INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Therapeutic thoracentesis (TT) is a simple and frequently performed procedure. Usually, its aim is to remove the maximum amount of pleural fluid to improve dyspnea (thus assessing whether pleurodesis is indicated) or to facilitate diagnostic techniques, such as bronchoscopy or radiological workup in patients with large effusions (1). Moreover, the withdrawal of large quantities of fluid may improve lung volumes and arterial blood gases (2). Despite these advantages, the amount of fluid removed is usually limited to 1-1.5 L because of complications, mainly reexpansion pulmonary edema and hypovolemia (1, 3). These complications are most likely caused by the negative pleural pressure (6) (PlP) and an associated increase in vascular permeability, although an inflammatory mechanism may also be involved (7). The measurement of PlP during TT has been previously reported by Light and coworkers (6). During TT, changes in PlP are a function, among others, of the volume of fluid recovered. A plot of PlP against volume displays a curve, the PlP curve. We think that appropriate knowledge of the PlP curve might be helpful to improve the clinical management of patients with large pleural effusions. However, to our knowledge, a detailed study of the PlP curve and its characteristics in different clinical situations has never been accomplished.

There are four objectives of this study:

1. To ascertain whether PlP measurement provides any clue to the etiology of the effusion.

2. To identify any variable that might predict the amount of fluid that can be removed to avoid repeated measurements of PlP.

3. To obtain insight into the relationship between PlP and volume of fluid withdrawn, the so-called PlP curve, as well as into the behavior of pleural elastance during the TT.

4. To describe the complications of therapeutic thoracentesis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sixty-one unselected patients underwent TT. In 57 patients the aim was to remove as much fluid as possible. In four patients the procedure was performed to reinforce a diagnosis of presumed trapped lung. Trapped lung is a condition characterized by the presence of a fibrous peel over the visceral pleura that prevents the underlying lung from expanding (8). A diagnosis of trapped lung was presumed when the radiological size and shape of the effusion had been unchanged during at least the previous year, the pleura seemed to be thick, the patient was clinically stable, and the patient had suffered from a known cause of trapped lung (e.g., uremic or parapneumonic effusions, chronic transudate, hemothorax). All patients gave their written consent before TT.

Therapeutic thoracentesis was performed with the patient sitting down, by puncture into the lowest zone of the effusion determined either by auscultation or by sonographic guidance. The device for measuring the PlP was similar to that described by Light and coworkers (6) except for some modifications. We used a thoracentesis kit needle (Mill-Rose), a pleural biopsy needle (Cope or Castelain), or an angiocath catheter instead of an Abram's biopsy needle, although we prefer the first one because its handling is easier and safer. In addition, the measurement of PlP was accomplished by a water column manometer designed for monitoring central venous pressure (Figure 1), with its scale reshaped to a 25 to +10 cm H₂O range; value 0 was set at the thoracic puncture level. The manometer is attached to the needle with a connecting catheter filled with saline.

View larger version (49K): [in this window] [in a new window]   Figure 1.   Water column manometer. The scale was reshaped to show a range from 25 to +10 cm H2O.

The value of PlP taken was the mean of the inspiratory and expiratory values. It was measured before removing any fluid, after the removal of every 500 ml for the first liter, then after the withdrawal of every 200 ml for the second liter, and every 100 ml thereafter until the procedure was completed. Whenever trapped lung was suspected or PlP was considered to drop fast, measurements were performed every 50 ml from the start. Clinical signs and symptoms such as chest pain, cough, and chest tightness were appraised and recorded every time PlP was obtained. Thoracentesis was discontinued when no more fluid could be obtained, the patient developed symptoms related to the removal of fluid (i.e., chest pain, cough, or chest tightness), or PlP became 20 cm H₂O or lower.

A chest X-ray was performed in the first 24 h following the procedure to detect pneumothorax, increased effusion, or new lung infiltrates suggesting pulmonary edema.

A PlP curve was obtained by plotting PlP against volume of fluid removed. Pleural space elastance (PE) has been defined as the change observed in PlP divided by the amount of fluid removed (6). PE_0.5 is referred to the PE while the first 0.5 L of fluid is aspirated, PE_E to the PE during the last 0.5 L, and PE_M to the PE in the interval between PE_0.5 and PE_E.

The Kolmogorov-Smirnov test was used to check the normal distribution of the data. Statistical comparisons were performed by Student's t test or one-way analysis of variance. Correlations were studied using the Pearson correlation coefficient. A p value < 0.05 was considered significant. Patients with a presumed diagnosis of trapped lung have been included in the etiologic analysis only because we did not proceed to aspirate the maximum amount of fluid.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There were 61 patients included in the study (45 males and 16 females); their mean age was 65 yr (range 16-93 yr). The mean pleural fluid aspirated was 1.45 L (standard deviation 0.71 L, range 0.075-3.2 L). More than 1 L was obtained in 44 patients and more than 1.5 L in 29 patients. Etiologies of the effusions and amount of pleural fluid aspirated in the different etiological groups are shown in Table 1. There was only a difference between the volume aspirated from patients with presumed trapped lung and the other etiologies. One patient with a presumed trapped lung had a small effusion (the costophrenic sulcus was obliterated, but the diaphragm was not completely covered); 11 had a medium effusion (the fluid covered up to one-third of the distance between the mediastinum and the lateral chest wall at the level of the hilar region); and 49 patients had a large one (larger than medium). No difference was found in the amount of pleural fluid aspirated in patients with medium or large effusions.

In the 57 patients without trapped lung, TT was performed with a mean delay of 26 d (range 3 to 240 d) since radiological diagnosis; in 35 of the 57 patients the delay was less than 20 d.

Therapeutic thoracentesis had to be discontinued in 29 cases because patients developed clinical symptoms related to the procedure. In 5 of them a decrease in PlP would have also justified the interruption. In 16 patients the cause of interruption was the decrease in PlP without any significant symptoms. In addition, in 10 patients TT was stopped because no more fluid could be obtained, and in 2 cases because the physician considered that too much fluid had been evacuated. Table 2 shows the values of total amount of fluid aspirated for initial PlP, PE, and PE_0.5 in the above-described groups.

Etiology of the Effusions

Figure 2 shows the initial PlP and PE for each patient within diagnostic groups. In the analysis of variance a difference (p < 0.05) was found between the mean of initial PlP of the group of patients with trapped lung on the one hand and the neoplastic or the miscellaneous groups on the other. Regarding the mean PE, it was also significantly different between the patients with trapped lung and all the other groups (neoplasms, tuberculosis, miscellaneous and unknown etiology).

View larger version (18K): [in this window] [in a new window]   Figure 2.   Etiology of the effusions related to initial pleural pressure (top) and pleural elastance (bottom). Triangles represent the patients with a neoplasm and a lobar bronchus obstructed. Trapped = trapped lung.

Prediction of the Amount of Pleural Fluid That Can Be Withdrawn

To determine predictors of the amount of pleural fluid that can be removed in a TT, we studied the correlation between this variable and initial PlP or PE_0.5. There was only a significant but weak correlation (r = 0.52) between PE_0.5 and the volume of fluid aspirated. Despite this finding, as shown in Figure 3, PE_0.5 does not reliably predict the feasibility of removing more than 1.5 L. 

View larger version (12K): [in this window] [in a new window]   Figure 3.   Pleural elastance for the first 0.5 L of fluid aspirated (PE0.5) and amount of fluid withdrawn.

A bronchoscopy was performed in 24 of the 32 patients with carcinoma, showing a lobar bronchial obstruction in four of them. The mean amount of fluid removed, initial PlP, PE, and PE_0.5 were similar in the groups with (n = 4) and without (n = 20) bronchial obstruction. We did not find any main bronchus obstruction.

Pleural Pressure Curve

Figure 4 shows the PlP curves according to the cause of discontinuation of thoracentesis. Five of 29 patients with symptoms had chest tightness or pain, 20 had cough, and 4 had both.

View larger version (28K): [in this window] [in a new window]   Figure 4.   PlP curves in the subgroups with different causes of interruption of the therapeutic thoracentesis. (A, B) Patients in whom it was stopped because of clinical symptoms. This group has been split into two parts to make interpretation easier. (C ) Decreased PlP. Dashed lines represent four patients with lobar bronchus obstruction. (D) Inability to obtain more fluid (solid lines) and other causes (dotted lines).

The analysis of partial PE (PE_0.5, PE_M, and PE_E) in patients who underwent aspiration of more than 1.5 L suggests that PE varies along the thoracentesis, showing higher values at the earlier (mean PE_0.5 10 cm H₂O/L) and later phases (mean PE_E 14 cm H₂O/L) of the procedure, compared with medium phase (mean PE_M 7.5 cm H₂O/L).

Complications

Chest X-rays performed after the procedure showed pneumothorax in 9 of the 57 patients. In three of them the air replaced the pleural space previously occupied by fluid. The mean lowest PlP in these three patients was 17.3 cm H₂O, compared with 10.8 cm H₂O in the remaining six. Although not suitable for statistical analysis, this fact might convey some physiopathological meaning. A chest tube was inserted in three patients, who actually did not have the fluid replaced by air.

Neither hypovolemia, reexpansion edema, or other significant complications occurred in any patient. Arterial blood pressure was measured immediately before and after the thoracentesis, and after 4, 8, 16, and 24 h in the first 13 patients. Afterward, it was measured every 8 h in the other patients. None of the patients developed symptoms of hypotension or a clinically significant decrease in blood pressure. Likewise, none of them developed new pulmonary shadows, suggesting reexpansion pulmonary edema in the chest X-ray following the procedure. In the first 22 patients, oxygen saturation was measured before and during the TT and during the next 8 h. No desaturation greater than 2% was observed, so the oxygen saturation was not checked in the remainder.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although the level of PlP potentially dangerous to human beings has not been accurately determined, there is clinical agreement regarding the association of a significant drop in PlP with outcomes such as pneumothorax, pulmonary edema, or hypovolemia (1). According to this, the measurement of PlP during TT can be helpful to prevent the side effects of removing large amounts of pleural fluid.

A simple method to monitor PlP during TT was described some years ago (6), but no additional information has been reported since. We systematically monitor PlP alongside the performance of TT, and use a device similar to that proposed by Light and coworkers (6) modified to simplify the procedure.

This research has found, as expected, that initial PlP was lower and PE higher in the group in which the reason for interrupting the thoracentesis was a decreased PlP. Nevertheless we have not been able to show either any within-subject relationship between the decrease in PlP and development of symptoms, difficulty in the removal of fluid, or any other premonitory sign of decrease in PlP.

We are aware that our requirements for stopping aspiration have not been grounded in strong evidence and could be somewhat conservative. More research is needed to define PlP and clinical thresholds that might allow aspirating more fluid without additional complications.

Etiology of the Effusions

In our study, the initial PlP level was not associated with any particular etiology. Nevertheless, every patient with both an initial PlP lower than 4 cm H₂O and a PE greater than 33 cm H₂O/L had a suspected trapped lung. These data are similar to those reported by Light and coworkers (6, 8), although they chose 5 cm H₂O or 10 cm H₂O for the initial PlP and 25 cm H₂O/L for the PE as suggestive of trapped lung. In a study of 65 patients with neoplastic effusions Lan and coworkers (9) defined trapped lung as a poor approximation of the pleurae, and reported that a PE_0.5 greater than 19 cm H₂O/L was more likely to be found in those with trapped lung. In our study, the analysis of PE_0.5 did not substantially modify the information obtained from initial PlP and PE values. We believe that the term "trapped lung" probably includes a gradation of stiffness of the lung and pleura. Higher grades would be found in those patients with no distension of the pleura; however, there are other patients with a "partially trapped lung" in whom, although with some difficulty, a partial or complete approximation of the pleurae after suction might be achieved. We think that this hypothesis may help to explain why the authors mentioned reported different thresholds to define trapped lung. If pleural septa are suspected, physicians should be cautious when aspirating fluid, since the PE might reflect the pressures of only a part of the pleural space.

Our data are discordant with those of Light and coworkers (6), who did not find any infectious effusion with a negative initial PlP. Actually, three of our patients with tuberculosis had a negative initial PlP (Figure 2).

Prediction of the Amount of Pleural Fluid That Can Be Withdrawn

We have found that neither initial PlP nor PE_0.5 were reliable predictors of the amount of pleural fluid to be withdrawn. A lobar bronchus obstruction did not preclude us from removing a large amount of pleural fluid in some patients. Nonetheless, if a main bronchus had been obstructed the results would have been different.

Pleural Pressure Curve

We believe that the PlP curve we obtained must be related to several factors such as visceral pleura stiffness, alveolar distensibility, bronchial obstruction (if any), and mechanical characteristics of the chest wall. Our study was conducted in a clinical, nonexperimental setting, and it is not possible to analyze the individual contributions of each factor.

The PlP during the thoracentesis decreases progressively with different slopes for each patient. In most cases, we can differentiate three parts of the curve. In the first one, there is a rapid decrease of the pressure, likely due to the resistance of lung and pleura to be distended from the position they have kept for some time. In the second, the curve tends to decrease steadily. In the third, which usually takes only a few milliliters of fluid, the PlP curve decreases very quickly. We believe that the duration of each part depends mainly on the stiffness of the pleuroparenchymatous structures. In some patients, likely because of a stiff pleura, the steady-decrease phase is absent or has been undetected by the sequence of measurements we have performed. Because a fast decrease in PlP is more likely to be found when larger amounts of fluid are withdrawn, we strongly recommend a measurement of PlP at least every 200 ml during the second liter and every 100 ml thereafter.

Complications

As in a previous study (6), the procedure has been useful in selecting patients for safer large volume thoracentesis. None of our patients developed clinically significant hypotension or pulmonary edema. The rate of pneumothoraces was similar to that previously reported (1, 10). We think that some of these could have been avoided if a sharp needle had not been used, and that it is safer to perform the TT with a needle catheter. In three patients air replaced the pleural fluid. In those patients the pneumothorax was more likely due to a conjoint effect of a noncompliant pleural surface (11, 12) and a negative PlP than to lung parenchyma needle damage. This air is likely to originate either from pulmonary venules or from subpleural blebs or bullae (12).

In conclusion, by measuring PlP during the TT, a large amount of pleural fluid can be aspirated with few and mild complications and the preliminary diagnosis of trapped lung can be reinforced. Initial PlP, PE_0.5, and lobar bronchial obstruction were not reliable predictors of the amount of fluid withdrawn. The downward PlP curve usually shows three parts, with the deepest slopes in the first and last phases of the thoracentesis, possibly reflecting the interaction of the thoracic structures.