Disease-modifying therapies and progressive multifocal leukoencephalopathy in multiple sclerosis: A systematic review and meta-analysis

Abstract

Background

High efficacy disease modifying therapies (DMT) in the management of Multiple Sclerosis (MS) have a favorable effect on relapse rate and disability progression; however, they can expose patients to significant risks, such as progressive multifocal leukoencephalopathy (PML).

Objective

The study aims to investigate prognostic factors that can determine outcome in MS-related PML patients.

Methods

We conducted a literature review and meta-analysis of 194 patients from 62 articles in PubMed, SCOPUS and EMBASE.

Results

Out of 194 patients (66.5% women, 33.5% men), 81% had progression in their EDSS score by at least 1 point from the time of PML diagnosis (EDSS-P group). The remaining patients had either stable or improved EDSS (EDSS-S group). In univariate analysis, older age at the time of PML diagnosis was associated with higher probability of disability accumulation and worsening of EDSS by at least 1 point (mean age = 44.8, p = 0.046). After adjusting for other variables, age at time of PML diagnosis remained a significant predictive variable in the multivariable logistic model (OR = 0.93, 95% CI: 0.88–0.99, p = 0.037). Natalizumab is the most commonly associated DMT linked to PML, followed by fingolimod and others including dimethyl fumarate, ocrelizumab, alemtuzumab. Among the different treatments used, no therapeutic agent was found to be superior in improving post-PML EDSS.

Conclusions

Younger age and lower JCV viral load at the time of PML diagnosis were associated with better outcome in MS-associate PML, while none of the PML therapies was superior over the others or associated with favorable outcome.

1. Introduction

Multiple sclerosis (MS) is a chronic demyelinating autoimmune and inflammatory disorder of the central nervous system (CNS) (Confavreux et al., 2000; Brownlee et al., 2017; Filippi et al., 2018). Demyelination followed by axonal damage through anterograde or retrograde degeneration leads to neurological disabilities and comorbidities (Confavreux et al., 2000; Filippi et al., 2018). Disease modifying therapies (DMT) have been widely and successfully implemented in clinical practice to prevent disease progression and decrease relapse rate (Dargahi et al., 2017).

Progressive multifocal leukoencephalopathy (PML) is a chronic demyelinating disorder of the CNS caused by JC-Virus (JCV), which is a ubiquitous virus that can become neurotropic and cause PML in rare patients with chronic cellular immunodeficiency (Tan and Koralnik, 2010; Khalili et al., 2007; Cortese et al., 2020). Studies have reported that serum antibodies against the JCV are present in approximately 50 to 90% of the general population (Bohra et al., 2017; Paz et al., 2018). A reactivated infection potentially leading to PML occurs nearly exclusively in immunocompromised individuals, thus making MS patients taking immunosuppressive drug therapy a particularly at-risk group (Bohra et al., 2017; D’Amico et al., 2016). PML presents with a wide range of neurologic deficits including limb ataxia, gait ataxia, diplopia, visual field defects, altered mental status, hemiparesis, and monoparesis. PML is a white matter disease of the brain, so its clinical presentation depends on the anatomic location of white matter lesions (Tan and Koralnik, 2010; Du Pasquier et al., 2004). The first reported case of DMT-associated PML dates back to 2005, and was associated with a combination of natalizumab and interferon beta-1 (Kleinschmidt-DeMasters and Tyler, 2005). Many other cases of PML have been reported as a complication of DMT, including natalizumab (Kleinschmidt-DeMasters and Tyler, 2005; Lindå et al., 2009; Dahlhaus et al., 2013), fingolimod (Nakahara et al., 2019; Berger et al., 2018), dimethyl fumarate (DMF) (Motte et al., 2018), alemtuzumab (Gerevini et al., 2019), and ocrelizumab (Sul et al., 2020).

Over the years, several therapeutic agents have been used in the management of PML including nucleoside analogues (cytarabine, cidofovir, ribavirin, adenine arabinoside, iododeoxyuridine), CCR5 antagonist (maraviroc), cytokines (interferon gamma), enzyme inhibitors (Camptothecin or topotecan), and JCV receptor blockers (5HT-2a receptor antagonists like chlorpromazine and mirtazapine) (Khalili et al., 2007; Pavlovic et al., 2015). Other commonly used therapies are intravenous immunoglobulin (IVIG), intravenous methylprednisone (IVMP), plasmapheresis, immunoadsoiption (IA), and antimalarial (mefloquine). Less frequently, agents such as granulocyte colony stimulating factor (filgrastim), monoclonal antibody (pembrolizumab) are used (Pavlovic et al., 2015; Castle and Robertson, 2019).

This meta-analysis aims to examine prognostic factors associated with PML, such as age, type of DMT, number of JCV copies in CSF at the time of diagnosis, treatment options for PML, and their impact on post-PML outcomes. Results of this study may influence treatment decisions and facilitate discussions with patients and families.

2. Material and methods

2.1. Study design

Using the keywords “PML treatment in MS” and “Progressive multifocal leukoencephalopathy and Multiple Sclerosis”; we searched PubMed, EMBASE, and Scopus from January 01, 2005, to December 31, 2020. Two reviewers independently performed the literature search and missing data was sought by discussion. We used the preferred reporting items for systematic review and meta-analysis (PRISMA) for the study (Moher et al., 2009). The review protocol was registered with PROS-PERO with registration number CRD42020182453.

Based on the search criteria, we found a total of 786 articles within PubMed (n = 399), EMBASE (n = 217), and Scopus (n = 170), of which 330 were identified as duplicates. We screened 379 articles based on their titles and abstracts and reviewed 103 full-text manuscripts in accordance with our study objective (refer to Fig. 1). In accordance with our inclusion and exclusion criteria, we identified 62 articles (194 patients) out of which 17 were case series (refer to Table S1), 44 were case reports and 1 was a retrospective systemic review. We then conducted a systemic review and quantitative meta-analysis of these sources. During the search, all results from the search algorithm were double-checked carefully to avoid duplicate or overlapping publications.

Fig. 1.

PRISMA flow diagram of systemic review. The flow diagram depicts the flow of information through the different phases of the systematic review. It maps out the number of records identified, included and excluded, and the reasons for exclusions.

2.2. Inclusion criteria

The inclusion criteria for our review included: a) confirmed diagnosis of PML based on AAN guidelines (Berger et al., 2013), b) MS diagnosis based on McDonald criteria (2005, 2010 and 2017, based on the time of the publication of each article) (Polman et al., 2005; Polman et al., 2011; Thompson et al., 2018), c) age > 18,d) PML cases associated only with FDA-approved DMT for MS.

2.3. Exclusion criteria

The exclusion criteria included: a) PML cases not related to DMT in MS, b) studies that did not specify the treatment of PML, c) PML cases related to off label use of DMT such as rituximab and cyclophosphamide, d) duplicates, and e) lack of EDSS documentation at the time of PML diagnosis and post-PML treatment.

2.4. Quality assessment

The critical appraisal checklist for case reports provided by the Joanna Briggs Institute (JBI) was used to assess the overall quality of case series and case reports (Joanna, 2019).

2.5. Data acquisition

Following identification of cases of interest, we extracted information for each patient: study type, publication date, age, gender, type and duration of MS medications, history of previous DMT, EDSS score at baseline, at time of PML diagnosis and following PML treatment, CSF JCV copies/mL at the time of PML diagnosis, histopathology/autopsy of the brain where applicable, treatment regimen used for PML management, and mortality.

2.6. Statistical analysis methods

Disability progression was defined as an increase in the EDSS score by ≥1.0 following treatment for PML as compared to 3–6 months prior to PML diagnosis. The primary response variable was defined as a binary outcome variable based on the changes in EDSS score as follows: i) Patients who showed progression as measured by worsening of the EDSS scores ≥1.0 (EDSS-P), ii) Patients who demonstrated improvement and who either had stable EDSS (change in EDSS <1.0) or decrease in EDSS score after being treated for PML (EDSS-S). Data analysis was performed with patient-based data since the patient characteristics and the outcome variable were available for each patient in this literature review study. Statistical analysis was performed using SAS (version 9.2) and R software (version 3.6.3, R Foundation, Vienna, Austria). Pearson correlation test was used to assess the correlation between continuous variables. Fisher exact test and Wilcoxon rank-sum test were used in the univariate data analysis for categorical and continuous variables, respectively, while logistic regression model was used in the multivariate data analysis on EDSS-S vs EDSS-P, adjusting for any potential confounding variables. All statistical tests were two-sided and a p-value <0.05 was statistically significant.

3. Results

We summarize a total of 194 PML individual cases, 129 women (66.5%) and 65 men (33.5%). The majority of the patients (99%) had relapsing-remitting multiple sclerosis (RRMS), while only 1% had primary progressive MS (PPMS). The cases were categorized into two groups, i) EDSS-P (n = 158) which includes participants with disease progression as measured by increasing the Expanded Disability Status Scale (EDSS) scores ≥1.0, and ii) EDSS-S (n = 36) which includes participants who demonstrated either stability (change in EDSS <1) or improvement (decrease in EDSS by at least 1 point) after the PML treatment (EDSS-S) (refer Table 1).

Table 1.

Study participant characteristics with progressive Multifocal leukoencephalopathy (PML) on Multiple Sclerosis medications.

	Total	Patients with disease progression (EDSS-P)b	Patients with stable or improved disability scores (EDSS-S)b	p-value
		n = 158	n = 36
		Mean (SD)	Mean (SD)
Age	194	44.8 (9.2)	40.9 (9.7)	0.046a
Gender
Female	129	107 (82.9%)	22 (17.1%)	0.44
Male	65	51 (78.5%)	14 (21.5%)
Medication at the time of PML Diagnosis
Natalizumab	165	141 (85.5%)	24 (14.5%)	0.003
Fingolimod	20	11 (55%)	9 (45%)
Others	9	6 (66.7%)	3 (33.3%)
Duration of medication
≤ 24 months	22	14 (63.6%)	8 (36.4%)	0.038
> 24 months	171	143 (83.6%)	28 (16.4%)
Management of PML
Mirtazapine
No	88	70 (63.6%)	18 (20.5%)	0.58
Yes	104	86 (82.7%)	18 (17.3%)
Mefloquine
No	91	68 (74.7%)	23 (25.3%)	0.041
Yes	100	87 (87%)	13 (13%)
Plasmapheresis
No	51	24 (47.1%)	27 (52.9%)	<0.0001
Yes	141	132 (93.6%)	9 (6.4%)
Immunoadsorption
No	131	96 (73.3%)	35 (26.7%)	<0.0001
Yes	62	61 (98.4%)	1 (1.6%)
Methylprednisolone
No	45	37 (82.2%)	8 (17.8%)	0.99
Yes	147	119 (81%)	28 (19%)
Maraviroc
No	140	109 (77.9%)	31 (22.1%)	0.061
Yes	52	47 (90.4%)	5 (9.6%)
CSF JCV DNA copies (copies / mL) (Median = 100)c
< 100	57	41 (71.9%)	16 (28.1%)	0.022
≥ 100	113	98 (86.7%)	15 (13.3%)
Patients with IRISd
No	15	15 (100%)	0 (0%)	0.078
Yes	157	124 (79%)	33 (9.6%)

Abbreviations: CSF, Cerebrospinal fluid; JCV = John Cunningham Virus; IRIS, Immune reconstitution inflammatory syndrome; PML, Progressive multifocal leukoencephalopathy; EDSS, Expanded disability status scale.

^aWilcoxon Rank Sum test (for continuous variables) and Fisher exact test (for categorical variables) were applied to assess statistically significant difference. p value reported in the table are measured using Fisher’s Exact Test unless otherwise specified.

^bEDSS-P stands for participants which showed progression of the disease as measured by worsening of the Expanded Disability Status Scale (EDSS) scores ≥1.0, and EDSS-S stands for participants which demonstrated either stability (change in EDSS <1) or improvement in EDSS score (≤ 1) after the PML treatment (EDSS-S).

^c24 patients did not have CSF JCV results reported.

^d22 cases IRIS was not documented or not available.

The mean age at the time of PML diagnosis in the EDSS-P group was 44.8 ± 9.23 years, whereas in EDSS-S was 40.9 ± 9.73 (p = 0.046 Wilcoxon stun rank test) (Fig. 2B and Table 1). There was a statistically significant correlation between age and change of EDSS (Pearson correlation ρ = − 0.26, p < 0.001), which implies better outcome in a younger population (Fig. 2A). There was no statistically significant association between gender and disease progression. In particular, of the 129 females (66.5%) included in our study, 107 of them (82.9%) were found to have a disability accumulation of ≥1 point in EDSS (EDSS-P) in comparison to 78.5% for males (p = 0.44, Fisher’s exact test) (refer Table 1). The mean duration of DMT exposure until diagnosis of PML was 36 months (SD = 17) with the majority of participants (88%) had a duration of exposure >24 months.

Fig. 2.

(A) Scatter plot distribution of change in expanded disability status scale (EDSS) at Progressive Multifocal Leukoencephalopathy (PML) diagnosis and post-PML management versus age of the participant. Y-axis represents age of the participant in years and X-axis represents the absolute change in EDSS score from the time of diagnosis to post PML management. Blue line indicates fitted regression line. (B) Distribution Plots representing the Mean Age and Standard Deviation of EDSS-S and EDSS-P groups. (C) Bar Graph representation of Percentage EDSS changes vs. JCV DNA Copies. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The majority of the reported cases of EDSS-P were associated with the use of natalizumab (141 of 165, 85.5%), in comparison to fingolimod (11 of 20, 55%) and other medications, such as dimethyl fumarate, ocrelizumab, alemtuzumab (6 of 9, 66.7%) in EDDS-P group. There was a statistically significant association between medications and EDSS-P (p = 0.003, Fisher’s exact test). Most of the natalizumab-associated PML patients progressed, followed by “others”, while fingolimod-associated PML patients showed less progression compared with the other groups (Table 1).

In the multivariate data analysis, age at the time of PML diagnosis was still a significant predictive variable in the multiple logistic model (OR = 0.93, 95% CI: 0.88–0.99, p = 0.037), adjusting for all potential confounding variables (refer Table 2).

Table 2.

Multivariable logistic model on EDSS-S vs EDSS-P, adjusting for all potential confounding variables.

Variables	Estimate	Std Error	Odds ratio (OR)	95% CI of OR		z-value	p-value
Age	−0.68	0.033	0.93	0.88	0.99	−2.085	0.037*
Fingolimod vs Natalizumab	0.929	0.963	2.53	0.38	16.70	0.965	0.335
Others vs Natalizumab	0.589	1.397	1.80	0.12	27.83	0.422	0.673
Duration of MS medication	−1.119	0.891	0.33	0.06	1.87	−1.255	0.209
Oral steroid	1.368	1.035	3.93	0.52	29.89	1.321	0.186
Ribavirin	2.563	2.268	12.98	0.15	100+	1.130	0.258
Methyl prednisone	1.332	0.831	3.79	0.74	19.32	1.601	0.109
Immunoadsorption	−3.606	1.240	0.03	0.00	0.31	−2.908	0.004*
Plasmapheresis	−2.590	0.629	0.08	0.02	0.26	−4.117	<0.001*
IVIG	1.300	1.066	3.67	0.45	29.67	1.219	0.223
Cytarabine	13.1	1455.4	100+	0.00	100+	0.009	0.993
Maraviroc	−0.894	0.684	0.41	0.11	1.56	−1.307	0.191
Mefloquine	−1.359	0.716	0.26	0.06	1.05	−1.898	0.058
Mirtazapine	−0.278	0.721	0.76	0.18	3.11	−0.385	0.700

Abbreviations: IVIG - Intravenous immunoglobulin;EDSS, Expanded disability status scale.

EDSS-P stands for participants which showed progression of the disease as measured by worsening of the Expanded Disability Status Scale (EDSS) scores ≥1.0, and EDSS-S stands for participants which demonstrated either stability (change in EDSS <1) or improvement in EDSS score (≤ 1) after the PML treatment.

^*p-value <0.05 underlines significant difference.

Overall CSF JCV DNA (copies per ml) were available in 170 patients. Ninety-eight cases in the EDSS-P group (n = 98 of 113, 86.7%) and fifteen cases in the EDSS-S (13.3%) group had CSF JCV DNA >100 copies/ml. Forty-one patients in EDSS-P group (n = 41 of 57, 72%) and sixteen patients in EDSS-S group (n = 16 of 57, 28%) had <100 copies/ml CSF JCV DNA. There was significant difference in outcomes (p = 0.02, Fisher’s exact test) between patients with ≥100 copies/ml and patients with <100 copies/ml (refer Table 1, Fig. 2C).

Treatment options used in the management of PML in our cohort are summarized in Table 1. Plasmapheresis, methylprednisolone, mirtazapine, and mefloquine in various combinations were the most widely used. In multivariable analysis on EDSS-S vs EDSS-P, there was not a significant difference in the outcomes between treatments given for PML, such as plasmapheresis, or a combination of plasmapheresis with IV methylprednisolone, immunoadsoiption with IV methylprednisolone, or a combination with either anti-malarial (mefloquine), antivirals (cytarabine, ribavirin, maraviroc, cidofovir), or 5 HT2A receptor antagonist (mirtazapine). There was no particular treatment that significantly improved the EDSS outcome based on the analysis performed for PML (refer Table 2). We also noted that plasmapheresis (p < 0.001, Fisher exact test) and immunoadsorption (p < 0.001, Fisher exact test) were not associated with improvement in EDSS scores (Tables 1 and 2).

In terms of fatalities there were in total 25 fatal outcomes among 194 cases (12.88%). The incidence of fatal outcomes was 3.8% (3/78) in patients who were below the mean age of 44 years and 19% in patients who were above or equal to the mean age of 44 (p = 0.002, Fisher’s exact test in Table 3b). There was a significant difference in post-PML treatment EDSS scores between fatal and non-fatal groups, with mean EDSS of 10 (SD 1.0) versus 5.59 (SD 2.23) (p < 0.001, Wilcoxon sum-rank test). There was a significant difference on EDSS score change from baseline to post-PML treatment between fatal and non-fatal cases (p < 0.001, Wilcoxon stun rank test), with a mean of 4.5 (SD 1.95) and 2.35 (SD 1.39) respectively. Similarly, there was a statistical significance in EDSS change from the time of PML diagnosis to post-PML between fatal and non-fatal group (p < 0.001), with a mean of 4.54 (SD 2.19) and 0.56 (SD 1.47) respectively (refer Table 3a).

Table 3b.

Distribution of mortality in terms of age variation.

Age	Sample size	Non-Fatal cases	Fatal cases
Age < 44	94	91 (96.8%)	3 (3.2%)	0.001
Age ≥ 44	133	111 (83.5%)	22 (16.5%)

Table 3a.

Distribution of EDSS scores in PML patients based on mortality outcomes.

	Non-fatal cases (n = 189)		Fatal cases (n = 25)		Wilcoxon Rank Sum test (p value)
	Mean (SD)	Total	Mean (SD)	Total
EDSS following management PML	5.48 (2.29)	189	10 (0)	25	<0.001
EDSS at PML diagnosis	4.92 (1.93)	189	5.32 (2.07)	25	0.55
Change in EDSS at baseline and Post PML treatment	2.33 (1.4)	143	4.5 (1.95)	15	<0.001
Change in EDSS at PML diagnosis and post PML treatment	0.56 (1.47)	189	4.54 (2.19)	25	<0.001

Abbreviations: EDSS - Expanded disability status scale, PML - Progressive Multifocal Leukoencephalopathy.

4. Discussion

We conducted a literature review and meta-analysis of 194 MS-related PML cases to examine the factors that affect the clinical outcome of PML-affected MS patients. We specifically defined clinical outcomes as progression of disability, fatality or recovery. All participants had been treated with one or more immunomodulatory medications associated with PML. Immunomodulators included natalizumab, fingolimod, dimethyl fumarate, alemtuzumab and ocrelizumab.

DMT induce immune system changes that mimic immunosenescence during treatment and even after drug cessation, further increasing the risk of PML, as effects of several DMT are not readily reversible (Aiello et al., 2019; Foley et al., 2016). As expected, natalizumab was associated with the majority of cases with poor outcomes, followed by fingolimod (Table 1). As of September 2020, natalizumab appears to be the leader of DMT-induced PML with 839 reported cases and an overall global incidence of 3.94 per 1000 patients. Fingolimod is next in line with 40 reported cases of treatment-associated PML. Dimethyl fumarate-associated PML was reported in 11 patients with risk estimates for fingolimod (0.13 per 1000 patients) and risk estimate for dimethyl fumarate (0.02 per 1000 patients) (Fox et al., 2020; Jordan et al., 2020), followed by 1 case of alemtuzumab-associated PML and 1 case of ocrelizumab-associated PML (Table 4). Lorefice L et al. described a patient who developed PML during a 3-month teriflimomide therapy, who had previously been on natalizumab. This case was attributed to natalizumab rather than teriflimomide (Lorefice et al., 2017). Similarly, nine cases of PML were reported in patients on ocrelizumab, however, only one case occurred in a patient who had not received prior DMT, while 7 of these patients received natalizumab previously and one patient received fingolimod. Even though non- statistically significant, there was a trend between PML diagnosis and duration of DMT exposure, with a higher percentage of cases being diagnosed after 2 years of treatment. The mean duration of DMT exposure till diagnosis of PML was 36 months (SD = 17) with 88%(n = 171) cases had exposure ≥ 25 months. As per this systematic review the majority of the patients (99%) had RRMS, while only 1% had PPMS. Among two cases of PML reported in PPMS, the first patient was a 78-years-old male treated with ocrelizumab and never exposed to any DMT by Patel A et al. (Patel et al., 2021). The second case of PPMS, reported by Bahamoori M et al., was a 64-years-old male was treated with Dimethyl Fumarate (Baharnoori et al., 2016).

Table 4.

Summary of progressive Multifocal Leukoencephalopathy (PML) cases with different disease modifying therapies.

Medication	Total Cases of PML
Natalizumab	839a
Fingolimod	40b,e
Dimethyl Fumarate	11a,e
Ocrelizumab	1c,e
Alemtuzumab	1d,e

The data is not published and was available on request by

^aBiogen,

^bNovartis,

^cGenentech and

^dSanofi till September 2020.

^eCases of PML not attributable to prior natalizumab exposure.

We demonstrated a modest, but significant correlation between age and outcome. Patients who were equal to or older than 44 years of age were associated with worse outcomes compared to younger patients (p = 0.037). The cause of this association is complex and multifactorial. Aging may lead to a reduced number of naïve T cells and increased terminally differentiated late effector memory T cells (T_EMRA) (Mills and Mao-Draayer, 2018). Naïve T cells have a reduced number of IL-2 receptors causing impairment in effector cells differentiation along with poor activation of the T-cell receptor signaling mechanism and thereby collectively hamper viral clearance. Reduced T-cell priming and activation by antigen-presenting cells (APCs) is mostly observed. T-cell-mediated immunity plays an essential role in response against pathogens. Aging is associated with a decline in immune system function due to the decrease in the phagocytic activity of macrophages and neutrophils, reduced antigen-presenting activity of dendritic cells as well as a reduction in the number of monocytes. The thymic involution at puberty also contributes to this decline as there is a reduction in the T cell receptor (TCR) repertoire and accumulation of senescent T cells that are not functional Studies have reported impaired dendritic cell maturation, which leads to reduced antigen uptake and presentation capacity (Dahlhaus et al., 2013; Longbrake et al., 2015). Loss of costimulatory signal with APCs can also contribute to impaired T-cell activation seen with aging; for example, loss of CD28 in auto-immune diseases suggests premature aging of the immune system (Aiello et al., 2019). In addition, aging and DMT share some common pathways contributing to an increased risk of PML. Both aging and DMT can reduce the number and functionality of the lymphocytes which can predispose to the higher risk of PML secondary to lymphopenia. DMT may further impair immune response in older patients which makes them vulnerable to chronic infections. Therefore, the mechanism by which DMT and aging impair the functions of APCs is multifactorial, and most likely due to poor functioning of the immune surveillance cellular mechanism, despite an adequate amount of T cells (Foley et al., 2016; Mills and Mao-Draayer, 2018).

We also demonstrated a modest but significant correlation between JCV viral load and outcome. Few studies have highlighted the association of high viral loads in CSF at the time of diagnosis with poorer outcomes (Dong-Si et al., 2015; Hoepner et al., 2017; Sanjo et al., 2019; Vermersch et al., 2011). Our results are in line with previous findings and emphasize the importance of an early diagnosis and intervention when the viral load is still low. Current recommendations for natalizumab-treated patients suggest anti-JCV antibody testing every 6 months for anti-JCV antibody-negative patients or antibody-positive patients with index levels ≤1.5. For those with an anti-JCV antibody index >1.5, further index testing is not mandatory (McGuigan et al., 2021). However, with newer DMTs being commonly used, PML risk evaluation should be considered in them as well based on theoretical possibilities. Further studies evaluation biomarkers and guidelines predicting PML risks regarding a wider range of DMTs need to be done (Wiendl, 2016; Subei and Ontaneda, 2015; Klotz et al., 2019).

Furthermore, we analyzed therapies used for the management of PML. Surprisingly, we found non-significant correlation between any specific therapy and improvement of PML-associated disability (Table 2). In the univariate analysis, both oral steroid and mefloquine were significantly associated with PML improvement. However, after adjusting for co-occurring medications such as cytarabine, maraviroc, mirtazapine, IVIG, etc., neither oral steroids nor mefloquine achieved statistical significance (Tables 1 and 2). We also noted that plasmapheresis and immunoadsorption were not associated with improvement in EDSS scores (Tables 1 and 2). Taking into consideration that the patients with the most severe presentation and subsequently worst prognosis were treated with plasmapheresis and immunoadsorption, we need to be extremely cautious to interpret the above results and question the efficacy of the above regiments in the treatment of DMT-associated PML. Our analysis was not adjusted for the severity of PML, given the complexity of the PML presentation, the lack of a detailed description of severity of symptoms and the shortcomings of the EDSS. Despite the results of a large study by Landi D et al. demonstrating the lack of evidence of a beneficial effect of PLEX treatment in outcomes of PML patients, plasma exchange has been used widely over the years by different practices with variable success (Landi et al., 2017).

Interestingly, immune reconstitution inflammatory syndrome (IRIS) can be seen after treatment of PML (Tan et al., 2011). In the study cohort, there were 157 cases reported to have IRIS, 124 (79%) of which were patients in EDSS-P group and 33 (9.6%) were patients in the EDSS-S group. Among the 157 IRIS cases 19 were fatal and 138 were non-fatal (refer to Tables 1 and 3b). In the remaining cases, IRIS information was not available. Among IRIS 157 cases, treatment information was available for 153 cases with 130 cases of IRIS were treated with IVMP, with 112 (86.2%) cases of IRIS in EDDS-P group and 18 (13.8%) in EDDS-S group. The remaining cases of IRIS were treated either with PLEX, or combination of PLEX and IVMP, or combination of PLEX, IVMP, and IVIG or maraviroc or mirtazapine (refer to Table S2).

Lastly, in our review, older age and greater change in EDSS scores between baseline and post-PML treatment as well as between the time of PML diagnosis and post-PML treatment were significant predictors of mortality (refer to Tables 3a and 3b). Unfortunately, no EDSS score was available 3 or 6-month post-PML diagnosis.

Our study had several strengths. This is one of the few comprehensive reviews focused on comparing the clinical presentation, management and outcomes in MS-related PML. We used strict inclusion and exclusion criteria and robust statistics. We identified predictors of disability progression and mortality. The cohort was selected through a comprehensive search of databases using a systematic search strategy. However, despite the set criteria, there is a possibility of missing out studies because of the incidence of new cases of PML. We acknowledge that the results should be considered in light of several limitations.

One of the limitations is the small sample size of the cohort, possibly due to the limited data published on the area of focus. Data reporting discrepancy was also observed, as all the focused variables were not reported by all the studies. Several large-scale studies were excluded based on the exclusion criteria, which might have caused estimation bias in this result. Furthermore, not all PML cases were reported and of those reported, not all information was available, such as clinical characteristics, management, and outcome. JCV serological measurements and risk stratification guidelines with natalizumab use have significantly decreased the incidence of PML (Vukusic et al., 2020). Given that the current guidelines mainly for patients on natalizumab, it is necessary to develop a biomarker-linked PML risk assessment for newer and emerging DMTs other than natalizumab. There has been evidence to suggest that higher polyoma antibody titers can be associated with an increased risk of viremia and viral load. Therefore, a higher JCV titer can reflect a general increased susceptibility for PML under conditions that can increase risk such as older age (Reuwer et al., 2017).

Our analysis reflects the inefficiency of the current treatment modalities for PML, as a higher number of the treated patients progressed rather than stabilized. Although we identified a full spectrum of studies worldwide, differences in therapeutic modalities for the management of PML and the lack of randomization might have affected our results. Subsequently, our results should be cautiously interpreted. As there are not standardized treatment guidelines for PML prevention and management, there is an unmet need to be developed. Large-scale, prospective, randomized, multicenter studies are needed to validate these results and lead to a better understanding and management of DMT-related PML in the MS population.

5. Conclusion

In conclusion, younger patients and a lower number of CSF JCV copies are associated with better outcomes and prognosis, though the most efficient treatment strategies for PML are yet to be determined. We hope that this finding will alarm clinicians to evaluate patients for possible PML as early as possible, to monitor frequently and carefully patients who are treated with high-risk DMT and possibly reduce the length of exposure to natalizumab, especially in older patients.

Abbreviations:

PML

Progressive Multifocal Leukoencephalopathy

MS

Multiple Sclerosis

RRMS

Relapsing Remitting Multiple Sclerosis

PPMS

Primary Progressive Multiple Sclerosis

EDSS

Expanded Disability Status Score

IRIS

Immune reconstitution Inflammatory Syndrome

DMT

Disease modifying therapy

JCV

JC Polyomavirus

CSF

Cerebrospinal fluid

CNS

Central nervous system

PLEX

Plasmapheresis

IA

Immunoadsorption

IVIG

Intravenous immunoglobulin

AAN

American Academy of Neurology

CCR5

C-C chemokines receptor type 5

T_EMRA

terminally differentiated late effector memory T cells

APC

Antigen presenting cell

DMF

Dimethyl fumarate