause is being investigated.
Newer Biomarkers of Infection
In addition to the tests described earlier, several
CRP
promising biomarkers of infection have been
The fluid in parapneumonic effusions is invariably used in research, although not in clinical prac-
an exudate, with predominantly polymorphonu- tice.21 They reflect different stages of the inflam-
clear leukocytes in more than 80% of the cases,15 matory cascade triggered by microorganisms
a feature that is not unique for this condition. and can be measured by commercially available
Approximately 10%, 20%, and 60% of tubercu- immunoassays. Some reported pleural fluid
lous, malignant, and pulmonary emboli–associ- biomarkers with acceptable performances for
ated pleural fluids, respectively, also reveal discriminating between nonpurulent complicated
neutrophilic predominance.15 However, if a neutro- and uncomplicated parapneumonic effusions are,
philic pleural fluid has CRP concentrations greater in decreasing order of positive LRs, tumor necrosis
than 45 mg/L, it is most likely parapneumonic factor a greater than or equal to 80 pg/mL
(LR 5 7.7).18 (LR 6.8),22 myeloperoxidase greater than 3000
More crucial than correctly diagnosing para- mg/L (LR 5.9),23 matrix metalloproteinase-2 less
pneumonic effusions is to identify those of nonpur- than or equal to 343 ng/mL (LR 5.5),24 neutrophil
ulent appearance, which require chest drainage elastase greater than 3500 mg/L (LR 4.6),25
(ie, complicated parapneumonic effusions). A interleukin-8 greater than or equal to 1000 pg/mL
delay in implementing the necessary drainage is (LR 4.6),26 lipopolysaccharide-binding protein
associated with significant morbidity and greater than or equal to 17 mg/mL (LR 4),27 and
mortality. Pleural fluid pH or, alternatively, glucose soluble triggering receptor expressed on myeloid
has a key role in defining complicated parapneu- cell greater than or equal to 180 pg/mL
monic effusions.19 The lower the pH or glucose (LR 3.9).27 Other biomarkers may be clinically
levels, the more likely the need for pleural meaningful because of their high sensitivity and
drainage. According to existing guidelines, a pH extremely low LR negative for complicated effu-
value of less than 7.20 or a glucose less than 60 sions. Examples are pleural interleukin 1b greater
mg/dL is recommended as an action threshold than 3.9 pg/mL (LR negative 0.02),28 terminal
for chest tube insertion,20 but decisions should complement complex SC5b-9 greater than
be made on an individual basis. A recent retro- 2000 mg/L (LR negative 0.03),29 and 8-isoprostane
spective review that evaluated the capacity of greater than 35.1 pg/mL (LR negative 0.04)30;
several pleural fluid biochemistries (ie, pH, glucose values lower than the reported thresholds make
and CRP) of 340 nonpurulent parapneumonic effu- a complicated parapneumonic effusion highly
sions to discriminate complicated from uncompli- unlikely. Procalcitonin, a prohormone used to
cated effusions showed that none could be confirm or exclude the diagnosis of severe bacte-
identified as being superior to the others (all area rial infection, lacks the ability to separate compli-
under the curve w0.80).18 Values of pH less than cated from uncomplicated effusions when
7.20, glucose <60 mg/dL, and CRP greater than evaluated in pleural fluid.27
30 Porcel

Caveats of Complicated Parapneumonic evaluated diagnostically in tuberculous pleurisy,

Marker Studies but none have been found to be superior to pleural
fluid ADA or IFN-g levels.33
Overall, there are several problems when trying to
assess the discriminatory properties of biomarker
measurements for pleural infection in the pub- Adenosine Deaminase
lished literature.19,21 First, there is no errorless
The ADA assay is inexpensive, rapid, simple to
gold standard for judging whether the physician’s
perform, and has a pivotal role for the immediate
decision to insert a chest tube was correct,
presumptive diagnosis of tuberculous pleuritis. In
possibly resulting in a misclassified complicated
the largest series from a single center published
effusion. Second, an incorporation bias can occur
to date, ADA levels were evaluated in the pleural
when the clinician makes the diagnosis of a compli-
fluid of 2104 consecutive patients, of whom 221
cated effusion after considering the results of
(10.5%) had tuberculosis.34 The pleural fluid ADA
routinely used tests, such as pleural fluid pH (ie,
level exceeded 35 U/L in 93% of pleural tubercu-
the index test forms part of the reference standard,
losis. In contrast, only 10% of lymphocytic
thereby overestimating its diagnostic accuracy).
exudates from patients with other diagnoses had
This bias may partly justify the nonsuperiority of
such high levels.34 These results are similar to
newer laboratory markers compared with tradi-
those from a meta-analysis of 63 studies that
tional ones. In addition, the reported operating
included 2796 patients with tuberculous pleuritis
characteristics of many investigative biomarkers
and 5297 with nontuberculous effusions.35 The re-
are based on studies with small sample sizes,
ported sensitivity, specificity, positive LR, negative
which result in wide 95% confidence intervals.
LR, and odds ratio of pleural fluid ADA levels in the
Because the new biomarkers provide the same
diagnosis of pleural tuberculosis were 92%, 90%,
information as the widely established pH and
9, 0.10, and 110, respectively.35 The most widely
glucose, or the easily performed CRP, it is not
accepted cutoff value for pleural ADA is 40 U/L.
anticipated that they will replace or be added to
Although one-third of parapneumonic effusions
those currently being used for clinical decision
and 70% of empyemas exhibit ADA levels above
making in parapneumonic effusions.
this threshold, they can be easily distinguished
from pleural tuberculosis by the clinical picture,
BIOMARKERS OF TUBERCULOSIS the fluid appearance, and because parapneu-
monic effusions predominantly have neutrophils
The need for biomarkers in pleural tuberculosis in the pleural fluid.32,34 High pleural fluid ADA
(TB) is justified by the low yield of conventional levels have also been reported in more than half
microbiological studies caused by the paucity of of lymphomatous effusions.34 An extremely high
Mycobacterium tuberculosis in pleural fluid, and ADA activity in pleural fluid (>250 U/L) should raise
the 6 to 8 weeks required to obtain results. For the suspicion of empyema or lymphoma rather
example, in a retrospective analysis of 214 than tuberculosis.34
patients with pleural tuberculosis, solid culture There are 2 different molecular forms of ADA:
media for mycobacteria were positive in just 28% ADA1 and ADA2. The first is ubiquitous and found
of sputum and 15% of pleural fluid samples.31 in many cells, whereas the second is produced
The respective figures for acid-fast bacilli staining, mainly by monocyte/macrophages and is respon-
a more rapid determination technique, were 14% sible for most of the increase in ADA activity in
and 3%.31 Pleural biopsy may show granulomas tuberculous pleural fluids.32 ADA probably retains
in approximately 80% of cases, but it is invasive its usefulness in patients positive for human immu-
and its yield and complication rate depend on nodeficiency virus with low CD4 cell counts
operator skills.32 because monocytes are not significantly affected
The presence of mycobacterial antigens in the by the retroviral infection.36 Tuberculous fluids
pleural space elicits an intense immune response, with polymorphonuclear predominance show
initially by neutrophils and subsequently by macro- higher pleural ADA levels, at the expense of both
phages and T-lymphocytes, which results in ADA1 and ADA2 isoenzymes, than their lympho-
a lymphocyte-predominant exudative effusion. cytic counterpart.31 As a result, the diagnostic
The cell-mediated immune response provokes performance of ADA is reliable even in the early
the release of T-lymphocyte enzymes, such as stage of tuberculous pleuritis.
adenosine deaminase (ADA), and proinflammatory Pleural fluid ADA is routinely used in the diag-
cytokines (eg, g-interferon [IFN-g]) into the pleural nostic work-up in countries where the prevalence
space. Numerous nonspecific inflammatory of tuberculosis as a cause of exudative effusions
and immune response biomarkers have been is high or moderate. In countries where it is low
 Pleural Fluid Biomarkers 31

(eg, 1%), the estimated positive predictive value of perform the assay,40 their regular use cannot be
the ADA test may be as low as 7%. Even so, the supported.
negative predictive value remains high (99.9%).34
For this reason, ADA can still be used to confi- Nucleic Acid Amplification Tests
dently exclude the disease.
Assays for the amplification and detection of M
tuberculosis nucleic acids in pleural fluid have
IFN-g high specificity (>95%), but varying and generally
disappointingly low sensitivities (w60%) that, in
IFN-g is a cytokine released by activated CD41 T
addition to the high cost of the tests, render
lymphocytes and increases the mycobactericidal
them inappropriate for ruling the disease in or out
activity of macrophages. There exist 2 different
in other than investigational settings.41 In-house
methods to evaluate it in the pleural fluid. The first
nucleic acid amplification tests produce highly
is the measurement of unstimulated IFN-g (most
inconsistent results compared with those that are
commonly by enzyme-linked immunosorbent
commercial and standardized. The low test sensi-
assay [ELISA]), and the second consists of the
tivity is mainly the result of the technical aspects of
detection of IFN-g release by pleural fluid mononu-
nucleic acid extraction, the presence of inhibitors
clear cells under stimulation with specific antigens
in the pleural fluid, and the paucibacillary nature
of M tuberculosis (IFN-g–releasing assays [IGRA]).
of the disease.
Concerning unstimulated IFN-g, a meta-analysis
of 22 studies that included 782 patients with tuber-
culous and 1319 with nontuberculous effusions re- BIOMARKERS OF MALIGNANCY
vealed that the mean sensitivity of the test was
The diagnosis of malignant pleural effusions is
89%, the mean specificity was 97%, and maxi-
most easily established by showing malignant
mum joint sensitivity and specificity was 95%.37
cells in the pleural space. However, pleural fluid
As in the ADA assays, hematologic malignancies
cytology is positive in only 60% of cases, leading
and empyemas can cause increased IFN-g levels
to the need for further diagnostic tests.15 The
in pleural fluid. Studies that have directly com-
cytology is more likely to be positive with adeno-
pared ADA and IFN-g in patients with pleural
carcinoma than squamous cell carcinoma or
tuberculosis have reported a slightly higher, but
lymphoma. Moreover, it provides a definitive diag-
nonsignificant, accuracy of the latter.38 However,
nosis of mesothelioma in only one-third of cases
because there is little difference in performance
and a suspected diagnosis in a further 20%.42
between the two and ADA is both cheaper and
simpler, ADA is considered to be the preferred
Classic Tumor Markers
test.32 A major limitation of ADA is that it does
not provide culture and drug sensitivity informa- Many articles have suggested the possibility of
tion, which may be vital in countries with a high diagnosing pleural malignancy when increased
degree of resistance to antituberculous drugs. A levels of tumor markers in the pleural fluid are
tissue diagnosis with accompanying culture and found. A myriad of tumor markers have been eval-
sensitivity data should be pursued if the patient uated. In a systematic review of 45 publications,
showing a lymphocytic exudative effusion with comprising 2834 patients with malignant and
a high ADA concentration is from a region with 3251 patients with nonmalignant effusions, the
high levels of mycobacterial resistance. summary estimates of pleural fluid carcinoem-
IGRAs were primarily designed to detect latent bryonic antigen (CEA) for identifying the former
tuberculosis, but their role in the work-up of sus- were sensitivity 54%, specificity 94%, positive
pected tuberculous effusions is now being LR 9.5, and negative LR 0.49.43 In an additional
debated. In a meta-analysis of 7 publications, meta-analysis by the same investigators, the
totaling 213 patients with tuberculous and 153 pooled sensitivity, specificity, positive LR, and
with nontuberculous effusions, the summary esti- negative LR of the following 4 pleural markers for
mates of sensitivity, specificity, positive LR, and differentiating malignant from benign effusions
negative LR for pleural IGRA assays (T-SPOT-TB were cancer antigen (CA) 125, 0.48/0.85/5.9/
and Quanti-FERON-TB) in diagnosing tuberculosis 0.54; CA 15-3, 0.51/0.96/11.7/0.52; CA 19-9,
were 75%, 82%, 3.5, and 0.24, respectively.39 0.25/0.96/10.4/0.7; and cytokeratin fragment
Because IGRAs are technically complex and (CYFRA) 21-1 0.55/0.91/6.5/0.43.44 For tumor
expensive tests that usually produce an unaccept- markers to be useful diagnostically, selected
able number of false-positive and false-negative cutoff points have to be 100% specific. However,
results, and 15% of cases are limited in their the adoption of a cutoff level that is not exceeded
inability to isolate sufficient mononuclear cells to by any of the benign effusions generates
32 Porcel

insensitive tests. For example, one study mea- a specificity of 95%, approximately 70% of
sured the pleural fluid levels of CEA, CA 15-3, CA patients with early mesothelioma would have re-
125, and CYFRA 21-1 in 416 patients, including mained undetected. This unacceptably low sensi-
166 with definitive malignant effusions, 77 with tivity limits the use of mesothelin in early diagnosis.
probable malignant effusions, and 173 with benign Serum mesothelin has also been suggested as
effusions.45 When cutoff levels that were 100% a complementary tool for detecting tumor progres-
specific were selected (ie, CEA>50 ng/mL, CA sion and evaluating response to treatment,50–52
15-3 >75 U/mL, CA 125 >2800 U/mL, and CYFRA although it has not yet been implemented in
21-1 >175 ng/mL), 54% of the malignant effusions routine clinical practice for this purpose.
were classified as such, and more than one-third Similar to serum mesothelin, pleural fluid meso-
of the cytology-negative malignant pleural effu- thelin levels are increased in patients with meso-
sions were identified by at least 1 marker. None thelioma relative to levels in effusions from other
of the 22 lymphomas and 7 sarcomas exceeded causes. In several studies, pleural mesothelin
the cutoff points set by the panel of tumor markers. has yielded approximately 65% sensitivity, 90%
The same was true for CEA in the 11 mesotheli- specificity, positive LR of 7, and AUC greater
omas, giving support that high pleural fluid than 0.80 for discriminating between mesothe-
concentrations of CEA are valuable in ruling out lioma and other malignancies or nonmalignant
this tumor type.45 diseases.53–57 Different accuracies in various
An increased level of some tumor markers in reports may have been influenced by the propor-
pleural fluid not only suggests malignancy, but is tion of sarcomatoid mesotheliomas (which do not
also an independent indicator of life expectancy. express mesothelin), or the type of effusions in
In a series of 224 confirmed metastatic pleural the nonmesothelioma groups (eg, mesothelin is
malignancies caused by adenocarcinoma or squa- increased in ovarian and pancreatic cancer). In
mous cell carcinoma, those with pleural CA 125 general, pleural mesothelin levels greater than
greater than 1000 U/mL and CYFRA 21-1 greater 20 nM strongly suggest mesothelioma. In one
than 100 ng/mL survived 7 months less.46 The clin- study of 105 cytology-negative effusions, 9 (60%)
ical applicability of measuring tumor markers in of 15 mesotheliomas had pleural fluid mesothelin
pleural fluid is limited because, even with high concentrations above this threshold, in contrast
concentrations, further confirmatory cytohisto- with only 3 (3%) of 90 nonmesothelioma effu-
logic diagnosis is necessary. sions.56 Although the use of pleural mesothelin as
a biomarker may reinforce the suspicion of meso-
thelioma, histopathologic diagnosis remains the
Mesothelin
gold standard.
Mesothelioma can be difficult to diagnose. Cyto-
logic examination of the pleural fluid often shows Fibulin-3
equivocal results, and international guidelines do
Fibulin-3 is an extracellular glycoprotein that medi-
not recommend it as the sole pathologic
ates cell-to-cell interactions and has variable
method.47 Thoracoscopic biopsy provides the
angiogenic effects. In a unique prospective study
most accurate definitive diagnosis. The search
that included a subsequent blinded validation
for a single diagnostic marker of mesothelioma
cohort, pleural fluid fibulin-3 levels were measured
has so far been evasive. Mesothelin, a 40 kDa
by using a commercially available ELISA in 74
membrane-bound protein attached to the meso-
patients with mesothelioma, 39 with benign effu-
thelial cell surface by phosphatidylinositol, has
sions, and 54 with nonmesothelioma malignant
been the most extensively investigated.48 It can
effusions.58 At the best cutoff value of greater
be measured by ELISA in either serum or pleural
than 346 ng/mL, fibulin-3 discriminated mesothe-
fluid.
lioma effusions from those with other causes
An individual patient data meta-analysis of 4491
with 84% sensitivity, 92% specificity, LR positive
patients, among whom there were 1026 with
of 11, LR negative of 0.17, and AUC of 0.93. If
malignant mesothelioma (MM), examined the
confirmed in future studies, this biomarker will
accuracy of serum mesothelin in the diagnosis of
become a useful addition to the mesothelioma
mesothelioma.49 When applying a common
diagnostic process.
threshold of 2 nmol/L, the summary estimate
sensitivity was 47% and specificity was 96%.
Immunocytochemical Markers
The use of mesothelin in early diagnosis was eval-
uated for differentiating 217 patients with stage I or In the examination of pleural fluid, a cytomorpho-
II epithelioid and biphasic mesothelioma from logic distinction between reactive mesothelial
1612 symptomatic or high-risk controls.49 At cells, MM, and metastatic adenocarcinoma can
 Pleural Fluid Biomarkers 33

often be difficult because of significant overlap- Another indication for effusion immunocyto-
ping cytologic features. Immunocytochemical chemistry is to establish the primary site of a malig-
staining on cell block preparations assists in es- nant effusion in patients with an occult primary or
tablishing such discriminations through the appli- multiple primaries. For example, determination of
cation of panels consisting of a variety of the primary origin of adenocarcinomas, the most
antibodies (Table 1).59 However, there is no agree- commonly found malignancy in effusion samples,
ment on the ideal antibody combination, but may necessitate the use of the following immuno-
a sensitivity of about 80% is desirable for inclusion cytochemical markers: TTF-1 (lung cancer),
in the panel.60 cytokeratin 7 (upper gastrointestinal and pancrea-
To confidently validate the diagnosis of epithe- tobiliary tract), cytokeratin 20 (colorectal) and, in
lioid mesothelioma as opposed to adenocarci- female patients, estrogen receptors (breast, female
noma, 2 positive mesothelioma markers along genital tract), mammaglobin (breast), and Wilms
with 2 negative adenocarcinoma markers are tumor gene 1 (WT-1; female genital tract).59
required (Table 1).47 The panel can be expanded Different molecular tests, such as fluorescent in
if the results are inconclusive. This author’s recom- situ hybridization and gene expression, may
mendation is to initially use small panels, which complement cytology in diagnosing malignant
include epithelial membrane antigen (EMA), calre- effusions, but require specialized equipment and
tinin, CEA, and thyroid transcription factor-1 personnel, which limits their introduction into
(TTF-1).61–63 routine clinical practice.64

Table 1
Immunocytochemical profile for differentiating between benign and malignant effusionsa

Marker Reactive Mesothelial Cells Mesothelioma Adenocarcinoma

Markers of Malignancy
EMA (clone E29) Negative Positive Positive
GLUT-1 Negative Positive Positive
Mesothelial Markers
Desmin Positive Negative Negative
Calretinin Positive Positive Negative
CK 5/6 Positive Positive Negative
WT-1 Positive Positive Negativeb
Mesothelin Positive Positive Negative
HBME-1 Positive Positive Negative
Podoplanin Positive Positive Negative
D2-40 Positive Positive Negative
Carcinoma Markers
CEA Negative Negative Positive
MOC-31 Negative Negative Positive
Ber-EP4 Negative Negative Positive
Leu-M1 (CD15) Negative Negative Positive
B72.3 Negative Negative Positive
BG8 (Lewis) Negative Negative Positive
TTF-1 Negative Negative Positivec
Napsin A Negative Negative Positived
Estrogen receptor (ER-1D5) Negative Negative Positivee

a
These are the expected results, but there may be deviations between cases.
b
Except in ovarian cancer.
c
Only in non–squamous cell lung and thyroid cancers.
d
In lung adenocarcinomas and renal cell carcinomas.
e
Only in breast and female genital tract carcinomas.
 34
Porcel
Fig. 1. Biomarker-based etiologic diagnosis of pleural effusions. The request for these pleural fluid biomarkers should be guided by clinical data.
 Pleural Fluid Biomarkers 35

SUMMARY 8. Porcel JM, Vives M, Cao G, et al. Measurement of

pro-brain natriuretic peptide in pleural fluid for the
Accurate diagnosis of the cause of a pleural effu- diagnosis of pleural effusions due to heart failure.
sion can be challenging. Analysis of soluble Am J Med 2004;116:417–20.
biomarkers from effusions may be a useful adjunc- 9. Zhou Q, Ye ZJ, Su Y, et al. Diagnostic value of
tive.47 Ideal biomarkers should be sensitive and N-terminal pro-brain natriuretic peptide for pleural
specific to the disease state being examined, effusion due to heart failure: a meta-analysis. Heart
and available at the bedside. These characteristics 2010;96:1207–11.
are met by the natriuretic peptide NT-proBNP and 10. Janda S, Swiston J. Diagnostic accuracy of pleural
the enzyme ADA. Pleural fluid NT-proBNP levels fluid NT-pro-BNP for pleural effusions of cardiac
more than 1500 pg/mL are practically diagnostic origin: a systematic review and meta-analysis.
of HF, whereas an ADA activity greater than 35 BMC Pulm Med 2010;10:58.
to 40 U/L in the pleural fluid provides strong 11. Porcel JM, Martı́nez-Alonso M, Cao G, et al.
evidence for tuberculosis. None of the many re- Biomarkers of heart failure in pleural fluid. Chest
ported biomarkers for pleural infection have yet 2009;136:671–7.
been shown to be more effective than pH in pre- 12. Porcel JM, Bielsa S, Morales JL, et al. Comparison of
dicting which parapneumonic effusions are pleural fluid NT-proBNP, MR-proANP and MR-
complicated. The value of pleural fluid tumor proADM for the diagnosis of cardiac effusions.
markers in diagnosing malignancy remains limited Respirology, in press.
because further pathologic examination is war- 13. Porcel JM, Chorda J, Cao G, et al. Comparing serum
ranted. In patients with effusions of undetermined and pleural fluid pro-brain natriuretic peptide
cause and negative cytology, an increased pleural (NT-proBNP) levels with pleural-to-serum albumin
fluid mesothelin or fibulin-3 level strongly favors, gradient for the identification of cardiac effusions mis-
although it is not definitive of, mesothelioma. classified by Light’s criteria. Respirology 2007;12:654–9.
Immunocytochemistry is essential for the diag- 14. Hooper C, Lee YC, Maskell N. BTS Pleural Guideline
nosis of mesothelioma and should be applied as Group. Investigation of a unilateral pleural effusion in
a panel of antibodies on cell block sections, adults: British Thoracic Society pleural disease
including at least 2 mesothelial cell markers and guideline 2010. Thorax 2010;65(Suppl 2):ii4–17.
2 other appropriate antibodies for the alternative 15. Porcel JM. Pearls and myths in pleural fluid analysis.
pathology being considered, either reactive meso- Respirology 2011;16:44–52.
thelial cells or metastatic adenocarcinoma. A sug- 16. Porcel JM, Esquerda A, Bielsa S. A specific point of
gested algorithm for the investigation of pleural care screen for infectious effusions using reagent
effusions, based on the inclusion of widely avail- strips. Eur Respir J 2011;37:1528–30.
able and evidence-supported biomarkers, is 17. Porcel JM, Ruiz-González A, Falguera M, et al.
shown in Fig. 1. Contribution of a pleural antigen assay (Binax
NOW) to the diagnosis of pneumococcal pneu-
REFERENCES monia. Chest 2007;131:1442–7.
18. Porcel JM, Bielsa S, Esquerda A, et al. Pleural fluid
1. Porcel JM, Light RW. Diagnostic approach to pleural C-reactive protein contributes to the diagnosis and
effusion in adults. Am Fam Physician 2006;73:1211–20. assessment of severity of parapneumonic effusions.
2. Bielsa S, Porcel JM, Castellote J, et al. Solving the Eur J Intern Med 2012;23:447–50.
Light’s criteria misclassification rate of cardiac and 19. Tobin CL, Porcel JM, Wrightson JM, et al. Diagnosis
hepatic transudates. Respirology 2012;17:721–6. of pleural infection: state of the art. Curr Respir Care
3. Gary Lee YC, Davies RJ, Light RW. Diagnosing Rep 2012;1:101–10.
pleural effusion: moving beyond transudate- 20. Davies HE, Davies RJO, Davies CWH. Management
exudate separation. Chest 2007;131:942–3. of pleural infection in adults: British Thoracic Society
4. Henry NL, Hayes DF. Cancer biomarkers. Mol Oncol pleural disease guideline 2010. Thorax 2010;
2012;6:140–6. 65(Suppl 2):ii41–53.
5. Porcel JM. Pleural effusions from congestive heart 21. Porcel JM. Pleural fluid tests to identify complicated
failure. Semin Respir Crit Care Med 2010;31:689–97. parapneumonic effusions. Curr Opin Pulm Med
6. Porcel JM. The use of probrain natriuretic peptide in 2010;16:357–61.
pleural fluid for the diagnosis of pleural effusions re- 22. Porcel JM, Vives M, Esquerda A. Tumour necrosis
sulting from heart failure. Curr Opin Pulm Med 2005; factor-a in pleural fluid: a marker of complicated par-
11:329–33. apneumonic effusions. Chest 2004;125:160–4.
7. Porcel JM. Utilization of BNP and NT-proBNP in the 23. Alegre J, Jufresa J, Segura R, et al. Pleural-fluid myelo-
diagnosis of pleural effusions due to heart failure. peroxidase in complicated and noncomplicated para-
Curr Opin Pulm Med 2011;17:215–9. pneumonic effusions. Eur Respir J 2002;19:320–5.
36 Porcel

24. Oikonomidi S, Kostikas K, Kalomenidis I, et al. Matrix 39. Zhou Q, Chen YQ, Qin SM, et al. Diagnostic accu-
metalloproteinase levels in the differentiation of par- racy of T-cell interferon-g release assays in tubercu-
apneumonic pleural effusions. Respiration 2010;80: lous pleurisy: a meta-analysis. Respirology 2011;16:
285–91. 473–80.
25. Alemán C, Alegre J, Segura RM, et al. Polymorpho- 40. Dheda K, van Zyl-Smit RN, Sechi LA, et al. Utility of
nuclear elastase in the early diagnosis of compli- quantitative T-cell responses versus unstimulated
cated pyogenic pleural effusions. Respiration interferon-g for the diagnosis of pleural tuberculosis.
2003;70:462–7. Eur Respir J 2009;34:1118–26.
26. Porcel JM, Galindo C, Esquerda A, et al. Pleural fluid 41. Pai M, Flores LL, Hubbard A, et al. Nucleic acid
interleukin-8 and C-reactive protein for discrimi- amplification tests in the diagnosis of tuberculous
nating complicated non-purulent from uncompli- pleuritis: a systematic review and meta-analysis.
cated parapneumonic effusions. Respirology 2008; BMC Infect Dis 2004;4:6.
13:58–62. 42. Rakha EA, Patil S, Abdulla K, et al. The sensitivity of
27. Porcel JM, Vives M, Cao G, et al. Biomarkers of cytologic evaluation of pleural fluid in the diagnosis
infection in pleural fluid for the differential diagnosis of malignant mesothelioma. Diagn Cytopathol
of pleural effusions. Eur Respir J 2009;34:1383–9. 2010;38:874–9.
28. Marchi E, Vargas FS, Acencio MM, et al. Pro- and 43. Hsi HZ, Liang QL, Jiang J, et al. Diagnostic value of
anti-inflammatory cytokines levels in complicated carcinoembryonic antigen in malignant pleural effu-
and non-complicated parapneumonic effusions. sion: a meta-analysis. Respirology 2008;13:518–27.
Chest 2012;141:183–9. 44. Liang QL, His HZ, Qin XJ, et al. Diagnostic accuracy
29. Vives M, Porcel JM, Gázquez I, et al. Pleural SC5b-9: of tumour markers for malignant pleural effusion:
a test for identifying complicated parapneumonic a meta-analysis. Thorax 2008;63:35–41.
effusions. Respiration 2000;67:433–8. 45. Porcel JM, Vives M, Esquerda A, et al. Use of a panel
30. Tsilioni I, Kostikas K, Kalomenidis I, et al. Diagnostic of tumor markers (carcinoembryonic antigen,
accuracy of biomarkers of oxidative stress in para- cancer antigen 125, carbohydrate antigen 15-3,
pneumonic pleural effusions. Eur J Clin Invest and cytokeratin 19 fragments) in pleural fluid for
2011;41:349–56. the differential diagnosis of benign and malignant
31. Bielsa S, Palma S, Pardina M, et al. Comparison of effusions. Chest 2004;126:1757–63.
polymorphonuclear- and lymphocytic-rich tubercu- 46. Bielsa S, Esquerda A, Salud A, et al. High levels of
lous pleural effusions. Int J Tuberc Lung Dis 2013; tumor markers in pleural fluid correlate with poor
17:85–9. survival in patients with adenocarcinomatous or
32. Porcel JM. Tuberculous pleural effusion. Lung 2009; squamous malignant effusions. Eur J Intern Med
187:263–70. 2009;20:383–6.
33. Trajman A, Pai M, Dheda K, et al. Novel tests for 47. Scherpereel A, Astoul P, Baas P, et al. Guidelines of
diagnosing tuberculous pleural effusion: what works the European Respiratory Society and the European
and what does not? Eur Respir J 2008;31:1098–106. Society of Thoracic Surgeons for the management of
34. Porcel JM, Esquerda A, Bielsa S. Diagnostic perfor- malignant pleural mesothelioma. Eur Respir J 2010;
mance of adenosine deaminase activity in pleural 35:479–95.
fluid: a single-center experience with over 2100 48. Tung A, Bilaceroglu S, Porcel JM, et al. Biomarkers
consecutive patients. Eur J Intern Med 2010;21: in pleural diseases. US Respir Dis 2011;7:26–31.
419–23. 49. Hollevoet K, Reitsma JB, Creaney J, et al. Serum
35. Liang QL, Shi HZ, Wang K, et al. Diagnostic mesothelin for diagnosing malignant pleural meso-
accuracy of adenosine deaminase in tuberculous thelioma: an individual patient data meta-analysis.
pleurisy: a meta-analysis. Respir Med 2008;102: J Clin Oncol 2012;30:1541–9.
744–54. 50. Creaney J, Francis RJ, Dick IM, et al. Serum soluble
36. Baba K, Hoosen AA, Langeland N, et al. Adenosine mesothelin concentrations in malignant pleural
deaminase activity is a sensitive marker for the diag- mesothelioma: relationship to tumor volume, clinical
nosis of tuberculous pleuritis in patients with very stage and changes in tumor burden. Clin Cancer
low CD4 counts. PLoS One 2008;3:e2788. Res 2011;17:1181–9.
37. Jiang J, Shi HZ, Liang QL, et al. Diagnostic value of 51. Hollevoet K, Nackaerts K, Gosselin R, et al. Soluble
interferon-gamma in tuberculous pleurisy: a meta- mesothelin, megakaryocyte potentiating factor, and
analysis. Chest 2007;131:1133–41. osteopontin as markers of patient response and
38. Kalantri Y, Hemvani N, Chitnis DS. Evaluation of real- outcome in mesothelioma. J Thorac Oncol 2011;6:
time polymerase chain reaction, interferon-gamma, 1930–7.
adenosine deaminase, and immunoglobulin A for 52. Franko A, Dolzan V, Kovac V, et al. Soluble mesothelin-
the efficient diagnosis of pleural tuberculosis. Int J related peptide levels in patients with malignant meso-
Infect Dis 2011;15:e226–31. thelioma. Dis Markers 2012;32:123–31.
 Pleural Fluid Biomarkers 37

53. Scherpereel A, Grigoriu B, Conti M, et al. Soluble 59. Ganjei-Azar P, Jordà M, Krishan A. Effusion cytology.
mesothelin-related peptides in the diagnosis of A practical guide to cancer diagnosis. New York:
malignant pleural mesothelioma. Am J Respir Crit Demos Medical; 2011.
Care Med 2006;173:1155–60. 60. Zeren EH, Demirag F. Benign and malignant
54. Creaney J, Yeoman D, Naumoff LK, et al. Soluble mesothelial proliferation. Surg Pathol 2010;3:
mesothelin in effusions: a useful tool for the diagnosis 83–107.
of malignant mesothelioma. Thorax 2007;62:569–76. 61. van der Bij S, Schaake E, Koffijberg H, et al.
55. Alemán C, Porcel JM, Segura RM, et al. Pleural fluid Markers for the non-invasive diagnosis of mesothe-
mesothelin for the differential diagnosis of exudative lioma: a systematic review. Br J Cancer 2011;104:
pleural effusions. Med Clin (Barc) 2009;133:449–53. 1325–33.
56. Davies HE, Sadler RS, Bielsa S, et al. Clinical impact 62. Betta PG, Magnani C, Bensi T, et al. Immunohisto-
and reliability of pleural fluid mesothelin in undiag- chemistry and molecular diagnostics of pleural
nosed pleural effusions. Am J Respir Crit Care malignant mesothelioma. Arch Pathol Lab Med
Med 2009;180:437–44. 2012;136:253–61.
57. Fujimoto N, Gemba K, Asano M, et al. Soluble 63. Porcel JM. Immunocytochemical markers for meta-
mesothelin-related protein in pleural effusion from static lung adenocarcinoma. Int Pleural Newsl
patients with malignant pleural mesothelioma. Exp 2012;10:8–9. Available at: http://www.musc.edu/
Ther Med 2010;1:313–7. pleuralnews/IPN/.
58. Pass HI, Levin SM, Harbut MR, et al. Fibulin-3 as 64. Sriram KB, Relan V, Clarke BE, et al. Diagnostic
a blood and effusion biomarker for pleural mesothe- molecular biomarkers for malignant pleural effu-
lioma. N Engl J Med 2012;367:1417–27. sions. Future Oncol 2011;7:737–52.