Download The Usefulness of Diagnostic Tests on Pericardial Fluid*

The Usefulness of
Pericardial Fluid*
David G. Meyers, MD, FCCP;
Thomas W. Prendergast, MD
Diagnostic Tests on
Rayma E. Meyers, BA; and
Study objectives: To determine the physical, chemical, and cellular characteristics of pericardial
fluid in various disease states and to assess their diagnostic accuracies.
Setting: A metropolitan university hospital.
Design: Consecutive case series.
seventy-five hospital patients, aged 1 month to 87 years, who had
undergone pericardiocentesis (n=165) or control subjects who had undergone open heart
surgery (n=10) between 1984 and 1996.
Measurements: The appearance of pericardial fluid and results of chemistry tests, cell counts,
and microbial cultures were obtained by chart review. The
cytologic studies, Gram's stain,
etiology of each pericardial fluid sample was determined using prospective diagnostic criteria.
Results: Exudates differed from transudates by higher leukocyte counts and ratios of fluid to
serum lactate dehydrogenase levels. Fluid glucose levels were significantly less in exudates.
Sensitivity for detecting exudates was high for specific gravity > 1.015 (90%), fluid total protein
>3.0 g/dL (97%), fluid to serum protein ratio >0.5 (96%), fluid lactate dehydrogenase ratio >0.6
(94%), and fluid to serum glucose ratio <1.0 (85%). None of these indicators were specific. Fluid
total protein and specific gravity were moderately correlated (r=0.56). Fluid cytologic study had
a sensitivity of 92% and specificity of 100% for malignant effusion. No other test was diagnostic
for a specific etiology. Among infection-associated effusions, culture-positive fluid had more
neutrophils, higher lactate dehydrogenase levels, and lower ratios of fluid to serum glucose than
culture-negative (parainfective) fluid.
Conclusions: Evaluation of pericardial fluid might be limited to cell count, glucose, protein, and
lactate dehydrogenase determinations plus bacterial culture and cytology. While not used
routinely, other tests that may be highly specific for particular diseases should be ordered only
to confirm a high clinical suspicion.
(CHEST
1997; 111:1213-21)
Patients: One hundred
Key words: culture; cytology; diagnosis; glucose; lactate dehydrogenase; protein
effusion is present in a variety of
pathologic conditions. Often the etiology of the
effusion is uncertain and cannot be clearly defined
on the basis of clinical assessment or even autopsy.
"P ericardial
-¦-
For editorial comment
see
page 1156
With few validated tests available, investigators have
successfully made correct diagnoses based on peri¬
cardial fluid analysis in 24 to 93% of cases in
reported series.1-2
The laboratory
analysis
of
pleural
fluid is well
Departments of Internal Medicine (Cardiology [Dr.
Meyers and Mr.
Meyers]) and Surgery (Cardiothoracic [Dr.
Kansas University Medical Center, Kansas City.
Prendergast]),
Manuscript received August 16, 1996; revision accepted Novem¬
ber 1, 1996.
Dr. Meyers, Division of Cardiology, Kansas
Reprint requests:
Medical Center, 3901 Rainbow Blvd., Kansas City, KS
University
66160-7378
*From the
established. For instance, exudative pleural effusions
have protein levels >3.0 g/dL, specific gravity
> 1.015, a pleural fluid to serum protein ratio >0.5,
and a pleural fluid to serum lactate dehydrogenase
ratio >0.6.35 If the pleural fluid has none of these
characteristics, it is a transudate. Grossly bloody fluid
with an RBC count > 100,000/mm3 is suggestive of
trauma, malignancy, or pulmonary embolism. Pleu¬
ral fluid leukocyte counts > 10,000/mm3 are most
common with empyema, parapneumonic effusions,
pancreatitis, pulmonary infarction, collagen vascular
diseases, malignancy, and tuberculosis. Pleural fluid
protein level >6.0 g/dL often indicates tuberculosis
or parapneumonic effusion. An elevated fluid lactate
dehydrogenase with normal fluid protein is most
likely due to malignancy. Low pleural fluid glucose
level (<60 to 80 mg/dL) may be due to parapneu¬
monic, rheumatoid, tuberculous, or malignant effu¬
sion.
CHEST/111 /5/MAY, 1997
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
1213
In contrast to the well-documented ability of tests
identify the cause of pleural
pleural
effusion, to our knowledge there has been no com¬
parable systematic evaluation of diagnostic test use¬
fulness when applied to pericardial fluid. This report
examines the diagnostic and discriminant power of
several chemistry, cellular, and microbiological tests
on pericardial fluid.
on
fluid
to
Materials
and
Methods
The charts of patients admitted to Kansas University Medical
Center who underwent pericardiocentesis between January 1,
1984 and January 30, 1992 were identified by a computer search
of medical records for pericardiocentesis (International Classifi¬
37.0). One hundred sixty-seven cases
were identified. Two were excluded because no laboratory
studies had been obtained.
Each hospital discharge summary was reviewed for the cause
of the pericardial effusion as originally determined by the
patient's attending physician. An investigator (D.G.M.) then
reviewed the entire hospital record to confirm the cause of the
effusion using the following prospectively defined criteria: ma¬
lignant.a histologically confirmed diagnosis of any type of
cancer, known to metastasize or extend to the pericardium, made
within 1 year preceding the pericardiocentesis; bacterial.cul¬
ture-proved bacterial infection of blood, lung, or pericardium;
viral.clinical presentation suggesting a \iral syndrome; radia¬
tion.prior mediastinal irradiation; uremic.current renal dialy¬
sis or serum creatine value >2.0 mg/dL; rheumatologic.Amer¬
ican Rheumatologic Association criteria6-7 for rheumatoid
cation of Diseases code
scleroderma, or cur¬
rent exposure to hydralazine or procainamide; postpericardiotomy.pericardiotomy within the previous 3 months; traumat¬
arthritis, systemic lupus erythematosus,
or
ic.chest trauma within the 7 days preceding pericardiocentesis;
and hypothyroid.serum thyroxine value <5 mg/dL and thyroid
stimulating hormone value >5 ixIU/mL.
If more than one of the above causes of pericardial effusion
were present or if the attending physician and the reviewer did
not identify the same cause, then the etiology for that patient was
classified as undetermined. Diagnoses were confirmed by au¬
topsy in eight cases.
Control patients were evaluated by obtaining pericardial fluid
artery7 bypass
during routine open heart surgery7 (nine coronary
one aortic valve replacement) between February
procedures and
and March 1996 by open pericardial aspiration. None of these
patients had diseases or were receiving medications known to
cause pericardial effusion.
Between 1984 and 1992, virtually all pericardial fluid samples
were submitted for the same battery of laboratory7 tests, including
the following: appearance; volume; cell count; specific gravity;
protein; lactate dehydrogenase; glucose; Gram's stain; culture for
bacteria mycobacteria, virus, and fungus; and cytology.
Data were abstracted from medical records, entered into a
computer database, and analyzed using software (STATA version
4.0 statistical software package; STATA Corporation; College
Station, Tex). Because of unequal variance, comparisons of
discrete data used the x2 tests, while comparisons of continuous
variables used either the Wilcoxon rank-sum test or the KruskalWallis test with Bonferroni post hoc testing. Significance was
defined as a two-tailed alpha level <0.10. Sensitivity was defined
as
true-positives/true-positives + false-negatives. Specificity7 was
as true-negatives/true-negatives + false-positives.
defined
Results
The results from 82 male and 93 female subjects
aged 46.0±21.6 years (range, 1 month to 87 years)
are included. Exudative effusions included those
diagnosed as malignant (n=37), viral (n=19), bacte¬
rial (n=9) or associated with infection but with no
growth on cultures (parainfective) (n=6), rheumato¬
logic (n=8), or postpericardiotomy (n=8) based on
the degree of inflammation inherent in each disease
process.3-5 Transudates included those effusions due
to radiation (n 15), uremia (n=15), trauma (n=4),
hypothyroidism (n=3), and normal controls (n=10).
The 41 unclassified cases were analyzed separately.
The initial analysis compared exudates with trans¬
udates (Table 1). Neither the volume nor the appear¬
ance of the pericardial fluid differed between exu¬
dates and transudates. While RBC counts were
similar, exudates had significantly higher fluid leuko¬
cyte counts (14,116±37,446 vs 2,210± 1,916 cells
per milliliter; p=0.0123). The proportions of neutro¬
phils and monocytes did not differ. Few differences
between exudates and transudates occurred among
the chemistry tests evaluated. The ratio of fluid
lactate dehydrogenase to serum lactate dehydroge¬
nase was greater with exudates (11.6± 16.4 vs
2.6±2.8; p=0.0668). Fluid glucose level was lower
with exudates (77.9±41.9 vs 96.1±50.7 mg/dL;
p 0.0218). Specific gravity and total protein of
pericardial fluid are moderately correlated, with
specific gravity changing by 0.0025 for each 1.0 g
change in total protein (r=0.56, p<0.001) (Fig 1).
The ability of each test to correctly classify fluid as
exudate or transudate was evaluated using the cut
points generally used for pleural fluid.35 Figure 2
shows the individual data points with the horizontal
line denoting the cut point and open symbols mark¬
ing correct classifications. Exudate was correctly
detected (sensitivity) by specific gravity > 1.015 in
90% of cases. Similarly, a fluid total protein level
>3.0 g/dL had a sensitivity of 97%, and a fluid to
serum protein ratio >0.5 had a sensitivity of 96%.
The most sensitive marker of exudate was fluid
lactate dehydrogenase >200 mg/dL, giving a sensi¬
tivity of 98%. The ratio of fluid to serum lactate
>0.6 had a sensitivity of 94%. Spec¬
dehydrogenase
tests. Conversely, the most
ificity wastestslow for these
specific (those best<identifying transudate) were
fluid leukocyte count 10,000 cells per milliliter
(89%) and fluid glucose level >60 mg/dL (76%).
Other cut points did not materially improve diagnos¬
tic accuracy for any test. The test that most correctly
classified fluids as exudate or transudate was a fluid
to serum lactate dehydrogenase ratio >0.6 (diagnos¬
tic accuracy of 87%). While results of fluid-specific
=
=
1214
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
Clinical
Investigations
5,000,000
pfRBC
140,000 |-
pfWBC
1.05
120,000
4,000,000
pfSpGrav
pfTP
pfTP/sTP
1.04
100,000
3,000,000
80,000
1.03
2,000,000
60,000
1.02
40,000
1.01
1,000,000
20,000
0
Sn=58%
Trans.
Sn=43%
Ex. Trans.
Sn=91%
Sp=65%
Sp=89%
Sp=30%
pfLDH
pfLDH/sLDH
pfGluc
Ex. Trans.
DA=56%
10,000
Ex.
DA=52%
DA=77%
DA=82%
pfGluc/sGluc
pfPH
8,000 [-
Sn=94%
Ex. Trans.
Sn=25%
Ex. Trans.
Sn=85%
Ex. Trans.
Sn=56%
Sp=40%
Sp=76%
Sp=12%
Sp=40%
Ex. Trans.
DA=87%
Figure 2. Distributions of
DA=36%
pericardial
fluid characteristics.
DA=69%
Open triangle=exudate;
DA=52%
open
circle=transudate; open symbols=correct classification; closed symbols=incorrect classification; horizontal
line represents the cut point for discriminating exudate from transudate with the numeric value of the cut
point to its right; Ex=exudate; Trans=transudate; Sn=sensitivity; Sp=specificity; DA=diagnostic accuracy;
pfWBC =pericardial fluid WBC count; pfSpGrav=pericardial fluid
pfRBC^pericardial fluid RBC count;
fluid total protein; pfTP/sTP== ratio of pericardial to serum total protein;
specific gravity; pfTP=pericardial
pfLDH/sLDH=ratio of pericardial to serum lactate
pfLDH=pericardial fluid lactate dehydrogenase;
fluid glucose; pfGluc/sGluc=ratio of pericardial to serum glucose;
dehydrogenase; pfGluc=pericardial
glucose, pfPH=pericardial fluid pH.
high¬
diagnostic groups (Table 3). WBC counts were those
among inflammatory diseases, particularly
with bacterial (41,410±77,122 cells per milliliter)
and rheumatologic (12,100±21,636 cells per millili¬
ter) etiologies. Very low WBC counts were seen in
per milliliter).
hypothyroid effusions (625±247 cellshad
the highest
and
effusions
Hypothyroid malignant
and
of
74±26%,
(79±27%
proportions monocytes
and
bacterial
effusions
Rheumatoid
respectively).
had the highest proportions of neutrophils (78±20%
and 69±32%, respectively).
Two chemistry tests differed among the 11 diag¬
nostic groups (Table 4). The ratio of fluid to serum
total protein significantly differed between the 10
control subjects and the 10 patients with bacterial or
vs 0.85±0.10, respec¬
parainfective casesAs(0.55±0.07traumatic
effusion had
tively; p<0.001). expected,
est
pericardial fluid to serum protein ratio approaching
significantly
differed between control subjects and subjects with
both bacterial-parainfective and malignant fluids
(49±18 vs 121±20 and 117±33 mg/dL, respectively;
with
p=0.09). Lowest pH measurements were noteddiffer
infected fluids but this did not statistically
from other causes. The overlap between most diag¬
nostic groups is so large with all chemistry tests that
discriminating
ability is lost.
Among the 15 cases clinically classified as having
bacterial infection-associated pericardial effusion,
nine had positive bacterial cultures of pericardial
fluid and six had no growth on aerobic bacterial
culture. There were no positive viral, fungal, or
mycobacterial cultures in the entire cohort. Based on
our clinical definition of bacterial infection-associa
1.0. Pericardial fluid cholesterol levels
1216
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
Clinical
Investigations
5,000,000
pfRBC
140,000 |-
pfWBC
1.05
120,000
4,000,000
pfSpGrav
pfTP
pfTP/sTP
1.04
100,000
3,000,000
80,000
1.03
2,000,000
60,000
1.02
40,000
1.01
1,000,000
20,000
0
Sn=58%
Trans.
Sn=43%
Ex. Trans.
Sn=91%
Sp=65%
Sp=89%
Sp=30%
pfLDH
pfLDH/sLDH
pfGluc
Ex. Trans.
DA=56%
10,000
Ex.
DA=52%
DA=77%
DA=82%
pfGluc/sGluc
pfPH
8,000 [-
Sn=94%
Ex. Trans.
Sn=25%
Ex. Trans.
Sn=85%
Ex. Trans.
Sn=56%
Sp=40%
Sp=76%
Sp=12%
Sp=40%
Ex. Trans.
DA=87%
Figure 2. Distributions of
DA=36%
pericardial
fluid characteristics.
DA=69%
Open triangle=exudate;
DA=52%
open
circle=transudate; open symbols=correct classification; closed symbols=incorrect classification; horizontal
line represents the cut point for discriminating exudate from transudate with the numeric value of the cut
point to its right; Ex=exudate; Trans=transudate; Sn=sensitivity; Sp=specificity; DA=diagnostic accuracy;
pfWBC =pericardial fluid WBC count; pfSpGrav=pericardial fluid
pfRBC^pericardial fluid RBC count;
fluid total protein; pfTP/sTP== ratio of pericardial to serum total protein;
specific gravity; pfTP=pericardial
pfLDH/sLDH=ratio of pericardial to serum lactate
pfLDH=pericardial fluid lactate dehydrogenase;
fluid glucose; pfGluc/sGluc=ratio of pericardial to serum glucose;
dehydrogenase; pfGluc=pericardial
glucose, pfPH=pericardial fluid pH.
high¬
diagnostic groups (Table 3). WBC counts were those
among inflammatory diseases, particularly
with bacterial (41,410±77,122 cells per milliliter)
and rheumatologic (12,100±21,636 cells per millili¬
ter) etiologies. Very low WBC counts were seen in
per milliliter).
hypothyroid effusions (625±247 cellshad
the highest
and
effusions
Hypothyroid malignant
and
of
74±26%,
(79±27%
proportions monocytes
and
bacterial
effusions
Rheumatoid
respectively).
had the highest proportions of neutrophils (78±20%
and 69±32%, respectively).
Two chemistry tests differed among the 11 diag¬
nostic groups (Table 4). The ratio of fluid to serum
total protein significantly differed between the 10
control subjects and the 10 patients with bacterial or
vs 0.85±0.10, respec¬
parainfective casesAs(0.55±0.07traumatic
effusion had
tively; p<0.001). expected,
est
pericardial fluid to serum protein ratio approaching
significantly
differed between control subjects and subjects with
both bacterial-parainfective and malignant fluids
(49±18 vs 121±20 and 117±33 mg/dL, respectively;
with
p=0.09). Lowest pH measurements were noteddiffer
infected fluids but this did not statistically
from other causes. The overlap between most diag¬
nostic groups is so large with all chemistry tests that
discriminating
ability is lost.
Among the 15 cases clinically classified as having
bacterial infection-associated pericardial effusion,
nine had positive bacterial cultures of pericardial
fluid and six had no growth on aerobic bacterial
culture. There were no positive viral, fungal, or
mycobacterial cultures in the entire cohort. Based on
our clinical definition of bacterial infection-associa
1.0. Pericardial fluid cholesterol levels
1216
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
Clinical
Investigations
Type ofEffusion
Purulent, %
Hemorrhagic, %
Table 2.Pericardial Fluid Characteristics by
Diagnoses
p Value
Volume, mL
Unclassified
Serous, %
p Value
666±421
618±565
535 ±649
448±379
429 ±680
365±267
354±263
318±217
217±208
200±100
Uremic
Radiation
Infection
Postpericardiotomy
Malignant
Viral
Rheumatic
Trauma
Hypothyroid
0.001
Controls
Table 3.Cell Counts by
Malignant
Viral
Bacterial
RBC
WBC
Neutrophils, %
Monocytes, %
484,446 ±753,594
9,280: ±21,967*
6,999: ±11,276
41,410: ±77,122
2,663: ±2,432
12,100: ±21,636
3,846: ±7,581
2,266: ±1452
3,276: ±3,168
26±26
46±32
69±32
27±20
78±20*
39±36
60±21
32±38
21±27*
32±26
<0.001
74±26
41 ±32
31±32*
70±21
22±20*
61 ±36
38 ±24
68 ±38
79 ±27*
66±28
<0.001
482,691±855,111
766,848±950,206
940,547± 1,044,064
257,840±425,521
617,500± 1,478,254
687,515± 1,186,049
2,107,585± 1,835,324
48,333 ±79,425
5,021±11,190
Postpericardiotomy
Rheumatic
Radiation
Uremic
Trauma
Hypothyroid
Unclassified
Kruskal-Wallis
p Value
Type ofEffusion
625: ±247*
14,690: ±3,605
0. 008
NS
*Significant pairwise difference.
Table 4.Pericardial Fluid
SpGR
Bacterial
(parainfective)
Postpericardiotomy
Rheumatologic
pfLDH
1.020±.003 3,503±3,281
Chemistry by Type ofEffusion*
PfLDH/LDH
pfTP
PfTP/TP
pfGluc
PfGlu/Gluc
pfChol
8.6±7.6
4.6±0.5
0.85±0.10f
67.1±39.4
0.64±0.50
121±20+ 0.79±0.37
PfCh/Chol
0.98
63.5== 9.3 0.64 j :0.21 63
4.9d 1.4 0.83d 0.20
3.5±3.2
66.4 j 31.9 0.63d :0.34
84 = :35
4.5d 1.0 0.64== 0.14
6.7±4.1
0.98±0.02
126d :58.8 1.03d :0.30
99
4.1d 1.2 0.66d 0.15
2.9±3.4
Radiation
80.8d 37.5 0.69d :0.29
4.8d 1.2 0.80== 0.16
1.3±1.5
Uremic
1.07
43.7== :247 0.43d :0.25
4.5±2.5 0.94±0.29
1.4±1.7
Trauma
164== 83.2 0.83d 0.25 100
1.10
3.5±1.7 0.48±0.39
3.4±5.0
Hypothyroid
4.3±1.5 0.70±0.24 106.2±63.7 0.83±0.26 64±57 0.99±0.03
2.4±2.9
Unclassified
*SpGr=specific gravity; PfLDH=pericardial fluid lactate dehydrogenase; PfLDH/LDH=ratio of percardial fluid lactate dehydrogenase to serum
lactate dehydrogenase; PfTP.pericardial fluid total protein; PfTP/TP=ratio of pericardial fluid total protein to serum total protein;
PfGluc=pericardial fluid glucose; PfGluc/Gluc=ratio of pericardial fluid glucose to serum glucose; PfChol=percardial fluid cholesterol;
PfChol/Chol=ratio of percardial fluid cholesterol to serum cholesterol.
f Denotes
significant pairwise differences. Blank cells had insufficient numbers of tests or no tests performed.
1.021 = :.001 1,552± 1,422
1.019d :.006 3,677±3,631
1.019= :.004
672±1,015
1.021== :.008 2,820 ±2,255
1.019== .008 1,285± 1,330
1.024=^ .003 1,376± 1,937
1.019±.005 2,332± 1,328
ated effusion, which would include septicemia and
pneumonitis, the six culture-negative cases could
have parainfective pericardial effusions, (similar to
parapneumonic effusion). As shown in Table 5, those
with infective cases (positive cultures) were signifi¬
cantly younger, trended toward higher fluid WBC
counts, had significantly more neutrophils, and
higher lactate dehydrogenase levels. They also had
significantly lower fluid glucose levels and fluid to
serum glucose ratios than those with parainfective
cases.
The diagnostic accuracy of Gram's stains was
evaluated using growth on bacterial cultures of peri¬
cardial fluid as the reference standard. Stained bac¬
teria were noted in three of the eight culture-positive
fluid samples and in 1 of 83 culture-negative samples
CHEST/111 /5/MAY, 1997
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
1217
Table
5.Comparison ofInfective and Parainfective
Pericardial Effusion*
Infective
Parainfective
9
Male/female
Age, yr
7/2
28.8±20.0
293 ±135.88
Volume, mL
Appearance (%)
Purulent
Hemorrhage
Clear
RBC
WBC
Neutrophils, %
Monocytes, %
LDH, mg/dL
Total protein,
g/dL
pfTP/serum TP
Glucose, mg/dL
pfGluc/serum glue
pH
Gram stain
571.67±560.91
2(22)
2(33)
4(45)
3(50)
3(33)
1(17)
345,400±429,645 700,237±943,171
NS
0.05
NS
0.066
31,775±36,092
8,192± 11,216
NS
NS
89±5
11±5
52±37
48 ±37
0.015
0.015
7,101 ±3,675
1,704±706
4.15±1.44
4.66±0.63
0.032
NS
0.78±0.28
47.3±25.3
0.28±0.14
7.035±0.445
0.81±0.12
102.5±35.6
0.84±0.23
7.082±0.732
NS
0.014
0.005
NS
2(33)
3(50)
1(17)
NS
h)
2(25)
3(375)
3(375)
Normal
WBCs
Bacteria
*
6
6/0
45.2±24.3
p Value
LDH=lactate dehydrogenase; pfTP=pericardial fluid total protein;
TP=total protein; pfGlue/serum glue=ratio of pericardial fluid
glucose to serum glucose; NS=not significant.
(Table 6). Presence of stained bacteria had a sensi¬
tivity of 38% and absence of bacteria had a specificity
of 99%. Leukocytes without bacteria were present in
the stained pericardial fluid of three culture-positive
cases and 42 culture-negative cases. Either bacteria
or leukocytes were present in 64% of exudates while
no abnormalities on Gram's stain were seen in 58%
of samples. Presence of leukocytes on Gram's stains
(false-positive) occurred in five transudates: four
with uremia and one case with hypothyroidism.
Among 110 mycobacterial, 62 viral, and 120 fungal
cultures of pericardial fluid, there were no positive
results. Viral syndromes had been diagnosed by
Table
6.Diagnostic Accuracy of Gram's Stains
Bacterial Culture
40
1
42
Exudate
19
4
31
Transudate
clinical
of these patients. No
mycobacterial or fungal diseases had been suspected
presentation
in 19
clinically.
Cytologic tests were obtained in 36 cases with
malignancy and 101 cases with no clinical evidence
of malignancy at the time of pericardiocentesis.
Malignant cells were noted in 33 of 36 cases with
malignancy (sensitivity of 92%) and in none of 101
nonmalignant cases (specificity of 100%).
Discussion
of any other studies that have
evaluated
the usefulness of laboratory
systematically
tests performed on pericardial fluid. In the era of
outcomes analysis, such critical evaluation is impor¬
tant. Since a battery of tests including microbial
cultures, cytology, and chemical tests were obtained
almost routinely, selection bias is minimal. The
clinical diagnosis was generally not confirmed histo¬
thus allowing some risk of misclassification
logically,
of diagnoses.
To our knowledge, no available literature discrim¬
inates pericardial fluids into transudate, exudate, or
complicated exudate, as is done with pleural fluid.35
An article reporting three ill-defined cases with
transudative pericardial fluid noted spe¬
purported
cific gravities of 1.018 to 1.022, protein level of 4.4 to
5.0 mg/dL, and glucose level of 70 to 90 mg/dL.10
One case series noted that inflammatory pericardial
fluid had a pH of 7.06 ±0.07 compared with nonin¬
fluid pH of 7.42±0.06.n The current
flammatory
series shows that exudate and transudate can be best
discriminated by specific gravity > 1.015, fluid pro¬
tein level >3.0 mg/dL, fluid to serum protein ratio
>0.5, fluid lactate dehydrogenase value >300 U/dL,
and fluid to serum lactate dehydrogenase ratio >0.6.
Case reports have described fluid characteristics in
several specific diseases. In rheumatoid arthritis,
pericardial fluid has been described in 41 cases.1218
The fluid is always serosanguineous and sometimes
viscous. A single report noted an RBC count of
30,100 cells per milliliter.13 Leukocyte counts have
300 to 88,100 cells per milliliter.131718
ranged from were
the predominant cell type (72 to
Neutrophils
Total
80%).
protein level was 4.2 to 6.4 mg/dL. Fluid
level
has ranged from 0 to 66 mg/dL, with
glucose
most reported cases having glucose levels <30 mg/
(JL 1417 Qur series observed similar fluid appear¬
ance, cell counts, and protein levels. But none of our
seven cases had fluid glucose levels <75 mg/dL.
With pleural fluid, it is believed that a diffusion block
of glucose occurs in the inflamed pleura.35 Our cases
may have had lower intensities of pericardial inflam¬
mation than those other reported cases that had
We
are unaware
1218
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
Clinical
Investigations
lower glucose levels. Fluid characteristics have been
in four patients with systemic lupus ery¬
reported
thematosus and three patients with drug-induced
colored
lupus.19-25wereThree fluid samples wereRBCstraw
counts
have
and two
serosanguineous.20-23
from
to
cells
milliliter.2033,750
229,000
per
ranged
22,24 Leukocyte counts of 544 to 199,600 cells per
milliliter have been reported, with 76 to 96% neu¬
With specific gravity of 1.02821 and
trophils.20-2224
of
4.9
to 7.5 mg/dL,20-25 this fluid is an
protein
exudate. Lactate dehydrogenase levels were 2,085
and 4,688 U/dL in two cases, while glucose level has
20 to 100 mg/dL.21-24 For rheumatoid
ranged from
the
arthritis, most specific tests on pericardial fluid
appear to be latex agglutination of rheumatoid anti¬
gen,15 immunoglobin complexes,17 diminished fluid
complement,14-26 and cytologic identification of char¬
acteristic "RA cells."13 High pericardial fluid titers of
antinuclear antibody have been reported in systemic
lupus
erythematosus2425 as have "LE cells."19-21'23-25
Our series utilized none of these latter highly specific
tests.
In one of the only case series published on puru¬
lent pericarditis, detailed analysis was performed on
pericardial fluid from five patients.27 Fluid leukocyte
counts ranged from 6,100 to 241,000 cells per milli¬
liter and were >50,000 cells per milliliter in four of
the five cases with infected fluid. There were >90%
in all cases. Fluid glucose level was <35
neutrophils
in
four
mg/dL
patients. Protein level ranged from 3.3
to 6.2 g/dL. Lactate dehydrogenase levels were 4,800
and 6,700 U/mL in two patients. Among the cur¬
rently reported culture-positive patients, nearly all
test results were similar to these reports. Our ob¬
served fluid glucose levels were not as low (mean of
47.3 mg/dL) as the prior report and two of the five
determinations were >60 mg/dL. Similar to rheu¬
matoid diseases, infective pleuritis is believed to
to glucose.3-5 To our
produce a diffusion block
of
the
stains of pericar¬
Gram's
knowledge, accuracy
dial fluid has not been evaluated previously. Limited
by the fact that only five culture-positive samples in
the current series had Gram's stains, we found a
sensitivity of 38% and a specificity of 99% for
presence of bacteria on the stain. Using either
leukocytes or bacteria on Gram's stain as indicators
was prohibitively nonspecific for identifying infective
pericardial fluid, but was more useful for discrimi¬
nating exudate from transudate. Thus, a fluid to
serum glucose ratio <1.0 may be useful in differen¬
tiating exudate from transudate and infective from
parainfective effusions. Bacterial culture, but not
Gram's stain, is accurate for defining infective effu¬
sions and identifying causative agents.
With tuberculous pericarditis, others have re¬
ported a low yield from fluid culture or pericardial
biopsy and culture.2829 Fluid cell counts and chem¬
istry have not been described. Recently, adenosine
deaminase levels >40 U/L have demonstrated a
sensitivity of 93% and a specificity of 97% for
tuberculous pericarditis.3031 Nested polymerase
chain reaction from as little as 1 jjlL of pericardial
fluid can establish the diagnosis.32
Little has been published regarding viral pericar¬
ditis. To our knowledge, no cell counts or chemistry
results have been reported. Intranuclear inclusions,
multinucleated giant cells, and atypical mesothelial
cells have been observed with Papanicolaou-stained
pericardial fluid in one case each of cytomegalic and
herpes simplex pericarditis.33 We found no distin¬
guishing results of any tests.
To our knowledge, no prior reports have described
parainfective pericardial effusion. In the cases be¬
lieved associated with bacterial infection, but with
negative pericardial fluid cultures, it is likely that the
pericardial fluid was an inflammatory response to
adjacent infection, eg, parainfective pericarditis. We
noted low fluid leukocyte counts (four of six cases
had < 10,000 cells per milliliter), a high proportion of
monocytes (mean=:48%), lactate dehydrogenase lev¬
els < 1,550 U/dL in four of five cases, and glucose
levels always >50 mg/dL. These observations are
very similar to those in parapneumonic pleural effu¬
3-5
sions.
The fluid characteristics of one patient with postmyocardial infarction pericarditis34 and nine patients
with postpericardiotomy syndrome35 have been re¬
ported. The fluid appears serous or serosanguineous
with a mean of 5,628 leukocytes per milliliter, pre¬
dominantly monocytes and lymphocytes, and a mean
protein level of 6.1 mg/dL (range, 4.3 to 7.2). Our
results closely resemble the results of these re¬
ports.3435
In 15 cases of uremic pericarditis, all had serosan¬
guineous fluid with hematocrits of 1 to 24%.3637 The
leukocyte count was 2,580 cells per milliliter,
primarily consisting of neutrophils.
Mean total protein value has been reported at 4.08
g/dL (80% that of serum levels) and a cholesterol
level of 94 mg/dL (50% that of serum levels). In
three patients, the mean pericardial fluid pH was
7.08±0.1.37
mean
Our results
are
quite similar for each
parameter.
In one previously described patient with myxedema, the fluid was amber color with a leukocyte
count of 1,800 cells per milliliter, total protein level
of 6.0 g/dL (similar to serum), and a cholesterol level
of 76 mg/dL.38 The protein levels we observed in two
patients were much lower and only half that of
serum.
Our observed cholesterol levels
were
higher
CHEST/111 /5/MAY, 1997
1219
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
and similar
to
serum.
These differences
are unex¬
Caution is warranted given the few cases
plained.
examined.
To our knowledge, the fluid characteristics of
radiation-induced pericardial fluid have not been
described previously. We failed to identify any dis¬
tinctive characteristics.
Meta-analysis of 113 cases of malignant pericardial
effusion demonstrates a sensitivity of 95% for peri¬
cardial fluid cytologic study and a specificity of 100%
in 82 samples from patients not having cancer.39-49
Our sensitivity of 92% and specificity of 100% in 36
cancer cases and 101 noncancer cases
with
agree
previous reports. Since our cases, as with many other
reports, did not always have histologic confirmation
of malignant pericardial involvement, a sensitivity
<100% would be expected. Additionally, singlesample cytologic study is known to allow falsenegative results.50
One report has examined normal pericardial fluid,
obtained at open heart surgery in 11 patients. These
serum: sodium, 138±4 mol/L;
samples resembled
4.5±1
mol/L; chloride, 109±5 mol/L;
potassium,
and bicarbonate, 25 ±6 mol/L.51 Total protein level
was 3.1±0.6 g/dL. Fluid pH was 7.57±0.11. We
noted a slightly higher protein level in our 10 control
patients. The other parameters were not measured.
Conclusions
evaluation of pericardial fluid should
with
differentiation
between exudate and tran¬
begin
sudate. The identification of an exudate is best done
with one or more of four tests using the following cut
points: fluid total protein level >3.0 g/dL; fluid to
serum total protein ratio >0.5; fluid to serum lactate
ratio >0.60; and fluid lactate dehy¬
dehydrogenase
<300
U/dL.
drogenase
Two other tests should also be ordered routinely
based on their high specificity and the crucial impor¬
tance of detecting a specific disease: bacterial culture
and fluid cytology.
Other tests either do not have significant discrim¬
inating power, do not identify a prevalent disease, or
closely parallel results of another test. Thus, routine
testing should be limited to the above six tests.
Other tests should be ordered only when a high
level of clinical suspicion of a specific disease war¬
rants confirmation. These tests might include the
for rheumatoid antigen,
following: fluid latex fixation
fluid complement levels
complexes, andfluid
7-globulin
antinuclear antibody
for rheumatoid arthritis;
titers for systemic lupus erythematosus; and fluid
adenosine deaminase and nested polymerase chain
reaction for tuberculosis.
Diagnostic
To our knowledge, no diagnostic tests of pericar¬
dial fluid exist that are specific for effusion associated
with postpericardiotomy syndrome, radiation or ure¬
mic
pericarditis, hypothyroidism, or trauma.
References
1 Kirkorian
JG,
Hancock EW. Pericardiocentesis. Am
J
Med
1978; 65:808-14
2 Corey GR, Campbell PT, Van Trigt P, et al. Etiology of large
pericardial effusions. Am } Med 1993; 95:209-13
3 Light RW. Pleural effusions. Med Clin North Am. 1977;
61:1339-52
4
5
Jay SJ. Pleural effusions: I. Preliminary evaluation.recogni¬
tion of the transudate. Postgrad Med 1986; 80:164-77
Jay SJ. Pleural effusions: II. Definitive evaluation of the
exudate.
6 Arnett
Postgrad Med 1986; 80:181-88
Edworthy SM, Bloch DA, et al. The American
FC,
Rheumatism Association 1987 revised criteria for the classi¬
fication of rheumatoid arthritis. Arthritis Rheum 1988; 31:
315-24
7 Schumacher HR,
Klippel JH, Koopmen WJ. Primer on the
appendices. 10th ed. Atlanta: Arthritis
rheumatic diseases:
Foundation, 1993; 328-32
8
Armitage P, Berry G. Statistical methods in medical research.
2nd ed. Oxford: Blackwell Scientific Publications, 1987;
411-17
9 Sokal RR, Rohlf FJ. Biometry: the principles and practice of
statistics in biological research. 2nd ed. San Francisco: WH
Freeman, 1981
10 Mokhobo KP. Serous effusions: an attempt at a clinicalpathological correlation. S Afr Med J 1974; 49:865-68
11 Kindig JD, Goodman MR. Clinical utility of pericardial fluid
pH determination. Am J Med 1983; 75:1077-79
12 Stern JB, Sobel HJ. Hemorrhagic rheumatoid pericarditis.
Am J Cardiol 1961; 8:670-73
13 Berger HW7, Seckler SG. Pleural and pericardial effusions in
rheumatoid disease. Ann Intern Med 1966; 64:1291-97
14 Franco AE, Levine HD, Hall AP. Rheumatoid pericarditis:
report of 17 cases diagnosed clinically. Ann Intern Med 1972;
77:837-41
15 Calick A, Bishop R. Pericardial effusion in rheumatoid arthri¬
tis. Chest 1973; 64:778-79
16 Thadani U, Iveson JMI, Wright V. Cardiac tamponade,
constrictive pericarditis and pericardial resection in rheuma¬
toid arthritis. Medicine 1975; 54:261-70
17 Ball GV, Schrohenloher R, Hester R. Gamma globin com¬
plexes in rheumatoid pericardial fluid. Am J Med 1975;
58:123-28
18 Richards AJ, Koehler BD, Broder I, et al. Rheumatoid
pericarditis: comparison of immunologic characteristics of
pericardial fluid, synovial fluid, and serum. J Rheumatol 1976;
3:275-82
19 Seaman AJ, Christerson JW. Demonstration of LE cells in
pericardial fluid. JAMA 1952; 149:145-47
20 Lerer RJ. Cardiac tamponade as an initial finding in systemic
lupus erythematosus. Am J Dis Child 1972; 124:436-37
21 Wolkove N, Frank H. Lupus pericarditis. Can Med J 1974;
111:1331-33
22
23
Goldenberg DL, Leff G, Grayzel AL Pericardial tamponade
in systemic lupus erythemtosus. NY State J Med 1975; 910-12
Carey RM, Coleman M, Feder A. Pericardial tamponade: a
major presenting manifestation of hydralazine-induced lupus
syndrome. Am J Med 1973; 54:84-7
24 Stein
E, Deduce A, Lawson L,
et
1220
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
al. Procainamide-induced
Clinical
Investigations
a case with a large pericardial
lupus erythematosus: report ofRheumatol
effusion and fluid analysis. J
1979; 6:543-48
25 Goldberg MJ, Husain M, Wajszczuk WJ, et al. Procainamideinduced lupus erythematosus pericarditis encountered during
coronary bypass surgery. Am J Med 1980; 69:159-62
26 Hunder GG, Mullen BJ, McDuffie FC. Complement in
fluid of lupus erythematosus. Ann Intern Med
pericardial
1974; 80:453-58
27 Rubin RH, Moellering RC. Clinical microbiologic and ther¬
apeutic aspects of purulent pericarditis. Am J Med 1975;
59:68-78
28 Barr JF. The use of pericardial biopsy in establishing etiologic
diagnosis in acute pericarditis. Arch Intern Med 1955; 39:
693-86
29 Sagrista-Sauleda J, Permanyer-Niralda G, Soler-Soler J. Tu¬
berculous pericarditis: ten year experience with a prospective
protocol for diagnosis and treatment. J Am Coll Cardiol 1988;
11:724-28
30
Martinez-Vazquez JM, Ribera E, Ocana I, et al. Adenosine
activity in tuberculous pericarditis. Thorax 1986;
41:888-89
Koh KK, Kim EJ, Cho CH, et al. Adenosine deaminase and
carcinoembryonic antigenic pericardial effusion diagnosis,
deaminase
31
especially in suspected tuberculous pericarditis. Circulation
1994; 89:2728-35
32 Seino Y, Ikeda U, Kawaguchi K, et al. Tuberculous pericar¬
ditis presumably diagnosed by polymerase chain reaction
analysis. Am Heart J 1993; 126:249-51
33 Goodman ZD, Gupta PK, Frost JK, et al. Cytodiagnosis of
\iral infections in body cavity fluids. Acta Cytol 1979; 23:
204-08
34 Soloff LA. Pericardial cellular response during the postmyocardial infarction syndrome. Am Heart J 1971; 82:812-16
35 Ofori-Krakye SK, Tyberg Tl, Geha AS, et al. Late cardiac
incidence, role of anti¬
tamponade after open heart surgery/:
its
and
in
its
relationship to the
pathogenesis
coagulants
1981; 63:1323-28
postpericardiotomy syndrome. Circulation
36 Alfred AC, Goss JE, Ogden DA, et al. Uremic hemopericar¬
dium. Am J Med 1968; 45:391-400
37 Mann W, Millen JE, Glauser FL. Bloody pericardial fluid: the
value of blood gas measurements. JAMA 1978; 239:2151-52
38 Davis PJ, Jacobson S. Myxedema with cardiac tamponade and
pericardial effusion of "gold paint" appearance. Arch Intern
Med 1967; 120:615-19
39 Zipf RE, Johnston WW. The role of cytology in the evaluation
of pericardial effusions. Chest 1972; 62:593-96
40 King DT, Nieberg RK. The use of cytology to evaluate
pericardial effusions. Ann Clin Lab Sci 1979; 9:18-23
41 Yazdi HM, Hajdu SI, Melamed MR. Cytopathology of peri¬
cardial effusions. Acta Cytol 1980; 24:401-12
42 Posner MR, Cohen GI, Skarin AT. Pericardial disease in
patients with cancer. Am J Med 1981; 71:407-13
43 Reyes CV, Strinden C, Banerji M. The role of cytology in
neoplastic cardiac tamponade. Acta Cytol 1982; 26:299-302
44 Meyers DG, Bouska DJ. Diagnostic usefulness of pericardial
fluid cytology. Chest 1989; 95:1142-43
45 Edoute Y, Kuten A, Ben-Haim S, et al. Symptomatic pericar¬
dial effusion in breast cancer patients: the role of fluid
cytology. J Surg Oncol 1990; 45:265-69
46 Edoute Y, Malberger E, Kuten A, et al. Symptomatic peri¬
cardial effusion in lung cancer patients: the role of fluid
cytology. J Surg Oncol 1990; 45:121-23
47 Wiener HG, Kristensen IB, Haubek A, et al. The diagnostic
value of pericardial cytology. Acta Cytol 1991; 35:149-53
48 Campbell PT, Van Trigt P, Wall'TC, et al. Subxiphoid
of large
diagnosis and management
pericardiotomy in theassociated
Chest
with
effusions
1992;
malignancy.
pericardial
101:938-43
49 Hallman JR,
peritoneal
Gersinger KR. Cytology of fluids from pleural,
and
pericardial
cavities in children. Acta
Cytol
1994; 38:209-17
50 Garcia LW7, Ducatman BS, Wang HH. The value of multiple
fluid specimens in the cytological diagnosis of malignancy.
Mod Pathol 1994; 7:665-68
51 Hutchin P, Nino HV, Suberman R. Electrolyte and acid-base
composition of pericardial fluid in man. Arch Surg 1971;
102:28-30
CHEST/111 /5/MAY, 1997
Downloaded From: http://journal.publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21747/ on 04/30/2017
1221