Soluble CD27 Levels in Cerebrospinal Fluid as a Prognostic Biomarker in Clinically Isolated Syndrome

Key Points

Question  What is the value of soluble CD27 in predicting long-term prognosis in patients with clinically isolated syndrome?

Findings  In this prospective study that included 77 patients with clinically isolated syndrome, soluble CD27 was independently associated with MS diagnosis. Additionally, soluble CD27 was associated with a 5.5-fold higher annualized relapse rate.

Meaning  Soluble CD27 in cerebrospinal fluid of patients with clinically isolated syndrome could be used to predict which patients will have an active disease course at the time of a first attack.

Importance  There is a growing number of therapies that could be administered after the first symptom of central nervous system demyelination. These drugs can delay multiple sclerosis (MS) diagnosis and slow down future disability. However, treatment of patients with benign course may not be needed; therefore, there is a need for biomarkers to predict long-term prognosis in patients with clinically isolated syndrome (CIS).

Objective  To investigate whether the T-cell activation marker soluble CD27 (sCD27) measured in cerebrospinal fluid of patients at time of a first attack is associated with a subsequent diagnosis of MS and a higher relapse rate.

Design, Setting, and Participants  This prospective study included 77 patients with CIS between March 2002 and May 2015 in a tertiary referral center for multiple sclerosis, in collaboration with several regional hospitals. Patients with CIS underwent a lumbar puncture and magnetic resonance imaging scan within 6 months after first onset of symptoms.

Main Outcomes and Measures  Soluble CD27 levels were determined in cerebrospinal fluid using a commercially available enzyme-linked immunosorbent assay. Cox regression analyses was used to calculate univariate and multivariate hazard ratios for MS diagnosis. Association between sCD27 levels and relapse rate was assessed using a negative binomial regression model.

Results  Among 77 patients with CIS, 50 were female (79.5%), and mean (SD) age was 32.7 (7.4) years. Mean (SD) age in the control individuals was 33.4 (9.5) years, and 20 were female (66.7%).Patients with CIS had higher cerebrospinal fluid sCD27 levels than control individuals (geometric mean, 31.3 U/mL; 95% CI, 24.0-40.9 vs mean, 4.67 U/mL; 95% CI, 2.9-7.5; P < .001). During a mean (SD) follow-up of 54.8 (35.1) months, 39 of 77 patients (50.6%) were diagnosed as having MS. In a model adjusted for magnetic resonance imaging and cerebrospinal fluid measurements, sCD27 levels were associated with a diagnosis of MS (hazard ratio, 2.4 per 100 U/mL increase in sCD27 levels; 95% CI, 1.27-4.53; P = .007). Additionally, patients with MS with high sCD27 levels (median, >31.4 U/mL) at the time of CIS had a 5.5 times higher annualized relapse rate than patients with low sCD27 levels (annualized relapse rate, 0.06 vs 0.33; P = .02).

Conclusions and Relevance  Soluble CD27 in cerebrospinal fluid of patients with CIS was associated with MS diagnosis and a high relapse rate. Therefore, sCD27 is an activation molecule directly related to the immunopathology of the disease and is a potential clinical marker to help in treatment decisions after a first attack of suspected MS.

Multiple sclerosis (MS) is a chronic inflammatory disease of the central nervous system with large heterogeneity in severity and prognosis.¹ There is a growing number of therapies that could be administered after the first symptom of central nervous system demyelination (clinically isolated syndrome [CIS]), even before MS diagnosis. Disease-modifying therapies (DMTs) delay MS diagnosis and have a potential to prevent future disability.^2-7 However, these therapies have adverse effects; therefore, treating patients with CIS who will not reach MS diagnosis should be prevented. This underscores the need for biomarkers to predict long-term prognosis at an early phase of the disease.^8,9 Many studies have shown the crucial role for T cells and B cells in the pathogenesis of MS.^10-13 T cells activated by the T-cell receptor/CD3 complex release a soluble form of CD27 (sCD27).¹⁴ CD27 and sCD27 costimulate maturing of T cells and induce activation and proliferation of T and B cells.^15,16 Increased sCD27 levels have been reported in various autoimmune diseases such as systemic lupus erythematosus, rheumatoid arthritis,^17-20 and MS.^20,21 Soluble CD27 is considered a biomarker of active intrathecal T-cell–mediated inflammation. For distinguishing MS from noninflammatory diseases, sCD27 showed a higher discriminatory power than common humoral markers for active intrathecal inflammation such as IgG index or presence of oligoclonal bands (OCB).²² Soluble CD27 levels in serum did not discriminate normal individuals from patients with MS.²¹ It is not known whether cerebrospinal fluid (CSF) sCD27 levels at time of CIS have a predictive value for disease course. The aim of this study was to examine whether CSF sCD27 levels in patients with CIS were associated with MS diagnosis and relapse rate.²³

For this study, data were prospectively collected between March 2002 and May 2015 from patients with a first episode of demyelinating disease who visited the neurological clinic of Erasmus MC University Hospital in Rotterdam, a tertiary referral center for patients with MS, in collaboration with several regional hospitals. Patients with CIS were eligible for this study if they were between 18 and 50 years of age. Patients with alternative diagnoses and patients who had life-threatening comorbidities (ie, malignancies or AIDS) were excluded from analysis. At baseline, patients underwent a magnetic resonance imaging (MRI) scan with gadolinium and routine laboratory tests to rule out alternative diagnoses.²⁴ If a lumbar puncture was performed, extra CSF was collected for the study. Cerebrospinal fluid samples were aliquoted and stored at −80°C. For this study, we included patients with a CSF sample collected within 6 months after onset of CIS and at least 6 months follow-up after CIS. After inclusion, patients were reassessed at least annually at the neurology outpatient department. For a control group, we used CSF from symptomatic control individuals, patients with neurological symptoms but who had no objective clinical findings to define a specific neurological disease, as defined by Teunissen et al.²⁵ Control group patients underwent a lumbar puncture at the neurology department of Erasmus MC between January 2002 and April 2014 for reasons other than neuroinflammatory diseases. Those reasons were acute headache with no subarachnoid hemorrhage, no idiopathic intracranial hypertension and no inflammation (n = 15), exclusion of neurosyphilis (n = 7), and other (eg, muscle pain or sensory disturbances without central nervous system pathology) (n = 8).

The study protocol was approved by the ethics committee of the Erasmus MC University Hospital. All prospectively included patients provided written informed consent.

An exacerbation of MS was defined as worsening of existing symptoms or new symptoms after 30 days of improvement or stable disease and no evidence of alternative diagnosis. To be regarded as exacerbation, symptoms should exist for more than 24 hours and not be preceded by fever. All exacerbations were confirmed by neurological examination.²⁶ The diagnosis of MS in all patients was made according to the McDonald criteria (revised in 2010)²⁷ based on a second clinical relapse or the baseline MRI scan.²⁷ All patients underwent a baseline MRI scan with gadolinium. A follow-up MRI scan was not performed regularly; therefore, we did not take the follow-up MRI scan into account when defining MS diagnosis. Patients with monophasic CIS did not fulfill McDonald criteria (revised in 2010) criteria during follow-up. Disability was measured using the Expanded Disability Status Scale (EDSS).²⁸ When patients experienced a second attack, EDSS was performed annually. Expanded Disability Status Scale scores performed within 3 months after exacerbation were not used in the analyses. Follow-up was calculated by subtracting CIS date from the last visit date.

Collected CSF samples were immediately centrifuged for 10 minutes at 3000 rpm to separate the supernatant from cells and cellular elements. After centrifugation, CSF was aliquoted and stored at −80°C until use for this study.²⁹ Routine CSF diagnostics, including IgG index, OCBs, cell count, and total protein, were performed. Soluble CD27 levels were measured twice for each sample using the commercially available enzyme-linked immunosorbent assay kit (PeliKine compact human sCD27 kit; Sanquin).¹⁴ The enzyme-linked immunosorbent assay was performed according to the manufacturer’s instructions. Concentrations were expressed in units per milliliter by reference to a standard curve supplied with the enzyme-linked immunosorbent assay kit. The analysts who performed the experiments were blinded for the clinical diagnosis of patients. The detection limit of the enzyme-linked immunosorbent assay was 5 U/mL.

Statistical analyses were done using SPSS, version 21.0 (SPSS Inc) for Windows and GraphPad Prism5 (GraphPad) for Windows. To assess normality of data distribution, we performed the Kolmogorov-Smirnov test. The distribution of sCD27 was nonparametric, and after log transformation the data were normally distributed. Therefore, geometric means were calculated. For group comparison of continuous variables, we applied 2-tailed t test (age, follow-up, and sCD27) or nonparametric Mann-Whitney U test (IgG index, time between CIS, and lumbar puncture). Nominal data comparison between groups was done using χ² or Fisher exact test (sex, CIS type, treatment before MS diagnosis, and OCB). Time to MS was calculated from onset of first symptoms to MS diagnosis (McDonald criteria, revised in 2010).²⁷ Patients who were not diagnosed as having MS during follow-up were considered as censored observations. Cox proportional hazard regression analysis was used to calculate univariate and multivariate hazard ratios. The proportional hazard assumption was tested by including time dependent covariate in the model. Odds ratios for MS in the subgroup with at least 0 subclinical T2 lesions on the MRI scan were calculated using binary logistic regression. Annualized relapse rate (ARR) was compared between groups using a negative binomial regression model with the natural logarithm of number of follow-up years after a second clinical attack as offset. This offset corrects for the difference in follow-up between included patients. Because of overdispersion of data, we did not use a Poisson regression model. P values less than .05 were considered significant.

Inclusion criteria for this study were met by 77 patients with CIS. The mean (SD) follow-up time for the total CIS group was 52.4 (34.5) months. The median time from CIS to MS diagnosis (McDonald criteria, revised in 2010) for 39 patients was 7.8 months (IQR, 2.0-41.9). Nine patients (11.4%) received DMT (Glatiramer acetate [n = 3] or interferon beta [n = 6]) in the period between CIS and MS diagnosis. No patients were treated with DMT at time of lumbar puncture. Thirty symptomatic controls were age and sex matched. Patient and control characteristics are shown in the Table. Patients with CIS were further stratified into MS and monophasic.

We first compared sCD27 levels in CSF between patients with CIS and control individuals. After log transformation, sCD27 levels were normally distributed. Soluble CD27 levels were significantly higher in patients with CIS than in control individuals. The geometric mean was 31.3 U/mL (95% CI, 24.0-40.9) vs 4.7 U/mL (95% CI, 2.9-7.5), respectively (P < .01) (Figure 1A).

Next, we compared patients with CIS who were diagnosed as having MS (n = 39) (mean [SD] follow-up time, 52.4 [34.5] months) with patients who remained CIS (n = 38) during follow-up. Soluble CD27 levels were significantly increased in patients with CIS with a future MS diagnosis (geometric mean, 42.0 U/mL; 95% CI, 29.1-50.6 vs 23.2 U/mL; 95% CI, 15.8-33.9, respectively; P = .03) (Figure 1B).

We stratified data by sCD27 levels using the median sCD27 level (31.4 U/mL) of patients with CIS as the cutoff. In Figure 2, we show that patients with CIS with high levels of sCD27 had a shorter time to MS diagnosis than patients with low sCD27 levels (log-rank test: mean, 37.59; 95% CI, 25.73-49.45 vs 56.91; 95% CI, 45.23-68.59; P = .02). After adjustments (>9 T2 lesions, contrast enhancement on baseline MRI scan, presence of OCB and IgG index), multivariate Cox regression analyses showed that sCD27 is an independent predictive factor for time to MS diagnosis. The hazard ratio (HR) was 2.4 per 100 U/mL increase of sCD27 (95% CI, 1.27-4.53; P = .007).

We performed a subanalysis where we excluded monophasic patients with CIS with less than 2 years of follow-up (n = 12). The HR in univariate and multivariate Cox analyses remained significant (HR, 2.3 per 100 U/mL; 95% CI, 1.23-4.31; P = .009).

Nine patients (11.8%) received DMT before MS diagnosis. Seven of 9 patients (77.8%) had sCD27 levels higher than 2 times the maximum level of the control group. Four of these patients (44.4%) were diagnosed as having MS. The log-rank test for time to MS diagnosis was more significant when we excluded these 9 patients with DMT before MS diagnosis (mean, 30.91; 95% CI, 17.97-43.86 vs 57.81; 95% CI, 45.87-69.75; P = .004). The corrected HR for MS diagnosis was 2.4 per 100 U/mL increase of sCD27 (95% CI, 1.27-4.60; P = .007).

The total number of relapses in patients who experienced a second clinical attack (n = 31) was 26 in a total of 120.9 person-years at risk. The group with high sCD27 levels at time of CIS (n = 18, using the median sCD27 level in patients with CIS as the cutoff [31.4 U/mL]) had a 5.5 times higher ARR during follow-up after the second attack, estimated by negative binomial regression analysis with log link, corrected for follow-up time (ARR, 0.06 vs 0.33; P = .02). Other possible factors effecting ARR, such as contrast enhancement, dissemination in space (McDonald criteria, revised in 2010), more than 9 T2 lesions on the first MRI scan, unique OCBs in CSF, and proportion of time on DMT, did not have significant influence on ARR. However, IgG index was significantly correlated with ARR. One point elevation in IgG index was associated with 3.7 times higher ARR (95% CI, 1.45-9.52; P = .006).

Soluble CD27 titer did not correlate with EDSS later in the disease course (data not shown). We collected EDSS scores of 24 of 31 patients with a second attack. During follow-up, only 3 patients reached an EDSS score of 4 or more. Of these, all had high sCD27 levels (greater than the median of 46.3 U/mL).

There was a positive correlation between IgG index and sCD27 in CSF at time of CIS (Spearman ρ, 0.68; P < .001). Patients with CIS with more than 1 unique OCB in CSF (n = 54) had significantly higher sCD27 levels than CIS patients with no OGB (n = 19) (geometric mean, 42.3 U/mL; 95% CI, 37.4-61.3 vs 9.1 U/mL; 95% CI, 6.2-15; P < .001) (Figure 3).

Patients with more than 9 T2 lesions on the baseline MRI scan (n = 24) had higher sCD27 levels (geometric mean, 52.3U/mL; 95% CI, 35.6-73.7 vs 24.8 U/mL; 95% CI, 18.0-34.4; P = .009). Patients with no subclinical T2 lesion on the brain MRI at time of CIS (n = 14) had lower sCD27 levels than patients with at least 1 subclinical T2 lesion (geometric mean, 14.8 vs 37.0; P = .008). In a subanalysis, we selected patients with at least 1 subclinical T2 lesion (n = 63). Within this group, the odds ratio of high sCD27 (using the median sCD27 level in CIS patients as a cutoff [31.4 U/mL]) for MS diagnosis was 3.0 (95% CI, 1.07-8.5; P = .04). In the same subgroup, the odds ratio for presence of OCBs did not reach significance (odds ratio, 2.9; 95% CI, 0.8-9.9; P = .09). Contrast enhancement on the MRI scan at time of CIS was not associated with sCD27 levels (Figure 3).

The rationale behind this study is based on the observation that CSF levels of sCD27 are significantly elevated in patients with MS vs symptomatic control individuals.^21,22,30 We reasoned that an MS-associated biomarker could perhaps predict disease course at first attack, in analogy with MS-related MRI abnormalities that can predict future definite MS.^27,31,32 Indeed, we demonstrate here that high sCD27 is associated with a subsequent diagnosis of MS and is associated with higher attack frequency after diagnosis. Twenty-one of 77 patients with CIS (27%) had sCD27 levels less than the maximum level of controls. These patients had a less active disease course.

Soluble CD27 is a stable molecule, and the immune assay is quite reproducible and performed consistently in the literature.^21,22,30 It carries some extra attraction because this T-cell activation marker appears directly related to the core of immunopathology in MS.^10,11,33 It is of note that Komori et al²² have found a clear relation between sCD27 levels and the presence of intrathecal T cells. Among an impressive set of CSF inflammatory parameters, sCD27 performed best in identifying local inflammation and appeared correlated with progression. It seems not farfetched to suggest that higher T-cell activation is related to higher lesion load and a more active disease course.

The number of T2 lesions on brain MRI at the time of CIS is currently the most predictive tool for MS diagnosis. A larger number of T2 lesions result in a shorter time to MS diagnosis. Also, unique OCBs in CSF are predictive for a second attack in patients with CIS.³

In a multivariate analysis, we found that the association of sCD27 with future MS diagnosis remained after adjustment for known predictive factors such as OCB, IgG index, and MRI abnormalities (more than 9 T2 lesions and contrast enhancement on baseline MRI scan).

In line with previous studies, we observed a correlation between sCD27 and IgG index^21,22,26 (Figure 3). Whether there is a direct role for sCD27 released by T cells on IgG production by B cells is speculative, but such a functional effect has been shown in vitro by Dang et al.³⁴ They showed that the binding of sCD27 to the CD27 ligand CD70 on memory B cells stimulates those B cells to differentiate into IgG-secreting plasma cells.^20,34

At time of lumbar puncture, none of the patients were treated with DMT in this study cohort. Nine patients (11.8%) received DMT before MS diagnosis. Excluding these 9 patients from the survival analyses does not have major effects on our results, although the log-rank test for time to MS diagnosis is more significant when we exclude the patients with DMT before MS diagnosis.

In a subgroup of patients with at least 1 subclinical T2 lesion, high sCD27 was associated with MS diagnosis. The sample size was too small for analyses in more subgroups. In combination with MRI findings, sCD27 quantification could be helpful in decision making for starting immunomodulatory therapy because sCD27 is a marker for neuroinflammation and predicts MS diagnosis and subsequent relapse rate.

There are some limitations of this study. First, the follow-up time has a wide range. To address this problem, we corrected for follow-up in the statistical analysis. We also did a subanalysis in which we excluded monophasic patients with CIS with less than 2 years follow-up. The HR in univariate and multivariate Cox analyses remained significant. Second, all patients underwent a baseline MRI scan with gadolinium, but a follow-up MRI scan was not performed uniformly according to a strict protocol. Therefore, we did not take the follow-up MRI scan in account when defining patients with MS but used the classic Poser criteria based on clinical manifestations. Third, although the sample size was sufficient for multivariate analysis, validation in a bigger cohort is required.

This study demonstrates that sCD27 levels in CSF of patients with CIS are associated with MS diagnosis and a high relapse rate. Therefore, CSF sCD27 could be an activation marker directly related to the immunopathology of the disease, with potential value for selecting patients with higher subsequent disease activity.

Corresponding Author: Rogier Q. Hintzen, MD, PhD, Erasmus Medical Centre, Department of Neurology and Immunology, PO Box 2040, 3000 CA Rotterdam, the Netherlands (r.hintzen@erasmusmc.nl).

Accepted for Publication: October 12, 2016.

Published Online: January 3, 2017. doi:10.1001/jamaneurol.2016.4997

Author Contributions: Dr Hintzen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: van der Vuurst de Vries, Hintzen.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: van der Vuurst de Vries.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: van der Vuurst de Vries.

Supervision: Hintzen.

Conflict of Interest Disclosures: Dr Jafari received honoraria for giving a lecture from Biogen-Idec and received personal fees from Novartis outside the submitted work. Dr Hintzen received honoraria for serving on advisory boards for Biogen Idec, Roche, and Sanofi. He participated in trials with BiogenIdec, Merck-Serono, Roche, Genzyme, and Novartis. No other disclosures are reported.

Funding/Support: The study was supported by the Dutch MS Research Foundation.

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We thank the patients, the participating centers of the Predicting the Outcome of a Demyelinating Event study, and the analysts who performed the enzyme-linked immunosorbent assay (M.J. Melief, BSc, and A.F. Wierenga-Wolf, BSc, Department of Immunology, MS Center ErasMS, Erasmus MC, Rotterdam, the Netherlands).