Biological variability dominates and influences analytical variance in HPLC-ECD studies of the human plasma metabolome

Table 1 Analytical Parameters

	Before polishing data	After polishing data
Range, median CVs for 66 variables	2–54%	2–36%
Range, mean CVs for 66 variables	2–58%	3–39%
Median CV for all 66 variables	12%	12%
Mean CV for all 66 variables	19%	17%
By 13 Pairs, overall median CV	9%	9%

"Before polishing data", primary (raw) analysis; "After polishing data", final analysis after manual re-inspection (see text). The most critical data are the median coefficients of variation (CV); our median CV was ~10%, well within the levels necessary to conduct the proposed study. We also note that upon re-examination of the dataset for quality control purposes, 1% (11 peaks) was left unscored due to analytical concerns, and 3% (25 peaks) had scores that differed by >10% from the original call. These data suggest that, even in the absence of any optimization, we can consistently measure >96% of all metabolites with a median ± 10% precision. For the last line "By 13 pairs", we first determined the median CV for all 66 variables in each of the pairs, and then determined the mean/median of these values. By definition, overall mean CVs are unchanged by this difference in calculation procedure, and are not shown.

ISSN: 1472-6890