**2. Influence of confounding factors in pre-analytical phases on the analysis of AD biomarkers**

The confounding factors in pre-analytical phases have a great importance to biochemical analysis and can affect the reliability of the results. Specially in the context of biomarkers of AD in CSF, there are some experimental studies that support this proposition [10,11,12]. Those factors are classically dichotomized in two different groups, «in vivo» and «in vitro». The «in vivo» factors are those biological factors that are linked directly to the patient, the «in vitro» factors are linked to the procedure of sample handling and processing.

#### **2.1.** *In vivo* **factors**

in a patient presenting the core clinical diagnosis [4]. The CSF biomarker panel of AD is a picture of the neurodegeneration, the neuronal loss, the tangle formation and Aβ-amyloid<sup>42</sup> (Aβ42) peptide accumulation in the brain. Indeed, the core CSF biomarkers for AD diagnosis are a decrease of Aβ42 levels and more recently a decrease of the ratio of Aβ-amyloid42 / Aβamyloid40 (Aβ42/Aβ40) which reflect senile plaques pathology as well as an increase of total tau (T-tau) and phosphorylated tau (P-tau) which reflect axonal degeneration [5,6]. The use of AD biomarker tests for routine diagnostic purposes at the present time, is only proposed as optional for use in patients with dementia when deemed appropriate by the clinician. From the several reasons for this limitation, the workgroup with the task of revising the 1984 criteria for Alzheimer's disease (AD) dementia, highlighted the limited standardization of biomarkers from one locale to another [4]. Despite a decrease in the number of side effects associated with the puncture, lumbar puncture remains an invasive procedure that is clearly the main factor preventing the wide dissemination of these biomarkers in the routine. However, we cannot ignore that the significant variability in measured biomarkers levels found in various studies, resulting in a high variability of both the diagnostic accuracy [7] and of the clinical cut-off for the diagnostic of AD [8], is a hindrance to the spread of these markers and their integration in the diagnostic criteria [3]. The cut-offs obtained in Europe for CSF total tau and beta-amyloid measured by the ELISA assays from the same manufacturer, were reported highly diverse, with two to three fold differences between the highest and lowest reported values [8]. Three major explanations are proposed in this report: first, the inter-laboratory comparisons are very difficult, as some laboratories have adopted the cut-off values from the research literature whereas others have established their own controls, these last controls being likely different in neuropsychology evaluation, neuroimaging and the follow-up. Secondly, the lack of standardized material between the different assays but also the lack of standardized protocols, seem to be a major source of this variation. Finally, pre-analytical factors are those factors that contribute to the variation of the laboratory results before the analysis of the sample. One consensus report has already established the main pre-analytical factors that should be standardized for CSF AD biomarkers analysis [9]. However, the importance of some preanalytical confounding factors highlighted in this report remained to be elucidated. The aim of this report is to discuss and focus on main critical points in the different preanalytical steps likely to be responsible of data variability. For analytical steps, the introduction since 2009 of an external quality control at a large scale gave an overview of the «desaster», in the same line that prior results. We will discuss rapidly the prior results reported in 2011 and we will

**2. Influence of confounding factors in pre-analytical phases on the analysis**

The confounding factors in pre-analytical phases have a great importance to biochemical analysis and can affect the reliability of the results. Specially in the context of biomarkers of AD in CSF, there are some experimental studies that support this proposition [10,11,12]. Those factors are classically dichotomized in two different groups, «in vivo» and «in vitro». The «in

underline the urgent need for standardization.

**of AD biomarkers**

178 Understanding Alzheimer's Disease

#### *2.1.1. Is there a specific time of day needed to collect the CSF?*

Answering this issue needs to know if a nycthemeral cycle exists that could modify the concentrations levels of AD CSF biomarkers during the day. Although a lack of standardiza‐ tion in the diagnostic strategy of the patient still exists, in most cases, after a first examination including a clinical and a neuropsychological evaluation, if needed, the lumbar puncture is generally scheduled in a second visit with morphological brain imaging in the same time, with the aim to minimize the duration of the hospitalization. As the time of the lumbar puncture is highly dependent of the coordinated organization of the clinical memory centre, of the biological laboratory and of the imaging department associated with it (waiting homeostasis results, scheduling imaging...), this question is highly relevant.

Previous results have suggested the existence of a large diurnal variability in Aβ levels during a time period of 36 hours, but without significant differences between the hours all along the day period [13]. Following these amazing and unexplained data, recent studies were unable to demonstrate the existence of a temporal fluctuation in CSF biomarker levels, not only for Aβ, but also for T-tau and P-tau [10, 14, 15]. Therefore, there is no need to standardize a specific time interval during the day for CSF collection dedicated to the AD biomarkers assays.

#### *2.1.2. Is fasting able to modify the concentrations levels of AD biomarkers ?*

At our knowledge, there are no study that has analyzed the influence of fasting on AD CSF biomarkers. The comparison of patients with and without fasting would give a set of indirect and biased data without clear conclusion. Moreover, for ethical reasons, it seems to be impossible to start a research study focused on this topic, as this study would imply a protocol with the realization of successive lumbar punctures in a short delay. Therefore, it is not possible to answer scientifically this issue. Nevertheless, it has been shown that, independently of the patient food intake, Aβ levels in plasma are very stable [10]. As there is a lack of data concerning this topic, as those kind of data could probably never be obtained, and taking account of the large diversity in the locale organization, it is not logical to recommend fasting for the analysis of AD biomarkers in CSF.

#### **2.2.** *In vitro* **factors**

#### *2.2.1. Localization of the puncture*

Due to the possible decreased rostro-caudal concentration gradient, the site of CSF withdrawal must be also standardized. At our knowledge, there is no study reporting any difference between AD biomarkers concentrations obtained by a ventricular puncture and those obtained by lumbar puncture. Therefore, it is not recommended to analyse these markers in the ventricular punctures obtained during neurosurgical interventions. Nowadays, diagnostic CSF is usually obtained by LP between the L3/L4 and L4/L5 intervertebral space.

hydrophobic balance is a important point in protein adsorption [11, 23]. The variability of adsorption intensity of proteins onto the plastic of the tube is the result of the incredible jungle of the manufacturing of different tubes called PP: difference in the nature and in the percentage of the copolymers in the plastic, presence of additives, surface treatments, modification of the surface by the sterilization process... The possibility of modifying the protein adsorption by additives or surface treatments was underlined by different reports. First, when Tween-20 was added in the tube containing the CSF, the adsorption of amyloid peptides was significantly reduced [22]. Secondly we recently reported similar results using various plasma treatments of the tube surface, able to modify the adsorption of different proteins like prion protein, Tau and alpha synuclein [23]. These data highlight the need to standardize also the type of test tube used since the great variability found could even lead to a possible AD misdiagnosis. In our laboratory, we shifted to the best tube that we found in this study. This shift has introduced an averaged increase of 25 % of Aβ42 levels leading to a modification of our cut off diagnostic value from 500 ng/L to 700 ng/L (data submitted). Currently the members of the Joint Pro‐ gramming Neurodegenerative Disease research (JPND*)* are performing a study which includes the analysis of the most suitable type of tube for AD CSF biomarkers research. Therefore, it is not reasonable to follow the actual guidelines recommending the use of generic PP tubes. Since the data of the JPND collaboration will probably not be available before 2 or 3 years, the best compromise would be that each laboratory concerned by these markers, compares its local tube with the best tubes identified in our study, which are easily available in the commercial

Pre-Analytical and Analytical Critical Factors Influencing the High Variability of the Concentrations Levels of...

http://dx.doi.org/10.5772/55512

181

This is an issue difficult to standardize due the high variety of existing procedures and its probable dependance of confounding factors (hemorragic puncture, hemolyzed samples, high levels of total protein, one sampling tube for AD biomarkers and various markers of others pathologies...) which could modify the stability of the biomarkers during this critical

For that, we will discuss first the need to centrifuge and the protocol of centrifugation. This step is able to avoid the presumed influence of the blood cells introduced by the hemorragic puncture. These hemorragic punctures occur in 14-20% cases of lumbar puncture. Bjerke et al. were unable to detect any difference in Aβ42 levels when up to 5000 erythrocytes/µl were spiked to the CSF. This value was found ten fold higher than those recommended in the regulation's document included in the Innogenetics kits. However, they found significant decreased Aβ<sup>42</sup> levels in CSF when plasma was added which was attributed to the binding of Aβ42 to different plasma proteins [10]. We cannot also neglect the presence of plasmatic proteases able to digest the peptides since it has been shown that blood contamination of CSF can also lead to protein degradation [25]. The guidelines of Vanderstichele et al. pointed out the absence of difference on the levels of Aβ42, T-tau and P-tau between centrifuged and non-centrifuged samples (N. Le Bastard, unpublished data) [9] which could be explained by the fact that they used clear CSF samples. In these guidelines, it was pointed out that spinning speed did not modify significantly the concentration levels of the biomarkers. More recently, it was reported that the

market.

period.

*2.2.5. Time delay between CSF collection and storage before assay*

#### *2.2.2. Does a CSF gradient of AD biomarkers exist ?*

Most brain-derived proteins have a decreased rostro-caudal concentration gradient [16]. Therefore, the volume of CSF taken can influence protein concentration. Using unpublished data from Le Bastard et al., Vanderstichele et al reported the absence of a gradient effect in AD CSF biomarkers concentrations during [9]. It was confirmed by another experimental study analyzing the gradient effect in the spinal cord on Aβ<sup>42</sup> [10]. Therefore, there is no reason to recommend any specific fraction of CSF volume for the assay of AD biomarkers.

#### *2.2.3. What kind of needle for the puncture ?*

The type of needle is likely known to influence the percentage of side effects in patients and to be a factor leading to the presence of red cells [17, 18]. Therefore, the needle could influence the biomarkers concentrations. It has been shown that post-lumbar puncture headache (PLPH) severity was significantly decreased when a 22G needle was used instead of a 20G needle [18]. Moreover, using a 22G atraumatic needle it was also observed a remarkably decrease of PLPH in comparison with 22G traumatic needles [19]. Finally, as lumbar puncture is sometimes difficult with 25G needle in elderly people, a korean group has compared the prevalence of PLPH using 23G and 25G needles. They concluded that the choice of a 23 or 25 gauge Quincke needle has no significant influence on post-dural puncture headache for Korean patients greater than 60 years old. Therefore, the 23 gauge Quincke needle is an option for lumbar punctures in this patient population [20].

#### *2.2.4. Types of sampling tubes*

It was established that polypropylene (PP) tubes should be preferred to glass or polystyrene tubes for collection of the CSF since Aβ peptides, but also T-tau and P-tau, bind in a non specifically manner to the polystyrene tubes and to the glass tubes [10, 21]. However, two independent studies reported significant differences on Aβ<sup>42</sup> levels (up to 50 % compared to basal values !) when CSF was collected in PP tubes from different suppliers [11, 22]. For Aβ42, we found that adsorption was effective in a contact time less than 15 minutes, the loss of Aβ<sup>42</sup> levels being highly significant [11]. Moreover the adsorption intensity was highly dependent on the levels of total proteinorachia, since we abolished this phenomenon when we spiked the CSF with solutions of bovine serum albumin. Amazingly, we also shown that, whereas all the tubes that we studied were commercialized by the providers as tubes in PP, a calorimetry and a spectroscopy analysis revealed that just one out of 11 tubes was pure PP while the others were copolymers made of PP and polyethylene (PE) [11]. Moreover, we also shown that the pure PP causes more adsorption of amyloid peptides than tubes in copolymers of PE and PP, with or without treatment surface, and that some tubes in copolymers could be worst than classical polystyrene: these highly striking results were reproducible in the independent laboratories which have collaborated in this study [11]. Moreover, it was also observed that the tubes that performed better for Aβ<sup>42</sup> were the worst for P-tau suggesting that hydrophilichydrophobic balance is a important point in protein adsorption [11, 23]. The variability of adsorption intensity of proteins onto the plastic of the tube is the result of the incredible jungle of the manufacturing of different tubes called PP: difference in the nature and in the percentage of the copolymers in the plastic, presence of additives, surface treatments, modification of the surface by the sterilization process... The possibility of modifying the protein adsorption by additives or surface treatments was underlined by different reports. First, when Tween-20 was added in the tube containing the CSF, the adsorption of amyloid peptides was significantly reduced [22]. Secondly we recently reported similar results using various plasma treatments of the tube surface, able to modify the adsorption of different proteins like prion protein, Tau and alpha synuclein [23]. These data highlight the need to standardize also the type of test tube used since the great variability found could even lead to a possible AD misdiagnosis. In our laboratory, we shifted to the best tube that we found in this study. This shift has introduced an averaged increase of 25 % of Aβ42 levels leading to a modification of our cut off diagnostic value from 500 ng/L to 700 ng/L (data submitted). Currently the members of the Joint Pro‐ gramming Neurodegenerative Disease research (JPND*)* are performing a study which includes the analysis of the most suitable type of tube for AD CSF biomarkers research. Therefore, it is not reasonable to follow the actual guidelines recommending the use of generic PP tubes. Since the data of the JPND collaboration will probably not be available before 2 or 3 years, the best compromise would be that each laboratory concerned by these markers, compares its local tube with the best tubes identified in our study, which are easily available in the commercial market.

#### *2.2.5. Time delay between CSF collection and storage before assay*

ventricular punctures obtained during neurosurgical interventions. Nowadays, diagnostic

Most brain-derived proteins have a decreased rostro-caudal concentration gradient [16]. Therefore, the volume of CSF taken can influence protein concentration. Using unpublished data from Le Bastard et al., Vanderstichele et al reported the absence of a gradient effect in AD CSF biomarkers concentrations during [9]. It was confirmed by another experimental study analyzing the gradient effect in the spinal cord on Aβ<sup>42</sup> [10]. Therefore, there is no reason to

The type of needle is likely known to influence the percentage of side effects in patients and to be a factor leading to the presence of red cells [17, 18]. Therefore, the needle could influence the biomarkers concentrations. It has been shown that post-lumbar puncture headache (PLPH) severity was significantly decreased when a 22G needle was used instead of a 20G needle [18]. Moreover, using a 22G atraumatic needle it was also observed a remarkably decrease of PLPH in comparison with 22G traumatic needles [19]. Finally, as lumbar puncture is sometimes difficult with 25G needle in elderly people, a korean group has compared the prevalence of PLPH using 23G and 25G needles. They concluded that the choice of a 23 or 25 gauge Quincke needle has no significant influence on post-dural puncture headache for Korean patients greater than 60 years old. Therefore, the 23 gauge Quincke needle is an option for lumbar

It was established that polypropylene (PP) tubes should be preferred to glass or polystyrene tubes for collection of the CSF since Aβ peptides, but also T-tau and P-tau, bind in a non specifically manner to the polystyrene tubes and to the glass tubes [10, 21]. However, two independent studies reported significant differences on Aβ<sup>42</sup> levels (up to 50 % compared to basal values !) when CSF was collected in PP tubes from different suppliers [11, 22]. For Aβ42, we found that adsorption was effective in a contact time less than 15 minutes, the loss of Aβ<sup>42</sup> levels being highly significant [11]. Moreover the adsorption intensity was highly dependent on the levels of total proteinorachia, since we abolished this phenomenon when we spiked the CSF with solutions of bovine serum albumin. Amazingly, we also shown that, whereas all the tubes that we studied were commercialized by the providers as tubes in PP, a calorimetry and a spectroscopy analysis revealed that just one out of 11 tubes was pure PP while the others were copolymers made of PP and polyethylene (PE) [11]. Moreover, we also shown that the pure PP causes more adsorption of amyloid peptides than tubes in copolymers of PE and PP, with or without treatment surface, and that some tubes in copolymers could be worst than classical polystyrene: these highly striking results were reproducible in the independent laboratories which have collaborated in this study [11]. Moreover, it was also observed that the tubes that performed better for Aβ<sup>42</sup> were the worst for P-tau suggesting that hydrophilic-

CSF is usually obtained by LP between the L3/L4 and L4/L5 intervertebral space.

recommend any specific fraction of CSF volume for the assay of AD biomarkers.

*2.2.2. Does a CSF gradient of AD biomarkers exist ?*

180 Understanding Alzheimer's Disease

*2.2.3. What kind of needle for the puncture ?*

punctures in this patient population [20].

*2.2.4. Types of sampling tubes*

This is an issue difficult to standardize due the high variety of existing procedures and its probable dependance of confounding factors (hemorragic puncture, hemolyzed samples, high levels of total protein, one sampling tube for AD biomarkers and various markers of others pathologies...) which could modify the stability of the biomarkers during this critical period.

For that, we will discuss first the need to centrifuge and the protocol of centrifugation. This step is able to avoid the presumed influence of the blood cells introduced by the hemorragic puncture. These hemorragic punctures occur in 14-20% cases of lumbar puncture. Bjerke et al. were unable to detect any difference in Aβ42 levels when up to 5000 erythrocytes/µl were spiked to the CSF. This value was found ten fold higher than those recommended in the regulation's document included in the Innogenetics kits. However, they found significant decreased Aβ<sup>42</sup> levels in CSF when plasma was added which was attributed to the binding of Aβ42 to different plasma proteins [10]. We cannot also neglect the presence of plasmatic proteases able to digest the peptides since it has been shown that blood contamination of CSF can also lead to protein degradation [25]. The guidelines of Vanderstichele et al. pointed out the absence of difference on the levels of Aβ42, T-tau and P-tau between centrifuged and non-centrifuged samples (N. Le Bastard, unpublished data) [9] which could be explained by the fact that they used clear CSF samples. In these guidelines, it was pointed out that spinning speed did not modify significantly the concentration levels of the biomarkers. More recently, it was reported that the sample temperature was always similar to the temperature set up in the centrifuge showing that temperature is not increased by spinning itself [26]. We can then recommend, that centrifugation should be performed at 2,000 g during 10 minutes at room temperature (RT) following the standardized protocol [26].

These data confirmed those obtained for sampling tubes, although the variability was lower than those found with these last tubes. The effect was present after 15 min, but increasing the incubation time to 24h at 2-8°C, the values did not significantly change compared to 15 minutes

Pre-Analytical and Analytical Critical Factors Influencing the High Variability of the Concentrations Levels of...

http://dx.doi.org/10.5772/55512

183

The next step consists to standardize if needed the temperature, i.e. the speed, of freezing.

Freezing temperatures may affect CSF proteins concentrations as it has previously been reported for cystatin C, which undergoes a proteolysis at -20°C but not at -80°C [31]. Recently, the levels of T-tau and P-tau were reported significantly lower when CSF samples were immediately frozen at -20°C instead of -80°C (N. Le Bastard, unpublished data) [19]. However, this group did not find any difference for the Aβ<sup>42</sup> levels when the CSF were frozen at -20°C or -80°C, confirming previous results [10]. Therefore, freezing and storage at -80°C the CSF

Aliquoting the supernatant of CSF is absolutely necessary since it avoids different Freeze/thaw cycles (see below). Although we did not realize a study designed to evaluate the possible synergy between the ratio volume/surface and the speed of freezing onto the absorption phenomenon in these storage tubes (total volume less than 1.5 ml), some procedures issued from previous reported guidelines can be logically applied [27]. They pointed out the need to use small volumes (never more than 0.5 ml), which would allow: a/to realize at least the assay of the 3 classical AD biomarkers and if needed the assay of Aβ40, b/ to prevent freeze/thaw cycles and c/ fill the tube up to 75% to minimize the adsorption and the evaporation effect, this

As mentioned before, the guidelines recommend separating the supernatant in several fractions, that which will reduce the numbers of freeze/thaw cycles since freezing was shown able to affect protein stability [32]. Some studies have already analyzed the influence of freeze/ thaw cycles on AD CSF biomarkers. Most studies using an ELISA format no have found any change on Aβ42 and Tau CSF levels after one freeze/thaw cycle [10, 12, 30, 33], whereas a significant loss of Aβ42 was found after one single cycle in one study using a semi-quantitative method [34]. Increasing the number of cycles was reported able to modify the stability of Aβ<sup>42</sup> CSF levels. However, about the exact numbers of cycles able to impact the levels, no real consensus was found between the different studies. If the Tau CSF levels seem to be unaffected by 3 or 6 freeze/thaw cycles [30, 12], the Aβ<sup>42</sup> CSF levels were found either stable after 3 cycles [30], either were significantly decreased after the third cycle [12]. In case of immunoassay analysis, it is logically recommended to limit the number of freeze/thaw cycles up to two as

Finally, the length of storage at -80°C does not seem to present a major influence on stability of CSF AD biomarkers, at least for 2 years [30] according to unpublished data from Blennow K. et al., referenced in the guideline published by Vanderstichele et al. [19]. Moreover, the levels of Aβ42 and T-tau but not Aβ40, remained stable up to 6 years [35]. In summary, we can conclude that CSF can be stored up to 2 years at -80°C as previously

last effect being negligible when the sample is stored frozen at -80°C [26].

incubation.

samples, seem to be logical.

maximum [9].

reported [19].

If several publications and recommendations are related to the delay between sampling and storage [27], it seems that there is a lack of conclusive data about the influence of the delay between sampling and centrifugation for AD biomarkers, mainly for hemorragic puncture. Nevertheless, it was reported significant changes of various metabolites, various amino acids and proteins in presence of white blood cells in the CSF, using a proteomics approach when the CSF were left at RT in the first 30 minutes [28].These data could explain the apparent discrepancy between the study of Kaiser et al, describing a significant increase of the levels of Aβ42 after 24 hours [29] and those of Bjerke, describing that Aβ42 concentrations remained stable up to 24 hours after the sampling (storage at RT) [10]. The lack of centrifugation prior incuba‐ tion is likely the reason of the increase in Aβ42 previously observed. Taken all together, all these data highlight the importance of centrifugation to be realized, as soon as possible after sampling, for CSF biomarker analysis.

Although the aspect of the CSF was not always indicated, we can imagine that the different studies which have reported a stability of the CSF levels of Aβ42, Aβ40, T-tau and P-tau over a period of 24 h at least, were done with clear CSF. Thus, the concentrations of Aβ42 were found stable 24 h [10], 72 h when the sample was stored at 4°C [12] and up to 7 days after LP at RT [30]. It was the same for the concentrations of T-tau [10, 12, 29, 30]. Regarding the temperature during the time delay, no significant difference was found between the storage of the CSF samples at RT, 4°C or frozen in any of the studies performed [9, 10].

#### *2.2.6. Freezing process*

This process is complex since different factors could influence the biomarkers concentra‐ tions: although it seems clear that heterogeneity also exists for storage tubes, the tempera‐ ture of freezing, the volume of the aliquots, the length of the storage and the possible effect of freezing / thawing cycles are potential factors to evaluate. Moreover, these factors can be synergistic: the adsorption of proteins onto the tube walls could be increased by the lower volume of the aliquot and mainly by the ratio volume / surface, or by the tempera‐ ture of freezing (-20 versus -80°C).

The first step is to choose a storage tube. In parallel to the test realized with 11 sampling tubes [11], we selected 9 different commercially available polypropylene storage tubes (Table 1, tubes 13 to 21), some of them being used by different clinical teams in the AD field. The volume capacity was ranged from 0,5 to 1,5 mL. We performed an analysis of the surface polymer composition using differential scanning calorimetry and Fourier Transformed Infrared spectroscopy. This revealed the same surprising results than obtained with the sampling tubes [11]: only one tube was constituted by pure polypropylene, the others being copolymers with at least polyethylene, with or without surface treatment. Using the same protocol as described for the sampling tubes [11], biomarkers concentrations showed variations that were signifi‐ cantly different for Aβ<sup>42</sup> peptide. Median values for Aβ<sup>42</sup> peptide varied from 94 % to 127 %. These data confirmed those obtained for sampling tubes, although the variability was lower than those found with these last tubes. The effect was present after 15 min, but increasing the incubation time to 24h at 2-8°C, the values did not significantly change compared to 15 minutes incubation.

sample temperature was always similar to the temperature set up in the centrifuge showing that temperature is not increased by spinning itself [26]. We can then recommend, that centrifugation should be performed at 2,000 g during 10 minutes at room temperature (RT)

If several publications and recommendations are related to the delay between sampling and storage [27], it seems that there is a lack of conclusive data about the influence of the delay between sampling and centrifugation for AD biomarkers, mainly for hemorragic puncture. Nevertheless, it was reported significant changes of various metabolites, various amino acids and proteins in presence of white blood cells in the CSF, using a proteomics approach when the CSF were left at RT in the first 30 minutes [28].These data could explain the apparent discrepancy between the study of Kaiser et al, describing a significant increase of the levels of Aβ42 after 24 hours [29] and those of Bjerke, describing that Aβ42 concentrations remained stable up to 24 hours after the sampling (storage at RT) [10]. The lack of centrifugation prior incuba‐ tion is likely the reason of the increase in Aβ42 previously observed. Taken all together, all these data highlight the importance of centrifugation to be realized, as soon as possible after

Although the aspect of the CSF was not always indicated, we can imagine that the different studies which have reported a stability of the CSF levels of Aβ42, Aβ40, T-tau and P-tau over a period of 24 h at least, were done with clear CSF. Thus, the concentrations of Aβ42 were found stable 24 h [10], 72 h when the sample was stored at 4°C [12] and up to 7 days after LP at RT [30]. It was the same for the concentrations of T-tau [10, 12, 29, 30]. Regarding the temperature during the time delay, no significant difference was found between the storage of the CSF

This process is complex since different factors could influence the biomarkers concentra‐ tions: although it seems clear that heterogeneity also exists for storage tubes, the tempera‐ ture of freezing, the volume of the aliquots, the length of the storage and the possible effect of freezing / thawing cycles are potential factors to evaluate. Moreover, these factors can be synergistic: the adsorption of proteins onto the tube walls could be increased by the lower volume of the aliquot and mainly by the ratio volume / surface, or by the tempera‐

The first step is to choose a storage tube. In parallel to the test realized with 11 sampling tubes [11], we selected 9 different commercially available polypropylene storage tubes (Table 1, tubes 13 to 21), some of them being used by different clinical teams in the AD field. The volume capacity was ranged from 0,5 to 1,5 mL. We performed an analysis of the surface polymer composition using differential scanning calorimetry and Fourier Transformed Infrared spectroscopy. This revealed the same surprising results than obtained with the sampling tubes [11]: only one tube was constituted by pure polypropylene, the others being copolymers with at least polyethylene, with or without surface treatment. Using the same protocol as described for the sampling tubes [11], biomarkers concentrations showed variations that were signifi‐ cantly different for Aβ<sup>42</sup> peptide. Median values for Aβ<sup>42</sup> peptide varied from 94 % to 127 %.

samples at RT, 4°C or frozen in any of the studies performed [9, 10].

following the standardized protocol [26].

182 Understanding Alzheimer's Disease

sampling, for CSF biomarker analysis.

ture of freezing (-20 versus -80°C).

*2.2.6. Freezing process*

The next step consists to standardize if needed the temperature, i.e. the speed, of freezing.

Freezing temperatures may affect CSF proteins concentrations as it has previously been reported for cystatin C, which undergoes a proteolysis at -20°C but not at -80°C [31]. Recently, the levels of T-tau and P-tau were reported significantly lower when CSF samples were immediately frozen at -20°C instead of -80°C (N. Le Bastard, unpublished data) [19]. However, this group did not find any difference for the Aβ<sup>42</sup> levels when the CSF were frozen at -20°C or -80°C, confirming previous results [10]. Therefore, freezing and storage at -80°C the CSF samples, seem to be logical.

Aliquoting the supernatant of CSF is absolutely necessary since it avoids different Freeze/thaw cycles (see below). Although we did not realize a study designed to evaluate the possible synergy between the ratio volume/surface and the speed of freezing onto the absorption phenomenon in these storage tubes (total volume less than 1.5 ml), some procedures issued from previous reported guidelines can be logically applied [27]. They pointed out the need to use small volumes (never more than 0.5 ml), which would allow: a/to realize at least the assay of the 3 classical AD biomarkers and if needed the assay of Aβ40, b/ to prevent freeze/thaw cycles and c/ fill the tube up to 75% to minimize the adsorption and the evaporation effect, this last effect being negligible when the sample is stored frozen at -80°C [26].

As mentioned before, the guidelines recommend separating the supernatant in several fractions, that which will reduce the numbers of freeze/thaw cycles since freezing was shown able to affect protein stability [32]. Some studies have already analyzed the influence of freeze/ thaw cycles on AD CSF biomarkers. Most studies using an ELISA format no have found any change on Aβ42 and Tau CSF levels after one freeze/thaw cycle [10, 12, 30, 33], whereas a significant loss of Aβ42 was found after one single cycle in one study using a semi-quantitative method [34]. Increasing the number of cycles was reported able to modify the stability of Aβ<sup>42</sup> CSF levels. However, about the exact numbers of cycles able to impact the levels, no real consensus was found between the different studies. If the Tau CSF levels seem to be unaffected by 3 or 6 freeze/thaw cycles [30, 12], the Aβ<sup>42</sup> CSF levels were found either stable after 3 cycles [30], either were significantly decreased after the third cycle [12]. In case of immunoassay analysis, it is logically recommended to limit the number of freeze/thaw cycles up to two as maximum [9].

Finally, the length of storage at -80°C does not seem to present a major influence on stability of CSF AD biomarkers, at least for 2 years [30] according to unpublished data from Blennow K. et al., referenced in the guideline published by Vanderstichele et al. [19]. Moreover, the levels of Aβ42 and T-tau but not Aβ40, remained stable up to 6 years [35]. In summary, we can conclude that CSF can be stored up to 2 years at -80°C as previously reported [19].
