Determination of serum cholesterol by isotope dilution mass spectrometry with a benchtop capillary gas chromatograph/mass spectrometer: comparison with the National Reference System's Definitive and Reference Methods.

We developed an isotope dilution mass spectrometric cholesterol method with [25,26,27-13C]cholesterol as internal standard and a benchtop gas chromatograph/mass spectrometer (GC/MS) that is much easier and less time consuming than previously described Reference and Definitive Methods for cholesterol. The internal standard, cholesterol standards, and unknown specimen are delivered volumetrically with an automated dilutor and the saponifying reagent. After saponification, extraction, and derivatization, specimens are injected into a benchtop quadrupole MS with an autosampler. Unknown cholesterol concentrations are calculated automatically by comparing the peak area ratio of the m/z = 368, 371 ion pair with the ratios for the cholesterol standards (0 to 12.93 mmol/L). We found within-run and day-to-day (overall) imprecision of 0.44% and 0.95%, respectively, when specimens were assayed singly. In several lyophilized and frozen Standard Reference Material (SRM) pools, cholesterol results with our GC/MS method averaged 0.4% less than the National Institute for Standards and Technology definitive GC/MS result performed about three years earlier. Our GC/MS results averaged 1.3% and 2.0% less than results by the National Reference System (NRS) Abell-Levy-Brodie-Kendall (ALBK) results from clinical specimens and the SRM pools, respectively. These results are consistent with the previously reported bias between the NRS Reference and Definitive Methods and the 0.1% per year decrease in cholesterol concentrations in SRM pools as determined by GC/MS analysis. These results further emphasize the small but consistent bias between cholesterol results by isotope dilution mass spectrometry and the ALBK Reference Method, the latter being the basis for the National Cholesterol Education Program guidelines and population reference values.