Cerebrospinal Fluid Collection Tubes: A Critical Issue for Alzheimer Disease Diagnosis

To the Editor: Total tau protein (hTau),1 its phosphorylated isoform (p-Tau181P), and Aβ1–42 peptides are the currently accepted cerebrospinal fluid (CSF) biomarkers used as aids in the diagnosis of Alzheimer disease (1). Although polypropylene (PP) was previously reported as the best material for CSF collection tubes (2), heterogeneity in CSF Aβ1–42 values was observed with different PP sampling tubes (3). Because the recommendation to use PP tubes did not lead to standardization of clinical cutoff values (4), we decided to fully address this issue by comparing various types of tubes within an actual clinical work flow and by analyzing the material of different commercially available PP tubes. In the framework of an ethically approved study, we collected CSF samples from 12 patients directly (from the lumbar puncture needle) into 2 PP tubes [BD catalog no. 352096 (BD-PP); Sarstedt catalog no. 62.610.201 (ST-PP)], 1 hemolysis polyethylene tube [Fisher Scientific catalog no. ref.W1773X (HE-PE)], and 1 polystyrene tube [BD catalog no. 352095 (BD-PS)]. CSF biomarker concentrations were measured in parallel in these 4 types of tubes with Innogenetics INNOTEST® kits. We extended this analysis by comparing the results obtained with 11 different commercially available collection tubes labeled as “PP” for a series of fresh (unfrozen) CSF samples from 6 patients (Table 1). Some of these tubes are being used by different clinical teams in the Alzheimer disease field. The material of these 11 tubes was also analyzed by differential scanning calorimetry and Fourier infrared spectroscopy. Although small but significant differences (P < 0.01) for p-Tau181P and hTau were observed in the comparison of the initial 4 tubes, the major discrepancies were for Aβ1–42, with median values being much higher for ST-PP tubes (137%, P < 0.001) compared with the others (80%, 81%. and 99% in BD-PP, BD-PE, and HE-PS tubes, respectively). This result of the superiority of the ST-PP tube is consistent with previous observations (2), but it has raised a question about the homogeneity of PP tubes. The data obtained in the second experiment revealed significant differences among the 11 types of PP tubes, with maximum median variation values of −48% to +31%, −8% to +8%, and −4% to +6% for Aβ1–42, hTau, and p-Tau181P, respectively (Table 1). Although hTau and p-Tau181P values were in the range of the acceptable and observed analytical imprecision for these biomarkers, differences in Aβ1–42 concentrations clearly exceeded them. The effect was present after 15 min, and an additional 24 h of incubation at 2 °C–8 °C did not change these values significantly. These results confirmed and extended previous observations (3), and they indicated that the use of different PP tubes could produce major preanalytical differences that could possibly lead to misclassification of patients. Indicated are peak maximum (°C) and peaks superposition results produced by differential scanning calorimetry and Fourier transform infrared spectroscopy analyses, which allowed determination of the polymer composition of the tubes. Six CSF supernatants from 6 freshly collected samples were placed in 6 tubes for each of the 11 tube types for 15 min. Then Aβ1–42, hTau, and p-Tau181P were measured with commercially available ELISAs. For each sample, the measured concentration in each of the 6 tubes was converted to a percentage of the mean of the values obtained for the 11 tube types. The median percentage for the 6 tubes is reported in the table for each tube type. The P value reported for each biomarker column is the result of the nonparametric Kruskal–Wallis test result for the comparison of tube A with the results for the 10 other tubes. P values <0.05 are in boldface. PE, polyethylene. Indicated are peak maximum (°C) and peaks superposition results produced by differential scanning calorimetry and Fourier transform infrared spectroscopy analyses, which allowed determination of the polymer composition of the tubes. Six CSF supernatants from 6 freshly collected samples were placed in 6 tubes for each of the 11 tube types for 15 min. Then Aβ1–42, hTau, and p-Tau181P were measured with commercially available ELISAs. For each sample, the measured concentration in each of the 6 tubes was converted to a percentage of the mean of the values obtained for the 11 tube types. The median percentage for the 6 tubes is reported in the table for each tube type. The P value reported for each biomarker column is the result of the nonparametric Kruskal–Wallis test result for the comparison of tube A with the results for the 10 other tubes. P values <0.05 are in boldface. PE, polyethylene. Our subsequent physical analysis revealed surprising results, in that only 1 tube consisted of pure PP. The other tubes were copolymers with at least polyethylene (Table 1). Adsorption of the biomarkers on the tube surface was the most likely explanation for this “tube effect.” This hypothesis was supported by the fact that biomarker concentrations in CSF with high concentrations of proteins (>1 g/L obtained in clinical practice, or artificially by adding albumin) were not sensitive to the tube effect. This result was also consistent with the observation that this effect on the Aβ1–42 concentration was present as early as 5 minutes after contact with the tube and was iterative (values decreased again when the sample was put in a new tube). The fact that the impact of the tube was major with Aβ1–42 and not with p-Tau181P suggests that the hydrophilic/hydrophobic balance of the analytes plays an important role in this phenomenon (5). Surprisingly, the pure PP tube (tube K) did not give the best results, and the differences between the tubes suggested that additional surface treatment could also change the adsorption properties of the tubes. The high preanalytical sensitivity of Aβ to different polymer/plastic surfaces is an important observation to take into account. Transfer of CSF into different tubes during processing or storage can cause a 20% to 60% decrease in measured concentrations. In addition, analytical protocols often contain intermediate steps involving transfer of CSF samples into new tubes or plates, a procedure that could also produce possible errors. This effect is true not only for Aβ1–42 but also for other Aβ peptides (Aβ1–40, Aβ1–38), which are of interest in the diagnosis of other neurologic diseases. In summary, our data indicate that the preanalytical impact of sampling/processing tubes has to be particularly well optimized and harmonized for multisite studies and for defining relevant and worldwide cutoff values for Alzheimer disease biomarkers. One way to go is probably to define a consensus protocol that specifically links cutoff values to given sampling tubes and handling/analytical protocols. total tau protein phosphorylated tau isoform cerebrospinal fluid polypropylene BD PP tube, catalog no. 352096 Sarstedt PP tube, catalog no. 62.610.201 Fisher Scientific hemolysis polyethylene tube, catalog no. ref.W1773X BD polystyrene tube, catalog no. 352095.