Recommendations for cerebrospinal fluid collection for the analysis by ELISA of neurogranin trunc P75, α-synuclein, and total tau in combination with Aβ(1–42)/Aβ(1–40)

CSF was collected prospectively from patients undergoing lumbar puncture (LP) due to clinical suspicion of normal pressure hydrocephalus at the Memory Clinic, Skåne University Hospital, Sweden. In these patients, 40 mL of CSF was collected as part of the clinical routine investigation. Anonymized CSF samples from 19 individuals were investigated. The detailed procedure for CSF collection and storage is shown in Fig. 1a. In addition, Fig. 1 gives an overview of the factors that were evaluated in the present study (Fig. 1b) and information on the selected recipients (Fig. 1c).

For each subject, 20 different SOPs were applied. Several vials with 5 mL of CSF were collected during the LP from the same subject in two types of tubes (low protein binding (LoB) or polypropylene (PP)) with or without additives (Tween-20 (Tw20; Panreac AppliChem, MO, USA) or Triton-X100 (TrX100; VWR Chemicals, Stockholm, Sweden)). Each tube was graduated in order to obtain the exact collection volume. A possible interference by a gradient effect was overcome by randomization of the order in which the different samples were collected. In a subset of the collection tubes, 25 μl of 10% Tw20 or TrX100 was added (final concentration 0.05% v/v). All vials were mixed by inverting them 25 times. The vials were kept at room temperature (RT) for 20 min, followed by centrifugation (2000 g, RT, 10 min). From these original tubes, CSF was transferred into microtubes (0.5 ml/1.5 ml, LoB/PP). After 1 week of storage at −80 °C, one 1.5-mL PP or LoB tube containing 1.3 mL CSF was thawed, aliquoted, and stored at −80 °C before testing. Among the tested procedures, there was no SOP using a collection LoB tube and a PP storage tube. All samples were analyzed after one freeze/thaw cycle. For the cases included in the study, CSF collection started on 23 September 2014 and ended on 2 December 2014. Sample analysis was performed in the first quarter of 2016. The measured concentrations for each analyte and subject included in the study are given in Additional file 1: Table S1.

CSF proteins were quantified by a CE-marked enzyme-linked immunosorbent assay (ELISA) for total tau (Euroimmun, Lübeck, Germany) and research ELISAs for α-synuclein or neurogranin trunc P75 (produced in the facilities of ADx NeuroSciences, Gent, Belgium). All assays were designed with a combination of two well-characterized monoclonal antibodies (mAbs). Some details of the test procedures are described in Additional file 1: Table S2.

Before analysis, CSF samples were pipetted into pre-blocked 96-well polypropylene plates as described in detail previously [15]. The impact of the addition of detergents to calibrators and/or CSF on analyte concentrations, as measured in the current assay designs, was verified before analyzing the samples of the study.

CSF and run-validation samples (i.e., calibrators added to phosphate-buffered solutions) were analyzed in duplicate in one lab (ADx) by one operator with one lot number of each ELISA. CSF samples from each individual subject were tested on the same 96-well plate in order to limit inter-plate variability. Reported values (mean of two optical density (OD) values) were used to calculate concentrations using a four parameter logistic (4-PL) curve fit algorithm. Acceptance criteria for run validation included: (i) re-calculated values of each calibrator concentration not to exceed 15% of the nominal concentration; (ii) kit controls within pre-defined specifications; and (iii) the variability between replicates must be lower than 20%.

The analytical specificity of the new neurogranin assay was verified in sandwich immuno-assay formats with synthetic peptides covering either the full length of the protein (Sequence: UniProtKB—Q92686) or peptides truncated at the amino- or carboxy-terminus. Phosphorylation at amino acid Serine 36 was included as an additional confounding factor. Several approaches were used to determine the assay selectivity, including but not limited to the replacement in the assay design of either the capture mAb or the detector mAb by a non-neurogranin-specific mAb or by the addition of proteins known to be present in CSF.

Parallelism was verified according to CLSI guideline EP06-A. Two neat CSF samples were diluted up to 1/10 (in the working range of the assay) with sample diluent and tested in quadruplicates. Concentrations were calculated taking into account the dilution factor.

Precision was evaluated using native CSF samples.

The specificity of the assay for the detection of α-synuclein (as compared to β-, or γ-) was verified with recombinant proteins (Source: rPeptide, Fremont, California, USA). Parallelism was verified according to EP06-A. Four native CSF samples were serially diluted with sample diluent up to 1/20 and tested in duplicate. Concentrations were calculated taking into account the dilution factor. Precision was evaluated using native CSF samples.

The analytical performance characteristics for the total tau assay have been summarized previously [16].

The relative difference in adsorption rate between analytes (individual proteins, ratio of proteins) were visualized by plotting the mean relative differences for all subjects when a comparison was made between the selected reference protocol (Protocol 7) and the protocol that always resulted in the highest relative difference (Protocol 6) or between the reference protocol and a protocol including CSF collection in PP tubes (Protocol 1).

The SOP experiment provided repeated measures data. The different measurements recorded on CSF collected from the same subject form a multivariate response. Measurements made with samples of the same subject are correlated, while measurements between different subjects are considered independent (with CSF of different subjects tested in different assay runs). More details on the analysis are described in reference [15] and Additional file 1: Figure S1. For each analyte, the impact of the analyte concentration on the calculated relative difference was verified by regression analysis.

An overview of the performance characteristics of each immuno-assay is essential to qualify the value of the results obtained in this study on standardization of the process for collection and storage of CSF. Part of the assay performance characteristics for the neurogranin trunc P75 and α-synuclein assay are described in Additional file 1: Figure S2 and Figure S3, respectively.

The addition of detergent (Tw20, Trx100) in the assay during sample incubation affects the outcome of the assay (= concentration) differently as a function of the analyte, the detergent, or OD values for calibrators and samples (Fig. 2). For calibrators, no effect on OD values were shown for neurogranin trunc P75 and α-synuclein, while OD values were higher for tau calibrators added to a buffer in which Tw20 was added. For CSF, slightly higher OD values were noted after the addition of detergent into the neurogranin trunc P75 and α-synuclein assay, but not for the tau assay. For α-synuclein, we observed a possible difference in function of the selected detergent type.

In order to standardize the measurement of the analytes in the study, and since our study design included different CSF compositions (no detergent, Tw20, TrX100), we performed the simultaneous incubation of the sample (CSF, calibrator, kit controls) with biotinylated antibodies dissolved in a buffer containing 0.05% (final concentration) Tw20. We observed no (major) difference in function of the selected detergent.

Figure 3 shows standardized mean concentrations for each CSF protocol, ordered as a function of the protocol number. Procedure 7 was considered previously as a possible reference protocol for further analysis. This was based on results obtained for the ratio Aβ(1–42)/Aβ(1–40). Details of the pair-wise differences between SOPs and p values are described in Additional file 1: Table S3. When looking at differences among procedures without the addition of detergent to CSF at the time of collection (Fig. 3; SOPs 1–10, left-hand side, blue scales), procedure 8 resulted in the highest observed mean concentration, followed by procedure 7. Both SOPs 7 and 8 use LoB collection and storage tubes and have no additional freeze-thaw cycle. For all analytes, the lowest observed mean concentration was with SOP 6 that combines a PP collection tube, a high volume PP storage tube (= higher percentage of CSF versus recipient surface area), and an additional freeze/thaw cycle. All in all, the differences among SOPs are modest, with the largest effects seen for neurogranin. The difference in mean concentration between SOP 7 and SOP 6 (the maximal difference, lower limit, upper limit) was estimated to be a decrease of 16.9% (−20.2%, −13.5%) for neurogranin (p < 0.0001), 5.7% (−9.4%, −1.7%) for α-synuclein (p = 0.005), and 8.8% (−11.7%, −5.8%) for total tau (p < 0.001). In the presence of detergent at the time of collection (Fig. 3; SOPs 11–20, right-hand side, red color), there was only a slight decrease of 4.3% (−7.0%, −1.4%) for total tau (p = 0.004) between SOP 11 and SOP 17. All other differences were not significant.

In addition, it was calculated that the observed differences of protocol 6 as compared to the reference method (= protocol 7) was not dependent on the protein concentration in the CSF.

In the second analysis (Fig. 4; Additional file 1: Table S4), the effects of the different procedural factors and their relationship were estimated for each analyte. In the absence of detergent (SOPs 1–10), the same pattern of effects was revealed for neurogranin and total tau, albeit with somewhat larger effects for neurogranin. For all analytes, the collection of CSF in PP tubes resulted in lower concentrations compared to collection in LoB tubes, amounting to a decrease of 2.7% (−0.3, −5.1%; p = 0.030), 1.7% (0.1, −3.5%; p = 0.063), and 0.5% (1.5, −2.5%; p = 0.574) for neurogranin, α-synuclein, and total tau, respectively.

The transfer to a PP storage tube had a larger effect and decreased the concentration on average by 5.3% (−7.8%, −2.8%; p = 0.0004), 2.4% (−4.7%, 0.0%; p = 0.052), and 2.5% (−4.8%, −0.1%; p = 0.043) for neurogranin, α-synuclein, and total tau, respectively.

Storage tube volume had no effect on the measured α-synuclein concentrations. For neurogranin and total tau, the effect of tube volume was dependent on the storage tube type, with larger and significant effects for PP tubes. For neurogranin, a large PP tube volume decreased the concentration by 8.1% (−10.7%, −5.5%; p < 0.0001) whereas for a LoB storage tube the estimated decrease was only 1.2% (−3.4%, 1.0%; p = 0.275). For total tau, a smaller volume of PP storage tube decreased the concentration significantly by 4.5% (−6.7%, −2.2%; p = 0.001) whereas for a larger volume of the storage tube, the estimated decrease was only 1.6% (−3.4%, 0.2%; not significant (p = 0.073)).

For all three analytes, the effect of an additional freeze-thaw cycle was dependent on the storage tube volume. The effect was more pronounced for tubes with a larger volume. For low volume tubes, the concentration was significantly decreased by an additional freeze-thaw cycle only for neurogranin, amounting to a decrease by 3.4% (−5.8%, −0.8%; p = 0.013). For α-synuclein and total tau, estimated differences in concentration after freezing were below 1%. For the larger storage tubes, the effect of an additional freeze-thaw cycle was a significant decrease in concentration of 7.7% (−10.1%, −5.3%; p < 0.0001) for neurogranin, 3.0% (−4.9%, −1.0%; p = 0.0056) for α-synuclein, and 5.8% (−7.9%, −3.8%; p < 0.0001) for total tau.

With the addition of detergent at the time of collection of CSF (SOPs 11–20), measured concentrations increased for all three analytes ( Fig. 4; right-hand side, red scales) with all standardized mean concentrations for SOPs 11 to 20 being above 100. The differences among SOPs, on the other hand, were largely attenuated. For neurogranin and α-synuclein, none of the factors affected the analyte levels significantly. For total tau, the collection and storage tube type affected the concentration levels significantly even in the presence of detergent. The use of a PP collection tube decreased the total tau concentration by 2.5% (−4.3%, −0.8%; p = 0.008) and the use of a PP storage tube increased the concentration by 1.9% (0.1%, 3.7%; p = 0.036) compared to a LoB tube.

The overall difference in adsorption rate of individual proteins or ratio of CSF proteins is shown in Fig. 5. A comparison was made between protocol 7 (= the reference protocol) and protocol 6 or protocol 1 and as a function of the inclusion of detergent at the time of collection of CSF.

In the absence of detergent at the time of collection (SOPs 1–10), the biggest difference with collection in LoB tubes is obtained with Aβ(1–42), followed by Aβ(1–40) and Aβ(1–38) (data modified according to [15]). Results for Aβ(1–42) and Aβ(N–42) are identical (the latter is quantified by replacement in the assay design of mAb 3D6 (specific for Aβ1) by mAb 1A1 (kindly provided by Eli Lilly; the mAb is specific for AβN)). Changes in concentration of neurogranin as a function of collection protocol are lower than for the Aβ isoforms. Effects are even lower for tau or α-synuclein, but are still significant. Adsorption problems linked to the use of Aβ(1–42) can be normalized in part by the ratio Aβ(1–42)/Aβ(1–40), but not by using the ratio Aβ(1–42)/tau.

In the presence of detergent at collection (SOPs 11–20), all analytes or ratio of analytes show comparable results to those obtained with the reference protocol.