Handbook of Clinical Neurology, Vol. 122 (3rd series) Multiple Sclerosis and Related Disorders D.S. Goodin, Editor © 2014 Elsevier B.V. All rights reserved
Cerebrospinal fluid analysis GAVIN GIOVANNONI* Centre for Neuroscience and Trauma, Blizard Institute, Barts and The London School of Medicine and Dentistry, London, UK
INTRODUCTION The cerebrospinal fluid (CSF) is the accessible body fluid that is closest to the pathology of multiple sclerosis (MS) and is therefore thought of as a “liquid biopsy.” CSF is used for performing diagnostic and exploratory biomarker studies in MS, which have provided important insights to the pathogenesis of MS. A biomarker is generally defined “as a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention” (NIH Biomarkers Definitions Working Group, 2001; Floyd and McShane, 2004). CSF biomarkers are studied in MS for several reasons (Table 30.1). The best example of this is the diagnostic and prognostic value of measuring the intrathecal synthesis of oligoclonal immunoglobulin G (IgG) in MS (Andersson et al., 1994; Freedman et al., 2005). Currently, the evaluation of the CSF is being used less and less in the diagnosis of MS, despite its obvious and well-established utility. However, the potential of CSF studies to help us understand and monitor the pathogenesis of MS has yet to be fully appreciated and explored. As a prelude to a detailed and comprehensive discussion on MS CSF biomarkers relevant background information in relation to CSF, CSF analysis and MS in general will be presented.
Brief considerations with regard to cerebrospinal fluid analysis in MS CSF is an admixture of fluids derived predominantly from the choroid plexus and interstitial fluid of the brain, spinal cord, meninges, and blood vessels which traverse the subarachnoid space. The total volume of CSF in an adult is approximately 125–150 mL. This CSF volume turns over approximately four times a
day; the daily volume of CSF produced is approximately 600 mL. CSF production and diurnal aqueductal flow rates of ventricular fluid can vary by as much as a factor of 3.5, with minimum CSF production occurring around 18 h00 (12 7 mL/hour) and maximum production at approximately 02 h00 (42 2 mL/hour) (Nilsson et al., 1992). The blood–CSF barrier consists of several anatomic structures that act as a physical barrier to the diffusion and filtration of proteins from the blood into the CSF. The integrity of these barriers and the flow of CSF determine the qualitative and quantitative content of the total CSF protein (Thompson, 1988; Reiber, 1994). The CSF constituents are not uniform and depend on the site from which the CSF is sampled. There is a rostrocaudal concentration gradient for total protein along the neuraxis, with the lowest concentrations in ventricular fluid and the highest concentrations in the fluid from the lumbosacral thecal sac (Thompson, 1988). The latter is the site from which most CSF is taken for analysis. For example, a significant decrease in the albumin quotient (Qalb) (see formulae below) is observed in successive aliquots of CSF obtained from spinal or lumbar tap (Blennow et al., 1994). In comparison, proteins that are synthesized intrathecally decrease (rather than increase) in concentration rostrocaudally. Finally proteins that are synthesized in both the systemic and intrathecal compartments have a relatively uniform CSF concentration, although this varies depending on their rate of synthesis and molecular size (Thompson, 1988). The CSF protein concentration is high in neonates, decreases gradually during the first year of life, with low levels being maintained throughout childhood, and then increases with age in adulthood (Eeg-Olofsson et al., 1981; Statz and Felgenhauer, 1983). In addition to CSF total protein the ratio of the CSF to serum
*Correspondence to: Professor Gavin Giovannoni, Centre for Neuroscience and Trauma, Blizard Institute, Barts and The London School of Medicine and Dentistry, Blizard Building, 4 Newark Street, London E1 2AT, UK. Tel: þ44-20-7882-8954, E-mail: [email protected]
Table 30.1 Reasons for performing cerebrospinal fluid biomarker studies in multiple sclerosis Diagnostic testing Positive and negative predictive testing Pathogenic studies Immunology Etiology Disease progression and recovery Disease heterogeneity Monitoring of disease processes Defining prognosis Monitoring effects of therapeutic interventions Monitoring for complications of therapeutic interventions
albumin concentration, or Qalb, is also used to evaluate blood–CSF barrier integrity (Andersson et al., 1994; Freedman et al., 2005). Albumin quotient ðQalb Þ ¼ CSF albumin ðCSF alb Þ= serum albumin ðserumalb Þ 100 The Qalb is not influenced by the rate of intrathecal protein synthesis as the CSF level is corrected for by the plasma concentration of albumin. Whether or not Qalb is a better indicator of the blood–CSF barrier integrity, and is more useful clinically than total CSF protein, remains a moot point. In practice, however, they provide similar information. Physical activity has an important influence on CSF protein concentrations; subjects who are supine for prolonged periods have higher CSF protein concentrations than active subjects (Seyfert et al., 2002). It is important to consider this when interpreting results of CSF analyses from subjects who have been supine for prolonged periods of time, typically more than 24 hours. Despite this, levels of greater than 1 g/L should not be ascribed to the effects of being supine and should be considered pathologic and interpreted in the clinical context of the patient. It would be very unusual to have a total protein concentration of greater than 1 g/L in subjects with MS; such a high level should raise the possibility of the presence of comorbidity or an alternative diagnosis. As the lumbar subarachnoid space is a cul-de-sac, CSF obtained from lumbar punctures does not necessarily provide accurate information on inflammatory or pathologic events in the brain which occur distal to the CSF outflow foramina of the fourth ventricle. Pathology in relation to the surface of the brainstem, cranial nerves, and cerebral hemispheres may not be as well reflected in the spinal fluid as pathology in the spinal cord. In addition, the extracellular space of intraparenchymal lesions may not necessarily communicate with all parts of the free CSF space (Jacobi et al., 1986).
A definition of multiple sclerosis MS is defined pathologically (Flier and de Vries Robbe, 1999) as an inflammatory demyelinating disease of the central nervous system (CNS) characterized by multifocal areas of demyelination and variable degrees of axonal loss and gliosis (Lassmann et al., 1998). However, this conventional definition is seldom used clinically and we therefore have to depend on a pretheoretic definition (Flier and de Vries Robbe, 1999) consisting of a set of polythetic clinical criteria which define the disease. The underlying principles of these criteria involve demonstrating involvement of more than one white-matter structure in the CNS (anatomic dissemination), which are separated in time (temporal dissemination). Despite appearing non-specific, the principles underlying these criteria are reasonably specific in identifying MS during life (Engell, 1988). The first set of polythetic criteria were clinical, formulated by Schumacher and colleagues in the 1960s (Schumacher et al., 1965). The Schumacher criteria were modified in 1982 (Poser et al., 1983) to include evoked potentials and CSF examination and subsequently in 2000 (McDonald et al., 2001), 2005 (Polman et al., 2005), and 2010 (Polman et al., 2011) to include and refine the use of magnetic resonance imaging (MRI) in contributing towards a diagnosis of MS. How we define MS has implications that go beyond day-to-day clinical practice and impacts many factors, including prognosis, clinical trial design, and personal issues for the individuals involved; for example, a diagnosis of MS makes it very difficult for individuals to obtain personal insurance cover at reasonable rates. By shifting patients previously diagnosed as having clinically isolated syndromes (CIS) compatible with MS into the category of MS alters the natural history of both conditions; i.e., the overall prognosis of CIS and relapsingremitting MS (RRMS) improves (Sormani et al., 2008). This phenomenon is called the Will Rogers effect and has already had an impact on clinical trials using the new criteria; contemporary cohorts of trial patients appear to have a more benign course in comparison to cohorts recruited using the Poser criteria (Sormani et al., 2008, 2009). Extending this further, it seems appropriate to consider, also, CIS and asymptomatic MS as part of the diagnostic spectrum of MS. The latter situation typically arises when white-matter lesions, compatible with demyelination, are detected incidentally on MRI performed for another reason. The term radiologically isolated syndrome has been used to define this syndrome (Lebrun et al., 2009; Okuda et al., 2009). One way to improve the sensitivity and specificity of the diagnostic criteria in these emerging situations would be to include CSF analysis as part of diagnostic criteria. Unfortunately, the opposite has occurred; in the era of MRI
CEREBROSPINAL FLUID ANALYSIS CSF analysis has become less important diagnostically. A good illustration of this is primary progressive (PP) MS in which the definition of dissemination in time and space is more difficult. In the original McDonald criteria PPMS could not be diagnosed without an abnormal CSF examination, defined as the intrathecal synthesis of IgG (McDonald et al., 2001). However, in the subsequent 2005 Polman revision of McDonald criteria an abnormal CSF was not essential for the diagnosis of PPMS (Polman et al., 2005). At first glance this subtle and relatively minor change may seem innocuous. However, in the glatiramer acetate PPMS phase III study (PROMISE study) (Wolinsky, 2004), subjects with abnormal CSF had a worse prognosis than subjects with a normal CSF (Wolinsky, 2004). The latter observation suggests that CSF-negative PPMS may have a different disease course compared to subjects with CSF-positive PPMS. Hopefully, this observation will be reflected in the future renditions of the international diagnostic criteria.
Key pathologic processes in MS Figure 30.1 summarizes the key pathogenic processes that are thought to occur in MS. Autoimmune-mediated inflammation is considered by many to be the key pathogenic process. The role of inflammation is complex and has been questioned based on the observation that aggressive immunotherapy fails to prevent disease progression in the progressive phase of the disease (Coles et al., 2006; Samijn et al., 2006). The inflammation predominates in a perivascular distribution and consists
predominantly of macrophages, activated microglia, T and B lymphocytes, and occasional plasma cells. The B cells and plasma cells are responsible for the production of oligoclonal Ig within the CNS. The detection of this abnormal IgG in the CSF is one of the paraclinical criteria supporting the diagnosis of MS. Despite the discovery of CSF oligoclonal immunoglobulins in 1960 (Lowenthal et al., 1960), their significance and role in the pathogenesis of MS remain speculative despite efforts to identify a pathogenic role in the course of MS.
HISTORY OF CSFANALYSIS IN RELATION TO MS The earliest report of a systematic CSF study in MS dates back to the 1934 paper by Houstin-Merritt published in Brain, in which he describes the opening CSF pressure, cell count, protein content, and colloidal gold reactions in 100 cases of MS. Elvin Kabat showed that CSF differed from serum in that prealbumin, which he termed protein X, was specific to spinal fluid (Kabat et al., 1948) (Fig. 30.2). For these analyses, Kabat required up to 70 mL of CSF, hence he used mainly pooled CSF samples. The next major step in CSF analysis by Kabat was the quantitative immunochemical precipitation of albumin and IgG, thereby reducing the volume of CSF required to about 2 mL. The latter heralded the beginning of routine CSF analysis and allowed large numbers of patients to be studied, which led to practical aids for the diagnosis of individual patients.
CSF biomarkers Macrophages & Microglia: sCD14, ferritin, neopterin B cells: OCBs, FLCs
CSF biomarkers S-100b, GFAP
Inflammation CSF biomarkers NF, Tau, 14-3-3, enolase
axonal damage (conduction block)
axonal & neuronal loss (reduced reserve)
myelin damage (demyelination) CSF biomarkers MBP
axonal recovery (remyelination)
CSF biomarkers Gap43 & NCAM
central adaptation (plasticity)
Fig. 30.1. Key pathogenic changes. Multiple sclerosis is an inflammatory demyelinating disease of the central nervous system with relative axonal sparing. Functional recovery occurs by remyelination and/or axonal plasticity in which sodium channels spread into demyelinated internodes to restore conduction, converting a myelinated into a non-myelinated axon. Central adaptation, for example, cortical plasticity, also occurs and these processes probably contribute to a significant degree of recovery. The safety factor of conduction in these fibers is reduced, making them liable to conduction block. This contributes to fatigue, heat sensitivity, and intermittent symptoms. Variable degrees of neuroaxonal loss and gliosis occur, presumably as a consequence of inflammation. These latter processes account for permanent neurologic deficit or impairment. CSF, cerebrospinal fluid; OCBs, oligoclonal immunoglobulin G bands; FLCs, GFAP, glial fibrillary acidic protein; MBP, myelin basic protein; NCAM, neural cell adhesion molecule.
G. GIOVANNONI A
Case number 661981–Idiopathic grand mal (normal) 647635–Anxiety state (normal) 652282–Lymphopathia venereum 661095–Multiple myeloma WH 138–Multiple sclerosis 655751–Neurosyphilis 635202–Diabetic neuritis 652371–Left frontal cystic astrocytoma
K A C E G I J K L
L Spinal fluid Spinal fluid Spinal fluid Spinal fluid Spinal fluid Spinal fluid Spinal fluid Cyst fluid
B Serum D Serum F Serum H Serum
Fig. 30.2. Electrophoretic patterns. Protein electrophoretic patterns identified in concentrated cerebrospinal fluid and their relationship to serum proteins. Case WH 138 has multiple sclerosis and you can see the increase in g-globulins, similar to case 655751 with neurosyphilis. (Republished with permission of American Society for Clinical Investigation, from An electrophoretic study of the protein components in cerebrospinal fluid and their relationship to the serum proteins, Elvin A. Kabat, Dan H. Moore, Harold Landow, Vol 21(5), 1942; permission conveyed through Copyright Clearance Center, Inc.)
Lowenthal et al. (1960) were the first to demonstrate bands in the gamma region in subjects with MS. At the same time Tourtellotte and Parker (1966, 1967) showed that MS plaques contained higher amounts of IgG than could be found in the corresponding CSF and that these levels, in turn, we higher than in parallel serum. These two observations lead to the seminal notion of local synthesis of IgG within the brain of subjects with MS. Tourtellotte et al. (1971) also used the Laurell rocket technique to estimate both albumin and IgG on the same agar dish and were among the first to recognize the importance of immunofixation for the study of CSF proteins in MS (Cawley et al., 1976). Klaus Felgenhauer (1970) showed that high-molecular-weight haptoglobin oligomers were not present in normal CSF. This was later used as an adjunct in the diagnosis of MS (Chamoun et al.,
2001). Hans Link (1967) then showed that the bands in the gamma region on CSF electrophoresis were IgG. Magnhild Sandberg-Wollheim (1974) showed that CSF lymphocytes from patients with MS synthesized radiolabeled proteins which migrated in the gamma region after incubation of the CSF cells in tissue culture. Christian Laterre et al. (1970) undertook an extensive survey of abnormalities of CSF proteins in several diseases, including MS, giving the differential diagnosis for the presence of oligoclonal bands; it was Laterre who coined the term “oligoclonal.”
DIAGNOSTIC TESTING The detection of intrathecal synthesis of oligoclonal immunoglobulins can be very helpful in making a
CEREBROSPINAL FLUID ANALYSIS diagnosis of MS. The Polman or “revised McDonald” criteria (Polman et al., 2005) were formulated to simplify the McDonald criteria (McDonald et al., 2001), whilst maintaining diagnostic sensitivity and specificity. The McDonald and Polman criteria specifically focused on the use of MRI to demonstrate dissemination of the disease process in time and space. With regard to CSF analysis, the Polman criteria differ from McDonald criteria in that they do not require a positive CSF analysis as an absolute requirement for a diagnosis of PPMS (Polman et al., 2005, 2011). The latter change is based on adopting the findings of a large multicenter PPMS study (Wolinsky, 2003). Unfortunately, the methods of CSF analysis performed in this study were not standardized and allowed the use of quantitative indices to determine intrathecal synthesis of IgG. Quantitative indices are less sensitive than the recommended qualitative methods (Freedman et al., 2005). This may explain why only 79% of subjects with PPMS had intrathecal synthesis of IgG in this study (Wolinsky, 2003). Had the earlier criteria for diagnosing PPMS been used (Thompson et al., 2000; McDonald et al., 2001), this figure would have needed to be 100%. With isoelectric focusing (IEF) and immunofixation, 95–98% of subjects with clinically definite MS (Poser et al., 1983) have demonstrable intrathecal synthesis of oligoclonal IgG (Andersson et al., 1994; Freedman et al., 2005). The intrathecal synthesis of oligoclonal IgG is therefore an almost invariable feature of MS and remains a useful diagnostic aid, particularly where there is some doubt about the clinical diagnosis. A negative result suggests the possibility of an alternative diagnosis whereas a positive result strongly supports a diagnosis of MS. The detection of increased free immunoglobulin light chains has also been used to improve the diagnostic sensitivity of IEF (see section on quantitative methods, below).
B-cell immunology The humoral response to a specific antigenic challenge is a complex process involving antigen recognition and the initial production of IgM. This IgM-producing step may be T-cell-independent. IgM production is then followed by a process of immunoglobulin isotype switching and affinity maturation, which require T-cell help. T-cell help involves cross-talk between T and B cells via cell surface molecules and cytokines. The production of immunoglobulins can be divided into the stages of activation, proliferation, and differentiation of B cells. These steps requires the antigen-specific interactions of T and B cells via the T-cell receptor (TCR)–major histocompatibility complex (MHC) II receptor complex, and important non-antigen-specific costimulatory signals via other receptor ligand pairs, like LFA1–ICAM1, LFA3–CD2, and CD28-B7. This results in T-cell activation and the
production of the T-cell cytokines which induce B-cell proliferation and differentiation with isotype switching and affinity maturation. Activation of B cells occurs via antigen-specific surface IgM receptors in the presence of the B-cell-activating cytokines (see Meinl et al. (2006) for review). The antigen bound to the IgM receptors is then internalized, processed, and presented in the context of MHC class II molecules. Proliferation needs antigen-specific T-cell help in the form of cytokines interleukin-2 (IL-2), IL-4, IL-5, B-cell-activating factor (BAFF), and APRIL (a proliferation-inducing ligand). The tumor necrosis factor (TNF) superfamily receptors that bind BAFF and APRIL B-cell maturation antigen (BCMA) and TACI (TNFR homolog transmembrane activator and Ca2þ modulator and CAML interactor), are expressed almost exclusively by cells of the B-cell lineage. Depending on the ligand, this then results in signal transduction that involves activation of NF-kB, caspases, JNK, and ERK. BAFF and APRIL promote the differentiation and proliferation of B cells and augment immunoglobulin production and upregulate the expression of several B-cell effector molecules. Interestingly, a clinical trial of atacicept, a human recombinant fusion protein that comprises the binding portion of a receptor for both BAFF and APRIL, selectively blocking the late stages of B-cell and plasma cell development, whilst sparing B-cell progenitors and memory cells, was stopped prematurely due to an unexpected increase in inflammatory activity in RRMS. The type of T-cell help influences the immunoglobulin isotype and subclass switching that occurs. In general the Th2-like cytokines result in the production of IgE, IgG2, and IgG4, whereas Th1-like cytokines IL-12 and interferon-g induce IgG1 and IgG3 production. The net result of these processes is the formation of mature clones of plasma cells, each producing a specific antibody. This antibody is characterized by the combination of a single class of light and heavy chain having unique variable and hypervariable regions defining the molecule’s antigen specificity. The microenvironment that exists in the inflamed CNS supports the survival of long-lived plasma cells and explains why the oligoclonal IgG bands (OCBs) persist in the CSF of subjects with MS (Meinl et al., 2006).
Antibody specificity and affinity Antibody specificity can either be viewed as a measure of the goodness of fit between the antibody-combining site (paratope) and the corresponding antigenic determinant (epitope), or the ability of the antibody to discriminate between similar or even dissimilar antigens (Candler et al., 2006). This binding specificity is associated with particular functional consequences, which are due to the properties conferred on the immunoglobulin
molecule by its non-antigen-binding sites, for example, the ability to activate complement. The biologic functions are isotype- and subclass-dependent. In comparison with specificity, affinity of an antibody is a measure of the strength of the binding between antibody and antigen, such that a low-affinity antibody binds weakly and high-affinity antibody binds firmly. The process of affinity maturation preferentially selects for survival of B cells and subsequently plasma cells which produce high-affinity antibodies rather than those producing low-affinity antibodies. High-affinity antibodies to a specific antigen are a good indicator that the antigen is directly involved in driving the humoral response (Giovannoni, 2006). Conversely, the presence of lowaffinity antibodies usually represents cross-reactivity or an anamnestic response.
Type of humoral response Using electrophoresis it is possible to classify a humoral response according to the number of antibody clones produced: a monoclonal antibody results from a single plasma cell clone, oligoclonal antibodies due to several clones, and a polyclonal antibody response representing a general increase in immunoglobulin production with no specific discernible clones noted above the background (Freedman et al., 2005).
MONOCLONAL RESPONSE A monoclonal response can represent the initial stage of an oligoclonal response, before the other antibody clones become visible or, more commonly, a single abnormal clone of plasma cells associated with a B-cell or plasma cell dyscrasia (Davies et al., 2003; Trip et al., 2003; Freedman et al., 2005). Methods such as IEF, which are more sensitive than standard agarose gel electrophoresis, often detect bands that are invisible by methods which have a lower resolution. Monoclonal gammopathies detected by IEF, but not by agarose electrophoresis, have not been studied longitudinally. However, they should be followed longitudinally in the event that they represent the early stages of a plasma cell dyscrasia.
OLIGOCLONAL RESPONSE An oligoclonal response represents an immunologic response to a specific antigen or set of antigens, and is associated with infections and several putative autoimmune and inflammatory conditions (Table 30.2). The oligoclonal response to a set of known antigens can be used as a specific diagnostic tool, e.g., viral encephalitis (Morris et al., 2006). When the eliciting antigens are unknown, as in MS, for example, the presence of an oligoclonal pattern is presumed to be non-specific,
Table 30.2 Inflammatory diseases of the central nervous system associated with the detection of cerebrospinal fluid oligoclonal immunoglobulin G bands Approximate incidence of oligoclonal bands (%)
Disorder Multiple sclerosis Autoimmune Neuro-SLE Neuro-Behc¸et’s Neuro-sarcoid Harada’s meningitisuveitis Infectious Acute viral encephalitis (