In this clinical trial, we evaluated whether intravenous magnesium sulfate administered after induction of anesthesia could reduce intraoperative hypoxemia in patients undergoing thoracotomy with OLV. The principal finding was that magnesium sulfate was associated with a significantly lower incidence of intraoperative hypoxemia than with normal saline, with events occurring in 20.6% of patients in the intervention group and 45.7% in the control group. The intervention was also associated with a lower total hypoxemia-duration burden after adjustment for smoking status, although the duration and number of hypoxemic episodes among patients who developed hypoxemia did not differ significantly between groups. In the time-to-event analysis, the protective association was significant in the unadjusted model but was attenuated after adjustment for smoking; smoking itself was associated with a higher hazard of hypoxemia.
The lower incidence of hypoxemia in the magnesium group should be interpreted in the physiological context of OLV. During OLV, an obligatory intrapulmonary shunt develops because only one lung is ventilated. Hypoxemia during OLV is primarily related to ventilation–perfusion mismatch, with additional contributions from patient position, anesthetic technique, cardiac output, pre-existing lung disease, and modulation of hypoxic pulmonary vasoconstriction (
3). Because oxygenation during OLV is multifactorial, our findings should not be interpreted as evidence of a single mechanism, but rather as a clinically observed reduction in hypoxemia. Direct comparison with previous magnesium studies is limited because earlier clinical studies mainly evaluated arterial oxygenation, pulmonary mechanics, postoperative pulmonary function, or analgesic outcomes rather than SpO₂-defined intraoperative hypoxemia during open thoracotomy with OLV (
14,
16-
18). In patients with moderate COPD undergoing general anesthesia, Ahmed et al. reported that magnesium sulfate preserved partial pressure of arterial oxygen (PaO₂) and PaO₂/FiO₂ and improved airway resistance, dead space, airway pressures, and dynamic compliance (
16). Similarly, Mowafi et al. reported favorable effects on PaO₂/FiO₂, dynamic compliance, and dead space in morbidly obese patients undergoing bariatric surgery, although the authors considered the clinical relevance limited (
18). In thoracic surgery, Sohn et al. found that intraoperative magnesium sulfate improved early postoperative forced expiratory volume in 1 second (FEV₁) and forced vital capacity (FVC) after VATS and reduced requirements for rocuronium and patient-controlled analgesia (
17). Our study extends this evidence by focusing on immediate intraoperative oxygenation during open thoracotomy with OLV rather than on postoperative pulmonary function or oxygenation during two-lung ventilation.
Several mechanisms may explain the observed reduction in hypoxemia. Magnesium acts as a calcium antagonist and modulates neuromuscular transmission and NMDA receptor activity, which may influence anesthetic requirements, muscle relaxation, and pain-related respiratory mechanics (
12). Improved dynamic compliance and reduced airway resistance could theoretically reduce dependent-lung atelectasis or improve ventilation distribution during OLV. Magnesium may also reduce perioperative requirements for analgesics and neuromuscular blockers, which could indirectly support respiratory mechanics, although this pathway is more relevant to postoperative pulmonary recovery than to early OLV hypoxemia (
11). Importantly, oxygenation during OLV is strongly affected by intrapulmonary shunt and pulmonary vascular responses; therefore, the favorable association observed here should not be interpreted as proof that magnesium directly improved hypoxic pulmonary vasoconstriction or reduced shunt. A systematic review and network meta-analysis of pharmacologic interventions during OLV showed that anesthetic and adjuvant agents can modify intrapulmonary shunt and PaO₂/FiO₂, but the certainty of the evidence varied, and magnesium sulfate was not established as a standard shunt-modifying therapy (
19).
Among patients who developed hypoxemia, the median duration was numerically shorter in the magnesium group than in the control group (2.00 versus 3.00 minutes), but the difference was not statistically significant. However, when the entire cohort was considered and patients without hypoxemia were coded as having zero minutes of hypoxemia, the intervention was associated with a significantly lower total hypoxemia-duration burden after adjustment for smoking status. This distinction is important because an intervention that prevents hypoxemia in some patients may reduce the overall burden even if the duration of episodes among those who still become hypoxemic is not markedly different. Recent evidence supports considering duration and severity together. In a large retrospective cohort of 2748 thoracic surgery patients, Zorrilla-Vaca et al. found that cumulative hypoxemia, defined as the area under the curve for SpO₂ < 90%, was independently associated with severe postoperative pulmonary complications, with an odds ratio of 1.21 per 1% × hour increase (
20). In our cohort, hypoxemic episodes were generally short, with median durations of only a few minutes, so it remains uncertain whether the observed reduction in burden translates into postoperative benefit.
Smoking status was an important modifier in the present analysis. Although smoking prevalence did not differ statistically at baseline, smoking was more frequent in the control group and was associated with a higher hazard of hypoxemia in the adjusted Cox model. This finding is consistent with evidence that smoke exposure can impair oxygenation during OLV. Ozkan et al. found that patients exposed to secondhand smoke who underwent lobectomy via open thoracotomy had substantially lower PaO₂ after 10 minutes of OLV than nonsmokers (186.7 ± 66 mmHg versus 285.5 ± 90 mmHg) and also had higher partial pressure of arterial carbon dioxide (PaCO₂) values (
21). However, the relationship between smoking, position, and oxygenation during OLV is not consistent across studies. In a pilot study, Sabetian et al. reported no significant differences in PaO₂ or SpO₂ between smokers and nonsmokers under supine and lateral ventilation conditions, although position-related changes in PaO₂ were observed (
22). In our study, attenuation of the treatment effect after adjustment for smoking suggests that the smoking imbalance may have contributed to the apparent protective association, but the direction of the effect remained protective.
The lower MAP and heart rate observed in the magnesium group are consistent with the known pharmacologic effects of magnesium sulfate. Magnesium can attenuate catecholamine release and cardiovascular responses to airway stimulation, and it may cause cardiovascular depression through calcium-channel antagonism (
11,
23). Kotwani et al. similarly reported that intravenous magnesium sulfate attenuated heart rate and systolic blood pressure responses to laryngoscopy and intubation, with 30 mg/kg providing adequate cardiovascular control without reported complications (
24). In our study, mean MAP was approximately 3.5 mmHg lower and mean heart rate was approximately 7.9 beats/min lower in the intervention group. Although these differences were statistically significant, they should be interpreted as hemodynamic effects of magnesium sulfate rather than evidence of improved hemodynamic stability. Importantly, no intervention-related adverse events required discontinuation of the study infusion.
These findings should be interpreted in the context of established OLV management strategies. Protective ventilation, appropriate PEEP, recruitment maneuvers, accurate lung isolation, and correction of device malposition remain central to preventing and treating hypoxemia during thoracic surgery. Blank et al. emphasized that low tidal volume alone may not be sufficient during OLV unless accompanied by adequate PEEP, and that driving pressure and ventilatory management are relevant to postoperative outcomes (
25). Therefore, magnesium sulfate should be considered only as a potential adjunct to standard OLV management.
Several limitations should be considered. First, the modest sample size may have limited the statistical power and precision of the estimated treatment effect, particularly in adjusted and time-to-event analyses. This limitation is reflected in the relatively wide confidence interval around the adjusted hazard ratio for hypoxemia. Second, the smoking imbalance between groups may have influenced the treatment-effect estimate. This is particularly important because the protective association in the Cox model was no longer statistically significant after adjustment for smoking. Importantly, arterial blood gas measurements were not obtained systematically; therefore, PaO₂/FiO₂ or another FiO₂-normalized arterial oxygenation index was unavailable. Because FiO₂ was adjusted according to clinical requirements during OLV, mean SpO₂ should be interpreted only as a descriptive saturation measure, not as a direct measure of oxygenation efficiency. Consequently, the present study cannot determine whether magnesium sulfate improved arterial oxygenation or gas exchange independently of changes in inspired oxygen concentration. Our conclusions are therefore limited to the occurrence and temporal burden of SpO₂-defined hypoxemia. Third, the study did not directly measure airway resistance, lung compliance, dead space, intrapulmonary shunt, or serum magnesium concentration. Fourth, although pulmonary recruitment was available for group-level analysis, detailed FiO₂ exposure and the frequency of individual rescue measures, such as FiO₂ escalation, PEEP adjustment, and CPAP use, were not captured at sufficient resolution for reliable between-group comparison. These variations in rescue management may have influenced the recorded occurrence and burden of hypoxemia. Fifth, postoperative pulmonary complications were not assessed as outcomes, so the study cannot determine whether reduced intraoperative hypoxemia translated into improved postoperative recovery. Sixth, detailed event-level safety data, including the frequency of hypotension and bradycardia, vasopressor or atropine administration, and recovery- or neuromuscular-related adverse events, were not systematically quantified, limiting a more detailed assessment of the clinical significance of the observed hemodynamic differences. Finally, the cohort included heterogeneous surgical indications and was limited to a single center, which may reduce generalizability to VATS, robotic thoracic surgery, pneumonectomy, or centers using different OLV protocols.
In conclusion, intravenous magnesium sulfate administered after induction of anesthesia was associated with a lower incidence of intraoperative hypoxemia, the prespecified primary outcome, in patients undergoing thoracotomy with OLV. Secondary analysis also demonstrated a lower cumulative hypoxemia-duration burden across the full cohort, whereas episode number, hypoxemia duration among affected patients, and the smoking-adjusted time-to-first-hypoxemia association were not statistically significant. Given the modest sample size, single-center design, and baseline imbalance in smoking status, these findings should be interpreted cautiously and require confirmation in larger multicenter randomized trials.