Developments in reference measurement systems for C-reactive protein and the importance of maintaining currently used clinical decision-making criteria

Of current inflammatory markers identified, C-reactive protein (CRP) has the assay and biological characteristics most conducive to use in clinical practice [1]. Its utility, strongly supported by several consensus reports [2], has been recently confirmed during the SARS-Cov-2 pandemic to both predict in-hospital death and exclude the need for admission in intensive care unit in coronavirus disease 2019 patients [3, 4]. As described in more detail in another contribution published on this journal issue [5], the road towards harmonization of CRP measurements has represented a challenging enterprise asking for about 20 years of efforts before to see a positive outcome on marketed assays [6]. Figure 1 displays the calibration hierarchy currently in place, using as common calibrator a commutable liquid frozen reference material with CRP entirely in the native status (ERM-DA474/IFCC), traceable to CRM 470 (later renamed to ERM-DA470) in turn calibrated with the WHO International Standard 85/506 (human CRP in pooled normal human serum) to which values had been assigned by using a pure protein as calibrant. Using this metrological traceability approach, the inter-assay variability, as witnessed by published data, is currently very good, with CV values in External Quality Assessment programs permanently below 10 %, and values close to 3 % if only the widely used commercial measuring systems are considered [7], [8], [9]. Importantly, this translates to the practical possibility to use common decision thresholds for the CRP clinical use minimizing result misinterpretation (Figure 2).

Figure 1:

Metrological traceability chain currently adopted by the great majority of in vitro diagnostic manufacturers providing commercial assays for C-reactive protein (CRP).

Figure 2:

Decision thresholds (in mg/L) for C-reactive protein as a marker of inflammation.

In the recent years, new CRP candidate reference materials (RM) were nominated to be listed in the Joint Committee for Traceability in Laboratory Medicine (JCTLM) database. JCTLM was however concerned about the impact of the materials for the laboratory medicine community that the JCTLM database serves. It was indeed unclear in which position of the current traceability chain for CRP shown in Figure 1 should these RM be inserted or if they belong to newly proposed traceability chains in which these materials are traced to the International System of Units (SI) through, e.g., the amino acids RM and the hydrolysis-isotope dilution mass spectrometry (IDMS) procedure. In other words, it was unclear if the new CRP candidate RM providers proposed an alternative calibration hierarchy to that currently used in the clinical community, already listed on the JCTLM database [10]. If this is the case, JCTLM recommended that further investigations should be carried out to understand the implications of introducing a new traceability chain for CRP for patient clinical results before proceeding with adoption of newly characterized RM and listing them on the JCTLM database.

The definition of the measurand, as quantity intended to be measured [11], is one of the major prerequisites needed for the application of the metrological traceability theory [12]. Once defined, the analytical selectivity towards the measurand should be maintained throughout procedures used in the entire traceability chain, and employed primary and secondary RM should contain the same ‘analytical target’. Concerning CRP, an agreement about the measurand definition can be easily reached: it is a non-covalent pentamer with identical subunits, each consisting of a single 206 amino acid chain and a relative molecular mass of ∼23,050, with minimal known post-translational modifications. However, problems observed for ‘CRP measurand’ in different type of candidate RM have been reported. Rzychon et al. showed that the lyophilization process resulted in a loss of ∼20 % of measurable CRP compared to the non-lyophilized material, even if both formats appeared commutable [13]. The presence of a monomeric form of CRP only in lyophilized materials, due to the partial dissociation of CRP, affected the signal response of different commercial immunoassays showing the centrality of the CRP aggregation status in the accurate measurand recovery. The lesson to be learned was clear: for RM to be used in standardization of CRP, changes in the processing steps may have a pronounced impact on the measurement results and would consequently lead to a biased calibration, even if the RM is commutable [14, 15]. Hence, a careful evaluation of the impact of different RM formats and processing steps on the properties of the RM and its suitability for calibration is a crucial requirement. Furthermore, the presence of a multimeric heterogeneity may affect different measurement procedures to a different extent: for instance, in IDMS-based procedures all polypeptide chains are measured, including ‘non-native’ ones [16]. This would not describe the properties of native CRP as ‘target analyte’ completely, leading to a significant bias if IDMS methods would have been used anywhere in the traceability chain, e.g., as secondary reference measurement procedure (RMP) [14, 17].

Table 1 shows the requirements to be considered and correctly fulfilled in the preparation of a CRP RM to be used as common calibrator of commercial assays. In addition to control all relevant parameters discussed above that may modify the measurand definition, the commutability of candidate RM should be assessed by following the available recommended approaches [18], [19], [20]. Commutability is the ability of a RM to show inter-assay properties comparable to those of human samples. Only commutable materials can be used for direct value assignment to manufacturers’ calibrators, having large importance to ensure an unbroken traceability chain [21]. Non-commutability of a material means that using it for calibration will introduce a bias in the calibrated procedure, with wrong results for clinical samples and the risk of incorrect medical decisions. In developing higher-order RM, focusing on allowable measurement uncertainty (MU) is also important. As discussed elsewhere, the MU specifications of RM are defined by the performance specifications of the MU on clinical samples [10, 22, 23]. Finally, an investigation on possible effects for patient result interpretation with the introduction of new higher-order references for CRP is mandatory, and a comparison of measurement results obtained with the new calibration hierarchy to results of measurements obtained with the previous calibration should be established [24]. Already 15 years ago, I highlighted the importance of maintaining currently used clinical decision-making criteria and related clinical experience even when a new traceability chain incorporating theoretically better metrological aspects is proposed [12].

In summary, new candidate RM or RMP for CRP cannot stand alone, but it must be integral parts of a reference measurement system, the current in use or a new one, and this must be clearly disclosed. CRP is a pentameric protein, and the pentamer is the clinical measurand. In its monomeric form CRP tends to aggregate. As most of the RM characterizations are carried out by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and matrix-assisted laser desorption/ionization-time of flight (MALDI-ToF) on the monomeric form, it is recommended using other methods to check for the pentamer. The RM intended use requires validation of its usability for the purpose. If the intended use of new candidate RM includes the calibration of field methods used in medical laboratories, this requires commutability studies involving parallel measurements of the candidate RM and clinical samples for all commercial assays with which it may be potentially used as calibrator for implementing metrological traceability. However, commutability is an indication for the degree of harmonization achievable but not necessarily for the absence of bias between different calibration hierarchies, an aspect that should be tested by appropriate extent-of-equivalence studies.