Main

During exercise below the ventilatory anaerobic threshold ventilation is closely and linearly related to ˙Vco2 both in children(1) and adults(2). It has been shown that the vigor of the ventilatory response to exercise is correlated with the HCVR in adults(3–5) and in children(6). Marcus et al.(7) recently demonstrated a fall in strength of the HCVR from childhood to adulthood. The ventilatory response to exercise also falls with increasing age throughout childhood(1, 8). This raises the possibility that age-related decrements in HCVR account, in part, for the observed difference in the ventilatory response to exercise between older versus younger children. Because ventilation is inversely related to arterial Pco2, regulation of Paco2 at a lower set-point could also explain the age-related fall in exercise ventilation during childhood(9). Cooper et al.(1) found that end-tidal Pco2 (Petco2) was lower in younger compared with older children during exercise, though another study found no difference in Petco2 between prepubertal boys and adults(6).

Some patients with CF develop hypercapnia with exertion(10) which may have prognostic significance(11). Elevations in Petco2 at peak exercise were associated with earlier mortality in CF by univariate analysis, although this association was no longer significant when one accounted for FEV1 in multivariate analysis. We speculated that these findings could be explained if exertional hypercapnia were determined both by mechanical limitation of the respiratory apparatus caused by advanced lung disease, and reduced ventilatory responsiveness to CO2. That is, a low intrinsic HCVR is a risk factor for development of exertional hypercapnia in patients with CF because such patients have a lower exercise ventilatory response, but elevations in Petco2 do not occur until pulmonary disease limits the ventilatory response to exercise, resulting in hypercapnia. Furthermore, it is conceivable that the age-related changes in HCVR described above can be altered in the presence of pulmonary disease. In this scenario, genetically predetermined hyporesponsiveness to CO2 could be reduced further by advancing pulmonary disease. The aim of the present study was to elaborate on regulation of breathing during exercise in children. We tested the hypothesis that a greater ventilatory response (Δ˙Ve/Δ˙Vco2) during progressive exercise is correlated with a more vigorous HCVR in healthy children and in children with CF. We further postulated that in younger children Paco2 is regulated at lower levels, which requires a greater ventilatory equivalent, in younger compared with older children. Lastly, we sought to determine what influence, if any, evolving obstructive lung disease could have on regulation of breathing both in response to CO2 inhalation and to exercise. We speculated that, despite relatively mild abnormalities in pulmonary function due to CF, the disease alters development of respiratory control during childhood.

METHODS

This study was approved by the University of Manitoba Committee for Research on Human Subjects. Healthy children with normal pulmonary function were recruited from among friends of hospital staff. The CF patients were followed regularly at the Winnipeg Children's Hospital and studied when clinically stable. Diagnosis of CF was based on typical clinical findings and confirmed by repeatedly elevated sweat chloride levels obtained by pilocarpine iontophoresis. Two groups of controls and CF patients participated in the study (Table 1). The first group (part I, n = 16 controls and 16 children with CF) participated in the arm of the study related to HCVR and ventilatory response to progressive exercise. The second group (part II, n = 28 controls and 23 CF children with CF) performed steady state exercise with measurement of Paco2 to assess the relationship between the ventilatory equivalent (˙Ve/˙Vco2) for CO2 during exercise and Paco2. Ten CF patients participated in both parts, but as parts I and II were done many months apart, data were analyzed separately. Patients with CF in part II were shorter (p = 0.008) and lighter (p = 0.005) than control subjects, reflecting chronic disease in this group.

Table 1 Characteristics of subjects (mean and range)
Full size table

Apparatus. ˙Ve was measured with a pneumotachograph on a Jaeger spirometer (Transferscreen II) modified to perform exercise studies. The flow signal was integrated to give tidal volume. For measurement of the HCVR, subjects breathed through a mouthpiece connected directly to the pneumotachograph. During exercise, subjects breathed through a Hans-Rudolph 2600 or 2700 valve connected to their choice of mouthpiece and noseclips, or face mask. Instrument dead space was subtracted from minute volume during exercise. During each minute of progressive exercise, and from 4-6 min of steady state exercise, exhaled gases were collected in large mixing bags. Fractional concentrations of O2 and CO2 in exhaled air were measured with zirconium oxide and infrared analyzers, respectively (Ametek models S-3A, CD-3A), to compute oxygen uptake and carbon dioxide output. They were calibrated with reference gases (room air; 5% CO2, 15% O2, balance N2; and 13% CO2, balance O2) before and after testing. End-tidal Pco2 was measured at the mouthpiece with the same analyzer for 4-6 breaths at the midpoint of each work rate during progressive exercise. Signals were recorded on a strip chart recorder running at 5 mm·s-1 (Gould TA2000). The ventilatory response to progressive exercise was quantitated by Δ˙Ve/Δ˙Vco2 derived from work rates up to the ventilatory anaerobic threshold determined by the V-slope method(12), whereas the ventilatory response to steady state exercise was quantitated by the ventilatory equivalent for CO2 (˙Ve/˙Vco2). Physiologic dead space was calculated from the Bohr equation with Paco2, Peco2, and tidal volume (less valve dead space) measured during steady state exercise.

Measurement of the HCVR(13). Subjects were seated and breathed through a mouthpiece while wearing a noseclip. The bag volume was chosen as the subject's vital capacity plus 10%, and the effluent from the CO2 analyzer was returned to the bag. Attention was paid to ensure the bag was not being emptied at high levels of ventilation. For most subjects, 7% CO2 in O2 as the initial inspirate from the rebreathing bag gave an acceptable plateau in Petco2. Subjects were encouraged to continue rebreathing until they felt they could not continue, or until 4 min had elapsed, whichever came first. Duplicate runs were successfully obtained in 12 out of 16 controls and in 10 out of 16 CF patients. Repeat attempts were unsuccessful either because of unwillingness to continue or because of leaks around the mouthpiece detected on the Petco2 recorder tracing. The mean of the two trials was used in subsequent analysis. Subjects rested to their satisfaction (≅10-20 min) between attempts. Calculation of the HCVR was done as follows(14). The first 45 s of the rebreathing maneuver were not included in the analysis. Rebreathing time beyond this point was divided into 20-s epochs, and the value of Petco2 at the mid-point of each epoch was taken as the Pco2 value. Ventilation was measured over each 20-s epoch, spanning 10 s before and 10 s after the nominal Pco2. In this way we obtained a series of values of Pco2 and matching ˙Ve. The slope of the regression of ˙Ve on Pco2, termed the HCVR, was scaled by normalizing for vital capacity; hence the units were VC·min-1·mm Hg-1(15).

Protocol I. Upon arrival at the laboratory, all subjects underwent routine spirometric measurements (model AT-6, Schiller AG), with results expressed as percent predicted(16). By design, we selected patients with mild or no abnormalities on pulmonary function testing for part I of the study, to avoid the potential constraints that abnormalities of ventilatory mechanics can have in determining the HCVR(17, 18). They next performed a Read's rebreathing test(13) to measure HCVR. After a brief rest, subjects did an incremental exercise test to exhaustion on an electronically braked ergometer (Quinton Excalibur 400), with increments of 8, 16, or 25 W each minute, depending on the subject's height.

Protocol II. Upon arrival at the laboratory, all subjects underwent routine spirometric measurements, with results expressed as percent predicted(16). They then performed a 7-min period of steady state exercise at 0.5 W·kg-1. At the 4-min mark arterialized capillary blood was collected in glass tubes by pricking a warmed fingertip. Blood gases were put on ice and analyzed within 20 min on a Radiometer ABL-500 analyzer. Only blood gases with normal pH and base excess were used in subsequent analysis, to ensure that subjects were exercising at levels below their ventilatory anaerobic threshold.

Statistical analysis. CF patient and control groups were compared with respect to ventilatory variables using unpaired t tests, and regression analysis was used to determine relationships between various parameters. To assess any possible influence of abnormal pulmonary function of the HCVR, we performed univariate regression analysis of HCVR against the spirometric variables, FVC, FEV1, and FEF25-75, as well as multivariate regression using the following model: ventilation =α·age + β·FEF25-75, where age reflects a maturational factor and FEF25-75 reflects a disease factor. Computations were done on a Macintosh LCIII using Statview 4.02 (Abacus Concepts, Inc., 1993).

RESULTS

Baseline spirometry and HCVR. Healthy children all had normal spirometry. Subjects with CF in part I had (mean, range) FVC of 100% (66-122) predicted, FEV1 of 89% (61-124) predicted, and FEF25-75 of 77%(34-136) predicted. Subjects for whom two successful rebreathing runs were obtained (12 controls and 10 CF patients) showed consistent responses-the mean difference between first and second runs averaged 13% of the mean value for HCVR in controls, and 5% in CF patients. Younger children found it difficult to continue beyond 2.5 min, as Petco2 approached 75 mm Hg (10 kPa) by that time. As a result, in these individuals regression of ˙Ve on Pco2 were based on only 4 points, but in no instance was the r2 value for linear regression less than 0.75.

There was a striking relationship between age and the HCVR in healthy controls and in children with CF (Fig. 1). There was no significant relationship between between age and any spirometric variable with univariate analysis in the CF children. However, there was a significant correlation between the strength of the HCVR and FEF25-75 by simple linear regression (r2 = 0.30, p = 0.027), indicating that patients with greater airflow obstruction had less vigorous HCVR. Multiple linear regression of HCVR on two predictors showed a significant effect of both age and FEF25-75 on strength of HCVR(r2 = 0.48, p = 0.014).

Figure 1
figure 1

Plots of the slopes ventilatory response to inhaled CO2 (HCVR) vs age for control and CF boys and girls. Linear regression analysis yielded the following: (controls) y = 0.90 - 0.035x, r2 = 0.77, p < 0.0001; and(CF) y = 0.90 - 0.04x, r2 = 0.29,p = 0.03.

Full size image

Ventilation during progressive exercise and HCVR (part I). In control children there was an inverse relationship between age andΔ˙Ve/Δ˙Vco2 (Fig. 2), and a positive correlation between HCVR and Δe˙V/Δ˙Vco2 during progressive exercise (Fig. 3). In CF children, despite a significant correlation between age and HCVR (Fig. 1), there was no correlation between age andΔ˙Ve/Δ˙Vco2 (Fig. 2) or between HCVR and Δ˙Ve/Δ˙Vco2 (Fig. 3). There were no sex differences in either the ventilatory response to exercise or to hypercapnia. There was no correlation between Petco2 at maximal exercise in either healthy control or CF subjects. On the other hand, there was a significant correlation between the highest Petco2 measured during progressive exercise and the HCVR in healthy children (r2 = 0.455, p = 0.004) but not in children with CF (r2 = 0.04, p = 0.5).

Figure 2
figure 2

Plot of ventilatory response to progressive exercise(Δ˙Ve/Δ˙Vco2) vs age for control and CF boys and girls. The regression equation for control subjects was: y= 34.4 - 0.76x, r2 = 0.43, p = 0.006.

Full size image
Figure 3
figure 3

Plot of ventilatory response to progressive exercise(Δ˙Ve/Δ˙Vco2) vs HCVR (Read's method), in control and CF subjects. The regression equation for control subjects was:y = 16.4 - 19.2x, r2 = 0.45, p= 0.004.

Full size image

Steady state exercise ventilation and Paco2 (part II). Subjects with CF in part II of the study, in whom blood gas sampling was done, had a mean FVC of 95% (66-127) predicted, a mean FEV1 of 90% (63-125) predicted, and FEF25-75 of 82% (32-128) predicted. The relationship between ˙Ve/˙Vco2 during steady state exercise and age mirrored the findings from incremental exercise, with a significant inverse correlation(r2 = 0.337, p = 0.001) in controls, and no correlation in CF children. There was a significant inverse correlation between ˙Ve/˙Vco2 and Paco2 in control(r2 = 0.40, p = 0.0003), but not in CF subjects. When the ventilatory equivalent was expressed in terms of alveolar ventilation(˙Va/˙Vco2), by subtracting physiologic dead space, CF and control children showed similar results. There were significant inverse correlations between age and ˙Va/˙Vco2 in CF patients(r2 = 0.58, p < 0.0001) and in control subjects(r2 = 0.40, p = 0.0003) (Fig. 4). Similarly, there were highly significant inverse correlations between˙Va/˙Vco2 and Paco2 in control children(r2 = 0.68, p < 0.0001) and in CF(r2 = 0.42, p = 0.0006). Dead space was greater in CF than in control children (3.8 versus 3.1 mL·kg-1,p = 0.025). As there were no significant differences between CF and control subjects in terms of age, Paco2, or ˙Va/˙Vco2, they were pooled to perform simple linear regression of Paco2 on age. There was a direct correlation between Paco2 and age in the combined study population (r2 = 0.19, p = 0.001). To assess the relative contributions of age and Pco2 “set point” on the steady state ventilatory response to exercise in children, stepwise logistic regression was done using ˙Va/˙Vco2, as the dependent variable and Paco2 and age as independent variables. Paco2 accounted for 57% of the variance in ˙Va/˙Vco2, whereas addition of age improved the accuracy of the regression, as together they accounted for 71% of the variance in ˙Va/˙Vco2.

Figure 4
figure 4

Plot of ventilatory response to steady state exercise(˙Va/˙Vco2) vs age in healthy children and in children with CF.

Full size image

DISCUSSION

The findings in the present study elaborate on previous work done by different groups of investigators over the last few decades. Godfrey et al.(8) demonstrated that smaller girls had greater ventilation for a given ˙Vo2 than did larger girls, but found no such trend in boys. Cooper et al.(1) found a small but significant decrease in Δ˙Ve/Δ˙Vco2 with increasing age, height, or weight, in a group of healthy children, and that Petco2 was significantly lower in younger children than in teenagers. Gratas-Delamarche et al.(6) concluded that, compared with adults, children had greater CO2 chemosensitivity, particularly during exercise, which could explain their higher ventilatory requirement. These investigators found no difference in Petco2 between children and adults. In any case, Petco2 and Paco2 cannot be equated because the relation between them changes throughout exercise. Most recently, Marcus et al.(7) found a decline in HCVR from childhood to adulthood.

These earlier studies formed the basis for the hypotheses of the present study. We found that children with a more brisk HCVR breathed more for a given metabolic requirement during progressive exercise than did children with a less brisk HCVR, and that there is an age-related decline in the strength of the HCVR. We have shown that the magnitude of the ventilatory response to exercise is linked to Paco2 in that lower Paco2 was associated with higher CO2 ventilatory equivalent in steady state work. Which of these (Paco2 or ˙Va/˙Vco2) is cause and which is effect cannot be determined, but our observations are predictable from the modified alveolar gas equation (κ = a constant and˙Vd/˙Vt is dead space to tidal volume ratio): If one assumes that Paco2 is the independent variable, and ˙Vco2 the dependent variable (determined by work rate and fuel utilization), then ˙Va is simply their quotient. We have demonstrated that there is an age-related decline in ˙Va/˙Vco2 throughout childhood and moreover that Paco2 in mild exercise varies inversely with age. Taken together, it appears that age and Paco2 explain a large measure of the variance in ventilation in children at a given level of exercise. This begs the“chicken and egg” question whether children breathe more because they must to regulate their Paco2 at a lower set point;or is their Paco2 lower because they breathe more (than adolescents) for other, unknown, reasons?

It is tempting to speculate that a lower Paco2 set point is responsible for this observation and for the higher HCVR in younger children. We could not test the latter hypothesis because we sampled blood gases and measured HCVR in different groups of children. Marcus et al.(7) postulated that a relative increase in basal metabolic rate could explain the higher HCVR in children than in adults. In this context it is worth noting that smaller children tended to reach ≥75 mm Hg (10 kPa) Pco2 concentrations after only 2.5-3 min of rebreathing in the present study. Read(13) discussed the theoretical basis of the rebreathing maneuver as it pertains to rate of rise of Pco2 in the bag-lung system, principally bag size and basal metabolic rate. Because we standardized the volume of the rebreathing mixture to the subject's vital capacity plus 10%, the rate of change of Pco2 in the system must have been related to metabolic rate. The rate of rise of Pco2 in the system varied from 5 mm Hg/min (0.67 kPa) in the largest subjects to almost 11 mm Hg (1.47 kPa)/min in the smallest subjects, and was inversely correlated with age and weight (r2 ≈ 0.40). In fact, the HCVR was significantly correlated with the rate of rise of Pco2 in healthy control children (r2 = 0.69,p < 0.0001). This provides inferential evidence that metabolic rate could account for changes in HCVR with age. The rate of rise might also reflect short-term CO2 storage capacity(19), but recent studies suggest CO2 stores do not differ between children and adults(20). It therefore seems reasonable to conclude that both a lower Paco2 and relatively higher basal metabolic rate explain the greater HCVR in childhood.

The normalization of the HCVR according to vital capacity, as opposed to height or weight, was chosen because this manipulation has been shown to halve the range of the HCVR in healthy adults(14). Furthermore, it seems more appropriate to normalize this response to reflect the dimensions of the effector mechanism, i.e. the respiratory apparatus, than to a simple bodily dimension such as height or weight, because the relationship between height and vital capacity is not linear during the growth period(21). This choice is also appropriate for our CF patients because they tended to be physically smaller, and their growth spurts would likely have been delayed, compared with healthy control children. It has previously been shown that the tidal volume response of children with CF in progressive exercise is best correlated with the ratio of residual volume to total lung capacity(22) (which clearly relates to vital capacity). Rebuck et al.(15) found that the magnitude of the tidal volume response has the greatest effect in determining the HCVR. As pulmonary function (particularly FVC) was relatively well preserved in our patients with CF, we believe that biases in normalization were kept to a minimum this way.

Our inclusion of children with CF in this study was aimed at elucidating the reasons for development of exertional hypercapnia in some patients with CF. This question has been studied most extensively by Coates et al.(10, 17). They speculated that, among patients with advanced disease, those who retain CO2 during exercise do so in part because of their innate low ventilatory response to CO2(10). It is recognized that the ventilatory response to exercise is not isocapnic, but that slight increments in Paco2 occur in light exercise(23). In the present study we demonstrated that the ventilatory response to exercise is correlated with the HCVR within individuals. We have also found evidence suggesting an interaction of the disease process on the maturation of ventilatory control in as much as a lower HCVR was associated with greater airways obstruction, independent of age. Lwin and Giammona(18) found that patients with maximum mid-expiratory flow rates ≥70% predicted had HCVR similar to controls, but those with evidence of greater airways obstruction had lower values for HCVR. This supports the hypothesis that a combined effect of premorbid hyporesponsiveness to CO2 and advancing pulmonary disease, superimposed on the normal tendency to allow Paco2 to rise slightly during moderate exercise, results in exertional hypercapnia in CF patients. This could explain the findings of Coates et al.(10) that Paco2 rose early in progressive exercise and remained elevated throughout the test, as severe pulmonary disease precluded the ability to mount an effective ventilatory defense against rising Paco2. As part II of our study revealed, the ventilatory response to exercise is not qualitatively different in a more diverse CF patient population than in healthy controls. One could therefore speculate that given two individuals with CF and moderate to severe airflow limitation, the patient with exertional hypercapnia mounts a less vigorous ventilatory response because he or she has a lower HCVR than the patient who maintains normocapnia or who hyperventilates.

In summary, there is an age-related decline in the ventilatory response to exercise throughout childhood, paralleling the decline in HCVR. Both phenomena are likely due to changing set point of Paco2 regulation, although the possibility of relative increase in metabolic rate in younger (compared with older) children contributing to a more vigorous HCVR remains. CF patients with mild pulmonary disease are not unlike healthy controls in that their HCVR also falls with age during childhood, but how this changing chemosensitivity affects minute ventilation during exercise is more difficult to measure. Qualitatively their ventilatory response to exercise is similar to controls, but quantitatively it is influenced by physiologic dead space. Evolving pulmonary disease exerts an additional influence on chemical control of breathing in CF. A relatively low HCVR explains why some patients experience exertional hypercapnia, whereas others do not despite their having similar degrees of airflow limitation.