Serum free light chain specificity and sensitivity: a reality check.

In 2001, a commercial test consisting of 2 separate measurements to quantify and free light chains (FLCs) was reported in Clinical Chemistry (1 ). The antisera specificities in this assay were reported to be 10 000-fold higher for FLCs than for light chains bound to immunoglobulin heavy chains. FLCs were thought to be associated with imbalances in heavy and light chain production in monoclonal plasma cell populations, and their quantifiability in the presence of the bulk of serum immunoglobulin opened new opportunities for characterizing plasma cell proliferation disorders. The reference interval for polyclonal FLCs was documented, and the reference interval for the free light chain -to(FLC K/L) ratio was demonstrated to be a sensitive indicator for excess (e.g., clonal) FLC production (1, 2). The gold standard for detection of monoclonal proteins is immunofixation electrophoresis (IFE). Several retrospective studies, however, showed that serum FLC had substantially higher detection limits than serum and urine IFE for diagnosis of the monoclonal light chain diseases of primary light-chain amyloidosis (3, 4) and light chain deposition disease (2 ), as well as nonsecretory multiple myeloma (5 ). This increase in diagnostic detection limit for this subset of monoclonal gammopathies indicates that the serum FLC assay is a natural addition to serum and urine IFE for diagnostic testing in the monoclonal gammopathies. The serum FLC assay specificity for the monoclonal plasma cell proliferative disorders resides in the FLC K/L. Although several disorders lead to abnormal concentrations of and FLC, the FLC K/L ratio appears to remain within the reference range except in the monoclonal plasma cell proliferative disorders. Patients with polyclonal hypergammaglobulinemia or decreased renal clearance have increased concentrations of FLC, but their FLC K/L ratio is within the reference range. We published a reference interval for FLC serum concentrations (2 ) and subsequently confirmed that patients without plasma cell or B-lymphocyte proliferative disorders all had FLC K/L ratios within reference intervals (6 ). In addition to these results on sensitivity and specificity, the above studies showed that when urine IFE results are abnormal, serum FLC assay is positive, calling into question the diagnostic necessity of urine studies (7 ). The current diagnostic algorithm for cases of suspected multiple myeloma and related disorders, however, recommends both serum and urine protein electrophoresis (PEL) and IFE (8 ). In this issue of Clinical Chemistry, Hill et al. (9 ) report their experience with 923 diagnostic samples for which serum PEL and FLC were performed, 370 of which had matched urine samples. Because their study draws from a general hospital population, it addresses key real-world challenges to the performance of the serum FLC assay. The report of Hill et al. (9 ) contains several important lessons that resonate with our own clinical laboratory experience, which involved a large dysproteinemia clinical practice, and our publications on diagnostic detection limit, which were from clinical trials and/or studies with well-defined patient groups. The first lesson regards specificity. Hill et al. (9 ) used serum PEL and FLC assays as the diagnostic screen and performed serum IFE only in cases in which a monoclonal band or suspicion of a band was detected on PEL, hypogammaglobulinemia was present, or the FLC K/L ratio was abnormal. Seven additional monoclonal gammopathies were detected in the group of patients with normal PEL and abnormal FLC K/L, and these included 2 cases of light chain multiple myeloma. There were also 35 cases in which the PEL was normal and the FLC K/L was abnormal, but the reflexed IFE was normal. These 35 cases indicate a potential false-positive rate of 3.8% and a specificity in this general hospital population of 96%. Bakshi et al. (10 ) assessed 1003 diagnostic sera with capillary electrophoresis and FLC assays, and in this screening algorithm, they had a specificity of 99%. The different specificities of 96%, 99%, and 100% in the studies by Hill et al. (9 ), Bakshi et al. (10 ), and Katzmann et al. (6 ), respectively, serve as a reminder that for the FLC assay, as for all laboratory tests, specificity is dependent on the context in which the test is ordered. After years of experience with the IFE assay, we have learned that there exist indeterminate and fuzzy bands that are of unknown analytic and clinical significance. On subsequent testing, these indeterminate results sometimes are demonstrated to be transient. The serum FLC assay generates falsepositive and borderline abnormal results that must be interpreted cautiously. Like IFE assays, FLC assays must be interpreted in a broader context. The second lesson regards the need for urine samples as part of diagnostic testing. Among the 370 urine samples tested with IFE, Hill et al. (9 ) detected monoclonal light chains in 15. Twelve of these patients also had an abnormal serum PEL or FLC K/L ratio, but 3 patients with urine monoclonal proteins had normal serum PEL and FLC results, representing a detection limit of only 80% for serum PEL and FLC in this small group of patients with positive urine studies. These authors also report that 2 of these patients had negative urine IFE results on followup. The third patient had persistent urine abnormality without clinical disease, and presumably had idiopathic Bence Jones proteinuria. Although the serum FLC assay for this small group of 15 abnormal urine samples apparently did not miss any serious disease, the false-negative rate was 20%. Because of the difficulty in obtaining paired urine samples and the diagnostic detection limit of serum PEL, IFE, and FLC studies, many clinicians would like to conclude that urine is not needed for detecting monoclonal proteins. The study by Hill et al. (9 ) is a first test of this hypothesis, but more evidence is needed before diagnostic urine studies are eliminated. If we want to add serum FLC and eliminate urine Editorials