INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary complications are common in patients infected with human immunodeficiency virus (HIV) (1, 2). A frequently found noninfectious complication of HIV infection is of lymphocytic alveolitis, seen in up to 60% of individuals without pulmonary symptoms, and in even a greater percentage of individuals with such symptoms (3). Importantly, the presence of lymphocytic alveolitis in HIV-infected subjects has been shown to be associated with a worse prognosis (4).

Throughout most of the course of HIV infection, the increased lung lymphocyte population in patients with lymphocytic alveolitis consists of CD8⁺ cytotoxic T lymphocytes (CTL) (3, 5). These CTL have been shown to have specific activity against HIV-infected cells, especially alveolar macrophages (6). Based on this finding and previous work in our laboratory showing that the cells needed to generate a specific immune response exist in the lungs of HIV-infected individuals (7), we have hypothesized that lymphocytic alveolitis in this population represents an appropriate immune response to HIV-infected cells in the lung. If this hypothesis is true, then one would speculate that individuals with a greater HIV burden in the lungs would be more likely to have a lymphocytic alveolitis than those patients without a significant viral burden. As a corollary to this hypothesis, the poorer prognosis for HIV-infected subjects with lymphocytic alveolitis could merely reflect higher viral loads at the tissue level.

With the advent of simpler methods for detecting HIV RNA, the ability to measure tissue viral load levels has improved. These methods have afforded the ability to quantify the presence of HIV in stored bronchoalveolar lavage fluid (BALF) from participants in prior studies. In the present study we analyzed HIV viral loads as reflected in BALF and correlated them with the presence of lymphocytic alveolitis and clinical outcomes in a well defined HIV-infected population that initially underwent bronchoscopy from 1990 through late 1993.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Twenty-one HIV-infected subjects who underwent bronchoscopy with bronchoalveolar lavage (BAL) for the first time between April 9, 1990 and November 22, 1993 constituted the patient population for the study. Ten subjects had bronchoscopy performed on more than one occasion separated by a minimum of 3 mo (mean interval: 11 ± 9.7 mo; range: 3 to 32 mo). Viral load studies on BALF were performed on 17 subjects who had their first bronchoscopy between December 29, 1992 and November 22, 1993. All 21 subjects underwent bronchoscopy solely for research purposes, and had no evidence of active pulmonary disease, either clinically or radiographically. All of the subjects were participating in a human study protocol approved by the Institutional Review Board at Indiana University-Purdue University in Indianapolis (IUPUI). Written informed consent in triplicate was obtained from all subjects. The average age of subjects at the time of bronchoscopy was 35.1 ± 5.5 yr. Based on criteria of the Centers for Disease Control and Prevention (10), five patients were in class A1, five in A2, one in A3, one in B1, two in B2, four in B3, one in C2, and two in C3. For analysis purposes, subjects were grouped according to absolute CD4⁺ cell counts as follows: > 500/µl: Group I (n = 6); 200 to 500 /µl: Group II (n = 8); and < 200/µl: Group III (n = 7).

BAL

After anesthetizing the subjects' upper airways with 2% topical lidocaine, we performed BAL through a fiberoptic bronchoscope wedged in subsegmental bronchi in the right middle and right lower lobes. A volume of 100 ml of normal saline at room temperature was instilled in 20-ml aliquots into three separate bronchi. Typically, 250 to 300 ml was instilled to obtain a return of 125 to 200 ml of BALF. Recovered lavage fluid was kept on ice until processed. Lavage fluid was filtered through 100-µ Nylon mesh (Tetko Inc., Elmsford, NY) to remove debris, and was centrifuged at 400 × g for 10 min. The acellular BALF component was removed and stored at 70° C. The cell pellet was washed twice and resuspended in medium. Cell differential counts were made on 200 cells. The presence of > 15% lymphocytes was considered indicative of lymphocytic alveolitis.

Determination of HIV RNA in BALF

HIV RNA was measured with the reverse transcription-polymerase chain reaction (RT-PCR), using the Amplicor HIV-1 Monitor Test (Roche Diagnostic Systems, Inc., Branchburg, NJ). This is an in vitro nucleic acid amplification test for the quantitation of HIV RNA in human specimens. Briefly, 200 µl of BALF was added to lysis buffer containing guanidine thiocyanate to lyse viral particles. After ultracentrifugation, RNA was precipitated in 70% ethanol and resuspended in 400 µl of specimen diluent. Reverse transcription and PCR amplification were then conducted in the same reaction mixture, using recombinant Thermus thermophilus DNA polymerase and the primers SK431 and SK462. These primers detect a 142-base sequence located in a highly conserved region of the HIV-1 gag gene (11), and have been biotinylated to permit enzymatic detection of amplified products. A known amount of a second gene target sequence (quantitation standard [QS]) was added to samples to control for variations in amplification and to allow for quantitation of viral load. Specific amplified HIV RNA was isolated by adding the reaction mixture to microwell plates coated with HIV-1-specific and QS-specific oligonucleotide probes. Bound RNA was detected enzymatically with an avidin-horseradish peroxidase conjugate and a substrate containing H₂O₂ and 3,3',5,5'-tetramethylbenzidine. HIV RNA copies/ml were determined by comparing amplified HIV RNA with the amplified QS. This test can detect fewer than two copies of HIV RNA per reaction, and 7.5 or more copies of HIV RNA are detected 100% of the time. Although measured viral copy numbers ranged from 0 to 21,015 copies/ml BALF, only samples containing more than 100 copies/ml were considered positive.

The appropriateness of frozen BALF for HIV RNA detection was validated in two ways. First, some samples were assayed on two separate occasions (with freezing between tests), and gave similar results. Second, a fresh BALF specimen underwent freeze-thawing five times over a 5-d period. The amount of HIV RNA detected was similar regardless of the number of freeze-thaws the sample had undergone (data not shown).

Follow-Up

Information about subjects beginning on June 1, 1997 was sought through multiple sources, including direct patient contact, consultation with the patients' physicians or family members, examination of medical records from hospitals within the Indiana University Medical Center complex, and through records kept at the Damien Center in Indianapolis. This last facility is a support center for HIV-infected subjects in Indianapolis, where many of the subjects were initially recruited. Information sought included whether the subject was alive or dead, current CD4⁺ cell count, use and types of antiretroviral therapy, and clinical course since initially studied. With this information we were able to assign each patient a current CDC classification status. Patients were then classified as either having progressive or stable HIV infection. Progressive disease was defined as worsening of the CDC classification or death caused by an AIDS-related illness. Stable disease referred to patients whose CDC classification remained unchanged during the follow-up period. Follow-up information was complete for all 21 subjects. The mean length of follow-up from initial bronchoscopy was 46.4 ± 18 mo.

Statistical Analysis

Data are expressed as means ± SEM. Chi-square analysis, done with a 2 × 3 contingency table, was used to compare the frequency of detectable HIV in the three patient subgroups. The Mann-Whitney U test for nonparametric data was used to compare the percentage of BALF lymphocytes in the three groups. Linear regression analysis was used to examine the relationship between the percentage of BALF lymphocytes and BALF HIV viral load. Chi-square analysis done with a 2 × 2 contingency table was used to compare the relationship between the presence of lymphocytic alveolitis or the presence of HIV in BALF and HIV disease progression. Values of p  0.05 were considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Seventeen of the 21 subjects had stored BALF available for analysis of HIV viral load. Nine out of seventeen subjects had measurable HIV RNA in BALF. There was a strong correlation between peripheral blood CD4⁺ cell count and the presence of HIV in BALF. No subjects in Group I, four of seven subjects in Group II, and five of six subjects in Group III had measurable HIV RNA in their BALF (Figure 1). Viral loads varied from 135 to 21,000 copies/ml of BALF. When one takes into account the estimate that epithelial lining fluid (ELF) is diluted 1:100 in BALF (12), estimates of HIV viral load in ELF are between 13,500 and 2,100,000 copies/ml. In 10 subjects we had simultaneous serum samples in which viral loads were determined. In nine of these 10 subjects, serum viral loads were higher than BALF viral loads, even after correcting for BALF dilution. However, viral loads were within one log difference between the two groups (serum viral load: 43,608 ± 15,244 copies/ml: adjusted BALF viral load: 19,220 ± 7,751 copies/ml).

View larger version (16K): [in this window] [in a new window]   Figure 1.   Frequency of BALF samples positive for HIV viral RNA in patients at different stages of HIV infection as defined by peripheral blood CD4+ cell counts. HIV viral RNA was determined in acellular BALF with the RT-PCR technique. More than 100 RNA copies/ml BALF was considered positive. Significantly more subjects in Group II (CD4+ cell count 200 to 500/µl) and Group III (CD4+ cell count < 200/µl) had HIV detected in BALF than did subjects in Group I (CD4+ count > 500/µl). Group I, n = 4; Group II, n = 7; Group III, n = 6. *p = 0.03.

We next analyzed the prevalence of lymphocytic alveolitis in HIV-infected subjects at different stages of their disease. The intensity of lymphocytic alveolitis was mild in Group I subjects, increased significantly in subjects with CD4⁺ cell counts between 200/µl and 500/µl, and then declined slightly in the most advanced group (Figure 2). Lymphocytic alveolitis, defined as more than 15% lymphocytes in BALF, was present in three of six Group I subjects, seven of eight Group II subjects, and four of seven Group III subjects. In 10 subjects, bronchoscopy was repeated after a mean 1-yr follow-up period. Interestingly, the percentage of BALF lymphocytes and the prevalence of lymphocytic alveolitis were relatively stable over this period (Figure 3). Only two patients who did not have lymphocytic alveolitis at the first bronchoscopy developed it by the second procedure. No subject who had lymphocytic alveolitis initially lost it by the second procedure.

View larger version (16K): [in this window] [in a new window]   Figure 2.   Percentage of BALF lymphocytes in HIV-infected subjects at different stages of infection as defined by peripheral blood CD4+ cell counts. The percentage of BALF lymphocytes was significantly greater in Group II (CD4+ cell count 200 to 500/µl) subjects than in subjects in Group I (CD4+ cell count > 500/µl). No significant difference existed between Group I and Group III or between Group II and Group III. Group I, n = 6; Group II, n = 8; Group III, n = 7. *p < 0.05.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Percentage of BALF lymphocytes in 10 HIV-infected subjects who had a second bronchoscopy (white bars) performed at a mean of 11 ± 9.7 mo (range: 3 to 32 mo) after a first bronchoscopy (black bars). Only two patients who did not have lymphocytic alveolitis at the first bronchoscopy developed it by the second procedure (indicated by *). No subject who initially had lymphocytic alveolitis lost it by the second procedure.

We have hypothesized that the presence of lymphocytic alveolitis in HIV infection represents an appropriate immune response by CTL to HIV-infected cells in the lung, especially in otherwise asymptomatic individuals. A corollary to this hypothesis is that higher BALF concentrations of HIV (or the antigen load) should be associated with a more vigorous immune response by CTL and thus with greater numbers of alveolar lymphocytes. We examined the relationship between the percentage of BALF lymphocytes and BALF viral load in HIV-infected subjects. Figure 4A shows a weak but significant correlation between the percentage of BALF lymphocytes and BALF viral loads when all HIV-infected subjects were included in the analysis (R² = 0.376, p = 0.01). However, the alveolar lymphocyte population changes from primarily a CTL population in patients with early and middle stage HIV infection to CD8⁺ suppressor cells in patients with advanced disease (13, 14). Since we have hypothesized that HIV induces a CTL response in the lung, the relationship between BALF lymphocytes and viral load should be especially strong in individuals whose alveolar lymphocytes consist primarily of CTL. Indeed, Figure 4B shows a very strong and highly significant correlation between the percentage of BALF lymphocytes and BALF viral load in subjects with early and middle stage HIV disease as determined by peripheral blood CD4⁺ cell counts exceeding 200/µl (R² = 0.744, p < 0.01).

View larger version (17K): [in this window] [in a new window]   Figure 4.   Correlation between BALF HIV viral load and percent BALF lymphocytes in all HIV-infected subjects (A) and in those individuals with peripheral blood CD4+ cell counts above 200/µl (B). There was a weak correlation between BALF lymphocytes and BALF viral load in HIV-infected subjects. However, when the analysis was confined to individuals whose BALF lymphocytes probably represented CTL (patients with CD4+ cell counts > 200/µl), there was a highly significant correlation between percent BALF lymphocytes and BALF viral load.

We also examined the long-term outcome in this group of 21 subjects who were initially studied in 1993 or earlier. During the nearly 4-yr follow-up period, eight of the 21 subjects died. Two of the deaths were from causes unrelated to HIV infection. The CD4⁺ cell count of one individual was declining at the time of death, and he was therefore considered to have had progressive HIV disease. The CD4⁺ cell count of the second individual was increasing at the time of his death, and there had been no opportunistic infections documented prior to death. He was therefore considered stable with regard to HIV disease. In total, nine of 21 patients had stable HIV infection. Disease progression was documented in the remaining 12 individuals. Of the 13 patients who remained alive at the end of the follow-up period, 11 were taking highly active antiretroviral therapy consisting of three drugs, one of which was a protease inhibitor.

Previous investigators have shown that lymphocytic alveolitis in HIV-infected subjects is associated with a poor prognosis (4). In accord with this, we found that 64% (nine of 14) of individuals with lymphocytic alveolitis showed HIV disease progression over 4 yr of follow-up. Progression was seen in only 43% (three of seven) individuals without lymphocytic alveolitis. Despite these observations, chi-square analysis showed that the presence of lymphocytic alveolitis was a relatively poor predictor of HIV disease progression (p = 0.64). However, we have postulated that lymphocytic alveolitis represents an appropriate immune response to HIV-infected cells in the lung. Thus, the poor prognosis associated with lymphocytic alveolitis could merely reflect a higher viral load in patients with this finding. Indeed, we found the presence of HIV viral particles in BALF to be a much stronger predictor of disease progression than the presence of lymphocytic alveolitis (p = 0.089 by chi-square analysis). Eighty-nine percent (eight of nine) individuals with measurable HIV in their BALF demonstrated disease progression, as compred with only 38% (three of eight) of individuals without this finding.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study we showed that the ability to detect HIV in BALF increases with disease progression. Lymphocytic alveolitis can be seen at any stage of HIV infection, but appears to be most intense in patients with middle stage disease. We found a strong correlation between BALF HIV viral load and the percentage of alveolar lymphocytes in HIV-infected subjects with peripheral blood CD4⁺ cell counts above 200/µl (individuals with early and middle stage disease). Including patients with late stage HIV disease weakened the correlation between these two factors. We also showed that readily detectable free virus in BALF is a stronger predictor of HIV disease progression than is the presence of lymphocytic alveolitis.

Numerous other reports have documented HIV in the lungs of HIV-infected subjects (15). However, techniques described in these reports have been cumbersome and time consuming. We measured BALF HIV with the same RT-PCR technique used to measure plasma viral loads in clinical samples. This test was easy to perform and had a rapid turnaround time. No evidence of assay inhibitors in BALF fluid was found. Moreover, the assay could be performed on frozen BALF samples with a high degree of reproducibility. Thus, the availability of commercially available RT-PCR diagnostic tools should facilitate studies of the effects of HIV in the lung.

In most subjects, serum viral loads were slightly greater than BALF viral loads, even after correction for the dilution factor of BAL. All BALF samples were obtained before the advent of highly active antiretroviral therapy (HAART), which can dramatically reduce plasma viral loads (20, 21). However, even in the presence of substantial decreases in viremia, tissue viral levels can remain high (22, 23). It is not unreasonable to speculate that the lung may also serve as a reservoir for HIV during HAART. In 1998, five years after their initial bronchoscopy, we had the opportunity to restudy two individuals from our original population. Both were receiving HAART. One individual who had no detectable HIV in BALF in 1993 continues to be negative for HIV in both blood and BALF. The other individual had the highest BALF viral load that we have measured (21,015 copies/ml BALF) in 1993. Today, despite receiving HAART, his BALF viral load is 9,558 copies/ml, with a corresponding plasma viral load of 2,463 copies/ml. The issue of whether HAART can affect tissue viral loads is an area of intense investigation. Trials are currently underway to determine the effect of HAART on lymphoid tissue and BALF viral loads.

There was a significant correlation between the percentage of BALF lymphocytes and BALF HIV viral load in the study population. One could argue that this merely reflects the fact that lymphocytes are the usual source of infected cells. Two points argue against this as the explanation. First, most lymphocytes in BALF from HIV-infected subjects are CD8⁺ cells (3, 5), which are not normally infected with HIV. Second, if BALF viral load was correlated only with BALF lymphocyte number, then one would predict this relationship to extend across all HIV-infected subjects. Although there was a significant correlation between BALF viral load and percent lymphocytes in HIV-infected subjects as a whole, this correlation was relatively weak, as demonstrated by the low correlation coefficient (R² = 0.376). In fact, the correlation between these variables is substantially greater only if HIV-infected subjects with CD4⁺ cell counts above 200/µl are considered (R² = 0.744). We speculate that this reflects a major difference in the CD8⁺ T-cell phenotype between patients with late stage HIV disease (i.e., CD4⁺ cell count < 200/µl) and individuals with middle and early stage disease (CD4⁺ count  200/µl). Other investigators have shown that in late HIV infection, CD8⁺/ CD57 CTL are replaced by CD8⁺/CD57⁺ suppressor T cells (13, 14). We have also found that the function of alveolar lymphocytes changes with HIV disease progression. In particular, we have found that alveolar lymphocytes from middle stage HIV infection secrete significantly more interferon-, a cytokine known to be secreted by CTL (24, 25), than do alveolar lymphocytes from patients with CD4⁺ cell counts below 200/µl (26). Since we have postulated that lymphocytic alveolitis represents an appropriate in situ immune response to HIV-infected cells in the lung, then a higher antigenic load (as reflected by BALF HIV viral load) should lead to a more vigorous immune response. However, this will be true only for those individuals whose CD8⁺ cells are primarily CTL: namely, those individuals with CD4⁺ cell counts over 200/µl. This hypothesis would explain why the correlation between BALF HIV viral load and BALF lymphocytes is significantly stronger if only subjects with CD4⁺ counts  200/µl are considered.

The positive correlation between BALF viral load and alveolar lymphocytes in patients with CD4⁺ cell counts above 200/µl suggests that alveolar CTL are ineffective in controlling HIV in the lung. There are many potential explanations for the inability of HIV-specific CTL to eradicate HIV infection (27). HIV may undergo antigenic drift too rapidly for the immune response to keep up. It has also been suggested that prolonged, high level antigen exposure may exhaust the CTL response. There may be a reservoir of infected cells that produce infectious virus that is not susceptible to attack by CTL, possibly through alteration of surface markers important in immune recognition. It is also possible that CTL that are present are impaired by the presence of suppressor cells. Furthermore, it is not clear that effective killing of HIV-infected cells actually results in viral death as well. Lysis of cells containing large amounts of intracellular virus (i.e., macrophages) may simply release viable virions into the surrounding environment.

Previous investigators have shown that the presence of lymphocytic alveolitis portends a poor prognosis in HIV- infected subjects (4). In our study we found the presence of HIV in BALF to be a better predictor of disease progression than were high BALF lymphocyte counts. On the basis of the positive correlation between BALF HIV viral load and percent BALF lymphocytes demonstrated in this study, we believe that lymphocytic alveolitis in HIV infection is merely a surrogate marker for a high viral burden. This finding is congruent with recent data suggesting that plasma viral loads are powerful predictors of HIV disease progression, even more so than peripheral blood CD4⁺ cell counts (28). We speculate that an increasing BALF viral load with advancing HIV disease reflects progressive impairment of protective pulmonary immune responses, manifested primarily as a loss of HIV-specific CTL.

An important limitation of this study was the relatively small patient population. Accordingly, the data should be interpreted cautiously. Most HIV-infected subjects are currently receiving protease inhibitors, thus making it difficult to recruit more patients for a prospective study. However, we have shown that the RT-PCR technique for measuring BALF viral loads can be applied to archived BALF samples. Therefore, other investigators with stored patient samples could address similar questions in their patient populations. An advantage to this approach is that the lengthy necessary follow-up period will have already accrued at the time of RT-PCR assay.

In conclusion, advanced HIV disease is characterized by easily detectable HIV in BALF, by the presence of lymphocytic alveolitis, and by altered phenotype and function of alveolar lymphocytes. We speculate that persistent HIV in the lung compartment, and the progressive impairment in pulmonary immunity in HIV infection, are intimately related. A better understanding of how HIV chronically evades host immune responses will hopefully lead to new therapeutic strategies aimed at reducing viral burden and restoring normal pulmonary immune function.