Immunoassays for free light chain measurement

From Wikilite

4.1. Introduction

There are numerous assays for monoclonal free light chains (FLCs) in urine (Table 4.1). Some, such as protein precipitation, are simple screening tests but are insensitive and non-specific, while others such as immunofixation electrophoresis (IFE) are more sensitive but laborious and may be difficult to interpret.

Some commonly used tests for FLCs are assessed in Table 4.1. Each assay has features that are to some degree satisfactory, but all fail to detect FLCs in serum at concentrations within the normal range (3-25mg/L), or even at concentrations that are several times higher than the normal range. Dipstick assays, based upon dye uptake, are particularly unreliable for measuring cationic proteins such as FLCs and should not be used.

One modern solution to the problems of inadequate assay specificity and sensitivity is to use antibody-based methods. The following chapter describes the development of sFLC immunoassays and their validation on a variety of laboratory analysers.

4.2. Immunoassays for free light chains

A. Polyclonal antisera versus monoclonal antibodies

Figure 4.1 Immunoelectrophoresis showing the specificity of κ FLC antisera. The anti ĸFLC antibody shows no cross reaction to proteins in normal human serum, including intact immunoglobulins (bottom well). Good anti-κ FLC activity is demonstrated by the presence of the arc against the purified κ chains. (Serum = normal human serum, κ chain = purified κ light chains).

Figure 4.2 Western blots showing the specificity of polyclonal FLC antisera compared with Mabs, and the reaction of polyclonal antisera against FLC monomers and dimers; all separated by non-reducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). (Lane 1, molecular weight markers. Lanes 2 & 3, urine containing κ FLCs and Lanes 4 & 5, normal serum; all probed with mono- and polyclonal anti-κ. Lanes 6 & 7, urine containing λ FLCs and Lanes 8 & 9, normal serum; all probed with mono- and polyclonal anti- λ. Lanes 10 & 11, polyclonal FLC antisera reacting with monomers and dimers of κ and λ).

Figure 4.3 Haemagglutination assays showing the specificity of κ FLC antisera against RBC coated with purified FLCs and IgG.

Figure 4.4 Specificity of (A) κ and (B) λ FLC antisera assessed by interference with the results of typical nephelometric assays upon the addition of various substances. Mean and 95% confidence limits for each added substance are shown.

It is essential that FLC immunoassays utilise antibodies that have high specificity and affinity. Early FLC immunoassays (Chapter 5) used polyclonal antisera but good specificity was difficult to obtain. Monoclonal antibodies (Mabs) seemed to be the obvious solution to the problem although considerable effort by my colleagues in the Immunology Department at The University of Birmingham failed to produce antibodies that reliably recognised a full range of monoclonal FLCs. Others have also reported Mab-based immunoassays that failed to have clinical impact, possibly for similar reasons ^{[15][16][17][18][19]}.

For practical reasons, our research focussed on optimising polyclonal FLC antisera. The following description is an outline of the successful procedures involved in the development of Freelite™ sFLC assays. Details of the many steps involved are proprietary and are not therefore available. However, in summary, sheep were immunised with κ or λ molecules that had been purified from urine samples containing Bence Jones proteins. The resultant antisera were adsorbed against purified IgG, IgA and monoclonal proteins and then affinity purified against mixtures of the respective FLCs that had been immobilised onto Sepharose. Antisera requiring further adsorption, as judged by the tests described below, were recycled through the adsorption and testing procedures until deemed satisfactory for assay use.

B. Antisera specificity

Specificity is the most important aspect of the immunoassays and was evaluated using several techniques.

1. Immunoelectrophoresis

The antibodies were purified until they showed no cross-reactions by immunoelectrophoresis with the alternate FLC and intact immunoglobulin molecules (Figure 4.1).

2. Western blot analysis

This sensitive technique was used to assess the reactivity of the antisera against immunoglobulin fragments and FLC polymers. The results showed that both κ and λ FLC antisera reacted strongly, with two closely migrating bands at 25-30kDa, and weakly with several larger and smaller molecular weight fragments. Similar staining patterns were observed using Mabs. The FLC antisera were readily able to detect monomers and dimers of both κ and λ molecules (Figure 4.2).

3. Haemagglutination assays

These assays are far more sensitive than immunoelectrophoresis and provide better assessment of specificity. Sheep red blood cells (RBCs) were sensitised with individual FLCs, and purified IgG, IgA and IgM and tested against the FLC antisera. The results showed that κ and λ FLC antibodies reacted with the appropriately labelled cells at >1:16,000 dilution and at <1:2 against cells coated with the alternate FLCs or intact immunoglobulins (Figure 4.3).

4. Nephelometry

Latex-conjugated FLC antisera were tested for specificity by nephelometry. Potentially interfering substances were added to serum containing known concentrations of FLCs and the changes in values indicated the effect on the assays (Figure 4.4). Overall, the specificity assessments showed that FLC antisera had minimal reactivity with light chains on intact immunoglobulins and other potentially interfering substances (0.2-0.01%). These values are within the purity specification for FLC contamination in the tested interfering materials.

There have been no published independent specificity analyses of the nephelometric Freelite latex reagents. Nakano et al. ^[18] reported an evaluation but, in error, only tested FLC antisera that were manufactured for IFE, where specificity requirements are less demanding ^[20].

C. Accuracy and standardisation

Assay accuracy is defined as the degree of closeness of achieved results relative to their absolute values. Unfortunately, international standards do not exist for FLC measurements, so there are no reference points from which to assess the accuracy of results. Furthermore, each monoclonal light chain is unique, with its own special set of surface epitopes, so accurate measurements are difficult to obtain. Nevertheless, the light chain constant region domains have little structural variability, so they are good antibody targets.

In order to ensure accuracy in FLC immunoassays, a suitable basis for standardisation and calibration was required. It was considered that polyclonal FLCs should be used in order to minimise any potential problems that might arise from the use of unique monoclonal proteins ^[21].

Figure 4.5 Flow chart for the production of standards for sFLC immunoassays. (SLE: systemic lupus erythematosus, y: years)

This was achieved in the following manner (Figure 4.5):

1. Production and accurate quantification of pure polyclonal, “primary” FLC standards.

2. Production of secondary and “working” reference materials, and comparison of these with the primary standards.

3. Production of calibration materials for use in the FLC assays that were referenced against the “working” standards.

4. Analysis of a variety of normal and abnormal samples using a reference nephelometric method.

5. Comparison of results from other instruments with the reference method.

The various materials were manufactured and purified in accordance with established procedures ^[22]. Each primary FLC preparation was found to be greater than 99% pure by silver-stained SDS-PAGE while the alternate FLC was not detected by haemagglutination inhibition and dot blot assays. The amino acid content of each primary standard was then determined in order to produce an accurate estimation of the protein content.

Secondary reference materials were prepared from pools of different monoclonal κ and λ proteins. These were not considered ideal for use as working calibrators, so additional reference materials were prepared from SLE sera that contained elevated polyclonal FLCs (Figure 4.5).

The pure FLC preparations were used to assign κ and λ concentrations to the secondary reference preparations. Subsequently, κ and λ values were assigned to the serum pool by nephelometry. Each stage of the value transfer was completed at three dilutions and repeated three times. All protein preparations were stabilised and stored at -80°C until required. The final FLC values for the “working” reference preparations were 46.0mg/L for κ and 71.4mg/L for λ. These calibration values were used for all subsequent laboratory and clinical studies.

An alternative reference material based upon polyclonal FLCs extracted from urine has been proposed by Nakano et al. ^[18]. However, sFLCs may differ chemically from urine FLCs and concentration measurements used in that study were based upon inappropriate dye uptake comparisons with albumin rather than amino acid analysis. As yet, these two preparations have not been compared. Clearly, there is a need for international agreement on standardisation of κ and λ FLCs, as well as the availability of a suitable international reference material.

D. Assay sensitivity

Assay sensitivity (the lowest limit of antigen detection) is somewhat dependent upon the types of samples measured. FLC assays can detect less than 0.1mg/L of κ and λ FLCs in undiluted, clear urine samples and CSF (Chapter 22). Serum sensitivity is typically 5-fold worse because of interference from lipids and other light-scattering particles (Table 4.2 and Chapter 6). Concentrations at 1-2mg/L are below the normal serum range so the FLC immunoassays allow extreme κ/λ concentration ratios to be established with considerable accuracy. This is important for monitoring a patient's treatment because FLC concentrations may be below the normal range.

Table 4.2. Representative routine sensitivity levels of frequently used FLC assays.

E. Assay ranges

Figure 4.6 Calibration curve for κ FLC using the Roche Modular P.

Figure 4.7 Calibration curve for λ FLC using the Roche Modular P.

The measuring ranges of the FLC assays are dependent upon two factors: the slope of the respective calibration curve and the portion selected for the assay (Figures 4.6 and 4.7). The latter should be chosen to allow the maximum number of normal and abnormal clinical samples to be measured at the initial sample dilution. A typical analytical range for κ is 3-150 mg/L and for λ, 5-200 mg/L (Chapter 27). Samples containing higher concentrations require further dilution (see Antigen excess, below).

Patients with bone marrow suppression may have low concentrations of the alternate FLC, where the precision of the calibration curves is relatively poor. Consequently, some consideration should be given to the accuracy of κ/λ ratios that include low FLC concentrations.

F. Antigen excess

Figure 4.8 Diagram illustrating the mechanism of antigen excess. The light scattering signal falls in antigen excess because of smaller immune complexes.

Figure 4.9 Demonstration of antigen excess capacity for FLC assays using serum containing monoclonal FLCs. Addition of high concentrations of the patients’ sera caused the assays results (in light scattering units) to plateau.

Antigen excess causes immunoassays to underestimate very high concentrations of protein (Figure 4.8). In the case of FLC measurements, sample concentrations can range from <1 mg/L to >100,000 mg/L. This is a greater range than almost any other serum protein test. Consequently, a small proportion of samples containing very high levels of FLCs may be underestimated because of antigen excess. In addition, the amino acid composition of the FLC produced by an individual B cell clone will influence the level at which a sample may show antigen excess. Therefore, a minority of monoclonal FLCs may exhibit antigen excess at fairly low FLC concentrations (see Chapter 15).

Antigen excess is an important issue. During the development of the Freelite™ sFLC assays on the Siemens BNII analyser, results indicated linear quantification at up to 200 mg/L with antigen excess at higher levels (Figure 4.9). Subsequently, monoclonal FLCs from 304 patients were assessed. At a 1:20 serum dilution, 6 high samples were incorrectly identified as moderately elevated and a further two were misclassified as normal. When the samples were tested at a dilution of 1:100, all 8 samples were correctly identified as highly elevated. Later assays were modified to use a starting serum dilution of 1:100 ^[23].

Freelite™ sFLC assays available on some instruments now include automatic antigen excess detection parameters. For example, the Binding Site SPAPLUS instrument monitors the initial reaction kinetics of each sample and compares the results with reaction limits set by the manufacturer through testing of an extensive myeloma library. Samples detected as being in antigen excess are automatically flagged by the instrument and retested at a higher sample dilution. A very small proportion of samples in antigen excess have normal reaction kinetics so will not prompt the flag. Therefore, it is recommended that the following statement accompanies all FLC results: "Undetected antigen excess is a rare event but cannot be excluded. If these FLC results do not agree with other clinical or laboratory findings, or if the sample is from a patient that has previously demonstrated antigen excess, the result must be checked by retesting at a higher sample dilution."

Freelite™ sFLC assays available on other instruments, such as the Siemens BNII, do not include automatic antigen excess checks. Samples should be tested for antigen excess (using the dilution protocol in the product insert) when the sample exhibits an abnormal FLC concentration or κ/λ ratio, if the patient has previously demonstrated antigen excess or when FLC results do not agree with other clinical or laboratory findings.

Several large studies have evaluated the incidence of antigen excess in large numbers of consecutive patients. Murata et al. studied 7,538 serum samples over a 4-month period using 1:100 and 1:400 sample dilutions on a Siemens BNII ^[24]. There were 9 samples with κ antigen excess but no samples with λ antigen excess giving an incidence of 1/840 (0.12%). Of importance, all the κ and λ samples had elevated FLC concentrations or abnormal κ/λ ratios at the initial dilution of 1:100. Bosmann et al. ^[25] studied the incidence of antigen excess in 91 patients. Samples from 2 patients (2.2%) exhibited antigen excess: one, a κ FLC patient with a known IgAκ monoclonal gammopathy and the other, a patient with λ FLC-monoclonal gammopathy of undetermined significance (Chapter 19). The authors conclude that the interpretation of FLC measurements is facilitated in many cases, when considering the data in the context of electrophoresis results and clinical information. Vercammen et al. ^[26] studied 865 patients using 1:100 and 1:2000 sample dilutions on a Siemens BNII. Antigen excess was defined as a greater than 4-fold difference between the results obtained at the two dilutions. A total of 5.4% (44/811) and 1.2% (9/773) of κ and λ samples exhibited antigen excess respectively. It is unclear why the incidence of antigen excess reported in this study is much higher than that reported by Murata et al. and Bosmann et al. The authors highlight the importance of selecting the correct dilution for reporting results, and state that if the result at the 1:2000 dilution is greater than 4 times the result obtained at the 1:100 dilution, then the 1:2000 result should be reported; if the result at the 1:2000 dilution is less than 4 times the result obtained at the 1:100 dilution, then the 1:100 result should be reported. This approach improves the consistency of reporting FLC values.

G. Precision

Figure 4.10 κ/λ log plot of approximately 4,000 abnormal sFLC samples showing “gaps” at the lower and upper ends of the instrument’s assay range. Samples in the normal range were excluded.

Figure 4.11 As for Figure 4.10 but showing κ sFLC results alone. The “steps” in the plot occur at the ends of the instrument’s assay range, where there is a change in sample dilution. Similar results occured for λ.

The precision of the sFLC assays is good, with percentage coefficients of variation typically less than 8%. However, samples measured at low concentrations, the use of poorly maintained instruments, and operation by inexperienced staff will be less precise. Results from the manufacturer are produced by expert staff under ideal conditions so that minor variations between production batches can be identified and corrected. Similar results are achievable in busy, routine laboratories. If they are less good, it is appropriate to seek assistance so that the causes can be identified. Results of precision studies for individual instruments are shown in Chapter 27.

Some years after the introduction of the FLC assays it became apparent that samples with results just outside the lower and upper ends of the instrument ranges were slightly lower and higher respectively, when repeated at the new dilutions. This is due to an uncharacterised substance(s) that interferes with the latex assays. Since the concentration of the interfering substance is different at the second dilution, the results are altered. This gives a “gap” in results of 2-3mg/L at the lower end of the curve (at 4-7 mg/L) and 8-9 mg/L at the upper end of the curve (at 62-70 mg/L) on, for example, the Roche Modular P (Figures 4.10 and 4.11).

In order to reduce the frequency of samples falling at either end of the curves, the assay ranges have been extended downwards and upwards so that fewer normal samples require re-dilution. This has improved the accuracy of results at this borderline region of the analytical range.

H. Linearity

Assays are considered to be linear when the results obtained from measurement of a sample at a series of dilutions are equivalent to the results expected from the original measurement. Non-linearity is due to a number of factors including: poor antibody specificity; inability of the antibody to recognise different forms of the antigen; nonspecific assay interference from lipids or fibrin; unsuitable materials for the calibrators or standards etc. Assessment of linearity forms an important part of immunoassay development.

FLC assays are potentially prone to non-linearity. Light chains are monoclonal proteins that exist as a number of sub-groups, in polymeric forms and contain unique combinations of hypervariable regions. Inevitably, these features cause non-linearity in some samples. Examples of good linearity are shown in Figure 4.12.

I. Polymerisation

Figure 4.12 Linearity of (A) κ and (B) λ assays on a Hitachi 911.

Figure 4.13 Stability of FLCs over different times and at different temperatures. (κ [square]; λ [triangle]. Standard deviations of results are mostly within the symbols).

FLC molecules are usually monomers or dimers, but higher polymeric forms frequently occur ^{[27][28][29][30][31][32][33]}. These act as multi-antigenic targets in immunoprecipitation assays, accelerating the formation of aggregates and leading to over-estimation of antigen concentrations. This occurs in patients with nonsecretory multiple myeloma (NSMM) who have undetectable concentrations of sFLCs by IFE but high concentrations by nephelometry (Chapter 9). Sölling ^[30][31] indicated with his assays that monomers and dimers were detected equally using antibodies against whole light chains, but antibodies directed only against FLC epitopes may preferentially detect dimers ^[32].

In order to determine the effect of polymerisation on FLC quantification, FLC monomers, dimers and polymers were purified from myeloma sera and concentrations compared with total protein measurements. It was apparent that purified dimers were over-estimated by 1.5-fold and higher polymers by 1.5 to 3.5-fold ^[34]. However, in a study by Émond et al.,^[33] a greater than 7-fold over-estimation was observed in two samples (when compared with CZE) in association with polymers of up to 200kDa.

An additional factor is that SPE tests can underestimate FLC concentrations. Variable polymerisation may cause smearing of monoclonal bands on the gels so that only a proportion of the monoclonal protein is measured (Figure 9.2). FLCs also take up less protein stain than albumin so their concentration is underestimated ^[2]. It is likely that a combination of these factors causes the ten-fold or more overestimations that are seen in the sera of some patients.

Whatever the explanation for the unexpectedly high results, it will be difficult to develop FLC assays that measure all molecular forms equally. Perfect quantification will be elusive.

J. Stability

FLC molecules are very stable in serum and urine (Figure 4.13). Tencer et al.^[35][36] showed that there was little variation in the concentrations of FLCs in samples stored at -20°C over a two-year period. We have found similar concentrations of FLCs in both fresh and old normal sera stored for 20 years at -20°C. Clearly, these results indicate that FLCs do not dissociate from intact immunoglobulins, or fragment, over prolonged periods.

Stability of the FLC antisera is also important. “Open-vial stability” refers to the shelf-life of the antisera after their first use. This may be short because pipetting procedures can introduce FLC molecules into the vials that react with the remaining antisera and reduce its activity. Care should be taken not to contaminate antisera vials with sera from previously pipetted samples.

K. Serum and plasma comparisons

Figure 4.14 Comparison of serum and acid citrate dextrose plasma samples from 50 blood donors and 20 MM patients using a Beckman Immage.

Figure 4.15 Comparison of serum samples on different platforms for κ (left) and λ (right).

Some laboratories prefer to analyse blood proteins in plasma rather than sera. In one study performed to directly compare the concentrations of FLCs in plasma and serum, 50 paired serum and plasma samples from blood donors and 20 paired samples from patients with multiple myeloma (MM) were studied on a Beckman Immage, a Siemens Dade Behring BNII and a Hitachi 911. Plasma samples were collected in acid citrate dextrose (ACD) or heparin and the sera were prepared by adding fibrin. Table 4.3, Figure 4.14 and Figure 4.15 show comparison data obtained on the different instruments ^[37].

Table 4.3. Comparison of serum and acid citrate dextrose plasma FLC concentrations from 50 blood donors and 20 MM patients.

4.3. Comparison of free light chain immunoassays on different instruments

sFLC kits are made for many laboratory instruments. A complete list of those currently available is given in Chapter 27. Comparison of FLC results using a Hitachi 911 and a Siemens Dade Behring BNII are shown in Figure 4.15 and the characteristics of four different instruments are shown in Table 4.4. Comparisons of normal ranges are discussed in Chapter 5. There are no significant differences between the results on different instruments ^[38].

Table 4.4. Summary of the characteristics of FLC assays on different instruments. (C.V. = coefficient of variation in precision studies. Numbers in brackets refer to sample dilutions). See Chapter 27 for more details.

4.4. Comparison of different serum free light chain immunoassays

Figure 4.16 Batch-to-batch analysis for the Freelite κ and λ assays. A. Four lots of the Freelite κ kit (1-4); and B. four lots of the Freelite λ kit (1-4) were compared using an in-house quality control panel (32 samples; κ range: 0.4-3500 mg/L; λ range: 0.4-2800 mg/L). In each analysis, values obtained from a predicate batch (1) were compared with values from three separate kits (2-4). Measurements were repeated in triplicate and mean values are presented. Continuous lines indicate the lines of best fit, and the R² values are presented.

Figure 4.17 Analysis of outcome of customer enquiries.

Figure 4.18 Numbers of units (100 tests) sold per customer enquiry.

Figure 4.19 AKTA biopilot system for controlling column

The Binding Site serum FLC assay (Freelite™) is the only assay system validated for routine serum measurements. New Scientific Company (NSC), from Italy, sells a urine assay for FLCs that it claims can be used for serum measurements. However, since the antisera are not latex-conjugated, there is insufficient sensitivity to measure normal sFLC concentrations. Furthermore, in this system the antisera cross-react with intact immunoglobulins, so that normal samples produce higher results than the Freelite assays. Consequently, sera containing elevated monoclonal FLCs may be classified as normal. In a study of 24 patients with monoclonal gammopathies, 15 were abnormal by Freelite but only 5 using NSC reagents. Of the 7 patients with AL amyloidosis in the cohort, all were positive by the Freelite assay but only 2 using NSC reagents. Furthermore, fluctuations of FLC concentrations were observed with Freelite reagents during monitoring that were undetectable by NSC reagents ^[39][40]. Thus, by scientific and clinical criteria, such unvalidated reagents should not be used for sFLC measurements.

4.5. Quality control of free light chain antisera and kits

Maintaining batch-to-batch consistency is essential as the assays may be used for monitoring individual patients over many years. Effective quality control is ensured using a variety of techniques, two of which are described briefly here. External quality assurance schemes are described in Chapter 28.

Specificity is controlled by comparing a set of test results from each new batch of antisera with results from previous batches. Typically, the panel of samples includes normal sera, sera with elevated polyclonal FLCs and myeloma sera. The results are compared using Passing-Bablok analysis and are considered acceptable when they fall within a defined set of criteria (Figure 4.16).

Once suitable antisera have been selected and attached to latex particles, the other kit components are selected and assembled. Similar materials are used for calibrators and controls and comprise sera containing high concentrations of polyclonal FLCs. The first step is to assign a value to the kit calibrators using the working reference material. This is achieved using 100 separate assays and 10 separate calibration curves. The second step is to assign values to the control reagents using similar procedures.

Analytical comparisons are also made using normal sera. A typical evaluation of a panel of 90 normal samples on the Binding Site SPAPLUS produced the following results: mean κ = 8.86 mg/L (range 4.32 - 20.6 mg/L), mean λ = 11.85 mg/L (range 3.77 - 28.77 mg/L).

The long-term effect of technical improvements, together with more experience of kit production, customer feedback (Figure 4.17) and larger batch production, has improved quality. This is apparent from the considerable reduction in customer enquiry rates per unit sold since the kits were first introduced 10 years ago (Figure 4.18) ^[41]. Upward trends typically occur as larger reagent batches are manufactured. These are advantageous because there is more material that can be used for assignment of reference values, stability checks, calibrator checks, specificity and sensitivity analysis, etc. Downward trends are associated with the introduction of new kits such as for the SPAPLUS instrument, in Autumn 2007. Newly introduced assays always have teething problems so it is of note that issues with the early kits have been largely resolved.

4.6. Scale-up of antiserum production

Scale-up involves producing larger batches, both by pooling antisera from more sheep and by using larger and more automated laboratory equipment. The polyclonal reagents have to be adsorbed to complete specificity and then affinity purified by passage through large chromatography columns. This is controlled using an AKTA biopilot system that can automatically load, wash, elute and collect antibody peaks (Figure 4.19). Batch sizes have increased a thousand fold, from a few hundred millilitres to hundreds of litres. In the life-cycle of diagnostic tests, however, the FLC immunoassays are still in their youth. Improvements will be seen in sensitivity, specificity, reactivity, precision, utility, costs, etc.

References