Reproducibility of NMR Analysis of Urine Samples: Impact of Sample Preparation, Storage Conditions, and Animal Health Status

Table 2

Method-related variance.

	Glucose	Citrate	Lactate	Creatinine	PAG	Urea
	Synthetic samples and rat urines (stability test series)
𝑅^2	0.994	0.991	0.998	0.992	0.997	0.988
RMSE	0.048	0.038	0.033	0.050	0.021	0.016
	Rat urines (stability test series)
CV [%]	6.9	3.4	4.9	3.5	3.0	2.1

Synthetic and urine samples as described in Figure 1 were used to calculate different measures of accuracy. 𝑅² and RMSE values were calculated for all data from synthetic and urine samples together, which were nominally/with first measure above LLOQ (including 100 synthetic samples and 54 urine samples with 8 repetitions, i.e., maximal n = 532). Additionally, CV values were calculated for the repeated measures of the 54 different rat urine samples, each measured in total 9 times. Again, values below LLOQ at the first measure were excluded (but values below LLOQ from repeated measures were included in the evaluation), thereby reducing the number of samples to 25 for lactate and to 24 for glucose. CV values for each metabolite are given as a mean of CV values for the different treatment groups (natrosol, furosemide, or HCBD) and effectors (pH 3 to pH 9, salt up to 20-fold), that is, first calculating the CV values for specific samples, for example, the sample natrosol treatment, pH 3, salt 4-fold, and thereafter calculating mean CV values for each metabolite.