The saturation of haemoglobin with oxygen in the pulmonary artery is known as the mixed venous oxygen saturation (SvO₂), which reflects the balance between the amount of oxygen delivered to the tissues and how much is used. It enables an estimate of the oxygen supply/demand balance to be made and hence enhances our comprehension of physiological concepts of hemodynamics and tissue oxygenation in critically ill patients. However, the routine measurement of SvO₂ requires the placement of a pulmonary artery catheter (PAC), which may not always be feasible. Furthermore, a substantial review of literature suggests at present that the use of PAC may lead to an overall increase in morbidity and mortality in critically ill patients 12, stimulating the quest for a micro-invasive tool for assessing SvO₂.

Pulse oximetry has been widely adopted in anesthesia and critical care medicine to provide noninvasive information about arterial oxygen saturation (SaO₂). Several studies have demonstrated that oximeters placed in deep, vessel-rich areas such as the esophagus 3, pharynx 4, and trachea 5 seemed to provide more accurate readings than superficial oximetry. The tissue being sampled was once assumed to be the surrounding mucosa 3, but recent studies have shown that the signals were derived primarily from deeper tissues, such as underlying large vessels around the esophagus and trachea 56.

The pulmonary artery lies close to the bronchus, with nothing but some connective tissues between them, raising the possibility that an appropriately located and directed bronchial oximetry probe might be able to derive oximetry readings from mixed venous blood (Figure 1). The present study was undertaken to test the feasibility of measuring SvO₂ through the left main bronchus (SpO_2trachea), and to compare SpO_2trachea with oxygen saturation from pulmonary artery samples (SvO_2blood) in a healthy hybrid pig to improve our understanding of the hypothesis that bronchial oximetry readings are derived primarily from the pulmonary artery, not from the tracheal mucosa. Furthermore, the stability and accuracy of SpO_2trachea were evaluated by assessing the impact of altered cardiac output on tracheal SpO₂ in hemorrhagic shock status.

SvO₂ reflects the balance between oxygen delivery and demand. It decreases when oxygen delivery has been compromised or systemic oxygen demands have exceeded supply. Its ability to give a real-time indication of tissue oxygenation makes it a preferred parameter for monitoring the adequacy of hemodynamics. In comparison with traditional parameters such as arterial oxygen saturation and cardiac output, SvO₂ allows a more precise understanding of the adequacy of cardiac and pulmonary function. Declines in SvO₂ precede the onset of inadequate myocardial function, shock, or the development of arrhythmias, even though vital signs may be normal. Its use as an end point for determining the adequacy of hemodynamics (blood pressure, cardiac output/cardiac index), measurement of right to left shunt, and prediction of potential hemodynamic instability makes this parameter invaluable for the knowledgeable clinician. There is now evidence that the timing of diagnostic and therapeutic intervention using this technology may be a critical determinant of outcome 7.

The PAC, otherwise known as the Swan–Ganz catheter, was developed by cardiologists HJC Swan and William Ganz in 1970. It is a flexible balloon-tipped flow-directed catheter that, when inserted via central venous access, can be guided into a branch of the pulmonary artery. Its ability to provide continuous measurements of SvO₂ in critically ill patients makes its use invaluable in the provision of quality medical care. However, controversy surrounding the efficacy and safety of the PAC has been going on for many years. The complications can be categorized as those of the initial venous cannulation (subclavian or carotid artery laceration, pneumothorax, thoracic duct laceration, phrenic nerve injury, and air embolism) and those due to the catheter itself (ie, arrhythmias, infection, valvular damage, thrombosis, pulmonary infarction, and rupture of the pulmonary artery). At the same time, the device requires a trained operator and is time-consuming. Moreover, it is expensive, bringing high healthcare costs.

There is therefore a powerful need for a method to measure SvO₂ more safely. Other researchers have developed the technique of deriving oximetry readings of arterial blood through the trachea, or right and left ventricular oximetry through the esophagus 58. The pulmonary artery is known to lie just proximal to the left bronchus. This evaluation of the anatomy made it practical to measure oximetry readings from the mixed venous circulation through the left main bronchus. However, so far no such studies have been reported. The present study establishes the first investigation to assess SvO₂ microinvasively according to the above anatomic and technological bases.

Waterproofing is crucial for the proper function of oximeters in the humid environment of the trachea. Our experiment employed medical silica gel as a surface coat, because silica gel is waterproof and is nontoxic to humans. It can solidify fully at normal temperature, thus avoiding potential damage to the oximeter caused by thermal treatment. Moreover, it can endure a certain level of friction and tension after solidification. Because the pulmonary artery and the bronchus run nearly parallel, with sufficient overlapping area in the longitudinal direction, the light emitter and sensor of the oximeter are affixed along the same direction on the tracheal tube. As a result, the probe turned from a penetrating model (the light emitter and sensor being aligned opposite each other) into a reflecting model (the two terminals lying side by side). Experimental results indicate that the optimum distance between the emitter and sensor should be close to 1 cm. If the two terminals are too close, transmitting signals will be attenuated, which will affect the stability and accuracy of the data. Conversely, an increase in distance will negatively affect the reception efficiency of the infrared reflection signal.

Despite the above changes to the oximetry probe, high-quality signals were still available. We found that SpO_2refit of the tongue was accurate at different inspiratory oxygen concentrations, in different head and neck positions, and over a prolonged period, suggesting good stability and sensitivity of the refitted probe.

The ability to localize the oximetry probe accurately is pivotal to the experiment. An experiential position 2 to 3 cm deep in the bronchus and an orientation of 15 to 20° left-leaning to the midline for the tracheal tube was found in our pilot study. To ensure that the tube was advanced to the optimal location, the animal should be fixed beforehand, and the position of the tube should be confirmed by electrocardiography.

Supported by the foregoing statement, our data showed that the reading of SpO_2trachea was close to SvO_2blood in stable physiological situations at 10 to 60 cmH₂O cuff pressure. The readings obtained at zero cuff pressure were probably low because of a lack of contact between the probe and the trachea. The SpO_2trachea was thought not to be derived primarily from the tracheal mucosa, because tracheal mucosal perfusion ceases when the intracuff pressure exceeds 50 cmH₂O, and there was no decrease in the accuracy of SpO_2trachea with increasing intracuff pressure. The blood flowing through the left pulmonary artery was speculated to be the mass of tissue sampled by the tracheal oximetry probe. At the same time, our study showed that SpO_2trachea was consistent with SvO_2blood in low cardiac output status during the hemorrhagic shock period. This measurement demonstrated that the precision of measuring SvO₂ through the left main bronchus was not influenced in a pathological state, suggesting great reliability of this technique in operation and for patients in intensive care units. Although ventilation with a double-lumen tube is itself an invasive procedure, its advantage in causing much fewer lesions than PAC cannulation, and in avoiding the multiple complications that accompany the PAC device, makes this technique particularly appropriate for critically ill patients.

However, several limitations of the present investigation should be noted. First, our device was homemade, with the oximeter probe fixed to the endoscope by tape. Damage to the mucosa of the trachea is possible, and accidental inhalation would occur if the probe exfoliated. Furthermore, to reduce complications, a small tracheal tube and thin wire were required. However, it would be possible to incorporate the oximeter within the cuff and the wire within the tube and in so doing to reduce the complication of damage or accidental inhalation and allow a larger tube to be used to decrease the risk of trauma. Secondly, there were difficulties with locating the probe in the left bronchus. In addition to adjusting the tube repeatedly, ultrasound is required to confirm the position of the oximeter. The technique for location merits further investigation.

Measurement of SpO₂ via the left main bronchus is feasible and provides similar readings to SvO_2blood in both hemodynamically stable status and hemorrhagic shock status. Tracheal oximetry readings are not derived primarily from the tracheal mucosa. This technique is capable of providing continuous and microinvasive measurements of SvO₂ despite the difficulty in achieving proper location of the probe. Further improvement is required for convenience of operation.

• An appropriately located and directed bronchial oximetry probe is able to derive oximetry readings from the pulmonary artery, because the artery lies in close proximity to the bronchus with only some connective tissues in between, thus providing a microinvasive tool for the assessment of mixed venous oxygen saturation (SvO₂).

• The mixed venous oxygen saturation via the left main bronchus (SpO_2trachea) was thought not to derive primarily from the tracheal mucosa, because it was lower than the oxygen saturation from pulmonary artery samples (SvO_2blood) at zero cuff pressure.

• SpO_2trachea was the same as SvO_2blood in hemodynamically stable status.

• Sp_O2trachea also provides similar readings to SvO_2blood in hemorrhagic shock status, suggesting great reliability of this technique in operation and for patients in intensive care units.

PAC = pulmonary artery catheter; SaO₂ = arterial oxygen saturation; SpO_2origin = pulse oximetry obtained with the original oximetry probe; SpO_2refit = pulse oximetry obtained with the refitted oximetry probe; SpO_2trachea = SvO₂ through the left main bronchus; SvO₂ = mixed venous oxygen saturation; SvO_2blood = oxygen saturation from pulmonary artery samples.

The study was funded by 'The Third Period of Hundred People Project, Shanghai City'.

XW conceived the study, participated in the design and execution of the study, and finalized and revised the manuscript. YZ participated in the animal experiments, performed the statistical analysis, and was involved in drafting the manuscript. JT participated in study design, interpretation of the results, and writing the manuscript. ZW and ZP participated in the animal experiments. All authors read and approved the final manuscript.