Introduction

Multiple sclerosis (MS) is a chronic inflammatory disorder of the central nervous system (CNS) of putative autoimmune aetiology. The diagnosis of MS is hampered by the lack of a specific laboratory marker. Intrathecal production of IgG as detected by calculation of the IgG cerebrospinal fluid (CSF)/serum ratio (QIgG) or by testing for CSF-restricted oligoclonal bands (OCBs) is considered a hallmark of MS. However, OCBs and an elevated IgG ratio are also found in a plethora of other autoimmune and infectious CNS disorders and are thus of low specificity for MS. It has been known for many decades that the intrathecal IgG response in patients with MS comprises antibodies to a broad panel of neurotropic viruses. As antibodies to measles virus (M), rubella virus (R) and varicella zoster virus (Z) are its most frequent constituents, it has been referred to as the MRZ reaction (MRZR) [2, 17, 18, 30, 49, 60]. That anti-viral antibody response is not thought to be directly involved in the pathogenic process but has been suggested to reflect non-specific bystander activation of B cells or, more recently, to be the result of nonsense activity of immortalised B cell clones.

The potential diagnostic relevance of MRZR testing as a ‘rule-in’ marker of MS, as opposed to total IgG OCB or QIgG testing, which are rather ‘rule-out’ markers, was pointed out in a European consensus statement on CSF diagnostics [3]. MRZR testing in patients with suspected MS is also been recommended by the German Society for Cerebrospinal Fluid Diagnostics and Clinical Neurochemistry in its current diagnostic guidelines [53]. Finally, the potential diagnostic relevance of MRZR testing has been stressed by a panel of experts on the occasion of the latest critical revision of the diagnostic criteria for MS [72]. However, no systematic review of the existing literature on the specificity and sensitivity of MRZR testing in MS exists so far.

For this study, we carried out a structured review of the entire existing English-, German-, and Spanish-language literature on MRZR. In addition, we evaluated MRZR in a cohort of patients with MS and other neurological disorders as a proof of principle.

Methods

MRZR testing

Matched serum and CSF samples from 43 unselected patients with relapsing-remitting MS (RRMS) (n = 9) or other CNS disorders (n = 34) were tested for MRZR as described before [30, 32]. None of these patients had been tested for MRZR before. The diagnoses in the control group included, among others, AQP4-IgG-positive neuromyelitis optica spectrum disorders (NMOSD), neuroborreliosis, neuro-lupus, neuro-Behçet, CNS vasculitis, migraine, trigeminal neuralgia and depression. The demographic and clinical features of all patients are summarised in Table 1. Behcet’s disease was diagnosed according to international consensus criteria [1]. All patients diagnosed with neuroborreliosis had an increased Borrelia burgdorferi-specific IgG antibody index (AI), and most had an increased Borrelia burgdorferi-specific IgM AI (Table 1). Systemic lupus erythematosus (SLE) was diagnosed according to the American College of Rheumatology criteria [24, 71]. Virus-specific antibody levels in CSF and serum were determined using a commercially available enzyme-linked immunosorbent assay (Siemens Healthcare/Dade Behring, Germany) according to the manufacturer’s instructions. Total IgG and total albumin concentrations in CSF and serum were determined nephelometrically (BN ProSpec, Siemens Healthcare/Dade Behring, Germany). The intrathecal synthesis of antibodies to M, R and Z was detected by calculation of the corresponding virus-specific AIs: AI = Q IgG[spec]/Q IgG[total], if Q IgG[total] < Q lim, and AI = Q IgG[spec]/Q lim, if Q IgG[total] > Q lim, with Q IgG[spec] = IgGspec[CSF]/IgGspec[serum], and Q IgG[total] = IgGtotal[CSF]/IgGtotal[serum]) [60]. The upper reference range of Q IgG, Q lim, was calculated according to Reiber’s formula [57]:

Table 1 Patient characteristics, antibody indices and MRZR results from nine patients with multiple sclerosis, five patients with neuro-Behçet, nine patients with neuro-borreliosis, five patients with systemic lupus erythematosus and CNS involvement, two patients with vasculitis and five patients with non-inflammatory CNS diseases
Full size table
$$Q_{{{ \lim }({\text{IgG}})}} = 0. 9 3\sqrt {\left( {Q_{\text{Alb}} } \right)^{ 2} + { 6 } \times { 1}0^{ - 6} } - 1. 7 { } \times { 1}0^{ - 3}$$

AI values >1.5 were considered to be indicative of intrathecal IgG production against the respective pathogen [60]. All samples were stored at −80 °C until testing. The study was approved by the institutional review board of each participating centre, and all patients gave written informed consent. If no consent could be obtained retrospectively, samples were tested in a strictly anonymised fashion as requested by the institutional review board of the University of Heidelberg. All samples were tested as part of a larger project on the differential laboratory diagnosis of MS, NMOSD and related disorders.

Review criteria

References were identified by searches of the databases of the National Library of Medicine (http://www.ncbi.nlm.nih.gov/pubmed) and of Thompson Reuters® (http://www.webofknowledge.com) for articles published between 1985 (i.e. the year in which Felgenhauer’s seminal work “Cerebrospinal fluid virus antibodies. A diagnostic indicator for multiple sclerosis?” [18] appeared) and September 2016 using the following search expression: “MRZ reaction” OR (IgG OR antibody OR antibodies OR IgG) AND (CSF OR “cerebrospinal fluid” OR intrathecal) AND ((measles AND rubella AND varicella) OR (“neurotropic viruses” OR “neurotropic virus”)). All studies that investigated the intrathecal production of antibodies to measles virus, rubella virus and varicella zoster virus and which contained information on the proportion of patients with a positive MRZR were considered eligible for this review. A positive MRZR was defined as intrathecal synthesis of antibodies against at least two of the three viral species defining the MRZ spectrum (measles, rubella, varicella zoster). Accordingly, studies that either tested for only one or two of the three reactivities or reported only on the frequency of each single antibody reactivity in the total cohort but not on the proportion of patients with a polyspecific (i.e. bi- or trispecific) reaction were excluded (Fig. 1), as were studies that tested for MRZR exclusively in preselected OCB-negative MS subgroups [5, 69]. To reduce the risk of publication bias, a search of the Thomson Reuters ® Web of Knowledge database of meeting abstracts was performed using the same search expression as stated above.

Fig. 1
figure 1

Data retrieval. Using the search expression given in the methods section, 65 publications were identified, 30 of which met the inclusion criteria specified in the methods section

Full size image

Statistical analysis

Sensitivity was calculated as true positives/(true positives + false negatives), specificity as true negatives/(true negatives + false positives). The positive likelihood ratio (pLR) was calculated as (true positives/(true positives + false negatives))/(1 − (true negatives/(true negatives + false positives))), the negative likelihood ratio (nLR) as (1 − (true positives/(true positives + false negatives)))/(true negatives/(true negatives + false positives)). Fisher’s exact test (two-tailed) was used to analyse contingency tables. The Mann–Whitney U test was used to test for significant differences in median AI values between groups. Ninety-five per cent confidence intervals were calculated for all test parameters evaluated.

Results

MRZR in the study cohort

MRZR as defined by a combination of at least two positive AIs was present in 7/9 (78%) patients with RRMS, but only in 1/34 (3%) patients with other CNS disorders (p < 0.00001) (Table 1). The only MRZR-positive control patient had been diagnosed with neuroborreliosis and showed a bispecific intrathecal antibody reaction to measles virus (AI 3.16) and varicella zoster virus (AI 4.22). MRZR was negative in all patients with neuro-Behçet and in all patients with neuro-lupus. One patient with neuro-Behçet and one with neuroborreliosis showed a monospecific intrathecal reaction to varicella zoster virus (2.05 and 4.81, respectively) and one patient with neuro-SLE showed a borderline reaction to rubella virus (1.56). Median AI values for measles differed significantly between patients with MS and those with CNS disorders other than MS [measles virus-specific AI: median 5.45, range (1st–9th percentile) 2.09–8.09, vs. 0.94, range 0.71–1.51, p < 0.0001; rubella virus-specific AI: median 4.1, range 0.78–5.78, vs 0.92, range 0.77–1.4, p < 0.006; varicella zoster virus-specific AI: median 4.0, range 0.68–6.06, vs. 0.82, range 0.72–1.43, p < 0.02] (Table 1; Fig. 2).

Fig. 2
figure 2

Antibody indices for measles (M), rubella (R) and varicella zoster (Z) virus in patients with MS (n = 9) and patients with other neurological disorders (n = 34). The boundary of the box closest to zero indicates the 25th percentile, the line within the box marks the median, and the boundary of the box farthest from zero indicates the 75th percentile

Full size image

Literature review

A structured literature search of PubMed and of Thompson Reuter’s Web of Knowledge® database retrieved 65 publications (Fig. 1). Among these, 30 studies on MRZR were identified that met the inclusion criteria (see Table 2 for references). In these and the present study, results of 1478 MRZR tests were reported. A positive MRZR was found in 458/724 (63.3%) tests performed in patients with MS but only in 19/754 (2.5%) tests performed in control patients (p < 0.000001) (Tables 2, 3). This corresponds to cumulative specificity of 97.48% (CI 95% 96–98.4) for MS, cumulative sensitivity of 63.3% (CI 95% 59.6–66.8), pLR of 25.1 (CI 95% 16–39.3) and nLR of 0.38 (CI 95% 0.34–0.41). The difference between MS patients and non-MS patients remained highly significant (p < 0.000001) after exclusion of all healthy and non-inflammatory controls.

Table 2 Frequency of MRZR positivity, as defined by a combination of at least two positive antibody indices (AI), in various CNS disorders as reported in the previous literature and the present report
Full size table
Table 3 Proportions of MRZR-positive and -negative patients in the various disease groups and corresponding likelihood ratios
Full size table

The frequency of positive MRZR was higher in adults with MS (67.4%, CI 95% 63.5–71.1; 14 studies; 617 tests) than in children with MS (39.3%, CI 95% 30.1–49.2; 3 studies; 107 tests) (p < 0.000001) (Table 4), and was highest in adult MS patients from Western Europe (69.4%, CI 95% 65.3–73.2; 11 studies; 546 tests) (Table 2).

Table 4 Frequency of MRZR in adult patients and in paediatric patients with MS
Full size table

Notably, the cumulative sensitivity in adult patients (67.4%; n = 617; 14 studies) was in close agreement with the sensitivity found in the largest single study performed in adult MS (67.2%; n = 177) [60].

Apart from MS, MRZR positivity was mainly reported in patients with rheumatic disorders and CNS symptoms (RD/CNS), which included cases of SLE, Sjögren’s syndrome, and Wegener’s granulomatosis (Table 2). If patients with MS and patients with RD/CNS were considered together, only 1.8% of the remaining controls were positive for MRZR (p < 0.000001), corresponding to specificity of 98.2% (CI 95% 96.9–99) and pLR of 35.1 (CI 95% 20.4–60.3) (Table 3). If patients with MS or RD/CNS were compared with patients with other inflammatory neurological diseases of the CNS (see Table 2 for details), cumulative specificity of 96.21% (CI 95% 93.3–97.9) and pLR of 16.2 (CI 95% 9.3–28.3) resulted.

Only 1 of the 305 patients with non-inflammatory diseases of the CNS or inflammatory diseases of the peripheral nervous system and none of the 99 healthy controls had a positive MRZR.

Of particular note, MRZR was absent in 52/53 patients with neuromyelitis optica or MOG-IgG-positive encephalomyelitis [5, 32, 35, 37, 54] (Table 2), two conditions that were considered variants of MS in the past and have only relatively recently been recognised as disease entities distinct from MS based on immunopathophysiological and neuropathological grounds.

Discussion

MRZR testing is currently performed at most CSF laboratories in Germany as a complementary test to OCB and QIgG testing in patients with suspected MS, and has recently been proposed in the current guidelines of the German Society for Cerebrospinal Fluid Diagnostics and Clinical Neurochemistry as a ‘rule-in’ test for the diagnosis of MS. Our results indicate that MRZR may in fact be a highly specific laboratory marker, which if present substantially increases the likelihood for a diagnosis of MS.

High specificity makes MRZR a typical ‘rule-in’ test

Currently, there is no other laboratory marker with a similarly high specificity for MS (97.5%; CI 95% 96–98.4). While testing for total IgG OCBs (≥95%, probably with a latitudinal gradient [16]) or QIgG is highly sensitive in MS, total IgG OCBs are present in a plethora of other autoimmune and infectious conditions and are thus not very specific for MS [7, 10, 46]. Conversely, MRZR testing is highly specific for MS but only moderately sensitive (67.4%, CI 95% 63.5–71.1, in adults). This translates into a very high pLR and a relatively weak nLR (Table 3), which makes MRZR a ‘rule-in’ test rather than a ‘rule-out’ test. Conversely, the high sensitivity and low specificity of total IgG OCB or QIgG testing results in a weak pLR but strong nLR, making OCBs a better ‘rule-out’ than ‘rule-in’ test for MS. Parallel testing for OCBs and MRZR might therefore substantially strengthen the diagnostic relevance of CSF analysis in patients with suspected MS.

MRZR is present at disease onset and may predict conversion from CIS to MS

Given that early treatment is thought to be of high prognostic impact in MS [8, 9, 29, 40–42], it is of clinical relevance that MRZR was demonstrated to be present early in the disease course and to predict later conversion to MS in patients with a clinically isolated syndrome (CIS) suggestive of MS, including in patients with acute monosymptomatic optic neuritis [6, 32, 73]. MRZR could thus assist physicians in making early treatment decisions. The combination of MRZR testing with OCB testing and brain MRI was suggested to further increase the predictive value of MRZR seropositivity for a diagnosis of MS in patients with CIS [6, 32, 73].

Repeat testing increases sensitivity of MRZR for MS

In MRZR-negative patients, repeat lumbar puncture was reported to increase the sensitivity of MRZR for MS. Petereit and Reske (2005) found a frequency of MRZR positivity of 63% in a cohort of 70 MS patients if only the first lumbar puncture (LP) was considered, but in 79% if follow-up samples (median 1, range 1–6) were also taken into account (unpublished data from Ref. [52]). This is in accordance with the fact that most patients with MS who do not show a positive MRZ reaction as defined by two or three positive AIs showed at least a monospecific, so-called incomplete response to one of the three viruses at first LP (89% [60] and 94% [17] in the two largest studies using enzyme-linked immunosorbent assay [ELISA] for determining antibodies to M, R and Z and 100% in a study using affinity-mediated capillary immunoblotting, an alternative method for studying the intrathecal production of pathogen-specific antibodies [67]).

MRZR is detectable also in a subset of OCB-negative patients with MS

Determination of antigen-specific IgG by AI calculation or affinity blotting has been repeatedly reported to be more sensitive than total IgG OCB and QIgG determination in patients with infectious conditions [17, 20, 31, 38, 39, 68]. In accordance with these observations, two recent studies independently found a positive MRZ reaction also in a subset of patients with RRMS, secondary progressive MS and primary progressive MS negative for total IgG OCBs [5, 67, 70]. Stich et al. recently demonstrated a positive MRZ reaction as defined by a response to at least two of the three antigens in 4/17 (18%) OCB-negative patients with MS according to McDonald et al. (2001/2005) (with six additional OCB-negative patients displaying a monospecific reaction) and in 2/17 (12%) by means of affinity blotting employing recombinant viral antigens and a highly sensitive chemiluminescence detection technique, but in none of 11 controls [69, 70]. Similarly, Brecht et al. found a positive MRZ reaction in 11/46 (24%) OCB-negative patients with MS according to McDonald et al. (2005) using the same standardised ELISA test used in the present study, but in none of 37 controls [5]. The authors did not report if any distinctive clinical features were present in these patients.

MRZR is part of the polyspecific intrathecal humoral immune response in MS

The intrathecal humoral immune response in MS comprises antibodies to a broad panel of viral and bacterial agents such as herpes simplex virus (HSV), Epstein Barr virus (EBV), human herpes virus 6, mumps virus and Chlamydia pneumoniae [14, 15, 50, 54, 63, 65]. However, antibodies to measles virus, rubella virus and varicella zoster virus are considered its most common constituents and are thus best evaluated. Whether inclusion of antibody reactivities other than those against M, R and Z would improve the sensitivity or specificity has not been studied systematically. However, Reiber et al. (1998) found no significant increase in sensitivity when antibodies to HSV were tested in addition to antibodies to M, R and Z [60]. Instead, inclusion of HSV in the diagnostic panel could lower the specificity because of a possible cross-reactivity between varicella zoster virus and HSV antibodies [13, 66, 67]. In children with MS, Rostasy et al. (2003) found slightly higher sensitivity after inclusion of Chlamydia pneumoniae (52 versus 44%) without loss of specificity; however, the control group was relatively small (n = 10).

Conversely, it seems inadvisable to limit the analysis to two of the three parameters (e.g. for economic reasons), since this would result in substantial loss of sensitivity. Although some studies have found intrathecal antibodies to Z to be slightly less frequent in MS than those to M and R, the proportion of patients with a bispecific reaction that included Z (i.e. either M + Z or R + Z) was still 16% in two large cohorts [52, 60]. Instead of changing the test panel, weighting of the three antibody specificities (M, R and Z) might be useful: In a recent study, Brettschneider et al. (2009) found that the use of a scoring system (with different scores for M, R and Z), established by means of logistic regression analysis, may possibly further increase the predictive value of MRZR for conversion to MS within 2 years in patients with CIS [6].

Pathophysiological implications of MRZR positivity

The exact reason for the presence of the polyspecific humoral immune in the CSF of patients with MS, which is detectable by OCB, QIgG and MRZR testing, is still not well understood. As simultaneous infection with several neurotropic viruses is highly unlikely, and the polymerase chain reaction (PCR) for measles virus, rubella virus and varicella zoster virus has been shown to be negative in MRZR-positive patients with MS [22], MRZR is thought to represent non-specific, so-called bystander activation of B cells (e.g. long-lived plasma cells) within the CNS in the absence of viral replication. This is further corroborated by recent data demonstrating that the virus-specific fraction of total intrathecally synthesised IgG is significantly (20- to 60-fold) lower in MS than typically found during acute viral infection [28]. The total intrathecally synthesised M + R + Z antibody concentration in the CSF was shown to represent less than 2% of the total intrathecally synthesised IgG [60].

The polyspecific intrathecal IgG response in MS may indicate an enhanced B cell-promoting environment in the CNS of patients with MS, which is also suggested by the life-long persistence of OCBs and the recent observation of B cell follicles in the meninges of patients with MS [47]. Whether the presence of EBV-infected B cells in these intrameningeal follicles (as well in white matter lesions) provides an explanation for the continuous B cell activation in MS is currently a matter of debate [19]. EBV can efficiently immortalise B cells, and thus establish lymphoblastoid cell lines in vitro [47].

Of note, the presence of intrathecally produced polyspecific anti-viral antibodies in MS as quantitatively evidenced by ELISA has been confirmed qualitatively by the discovery of CSF-restricted anti-viral OCBs (as opposed to total-IgG OCBs) [20]. Sindic (1998) found OCBs to measles, rubella, varicella zoster, and mumps virus in 18/18 patients with MS using an antigen-driven capillary blot technique, 15 of whom (83%) showed a polyspecific reaction [20, 67]. Interestingly, these bands did not correspond to the main OCBs present in the same patients, indicating that M, R and Z are not the main targets of the intrathecal IgG response in MS [20, 67].

The fact that MRZR was shown to be mostly absent in patients with other well-established autoimmune conditions of the CNS, such as paraneoplastic neurological disorders, neuromyelitis optica and MOG-IgG-associated encephalomyelitis, or chronic infectious diseases, including neuro-borreliosis, neuro-syphilis and neuro-tuberculosis (Table 2), suggests that MRZR is not generally associated with CNS autoimmunity or a general result of chronic CNS inflammation, but may be more specifically linked to the immunopathophysiology of MS.

Of interest, antibodies to measles, rubella and varicella zoster have been demonstrated not only in the CSF but also in extracts of brain tissue from patients with MS [64].

MRZR-negative multiple sclerosis

It is not fully understood why some patients with MS lack a positive MRZ reaction. This could simply reflect variations—or an increase over time—in the amount of intrathecally produced IgG or in IgG affinity between patients and, thus, limited sensitivity of the immunoassays used [59, 60]. This view is supported by the findings that repeat lumbar puncture was found to result in increased sensitivity of MRZR testing [52]. Alternatively, the lack of positive MRZR could reflect interindividual differences in history of previous infections or immunisations, as suggested by differences between populations from countries with different rubella prevalences and/or vaccination schemes [47]. Similarly, regional differences (with a latitudinal gradient) in the frequency of OCBs in MS have been reported [16]. Finally, the lack of positive MRZR could be due to real pathophysiological differences among patients with MS, a condition considered by some to be histopathologically heterogeneous [43]; studies of MRZR in histopathologically characterised patients with MS are currently in progress. Moreover, patients with MOG-IgG-positive encephalomyelitis (MOG-EM) [34–36, 44, 51] or NMOSD [37, 74], which are now recognised as disease entities in their own right immunopathophysiologically distinct from MS, sometimes meet the current clinicoradiological criteria for MS and, in consequence, were frequently misdiagnosed with MS in the past. Of interest, MRZR was negative in virtually all (52/53) patients with MOG-EM or NMOSD analysed so far [5, 32, 35, 37, 54] (Table 2). Systematic studies on the frequency of MOG-IgG and AQP4-IgG in MRZR-negative patients diagnosed with MS seem warranted.

Differential diagnostic considerations

Apart from MS, MRZR positivity has been observed mainly in patients with rheumatic diseases and CNS involvement (RD/CNS) (Table 2). RD/CNS has thus to be considered as a potential differential diagnosis of MS in MRZR-positive patients. However, the number of cases detected in previous studies was low (n = 6), confirmatory data from large and well-defined cohorts is missing so far, and the possibility of co-existing MS was not excluded in some of those cases. Moreover, well-established standardised serological tests as well as diagnostic criteria for RD exist, which can be applied in patients with suspected MS to exclude RD/CNS. Finally, it should be taken into account that CNS involvement due to RD is very rare compared with MS.

Besides its diagnostic implications, the fact that a positive MRZR is present also in a subset of patients with RD/CNS is interesting from an immunological point of view, as it could indicate possible similarities in the pathophysiology of MS and RD/CNS (e.g. presence of an enhanced B cell-promoting environment [47]).

A positive MRZ reaction has also been reported in a few patients with autoantibody-associated CNS disorders (3/113), neuroborreliosis (2/54) or neuro-sarcoidosis (2/26) (Table 2; Fig. 3). As a limitation, however, the latter two conditions are difficult to diagnose and, in addition, may well co-exist with classical MS in some patients. Unfortunately, it is unknown whether the few MRZR-positive control patients reported in the literature met the current diagnostic criteria for MS.

Fig. 3
figure 3

Stacked bar graph showing the proportion of MRZ-positive (grey) and MRZ-negative (white) test results in various disease groups as reported in the literature (see Table 2 for a list of references). MS multiple sclerosis, RD/CNS rheumatic diseases (including vasculitis) with CNS involvement, NMO neuromyelitis optica, MOG-EM myelin oligodendrocyte glycoprotein-associated encephalomyelitis, Other Ab/CNS other autoantibody-associated disorders of the CNS (including paraneoplastic neurological disorders), ADEM acute disseminated encephalomyelitis, NB neuroborreliosis, Sy/Tb/Sa neuro-syphilis/neuro-tuberculosis/neuro-sarcoidosis, CJD Creutzfeldt–Jakob disease, FNP peripheral facial nerve palsy, others other neurological disorders (see the footnotes to Table 2 for a list of diagnoses); HC healthy controls. P values were corrected for multiple testing according to Bonferroni

Full size image

Diagnostic caveats

Some caveats need to be considered when dealing with MRZR results. First, given that a positive MRZR has also been observed in a few patients with diseases other than MS, it must be kept in mind that a positive MRZR alone does not prove the presence of MS in a given patient. Rather, it increases the pre-test odds for a diagnosis of MS by a factor indicated by the test-specific pLR (e.g. by a factor of 27 in adults, see Table 3; conventionally, laboratory tests with a pLR of 10 are considered useful [11]), with the pre-test odds depending mainly on the reliability of the diagnostic criteria used.

Second, only an intrathecal reaction against at least two of the three viral agents (M + R, M + Z, R + Z or M + R + Z) may be taken into account. Monospecific reactions against M, R or Z are present in many patients with MS who lack the typical bi- or trispecific pattern (likely representing a forme fruste of the complete MRZR) but are non-specific, since they are frequently found in conditions other than MS and may indeed indicate acute infection with the respective virus. While absolute AI levels do not permit discrimination between a microorganism-driven process and the bystander or ‘nonsense’ activation found in MS, calculation of the virus-specific intrathecal IgG fraction, F(s), can be helpful in selected cases [28, 50], alongside PCR testing of paired CSF and serum samples.

Third, the rate of MRZR positivity in a given population may depend on the natural prevalence of measles, rubella and varicella zoster virus as well as on the local vaccination coverage. The frequency of MRZR in MS has been suggested to be lower in patients from tropical or subtropical regions than in Western European patients [61].

Finally, it should be kept in mind that the frequency of bispecific MRZR seropositivity is lower in children and adolescents (Table 4), particularly before puberty. This possibly reflects the pre- vs. postpubertal prevalence of rubella virus antibodies [59].

Limitations

This study has obvious limitations, some of which are inherent to the study design. First, patient ethnicity, median age, pretreatment and diagnostic criteria, and laboratory methods differed among the various studies analysed here. This may in part explain the interstudy variations in MRZR positivity rates (Fig. 4). Notably, however, the cumulative sensitivity of all studies in adult MS patients found in the present literature review (67.4%) was in almost perfect accordance with the sensitivity found in the largest single study (67.2%), suggesting that this analysis may be robust despite those differences, owing to the high number of studies and patients included. Second, negative studies are generally less likely to be published, which could hypothetically have introduced a bias towards studies that found a positive MRZ reaction in MS. However, the fact that a positive MRZ reaction in MS was consistently observed in so many studies over a period of three decades, by independent groups and by independent methods, argues against a strong influence of such bias. Moreover, we also screened congress proceedings to reduce the risk of publication bias. Finally, despite the high total number of non-MS controls tested for MRZR (n = 754), patient numbers in individual control groups were rather small. Studies that systematically evaluate the rate of MRZR positivity in large, homogeneous control cohorts are warranted.

Fig. 4
figure 4

Forest plot showing the frequency of MRZR in adult patients with MS (centre value and upper and lower limits of the 95% confidence interval) as observed in the previous literature. Studies with fewer than seven patients are not shown. The uppermost red box and whiskers refer to the cumulative sensitivity (red box and red dotted line) and 95% confidence interval (red whiskers) of the total cohort (N = 617 tests). The black dotted vertical line indicates the frequency of MRZR reported in adult patients with diseases other than MS (2.5%; N = 754 tests). *Studies from non-Western European countries (MRZR frequency in a given population depends on local vaccination schemes and history [40])

Full size image

Conclusion and outlook

There is a need for a highly specific laboratory marker of MS. Established CSF markers such as OCBs and QIgG are sensitive but rather unspecific. Accordingly, a huge number of differential diagnoses have to be excluded before MS can formally be diagnosed according to the current criteria [48, 55]. MRZR is the most specific routine laboratory marker of MS available so far and substantially increases the likelihood of that diagnosis. Our results provide a strong rationale for prospective and systematic studies on the diagnostic and prognostic impact of MRZR testing in MS and CIS.