Matthew Crawford (Presenter)
LabCorp
Bio: I've been in the clinical industry for 10 years with 7 of them focused on research and development. I primarily focus on LC-MS/MS method development and validation of high-throughput small molecule clinical assays.
Authorship: Matthew Crawford, Christopher Shuford, Ph.D., Stacy Dee, Martin Green, Patricia Holland, Yvonne Wright, Mary Morr, Russell Grant, Ph.D.
Laboratory Corporations of America Inc, 1447 York Court, Burlington, NC 27215
| Short Abstract LC-MS/MS calibration can have compounding error as they rely on gravimetric weighing of materials and the quality of material. As such in-house gravimetry was undertaken with materials from 3 different vendors and compared against the reference method via samples of assigned concentrations. Beyond gravimetry, internal standard based calibration will be shown by setting IS response at 6 clinical cut-offs. Finally, the concept of calibration using normal patient sample pools will be demonstrated. Pools were collected for age and gender specific reference intervals. Value assigned for each population pool will be explored. To conclude all mechanisms of calibration were challenged against value assigned reference method samples. |
Long Abstract
Introduction
LC-MS/MS assay calibration has proven to vary based on individual components in the manufacturing of materials. Standardization programs have been implemented in an effort to ensure accuracy of methods across laboratories. As the acceptance criteria for these programs can be stringent (± 6.4% mean bias), all areas for potential error need to be explored. These methods of calibration are subject to compounding error as they rely on gravimetric weighing of materials and consideration of the quality of material, together with multiple dilution steps. In this paper, we explore the use of internal standard response based calibration and propose the expanded use of “clinical calibration” where multiple reference intervals are expected, together with the use of normal population mean for internal standard “concentration” assignment and thus assay calibration.
Method
Analysis was performed on a Thermo Transcend TLX-4 coupled to a Sciex API 5000 or 5500. In-house gravimetry was undertaken with certified materials from 3 different vendors and compared against the reference method via samples of assigned concentrations (CDC). Moving from gravimetry, the continuing evolution of internal standard based multi-point calibration will be shown by setting IS response (response factor, RF) at 6 clinical cut-offs. Finally, the concept of calibration using normal patient sample pools as a component in universal calibration will be demonstrated. Clinical patient pools (n=400/pool) were collected for age and gender specific reference intervals and concentration assigned by retrospective analysis of >10000 specimen results within each “reference” pool. This model of calibration was then compared to the reference method.
Results
While preparing multiple iterations of calibration curves for testosterone it was noted that mean bias of reference material varied depending on the manufacturing components used following correction for listed purity. Alternate sources of stripped matrices/analyte reference material as well as intermediate stock composition impacted calibrator accuracy. Analysis of 6 calibration iterations compared against reference method (n=40) provide Deming slopes between 0.957 to 1.059 and mean bias of -5.21% to +5.91%. Final calibrator materials were selected when the average mean bias to the reference method was <1% and Bland-Altman plots indicated equal bias distribution within the expected heteroscedastic variation of the assay. Final calibrator correlation to the reference method produced a Deming slope of 0.998, correlation coefficient of 0.9996, and mean bias of -0.49%. After analysis of over 400 samples through the reference method certification program (over 2 ½ years) a fixed positive bias of approximately 3% was observed. Subsequently, reference materials (identical source and providence of reference method materials) were sourced and re-analysis indicated a H2O content change of 4% (Karl Fisher). We have explored alternate modes of calibration using contemporaneous and sporadic response factor (RF) techniques in isotope dilution assays. Analysis of methylmalonic acid (MMA) without calibration curves but by setting IS response factors at 2 medical decision points and back-calculating sample concentration for MMA for over 2000 samples over 8 days yielded a Deming slope of 1.016, correlation coefficient of 0.9995, and mean bias of 0.54% when compared against classical isotope dilution/multi-point analyte calibrators. This same mode was evaluated for testosterone using 6 medical decision points due to the age and gender distribution observed clinically. It will be shown that when the approximately matched IS response factors are used to calculate results that have been informed by reference method samples, the correlation to CDC reference method results yields a Deming slope of 1.001, correlation coefficient of 0.9995, and mean bias of -0.869%. Research has been performed to remove analyte and internal standard gravimetry and response function modes entirely and relying solely on very large population pools to be used as calibrators. Preliminary assessment suggests pools of at least 400 specimens per level may enable gravimetry free calibration to be deployed for endogenous analyte measurement, further resolving issues regarding matrix matching calibrators and samples. Multiple means of value assigned for each population pool were explored and compared to standard isotope dilution. To conclude all mechanisms of calibration discussed above were challenged against value assigned reference method samples. Mean bias observed from this litmus test of calibration mechanisms ranged from -9.902% to +1.395%.
Conclusion
Accurate calibration is key to clinical diagnostics and alternative calibration protocols may be accurate by comparison and a viable alternative.
References & Acknowledgements:
[1] Clinical Calibration, Moving away from the Line, MSACL US, 2013, Matthew Crawford
| Description | Y/N | Source |
| Grants | no | |
| Salary | yes | LabCorp |
| Board Member | no | |
| Stock | yes | LabCorp |
| Expenses | no |
IP Royalty: no
| Planning to mention or discuss specific products or technology of the company(ies) listed above: | no |