|Home | About | Journals | Submit | Contact Us | Français|
There is a long history of research into body fluid biomarkers in neurodegenerative and neuroinflammatory diseases. However, only a few biomarkers in CSF are being used in clinical practice. One of the most critical factors in CSF biomarker research is the inadequate powering of studies because of the lack of sufficient samples that can be obtained in single-center studies. Therefore, collaboration between investigators is needed to establish large biobanks of well-defined samples. Standardized protocols for biobanking are a prerequisite to ensure that the statistical power gained by increasing the numbers of CSF samples is not compromised by preanalytical factors. Here, a consensus report on recommendations for CSF collection and biobanking is presented, formed by the BioMS-eu network for CSF biomarker research in multiple sclerosis. We focus on CSF collection procedures, preanalytical factors, and high-quality clinical and paraclinical information. The biobanking protocols are applicable for CSF biobanks for research targeting any neurologic disease.
There is a long history to the search for body fluid biomarkers in neurodegenerative and neuroinflammatory diseases, such as multiple sclerosis (MS). CSF has major advantages in the study of neurologic conditions, although sampling CSF is more invasive than sampling blood or urine.1 Because of its close proximity to the CNS, the CSF may more accurately reflect ongoing pathology of the brain, spinal cord, and meninges, and therefore may provide important and novel information.
Currently, the most frequently used CSF biomarker in MS is the detection of oligoclonal immunoglobulin G (IgG) bands or quantitative intrathecal IgG synthesis. Despite extensive research efforts, no other markers have been adopted into clinical practice in MS. Reviews on the state-of-the-art of biomarker research in MS have shown that the majority of studies are underpowered.2,3 One of the most critical is the lack of sufficient CSF samples that can be obtained by a single research center. Therefore, collaboration between investigators is needed.
Standardized collection protocols should be established to ensure that the statistical power gained by large numbers of samples is not compromised by preanalytical factors. Furthermore, standardization of collection protocols allows investigators to replicate studies with samples that match the initial pilot data.
Here, we provide protocols for the standardized collection, biobanking, and exchange of CSF samples. This is a consensus protocol obtained during meetings of the European network for biomarkers in MS, BioMS-eu, held in London in March 2007. Large differences were present between collection protocols (figure and table 1). In the discussions, we have sought a balance between practicality and scientific rationale. Particular attention has been focused on preanalytic procedures, because errors in the collection, storage, and exchange of biofluids account for 60% of total laboratory errors.4 Last, for optimal CSF research in MS, high-quality clinical and paraclinical data such as MRI are also needed. Such data will have great importance for the estimation of the prognostic value of a candidate marker.
We would like to stress that researchers should adhere to these protocols for optimal collaboration in the field of CSF biomarker research. We suggest using tables 2 and 3 as a checklist for CSF biomarker research and recommend that future studies of CSF biomarker take these issues into account. In discovery-based biomarker research, all these items should be considered carefully before initiating a study. Although some procedures may not be possible in everyday clinical practice and less stringent requirements may suffice for specific research questions, careful documentation of these issues is crucial to facilitate retrieval of appropriate samples dictated by specific study aims.
Importantly, the procedures for withdrawal and storage of CSF (table 2) are broadly applicable for any neurologic disease.
The CSF volume taken can influence the concentration of biomarkers. Most molecules and cell numbers have a rostrocaudal concentration gradient.5,6 If a small volume is taken, the CSF will reflect the composition of the lumbar dural sac, whereas large volumes may reflect the rostral spinal or even ventricular CSF. Therefore, if biomarker concentrations in a sample from a puncture of 2 mL are compared with that in a puncture of 15 mL, this can lead to erroneous results. Also, collecting different portions of the CSF for biobanking (e.g., initial and final volumes of the puncture) may introduce errors. Thus, a standard volume of CSF should be collected during lumbar puncture; the first 2 mL can be used for basic CSF analysis (item 33), and the remainder of the sample should be pooled before being divided into aliquot parts. At least, the procedure must be recorded. The volume of collected CSF does not correlate with the risk of post–lumbar puncture headache.7,8
Usually, diagnostic CSF is obtained by lumbar puncture. Because of the increasing gradient in protein concentration from ventricular to lumbar CSF,9 the site of CSF withdrawal must be recorded. When CSF is taken from other locations, such as the cervical cisterns or the lateral ventricles (e.g., ventricular drainage), this should be documented.
A traumatic tap causing blood contamination of CSF occurs in approximately 14% to 20% of standard lumbar punctures.10 For biomarkers that have high serum concentrations, such as coagulation factors, blood contamination can lead to false-positive results. In addition, blood proteins lead to suppressed matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF)/MS proteomics patterns in CSF. This suppression by blood proteins is, however, highly reduced after removal of the blood cells by centrifugation before initial freezing.11,12 Recording of erythrocyte count is essential to select CSF samples appropriate for these measurements. CSF samples with an erythrocyte count above 500/μL should not be used for biomarker studies.
There is no evidence that the type of lumbar puncture needle influences biomarker concentrations. However, atraumatic needles are best tolerated by patients and are associated with a lower risk for post–lumbar puncture headache, i.e., approximately 12% for a needle size of 20–22 gauge compared with approximately 70% for a needle size of 16–19 gauge.13,14
There are several reports showing that the type of collection tube influences biomarker outcomes, e.g., total tau proteins and amyloid-β peptides.15 Therefore, standardization is important. We propose to use polypropylene tubes, with their low protein binding potential, for collecting CSF. No additives should be used. Glass tubes should be avoided because of safety reasons for personnel.
For biomarkers that are influenced by circadian rhythms, time of withdrawal is important.16 Because it is often difficult to accomplish standardization of withdrawal time in everyday clinical practice, documentation is necessary to select the appropriate samples to minimize the effect of this variable.
It is important to collect matched serum and/or plasma samples for evaluation of CSF biomarkers because the concentration of the marker in blood often influences that in CSF.1 Further, serum/plasma pairs are essential to study the intrathecal origin of a biomarker and its CNS specificity. Furthermore, the presence of CNS markers in serum/plasma may aid in disease monitoring. Vacuum tubes that use ethylenediaminetetraacetic acid (in dried format) are preferred over those that use citrate (in solution), because if tubes containing a standard volume of citrate are filled incompletely, the final biomarker concentration is diluted artificially. Depending on the type of biomarkers and methods of study, we recommend collecting both serum and plasma17; for some methods, plasma is preferred over serum and vice versa. Serum/plasma samples should not be hemolytic.
Last, DNA collection expands the possibilities for studying the phenotypes and genotypes within individuals. A protocol for storage and handling of DNA can be found in appendix e-1 on the Neurology® Web site at www.neurology.org.
For CSF, there are no data available yet that support a preference for leaving the samples at room temperature or at 4°C until processing. For serum/plasma, it is more crucial. To avoid platelet activation,18 serum/plasma samples should be kept at room temperature before centrifugation. Therefore, processing at room temperature for both serum/plasma and CSF, including during and after spinning, is suitable for most studies. Relatively few systematic studies have been performed on this issue. We would recommend exploratory studies to define the effect of temperature on specific biomarkers.
We propose to adhere to a standardized spinning protocol of 400 g for 10 minutes at room temperature when fragile cells need to be preserved for RNA of cell isolation, and otherwise at 2,000 g. For serum, we propose to spin at 2,000 g for 10 minutes at room temperature. Standardization of spinning temperature and speed may be important for some biomarkers, although no studies have addressed these specific preanalytical variables for CSF. For plasma and serum, temperature of processing is known to be critical for specific biomarkers.19 After centrifugation, the supernatant must be divided into aliquots and stored immediately. If this is not done, the processing time should be documented.
Studies of the effects of preanalytical variables by MALDI-TOF/MS proteomics (proteins/peptides <20 kD) have shown that the time between sampling and storage is more crucial for specific serum proteins or peptides than for CSF.11,12,20 For CSF, it was observed that processing within 2 hours does not lead to artifactual results.11,12 For serum, it was observed that small differences in processing time (approximately 10–30 minutes) can result in different proteomics spectra.17 Some biomarkers, such as antibodies or specific cytokines, are not very sensitive to sampling and storage conditions.21 For practical reasons and in view of the standard of 30- to 60-minute clotting time for serum, we recommend a time delay of 1.5 hours (±30 minutes) for both matrices. When CSF cells are to be preserved, processing as soon as possible is advised because cell numbers decrease quickly. However, in most of the centers, processing of the body fluid samples within 1 hour is not common practice. Therefore, documentation of time of withdrawal and storage is required to select uniform samples. For newly discovered biomarkers, these preanalytical variables should be evaluated.
Because of the same rationale as for CSF withdrawal (item 5), we recommend that polypropylene tubes should be used for division into aliquots and storage. Furthermore, vials with screw caps should be used for a secure sealing.
Freeze/thaw cycles can influence biomarker concentrations.22 For example, 1-time freezing of CSF samples can lead to a highly significant loss of amyloid-β, which is decreased a further 20% after 3 more thawing cycles.23,24 By contrast, no effects on CSF proteome profiles obtained by MALDI-TOF/MS have been observed after up to 4 freeze/thaw cycles.12
In principle, repeated freezing/thawing of samples should be avoided, because data addressing this topic are available for only a few biomarkers and the response to freeze/thaw cycles of new biomarkers is not known. Thus, splitting the pooled sample into multiple small aliquots is optimal, and eventual freeze/thaw cycles should be recorded.
Small aliquot volumes are optimal to avoid freezing/thawing and to avoid waste of CSF. The proposed tube sizes are 0.25, 0.5, and 1 mL. Tubes should be filled up to 75% to prevent freeze-drying within the tube, which will affect the concentration of biomarkers, although it may only be a problem if the seal of the cryogenic tubes are not airtight. This issue has not been formally studied and is not referred to in related standard operating procedures.25
Unique codes are necessary to track samples and pair with clinical data. Ideally, bar codes should be used to facilitate searching, to aid in blinding the analysis, and to protect the privacy of patients. It is important to have center-unique codes, to track data retrospectively. Labels must be water and frost (−80°C) resistant.
Proteins may not be stable at −20°C for years. In one study, the effect of storing CSF at −20°C and −80°C on cystatin C, an abundant CSF protein, was investigated. Cleavage of this protein occurred in all samples stored at −20°C but not in samples stored at −80°C.26 Apart from the cystatin C truncation, changes in the low molecular weight polypeptide profile due to CSF sample storage at −20°C for 3 months seemed to be minimal.11,12 Oligoclonal bands in CSF may be recovered after several years of storage at −20°C, indicating a high stability of immunoglobulins.27 No data are available showing the benefit of storage of CSF or serum in liquid nitrogen. Because this is expensive and not practical for CSF biobanking, there is no basis yet to recommend storage in liquid nitrogen.
Taking these data together, we recommend that samples are stored at −80°C to ensure long-term stability of biomarkers.
To enable easy tracking and fast relocation of samples, storage information should include freezer location, freezer identification, and sample location within the freezer.
Freezers should be alarm controlled, and a sample rescue plan should be established and documented. All freezers must be registered in a freezer log file. Ideally, daily temperature logs should be available for all freezers. Aliquots of samples should be distributed among different freezers. Ideally, an empty backup freezer should be available.
Information on the age at sampling is needed to allow comparability to age-matched reference values, because many proteins show age-dependent changes, e.g., albumin or IgG.28 Ideally, date of birth and date of sampling are recorded (table 3). Gender must be provided because of variability of markers influenced by hormones.
Reference ranges of biomarkers can be influenced by the genetic status.29 For example, a recent study observed a higher IgG index in African Americans than in Caucasians, unrelated to socioeconomic status.30 Criteria for race and ethnicity are available via the Web site of the NIH.31
A goal of biomarker research is to identify surrogate endpoints for relevant disease characteristics such as clinical subtype, disease duration, disease activity, and progression. Therefore, it is crucial that diagnosis is made according to standard criteria32,33 and that the correct disease subtype at sampling is recorded. The date of first and second events should also be recorded to define the date of conversion from clinically isolated syndrome (CIS) to clinically definite MS.32,33
The following clinical subtypes (stages) of MS are commonly differentiated: CIS, relapsing-remitting MS, relapsing secondary progressive MS (SPMS), nonrelapsing SPMS, primary progressive MS, and progressive relapsing MS.34 The time point of conversion to SPMS may be helpful to distinguish between slow and rapidly progressive MS or to develop biomarkers to predict conversion to the SPMS phase.
We recommend using appropriate and well-validated outcome measures. Although Expanded Disability Status Scale (EDSS) scores and functional scores35 are not linear disease severity scores and do not necessarily reflect progression rate, inflammatory activity, and lesion load,36,37 they are important for stratification purposes and as a measure of severity. However, it is advised to include other clinical scales too, such as the Multiple Sclerosis Functional Composite (MSFC) and its subscores.38,39 The MSFC also accounts for cognitive functioning, which is affected in 30% to 70% of MS patients.40
Follow-up EDSS, functional system scores, and preferably MSFC and its subscores are helpful to determine disease progression rate, which is an important clinical endpoint required for evaluation of candidate biomarkers.
Numbers of relapses is an indicator for retrospective clinical activity that may influence biomarker status. Relapses should be defined according to the Schumacher criteria,41 and the dates should also be recorded to determine the time interval between the start of the last relapse and the time of sampling. It is important to relate number of relapses to a defined retrospective time window, preferably 2 years, to obtain comparable and reliable data on annual relapse frequency.
Because annual relapse rates are often used to describe the clinical activity and most MS therapies mainly affect the number of relapses, this is an important target variable for biomarker research.42,43 Ideally, dates of relapses should be recorded, which allows for studying whether a biomarker predicts early vs late relapses.
It is well known that commonly used drugs for treatment of MS, including immunomodulatory agents and methylprednisolone for treatment or prevention of relapses, have an influence on expression of biomarkers.44,45 Therefore, type and duration of treatment, also not related to MS therapy, should be documented in detail, preferably beginning at least 1 year before CSF collection.
To enable stratification of patients according to their CSF findings and to evaluate suitability of samples for further analysis, results of basic CSF analysis should be recorded. Primarily, the CSF profile serves for exclusion of other diseases. In addition, quantitative changes of immunologic markers are likely to occur, depending on disease stage, relapse activity, and medication. Inflammatory processes may influence the blood-CSF barrier function and thereby biomarker concentrations.1
Oligoclonal IgG bands are important to test the value of a new diagnostic biomarker. The sensitivity of oligoclonal IgG bands is strongly dependent on the method used. We strongly recommend isoelectric focusing followed by immunoblotting and staining for IgG.46,47 Preferably, the methods of all routine diagnostic procedures, including oligoclonal banding, should be documented.
This information should contain the following items: whether a brain and/or spinal cord MRI has been performed, the date the study was performed, and where it was performed (to access the source data if required). Preferably, the data set should be available for later analysis. It is questionable whether the number of lesions and total lesion load are reliably recorded with routine diagnostic studies, but these are valuable parameters for estimating disease activity. There is great variation between different imaging centers regarding which routine sequences are obtained (axial vs sagittal, T2 vs proton density vs fluid-attenuated inversion recovery techniques, slice thickness, and so on). A valuable and reliable parameter to collect is the number of gadolinium enhancing lesions, because the presence of gadolinium enhancing lesions reflects the inflammatory activity at the time of the scan and ideally also of the CSF or blood collection. In the context of studies on biomarkers for neurodegeneration and neuroprotection in MS, information regarding brain atrophy is valuable, although several outcome measures of atrophy exist in various centers. A common MRI protocol would be desirable.48,49
Longitudinal MRI scans would be desirable to monitor disease progression, and they may be possible in some centers. However, because of the lack of using standardized MRI protocols in routine investigations, regular follow-up MRIs cannot be recommended at present.
The mask on the database screen could be in the local language, but the underlying files will need to be in English. It is strongly suggested to use a commercially available program, if not a common database for networks such as BioMS-eu. The database should also adhere to standardized international units.
For collaboration in Biomarker research, the presence of good ethical protocols that comply with national and international ethical and other legal regulations is of utmost importance. Most, if not all, centers already have a good ethical protocol for CSF biobanking and biomarker research. The most important issues are that the protocol allows exchange of (coded or pseudoanonymized) samples and relevant patient information, and patient consent. To address the needs of investigators, samples should initially be pseudoanonymized, i.e., stored with a code that can be linked to subject identifiers so that clinical details could be updated after the sample has been collected. However, when samples are extracted from the biobank for the purposes of research, they should be released as anonymized samples, i.e., stripped of their link with personal identifiers. This is to protect the privacy of individuals participating in research.
Researchers may have to give notice of new research projects to their ethical office and make material transfer agreements. Because rules differ among centers and countries, it is the responsibility of the research centers to adapt their procedures according to local rules.
The lists provided in table 2 can be used as an easy checklist for CSF biobanking for any CNS disease, applicable during setup of the procedures and also as a checklist for recording sample characteristics. It is expected that these standardizations will pave the way for large biomarker studies and fruitful collaborations. Ultimately, these endeavors are to arrive at validated biomarker assays for diagnosis, prognosis, and treatment of CNS diseases and a potential to elucidate relevant disease mechanisms.
Other participants of the BioMS-eu meeting in London, March 2007, are acknowledged for their contribution to the discussions resulting in this guideline: H. Abderrahim, S. Al-Izki, D. Baker, R. Banks, Y. Ben-Schloma, A. Berthele, A. Bertolotto, R. Farrell, J. Furby, R. Gani, F. Gilli, S. Gnanapavan, B. Gomez, B. Greco, E. Hauben, E. Havrdova, T. Hayton, E. Iacobaeus, A. Jaber, N. Jafari, S. Jackson, C.R. Jimenez, R. Kapoor, G. Keir, J. Killestein, A. Kok, A. Konieczny, A. Kroksveen, V. Lampasona, R. Lindberg, A. Lockhart, C. Luk, C. Maggiore, M. Mauritzio, A. Millonig, H. Parkes, T. Plitz, D. Sadovnik, A. Sala, R. Sachse, K. Schmierer, K. Smith, S. Suessmith, P. Thompson, H.L. Weiner, B. Wilson, D. Wright, J. Zajicek, and P. Zaratin.
Dr. Teunissen and Dr. Giovannoni report no disclosures. Dr. Bennett has received honoraria from the NIH and the Armed Services Institute; has served as a consultant to EMD Serono, Inc., Biogen Idec, and Teva Pharmaceutical Industries Ltd.; has served on speakers' bureaus for EMD Serono, Inc. and Pfizer Inc; and receives research support from EMD Serono, Inc., the NIH/NEI [R03EY014573 (PI)], and the National Multiple Sclerosis Society. Dr. Berven reports no disclosures. Dr. Brundin has received funding for travel and royalties for publishing from the Karolinska Institutet; received honoraria from Bayer Schering Pharma; and receives research support from the Torsten and Ragnar Söderberg Foundation, the Harvard Foundation, and the Swedish Medical Research Council. Dr. Comabella and Dr. Franciotta report no disclosures. Dr. Frederiksen serves on a scientific advisory board for Biogen Idec; and served on a speakers' bureau for Merck Serono. Dr. Fleming receives research support from the National Multiple Sclerosis Society. Dr. Furlan has served on a scientific advisory board for UCB; has received funding for travel from Biogen-Dompè AG; has received speaker honoraria from Sanofi-aventis and Biogen Idec; and receives research support from Sanofi-aventis, Merck Serono, Novartis, Teva Pharmaceutical Industries Ltd., the Italian Ministry of Health, the Italian Ministry of Research and Education, and the Italian Foundation for Multiple Sclerosis (FISM). Dr. Hintzen reports no disclosures. Dr. Hughes receives royalties from publishing Complete Data Interpretation for MRCP (Churchill Livingstone Press, 2001); is employed as Director of Medical Research for Biogen Idec; and receives stock and stock options in Biogen Idec as part of his compensation package. Dr. Johnson has received funding for travel and speaker honoraria from Schering-Plough Corp., Biogen Idec, Merck Serono, and Teva Pharmaceutical Industries Ltd.; has received research support from Leeds Teaching Hospitals Special trustees; holds stock in GlaxoSmithKline; and has a medicolegal practice in the clinical negligence field and receives fees for expert neurological opinions. Dr. Krasulova has received funding for travel for lectures or educational activities not funded by industry. Dr. Kuhle, Dr. Magnone, Dr. Petzold, Dr. Rajda, and Dr. Rejdak report no disclosures. Ms. Schmidt has served as a consultant to the Robert Wood Johnson Foundation. Dr. van Pesch has received funding for travel from Bayer Schering Pharma and Merck Serono; and received FRSM funding from the Catholic University of Louvain. Dr. Waubant reports no disclosures. Dr. Wolf is a salaried employee of UCB S.A., Belgium; and holds stock in PAION AG and Sanofi-aventis. Dr. Tumani has received funding for travel from Bayer Schering Pharma, Novartis, Biogen Idec, Merck Serono, and Teva Pharmaceutical Industries Ltd.; has received speaker and/or consulting honoraria from Bayer Schering Pharma, Novartis, Biogen Idec, Merck Serono, and Teva Pharmaceutical Industries Ltd.; and receives research support from University of Ulm, Biogen Idec, Bayer Schering Pharma, Merck Serono, Roche, and Genzyme Virotech. Dr. Hemmer has served on scientific advisory boards for Novartis, Bayer Schering Pharma, and Genzyme Corporation; has received funding for travel from Bayer Schering Pharma, Novartis, Biogen Idec, Merck Serono, Teva Pharmaceutical Industries Ltd., Roche, and Genentech, Inc.; serves on the international advisory board of Archives of Neurology; has received speaker honoraria from Bayer Schering Pharma, Novartis, Biogen Idec, Merck Serono, and Teva Pharmaceutical Industries Ltd.; serves as a consultant to Genentech, Inc., Micromet, Inc., Roche, and Novartis; and receives research support from Biogen Idec, Bayer Schering Pharma, and Merck Serono. Dr. Deisenhammer has received funding for travel from Biogen Idec, Bayer Schering Pharma, Merck Serono, and Sanofi-aventis; serves as Associate Editor of Journal of Laboratory Medicine; has received speaker and/or consulting honoraria from Bayer Schering Pharma, Biogen Idec, Merck Serono, and Sanofi-aventis; and receives research support from Bayer Schering Pharma, Biogen Idec, Merck Serono, and the European Union (NABINMS project: www.nabinms.eu).
Address correspondence and reprint requests to Dr. Charlotte E. Teunissen, Neurology Laboratory Clinical Chemistry, VU University Medical Center, FdG, PO Box 7057, 1007 MB Amsterdam, the Netherlands firstname.lastname@example.org.
Supplemental data at www.neurology.org.
Disclosure: Author disclosures are provided at the end of the article.
Received November 17, 2008. Accepted in final form September 17, 2009.