= Emerging. More than 5 years before clinical availability. (24.37%, 2023)
= Expected to be clinically available in 1 to 4 years. (39.50%, 2023)
= Clinically available now. (36.13%, 2023)
MSACL 2023 : Geistanger

MSACL 2023 Abstract

Self-Classified Topic Area(s): Data Analytics > Assays Leveraging MS

Podium Presentation in Steinbeck 3 on Thursday at 15:15 (Chair: Stephen Master / Bo Burla)

Uncertainty Calculations for Reference Measurement Procedures

Andrea Geistanger (1), Galina Babitzki (1), Friederike Bauland (2), Judith Taibon (1)
(1) Roche Diagnostics GmbH, Nonnenwald 2, 82377 Penzberg, Germany,(2) Chrestos Concept GmbH & Co. KG, Girardetstr. 1-5, 45131 Essen, Germany

Andrea Geistanger, PhD (Presenter)
Roche Diagnostics GmbH

Presenter Bio: Andrea Geistanger is Head of MassSpec Biostatistics, at Roche Diagnostics in Germany. Her department of biostatisticians supports system and assay development through the whole life cycle of Roche’s cobas products. Her team is involved in the early development phases, including biomarker search projects with machine learning and multivariate statistics analysis. During product development phases, Andrea’s data analysts support scientists in experimental planning with Design of experiments, as well as in the experiment of validation studies according to regulatory requirements. Furthermore, they develop standardization schemes and calibration concepts for cobas analyzers. Throughout the development phase, software tools are designed and developed as needed. These programs are also made available to a broader community through open software projects.

Andrea Geistanger recently gave a talk at MSACL Connect on the mcr and VCA R packages for method comparison and precision analysis. That talk was dedicated to statistical tools, the actual one will address the soft topics of these experiments, as study design, analysis and interpretation.


Quantitative results based on reference measurement procedures require the calculation of the uncertainty of this value. To obtain this uncertainty the different error sources have to be taken into account and combined.

This talk will explain different ways for calculating the uncertainty of a liquid chromatography/mass spectrometry (LC/MS) reference measurement procedure, showing the differences between Type A and Type B uncertainties and best strategies to choose from them. In addition, combinations of results from subsequent steps to a final overall uncertainty are presented.

A LC/MS reference measurement procedure requires as a starting point a primary material with known purity of the analyte under consideration. Nowadays this purity assessment is often done with NMR technology. However, as NMR is a measuring method, the purity estimate carries an uncertainty. Here we use the Type A uncertainty approach, based on replicate NMR measurements. In addition, we will shortly introduce a Bayesian-approach for NMR uncertainty calculation and do a comparison of both techniques.
With this primary material a first set of primary calibrators is produced, often through multiple steps of preparation of stock solutions, weighings, dilutions etc. For these steps balances, volumetric flasks or pipettes are used, all of them coming with an error budget which has to enter the uncertainty budget of the primary calibrator preparation. We will show how this can be done through Type B uncertainty calculation.

Based on the primary calibrators the LC/MS instrument is calibrated and measurements of individual samples are carried out. We will introduce a hierarchical variability experiment, which takes into account the most important error sources and results in a Type A uncertainty for the LC/MS measurements.
Finally, the errors of the LC/MS measurements have to be combined with those of the primary calibrator preparation to obtain the overall uncertainty of an LC/MS reference measurement procedure.
RESULTS: We will show the approaches and calculation strategies based on an analyte of the Roche LC/MS Reference Measurement Procedure development project.

Uncertainty calculation for LC/MS reference measurement procedures requires the modularisation of the different execution steps, starting from the primary reference material until the final concentration value. Different strategies have to be applied and combined to obtain the final uncertainty. We have developed a sound framework for the different steps, including the management of the different input variables up to comprehensive reports of the calculations.

Financial Disclosure

SalaryyesRoche Diagnostics GmbH
Board Memberno
Stockyes Roche Holding
IP Royaltyno

Planning to mention or discuss specific products or technology of the company(ies) listed above: