Measurement of S100B protein: evaluation of a new prototype on a bioMérieux Vidas® 3 analyzer

Abstract Background The addition of S100B protein to guidelines for the management of mild traumatic brain injury (mTBI) decreases the amount of unnecessary computed tomography (CT) scans with a significant decrease in radiation exposure and an increase in cost savings. Both DiaSorin and Roche Diagnostics have developed automated assays for S100B determination. Recently, bioMérieux developed a prototype immunoassay for serum S100B determination. For the first time, we present the evaluation of the S100B measurement using a bioMérieux Vidas® 3 analyzer. Methods We evaluated the matrix effects of serum and plasma, and their stability after storage at 2–8 °C, −20 °C and −80 °C. The new measurement prototype (bioMérieux) was compared with an established one (Roche Diagnostics), and a precision study was also conducted. Lastly, clinical diagnostics performance of the bioMérieux and Roche Diagnostics methods were compared for 80 patients referred to the Emergency Department for mTBI. Results Stability after storage at 2–8 °C, −20 °C, and −80 °C and validation of the serum matrix were demonstrated. The bioMérieux analyzer was compared to the Roche Diagnostics system, and the analytical precision was found to be efficient. Clinical diagnosis performance evaluation confirmed the predictive negative value of S100B in the management of mTBI. Conclusions The study’s data are useful for interpreting serum S100B results on a bioMérieux Vidas® 3 analyzer.


Introduction
Computed tomography (CT) is the gold standard diagnostic tool for diagnosing mild traumatic brain injury (mTBI).However, the use of CT scans is limited due to radiations risks (1)(2)(3) and economic implications (4).Moreover, most CT scans could be avoided because only a small number of mTBI patients have intracerebral lesions (5).An alternative strategy to using CT scan has been developed based on blood protein biomarkers of brain damage, and S100B protein has emerged as a candidate biomarker in children and adults (6,7).The addition of S100B detection to the Scandinavian guidelines for the management of mTBI decreased the need for CT scans among adults by one-third, using a cut-off of 0.10 µg/L (8).
S100B protein is one of the calcium binding proteins found in glial cells.It is a small dimeric cytosolic protein (21 kDa) consisting of ββ or αβ chains.It is involved in a variety of intracellular and extracellular regulatory activities (9,10).Because S100B has a short half-life, a maximal interval of 3 hours is recommended between trauma and blood sampling (11).
Both DiaSorin and Roche Diagnostics have developed automated assays for serum S100B protein determination.In comparison to manual assays, automated assays provide better analytical results with regard to precision, linearity, and accuracy, and they seem to be a preferable option for S100B measurement (12,13).On a pre-analytical level, only one study has evaluated and compared the matrix effect between plasma and serum (14).A few studies have evaluated the stability of S100B after storage at room temperature or at 4°C (15,16).
However, the stability in frozen samples has never been clearly identified in the literature.On an analytical level, the precision (12,13) of DiaSorin and Roche Diagnostics S100B assays as well as a comparison (11)(12)(13)17,18) of both methods have been reported.With regard to diagnostics clinical performance, it is essential that the performance of S100B assays are comparable in order to reduce the use of CT scans after mTBI.
The clinical utility of S100B protein has attracted new companies that specialize in in vitro diagnosis.One company, bioMérieux, has developed an automated prototype for S100B protein assay on a Vidas ® 3 analyzer.To date, no study has evaluated this new technique.For the first time, we set out to evaluate the pre-analytical (matrix effect of serum and plasma, and their stability at 2-8°C, -20°C, and -80°C), analytical (imprecision and comparison), and diagnostics clinical performance in context of mTBI (CT scans reduction in adults with mTBI) of the bioMérieux S100B prototype assay.

Study design and patients
This study was conducted at the Clermont-Ferrand teaching hospital and in accordance with the Declaration of Helsinki principles for ethical medical research involving human subjects.The project was also approved by the institutional ethic review board of the hospital.
Patients and their families were informed of their right to express their disagreement regarding the use of their biological samples.Venous blood samples were taken from 71 patients from the Neuro-Critical Care Unit (NCCU) and 80 patients from the Emergency Department (ED).All study subjects were ≥18 years old.All patients in NCCU (referred in different contexts as subarachnoid hemorrhage, neurosurgical postoperative …) benefit from an S100B assay in their first biological assessment and were included only to evaluate the matrix effect of serum and plasma, the stability after storage at 2-8°C, -20°C, -80°C and the comparison between bioMérieux and Roche Diagnostics assays (Table 1).Three groups of S100B ranges of equal sizes were formed: ≤ 0.10 µg/L; 0.11 to 0.50 µg/L; > 0.50 µg/L for their clinical interest.The concentration of 0.10 µg/L corresponds to the worldly used decision threshold of patients with mTBI (19) whereas concentrations greater than of 0.50 µg/L is relevant for moderate to severe injury (20).Patients from ED were included for the comparison between bioMérieux and Roche assays and the study of clinical diagnostics performance in context of mTBI (Table 1).Patient samples from the ED were drawn into serum gel separator tubes (tube a).Samples from patients in the NCCU (n = 44) were drawn into tube a, serum separator tubes without gel (tube b), and lithium heparinate gel separator tubes (tube c).For the other 27 patients from the NCCU, the samples were drawn into tube a and tube c.

S100B assay
Samples were sent to the laboratory for processing within one hour, then they were centrifuged (2200 g, 15 min).S100B concentrations were determined using an electrochemiluminescence immunoassay on a Cobas e411 ® analyzer (Roche Diagnostics, Meylan, France) and an automated enzyme-linked fluorescence assay on a Vidas ® 3 analyzer (bioMérieux, Marcy l'Etoile, France).The lower limit of detection is 0.005 μg/L for the Roche Diagnostics assay and 0.012 μg/L for the bioMérieux assay.The test result is available in 18 minutes for the Roche Diagnostics assay and 20 minutes for the bioMérieux assay.Both assays detect S100 dimers that contain S100B (S100BB and S100A1B).For bioMérieux assay, one calibrator was used for every 28-days recalibration.For Roche Diagnostics assay, 2 calibrators were used for every 3-months recalibration.For both assays, two control levels were used.

Matrix effect and storage stability
Blood samples from 44 were collected using three different tubes: tube a, tube b, and tube c.The samples were analyzed immediately after centrifugation (T0).Then, the samples were stored 24 h (8 h at room temperature and 16 h at 2-8°C) and analyzed (T24).The samples were again stored at 2-8°C for 24 h and reanalyzed (T48) (Figure 1A).Blood samples from 71 subjects were collected using two different tubes: tube a and tube c.The samples were analyzed immediately after centrifugation, and they were then divided into 5 aliquots to measure the S100B concentration after storage at -20°C (for 1, 3, and 6 months) and -80°C (for 3 and 9 months) (Figure 1B).

Precision and comparison of the methods
The repeatability (n = 30 in 1 day) and reproducibility (n = 30 over 15 days) were determined for the Vidas ® 3 analyzer S100B assay using two quality controls (ref #415739) and two laboratory-made serum pools at clinically-relevant decision levels, including pools for the median and highly pathological range within the reference range.For each pool, 30 aliquots were prepared and frozen at -80°C.For the measurements (made twice daily over 15 days), the aliquots were freshly defrosted, centrifuged, and measured.
For comparison, parallel measurements were performed in the 151 serum samples obtained from the patients (80 from the ED and 71 from the NCCU) using both Roche Diagnostics Cobas e411 ® and bioMérieux Vidas ® 3 analyzers.

Clinical diagnostics performance in context of mTBI
Eighty patients admitted to the ED with mTBI were enrolled in this study.Inclusion criteria for the use of S100B determination in mTBI management were: ≥18 years of age, blood sample draw performed within 3 h after trauma, an mTBI with a medium risk of complications (i.e., with a Glasgow Coma Scale [GCS] equal to 15 and antiplatelet treatment or loss of consciousness or retrograde amnesia 30 min after injury) (21).Exclusion criteria were: an mTBI with a low risk of complications (i.e., an asymptomatic patient without medium-or high-risk criteria and a GCS of 15), an mTBI with a high risk of complications (i.e., a GCS <15 within 2 h post-injury, or focal neurology, post-traumatic convulsion, open skull fracture, embarrassment, signs of skull base fracture, anticoagulant intake, or repeated vomiting) (21).
At the Clermont-Ferrand ED, S100B serum is measured on a Roche Diagnostics Cobas e411 ® assay and integrated into a decision algorithm from the French Society of Emergency Medicine (SFMU) to manage patients with mTBI (21).If the S100B value is <0.10 µg/L, a CT scan is not prescribed; however, a CT scan is recommended if the S100B value is ≥0.10 µg/L (21).In our study, serum S100B concentrations were determined using a Roche Diagnostics Cobas e411 ® assay (value communicated to the physician) and immediately measured on a Vidas ® 3 bioMérieux assay (value not communicated to the physician).The same cut-off (0.10 µg/L) was used for both assays.Clinical data were collected for each patient (CT scan prescription or no CT scan prescription, normal CT scan or CT scan with intracerebral lesion).

Statistics
Statistical analysis was performed using SAS 9.1.3software (SAS Institute, Cary, NC, USA).For quantitative data, normality was verified using the Shapiro-Wilk test.Data were presented as median (minimum [min] and maximum [max], interquartile range [IQR]), as it was not normally distributed.Comparisons of the S100B median concentrations (for the matrix effect and stability study) were interpreted using the Kruskal-Wallis test followed by the Dunn post-test.A P value <0.05 was considered to be statistically significant.To study the clinical performance for patients admitted to the ED, differences in the positive and negative S100B values between the Roche Diagnostics Cobas e411 ® and the bioMérieux Vidas ® 3 assays were compared using McNemar's test.Then, a Fisher's test was used to compare the sub-categories (CT scan prescription, no CT scan prescription).A comparison of the Roche Diagnostics and bioMérieux methods was performed using linear regression and a Bland-Altman plot.The Bland-Altman plot was also used to compare the serum S100B levels (tube a) and plasma S100B concentrations (tube c).

Results
Patient demographics were summarized in Table 1.

Matrix effect and storage stability
At T0, the S100B median concentrations determined in the tube a samples, the tube b samples, and the tube c samples (plasma) were 0.20 µg/L (min: 0.03; max: 4.37; EI: 0.08-0.56),0.20 µg/L (min: 0.03; max: 4.59; EI: 0.08-0.54),and 0.23 µg/L (min: 0.05; max: 4.77; EI: 0.10-0.61),respectively.The difference between these three medians was not statistically significant (p = 0.49).However, the Bland-Altman plot showed higher values for the plasma concentrations.The mean of the differences between the tube c samples and the tube a samples was 0.07 µg/L (Figure 2).Global overestimation of the plasma concentrations was approximately 20%.
For samples of tubes a, b, and c, the median concentrations determined at T0, T24, and T48 are summarized in Table 2.No significant difference was identified, between the three times, for tube a samples (p = 0.83), tube b samples (p = 0.95), and tube c samples (p = 0.56) (Table 2).
No statistically significant difference was found between the median concentrations of S100B after storage at -20°C (Table 3) or at -80°C (Table 4).

Precision and comparison of the methods
The coefficients of variation of repeatability and reproducibility were summarized in Table 5 for quality controls and laboratory-made serum pools.Repeatability was evaluated by 30 repeated measurements of the same sample whereas reproducibility was evaluated by 30 measurements taken over 15 days.
The Bland-Altman plot showed a mean difference between the two methods of -0.06 µg/L (Figure 3B).

Clinical diagnostics performance in context of mTBI
S100B measurement was indicated for the management of each of the 80 patients with mTBI admitted to the ED.The median age was 42 years (min: 18; max: 91; EI: 21 -65) and the sex ratio male/female was 1.58 (Table 1).S100B concentrations were measured using both the Roche Diagnostics and bioMérieux methods.The percentages of negative S100B (<0.10 µg/L) were 40% and 37.5%, for the Roche Diagnostics and bioMérieux methods, respectively.No statistically significant difference was found between these two techniques (p = 0.41) (Table 6).For the three patients with intracerebral lesions on CT scan, the S100B concentrations were positive (≥0.10 µg/L) with the Roche Diagnostics and bioMérieux methods.The ability of S100B to identify intracerebral lesions was not significantly different between the two analyzers (p = 0.53) (Table 6).The physicians did not use the decision algorithm for 10% of the patients.For 5% of the patients, a CT scan was prescribed despite a negative S100B.For an additional 5% of the patients, a CT scan was not prescribed, even though the S100B values were positive (Table 6).

Discussion
Currently, two companies specialize in in vitro diagnosis: Roche Diagnostics and DiaSorin.Both companies offer automated analyzers that can determine the S100B protein concentration in serum.For the first time, the present study evaluated the analytical performances of an automated prototype for an S100B protein assay: the bioMérieux Vidas ® 3. It should be noted that the Vidas ® analyzer is the most widely used immunoassay system in clinical laboratories, worldwide.Compared with the Cobas ® analyzer, the Vidas ® offers several advantages such as the fast adoption (easier-to-use assay), the cost avoidance in low throughput samples (1 test for 1 patient), the random access with up to 4 independent sections and no daily maintenance.
It is essential to obtain data on the influence of the matrix effect and storage in order to avoid pre-analytical errors that may affect the interpretation of the test result.Evaluation of the matrix effect showed an overestimation of ≈ 20 % for the plasma values in comparison to the serum values.This difference was also reported by Tort et al. (14).Two other publications evaluated the impact of heparin treatment on S100B measurement (22,23).Wang et al.
showed a rapid increase in serum S100B (1.74 fold) within 15 min of unfractionated heparin administration (1.74 fold) and 3 h after low molecular weight heparin injection (1.44 fold) (23).While heparin constantly appears to increase the immuno-reactivity for S100B, the presence of the other anticoagulants in blood samples interferes chaotically with the S100B measurement.Indeed, ethylenediaminetetraacetic acid (EDTA) and citrate are calcium chelators, whereas heparin action is independent of calcium (14).Our stability study showed no difference between the results of the S100B measurements after storing the samples at room temperature for 8 h and at 2-8°C for 48 h.Our findings are in agreement with Raabe et al. who demonstrated that storage at 4°C for 48 h did not affect the measurement result (15).
In contrast, Djukanovic et al. showed that storing samples at room temperature after 3 h was associated with an increase in the S100B values (16).In terms of S100B measurement after freezing, we demonstrated, for the first time, stability after storing the samples at -20°C for 6 months and at -80°C for 9 months.However, three plasma samples showed unsatisfactory storage with post-freeze concentrations 7-to 10-times higher than initial concentrations.
Knowing that S100B is expressed in lymphocyte cells (13,24), we hypothesized that a release of protein after lysis was caused by freezing of residual lymphocytes.Indeed, despite using a separator gel, a significant proportion of the cells (leukocytes, red blood cells, and platelets) were found in the plasma (25).
The precision study (repeatability and reproducibility) results demonstrated that the bioMérieux method was very satisfactory, which is consistent with the data reported by the manufacturer.The within run CVs (1.7-2.2%) and between run CVs (2.8-3.9%) of the bioMérieux method were less than the CVs of the DiaSorin method (within run CV: 4.0-5.6%;between run CVs: 3.7-6.1%),and they were very similar to those of the Roche Diagnostics method (within run CV: 1.1-2.0%; between run CVs: 1.8-2.1%)(13).Therefore, the precision of the Vidas ® 3 S100B prototype is comparable to the two analyzers that are routinely used in clinical settings.
Moreover, the bioMérieux and Roche Diagnostics methods showed good correlation (r = 0.97) for S100B protein determination.However, the two methods were not interchangeable; the mean values were 13% higher with the bioMérieux assay.Moreover, the gap between these two techniques widened as the S100B concentrations increased.Overall, the results of this study are in line with the findings reported in the literature; that is, the results of these two immunoassays are correlated, but not interchangeable.In the literature, comparisons of the Roche Diagnostics and DiaSorin methods showed an overestimation close to 30% for the DiaSorin assay (11)(12)(13)17,18).However, the three methods detect S100 dimers that contain S100B (S100BB and S100 A1B).The use of a single S100B protein assay technique is essential for patient management and follow-up.However, currently, no diagnostic test is registered for patient follow-up.The absence of a standardized S100 immunoassay could clearly explain the observed difference reported in the literature.
In our study, the addition of S100B to the SFMU's algorithm could reduce the use of CT scans by 40% and 37.5% for the Roche Diagnostics and bioMérieux assays, respectively.This better specificity than that usually described (33%) is probably due to an age effect.Indeed, in our study, the proportion of people over 65 years is lower than in a previous study showing that the specificity is lower on this age group (21).
There are some limitations to our study.The cutoff of 0.10 µg/L was evaluated for the management of adults with mTBI (n = 80).However, interventionnal studies on larger cohorts are required.Moreover, it would be interesting to study S100B in other indications than mTBI like subarachnoid hemorrhage and severe trauma.Reference ranges of serum S100B concentration should also be made for children as for Roche Diagnostics (26) and DiaSorin (13) technology.
In conclusion, this study provided valuable new data for the concerted interpretation of S100B assay results by biologists and clinicians.Pre-analytical stability and validation of serum matrix were demonstrated.The bioMérieux analyzer was compared to the Roche Diagnostics system.The precision study results for the bioMérieux Vidas ® 3 analyzer showed that this method had very satisfactory results.Clinical diagnostics performance confirmed the predictive negative value of the S100B biomarker in the management of mTBI in adults.
mTBI: mild Traumatic Brain Injury NCCU: Neuro-Critical Care Unit SD: Standard Deviation SFMU: French Society of Emergency Medicine

Figure 1
Figure 1 Pre-analytical study: matrix effect and stability.(A) Storage at room temperature and at 2-8°C.(B) Storage at -20°C and -80°C.Tube a: serum separator tube with gel; tube b: serum separator tube without gel; tube c: lithium heparinate gel separator tube; T0: measurement after centrifugation; T24: measurement after storage for 24 h (8 h at room temperature and 16 h at 2-8°C); T48: measurement after storage for 48h (8 h at room temperature and 40 h at 2-8°C).

Figure 2
Figure 2 Plot of the differences in the S100B concentrations between the serum and plasma samples against the average of the two samples.Tube a: serum gel separator tubes; Tube c: lithium heparinate gel separator tubes; SD: standard deviation.

Table 2
Stability of the S100B protein after storage for 48 h (8 h at room temperature and 40

Table 5
Precision study of the S100B prototype assay on a bioMérieux Vidas ® 3 analyzer.