Which pre‐analytical confounder matters the most in the comparison of two cohorts? Tubes and storage fill volume put to the test

Abstract

Background The tubes used for collection and storage and their fill volumes have been identified as critical elements for standardized CSF collection as they have proven to significantly influence recovery of CSF biomarker levels and β‐amyloid1‐42 (Aβ1‐42) in particular. The aim of the current study was to compare different collection protocols in terms of CSF biomarker recovery to assess the potential impact on studies establishing cut‐off values. Method Two pre‐analytical protocols, used in two experienced CSF biomarker testing sites in Europe, were compared by exchanging the collection and storage tubes. In addition, two different fill volumes were tested for the largest storage tube. In total, six protocols were compared of which protocol 1 and 2 represent the routine practice in the respective sites (Table 1). Twelve consecutive patients were sampled according to all six protocols and the order of the protocols was sequentially assigned per patient. CSF biomarker analyses were done using the LUMIPULSE G platform and related assays. A linear mixed effects model was applied using R. Result Protocols 1 and 3, both of which used storage tube 1 with a 0.5 mL fill volume, resulted in significantly lower CSF Aβ1‐42 and CSF Aβ1‐42/Aβ1‐40 ratio compared to all other protocols (p<0.001). Compared to protocol 6, which gave the highest recovery for the Aβ markers, the relative difference in Aβ1‐42 was ‐13.9% (95% CI: ‐16.1% to ‐11.6%) for protocol 1 and ‐14.8% (95% CI: ‐17.1% to ‐12.6%) for protocol 3. For the Aβ1‐42/Aβ1‐40 ratio, this resulted in a difference of ‐6.7% (95%CI: ‐8.8% to ‐4.5%) and ‐6.3% (95%CI: ‐8.5% to ‐4.1%) for protocol 1 and 3, respectively. No significant differences were observed for total Tau and hyperphosphorylated Tau. Conclusion In the current study, and with the tubes studied in these protocols, CSF biomarker recovery did not depend on the tubes in which the samples were collected or stored. However, fill volume did have a significant impact on the Aβ1‐42 levels and the Aβ1‐42/Aβ1‐40 ratio, implying that the combination of tube plus fill volume is a factor that needs to be taken into account when comparing studies calculating cut‐off values.