Comparison of liquid chromatography-mass spectrometry, radioimmunoassay, and enzyme-linked immunosorbent assay methods for measurement of urinary estrogens

Jessica M. Faupel-Badger, Barbara J. Fuhrman, Xia Xu, Roni T. Falk, Larry K. Keefer, Timothy D. Veenstra, Robert N. Hoover, and Regina G. Ziegler

Source: Cancer Epidemiol Biomarkers Prev. 2010 January ; 19(1): 292–300. doi:10.1158/1055-9965.EPI-09-0643.


Absolute and relative concentrations of estrogens and estrogen metabolites (EM) are important for clinical decisions, as well as for epidemiologic, experimental, and clinical research on hormonal carcinogenesis. Radioimmunoassays (RIA) and enzyme-linked immunosorbent assays (ELISA) are routinely used for measuring EM in blood and urine due to efficiency and low cost. Here we compare absolute and ranked concentrations of estrone, estradiol, and estriol measured by indirect RIA and of 2-hydroxyestrone and 16α-hydroxyestrone measured by ELISA to the concentrations obtained using a novel liquid-chromatography-tandem mass spectrometry (LC-MS/MS) method which measures 15 EM concurrently.

We used overnight urine samples collected from control women (362 premenopausal, 168 postmenopausal) participating in a population-based case-control study of breast cancer among Asian-American women ages 20–55 years. When comparing RIA or ELISA levels to LC-MS/MS, absolute concentrations for the five EM ranged from 1.6–2.9 and 1.4–11.8 times higher in premenopausal and postmenopausal women, respectively, (all p<0.0001). However, LC-MS/MS measurements were highly correlated [Spearman r (rs) =0.8–0.9] with RIA and ELISA measurements in premenopausal women, and moderately correlated (rs=0.4–0.8) in postmenopausal women. Measurements of the 2-hydroxyestrone:16α-hydroxyestrone ratio, a putative biomarker of breast cancer risk, were moderately correlated in premenopausal women (rs=0.6–0.7) but only weakly correlated in postmenopausal women (rs=0.2). LC-MS/MS had higher intraclass correlation coefficients (≥99.6%) and lower coefficients of variation (≤9.4%) than ELISA (≥97.2% and ≤14.2%) and RIA (≥95.2% and ≤17.8%).

Comparison with the LC-MS/MS method suggests that the widely used RIA and ELISA EM measures may be problematic, especially at low EM levels characteristic of postmenopausal women.

Link to Study