Individuality of breathing patterns during hypoxia and exercise J. H. EISELE, B. WUYAM, G. SAVOUREY, J. ETERRADOSSI, J. H. BITTEL, AND G. BENCHETRIT Laboratoire de Physiologie, Faculte’ de Mkdecine de Grenoble, and Centre de Recherches du Service de Sante’ des Armkes, Grenoble, France EISELE, J. H., B. WUYAM, G. SAVOUREY,J. ETERRADOSSI, J. H. BITTEL, ANDG. BENCHETRIT.Individuality of breathing patterns during hypoxia and exercise. J. Appl. Physiol. 72(6): 2446-2453, 1992.-Breathing was recorded via a pulsed ultrasonic flowmeter in 11 healthy subjects, at rest and during steady-state exercise (at 50% of their maximal 0, consumption) at both sea level (200 m) and simulated altitude (4,500 m in a hypobaric chamber). The pattern of breathing was quantified breath by breath in terms of classical respiratory variables (tidal volume and inspiratory and expiratory times), and the shape of the entire airflow profile was quantified by harmonic analysis. Statistical tests were used to compare the within-individual with the between-individual variations. In comparing the sea level vs. altitude rest (16% increase in ventilation) and sea level vs. altitude exercise (40% increase in ventilation) airflow profiles, we found a significantly greater resemblance within the individual than between individuals. Comparisons of sea level rest and exercise (295% increase in ventilation) and altitude rest and exercise (375% increase in ventilation) revealed no similarity within individuals. Despite airflow profile changes between rest and exercise, it is still possible to attest to a diversity of flow profile between individuals during exercise. Hypoxia at rest or during exercise does not alter the phenomenon of the individuality of breathing patterns. control of breathing; breathing pattern
human;
airflow
profile;
reproducibility
of
THE SHAPE OFRESPIRATORYCYCLES derived from flow signals reveals characteristic, individual patterns during quiet breathing (2, 5, 7, 12, 13). Previous studies have shown that these so-called “breath prints” remain unchanged over time in adult human subjects (3) and are very similar for identical twins (14). These studies were performed on resting subjects, and the extent to which the individuality is maintained when ventilation is increased remained to be elucidated. Proctor and Hardy (13) reported a triangular or a quasi-rectangular pattern of the maximum effort pneumotachograms. Bradley and Crawford (4) concluded that the airflow pattern varied considerably between subjects but was fairly constant for any one subject even during exercise. Several studies (5, 10, 11,16) that aimed to obtain the criterion of optimization of the airflow shape reached the conclusion that a rectangular airflow pattern was one of the most economical for most of the criteria functions utilized. When the observed patterns of breathing are considered, only those recorded during exercise are of the rectangular type, and 2446
0161-7567192 $2.00 Copyright
the criteria retained cannot explain individual variations in breathing pattern. Furthermore, these criteria were chosen to describe a general optimal pattern, which could then be selected by a physiological control system for changes in ventilation. We began with the assumption an individuality of breathing pattern exists at rest and designed the present study to determine the individual airflow profiles at different levels of ventilation. Hence breathing was recorded for subjects at rest at sea level (200 m), exposed to a simulated altitude of 4,500 m, and during exercise at sea level and at 4,500 m. The aim of the study was to investigate whether the individuality of breathing pattern is conserved when ventilation is increased. METHODS Subjects. Eleven healthy male rescue workers (aged 23-52 yr) participated in the study. All resided close to sea level and had some experience in mountaineering. A few days before the study they underwent pulmonary function tests and a maximal exercise test on a bicycle ergometer to determine their maximal 0, consumption (VO 2,,,). Their physical characteristics and pulmonary function data are given in Table 1. Protocol and measurement. The study was carried out on each subject between 8 and 11 A.M. The first recording was performed in Grenoble (200 m altitude) in the seated position in a comfortable chair for - 15 min. This recording has been labeled Sl rest. It was followed by 5-10 min of exercise on a bicycle ergometer set for a work load equivalent to 50% of the subject’s VO~,~, as determined before the study (Sl exercise). Within 0.5 h the subjects (two at a time) entered a hypobaric chamber at a simulated altitude of 4,500 m; the pressure in the chamber was 443 Torr [equivalent inspired 0, fraction (FI,J = 0.12 at barometric pressure (PB) of 760 Torr]. The chamber had a volume of 31 m3, the air within it was continuously renewed, and its temperature was maintained at the ambient outside temperature. The duration of hypoxic exposure before the beginning of the recording was between 30 min and I h in all subjects. The same sequence of recordings, rest (Al rest) and exercise (Al exercise) at 50% of sea level VO, max,was then performed. Airflow was measured with a pulsed ultrasonic flowmeter (Branta Biometrics) connected to a face mask (Hans Rudolph). 0, and CO, were measured by mass
G3 1992 the American
Physiological
Society
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
AIRFLOW
TABLE
1. Anthropometric
PATTERN
DURING
HYPOXIA
AND
2447
EXERCISE
data, exercise performance, and respiratory function of the subjects vc
Subj
Yr
Height, m
AP BP DP LE LA LT MD PM SG TJC BV
23 29 33 31 52 29 37 44 29 40 32
1.69 1.80 1.78 1.83 1.68 1.75 1.62 1.74 1.75 1.86 1.82
Age,
vo
2max9
maximal
0, consumption
Weight, kg
64.0 73.3 73.0 80.2 76.0 66.0 59.0 70.0 77.3 84.5 86.5 determined
VO
ml
l
kg-’
l
min-’
liters
STPD
58.5 50.7 62.0 43.4 50.6 53.0 64.2 51.6 59.6 49.8 42.0 at sea level;
FEV,
2max7
VC, vital
spectroscopy (Mediflex SX 200, VG Instrument) of a gas sample from the mask, with the signals displayed continuously on an oscilloscope. In addition, arterial 0, saturation (Sao,) was measured continuously with a finger pulse oximeter (Lifestat 1600, PhysioControl). Five minutes were allowed to elapse for a steady state of ventilation to be achieved at rest or during exercise. Then 2100 breaths were recorded on magnetic tape, along with 0, and CO, fraction of the gas sampled from the mask. Analysis of recordings. For each subject all recorded breaths were digitized at a frequency that gave >lOO points per breath. Breath-by-breath analysis of respiratory cycles was performed off-line. For each condition, 50-100 cycles were examined, and abnormal breaths, such as accompanying swallowing and the three breaths after such disturbances, were discarded. A quantitative analysis of the shape of the airflow signal for each breath was performed using a harmonic analysis (fast-Fourier transform of airflow signal normalized for breath duration) as previously described for airflow signals recorded with a Fleisch head (1). In agreement with that study, we confirmed that, for each breath, the first four harmonics retained >95% of the information about the shape of the breath, whatever the condition of recording (rest, altitude, or exercise). The vectorial representation of these four harmonics termed ASTER, and the Cartesian coordinates of the vectors provided eight figures that describe quantitatively the shape of the airflow. After the breathby-breath analysis, a mean ASTER was calculated for each recording by averaging each coordinate across all breaths in the data set of this recording. In addition to the multivariate ASTER, the following variables were measured for each breath: inspiratory time (TI), expiratory time (TE), total breath duration ( TT), tidal volume (VT), and the time indexes (VT/TI and TUTI, TE, and VT were taken together as a trivariate description of each breath’s volume shape, which was termed a TRIAD as previously described (3). Statist ical analysis. For each recording, the means t SD were calculated fo r each variable and for the multivariate ASTER and TRIAD. The analysis used was designed to test whether an individual’s “respiratory personality” is maintained during different conditions (e.g., hypoxia, exercise) relative to the differences occurring between different individuals in the same group (11 subjects) between the same conditions. Such an analysis al-
%pred
BTPS
5.77 5.64 5.99 5.54 4.99 4.92 5.08 5.83 5.15 5.80 5.41 capacity;
FEV,,
forced
liters
4.44 5.28 5.15 4.70 4.11 3.57 3.95 4.67 4.15 4.67 4.40
114
100 112 99 120 91 120 122 97 105 102 expiratory
volume
BTPS
%pred
107 118 121 105 127 84 116 124 98 108 100
in 1 s.
lows for differences within an individual under the different conditions but implies that the individual has a certain breathing personality that can be distinguished from that of other individuals. For example, the statistical analysis was performed by comparing differences between the Sl rest and Al rest recordings within individuals with those differences observed between random pairs of recordings in these two conditions from the same 11 subjects. The differences between recordings were expressed in terms of distance. For a single variable this distance is calculated as the square of the difference between the means divided by the pooled variance. For the ASTERS and TRIADS it was necessary to employ a multivariate determination of this distance in which the covariances between the different variables were also taken into account. The null hypothesis of the test is that the sum of the differences within individuals is not different from the sum of the differences between pairs of individuals taken randomly; the alternative hypothesis is that the sum of the differences within individuals is significantly less than that between a random pair of individuals. A full description of the test has been given previously by Benchetrit et al. (3) and can be applied to single respiratory variables (e.g., VT, TI, TE) or to the multivariate ASTER and TRIAD. When we compared the airflow profiles between two conditions, we normalized the ASTERS for amplitude; i.e., each coordinate was divided by the square root of the sum of the square of all eight coordinates. Thus, for these comparisons, each breath was normalized for amplitude and breath duration (harmonic analysis). This allows the comparisons to refer to the shape only. A paired t test was employed to compare ventilation in the different conditions of recording. RESULTS
The subjects were healthy volunteers with normal respiratory functions (Table 1). Chamber hypoxia at a simulated altitude of 4,500 m was well tolerated by all subjects, who were able to complete the exercise (for a work load equivalent to 50% of sea level Vozmax) during hypoxia without clinical evidence of any ill effects. SL rest US.Al rest. The degree of hypoxia at simulated altitude was shown by the fall in mean SaoQfrom 97 t 1 to
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
2448
AIRFLOW
PATTERN
DURING
2. Minute ventilation at rest and during exercise at sea level and altitude TABLE
Rest
Exercise
Subj
Sl
Al
Sl
Al
AR BP LIP LA LE LT MI) PM
8.72
7.86
9.09
8.59
8.16
8.91
38.71 50.93 51.43 22.06 41.69 39.39 41.11 50.28 52.45
43.91 I
8.01 6.44 6.18 8.66 8.49 7.32 7.48
9.99 8.07 7.24 12.78 8.50 10.73 9.33
22.83 27.81 36.23 15.83 33.73 28.86 27.55 42.74 40.00 39.60 32.04
7.97 I
9.24 I
31.57 I
9.08
SG
TJ VB Mean
9.60
P = 0.02
51.00
P < 0.005
Minute ventilation (\iE) values are in l/min ATP at sea level (Sl) and altitude (Al). Exercise intensity for each subj at 50% VO, max was measured at Sl. * Significant increase in VE at Al, by paired t test on 10 subjs.
84 t 3% at 4,500 m. Ventilation
at rest increased during hypoxic exposure in eight subjects, remained almost unchanged in one subject, and decreased slightly in two subjects. The average increase in ventilation was 15.9% (Table 2); a paired t test performed on the mean values of ventilation calculated over the analyzed breaths for the 11 subjects shows a significant increase (P = 0.02). Table 3 lists the values of VT, TI, and TE in all subjects for the four different conditions of recording. The increase in ventilation from Sl rest to Al rest was achieved mainly by an increase in VT (7 of 11 subjects). The airflow analysis is illustrated in Fig. 1 for subject DP for each condition of recording: the mean ASTER (normalized in amplitude), the airflow profile reconstructed from this ASTER, the airflow profile with its actual duration, and finally the TRIAD drawn to show the actual mean values of VT, TI, and TE. In Fig. 1, the resemblance can be seen between the ASTERS and reconstructed flows in the two conditions of resting ventilation at sea level and altitude, despite changes in the TRIADS. The reconstructed flows for all 11 subjects are shown TABLE
HYPOXIA
AND
EXERCISE
in Fig. 2. The following observations can be made. 1) When airflow at Sl rest for each subject is considered, the difference between subjects represents the individuality of breathing that has previously been reported (2, 3). 2) When airflow at Al rest is considered, the variations among the subjects suggest that there is still an individuality of breathing under hypoxia. 3) The resemblance between the two flows at rest for each subject suggests that the individuality of breathing is maintained during hypoxia. The statistical analysis performed to test the similarity of the ASTERS indicates that the differences within individuals are significantly less than those between individuals. The same test performed on the other ventilatory variables and the trivariate TRIAD indicates that the similarity between the breathing patterns is best evidenced when the multivariate ASTER is used; the P values of these tests are given in Table 4. Hence, the airflow shape appears to be the breathing pattern characteristic that is most conserved when ventilation is changed in subjects exposed to a simulated altitude of 4,500 m. Sl rest vs. Sl exercise. During exercise at a work load of 50% vo, max, ventilation increased in all subjects (Table 2). This augmentation in ventilation was achieved by an increase in VT and decreases in TI and TE (Table 3). The changes in airflow brought about by exercise are illustrated for subject DP in Fig. 1. The ASTER and the reconstituted flow profile (normalized for breath amplitude and duration) of this subject are quite different between Sl exercise and Sl rest, corresponding with the more rectangular shape of inspiration during exercise. For subject DP, despite the decrease in TE, little change can be seen in the shape of expiration. In Fig. 2, for each subject the flow profile shows a general change in shape from Sl rest to Sl exercise. There is a more rectangular shape of the inspiration in almost all subjects, but (except for subject DP, Fig. 1) there are changes also in the shape of expiration in most of the subjects. However, the flows of the different subjects at Sl exercise are not all the same, and there still appears to be an individuality of breathing pattern, although different from that at rest. Indeed, the statistical comparison of the similarity of the ASTERS between rest and exercise at sea level showed that the within-individual difference was not less than
3. VT, TI, and TE at rest and during exercise at sea level and altitude Rest
Exercise
Sea T,evel Subj
AR RP nr LA LE LT MD PM SG TJ VB
VT, ml ATP
420+60 506+50 507k47 484k60 506+35 338-t75 581k133 6252218 611+121 433231 540+107
VT, tidal volume;
Altitude
TI, s
TE, s
1.22kO.10 1.40k0.12 1.38kO.18 1.41kO.12 1.58kO.14 1.29kO.26 2.4OkO.55 1.6920.27 1.8OkO.71 1.44kO.11 1.96kO.40
1.66kO.21 1.93L0.18 2.36kO.48 1.79kO.32 2.21kO.20 1.87kO.34 3.24kO.80 2.65kO.43 2.52kO.98 2.11kO.29 2.3720.51
TI, inspiratory
time;
VT, ml ATP
491k164 500225 6OOk56 5Olk95 666+123 55Ok97 455+133 954+210 587-t91 665-t94 714+142
TE, expiratory
Sea Level
TI, s
TE, s
1.51kO.40 1.26kO.34 1.47t0.21 1.49kO.19 1.81k0.43 1.83kO.43 1.6OkO.45 1.63kO.28 1.55kO.31 1.54kO.11 2.06t0.37
2.25k0.89 2.24kO.80 2.57kO.42 1.65kO.29 2.18k0.32 2.26kO.31 2.17kO.40 2.85k0.54 2.6OkO.53 2.18kO.39 2.52kO.39
VT , ml ATP
956k214 1,140+201 1,280*208 550+109 1,220+67 1,010+135 1,180+212 2,030*187 1,180+130 1,540&117 1,180+126
Altitude
TI, s
TE, s
1.10+0.34 1.06kO.19 0.87kO.20 0.98-tO.11 0.93kO.17 0.9720.12 1.09kO.22 1.43k0.24 0.81+0.14 1.15LO.16 1.11+0.12
1.4 120.62 1.4O-tO.24 1.2520.25 1.1220.12 1.24k0.17 1.14kO.12 1.48k0.28 1.42k0.23 0.96kO.14 1.18t0.17 1.10+0.14
VT, ml ATP
TI, s
TE, s
1,200+176 1,460+156 1,560+210 739+ 150 1,480-+86 1,300+95 788-t-95 2,430?166 1,530+100
0.80+0.16 0.79kO.21 0.7920.15 0.99&O. 19 0.92&O. 13 0.86kO.10 0.50+0.06 1.35kO.15 0.82kO.09
1.05LO.24 0.9320.18 1.03t0.22 1.02kO.16 1.22k0.17 1.12kO.12 0.65kO.06 1.5520.23 0.94+0.10
1,700+88
1.03kO.09
0.97&O. 13
time.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
AIRFLOW
PATTERN
DURING
HYPOXIA
2449
AND EXERCISE 1 sec.
Subject
DP
“I
500 ml
0
2
4
sec.
6
A
VT
TI
n
b
Altitude Rest (72 Breaths)
0 b
;
2
6 sec.
A VT TI
n
Sea Level Exercise (56 Breaths)
Altitude Exercise (65 Breaths)
TI
n
1. Pattern of breathing in 4 different conditions of recording in subj DP in each condition. Left to right: vectorial representation of 4 harmonics in breath-by-breath analysis (ASTER, calculated over no. of breaths in parentheses), airflow profile reconstructed from ASTER, ASTER represented with its actual duration, and each breath’s volume shape (TRIAD). Reconstructed airflow from ASTER is normalized (see METHODS) for breath duration and amplitude and is the best representation of breath shape. Changes in ventilation in different conditions can be judged from TRIADS. FIG.
the between-individual difference (P = 0.57). This indicates that, in the group of 11 subjects, it was not possible to match the airflow profile of each subject at rest and exercise, indicating that the shape of the airflow was significantly modified by exercise. Al rest us. Al exercise. The mean Sa,, levels at altitude fell to 84 t 3% at rest and decreased further to 77 t 5% (P < 0.0001) during exercise. The average increase in ventilation from rest to exercise at altitude was from
9.24 t 1.54 to 43.91 t 9.54 l/min (Table 2). Mean values for VT, TI, and TE are given in Table 3. The changes in the airflow profile (Figs. 1 and 2) parallel those at sea level, with a tendency toward a more rectangular shape of the inspiration during exercise. The same conclusions were inferred from the statistical analysis, which tested for the preservation of the individuality between rest and exercise at altitude (P = 0.98). Hence, the same individuality is not preserved, inasmuch as the shape of the air-
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
2450
AIRFLOW
PATTERN
DURING
HYPOXIA
DP
AR
LE
AND EXERCISE
/\
Sf Rest Al Rest
Sl Exer
Al Exer
LT
BP Sl Resz Al Rest z Sl Exer
Al Exer z PM
SG
Sl Rest
FIG. 2. Reconstructed airflow profiles normalized for breath duration and breath amplitude (see METHODS) are represented in 4 different conditions of recording in all subjects. Sl, sea level; Al, altitude; Exer, exercise. Note diversity of breathing pattern between individuals at rest and during exercise and similarity for each subject between the 2 recordings at rest and those during exercise.
Al Rest
Sl Exer
Al Exer
VB
TJ
Sl Rest Al Rest r Sl Exer
\
1
Al Exer
flow profile is significantly modified between Al rest and Al exercise. SI exercise us. Al exercise. The mean Sa,, levels at altitude fell from 84 to 77% with exercise, whereas there was no change in Sa,, levels during exercise at sea level. There was a significant increase in ventilation between Sl exercise and Al exercise for the same work load (Table 2): ventilation was 42.7% higher for exercise at altitude. The increase in ventilation between the two conditions was caused by an increase in VT and a decrease in TI and TE (Table 3). The airflow profile changes between the two conditions of exercise are illustrated in Fig. 1 for one subject. The ASTERS appear slightly different: all the vectors are closer to the vertical axis, perhaps due to the higher TI/ TE in Al exercise (0.77) than in Sl exercise (0.69). How-
ever, the four vectors occupy the same relative position. This resemblance among airflow shapes is more striking in the reconstructed flow profiles, even when they are represented with their actual duration. The TRIADS show the changes in volume and timing of the breaths between these two conditions. The resemblance is worth noting (Fig. 1) between the two airflow profiles at rest, between those during exercise, and between rest and exercise. This observation can be made for all subjects in Fig. 2. For each subject there is a resemblance between the two flow profiles at rest and during exercise. In addition, as at rest, the flow profiles during exercise differ among the subjects. We tested the similarity of the respiratory variables between Sl exercise and Al exercise. The results are given in Table 4; when ASTERS are considered, there is signifi-
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
AIRFLOW
PATTERN
DURING
4. statistical comparisons of similarity of respiratory variables at sea level and altitude in conditions of rest and exercise TABLE
Sl Rest vs. Al Rest
ASTER TRIAD VT TT TI TE TI/TT VT/TI
(n = 11)
Sl Exercise vs. Al Exercise (n = 10)
P < 0.00025 P = 0.046 P = 0.514 P = 0.011 P = 0.089 P = 0.013 P = 0.107 P = 0.680
P < 0.00025 P = 0.120 P = 0.174 P = 0.083 P = 0.015 P = 0.043 P = 0.015 P = 0.300
ASTER, vectorial representation of 4 harmonics in breath-bybreath analysis; TRIAD, trivariate description of each breath’s volume shape based on VT, TI, and TE; TT, total breath duration; TI/TT, inspiratory duty cycle; VT/TI, mean inspiratory flow. At 0.01 level of significance, only multivariate ASTER shows that there is significantly less difference within individuals than between individuals (see METHODS).
cantly less difference within than among individuals, indicating that, in these two conditions of exercise, it is possible to match the two airflow profiles of each subject. DISCUSSION
The major finding of our study was that the airflow shape changed from rest to exercise. However, not all the subjects adopt the same pattern, and it is still possible to observe a diversity in flow patterns among individuals during exercise. In addition, hypoxia at rest or during exercise does not significantly change the individual flow pattern. Recording methods and signal analysis. Ventilation was measured with a calibrated ultrasonic flowmeter mounted on a face mask. Although this system could have modified the pattern of breathing, introducing an additional dead space, any effect on the shape of the airflow has been kept constant. The shape analysis used (1) was based on a Fourier analysis of cycles normalized for breath duration. The sum of the first four harmonics was found to retain >95% (mean value 98%) of the information about the shape for each breath in any of the four conditions of recording. Lafortuna et al. (11) used a similar analysis on the inspiratory flow profile alone; harmonic analysis was performed on inspiration only, with seven harmonics characterizing the shape of the flow signal. However, as in our study, the index they utilized to estimate the appropriateness of their analysis is clearly >95% when only four harmonics are used (whatever the level of ventilation at rest or during exercise, as in our study). This justifies the use of the four harmonics (8 variables) to describe the shape of the breaths and to perform the statistical analyses. However, when the flow signal with four harmonics is reconstructed, the parts of the breath with a constant flow (flat regions) will appear slightly bumpy, whereas the steep parts of the shapes, such as rapid changes in the flow, are very precisely described. In the harmonic analysis, breaths are normalized for duration. For the illustrations we also normalized the mean ASTERS for amplitude to eliminate the effect of
HYPOXIA
AND EXERCISE
2451
changes from differences in VT in the four conditions of recording and therefore to focus only on the shape. In addition, we normalized each breath for amplitude for statistical comparisons between conditions. Lafortuna et al. (11) also used normalizations in amplitude to 1) compare shapes at different levels of ventilation and 2) add shapes for different subjects. Normalization in breath duration and amplitude is one such dimensionless transformation that was suggested by Gray and Grodins (9) as the first step in analyzing the significance of the shape of flow profiles. Ventilation during hypoxia and exercise. The level of hypoxia was chosen to provide 1) a hypoxic stimulus at rest and 2) an environment compatible with an exercise at 50% of sea level ‘ire, M8Xfor all subjects, given their physical fitness. Indeed, all subjects completed exercise at this altitude without exhaustion or fatigue. Exercise data for one subject (Table 3) were missing because of technical problems during the off-line analysis of the recording. At 4,500 m, where resting Sa,, fell to 84%, the mean increase in ventilation was 15.9% at rest, with an interindividual variability in the response to the hypoxic exposure (Table 2). The mean increase in ventilation observed is comparable with the 19% increase in ventilation observed by Easton et al. (8) in 20 young adults exposed to moderate hypoxia (Sa,, of 80%). In that study, after an immediate initial rise in ventilation, a secondary “plateau” of ventilation (at 119% of the control ventilation) was maintained over 21 h. Our recordings were performed between 30 min and 2 h after altitude exposure, when we considered that a steady state of ventilation was reached. In fact, we performed all recordings in a steady state so that we could calculate a mean cycle representative of the ventilation at the environmental condition under study. Ventilation during exercise at sea level was 31.75 l/min and was further increased (by 40%) to 43.91 l/min under the effect of hypoxia. The higher level of ventilation for the same work load during exercise at altitude than at sea level has already been reported (7). This increase may partly be a response to a further fall in Sa,, during exercise from 84 to 77%. Breathing pattern during hypoxia and exercise. Most of the previous studies on the individuality of the breathing pattern were performed on resting subjects (2, 3, 6, 14, 15). In the present study, the increase in ventilation during hypoxic exposure, although moderate, was significant. We observed that the diversity of the flow shapes was also manifest at Al rest (Fig. 2) and that for each subject they were similar to those airflow profiles seen at Sl rest. In other words, the individuality of the shape of the airflow previously reported as an individual characteristic in humans at rest (2,3) is maintained under conditions of hypoxic “stress” of that intensity. According to the statistical tests, the similarity between the ASTERS of Sl rest and Al rest is much greater than that of the other ventilatory variables. Some variables, such as VT and VT/TI, are clearly dissimilar between the two conditions. These results are somewhat different from those observed when the individuality over time was assessed, in which all variables were similar but the test had the
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
2452
AIRFLOW
PATTERN
DURING
highest value for the ASTERS (3). This suggests that, in the present study, there were effective changes in ventilatory characteristics due to hypoxia despite the persistence of the individual airflow shapes. The changes in airflow shape induced by exercise have been investigated by Bradley et al. (4) in healthy subjects and patients. They noted that the shape of the pneumotachogram observable by eye remained essentially unchanged between rest and exercise in the majority of the normal subjects (21 of 38) and patients (23 of 24) and observed a more rectangular shape of the flow patterns during exercise in only a few patients. Our results show that the shape of the airflow is different between rest and exercise, inasmuch as it was not possible to match the airflow pattern at rest and during exercise for each subject. Changes in airflow profile have been reported previously by several authors. Lafortuna et al. (1 l), using the harmonic analysis of the flow signal during inspiration, reported changes between the flow at rest and during exercise. Using this quantitative analysis, they reported that inspiratory airflow has a tendency toward a rectangular shape during exercise despite an asymmetry between the beginning and the end of inspiration. They claim that this rectangular shape would be more economical than a sinusoidal one for the same inspiratory ventilation. A theoretical approach to the regulation of the airflow shape by Yamashiro and Grodins (17) concluded that the human airflow pattern appears to be regulated for minimum work conditions at high work rate levels but not at rest. At rest, the criterion of optimality appears to be based on volume acceleration. These studies show clearly that there is a difference between the airflow shapes at rest and during exercise. In addition to this difference between rest and exercise, our results show that a diversity of breathing patterns can still be observed during exercise. The different subjects have not adopted the same airflow profile to respond to the increase in ventilation (at least at the level of ventilation induced by the exercise at 50% of their sea level VO, max). If we assume that this level of exercise induces the same metabolic demand in each subject, the resulting airflow shape is different between individuals. This observation differs from the hypothesis of several theoretical works in which an identical airflow shape for all individuals was assumed and an optimal pattern of breathing was calculated from the chosen criterion of optimization. Because of the persistence of some diversity among subjects, it appears that further investigation with the aim of explaining the optimization of breathing pattern should take into account the characteristics of the modifications of the shape from rest to exercise, rather than the characteristics of the shape itself. Another hypothesis that can be postulated from the diversity during exercise is that there exists for each subject a characteristic shape during exercise at 50% sea level VO, m8X that is different from the characteristic shape at rest. This implies that when the individuality of the breathing pattern is discussed, the condition (rest or exercise) should be specified. This makes the individuality of breathing pattern different from the fingerprints to which they have been compared (14). It suggests that the
HYPOXIA
AND
EXERCISE
individuality of breathing patterns is a concept rather than a fixed pattern: each individual has his own way of breathing in any one condition. The within-individual similarity of the airflow shapes between the two conditions of exercise supports this hypothesis. The fact that the individual shapes are conserved under the effect of hypoxia and not during exercise may be interpreted as resulting from either a relatively moderate increase in ventilation during hypoxia or the absence of effect of hypoxic stimulation on the airflow pattern formation. During altitude exercise, ventilation was increased by 40% compared with sea level exercise, and the airflow shapes between these two conditions are similar. This result favors the second hypothesis, an absence of effect of the hypoxic stimulation on airflow pattern formation. However, this increase is far less than the nearly 300% increase in ventilation from rest to exercise at sea level, and the importance of the extent of increase in ventilation on the mechanisms acting on the airflow pattern formation cannot be ruled out. In conclusion, exercise and hypoxia have different effects on the individual’s airflow profile: exercise modifies it, whereas hypoxia would not seem to change it either at rest or during exercise. We thank Drs. S. A. Shea and T. Pham Dinh for helpful of the study. Present address of J. H. Eisele: Dept. of Anesthesiology, of California, Davis, CA 95616. Address for reprint requests: G. Benchetrit, Laboratoire logie, Faculte de Medecine de Grenoble, 38 700 LaTronche, Received
5 September
1991; accepted
in final
form
discussions University de PhysioFrance.
27 January
1992.
REFERENCES 1. BACHY, J. P., A. EBERHARD, P. BACONNIER, AND G. BENCHETRIT. A program for cycle-by-cycle shape analysis of biological rhythms. Application to respiratory rhythm. Comput. Methods Program Biomed. 23: 297-307, 1986. 2. BENCHETRIT, G., P. BACONNIER, J. DEMONGEOT, AND T. PHAM DINH. Flow profile analysis of human breathing at rest. In: Concepts and Formalizations in the Control of Breathing, edited by G. Benchetrit, P. Baconnier. and J. Demongeot. Manchester, UK: Manchester University Press, 1987, p. 207-216. 3. BENCHETRIT, G., S. A. SHEA, T. PHAM DINH, S. BODOCCO, P. BACONNIER, AND A. GUZ. Individuality of breathing patterns in adults assessed over the time. Respir. Physiol. 75: 199-210, 1989. 4. BRADLEY, G. W., AND R. CRAWFORD. Regulation of breathing during exercise in normal subjects and in chronic lung disease. Clin. Sci. Mol. Med. 51: 575-582, 1976. 5. BRETSCHGER, H. J. Die Geschwindigkeitskurve der menschlichen Atemluft (Pneumotachogramm). Pfluegers Arch. Ges. Physiol. 210: 134-148, 1925. 6. DEJOURS, P., Y. BECHTEL-LABROUSSE, P. MONZEIN, AND J. RAYNAUD. Etude de la diversite des regimes ventilatoires chez l’homme. J. Physiol. Paris 53: 320-321, 1961. 7. DEJOURS, P., R. H. KELLOGG, AND N. PACE. Regulation of respiration and heart rate response in exercise during altitude acclimatization. J. Appl. Physiol. 18: 10-18, 1963. 8. EASTON, P. A., L. J. SLYKERMAN, AND N. R. ANTHONISEN. Ventilatory response to sustained hypoxia. J. Appl. Physiol. 61: 906-911, 1986. 9. GRAY, J. S., AND F. S. GRODINS. Respiration. Annu. Reu. Physiol. 13: 217-232, 1951. 10. H~;M;~L&NEN, R. P. Optimization of respiratory airflow. In: Modeling and Control of Breathing, edited by B. J. Whipp and D. M. Wiberg. New York: Elsevier, 1983, p. 181-188.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
AIRFLOW
PATTERN
DURING
11. LAFORTUNA, C. L., A. E. MINETTI, AND P. MOGNONI. Inspiratory flow pattern in humans. J. Appl. Physiol. 57: 1111-1119, 1984. 12. MORROW, P. E., AND R. E. VOSTEEN. Pneumotachographic studies in man and dog incorporating a portable wireless transducer. J. Appl. Physiol. 5: 348-360, 1953. 13. PROCTOR, D. F., AND J. B. HARDY. Studies of respiratory airflow. 1. Significance of the normal pneumotachogram. Bull. Johns Hopkins Hosp. 85: 253-280, 1949. 14. SHEA, S., G. BENCHETRIT, T. PHAM DINH, R. D. HAMILTON, AND
HYPOXIA
AND
EXERCISE
2453
A. GUZ. The breathing patterns of identical twins. Respir. Physiol. 75: 211-224, 1989. 15. SHEA, S. A., R. HORNER, G. BENCHETRIT, AND A. Guz. Persistence of a respiratory “personality” into stage IV sleep in man. Respir. Physiol. 80: 33-44, 1990. 16. YAMASHIRO, S. M., AND F. S. GRODINS. Optimal regulation of respiratory airflow. J. Appl. Physiol. 30: 597-602, 1971. 17. YAMASHIRO, S. M., AND F. S. GRODINS. Respiratory cycle optimization in exercise. J. Appl. Physiol. 35: 522-525, 1973.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
human;
airflow
profile;
reproducibility
of
THE SHAPE OFRESPIRATORYCYCLES derived from flow signals reveals characteristic, individual patterns during quiet breathing (2, 5, 7, 12, 13). Previous studies have shown that these so-called “breath prints” remain unchanged over time in adult human subjects (3) and are very similar for identical twins (14). These studies were performed on resting subjects, and the extent to which the individuality is maintained when ventilation is increased remained to be elucidated. Proctor and Hardy (13) reported a triangular or a quasi-rectangular pattern of the maximum effort pneumotachograms. Bradley and Crawford (4) concluded that the airflow pattern varied considerably between subjects but was fairly constant for any one subject even during exercise. Several studies (5, 10, 11,16) that aimed to obtain the criterion of optimization of the airflow shape reached the conclusion that a rectangular airflow pattern was one of the most economical for most of the criteria functions utilized. When the observed patterns of breathing are considered, only those recorded during exercise are of the rectangular type, and 2446
0161-7567192 $2.00 Copyright
the criteria retained cannot explain individual variations in breathing pattern. Furthermore, these criteria were chosen to describe a general optimal pattern, which could then be selected by a physiological control system for changes in ventilation. We began with the assumption an individuality of breathing pattern exists at rest and designed the present study to determine the individual airflow profiles at different levels of ventilation. Hence breathing was recorded for subjects at rest at sea level (200 m), exposed to a simulated altitude of 4,500 m, and during exercise at sea level and at 4,500 m. The aim of the study was to investigate whether the individuality of breathing pattern is conserved when ventilation is increased. METHODS Subjects. Eleven healthy male rescue workers (aged 23-52 yr) participated in the study. All resided close to sea level and had some experience in mountaineering. A few days before the study they underwent pulmonary function tests and a maximal exercise test on a bicycle ergometer to determine their maximal 0, consumption (VO 2,,,). Their physical characteristics and pulmonary function data are given in Table 1. Protocol and measurement. The study was carried out on each subject between 8 and 11 A.M. The first recording was performed in Grenoble (200 m altitude) in the seated position in a comfortable chair for - 15 min. This recording has been labeled Sl rest. It was followed by 5-10 min of exercise on a bicycle ergometer set for a work load equivalent to 50% of the subject’s VO~,~, as determined before the study (Sl exercise). Within 0.5 h the subjects (two at a time) entered a hypobaric chamber at a simulated altitude of 4,500 m; the pressure in the chamber was 443 Torr [equivalent inspired 0, fraction (FI,J = 0.12 at barometric pressure (PB) of 760 Torr]. The chamber had a volume of 31 m3, the air within it was continuously renewed, and its temperature was maintained at the ambient outside temperature. The duration of hypoxic exposure before the beginning of the recording was between 30 min and I h in all subjects. The same sequence of recordings, rest (Al rest) and exercise (Al exercise) at 50% of sea level VO, max,was then performed. Airflow was measured with a pulsed ultrasonic flowmeter (Branta Biometrics) connected to a face mask (Hans Rudolph). 0, and CO, were measured by mass
G3 1992 the American
Physiological
Society
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
AIRFLOW
TABLE
1. Anthropometric
PATTERN
DURING
HYPOXIA
AND
2447
EXERCISE
data, exercise performance, and respiratory function of the subjects vc
Subj
Yr
Height, m
AP BP DP LE LA LT MD PM SG TJC BV
23 29 33 31 52 29 37 44 29 40 32
1.69 1.80 1.78 1.83 1.68 1.75 1.62 1.74 1.75 1.86 1.82
Age,
vo
2max9
maximal
0, consumption
Weight, kg
64.0 73.3 73.0 80.2 76.0 66.0 59.0 70.0 77.3 84.5 86.5 determined
VO
ml
l
kg-’
l
min-’
liters
STPD
58.5 50.7 62.0 43.4 50.6 53.0 64.2 51.6 59.6 49.8 42.0 at sea level;
FEV,
2max7
VC, vital
spectroscopy (Mediflex SX 200, VG Instrument) of a gas sample from the mask, with the signals displayed continuously on an oscilloscope. In addition, arterial 0, saturation (Sao,) was measured continuously with a finger pulse oximeter (Lifestat 1600, PhysioControl). Five minutes were allowed to elapse for a steady state of ventilation to be achieved at rest or during exercise. Then 2100 breaths were recorded on magnetic tape, along with 0, and CO, fraction of the gas sampled from the mask. Analysis of recordings. For each subject all recorded breaths were digitized at a frequency that gave >lOO points per breath. Breath-by-breath analysis of respiratory cycles was performed off-line. For each condition, 50-100 cycles were examined, and abnormal breaths, such as accompanying swallowing and the three breaths after such disturbances, were discarded. A quantitative analysis of the shape of the airflow signal for each breath was performed using a harmonic analysis (fast-Fourier transform of airflow signal normalized for breath duration) as previously described for airflow signals recorded with a Fleisch head (1). In agreement with that study, we confirmed that, for each breath, the first four harmonics retained >95% of the information about the shape of the breath, whatever the condition of recording (rest, altitude, or exercise). The vectorial representation of these four harmonics termed ASTER, and the Cartesian coordinates of the vectors provided eight figures that describe quantitatively the shape of the airflow. After the breathby-breath analysis, a mean ASTER was calculated for each recording by averaging each coordinate across all breaths in the data set of this recording. In addition to the multivariate ASTER, the following variables were measured for each breath: inspiratory time (TI), expiratory time (TE), total breath duration ( TT), tidal volume (VT), and the time indexes (VT/TI and TUTI, TE, and VT were taken together as a trivariate description of each breath’s volume shape, which was termed a TRIAD as previously described (3). Statist ical analysis. For each recording, the means t SD were calculated fo r each variable and for the multivariate ASTER and TRIAD. The analysis used was designed to test whether an individual’s “respiratory personality” is maintained during different conditions (e.g., hypoxia, exercise) relative to the differences occurring between different individuals in the same group (11 subjects) between the same conditions. Such an analysis al-
%pred
BTPS
5.77 5.64 5.99 5.54 4.99 4.92 5.08 5.83 5.15 5.80 5.41 capacity;
FEV,,
forced
liters
4.44 5.28 5.15 4.70 4.11 3.57 3.95 4.67 4.15 4.67 4.40
114
100 112 99 120 91 120 122 97 105 102 expiratory
volume
BTPS
%pred
107 118 121 105 127 84 116 124 98 108 100
in 1 s.
lows for differences within an individual under the different conditions but implies that the individual has a certain breathing personality that can be distinguished from that of other individuals. For example, the statistical analysis was performed by comparing differences between the Sl rest and Al rest recordings within individuals with those differences observed between random pairs of recordings in these two conditions from the same 11 subjects. The differences between recordings were expressed in terms of distance. For a single variable this distance is calculated as the square of the difference between the means divided by the pooled variance. For the ASTERS and TRIADS it was necessary to employ a multivariate determination of this distance in which the covariances between the different variables were also taken into account. The null hypothesis of the test is that the sum of the differences within individuals is not different from the sum of the differences between pairs of individuals taken randomly; the alternative hypothesis is that the sum of the differences within individuals is significantly less than that between a random pair of individuals. A full description of the test has been given previously by Benchetrit et al. (3) and can be applied to single respiratory variables (e.g., VT, TI, TE) or to the multivariate ASTER and TRIAD. When we compared the airflow profiles between two conditions, we normalized the ASTERS for amplitude; i.e., each coordinate was divided by the square root of the sum of the square of all eight coordinates. Thus, for these comparisons, each breath was normalized for amplitude and breath duration (harmonic analysis). This allows the comparisons to refer to the shape only. A paired t test was employed to compare ventilation in the different conditions of recording. RESULTS
The subjects were healthy volunteers with normal respiratory functions (Table 1). Chamber hypoxia at a simulated altitude of 4,500 m was well tolerated by all subjects, who were able to complete the exercise (for a work load equivalent to 50% of sea level Vozmax) during hypoxia without clinical evidence of any ill effects. SL rest US.Al rest. The degree of hypoxia at simulated altitude was shown by the fall in mean SaoQfrom 97 t 1 to
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
2448
AIRFLOW
PATTERN
DURING
2. Minute ventilation at rest and during exercise at sea level and altitude TABLE
Rest
Exercise
Subj
Sl
Al
Sl
Al
AR BP LIP LA LE LT MI) PM
8.72
7.86
9.09
8.59
8.16
8.91
38.71 50.93 51.43 22.06 41.69 39.39 41.11 50.28 52.45
43.91 I
8.01 6.44 6.18 8.66 8.49 7.32 7.48
9.99 8.07 7.24 12.78 8.50 10.73 9.33
22.83 27.81 36.23 15.83 33.73 28.86 27.55 42.74 40.00 39.60 32.04
7.97 I
9.24 I
31.57 I
9.08
SG
TJ VB Mean
9.60
P = 0.02
51.00
P < 0.005
Minute ventilation (\iE) values are in l/min ATP at sea level (Sl) and altitude (Al). Exercise intensity for each subj at 50% VO, max was measured at Sl. * Significant increase in VE at Al, by paired t test on 10 subjs.
84 t 3% at 4,500 m. Ventilation
at rest increased during hypoxic exposure in eight subjects, remained almost unchanged in one subject, and decreased slightly in two subjects. The average increase in ventilation was 15.9% (Table 2); a paired t test performed on the mean values of ventilation calculated over the analyzed breaths for the 11 subjects shows a significant increase (P = 0.02). Table 3 lists the values of VT, TI, and TE in all subjects for the four different conditions of recording. The increase in ventilation from Sl rest to Al rest was achieved mainly by an increase in VT (7 of 11 subjects). The airflow analysis is illustrated in Fig. 1 for subject DP for each condition of recording: the mean ASTER (normalized in amplitude), the airflow profile reconstructed from this ASTER, the airflow profile with its actual duration, and finally the TRIAD drawn to show the actual mean values of VT, TI, and TE. In Fig. 1, the resemblance can be seen between the ASTERS and reconstructed flows in the two conditions of resting ventilation at sea level and altitude, despite changes in the TRIADS. The reconstructed flows for all 11 subjects are shown TABLE
HYPOXIA
AND
EXERCISE
in Fig. 2. The following observations can be made. 1) When airflow at Sl rest for each subject is considered, the difference between subjects represents the individuality of breathing that has previously been reported (2, 3). 2) When airflow at Al rest is considered, the variations among the subjects suggest that there is still an individuality of breathing under hypoxia. 3) The resemblance between the two flows at rest for each subject suggests that the individuality of breathing is maintained during hypoxia. The statistical analysis performed to test the similarity of the ASTERS indicates that the differences within individuals are significantly less than those between individuals. The same test performed on the other ventilatory variables and the trivariate TRIAD indicates that the similarity between the breathing patterns is best evidenced when the multivariate ASTER is used; the P values of these tests are given in Table 4. Hence, the airflow shape appears to be the breathing pattern characteristic that is most conserved when ventilation is changed in subjects exposed to a simulated altitude of 4,500 m. Sl rest vs. Sl exercise. During exercise at a work load of 50% vo, max, ventilation increased in all subjects (Table 2). This augmentation in ventilation was achieved by an increase in VT and decreases in TI and TE (Table 3). The changes in airflow brought about by exercise are illustrated for subject DP in Fig. 1. The ASTER and the reconstituted flow profile (normalized for breath amplitude and duration) of this subject are quite different between Sl exercise and Sl rest, corresponding with the more rectangular shape of inspiration during exercise. For subject DP, despite the decrease in TE, little change can be seen in the shape of expiration. In Fig. 2, for each subject the flow profile shows a general change in shape from Sl rest to Sl exercise. There is a more rectangular shape of the inspiration in almost all subjects, but (except for subject DP, Fig. 1) there are changes also in the shape of expiration in most of the subjects. However, the flows of the different subjects at Sl exercise are not all the same, and there still appears to be an individuality of breathing pattern, although different from that at rest. Indeed, the statistical comparison of the similarity of the ASTERS between rest and exercise at sea level showed that the within-individual difference was not less than
3. VT, TI, and TE at rest and during exercise at sea level and altitude Rest
Exercise
Sea T,evel Subj
AR RP nr LA LE LT MD PM SG TJ VB
VT, ml ATP
420+60 506+50 507k47 484k60 506+35 338-t75 581k133 6252218 611+121 433231 540+107
VT, tidal volume;
Altitude
TI, s
TE, s
1.22kO.10 1.40k0.12 1.38kO.18 1.41kO.12 1.58kO.14 1.29kO.26 2.4OkO.55 1.6920.27 1.8OkO.71 1.44kO.11 1.96kO.40
1.66kO.21 1.93L0.18 2.36kO.48 1.79kO.32 2.21kO.20 1.87kO.34 3.24kO.80 2.65kO.43 2.52kO.98 2.11kO.29 2.3720.51
TI, inspiratory
time;
VT, ml ATP
491k164 500225 6OOk56 5Olk95 666+123 55Ok97 455+133 954+210 587-t91 665-t94 714+142
TE, expiratory
Sea Level
TI, s
TE, s
1.51kO.40 1.26kO.34 1.47t0.21 1.49kO.19 1.81k0.43 1.83kO.43 1.6OkO.45 1.63kO.28 1.55kO.31 1.54kO.11 2.06t0.37
2.25k0.89 2.24kO.80 2.57kO.42 1.65kO.29 2.18k0.32 2.26kO.31 2.17kO.40 2.85k0.54 2.6OkO.53 2.18kO.39 2.52kO.39
VT , ml ATP
956k214 1,140+201 1,280*208 550+109 1,220+67 1,010+135 1,180+212 2,030*187 1,180+130 1,540&117 1,180+126
Altitude
TI, s
TE, s
1.10+0.34 1.06kO.19 0.87kO.20 0.98-tO.11 0.93kO.17 0.9720.12 1.09kO.22 1.43k0.24 0.81+0.14 1.15LO.16 1.11+0.12
1.4 120.62 1.4O-tO.24 1.2520.25 1.1220.12 1.24k0.17 1.14kO.12 1.48k0.28 1.42k0.23 0.96kO.14 1.18t0.17 1.10+0.14
VT, ml ATP
TI, s
TE, s
1,200+176 1,460+156 1,560+210 739+ 150 1,480-+86 1,300+95 788-t-95 2,430?166 1,530+100
0.80+0.16 0.79kO.21 0.7920.15 0.99&O. 19 0.92&O. 13 0.86kO.10 0.50+0.06 1.35kO.15 0.82kO.09
1.05LO.24 0.9320.18 1.03t0.22 1.02kO.16 1.22k0.17 1.12kO.12 0.65kO.06 1.5520.23 0.94+0.10
1,700+88
1.03kO.09
0.97&O. 13
time.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
AIRFLOW
PATTERN
DURING
HYPOXIA
2449
AND EXERCISE 1 sec.
Subject
DP
“I
500 ml
0
2
4
sec.
6
A
VT
TI
n
b
Altitude Rest (72 Breaths)
0 b
;
2
6 sec.
A VT TI
n
Sea Level Exercise (56 Breaths)
Altitude Exercise (65 Breaths)
TI
n
1. Pattern of breathing in 4 different conditions of recording in subj DP in each condition. Left to right: vectorial representation of 4 harmonics in breath-by-breath analysis (ASTER, calculated over no. of breaths in parentheses), airflow profile reconstructed from ASTER, ASTER represented with its actual duration, and each breath’s volume shape (TRIAD). Reconstructed airflow from ASTER is normalized (see METHODS) for breath duration and amplitude and is the best representation of breath shape. Changes in ventilation in different conditions can be judged from TRIADS. FIG.
the between-individual difference (P = 0.57). This indicates that, in the group of 11 subjects, it was not possible to match the airflow profile of each subject at rest and exercise, indicating that the shape of the airflow was significantly modified by exercise. Al rest us. Al exercise. The mean Sa,, levels at altitude fell to 84 t 3% at rest and decreased further to 77 t 5% (P < 0.0001) during exercise. The average increase in ventilation from rest to exercise at altitude was from
9.24 t 1.54 to 43.91 t 9.54 l/min (Table 2). Mean values for VT, TI, and TE are given in Table 3. The changes in the airflow profile (Figs. 1 and 2) parallel those at sea level, with a tendency toward a more rectangular shape of the inspiration during exercise. The same conclusions were inferred from the statistical analysis, which tested for the preservation of the individuality between rest and exercise at altitude (P = 0.98). Hence, the same individuality is not preserved, inasmuch as the shape of the air-
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
2450
AIRFLOW
PATTERN
DURING
HYPOXIA
DP
AR
LE
AND EXERCISE
/\
Sf Rest Al Rest
Sl Exer
Al Exer
LT
BP Sl Resz Al Rest z Sl Exer
Al Exer z PM
SG
Sl Rest
FIG. 2. Reconstructed airflow profiles normalized for breath duration and breath amplitude (see METHODS) are represented in 4 different conditions of recording in all subjects. Sl, sea level; Al, altitude; Exer, exercise. Note diversity of breathing pattern between individuals at rest and during exercise and similarity for each subject between the 2 recordings at rest and those during exercise.
Al Rest
Sl Exer
Al Exer
VB
TJ
Sl Rest Al Rest r Sl Exer
\
1
Al Exer
flow profile is significantly modified between Al rest and Al exercise. SI exercise us. Al exercise. The mean Sa,, levels at altitude fell from 84 to 77% with exercise, whereas there was no change in Sa,, levels during exercise at sea level. There was a significant increase in ventilation between Sl exercise and Al exercise for the same work load (Table 2): ventilation was 42.7% higher for exercise at altitude. The increase in ventilation between the two conditions was caused by an increase in VT and a decrease in TI and TE (Table 3). The airflow profile changes between the two conditions of exercise are illustrated in Fig. 1 for one subject. The ASTERS appear slightly different: all the vectors are closer to the vertical axis, perhaps due to the higher TI/ TE in Al exercise (0.77) than in Sl exercise (0.69). How-
ever, the four vectors occupy the same relative position. This resemblance among airflow shapes is more striking in the reconstructed flow profiles, even when they are represented with their actual duration. The TRIADS show the changes in volume and timing of the breaths between these two conditions. The resemblance is worth noting (Fig. 1) between the two airflow profiles at rest, between those during exercise, and between rest and exercise. This observation can be made for all subjects in Fig. 2. For each subject there is a resemblance between the two flow profiles at rest and during exercise. In addition, as at rest, the flow profiles during exercise differ among the subjects. We tested the similarity of the respiratory variables between Sl exercise and Al exercise. The results are given in Table 4; when ASTERS are considered, there is signifi-
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
AIRFLOW
PATTERN
DURING
4. statistical comparisons of similarity of respiratory variables at sea level and altitude in conditions of rest and exercise TABLE
Sl Rest vs. Al Rest
ASTER TRIAD VT TT TI TE TI/TT VT/TI
(n = 11)
Sl Exercise vs. Al Exercise (n = 10)
P < 0.00025 P = 0.046 P = 0.514 P = 0.011 P = 0.089 P = 0.013 P = 0.107 P = 0.680
P < 0.00025 P = 0.120 P = 0.174 P = 0.083 P = 0.015 P = 0.043 P = 0.015 P = 0.300
ASTER, vectorial representation of 4 harmonics in breath-bybreath analysis; TRIAD, trivariate description of each breath’s volume shape based on VT, TI, and TE; TT, total breath duration; TI/TT, inspiratory duty cycle; VT/TI, mean inspiratory flow. At 0.01 level of significance, only multivariate ASTER shows that there is significantly less difference within individuals than between individuals (see METHODS).
cantly less difference within than among individuals, indicating that, in these two conditions of exercise, it is possible to match the two airflow profiles of each subject. DISCUSSION
The major finding of our study was that the airflow shape changed from rest to exercise. However, not all the subjects adopt the same pattern, and it is still possible to observe a diversity in flow patterns among individuals during exercise. In addition, hypoxia at rest or during exercise does not significantly change the individual flow pattern. Recording methods and signal analysis. Ventilation was measured with a calibrated ultrasonic flowmeter mounted on a face mask. Although this system could have modified the pattern of breathing, introducing an additional dead space, any effect on the shape of the airflow has been kept constant. The shape analysis used (1) was based on a Fourier analysis of cycles normalized for breath duration. The sum of the first four harmonics was found to retain >95% (mean value 98%) of the information about the shape for each breath in any of the four conditions of recording. Lafortuna et al. (11) used a similar analysis on the inspiratory flow profile alone; harmonic analysis was performed on inspiration only, with seven harmonics characterizing the shape of the flow signal. However, as in our study, the index they utilized to estimate the appropriateness of their analysis is clearly >95% when only four harmonics are used (whatever the level of ventilation at rest or during exercise, as in our study). This justifies the use of the four harmonics (8 variables) to describe the shape of the breaths and to perform the statistical analyses. However, when the flow signal with four harmonics is reconstructed, the parts of the breath with a constant flow (flat regions) will appear slightly bumpy, whereas the steep parts of the shapes, such as rapid changes in the flow, are very precisely described. In the harmonic analysis, breaths are normalized for duration. For the illustrations we also normalized the mean ASTERS for amplitude to eliminate the effect of
HYPOXIA
AND EXERCISE
2451
changes from differences in VT in the four conditions of recording and therefore to focus only on the shape. In addition, we normalized each breath for amplitude for statistical comparisons between conditions. Lafortuna et al. (11) also used normalizations in amplitude to 1) compare shapes at different levels of ventilation and 2) add shapes for different subjects. Normalization in breath duration and amplitude is one such dimensionless transformation that was suggested by Gray and Grodins (9) as the first step in analyzing the significance of the shape of flow profiles. Ventilation during hypoxia and exercise. The level of hypoxia was chosen to provide 1) a hypoxic stimulus at rest and 2) an environment compatible with an exercise at 50% of sea level ‘ire, M8Xfor all subjects, given their physical fitness. Indeed, all subjects completed exercise at this altitude without exhaustion or fatigue. Exercise data for one subject (Table 3) were missing because of technical problems during the off-line analysis of the recording. At 4,500 m, where resting Sa,, fell to 84%, the mean increase in ventilation was 15.9% at rest, with an interindividual variability in the response to the hypoxic exposure (Table 2). The mean increase in ventilation observed is comparable with the 19% increase in ventilation observed by Easton et al. (8) in 20 young adults exposed to moderate hypoxia (Sa,, of 80%). In that study, after an immediate initial rise in ventilation, a secondary “plateau” of ventilation (at 119% of the control ventilation) was maintained over 21 h. Our recordings were performed between 30 min and 2 h after altitude exposure, when we considered that a steady state of ventilation was reached. In fact, we performed all recordings in a steady state so that we could calculate a mean cycle representative of the ventilation at the environmental condition under study. Ventilation during exercise at sea level was 31.75 l/min and was further increased (by 40%) to 43.91 l/min under the effect of hypoxia. The higher level of ventilation for the same work load during exercise at altitude than at sea level has already been reported (7). This increase may partly be a response to a further fall in Sa,, during exercise from 84 to 77%. Breathing pattern during hypoxia and exercise. Most of the previous studies on the individuality of the breathing pattern were performed on resting subjects (2, 3, 6, 14, 15). In the present study, the increase in ventilation during hypoxic exposure, although moderate, was significant. We observed that the diversity of the flow shapes was also manifest at Al rest (Fig. 2) and that for each subject they were similar to those airflow profiles seen at Sl rest. In other words, the individuality of the shape of the airflow previously reported as an individual characteristic in humans at rest (2,3) is maintained under conditions of hypoxic “stress” of that intensity. According to the statistical tests, the similarity between the ASTERS of Sl rest and Al rest is much greater than that of the other ventilatory variables. Some variables, such as VT and VT/TI, are clearly dissimilar between the two conditions. These results are somewhat different from those observed when the individuality over time was assessed, in which all variables were similar but the test had the
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
2452
AIRFLOW
PATTERN
DURING
highest value for the ASTERS (3). This suggests that, in the present study, there were effective changes in ventilatory characteristics due to hypoxia despite the persistence of the individual airflow shapes. The changes in airflow shape induced by exercise have been investigated by Bradley et al. (4) in healthy subjects and patients. They noted that the shape of the pneumotachogram observable by eye remained essentially unchanged between rest and exercise in the majority of the normal subjects (21 of 38) and patients (23 of 24) and observed a more rectangular shape of the flow patterns during exercise in only a few patients. Our results show that the shape of the airflow is different between rest and exercise, inasmuch as it was not possible to match the airflow pattern at rest and during exercise for each subject. Changes in airflow profile have been reported previously by several authors. Lafortuna et al. (1 l), using the harmonic analysis of the flow signal during inspiration, reported changes between the flow at rest and during exercise. Using this quantitative analysis, they reported that inspiratory airflow has a tendency toward a rectangular shape during exercise despite an asymmetry between the beginning and the end of inspiration. They claim that this rectangular shape would be more economical than a sinusoidal one for the same inspiratory ventilation. A theoretical approach to the regulation of the airflow shape by Yamashiro and Grodins (17) concluded that the human airflow pattern appears to be regulated for minimum work conditions at high work rate levels but not at rest. At rest, the criterion of optimality appears to be based on volume acceleration. These studies show clearly that there is a difference between the airflow shapes at rest and during exercise. In addition to this difference between rest and exercise, our results show that a diversity of breathing patterns can still be observed during exercise. The different subjects have not adopted the same airflow profile to respond to the increase in ventilation (at least at the level of ventilation induced by the exercise at 50% of their sea level VO, max). If we assume that this level of exercise induces the same metabolic demand in each subject, the resulting airflow shape is different between individuals. This observation differs from the hypothesis of several theoretical works in which an identical airflow shape for all individuals was assumed and an optimal pattern of breathing was calculated from the chosen criterion of optimization. Because of the persistence of some diversity among subjects, it appears that further investigation with the aim of explaining the optimization of breathing pattern should take into account the characteristics of the modifications of the shape from rest to exercise, rather than the characteristics of the shape itself. Another hypothesis that can be postulated from the diversity during exercise is that there exists for each subject a characteristic shape during exercise at 50% sea level VO, m8X that is different from the characteristic shape at rest. This implies that when the individuality of the breathing pattern is discussed, the condition (rest or exercise) should be specified. This makes the individuality of breathing pattern different from the fingerprints to which they have been compared (14). It suggests that the
HYPOXIA
AND
EXERCISE
individuality of breathing patterns is a concept rather than a fixed pattern: each individual has his own way of breathing in any one condition. The within-individual similarity of the airflow shapes between the two conditions of exercise supports this hypothesis. The fact that the individual shapes are conserved under the effect of hypoxia and not during exercise may be interpreted as resulting from either a relatively moderate increase in ventilation during hypoxia or the absence of effect of hypoxic stimulation on the airflow pattern formation. During altitude exercise, ventilation was increased by 40% compared with sea level exercise, and the airflow shapes between these two conditions are similar. This result favors the second hypothesis, an absence of effect of the hypoxic stimulation on airflow pattern formation. However, this increase is far less than the nearly 300% increase in ventilation from rest to exercise at sea level, and the importance of the extent of increase in ventilation on the mechanisms acting on the airflow pattern formation cannot be ruled out. In conclusion, exercise and hypoxia have different effects on the individual’s airflow profile: exercise modifies it, whereas hypoxia would not seem to change it either at rest or during exercise. We thank Drs. S. A. Shea and T. Pham Dinh for helpful of the study. Present address of J. H. Eisele: Dept. of Anesthesiology, of California, Davis, CA 95616. Address for reprint requests: G. Benchetrit, Laboratoire logie, Faculte de Medecine de Grenoble, 38 700 LaTronche, Received
5 September
1991; accepted
in final
form
discussions University de PhysioFrance.
27 January
1992.
REFERENCES 1. BACHY, J. P., A. EBERHARD, P. BACONNIER, AND G. BENCHETRIT. A program for cycle-by-cycle shape analysis of biological rhythms. Application to respiratory rhythm. Comput. Methods Program Biomed. 23: 297-307, 1986. 2. BENCHETRIT, G., P. BACONNIER, J. DEMONGEOT, AND T. PHAM DINH. Flow profile analysis of human breathing at rest. In: Concepts and Formalizations in the Control of Breathing, edited by G. Benchetrit, P. Baconnier. and J. Demongeot. Manchester, UK: Manchester University Press, 1987, p. 207-216. 3. BENCHETRIT, G., S. A. SHEA, T. PHAM DINH, S. BODOCCO, P. BACONNIER, AND A. GUZ. Individuality of breathing patterns in adults assessed over the time. Respir. Physiol. 75: 199-210, 1989. 4. BRADLEY, G. W., AND R. CRAWFORD. Regulation of breathing during exercise in normal subjects and in chronic lung disease. Clin. Sci. Mol. Med. 51: 575-582, 1976. 5. BRETSCHGER, H. J. Die Geschwindigkeitskurve der menschlichen Atemluft (Pneumotachogramm). Pfluegers Arch. Ges. Physiol. 210: 134-148, 1925. 6. DEJOURS, P., Y. BECHTEL-LABROUSSE, P. MONZEIN, AND J. RAYNAUD. Etude de la diversite des regimes ventilatoires chez l’homme. J. Physiol. Paris 53: 320-321, 1961. 7. DEJOURS, P., R. H. KELLOGG, AND N. PACE. Regulation of respiration and heart rate response in exercise during altitude acclimatization. J. Appl. Physiol. 18: 10-18, 1963. 8. EASTON, P. A., L. J. SLYKERMAN, AND N. R. ANTHONISEN. Ventilatory response to sustained hypoxia. J. Appl. Physiol. 61: 906-911, 1986. 9. GRAY, J. S., AND F. S. GRODINS. Respiration. Annu. Reu. Physiol. 13: 217-232, 1951. 10. H~;M;~L&NEN, R. P. Optimization of respiratory airflow. In: Modeling and Control of Breathing, edited by B. J. Whipp and D. M. Wiberg. New York: Elsevier, 1983, p. 181-188.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
AIRFLOW
PATTERN
DURING
11. LAFORTUNA, C. L., A. E. MINETTI, AND P. MOGNONI. Inspiratory flow pattern in humans. J. Appl. Physiol. 57: 1111-1119, 1984. 12. MORROW, P. E., AND R. E. VOSTEEN. Pneumotachographic studies in man and dog incorporating a portable wireless transducer. J. Appl. Physiol. 5: 348-360, 1953. 13. PROCTOR, D. F., AND J. B. HARDY. Studies of respiratory airflow. 1. Significance of the normal pneumotachogram. Bull. Johns Hopkins Hosp. 85: 253-280, 1949. 14. SHEA, S., G. BENCHETRIT, T. PHAM DINH, R. D. HAMILTON, AND
HYPOXIA
AND
EXERCISE
2453
A. GUZ. The breathing patterns of identical twins. Respir. Physiol. 75: 211-224, 1989. 15. SHEA, S. A., R. HORNER, G. BENCHETRIT, AND A. Guz. Persistence of a respiratory “personality” into stage IV sleep in man. Respir. Physiol. 80: 33-44, 1990. 16. YAMASHIRO, S. M., AND F. S. GRODINS. Optimal regulation of respiratory airflow. J. Appl. Physiol. 30: 597-602, 1971. 17. YAMASHIRO, S. M., AND F. S. GRODINS. Respiratory cycle optimization in exercise. J. Appl. Physiol. 35: 522-525, 1973.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
