INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pleural effusion (PE) is a common clinical problem that frequently causes dyspnea and abnormal arterial oxygenation (1). However, the mechanisms underlying this latter effect remain unsettled. Although lung collapse due to the accumulation of pleural fluid is a likely contributor, measurements of pulmonary mechanics in laboratory animals (2) and in patients (3) suggest that the lungs may actually "float" on the PE, an effect that would minimize the impact of atelectasis and subsequent shunting. That the size of the effusion has a small effect on pulmonary gas exchange (4) appears to support this latter hypothesis. To date, however, ventilation-perfusion (A/) distributions in patients with PE have not been adequately characterized. Only a case report (4) has used the multiple inert gas elimination technique (MIGET) (7, 8) to define them appropriately in a single patient.

Alternatively, the drainage of PE by thoracentesis is very effective in relieving dyspnea (1, 9), but its effect upon arterial oxygenation is variable. Pa _O2 can improve (10), remain unchanged (13), or deteriorate (14, 15) after drainage. These contrasting findings have been explained on the basis of the presence or absence of concomitant underlying lung parenchymal disease, nature and chronicity of the PE, and/or differences in the technique used to drain it, timing of drainage, or volume drained (10, 15). Nonetheless, its effect upon the A/ distributions have not been assessed as yet. In this investigation, we used the MIGET in patients with PE of recent clinical onset to characterize the A/ distributions and also to assess their response to the effects of PE drainage by thoracentesis. We hypothesized that pleural effusion would cause regions of intrapulmonary shunt or of low A/, or both, and that the drainage would induce a variable degree of gas exchange improvement, other things being equal.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

We studied nine consecutive patients (two females) with PE of recent clinical onset ( < 2 wk of symptoms). A positive diagnosis of tuberculosis was obtained in six patients, one patient had PE related to rheumatoid arthritis, and the remaining two patients had PE of unknown cause. None of them had cardiac failure or clinical evidence of any other disease, such as chronic obstructive pulmonary disease, asthma, lung fibrosis, and/or liver cirrhosis that could potentially confound pulmonary gas exchange. The main clinical and functional characteristics of the patients are shown in Table 1. There was a moderate restrictive ventilatory defect with a moderate low-transfer factor (DL_CO). This study was approved by the Ethics Committee of the Hospital Clinic. All patients gave written consent after being fully informed of the purpose, characteristics, and nature of the study.

Study Design

Patients were always studied at the same time of the day (between 9:00 A.M. and 12:00 P.M.). The day of the study, a forced spirometry (Datospir-2000; Siebel-Med, Barcelona, Spain) was obtained in all patients according to ATS recommendations (18). While the patient was sitting comfortably and under steady-state conditions (see below), a complete set of measurements was obtained before and 30 min after draining the PE, according to the following sequence; (1) heart rate and systemic arterial pressure; (2) respiratory frequency and tidal volume; (3) simultaneous arterial blood and mixed expired gas sampling for both inert and respiratory gas determinations; and (4) cardiac output. Thoracentesis was performed in the sitting position, after local anesthesia, using a 18-gauge needle with an internal diameter of 1.2 mm connected to a standard suction machine (A-70; Industrial Ordisi, Barcelona, Spain) through a three-way stopcock. This setup allowed us the direct measurement of pleural pressure (at the level of the fifth intercostal space) before and after withdrawing PE (HP 7754 B recorder; Hewlett Packard, Waltham, MA) (Table 1). Pleural fluid was evacuated at a constant rate of about 150 ml/min until no more fluid could be obtained or the patient showed discomfort (cough, pain, dyspnea). The amount of volume drained and the duration of drainage were recorded and, from these data, drainage flow was calculated (Table 1). Within three h of thoracentesis, a chest roentgenogram was obtained to rule out the presence of pneumothorax, gross pulmonary edema, or other potential abnormalities.

Measurements and Calculations

Blood samples were anaerobically collected through a polyethylene catheter (Seldicath; Plastimed, Saing-Lou-La-Foret, France) inserted into the radial artery of the nondominant hand, after local anesthesia. Pa_O2, Pa_CO2, and pH (IL 1302; Izasa, Barcelona, Spain) as well as hemoglobin concentration and oxyhemoglobin saturation (IL 482; Izasa) were analyzed in duplicate. The alveolar-arterial PO₂ difference (AaPO₂) was calculated according to the standard formula, using the measured respiratory exchange ratio (R). A low-dead space, low-resistance, nonrebreathing valve (No. 1500; Hans Rudolph, Kansas City, MO) connected to a heated metal mixing chamber was used to collect the mixed expired gas. Oxygen uptake (O₂) and CO₂ production (CO₂) were calculated from measured mixed expired O₂ and CO₂ by mass spectrometry (Multigas monitor MS2; BOC-Medishield, London, UK). Minute ventilation (E) and respiratory rate were measured using a calibrated Wright spirometer (Respirometer MK8; BOC-Medical, Essex, UK). A three-lead EKG, heart rate, and systemic arterial pressure were continuously recorded throughout the study (HP 7830 A monitor and HP 7754 B recorder; Hewlett-Packard). Cardiac output (T) was measured in duplicate or triplicate using a 5-mg bolus of indocyanine green injected into the peripheral venous line used for inert gas infusion (see below), as previously shown (19). Tracings of dye concentration versus time were derived from a densitometer used in conjunction with a DC-410 cuvette transducer (Waters Instruments Inc., Rochester, NY) at the peripheral artery site. We obtained the distribution of A/ ratios using the MIGET, as previously described in our laboratory (19). Briefly, a mixture of six inert gases with different solubilities (sulfurhexafluoride [SF6], ethane, cyclopropane, enflurane, ether, and acetone) dissolved in saline was given at 3 to 5 ml/min through a peripheral vein. Steady-state conditions were assured by monitoring end-tidal PCO₂ and PO₂, respiratory frequency, tidal volume, heart rate, and systemic arterial pressure kept within ± 5% of variations. Duplicates of arterial blood and mixed expired gas samples were obtained and the concentrations of inert gases were measured (HP 5880 A; Hewlett-Packard) (19). Mixed venous inert partial pressures of each inert gas were calculated from arterial and expired gases using the Fick principle (7, 8, 19). Results of A/ indices are the mean of each duplicate.

Statistical Analysis

Results are shown as mean ± SD. The effects of thoracentesis were analyzed using a two-tailed paired t test. Potential correlations between variables of interest were assessed using the Spearman's correlation coefficient (rho). A p value < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The hemodynamic, ventilatory, and gas exchange data in these patients, before and after thoracentesis, are shown in Table 2. Before thoracentesis, heart rate, mean systemic arterial pressure, T, E, respiratory rate, O₂, and CO₂ were within the normal range. Mean Pa_O2 (range: 69 to 104 mm Hg), Pa_CO2, and pH were also within normal limits (Table 2); the AaPO₂ was slightly increased (range: 9 to 46 mm Hg). A representative A/ distribution in one of the patients studied is illustrated in Figure 1. All subjects had broadened unimodal A/ distributions, with small amounts of blood flow perfusing units with low A/ ratios ( < 0.1, excluding shunt) (1.4 ± 2.2%; range: 0% to 6%). The former abnormality caused an increased dispersion of the perfusion distribution (Log_SD Q), whose mean value was 0.72 ± 0.29 (normal values < 0.6) (19) (range: 0.33 to 1.32). Besides, most of them had mild intrapulmonary shunt (A/ ratios < 0.005: 6.9 ± 6.7%; range: 0% to 23%). The ventilation distribution was normal. Predicted Pa_O2 (from the degree of both shunting and A/ mismatch) and measured Pa _O2 were not significantly different, indicating no impairment in the diffusion of O₂ from the alveolar air to the capillary blood (20). The degree of arterial hypoxemia before thoracentesis was significantly related to the intrapulmonary shunt value (rho = 0.82; p < 0.01) but did not correlate (rho = 0.40) with the amount of blood flow perfusing lung units with low A/ ratios ( < 0.1).

View larger version (14K): [in this window] [in a new window]   Figure 1.   Distribution of ventilation-perfusion (A/) ratios in a representative patient, before ( left panel ) and after (right panel ) thoracentesis. For further explanation, see text.

There were no clinical complications associated with thoracentesis. The mean volume of pleural fluid drained was 693 ± 424 ml. In seven patients, there was a noticeable change in chest X-ray after thoracentesis. However, all patients still showed some evidence of PE after thoracentesis. Pleural pressure fell significantly after thoracentesis (from 3.6 ± 7.2 to 7.1 ± 13.0 mm Hg; p < 0.01) (Table 1). Yet, most of the other physiologic variables remained unchanged (Table 2). In particular, neither Pa_O2 nor AaPO₂ varied after thoracentesis. As shown in Figure 2, Pa_O2 after thoracentesis increased in four patients, decreased in four others, and remained unchanged in one; in absolute terms, however, changes were of small magnitude. Similarly, the amount of intrapulmonary shunt did not vary significantly after thoracentesis (Table 2, Figure 2). In contrast, the perfusion of low A/ units increased slightly in the majority of patients (p < 0.05), although, in absolute terms, changes were again of small magnitude (Figure 1). Likewise, the dispersion of the blood flow distribution (Log_SD Q) mildly worsened after thoracentesis in most patients (Table 2, Figure 2). Changes in pulmonary gas exchange variables (Pa_O2, AaPO₂, shunt, and/or Log_SD Q) after thoracentesis did not correlate with the fall in pleural pressure (Figure 3), the volume drained, the timing of drainage, and/or flow of drainage. Also, predicted Pa_O2 and measured Pa_O2 were close after thoracentesis, again suggesting the lack of alveolar-capillary diffusion limitation or post-pulmonary shunt as factors explaining arterial hypoxemia. The remaining sum of squares (RSQ) in the present study was 2.4 ± 1.5 (range: 0.65 to 6.8).

View larger version (25K): [in this window] [in a new window]   Figure 2.   Individual and mean (solid bars) values of several pulmonary gas exchange variables before and 30 min after draining pleural effusion by thoracentesis. For further explanation, see text.

View larger version (11K): [in this window] [in a new window]   Figure 3.   Correlation between the change in pleural pressure seen after thoracentesis (bottom axes) and several indices of pulmonary gas exchange (n = 8). For further explanation, see text.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our study demonstrates that intrapulmonary shunt is the principal mechanism underlying defective arterial oxygenation in patients with PE of recent clinical onset and that thoracentesis caused very small short-term (within 30 min) changes in pulmonary gas exchange, despite the relatively large amounts of PE drained and the significant fall in pleural pressure induced. Only the amount of blood flow perfusing low A/ units, hence that of Log _SD Q, increased significantly after thoracentesis. However, this change was of small magnitude and did not influence arterial oxygenation. These observations, though, apply to the particular patients studied herein and may not be readily extrapolated to patients with chronic PE.

Several studies have investigated the effects of PE on pulmonary gas exchange (5, 11, 12, 15, 21), but only a single case report (4) unraveled the underlying mechanisms using the MIGET to characterize the distribution of A/ ratios. Our study extends this case report to a group of patients with PE of recent clinical onset, carefully selected to avoid potential confounding clinical factors. Our results show that A/ distributions in these patients are characterized by the presence of mild intrapulmonary shunt and the virtual absence of areas with low A/ units, likely related to different degrees of lung collapse caused by the accumulation of fluid in the pleural cavity. Further, we observed that the degree of arterial hypoxemia was significantly related to the value of shunt but not to the amount of blood flow perfusing low A/ units. These findings demonstrate that the main mechanism underlying abnormal oxygenation in patients with PE is the presence of intrapulmonary shunt.

Drainage of PE by thoracentesis is a common procedure in clinical practice to alleviate dyspnea (1). Yet, there is controversy concerning its effects upon arterial oxygenation because Pa _O2 can improve (10), remain unchanged (13), or deteriorate (14, 15) after it. The presence of other concurrent cardiopulmonary diseases, the etiology and timing of the PE, and/or methodologic differences have been quoted as potential mechanisms/factors to explain these conflicting findings (10, 15). We selected our patients to avoid some of these potential biases. Further, this is the first study that has used the MIGET to assess the effects of thoracentesis upon the distribution of A/ ratios. In our patients, despite the relatively large amounts of PE drained and the significant fall in pleural pressure induced by thoracentesis, neither Pa _O2, AaPO₂, or shunt changed significantly 30 min after thoracentesis. By contrast, the percentage of blood flow perfusing poorly ventilated lung units increased in most patients, although changes were small in absolute terms and not related to the volume of pleural fluid drained, the timing of drainage, and/or the drainage flow. Collectively, these observations admit, as least, three different explanations.

Firstly, re-expansion of the lung after thoracentesis may not be immediate (11, 12, 22). When fluid is aspirated from the thoracic cavity, the net gain of pulmonary volume is set by the balance between the compliance of the lung parenchyma and that of the thoracic cavity (11). Therefore, the amount of pleural fluid drained by thoracentesis does not necessarily reflect the net gain in lung volume after the procedure (9, 22). In fact, previous studies reported a lack of relationship between the volume of pleural fluid drained and changes in arterial oxygenation (11, 12). We cannot prove this hypothesis because, unfortunately, we did not have sequential A/ or lung volume measurements over a period of a few hours after thoracentesis. However, pleural pressure became quite negative after thoracentesis without a parallel change in Log_SD Q (Figure 3), hence indicating that the underlying lung was not expanding after drainage.

Secondly, previous studies of lung mechanics in patients with PE have suggested that the lung actually floats in the fluid of the pleural cavity, hence minimizing the impact of pulmonary collapse (2). If this was true, effective drainage of PE by thoracentesis should be expected to cause small changes in the distribution of A/ ratios despite reducing the pleural pressure significantly, as it actually occurred.

Thirdly, an alternative mechanism invoked to explain impaired arterial oxygenation after thoracentesis is the development of ex vacuo pulmonary edema (14, 15, 22). Although the increase in the percentage of blood flow perfusing low A/ units ( < 0.1) after thoracentesis is compatible with this phenomenon, the lack of change in intrapulmonary shunt after the procedure does not seemingly support this contention; further, the chest roentgenogram obtained after the study did not show the presence of alveolar infiltrates suggestive of pulmonary edema. However, our results cannot rule out completely this potential additional mechanism. Unfortunately, other types of information that may have potentially provided relevant data on this subject, such as a CT scan after thoracentesis, were not available.

In summary, this study shows that mild intrapulmonary shunt is the cardinal mechanism underlying arterial hypoxemia in patients with PE and that thoracentesis increases slightly the amount of blood flow perfusing low A/ units in the short term. Yet, the latter finding appears to be dissociated from the volume of pleural fluid drained, the timing of drainage, or the drainage flow. This is compatible with delayed pulmonary volume re-expansion after thoracentesis with or without the hypothesis of mild ex vacuo pulmonary edema developing after thoracentesis.