Marathoners’ breathing pattern protects against lung injury by mechanical ventilation: a pilot study

Background: Marathoners use the 2:2 breathing pattern. We hypothesized that this ventilation method may protect against ventilator-induced lung injury. Methods: We studied the 2:2 breathing pattern as a mechanical ventilation mode, by assessing the gas exchange in intact rabbits and the pulmonary protective effects in isolated rabbit lungs. The typical setting of this breathing method was 30 cycles/min of respiratory frequency. The time allocation for one cycle was as follows: 1st inspiratory period, 2nd inspiratory period, 1st expiratory period, and 2nd expiratory period (all 0.3 s long) with intermittent resting periods (all 0.2 s). Results: The 2:2 breathing pattern caused no problems regarding the eciency of oxygen uptake and carbon dioxide elimination. The wet-to-dry weight ratio of the lung was lower for the proposed 2:2 breathing pattern than with the inversed ratio ventilation (both inspiratory:expiratory ratio 1:1). Conclusions: The marathoners’ breathing pattern may be a novel method to provide protection against ventilator-induced lung injury in clinical settings.

metabolism is increased over an extended time, it has been suggested that the 2:2 breathing pattern provides su cient capacity for CO 2 release [13].
In the present study, we developed a ventilator prototype that can perform intermittent positive pressure ventilation, mimicking the breathing cycle of the 2:2 breathing pattern (for technical details see Additional le 1). This mode of ventilation was named the marathoners' breathing pattern ventilation (MBV). We assumed that the MBV is a mechanical ventilation method that divides the tidal volume required for conventional mechanical ventilation into the inspiratory volume of 1 st and 2 nd inspiratory periods. Since the prototyped ventilator was designed to deliver a constant inspiratory ow rate, and the durations of the 1 st and 2 nd inspiratory periods were set to be the same, the inspiratory volumes of the two periods were the same. Thus, each inspiratory volume can be reduced in the MBV. In addition, the MBV had an I:E ratio of 1:1. Therefore, MBV can be considered as a ventilation mode that combines the features of low tidal volume ventilation and IRV. In this study, the following experiments were conducted for MBV: Experiment 1) The oxygen uptake and carbon dioxide emission of healthy lungs were compared in vivo between the MBV and IRV (I:E ratio 1:1) in rabbits. Experiment 2) The effects of the MBV on the pulmonary pre-edema model were compared with those of the IRV (I:E ratio 1:1) in isolated perfused rabbit lungs.
In the MBV method, the 1 st inspiratory, 2 nd inspiratory, 1 st expiratory, 2 nd expiratory phases taken together, de ned one cycle. In the IRV method, the inspiratory and subsequent expiratory phase together de ned one cycle. In both MBV and IRV methods, tidal volume was de ned as the total ventilation volume during one cycle. In both MBV and IRV methods, the unit of respiratory rate was cycles/min.

Methods
All experimental protocols regarding the use and care of animals in the present study were approved by the Laboratory Animal Care Committee of the Faculty of Medicine at Tottori University. Adult female Japanese white rabbits (2.2-3.1 kg) were purchased from Oriental Yeast Co. (Tokyo, Japan). The rabbits were kept under standard housing conditions with free access to food and water.
2.1. Experiment 1: Assessment of pulmonary oxygenation e ciency 2.1.1. Experimental design Rabbits (n = 9; body weight, 2.6 ± 0.3 kg) were anesthetized with 30 mg/kg pentobarbital (Abbott Laboratories, North Chicago, IL, USA) intravenously and 25 mg/kg ketamine (Daiichi Sankyo, Tokyo, Japan) intramuscularly. The experiment was conducted with the rabbits in supine position. The animals were tracheotomized, and a catheter was placed into the common carotid artery to measure the arterial blood pressure and heart rate, as well as to allow for blood gas sampling. Another catheter was inserted into the external jugular vein ipsilaterally for continuous infusion of lactated Ringer's solution (Otsuka Pharmaceutical, Tokyo, Japan) (10 ml/kg/h) containing pancuronium bromide (MSD K.K., Tokyo, Japan) (0.5 mg/kg/h) and pentobarbital (5 mg/kg/h). Anesthesia, muscle relaxant dosage, and uid therapy were according to Maeda et al. (2004) [14] with slight modi cations. Animals were ventilated through a 3ml arti cial nose (total volume of the circuit was about 7 ml) using a Harvard Rodent Ventilator 683 (Harvard Apparatus, Holliston, MA, USA) which had a tidal volume of 6 ml/kg, a respiratory frequency of 40 breaths/min, and a positive end-expiration pressure (PEEP) of 2 cmH 2 O. To prevent VILI, the tidal volume was set to a low value of 6 ml/kg during the stabilization period [15].
After a stabilization period of 30 min under these conditions, normal blood acid-base balance was con rmed. Then, the MBV and IRV methods were alternately conducted for 20 min. The MBV method was carried out with the ve patterns shown in Table 1 The IRV method was conducted with a tidal volume of 8 ml/kg using the Harvard Ventilator 683, and the respiratory rate was adjusted to one cycle/min increments until the PaCO 2 value was in the normal range.
The I:E ratio of the Harvard Ventilator 683 is xed at 1:1, as the factory default setting. The tidal volume in rabbits under conventional mechanical ventilation is considered to be 8-10 ml/kg -(built into the prototype system) [16], the respiratory frequency of ACOMA AR 300 was set to be the same as the respiratory rate, which was allocated to the MBV method, the tidal volume of ACOMA AR 300 was xed at 8 ml/kg, and the I:E ratio of ACOMA AR 300 was xed at 1:1.
The tidal volumes in the MBV method were determined from the average gas volume of several exhalation cycles collected by the classic water displacement method [17]. According to ndings in a preliminary experiment, the dissolution-dependent decrease in volume after 10 min was less than 0.05 ml per 10 ml of a gas mixture containing 5% CO 2 .

Measured variables
Arterial blood pressure and airway pressure were recorded using a PowerLab system (AD Instruments, New South Wales, Australia; software, Chart ver. 5) with transducers (P23 ID, Gould, Oxford, CA, USA) connected to ampli ers (Model 2238; San-ei, Tokyo, Japan). The mean arterial blood pressure (MBP), heart rate, respiratory rate, peak inspiratory pressure (PIP), mean airway pressure (MAP), and minute volume were calculated from the recorded data. Arterial blood gas analysis was performed using iSTAT (iSTAT Corp., East Windsor, NJ, USA) and tidal volume was measured at the end of each ventilatory mode.
Regarding MBV, because different ow volumes and time conditions coexist at the same respiratory rate in the same subject, the highest PaO 2 levels in the normocapnic MBV data were compared to determine the ventilation e ciency, regardless of the abovementioned conditions.
At the end of the experiment, the rabbits were euthanized with bolus injections of an overdose of sodium pentobarbital (100 mg/kg).
2.2. Experiment 2: Assessment of pulmonary protective effects 2.2.1 Experimental design The isolated perfused rabbit lung model (n = 14) was used, and the degree of pulmonary damage during ventilation was compared between MBV and IRV. The isolated perfused rabbit lungs were prepared using the method described in detail by Liu et al. (2001) [18] with minor modi cations. Brie y, the rabbits were anesthetized with pentobarbital 30 mg/kg intravenously followed by ketamine 25 mg/kg intramuscularly and anticoagulation with heparin 500 u/kg intravenously. After local anesthesia of the anterior neck and the sternum region with 1% lidocaine, tracheal intubation was performed through a tracheostomy and the rabbits were ventilated mechanically. A median sternotomy was performed, and an incision was made into the right ventricle. The rabbits were euthanized by rapidly exsanguinating whole blood (70ml) from the incision site in the right ventricle. The pulmonary artery and the left atrium were cannulated via the right and left ventriculotomies, respectively. Finally, the lungs were removed en bloc and enclosed in a humidi ed chamber. If an isolated rabbit lung is perfused with Krebs Ringer solution supplemented with 4% (w/v) albumin without the addition of red blood cells to the perfusate, it will develop pulmonary edema within 2 h [19].
Ficoll® PM70, as well as albumin, provide a normal colloidal osmotic pressure at 4% (w/v) [20]. If isolated murine lungs are perfused with Dulbecco's Modi ed Eagle's Medium containing 4% (w/v) Ficoll® P70 without the addition of red blood cells to the perfusate, an edema will form in the lung interstitium within 1 h [21]. In the current study, a pulmonary pre-edema model was created in the isolated rabbit lung via perfusion with bicarbonate-buffered PSS containing 3% (w/v) Ficoll® PM70 without the addition of red blood cells to the perfusate.
Pulmonary arterial pressure, left atrial pressure, airway pressure, and perfusate ow rate were recorded using a PowerLab system with an electromagnetic owmeter (MF-1200; Nihon Kohden, Tokyo, Japan). The left atrial pressure was set to 4 mmHg by regulating the height of a reservoir connected to the venous circuit. After a 20-min stabilization period, the baseline values of the pressure-volume (PV) curve and airway pressure were measured as described in section 2.2.2.
The 14 isolated rabbit lung preparations were randomly divided into the IRV and MBV groups. In the IRV group, rabbit lungs (n = 7; body weight, 2.5 ± 0.2 kg) were ventilated using a Harvard Ventilator 683 with a tidal volume of 8 ml/kg, a respiratory rate of 30 cycles/min, and a PEEP of 2 cmH 2 O. In the MBV group, rabbit lungs (n = 7; body weight, 2.4 ± 0.2 kg) were ventilated using the prototype ventilator described in Experiment 1 with a PEEP of 4 cmH 2 O ( rst step) and 2 cmH 2 O (second step). Before the experiment, the tidal volume was adjusted to 6 ml/kg using a test lung and measured by the water displacement method. In Experiment 2, the MBV method had a respiratory rate of 30 cycles/min, and the time allocation for one cycle had exclusively the following pattern:

Measured variables
The airway pressure was recorded using the same device described in Experiment 1, and mean airway pressure (MAP) and peak inspiratory pressure (PIP) were calculated based on the recorded values.
The in ation PV curve was measured using the quasi-static method and a syringe containing mixed gas, while ventilation and perfusion were temporarily removed [4]. This method involves the measurement of airway pressure as the lungs are gradually in ated in 5-ml steps, until a volume of 40 ml is reached. The total inhalation volume was assessed until 40 ml to prevent injuries caused by the measurement method itself. Each in ation interval was set at 15 s to obtain a plateau pressure. The de ation PV curve was not measured.
At the end of all measurements, lungs were clamped in the end-inspiratory state. The left lung was excised, and its wet weight was measured. It was dried at 60 °C in an oven for two weeks, and its dry weight was measured to determine the lung wet-to-dry ratio (W/D) using the formula: W/D = wet weight / dry weight [18]. The right lung was washed three times with 5 ml of sterile saline. The lavage uid was centrifuged at 200 × g for 10 min at 4 °C, and the cell-free supernatant was stored at -70 °C as bronchoalveolar lavage uid (BALF) for further chemical analyses. The BALF was used to measure total protein concentration and myeloperoxidase (MPO) activity. Total protein concentration was measured using BAC Protein Assay Reagent Kit (Pierce, Rockford, IL, USA). MPO activity was measured using the method of o-dianisidine dihydrochloride oxidation [22]. MPO activity was expressed as the change in optical density (∆OD) per min and per ml of BALF. W/D, total protein concentration in BALF, and MPO activity in BALF were used to determine histochemical lung injury.

Statistical analysis
Since no previous published studies were available regarding the study outcome, we could not get an idea regarding the standard deviation. Moreover, as we were interested in nding any level of difference between MVB and IRV, we measured multiple endpoints. Therefore, we used the so-called "resource equation" method, which depends on the law of diminishing returns for sample size calculation [23]. Experiment 1 was performed using a crossover design to avoid unnecessary waste of resources.
All data are expressed as the mean ± standard deviation. Prism® ver. 4 (GraphPad Software, San Diego, CA, USA) software was used for statistical calculations and gure preparations. Data were compared using Welch's t-test; p-values less than 0.05 were considered to be statistically signi cant.

Experiment 1
Nine rabbits completed Experiment 1. Blood gas analysis and circulatory parameters are shown in Table  2. There were no differences observed in either blood gas composition or circulatory dynamics between the two ventilation methods.

Experiment 2
Fourteen rabbits completed Experiment 2. Fig. 4 shows PIP and MAP values determined in Experiment 2. PIP in the IRV group was signi cantly increased after 60 min perfusion (p < 0.05). A statistically signi cant difference in PIP after 60 min perfusion was also observed between the two groups (p < 0.05).
By contrast, PIP scarcely changed in the MBV group after 60 min perfusion (p = 0.93). No signi cant differences in MAP were observed between the two groups either at baseline (p = 0.27) or after 60 min perfusion (p = 0.40).
The pulmonary PV curves of both groups at baseline and after 60 min perfusion are shown in Fig. 6, in which the pressure is expressed as the average value for the corresponding volume, since no signi cant differences in pressure values were detected between the two groups.
In comparison to the PV curves of the MBV group, those of the IRV group were slightly shifted towards higher pressures.

Discussion
Many marathon runners have adopted the 2:2 breathing pattern. We have named an intermittent positive pressure ventilation that mimics this breathing rhythm of marathon runners as MBV pattern. The ability of the MBV to take in oxygen and release carbon dioxide in healthy lungs was examined in rabbits in vivo, and the lung-protective effect of MBV on the pulmonary pre-edema model was examined using an isolated perfused rabbit lung. In healthy lungs, even if the total ventilation volume for one respiratory cycle was set lower in the MBV than in the IRV, there was no difference in MAP values between the two ventilation patterns. In healthy lungs, there were also no signi cant differences in hemodynamics and blood gas values between MBV and IRV. In the pulmonary pre-edema model, MBV was, in comparison to IRV, able to reduce the increase in PIP values, the W/D ratio at the end of the experiment was lower, and a protective effect against deterioration of the lung edema was observed.

Lung-protective ventilation strategy
In recent years, low tidal volume ventilation has secured a leading position among lung-protective ventilation strategies [6]; additionally, HFOV [8] and IRV [7] are also listed as suitable candidates.
Low tidal volume ventilation has a serious disadvantage in that it sometimes causes respiratory acidosis at pH < 7.2 [6]. In a piglet model of VILI, it was found that even with low tidal volume ventilation, high respiratory rates activate transforming growth factor β pathways and exacerbate pulmonary edema [24].
In HFOV, lung volume is secured by the MAP. If the MAP setting is too low, the lung volume decreases, and su cient oxygenation cannot be obtained; instead, setting the MAP too high can result in decreased cardiac output and increased pulmonary vascular resistance [25]. In the OSCILLATE trial [8], an RCT for ARDS, HFOV did not prove to be superior to low tidal volume ventilation. In the same trial, the high mortality rate found for ARDS patients with HFOV was caused by circulatory suppression due to a high MAP [26]. Moreover, when CO2 is retained under HFOV, the only possible strategy is to decrease the frequency and increase the tidal volume [27].
In the IRV, PIP values are kept low, but the MAP is maintained at high values [28]. The improved oxygenation capacity by IRV is attributed to the increase in MAP [27] and the occurrence of intrinsic PEEP [29] due to the decrease in expiratory time. However, in mouse experiments with high tidal volume ventilation, lung injury was rather induced by the IRV than by conventional mechanical ventilation [30].

MBV
In Experiment 1, the total ventilation volume for one respiratory cycle was set lower in the MBV than in the IRV; however, there was no difference in MAP values between the two ventilation patterns. In the MBV, the MAP could be maintained to the same extent as in the IRV. There was no difference in PaO 2 between MBV and IRV, presumably because the MAP was maintained in the MBV. Moreover, PaO 2 was also maintained in the MBV, as division of the expiration into two phases separated by a resting phase generated positive pressure at the end of the expiration, thus possibly reducing the extent of alveolar collapse. There was no difference in mean blood pressure and HR between the MBV and the IRV, since there was no difference in MAP values between the two methods, as well as no difference in the intrathoracic pressure under muscle relaxant administration. Although the MBV had lower minute volume values compared to the IRV, there was no difference in PaCO 2 between the two methods. This was because in the former, redistribution of the so-called "pendelluft" [31] occurred for the inhaled gas from the short-time constant alveoli to the long-time constant alveoli during the resting phase between the 1 st and 2 nd inspiration, decreasing the dead space. Similarly, providing an end-inspiratory pause in healthy lungs of piglets [32] and in adult acute respiratory insu ciency [33] decreases the dead space/tidal volume and PaCO 2 .
In Experiment 2, the MBV in the pulmonary pre-edema model reduced the increase in the PIP, and the W/D at the end of the experiment was lower using MBV than using IRV. We suggest that alveolar overdistension was prevented in the MBV because the total ventilation volume for one respiratory cycle was divided into two fractions. In high tidal volume ventilation of isolated perfused rat lung, active sodium transport and Na-K-ATPase activity of the alveolar epithelium are impaired, and lung edema clearance is reduced [34]. According to our results, the MBV reduces pulmonary edema.

Marathon
During intense exercise, the O 2 demand increases beyond the limit of pulmonary diffusion capacity, thus increasing the alveolar-arterial oxygen difference and inducing hypoxemia [35]. It was shown that the HR increases to 145-180/min during a marathon [36]. In an exercise that corresponds to 80% of the maximum oxygen uptake rate, the mean pulmonary artery pressure increases up to 38 mmHg in young people, while the left atrial pressure increases up to 25 mmHg [37]. Subclinical interstitial pulmonary edema is found in 17% of runners after a marathon [38]. The fact that the 2:2 breathing pattern is favored empirically in situations such as a marathon, which has a long exercise load, suggests that this respiratory technique works effectively to prevent further progression of hypoxia and pulmonary edema. Moreover, the fact that the end-tidal CO 2 is not retained during a marathon, along with the increased metabolism over a long time [13], suggests that the 2:2 breathing pattern provides a su cient capacity for CO 2 release.

Experimental models
In Experiment 1, both examined ventilation methods did not utilize PEEP to prevent their in uence on the comparison of O 2 uptake and CO 2 elimination in the MBV and IRV. In low tidal volume ventilation, atelectasis progresses [39] even if PEEP is set 2 cmH2O higher than the lower in ection point. Every time the ventilation mode was changed in Experiment 1, an alveolar recruitment maneuver was initially performed to overcome the atelectasis, and values were measured after 20 min. In early ARDS, the alveolar recruitment maneuver improves oxygenation after 3 min; however, this effect does not persist, and within 30 min, the oxygenation capacity decreases again [40]. Because arterial blood gas analysis was performed 20 min after changing the ventilation mode in the present study, we believe that the assessment of superiority between IRV and MRV were not affected by recruitment maneuvers.
In Experiment 2, the isolated perfused rabbit lung was used as an experimental model. The PV curves at the end of the experiment were not different between MBV and IRV. The pulmonary pre-edema model we employed can be considered as an early stage of lung injury, since no changes are observed in lung mechanics at early stages of lung injury, and lung mechanics are not a valid marker for early-stage lung injury [41].
In small animals, acute extreme lung stretching rapidly develops into an increased permeability lung edema. The mechanism is thought to involve the destruction of vascular endothelial cells, which induces direct contact between the basement membrane and polymorphonuclear cells, rather than involving the recruitment of in ammatory cells [2]. In Experiment 2, there was no signi cant difference in MPO activity in the BALF between MBV and IRV, suggesting that polymorphonuclear cells had not yet in ltrated into the alveolus.

Limitations
The average frequency of steps taken during a marathon is about 180 per minute [42]. In the 2:2 breathing pattern, the average respiratory rate during a marathon is 45 cycles/min, given that one cycle consists of a 1 st inspiratory period, 2 nd inspiratory period, 1 st expiratory period, and 2 nd expiratory period. The resting respiratory rate in humans is 12-20 breaths/min, whereas in rabbits, it is 32-60 breaths/min [43]. In the current study, we examined the effects of MBV with respiratory rates of approximately 30 cycles/min in rabbits. Higher respiratory rates should also be examined. O 2 uptake and CO 2 elimination in healthy lungs exposed to MBV, which should also be determined for diseased lungs in future studies.
Furthermore, since our experiments do not directly explain how the time allocation of inspiratory, expiratory, and resting phases within a single MBV cycle increases the O 2 uptake and CO 2 elimination e ciency, future studies should address this point.

Clinical application
Page 13/29 The 2:2 breathing pattern in a marathon differs signi cantly from that in MBV because the former is a spontaneous ventilation, whereas the latter is an intermittent positive pressure ventilation. Thus, before the MBV approach can be applied in the clinic, its effects have to be examined in larger animals which have breathing cycle values that are closer to those in humans during a marathon. In addition, MBV implementation requires the administration of sedatives or muscle relaxants, which needs to be taken into consideration prior to MBV clinical use.

Conclusions
The marathoners' breathing pattern ventilation method that we have developed appears to be a novel way of protecting against ventilator-induced lung injury in at-risk patients, thereby tending to decrease mortality rates in such patients. Authors' contributions: YO -analysis and interpretation of data, revising the manuscript; NO -acquisition of data, drafting the manuscript; AO -conception of the work, revising the manuscript; ST -analysis and interpretation of data, revising the manuscript; TH -analysis and interpretation of data, revising the manuscript; YIconception of the work, revising the manuscript. All authors have read and approved the manuscript and have ensured that the above-mentioned information is true. Figure 1 Comparison of respiratory rate, tidal volume, and minute volume between the IRV and MBV methods. In both the MBV and IRV methods, the inspiratory and expiratory phases are determined as one cycle, and tidal volume is determined to be the total ventilation volume during one cycle. In both the methods, the unit of respiratory rate is cycles/min. Signi cant differences are observed in tidal volume and minute volume between the two methods (*p < 0.01). IRV, inversed ratio ventilation; MBV, marathoners' breathing pattern ventilation Figure 2 Comparison of peak inspiratory pressure (PIP) and airway pressure between the IRV and MBV methods. A signi cant difference is observed in the PIP between the two methods (*p < 0.01). IRV, inversed ratio ventilation; MBV, marathoners' breathing pattern ventilation Figure 3 Typical waveforms of the airway pressure for IRV and MBV methods in the same rabbit. In the MBV method (lower panel), the airway pressure transiently decreases immediately after the end of the 1st inspiratory period and transiently increases immediately after the end of the 1st expiratory period. IRV, inversed ratio ventilation; MBV, marathoners' breathing pattern ventilation Pulmonary PV curves at baseline and 60 min after perfusion in IRV and MBV groups. Pressure is expressed by the average value against each volume, since no signi cant differences are observed between groups at each pressure. Compared with the PV curves in the MBV group, the curves in the IRV group have slightly shifted to the high-pressure side. IRV, inversed ratio ventilation; MBV, marathoners' breathing pattern ventilation; PV, pressure-volume.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.