Journal of Applied Physiology
Vol. 82, No. 3, pp. 988-997, March 1997

Oxygen pulse in guinea pigs in hyperbaric helium and hydrogen

Susan R. Kayar and Erich C. Parker

Albert R. Behnke Diving Medicine Research Center, Naval Medical Research Institute, Bethesda, Maryland 20889-5607

ABSTRACT

INTRODUCTION

MATERIALS AND METHODS

RESULTS

DISCUSSION

ACKNOWLEDGEMENTS

FOOTNOTES

REFERENCES

ABSTRACT

Kayar, Susan R., and Erich C. Parker. Oxygen pulse in guinea pigs in hyperbaric helium and hydrogen. J. Appl. Physiol. 82(3): 988-997, 1997.We analyzed O₂ pulse, the total volume of O₂ consumed per heart beat, in guinea pigs at pressures from 10 to 60 atmospheres. Animals were placed in a hyperbaric chamber and breathed 2% O₂ in either helium (heliox) or hydrogen (hydrox). Oxygen consumption rate (O₂) was measured by gas chromatographic analysis. Core temperature and heart rate were measured by using surgically implanted radiotelemeters. The O₂ was modulated over a fourfold range by varying chamber temperature from 25 to 36°C. There was a direct correlation between O₂ and heart rate, which was significantly different for animals in heliox vs. hydrox (P = 0.003). By using multivariate regression analysis, we identified variables that were significant to O₂ pulse: body surface area, chamber temperature, core temperature, and pressure. After normalizing for all nonpressure variables, the residual O₂ pulse was found to decrease significantly (P = 0.02) with pressure for animals in heliox but did not decrease significantly (P = 0.38) with pressure for animals in hydrox over the range of pressures studied. This amounted to a roughly 25% lower O₂ pulse for normothermic animals in 60 atmospheres heliox vs. hydrox. These results suggest that reduction of cardiovascular efficiency in a hyperbaric environment can be mitigated by the choice of breathing gas.

diving; heart rate; high-pressure neurologic syndrome; metabolic rate; modeling; oxygen consumption; telemetry; thermoregulation

INTRODUCTION

CHANGES IN HEART RATE are a classic response to diving (16). Bradycardia can be quite pronounced in naturally diving animals: a seal may slow its heart rate from over 100 to 10 beats/min as it submerges (15), and heart rate may continue to drop throughout a dive (2). In an aquatic animal, this bradycardia is presumed to reflect a reduction in the volume of the perfused vascular bed, as the animal conserves blood flow to hypoxia-sensitive central organs but reduces perfusion to peripheral tissues (15).

Even animal species that are not evolutionarily adapted to diving can demonstrate a marked reduction in heart rate during immersion, although of a smaller magnitude than that shown by diving species (2). Apnea evidently slows the heart by stimulating a vagal response, but immersion of the face in water (13) and hyperbaric compression of the thorax (21) may have additional depressive effects on heart rate. Bradycardia has been noted in freely breathing human divers compressed in a dry chamber (4, 19, 22), but its magnitude and duration are apparently highly variable (4). Human divers have been observed to lose their initial bradycardia and return to predive heart rates (19) and also to regain a bradycardia later in the hyperbaric exposure (22). Pressure has been found to have a depressive effect even on the contraction rate of isolated atrial preparations, with the magnitude of this depression varying with the composition and concentration of gases in the superfusing fluids (9). Pressure has also been reported to alter other aspects of cardiac function, including contractility (23), cardiac output (14), and electrical repolarization (19) in nondiving mammals breathing hyperbaric gases. Some of these effects may be a consequence of the high intrathoracic pressures needed to ventilate in hyperbaria due to high gas density (19). These observations suggest that the relationship between heart rate and ambient pressure is complex. This complexity makes it difficult to draw conclusions, unless multiple factors are taken into consideration simultaneously. Stepwise multivariate analysis is one approach to identifying correlations among variables in complex situations.

Cold, exercise, and general stress are additional factors often present in diving that may have a significant effect on heart rate and metabolism and may change the onset or intensity of a diving bradycardia. To differentiate among these factors, it is often useful to compute heart rate relative to metabolic rate. The total volume of O₂ consumed by an animal per heart beat, known as the O₂ pulse, is important because a change in this ratio implies a change in cardiovascular efficiency (20). Little is known about the O₂ pulse as heart rate changes during diving. This dearth of correlated physiological data on cardiac function and oxygen transport is undoubtedly because of technical difficulties in measuring both O₂ consumption rate (O₂) and heart rate simultaneously in immersed and swimming subjects. We, therefore, measured O₂ pulse in a nondiving mammal in a dry hyperbaric chamber.

In this study, guinea pigs were instrumented with radiotelemeters to transmit electrical signals from the heart. The animals were placed in a dry chamber that was pressurized to generate hyperbaric conditions of up to 60 atmospheres (atm; pressure equivalent of a dive to 590 m) by using either a He-O₂ or a H₂-O₂ gas mixture. O₂ of the animals was measured by gas chromatographic analysis of the chamber gas throughout the dive. Chamber temperature was varied as a means of generating a range of O₂ and heart rate values at each pressure. The radiotelemeters also transmitted body temperature from their implant site in the abdomen.

We chose to use H₂ in these experiments because it has undergone analysis by the United States Navy and others (1, 8, 9, 19) as a major component of a breathing mixture for deep dives. Advantages to using H₂ instead of He include its greater worldwide availability and its lower density and, therefore, lower breathing resistance. Hyperbaric H₂ also possesses narcotic properties that diminish symptoms of high-pressure neurologic syndrome (HPNS) (1, 8), a debilitating condition that appears to be caused by interference with synaptic transmission at pressures exceeding 20 atm (10).

Body heat losses are particularly high in a hyperbaric environment because of the increases in mass-specific heat capacity and thermal conductivity with increasing density. Animals compensate for this increased heat loss by increasing O₂ over a range of chamber temperatures that varies with pressure (18). Because H₂ has a higher molar heat capacity and thermal conductivity than He (12), body heat losses are even greater in a hyperbaric H₂ environment than in one containing He. Thus our use of these two breathing gas mixtures was intended to vary O₂ in a predictable manner as well as to offer us an opportunity to further study the physiological consequences of hyperbaric H₂ exposure.

The values we measured for O₂ pulse were mathematically modeled as a function of physical and biological terms related to metabolism and thermoregulation in these two hyperbaric gas mixtures. Multivariate regression analysis allowed us to clearly demonstrate the relationship between O₂ pulse and pressure by separating the simultaneous effects of the measured thermal variables on O₂ and heart rate.

MATERIALS AND METHODS

Animal preparations. Guinea pigs (Cavia porcellus, male, Hartley strain, n = 20) were housed in an accredited, professionally staffed animal-care facility and had ad libitum access to food and water before experiments. All experiments were approved by the institute's Animal Care and Use Committee and were conducted according to the principles described in the the National Institutes of Health "Guide for the Care and Use of Laboratory Animals," [DHEW Publication No. (NIH) 92-3415, 1992, Office of Science and Health Reports, DRR/NIH, Bethesda, MD 20892].

Radiotelemeters were used to sense and transmit core body temperature and electrical signals from the heart. The telemeters (model TA11ETA F40-L20, Data Sciences International, St. Paul, MN) had been specially modified by the manufacturer to withstand repeated compression and decompression by filling all empty space inside the telemeter case with a gel to eliminate any trapped pockets of gas surrounding the electronics. Aseptic surgery was performed to implant a telemeter in the peritoneal cavity of each animal, following the general procedures recommended by the manufacturer. Animals were anesthetized by halothane inhalation (4 l/min of 4% halothane to induce anesthesia, 1 l/min of 2% halothane to maintain it). A midline abdominal incision was made in the skin and muscle layers. The body of the telemeter was laid gently onto the intestines and attached to the overlying muscle wall with nonabsorbable sutures. The muscle wall was then closed with absorbable sutures. The telemeter was equipped with two leads for sensing the heart electrical activity; these leads were run through small punctures made in the muscle wall to exteriorize them from the peritoneal cavity. A small trochar was slid between the skin and muscle of the upper abdomen and chest to form two narrow subdermal tracks for the leads to lie in. The final position of the telemeter leads was with the negative lead tip near the right shoulder and the positive lead tip below the left axilla; this simulated a conventional lead II. The abdominal skin was then closed with surgical staples, and the animal was allowed to recover.

Animals were used in experiments 1-6 wk after surgery. In all cases, at the time of use in experiments, animals had sufficiently recovered so that their immediate postsurgery weight loss was regained, the incision was sufficiently healed to remove the surgical staples, and some of the hair shaved from the surgical field had regrown. The animals appeared to tolerate the telemeter implants well; there was no postoperative fever or indications of discomfort, i.e., the animals did not exhibit tooth grinding, unusual vocalizing, struggling, or lethargy when handled, and food and water consumption appeared normal.

To monitor the output from the telemeter, an animal was placed inside a 7.6-liter Plexiglas box. The box was wrapped externally with two electrical wires to form the antenna for the telemetry system. The telemetry system sent out heart electrical signals at a frequency of ~200 Hz and core temperature data at a frequency of ~1 KHz. These data were automatically filtered (values exceeding ±2 SDs were discarded), averaged, and recorded each minute for the duration of an experiment by using a CTR86 receiver, a BCM-100 consolidation matrix, and a Dataquest III analysis system (Data Sciences International).

Dive protocol. Animals were selected randomly for diving in the He [n = 10; 761 ± 43 (SE) g mean body mass] or H₂ (n = 10; 787 ± 30 g) gas mixtures.

For each dive, an animal was placed inside the antenna box, which was set inside a 140-liter hyperbaric chamber (Bethlehem, Bethlehem, PA). The box had a small opening in one end to admit chamber gas and was connected by a hose at the other end to a port on the chamber. Temperature inside the box was monitored and regulated to within 1.5°C (Fig. 1) by inserting the chamber thermostat controller into a small opening in the top of the box. Chamber temperature was regulated by means of a heat pump that circulated freon through coils lining the inside of the dive chamber.

For the animals to be dived in the H₂ gas mixture, the chamber was pressurized at 1-2 atm/min with pure He to 10 atm. The O₂ concentration thus fell to 2%, but the PO₂ remained near 0.2 atm from the 1-atm air initially enclosed in the chamber. This initial pressurization with He was needed to dilute the O₂ in the chamber to avoid an explosive gas mixture when introducing H₂; nonexplosivity limits for mixtures of H₂ and O₂ are 0-4% O₂ in H₂ and 0-4% H₂ in O₂ (7). The chamber was then flushed with a mixture of 2% O₂ in H₂ (hydrox) until the N₂ content of the chamber gas dropped to <0.5%, and the He content was <4%, as measured by a gas chromatograph (Shimadzu GC-9A, Columbia, MD). The animal was maintained at 10 atm in hydrox (0.2 atm PO₂) for ~1 h at a selected temperature between 25 and 36°C. A constant stream of gas flowed through the animal's box to the gas chromatograph. Hydrox was added to the chamber as needed to maintain constant pressure (±0.15 atm).

The O₂ of the animal was computed from the gas flow rate, and the O₂ content difference between the chamber gas and the gas stream leaving the animal's box, once a steady state in chamber temperature and excurrent gas O₂ content was reached after ~30 min (Fig. 1). The excurrent O₂ content used for these calculations was the mean of three consecutive gas chromatographic readings. SEs for these O₂ readings were typically <1%. Gas flow rate was measured with a floating ball-type meter that had been calibrated with hydrox. A constant flow rate of 11.8 ± 0.2 l/min was selected to allow excurrent O₂ content to be never >1.95% and never <1.70%. This flow rate was thus low enough to permit an easily detectable O₂ extraction by the animal but high enough to prevent the animal from becoming hypoxic or hypercapnic (minimum PO₂ of 0.17 atm and maximum PCO₂ of 0.03 atm). Water vapor was scrubbed from the excurrent gas with an absorbent (anhydrous CaSO₄, Drierite, Hammond, Xenia, OH), and CO₂ production was assumed equal to O₂ for the sake of mass balance calculations. If the ratio of CO₂ production to O₂ were actually 0.7, our O₂ calculations would be in error by not more than 2%. The precision of the O₂ measurements was ultimately set primarily by the precision of the calibration of the gas flowmeter (±2%).

The chamber was subsequently pressurized at 1-2 atm/min with hydrox to 20, 40, and 60 atm (Fig. 1). Because the hydrox always contained 2% O₂, PO₂ increased throughout the experiments to 0.4, 0.8, and 1.2 atm at 20, 40, and 60 atm chamber pressure, respectively. Each pressure was maintained for ~1 h to measure O₂ during the second half of the hour. A different temperature between 25 and 36°C was randomly selected for each pressure. Thus each animal was measured at all four pressures, with a variable sequence to the temperature settings at each pressure. This approach allowed us to generate a matrix of temperatures and pressures sampled. At the higher pressures, the lower end of the temperature range had to be limited to prevent the animals from becoming severely hypothermic; at 60 atm, 30°C was the coldest chamber temperature sampled. Core temperature was reported as the final value measured at the end of the hour at each pressure; this value was known with a precision of ±0.1°C (Fig. 1).

Heart rate values were also computed near the end of the hour at each pressure (Fig. 1). In many animals, the telemeter signal was sufficiently strong and free of electrical artifacts from skeletal muscles so that 20 consecutive heart rate measurements, each representing the mean value for 1 min, were averaged to obtain the value we reported. However, on occasion, the telemetry system did not recognize the signal and did not record a heart rate value for a minute, or the value recorded was clearly deviant from successive values by 50 beats/min or more. Heart rate values reported here were from at least 10 recorded values, with SE values from individual animals of 1-5%.

The dive profile for animals in the He gas mixture [2% O₂ in He (heliox)] was identical to that used for the animals in hydrox. The meter for measuring gas flow through the animal's box was recalibrated with heliox (8.5 ± 0.1 l/min; similar range of excurrent O₂ contents to hydrox). Initial pressurization to 10 atm was with pure He. The chamber was then flushed with heliox until the N₂ content fell below 0.5%. Animals spent ~1 h at 10, 20, 40, and 60 atm in heliox (0.2, 0.4, 0.8, and 1.2 atm PO₂, respectively).

At the end of all experiments, animals were rapidly decompressed within 1-2 min to 10 atm and killed by an addition of 1.5 atm CO₂ to the chamber. It was necessary at this point to kill the animals to prevent them from asphyxiating, since the chamber PO₂ dropped below 0.2 atm at chamber pressures <10 atm. Death at 10 atm also prevented animals from suffering from decompression sickness (DCS). Pain from DCS is expected only from decompressions at pressures <10 atm, whereas with explosive decompression from greater pressures, symptoms of DCS are typically numbness, paralysis, and cardiac arrest and require 5-10 min to develop. The chamber was then further depressurized and, in the case of hydrox dives, flushed with pure He for several minutes before opening to ensure safe elimination of H₂.

From the 20 animals used, a total of 78 data points (39 in each gas mixture) were included in the analysis. Values are missing from one animal in heliox at 60 atm (weak telemeter signal) and from one animal in hydrox at 60 atm (gas shortage prevented completion of the dive profile).

Statistical analysis. Data were analyzed by stepwise least squares regression. Regressions were compared with each other by F-tests, with significant difference assigned at the P = 0.05 level. The data from each gas mixture were pooled to generate a multivariable model of O₂ pulse, by using chamber temperature, pressure, and the various measurements of biological variables collected during the experiments. For least squares regressions of the model, these terms were analyzed sequentially, starting from only the y-intercept term and adding each successive variable in order of greatest statistical significance (forward step). After adding each variable, an F-test was performed to determine whether there had been a significant improvement. We then started with all variables and eliminated any that were not significant (backward step). There was no difference in the best-fit parameter estimates in the forward vs. backward step analyses. The functions fitted in the analysis were of the form y = a + bx. More complex terms in x, including squared terms and cross terms, were tested and were not found to contribute significantly to the analysis, as confirmed by an examination of residuals (6).

RESULTS

For animals in both heliox and hydrox, when we used data collected at all temperatures and pressures together, there was a direct correlation between O₂ and heart rate. This relationship was significantly different [F(1,75) = 9.295, P = 0.003] for animals in heliox vs. hydrox, with the data for hydrox animals best fit by a greater slope (Fig. 2A). The same data are shown plotted separately for each pressure, with the regression lines from the pooled data superimposed (Fig. 2B). To analyze the effects of pressure and to explore the reasons for the differences in O₂ pulse between animals in hydrox and heliox, we then constructed a multivariable model.

We reasoned that the heat drain of a small animal in hyperbaria should have a major impact on metabolism, heart rate (f_H), and, potentially, on O₂ pulse. The most complex model with statistically significant parameters that we identified was the following

(1)

where S (m²) is estimated body surface area of an animal [computed from body mass (M_b; in g) as 9 ^· 10⁴ M^2/3_b] (11), T_cham is chamber temperature (°C) measured inside the animal's box, T_core is the temperature (°C) registered by the telemeter in the peritoneal cavity, and P is chamber pressure (atm). The  parameters were estimated by fitting Eq. 1 to data from animals in the two gases separately (Table 1).

Table 1. Estimated values for parameters of Eq. 1, their SE values, and level of significance P Parameter Heliox Hydrox Estimated value     SE P Estimated value SE P  0 0.378 × 103 0.57 × 104 <0.001 0.231 × 103 * 0.26 × 104 <0.001  surf  0.520 × 103 0.22 × 103 0.022  0.681 × 103 0.27 × 103 0.018  cham  0.234 × 105 0.84 × 106 0.009  0.309 × 105 0.56 × 106 <0.001  core  0.461 × 105 0.17 × 105 0.011 * 0.275  P  0.419 × 106 0.12 × 106 0.002 * 0.362 Marked parameters (*) are significantly different (P < 0.05) in 2% O2 in hydrogen (hydrox) vs. 2% O2 in helium (heliox). See Eq. 1 and text for symbol definitions.

For animals breathing heliox, O₂ pulse decreased with increasing animal size, increasing chamber temperature, increasing core temperature, and increasing pressure (Table 1). For animals in hydrox, neither core temperature nor pressure was a significant variable (Table 1). The O₂ pulse in hydrox decreased only with increasing chamber temperature and animal size, with parameter estimates for these variables not significantly different [F(2,70) = 0.447, P = 0.64] from those in heliox (Table 1).

DISCUSSION

Our measured values for O₂, from ~0.7 ml O₂ ^· g^{1 ·} h¹ and increasing fourfold in response to severe cold, are in agreement with those previously reported for guinea pigs at 1 atm from 26 to 5°C (3). This increasing metabolic rate with declining environmental temperature is a standard thermoregulatory response (11), in which heart rate is expected to increase proportionally with O₂, at least so long as animals remain normothermic. Indeed, when exercise intensity is used to modulate O₂, a direct correlation between heart rate and O₂ is so well accepted that it is common practice to estimate O₂ for a given subject and activity from heart rate alone (20). However, factors such as temperature extremes and emotional state are recognized to alter heart rate without necessarily changing O₂ proportionally (20). Pressure and dive gas mixture are expected to have an additional but unpredictable effect (4, 9, 19). Consequently, we expected that multiple factors were acting simultaneously to vary O₂ and heart rate in a manner that should be examined most effectively as a multivariable model.

The model analysis confirmed that there were significant differences in the relationship between heart rate and O₂ for animals breathing heliox vs. hydrox, with these differences attributable to heat loss and pressure effects. Because this relationship involves up to five parameters per gas (Table 1), it is too complex to represent graphically. We can examine O₂ pulse with regard to individual independent variables to illustrate the primary features of the data. However, it must be recognized that these two-dimensional analyses are incomplete and, therefore, can appear to be somewhat misleading (5).

The O₂ pulse decreased with increasing body surface area (Fig. 3; regression line is from data at all pressures combined). In this two-dimensional analysis, as well as in the full model, there was no statistically resolvable difference in the value of the slope for heliox vs. hydrox animals [F(2,74) = 0.336, P = 0.72]. This inverse correlation between O₂ pulse and body surface area was attributable to a significant decrease in O₂ with increasing body surface area (P = 0.025) and no significant correlation between heart rate and body surface area (P = 0.19). Body surface area was computed from body mass as a variable that is associated with body heat loss and as a means of normalizing for animal size. O₂ per unit body mass is generally found to decrease with increasing animal size for a wide variety of animal species (24). Heart rate also has been found to scale inversely with body size in many species, but significant differences in heart rate usually require a wider range of body sizes than included in this study (17). The similarity between heliox and hydrox data suggests that within this study gas mixture did not contribute significantly to the scaling of O₂ and, therefore, O₂ pulse with body size.

The O₂ pulse decreased significantly with increasing chamber temperature (Fig. 4; regression line is from data at all pressures combined). As a two-dimensional analysis, as well as in the full model, there was no statistically resolvable difference in slope between heliox and hydrox data [F(1,74) = 2.08, P = 0.15]. Similarly, O₂ pulse decreased significantly with increasing core temperature (Fig. 5; regression line is from data at all pressures combined), with no statistically resolvable difference in slope between the two gases in this two-dimensional analysis [F(1,74) = 0.24, P = 0.62]. Note that this is in contrast to the results from the full five-parameter model in which core temperature required separate slopes in the two gas mixtures, with a slope not distinguishable from zero for animals in hydrox (Table 1). This illustrates the point that two-dimensional projections do not always reveal multidimensional relationships (5). Because we were able to compute significant parameters for both core and chamber temperatures for animals in heliox, but a significant temperature parameter only for the chamber in hydrox (Table 1), this suggests that animals in heliox were able to remain more thermally independent of their environment than animals in hydrox. This is consistent with the lower heat capacity and thermal conductivity of heliox vs. hydrox (12).

The O₂ pulse decreased significantly [F(1,75) = 11.29, P = 0.001] with increasing pressure for animals in heliox in this two-dimensional analysis (Fig. 6). However, for animals in hydrox, O₂ pulse and pressure were not significantly correlated [F(1,74) = 0.321, P = 0.57] (Fig. 6). This is consistent with the five-parameter model, in which there was a significant regression with pressure for heliox but not for hydrox (Table 1).

To examine the effect of pressure on O₂ pulse independently of the other concurrent factors, we computed the regression terms for all the significant nonpressure variables of the model (₀, _surf, _cham, and _core for heliox; ₀, _surf, and _cham for hydrox; Table 1). This is equivalent to illustrating the last step in a stepwise regression analysis of these data, in which pressure is the last term to be tested for significance. The residuals of this fit (i.e., the mean difference, at each pressure, between the observed O₂ pulse and the O₂ pulse predicted by the other terms of Eq. 1) indicate the presence or absence of a pressure dependence after accounting for other significant terms. We plotted mean residuals (Fig. 7) for clarity only; all analyses were carried out on individual data points. Negative values for the mean O₂ pulse residuals indicated those pressures at which the model, in the absence of a pressure term, overestimated O₂ pulse. Similarly, positive residuals indicated underestimation (6).

The differing effect of pressure on O₂ pulse in heliox vs. hydrox is now clearly illustrated (Fig. 7), taking into account other significant factors, as opposed to the statistically significant but not graphically convincing two-dimensional analysis shown previously (Fig. 6). There was a significant regression of O₂ pulse residuals vs. pressure for the heliox data (Fig. 7; P = 0.018). This indicated that pressure was indeed a necessary variable in the model for heliox. The O₂ pulse residuals in a regression with pressure in hydrox (Eq. 1) were not significantly correlated with pressure (P = 0.38; Fig. 7), suggesting that the nonpressure terms explained our hydrox data adequately. The slopes of these regressions in heliox vs. hydrox were significantly different from each other [F(1,75) = 4.960, P = 0.029).

We tested a variety of other mathematical models before concluding that the terms shown in Eq. 1 were an adequate representation of the relationships among the variables we measured. When we included parameters for both chamber and core temperature (as performed in Eq. 1), a significant value for animals in hydrox was obtained only for the chamber temperature term (Table 1). A significant core temperature parameter in hydrox (P = 0.007, _core = 0.394 × 10⁵) could be obtained only by eliminating chamber temperature from the analysis, which resulted in a substantially poorer fit to the data (71% larger residual sum of squared errors). We also explored alternative temperature variables, such as the difference between core and chamber temperature and the difference between actual core temperature and normal core temperature (38.5 ± 0.5°C). We found no statistical support for using these compound variables.

We also tested terms for both pressure and the square of pressure, to determine whether there was a statistically significant curvature in the relationship between O₂ pulse and pressure, particularly in the hydrox data (Fig. 7). Neither pressure (P = 0.36) nor the square of pressure (P = 0.59) was found to be significant in hydrox. For the heliox data, the simultaneous presence of both pressure and square of pressure terms was not supportable; pressure and square of pressure were essentially interchangeable in fits to the heliox data. The square of pressure term (_P² = 0.589 × 10⁸ ± 0.170 × 10⁸, P = 0.001) provided a slight curvature that did not materially affect either the goodness of fit (residual sum of squared errors with _P² smaller by 0.2%) or the values of the other parameters. Computations of O₂ pulse at sample pressures and temperatures using the _P² term were only different by 1 nl O₂ ^· g^{1 ·} beat¹ from those computed using the _P term. Similarly, testing squared terms for the nonpressure variables and cross-terms between the various variables did not yield useful improvements in the model fit.

In Fig. 7, there is an appearance of increasing residual O₂ pulse at 10-40 atm, and decreasing O₂ pulse from 40 to 60 atm in hydrox. This kind of curvature would not be adequately tested for by using the pressure term in Eq. 1. Inclusion of both pressure and the square of pressure did not prove to be statistically significant in hydrox. The preferred approach to testing for this curvature would be to include a model parameter that estimated the value of a pressure maximum as, for example

(2)

This model failed to estimate a value for the pressure at which O₂ pulse was at a maximum (_{P max}), because the model was attempting to estimate too many parameters for the number of discrete pressures sampled (four). The _{P max} term never appears independently of _P in Eq. 2. Lacking a robust estimate for _P, _{P max} cannot be estimated with the available data.

We then attempted to force a value for _{P max} of 40 atm in hydrox and 20 atm in heliox, based on the appearance of curvature in Fig. 7, giving the following models

(3)

(4)

This is not the optimal approach to examining curvature in Fig. 7 because we had no a priori physiological or methodological reasons for selecting these pressures and we cannot be sure that 20 and 40 atm are correctly chosen, given that only four pressures were sampled. In heliox, we found no statistical support for the curve fit of Eq. 4 (larger residual sum of squared errors than for parameter fits of Eq. 1). In hydrox, we found that there was a significant parameter estimate for the _P term of (P = 0.049) and that Eq. 3 was a better fit to the data than Eq. 1 (smaller residual sum of squared errors using Eq. 3). This suggests that there may be an underlying small increase in O₂ pulse with increasing pressure in the general range of 10 to 40 atm and a subsequent decrease in O₂ pulse at pressures exceeding 40 atm in hydrox.

We estimated the residuals for O₂ and heart rate by normalizing for all the nonpressure terms and we regressed them with pressure. In both hydrox and heliox, these regressions had a curved appearance as well, lending further support to this speculation. This "pressure maximum" phenomenon, if valid, would raise the interesting question of what physiological mechanisms were being stimulated by pressure in the lower pressure ranges and inhibited by pressure at higher pressures. The pressure maximum may also be lower in heliox than in hydrox; the disparate physiological properties of these gases may lend clues to the basis for this phenomenon.

Accounting for individual variability between animals did not change the results of this analysis. A model that allowed for each animal to have its own intercept term, effectively a correction for each individual, resulted in the same relationship of O₂ pulse to pressure: zero slope for hydrox and a significant negative slope for heliox. This slope for heliox was not statistically different from that shown in Table 1. Moreover, this more elaborate model could predict an O₂ pulse response for given environmental conditions only for the twenty animals used in the present study. The model of Eq. 1 and Table 1 is more generally applicable and is therefore preferred.

As an example of how this model can be used, consider a representative guinea pig of 775 g body mass: its surface area computed from mass is 7.59 × 10² m². From the raw data, we find that 34°C is a chamber temperature that will allow a guinea pig to maintain a stable core temperature of 38.8°C over the 10- to 60-atm range in both heliox and hydrox. From the parameters in Table 1, we compute the O₂ pulse for this animal in heliox at 10 and 60 atm to be 75.9 nl O₂ ^· g^{1 ·} beat¹, and 55.0 nl O₂ ^· g^{1 ·} beat¹, respectively. For the same animal in hydrox, the O₂ pulse at any pressure is 74.3 nl O₂ ^· g^{1 ·} beat¹. Thus after normalizing for significant thermoregulatory factors, the effect of increasing pressure from 10 to 60 atm is a 28% [(75.9  55)/75.9] reduction in O₂ pulse for animals in heliox in this example, but no change for animals in hydrox.

The observed decrease in O₂ pulse in heliox could theoretically be due to an increase in relative heart rate, a decrease in relative O₂, or some combination of these. To resolve this issue, we computed residual heart rate and residual O₂ for animals in each gas mixture in the same manner in which we computed O₂ pulse residuals and regressed them with pressure. For animals in hydrox, neither the heart rate nor O₂ residuals varied significantly with pressure (r = 0.158, P = 0.34 and r = 0.133, P = 0.42, respectively). However, for animals in heliox, the significant decrease in O₂ pulse residuals with pressure resulted from the combined effects of a nonsignificant decrease in O₂ residuals (r = 0.151, P = 0.36) and a nonsignificant increase in heart rate residuals (r = 0.139, P = 0.40) with pressure. This underscores the importance of O₂ pulse as a subtle measure of cardiovascular efficiency. The reduced O₂ pulse at high pressures in heliox can be described as an elevated heart rate relative to the amount of O₂ consumed, when pressure and temperature effects are taken into consideration.

We did not measure O₂ in these guinea pigs at 1 atm before the dive. However, we have measured O₂ in other individuals of similar body size in 1 atm of air and at room temperature (25°C); mean O₂ was 0.73 ± 0.01 ml O₂ ^· g^{1 ·} h¹ (n = 5; unpublished observation). From the regression of heart rate and O₂ for animals in hydrox, we would predict that heart rate at this O₂ should be 247 ± 9 beats/min (Fig. 2). Some of the animals studied here (n = 10) remained undisturbed in the chamber, breathing 1 atm air at room temperature (23-30°C) for 30-90 min before the start of the dive (Fig. 1); mean heart rate in these animals was 243 ± 9 beats/min. Given the scatter in these data, it is inappropriate to place too much emphasis on a match between the 1-atm air estimate and the hyperbaric hydrox data. However, the similarity in these calculations argues against the presence of a general diving bradycardia in the hydrox animals.

We can only speculate on the cause of the decreasing O₂ pulse with increasing pressure that we report here for animals breathing heliox but no difference (t-test, P = 0.57) in O₂ pulse at 10 vs. 60 atm in hydrox. From the Fick equation, we know that in general O₂ may decrease due to a reduction in arterial O₂ content, arteriovenous O₂ extraction, cardiac output, or some combination of these. We have no data to support a change in any of these variables. We do not know what effect increasing O₂ pressure from 0.2 to 1.2 atm had on blood gases or blood flow with increasing chamber pressure in these experiments. Breathing pure O₂ at 1 atm is believed to elicit modest increases in arteriovenous O₂ extraction (20), whereas cardiac output has been reported to decrease significantly in hyperoxia (14). However, since the pressure of O₂ was the same in heliox and hydrox at any chamber pressure, it is unlikely that hyperoxia was a primary factor involved in the decreasing O₂ pulse in heliox.

Bradycardia is the commonly expected response of the heart to increasing pressure (4, 16), whereas our analysis suggested that the animals in heliox were experiencing a slight tachycardia, and hydrox animals were not altering heart rate in a consistent manner in response to pressure over the pressure range we studied. The presence of a hyperbaric bradycardia in nondiving mammals is highly variable, and its origin and physiological significance are unexplained (4). There is evidence to suggest that electrical repolarization of the heart is slowed as pressure increases, and this effect may be due to the high intrathoracic pressures generated in order to ventilate the lungs with dense gases (19). Because hydrox is less dense than heliox at any given pressure, gas density differences may be related in some complex manner to heart rate differences in our animals in heliox vs. hydrox. However, Gennser and Örnhagen (9) demonstrated that in preparations of isolated atria of rats, beating frequency declined in response to pressure, and this bradycardia could be mitigated by superfusing the atria with solutions containing H₂. Superfusion with He caused only a slight increase in beating frequency (9). This partial reversal of bradycardia by H₂ was attributed to its narcotic properties, which, in turn, are generally attributed to the relative solubility of H₂ in the lipid component of neurological tissues (9). Hydrogen is more soluble in lipids than is He, and one of the perceived benefits to diving with hydrox is the reduction in symptoms of HPNS observed in subjects breathing hydrox (1, 8). The animals in this study were conspicuously calmer in hydrox than in heliox; at 40 and 60 atm in heliox, animals were observed to be generally nervous and agitated, and to have moderate to severe tremors.

Thus the reduced O₂ pulse observed at high pressures for animals in heliox may be another manifestation of HPNS, with the consequent conclusion that cardiovascular efficiency is reduced when subjects undergo this additional diving stress. Reduced cardiovascular efficiency with increasing ambient pressure is apparently not obligatory, since O₂ pulse was not monotonically correlated with pressure in animals breathing hydrox and, therefore, partially protected from HPNS (Fig. 7). The ambiguity of the optimal curve fit for O₂ pulse in hydrox (Eq. 1 vs. Eq. 3) may reflect a delayed onset for HPNS manifestations in hydrox.

We conclude that there is a difference in the amount of O₂ consumed per heartbeat in guinea pigs in a hyperbaric hydrogen vs. He environment. The results of our model showed that body surface area, ambient temperature, body temperature, and pressure all play roles in this difference. When O₂ pulse was analyzed as a function of all these variables simultaneously, we found that pressure significantly depressed O₂ pulse for animals breathing heliox, whereas O₂ pulse in hydrox was similar at 10 and 60 atm. The explanation we offer is that animals in heliox at higher pressures have reduced cardiovascular efficiency as a symptom of HPNS. Animals breathing hydrox, which elicits a narcotic suppression of HPNS, do not experience this systematic decline in O₂ pulse over the range of pressures we examined. This demonstrates that a change in relative cardiovascular efficiency, when O₂ pulse is used as an index, may be present but is not an obligatory part of deep hyperbaric exposure.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the technical support provided by Tom James, Walter Long, Jr., William Porter, David Schoenauer, and the staff of electrical technicians. The installation and support for the telemetry system was by Pankaj Karnik, whose efforts are always much appreciated. Animal care and surgery were expertly performed by Eugenia O. Aukhert, John Braisted, and Tracy Cope. We thank Joe Ahiers and James Huhn of Data Sciences International for working through our problems of designing unbendable telemeters. Susan Mannix provided editorial services. An anonymous referee proposed the analysis associated with Eqs. 3 and 4.

FOOTNOTES

Address for reprint requests: S. R. Kayar, Code 0512, Albert R. Behnke Diving Medicine Research Center, Naval Medical Research Institute, 8901 Wisconsin Ave., Bethesda, MD 20889-5607.

Received 14 December 1995; accepted in final form 28 October 1996.

REFERENCES