INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ozone (O₃) pollution occurs on warm sunny days within and downwind of most urban areas in North America. Short-term respiratory irritant symptoms, lung dysfunction, and bronchoalveolar inflammation have been well documented in controlled O₃ exposures (1) and in natural outdoor exposures (4). Increased daily mortality rates have been associated with increased ambient O₃ levels (5). Although specific pulmonary and extrapulmonary causes or mechanisms for "excess" deaths are difficult to identify, cardiovascular diseases constitute a large proportion of the deaths. Acute O₃-related cardiovascular morbidity may result from primary alterations in cardiovascular function or pulmonary vascular integrity, the release of hormonal, cytokine, or oxidative products, and/or secondary effects from cardiopulmonary dysfunction for which an already compromised cardiovascular system might be unable to compensate.

Hemodynamic results from controlled O₃ exposure studies of humans are mixed. In healthy human volunteers, cardiac output (CO) and other hemodynamic variables did not change significantly after acute oxidant exposure (8, 9), although the combination of 0.60 ppm nitrogen dioxide (NO₂) and 0.45 ppm O₃ was associated with a lower CO than that measured with O₃ alone and filtered air (FA) exposures. No significant O₃ effects on heart rate (HR), blood pressure (BP), rate-pressure product (RPP), and electrocardiogram (ECG) were found in volunteers with coronary artery disease (CAD) after 40 min exercise in 0.2 or 0.3 ppm O₃ (10). However, O₃ exposure can reduce maximal oxygen uptake by unclear mechanisms which may involve cardiac and/or ventilatory limitations (11, 12).

The above human studies utilized noninvasive hemodynamic monitoring and did not directly measure arterial oxygenation. Invasive physiologic monitoring may be more sensitive than noninvasive methods to detect acute hemodynamic effects of O₃ in humans. We hypothesized that (1) O₃ exposure can acutely affect cardiovascular function in humans, and (2) subjects with a clinical cardiovascular disease (e.g., hypertension) respond differently from healthy subjects. To test these hypotheses, we performed controlled O₃ exposures of two groups of volunteers (healthy and hypertensive) with right heart and radial artery catheterizations, allowing repeated direct measurements of central and peripheral hemodynamics and blood sampling for cardiac enzymes, catecholamines, and atrial natriuretic factor. This is the first clinical study to report directly measured responses of hemodynamics, cardiac enzymes, and circulating hormones in humans exposed to an air pollutant (O₃), to our knowledge.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The study was approved by the institutional review board and hospital administration. Written informed consent was obtained from each volunteer, who received a participation fee. We recruited 10 subjects with stable essential hypertension and six healthy control subjects, all of whom were male, > 40 yr of age, and nonsmoking > 2 yr (Table 1). The patient group had physician-diagnosed essential hypertension, treated either pharmacologically for > 1 yr or by nonpharmacologic methods. No subject had cardiopulmonary symptoms, accelerated or uncontrollable hypertension, left ventricular hypertrophy (by ECG), congestive heart failure (CHF), myocardial infarction, cardiac dysrhythmias, or pulmonary diseases, including asthma. Screening tests included a medical and environmental exposure history; physical examination; resting 12-lead ECG; spirometry; venous hemogram, platelet count, prothrombin time, and partial thromboplastin time; urine screening for illicit drugs that might affect hemodynamics; echocardiogram; and cardiac stress test (treadmill) according to a modified Bruce protocol. Hypertensive subjects tended to be older, taller, and heavier than the control subjects (Table 1). All subjects showed unremarkable blood and urine tests, spirometry, ECG, and echocardiogram except for a mild obstructive ventilatory defect in one ex-smoker (No. 2261). Treadmill results were unremarkable, except for one subject (No. 2263) who demonstrated exercise-induced ischemic ECG changes.

Hypertensive subjects continued receiving their antihypertensive medication(s) during screening. Subsequently, antihypertensive medications, nonsteroidal anti-inflammatory drugs, vitamin supplements, caffeine, and alcohol were withheld for at least 7 d prior to and during the exposure sessions. These subjects were monitored during individualized medication withdrawal periods to obtain a target BP of 140 to 180 mm Hg systolic (BPs) and 90 to 110 mm Hg diastolic (BPd) at rest.

Protocol

Exposures. Subjects were studied one at a time year-round. They were advised to avoid ambient respiratory irritants prior to the study. Each subject was exposed for 3 h to filtered air (FA) alone on one day and to 0.30 ppm O₃ in FA on the next day. Ozone was generated from medical-grade oxygen (O₂) by high-voltage discharge (Model IV ozonator; Sander, Germany). Chamber concentrations of O₃ and nitrogen oxides (NOx) were monitored continuously during exposures by ultraviolet photometers (Dasibi Environmental Corp., Glendale, CA) and chemiluminescent analyzers (ThermoElectron [Franklin, MA] and Monitor Labs [San Diego, CA]), respectively, with calibrations traceable to the local air-monitoring authority.

Each subject was admitted to the hospital on a Wednesday morning (Day 1). Symptoms, vital signs, 12-lead ECG, and spirometry were initially recorded at rest in a special procedures room adjacent to the exposure laboratory. Telemetry with ECG was initiated, and blood was collected from an arm vein for determination of precatheterization catecholamine levels. Once the vascular catheters were passed and secured, baseline supine measurements were recorded after a 60-min stabilization period (see below). The subject then underwent a 3-h single-blinded exposure to FA inside an environmentally controlled chamber (13) at 22° C and 50% relative humidity. Approximately 0.05 ppm O₃ was transiently generated in the chamber prior to the subject's entry to produce the characteristic odor. Throughout exposure, the subject alternately exercised on an electric cycle ergometer (Corival; Quinton, Seattle, WA) for 15 min (target ventilation, 30 to 40 L/min) and rested for 15 min. A physician was continuously inside the chamber to monitor the subject. Symptoms and spirometry were recorded every 30 min and hourly, respectively. The subject's breathing was unencumbered except for 3-min intervals during the last rest and exercise periods of each hour, when minute ventilation (E) was measured via a mouthpiece and nonrebreathing valve. For seven hypertensive and four healthy subjects, HR, BP, and pulmonary artery pressure (Ppa) were recorded in the upright position during the final minutes of the first and final exercise/rest periods. Heart rate and BP were recorded by ECG and cuff measurement, respectively, for the other subjects. The subject consumed fluids ad libitum and ate lunch during the exposure. Within 30 to 60 min after completion of exposure, the subject underwent hemodynamic measurements, 12-lead ECG, blood collection, and symptom scoring.

The subject stayed overnight in the hospital's observation unit with ECG telemetry and indwelling catheters. When necessary, catheter discomfort was managed with oral acetaminophen and temporary ice packs; caffeinated beverages and sedating and antihypertensive medications were withheld between and during exposures, with one exception (see COMPLICATIONS).

On Day 2, the subject underwent the same exposure and measurement sequence as on Day 1, except with 0.3 ppm O₃ added to the chamber atmosphere. Catheters were removed after completion of the protocol, and catheter sites were cleaned and bandaged with adequate hemostasis. After evaluation by the study physician, the subject was discharged and resumed his antihypertensive medication, if any. The subject returned to the laboratory on the following day (Day 3) for medical reevaluation, including symptom ratings, vital signs, physical examination, 12-lead ECG, spirometry, and venous blood sampling for measurements of routine blood biochemistries and cardiac enzymes.

Catheterization and hemodynamic measurements. Pulmonary artery and radial artery catheters were placed percutaneously after local infiltration with 1% lidocaine under sterile conditions. An 8.5-French sheath introducer (Arrow International, Reading, PA) was placed in the right internal jugular vein, followed by a 7.5-French flow-directed thermodilution pulmonary artery catheter (Swan-Ganz VIP with Thromboshield coating; American Edwards Laboratories, Santa Ana, CA) with a catheter contamination shield (Cath-Gard; Arrow) connecting the sheath adapter and proximal catheter. The catheter tip was then passed into the central bloodstream, with reference to characteristic right-heart pressures and wave forms, until the pulmonary artery wedge pressure (Ppaw) was determined. Pressure transducers (Telos Medical, Upland, CA) were zero-referenced to the atmosphere at the fourth intercostal space and midaxillary line (phlebostatic axis) before, during, and after exposures. Cardiac output was measured at least in triplicate by the thermodilution method with infusions of iced saline. Core temperature was continuously measured by the thermistor in the pulmonary artery catheter. A 20-gauge radial artery catheter (Arrow) was inserted into a radial artery (after an Allen test) for arterial pressure monitoring and blood sampling. In the few instances in which arterial catheterization was unsuccessful, locally anesthetized brachial artery punctures supplied the necessary samples. Catheter patency was maintained with continuous infusion of saline containing heparin (6 units/h). A portable postcatheterization chest radiograph confirmed the vascular location of the catheter tip (always in the right pulmonary artery).

The two indwelling catheters were connected to an integrated physiologic monitoring and recording system (Marquette Eagle monitor with 7025 software; Marquette Electronics, Milwaukee, WI), which displayed real-time ECG, HR, hemodynamic pressures, CO, arterial-catheter (a-line) BP, noninvasive (cuff) BP, oximetric O₂ saturation (Sa_O2), and core temperature. Additional hemodynamic variables [including cardiac index, stroke volume, pulmonary and systemic vascular resistances, left and right ventricular stroke-work indices, and rate-pressure product (= HR × BPs)] were calculated according to standard formulas. An automated oscillometric cuff sphygmomanometer on the noncatheterized arm measured BP every 5 to 10 min. The physiologic parameters were recorded on hard-copy tracings. Cuff BP data were more complete than a-line data, which were unobtainable or incomplete in seven subjects.

Blood analyses. Blood samples were collected from both catheters (after discarding heparinized catheter volume) immediately before and after each exposure. Concentrations of serum lactate dehydrogenase (LDH) and creatine kinase (CK) and percentages of their isoenzymes were measured in the hospital's chemistry laboratory, according to standard methods. Aliquots were also frozen at 70° C for subsequent quantitative assays of serum troponin T (from both catheters) and plasma atrial natriuretic factor (ANF), prohormone-ANF peptides 1 to 30 and 31 to 67, epinephrine (EPI), and norepinephrine (NOREPI) (from pulmonary artery catheter only), according to enzyme immunologic (Boehringer Laboratories, Wynnewood, PA), radioimmunologic (14, 15), or radioenzymatic (16) methods. Routine blood biochemistries (Chemzyme Plus; SmithKline Beecham Clinical Laboratories, Van Nuys, CA) were measured before and after exposures and on Day 3. The laboratory technicians were blinded to the study conditions.

Respiratory testing. FVC and FEV₁ were measured immediately before and after each exposure with a Spirotech (Model S400; Atlanta, GA) rolling seal spirometer, calibrated daily with a 3-L volumetric syringe. Respiratory rate was counted visually and E was determined from a dry gas meter. Arterial blood samples were anaerobically collected, kept on ice, and analyzed within an hour after each collection for blood gas parameters (including derived Sa_O2) in the hospital's pulmonary function laboratory (ABL520 analyzer; Radiometer, Copenhagen, Denmark) by a technician blinded to study conditions. The abbreviated alveolar air equation was used to calculate the alveolar- arterial oxygen tension difference (AaPO₂). Noninvasive Sa_O2 was continuously monitored with a finger pulse oximeter (Nellcor, Hayward, CA) interfaced with the Eagle monitor. Statistical results from pulse oximetry and derived Sa_O2 data were generally similar. The derived Sa_O2 data are reported here since they were more complete than the pulse oximeter data.

Symptom assessment. Immediately before each exposure and at the end of each rest/exercise cycle, the subject completed a standardized symptom list, which recorded 11 respiratory and systemic symptoms commonly assessed in clinical studies of respiratory irritants (17), plus four cardiovascular symptoms and an item for other (miscellaneous) symptoms. The subject scored each symptom as to its intensity, ranging from zero (not present) to 40 (incapacitating). The individual symptom scores were summed to determine a total score and subtotals for respiratory and cardiovascular symptoms.

Statistical Analysis

BMDP-Dynamic statistical software (SPSS, Inc., Chicago, IL) was used in all analyses. Data for many physiologic and biochemical end points were available, raising the possibility of spurious "significant" results among a large number of statistical tests. Accordingly, analysis focused on a subset of "primary" cardiovascular end points which were considered most relevant to the experimental hypotheses: cardiac index (CI), mean pulmonary artery pressure (), Ppaw, HR, BPs, BPd, mean arterial pressure (), and rate-pressure product (RPP). In addition, respiratory variables previously found to change in response to O₃ stress were analyzed: FVC, FEV₁, AaPO₂, and symptom scores. (Analyses of the remaining "secondary" experimental variables yielded only a few additional meaningful findings, described in RESULTS.) To test Hypotheses 1 and 2 simultaneously for a given variable, we first calculated its change (postexposure minus preexposure) on each day. We then performed analysis of variance with day (FA versus O₃) as a within-subjects factor (repeated measure), and clinical status (hypertensive versus control) as a grouping factor. A significant (p < 0.05) day effect would support Hypothesis 1 that O₃ affects cardiovascular function, whereas a significant interaction of day with clinical status would support Hypothesis 2 that hypertensive and healthy subjects respond to O₃ differently. If the main and interactive effects of status were nonsignificant, the analysis was repeated for all subjects pooled. Additional analyses of variance were performed to test the overall differences between hypertensive and control subjects (independent of experimental exposures), and to test the O₃ responses of each clinical subgroup separately.

Data were incomplete for certain variables because of occasional technical problems. These variables were analyzed as described above, excluding subjects with incomplete data, and were reanalyzed with maximal likelihood estimation (MLE) of missing data using the Newton-Raphson algorithm (18). The conventional analysis and the reanalysis with MLE gave similar statistical conclusions.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Exposure Conditions

The 3-h-average O₃ concentrations during all O₃ chamber exposures (recorded at 5-min intervals) ranged from 0.294 to 0.312 ppm. Standard deviations of O₃ levels did not exceed 0.027 ppm. Chamber O₃ levels on FA exposure days did not exceed 0.01 ppm. Total NOx did not exceed 0.02 ppm. Chamber temperature and relative humidity during the 3-h exposures ranged from 22 to 25° C and 46 to 64%, respectively. Intercurrent ambient exposures to O₃ were low in comparison with experimental exposure levels. At monitoring stations near the laboratory, maximal hourly average concentrations on the study days or the immediately preceding days were 0.08 ppm or lower for 13 of 16 subjects. For the three potentially most exposed subjects, maximal hourly averages were 0.11 to 0.13 ppm.

Cardiovascular Comparison of Control and Hypertensive Subjects

Screening cuff BP averaged 122/82 and 147/88 mm Hg in the control and hypertensive subjects (receiving medication), respectively. Precatheterization BP on Day 1 averaged 119/74 and 147/91 mm Hg in the control and hypertensive (not receiving medication) subjects, respectively. Three of 10 hypertensive subjects had elevated screening BPd ( 90 mm Hg), whereas six of 10 had diastolic hypertension prior to catheterization on Day 1. Thus, the withdrawal of antihypertensive therapy modestly increased BPd without changing BPs. Differences between screening and precatheterization BP were nonsignificant (p < 0.3, ANOVA) for the hypertensive subjects and for all subjects. Overall differences in BPs and BPd were significant (p = 0.005 and p = 0.04, respectively) between control and hypertensive subjects.

In general, expected differences in resting hemodynamics between the healthy and hypertensive subjects (19) were observed. Overall mean BPs from preexposure and postexposure measurements on both exposures was 154 mm Hg in hypertensive subjects compared with 128 mm Hg in normal subjects (p = 0.01). Mean BPd tended to be higher in hypertensive than in control subjects (83 versus 70 mm Hg, respectively; p = 0.07). Mean arterial pressure was higher in hypertensive (107 mm Hg) than in control subjects (92 mm Hg; p = 0.05). Corresponding was 17 mm Hg for hypertensive and 12 mm Hg for control subjects (p = 0.03). The initial (pre-FA) CI was higher in hypertensive than in control subjects (mean, 3.67 versus 2.68 L/min/m²; p = 0.003), although the overall means were less divergent (3.69 for hypertensive versus 3.21 L/min/m² for control subjects; p = 0.07). Other cardiovascular end points did not show statistically significant differences between the two groups.

Cardiovascular Responses

The primary hemodynamic results for hypertensive subjects, control subjects, and both groups combined are shown in Table 2. Within the limitations of this small data set, the hypertensive and control subjects appeared to respond similarly to exposures (i.e., the day-status interaction was nonsignificant for most of the primary end points). Changes preexposure to postexposure in BPs, BPd, , , Ppaw, and CI were not significantly different between FA and O₃. However, resting HR increased after O₃, relative to its preexposure value, significantly more than after FA. The increase in HR, with little change in BPs, gave rise to a significant increase in RPP. If RPP was calculated using pulse pressure (thus allowing influence by BPd as well as BPs), the statistical conclusions did not change. There were no clinically significant changes in 12-lead ECGs or on ECG telemetry throughout the exposures (see COMPLICATIONS). During both exposures, mean core body temperature significantly increased by 0.2 to 0.3° C from a baseline of 36.4 to 36.5° C, similarly in hypertensive and control subjects.

Physiologic results obtained during the first and last exercise/rest periods in both groups are summarized in Table 3. Exercise significantly increased BPs (but not BPd), HR, , and E, as expected. Overall BP during exercise averaged 145/80 mm Hg in control subjects and 169/88 mm Hg in hypertensive subjects. Neither exercise BP nor RPP showed statistically significant day, time, or interaction effects in 2-factor ANOVA. Heart rate tended to rise during the final measurements, to a greater degree in O₃ than in FA (p = 0.06). This response pattern was consistent with that for preexposure and postexposure resting HR, which showed a statistically significant O₃ effect, as described previously.

Cardiac Enzymes

Enzyme values were not statistically different between the pulmonary artery and arterial sources. Values from the pulmonary artery catheter (representing mixed venous blood) were used in the statistical analyses since clinical reference values are based on venous samples. The concentrations of CK, LDH, the percentages of their respective isoenzymes, and troponin T were within the normal clinical range in all samples, including those on the follow-up day (Day 3). Total CK was significantly higher overall in hypertensive than in control subjects. The isoenzyme CK-BB was never detected (normal range, 0%) and, in most samples, CK-MM (normal range, 95 to 100%) constituted 100% of CK. LDH-3 was the only LDH isoenzyme to show a statistically significant change related to O₃, rising from 23.4 to 24.6% during FA exposure, and falling from 26.3 to 25.6% during O₃ exposure in all subjects. The heart-associated fraction of LDH (LDH-1) decreased during both exposures. The heart-associated fraction of CK (CK-MB) was detected in only six of 16 subjects, and in them, only at low levels. Most values for troponin T (normal range, 0.0 to 0.1 ng/ml) were below the limit of detection (0.04 ng/ml), and preliminary inspection suggested that the few samples with detectable (but low) concentrations occurred more or less at random, so no formal analysis was attempted.

Hormonal Responses

Both catecholamines were substantially above the normal basal range [NOREPI, 148 ± 45 (SD) ng/L; EPI, 42 ± 35 (SD) ng/L] throughout the study. The mean plasma NOREPI concentration ranged between 502 and 668 ng/L in the hypertensive subjects and between 574 and 683 ng/L in the control subjects at each preexposure and postexposure measurement. Similarly, mean EPI levels ranged between 58 and 106 ng/L in the hypertensive subjects and between and 77 and 102 ng/L in the control subjects. The catecholamine response pattern was not significantly different between the two groups. The EPI concentration decreased significantly, from 104 to 62 ng/L (p = 0.01) during O₃ exposure in all subjects combined, compared with essentially no change during FA exposure. The catheterization procedure (Day 1) had no statistically significant effect on circulating catecholamine levels for hypertensive subjects, control subjects, and all subjects combined, according to comparisons of blood obtained before and after catheterization. However, mean EPI rose from 45 to 82 ng/L in the control subjects. Atrial natriuretic factor and its components did not show significant O₃-induced changes in either group or for all subjects combined, except for a small decrease in pro-ANF 1-30 in the control subjects (p = 0.04).

Respiratory Responses

Mean resting E in FA and O₃ was similar overall (14 L/min) and in the hypertensive (15 L/min) and control (11 L/min) groups. Hypertensive subjects had higher overall exercise E (mean, 36 L/min) than control subjects (mean, 30 L/min) during exposures. Exercise E averaged 35 L/min in FA and 33 L/min in O₃ for all subjects combined. The diminution in E with O₃ was significant (p = 0.007, 2-factor ANOVA with all subjects combined) and contrasted with the significant rise in HR during O₃ exposures compared with that during FA (see above). Ventilation increased from early to later exercise sessions during both exposures (p = 0.012).

Both control and hypertensive subjects maintained essentially normal baseline lung function throughout both exposures. Mean FEV₁ decreased significantly in hypertensive and control subjects considered separately, and in all subjects considered together, with O₃ exposures as compared with FA response (Table 4). Mean FVC also decreased significantly in the entire group, and in the control group considered separately. For all subjects combined, both mean FVC and FEV₁ decreased by approximately 6% during O₃ exposure, but they changed little during FA exposure.

Arterial oxygenation was maintained by all subjects with both exposures. All subjects' mean Sa_O2 decreased significantly from preexposure to postexposure on both days, and was significantly lower overall on Day 2. The excess loss during O₃ exposure relative to FA (Sa_O2, 0.3%) was not significant. However, in pairwise comparison, the post-O₃ mean of 94.7% was significantly lower than the post-FA mean of 95.7% (p = 0.04). Thirteen of the 16 subjects showed increased preexposure to postexposure AaPO₂ during O₃ exposure, compared with FA (Figure 1). For all subjects, mean AaPO₂ rose < 1 mm Hg in FA but more than 10 mm Hg in O₃, from similar preexposure values (p  0.01). The difference between hypertensive and control subjects was nonsignificant.

View larger version (24K): [in this window] [in a new window]   Figure 1.   Preexposure to postexposure change in AaPO2 (AaPO2) in FA and O3 exposures. Open symbols represent individual subjects; solid squares represent group means; error bars represent 95% confidence limits of means.

Symptoms

The mean total symptom score changed little in FA, and increased during O₃ exposure to an extent that suggested a minimal (barely perceptible) worsening of one symptom in a typical subject. This increase was not statistically significant. However, the subtotal score for respiratory symptoms significantly increased during O₃ exposure for all subjects (p = 0.04). The difference between the hypertensive and control subjects was nonsignificant. The increase attributable to O₃ averaged 9 points, representing a mild worsening of one symptom. Nearly all reported symptoms were rated minimal or mild, although a severe sore throat was reported during O₃ exposure. The subtotal score for cardiovascular symptoms showed no significant variation during exposures, and any minimal or mild symptoms were reported only rarely.

Complications

There were no short- or long-term medical complications related to the temporary withdrawal of antihypertensive medications or during placement, maintenance, or removal of the catheters in the subjects who completed the protocol. Catheter-related discomfort was well tolerated with symptomatic management. Chest radiographs after catheterizations were unremarkable. Hypertensive subject No. 2263 performed only the first two exercise sessions on both study days because, on Day 1, he exhibited frequent ventricular ectopic beats during exercise, which resolved at rest. No clinically significant arrhythmias were observed subsequently. This subject also required a single oral dose of short-acting nifedipine (10 mg) for increasing diastolic hypertension over 2 h (Day 1) and 6.5 h (Day 2) prior to the preexposure measurements. There were no clinically significant changes in routine blood chemistries before and after the exposures in the 16 subjects.

Five additional hypertensive subjects were removed from the study early on Day 1 and their data were not used in the analysis. Two subjects declined to proceed shortly after beginning the initial catheterization. The protocol was terminated by the investigators in three subjects because the ballooninflated pulmonary artery catheter could not be successfully passed and/or transient but recurrent cardiac arrhythmias occurred (two subjects), and transient hypotension occurred during the initial exercise period (one subject). None of the subjects developed further complications.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study is the first report of detailed direct measurements of cardiovascular responses to O₃ in humans. The results indicate minimal clinically unfavorable effects of O₃ inhalation on the heart and vasculature. The effective dose level of O₃ was comparable to a worst-case Southern California ambient exposure, either in representatives of a presumed "at-risk" population (hypertensive) or in healthy volunteers. Thus, our results do not strongly support (but do not unequivocally refute) the hypothesis that O₃ exposure affects cardiovascular function in humans. As discussed later, the results provide important new evidence that pulmonary effects of O₃ potentially increase the hemodynamic load and oxygen demands of the heart, and thereby place patients with cardiovascular disease at risk whether or not O₃ directly affects the circulation.

We initially considered evaluating patients with stable CHF and CAD. However, patients with CHF were considered to be at increased risk from the procedures, could not ethically undergo drug withdrawal, and have complex neurohormonal interactions. Although patients with CAD were actively recruited, few candidates ultimately agreed to participate or fulfilled the entry criteria. Nonetheless, an estimated 60 million adults in the United States have essential hypertension, and they provide a practical and optimal volunteer group for this clinical investigation. Hypertension is both a dependent and an independent risk factor for serious cardiovascular morbidity and mortality (20).

Hemodynamics

Results from animal studies provide the strongest evidence for direct myocardial effects from exposure to realistic levels of O₃. Acute cardiovascular dysfunction (21), microscopic myocardial pathology (24), and abnormal myocardial protein synthesis (25) have been reported. Acute reductions in cardiac output have also been reported in anesthetized dogs (27), but not in conscious resting sheep (28, 29).

On the other hand, human studies have shown equivocal O₃-related effects on cardiovascular function. Our direct hemodynamic results generally support previous noninvasive observations in humans (8). Our subjects maintained myocardial function and systemic hemodynamics in the expected range during and after acute O₃ exposure. The statistically most convincing cardiovascular change attributable to O₃ was the greater increase in resting HR after O₃ than after FA, relative to preexposure levels. In addition, O₃ increased exercise HR slightly on Day 2, whereas E declined slightly. The accompanying increase in RPP with O₃ in the hypertensive subjects reflects increased myocardial O₂ consumption. The tachycardia and increased myocardial work with O₃ exposure may have been related to hypoxemia (see subsequent discussion) but did not elicit significant change in Ppaw. Accordingly, there was no clinical evidence of pulmonary congestion or left ventricular dysfunction. The observed hemodynamic changes are of uncertain clinical significance; they may represent homeostatic cardiovascular and/or hormonal responses to maintain tissue oxygenation in the face of lung dysfunction and decreased oxygenation.

Enzymes and Hormones

There was no biochemical evidence of myocardial injury with either exposure. Although the blood collection schedule for CK and LDH over the 3-d protocol may have missed early, transient rises, troponin T was included in this assay battery since it is a rapidly responding, highly sensitive and specific marker of myocardial-cell injury (30). Stable 12-lead ECGs and ECG telemetry corroborated the biochemical findings.

Adrenergic stimulation and elevation of catecholamines in response to O₃ exposure were expected and considered likely to exert active effects on BP, HR, coronary artery tone, and myocardial function (increasing O₂ consumption and demand), which might in turn precipitate cardiovascular events in hypertensive patients (10). The results in this study were generally inconsistent with these expectations, although elevated "baseline" catecholamine levels may have been a factor. Other indicators of stress such as HR and BP did not follow the catecholamine responses.

The ANF hormonal system derives primarily from the cardiac atria and lungs and has multiple biologic effects, including vasodilation, bronchorelaxation, permeability modulation, and natriuresis, resulting in reduced cardiac venous return, ventricular filling pressures, CO, and BP (31, 32). Vesely and colleagues (15, 33, 34) have reported increased concentrations of ANF and its prohormone peptides in the heart, lungs, and blood in O₃-exposed rats, suggesting that O₃ stimulates this hormonal system. Although decreased after O₃ exposure in our study, ANF and its prohormones did not correspondingly rise in either exposure or subject group.

Respiratory Responses

The modest FVC and FEV₁ losses with O₃ exposure were typical for middle-aged and older adults, whereas related symptoms were, if anything, disproportionately mild. Subjects' perception of O₃-related respiratory symptoms might well have diminished (relative to noncatheterized younger, healthy subjects in previous studies) because of competing symptoms associated with catheterization. However, only a few subjects reported symptoms relatable to the catheters. Chest pain was the only "cardiovascular" symptom to show even a slight, nonsignificant increase during O₃ exposure, relative to FA. In this context, chest pain might well have been a manifestation of respiratory irritation rather than a cardiac effect.

The post-O₃ increase in AaPO₂ likely represents a true effect of O₃, in light of the stability of AaPO₂ at all three measurement times prior to O₃ exposure, and prior evidence that short-term O₃ exposure may impair blood oxygenation. In the earliest study (35) to address that issue, 0.1 ppm O₃ reportedly increased AaPO₂ in healthy subjects who experienced no meaningful spirometric changes. Two subsequent studies (36, 37) failed to confirm that finding. The present observation of significantly increased AaPO₂ after O₃ exposure, accompanied by a small and equivocally significant loss in Sa_O2, does not resolve the uncertainty about previous findings, all at lower effective O₃ doses. Nevertheless, the present finding considerably strengthens the case that O₃ levels sufficient to cause modest FVC and FEV₁ losses and minimal respiratory symptoms can also impair alveolar-arterial gas exchange. Medically, the observed AaPO₂ effect implies an important risk of acute arterial hypoxemia for anyone with preexisting marginal oxygenating capacity exposed to similar O₃ levels. Mechanistic explanation of the increased AaPO₂ requires further investigation. Ozone-induced lung permeability and interstitial edema (which would not be readily detected with our methods), increased airway mucus volume, and/or enhanced smooth muscle tone could conceivably narrow airway caliber, producing ventilation-perfusion mismatching and hypoxemia (38). Our hemodynamic results argue against overt pulmonary edema secondary to left ventricular failure. Although inhaled O₃ decreased arterial oxygenation in our subjects, significant ECG changes or myocardial ischemia or injury (with enzyme elevations) did not occur since the subjects had normal baseline oxygenation (and lung function), modest decline in oxygenation, and sufficient myocardial function and reserve.

Limitations

Interpretation of the hemodynamic results is largely limited by the study design, the small number of subjects, and the selection of subjects with mild hypertension. These limitations reflected ethical and logistical aspects of the protocol, as well as the lack of direct precedent. Specifically, exposure order was not randomized because it was considered ethically questionable for subjects either to undergo two major stresses (catheterization and O₃ exposure) on the same day or to repeat the catheterization on a separate day. Thus, we cannot exclude the possibility that time-dependent confounding factors caused cardiopulmonary changes that were either mistakenly attributed to O₃ or masked real effects of O₃. Confounding seems unlikely with respect to the changes in FVC, FEV₁, and AaPO₂ after O₃ exposure. Accordingly, we conclude that these respiratory changes represent true responses to O₃ typical of middle-aged and older men without chronic respiratory disease. Conclusions concerning cardiovascular effects must be more tentative, in that the hemodynamic end points were generally less stable than respiratory variables. Occasional missing data required maximal likelihood estimation, and the interval between the end of exposure and the hemodynamic measurements may have precluded detection of possible transient hemodynamic responses. In addition, women were excluded from this study in order to avoid confounding hormonal effects (39) and to ensure a more homogeneous study population. Finally, the withdrawal of antihypertensive medications may have been less than adequate, and tissue drug levels may have persisted to some extent, possibly attenuating any BP effects of O₃. The investigators did not believe that prolonged medication withdrawal was medically prudent. As a compromise, abrupt drug stoppage or gradual withdrawal (with beta-blocking agents only), with subsequent monitoring of clinical status, was performed 1 to 3 wk prior to the initial exposure.

Overall, the subjects tolerated the insertion and maintenance of the two catheters well. There were no clinically significant acute or subsequent complications related to the catheters. None of the subjects voluntarily withdrew from the study once the catheters were successfully inserted and functioning. However, carefully performed procedures with sterile technique, continuous clinical monitoring of the subject, catheters, and hemodynamics, experienced medical and technical staff, and coordination of accessory physical and instrumentational resources are necessary for successful and safe implementation of this type of research.

Our study has demonstrated that invasive hemodynamic research can be effectively and safely conducted and may have implications for application in future studies (e.g., in controlled human exposures to inhaled particulates). Epidemiologic studies (40) indicate a consistent association between particulate air pollution and deaths from cardiovascular causes. Preliminary results (43) from an animal model suggest that acute cardiac electrophysiologic abnormalities are induced by exposure to inhaled particles. Some epidemiologic data also suggest that particulate air pollution may indirectly promote cardiac events via a respiratory mechanism (e.g., hypoxia), which is consistent with our results.

In summary, our results did not support the hypothesis of significant acute hemodynamic dysfunction in healthy humans or in subjects with chronic cardiovascular disease (hypertension) exposed to O₃. Overall, we did not find convincing evidence of major or numerous short-term cardiovascular effects from O₃ exposure. However, the increase in RPP in the hypertensive subjects exposed to O₃ may be a clue that such increased myocardial work could pose a clinically significant stress to patients with more severe hypertension. More importantly, the results support our hypothesis in a more general sense, in showing that O₃ can exert cardiovascular effects indirectly by impairing alveolar-arterial O₂ transfer and potentially reducing O₂ supply to the myocardium. Cardiopulmonary interactions may substantially increase the need for compensatory cardiovascular adjustments and the susceptibility to adverse cardiac sequalae in patients with severe or unstable cardiopulmonary disease (e.g., CHF, CAD). Thus, this study strengthens the argument that persons with preexisting cardiovascular disease, with or without concomitant respiratory disease, can be at risk from ambient O₃ pollution.