Judy Stone (Presenter)
Univ. of Calif. San Diego Health System
Bio: Judy Stone is Sr. Technical Specialist at the UCSD Toxicology/Mass Spectrometry Laboratory. Her research interests are LC-MSMS method development and automation for small molecule analytes and interfacing of laboratory information systems, middleware, liquid handlers and mass spectrometry instruments.
Authorship: Stone JA1, van Staveren DR 3, Fitzgerald R 1,2
1Univ. of Calif. San Diego Health System, 2Univ. of Calif. San Diego, Dept. of Pathology, 3Tecan Schweiz AG
| Short Abstract We validated an LC-MSMS testosterone method using the Tecan AC Extraction Plate and observed excellent precision and stability. We then evaluated (1)an historical calibration (from another batch within +/-14 days) or (2)partial calibration (only 2 standards) versus (3)calibration with 5-10 standards in each batch. CVs for between run precision ranged from 2.9-8.6% for 4 QC levels, 6 batches, across 14 days for the 3 calibration schemas. The bias for QC means and 100 patient samples between calibration schemas 1 or 2 versus 3 was <3.4ng/dL or 5.4%. Less intensive calibration schema to enhance productivity appears promising and deserves additional investigation. |
Long Abstract
Introduction:
The Toxicology/Mass Spectrometry Laboratory at the Univ. of Calif. San Diego Health System (UCSD) had the goal of developing and validating a highly automated, routine production, LC-MSMS method to report serum testosterone for pediatric, female and male patient samples in a hospital laboratory setting. Towards that end, UCSD and Tecan Schweiz AG collaborated to modify the existing Tecan method (1,2) for extracting serum testosterone with the Tecan AC Extraction Plate™ and automated liquid handling (ALH).
The Tecan AC Extraction Plate™ with ALH was appealing because of the minimal user intervention required (no off-deck extraction steps such as mixing, centrifugation or evaporation are needed) and the simplicity of automation (only pipetting and shaking steps are necessary). After intensive optimization of liquid handling parameters, we observed during development and validation that quantitation lower limits, LC retention times, stability of the standard curve, and precision were consistently better than target thresholds. In the interest of lowering costs and achieving better turn-around times we queried our data set to determine whether an historical calibration (from a batch on a previous day) and/or a minimal number (2) of calibrators (partial calibration) included in each batch could provide acceptable method performance compared to calibration with 5-10 standards in every batch (in-batch full calibration). We evaluated 9 pre-validation and validation batches from February 2016 through May 2016 that used the same calibration and internal standard lots on one Waters Acquity-XEVO TQS triple quadrupole LC-MSMS instrument. During this interval the TQS instrument was used daily for production with 6 other quantitative serum/urine small molecule methods and the LC column used for testosterone analysis was also in use for urine opiate and cocaine metabolite confirmation methods.
Method Parameters (3):
All pipetting and extraction steps were carried out automatically by a Tecan Freedom EVO 100 ALH. The serum sample volume added to the AC Extraction Plate™ was 100 µL. 25 µL of 50 ng/dL, 13C3-testosterone in 60:40 H2O:acetonitrile was pipetted to wells followed by mixing (2 min) on an orbital shaker. Extraction buffer (175 µL of 6.7:26.7:66.7 [vol:vol%] of H2O:acetonitrile:0.33 mol/L LiCl with 0.1% NH4OH) was pipetted to the plate followed by shaking for 10 min to extract testosterone into the proprietary coating on the wells. Two plate washes were performed to remove matrix (250 µL each of 0.2% NH4OH with shaking for 5 min). Testosterone was eluted from the coating on the wells with 100 µL of Elution Reagent (35:65 H2O:acetonitrile) and shaking for 5 min. The eluate was transferred by the pipetting robot to 700 µL glass inserts in a 96 well round bottom polypropylene plate for injection of 10 µL on a Waters Acquity-XEVO TQS triple quadrupole LC-MSMS. The LC run was 5.44 min, flow rate 0.4-0.6 mL/min with a gradient from 10:90 to 95:5 Mobile Phase A:B followed by re-equilibration. Column temperature was 45 °C. The LC column was a Waters XSelect HSS T3, C18, 2.1x150mm, 2.7 µm. Mobile Phase A was 2 mmol/L ammonium acetate/0.1% formic acid in H2O, mobile phase B was acetonitrile/0.1% formic acid. MSMS acquisition was performed in positive ESI mode using MRMs 289/97, 289/109, 292/100.
Validation Results:
The validated reportable range for the assay is 2-2,000 ng/dL. Mean matrix effect and recovery measured in 10 patient samples were 95% and 26% respectively with coefficients of variation (CV) of 5% and 10%. A within-run precision study (n=5) with full in-batch calibration yielded CVs of 0.7 – 4.4% for mean concentrations between 2.0 and 1,891 ng/dL. Between run precision for BioRad LiquiCheck and in-house QC materials (4 levels, means 6.9 to 938 ng/dL, n=17) had CVs from 3.8-8.6%. No interference was observed using serum separator tubes with clot activator, or from DHEA, 3-epitestosterone, androstenedione, 17-hydroxyprogesterone, cortisol, 11-deoxycortisol or progesterone. Mean bias was 1.2% for College of American Pathologist Accuracy Based Survey (CAP-ABS) samples after using NIST SRM 971 to value assign in-house calibrators. We found excellent agreement (mean bias: +3.4%, median difference: -2.9%) for a 100 patient sample set in comparison with 3 other clinical laboratories reporting testosterone by LC-MSMS.
Alternate Calibration Results:
We calculated concentrations in multiple batches using (1)an historical calibration with 5 standards (from another batch within +/-4 months) or (2)partial calibration (only 2 standards) versus (3)calibration with 5-10 standards in every batch. CVs for between run precision with all 3 calibration schema ranged from 2.9-8.6% for 4 QC materials in 6 batches, across 14 days and for 6 replicates of 4 CAP-ABS samples tested in 3 batches in February and May. The bias for QC means, CAP-ABS samples, and 100 patient samples between calibration schemas 1 or 2 versus 3 was <3.4ng/dL or 5.4%. As anticipated, the largest deviations between historical and partial calibrations versus in batch full calibration were seen at concentrations <10 and >1,000 ng/dL. Strategies for stronger anchoring of the extremes of the calibration curve and compensating for changes in the sensitivity of the MSMS over time are under investigation.
Good quantitation practice with LC-MSMS also requires, at a minimum, evaluation of MRM qualifier ratios, retention times and relative internal standard abundance for each quality control and patient sample compared to calibrators. Even when between batch quantitation accuracy is good, these quality parameters may not be sufficiently consistent across batches to use an historical calibration to calculate expected values. For the AC Extraction Plate method we found the CV for the 13C3-testosterone internal standard peak area across 4 months, all 9 batches (520 samples with the same injection volume) was only 15% despite a change in the LC run time (6.30 min to 5.44 min) between pre-validation and validation; and LC and MSMS annual preventative maintenance and a column change performed in March 2016. We established that optimizing liquid handling parameters for the pipetting of internal standard in 60:40 (vol:vol %) water:acetonitrile by the Tecan Freedom EVO 100 ALH was a key feature for improving precision. Absolute retention times per LC program (means = 4.09 min and 4.29 min for the 2 LC programs) and MRM qualifier ratios (mean ratio = 0.815) for all 537 samples within the reportable range in the 9 batches had CVs of <0.2% and 5% respectively. The calibration curve slope had a CV of 5.1% and all calibration curve R2 values were 0.997 or higher. The good long term reproducibility of these quality assurance parameters should allow the use of an historical calibration.
Conclusions:
Using a less resource intensive calibration schema than the recommended 5 or 6 non-zero calibrators in each batch appears promising with this method. Comparing the different calibration schema with additional batches tested over a longer interval and adding the variance of multiple analysts is necessary. However, the high degree of automation possible with the AC Extraction Plate and ALH reduces between analyst differences significantly, making it more likely that good precision and accuracy with historical/partial calibration can be achieved in routine production.
References & Acknowledgements:
1. Determination of testosterone in serum by automated sample preparation and ultra-fast LDTD-MS/MS in a cross validation study with real patient samples. Alex Birsan, Pierre Picard, Serge Auger, Annie-Claude Bolduc, Dave van Staveren, Roland Geyer and Jean Lacoursière. MSACL US, 2015, # 80267.
2. Sample preparation of three steroids for quantitative determination by LC-MS/MS – comparison of two extraction procedures. Patrizia Sebregondi, Christiane Zaborosch, Dave van Staveren, Roland Geyer, Michal Svoboda. MSACL US, 2015, # 80267.
3. Characterizing Matrix Depletion using a Novel 96-well Format Extraction Media - Tecan® AC Extraction Plate (AC Plate). Stone JA, van Staveren DR, Maisenhölder B, Fitzgerald R. MSACL US, 2016, #69.
We thank Tecan Schweiz AG for providing the Freedom EVO 100 ALH, AC Extraction Plates and EVO pipetting tips at no charge and for a highly productive collaboration. We thank Grace vanderGugten (St. Paul's Hospital, Vancouver, BC), Deborah French, Ph.D. (UCSF, San Francisco, CA) and Julia Drees, Ph.D. (TPMG Kaiser Regional Laboatories, Berkeley, CA) for the Ladies Testosterone Cooperative 1st Multi-laboratory patient sample comparison for testosterone.
| Description | Y/N | Source |
| Grants | no | |
| Salary | no | |
| Board Member | no | |
| Stock | no | |
| Expenses | no |
IP Royalty: no
| Planning to mention or discuss specific products or technology of the company(ies) listed above: | yes |