Better health care through mass spectrometry – better mass spectrometry through standardization

by Prof. Michael Vogeser, Dr Judy Stone and Prof. Alan Rockwood
While analytical standardization and metrological traceability are well-defined terms, ‘methodological standardization’ in clinical mass spectrometry is still in a developing stage. We propose a framework that facilitates the widespread implementation of this highly complex and very powerful technology and is based on two pillars – standardization of the description of LC-MS/MS methods and standardization of the release of clinical test results as a three-step sequence of method validation, batch validation and validation of individual measurements.
Mass spectrometry in the clinical laboratory
Mass spectrometry (MS)-based methods now play an important role in many clinical laboratories worldwide. To date, areas of application have focused especially on screening for hereditary metabolic diseases, therapeutic drug monitoring, clinical toxicology and endocrinology. In fact, these techniques offer significant advantages over immunoassays and photometry as basic standard technologies in clinical chemistry: high analytical selectivity through true molecular detection; wide range of applications without the need for specific molecular features (as in UV detection or specific epitopes); high multiplexing capacity and information-rich detection; and, in many cases, matrixindependent analyses, thanks to the principle of isotope dilution [1].
Various MS technologies – in particular tandem MS (MS/MS-coupling with molecular fragmentation), time-of-flight (TOF) MS and Orbitrap-MS – with front-end fractionation technologies such as HPLC or UPLC potentially allow very reliable analysis, but the technology itself is no guarantee of this: these techniques have a very high complexity and a wide range of potential sources of error [2] which require comprehensive quality assurance [3–5]. Indeed, the high degree of complexity is still the main hurdle for the application of MS in the special environment of clinical laboratories. Specific challenges of this type of laboratory – in contrast to research and development laboratories – include: heterogeneous mix of staff qualifications; requirement for maximum handling safety when operating a large number of analysis platforms; work around the clock; and direct impact on the outcome of the individual patient.
Indeed, after more than two decades of commercial availability of LC-MS/MS instruments, their application in a global perspective has remained very limited. The translation of MS into fully automated ‘black box’ instruments is underway, but still far from being realized on a large scale [6], with laboratory developed tests (LDTs) still dominating the field of clinical MS applications. Kit solutions for specific analytes provided by the in vitro diagnostics (IVD) industry are becoming increasingly available, but their application also requires a very high level of skills and competence from laboratories.
Two main differences of MS-based LDTs as opposed to standard ‘plug-and-play’ analysis systems in today’s clinical laboratories can be identified: first, the high heterogeneity of device configurations and second, the handling of large amounts of data, from sample list structures to technical metadata analysis.
In fact, the random access working mode is now so widespread in all clinical laboratories that the ‘analytical batch’ is no longer standard in laboratories. In the same way, modern analytical instruments no longer challenge the end users with extensive metadata (such as reaction kinetics or calibration diagrams). To achieve the goal of making the extraordinary and disruptive analytical power of MS fully usable for medicine to an appropriate extent, approaches to master the heterogeneity of platform configurations and to regulate the handling of batches and metadata are urgently needed – and standardization efforts seem to be crucial in this context.
Standardization of the method description
IVD companies manufacture many different instrument platforms, but each of these platforms is very homogeneous worldwide and is produced in large quantities for years. In contrast, MS platforms in clinical laboratories have to be individually assembled from a very large number of components from many manufacturers (sample preparation modules, autosamplers, high performance pumps, switching valves, chromatography columns, ion sources, mass analysers, vacuum systems, software packages, etc). As a result, hardly any two instrument configurations in different laboratories correspond completely with each other. This makes handling very demanding for operators, maintenance personnel, and service engineers.
Methods implemented on these heterogeneous platforms (e.g. instruments from various vendors) are in turn characterized by a very considerable number of variables, e.g. chromatographic gradients, labelling patterns of internal standards, purity of solvents, dead volume of flow paths, etc.
Taken together, the variety of assays referring to an identical analyte (such as tacrolimus or testosterone) is enormous, with an almost astronomical combinatorial complexity.
However, method publications are still traditionally written more or less in a case report approach: the feasibility and performance of a method realization is demonstrated for one individual system configuration. It is usually not clear which features are really essential for the method and which features can be variable between different implementations – and which second implementation can still be considered ‘the same’ method. This means that the question of the true ‘identity’ of a method has not yet been deepened by application notes or publications in scientific journals; thus the level of abstraction required here is missing.
In an attempt to standardize the description of MS/MS-based methods, we selected a set of 35 characteristics that are defined as essential for a method (see Table 1) [7], for example, main approach of sample preparation (e.g. protein precipitation with acetonitrile), main technique of ionization (e.g. electrospray ionization in negative mode); molecular structure of the internal standard; mass transitions; calibration range. In addition, we define 15 characteristics of a method that cannot or should not be realistically standardized in time and space (examples: manufacturer and brand of the MS detector; dead volume of the flow path; lot of analytical columns and solvents). These characteristics – identified as variable – should be documented in the internal report files.
We found it feasible to describe several exemplary MS/MS methods using this scheme and a corresponding matrix. On the basis of this matrix, the method transfer to different platforms and laboratories will be much easier and more reliable. Specifying the identity of a method in the proposed way has the essential advantage that a method revalidation can be transparently triggered by defined criteria, e.g. the use of a novel internal standard with a different labelling pattern.
The proposed scheme for method description may also be the basis of a comprehensive traceability report for any result obtained by an MS-based method in the clinical laboratory.
Standardization of batch release (Table 2)
While today’s routine analyser platforms essentially provide unambiguous final results for each sample, the process of generating quantitative results from primary data in MS is open and transparent. Primary data in MS are the peak areas of the target analyte observed in diagnostic samples. In addition to these primary data, a range of metadata is provided (e.g. internal standard area, peak height-to-area, peak skewness, qualifier peak area; metadata related to analytical batches, e.g. coefficient of variation (CV) of internal standard areas). This transparency and abundance of data is a cornerstone of the high potential reliability of MS-based assays and therefore their interpretation is very important [8, 9].
However, the evaluation of this metadata – related to individual samples and batches – is nowadays done very heterogeneously from laboratory to laboratory [10]; this applies to LDTs as well as to commercially available kit products. The structure of analytical batches is also very variable and there is no generally accepted standard (number and sequence of analysis of calibration samples in relation to patient and quality control samples, blank injections, zero samples, etc).
While the validation of methods – which is performed before a method is introduced into the diagnostic routine – is discussed in detail in the literature (and in practice), the procedures applied to primary data before release for laboratory reporting have not yet been standardized. Validation is generally defined as the process of testing whether predefined performance specifications are met. Therefore, quality control and release of analytical batches and patient results should also be considered a process of validation, and criteria for the acceptance or rejection of results should be predefined.
A three-step approach to validation, covering the entire life cycle of methods in the clinical laboratory, can be conceptualized: dynamic validation should integrate validation of methods, validation of analytical batches and validation of individual test readings. We believe that standardization of this process of batch and sample result validation and release is needed as a guide for developers of methods, medical directors, and technicians.
In a recent article published in Clinical Mass Spectrometry [11], we propose a list of characteristics that should be considered for batch and sample release. In this article we only mention figures for merits and issues to be addressed and do not claim to have specific numerical acceptance criteria. Therefore, this generic list of items is intended as a framework for the development of an individual series and batch validation plan in a laboratory. Furthermore, we consider this list to be a living document, subject to further development and standardization as the field matures.
We believe that it is essential to include basic batch and sample release requirements as essential characteristics in the description of a method [7]. Therefore, we believe that efforts to standardize method description and batch/sample release should be synergistically linked to facilitate the use of MS in routine laboratories.
The approach proposed to clinical MS in these two companion articles [7, 11] can be the basis for discussion and eventually for the development of official standards for these areas by the Clinical and Laboratory Standards Institute (CLSI) and/or International Organization for Standardization (ISO). We believe that these documents can provide a solid basis for internal and external audits of LC-MS/MS-based Quality Control April/May 2020 9 | LDTs, which will become particularly relevant in the context of the IVD Regulation 746 in the European Union [12].
Both approaches – standardized description of MS methods and standardization of batch release – aim at implementing methodological traceability. This corresponds to the analytical standardization and metrological traceability of measurements to higher order reference materials [13, 14].
Future perspectives
In the future, a commercialization of MS-based black-box instruments on a larger scale is expected. However, LC-MS/MS will remain a critical technique for LDTs, and the flexibility of MS to develop tests on demand – independent of the IVD industry on fully open LC-MS/MS platforms – will remain a key pillar of laboratory medicine.
Both publications, which this article puts into context [7, 11], have been published in Clinical Mass Spectrometry, the first and only international journal dedicated to the application of MS methods in diagnostic tests including publications on best practice documents. Both articles are freely available.
Clinical Mass Spectrometry is the official journal of MSACL (The Association for Mass Spectrometry: Applications to the Clinical Laboratory; www.msacl.org). MSACL organizes state-of-the-art congresses that focus on translating MS from clinical research to diagnostic tests (i.e. bench to clinic).
In summary, we advocate innovative approaches to methodological standardization of LC-MS/MS methods to master the complexity of this powerful technology and to facilitate and promote its safe application in clinical laboratories worldwide.
The authors
Michael Vogeser*1 MD, Judy Stone2 PhD, Alan Rockwood3 PhD
1 Hospital of the University of Munich (LMU), Institute of Laboratory Medicine, Munich, Germany
University of California, San Francisco Medical Center, Laboratory Medicine, Parnassus Chemistry, San Francisco, CA, USA
3 Rockwood Scientific Consulting, Salt Lake City, UT, USA

* Corresponding author
E-mail: Michael.Vogeser@med.uni-muenchen.de