MSACL 2024 Abstract

Introduction:
To evaluate the clinical effectiveness of conventional and emerging cardiovascular biomarkers, 23,376 serum samples of the Odyssey Outcomes Trial, were measured retrospectively on Cobas, LC-MS/MS and Diazyme analyzers.(1) Prior to the study, a fit-for-purpose process was designed, using the ISO15189 accredited laboratory infrastructure and the Total Testing Process (TTP), which was already in place for routine analyses, including but not limited to: standard operating procedures, laboratory information management system (LIMS), quality management system, track-and-trace registration and cross-trained laboratory technicians.(2) However, the demands of this study, comprising measurement of 23,376 samples on three analyzers in high-throughput, exceeded standard operations. Therefore, the already existing laboratory infrastructure needed to be enhanced with tailored additions regarding sample receipt, sample aliquotation, LC-MS/MS analysis and processing, and Diazyme analysis and processing.

Methods:
21 cardiovascular biomarkers of which 9 apolipoproteins were analyzed in 23, 376 frozen specimens from Acute Coronary Syndrome patients on three distinct analyzers: Roche Cobas 8000, Agilent 6495 triple quadrupole MS (2x 6495A and 2x 6495C), and a Diazyme DZ-Lite analyzer. Preanalytical precautions were taken so that Roche analyses were done in aliquots who underwent one freeze/thaw cycle whereas LC-MS/MS and Diazyme analyses underwent two freeze/thaw cycles.
The testing process entailed: sample receipt, sample registration, sample aliquoting, sample measurement, and data management. Quality assurance was performed with bi-level internal quality controls (IQC) using Levey-Jennings plots and Westgard rules for Cobas and Diazyme analysis. For LC-MS/MS, Westgard rules were adjusted to fit a multiplex test. In addition, system suitability testing (SST) to ensure instrument performance, and interpeptide agreement between quantifying and qualifying peptides to assess individual sample validity, were implemented as well. The rationale of the study to improve patient diagnostics was communicated to all stakeholders to foster an understanding of the necessity to conduct this study. Data was reported on a weekly basis with a custom R script and progress was communicated with all stakeholders. The traceability of results was ensured by adhering to FAIR principles.

Results:
23,376 serum samples were analyzed within 18 months on Cobas, Diazyme and LC-MS/MS instruments by 24 medical laboratory and research technicians. IQC evaluation for Cobas and Diazyme analyses resulted in a coefficient of variation (CV) of 1.7-5.5% and 7.2-10.0%, respectively. LC-MS/MS analysis is more complex with semi-automated sample preparation and independent analytical instrumentation, therefore special care was given to quality assurance. This involved three evaluation steps: (1) SST, (2) IQC evaluation, and (3) interpeptide agreement analysis.(3) 9.1% of the batches failed the system suitability check. In which case, samples were transferred to an alternative LC-MS/MS, while the malfunctioning LC-MS/MS was troubleshooted. IQC evaluation for LC-MS/MS was performed with adjusted Westgard rules to fit a multiplex test: (1) 2x2SD within the same IQC lot for the same peptide, (2) 10x2SD overall for the whole batch, and (3) 1x3SD for quantifying peptides only. 10.5% of the batches violated one of the rules, of which rule 1 was the most stringent (7.3% rejection). Repreparation of these batches indicated that 4.7% of the batches violating rule 1 were accurately rejected. Rule 2 and 3 did not have additional benefit. Overall, LC-MS/MS IQC evaluation of 41,448 datapoints showed an excellent average between-run CV of 2.9% (6495C) and 4.5% (6495A), ranging from 1.7-7.6% across nine apolipoproteins. 160 samples were reprepared based on interpeptide discrepancies; of these 152 samples were reconfirmed in a second measurement. Overall, an average interpeptide agreement of the 23,376 samples showed a Pearson R of 0.981, with all apolipoproteins above 0.975 except ApoC-I.

Cobas and Diazyme analyzers were bidirectionally connected to the LIMS and data transfer was automated. A tailored semi-automated R script was designed for the LC-MS/MS-based data to automatically evaluate the data against the performance criteria for SST, IQC limits and interpeptide agreement. This script automatically made the decisions based on the predefined criteria. Furthermore, a dashboard depicting the progress for all the different stages of the process was generated.

Conclusion
This is the first quantitative Clinical Chemistry Proteomics (qCCP) trial at this scale performed in a diagnostic medical laboratory, which meets both the test process requirements as well as the predefined analytical performance criteria that make medical tests fit-for-clinical purpose. Tailored additions to the TTP was essential and ensured high-quality data for Cobas, LC-MS/MS, and Diazyme measurements. In conclusion, our qCCP implementation research on CVD biomarkers successfully demonstrates the feasibility of high-volume LC-MS/MS analyses in a large clinical trial.

References
1. Szarek M, Reijnders E, Jukema JW, Bhatt DL, Bittner VA, Diaz R, et al. Relating Lipoprotein(a) Concentrations to Cardiovascular Event Risk After Acute Coronary Syndrome: A Comparison of Three Tests. Circulation. 2023;10.1161/CIRCULATIONAHA.123.066398.
2. Cobbaert C, Albersen A, Zwiers I, Schippers P, Gillis J. Designing a diagnostic Total Testing Process as a base for supporting diagnostic stewardship. Clin Chem Lab Med. 2020;59(3):473-89.
3. Smit NPM, Ruhaak LR, Romijn F, Pieterse MM, van der Burgt YEM, Cobbaert CM. The Time Has Come for Quantitative Protein Mass Spectrometry Tests That Target Unmet Clinical Needs. J Am Soc Mass Spectrom. 2021;32(3):636-47.