Effect of long-term frozen storage on stability of kappa free light chain index

Introduction

Kappa free light chains (κ-FLC) in the cerebrospinal fluid (CSF) serve as a biomarker in neurology indicating the presence of an intrathecal immunoglobulin synthesis [1]. κ-FLC are produced by plasma cells in excess over heavy chains and, thus, are secreted along with intact immunoglobulins. Similar to immunoglobulins κ-FLC accumulate in the CSF in several inflammatory neurological diseases [1].

An increased κ-FLC index has been shown, e.g., in multiple sclerosis (MS), Lyme disease, HIV infection and autoimmune encephalitis [2], [3], [4], [5]. The κ-FLC index has been proposed as diagnostic marker for MS [6]. In contrast to oligoclonal bands (OCB), the previous gold standard to demonstrate an intrathecal immunoglobulin synthesis, determination of the κ-FLC index offers significant methodological advantages while maintaining similar diagnostic sensitivity and specificity [6], 7]. Determination of κ-FLC is a fast, cost-effective, widely available and rater-independent method. Furthermore, there is evidence in patients with early MS that the κ-FLC index as a quantitative marker predicts future disease activity. Higher κ-FLC index determined at the time of disease onset was associated with various outcome measures including shorter time to relapse [8], new MRI activity [9], disability accrual [10] as well as with shorter time to cognitive decline [11].

With an increasing amount of κ-FLC research and its use in clinical routine, there is a need to clarify pre-analytical aspects such as sample storage. While κ-FLC concentrations in serum have been shown to minimally change after frozen storage over a period of approximately one year [12], the effects of frozen storage duration on κ-FLC concentrations in the CSF as well as on the κ-FLC index are not known. This is why we performed the present study.

Materials and methods

Patients at the Department of Neurology of the Medical University of Innsbruck, in whom κ-FLC index was determined in previous research (cohort I [13], and cohort II [8]), considered as time point 1 (T1) with enough remaining CSF and serum volume for a second measurement after frozen storage at −20 °C (T2), were eligible for inclusion into the present study.

Initially, all CSF samples were collected by lumbar puncture (LP), and serum samples concomitantly within 30 minutes by venous puncture. All samples were centrifuged at 2000 g for 10 min at room temperature before storage [14].

Results

A total of 70 patients with a median age of 31 years (25th-75th percentile: 26–39) comprising 44 (63 %) females were included in the study.

Before frozen storage, i.e. at T1, median CSF κ-FLC concentration was 0.25 (25th-75th percentile: 0.12–0.43) mg/dL, serum κ-FLC concentration was 1.05 (0.86–1.34) mg/dL, CSF albumin concentration was 17.7 (14.6–26.1) mg/dL and serum albumin concentration was 4230 (3898–4488) mg/dL. The calculated Q_κ-FLC was 0.22 (0.11–0.45) and Q_alb was 4.2×10⁻³ (3.4×10⁻³−6.0×10⁻³). The κ-FLC index was 44 (25–108).

At T2, i.e. after frozen storage duration of 0.5 years (cohort I) as well as of 4.5 years (cohort II), CSF and serum κ-FLC concentrations were significantly decreased (Figure 1A and B, Supplementary Table 1). As the decrease was more pronounced in CSF than in serum, also the Q_κ-FLC showed a slight decrease at T2 (Figure 1C, Supplementary Table 1). Similarly, CSF and serum albumin concentrations decreased with longer frozen storage duration (Figure 1D and E, Supplementary Figure S1A-B). Again, as the decrease of albumin in CSF was more pronounced than in serum, a decrease of Q_alb with longer frozen storage duration was observed (Figure 1F, Supplementary Figure S1C).

Figure 1:

Impact of frozen storage duration on κ-FLC and albumin in CSF and serum. Change in (A) CSF κ-FLC concentration, (B) serum κ-FLC concentration, (C) κ-FLC quotient, (D) CSF albumin concentration, (E) serum albumin concentration and (F) albumin quotient between the time points T1 and T2 are shown. CSF albumin, serum albumin and albumin quotient were adjusted to the fixed frozen storage duration of 0.5 and 4.5 years. CSF, cerebrospinal fluid; FLC, free light chain; p, p-value.

Altogether, the above-mentioned changes (i.e., decrease of Q_κ-FLC as well as of Q_alb) evened each other out when the κ-FLC index was calculated (i.e. Q_κ-FLC/Q_alb). The κ-FLC index showed similar values irrespective of frozen storage duration (Figure 2).

Figure 2:

Impact of frozen storage duration on the κ-FLC index. Albumin quotient used for calculating the κ-FLC index (and its percentage change) was adjusted to the fixed frozen storage duration of 0.5 and 4.5 years. FLC, free light chain; p, p-value.

Discussion

In the present study, we observed a decrease in the absolute concentrations of κ-FLC and albumin in CSF and serum with increasing frozen storage duration. However, the κ-FLC index remained stable.

Both CSF and serum κ-FLC concentrations were statistically significantly lower at T2 compared to T1, i.e. after a frozen storage duration of up to 4.5 years, with a more pronounced relative decrease in CSF than in serum leading also to a lower Q_κ-FLC at T2. Similarly, albumin concentrations in both CSF and serum decreased with increased frozen storage duration, also with a larger relative reduction in CSF than serum resulting also in a lower Q_alb at T2.

The observed decrease in serum κ-FLC concentrations aligns with findings from a previous study that reported a statistically significant reduction in serum κ-FLC concentrations after frozen storage of approximately one year (ranging between 193 and 568 days) at −20 °C [12]. With regard to long-term stability of κ-FLC in CSF, we provide evidence for the first time.

In contrast to a previous study which reported an increase of serum albumin at a rate of 0.5 % per year over a 25-year frozen storage duration, we observed a decrease at a rate of 1 % per year. This discrepancy might be explained by different detection methods. While we measured albumin by immunonephelometry, the authors of the above-mentioned study measured colorometrically by bromocresol green method. One might speculate that the reported increase is due to an unfolding of the protein due to freezing allowing more bromocresol green to be bound at the time of second measurement. This might explain a false-positive increase [18]. Another study reported that serum albumin in samples stored over different time periods remained stable. However, samples were drawn from different groups of individuals (matched by sex and age) at the different time points back in time, limiting its generalizability [19].

Why the relative decrease of κ-FLC and albumin concentrations is higher in the CSF than in serum is unknown. However, it explains why Q_κ-FLC and Q_alb decrease as well. The decrease in the CSF/serum quotients is already by far not that pronounced compared to the decrease of the absolute concentrations. Further, due to very similar decrease rate of Q_κ-FLC and Q_alb over time, their changes are evened out when the κ-FLC index is calculated. Accordingly, we did not observe a statistically significant change of the κ-FLC index due to frozen storage duration (Figure 2, Supplementary Figure S2).

There are several limitations of the present analysis. First, this was a retrospective study with all its inherent limitations due to this study design. We included samples that had been already used for κ-FLC measurements in previous research [8], 13] based on the availability of spare sample volume for another determination of κ-FLC index (after frozen storage). Therefore, an inclusion bias cannot be excluded, and we need to state that prospectively run studies are necessary to replicate our findings. Secondly, while κ-FLC concentrations were measured using the same method, albumin concentrations were determined using different platforms (even though both were nephelometry). However, an impact of different platforms would result in a different intercept in the regression analysis but would not explain a change in the slope. While a decrease in measurement is due to frozen storage duration (slope), an overall impact of the different platforms would result in time-independent level effects on the measurements (intercept). As the research question of the study addressed the slope, i.e. the impact of frozen storage duration on measurement, we are confident that the use of different platforms for albumin measurement had no impact on our findings. Thirdly, with regard to the observed decrease in absolute concentrations of κ-FLC and albumin over time, on might also speculate that this is due to changes in instrument performance over time (and not due to frozen storage). We provide evidence that instrument performance for both κ-FLC and albumin were stable over time (Supplementary Figure S3). Fourthly, frozen storage duration between each the two albumin measurements and the two κ-FLC measurements were different. While κ-FLC measurement was done for study purposes and thus at fixed time points (resulting in frozen storage duration of 0.5 and 4.5 years) [8], 13], the initial albumin measurements were done as part of clinical routine resulting in longer frozen storage duration (ranging from approximately 4 to 16 years). It cannot be excluded that albumin remained stable within the first years of frozen storage and decreased only thereafter. However, this scenario is very unlikely as changes have been reported already to occur within two years [19]. Finally, we would like to state that there were no additional freeze-thaw-cycles in our samples, and the temperature (−20 °C) of the freezer was stable over time (the temperature was monitored).

Overall, we provide evidence that κ-FLC index remains stable even after samples have been frozen for several years.