Harmonization of units and reference intervals of plasma proteins: state of the art from an External Quality Assessment Scheme

Units

Harmonized reporting of units remains an important challenge to be undertaken by the clinical laboratory community. The EFLM WG-H has begun a campaign, articulated in various steps, to promote the harmonization of the units of measure in the report (step 1: changing from mL to L as unit of volume; step 2: changing to the liter for reporting protein concentrations; step 3: promotion of the use of mmol/L for reporting electrolytes and minerals).

Particularly, step 2 foresees that all laboratories still reporting plasma proteins in mg/dL or g/dL must change to mg/L or g/L. As indicated by Dybkaer and Jorgensen 50 years ago [19], in fact, the recommended unit of volume is liter. EFLM WG-H proposed 31st October 2016 as the deadline for the implementation of this step. At a national level, the Scientific Societies of Clinical Chemistry and Laboratory Medicine play an important role in supporting this campaign [20].

Reference intervals (RIs)

The harmonization process of RIs seems to have an uphill road [21]: despite their role in providing clinical information to guide clinicians in their decision-making, there appears to be little conformity in laboratory practice, and differences in RIs persist between laboratories that use the same platforms and reagents [22], [23].

The same RI across different assays for a specific analyte could be optimally shared when there is sound calibration traceability, and evidence from a between-method comparison shows that bias would not prevent the use of a common RI [24], [25].

In the field of plasma proteins, after the use of the CRM470, which has resulted in significant changes in reference values for some proteins, studies on plasma proteins RIs began to be performed in different geographical areas [26], [27], [28], and in 1995, several professional societies and diagnostic companies agreed to use interim consensus reference ranges for all immunochemical methods, independently of the system or instrument used, until further studies documented more accurate values [29], [30]. As a part of the latter, the IFCC Committee for RIs and Decision Limits (C-RIDL) began a study to evaluate protein concentrations in serum for several racial and ethnic groups in different geographical locations [31] and recently arranged a global multicenter study to explore rational and harmonizable procedures for derivation of RIs and investigate the feasibility of sharing RIs, which includes also some plasma proteins such as ALB, IgA, IgG, IgM, C3, C4 and CRP [32], [33], [34], [35].

In addition, EFLM recently began a campaign to improve the harmonization of RIs used by European laboratories, whose first step is a survey to understand which are the origin of the RIs presently in use and if partitioning criteria for the use of RIs are the same all over Europe.

Aim

The aim of this work was to analyze the units and the RIs used by laboratories participating in the EQA program of the Centre of Biomedical Research (CRB) in reporting plasma proteins. Moreover, the inter-laboratory variability of the 13 certified proteins was evaluated to report the state of the art and changes in time of their measurement.

Units

Table 1 shows the units utilized by participants in 2017.

For ALB, ~66% of laboratories still use dL instead of L. Only for FLC the totality of laboratories report the result with the recommended units (mg/L), while for TLC and RBP, 37.1% and 9.1%, respectively, still utilize mg/dL. For CRP, mg/L and mg/dL are used equally; for B2M, four different units are used: mg/L (82.0%), mg/dL (8.0%), μg/mL (8.0%) and ng/mL (2.0%); for all the other proteins, most laboratories (~70%, range 63.9%–76.6%) still express the result in mg/dL.

In 2016, before the deadline, the state of the art was nearly identical: only two laboratories shifted from mg/dL to g/L.

Reference intervals

Table 2 reports the RIs utilized by the participants, related to an adult male (45 years old), for all the plasma proteins included in the EQA scheme.

For Igs, ~60% of laboratories utilized the same RI: 0.7–4.0 g/L for IgA, 7–16 g/L for IgG and 0.4–2.3 g/L for IgM.

The more commonly used RIs for the others certified proteins were: C3=0.9–1.8 g/L (71%), C4=0.1–0.4 g/L (80%), AAT=0.9–2.0 g/L (56%), TTR=0.2–0.4 g/L (73%), AAG=0.5–1.2 g/L (63%), A2M=1.3–3.0 g/L (50%), HPT=0.3–2.0 g/L (59%), TRF=2.0–3.6 g/L (66%) and CRP=<5 g/L (66%).

For ALB, it was not observed a RI employed with a large prevalence by participants; the RIs mostly used were: 35–50 g/L (26%), 35–52 g/L (23%) and 34–48 g/L (18%).

Only 19.4% of participants employed nephelometric or turbidimetric methods while 80.6% utilized a colorimetric method, precisely 59.6% the bromocresol green method (BCG) and 21.0% the bromocresol purple method (BCP). The more commonly used RI among the participants utilizing immunometric methods was 35–52 g/L (42%), while 35–50 g/L was the most used RI both for BCG (27%) and for BCP (31%) methods (Table 3).

Laboratories primarily used the RI reported in the manufacturers’ packaging inserts of the commercial kits (Table 4), but for each protein, we observed a wide variability of others RIs, also among the laboratories using the same analytical method (Figure 1).

Figure 1:

Graph reporting the RIs (blue rectangles) of participants to CRB External Quality Assessment program for plasma proteins and results (black circles) obtained on one control sample.

Inter-laboratory variability

For the 13 certified proteins, the results of the last five cycles of EQA showed the mean CV% reported in Table 5. Mean CV% was: <5% for IgG, C3, ALB, HPT and TRF; <10% for IgA, IgM, AAG, TTR, A2M and CPR; <13% for C4 and AAT.

HPT and ALB presented the lowest variability with a mean CV% of 3.82%±0.38% and 3.95%±0.88%, respectively.

The mean CV% of TTR resulted in 9.53% due to 2014 results (CV%=13.10%) because in the other years, it remains <10%. Instead, the inter-laboratory variability of AAT, despite the drop during 2013 to 2017 (CV%=16.41%, 13.17%, 11.33%, 11.22%, 9.62%, respectively), it remains the highest, with a mean CV%=12.35%. The inter-laboratory variability of C4, constantly over 11.5% until 2016, decreased to 8.5% in 2017.

The inter-laboratory variability over the time is reported in Figure 2. We observed a substantial and progressive reduction for C3, ALB, A2M, HPT and TRF.

Figure 2:

CRB External Quality Assessment program for plasma proteins: mean coefficients of variation (CVs).

AAT, α₁-antitrypsin; TTR, prealbumin or transthyretin; ALB, albumin; AAG, α₁-acid glycoprotein; A2M, α₂-macroglobulin; HPT, haptoglobin; TRF, transferrin, CRP, C-reactive protein.

Units

The harmonization in reporting measurands with the same units is a universal problem. On a local basis, a pragmatic solution should be sought, at least by different laboratories in the same geographical area, in order to obviate the reporting of confusing results.

At a European level, the EFLM WG-H campaign foresaw that all laboratories still reporting plasma proteins in mg/dL or g/dL would have to change to mg/L or g/L. In particular, it suggested to change from mg/dL to mg/L for B2M, κ and λ FLC and CRP; from g/dL to g/L for ALB and from mg/dL to g/L for AAT, AAG, A2M, C3, C4, CER, HPT, Ig A, G, M, TTR, RBP and TFR.

At a national level, the Società Italiana di Biochimica Clinica e Biologia Molecolare Clinica (SIBioC) supported this campaign [20] and invited all members to make these changes within 31st October 2016. At the same time, all EQAS providers were asked to change the units in their periodical reports. The CRB, which already used L as the unit of volume for all proteins, following the indications of the EFLM, changed the unit for C4, TTR and CER from mg/L to g/L, except for RBP that remained in mg/L. Moreover, the CRB communicated the reason of the changes to participants with a letter, encouraging them, at the same time, to make the change from mg/dL to g/L or mg/L.

Despite these suggestions, the census of participants’ units used in 2017 demonstrated that their harmonization is far from coming. In Italy, only ~30% of laboratories used the recommended units of measure, and different units were used to express the same concentration of protein, for example, mg/L, g/L and mg/dL for TTR. This may lead clinicians to wrong interpretations, risking patients’ safety.

Why are the laboratories so reticent to change? There are many aspects to be considered: first of all, changing old habits is difficult both for the laboratorist and for the clinician; second, a series of actions have to be undertaken before and during the change.

A particularly troublesome aspect is when the units cause a change in numeric values, which may lead physicians to misinterpretation. The change from dL to L introduces a 10- or 100-fold modification of the numbers and results will increase 10 times for B2M, CRP (from mg/dL to mg/L) and for ALB (from g/dL to g/L) while results will decrease by 100-fold for AAT, AAG, A2M, CER, HPT, Ig A, G, M, TTR and TFR.

Moreover, going from mg/dL to g/L, the values reported in whole numbers must be reported in decimal places and it could be an issue that hinders change, in particular for some of the lower concentration proteins, for example, C4, TTR, CER and RBP must be reported with three decimal places rather than in whole numbers.

The following planning and actions should be undertaken by laboratories when changing units from mg/dL to mg/L causing a 10-fold increase or when changing units from mg/dL to g/L causing a 100-time reduction: (1) synchronized adjustment of analyzer and computer systems; (2) communication and liaison with all service users; (3) updating of all documentation.

Communication to hospital users and general practitioners is fundamental; they should be informed of the intention to change units of measurement. EFLM suggests to insert a message with every report for a period of time prior to the change to provide advance notification: “Please, note: from XX.XX.XX, protein xyz, results will be reported in g/L (or mg/L) instead of mg/dL in line with the national and international guidelines. This means, for example, that a transferrin currently reported as 300 mg/dL will be reported as 3.0 g/L”. Moreover, from the moment of the change, a standard comment to every report sent out for a period of 12 months is suggested: “Please note new units and the change of reference intervals”.

Thus far, there has not been the desired improvement toward harmonization of units, but this should not let the scientific community and the EQAS providers give up encouraging the laboratories to implement all the activities necessary for the change.

Reference intervals

Harmonization of RIs is a continuing project that aims to create uniform interpretation of results and prevent misdiagnosis caused by a greater variation in RI than in the measurement result. However, the situation is complex: regulations, e.g. the European Directive on in vitro diagnostics device [36], lay down that manufacturers should mention RIs in the package inserts, and clinical laboratories, seeking accreditation for compliance with ISO 15189:2012, need to demonstrate that RIs communicated to all users of the laboratory service are appropriate for the patient population served and for the measurement systems used [37]. Laboratories should also verify if they use RIs from literature or manufacturers: in fact, although the International Organization for Standardization (ISO) and Clinical and Laboratory Standard Institute (CLSI) encourage every laboratory to establish its own RIs [38], producing RIs is actually too expensive and a heavy task for most laboratories. The use of the same RI for a specific analyte, for homogeneous populations shared by all laboratories within a country or a region is possible if there are some prerequisites to determine common RIs, one of them is the use of analytical methods traceable to a reference system [39]. If a reference system exists for calibration traceability and well-characterized RIs obtained with a well-standardized measurement procedure are published, the manufacturer can provide those RIs without the need of a new RI study, on condition that its measurements procedure has the same level of standardization [25].

The results of our study show that laboratories primarily used the RIs reported in the packaging inserts of the commercial kits, which, for the 13 certified proteins, should be the same since the manufacturers declared to be traceable to ERM-DA470k/IFCC, but the evidence demonstrates otherwise: Roche cobas, Beckman AU, Siemens BN and Siemens Vista all use IFCC-recommended RIs, but Abbott Architect and Beckman Immage RIs are different because they suggest RIs deriving from a manufaturer’s study on a small groups of subjects. These last RIs should be revised on the basis of larger studies.

Few multicenter RI studies on proteins, measured with a specific reagent system, are available and results are slightly different from the expected value in the manufacturer’s package [40], [41], but it can be seen, for example, that the results of IgA, IgG, IgM and C3 from Beckman AU system aligned well with the results of the Beckman Immage system; hence, for these proteins, common RIs could be given on the two platforms [41].

For ALB, a wide variability of RIs was observed, even among the laboratories using the same analytical method. In particular, the BCP method reads 2 g/L lower than the BCG method; thus, RIs should be different [42], whereas the most used RI for both colorimetric methods was the same: 35–50 g/L.

In addition, for each other protein, a large variability of other RIs exists, even among laboratories using the same analytical system.

To promote harmonization, CRB EQAS provides information on the RIs used by participating laboratories, thus enabling them to verify the appropriateness of their RIs. The laboratory can compare its own RI with those of laboratories using the same or different analytical systems.

We do not recommend using a specific RI, but when a laboratory highlights a lack of agreement between its RI and those of other laboratories using the same analytical system, CRB advises it to verify the correctness of RI employed. For example, a laboratory may change its analytical system, but fails to review the RIs applied, resulting in an incorrect clinical significance.

This extremely important aspect cannot be ignored by EQAS, and organizers must strongly encourage laboratories to carefully verify the reliability of their values [43].

Inter-laboratory variability

The overall inter-laboratory analytical variability, which included both within- and between-manufacturer variances, was very different among the studied proteins, ranging from 3.8% to 12.4%, and decreased in the years for most of them. Results of the present study and those previously found [44], [45], [46] are essentially in agreement with total between-laboratory variability observed in the UK National External Quality Assessment Scheme (UK NEQAS) until 2009 [17].

Certainly, the introduction of ERM-DA470 has resulted in a sudden reduction of the between-assay variation for the majority of the certified proteins [17]; however, persisting bias among manufacturers were reported shortly thereafter for some proteins [47].

Method discrepancy remains a significant gap between different analytical systems and laboratories: for example, results of a recent study highlight that harmonization of the results of three immunoassays (Beckman Immage, Siemens BNI and Roche cobas) for serum Ig measurement has not been completely achieved, even if their traceability to ERM-DA470 [48]. The causes of the differences observed among systems in the measurements of EQA materials and individual patient serum samples, excluding commutability problems, may be related to the nature of antibodies used in the assays with various differences in epitope recognition. In addition, the process of value transfer from the ERM-DA470k/IFCC to the manufacturer’s master calibrators, and subsequently to product calibrants, may somehow deviate from the IFCC protocol [49].

Additionally, for ALB, despite the low inter-laboratory analytical variability observed (CV%=3.95), studies on measurement accuracy and clinical use of ALB demonstrated that the dominating error component was bias for most measurement procedures and that significant differences among immunochemical, BCG and BCP methods compromise interpretations of serum ALB results [42], [50].