Roos Van Der Vuurst De Vries

Multiple sclerosis (MS) is a chronic inflammatory disease of the central nervous system (CNS), affecting genetically susceptible individuals, influenced by environmental factors. In 85% of cases, the disease presents itself with clinical relapses as a result of demyelination in different parts of the CNS. Because the course of MS is highly heterogeneous, prediction factors at time of the first attack (CIS) are needed. The need for reliable prognostic factors is even more urgent since there is a growing number of disease modifying therapies (DMT) available for MS. Early initiation of these therapies is favourable for the prevention of future relapses and disability.

The studies described in this thesis are executed within two prospective cohorts of adult and paediatric patients followed after a first attack of demyelination. We aimed to find prognostic factors for the disease course after a first attack of demyelination both in children and adults.

Chapter 1 gives a short introduction on MS and describes currently known and used clinical, imaging and biological factors for the prediction of disease course.

The first part of this thesis focused on the disease course after CIS in adult patients. In 2017, the International Panel on Diagnosis of Multiple Sclerosis proposed new revisions to the McDonald criteria for MS. The most important change was that with the new criteria dissemination in time (DIT) could be realized when unique oligoclonal bands (OCB) were found in the cerebrospinal fluid (CSF), this could replace DIT on MRI. The other change was that not only asymptomatic but also symptomatic lesions on MRI could contribute to DIT and dissemination in space (DIS). In chapter 2, we validated these criteria in 229 adult CIS patients. With these new criteria twice as many MS diagnoses could be made at time of the first attack (97 (54%) vs 46 (26%); p<0.01). We found a higher sensitivity for a second attack using the new 2017 criteria than when using the former 2010 criteria (68% vs 36%; p<0.01). However, the specificity was lower for the new criteria (61% vs 85%; p<0.01). These data imply that the McDonald 2017 criteria will lead to a higher number of MS diagnoses in patients with a less active disease course. Fatigue is one of the most reported and disabling symptoms for MS patients. Chapter 3 evaluates the long-term course of fatigue after CIS. First, we validated our previous finding that fatigue at time of CIS predicts a future CDMS diagnosis after adjustments for sex, age, ethnicity, localization of symptoms, gadolinium enhancement at baseline MRI, and DMT before CDMS diagnosis in 235 CIS patients (hazard ratio (HR) 2.6; p<0.01). After that, we showed that the fatigue scores (FSS) increased by 0.86 units after CDMS diagnosis (p=0.01). We demonstrated that fatigue started early in the disease course, already at time of CIS, and increased after a second attack. In chapter 4, we investigated whether the T-cell activation marker soluble CD27 (sCD27) measured in CSF from CIS patients predicts MS diagnosis and a high relapse rate. First we showed higher level of CSF sCD27 in 77 patients with CIS than in 30 control individuals. We showed that within the CIS group, high sCD27 levels were predictive for MS diagnosis independently of MRI and CSF measurements (HR 2.4 per 100 U/mL increase of sCD27; p<0.01). Interestingly, within the group that experienced a second attack, we observed that patients with high sCD27 levels at time of CIS had a 5.5 times higher annualized relapse rate (ARR) during follow-up after the second attack compared to patients with low sCD27 levels at time of CIS (ARR 0.33 vs 0.06; p=0.02). Soluble CD27 levels were also associated with IgG index, unique OCB in CSF, ≥9 T2 lesions and subclinical T2 lesions on baseline MRI. In chapter 5, we performed an extensive study, in different patient groups, on Th1-like Th17 cells (CCR6+CXCR3+), which is a subpopulation of T helper 17 cells (CCR6+), a key mediator in MS disease activity. We found that Th1-like Th17 cells, expressing IL-17 but also IFN-γ and GM-CSF, were decreased in peripheral blood from CIS patients who converted to CDMS within one year (CIS-CDMS) compared to CIS patients who remained monophasic for at least 5 years (CIS-CIS) (6% vs 11%; p=0.01). For Th1 (CCR6-CXCR3+) and Th17 (CCR6+CXCR3-) cells this reduction was not seen. A reduced effector memory (EM) to central memory (CM) ratio was found for the Th1-like Th17 subset in CIS-CDMS patients (0.3 vs 0.5; p<0.01). This ratio correlated with anti-EBNA1 IgG titres (p=0.01) and fatigue scores (FSS) (p<0.01), markers that are shown to be predictive for an early MS diagnosis. In blood from patients who were already diagnosed with MS, we found strongly reduced frequencies of Th1-like Th17 effector memory (EM) cells compared to CIS-CIS and healthy controls (0.7% vs 2.8% vs 3.3%; p<0.01). In MS patients, a higher proportion of Th1-like Th17 cells was positive for the activation markers CD38 and HLA-DR. This was not seen for Th1 cells. This pleads for a selective activation of Th1-like Th17 effector cells in blood from patients with MS. After that, we analysed paired CSF and blood samples from CIS and early MS patients. We found that the Th1-like Th17 proportion in CSF was 2- to 3-fold higher and produced more IFN-γ and GM-CSF than the paired blood Th17-like Th17 cells. Ex vivo experiments confirmed the local enrichment of Th1-like Th17 cells in the CNS, this subset was enriched in brain tissue from late-stage MS patients and not in brain tissue from non-demented controls. Next, we studied the migration of peripheral T helper subpopulations into the CNS. VLA-4 was the most abundant adhesion molecule on Th1-like Th17 cells, especially on the Th17.1 subset (CCR6+CXCR3+CCR4-). This specific subset accumulated selectively in blood of natalizumab (an anti-VLA-4 monoclonal antibody) treated patients. We saw this only in natalizumab-treated patients who remained free of relapses (relapse-free patients: pre-treatment 3.8% vs post-treatment 6.8%; p<0.01, patients with relapses: pre-treatment 3.2% vs post-treatment 4.0%). And downregulation of VLA-4 by natalizumab was most prominent on Th17.1 cells. In vitro trans-well migration assays confirmed that Th17.1 was the main Th17 subpopulation migrating across human brain endothelial layers towards the inflammatory mediator CXCL10. Furthermore, Th17.1 cells were enriched in CSF from CIS and early MS patients compared to paired blood samples. Also lower Th17.1 frequencies were seen in blood from CIS-CDMS vs CIS-CIS patients (p=0.02). These data demonstrate the ability of Th1-like Th17 cells, in particular Th17.1 cells to migrate to the CNS and influence disease activity in early MS. Cigarette smoking is shown to be a risk factor that influences disease course of MS patients. However, there are conflicting results for the predictive value of smoking for MS diagnosis in CIS patients. In chapter 6, we studied the association between smoking at time of CIS and a subsequent CDMS diagnosis in 250 CIS patients. We found a higher rate of CDMS diagnoses in CIS patients who smoked at time of CIS than in non-smoking CIS patients (67% vs 36%; p<0.01). And smoking at time of CIS was an independent predictor for CDMS diagnosis (HR: 2.3; p<0.01). Interestingly, non-smoking CIS patients who had a history of smoking did not have a higher risk of CDMS than CIS patients who had never smoked, suggesting that the harmful effects of smoking are reversible. Part 2 focused on the disease course after CIS in paediatric versus adult patients. In chapter 7, we compared disease course and clinical features at disease onset between 107 children and 276 adults who were followed after CIS. We divided the total group in three age groups (1-10, 11-17, 18-50 years). Lesion load on MRI was highest in the 11-17-year-old children. Children presented more often with polyfocal clinical symptoms than adults (32% vs 12%; p<0.01). We saw the highest rate of MS and CDMS conversion in the 11-17-year-old CIS group (84% vs 50%; p<0.01). This group also had the shortest time to MS diagnosis (HR: 3.2, 95%; p<0.01). Annualized relapse rates were evaluated in patients who were diagnosed with MS using a negative binomial regression model with log link. Without DMT, paediatric patients showed a higher ARR than adults (2.22 vs 0.91; p<0.01). However, after DMT administration ARR was the same for paediatric and adult MS patients (0.48 vs 0.40). These data showed a more inflammatory disease course in children after CIS than in adults, this argues for early initiation of DMT in 11-17-year-old children in line with current clinical practice in adult CIS patients. Based on our earlier finding in adult patients with CIS that high sCD27 levels were predictive for a future MS diagnosis, we evaluated this T-cell activation marker in 94 children with ADS in chapter 8. We found the same result as we found in adults described earlier in this thesis. High CSF sCD27 levels in paediatric ADS patients without encephalopathy were associated with a future CDMS diagnosis (HR 2.8 per 100U/mL increase in sCD27 levels; p<0.01). Children with ADS and encephalopathy did not differ in sCD27 levels from monophasic ADS patients without encephalopathy. However, ADS patients with encephalopathy had lower levels sCD27 than ADS patients without encephalopathy who were diagnosed with CDMS during follow-up. Also in children, sCD27 levels were associated with OCB and IgG index. It has been shown that neurofilament light chain (NfL), a marker for axonal damage, in CSF is predictive for MS diagnosis in adult CIS patients. In children, this was not studied yet. Therefore we investigated the predictive value of NfL in CSF from 65 paediatric and 88 adult CIS patients. In chapter 9, we first validated that high NfL levels were associated with CDMS diagnosis in adult patients with CIS (HR: 2.1; p=0.03). After that we demonstrated that NfL was also an independent predictor for CDMS diagnosis in paediatric CIS patients (HR: 3.7; p<0.01). Moreover, in the group with a future CDMS diagnosis, children had higher NfL levels than adults at time of CIS (4888 vs 2156pg/mL; p<0.01). We also showed that patients with a future CDMS diagnosis and T1-hypointense lesions on baseline MRI, which is a sign for axonal damage, had higher NfL levels than patients without T1-hypointense lesions, in both adults and children (adults: 3188 vs 1588pg/mL; p<0.01, children: 8920 vs 1668pg/mL; p<0.01). These results imply that children not only have a more inflammatory disease course than adults, as shown earlier in this thesis, but also that this young patient group has signs of neurodegeneration already early in the disease course. Chapter 10 discusses the main findings of this thesis and recommendations for future research.