The kappa index (K-Index), calculated by dividing the cerebrospinal fluid (CSF)/serum kappa free light chain (KFLC) ratio by the CSF/serum albumin ratio, is gaining increasing interest as a marker of intrathecal immunoglobulin synthesis. However, data on inter-laboratory agreement of these measures is lacking. The aim was to assess the concordance of CSF and serum KFLC measurements, and of K-index values, across different laboratories. KFLC and albumin of 15 paired CSF and serum samples were analyzed by eight participating laboratories. Four centers used Binding Site instruments and assays (B), three used Siemens instruments and assays (S), and one center used a Siemens instrument with a Binding Site assay (mixed). Absolute individual agreement was calculated using a two-way mixed effects intraclass correlation coefficient (ICC). Cohen's kappa coefficient (k) was used to measure agreement on positive (≥5.8) K-index values. There was an excellent agreement in CSF KFLC measurements across all laboratories (ICC (95% confidence interval): 0.93 (0.87-0.97)) and of serum KFLC across B and S laboratories (ICC: 0.91 (0.73-0.97)), while ICC decreased (to 0.81 (0.53-0.93)) when including the mixed laboratory in the analysis. Concordance for a positive K-Index was substantial across all laboratories (k = 0.77) and within S laboratories (k = 0.71), and very good (k = 0.89) within B laboratories, meaning that patients rarely get discordant results on K-index positivity notwithstanding the testing in different laboratories and the use of different platforms/assays.
Keywords: cerebrospinal fluid; free light chains; kappa index; multiple sclerosis.