Comparison of 13 published cytochrome c oxidase near-infrared spectroscopy algorithms

Conflicting patterns of change in cytochrome c oxidase (Cyt a,a3) redox status have been obtained between different near-infrared spectrophotometers when making measurements during tissue ischaemia. This study identifies possible sources of error that could be the cause of the discrepancy. A single set of optical density data was repeatedly analysed using each of the absorption spectra from 13 publications. In addition, changes in Cyt a,a3 redox status were calculated from the data set using three numerical methods, five computer software routines, eight displacements in wavelength, and ten incremental changes in the value of absorption or concentration coefficients. All Cyt a,a3 absorption spectra resulted in algorithms yielding similar patterns of change, regardless of the numerical method or computer process employed (0.9996 average r2, coefficient of correlation). However, a significantly different pattern of change in Cyt a,a3 redox status, resembling that reported by Piantadosi [Piantadosi CA (1993) Methods Toxicol. 2:107-126], was obtained when either the wavelengths, and/or the absorption values were altered to simulate erroneous values. This implies that all of the present algorithms are valid (including those of Piantadosi), but that microchip encoding errors may exist in the instrument used by Piantadosi.