INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Considerable evidence has accumulated that a diffusion impairment in the lungs is associated with human immunodeficiency virus (HIV) infection (1). Although exacerbated by cigarette smoking, intravenous drug use, and Pneumocystis carinii pneumonia (PCP), an HIV-related drop in diffusing capacity occurs independently of these factors (1). The pathophysiologic basis underlying this gas exchange abnormality remains obscure. However, it is possible that either an inflammatory pneumonitis resulting in infiltration of fluid, cells, or matrix creates a greater barrier to diffusion or that parenchymal lung damage secondary to HIV-related "inflammatory events" may be responsible (1, 4).

Recent evidence has demonstrated that a declining diffusing capacity of the lungs for carbon monoxide (DL_CO), independent of the blood CD4 count, represents a "risk factor" for the subsequent development of PCP (7). This connection between a declining DL_CO and opportunistic pneumonia suggests that factors resulting in regional immunodeficiency in the lung are also associated with alterations in gas exchange (7). As such, understanding the nature of HIV-related diffusion impairment may provide important insight regarding the effects of immune/inflammatory dysregulation on lung structure and function.

The purpose of the present study was to delineate the nature of this impairment in pulmonary gas exchange in individuals with no history of acquired immunodeficiency syndrome (AIDS)-related pulmonary complications. We investigated a group of HIV⁺ subjects with unexplained reductions in DL_CO using high-resolution computed tomography (HRCT) of the chest and a separation of diffusing capacity into its membrane (Dm) and capillary blood volume (Vc) components. We compared this group with low DL_CO to HIV⁺ subjects with more normal gas exchange and also with a group of HIV volunteers matched for age and smoking history.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

The current report is a baseline, cross-sectional analysis of a prospective, observational study. Clinically stable HIV-seropositive individuals (n = 243) were recruited from the central Ohio area for a study delineating the physiologic properties of the lung in individuals without prior AIDS-related pulmonary complications. Subjects were predominantly male (87%), with a mean (± SE) age of 34.6 ± 0.5 yr and mean CD4 count of 388.2 ± 17.8 cells/mm³. Sixty-seven percent had smoked cigarettes with a mean pack-year smoking history of 17.1 ± 1.2. Subjects with a history of PCP, pulmonary tuberculosis, Kaposi's sarcoma involving the lung, or other recognized pulmonary complications of AIDS were excluded. However, prior AIDS-defining diagnoses not involving the lungs were not used as exclusion criteria. The majority of subjects were recruited through newspaper advertisements, word of mouth, and via the Ohio State University Medical Center AIDS Clinical Trials Unit. The study was approved by the Ohio State University Human Subjects Institutional Review Board and informed consent was obtained from all individuals.

As a control group 30 HIV subjects were studied. Control subjects were similar in terms of sex (87% male), age (34.8 ± 1.8 yr), and smoking history (63% smokers; 20.3 ± 4.0 pack-years).

Testing Procedure

Upon entry into the study, all HIV⁺ subjects underwent complete pulmonary function testing. A subset of these individuals including those with unexplained decreases in diffusing capacity as well as those with normal values were targeted for more intensive investigation, including HRCT (n = 95), determination of Vc and Dm (n = 96), and bronchoalveolar lavage (BAL) (n = 62).

Among control subjects, 30 underwent spirometry and measurement of lung volumes; 29 had determination of DL_CO. In virtually all control subjects HRCT was performed (n = 29) and Vc and Dm measured (n = 27).

Baseline Pulmonary Function Testing

Pulmonary function studies were performed on a Keystone model pulmonary function analyzer (S&M Instrument Co. Inc., Doylestown, PA) or a Sensormedics Series V6200 Autobox (Sensormedics, Inc., Yorba Linda, CA) according to American Thoracic Society (ATS) standards. Reproducibility between the two machines was documented by testing a number of individuals on both systems. Lung volumes were determined using the helium dilution technique (Keystone) or nitrogen washout (Sensormedics); diffusing capacities were determined using the single-breath method and adjusted for hematocrit (9). The most appropriate reference equations for our laboratory and testing conditions were chosen by applying a number of reference equations to pulmonary function results of normal volunteers. As such, normal standards for spirometry were those of Knudson and coworkers (10); lung volumes, Goldman and Becklake (11); and diffusing capacity, Roca and coworkers (12).

Vc and Dm

The movement of carbon monoxide (CO) from the alveolar space to the blood (i.e., the diffusing capacity) is determined by the properties of the alveolar capillary membrane as well as by the volume of blood in the pulmonary capillaries and the reaction rate of CO with hemoglobin. The relative contribution of these two components can be resolved physiologically by the method of Roughton and Forster (13) which divides the resistance to diffusion (1/DL_CO) into the membrane resistance (1/Dm) and the resistance occurring in the vascular bed 1/Vc(). Here,  represents the reaction rate of CO with hemoglobin and varies with the PO₂, according to the following formula: 1/ = 0.59 + 0.0077 (PO₂) (13). By measuring the diffusing capacity and plotting 1/ versus 1/DL_CO at various inspired oxygen concentrations, Dm and Vc can be solved algebraically. 1/Vc equals the slope of the regression line and Dm equals the x intercept. Subjects performed a total of six diffusing capacity maneuvers, 2 each at low, medium, and high fraction of inspired oxygen (FI_O2). Before performance of the DL_CO measurements a sample of gas was taken and measured for CO to estimate baseline CO in equilibrium with blood carboxyhemoglobin. Normal values for capillary blood volume and membrane diffusing capacity were those of Georges and colleagues (14).

HRCT Technique

Computed tomographic (CT) scanning was performed on all subjects with either a GE 9800 CT scanner (GE Medical Systems, Milwaukee, WI) or a Picker PQ 2000 CT scanner (Solon, OH) with 1.5-mm collimation at 10-mm intervals through the chest during suspended inspiration at TLC with the patient in the supine position. Images were reconstructed using the high spatial frequency algorithm and photographed at lung (window width 1,500 Hounsfield units [HU], level 700 HU) and mediastinal (window width 400 HU, level 0 HU) windows. No intravenous contrast was administered.

HRCT Evaluation

All scans were interpreted by two experienced chest radiologists blinded to HIV status and physiologic data. The CT scans were evaluated for the presence and severity of lung parenchymal processes that might be associated with an abnormality in gas exchange and were specifically examined for the presence of emphysema or interstitial lung disease. The presence of emphysema was noted if there were bullae, thin-walled cystic spaces, or abnormal decreases in attenuation accompanied by vascular disruption. The presence of interstitial lung disease was noted if there was any evidence of abnormal linear densities, parenchymal banding, or reticular abnormalities. Scores of 0-10 were assigned for each lobe according to the percentage, if any, of each lobe affected. For the purposes of the study the lingula was considered a separate lobe. The total score represents the sum of the scores for the individual lobes.

Bronchoscopy with BAL

BAL was performed according to standard techniques (15). Five 20-ml aliquots of sterile saline were then instilled into the distal alveolar spaces of the right middle lobe and subsequently aspirated back into suction traps. BAL fluid was examined by an experienced microbiologist, blinded to HIV status for the presence of fungi, acid-fast bacilli (AFB), PCP and bacteria by special microbiologic staining. Viral inclusions consistent with cytomegalovirus (CMV) infection were also sought.

Data Analysis

We divided HIV⁺ subjects into three groups based on their DL_CO values, expressed as percentage of predicted. Subjects with DL_CO values in the bottom 25th percentile (< 72% of predicted) were considered Group 1; subjects with DL_CO values between the 25th and 75th percentile were considered Group 2 and those with DL_CO values > 75th percentile (> 95% of predicted) were considered Group 3. The fourth group examined was the HIV control subjects matched for age and smoking history.

One-way analysis of variance (ANOVA) was used to compare differences in mean values among the groups. Post hoc analysis was performed using the Dunnett procedure. We specifically compared clinical variables, results of physiologic testing, and HRCT scanning of the HIV⁺ subjects with the lowest values of DL_CO (Group 1) with the other groups tested. A p < 0.05 was considered statistically significant for all comparisons. Data are presented as group means ± SE.

To compare the frequency and relative degree of HRCT emphysema between groups, each individual was given a score of 0 (no emphysema), 1 (a positive emphysema score of 1-5), or 2 (a score > 5). A Pearson's chi-square test with the Bonferroni procedure was used to compare differences between groups.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Comparison Groups: Diffusing Capacity for Carbon Monoxide

The distribution of DL_CO values among the various study groups is demonstrated in Figure 1. As demonstrated in the figure, the lowest value recorded in the HIV group corresponds to the highest value among Group 1 subjects. Thus, 25% of the HIV⁺ subjects had DL_CO values lower than the lowest value recorded in our control population.

View larger version (8K): [in this window] [in a new window]   Figure 1.   Range of DLCO values for the various comparison groups. Horizontal lines indicate group means. HIV+ subjects with DLCO values in the bottom 25th percentile (< 72% of predicted) were considered Group 1 (n = 59); subjects with DLCO values between the 25th and 75th percentile were considered Group 2 (n = 124); and those with DLCO values > 75th percentile (> 95% of predicted) were considered Group 3 (n = 60). For HIV subjects, n = 29.

Dm and Vc

Table 1 demonstrates the components of diffusing capacity among various study groups for the subset of subjects undergoing these tests.

As demonstrated in the table, Group 1 subjects had reductions in both Dm and Vc compared with the other groups. With regard to Dm, Group 1 subjects were significantly different from HIV⁺ subjects in Group 3, but were not different from Group 2 subjects or the HIV group. On the other hand, Vc measurements in Group 1 subjects were significantly different from all other groups tested. This suggests that a loss of capillary blood volume is an important feature of HIV-associated diffusion impairment.

Spirometry/Lung Volumes

As demonstrated in Table 2, there was no significant difference among any of the groups with respect to FEV₁/FVC. Furthermore, while FEV₁ and TLC were higher in Group 3 subjects compared with the other groups, lung volumes were well maintained among all groups studied (Table 2). The relative absence of restrictive physiology among the HIV⁺ population is underscored by the fact that only 11 of 243 (4.5%) HIV⁺ subjects had TLC values consistent with a restrictive process, (i.e., TLC < 80% of predicted), whereas 27 (11.1%) had physiologic evidence of hyperinflation (TLC > 120% of predicted).

HRCT of the Chest

HRCT evidence of even early interstitial lung disease was virtually nonexistent in any of the groups studied. Specifically, no Group 1 subject had evidence of reticulations, linear densities, or any other changes consistent with early interstitial fibrosis. In fact, of 95 HIV⁺ subjects undergoing HRCT scanning, only one (normal DL_CO) had increased interstitial markings on CT scan, consisting of reticulations isolated to the right lower lobe. The virtual absence of any HRCT findings of linear densities, reticulations, or parenchymal banding, suggestive of early interstitial fibrosis is consistent with the physiologic finding of well-preserved total lung capacities in this population.

HRCT evidence of emphysema was identified and was correlated with decreased DL_CO. Indeed, among Group 1 subjects 50% demonstrated detectable emphysema on CT and nearly all subjects with CT emphysema scores > 5 were Group 1 subjects (Figure 2).

View larger version (9K): [in this window] [in a new window]   Figure 2.   Proportions of subjects in various groups with CT emphysema scores > 5. Group 1 significantly different from all other groups (p < 0.02). HIV+ subjects with DLCO values in the bottom 25th percentile (< 72% of predicted) were considered Group 1 (n = 22); subjects with DLCO values between the 25th and 75th percentile were considered Group 2 (n = 49); and those with DLCO values > 75th percentile (> 95% of predicted) were considered Group 3 (n = 24). For HIV subjects, n = 29.

Clinical Correlates

Table 3 demonstrates selected clinical variables among the various study groups. As expected, mean CD4 counts were significantly greater in the HIV group, but were not significantly different among the various HIV⁺ groups. There were also no significant differences among any of the groups with respect to the percentage of ever smokers, current smokers, or history of intravenous drug use. Conversely, whereas there was no significant difference in pack-year history of cigarette smoking between Group 1 subjects and HIV control subjects, HIV⁺ subjects in Group 1 had a significantly higher mean pack-year history than HIV⁺ subjects in Groups 2 or 3. These results suggest a potentially important role for cigarette smoking in accentuating the diffusion impairment associated with HIV.

Interestingly, the most important clinical variable associated with a decreased DL_CO among the HIV⁺ subjects appeared to be the body mass index, which was approximately 10% lower in the Group 1 subjects than in all the other study groups (Table 3).

Although not shown in Table 3, we studied other clinical variables in an attempt to further address the question as to why certain HIV⁺ individuals are predisposed to diffusion impairment. However, we found no significant difference among the four groups with respect to occupational exposure to dust or fumes, history of doctor-confirmed pneumonia, or family history of chronic obstructive pulmonary disease. Furthermore, we specifically compared the three HIV⁺ groups with respect to HIV transmission risk, duration of HIV infection, use of inhaled pentamidine, and use of antiretroviral medications. Again, we found no significant differences among the groups with respect to these variables. It should be pointed out that virtually no subjects were taking protease inhibitors as they were studied prior to the widespread availability of these agents.

BAL

No evidence of PCP, viral inclusions consistent with CMV or other pathogens was observed on cytologic examination of BAL fluid from any of the HIV⁺ subjects. This suggests that the observed physiologic and HRCT findings are not related to inapparent opportunistic infection.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Because diffusion impairment has been described by all investigators studying pulmonary function in the setting of HIV (1- 8) and because this finding appears to occur with increased frequency at later stages of infection (1, 4), it is possible that this abnormality may be an important physiologic manifestation of HIV-related immune dysfunction on respiratory structure and function. The potential relevance of this concept is underscored by the recent observation of Stansell and colleagues (7) who have shown that a decrease in the DL_CO is a powerful predictor of subsequent progression to PCP. This predictive ability of the DL_CO is independent of CD4 count and remains significant even if measurements within 60 d of PCP diagnosis are excluded, suggesting that the drop in DL_CO is not merely from incipient pneumonia (7). These investigators (7) and others (1) have hypothesized that an accelerated decline in lung function accompanies increasing immunosuppression in HIV⁺ individuals, even in the absence of clinically apparent opportunistic pneumonia. To date, the pathophysiologic nature of this accelerated decline in pulmonary function has remained totally obscure. Our physiologic data support a process which is associated with a loss of Vc, with overall preservation of TLC. HRCT data suggest a process consistent with early emphysema, a pathologic process that accommodates this pattern of physiologic findings.

The studies presented herein suggest that HIV-related diffusion impairment is not merely a result of interstitial infiltration of fluid, cells, or matrix, creating a greater barrier to diffusion. To begin with, the observed drops in diffusing capacity are accompanied by well-preserved lung volumes. If diffusion impairment were secondary to increased matrix deposition or alveolar capillary thickening, decreases in TLC would be expected to parallel drops in DL_CO. In addition, cellular infiltration alone would be expected to result in isolated drops in the membrane component of the diffusing capacity, whereas we have described significant reductions in Vc.

The demonstrated loss of Vc serves to distinguish diffusion impairment associated with HIV seropositivity with the additional decline in DL_CO resulting from PCP (16). Sankary and colleagues studied HIV⁺ patients with acute PCP and have found an isolated and reversible drop in the Dm suggesting a temporary alveolar-capillary block (16). Such a drop in the membrane component that reverses with therapy has also been described in early-stage pulmonary sarcoid (17) and with tropical eosinophilia (18). Presumably, infiltration of the alveolar space or interstitium secondary to fluid and/or inflammatory cells creates a greater barrier to diffusion (17). With therapy, the Dm returns to baseline values (17, 18). On the other hand, evidence exists that processes associated with parenchymal lung destruction, such as pulmonary emphysema are also associated with significant decreases in Vc (17).

The loss of Vc that we describe is of particular interest given several lines of evidence suggesting that HIV⁺ individuals are predisposed to pulmonary vascular abnormalities. For example, a number of reports have described an association between severe pulmonary hypertension and HIV infection (19, 20). Furthermore, recent evidence suggests that right ventricular dysfunction (presumed secondary to pulmonary vascular abnormalities) is extremely common in advanced HIV infection. Casalino and colleagues (21) performed cine-magnetic resonance imaging (MRI) of the heart in asymptomatic AIDS patients and found that compared with control subjects, AIDS patients had significantly decreased right ventricular ejection fractions, as well as significantly increased right ventricular end-diastolic and end-systolic volumes. Furthermore, right ventricular wall motion abnormalities were far more common than left ventricular abnormalities in HIV⁺ patients. The investigators concluded that the most likely pathophysiologic mechanism explaining their findings was a systolic overload to the right ventricle, secondary to a long-standing pulmonary circulation abnormality (21). Our data, in a population with less advanced HIV, support the concept that important alterations in the pulmonary vascular bed may be relatively common in HIV infection.

Although our study is thus far limited by the lack of histopathologic confirmation, a number of studies have demonstrated that HRCT is a reliable in vivo morphologic surrogate for parenchymal lung destruction as occurs in emphysema (22). Furthermore, recent physiologic and HRCT correlative studies have demonstrated that the diffusing capacity is the functional test that best correlates with HRCT evidence of emphysema (25). Although the changes we describe are generally mild, the significant association of emphysema score with diffusion impairment suggests the HRCT findings are of physiologic relevance. Because visual analysis of the CT scan may be relatively insensitive to microscopic emphysema (23), it is possible that certain HIV⁺ subjects with diffusion impairment and negative scans may have had microscopic changes beyond the resolution of our analysis.

We believe that an important negative finding of our study is the paucity of HRCT or physiologic evidence suggesting that HIV⁺ individuals are predisposed to increased interstitial matrix deposition in the lung. Indeed, virtually no evidence of linear densities, reticulations, or parenchymal banding was noted on HRCT. This radiographic finding is consistent with our physiologic data demonstrating that pulmonary overinflation is more common than lung volume restriction in HIV. The absence of even early fibrosis is especially intriguing in light of the high frequency of lung lymphocyte infiltration occurring in the setting of HIV. Both a CD8⁺ lymphocytic alveolitis and a nonspecific interstitial pneumonitis are commonly found in HIV⁺ individuals and occur at all stages of infection (26). In contrast, other clinical entities associated with lung lymphocyte infiltration are associated with restrictive ventilatory defects secondary to increased lung matrix deposition (30). For example, both sarcoidosis and hypersensitivity pneumonitis are characterized by increased numbers of lung lymphocytes, and this lymphocytic infiltration is felt to be important in the pathogenesis of the interstitial fibrosis that may develop (30).

If certain HIV⁺ individuals are at risk for accelerated lung damage with evidence of early emphysema, these individuals may provide insight into important factors regarding the pathogenesis of emphysema. Identifying the factors responsible for this susceptibility may shed important insight into the pathogenesis of emphysema. Of interest is the association of diminished diffusing capacity with decreased body mass index, suggesting a potential role of poor nutritional status in the development of HIV-associated gas-exchange abnormalities. Recent experimental work has shown that food restriction in laboratory animals leads to morphologic and physiologic changes similar to those occurring in emphysema. The basic mechanisms involved in these changes remain obscure, but a decreased rate of synthesis of connective tissue proteins could be responsible (33). It should also be noted that HIV⁺ subjects with diffusion impairment had a cumulative pack-year history of cigarette smoking similar to HIV subjects, but significantly greater than HIV⁺ subjects with normal DL_CO. This finding is consistent with an interaction between smoking and HIV disease, suggesting that HIV⁺ smokers may be particularly susceptible to parenchymal lung damage and abnormalities in gas exchange.

In conclusion, we report a detailed investigation of pulmonary physiology and HRCT scanning of the chest in a population of HIV⁺ individuals who have not yet developed AIDS-related pulmonary complications. Our results suggest that the previously reported impairment in gas exchange in this population of patients involves loss of the Vc, likely representing the development of early emphysema.