An evaluation of the accuracy of the folate microbiologic assay and comparison with liquid chromatography-tandem mass spectrometry

Abstract

Monitoring of folate status as part of population surveys requires accurate laboratory methods. Although the microbiologic folate assay is considered a “gold standard”, there is no standardised procedure. Laboratory-specific procedural differences in the assay can result in considerable variation in blood folate results, and reduce comparability over time and between laboratories. The aim of this research was to evaluate some of the analytical issues that may influence the accuracy of the chloramphenicol-resistant microbiologic assay procedure within our laboratory. A secondary objective was to compare whole blood folate concentrations measured by the microbiologic method with liquid chromatography-tandem mass spectrometry (LC-MS/MS).

EDTA blood samples were collected from 73 healthy adult participants aged 18- 65 years. Total folate in plasma and whole blood were measured by the microbiologic assay using different calibrators (5-methyltetrahydrofolate [5- methylTHF], Sigma folic acid and Merck folic acid), and sample dilutions of 1/40 and 1/80 for plasma folate and 1/800 and 1/1600 for whole blood folate. Whole blood folate species (5-methylTHF, 5,10-methenylTHF, THF, and folic acid) in the same samples were measured by LC-MS/MS. Ordinary linear regression and Pearson correlations were used to compare dilutions and to compare the microbiologic assay results with LC-MS/MS. Bland-Altman relative bias was also calculated to compare the microbiologic assay results with LC-MS/MS. 5-MethylTHF produced the highest microbiologic assay calibration curve, followed by Merck folic acid and then Sigma folic acid. The relationship between dilutions was significantly non-linear, with a second order polynomial regression best fitting the data (P < 0.01 for the b2 coefficient). Using 5-methylTHF as a calibrator, the lowest median folate concentrations were obtained with the higher plasma (1/80) and whole blood (1/1600) dilutions, compared with the Sigma folic acid calibrator and lower plasma (1/40) and whole blood (1/800) dilutions, which resulted in the highest median folate concentrations. Between these two conditions, there was a 2.8 fold difference in the median plasma folate concentration (9.9 nmol/L versus 28.1 nmol/L) and a 2.1 fold difference in the median whole blood folate concentration (220 nmol/L versus 467 nmol/L). No significant difference between LC-MS/MS and microbiologic assay whole blood folate values was found when 5-methylTHF was used as a calibrator and a dilution of 1/1600 was used (slope 1.05, P = 0.308, and intercept -21, P = 0.159). The poorest agreement between LC-MS/MS and the microbiologic assay was found when the Sigma folic acid calibrator and a dilution factor of 1/800 were used (slope 2.69, P < 0.001, and intercept -156, P < 0.001). In this case, the Bland-Altman relative bias was 100% (95% limit of agreement 34% to 166%). Differences in blood folate concentrations between dilutions indicate a sample matrix effect, which may be another source of variation in blood folate concentrations amongst laboratories. Such differences could lead to differences in folate status interpretation. Overall, our whole blood folate results showed good agreement between the LC-MS/MS and the microbiologic assay at higher dilutions and with the use of 5-methylTHF as the microbiologic assay calibrator. If the microbiologic assay is used, 5-methylTHF should be used as a calibrator, given that it is the main form of folate in blood. Better harmonisation of methods, and preferably standardisation of folate assays is needed in order to ensure comparability amongst laboratories.