Hypoxia simulation in humans has usually been performed by inhaling hypoxic gas mixtures. For this study, apneic divers were used to simulate dynamic hypoxia in humans. Additionally, physiological changes in desaturation and re-saturation kinetics were evaluated with non-invasive tools such as Near-Infrared-Spectroscopy (NIRS) and peripheral oxygenation saturation (SpO2).
In case of apnea, arterial partial pressure of oxygen (pO2) decreases, while partial pressure of carbon dioxide (pCO2) increases. To avoid damage to hypoxia sensitive organs such as the brain, compensatory circulatory mechanisms help to maintain an adequate oxygen supply. This is mainly achieved by increased cerebral blood flow. Intermittent hypoxia is a commonly seen phenomenon in patients with obstructive sleep apnea. Acute airway obstruction can also result in hypoxia and hypercapnia. Until now, no adequate model has been established to simulate these dynamics in humans. Previous investigations focusing on human hypoxia used inhaled hypoxic gas mixtures. However, the resulting hypoxia was combined with hyperventilation and is therefore more representative of high altitude environments than of apnea. Furthermore, the transferability of previously performed animal experiments to humans is limited and the pathophysiological background of apnea induced physiological changes is poorly understood. In this study, healthy human apneic divers were utilized to mimic clinically relevant hypoxia and hypercapnia during apnea. Additionally, pulse-oximetry and Near Infrared Spectroscopy (NIRS) were used to evaluate changes in cerebral and peripheral oxygen saturation before, during, and after apnea.
Clinically relevant acute hypoxia and concomitant hypercapnia is mostly seen in patients with obstructive sleep apnea syndrome (OSAS), acute airway obstruction or during cardiopulmonary resuscitation. Major limitations in the field of OSAS and other hypoxemic conditions include the limited transferable knowledge about the pathophysiology derived from animal studies and that human models are non-existent 1. To mimic hypoxia in humans, hypoxic gas mixtures have so far been used 2–7. However, these conditions are more representative of high altitude surroundings than of clinical situations where hypoxia, in general, is accompanied by hypercapnia. To monitor tissue oxygenation during cardiac arrest and resuscitation, animal studies have been performed 8 to investigate physiological compensatory mechanisms.
Apneic divers are healthy athletes capable of depressing the breathing impulse that is evoked by low arterial oxygen saturation 9 and an increased pCO2 10,11. We investigated apneic divers in order to mimic clinical situations of acute hypoxia and concomitant hypercapnia 12. This model can be used to evaluate clinical setups, improve the pathophysiological understanding of patients with OSAS or pathological breathing disorders, and reveal new possibilities for studying a potential counter balancing mechanism in cases of apnea. Furthermore, different techniques to detect hypoxia in humans can be tested for feasibility and accuracy in the case of dynamic hypoxia that is present in emergency situations (i.e., airway obstructions, laryngospasm or cannot intubate, cannot ventilate situations) or to simulate intermittent hypoxia in patients with OSAS.
Noninvasive techniques to detect hypoxia in humans are limited. Peripheral pulse oximetry (SpO2) is an approved tool in pre-hospital and hospital settings to detect hypoxia 13. The method is based on light absorption of hemoglobin. However, SpO2 measurement is limited to peripheral arterial oxygenation and cannot be used in cases of pulseless electrical activity (PEA) or centralized minimal circulation 14. In contrast, Near-Infrared Spectroscopy can be used to evaluate cerebral tissue oxygen saturation (rSO2) in real-time during PEA, during hemorrhagic shock or following subarachnoid hemorrhage 15–19. Its use is constantly growing 20 and methodological studies have revealed a positive correlation between SpO2 and rSO2 3,4.
In this study, we provide a model to simulate clinically relevant hypoxia in humans and present a step-by-step methodology to compare peripheral pulse oximetry and NIRS in case of de- and re-saturation. By analyzing physiological data in case of apnea, our understanding of counter balancing mechanisms can be improved.
Ethics statement
All procedures performed in studies involving human participants were in accordance with the ethical standards of the 1964 Helsinki declaration and its later amendments. The design of this study was approved by the local ethics committee of the University Hospital of Bonn, Germany.
NOTE: Ensure that subjects are in good and healthy condition, free of any anti-hypertensive medicine and at least 24 hours free of catecholamine inducing agents like caffeine or equal substances.
1. Preparation of the Test Subject
2. Data Collection
3. Apnea
4. Processing Data
5. Analyze Values
6. Statistical Processing
Figure 1 displays simultaneous recordings of SpO2 and NIRS values (NIRScerebral and NIRStissue) during apnea in one patient. Total apnea time was 363 sec. Following apnea NIRS and SpO2 values remained stable for approximately 140 sec. A decrease in SpO2 was detected after 204 sec by peripheral SpO2 whereas a decrease of NIRScerebral was detected after 238 sec. The lowest measured SpO2 following apnea was 58% and lowest measured NIRScerebral was 46%. At the end of apnea NIRScerebral increased after a time delay of 12 sec whereas SpO2 increased after a time delay of 30 sec.
In a recent study of ten apneic divers we showed a significant decrease in NIRScerebral values from 71% (range 85 – 55) to 54% (range 74 – 24) 12. Median SpO2 decreased from 98% (range 100 – 98) to 81% (range 94 – 67). Figure 2 displays the mean time delays between the beginning of apnea and decrease in NIRScerebral versus SpO2 values of these ten divers. Oxygen saturation measured by NIRScerebral decreased significantly later than oxygen saturation on the fingertip measured by SpO2 [175 sec; SD = 50 sec versus 134 sec; SD = 29 sec; (t(9) = 2.865, p = 0.019, r2 = 0.477)]. This can be taken as a sign for elevated cerebral blood flow and preferential oxygen supply of cerebral tissue during apnea.
After restart of respiration (Figure 2c), values of NIRScerebral increased significantly earlier than SpO2 values [10 sec; SD = 4 sec versus 21 sec; SD = 4 sec (t(9) = 7.703, p < 0.001, r2 = 0.868)]. Figures 2b and d display de- and re-saturation measured on the fingertip (SpO2) and above the musculus quadriceps femoris (NIRStissue) during apnea. NIRStissue values decreased significantly earlier than SpO2 values [39 s; SD = 13 sec versus a delay of 125 sec; SD = 36 sec (t(6) = 4.869, p = 0.003, r2 = 0.798)]. This time delay might show that peripheral vasoconstriction leads to a decrease in tissue oxygenation, even before a decrease in arterial oxygen saturation — visualized by SpO2 — is measureable. There was no difference in time delay after restart of respiration between NIRStissue and SpO2 [NIRStissue 30 sec; SD = 16 sec versus SpO2 27 sec; SD = 7 sec (t(6) = 0.631, p = 0.551, r2 = 0.062)]. This indicates, that the observed time delay is not caused by the different devices themselves.
To compare de- and re-saturation during individual apnea durations, we normalized SpO2-, NIRStissue– and NIRScerebral-baseline values to 100% (Figure 3). To compare individual apnea duration, total apnea duration of each subject was also set to 100%. 12
Figure 1: Time-course of NIRS, SpO2, and Heart Rate (HR) during Apnea. Raw data of one participant is displayed. Total apnea-time was 363 sec. Subject exhibited an earlier decrease in SpO2 than in cerebral rSO2. Please click here to view a larger version of this figure.
Figure 2: Time Delays during Apnea and Restart of Respiration. a) Mean time delay between beginning of apnea and decrease of NIRScerebral versus SpO2 values; b) Mean time delay between beginning of apnea and decrease of NIRStissue versus SpO2 values; c) Mean time delay between restart of respiration and an increase of NIRScerebral versus SpO2 values; d) Mean time delay between restart of respiration and an increase of NIRStissue versus SpO2 values. Error bars indicate standard error of the mean. Data and figure from Eichhorn et al. 2015 12. Please click here to view a larger version of this figure.
Figure 3: Temporal Progression of Normalized SpO2, NIRScerebral and NIRStissue Values: To equilibrate individual variations in apnea time, all apnea times were standardized to 100%. Thus the variations in the three plotted parameters are assigned to the relative apnea times. Baseline values measured prior to apnea were defined as 100%. Error bars indicate standard error of the mean. Data and figure from Eichhorn et al. 2015 12. Please click here to view a larger version of this figure.
The total apnea time is mainly caused by lung size and oxygen consumption per minute and influenced by an individuals' ability to withstand the breathing reflex caused by increasing pCO2 or decreasing pO2. Apnea divers are trained to maximize their breath-hold duration and are used to doing so in maximal inspiration. Therefore, the time until hypoxia is detectable differs between individuals and depends on the subject's physical condition and training status and might even vary by their daily state and willingness to withstand the breathing reflex. The subject's stress levels can be reduced by detailed education of protocol steps and a calm ambient environment.
There are many factors that influence total apnea time, which means that the testing environment should be standardized in order to get results that are reliable and repeatable. If researchers are interested in studying the catecholamine increase or sympathetic nerve activity, substances influencing both (i.e., caffeine, nicotine, food like bananas, nuts, or any medical substances like monoamine oxidase (MAO) inhibitors, etc.) should be avoided. Also the intravenous line should be established at least 20 min before apnea. A subjects' stress level will mainly influence catecholamine-levels and could falsify researchers' results of blood analysis. In general, researchers should create baseline levels of each subject to normalize the results because of the large inter-individual differences.
Non-invasive measurements of tissue oxygenation by NIRS technology uses semi-quantitative changes in oxygenated and deoxygenated hemoglobin 21. The use of NIRS is constantly growing 20 and it can detect saturation of cerebral and peripheral tissue, independent of pulsatile blood flow. NIRS values depend on the amount of venous and arterial vessels placed under the NIRS-electrodes. NIRS values can therefore differ significantly depending on the amount of venous versus arterial vessels under the electrode. Also, placement and contact pressure will influence the reliability of values. Values should be checked for stability before starting measurement. If NIRS signals vary during baseline measurements, replace the electrodes or check for total skin contact. For interpretation of the NIRS results, relative de- or increase of values compared to baseline values should be used (not absolute).
Due to the physical burden of a maximal breath-hold, the number of apneas per subject is limited. The preparation protocols should be equal for each subject and all devices should be double-checked before they are used. Do not modify the protocol in one cohort. Standardized setups are mandatory to create results that are reproducible. Although hyperventilation before maximal breath hold lowers arterial CO2 levels and delays the breathing stimulus, it also affects cerebral autoregulation and vasomotor reactivity22. Active hyperventilation should be avoided to minimize disruptive effects by the subject.
The overall goal of this model is to simulate hypoxia in humans by breath hold. Therefore, additional measurement devices can be established to get more detailed information about blood pressure (i.e., invasive blood pressure measurement) or sympathetic nerve activity. Blood pressure measurements can be used to estimate the burden of prolonged apnea to the vessel system. ECG signals can be used to calculate beat-to-beat variability in R-R interval or to detect cardiac arrhythmia. Furthermore, cortisol-levels in saliva or catecholamine-levels 29 in blood-samples can be measured at different time points during and after apnea. The kinetics of these values opens up a number of possible study opportunities. Still, a reliable detection of hypoxia is necessary to ensure hypoxic conditions caused by apnea. Values measured by different devices but in the same apneic session can be compared directly. Time differences (for instance, until blood pressure increase, desaturation starts, etc.) from different individuals should be normalized to total apnea time.
The respiratory reflex is one of the strongest stimulus of the human body. Acute hypoxia and hypercapnia is therefore only seen in patients with pathologies (i.e., OSA, emergency situations, laryngospasm, CPR, etc.). Mostly unforeseen, hypoxia is difficult to detect, always influenced by a triggering event and difficult to evaluate because of a subjects' comorbidities. Although total apnea time of divers and patients undergoing hypoxia should not be compared because of the completely different starting conditions, human compensatory mechanisms to avoid damage to the brain in case of hypoxia are identical 23–28. An extended voluntary breath-hold also empties the body's oxygen-storage and increases a subject's pCO2 29. Apneic divers were shown to generate reliable results during simulation of dynamic hypoxia in humans 12. We measured a minimum cerebral saturation only slightly higher than values seen in patients during cardiac arrest (42.2 ± 10.7% 15 and 37.2 ± 17.0% 14). This indicates that our model is able to mimic clinically relevant hypoxia. Although hypoxia causes serious health problems, the underling physiological mechanisms are yet not completely understood 1 and till now no relevant clinical human model existed to simulate acute hypoxia in humans. Using healthy apneic divers as a clinical relevant model to simulate hypoxia and hypercapnia in humans holds large potential for future investigations. This model allows scientists to study the compensatory mechanism to avoid hypoxic damage in a reproducible human model. It allows a clinically relevant simulation of hypoxic emergency situations such as laryngospasm or "cannot ventilate – cannot intubate". It might be used to prove the feasibility of new invasive or non-invasive tools for measuring human hypoxia. Furthermore, this model may help to understand the correlation of increased endogenous catecholamines and their impact on cardiac function (i.e., heart rate variability, cardiac output, etc.). By using different and new devices to observe hypoxia in apneic divers new parameters may be explored and may extend our understanding of hypoxia in the future.
The authors have nothing to disclose.
Special thanks to all volunteers who participated in the original study. The work of L. Eichhorn was supported through a scholarship of the Else-Kröner-Fresenius Foundation. The authors would like to thank Springer, Part of Springer Science+Business Media, for copyright clearance (License Number 3894660871180) and the kind permission of reusing previously published data.
SpO2 | Dräger Medical AG&CO.KG | SHP ACC MCABLE-Masimo Set | peripheral SpO2-Monitoring |
Non Invasive Blood Pressure (NIBP) | Dräger Medical AG&CO.KG | NIBP cuff M+, MP00916 | |
Electrocardiographic (ECG) | Dräger Medical AG&CO.KG | Infinity M540 Monitor | ECG monitoring |
Docking station | Dräger Medical AG&CO.KG | M500 Docking Station | connection of M540 to laptop |
NIRS | NONIN Medical’s EQUANOX | Model 7600 Regional Oximeter System | measuring of cerebral and tissue oxygenation |
NIRS diodes | EQUANOX Advance Sensor | Model 8004CA | suited for measuring cerebral and somatic oxygen-saturation |
Laptop | |||
DataGrabber | Dräger Medical AG&CO.KG | DataGrabber v2005.10.16 | software to synchronize M540 with laptop |
eVision | Nonin Medical. Inc. | Version 1.3.0.0 | software to synchronize NONIN with laptop |