The MRZ reaction in primary progressive multiple sclerosis

This is a retrospective study in which all patients were treated at the University Medical Centre Freiburg in Germany between 2003 and 2015. Lumbar puncture (LP) had already been performed for all patients—for clinical purposes only and after written consent. CSF and serum samples were taken on the same day and stored according to consensus protocol for the standardization of CSF collection and biobanking [13]. Haemolytic CSF specimens were excluded. Diagnosis of MS was established according to the 2010 revised McDonald criteria, with particularly careful exclusion of relevant differential diagnoses [4]. All PPMS patients in the MS cohort were included for whom an MRZR measurement had been recorded or could be performed with existing CSF and serum samples. The MS cohort was screened for those with a recorded MRZR result, and from this set of MS patients, 100 RRMS patients were randomly drawn. The OIND group consisted of 48 patients for whom MRZR results were available from previous research, and unlike MS groups the OIND group did not include patients treated in 2015 [9]. Data concerning the ethnicity and immunization status of study patients was not available. The ethics committee of University Medical Centre Freiburg approved the study.

All MRZR assessments were performed at the Department of Virology of the University of Freiburg between 2003 and 2015. All other CSF measurements relevant to this study were carried out in the CSF laboratory at the Department of Neurology. The CSF laboratory at the Department of Neurology regularly takes part in the External Quality Control of Diagnostic Assays and Tests (EQAS) for CSF diagnostics, including protein analytics, and the laboratory at the Department of Virology regularly takes part in the EQAS for virus diagnostics, including virus serology. Both EQAS procedures are performed by Instand e.V. (Gesellschaft zur Förderung der Qualitätssicherung in medizinischen Laboratorien e.V.; Düsseldorf; Germany).

Total immunoglobulin (Ig) concentrations in serum and CSF were detected nephelometrically (ProSpect System, Siemens, Germany), whereas measles-, rubella- and varicella-IgG (IgGspec) levels in CSF and serum were measured by enzyme linked immunosorbent assay (Serion classic ELISA, Germany), both according to the manufacturer’s instructions. MRZR was determined from the three respective virus-specific AI which were calculated as follows: 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 according to Reiber’s formula [14]. For a positive AI finding the threshold of AI ≥ 1.5 was applied [9, 10, 12, 15]. Previous studies have varied as to how many positive AIs are required for positive MRZR. In this study, MRZR-2 is used to refer to the MRZR definition requiring two or more positive AI, and MRZR-1 to refer to the MRZR definition requiring only one or more positive AI. Where an AI could not be calculated because no antibodies were detected in the CSF, AI was considered to be 1.0 (negative). CSF laboratory records were used which routinely include total CSF cell count, significant quantitative intrathecal antibody synthesis (defined as ≥10%), IgG index, QIgG, IgG concentration in CSF and presence or absence of oligoclonal bands (OCB) according to the Reibergrams and the CSF consensus report [16]. Detection of OCB for patients was performed using a highly sensitive isoelectric focusing technique on agarose gel followed by immunofixation (Hydragel Isofocusing, sebia, France) [17]. A positive OCB finding is defined as two or more OCB [16].

Statistical testing of differences between groups on gender, prevalence of positive AI, MRZR, intrathecal Ig synthesis and OCB was performed using Fisher’s exact test (two-tailed). Differences of mean values of AI, total CSF cell count, intrathecal Ig synthesis, QIgG, IgG concentrations in CSF and age between groups were tested using Student’s t test (two-tailed). A p value <0.05 was regarded as statistically significant. The correlation between MRZR and OCB status was measured using the Phi correlation coefficient (\(\phi\) ). A correlation coefficient between 0.2 and 0.4 was considered as weak, between 0.4 and 0.6 as intermediate and >0.6 as strong.

The complete 2003–2015 cohort consisting of 1668 patients with a recorded MS diagnosis was retrospectively screened for MS subtype, RRMS or PPMS, following the 2010 McDonald criteria [4]. 236 PPMS patients (14.1%) were found of whom 96 had to be excluded due to missing CSF/serum samples and 37 due to insufficient clinical data, resulting in a PPMS group of 103 well-characterized patients. A random sample of 100 RRMS patients already tested for MRZR for clinical reasons was drawn from the same MS cohort. Additionally, an existing group of 48 patients with OIND was drawn on for comparison. Twenty-two of this OIND group had been diagnosed with neurosarcoidosis (NS), 19 with autoimmune encephalitis (AIE) and 7 with acute disseminated encephalomyelitis (ADEM). Table 1 shows key demographic features of the three study groups.

Results of AI tests of the three study groups are shown in Table 2. No statistically significant differences were found between the two MS groups in respect of frequency of one, two or three positive AIs and mean values of any of the three AI (M, R and Z). However, a positive AI for M and R was statistically significantly more frequent in PPMS compared to RRMS. Compared to both MS subgroups, the OIND group showed lower mean AI values and less frequent positive AI for all three viruses.

In accordance with AI findings, positive MRZR was found in a minority of OIND patients (MRZR-2: 8.3%, MRZR-1: 22.9%), statistically significantly less than in either of the MS subtypes (PPMS MRZR-2: 54.4%, PPMS MRZR-1: 83.5%; RRMS MRZR-2: 43.0%, RRMS MRZR-1: 69.0%—see Fig. 1).

Merging the two MS subtypes (PPMS and RRMS, n = 203) confirms that frequencies of both positive MRZR-2 (48.8%) and MRZR-1 (76.4%) among MS patients are statistically significantly different to those found in the OIND group (p < 0.0001 each).

Table 3 presents MRZR results from the present study alongside those from previous studies. For these studies, findings from both MRZR definitions (MRZR-2 and MRZR-1) are listed where these could be reconstructed from the data provided in the relevant publication.

As can be seen from Fig. 2, the prevalence of positive OCB in CSF was similar in the two MS groups (93.2% for PPMS and 87.0% for RRMS) and significantly higher than in the OIND group (31.2%).

As presented in Tables 4 and 5, there was a weakly positive correlation between OCB and MRZR status (applying to both MRZR-2 and MRZR-1) in the combined MS group. Among the few OCB negative MS patients, 15.0 and 30.0% showed positive MRZR-2 and MRZR-1 respectively.

Inflammatory CSF changes were found in 93.2% of PPMS patients and in 87% of RRMS patients (n.s.). The most frequent pathological CSF finding was positive OCB (see above) followed by quantitative intrathecal IgG synthesis (51.5% in PPMS patients and 50.0% in RRMS; n.s.), elevated total CSF cell count (PPMS: 22.3%, RRMS: 43.0%; p < 0.01), quantitative intrathecal IgM synthesis (PPMS: 8.7%, RRMS: 14.8%; n.s.) and quantitative intrathecal IgA synthesis (PPMS: 7.8%, RRMS: 8.0%; n.s.). The data from the routine CSF analyses are presented in Table 6.

PPMS patients showed typical demographic features: a more balanced gender distribution and a higher average age than RRMS patients [3, 18]. It has to be mentioned that diagnosis of RRMS can usually now be made earlier than at the mean age of this cohort (39 years). The high average age of RRMS patients might be a consequence of selection bias, as discussed below.

Frequencies of positive AI and AI mean values in both MS groups (M > R > Z) were similar compared to a previous study but inverse in OIND patients (M < R < Z, Table 2) [15]. The rather low frequency of positive MRZR-2 in the OIND group indicates that MRZR-2 might be particularly helpful in distinguishing both MS subtypes from OIND. The less rigorous MRZR-1 is clearly less specific, showing more false positive results in OIND. This was also found in another recent study in tertiary care, where MRZ-1 was found to be positive in 19% of patients with OIND (well in line with the 23% in our OIND group) and 8% of patients with other non-inflammatory neurological diseases (OND), whereas none of these 53 patients were MRZR-2 positive [19]. Up to now, there is no indication of a pathophysiological role of any of the three MRZ viruses in MS pathogenesis, unlike in the case of the Epstein-Barr-virus (EBV) where it was shown that after EBV infection the risk of developing MS is increased [20].

Table 3 illustrates how inconsistent results of MRZR studies can appear if differences in definition (MRZR-2 vs. MRZR-1) are not considered. After clearly distinguishing between MRZR-2 and MRZR-1, the MRZR results from this study align quite well with most previous research. The study shows positive MRZR-1 in 76% of MS patients, which is less frequent than in some earlier studies [5, 10, 15, 21]. But, the frequency of positive MRZR-1 is closely in line with two other recent studies [22, 23]. Furthermore, several studies established a similar frequency of MRZR-2 positives of around 50% [10, 12, 21, 22, 24]. Exceptionally, Jarius et al. reported a considerably higher proportion of positive MRZR-2 (88%) in their MS cohort, which was used in two MRZR studies [7, 8]. Reasons for this divergence remain elusive, but selection bias is a conceivable explanation. Unfortunately, study reports did not include a detailed description of patient selection. In the present study, there may have been selection bias in the RRMS group, due to including patients with MRZR already previously performed for clinical reasons. Though such a bias might have contributed to their unexpectedly high mean age, this was actually similar to the mean age of MS patients in an even larger retrospective CSF study (not addressing MRZR) [25]. There is a lower risk of selection bias in the PPMS group because all available PPMS patients were included in the analysis, where possible. This is supported by the fact that our PPMS group matches demographic characteristics of PPMS patients studied elsewhere very well [1, 3, 18]. Limitations of this study include the monocenter cohort, the retrospective design and the lack of data concerning ethnicity and vaccination status of enrolled patients. Infection rates and vaccination status in respect of the three MRZ viruses can influence MRZR results, as has been demonstrated for the rubella virus in Cuba [21]. Therefore, verification of the MRZR results for PPMS patients with known vaccination status, and expansion to a multicenter prospective study, including populations from different parts of the world, would be a valuable next step.

As expected, a high and very similar prevalence of OCB of around 90% was found in PPMS and RRMS patients- confirming the high sensitivity of this CSF parameter for both MS subgroups. The prevalence of OCB in PPMS and RRMS patients found here matches the findings of a previous monocenter Canadian study investigating 451 MS patients with CSF analyses performed between 1993 and 2007 [25]. In that study positive OCB was reported in 91% of RRMS and in 85% of PPMS patients.

Though the high prevalence of OCB in MS patients can be useful, a diagnostic disadvantage of OCB is their high prevalence in related conditions of diagnostic importance, in particular OIND. For example, OCB can be found in 40% of patients with neurosarcoidosis and in 50% of patients with neurologic complications of systemic lupus erythematosus (neuro-SLE) [9, 26].

The correlation between MRZR and OCB positivity in MS patients was weak, showing that MRZR provides additional, independent diagnostic information as previously reported for clinical isolated syndrome (CIS) [12]. One important issue in this context is whether MRZR tests can potentially improve the accuracy of diagnosis based on the 2010 revised McDonald criteria [4]. Referring to Tables 4 and 5, it appears that a combination of MRZR and OCB criteria could extend diagnostic sensitivity for the small proportion of OCB negative MS patients with 15% MRZR-2 positive and 30% MRZR-1 positive results. This seemingly contradicts findings in a group of 177 MS patients in which no patient was found MRZR positive and OCB negative [15]. This led the authors to recommend testing for MRZR only in OCB positive patients. However, the very high OCB prevalence in that study group (98%) meant that there were only three or four OCB negative patients, so that it could easily be a result of chance that none of these patients were found MRZR positive. MRZR/OCB findings of the present study are in line with a recent study in which MRZR-2 positives made up even 24% of 46 OCB negative MS patients [19]. Apart from that, MRZ-specific OCB were found in nearly 50% of MS patients without routine OCB [17]. Taking these results together, testing for MRZR in OCB negative patients with suspected MS appears well justified.

Considering the other CSF routine parameters, only the total cell count differed between the two MS subgroups: RRMS patients more often showed an elevated total cell count and a higher mean total cell count compared to PPMS patients. All Ig parameters (total IgG concentration in CSF, QIgG and intrathecal IgG/M/A synthesis) were similar between PPMS and RRMS patients paralleling the MRZR-2 results and indicating that these two MS subgroups seemingly display similar humoral immune responses in CSF.