Maximal FiO₂ values (>95%) have been administered to horses to maintain or improve the arterial partial pressure of oxygen (PaO₂) and tissue oxygenation since the advent of inhalant anesthesia [1]. Maximal FiO₂ concentrations are used in the belief that their use optimizes arterial O₂ content (CaO₂) and O₂ delivery (DO₂) to tissues providing cardiac output (Q) is maintained [2]. Lower than expected (subnormal) PaO₂ values are frequently encountered during equine inhalant anesthesia despite the use of maximal FiO₂ [3–6]. Subnormal PaO₂ values usually develop during the first 30 to 90 min of anesthesia, continue for the duration of anesthesia, and resolve within 60 min of recovery to standing [4]. The 5 mechanisms producing subnormal PaO₂ values are decreased FiO₂, diffusion impairment, vascular shunting (true shunt), hypoventilation, and ventilation-perfusion mismatching (pseudo shunt) [4]. Decreased FiO₂, diffusion impairment, and vascular shunts are unlikely to contribute to subnormal PaO₂ during inhalation anesthesia in normal horses. Hypoventilation [arterial partial pressures of carbon dioxide (PaCO₂) >5.92 kPa (45 mm Hg)] is a common occurrence during equine anesthesia because virtually all anesthetic drugs cause respiratory depression. Hypoventilation is easily corrected by mechanical ventilation [7]. The use of mechanical ventilators, mechanical ventilation and increased FiO₂ probably reduces the incidence of subnormal PaO₂ values however a significant number of horses continue to experience subnormal PaO₂ despite normocapnia and increased FiO₂ [2, 5, 6].

The major cause of subnormal PaO₂ values in anesthetized horses is ventilation-perfusion mismatching within the lung parenchyma [8]. Functional residual capacity and residual volume are reduced in recumbent horses during inhalant anesthesia [3]. The distribution of ventilation is shifted to non-dependent areas of lung and altered by changes in diaphragmatic function and mechanical ventilation [9–13]. The shifting of the distribution of lung ventilation uncouples ventilation and perfusion because lung perfusion remains primarily caudal-dorsal regardless of body posture [14–16]. Ventilation-perfusion mismatching is exacerbated by atelectasis because collapsed alveoli continue to be perfused with little or no gas exchange. The two primary causes of atelectasis are compression (from overlying abdominal contents) and absorption (collapse due to absorption of the alveolar gas). Compression atelectasis is more pronounced in dorsally recumbent horses compared to other species due to its long, sloping diaphragm [17]. The development of compression atelectasis can be partially offset or delayed by initiating mechanical ventilation early during the anesthetic period but predicted PaO₂ values are not always achieved [2]. Ventilatory recruitment maneuvers including the use of positive end expiratory pressure (PEEP) have been investigated in horses, however improvements are inconsistent and only occur when PEEP values that compromise Q are employed [12, 18, 19].

The relationship between PaO₂ and hemoglobin saturation varies with blood temperature, pH and intracellular factors [13, 20]. Although equine hemoglobin is more than 90% saturated at PaO₂ values greater than 9.2 kPa (70 mm Hg), there is rapid desaturation at values below 7.9 kPa (60 mm Hg) [13]. One author has provided data from anesthetized horses demonstrating that DO₂ decreases when PaO₂ decreases, and that PaO₂ values less than 5.6 kPa (50 mm Hg) are associated with decreases in venous partial pressure of oxygen (PvO₂), an indicator of the adequacy of DO₂ [21].

The recognition that atelectasis contributes to ventilation-perfusion mismatching, intrapulmonary shunt and subnormal PaO₂ values has raised questions about the use of maximal FiO₂ [22, 23]. Absorption atelectasis is exacerbated by administering maximal FiO₂ because O₂ is more readily absorbed than other, less soluble, gases such as nitrogen. Further, studies in humans suggest that the use of maximal FiO₂ promotes alveolar damage, reduces cardiac index, and increases peripheral vascular resistance [24, 25]. The use of 0.5 FiO₂ could reduce absorption atelectasis and augment alveolar O₂ tensions (compared to ambient air) thereby improving gas exchange and DO₂. The purpose of this study was to compare the effects of 0.5 FiO₂ and >0.95% FiO₂ on pulmonary gas exchange, shunt fraction (Qs/Qt) and DO₂ in dorsally recumbent horses during inhalant anesthesia. Our hypothesis was that 0.5 FiO₂ would reduce ventilation-perfusion mismatching and increase the fraction of pulmonary blood flow that is oxygenated, thus improving CaO₂ and DO₂.

All horses completed all phases of the experiment. Initial flow rates for the 0.5 FiO₂ group were 3 L/min of O₂ and 6 L/min of air. Flow rates required to maintain 0.5 FiO₂ when the total flow rate was set at 9 mL/kg bwt/min ranged from 1.8 to 3 L/min for O₂ and 3 to 4 L/min for air. All horses were administered dobutamine during each trial (dose range 0.5-1 μg/kg/min). One horse developed atrial fibrillation 30 min after induction of anesthesia during the second procedure (>0.95 FiO₂). Atrial fibrillation continued throughout anesthesia and an electrocardiogram confirmed normal sinus rhythm 1 hr after the horse stood. Blood pressures and heart rates in this horse were similar to other horses in this treatment group and ranged from 30 to 36 bpm prior to the onset of atrial fibrillation and from 35 to 56 bpm after the onset of atrial fibrillation.

The use of 0.5 FiO₂ in dorsally-recumbent isoflurane-anesthetized horses did not reduce Qs/Qt or Vd/Vt compared to >0.95 FiO₂. Arterial partial pressures of O₂ and P_(A-a)O₂ were higher in horses breathing >0.95 FiO₂ but DO₂ was not different between groups during 3 hr of anesthesia or after anesthesia. Reduction of FiO₂ from >95% to 50% did not improve pulmonary gas exchange or oxygenation.

The results of this study support the hypothesis that the development of Qs/Qt mismatching from compression atelectasis is the most likely cause of impaired arterial oxygenation during inhalant anesthesia in the horse [8]. Shunt fraction, P_(A-a)O₂, and Vd/Vt increased in both groups after induction to anesthesia in our study. Our results differ from two previous reports that found a greater Qs/Qt in horses that breathed >0.95 FiO₂ compared to horses that breathed air during intravenous anesthesia [8] and an increase in Vd/Vt when >0.85 FiO₂ was compared to 0.35 FiO₂ in spontaneously breathing, halothane-anesthetized horses [26]. In agreement with our results, the administration of varying helium-oxygen mixtures did not change Qs/Qt but resulted in progressive increases in P_(A-a)O₂ and PaO₂ as FiO₂ increased [27]. Differences between our results and those of others may be due to the FiO₂ selected (0.5 versus 0.2 to 0.35), the effects of intravenous and different inhalant anesthetics (isoflurane versus no inhalant or halothane), the duration of anesthesia, the mode of ventilation (controlled ventilation versus spontaneous ventilation), or body position (dorsal recumbency versus lateral recumbency). Positive pressure ventilation and inhalant anesthetics alter the distribution of ventilation in the anesthetized horse but most have accepted this limitation of the two techniques in the interest of normalization of PaCO₂ and the maintenance of consistent anesthetic depth [9–12]. Body position is a primary determinant of oxygenation in anesthetized horses with dorsal recumbency consistently associated with larger decrements in pulmonary function and gas exchange [2, 9, 28, 29]. The cause for the decrements in pulmonary function has classically been attributed to a gravity-dependent shift in blood flow and a shift in ventilation to less dependent lung zones [30]. This description has been challenged by several studies that have demonstrated that only small shifts in blood flow occur in anesthetized horses after changes of body position [15, 29, 31]. We purposefully selected overweight horses and placed them in dorsal recumbency for an extended period in order to maximize the potential for changes in FiO₂ to improve gas exchange and reduce hypoxemia [5, 32–34].

Similar to others we confirmed that >0.95 FiO₂ is associated with greater PaO₂ and P_(A-a)O₂ values than the those produced by 0.2 to 0.5 FiO₂ and that CaO₂ and DO₂ may not be markedly different if arterial hemoglobin remains fully saturated [23, 26, 27]. Step-wise increases in FiO₂ to titrate PaO₂ and CaO₂ have been proposed [27] but such an approach requires multiple gas sources, the ability to determine FiO₂ and frequent sampling of arterial blood for arterial blood gas analysis. The differences in PaO₂ and P_(A-a)O₂ between the two groups in the face of no difference in calculated Qs/Qt is a function of the sigmoid shape of the oxyhemoglobin saturation curve and highlights the importance of considering both PaO₂ and CaO₂ when assessing the significance of changes in the ranges of PaO₂ and Qs/Qt routinely encountered in the anesthetized horse [35].

The incidence of subnormal PaO₂ values and hypoxemia in dorsally recumbent anesthetized horses is a recurring problem during equine anesthesia and has been associated with body shape, body weight, low pulse pressures, emergency case status, and male gender [5, 6, 33, 34]. It has been suggested that the application of controlled ventilation immediately after induction to anesthesia is more likely to maintain PaO₂ during anesthesia but controlled ventilation may decrease Q and thus DO₂ [2, 21]. Additional proposed methods of improving oxygenation in anesthetized horses include PEEP, inhalation of aerosolized bronchodilators such as albuterol, selective mechanical ventilation of dependent lung regions and breathing helium-oxygen mixtures [18, 27, 36, 37]. Positive end expiratory pressures of at least 10 cm of water may improve CaO₂ without changing Qs/Qt but Q is reduced unless fluids and cardiac inotropes (ex. dobutamine) are administered [17, 18]. Aerosolized albuterol is reported to increase PaO₂ but its effects are inconsistent in the authors' and other's experience regarding efficacy and consistency of effect [36, 38]. Selective mechanical ventilation combined with selective PEEP consistently increases DO₂, however it requires a surgical procedure (tracheostomy) and multiple ventilators [37].

The lowest individual PaO₂ measurements in our study ranged from 6.32 to 13.55 kPa (48 to 103 mm Hg) in the 0.5 FiO₂ group and from 9.21 to 42.1 kPa (70 to 320 mm Hg) in the >0.95 FiO₂ group. All horses in 0.5 FiO₂ group had PaO₂ values below 13.16 kPa (100 mm Hg) at some time during the anesthetic period and 2 of 5 horses had PaO₂ values below 7.89 kPa (60 mm Hg). Only one of 5 horses in the >0.95 FiO₂ group had a PaO₂ below 13.16 kPa (100 mm Hg). All horses in both trials had PaO₂ values below 7.89 kPa (60 mm Hg) ten minutes after discontinuation of anesthesia but CaO₂ did not change from preanesthetic levels and was not different between groups at any time suggesting there was no apparent advantage with either technique. Further, recent studies in laterally recumbent isoflurane anesthetized horses have determined that breathing >0.95 FiO₂ for 90 min does not affect erythrocyte membrane dynamics or structure, blood viscosity or muscular perfusion and that breathing room air (0.21 FiO₂) decreases skeletal muscle oxygenation [39].

The results of our study, although limited by a relatively small sample size, are similar and consistent with those of others [23, 26]. One of our horses developed atrial fibrillation, which could have influenced the development of V/Q and our results. The range of HR and Q for the horse that developed atrial fibrillation was similar in both trials so we do not believe that it affected our results since the development of atrial fibrillation in the absence of cardiac disease, is not necessarily associated with the deterioration of hemodynamics [40]. We did not have the capability of performing the multiple inert gas elimination technique which would have allowed potential differentiation between pulmonary units with low and high V/Q and shunt [23, 41, 42]. Regardless, the ability to more precisely identify V/Q matching and shunt would not have changes our results. Decreases in atelectasis, Qs/Qt, and P_(A-a)O₂ are interesting but unless they result in improvements in PaO₂, CaO₂, and DO₂, their impact is suspect. The risks of inhaling >0.95 FiO₂ are well documented in other species [43–45] but we could find no reports of pathology associated with the use of >0.95 FiO₂ in horses anesthetized for clinically relevant durations. To the contrary, isoflurane anesthetized horses breathing >0.95% O₂ for 90 min demonstrated minimal damage from reactive oxygen species and no alterations in muscular perfusion [39]. The use of submaximal FiO₂ requires additional equipment and more stringent monitoring thus increasing expense. In the absence of evidence of benefit, the authors do not recommend the use of submaximal 0.5 FiO₂ in anesthetized horses.

The purpose of breathing increased FiO₂ during equine anesthesia is to partially offset the loss of the adaptive mechanisms that are initiated by hypoxemia in awake horses such as increases in ventilation, Q and contraction of the spleen [3]. These mechanisms are obtunded or eliminated during inhalant anesthesia in horses. Our study suggests that decreasing FiO₂ to 0.5% from maximal levels does not improve pulmonary gas exchange or DO₂ during inhalant anesthesia in horses. Further studies investigating the benefits of alternative gas mixtures (ex. helium- O₂) and modes of ventilation in conjunction with changes in FiO₂ are warranted.