Noninvasive spectrophotometric oximetry was sensitive to changes in oxygen saturation in pigs and correlated with intravitreal pO
2 measurements and with femoral artery pO
2. This is in agreement with previous studies, which have shown a significant correlation between systemic and retinal oxygen saturation in systemic hypoxemia due to Eisenmenger syndrome
10 and chronic obstructive pulmonary disease,
11 as well as during induced systemic oxygen saturation changes on retinal saturation in healthy subjects.
4 Furthermore, studies using multispectral scanning laser ophthalmoscopy have found a strong correlation between femoral artery saturation and retinal artery saturation, suggesting that comparison of these values is a reasonable method for calibrating retinal oximetry.
12
Using porcine experiments offers the opportunity to test noninvasive oxygen saturation measurements at a much wider range of systemic oxygen saturation than is possible in humans.
The levels of inspiratory oxygen percentage used in the present study were similar to those used on human subjects by Beach et al.
5 in 1999 (5%–100% and 8%–100%, respectively), but resulted in considerably lower femoral artery saturation values (Table). This may partly be explained by the difficulties in ventilation of the anesthetized pigs, but pigs have also been shown to have lower oxygen levels than humans in conscious states and present lower oxygen saturation values at comparable levels of pO
2, as determined by the difference in human and porcine oxyhemoglobin dissociation curves.
9,13,14 In comparison to the human curve, the porcine dissociation curve is shifted to the right, resulting in a saturation difference between humans and pigs at the same pO
2 that is more pronounced at lower levels of pO
2 (
Fig. 4). The calculated mean difference is 13.9 percentage points for blood gas measurements made at 10% inspiratory oxygen, 2.8 percentage points for measurements made at room air, and virtually nonexistent at 100% inspiratory oxygen.
Pigs present a higher intraindividual variability in retinal vessel oxygen saturation compared to retinal saturation values in human subjects and a lower overall retinal saturation (Table,
Fig. 2B).
4,5,15 The higher variability may be due to the different optics of the porcine eye, which is shorter on the anteroposterior axis and has a substantial level of corneal astigmatism, making it difficult to get uniform focus over the porcine retina.
16,17
Vessel diameter on retinal images has previously been shown to have an artifactual effect on spectrophotometric retinal oximetry and is corrected for in the human version of Oxymap Analyzer.
4,15 The mean vessel diameter on our porcine fundus images taken at normal air (Table) is approximately two times greater than values reported for first-degree human retinal vessels.
18
Reports of normal retinal vessel diameter values in porcine eye are scarce. A study by Jeppesen et al.
19 suggests that first-degree retinal arteries are approximately 140 μm in diameter in enucleated porcine eyes. Human studies have reported an arterial diameter of approximately 120 μm in healthy subjects.
20,21 This could suggest that porcine retinal arteries are not significantly larger than human retinal arteries in vivo and that the observed difference is mainly due to a different level of magnification.
Porcine retinal vessels lie superficially in the inner retina, whereas they lie more deeply embedded in the nerve fiber layer in humans.
22 The superficial position of the porcine retinal vessels may cause an inconsistency in the paths of incident and transmitted light, especially since fundus photography of pigs usually requires that the fundus camera be tilted and that images be taken at a superior angle.
The generally low retinal venous saturation values may be due to a combination of the low femoral saturation values and the calibration method used for retinal saturation. Highly variable raw ODR values from retinal arteries were compared to femoral arterial values. Given the relatively small sample size, the linear regression is sensitive to random effects.
This is perhaps most prominent at 21% inspiratory oxygen fraction, where the reported arterio–venous difference is considerably higher than at 10% or 100% oxygen. Theoretically, one could expect the arterio–venous difference to be higher at room air than either at 10% oxygen, where the total amount of oxygen is limited, or at 100%, where the concentration of oxygen dissolved in plasma and the contribution of the choroid keep the hemoglobin oxygen saturation high even on the venous side. This physiological effect is most likely intensified by random effects of measurement noise.
One possible source of measurement error that should be taken into consideration is the anesthesia. Previous studies suggest that propofol causes a reduction in systemic blood pressure, heart rate, and cardiac output, mainly through peripheral vasodilatation.
23–25 Although altered, these vital signs are stable during propofol anesthesia and therefore unlikely to have an effect on our results.
In conclusion, spectrophotometric snapshot oximetry is sensitive to systemic oxygen saturation changes over a wide range in pigs, but the variability between subjects is high. Pigs have a lower systemic oxygen saturation than do humans at comparable levels of inspiratory oxygen fraction, which results in low retinal oxygen saturation values. The difference in optical properties between porcine and human eyes makes direct comparisons of measurements difficult and reduces the value of porcine eyes as a model for spectrophotometric oximetry in humans.