INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

CENTRAL NERVOUS SYSTEM (CNS) oxygen toxicity can appear in humans on exposure to oxygen pressures >180 kPa. CNS oxygen toxicity can occur as convulsions (similar to epileptic seizures, grand mal) and loss of consciousness, without any warning symptoms. CNS oxygen toxicity is a risk encountered in several fields of human activity, such as combat diving with closed-circuit breathing apparatus and diving with mixtures of nitrogen and oxygen (nitrox) or nitrogen, oxygen, and helium (trimix) in sport and professional diving to depths >30 m. The risk of oxygen toxicity is always considered when deep diving is planned. Hyperbaric oxygen (HBO) therapy may be employed for various indications, including CO poisoning, gas gangrene, nonhealing wounds in diabetic patients, and air embolism after surgery. During such HBO treatments, there is a risk of CNS oxygen toxicity.

Lin and Jamieson (10) suggested that humidity shortens the latency to CNS oxygen toxicity. They presented results demonstrating shortened latencies in a humid atmosphere (60% humidity) compared with a dry atmosphere (<10% humidity) in rats and mice at PO₂ of 585 kPa. Shortened latencies in a humid compared with a dry atmosphere were also demonstrated in mice at 550 and 515 kPa. Pulmonary oxygen toxicity, estimated by wet-to-dry weight, was also more severe in these animals on the humid hyperbaric protocol but less severe in a humid as opposed to a dry atmosphere in normobaric oxygen. The authors related this difference to the contribution of seizures under hyperbaric conditions to pulmonary edema. When there were no convulsions, breathing a humidified gas was found to affect the lung in different ways. There are contradictory findings regarding the effect of humidity on the lung. Ching et al. (5) evaluated the toxic effect of oxygen from the PO₂ in the pulmonary vein of the anesthetized dog at normobaric pressure and found that humidified oxygen aggravated pulmonary oxygen toxicity after 7 h of oxygen breathing. It may also be possible that dry normobaric oxygen is more deleterious to lung surface tension than humidified oxygen in the anesthetized dog after 12 h (11). Exposure of rats to dry normobaric oxygen for 48 h caused epithelial thickening in the lobar bronchi, which was not the case with humidified oxygen (12). In patients treated daily with 256 kPa of oxygen for 95 min, humidified oxygen increased the forced expiratory volume in the first second as opposed to dry oxygen (13). Similar effects of dry and humid conditions on expiratory flow were found at high pressure when the respiratory gas was air (14). Thus most of the findings (excluding the study on pulmonary vein PO₂) showed that humidified normobaric oxygen is beneficial to pulmonary function. Convulsions in humidified HBO cause deterioration of pulmonary function, whereas humidified HBO may improve expiratory flow.

It seems reasonable to us to find that prolonged breathing of dry or humidified oxygen may affect the lung by enhancing the edematous process, affecting bronchial resistance and the lung's inner surface lining. However, it is hard to envisage the effect of humidity on the much faster process of CNS oxygen toxicity. The alveolar gas is saturated with water vapor, whether the inspired gas is dry or humidified. Thus no effect may be expected on arterial PO₂. Only the upper airways may be affected by the humidity of the inspired gas. A cause-and-effect relationship between the upper airways and CNS oxygen toxicity remains somewhat obscure. If humidity does increase the risk of CNS oxygen toxicity, it may have serious implications for diving and for considerations affecting diving equipment and breathing mixtures.

Lin and Jamieson (10) conducted their hyperbaric exposure at an ambient temperature of 20-22°C. This is below the thermoneutral range for both rats and mice at normobaric pressure. The lower critical temperatures for rats and mice are 28 and 30°C, respectively (9). Twenty-two degrees Celsius is also below the thermoneutral range for the rat under hyperbaric pressure (1). Therefore, a cold-induced increase in metabolic rate could have been expected in their experiment. We recently showed that the reduction in the latency to CNS oxygen toxicity is linearly related to the CO₂ production (CO₂) rate (1). An ambient temperature of 22°C elevated the metabolic rate by 30-70% in different rats, which accounted for a 25-50% reduction in the latency to CNS oxygen toxicity. It is possible that bubbling of the oxygen through water inside the hyperbaric chamber (10) created small gas-born water droplets that increased the thermal conductance, thereby elevating the cold-induced metabolic rate of the animals. It is also possible that the increased metabolic rate caused reduced latency to CNS oxygen toxicity. We hypothesized that, when the main modulators of CNS oxygen toxicity (inspired CO₂ and metabolic rate) are controlled, there will be no effect of humidity on CNS oxygen toxicity. In the present study, we controlled the metabolic rate by selection of a thermoneutral temperature and were thus able to separate the effect of metabolic rate from that of the humidity.

Because of the difficulty in measuring oxygen consumption in an oxygen atmosphere, we took CO₂ as an estimate of metabolic rate. Thermoneutral ambient temperatures were used to obtain resting metabolic rates. For the determination of metabolic rate, CO₂ was allowed to increase. Because we have shown that even low levels of CO₂ affected the latency to CNS oxygen toxicity (2), we also checked for a possible contribution of CO₂. The first electrical discharge (FED) in the electroencephalograph (EEG), which precedes the clinical convulsions, is a well-defined phenomenon (8) and was used to define the end point.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals

Experimental System and Procedure

Experimental cage. The experimental cage was a metal, double-walled cage (25 × 11 × 12 cm). One wall for observation of the animal and the top cover, which can be opened, were made of Plexiglas (for details see Fig. 1 in Ref. 1). Thermoregulated water was pumped through the double wall to control the ambient temperature. The incoming gas flowed through a metal container attached to the cage wall for temperature equilibration before entering the cage. The metal walls of the cage were covered on the outside with thermal insulating material. A cable with a male miniconnector for EEG recording passes through the top cover. A humidity- and temperature-measuring device (EE20FT, EE Electronics) was inserted through the top of the cage. For the dry-exposure protocol, a four-walled metal grid measuring 18 × 8 × 10 cm (open at the top and the front wall for observation) was placed on a layer of anhydrous calcium sulfate (Drierite, Vacumed, Ventura, CA), and the space between the three walls of the grid and the cage walls was filled with the same water-absorbing material. Because we found that the compressed oxygen contained water vapor, before entering the cage it was passed through a canister filled with Drierite. The total gas volume in the cage (3,200 ml) was measured with water. For the humid protocol, the Drierite grains were replaced with wet limestone grains of similar size and shape. The metal container through which the inflowing gas passed for temperature equilibration was filled with water, and the incoming gas thus bubbled through the water before entering the cage.

Experimental system. The miniconnectors were mated, and the rat was placed in the experimental cage, which was placed in a 150-liter pressure chamber (Roberto Galeazzi, La Spezia, Italy). The flow of gas through the cage was controlled by needle valves and by observation of two flowmeters situated inside the pressure chamber: one high flow (0-10 l/min), used for fast flushing of the cage's atmosphere, and the other low flow (0-0.5 l/min), used for measuring CO₂. The outgoing gas exited via a bypass tube into the atmosphere of the pressure chamber. A small portion of the outgoing gas was directed out of the pressure chamber (this was controlled by another needle valve), passed through a flowmeter, and was sampled by a mass spectrometer (QP 9000, Morgan Medicals, Rainham, UK) for CO₂ and O₂ concentrations. Water hoses were connected to ports in the pressure chamber and to ports in the experimental cage for recirculation of the thermoregulated water (C/H Temperature Controller Bath and Circulator 2067, Forma Scientific, Marietta, OH). The EEG was recorded on a chart recorder (Gould, Cleveland, OH).

Calibration. The outlets of the flowmeters inside the pressure chamber were connected via a needle valve to the outside atmosphere, whereas the inlet ports were open to the pressure chamber atmosphere. The pressure chamber was filled with oxygen at the experimental pressures of 507 and 608 kPa. Using the needle valve, we diverted various flows out of the pressure chamber into a flowmeter calibrator (VOL-U-METER, Brooks Instrument, Hatfield, PA) and calibrated the marked flow rates on the flowmeter with the actual flow at normobaric pressure outside the pressure chamber. The relationship was linear for all pressures, with r² > 0.99. The flow rate of the compressed oxygen (_ox) in milliliters per minute (calculated from the normobaric flow rate) was lower but linear with the marked rate on the flowmeter scale (FS): at 507-kPa O₂, _ox = 54 + 0.43 × FS, and at 608-kPa O₂, _ox = 46 + 0.41 × FS.

Experimental procedure. The rat was placed in the experimental cage, the EEG miniconnectors were mated, and the cage was placed in the pressure chamber. The rat was unrestrained and could move about freely inside the cage. When the pressure in the chamber was being raised (at 100 kPa/min), the gas flowing through the cage was air (using the high-flow flowmeter). When the desired pressure was reached, a period of 20 min was allowed for acclimation to the experimental conditions (pressure and ambient temperature) in air. At the end of this period, the flow of air was immediately replaced by pure oxygen using the high-flow flowmeter at ~20 l/min for 3 min. Thus one volume of the cage was flushed in 10 s. The time at the end of the first 10 s of O₂ flushing was taken as the start of the oxygen exposure. After the first 3 min, the flow was reduced to 90-100 ml/min (compressed oxygen) using the low-flow flowmeter. This rate was intended to yield a measurable concentration of CO₂, which would rise in an exponential mode until equilibrium was reached (if enough time were allowed). This flow rate was low enough to allow the buildup of CO₂ for the calculation of CO₂ and high enough to prevent the accumulation of CO₂ to a level that markedly affects latency to the FED. The EEG signal was amplified and was recorded continuously on a chart recorder; CO₂ fraction (FCO₂), ambient temperature, and humidity were read and recorded at 1-min intervals. Oxygen concentration was read to ensure that the cage was completely flushed with oxygen. The rat was observed through a window in the pressure chamber for signs of clinical seizure activity. When the FED in the EEG that precedes the clinical convulsions was seen on the recorder, the time was noted and decompression was commenced (at 100 kPa/min). The cage was removed from the pressure chamber, and the rat was freed from the experimental system.

Calculation of CO₂. We calculated CO₂ as we reported previously (1) from the exponential rise in FCO₂. We used nonlinear regression (SAS Institute, Cary, NC) of the measured data for time and FCO₂, according to the following equation
where _ox and CO₂ are in milliliters per minute, P is the pressure in atmospheres absolute, t is the time in minutes, and V_cage is the total gas volume of the cage in milliliters. The calculated CO₂ is then corrected to STPD and expressed in milliliters per gram per hour.

Experimental protocol. Each rat was subjected to a different exposure every 2-3 days. In previous studies (3, 4), we showed that the preceding exposure had no effect on the time to the FED in the following one if there were a 2-day interval in between. This procedure enabled repeated measurements to be taken from the same animal with reduced variability, because we have shown that intra-animal variability is much lower than interanimal variability (3, 4). Each rat was subjected randomly to four exposures: a dry protocol and a humid protocol at 507- and 608-kPa oxygen. Because we had previously found that the sensitivity of CNS oxygen toxicity to any effector (metabolic rate, inspired CO₂, or cinnarizine) (1, 2, 5) increased as inspired PO₂ decreased, we chose two levels of inspired PO₂ to find out whether there is varying sensitivity to dry or humidified oxygen. We could not complete all four exposures in all of the rats due to disconnection of the miniconnector. Because we had to use a low flow of oxygen to enable the calculation of CO₂, we could not eliminate all of the vapor in the dry exposure. Humidity was 31 ± 1% in the dry exposures and 99% in the humid exposures. Therefore, in a second series of experiments, we suspended the measurement of CO₂ and used high gas flow to compare dry and humid conditions at 507 kPa. In the second series, humidity in the dry exposure was 14 ± 3% and in the humid exposure was 99%.

Statistics. Differences in CO₂, ambient temperature, PCO₂ at the end of the exposure, and latency to the FED between humid and dry exposures were tested using the paired t-test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of all of the exposures with the measurement of CO₂ in 20 rats are summarized in Tables 1 and 2 for a PO₂ of 507 and 608 kPa, respectively. In Tables 1 and 2, we present the calculated CO₂, latency to CNS oxygen toxicity, ambient temperature, PCO₂ at the end of the exposure, the humidity, and the difference in latency between the dry and humid exposures. An example of an EEG recording at the start of the electrical convulsion is shown in Fig. 1, where the low-amplitude-high-frequency waveform changes to high amplitude-low frequency. Paired t-test yielded no significant differences between dry and humid exposures for CO₂, ambient temperature, end point inspired PCO₂ (PI_CO2), and latency to CNS oxygen toxicity at 507 or 608 kPa. Ambient temperature was kept in the thermoneutral range. The difference in humidity, 31 vs. 99%, did not affect the latency to the FED. CO₂ was not affected by either humidity or pressure. A combined CO₂ of 0.81 ± 0.06 ml · g¹ · h¹ was the resting value, as expected at the thermoneutral temperature of 28 ± 1°C. At 507-kPa O₂, latency was 23 ± 5 and 22 ± 6 min in dry and humid conditions, respectively, and, at 608-kPa O₂, it was 15 ± 4 and 14 ± 3 min in dry and humid conditions, respectively.

View larger version (48K): [in this window] [in a new window]   Fig. 1.   Example of an electroencephalogram (EEG) at the start of convulsions at an inspired PO2 of 507-kPa O2. Traces are shown at 2 chart velocities. The normal EEG changes into a convulsive EEG at the first electrical discharge (FED).

The relationship between latency to CNS oxygen toxicity and CO₂ is shown in Figs. 2 and 3 for 608- and 507-kPa O₂, respectively. No clear trend with respect to slope and direction can be seen in the lines connecting dry and humid data for each animal (Figs. 2 and 3). There is also no reduction in latency as a function of CO₂ for any of the rats in dry or humid conditions. The only clear trend that may be seen is that both dry and humid latencies from the same animal are usually close to each other.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Latency to central nervous system (CNS) oxygen toxicity plotted against CO2 production at 608-kPa O2. Lines connect values obtained from the same rat for dry and humid exposures.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Latency to CNS oxygen toxicity plotted against CO2 production at 507-kPa O2. Lines connect values obtained from the same rat for dry and humid exposures.

The possible relationships between latency to CNS oxygen toxicity, humidity, and PI_CO2 are shown in Figs. 4 and 5 for a PO₂ of 507 and 608 kPa, respectively. As in Figs. 2 and 3, apart from the prevalent close latency times in dry and humid conditions for most of the rats, no clear trend (slope and direction) can be seen for the data of each animal with respect to PI_CO2. In all of the data, however, there is a general pattern of increased latency to CNS oxygen toxicity with respect to PI_CO2.

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Latency to CNS oxygen toxicity plotted against inspired PCO2 (PICO2) at the end of exposure to 507-kPa O2. Lines connect values obtained from the same rat for dry and humid exposures.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Latency to CNS oxygen toxicity plotted against PICO2 at the end of exposure to 608-kPa O2. Lines connect values obtained from the same rat for dry and humid exposures.

The results from 11 rats in the second series of experiments, in which a high flow of oxygen produced a drier atmosphere than the low flow required for the estimation of CO₂, are shown in Table 3. In this series, there was no significant difference in ambient temperature, and, although there was a greater difference in humidity (14 vs. 99%), there was no significant difference in latency to the FED between dry and humid conditions. Latencies in this series did not differ from those observed at 507 kPa in the first series.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study proves that humidity, at least at thermoneutral temperatures, does not increase the risk of CNS oxygen toxicity, as suggested in a previous paper (10). These discrepancies are explained by the control of other risk factors in the present study. The results agree with our view that it is impossible to adduce theoretical evidence for the effect of humidity on CNS oxygen toxicity.

Cooling of the brain by inhalation of dry oxygen, and the ensuing brain protection, may be another explanation for the results of Lin and Jamieson (10). Flushing of the nasal cavities with oxygen in the intubated rat on a heating pad increased the core-to-brain temperature difference (6). This cooling effect was related to the level of oxygen flow (250-1,000 ml/min). When we extrapolated Einer-Jensen and Khorooshi findings (6) to a moderate ventilatory rate of 150 ml/min, we obtained a reduction in brain temperature of <0.2°C. A reduction in brain temperature due to evaporative cooling of the nasal pathways is usually found when an increase in brain temperature might be expected, such as in exercise (7) or in heat load, but not during cold exposure. This is controlled by thermally sensitive shunting of facial venous flow (15). Therefore, it is questionable whether protection against CNS oxygen toxicity during dry-oxygen breathing may be affected by the small respiratory effect on brain cooling.

The predetermined low resting range of CO₂ in the present study had no effect on latency, and there was no difference between dry and humid conditions (Figs. 2 and 3). In our previous publication (1), we proved that latency to CNS oxygen toxicity decreased linearly with the increase in metabolic rate. There was no reduction in latency as a function of CO₂ for any of the rats in dry or humid conditions. This is related to the resting metabolic rate due to the ambient temperature, which was kept within the thermoneutral range compared with a wider range in the previous study (1). The only clear trend is that both dry and humid latencies from the same animal are usually close to each other. This reflects our well-established finding (1-5) that there are sensitive rats with a short latency to CNS oxygen toxicity and nonsensitive rats with a long latency.

The presence of CO₂ in the inspired oxygen reduces latency to CNS oxygen toxicity linearly with the increase in PI_CO2 (2), and even a small increase in PI_CO2 may cause CNS oxygen toxicity at oxygen pressures that would not produce this toxicity on their own (4a). When measurement of CO₂ is required, the presence of CO₂ in the inspired oxygen is unavoidable. In the present study, latency to CNS oxygen toxicity for each rat in dry and humid conditions was not affected by PI_CO2 (Figs. 4 and 5). If PI_CO2 were affecting latency, then the line connecting the data from each rat should have a negative slope, which is clearly not the case. The increased latency to CNS oxygen toxicity with respect to PI_CO2 in all of the data is in the opposite direction of the established effect of CO₂. In the present study, the CO₂ rose continuously, compared with the constant level in the previous report (2), and the PI_CO2 thus represents the end point and not the level during the exposure. Longer latencies enable greater accumulation of CO₂, because of the similarity in the rate of CO₂. This explains the positive relationship between latency and PI_CO2. The latencies to CNS oxygen toxicity at 507-kPa O₂ when CO₂ was present in dry and humid conditions were 23 ± 5 and 22 ± 6 min, respectively. These did not differ from the latencies in the series without CO₂ accumulation, which were 23 ± 6 and 21 ± 6 min in dry and humid conditions, respectively. We, therefore, conclude that, in the present study, there was no effect of CO₂ on latency to CNS oxygen toxicity. When there was no effect of metabolic rate or of CO₂, no effect of humidity was found either. Humidity does not on its own affect sensitivity to CNS oxygen toxicity.

Humans are exposed to high oxygen pressures in various situations in diving and hyperbaric treatments. Whereas in open-circuit diving the diver breathes dry gas, with a closed or semiclosed circuit the diver breathes humid gas. During hyperbaric treatment, a spontaneously breathing patient usually breathes dry gas, whereas humidifiers may be used in artificial respirators. CNS oxygen toxicity is a known risk in all of the above-mentioned situations. Whether or not humidity affects CNS oxygen toxicity has important implications for the design of such systems. The present study demonstrates that humidity does not increase the risk of CNS oxygen toxicity in thermoneutral conditions, and we cannot conceive of a reasonable explanation for an influence at colder temperatures. Our findings, therefore, ease the burden on hyperbaric oxygen protocols.