Clinical Manifestations and Useful Tests
Normally, the pleural space contains only a few milliliters of pleural fluid. If fluid in the pleural space is detected on a radiologic examination, it is abnormal. Many conditions can be associated with pleural fluid accumulation (Table 8.1). When pleural fluid is detected, an effort should be made to determine which of the many conditions listed in Table 8.1 is responsible. In this chapter, the clinical manifestations of pleural effusions are first discussed. Then, the various tests used in the differential diagnosis of pleural effusions are reviewed. In Chapter 8, recommendations are given for a systematic approach to the patient with an undiagnosed pleural effusion.
CLINICAL MANIFESTATIONS
The presence of moderate-to-large amounts of pleural fluid produces symptoms and characteristic changes on physical examination.
Symptoms
The symptoms of a patient with a pleural effusion are mainly dictated by the underlying process causing the effusion. Many patients have no symptoms referable to the effusion. When symptoms are related to the effusion, they arise either from inflammation of the pleura, from compromise of pulmonary mechanics, from interference with gas exchange, or, on rare occasions, from decreased cardiac output. A pleural effusion associated with pleuritic chest pain indicates inflammation of the pleura, specifically, the parietal pleura as the visceral pleura does not have pain fibers. Some patients with pleural effusions experience a dull, aching chest pain rather than pleuritic chest pain. This symptom is very suggestive that the patient has pleural malignancy (1). The presence of either pleuritic chest pain or dull, aching chest pain indicates that the parietal pleura is probably involved and that the patient has an exudative pleural effusion.
Ordinarily, the pain associated with pleural disease is well localized and coincides with the affected area of the pleura because the parietal pleura is innervated mostly by the intercostal nerves. At times, however, pleuritic pain is referred to the abdomen because intercostal nerves are also distributed to the abdomen. A notable exception to the localization of the pain occurs when the central portion of the diaphragmatic pleura is involved. The nerve supply to this portion of the parietal pleura is the phrenic nerve; therefore, inflammation of the central portion of the diaphragm is referred to the tip of the ipsilateral shoulder. Pleuritic pain felt simultaneously in the lower chest and ipsilateral shoulder is pathognomonic of diaphragmatic involvement.
A second symptom of pleural effusion is a dry, nonproductive cough. The mechanism producing the cough is not clear, although it may be related to pleural inflammation. Alternately, lung compression by the fluid may bring opposing bronchial walls into contact, stimulating the cough reflex.
The third symptom of pleural effusion is dyspnea. A pleural effusion acts as a space-occupying process in the thoracic cavity and therefore reduces all subdivisions of lung volumes. Small-to-moderate-sized pleural effusions displace rather than compress the lung and have little effect on pulmonary function (2). Larger pleural effusions obviously cause a significant reduction in lung volumes, but the improvement in pulmonary function following therapeutic thoracentesis is much less than what one would anticipate. We obtained spirometry before and 24 hours after thoracentesis in 26 patients from whom a mean of 1,740 mL pleural fluid was withdrawn (3). In these patients, the mean vital capacity improved 410 ± 390 mL. Patients in this
study, with higher pleural pressures after the removal of 800 mL pleural fluid and patients with smaller decreases in the pleural pressure after the removal of 800 mL pleural fluid, had greater improvements in the forced vital capacity (FVC) after thoracentesis.
study, with higher pleural pressures after the removal of 800 mL pleural fluid and patients with smaller decreases in the pleural pressure after the removal of 800 mL pleural fluid, had greater improvements in the forced vital capacity (FVC) after thoracentesis.
Associated parenchymal disease probably explains this small increase in pulmonary function following therapeutic thoracentesis. The degree of dyspnea is frequently out of proportion to the size of the pleural effusion. Often, this feature is the result of compromised diaphragmatic function due to the weight of fluid on the diaphragm. At times, the diaphragm becomes inverted and this usually results in disproportionate dyspnea (4). Either pleuritic chest pain, with the resultant splinting, or concomitant parenchymal disease can also be responsible for the disproportionate dyspnea. When the pleural effusion is large, ventricular filling may be impeded, leading to decreased cardiac output and dyspnea (5). Arterial blood gases usually remain at clinically acceptable levels whatever the size of the effusion (6) because of the reflex reduction in perfusion to the lung underlying the effusion.
Physical Examination
When a patient presents with chest symptoms, the physical examination is useful in suggesting whether pleural fluid is present (7). In such an examination, attention should be paid to the relative sizes of the hemithoraces and the intercostal spaces. If the pleural pressure is increased on the side of the effusion, that hemithorax will be larger, and the usual concavity of the intercostal spaces will be blunted or even convex. In contrast, if the pleural pressure on the side of the effusion is decreased, as with obstruction of a major bronchus or a trapped lung, the ipsilateral hemithorax will be smaller, and the normal concavity of the intercostal spaces will be exaggerated. In addition, with inspiratory efforts, the intercostal spaces retract. Enlargement of the hemithorax with bulging of the intercostal spaces is an indication for therapeutic thoracentesis to relieve the increased pleural pressure. Signs of decreased pleural pressure are a relative contraindication to therapeutic thoracentesis because the decreased pleural pressure can lead to reexpansion pulmonary edema (8). Of course, in many patients with pleural effusions, the hemithoraces are equal in size and the intercostal spaces are normal.
Palpation of the chest in patients with pleural effusions is useful in delineating the extent of the effusion. In areas of the chest where pleural fluid separates the lung from the chest wall, tactile fremitus is absent or attenuated because the fluid absorbs the vibrations emanating from the lung. Tactile fremitus is much more reliable than percussion for identifying both the upper border of the pleural fluid and the proper site to attempt a thoracentesis. With a thin rim of fluid, the percussion note may still be resonant, but the tactile fremitus is diminished. Palpation may also reveal that the cardiac point of maximum impulse is shifted to one side or the other. With large left pleural effusions, the cardiac point of maximum impulse may not be palpable. In patients with pleural effusions, the position of the trachea should always be ascertained because it indicates the relationship between the pleural pressures in the two hemithoraces.
The percussion note over a pleural effusion is dull or flat. The dullness is maximum at the lung bases where the thickness of the fluid is the greatest. As mentioned earlier, however, the percussion note may not be duller if only a thin rim of fluid is present. Light percussion is better than heavy percussion for identifying small amounts of pleural fluid. If the dullness to percussion shifts as the position of the patient is changed, one can be almost certain that free pleural fluid is present (9).
Auscultation over the pleural fluid characteristically reveals decreased or absent breath sounds. Near the superior border of the fluid, however, breath sounds may be accentuated and may take on a bronchial characteristic. This phenomenon has been attributed to increased conductance of breath sounds through the partially atelectatic lung compressed by the fluid (10). This accentuation of breath sounds does not mean that an associated parenchymal infiltrate is present. Auscultation may also reveal a pleural rub. Pleural rubs are characterized by coarse, creaking, leathery sounds most commonly heard during the latter part of inspiration and the early part of expiration, producing a to-and-fro pattern of sound. Pleural rubs, caused by the rubbing together of the roughened pleural surfaces during respiration, are often associated with local pain on breathing that subsides with breath-holding. Pleural rubs often appear as pleural effusions diminish in size, either spontaneously or as a result of treatment, because the pleural fluid is no longer present between the roughened pleural surfaces.
It is important to realize that an elevated hemidiaphragm can produce all the classic physical findings associated with a pleural effusion. Obviously, the chest is not the only structure that should be examined when evaluating a patient with a pleural effusion; clues to
the origin of the effusion are often present elsewhere. The effusion is probably due to congestive heart failure (CHF) if the patient has cardiomegaly, neck vein distension, or peripheral edema. Signs of joint disease or subcutaneous nodules suggest that the pleural effusion is due to rheumatoid disease or lupus erythematosus. An enlarged, nontender nodular liver or the presence of hypertrophic osteoarthropathy suggests metastatic disease, as do breast masses or the absence of a breast. Abdominal tenderness suggests a subdiaphragmatic process, whereas tense ascites suggests cirrhosis and a hepatothorax. Lymphadenopathy suggests lymphoma, metastatic disease, or sarcoidosis.
the origin of the effusion are often present elsewhere. The effusion is probably due to congestive heart failure (CHF) if the patient has cardiomegaly, neck vein distension, or peripheral edema. Signs of joint disease or subcutaneous nodules suggest that the pleural effusion is due to rheumatoid disease or lupus erythematosus. An enlarged, nontender nodular liver or the presence of hypertrophic osteoarthropathy suggests metastatic disease, as do breast masses or the absence of a breast. Abdominal tenderness suggests a subdiaphragmatic process, whereas tense ascites suggests cirrhosis and a hepatothorax. Lymphadenopathy suggests lymphoma, metastatic disease, or sarcoidosis.
SEPARATION OF TRANSUDATIVE FROM EXUDATIVE EFFUSIONS
The accumulation of clinically detectable quantities of pleural fluid is distinctly abnormal. A diagnostic thoracentesis (see Chapter 28) should be attempted whenever the thickness of pleural fluid on ultrasound or the decubitus radiograph is greater than 10 mm or whenever loculated pleural fluid is demonstrated with ultrasound unless the etiology of the effusion is known. A properly performed diagnostic thoracentesis takes less than 10 minutes and should cause no more morbidity than a venipuncture. The information available from examination of the pleural fluid is invaluable in the management of the patient.
Pleural effusions have classically been divided into transudates and exudates (11). A transudative pleural effusion develops when the systemic factors influencing the formation or absorption of pleural fluid are altered so that pleural fluid accumulates. The pleural fluid is a transudate. The fluid may originate in the lung, the pleura, or the peritoneal cavity (12). The permeability of the capillaries to proteins is normal in the area where the fluid is formed. Examples of conditions producing transudative pleural effusions are left ventricular failure producing increased pulmonary interstitial fluid and a resulting pleural effusion, ascites due to cirrhosis with movement of fluid through the diaphragm, and decreased serum oncotic pressure with hypoproteinemia.
In contrast, an exudative pleural effusion develops when the pleural surfaces or the capillaries in the location where the fluid originates are altered such that fluid accumulates. The pleural fluid is an exudate. The most common causes of exudative pleural effusions are pleural malignancy, parapneumonic effusions, and pulmonary embolism.
The first question to ask in assessing a patient with a pleural effusion is whether that effusion is a transudate or an exudate. If the effusion is a transudate, no further diagnostic pleural procedures are necessary, and therapy is directed to the underlying CHF, cirrhosis, or nephrosis. Alternately, if the effusion proves to be an exudate, a more extensive diagnostic investigation is indicated to delineate the cause of the effusion. It has been shown that pulmonary specialists are not very accurate at doing this on the basis of clinical history, physical examination, and radiographic findings (13).
For many years, a pleural fluid protein level of 3.0 g/dL was used to separate transudates from exudates, with exudative pleural effusions characterized by a protein level above 3.0 g/dL (14,15). Use of this one simple test led to the misclassification of approximately 10% of pleural effusions (14,15,16). Light et al. subsequently demonstrated that with the use of simultaneously obtained serum and pleural fluid protein and lactic acid dehydrogenase (LDH) values, 99% of pleural effusions could be correctly classified as either transudates or exudates (16). Exudative pleural effusions meet at least one of the following criteria, whereas transudative pleural effusions meet none (Light’s criteria):
Pleural fluid protein divided by serum protein greater than 0.5
Pleural fluid LDH divided by serum LDH greater than 0.6
Pleural fluid LDH greater than two thirds of the upper limit of normal serum LDH
Subsequent studies have demonstrated that Light’s criteria classify virtually all exudates correctly but misclassify about 25% of transudates as exudates.
In recent years, other tests have been proposed for the separation of transudates from exudates. The tests that have been proposed to indicate a pleural exudate have included a pleural fluid cholesterol greater than 60 mg/dL (17,18), a pleural fluid cholesterol greater than 45 mg/dL (19), a gradient of less than 1.2 g/dL for the difference in the pleural fluid and serum albumin level (20), a pleural fluid-to-serum bilirubin ratio above 0.6 (21), a high pleural fluid viscosity (22), a high level of oxidative stress markers (23), soluble leukocyte selectin (24), cytokines (25), uric acid (26), and a pleural fluid-to-serum cholinesterase ratio above 0.23 (27).
Two subsequent reports (28,29) have compared Light’s criteria with some of the other proposed tests and have concluded that Light’s criteria best separate exudates and transudates. In the study of Romero et al. (28) of 297 patients including 44 transudates and 253 exudates, Light’s criteria were superior to cholesterol measurement in making the distinction. In this study
with Light’s criteria, 98% of the exudates and 77% of the transudates were correctly classified (28). In a subsequent study of 393 patients including 123 with transudates and 270 with exudates from South Africa (29), Light’s criteria were found to be superior to the serum effusion albumin gradient, the effusion cholesterol concentration, and the pleura fluid-to-serum bilirubin ratio (29). Again in this study, Light’s criteria identified 98% of the exudates correctly, but they were less accurate in identifying transudates, misclassifying 19 of 112 (17%) (29). Two additional studies (30,31) have come to similar conclusions. It is unlikely that the pleural fluid cholesterol measurement will provide additional information to the ratio of the pleural fluid to the serum protein because the pleural fluid cholesterol level can be accurately predicted from the serum cholesterol and the ratio of the pleural fluid to the serum protein level (32).
with Light’s criteria, 98% of the exudates and 77% of the transudates were correctly classified (28). In a subsequent study of 393 patients including 123 with transudates and 270 with exudates from South Africa (29), Light’s criteria were found to be superior to the serum effusion albumin gradient, the effusion cholesterol concentration, and the pleura fluid-to-serum bilirubin ratio (29). Again in this study, Light’s criteria identified 98% of the exudates correctly, but they were less accurate in identifying transudates, misclassifying 19 of 112 (17%) (29). Two additional studies (30,31) have come to similar conclusions. It is unlikely that the pleural fluid cholesterol measurement will provide additional information to the ratio of the pleural fluid to the serum protein because the pleural fluid cholesterol level can be accurately predicted from the serum cholesterol and the ratio of the pleural fluid to the serum protein level (32).
The number of false positives and false negatives with any test depends upon the cutofflevel chosen for the identification of an exudate. If a high cutoff level is chosen, all transudates will be identified correctly, whereas if a low level is chosen, all exudates will be identified correctly. Light et al. originally developed Light’s criteria with the goal to identify all exudates correctly, and the criteria are remarkably effective in achieving this goal.
An alternative approach is to select the cutoff level that will correctly identify the highest percentage of patients. Using this approach, Heffner et al. (33) analyzed the data from eight studies with a total of 1,448 patients and concluded that the best cutoff levels for the different pleural fluid tests were as follows: protein ratio 0.5, pleural fluid LDH 0.45 of the upper limits of the normal for serum, LDH ratio 0.45.
As demonstrated in the preceding text, Light’s criteria identify approximately 25% of transudative effusions as exudates. This mislabeling occurs most commonly when patients with CHF are treated with diuretics before thoracentesis is performed (34). These mislabeled transudates barely meet the exudative criteria. The protein ratio is less than 0.65, the LDH ratio is less than 1.0 and the level of the LDH is less than the upper limit of normal. How can these mislabeled transudates be identified? One possible means is to examine the gradient between the serum and the pleural fluid protein levels. If this gradient is greater than 3.1 g/dL, one can presume that the fluid is actually a transudate (34). In a previous edition of this book, it was recommended that an albumin gradient of 1.2 g/dL (29) rather than the protein gradient of 3.1 g/dL be used. Bielsa et al. (35) reported that the albumin gradient identified more of these effusions correctly than did the protein gradient. It is suggested that the protein gradient first be examined because it is already available from Light’s criteria. If the protein gradient is not definitive, then one may use the albumin gradient or the NT-pro-BNP.
In the discussion in the preceding text, pleural effusions have been dichotomized into transudates or exudates on the basis of a single cutoff point. An alternative approach is to use likelihood ratios for identifying whether a pleural fluid is a transudate or an exudate (36,37). The idea behind this approach is that the higher a value, for example, the pleural fluid LDH, the more likely the effusion is to be an exudate and the lower the value, the less likely the effusion is to be an exudate. Heffner et al. have derived multilevel (36) and continuous (37) likelihood ratios for the usual biochemical tests used to differentiate transudates and exudates. When these likelihood ratios are used in conjunction with pretest probabilities using Bayes’ theorem, posttest probabilities can be derived (37). Difficulties in using this approach occur because the pretest probabilities vary significantly from physician to physician, and most physicians do not understand the mathematics involved. This approach does emphasize that it is important to take into consideration the absolute value of the measurements. Very high or very low values are almost always indicative of exudates and transudates, respectively, whereas values near the cutoff levels can be associated with either transudates or exudates.
The following approach is recommended for determining whether a pleural effusion is a transudate or an exudate. First assess the fluid with Light’s criteria. The higher the value for the protein ratio, the LDH ratio, and the absolute value of the LDH, the more likely the fluid is an exudate. If the fluid meets the criteria for a transudative effusion, it is a transudate. If the fluid meets the criteria for an exudative effusion by only a small margin and the clinical picture is compatible with a transudative effusion, measure the protein gradient between the serum and pleural fluid. If this value is greater than 3.1 g/dL, then the fluid can be relabeled a transudate. An alternative approach is to measure the brain natriuretic peptide (BNP) level in the pleural fluid or the serum. If this is greater than 1,500 pg/mL, the diagnosis of CHF is established (see the discussion on NT-pro-BNP later in this chapter).
Specific Gravity
The specific gravity of the pleural fluid as measured with a hydrometer was used in the past to separate transudates from exudates (38) because it was a simple
and rapid method of estimating the protein content of the fluid. A specific gravity of 1.015 corresponds to a protein content of 3 g/dL, and this value was used to separate transudates from exudates (38). At the present time, the specific gravity of pleural fluid is usually measured with a refractometer rather than a hydrometer. Unfortunately, the scale on the commercially available refractometers is calibrated for the specific gravity of urine rather than for pleural fluid. A reading of 1.020 on the urine specific gravity scale corresponds to a pleural fluid protein level of 3 g/dL. However, there is a scale on the same refractometer for protein levels that is valid for pleural fluid. Because the only reason to measure specific gravity is to estimate the protein level, and because the pleural fluid specific gravity measurement is extraneous and confusing, it should no longer be ordered (39). However, a rapid estimate of the pleural fluid protein content can be obtained at the patient’s bedside with the protein scale on the refractometer (39).
and rapid method of estimating the protein content of the fluid. A specific gravity of 1.015 corresponds to a protein content of 3 g/dL, and this value was used to separate transudates from exudates (38). At the present time, the specific gravity of pleural fluid is usually measured with a refractometer rather than a hydrometer. Unfortunately, the scale on the commercially available refractometers is calibrated for the specific gravity of urine rather than for pleural fluid. A reading of 1.020 on the urine specific gravity scale corresponds to a pleural fluid protein level of 3 g/dL. However, there is a scale on the same refractometer for protein levels that is valid for pleural fluid. Because the only reason to measure specific gravity is to estimate the protein level, and because the pleural fluid specific gravity measurement is extraneous and confusing, it should no longer be ordered (39). However, a rapid estimate of the pleural fluid protein content can be obtained at the patient’s bedside with the protein scale on the refractometer (39).
Other Characteristics of Transudates
Most transudates are clear, straw colored, nonviscid, and odorless. It takes a pleural fluid RBC count of more than 10,000/mm3 to give the pleural fluid a pinkish tinge. Approximately 15% have RBC counts above this level. Therefore, the discovery of bloodtinged pleural fluid does not mean that the fluid is not a transudate. Because RBCs contain a large amount of LDH, one might suppose that the LDH level in a blood-tinged or bloody transudative pleural effusion would be so elevated that it would meet the criteria for an exudative pleural effusion. Such does not appear to be the case, however. The LDH isoenzyme present in RBCs is LDH-1, and in one study of 23 patients with bloody pleural effusions (pleural fluid red cell counts greater than 100,000/mm3), the fraction of LDH-1 in the pleural fluid was only slightly increased (40).
The pleural fluid white blood cell (WBC) count of most transudates is less than 1,000/mm3, but approximately 20% have WBC counts that exceed 1,000/mm3 (41). Pleural fluid WBC counts above 10,000/mm3 are rare with transudative pleural effusions. The differential WBC count in transudative pleural effusions may be dominated by polymorphonuclear leukocytes, small lymphocytes, or other mononuclear cells. In a series of 47 transudative effusions, 6 (13%) had more than 50% polymorphonuclear leukocytes, 16 (34%) had predominantly small lymphocytes, 22 (47%) had predominantly other mononuclear cells, and 3 (6%) had no single predominant cell type (41). The pleural fluid glucose level is similar to the serum glucose level, and the pleural fluid amylase level is low (42). The pleural fluid pH with transudative pleural effusions is higher than the simultaneously obtained blood pH (43), probably because of active transport of bicarbonate from the blood into the pleural space (44).
Probrain Natriuretic Peptide (BNP)
The levels of NT-pro-BNP and BNP in the pleural fluid are useful in establishing that the etiology of the pleural effusion is CHF. When the ventricles are subjected to increased pressure or volume, these peptides are released into the circulation (45). The biologically active BNP and the larger amino terminal part NT-pro-BNP are released in equimolar amounts into the circulation (45). The serum levels of BNP are used to help establish the diagnosis of CHF. In clinical practice, levels above 500 pg/mL are considered diagnostic of CHF whereas levels below 100 pg/mL are thought to make the diagnosis of CHF unlikely (46).
Porcel et al. (47) first demonstrated that the pleural fluid levels of NT-pro-BNP are elevated in patients with heart failure. They measured NT-pro-BNP levels in 117 pleural fluid samples with the following diagnoses: CHF in 44 samples, malignancy in 25, tuberculous pleuritis in 20, hepatic hydrothorax in 10, and miscellaneous in 18. The mean NT-pro-BNP fluid level in the CHF patients (6,931 pg/mL) was significantly higher than the 551 pg/mL in the patients with hepatic hydrothorax and the 292 pg/mL in the patients with exudative pleural effusions (45). When a cutofflevel of 1,500 pg/mL was used, the sensitivity was 91% and the specificity was 93% for the diagnosis of CHF. We have compared the pleural fluid NT-pro-BNP levels in 10 patients each with effusions due to CHF, pulmonary embolism, postcoronary artery bypass surgery, and malignancy (48).
All the patients with CHF had NT-pro-BNP levels above 1,500 pg/mL, and none of the other patients had NT-pro-BNP levels this high (48). It should be emphasized that the serum or pleural fluid BNP and NT-pro-BNP cannot be used interchangeably in the diagnosis of pleural effusions due to CHF (49). The BNP levels are only about 10% of the NT-pro-BNP levels. There is not a close correlation between the BNP levels and the NT-pro-BNP levels (r = 0.78) (50). Moreover, the diagnostic usefulness of the NT-pro-BNP in making the diagnosis of heart failure is superior to that of the BNP (50,51). The accuracy of NT-pro-BNP in making the diagnosis of pleural effusion due to heart failure was attested to
in a meta-analysis of 10 studies with a total of 1,120 patients in which the pooled sensitivity and specificity were 94% and 94%, respectively (52).
in a meta-analysis of 10 studies with a total of 1,120 patients in which the pooled sensitivity and specificity were 94% and 94%, respectively (52).
The pleural fluid NT-pro-BNP is also superior to the BNP and the protein gradient in identifying patients with heart failure who meet Light’s criteria for exudates (50). In one study of 20 patients with heart failure who met Light’s criteria for exudates, 18 had NT-pro-BNP levels above 1,300, 16 had NT-pro-BNP levels above 1,500, but only 10 had serum pleural fluid protein gradient greater than 3.1 g/dL (50).
Other workers have demonstrated that there is a close relationship between the levels of NT-pro-BNP in the pleural fluid and serum. Han et al. (53) measured the NT-pro-BNP levels in 240 patients and reported that the correlation coefficient between the pleural and serum NT-pro BNP was 0.928. In a second study, Kolditz et al. (54) measured the serum and pleural fluid NT-pro-BNP levels in 93 patients including 25 with CHF. They confirmed the results of the above study in that the levels of serum and pleural fluid NT-pro-BNP were again closely correlated (r2 = 0.90). From the latter two studies it appears that measurement of the pleural fluid NT-pro-BNP levels provides no additional information beyond the serum measurements.
GENERAL TESTS FOR DIFFERENTIATING CAUSES OF EXUDATES
Appearance of Fluid
The gross appearance of the pleural fluid frequently yields useful diagnostic information. The color, turbidity, viscosity, and odor should be described. Most transudative and many exudative pleural effusions are clear, straw colored, nonviscid, and odorless. Any deviations should be noted and investigated.
A reddish color indicates that blood is present, and a brownish tinge indicates that the blood has been present for a prolonged period. If the pleural fluid is blood tinged, the pleural fluid RBC count is between 5,000 and 10,000/mm3. If the pleural fluid appears grossly bloody, a hematocrit should be obtained to determine whether the patient has a hemothorax (see Chapter 25).
On rare occasions, the pleural fluid can be black. Black pleural fluid has been reported with infection due to Aspergillus niger, infection due to Rhizopus oryzae, pigment laden macrophages following massive bleeding due to metastatic carcinoma (55) and melanoma (56).
Turbid pleural fluid can occur from either increased cellular content or increased lipid content. These two entities can be differentiated if the pleural fluid is centrifuged and the supernatant examined. If turbidity remains after centrifugation, it is in all probability due to increased lipid content, and the fluid should be sent for lipid analysis (see the discussion later in this chapter).
Alternately, if the supernatant is clear, the original turbidity was due to increased numbers of cells or other debris. The discovery of pleural fluid that looks like chocolate sauce or anchovy paste is suggestive of amebiasis with a hepatopleural fistula (57). This appearance is due to the presence of a mixture of blood, cytolyzed liver tissue, and small solid particles of liver parenchyma that have resisted dissolution.
A pleural fluid with a high viscosity is suggestive of malignant mesothelioma; the high viscosity is secondary to an elevated pleural fluid hyaluronic acid level. Of course, the fluid from a pyothorax is also viscid because of the large amounts of cells and debris in the fluid.
The odor of all pleura fluids should be noted. One can immediately establish two diagnoses by smelling the pleural fluid. A feculent odor indicates that the patient has a bacterial infection of the pleural space that is probably anaerobic. If the pleural fluid smells like urine, the patient probably has a urinothorax.
Red Blood Cell Count
Only 5,000 to 10,000 RBCs per mm3 need be present to impart a red color to pleural fluid. If a pleural effusion has a total volume of 500 mL and the RBC count in the peripheral blood is 5 million/mm3, a leak of only 1 mL blood into the pleural space will result in a blood-tinged pleural effusion. It is probably for this reason that the presence of blood-tinged or serosanguineous pleural fluid has little diagnostic significance. More than 15% of transudative and more than 40% of all types of exudative pleural fluids are blood tinged (41); that is, they have pleural fluid RBC counts between 5,000 and 100,000/mm3.
Occasionally, pleural fluid obtained by diagnostic thoracentesis appears grossly bloody. In such cases, one can assume that the RBC count in the pleural fluid is above 100,000/mm3. One should obtain a hematocrit on such pleural fluids to document the amount of blood in the pleural fluid. If the hematocrit of the pleural fluid is greater than 50% of the peripheral hematocrit, a hemothorax is present, and one should consider inserting a chest tube (see Chapter 25). Usually, the hematocrit of bloody pleural fluid is much lower than
what one would expect from its gross appearance. If a pleural fluid hematocrit is not available, its estimate can be obtained by dividing the pleural fluid RBC count by 100,000.
what one would expect from its gross appearance. If a pleural fluid hematocrit is not available, its estimate can be obtained by dividing the pleural fluid RBC count by 100,000.
The presence of bloody pleural fluid suggests one of three diagnoses, namely, malignant disease, trauma, or pulmonary embolization. In a series of 22 bloody pleural effusions that I observed on medical wards, 12 were due to malignant disease, 5 to pulmonary embolization, 2 to trauma, and 2 to pneumonia, and 1 was a transudative effusion secondary to cirrhosis (30). The traumatic origin of the pleural effusion may not be obvious, particularly when the patient is on a medical ward. The patient may have broken a rib while coughing or suffered trauma during an episode of inebriation that is not remembered.
At times, it is unclear whether blood in the pleural fluid resulted from or was present before the thoracentesis. If the blood is a result of the thoracentesis, the degree of red discoloration of the fluid frequently is not uniform throughout the course of aspiration. Examination of the fluid microscopically may also be useful. If the RBCs were present before the thoracentesis, the macrophages in the pleural fluid usually contain hemoglobin inclusions. Although the levels of D-dimer in the cerebrospinal fluid are useful in demonstrating whether blood in the fluid has a traumatic origin, pleural fluid levels of D-dimer are not useful in making this differentiation (58). Crenation of the RBC in the pleural fluid rarely occurs because the osmotic pressure of the pleural fluid is similar to that of serum.
White Blood Cell Count
Although WBC counts on the pleural fluid were traditionally performed manually, we have shown that the automated counters provide accurate pleural fluid WBC counts (59). The pleural fluid for cell counts and differentials should be collected in a test tube with an anticoagulant (59). If the pleural fluid is collected in plastic or glass tubes without an anticoagulant, the fluid may clot or the cells may clump, providing inaccurate cell counts and differentials (59).
The pleural fluid WBC count is of limited diagnostic use. Most transudates have WBC counts below 1,000/mm3, whereas most exudates have WBC counts above 1,000/mm3 (16). Pleural fluid WBC counts above 10,000/mm3 are most commonly seen with parapneumonic effusions, but they are also seen with many other diseases (41), as shown in Table 7.1. I have seen pleural fluid WBC counts above 50,000/mm3 with both pancreatic disease and pulmonary embolization. With grossly purulent pleural fluid, the pleural fluid WBC count is frequently much lower than what one would anticipate because debris, rather than cells, accounts for much of the turbidity.
Differential White Cell Count
Examination of a Wright’s stain of pleural fluid is one of the most informative tests on pleural fluid. Because the pleural fluid WBC count is frequently less than 5,000/mm3, it is useful to concentrate the cells before staining. This is easily accomplished by centrifuging approximately 10 mL of fluid and then resuspending the button of cells in approximately 0.5 mL of supernatant. After thorough mixing, slides that are similar to those for examining peripheral blood are made and stained in the usual way. Occasionally, large amounts of fibrinogen adhere to the cells. In such cases, resuspension in saline solution, followed by centrifugation, is indicated in order to evaluate cellular morphologic features. Automatic cell counters do not provide sufficiently accurate differential cell counts for clinical use (59,60), presumably because of the high number of mesothelial, lymphoid, and tumors cells in pleural fluid.
TABLE 7.1 ▪ Etiology of 25 Effusions Containing More than 10,000 WBCs/mm3 | ||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||||||||||||||||
Although most laboratories divide pleural fluid WBCs into polymorphonuclear leukocytes and mononuclear cells, I prefer to divide them into four categories—polymorphonuclear leukocytes, lymphocytes, other mononuclear cells, and eosinophils— because of the diagnostic significance of small lymphocytes (see the discussion on lymphocytes later in this chapter). The mononuclear cells include mesothelial cells, macrophages, plasma cells, and malignant cells. Excellent color plates demonstrating the morphologic and staining characteristics of the different cells in pleural effusions are contained in the monograph by Spriggs and Boddington (61).
Neutrophils
Because neutrophils are the cellular component of the acute inflammatory response, they predominate in pleural fluid resulting from acute inflammation as with pneumonia, pancreatitis, pulmonary embolization, subphrenic abscess, and early tuberculosis. Although more than 10% of transudative pleural effusions contain predominantly neutrophils, pleural fluid neutrophilia in transudates has no clinical significance (41). The significance of neutrophils in an exudative pleural effusion is that they indicate acute inflammation of the pleural surface.
Interleukin 8 (IL-8) appears to be one of the primary chemotaxins for neutrophils in the pleural space (62,63). The number of neutrophils in pleural fluid is correlated with the IL-8 level, and empyemas have the highest levels of IL-8. The addition of IL-8 neutralizing serum decreases the chemotactic activity for neutrophils in empyema fluids (63). The cellular source of IL-8 is unknown (62).
Examination of the pleural fluid neutrophils in patients with parapneumonic effusions is useful in identifying those that are infected. If pleural infection is present, the neutrophils undergo a characteristic degeneration. The nucleus becomes blurred and is no longer stained purple. The cytoplasm shows toxic granulation initially. Subsequently, the neutrophilic granules become indistinct and are then lost. Finally, only a smear cell remains (61).
Eosinophils
Approximately 7% of pleural fluids are characterized by pleural fluid eosinophilia (>10%) (64,65). Most clinicians believe that significant numbers of eosinophils (>10%) in pleural fluid should be a clue to the origin of the pleural effusion. In most instances, the pleural fluid eosinophilia is due to either air or blood in the pleural space and therefore does not contribute any diagnostic information in these situations. Charcot-Leyden crystals (66), as well as Curschmann’s spirals (67), are occasionally found in the pleural fluid of patients with pleural eosinophilia. Their presence appears to have no diagnostic significance.
The factors responsible for pleural fluid eosinophilia have been intensively studied, but there is still no unifying concept and it is likely that multiple factors are involved. In animals, it has been shown that the intrapleural injection of stem cell factor (68), plateletactivating factor (PAF) (69), endotoxin (70,71), bradykinin (72), and leukotriene B4 (69) all result in pleural fluid eosinophilia. The eosinophil influx is inhibited in some but not all situations by monoclonal antibodies (MAb) directed against IL-5.
In humans, pleural fluid from patients with eosinophilic pleural effusions stimulates bone marrow cells to form colonies of eosinophils (73,74). In addition, when eosinophils are incubated in the presence of eosinophilic pleural fluid, their survival is prolonged (73). Peripheral blood from patients with eosinophilic pleural effusions does not stimulate the bone marrow to form eosinophil colonies and does not prolong survival of eosinophils. The factor responsible for the increased colony-forming activity and the increased survival appears to be IL-5 (73,74), although IL-3 and granulocyte or macrophage colony-stimulating factor (GM-CSF) may also play a role (73). In patients with eosinophilic pleural effusions, the pleural fluid level of IL-5 is significantly correlated with the pleural fluid eosinophil count (r = 0.55) and the pleural fluid eosinophil percentage (r = 0.54) (75). Both bloody and nonbloody eosinophilic effusions have high levels of IL-5 (75). In patients with eosinophilic pleural effusions, the pleural fluid levels of vascular cell adhesion molecule (VCAM)-1 and eotaxin-3 are also significantly correlated with the eosinophil count and percentage of eosinophils in the pleural fluid (76). VCAM-1 is an adhesion molecule that is expressed on the endothelial surface and interacts with the β1-integrins expressed on the eosinophil surface to facilitate eosinophil tissue migration (76).
The source of the IL-5 appears to be the CD4+ lymphocyte in the pleural fluid (74), but the eosinophils in the pleural fluid may themselves also produce IL-5 (73). The source of the eosinophils in eosinophilic pleural effusions appears to be the bone marrow; no progenitor cells are present in the pleural space. It is not known what stimulates the CD4+ lymphocytes to produce the IL-5. However, it probably results
from another cytokine because the intrapleural injection of IL-2 into malignant pleural effusions results in an eosinophilic pleural effusion with a high level of IL-5 (77). There are factors other than IL-5 that recruit eosinophils to the pleural space. Antibodies to IL-5 eliminate the eosinophilic influx to an allergen but not to endotoxin in the mouse (78). However, antibodies to gamma delta lymphocytes eliminate the eosinophilic response to endotoxin (71).
from another cytokine because the intrapleural injection of IL-2 into malignant pleural effusions results in an eosinophilic pleural effusion with a high level of IL-5 (77). There are factors other than IL-5 that recruit eosinophils to the pleural space. Antibodies to IL-5 eliminate the eosinophilic influx to an allergen but not to endotoxin in the mouse (78). However, antibodies to gamma delta lymphocytes eliminate the eosinophilic response to endotoxin (71).
The most common cause of pleural fluid eosinophilia is air in the pleural space. In a series of 127 cases with more than 20% eosinophils in the pleural fluid, 81 (64%) were thought to have pleural fluid eosinophilia secondary to air in the pleural space (61). In a review of 343 pleural effusions with greater than 10% eosinophils, 95 (28%) had air in the pleural space (79). It is likely that the pleural fluid eosinophilia in many other patients in this series was also due to the introduction of air into the pleural space during a prior thoracentesis. On numerous occasions over the past three decades, I have seen patients who had no pleural fluid eosinophilia at the initial thoracentesis but who had many eosinophils at a subsequent thoracentesis. In each case, a small pneumothorax resulted from the initial thoracentesis. When patients with spontaneous pneumothorax undergo thoracotomy, a reactive eosinophilic pleuritis frequently exists in the resected parietal pleura (80). It should be noted, however, that in two recent series (64,81) the prevalence of eosinophilic pleural effusion did not increase after a thoracentesis or a pleural biopsy.
The mechanism responsible for the pleural fluid eosinophilia in response to air in the pleural space is unknown but is probably related to IL-5. Smit et al. (82) measured the percentage of eosinophils and the levels of IL-5 in 23 patients with pneumothorax and pleural fluid. They found that IL-5 level and the eosinophil concentration in the pleural fluid were highly correlated (r = 0.84) and that the eosinophil percentage tended to increase with time, with a mean of less than 5% in the first 24 hours, 20% at days 1 to 3, 40% at days 4 to 7, and 50% after day 7 (82). There was no relationship between the PAF level or the monocyte chemotactic protein-1 levels and the eosinophils. In these fluids, IL-8 was not detectable. When air is injected into the pleural space of a mouse, pleural fluid eosinophilia occurs within 30 minutes and peaks at 48 hours (83).
The second most common cause of pleural fluid eosinophilia is blood in the pleural space. Following traumatic hemothorax, pleural fluid eosinophils do not usually become numerous until the second week (61). There is frequently an associated peripheral blood eosinophilia that does not disappear until the pleural effusion is completely resolved (84). The pleural effusions associated with pulmonary embolization are frequently bloody and contain numerous eosinophils (85). Bloody pleural fluids due to malignant disease are not usually characterized by eosinophilia (41,61). In a study conducted by my colleagues and me of bloody pleural effusions, none of the 11 cases of malignant pleural effusions with pleural fluid RBC counts greater than 100,000/mm3 had more than 10% eosinophils (41). Patients with pleural effusions postcoronary artery bypass graft (CABG) surgery, frequently have bloody pleural effusions in the first few weeks after surgery. In these effusions, the pleural fluid IL-5 level is higher than the corresponding serum level and there is a significant correlation between the pleural fluid and serum IL-5 level (86). Moreover, the pleural fluid IL-5 levels are significantly correlated with the pleural fluid eosinophil counts (86). In addition, the pleural fluid eotaxin-3 levels are significantly higher than the serum levels and the pleural fluid eotaxin-3 levels significantly correlate with the pleural fluid eosinophil count (86).
If the patient has neither blood nor air in their pleural space, what is the significance of an eosinophilic pleural effusion? Kalomenidis and Light (87) reviewed the etiology of 392 cases of eosinophilic pleural effusions when cases associated with pleural air and/or blood were excluded. They reported that the most common diagnosis was idiopathic (39.8%), followed by malignancy (17%), parapneumonic (12.5%), transudates (7.9%), tuberculosis (5.6%), pulmonary embolism (4.3%), and others (12.8%). In a recent study of 135 patients with eosinophilic pleural effusions from a single institution, the following diseases were responsible: malignancy 34.8%, infections 19.2%, unknown 14.1%, posttraumatic 8.9%, and miscellaneous 23.0%. (64). The incidence of malignancy was significantly higher in patients with a lower (<40%) pleural fluid eosinophil percentage (64).
If neither air nor blood is present in the pleural space, several unusual diagnoses should be considered. Pleural eosinophilia is common in patients with asbestos-related pleural effusions. In a review of eosinophilic pleural effusions (79), 15 of 29 (52%) asbestos-related pleural effusions had more than 10% eosinophils in the pleural fluid. Patients with eosinophilic pneumonia frequently have pleural effusions and the mean eosinophil percentage in the pleural fluid was 38% in one study (88). The pleural effusions secondary to drug reactions are frequently eosinophilic.
Offending drugs include dantrolene, bromocriptine, and nitrofurantoin (see Chapter 22). Pleural effusions secondary to parasitic diseases (see Chapter 15) such as paragonimiasis (68), hydatid disease (89), amebiasis (61), or ascariasis (61) frequently contain a large percentage of eosinophils. Lastly, the pleural effusion associated with the Churg-Strauss syndrome (see Chapter 21) is eosinophilic (90).
Offending drugs include dantrolene, bromocriptine, and nitrofurantoin (see Chapter 22). Pleural effusions secondary to parasitic diseases (see Chapter 15) such as paragonimiasis (68), hydatid disease (89), amebiasis (61), or ascariasis (61) frequently contain a large percentage of eosinophils. Lastly, the pleural effusion associated with the Churg-Strauss syndrome (see Chapter 21) is eosinophilic (90).
If none of the foregoing rare diseases is causing the pleural effusion, the following statements are pertinent to patients with eosinophilic pleural effusions. If the patient has pneumonia and pleural effusion, the presence of pleural fluid eosinophilia is a good prognostic sign because such an effusion rarely becomes infected. The origin of approximately 40% of eosinophilic effusions is not established, and these effusions resolve spontaneously.
Basophils
Basophilic pleural effusions are distinctly uncommon. I have not seen a pleural effusion that contained more than 2% basophils. A few basophils are usually present in pleural effusions with eosinophils. Basophil counts greater than 10% are most common with leukemic pleural involvement (61). Basophil counts greater than 5% are most common with pneumothorax, pneumonia, and transudates (91).
Lymphocytes
The discovery that more than 50% of the WBCs in an exudative pleural effusion are small lymphocytes is diagnostically important because it means that the patient probably has a malignant disease, tuberculous pleuritis, or a pleural effusion after CABG surgery. In two series (41,92) studied before the advent of CABG surgery, 96 of 211 exudative pleural effusions had more than 50% small lymphocytes. Of these 96 effusions, 90 (94%) were due to tuberculosis or malignant disease.
When the foregoing series are analyzed, almost all of the effusions secondary to tuberculosis (43 of 46), but only two thirds of the effusions secondary to malignant disease (47 of 70), had predominantly small lymphocytes. In one series of 26 patients with chronic pleural effusions post-CABG, the mean percentage of lymphocytes in the pleural fluid was 61% (93). Approximately one third of transudative pleural effusions contain predominantly small lymphocytes (41), and a lymphocytic transudative effusion is not an indication for pleural biopsy.
Several papers have assessed the diagnostic utility of separating pleural lymphocytes into T and B lymphocytes (94,95,96,97). In general, this separation has not been useful diagnostically. With most disease states, the pleural fluid contains a higher percentage of T lymphocytes (70%), a lower percentage of B lymphocytes (10%), and a higher percentage of null cells (20%) than the corresponding peripheral blood (94,95). The partitioning of lymphocytes may be useful, however, when chronic lymphatic leukemia or lymphoma is suspected. In a report of four such patients, all had more than 80% B lymphocytes in their pleural fluid (96).
The development of MAb has permitted a further subdivision of T lymphocytes. In comparison to peripheral blood, in pleural fluid the ratio of the helper and inducer cells (CD4+) to the suppressor and cytotoxic cells (CD8+) is higher, regardless of the etiology of the pleural effusion (98,99,100). Therefore, this subdivision is not useful diagnostically.
Natural killer (NK) cells are lymphocytes derived from an unimmunized host that lyse certain tumor cell lines and virus-infected cells. In general, the percentage of T lymphocytes in pleural fluid that are identified as NK cells is approximately the same (˜15%) as in the peripheral blood (101). There are two types of NK cells, CD56bright and CD56dim. In the pleural fluid, there is a much higher percentage of CD56bright cells than C56dull cells (101). The CD56dim subset has cytotoxic activity that is superior to that of the CD56bright subset. However, there is a discrepancy between the number of NK positive cells and the NK activity of the cells when patients with tuberculosis are compared with patients with malignancy. Although the number of NK cells is comparable in the two populations, there is much more NK activity in the tuberculous pleural effusions (102). Differences in the NK subset percentages may explain the variation in the NK activity (103).
Mesothelial Cells
Mesothelial cells line the pleural cavities. They frequently become dislodged from the pleural surfaces and are present in the small amount of normal pleural fluid (104). These cells are usually 12 to 30 µm in diameter, but multinucleated forms may have diameters up to 75 µm. Their cytoplasm is light blue (Fig. 7.1A) and often contains a few vacuoles. The nucleus is large (9-22 µm) and stains purplish with a uniform appearance. The nucleus usually contains one to three bright blue nucleoli (61). Mesothelial cells are discussed in more detail in Chapter 1.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
