Determination of Serum Aldosterone
Determination of Serum Aldosterone
The d7-aldosterone internal standard (IS) (Isosciences, King of Prussia, Pennsylvania, USA) was prepared to a working solution of 110 nmol/l in 50:50 MeOH/H2O. A weighed-in stock calibrator was prepared using HPLC-purified aldosterone (Sigma, St. Louis, Missouri, USA) at a concentration of 1953.3 nmol/l in 50:50 MeOH/H2O with corrections for stated purity. The IS and calibrator solutions were aliquoted and stored at −70°C. For analysis, the calibrator was thawed and diluted by three serial 10-fold dilutions with 50:50 MeOH/H2O to prepare solutions of 195 330, 19 533 and 1953.3 pmol/l. Appropriate volumes of these solutions were spiked into double-stripped steroid-free human serum (Golden West Biologicals, Temecula, California, USA) to produce calibrators of 49, 98, 146, 195, 488, 976, 1950 and 3420 pmol/l. Spiking volume was limited to <10% of the total calibrator volume to minimise disturbances to the serum matrix. Calibrator concentrations were chosen to give optimal coverage in the range of clinical interest (~100–1000 pmol/l).
All pipetting was handled robotically using the Hamilton STARlet liquid-handling system. Sixty microlitres of IS working solution and 500 μl of serum/calibrator were pipetted into 5 ml polypropylene Eppendorf tubes (Argos Technologies, Elgin, Illinois, USA) followed by 2500 μl of MTBE. After a 3 min vortex, phase separation was assisted with 5 min centrifugation (3000 RPM, 5°C). The organic layer (2000 μl) was then pipetted into a 2.2 ml Eppendorf tube and dried down passively in a fume hood. Samples were reconstituted with 125 μl of 50:50 MeOH/H2O, vortexed for 30 s and transferred to a 96-well plate (Axygen, Union City, California, USA).
HPLC was performed using 2 mmol/l NH4Ac in H2O (A) and 2 mmol/l NH4Ac in MeOH (B) and a Gemini-NX 3u C18 110A 100×2.0 mm column preceded by a Gemini-NX Security Guard C18 4×2.0 mm guard-cartridge (Phenomenex, Torrance, California, USA) on a Shimazdu LC-20AD/CT-20A/SIL-20ACHT UFLC (column oven: 55°C). Solvent gradient: 0 min, 20% B; 1 min, 20% B; 5 min, 70% B; 6 min, 70% B; 6.5 min, 90% B; 7.5 min, 90% B; 7.6 min, 20% B; and 10 min, 20% B. The elution time for aldosterone was approximately 6.5 min.
Ionisation parameters and multiple reaction monitoring (MRM) transitions are shown in Table 1. Conditions were optimised for sensitivity using ABSCIEX Analyst® software (V.1.5, ABSCIEX, Foster City, California, USA).
MS data were analysed using Analyst® software and its Intelliquant algorithm. Calibration curves were constructed with 1/x-weighted linear regression.
Linearity was investigated using the CLSI EP6-A protocol. Response (ie, ratio of analyte to IS peak-area) versus concentration relationship was analysed using data from nine calibration curves. The mean-response/concentration curve was fitted using linear, quadratic and cubic least-squares regression. A linearity difference plot was prepared with a maximum acceptable linearity difference set at 1%.
Recovery was assessed by analysing high and low patient pools (high=1580 pmol/l, n=20; low=39 pmol/l, n=35), mixed in proportions of 80:20, 60:40, 50:50, 40:60 and 20:80 (v:v).
Accuracy was determined by analysis of external quality assurance specimens with GCMS assigned aldosterone values or GCMS certified aldosterone values available from Referenzinstitut fur Bioanalytik (Bonn, Germany). Material was reconstituted and analysed in triplicate (singleton analysis over three runs).
Precision was evaluated using the CLSI EP-5A2 protocol. Quadruplicate analysis of pooled human samples (n=10–15 per level) with concentrations of approximately 50 pmol/l, 100 pmol/l, 500 pmol/l and 1000 pmol/l was performed on 4 days (ie, 16 analyses per level). Identical analysis was performed on a commercial material (Bio-Rad Lyphocheck® Immunoassay Plus).
The limit of quantitation (LOQ) and limit of detection (LOD) were determined based on signal to noise (S:N) ratios of 10:1 and 3:1 using Analyst® software.
The method comparison cohort was comprised of 138 samples collected from 129 unique subjects. Specimens from subjects known to have chronic kidney disease, based on eGFR<60 ml/min/1.73 m were excluded because of overestimation of aldosterone by non-extraction RIAs, such as the Siemens Coat-a-Count® used in this study. After exclusion, 118 specimens on 111 unique subjects remained. Duplicate collections from the same subject were excluded in favour of the first-obtained specimen.
The effect of EDTA sample matrix was assessed by comparing 22 concomitantly collected gel-free serum and EDTA specimens, spanning ~50–1000 pmol/l.
Interferences were investigated by preparing 50:50 MeOH/H2O solutions of natural and synthetic steroids at supra-physiological concentrations (or 1000 nmol/l for synthetic compounds). Solutions were analysed using the LC-MS/MS method described. Subsequently, steroid-free serum was spiked with a steroid cocktail obtained from ABSCIEX to achieve mid-to-high physiological concentrations. This mixture was then subjected to extraction, reconstitution and analysis. Specific compounds investigated are discussed later.
Regression and statistics were performed using R version 2.10.1 (Free Software Foundation, Boston, Massachusetts, USA).
Materials and Methods
Calibrator and Internal Standards
The d7-aldosterone internal standard (IS) (Isosciences, King of Prussia, Pennsylvania, USA) was prepared to a working solution of 110 nmol/l in 50:50 MeOH/H2O. A weighed-in stock calibrator was prepared using HPLC-purified aldosterone (Sigma, St. Louis, Missouri, USA) at a concentration of 1953.3 nmol/l in 50:50 MeOH/H2O with corrections for stated purity. The IS and calibrator solutions were aliquoted and stored at −70°C. For analysis, the calibrator was thawed and diluted by three serial 10-fold dilutions with 50:50 MeOH/H2O to prepare solutions of 195 330, 19 533 and 1953.3 pmol/l. Appropriate volumes of these solutions were spiked into double-stripped steroid-free human serum (Golden West Biologicals, Temecula, California, USA) to produce calibrators of 49, 98, 146, 195, 488, 976, 1950 and 3420 pmol/l. Spiking volume was limited to <10% of the total calibrator volume to minimise disturbances to the serum matrix. Calibrator concentrations were chosen to give optimal coverage in the range of clinical interest (~100–1000 pmol/l).
Sample Preparation
All pipetting was handled robotically using the Hamilton STARlet liquid-handling system. Sixty microlitres of IS working solution and 500 μl of serum/calibrator were pipetted into 5 ml polypropylene Eppendorf tubes (Argos Technologies, Elgin, Illinois, USA) followed by 2500 μl of MTBE. After a 3 min vortex, phase separation was assisted with 5 min centrifugation (3000 RPM, 5°C). The organic layer (2000 μl) was then pipetted into a 2.2 ml Eppendorf tube and dried down passively in a fume hood. Samples were reconstituted with 125 μl of 50:50 MeOH/H2O, vortexed for 30 s and transferred to a 96-well plate (Axygen, Union City, California, USA).
HPLC Conditions
HPLC was performed using 2 mmol/l NH4Ac in H2O (A) and 2 mmol/l NH4Ac in MeOH (B) and a Gemini-NX 3u C18 110A 100×2.0 mm column preceded by a Gemini-NX Security Guard C18 4×2.0 mm guard-cartridge (Phenomenex, Torrance, California, USA) on a Shimazdu LC-20AD/CT-20A/SIL-20ACHT UFLC (column oven: 55°C). Solvent gradient: 0 min, 20% B; 1 min, 20% B; 5 min, 70% B; 6 min, 70% B; 6.5 min, 90% B; 7.5 min, 90% B; 7.6 min, 20% B; and 10 min, 20% B. The elution time for aldosterone was approximately 6.5 min.
MS Conditions
Ionisation parameters and multiple reaction monitoring (MRM) transitions are shown in Table 1. Conditions were optimised for sensitivity using ABSCIEX Analyst® software (V.1.5, ABSCIEX, Foster City, California, USA).
Analysis
MS data were analysed using Analyst® software and its Intelliquant algorithm. Calibration curves were constructed with 1/x-weighted linear regression.
Method Validation
Linearity was investigated using the CLSI EP6-A protocol. Response (ie, ratio of analyte to IS peak-area) versus concentration relationship was analysed using data from nine calibration curves. The mean-response/concentration curve was fitted using linear, quadratic and cubic least-squares regression. A linearity difference plot was prepared with a maximum acceptable linearity difference set at 1%.
Recovery was assessed by analysing high and low patient pools (high=1580 pmol/l, n=20; low=39 pmol/l, n=35), mixed in proportions of 80:20, 60:40, 50:50, 40:60 and 20:80 (v:v).
Accuracy was determined by analysis of external quality assurance specimens with GCMS assigned aldosterone values or GCMS certified aldosterone values available from Referenzinstitut fur Bioanalytik (Bonn, Germany). Material was reconstituted and analysed in triplicate (singleton analysis over three runs).
Precision was evaluated using the CLSI EP-5A2 protocol. Quadruplicate analysis of pooled human samples (n=10–15 per level) with concentrations of approximately 50 pmol/l, 100 pmol/l, 500 pmol/l and 1000 pmol/l was performed on 4 days (ie, 16 analyses per level). Identical analysis was performed on a commercial material (Bio-Rad Lyphocheck® Immunoassay Plus).
The limit of quantitation (LOQ) and limit of detection (LOD) were determined based on signal to noise (S:N) ratios of 10:1 and 3:1 using Analyst® software.
The method comparison cohort was comprised of 138 samples collected from 129 unique subjects. Specimens from subjects known to have chronic kidney disease, based on eGFR<60 ml/min/1.73 m were excluded because of overestimation of aldosterone by non-extraction RIAs, such as the Siemens Coat-a-Count® used in this study. After exclusion, 118 specimens on 111 unique subjects remained. Duplicate collections from the same subject were excluded in favour of the first-obtained specimen.
The effect of EDTA sample matrix was assessed by comparing 22 concomitantly collected gel-free serum and EDTA specimens, spanning ~50–1000 pmol/l.
Interferences were investigated by preparing 50:50 MeOH/H2O solutions of natural and synthetic steroids at supra-physiological concentrations (or 1000 nmol/l for synthetic compounds). Solutions were analysed using the LC-MS/MS method described. Subsequently, steroid-free serum was spiked with a steroid cocktail obtained from ABSCIEX to achieve mid-to-high physiological concentrations. This mixture was then subjected to extraction, reconstitution and analysis. Specific compounds investigated are discussed later.
Regression and statistics were performed using R version 2.10.1 (Free Software Foundation, Boston, Massachusetts, USA).
