Abstract

A new analytical method has been developed and implemented at the U.S. Geological Survey National Water Quality Laboratory that determines a suite of 20 steroid hormones and related compounds in filtered water (using laboratory schedule 2434) and in unfiltered water (using laboratory schedule 4434). This report documents the procedures and initial performance data for the method and provides guidance on application of the method and considerations of data quality in relation to data interpretation. The analytical method determines 6 natural and 3 synthetic estrogen compounds, 6 natural androgens, 1 natural and 1 synthetic progestin compound, and 2 sterols: cholesterol and 3--coprostanol. These two sterols have limited biological activity but typically are abundant in wastewater effluents and serve as useful tracers. Bisphenol A, an industrial chemical used primarily to produce polycarbonate plastic and epoxy resins and that has been shown to have estrogenic activity, also is determined by the method.

A technique referred to as isotope-dilution quantification is used to improve quantitative accuracy by accounting for sample-specific procedural losses in the determined analyte concentration. Briefly, deuterium- or carbon-13-labeled isotope-dilution standards (IDSs), all of which are direct or chemically similar isotopic analogs of the method analytes, are added to all environmental and quality-control and quality-assurance samples before extraction. Method analytes and IDS compounds are isolated from filtered or unfiltered water by solid-phase extraction onto an octadecylsilyl disk, overlain with a graded glass-fiber filter to facilitate extraction of unfiltered sample matrices. The disks are eluted with methanol, and the extract is evaporated to dryness, reconstituted in solvent, passed through a Florisil solid-phase extraction column to remove polar organic interferences, and again evaporated to dryness in a reaction vial. The method compounds are reacted with activated -methyl--trimethylsilyl trifluoroacetamide at 65 degrees Celsius for 1 hour to form trimethylsilyl or trimethylsilyl-enol ether derivatives that are more amenable to gas chromatographic separation than the underivatized compounds. Analysis is carried out by gas chromatography with tandem mass spectrometry using calibration standards that are derivatized concurrently with the sample extracts.

Analyte concentrations are quantified relative to specific IDS compounds in the sample, which directly compensate for procedural losses (incomplete recovery) in the determined and reported analyte concentrations. Thus, reported analyte concentrations (or analyte recoveries for spiked samples) are corrected based on recovery of the corresponding IDS compound during the quantification process. Recovery for each IDS compound is reported for each sample and represents an absolute recovery in a manner comparable to surrogate recoveries for other organic methods used by the National Water Quality Laboratory. Thus, IDS recoveries provide a useful tool for evaluating sample-specific analytical performance from an absolute mass recovery standpoint. IDS absolute recovery will differ and typically be lower than the corresponding analyte’s method recovery in spiked samples. However, additional correction of reported analyte concentrations is unnecessary and inappropriate because the analyte concentration (or recovery) already is compensated for by the isotope-dilution quantification procedure.

Method analytes were spiked at 10 and 100 nanograms per liter (ng/L) for most analytes (10 times greater spike levels were used for bisphenol A and 100 times greater spike levels were used for 3--coprostanol and cholesterol) into the following validation-sample matrices: reagent water, wastewater-affected surface water, a secondary-treated wastewater effluent, and a primary (no biological treatment) wastewater effluent. Overall method recovery for all analytes in these matrices averaged 100 percent, with overall relative standard deviation of 28 percent. Mean recoveries of the 20 individual analytes for spiked reagent-water samples prepared along with field samples and analyzed in 2009–2010 ranged from 84–104 percent, with relative standard deviations of 6–36 percent. Concentrations for two analytes, equilin and progesterone, are reported as estimated because these analytes had excessive bias or variability, or both. Additional database coding is applied to other reported analyte data as needed, based on sample-specific IDS recovery performance.

Detection levels were derived statistically by fortifying reagent water at six different levels (0.1 to 4 ng/L) and range from about 0.4 to 4 ng/L for 16 analytes. Interim reporting levels applied to analytes in this report range from 0.8 to 8 ng/L. Bisphenol A and the sterols (cholesterol and 3-beta-coprostanol) were consistently detected in laboratory and field blanks. The minimum reporting levels were set at 100 ng/L for bisphenol A and at 200 ng/L for the two sterols to prevent any bias associated with the presence of these compounds in the blanks. A minimum reporting level of 2 ng/L was set for 11-ketotestosterone to minimize false positive risk from an interfering siloxane compound emanating as chromatographic-column bleed, from vial septum material, or from other sources at no more than 1 ng/L.

First posted November 13, 2012