Normal ranges and reference intervals

From Wikilite

5.1. Serum free light chain normal ranges

Normal range data for serum free light chains (sFLCs) have been published many times: those considered to be the most reliable are shown in Table 5.1. Serum results using polyclonal antibodies have varied considerably, indicating a degree of cross-reactivity with bound light chains in some of the assays. Perhaps unsurprisingly, assays using monoclonal antibodies have also produced varied results, the discrepancy between reports being more than tenfold ^[1][2][3][4]. It is unclear whether the variations were due to specificity, calibration or matrix differences, or a combination of these factors.

Table 5.1. Selected publications reporting normal sFLC concentrations (mg/L) and κ/λ ratios ^[11][3]. (^apolyclonal antibody against total light chains, ^cpolyclonal antibody against FLCs, ^dmonoclonal antibody against FLCs, ^elatex reagent, ^f 95% range).

Figure 5.1. (a) κ, (b)λ and (c) κ/λ ratios versus age (years) in 282 normal serum samples together with (d) cystatin C results in the same patients. sFLC/cystatin C ratios (e) and (f) show no change with age, confirming renal deterioration as the cause of increased sFLC levels in elderly individuals. (red = fresh sera, black = frozen sera). (Courtesy of J. A. Katzmann) .

The most detailed study of FLC concentrations in normal individuals was published by Katzmann et al. ^[12], and showed similar results to those previously reported by Bradwell et al. ^[11] The same assay procedures were used in both studies but a wider age range of individuals was studied by Katzmann et al. Serum samples were obtained from 127 healthy blood donors (21-62 years) and 155 older, normal individuals (51-90 years). A summary of the median values and reference ranges in the Katzmann study is shown in Tables 5.1 and 5.2.

The results of κ and λ measurements for all individuals in the Katzmann study are shown in Figure 5.1. There were no significant differences between results obtained using either fresh or frozen sera but there was a trend towards higher concentrations in elderly people (Figure 5.1 and Table 5.3). This occurred for both FLCs and also for the renal function marker, cystatin C, which was measured in the same samples. Calculations of cystatin C/sFLC ratios and κ/λ ratios on each sample normalised the elevated values (Figure 5.1, c,e,f). Therefore, the higher sFLC values seen in older people can be explained by small reductions in glomerular filtration rate ^[13].

An alternative method of presenting normal range data is shown in Figure 5.2. This shows the serum κ and λ results for each person plotted on a logarithmic scale. Such data presentation inherently includes the κ/λ ratios and is useful for visualising results from individual patients. It also allows easy comparison between different disease groups and is used extensively in later chapters for clinical comparisons.

One discrepancy between the results of Bradwell et al. and Katzmann et al. and those of many earlier researchers is that serum κ concentrations were found to be lower than serum λ, whereas traditionally κ levels were higher than λ (Table 5.1). Since there are nearly twice as many κ- as λ-producing lymphoid cells, this observation had seemed reasonable. Certainly, the total serum κ and λ levels reflect the normal ratio of κ to λ FLC synthesis (Table 5.2).

A plausible explanation for the inverted serum κ/λ ratio relates to the kinetics of FLC clearance. Since κ molecules are normally monomeric (25kDa), their renal clearance is faster than dimeric λ molecules (50kDa); consequently they accumulate less in serum. Calculation of the κ/λ clearance rate from serum and urine FLC concentrations ([κ urine]/[λ urine] ÷ [κ serum]/[λ serum]) produced a result of 3.0 ^[11]. In a study on the movement of dextran polymers across capillary membranes, it was shown that molecules of 20kDa were cleared 3.2 times faster than 37kDa molecules ^[14]. Admittedly, polysaccharides are different in charge, shape and flexibility from globular proteins of similar molecular weight, but differential glomerular filtration probably accounts for the inverse κ/λ ratios. This may not have been observed in some of the earlier FLC studies due to poor antisera specificity. Recent studies have shown more consistency in serum κ/λ ratios although Nakano et al. ^[3][4] remain uncertain about actual FLC concentrations (Table 5.1).

Table 5.2. Median values and ranges for free and total light chain concentrations and κ/λ ratios in the sera of 282 normal individuals.

The κ/λ ratio is the most important parameter when distinguishing monoclonal from polyclonal increases in sFLCs. Polyclonal increases result from increased synthesis or decreased renal clearance of normal FLCs (Chapter 20-21). These two processes increase both κ and λ concentrations equally, thereby maintaining a fairly constant κ/λ ratio. In patients with renal failure, the half-life of both FLCs is prolonged from a few hours to several days, so concentrations increase 20-fold or more. However, κ/λ ratios remain within fairly narrow limits. In contrast, in monoclonal gammopathies, only one of the FLC concentrations increases. Thus, κ/λ ratios distinguish monoclonal from polyclonal diseases.

5.2. Normal ranges in children

Figure 5.2. κ and λ FLC concentrations in 282 normal sera. A: axis at the normal κ/λ ratio of 0.6. B: axis at a κ/λ ratio of 1.00.

Figure 5.3.Changes in sFLC during childhood. Modified from Sölling ^[15] to allow comparison with adult concentrations. (Reproduced with permission from Scand J Clin Invest)

Table 5.3. Median values for sFLCs and κ/λ ratios in different age groups.

Only one study of sFLCs in children has been reported. This was conducted by Sölling, ^[15] in 1977 (Figure 5.3), who noted that sFLC concentrations fall after birth and then increase with B cell development at six months. It is likely, therefore, that diseases such as agammaglobulaemia could be detected in the newborn from low sFLCs.

5.3. Variations in normal/reference ranges - hospital ranges

While decreasing renal function has little influence on κ/λ ratios, there is, nevertheless, a small but observable effect. As renal clearance falls the proportion of FLCs removed by other tissues increases (Chapter 3.4). Since this is by pinocytosis of serum, there is no distinction between κ and λ removal rates. Relative to λ, κ FLC molecules are therefore removed more slowly than by the kidneys and concentrations rise. Hence, κ/λ ratios increase with deteriorating renal function and are highest in complete renal failure (Chapter 20). The data in Table 5.3 show the higher κ/λ ratios in normal elderly individuals, who typically have reduced glomerular filtration. This has practical importance for identifying multiple myeloma (MM) patients in screening studies. Thus, by using the Katzmann FLC ratio range of 0.26–1.65, assay specificity for patients with MM is 93%. When the renal reference range of 0.37–3.1 is used, the assay specificity increases to 99% ^{[16][17][18][19][20]}.

When screening symptomatic patients in a hospital setting, many ill patients have inflammatory conditions (causing increased polyclonal immunoglobulins and FLCs) with associated renal impairment and they show small increases in κ/λ ratios. An example of such borderline results, seen in the context of a screening study, is shown in Figure 5.4 ^[21].

Figure 5.4. Frequency of borderline κ/λ ratios (37 in total) identified in a routine screening study for monoclonal gammopathies in 925 hospitalised patients. PIgs: polyclonal immunoglobulins (Courtesy of P Hill)

Figure 5.5. Frequency distributions of FLC κ (A), λ (B) and κ/λ ratios (C) in 66 normal urines.

This is an important issue as some samples will be from patients who have no evidence of monoclonal gammopathies but have previously unrecognised renal impairment that should be investigated. Also, borderline elevated κ/λ ratios should be assessed in the context of an accurate glomerular filtration marker such as cystatin C when monitoring patients with changing renal function (Chapter 20).

5.4. Borderline results

All tests have borderline results but they need to be considered in their clinical context. The FLC normal range of Katzmann et al. ^[12] included all samples in a population of 282 individuals, so the extreme values are approximately 5 standard deviations from the median. Since this range is far greater than that used for most analytes (typically 2-3 standard deviations), borderline results for FLCs are more likely to be clinically significant. Variations in instrument performance, differences between instruments and variations in kit manufacture etc (Chapter 4) are correspondingly of less significance. In addition, FLC results are usually visualised on κ/λ ratio log plots. This is essential because of the huge range of clinical results; however, such a diagrammatic representation exaggerates differences between low concentration results compared with those at high concentrations. Consequently, undue emphasis may be placed on borderline analytical differences compared with huge differences in clinical results that are highly elevated.

From the clinical perspective, borderline elevated results are due to a variety of causes. Increases in FLC concentrations from renal impairment are well documented (Chapter 20) as are those from infections, autoimmune diseases and diabetes mellitus (Chapter 21). Added to these are a variety of monoclonal diseases encompassing all the plasma cell dyscrasias, many lymphomas and leukaemias (Chapter 18), AL amyloidosis (Chapter 15) and FLC monoclonal gammopathy of undetermined significance (MGUS) (Chapter 19). Borderline abnormal results arising from these diseases may decrease the specificity of the test for light chain multiple myeloma (LCMM), for example, but need to be considered in the general context of why the serum sample was being assessed. In screening studies, borderline elevated results may indicate a significant disease process. Secondary tests such as serum creatinine for assessing renal impairment are then necessary, while follow-up FLC tests will help determine whether a disease has progressed.

5.5. Utility of serum free light chain assays for disease diagnosis

Using the reference intervals and diagnostic ranges shown above, Katzmann et al. ^[12], assessed the utility of sFLC measurements for identifying monoclonal FLC gammopathies (Table 5.4). Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy were estimated for the sFLC κ/λ ratios on the basis of both the central 95% interval and a diagnostic range that included 100% of normal results. Accuracy was calculated as the proportion of individuals classified correctly. PPV and NPV were calculated assuming a 15% prevalence of monoclonal proteins in the samples submitted for monoclonal protein studies. The patients in these studies are described in Chapter 6 for immunofixation electrophoresis (IFE), and in the respective chapters for MM, AL amyloidosis, light chain deposition disease (LCDD) and polyclonal hypergammaglobulinaemia. In the small group of samples selected, sFLC measurements had a higher sensitivity than IFE for detecting low concentrations of monoclonal sFLCs (Chapter 6).

The normal range data reported by Katzmann et al. ^[12] is the most detailed yet published and has been generally adopted (See Chapter 27 for implementation).

Table 5.4. Comparison of reference intervals and diagnostic ranges for sFLCs and κ/λ ratios. (Sera from the 282 reference individuals and 25 polyclonal hypergammaglobulinaemia patients as well as 66 sera from AL amyloidosis, LCDD and MM patients were used to calculate utility. (CL: confidence limits).

5.6. Urine free light chain normal ranges

Figure 5.6. Comparison of FLC measurements in serum (from Figure 5.2) and early morning urine samples from healthy individuals.

In general, FLC values in urine (uFLC) have been similar in all studies even when they have included random, 24-hour or early morning samples. Furthermore, κ concentrations were higher than λ in all publications apart from one recent study (Table 5.5 ^[3]). This latter study also reported exceptionally high serum FLC concentrations (Table 5.1) suggesting there were problems with antibody specificity. It is logical that urine contains more κ than λ molecules because of its faster renal clearance.

Table 5.5. Selected publications reporting normal urine FLC concentrations in mg/L ^[11][3]. (^apolyclonal antibody against total light chains, ^burine FLCs in mg/24 hours, ^cpolyclonal antibody against FLCs, ^dmonoclonal antibody against FLCs, ^e latex reagent, ^f 95% range, ^gearly morning urine samples, ^hrandom urine).

The overall similarity of uFLC results suggests that reported concentrations are reasonably accurate. This is in part because normal urine contains trivial amounts of intact immunoglobulin molecules that can interfere with accurate FLC quantification. The higher uFLC levels shown in the study by Bradwell et al. ^[11] for both κ and λ were explained by the use of early morning urine samples, which are typically 2-3-fold more concentrated than 24-hour urine samples. The mean (±SD) free κ concentration was 5.4 ± 4.95mg/L (n = 66; range, 0.36-20.3mg/L), and the mean (±SD) free λ concentration was 3.17 ± 3.3mg/L (n = 66; range, 0.81-17.3mg/L) (Figures 5.5 and 5.6). The mean κ/λ ratio was 1:0.54 (95% confidence interval, 1:2.17-1:0.25). The mean normal uFLC excretion was 3.7 mg/g of creatinine for κ and 2.0mg/g of creatinine for λ. There was a positive but non-significant correlation of urine creatinine concentrations with κ (r = 0.22) and λ (r = 0.17) measurements.

As expected, the range of uFLC concentrations was much wider than for serum, and κ/λ ratios were more variable. Presumably, this reflects minor differences in renal handling, urine dilution and variations in mucosal secretion of FLCs. The wide range of normal uFLC concentrations is another argument in favour of serum measurements.

References