Determination of urinary phytoestrogens by HPLC–MS/MS: A comparison of atmospheric pressure chemical ionization (APCI) and electrospray ionization (ESI)

A comparison of the analytical performance of atmospheric pressure chemical ionization (APCI) and electrospray ionization (ESI) for the quantitative determination of six urinary phytoestrogens (daidzein, O-desmethylangolensin, equol, enterodiol, enterolactone and genistein) by high-performance liquid chromatography–tandem mass spectrometry (HPLC–MS/MS) is presented here. Both APCI and ESI were suitable for the analysis of these compounds; however, ESI did improve measurement imprecision and sensitivity in certain cases. Method imprecision (between-run coefficients of variation [CVs] from duplicate analysis of three quality control [QC] urine pools across 20 runs) was 5.6–12% for ESI, as opposed to 5.3–30% for APCI. At low concentrations (3–60 ng/mL, analyte dependent) imprecision was lower with ESI, whereas both techniques were generally commensurate at high concentrations (200–1000 ng/mL, analyte dependent). Method accuracy (spiked analyte recovery from the QC pools) was comparable between techniques: 86–114% for ESI; 95–105% for APCI. Limits of detection (LODs) were equivalent or better with ESI compared to APCI, with the most significant LOD improvement observed for equol (ESI: 0.3 ng/mL; APCI: 2.7 ng/mL). This translated into a substantial increase in equol detection frequency (% of sample results above LOD) within a random patient sample subset (98% for ESI, compared to 81% for APCI, n = 378). Correlation (Pearson) and agreement (Deming regression, Bland-Altman bias) between ESI and APCI results in the patient subset was better in cases where imprecision and sensitivity was similar for both techniques (daidzein, enterolactone, genistein: r = 0.993–0.998; slope = 0.98–1.03; bias = −4.2 to −0.8%); correlation and/or agreement was poorer for analytes, where APCI imprecision and sensitivity were inferior (equol, O-desmethylangolensin, enterodiol). Baring significant factors arising from differences in ionization source design, these observations suggest that ESI is more appropriate for urinary biomonitoring of these compounds by LC–MS/MS.