Survey of regional cerebral oximetry in thoracic aortic replacement surgery: comparison of near-infrared cerebral oximeters

[Background and aims of the study]
It is reported that incidences of cerebral infarction during and of neurological deficit following aortic surgeries are 1-3% and 35%, respectively. The risk of cerebral ischemia is high especially in total arch replacement (TAR) with deep hypothermic circulatory arrest (DHCA). Near infrared spectroscopy (NIRS), used in near-infrared cerebral oximeters such as INVOSTM-4100 (Somentics/Covidien, Inc, Mansfield, MA)) and NIROTM-200X (Hamamatsu Photonics, Hamamatsu City, Japan), can detect some changes in cerebral oxygen saturation, and these machines are in common clinical use for that purpose. However, there are few evaluations of their accuracy or the dynamics of cerebral tissue oxygenation during DHCA. We prospectively observed the efficacy of NIRS of NIRO and INVOS especially during Retrograde Cerebral Perfusion (RCP) during DHCA.
[Methods]
The study included 19 patients who underwent conventional TAR from June to September 2016. After induction of anesthesia, 2 sensors of both INVOS and NIRO were placed over the forehead of the patients. We collected regional cerebral oxygen saturation (rSO2:%) and tissue oxygen index (TOI:%) data from INVOSTM and NIROTM respectively. Data collected during RCP was used for statistical analysis. Bland-altman analysis was used to compare the bias ± precision and polar plot analysis was also performed for concordance analysis of these devices either on left and right side of the forhead. Linear regression was applied to the bland-altman plots. Data were expressed as mean±SD. 
[Results]
The demographic characteristics of the 19 patients are as follows; age 71 years old (±9), height 166.2cm(±8.7), weight 66.4kg (±16.1), and BMI 23(±4.4). A total of 694 pairs of data were obtained (right 355, left 355). The bias of right was -2.2%, left was -1.8%. Precision of right and left was 6.3% and 9.0% between NIRO and INVOS. There is a correlation between the two devices (r=0.54 for right, 0.83 for left, p≦0.0001). Concordance rate was 47.2% and 53.3% for right and left respectively.
[Conclusion]
NIRO has a close correlation with INVOS during RCP. Bias of left and right were varied, however. Comparing the bias of the left to the right, right side had smaller bias than the left. Concordance rate was also low between the two. These differences were created because of the algorithms on which these two oxymeters based and nonphysiological hemodynamics during RCP.