INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: exercise test; exertion; reference values; aging

Cardiopulmonary exercise testing (CPET) provides a means of educing evidence of abnormal physiologic functioning which may not be apparent at rest and which may be pathognomonic of particular disease processes (1). Current techniques provide a means of spanning the tolerable work rate (WR) range with a single incremental test of a relatively short duration during which a high-density (e.g., breath-by-breath) computation and display of a range of physiologically relevant variables is available to the investigator for interpretation. The normalcy of response to such a test is usually considered with respect to particular functional indices, such as the peak oxygen uptake (peak O₂), the estimated lactate threshold (_l), and the maximum level achieved for ventilation and heart rate (HR) with respect to some expected limiting value (1). A large body of work has established normal values for these indices with respect to age, sex, body dimensions, and regular level of physical activity (4).

However, consideration of a single value for the variables of interest may be unsuitable as a frame of reference for the continuous, dynamic cardiopulmonary responses that develop throughout these tests. In this regard, the regressed baseline value (intercept or constant) and the rate of change (slope) are likely to be of substantially more importance. In addition, such an analysis maximizes the use of the massive amount of data generated during routine CPET. In fact, the recent European Respiratory Society's monograph on CPET stressed the importance of interpreting the trending of the physiologic response profiles; it also recognized the paucity of the extant information on the topic (1).

The trending of certain variables has been considered as a crucial component of the interpretative strategy. For example, the shift from a linearly increasing profile of oxygen uptake with respect to work rate (O₂/WR) to a more shallow rate of change has been shown to be indicative of circulatory dysfunction (10). Similarly, a steep increase of HR as a function of the metabolic demand could be regarded as indirect evidence of cardiac abnormalities or peripheral muscle impairment for oxygen utilization (HR/O₂) (2, 3, 13, 14). A high slope of the minute ventilation (E) change as a function of pulmonary CO₂ output (E/CO₂) is considered to be reflective of hyperventilation, an enlarged dead space fraction of the breath, or both (13, 15). Furthermore, ventilation could increase at the expense of a tachypneic breathing pattern with a reduced tidal volume (VT) to the ventilatory demand (VT/ E). This pattern will, itself, reduce the efficiency of the lung as gas exchanger, consequent to the high dead space fraction of the breath (2, 3). Little is known, however, about the normal values for these trending phenomena during cycle ergometry in randomly selected subjects. Additionally, in those few studies in which such values have been provided, the cutoff value for assessing normalcy is usually given as a discrete level (2, 4, 5, 14, 19) rather than recognizing that the function may well be age-dependent and sex-dependent (20).

We were therefore interested in establishing appropriate frames of reference for assessing the normalcy of the profiles of the physiologic responses to a rapidly incremental cycle ergometer test. To achieve this, we determined values for the dynamic response profiles of the most commonly used physiologic variables to an incremental exercise test in a randomly selected group of sedentary subjects, both male and female, with an age span of six decades.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design and Subjects

This study used a random sample of ancillary staff (clerical and manual work) from a large university population in a controlled, prospective design. The subjects were chosen randomly by electronic selection from this total population (n = 8,226). A total of 120 individuals (60 men, 60 women) evenly distributed in age groups were evaluated (20 to 39, 40 to 59, 60 to 80 yr) (Table 1) (9, 21). Informed consent (as approved by the institutional medical ethics committee) was obtained from all subjects.

Skinfold thickness was measured at four sites (biceps, triceps, subscapularis, and iliac crest) using a Harpenden skinfold caliper. Body subcutaneous fat and lean body mass were then estimated using the method of Durnin and Womersley (22). The questionnaire of Baecke and coworkers for epidemiologic studies (23) was used to detail and quantify information regarding occupation, sports activities, and leisure habits.

CPET

The exercise tests were carried out on an electromagnetically braked cycle ergometer (CPE 2000; Medical Graphics Corp.-MGC, St. Paul, MN) with gas exchange and ventilatory variables analyzed breath by breath using a computer-based exercise system (MGC-CPX System, MGC), calibrated as previously described (9). Periodically, the overall output data system was validated against a respiratory gas exchange simulator (24). During the exercise tests, the power (W) was increased to the limit of tolerance in a linear "ramp" pattern (10 to 25 W · min¹ in females and 15 to 30 W · min¹ in males). The following variables were determined: pulmonary oxygen uptake (O₂, ml · min¹); pulmonary carbon dioxide output (CO₂, ml · min¹); respiratory exchange ratio (R); E (L · min¹); VT (ml); respiratory rate (f, breaths/ min); ventilatory equivalents for O₂ and CO₂ (E/O₂ and E/CO₂); and end-tidal partial pressures of O₂ and CO₂ (PET_O2 and PET_CO2, mm Hg). The average O₂ for the last 15 s of the ramp was considered to be representative of the subject's peak O₂. The O₂ at the _L was estimated using both gas exchange (25) and ventilatory methods (26).

The following dynamic relationships were determined to characterize the metabolic, cardiovascular, ventilatory, and breathing pattern responses, respectively: (1) O₂/WR (ml · min¹ · W¹), i.e., normal values would indicate adequate metabolic response for a given power output (2, 10, 19); (2) HR/O₂ (beats · min¹ · L · min¹), i.e., a steeper HR response for a given metabolic demand would imply reduced stroke volume or low peripheral oxygen extraction (2, 4, 14); (3) subrespiratory compensation point E/CO₂ (L · min¹ · L · min¹), i.e., high values would indicate "excessive" ventilation to the metabolic stress (2, 15); and (4) VT as a function of the linearized E response (VT/lnE), i.e., shallow slopes would suggest a tachypneic breathing pattern (1) (Figures 1A to 1D).

View larger version (26K): [in this window] [in a new window]   Figure 1.   Procedures used to establish four dynamic indices of exercise function during incremental CPET in representative young (24-yr-old, left panels) and old (70-yr-old, right panels) subjects. (A) A metabolic index ( O2/WR, ml · min1 · W1). (B) A cardiovascular index (HR/ O2, beat · min1 · L · min1). (C ) A ventilatory index ( E/ CO2, L · min1 · L · min1). (D) A breathing pattern index (VT/ln E), derived from the nonlinear relationship between VT and E (inserted graph). These dynamic relationships were obtained by simple linear regression; arrows show the range of values considered for analysis. RCP = respiratory compensation point.

Data Analysis

Data are reported as mean values and standard deviations (SD). Association between variables was assessed by Pearson's linear correlation. Sex-grouped data were compared using Student's t test, and analysis of variance (ANOVA) was used to determine differences among age groups. Multiple linear regression was also performed with the dynamic relationships as dependent variables (27). The probability of a Type I error was established at 0.05 for all tests.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

O₂/WR

Data were pooled only after establishing the response linearity for each individual (Figure 1A). The age-corrected coefficient of variability [CV = (SD/mean) × 100] for this relationship was consistently below 10% in males and 15% in females (Table 2). We found that anthropometric characteristics and age (Table 2 and Figure 2A) did not influence O₂/WR, independent of sex (p > 0.05). On the other hand, males presented higher values than females in all age groups. The lower limit of normal at the 95% confidence limit, therefore, was sex-specific: 9.8 ml · min¹ · W¹ for men and 8.5 ml · min¹ · W¹ for women (Table 2). Interestingly, O₂/WR was positively correlated with both peak O₂ (r = 0.48 and 0.39 in males and females, respectively) and _L (r = 0.32 and 0.29, p < 0.01).

View larger version (32K): [in this window] [in a new window]   Figure 2.   The four submaximal relationships during incremental CPET as a function of age in males (n = 60, left panels) and females (n = 60, right panels). Although age did not influence O2/WR (A), aging was significantly associated with increased HR/ O2 (B) and E/ CO2 (C ) in both sexes. On the other hand, a negative effect of age on VT/ ln E was found in females (D). Females presented higher values of HR/ O2 and E/ CO2 but lower values of O2/WR and VT/ ln E than males. Regression lines are shown with their respective 95% confidence intervals for those relationships in which the variables were influenced by age. Regression coefficients and intercepts of the linear prediction equations are depicted with their respective standard errors. SEE = standard error of the estimate.

HR/ O₂

The CV for the individually determined HR/O₂ throughout the linear range of the responses (Figure 1B) was typically less than 15% in females and 20% in males (Table 2). We found that age and sex significantly influenced this relationship (Table 2), in such a way that aged females manifested the steepest slopes (p < 0.01); the descriptive equations are presented in Figure 2B. In addition, weight (kg) independently influenced this relationship, i.e., HR/O₂ (beats · min¹ · ml · min¹) = 0.42 (0.08) × age  0.53 (0.12) × weight + 73.5 (9.8), r² = 0.41, standard error of the estimate (SEE) = 11.2 in males, and 0.42 (0.09) × age  0.28 (0.10) × weight + 78.1 (10), r² = 0.31, SEE = 12.1 in females. Although body mass index, lean body mass, and the level of regular physical activity were also individually associated with lower HR/O₂ (p < 0.01), these variables did not appreciably improve the residuals dispersion when used instead of body weight. Not surprisingly, HR/O₂ was negatively correlated with peak O₂ (r = 0.44 and 0.47 in males and females, respectively), _L (r = 0.49 and 0.34), and O₂/WR (r = 0.45 and 0.31, p < 0.01).

E/ CO₂

Both age and sex significantly influenced the dynamic E/ CO₂ relationship, as was the case for HR/O₂. The CV for this relationship was within 10% for each age group, independent of sex (Table 2). In addition, E/CO₂ was negatively correlated with height in females (r = 0.31, p < 0.05). However, in a multiple regression analysis, only age remained an independent predictor for this relationship, independent of sex (Figure 2C, males on left, females on right). Interestingly, E/CO₂ was negatively related to both peak O₂ (r = 0.51 and 0.49) and _L (r = 0.35 and 0.30, p < 0.01) in males and females, respectively.

We also sought to characterize the relationship between PET_CO2 (mm Hg) and the metabolic demand (O₂, L/min) from the unloaded control condition to _L, and also from _L to the limit of tolerance (peak _O2). PET_CO2 was lower at each of these three points as a function of age: females tended to present lower values than males at _L (Figure 3, upper panels). The actual values (mean ± SD) at unloaded cycling, at _L, and at peak O₂ were in males: 41.5 ± 3.5, 46.6 ± 2.3, and 37.2 ± 5.3 mm Hg (at 20 to 39 yr); 38.8 ± 1.9, 43.4 ± 2.3, and 36.5 ± 4.9 mm Hg (at 40 to 59 yr); and 36.2 ± 2.5, 40.6 ± 2.8, 34.7 ± 3.5 mm Hg (at 60 to 80 yr). For the female group, these values were as follows: 40.5 ± 2.7, 43.5 ± 1.3, and 36.1 ± 3.7 mm Hg (at 20 to 39 yr); 39.5 ± 2.7, 42.7 ± 1.8, and 35.9 ± 3.3 mm Hg (at 40 to 59 yr); and 37.2 ± 1.6, 40.4 ± 2.1, 34.7 ± 3.0 mm Hg (at 60 to 80 yr) (Figure 3, upper panels). On the other hand, there was no significant effect of age on unloaded _L PET_CO2/O₂; males, however, did present significantly lower values of this relationship than females (9.30 ± 3.9 and 14.1 ± 4.5 mm Hg · L¹, respectively; p < 0.05) (Figure 3, lower panels). The negative PET_CO2/O₂ relationship between _L and peak O₂, however, did increase with age in both sexes, with females also presenting higher values than males, i.e., males = 6.6 ± 2.9, 6.9 ± 3.1, and 8.2 ± 3.9 mm Hg · L¹; females = 9.8 ± 3.5, 13.7 ± 4.1, and 17.1 ± 5.9 mm Hg · L¹ for the 20 to 39, 40 to 59, and 60 to 80 age groups, respectively (Figure 3, lower panels).

View larger version (25K): [in this window] [in a new window]   Figure 3.   PETCO2 in males (left panels) and females (right panels), age 20 to 39 (squares), 40 to 59 (circles), and 60 to 80 (triangles), respectively. The upper panels depict the values at the unloaded condition (UNLOAD), at the estimated lactate threshold (LT), and at peak O2 (PEAK), regardless of the metabolic demand. There was an inverse relationship between this variable and age, independent of sex. The lower panels show these values expressed as a function of the actual metabolic demand ( O2) at the same points (i.e., UNLOAD, LT, and PEAK). Note that steeper PETCO2/ O2 relationships, either before or after LT, were found in females and older (60- to 80-yr-old group) subjects.

VT/ln E

The CV for the individually determined VT/lnE (Figure 1D) was near 20% for each age group in both sexes (Table 2). We found that sex was a significant determinant in this relationship, with males presenting higher values than females in all age groups. In addition, we found a negative relation between VT/lnE and age only in females (Table 2, Figure 2D) but a positive association with height (r = 0.40, p < 0.05). As with E/CO₂, however, only age remained an independent predictor of VT/lnE. Figure 2D depicts the age-corrected prediction equation for this relationship in females. Furthermore, VT/lnE was negatively related to E/CO₂ in both sexes (r = 0.33 and r = 0.25, in males and females, respectively).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study presents a systematic evaluation of selected dynamic submaximal relationships for rapidly incremental cycle ergometry in a randomly selected sample of sedentary males and females, up to 80 yr of age, providing age- and sex-specific indices of metabolic (O₂/WR), cardiovascular (HR/O₂), and ventilatory (E/CO₂ and VT/lnE) function (Table 2, Figures 1 and 2). These normative data therefore provide a frame of reference for the normalcy of the submaximal responses during clinical incremental CPET for use in conjunction with the readily available discrete reference values.

O₂/WR

The linear phase of the O₂/WR relationship during rapidly incremental exercise has been demonstrated to be a useful noninvasive index of aerobic work efficiency in normal subjects (2, 10). In several patient groups, however, this index is lower (2, 10, 12, 19), indicating increased energetic contribution from anaerobic sources of ATP regeneration. In addition, the linearity of the O₂-WR relationship can be lost in patients with cardiac impairment, for example, because of decreased oxygen flow to the exercising muscles. As expected, this is more likely to occur above the anaerobic (lactic) threshold (10, 12).

Previous normative values for the O₂/WR relationship were obtained in volunteers, typically involving higher-than-average fit males with narrow ranges of age (10, 19). Although the average values obtained in women and aged men in this study are not appreciably different from those published by Hansen and coworkers (10, 19), our younger sedentary men clearly presented higher valuescomparable, however, with those reported by others (11). In addition, we confirm that although overweight does not change the response slope, it does displace this relationship upwards (data not shown). In reality, we recently demonstrated that this should be better related to leg than total body mass during cycle ergometric exercise (28).

Interestingly, this ratio has long been considered to be independent of age, sex or physical fitness (2); however, it would be expected that the less fit subjects (who would rely more on anaerobic energetic sources) would present lower slopes, as was the case in our study for female subjects (Table 2). Additionally, the fitness dependence of O₂/WR is consistent with the observed positive correlation between O₂/WR and the level of regular physical activity (p < 0.01) and peak O₂, and the inverse relationship between O₂/WR and HR/O₂ in both sexes.

HR/ O₂

The HR/O₂ might also be considered to represent a useful index of overall "cardiovascular fitness" because reductions on stroke volume and peripheral oxygen extraction are both expected to steepen this relationship, assuming the normal independence of training on the linear O₂-cardiac output relationship (2, 13, 14). This is particularly true in this study where the use of rigorous exclusion criteria allowed us to avoid other confounding factors such as anemia, carboxyhemoglobinemia, hypoxemia, or clinically significant shunts (9, 21). It is not surprising therefore that the prediction equations developed by Fairbarn and coworkers (14), who evaluated a group of volunteers, significantly underestimated the HR/O₂ slope of our randomly selected subjects (p < 0.01, comparison not shown). The inverse relationships between HR/O₂ and weight and lean body mass in both sexes are likely to be related to the well-known stroke volume-body mass relationship, a "training" effect of chronic overweight, and to the underlying relationship between lean body mass and regular physical activity.

E/ CO₂ Relationship

Although it is well known that the ventilatory response to muscular exercise as a function of metabolic rate increases with age (29), it is less known how it changes with aging and also with respect to rapidly incremental tests now common to clinical exercise testing. There is therefore a paucity of appropriate reference values for the E/CO₂ relationship; the few available studies also used somewhat fitter volunteers (20, 30).

Normal values for the slope of increase in E as a function of CO₂ (E/CO₂) during exercise are presented, as this approach is commonly used in the assessment of "abnormal" ventilatory response to exercise. However, this slope should be used with caution, as it is not the constant, but the variable E/CO₂ that is important in establishing arterial blood gas and acid-base status, i.e., E = mCO₂ + c, where m is the slope and c the intercept. Rearrangement of this equation yields E/ CO₂ = m + c/CO₂. It should be noted, however, that physiologically

E (BTPS) = 863 CO₂ (STPD)/Pa_CO2(1-VD/VT)

where VD = dead space ventilation. For a linear response, therefore, E/CO₂ at the lactate threshold will closely approximate E/CO₂ when "isocapnic buffering" begins at a high value of CO₂ but will be higher if it begins at low CO₂. In fact, our previously reported values of E/CO₂ at _L (in this population) (9) were higher than the E/CO₂ reported here in women and older subjects, i.e., the less fit subjects. This, we believe, is an important distinction.

In this context, it should be emphasized that CO₂and not O₂ (4, 5, 31)is the more appropriate independent variable in this relationship. In fact, the CV for the individually determined E/O₂ values (15 to 20%) in the present study was substantially higher than E/CO₂ (5 to 10%), independent of sex and age. As was the case in the studies of Poulin and coworkers (30) and Habedank and coworkers (20), we found that age presented a more definitive influence in reducing the ventilatory efficiency in men than women. Whether this relates to a steeper increase in the physiologic dead space or reduced CO₂ set-point in men remains to be determined. In fact, we found that women and older subjects tended to present lower sub-_L PET_CO2 values than men and younger subjects, respectively (see RESULTS). Importantly, however, age and female sex were related to a more tachypneic breathing patterna well-known negative influence on the end-tidal values (Figure 2D).

VT/ ln E Relationship

With respect to the breathing pattern, E seems to change as an effectively linear function of VT up to a critical value VT which has been shown to be related to an age-specific and height-specific fraction of the resting vital capacity (50 to 60%) (7, 8) or, more properly, the inspiratory capacity (6) (up to 85% in this population) (9). Hey and coworkers (32) termed this phase as "range 1" with the respiratory rate (f) contribution depending on a positive intercept in the E-VT relationship (Figure 1D, insert). In the "range 2," the E-VT relation is steeper as a result of the more dominant influence of f: the asymptote value of VT is thought to be linked to elastic work of breathing and a critical lung volume threshold for vagally mediated mechanoreception (32). We needed therefore to linearize the relationship over the entire work rate range before applying regression analysis: as shown in Figure 2D, females did present significantly shallow slopes (i.e., a more tachypneic breathing pattern). This relationship is likely to be useful in assessing malingering or subjectively mediated changes in breathing pattern during CPET (33), in addition to being reflective of thoracic mechanical function.

In summary, this study constitutes, we believe, the first characterization of reference values for certain widely recommended (1) submaximal indices of metabolic, cardiovascular, and ventilatory function for clinical exercise testing interpretation using incremental cycle ergometry and a randomly selected sample of adults up to 80 yr of age. Our results demonstrate that sex, age, and anthropometric characteristics should be considered in the assessment of the normalcy of these dynamic exercise responses. The use of a single cutoff value, therefore, may mislead the normalcy judgment of system functioning during exercise.