Cerebrospinal Fluid Xanthochromia Testing Simplified

We previously published a method that determines xanthochromia in cerebrospinal fluid (CSF) to detect subarachnoid hemorrhage (1). The method involved a spectrophotometric scan of the CSF in which the absorbances of hemoglobin (Hb; 415 nm) and bilirubin (440 nm) were determined by subtracting a background obtained by drawing a tangent from 360 to ∼530 nm. The net absorbance of bilirubin was then determined from an equation that compensated for Hb absorbance. A closer examination of this method showed that the Hb spectrum gave zero absorbance at 476 nm even at Hb absorbances up to 1.5 at 415 nm. We then tested 476 nm to measure net bilirubin absorbance because, although it is not the λmax for bilirubin, the mathematics for the calculation are much simpler. We compared results for 112 CSF samples submitted for CSF xanthochromia analyses by the older method (1), which compensates for Hb interference in the bilirubin estimation, with results of the new method, which measures the absorbance of bilirubin at 476 nm and therefore does not use compensation for Hb absorbance. The new method, like the old one, uses the tangent approach to measure the net absorbance at 476 nm. with r2 = 0.93 (P <0.001) and mean absorbances (SD) of 0.0145 (0.0131) and 0.0136 (0.0119) for the old and new methods, respectively. In this group of samples, there was one pair of discrepant results with a marginally positive absorbance of 0.017 (reference >0.015) by the old method and borderline positive (0.013; reference interval, 0.01–0.014) by the new. In general, patients who are either positive or borderline positive have the same clinical follow-up, namely, an angiogram to confirm the presence or absence of an aneurysm. This patient was subsequently found to have a CSF infection with associated inflammation; no aneurysm was detected. Because of the excellent correlation, we are now using the new method because it is simpler, thereby making the method potentially less prone to error. We have used the old method for >3 years now in >250 analyses.