INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Plasma levels of human immunodeficincy virus-1 (HIV-1) RNA are a surrogate marker for active viral replication, and represent a sensitive clinical tool for monitoring progression of HIV-1 disease and for gauging its response to highly active antiretroviral therapy (HAART) (1, 2). Plasma HIV-1 RNA can be detected at all clinical stages of HIV-1-related disease (3), and plasma RNA levels measured early in HIV-1 infection predict the temporal course of HIV-1 disease progression (4). HIV-associated opportunistic infections, including influenza (5), infection by Mycobacterium tuberculosis (6), Mycobacterium avium complex (MAC) (7), cytomegalovirus (7), Cryptococcus neoformans (7), and bacteria (7), as well as Pneumocystis carinii pneumonia (7, 8), may increase plasma HIV-1 RNA levels, and may thus contribute to the pathogenesis of HIV-1 disease and the associated immune dysfunction. However, the effect of these opportunistic infections on viral replication in other physiologic compartments is not well characterized.

Recent observations suggest that pulmonary infections in HIV-seropositive patients may enhance HIV-1 replication in the lungs. HIV-1 RNA levels in the bronchoalveolar lavage fluid (BALF) are increased in the setting of infection with with M. tuberculosis (9) or P. carinii (8). In the case of M. tuberculosis, BALF HIV-1 RNA levels exceed those in plasma, suggesting that the increase in HIV-1 RNA represents local production in the lungs (9). Such pulmonary infections may accelerate HIV-1 replication and contribute to the progression of HIV-1-related disease in the lungs, in analogy to the recently observed enhancement of HIV-1 replication in the lymph nodes of HIV-1-infected persons with P. carinii and other opportunistic infections (10).

Current understanding of the HIV-1 viral burden in the lungs, and its relationship to plasma viral load, remains incomplete. Whether levels of HIV-1 RNA in BALF parallel those in the serum of asymptomatic HIV-1-infected persons is unclear. Furthermore, whether only certain pulmonary opportunistic infections (such as with M. tuberculosis [9]) activate HIV-1 expression in the lungs, or whether the phenomenon is common to other causes of pulmonary disease, remains to be determined. The purpose of this study was to further characterize the HIV-1 viral burden in the lungs. Through the use of target amplification methods of molecular biology, and comparison of asymptomatic HIV-infected persons with HIV- infected persons with active pulmonary diseases, we sought: (1) to examine the relationship of viral burden as reflected by HIV-1 RNA in the cell-free BALF compartment to that in the serum; (2) to examine the relationship of HIV-1 RNA levels in cell-free BALF to latent HIV-1 proviral DNA in BALF cells; and (3) to examine the relationship of HIV-1 RNA expression in cell-free BALF to levels of selected cytokines in the lungs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Subjects

Asymptomatic HIV-seropositive individuals were prospectively recruited for research bronchoscopy. Clinical specimens were also available from HIV-seropositive patients undergoing diagnostic bronchoscopy for respiratory symptoms and radiographic infiltrates. For clinical specimens, only that portion of the serum or BALF that remained after clinical laboratory use was available for analysis. Archived paired serum and cell-free BALF supernatants (maintained at 80° C) from asymptomatic HIV-seropositive persons, and cell-free BALF supernatants from HIV-seropositive persons with active P. carinii pneumonia, were also available for this study. Demographic characteristics for all subjects were recorded on standardized forms. All protocols were approved by the institutional review boards of the Boston Medical Center and the Beth Israel Deaconess Medical Center.

Bronchoscopy

Pulmonary immune cells were obtained by bronchoalveolar lavage (BAL) performed on consenting adults, using standard technique (11). Briefly, following induction of anesthesia with topical 2% lidocaine applied to the oropharynx, a fiberoptic bronchoscope was passed into the airways and wedged in a subsegment of the right middle lobe. BAL was performed by instilling from four to eight 50-ml aliquots of warm, nonbacteriostatic normal saline (0.9%), followed by gentle suction after each aliquot was infused. The pooled BALF was transported to the laboratory on wet ice.

Isolation of BAL Cells and Lavage Supernatants

Cells in BALF were separated from the pooled BALF by centrifugation at 100 × g for 10 min at 4° C, were washed in cold RPMI-1640 supplemented with penicillin 100 U/ml and streptomycin 100 µg/ml (Sigma, St. Louis, MO), and were counted with a hemacytometer. Slides for cell differential counts were prepared by cytocentrifugation (Shandon, Pittsburgh, PA), stained with the modified Giemsa method (Diff-Quik; Sigma) and examined through light microscopy. The cell-free BALF supernatants were aliquoted and maintained at 80° C until assayed. None of the BALF specimens required lysis to eliminate contaminating red blood cells during processing.

Serum Samples

Peripheral blood specimens were obtained at the time of bronchoscopy by venipuncture with a sterile technique, and were collected into heparin-free containers. The serum was isolated after coagulation at room temperature and centrifugation, and 1.0-ml aliquots were maintained at 80° C. All patients were confirmed HIV-1 seropositive with an enzyme-linked immunosorbent assay (ELISA) (Abbott Diagnostics, North Chicago, IL) performed according to the manufacturer's instructions.

HIV-1 RNA Determination in Serum and Cell-Free BALF

Detection and quantitation of HIV-1 RNA was done with an in vitro nucleic acid amplification test (Amplicor HIV-1 Monitor Test; Roche Diagnostic Systems, Inc., Branchburg, NJ) according to the manufacturer's instructions. The lower level of detectability was < 200 HIV-1 RNA copies/ml biologic fluid. Testing was done on samples of serum and on diluted, cell-free BALF supernatants. To allow direct comparison with serum HIV-1 RNA levels, viral RNA measurements in the BALF supernatants (diluted by the BAL procedure) were normalized through the urea method (12) and the data were expressed as HIV-1 RNA copies/ml alveolar lining fluid.

Cellular HIV-1 Proviral DNA Determination in Mixed BAL Cells and Isolated Alveolar Macrophages

HIV-1 proviral DNA was detected separately through a DNA polymerase chain reaction (PCR) applied to both unseparated BAL cells and isolated alveolar macrophages (AM). Three primer pairs were utilized: SK38/39, located in the HIV-1 gag gene region; SK68/69, located in the HIV-1 env gene region; and SK 29/30, located in the HIV-1 long terminal repeat region (13). For unseparated BAL cells, stepwise 10-fold serial dilutions of target cells were performed in triplicate. AM were isolated by adherence for 3 d and removed by scraping in cold phosphate-buffered saline (PBS). Stepwise 10-fold serial dilutions of the isolated cells were then performed in duplicate. Isolated AM were > 99% nonspecific-esterase-positive. The dilutions containing 10 cells and 1 cell were verified directly by counting in microwells.

Templates were prepared by cell digestion with proteinase K for 1 h at 37° C. PCR was done with the three previously named specific primer pairs for 30 cycles of amplification, and PCR products were analyzed by agarose gel electrophoresis. DNA-free samples and uninfected cells were included as controls. Data were expressed as the percentage of cells containing HIV-1 proviral DNA. The proportion of infected cells was estimated by the lowest dilution of cells yielding a specific PCR product. For example, if template from 100 donor cells was positive but template from 10 donor cells was negative, the result was expressed as "1% positive." If a positive result was obtained by amplification of template from a single donor cell, then the result was expressed as "> 10% positive cells." A positive result with any primer at the lowest dilution of cells was considered to be the most accurate reflection of the proportion of HIV-1-infected cells. Specificity of the PCR product was confirmed in selected samples by Southern blotting with HIV-1 oligonucleotide probes.

Cytokine Determinations in BALF Supernatants

Measurements of tumor necrosis factor (TNF)-, granulocyte-macrophage colony-stimulating factor (GM-CSF), and interleukin (IL)-16 were made through ELISA on duplicate 100-µl aliquots of cell-free BALF (maintained at 80° C for < 6 mo) according to the manufacturer's instructions (R&D Systems, Minneapolis, MN). All BALF measurements were normalized through the urea method (12), and the data were expressed as picograms of cytokine per milliliter of alveolar lining fluid. The level of detectability for each ELISA assay was 5 pg/ml for TNF-, 2 pg/ml for GM-CSF, and 16 pg/ml for IL-16. Select samples of dilute BALF with undetectable cytokine levels were concentrated by 10- to 15-fold, using Centriprep centrifugation filters (Amicon, Inc., Beverly, MA), and subjected to second ELISAs for the three cytokines.

Statistical Analysis

HIV-1 RNA and cytokine levels were measured in duplicate or triplicate as indicated, and comparison of measurements was done with the Mann-Whitney U test. Correlation of HIV-1 RNA in BALF with cellular HIV-1 proviral DNA and with BALF cytokine levels for the different study subject groups was evaluated with Fisher's exact test. Statistical significance was accepted for values of p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Subject Characteristics

Specimens were obtained from a total of 77 HIV-seropositive individuals. These included serum, cell-free BALF supernatant, and BAL cell specimens obtained prospectively from 46 HIV-seropositive persons, including 24 asymptomatic individuals, and 22 clinical specimens from persons with active pulmonary disease. The 77 total samples also included 19 paired archived serum and cell-free BALF supernatant specimens from asymptomatic HIV-seropositive persons, and 12 archived cell-free BALF supernatant specimens from HIV-seropositive persons with confirmed P. carinii pneumonia (maintained in both cases at 80° C).

The 77 HIV-seropositive individuals studied included 59 males and 18 females. The ethnic background included 55 white, 17 black, and five Hispanic persons, aged 37.7 ± 7.2 yr (mean ± SD) (range: 26 to 75 yr). Tobacco use was recorded for 47 of the 77 subjects (61%). The range of peripheral CD4⁺ T-lymphocyte counts was from 0 to 700 cells/mm³. HIV-1 risk factors included men having sex with men (MSM), in 58 cases, injection drug use (IDU) in 15 cases, and heterosexual contacts in four cases. Antiretroviral agents were prescribed for a total of 37 cases of the 77 subjects (48%), with HAART prescribed for 25 of these 37 subjects (68%).

The asymptomatic study subjects were further identified according to early (peripheral blood CD4⁺ T-lymphocyte count > 200 cells/mm³) or advanced (peripheral blood CD4⁺ T-lymphocyte count < 200 cells/mm³) stages of HIV-related disease. The characteristics for these groups are presented in Table 1. The peripheral blood CD4⁺ T-lymphocyte counts were significantly lower in asymptomatic subjects with advanced disease and in subjects with active pulmonary disease than in asymptomatic subjects with early stages of HIV-related disease (p < 0.001). However, there was no difference in the peripheral blood CD4⁺ T-lymphocyte counts of asymptomatic subjects with advanced disease and subjects with active pulmonary disease (p > 0.05).

Pulmonary Diagnosis for Clinical Bronchoscopy Specimens

The pulmonary diseases identified for subjects from whom clinical bronchoscopy specimens were taken are included in Table 1. Of the study subjects with active pulmonary disease, 76% had P. carinii pneumonia. The diagnosis of P. carinii pneumonia was established through specific immunofluorescence staining with a monoclonal antibody in BALF specimens. For other pulmonary diseases, the diagnosis was determined by the identification of a pathogenic organism through culture or specific fungal, acid-fast, or immunofluorescence staining of BALF specimens. There were no presumptive diagnoses of pneumonia. The diagnosis of endobronchial Kaposi's sarcoma was made from the bronchoscopic appearance of characteristic endobronchial lesions (in a patient with cutaneous and gastrointestinal Kaposi's sarcoma) and the absence of an isolated pathogen.

BALF and BAL Cell Characteristics

For the asymptomatic subjects, the BALF return was 67 ± 7% (mean ± SD) of the instilled saline. The mean total BAL cell count was 37.5 × 10⁶ cells, and cell differential counts revealed 82 ± 12% AM, 17.1 ± 15% alveolar lymphocytes, and 0.8 ± 0.1% neutrophils. The BALF return and cell differential characteristics were not available for either the clinical specimens or the archived samples. Determinations of the dilution factors for the cell-free BALF supernatants (for subjects with corresponding serum samples), done with the urea method, revealed mean values of 71.8 ± 17 for the archived BALF samples from asymptomatic subjects, 99.9 ± 46 for the BALF specimens from prospectively enrolled asymptomatic subjects, and 76.7 ± 66 for the BALF specimens from prospectively enrolled subjects with active pulmonary disease. These mean values were used to determine the lower level of detectability of HIV-1 RNA in BALF for the respective groups. For reporting HIV-1 RNA levels in BALF, however, a specific dilution factor was determined and used for each individual specimen.

HIV-1 RNA Measurements in Serum of Asymptomatic Subjects

HIV-1 RNA was detected in the serum of 35 of 41 (85%) members of the group of asymptomatic HIV-seropositive individuals, for whom specimens were available, including all 19 (100%) archived serum specimens. For individuals with detectable HIV-1 RNA levels, the mean value was 6.02 × 10⁴ copies of HIV-1 RNA/ml serum (range: 530 to 263,074 HIV-1 RNA copies/ml serum). Nineteen individuals had more than 10,000 RNA copies/ml serum, and eight had more than 100,000 HIV-1 RNA copies/ml serum. Of the 26 subjects for whom antiretroviral agents were perscribed, 21 (81%) had detectable HIV-1 RNA in their serum.

HIV-1 RNA Measurements of Cell-Free BALF of Asymptomatic Subjects

HIV-1 RNA was detected in the cell-free BALF of seven of the 43 (16%) asymptomatic subjects. The mean value was 1.04 × 10⁵ HIV-1 RNA copies/ml alveolar lining fluid (range: 20,979 to 328,371 HIV-1 RNA copies/ml alveolar lining fluid). HIV-1 RNA could be detected in the cell-free BALF of five of 27 (18.5%) asymptomatic individuals with early clinical stages of HIV-related disease (Figure 1A), as compared with two of 16 (12.5%) asymptomatic individuals with advanced stages of HIV-related disease (Figure 1B) (p = 0.62).

View larger version (26K): [in this window] [in a new window]   View larger version (24K): [in this window] [in a new window]   View larger version (26K): [in this window] [in a new window]   Figure 1.   Viral load comparisons in paired samples of serum and BALF from (A) asymptomatic HIV-seropositive subjects with peripheral blood CD4+ T-lymphocyte counts > 200 cells/mm3 (n = 27), (B) asymptomatic HIV-seropositive subjects with peripheral blood CD4+ T-lymphocyte counts < 200 cells/mm3 (n = 16), and (C ) HIV-seropositive subjects with active P. carinii pneumonia or other lung diseases (n = 34). Data for the the cell-free BALF are corrected for the dilution factor for each sample, and are expressed as copies of HIV-1 RNA per milliliter of alveolar lining fluid. Data points connected by a line represent paired values for serum and BALF from the same individual. Data points included within the dashed areas represent values below the level of detectability for the assay (< 200 copies HIV-1 RNA/ml of serum; < 19,980 copies HIV-1 RNA/ml alveolar lining fluid for asymptomatic individuals; < 15,340 copies HIV-1 RNA/ml alveolar lining fluid for persons with active pulmonary disease). Horizontal bars for each column represent mean values. Note that the ordinate values are log scale. Data points contained within the dashed areas are centered and vertically displaced to more acurately represent the number of measurements with values below the level of detectability for the assay. Each data point included within these dashed areas represents a measurement below the level of detectability rather than a specific value at the level indicated (*p = not significant compared to serum from asymptomatic individuals; **p = 0.045 compared with serum from persons with active lung disease; ++p = 0.032 compared with BALF from asymptomatic individuals).

All seven asymptomatic subjects with HIV-1 RNA in their cell-free BALF had detectable HIV-1 RNA in their serum, and in two cases (28%) the quantity of HIV-1 RNA in the BALF exceeded that in serum by 3- to 52-fold (although the differences in mean values did not achieve statistical significance [p = 0.18]). BAL cell differential counts for these seven subjects were similar to those for asymptomatic persons with no detectable HIV-1 RNA in their BALF. Six of the seven subjects were not taking any antiretroviral medications, and a combination of reverse transcriptase inhibitors only (without protease inhibitors) was perscribed for one. Repeat determinations were performed on subsequent occasions for five of the subjects with HIV-1 RNA in both cell-free BALF and serum, and demonstrated detectable HIV-1 RNA in the BALF in all five cases. None of these subjects had received recent influenza vaccinations, none had experienced pulmonary symptoms during the 1 mo before the HIV-1 RNA determinations, and none developed pulmonary symptoms or pneumonia during the 2 mo after these determinations. All seven asymptomatic individuals with HIV-1 RNA in their cell-free BALF and in their serum had intravenous drug abuse as their risk factor for HIV infection, and all seven actively used tobacco products.

Excluding any effect of BALF constituents on the detection of HIV-1 RNA with the Amplicor HIV-1 RNA detection kit used in the study was the finding that BALF supernatants from healthy HIV-seronegative persons were negative for HIV-1 RNA and did not produce any interference signal for the assay (n = 6; data not shown). Furthermore, serial 10-fold dilutions of a standard HIV-1 RNA concentration with BALF supernatant from healthy HIV-1-seronegative persons provided HIV-1 RNA measurements comparable to those achieved with dilution with PBS (n = 3; data not shown). HIV-1 RNA measurements in BALF were stable over a 6-mo period (n = 8; data not shown).

HIV-1 RNA Measurements in Serum from Subjects with Active Lung Disease

Among the 34 HIV-seropositive subjects with active pulmonary disease, HIV-1 RNA was detected in the serum of 12 of 14 (86%) for whom measurements were available (Figure 1C). The mean value for these 12 persons was 1.81 × 10⁵ HIV-1 RNA copies/ml serum (range: 433 to 910,000 HIV-1 RNA copies/ml serum; p = 0.12 as compared with serum of asymptomatic subjects). Of these 12 individuals, 10 had more than 10,000 HIV-1 RNA copies/ml serum, and four had more than 100,000 HIV-1 RNA copies/ml serum. Of the 11 subjects for whom antiretroviral agents were prescribed, 60% had detectable HIV-1 RNA in their serum. Serum samples were not available among the archived specimens of subjects with P. carinii pneumonia.

HIV-1 RNA Measurements in Cell-Free BALF from Subjects with Active Lung Disease

BALF specimens were available for 30 of 34 subjects with active pulmonary disease (Figure 1C). HIV-1 RNA was detected in 67% of these subjects' specimens (p = 0.001 as compared with asymptomatic subjects), including those of 15 of 23 (65%) subjects with P. carinii pneumonia, one of three (33%) subjects with bacterial pneumonia (Streptococcus pneumoniae), one of two subjects with MAC pneumonia, and in each case of Aspergillus sp. pneumonia, Nocardia sp. pneumonia, and pulmonary Kaposi's sarcoma. The mean HIV-1 RNA value for these 20 individuals was 3.31 × 10⁶ HIV-1 RNA copies/ml alveolar lining fluid (range: 28,686 to 59,859,518 HIV-1 RNA copies/ml alveolar lining fluid; p = 0.032 compared with BALF of asymptomatic subjects). HIV-1 RNA was detected in the cell-free BALF of 69% of patients with serum viremia, and the mean values were higher in these individuals' BALF than in their serum (p = 0.045). BALF HIV-1 RNA levels exceeded serum levels in 78% of these patients by 1.4- to 66-fold. Two patients with P. carinii pneumonia had increased BALF HIV-1 RNA levels in the absence of detectable serum levels.

HIV-1 Proviral DNA Detection in Mixed BAL Cells and Isolated AM

For the purpose of correlating HIV-1 RNA in BALF with that in latently infected BAL cells, we performed HIV-1 proviral DNA determinations with semiquantitative PCR on unseparated BAL cells and isolated macrophages from 27 subjects (Figure 2). To validate this method, parallel experiments were conducted with the ACH-2 cell line (each cell containing a single copy of HIV-1 DNA). Quantitative PCR analysis showed that a positive signal could be detected to the level of a single ACH-2 cell (data not shown).

View larger version (39K): [in this window] [in a new window]   Figure 2.   Representative agarose-gel electrophoresis of PCR products of HIV-1 DNA from mixed bronchoalveolar cells and isolated AM from an HIV-seropositive subject with active P. carinii pneumonia, using the HIV-1 SK38-39 (gag) amplimer. Lane 1: Hae/HindIII ladder, and serial dilutions of mixed BAL cells in lanes; lane 2: 105 cells; lane 3: 104 cells; lane 4: 103 cells; lane 5: 102 cells; lane 6: 101 cells; and lane 7: 100 cells; (single cell). Also shown are serial dilutions of isolated AM; lane 8: 105 cells; lane 9: 104 cells; lane 10: 103 cells; lane 11: 102 cells; lane 12: 101 cells; and lane 13: 100 cells; (single cell). Controls are in lane 14: HIV + plasmid; and lane 15: amplimers only. For this particular example, the lowest dilution that was considered positive was 101 cells (10 cells) for the mixed BAL cells lane 6, and 103 cells (1,000 cells) for the AM (lane 10).

HIV-1 proviral DNA was detected in 62% of samples from the 13 asymptomatic subjects. Depending upon the specific HIV-1 gene primer pair employed, HIV-1 proviral DNA was detectable in from 0.01 to > 10% of unseparated BAL cells and from 0.01 to > 10% of isolated AM. Differences in the level of HIV-1 DNA detection in cells from a single individual with the three primer pairs used in the PCR probably reflected the influence of donor HIV-1 sequence variation on the efficiency of the PCR reaction. HIV-1 proviral DNA could be detected in three of seven (43%) asymptomatic individuals with peripheral CD4⁺ T-lymphocyte counts > 200 cells/mm³ (Table 2). Two of these seven individuals had proviral DNA in  10% of mixed BAL cells, whereas one of the seven had proviral DNA in up to 10% of isolated AM. In comparison, HIV-1 proviral DNA could be detected in five of six (83%) asymptomatic individuals with peripheral CD4⁺ T-lymphocyte counts < 200 cells/mm³. Four of these six individuals had proviral DNA in  10% of mixed BAL cells, whereas three of the six had proviral DNA in  10% of isolated AM.

Among 14 subjects with active pulmonary disease, HIV-1 proviral DNA was detected in 86% (p = 0.16 compared with asymptomatic subjects) in serial dilutions of mixed BAL cells or isolated macrophages (Table 3). Depending upon the specific HIV-1 gene primer pair used, HIV-1 proviral DNA was detectable in from 0.01 to > 10% of unseparated BAL cells and from 0.001 to > 10% of isolated AM.

Comparing subjects with early stages of HIV-related disease (asymptomatic subjects with peripheral CD4⁺ T-lymphocyte counts > 200 cells/mm³) and those with advanced stages of HIV-related disease (asymptomatic subjects with peripheral CD4⁺ T-lymphocyte counts < 200 cells/mm³ or with active pulmonary disease) revealed significantly more HIV-1 proviral DNA in the BAL cells of the group with advanced disease (43% versus 85%; p = 0.0496). Among the 20 HIV- seropositive subjects with detectable HIV-1 proviral DNA in their BAL cells, HIV-1 RNA was detected in the cell-free BALF of 25% of asymptomatic subjects and 50% of subjects with pneumonia (p = 0.26).

Measurement of Cytokines in Cell-Free BALF

Measurements for TNF-, GM-CSF, and IL-16 were made on cell-free BALF from 32 subjects, including 14 specimens from asymptomatic subjects and 18 speciments from subjects with active lung disease. These cytokines were selected because of the reportedly increased levels of TNF- and GM-CSF HIV-infected individuals (14, 15) and the augmenting effects of TNF- and GM-CSF on HIV-1 replication (16), and the IL-16-mediated inhibition of HIV-1 replication (17, 18). There was no correlation between detection of HIV-1 RNA in BALF and detection of TNF-, GM-CSF, or IL-16 in BALF or BALF concentrated by 10- to 15-fold (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The findings in this study provide evidence for in vivo HIV-1 replication (as measured by HIV-1 RNA levels) in the lungs of persons at various clinical stages of HIV-related disease. Sixteen percent of asymptomatic subjects had evidence of HIV-1 RNA in their cell-free BALF, independent of whether they were in early or advanced clinical stages of HIV-related disease. In comparison, 62% of subjects with active lung disease, such as P. carinii pneumonia, had HIV-1 RNA in their cell-free BALF, and mean values for this were significantly higher than those for BALF of asymptomatic individuals.

The enhanced HIV-1 replication in the lungs was compartmentalized relative to HIV in the serum. Whereas the proportion of individuals with detectable serum HIV-1 RNA and the mean serum values of HIV-1 RNA were similar for asymptomatic subjects and those with active lung disease (56% in both groups), subjects with active lung disease had higher BALF HIV-1 RNA levels (adjusted to allow direct comparison with those in serum) than serum HIV-1 RNA levels, with values up to 66-fold higher than their serum values. The higher HIV-1 RNA levels in these subjects' BALF than in their serum suggests enhanced local HIV-1 production rather than exudation of virus from serum.

The finding of HIV-1 RNA in the BALF of 16% of the asymptomatic HIV-seropositive subjects in our study suggests that the lungs may represent a source of persistent viral replication even during a period of apparent clinical latency of HIV-1 infection. Similar reports of HIV-1 RNA in the BALF of 10% of asymptomatic HIV-1-infected persons (9) and of 60% of acquired immune deficiency syndrome (AIDS) patients without evidence of active pulmonary disease (8) suggest that the dynamics of viral replication in the lungs of asymptomatic persons may be similar to those of active HIV-1 replication in lymph nodes of such persons (19, 20).

Among our subjects with active P. carinii pneumonia, the frequent identification of HIV-1 RNA in BALF, and the observation that BALF HIV-1 RNA levels were higher than those of asymptomatic individuals, are consistent with findings in a prior study (8). Furthermore, our observations that HIV-1 RNA was detected more frequently in the BALF of subjects with active P. carinii pneumonia (96% of whom had peripheral blood CD4⁺ T-lymphocyte counts < 200 cells/mm³), and exceeded the levels in BALF from asymptomatic subjects with advanced HIV-related disease (persons with peripheral blood CD4⁺ T-lymphocyte counts < 200 cells/mm³), suggests that pulmonary infection rather than an advanced clinical stage of HIV-related disease is associated with active HIV-1 replication.

Our findings also suggest that in vivo replication of HIV-1 in the lungs may not be specific to P. carinii pneumonia. Increased levels of HIV-1 RNA were measured in the BALF of persons with pneumonia caused by other organisms, including S. pneumoniae, Aspergillus sp., Nocardia sp., and MAC, and by those with pulmonary Kaposi's sarcoma. However, definitive conclusions cannot be drawn about this, since only limited numbers of cases of these latter diseases were available for study. Recent findings show that M. tuberculosis enhances HIV-1 RNA replication in the lungs (9).

The principal source for the HIV-1 RNA in BALF was not established in the present study. The frequent isolation of HIV-1 proviral DNA from BAL cells in this and other studies (21, 22) implicates alveolar lymphocytes and/or macrophages as the source of this RNA. Alveolar lymphocytes from persons with active P. carinii pneumonia release HIV-1 p24 antigen, but require in vitro stimulation to do so (23). A recent study demonstrated that macrophages are the source of HIV-1 replication in the peripheral lymph nodes of persons with active P. carinii pneumonia (10). Although HIV-1 proviral DNA was detected in > 10% of AM of some subjects in the present study, it remains to be determined whether macrophages serve a similar role in the lungs. The findings of an increased BALF HIV-1 RNA level without evidence for latent infection in BAL cells in one of our subjects (Subject A3, Table 2) may reflect the sensitivity of the specific primers used in our study, but also suggests that the source of HIV-1 RNA in BALF may be other pulmonary cells, such as alveolar epithelial cells or pulmonary dendritic cells.

The mechanism for enhanced HIV-1 replication in the lungs was not established in the present study. The finding that HIV-1 RNA in the lungs often exceeded serum levels of HIV-1 RNA suggests an important role for local or lung-specific factors in pulmonary HIV-1 replication. In persons with active lung disease, extracellular or intracellular pathogens may directly or indirectly enhance viral replication in HIV-infected cells (10). In asymptomatic individuals, viral replication may be influenced by HAART, tobacco use (24), or subclinical respiratory tract infections. The absence of a correlation of HIV-1 RNA levels with those of TNF-, GM-CSF, or IL-16 does not exclude possible effects of other cytokines. Alternatively, viral replication may be regulated by local expression of chemokines and chemokine receptors (25), or by coinfection with other pulmonary pathogens (such as cytomegalovirus), although these were not specifically investigated.

The finding of HIV-1 RNA in the cell-free BALF of 16% of asymptomatic subjects in the present study, but of HIV-1 proviral DNA in 62% of BAL cells from these same subjects suggests that replication of HIV-1 in the lungs is restricted. However, this finding may reflect limitations of HIV-1 RNA detectability, since dilute BALF reduced the sensitivity of HIV-1 RNA detection by 76- to 100-fold. Efforts to improve the sensitivity of HIV-1 RNA detection in BALF are currently in progress. Furthermore, in view of the limitations of the urea method, the reported values for HIV-1 RNA may be underestimated (26).

In conclusion, the findings in the present study show that replication of HIV-1 occurs in vivo in the lungs of asymptomatic HIV-1-infected persons, and is enhanced in persons with active pulmonary disease such as P. carinii pneumonia. HIV-1 replication is compartmentalized in the lungs and may be restricted, and is probably influenced by local factors even during the clinically latent phase of HIV-1 infection. Furthermore, enhanced HIV-1 replication in the lungs may not be specific for P. carinii pneumonia, but may also occur with other pulmonary diseases. Because the lungs represent a large reservoir for HIV-1, defining the mechanisms that control local viral replication will have important implications for understanding the pathogenesis of HIV-1 infection and the susceptibility of infected individuals to opportunistic lung infections. The compartmentalization of HIV-1 replication in the lungs also has important implications for monitoring antiretroviral therapy.