INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The effect of aging on the normal lung has received considerable attention over the years. Elastic recoil is progressively reduced (1), VC and maximal expiratory flow rates fall (2), closing volume is increased (3), and Pa_O2 diminishes (4). Since Pa_CO2 does not rise, the decrease in Pa_O2 must indicate an increase in the alveolar-arterial PO₂ difference (AaPO₂).

The AaPO₂ will increase if any one or more of three forms of gas-exchange defect develop: (1) right-to-left shunting of venous blood; (2) diffusion limitation of alveolar-capillary O₂ exchange; or (3) ventilation-perfusion (A/) inequality. Right-to-left shunting can occur within the lungs (or through defects in the atrial septum), and this form of defect can be conveniently termed intrapulmonary shunting. In addition, Pa_O2 can be reduced by so-called postpulmonary shunting, in which bronchial or thebesian venous blood enters the oxygenated blood downstream of the lungs. In the first case, mixed venous blood never sees alveolar gas; in the second case, normally oxygenated blood passes into the bronchial and myocardial vasculature before it reaches the systemic arterial vessels as deoxygenated blood.

Each of these varied causes of an increased AaPO₂ could occur in normal subjects. Very few studies have been performed with methods that can distinguish among them, and those few have focused mainly on young, active normal volunteers. Those studies incorporating older subjects have suggested that A/ inequality may increase significantly with age (5). Intrapulmonary shunts have been found to be either small (mainly less than 1% of the cardiac output) or nonexistent (5), postpulmonary shunts appear to be negligible (5), and no evidence for diffusion limitation has been found during rest (5- 8). For older subjects, these conclusions are, however, tentative, having been based on very small numbers of subjects (9).

Since the known changes in lung mechanical properties with age referred to earlier could cause A/ inequality to increase, the present study was designed to determine whether there was indeed an increase in A/ mismatching with age, and if so whether this accounts for any accompanying decline in Pa_O2. The principal tool used to answer these questions was the multiple inert-gas elimination technique (MIGET).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Selection of Subjects

Sixty four normal subjects were studied. All were residents of Barcelona, Spain, and were lifetime nonsmokers not employed in jobs exposing them to known pulmonary toxins. The subject's ages ranged from 18 to 71 yr; there were 21 females and 43 males. None had any history of respiratory, cardiovascular, or systemic disease other than occasional upper-respiratory infections. No such episode had occurred within 2 mo prior to study. Physical examination was within normal limits, and pulmonary function tests (Model 1070; Med-Graphics, St. Paul, MN) revealed normal forced spirometry (Table 1). No FEV₁ values were less than 80% predicted. Mean FEV₁ was 105 ± 11% predicted. The decrease in the ratio of FEV₁ to FVC with age was 0.2 per year, which was within the expected range for healthy subjects.

Subject Preparation

We performed clinical evaluations and pulmonary function testing after giving the subjects an explanation of all procedures and receiving their informed consent. Within a week, each subject then returned after an overnight fast for the principal study of gas exchange. After adequate ulnar collateral arterial bloodflow was assured, a 20-gauge cannula was placed in the radial artery of the nondominant hand. A catheter was placed into a vein on the contralateral arm. In both cases, sterile technique and local anesthesia (1% lidocaine) were used. Not less than 30 min after the beginning of an intravenous infusion of a sterile mixture of dissolved inert gases (for the MIGET; see the subsequent discussion), the subject was connected through a nonrebreathing valve to a heated 10 L volume metallic mixing box (10). Ventilation was continuously recorded with a calibrated Wright respirometer, and mixed expired O₂ and CO₂ concentrations were continuously measured with a calibrated respiratory mass spectrometer (SensorLab N/S V12866; Fisons, Cheshire, UK). These measurements provided a continuous record of O₂ uptake and CO₂ elimination. These data were also used to confirm a steady state of gas exchange both by constancy of values to within ± 5% and by values of the respiratory exchange ratio lying with physiologic limits, an important requirement for interpretation of any gas-exchange data. O₂ was used to estimate cardiac output according to the Fick principle and an assumed arteriovenous O₂ difference of 50 ml · dl¹. This was necessary for the seven subjects (of the total of 64) in whom direct cardiac output measurements required for applying the MIGET were unavailable; correlations of measured and computed cardiac output values for the 57 other subjects provided validation for the computed values as a reasonable substitute in these seven.

Cardiac Output Measurements

The indocyanine green dye technique was used to measure cardiac output. To inscribe the dye curve, 5 mg of dye in 1 ml of water was rapidly flushed into the venous catheter and arterial blood was withdrawn at a constant rate of 20 ml · min¹. In these resting subjects, adequate curves were obtained prior to recirculation for the conventional Stewart-Hamilton analysis with a cardiac output computer (DC-410; Waters Instruments Inc., Rochester, MN). Duplicate measurements requiring about 40 ml blood in all were made and the results averaged.

The MIGET

Using methods described previously from this laboratory (11), we used the MIGET to assess A/ relationships in subjects at rest in the seated upright position. To measure A/ inequality, we used previously described indices (10), principally the second moment of the distributions of ventilation and blood flow on a logarithmic scale log SDV and log SDQ, respectively. The analysis separately yields a value for the intrapulmonary shunt as the fraction of the cardiac output perfusing regions with a A/ ratio below 0.005, the lower limit of resolution of the method (12, 13). The total perfusion to areas with subnormal ventilation (low A/ areas) was defined as the percent perfusion to lung units with a A/ ratio above 0.005 and below 0.1. Duplicate or triplicate measurements were attempted in order to compute the intrasubject variance of the outcome variables with a previously described approach involving the pooling of normalized data from many subjects (9). Such sequential samples were taken 5 min apart, with the inert gas infusion continuing, and including catheter dead space, each sample required 10 ml of blood. In all, 159 samples were successfully taken and processed from the 64 subjects.

Arterial Blood-gas Measurements

Immediately after each inert gas sample was collected as described above, a 3-ml arterial blood sample was taken into a heparinized syringe. Bubbles were removed and the sample was iced prior to measurements, which were made within 30 min. P O₂, PCO₂, pH, O₂ saturation, and [Hb] were measured directly from each sample (BG3 and IL482; Instrumentation Laboratories, Milan, Italy). Values were corrected to body temperature measured once prior to connecting the subject to the mouthpiece of the spirometer.

Statistical Analysis

Results given in the text are expressed as mean ± SD. Linear regression was used to analyze relationships between principal variables. Measured and calculated values of cardiac output were compared with Student's paired t test. Both Pa_O2 expected from the measured amount of intrapulmonary shunting and A/ inequality were calculated (12) and compared with the measured Pa _O2 value for each subject with Student's paired t test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A/ Inequality

Figure 1 shows the principal dispersion data for the A/ distribution (log SDQ and log SDV). The two variables explicitly exclude contributions of intrapulmonary shunting, which were generally very small and which are analyzed later. Very few data lay above the previously reported reference limits for young normal subjects (9). Thus, in only two cases was log SDQ above the reference limit of 0.6, and in only six cases was log SDV above its limit of 0.65. Nevertheless, an in-depth analysis of extreme data points did not yield any insight into the mechanisms that were involved.

View larger version (22K): [in this window] [in a new window]   Figure 1.   Ventilation-perfusion (A/) inequality increases with age. Upper panel shows a slight increase in dispersion of the blood flow distribution (log SDQ) with age, and lower panel a similar increase for the ventilation distribution (log SDV). In each graph, the regression line is indicated by a continuous line.

The dispersion of A/ distributions increases with age. The rates of increase are similar for both log SDQ and log SDV, but it must be emphasized that the magnitude of the age effect is on average physiologically small and clinically not significant. For example, from the log SDQ regression line, mean values would be only 0.36 at age 20 yr and 0.47 at age 70 yr. As shown by West (14), these values of dispersion correspond to an AaPO₂ of only about 5 mm Hg and 15 mm Hg, respectively, or to arterial PO₂ values of about 95 to 100 mm Hg and 85 to 90 mm Hg, respectively, depending on the arterial PCO₂. Consequences for arterial O₂ saturation would be very small. In contrast, patients with severe lung disease often develop log SDQ values of 2.0 or greater (15); a perfectly homogeneous lung would have a log SDQ of 0.

The variance in both log SDQ and log SDV is considerable, as Figure 1 shows. The sources of this variance are discussed subsequently, but the low correlation coefficients of 0.35 and 0.24 for each variable indicate that only 12% of the total variance is age-related.

The impact of intrasubject variance due to a combination of experimental errors and biologic change in samples taken 5 min apart was examined by normalizing and then pooling the dispersion parameters for each subject (9). In the present study, this approach indicated that only 11% of the total variance of the dispersion indices (Figure 1) was due to intrasubject variation. The majority of the variance in A/ dispersion therefore remained unexplained, but must have been of intersubject origin. An obvious candidate for the cause of such intersubject variance is the range of spirometric values (Table 1). However, when log SDQ values were plotted against either FEV₁ as %predicted or against the FEV₁/FVC ratio, no evidence was found that these spirometric variables had any relationship to A/ inequality. We were therefore unable to account for the majority of the observed variance in A/ inequality through any physiologic variable or possible measured source of variability such as age, height, weight, expiratory flow rates, or experimental errors.

Since log SDQ and log SDV do not reflect unventilated regions (nor intrapulmonary shunting), we examined the amount of blood flow perfusing areas with a subnormal A/ ratio (i.e., areas of A/ < 0.1). It is these areas that contribute most to the size of the AaPO₂ and which would therefore cause a decrease in arterial PO₂. Expressed as a percentage of the cardiac output, this low A/ area perfusion is shown in Figure 2. In all but seven of the 64 subjects, there was no more than 0.75% of the cardiac output in such areas, a trivial amount in terms of arterial oxygenation. No subject had more than 3% of the cardiac output in low A/ regions.

View larger version (13K): [in this window] [in a new window]   Figure 2.   Combined perfusion of unventilated and low A/ ratio regions in percentage of cardiac output as a function of age. Most subjects had less than 0.75% perfusion in such abnormal areas. Only about 10% had more than 1% perfusion, and 5% had more than 1.5%; the maximal value is less than 3%.

Arterial Blood-gas Data

Pa _O2 fell with age as expected, but only slightly (Figure 3). Because of variance, the slope of the regression line was only just significantly negative (p = 0.05, one-tailed test), and indicated a mean decrease of about 6 mm Hg, from 102.3 mm Hg at age 20 yr to 96.5 mm Hg at age 70 yr. However, and unexpectedly, arterial PCO₂ also fell with age, by almost 4 mm Hg, from a mean of 38.0 mm Hg to 34.3 mm Hg over the same age range (p = 0.02). When the alveolar gas equation was used to calculate a Pa_O2 that would have existed had the arterial PCO₂ been 40 mm Hg in every subject, rather than tending to fall with age, a more clear-cut effect of age became apparent (Figure 3, lower left panel ). When this was done, the mean corrected arterial PO₂ fell by 10 mm Hg, from 100 mm Hg at age 20 yr to 90 mm Hg by age 70 yr. It is not apparent why older subjects hyperventilated, but it is evident that this should be considered in interpreting age-dependent changes in Pa_O2. Owing to the very large relative variance, the apparently positive slope of AaPO₂ with age (Figure 3, lower right panel ) was not significant. The variance of AaPO₂ is expected to be relatively large, since AaPO₂ is the generally small difference between two large numbers, alveolar and arterial PO₂. This consequence of normal error propagation also explains the negative values of AaPO₂, which are to be expected on occasion in the normal population. However, despite the large variance in PO₂ and in AaPO₂, the inert-gas data and those for O₂ are internally consistent with each other.

View larger version (22K): [in this window] [in a new window]   Figure 3.   Respiratory-gas data as a function of age, showing measured PaO2 and PCO2 (top panels) and PaO2 corrected in every subject to a normal PCO2 of 40 mm Hg (lower left). AaPO2 differences appear at lower right. PO2 decreases slightly but significantly with age, as does PCO2, PO2 corrected to a PCO2 of 40 mm Hg shows this more clearly. Although AaPO2 appears to increase, the large variance results in lack of significance.

Additionally, the mean value of Pa_O2 for all 64 subjects (100 ± 8 mm Hg) was not significantly different from the Pa_O2 predicted from the measured combination of shunting plus perfusion of low A/ areas as recovered from the inert-gas data (98 ± 11.6 mm Hg). Had there been significant alveolar- capillary diffusion limitation of O₂, and/or bronchial venous or thebesian (postpulmonary) shunts reducing Pa_O2, the measured Pa_O2 would have been systematically lower than the value predicted by the inert gas data. That this was not the case argues against the presence of such phenomena to any measurable degree.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A/ Inequality, O₂ Exchange, and Age: Main Findings

The principal findings of this study are: (1) that A/ inequality, but not intrapulmonary shunting, does indeed increase with age as previously expected; ( 2) that based on our group of 64 subjects aged from 18 to 71 yr, the increase over the span of about 50 yr is physiologically very small; (3) that most of the variance in A/ mismatch over the age range of the group is not due to aging, and remains unexplained; and (4) that the decrease Pa_O2 with age is also quite small but is internally consistent with the A/ changes measured independently.

A/ inequality was characterized in most subjects by a narrow distribution that widened slightly with age, together with a very small shunt of less than 1% of the cardiac output in 90% of cases. Both log SDQ and log SDV increased by about 0.1 between the ages of 20 yr and 70 yr. Thus, log SDQ increased from 0.36 to 0.47, which is consistent with a decrease in Pa _O2 of only about 6 mm Hg. The cause(s) of the increase in A/ mismatch with age were examined as far as possible from the data collected. Only about 10% of the total variance in dispersion was attributable to age. A similar amount was due to intrasubject variability, but none was due to variation in FEV ₁ % predicted, the FEV₁/FVC ratio, weight, or height. It is certainly possible that age could increase A/ inequality as a result of increases in closing volume (3), such that in older subjects some airways are closed during normal tidal breathing, reducing local ventilation and hence producing A/ mismatching. We did not measure closing volume, but since this mechanism is highly unlikely to compromise A/ relationships in young normal subjects, we would argue that closing volume is not a strong candidate for causing the variance in A/ matching. In a very much smaller number of subjects consisting of both young and middle-aged volunteers studied for other reasons, differences in closing volume were apparent as a function of age, but there were no evident effects on A/ mismatching of increasing closing volume by water immersion to the neck (16). When both dry and immersed, the older subjects had greater A/ mismatching than the younger subjects. Taken together, these findings also do not support a role for airway closure as an explanation for age-related A/ inequality.

Even though no cause for the effects of age were identified, approximate upper limits to A/ inequality could be defined across the 50-yr span from age 20 yr to 70 yr. Figure 1 shows that the 95% upper confidence limits found previously for young (20 to 40 yr old) normal subjects (9) can be retained. For log SDQ, this limit is 0.60, and for log SDV it is 0.65. With the 0.1 increase in both parameters by age 70 yr, a reasonable estimate of the 95% upper confidence limit at age 70 yr would be 0.1 units greater for each parameter, at 0.70 for log SDQ and 0.75 for log SDV. These estimates are compatible with the 95% confidence interval (CI) calculated from individual data in Figure 1, although greater precision would require a greater number of subjects, an undertaking that is probably not justified.

The establishment of such upper confidence limits is of clinical utility. Most studies of patients with lung disease are done with older subjects because most of the diseases of interest are more common as age increases.

It was something of a surprise that both Pa_O2 and A/ inequality changed so little with age, especially in light of prior work showing larger decreases in P O₂ with age (17). However, more recent data (18) are more in line with our findings. Whether the differences between earlier and more recent work reflect methodologic differences or differences in subject selection cannot be answered.

In summary, this study of 64 normal subjects aged 18 to 71 yr has confirmed earlier suspicions that A/ inequality increases with age. The effects of age are, however, very small, with dispersion (i.e., log SDQ) increasing on average by only about 0.1, from 0.36 at age 20 yr to 0.47 at age 70 yr. The data fit well with concurrently measured indices of arterial oxygenation, which showed a small decline of only about 6 mm Hg over this age range. There was far more variance in A/ inequality among the subjects than could be explained by age, and candidates for sources of this variance such as experimental errors and differences in spirometric indices, body weight, and height were excluded. The explanation remains to be found. Although previously established 95% upper confidence limits for the A/ dispersion parameters log SDQ and log SDV in young subjects continue to fit the younger subjects in the present group, the increase in A/ mismatching with age sugg ests that these limits of normality be raised for older subjects. Thus, at age 20 yr, the upper limit of reference for log SDQ is 0.60, whereas at age 70 yr it would be 0.70. The upper limits for log SDV are 0.65 at 20 yr and, if raised, 0.75 at 70 yr, and for subjects of intermediate age, linear interpolation between these values is reasonable. These limits should be useful from now on for interpreting A/ dispersion data in older subjects with cardiopulmonary diseases.