Pleural Effusions in Hospitalized Patients with AIDS

Methods

Patients

We reviewed the medical records of patients with human immunodeficiency virus (HIV) who were admitted to the Medical University of South Carolina Hospitals (Medical University Hospital, Veterans Affairs Medical Center, and Charleston Memorial Hospital) between January 1986 and January 1992. Of 350 HIV-positive patients identified, 222 who fulfilled the Center for Disease Control criteria for the diagnosis of AIDS were selected for the study. Their chest radiographic reports, clinical notes, and discharge summaries were searched for a diagnosis of pleural effusion. From this cohort, 64 patients were found to have pleural effusions from their chest radiographic reports or discharge summaries. Of 64 patients, chest radiographs were not available for 2 patients, and in 3 patients, the effusions were secondary to chest-tube placement for pneumothorax. Of 59 patients, all but 7 had lateral decubitus films. However, these seven patients all had chest radiographs that showed subsequent clearing of the effusions. Thus, all available chest radiographs, chest ultrasound reports, and computed tomographic scans of the thorax were reviewed for 59 patients.

The size and location of the pleural effusions were evaluated and, where available, the pleural fluid characteristics were recorded. The records were evaluated for primary diagnoses, diagnosis responsible for pleural effusion as well as serum albumin and CD4 counts. Additionally, the number of Pneumocystis carinii pneumonia (PCP) episodes per patient per year was obtained. Every case study was evaluated by an expert in pleural disease (SAS) who determined the cause of the pleural effusion.

The size of the pleural effusion on a chest radiograph was designated small if the costophrenic angle was obliterated; moderate if the lower zone was completely opaque; large if the lower and middle zones were opaque; and massive if all three zones were opaque.

A parapneumonic effusion was diagnosed based on a localized chest radiographic infiltrate, fever, and isolation of a bacterial organism from blood or sputum or a fourfold increase in mycoplasma titer or both (one patient); empyemas were diagnosed when pus was aspirated from the pleural space and the fluid was culture positive. Nocardiosis was diagnosed after recovery of the organism from the sputum or bronchoalveolar lavage fluid with resolution of the effusion with therapy. Pneumocystis carinii pneumonia was diagnosed if organisms were recovered in sputum or from bronchoalveolar lavage fluid with resolution of the pneumonia and effusion with specific therapy for PCP. Cryptococcal pneumonia was diagnosed after isolation of the organisms from blood, cerebrospinal fluid, or sputum. Tuberculous effusions were diagnosed after isolation of Mycobacterium tuberculosis from the pleural fluid or tissue.

Noninfectious causes of effusions were diagnosed as follows: Kaposi sarcoma based on pathologic results; atelectasis, when there were compatible chest radiographic results and spontaneous resolution of pleural effusions; congestive heart failure, after appropriate clinical presentation, chest radiographic results, and response to treatment; hypoalbuminemia, when patients had serum albumin concentrations 1.8 g/dL (18 g/L), no pulmonary parenchymal lesions on their chest radiographs, and no other cause for the effusion.

Statistics

Age, risk factors, serum albumin concentration, and CD4 counts in patients with and without pleural effusions were compared with the unpaired two-tailed Student t-test as well as the male to female ratio in patients with and without effusions with chi-square test using computerized software (Statworks, Cricket Software, Philadelphia, Pennsylvania).

Results

Pleural effusions occurred in 59 of 222 (27%) hospitalized AIDS patients. The mean age of the patients with effusions was 35 2 years (mean SE) and the age without pleural effusions was 36 4 years. The male-to-female ratio was 5:1 in patients with effusions and was 5:2 in patients without effusions. Risk factors for AIDS were intravenous drug abuse in 41% (24 of 59), homosexuality in 34% (20 of 59), and other risk factors in 25% (15 of 59). The mean age and risk factors of patients with pleural effusions were not significantly different from patients without pleural effusions (P > 0.2).

Pleural effusions (Table 1) were caused by infections in 39 of 59 patients (66%) and by noninfectious processes in 18 of 59 (31%); the cause was unknown in 2 of 59 patients (3%). The causative agent was Streptococcus pneumoniae in 8 of 16 (50%) patients with parapneumonic effusion, Legionella pneumophila in 1 patient, Mycoplasma pneumoniae in 1 patient, and unknown in 6 patients. One patient with Escherichia coli empyema required chest-tube drainage. The other patient with empyema due to Enterobacter species required drainage and decortication for resolution. Nine (15%) patients had pleural effusions caused by PCP. All were small; six were unilateral and three were bilateral. Analysis of the two PCP effusions that were sampled showed a straw-colored exudate with macrophages as the predominant cells. Special stains were not done on these samples to look for cysts or trophozoites of P. carinii.

Hypoalbuminemia was the leading noninfectious cause of pleural effusions in 11 of 59 patients (19%). It resulted from chronic liver disease in two patients, chronic renal failure with nephrosis in two, and progressive AIDS in seven. Serum analysis in 48 patients with nonhypoalbuminemic pleural effusions showed a lower albumin concentration (2.5 0.1 g/dL) (25 g/L) compared with 96 patients without pleural effusions (3.4 0.1 g/dL) (34 g/L) (P < 0.001). In 2 patients with serum albumin concentrations of 1.8 g/dL (18 g/L) and 1.1 g/dL (11 g/L), respectively, the effusions resolved with intravenous albumin therapy. The hypoalbuminemia patients were followed for a mean of 6 months, and no other cause could be established. There were 3 (5%) patients with cardiac failure; their effusions were bilateral and small, and resolved with appropriate treatment.

The pleural effusions were small in 46 patients, moderate in 4, large in 8, and massive in 1 patient with pleural tuberculosis (see Table 1). The effusions were associated with a unilateral bacterial pneumonia in 13 patients and were bilateral in 5 other patients. The CD4 counts were lower in the 59 patients with pleural effusions [72 12 cells/L] compared with 96 patients without effusions (274 26 cells/L) (P < 0.001). Pleural fluids were analyzed in 14 patients (Table 2). Effusions due to tuberculosis were all exudates with lymphocytes as the predominant cell. Three effusions resolved within 2 weeks after initiation of appropriate treatment, although one patient required surgical intervention for trapped lung. The patient with Kaposi sarcoma had bilateral progressive, large pleural effusions. Results from pleural fluid cytologic tests did not show abnormal cells. In two patients with hypoalbuminemic effusions who had thoracentesis, the fluids were transudates, and the cells were predominantly macrophages.

Discussion

Although pleural effusions are thought to be relatively uncommon [5], we found the incidence of pleural effusions to be 27% in all AIDS patients admitted to our hospitals. Infections accounted for the majority of pleural effusions; bacterial parapneumonic effusions accounted for 31% of the effusions compared with a 12% incidence of bacterial parapneumonic effusions in 300 patients from the pre-AIDS era [6]. This higher incidence of bacterial pneumonia in AIDS patients probably results from defective B-lymphocyte function [7-9]. The high incidence of parapneumonic pleural effusions appears to be related to the frequency with which S. pneumoniae causes effusions [10] rather than to a propensity for AIDS patients to develop parapneumonic effusions when infected with S. pneumoniae.

Tuberculosis accounted for 10% of the total number of patients with pleural effusions, which was similar to a previous report [11], and it was the most common cause for large-to-massive pleural effusions (50%). Pleural effusions in four patients were characteristically lymphocyte rich [12, 13], despite a low number of circulating blood lymphocytes.

Despite being the most common opportunistic pulmonary infection in patients with AIDS [14, 15], PCP remains an uncommon cause of pleural effusions. Suster and colleagues [16] in a retrospective analysis found one patient with pleural effusion among 59 patients with PCP. Stover and colleagues [5] observed small, transient pleural effusions in 4 of 35 (11%) patients with PCP. Two other patients had P. carinii isolated from the pleural space [17, 18]. Of 222 hospitalized patients, there were 145 episodes in patients without and 70 episodes in patients with pleural effusion. Pleural effusions were small and transient; they did not require thoracentesis. Although we could not prove a cause-and-effect relationship, all patients cleared their effusions after therapy for PCP. It is unlikely that the effusions result solely from pleural invasion by these organisms [15, 16]. Other factors such as increased negative intrapleural pressure from noncompliant lungs, impaired lymphatic clearance accompanying infection, increased permeability of the microcirculation resulting from products of inflammation, and low serum oncotic pressure from hypoalbuminemia may be responsible.

Pleural effusions accompanying AIDS-associated fungal infections are unusual. The most common are secondary to infection with Cryptococcus neoformans [19, 20], where an isolated pleural effusion may be the only presenting manifestation. Pleural effusions in AIDS patients have been associated with Nocardia, Histoplasma, and Toxoplasma infections [21, 22]. Typically they are small and respond to specific therapy.

The most frequent cause for a noninfectious pleural effusion was hypoalbuminemia. The albumin concentration below which extravascular sequestration of fluid occurs is estimated to be about 56 nmol/L (15 g/L); anasarca will invariably be present when the albumin concentration is < 37 nmol/L (10 g/L) [23]. Even when infectious causes of pleural effusion were evident, hypoalbuminemia may have been a contributing factor. Because hypoalbuminemia is a problem in progressive AIDS, hypoalbuminemic pleural effusions may become more prevalent as the infectious complications of AIDS are managed more effectively.

The other noninfectious causes of large pleural effusions in AIDS are Kaposi sarcoma and lymphoma [24, 25]. Our patient with Kaposi sarcoma had recurrent, bilateral, large, bloody pleural effusions and bilateral nodular infiltrates. As expected, pleural fluid was cytologically negative. Because progressive pleural effusion occurs in half of these patients [26], the presence of a pleural effusion as well as mediastinal and hilar lymphadenopathy should suggest the possibility of pleuropulmonary Kaposi sarcoma in a patient with AIDS [26]. Pleuropulmonary involvement in AIDS-related lymphoma is well recognized and hilar or mediastinal lymphadenopathy is often absent [27, 28]. Pleural effusion may be the sole presenting problem in these patients [26]. The diagnosis usually requires invasive procedures and flow cytometric examination of the pleural fluid may establish the diagnosis [29]. In summary, lymphoma, Kaposi sarcoma, and tuberculosis are the most probable causes of large pleural effusions in patients with AIDS. Generally, pleural effusions in advanced AIDS are small, do not result in symptoms, and do not require pleural-space drainage.