Trends in

OPEN ACCESS Immunology

Opinion

Multiple sclerosis: role of meningeal lymphoid

aggregates in progression independent of
relapse activity
1,2,
Richard M. Ransohoff *

The emphasis on mechanisms driving multiple sclerosis (MS) symptomatic wors- Highlights
ening suggests that we move beyond categorical clinical classifiers such as Descriptive categorical classifiers of mul-
relapsing–remitting MS (RR-MS) and progressive MS (P-MS). Here, we focus on tiple sclerosis (MS) – relapsing–remitting,
chronic-progressive – do not corre-
the clinical phenomenon progression independent of relapse activity (PIRA),
spond to mechanisms that explain
which begins early in the disease course. PIRA occurs throughout MS, becoming symptomatic worsening.
more phenotypically evident as patients age. The underlying mechanisms
for PIRA include chronic-active demyelinating lesions (CALs), subpial cortical Persons with MS simultaneously exhibit

demyelination, and nerve fiber injury following demyelination. We propose relapse activity with clinical attacks and
new magnetic resonance imaging (MRI)
that much of the tissue injury associated with PIRA is driven by autonomous lesions, as well as progression indepen-
meningeal lymphoid aggregates, present before disease onset and unresponsive dent of relapse activity (PIRA) to varying
to current therapeutics. Recently, specialized magnetic resonance imaging (MRI) degrees throughout the disease course.
Relapse activity has been successfully
has identified and characterized CALs as paramagnetic rim lesions in humans, treated, while PIRA has not.
enabling novel radiographic–biomarker–clinical correlations to further understand
and treat PIRA. PIRA is mechanistically associated with
chronic-active lesions (CALs), cortical
demyelination, and neurite degeneration.
Meningeal lymphoid aggregates are
causally associated with CALs and corti-
MS progression: an enduring conundrum cal demyelination. Conventional MRI
does not detect pathology associated
MS is a common cause of neurological disability in young adults, and has a worldwide preva- with PIRA, except as tissue atrophy.
lence of ~2.8 million, with a significant increase in the past decade [1]. MS is an autoimmune
disease that selectively affects central nervous system (CNS) myelin (see Glossary), with sec- Recent advances using MRI to detect
CALs may enable a personalized medi-
ondary damage to nerve fibers. This conclusion comes from investigating underlying genetic
cine approach to understanding and
architecture [2,3], neuropathology, and immunology of the disease. The disease was identified treating PIRA.
as a distinct nosological entity >150 years ago [4]. Even then, it was appreciated that MS is ini-
tially characterized by episodic neurological symptoms that appear abruptly and usually resolve Significance
(Box 1). After 10–25 years, this relapsing phase is typically followed by often-relentlessly-
Using disease-modifying therapies
increasing symptomatic severity, termed progression (Box 1). This unusual tempo of symptom (DMTs) to suppress multiple sclerosis
evolution prompted efforts at explanation, whose character has changed as research into the (MS) relapse pathology, a concurrent
understanding and treatment of MS has borne fruit. Recent careful delineation of the MS clinical process in many patients, termed
progression independent of relapse
course has highlighted progression independent of relapse activity (PIRA). Discovery that activity (PIRA), has been unmasked.
PIRA occurs early in persons with MS prompted investigation into its etiological mechanisms, Current DMTs are only modestly
which are distinct from those that drive relapsing pathology. In this Opinion, a mechanistic effective in moderating PIRA, and con-
aspect of MS progression is described; namely, that chronic inflammatory tissue injury is a sig- ventional magnetic resonance imaging
(MRI) and/or animal models of inflam-
nificant contributor to PIRA and is driven by meningeal inflammatory aggregates. We propose matory demyelination, have not been
that investigating biomarker and imaging correlates of PIRA will be invaluable support for ther- useful for understanding PIRA. This
apeutic development for the full range of MS clinical manifestations. This hypothesis brings to- Opinion advances the hypothesis that
a novel MRI technology for detecting
gether observations coming from rigorous longitudinal clinical cohorts, molecular and cellular
chronic-active brain lesions as
neuropathology, and MRI, with relevance for developing new therapies.

266 Trends in Immunology, April 2023, Vol. 44, No. 4 https://doi.org/10.1016/j.it.2023.02.002

© 2023 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Trends in Immunology
OPEN ACCESS

Box 1. MS clinical phenotypes paramagnetic rim lesions may be in-

Before DMTs became available in 1993, MS disease course was characterized through population-based longitudinal co- structive for clarifying mechanisms
hort studies, based in Lyon (France), Gothenburg (Sweden), and London, Ontario (Canada). Data from these studies were driving PIRA.
analyzed critically and summarized by Confavreux and Compston [8]. MS typically (~85%) begins with episodes of abrupt
worsening (relapse) succeeded by recovery and clinical quiescence, termed the relapsing–remitting phase. Onset typically
occurs during the 20s or early 30s of human adults, with women affected at a 2:1 ratio [8]. Within 20 years, about 80% of
RR-MS patients enter the phase of P-MS, usually presenting with fewer relapses and steady worsening, termed SP-MS. In 1
Third Rock Ventures, Boston, MA, USA
~15% of patients, the first MS symptoms constitute slow progression (termed PP-MS). Of this minority, as many as 40% 2
Abata Therapeutics, 100 Forge Road,
will experience subsequent relapses, suggesting an underlying commonality of pathogenesis in PP-MS and SP-MS. Con-
Suite 200, Boston, MA 02472, USA
sistent with this concept, the ages at which progression begins (typically late 40s to early 50s) and rates of progression do
not distinguish PP-MS from SP-MS [12]. Thus, the occurrence and severity of relapses – somewhat against expectation –
do not affect the overall timing or pace of disability accumulation in P-MS [8].

*Correspondence:
[email protected],
The curious relationship between relapse and progression in MS [email protected]
Until recently, it was considered tenable to propose that an uncharacterized neurodegenerative (R.M. Ransohoff).
process underlies P-MS; this notion was based on the lack of relationship between relapses –
which were explicitly inflammatory – and progression [5]. However, clarifying data came from
long-term follow-up of patients treated with low- or high-efficacy disease-modifying thera-
pies (DMTs), or left untreated; this report indicated that the likelihood of developing P-MS
within 5 years of follow-up was reduced in stepwise fashion by treatment with low- or high-
efficacy DMTs, whose evident effects were to suppress the occurrence of new inflammatory le-
sions detected by MRI, along with clinical relapses [6]. These observations were made possible
using high-efficacy DMTs that essentially eliminated relapse pathology – manifested as clinical
relapses and the occurrence of new lesions [gadolinium-enhancing (GdE) lesions or T2-
hyperintense lesions] on MRI (Figure 1) [7]. Therefore, the risk of developing P-MS was sur-
mised to correlate directly with the extent of relapse pathology occurring early during the dis-
ease course [8]. However, the link between relapse pathology and P-MS has remained
incompletely explained.

MS relapse pathology, both with respect to clinical attacks [9] and GdE-enhancing lesions [10]
lessens as patients age. Underlying explanations for this phenomenon are uncertain, and may be re-
lated to age-related changes in the immune system; some of them potentially MS specific [11]. Con-
versely, as patients age, their likelihood of entering the P-MS phase of disease increases [12].
Moreover, continued functional worsening in the absence of relapse pathology represents the
unsolved scientific problem of P-MS. One principal hypothesis to explain worsening during P-MS
is that demyelinated neurites are vulnerable to degeneration [13]. In autopsy tissue sections, neurite
degeneration is ten times increased in persistently demyelinated brain regions compared with unaf-
fected control white matter found in the same tissue sections [14]. By contrast, in remyelinated areas,
degenerating neurites are found approximately as frequently as in control white matter [14]. These
observations are more compelling since the pathological hallmark of degenerating neurites –
accumulation of amyloid precursor protein (APP) – is transient, lasting perhaps 1 month
[15,16]. Therefore, APP accumulation in these tissues represent neurites that degenerated within
approximately 1 month of death, highlighting that demyelinated neurites are continuously lost at
substantial numbers throughout the disease course.

Cortical demyelination, with its functional and neurodegenerative sequelae, is also associated
with progression. Subpial cortical demyelination (Figure 2) is specific for MS [17,18], and is topo-
graphically associated with lymphoid aggregates in the meninges [19]. Conventional MRI neither
reliably detects meningeal inflammatory aggregates nor subpial cortical demyelination; therefore,
their cumulative contributions to MS progression have been characterized mainly by MRI-based
quantification of cortical atrophy [20].

Trends in Immunology, April 2023, Vol. 44, No. 4 267

 Trends in Immunology
OPEN ACCESS

A group of investigators, on behalf of the International Advisory Committee on Clinical Trials in Glossary
Multiple Sclerosis, recently articulated a need to update classification of the clinical phenotypes Active progressive MS: disease
of MS and were joined by other expert commentators [21,22]. Traditional clinical classifiers course characterized by slowly-
(Box 1) are now viewed as complicating research into pathogenesis, as well as clinical trial worsening symptoms accompanied by
ongoing relapse pathology.
design and execution. The concern is most pressing in the context of therapeutic initiatives Amyloid precursor protein (APP):
for P-MS [21]. In place of categorical phenotypic classifiers, it is argued that MS clinical presen- substrate for fast axonal transport;
tations represent a continuous spectrum, resulting from interleaved pathogenic mechanisms accumulates at sites of cytoskeletal
that operate simultaneously and contribute differentially to MS clinical phenomena over time. disruption; used in immunohistochemical
studies as a sensitive indicator of axonal
Importantly, researchers propose that targeting mechanisms rather than clinical classifications injury.
will produce improved outcomes of trials of therapeutics for patients. Astrogliosis: glial cells responsible for
the maintenance of ionic and water
PIRA: flying under the radar balance, clearance of neurotoxic
concentrations of neurotransmitters
One line of research supporting this concept concerns PIRA [23–26]. For the purposes of this from synapses and support for
Opinion, PIRA is considered to comprise symptom worsening in the absence of relapse myelinating cells. The astrocyte reaction
pathology, which includes both clinical relapse and specific changes on conventional MRI imaging to loss of homeostasis entails decreased
function for some or all of these crucial
(new or enlarging T2-bright lesions or GdE lesions). During the early phase of MS, relapse pathology
activities.
dominates the phenotype. Nonetheless, natural history studies show that PIRA [23] occurs com- Blood–brain barrier: denotes a set of
monly, even as early as clinically isolated syndrome [27] – the initial stage of disease for most pa- attributes associated with endothelial
tients. However, major avenues to study MS, including conventional MRI and animal models, cells of the cerebral vasculature,
including tight intercellular junctions,
have largely been ineffective for understanding the CNS-compartmentalized pathological process
limited transcytosis, and efflux pumps.
by which meningeal lymphoid aggregates may drive the formation of chronic active lesions Under physiological conditions, the BBB
(CALs) and subpial cortical demyelination to produce PIRA. restricts entry of macromolecules and
cells into the CNS parenchyma.
Chronic active lesions (or mixed
The most convincing evidence for the clinical impact of PIRA comes from studies using high- active/inactive lesions): can be
efficacy DMTs, which virtually abolish relapse activity, and unmask PIRA. This research includes categorized according to the presence
analysis of clinical trial data in MS [e.g., ocrelizumab, CD20 monoclonal antibodies; [26]; OPERA I or absence of ongoing demyelination by
(NCT01247324)i and OPERA II (NCT01412333)ii], as well as registries that systematically track identification of myelin components
within phagocytic macrophages.
patients using high-efficacy DMTs (e.g., natalizumab, α4-integrin monoclonal antibodies; [25]). Inactive lesions show gliosis (astrocytic
In the clinical trial setting (OPERA I and OPERA II) where concurrent placebo control groups and microglial reaction) with varying
were available for comparison, high-efficacy DMTs demonstrated only a modest effect on PIRA degrees of infiltrating lymphocytes,
without ongoing demyelination. CALs
[26]. The two OPERA trials were identical and showed robust benefit: the primary outcome of an-
show myelin components within
nualized relapse rate – an indicator of relapsing pathology – was reduced by >45% when com- phagocytic macrophages, concentrated
pared to IFN β1a injections in patients [28]. at the lesion edge. Because myelin
components within macrophages turn
over within 2–4 weeks, demyelination is
During the progressive phase of MS, PIRA dominates the disease phenotype [29]. Relapse
deemed present at the time of tissue
pathology continues in a minority of patients [both primary progressive MS (PP-MS) and secondary sampling (autopsy, biopsy).
progressive MS (SP-MS)] (Box 1), giving rise to the category of active progressive MS [30]. Post- Disease-modifying therapies
hoc subtype meta-analysis showed that the ability of high-efficacy DMTs [e.g., ocrelizumab, (DMTs): modify the natural history of
MS. The first DMT was IFN-β1b. There
siponimod (a modulator of S1P1)] to slow disability progression in P-MS occurred in patients
are more than 20 approved DMTs for
with ongoing relapse pathology, as demonstrated by the comparison of hazard ratio for pro- MS treatment.
gression in patients with or without ongoing relapse pathology [31]. Inactive progressive MS Gadolinium-enhancing (GdE) lesion:
characterizes P-MS patients who lack relapse pathology; a small fraction of these patients remain Gd is a paramagnetic contrast agent,
used to identify foci of BBB disruption
clinically stable, while the remainder worsen steadily, consistent with a PIRA-dominant phenotype with MRI. After intravenous injection, Gd
[30,32]. In summary, relapse pathology and PIRA are manifestations of distinct mechanisms of tis- is excluded from the brain parenchyma
sue injury. They coexist during MS, with relapse pathology dominating the phenotype in initial unless BBB function is compromised, as
phases of disease and PIRA subsequently taking the primary role. in acute MS lesions. GdE lesions
represent the earliest stage of most MS
lesions. Within approximately 1 month
From PIRA to CALs, to paramagnetic rim lesions after its first appearance, GdE resolves.
Relapse pathology is characterized by the appearance of new inflammatory lesions on MRI GdE lesions evolve into one of several
distinct types of lesions during the
(Figure 1), with evident disruption of blood–brain barrier integrity; indeed, new lesions are

268 Trends in Immunology, April 2023, Vol. 44, No. 4

Trends in Immunology
OPEN ACCESS

temporally related to relapses, indicating the clinical relevance of MRI findings [33]. Recently, spe- subsequent few months, representing
varying forms and extent of tissue injury.
cial MRI pulse sequences were used to detect a paramagnetic rim (Figure 1) in a subset of chronic
Inactive progressive MS: indicates a
MS lesions (termed paramagnetic rim lesions; PRLs) [34]. Moreover, radiographic–pathological disease course typified by continual
correlation studies showed that CALs (also termed mixed active/inactive lesions), at a first level of worsening without relapse pathology.
approximation, were those detected by MRI as PRLs [21,35], allowing researchers to establish Intrathecal: intrathecal space is
delimited by the meninges, where CSF is
relationships between CALs and clinical course during life. produced and circulates.
Meninges: membranes associated
Are CALs related to PIRA? CALs are cardinal features of MS neuropathology [36], characterized by with the extraparenchymal structures of
an inactive lesion core and ongoing macrophage-mediated demyelination at the lesion edge. In a the CNS. Pachymeninges indicates the
dura mater, applied to the undersurface
large, rigorous brain-bank-based study, CALs were present in 78% of MS autopsy cases (similar of the skull. Leptomeninges, composed
to the overall prevalence of PRLs [37]) and were more common in P-MS than in RR-MS, suggest- of the arachnoid membrane and pia-
ing that tissue injury caused by CALs worsens the clinical course [38]. Furthermore, in a prior study, arachnoid are associated with the
undersurface of the dura and the surface
injured axons were detected in CALs ~100 times more frequently than in control white matter, sug-
of the parenchyma, respectively. Arach-
gesting that chronic demyelination is associated with nerve fiber injury, albeit less dramatically than noid and pia define the intrathecal com-
that seen in acute demyelinating lesions [14]. Also, in clinical/postmortem correlations, CALs partment and confine the cerebrospinal
showed a direct, strong quantitative relationship with worse clinical course during life, as evidenced fluid flow pathways.
Myelin: In the CNS, myelin is produced
by shorter time from onset to requiring a cane or walker for ambulation [38].
and maintained by oligodendrocytes;
multilamellar structure composed of lipid
These observations suggest that studies focused on PRLs – the MRI correlate of CALs – might be along with specialized proteins largely
robustly informative of the pathological basis for PIRA. Indeed, the evaluation of PRLs, along with found only within myelin. The myelin
membrane serves multiple functions: it
radiographic–clinical correlation, establishes the temporal appearance of CALs during life and en-
facilitates rapid conduction along nerve
ables researchers to define real-time relationships between numbers of CALs and disease course fibers through its insulating function;
[34]. The clinical relationships of PRLs have supported the value of this line of research [35,39]. restricts access to the neurite plasma
membrane; myelinated fibers obtain
protection from noxious stimuli and
PRLs are specific for MS [40] and show a direct semiquantitative relationship with clinical severity metabolic support from the myelinating
[34]. PRLs arise from new GdE lesions, as a subset of T1-dark lesions (Figure 1) [34,41,42]. Im- oligodendrocyte.
portantly, PRLs increase in volume over time [43], whereas individual T1-dark lesions that lack Neurites: axons and dendrites are
PRLs tend to decrease in volume over time [34]. During the first decade of a typical course of projections from neuronal cell bodies,
collectively termed neurites and involved
MS, PRLs are less likely to result from new GdE lesions on a per-lesion basis [41]. Later in the dis- in neuron-to-neuron communication.
ease course, PRLs occur more likely as an outcome of new GdE lesions, although the frequency Neurofilament light chain (NfL):
of new GdE lesions at that stage is less than during the first 5–10 years after onset [10]. PRLs cytoskeletal component of neurons.
With neurite injury, NfL is released into
show a median duration of ~7 years from appearance to vanishing [44]. Preliminary evidence sug-
extracellular interstitial fluid and can be
gests that current therapies do not promote resolution of PRLs [34,45]. These characteristics are quantified in CSF or blood. NfL shows
consistent with the proposal that PRLs – the MRI correlate of CALs – can contribute to the path- considerable promise as a biomarker of
ogenesis of PIRA (see Outstanding questions). nervous system pathology, being
extensively applied as an indicator of
potential treatment effect in clinical trials
Are meningeal lymphoid aggregates associated with progressive pathology? for acute and chronic conditions. Ele-
Progressive pathology and PIRA are caused by multiple interactive processes, including neurite vated NfL has distinct implications in
injury, CALs, and subpial cortical demyelination, as described below. Additionally, loss of synap- different CNS pathologies. High NfL is
also observed with injury to peripheral
ses and neuronal cell death are implicated in symptomatic worsening. Adverse consequences of
nervous system nerve fibers.
astrogliosis are also strongly suggested by the finding that elevated serum glial fibrillary acidic Oligoclonal bands (OCBs): increased
protein (GFAP) – an indicator of astrocyte injury or reaction – can help predict symptomatic wors- intrathecal immunoglobulin synthesis is
ening in P-MS patients [46]. Meningeal aggregates [47], found in most postmortem MS tissues typical of MS, and its detection aids in
MS diagnosis. OCBs are not specific for
that are appropriately sampled, may also be considered likely central players in PIRA (Figure 2, MS, being present in other neuroinflam-
Key figure). In autopsy studies, CALs correlated directly with high meningeal lymphocytic inflam- matory disorders including infection.
mation, both in group analysis and in individuals [48]. Meningeal aggregates are also topograph- Several technologies measure increased
ically associated with cortical demyelination (Figure 2) [19,49], representing sites of prominent intrathecal immunoglobulin synthesis,
including quantification of IgG index or
neurite injury [50], which are other key drivers of PIRA [32]. It is important to note that meningeal free kappa light chains as well as OCBs.
aggregates are present before the onset of MS, and are poised to contribute to PIRA early in dis- OCBs of diagnostic significance are
ease course [47]. Studies of biopsy tissues obtained at a mean of 1 month from symptom onset immunoglobulins of restricted

Trends in Immunology, April 2023, Vol. 44, No. 4 269

 Trends in Immunology
OPEN ACCESS

demonstrated the presence of meningeal aggregates in association with cortical demyelination heterogeneity, detected uniquely in CSF
but not blood.
[49]. Additional evidence for the early presence of meningeal aggregates comes from the detec-
Paramagnetic rim lesions (iron rim
tion of oligoclonal bands (OCBs) from the onset of MS, as well as at the clinically isolated syn- lesions): PRLs are detected using MRI,
drome (CIS) [51,52]) stage of disease. These observations are relevant to the pathogenic role of via specialized susceptibility-weighted
meningeal aggregates because OCBs are produced by intrathecal clonally expanded B lym- pulse sequences which are highly sen-
sitive to tissue components (including
phocytes [53], which are components of meningeal aggregates [54]. iron) that exhibit paramagnetic proper-
ties; represent a subset of chronic MS
How do meningeal aggregates promote tissue injury in remote CALs? Single-cell RNA-sequence lesions showing ongoing macrophage-
analysis of autopsy tissues characterized the rim macrophages in CALs (equivalent to PRLs) and mediated demyelination at the lesion
edge. PRLs evolve from acute lesions
showed two transcriptomic phenotypes, one of which represented iron-laden macrophages that over a period of approximately 3
generated the paramagnetic signal [44]. Compared with macrophages from normal appearing months.
(internal control) white matter, differentially expressed genes (DEGs) in PRL macrophages of Progression independent of relapse
activity (PIRA): Clinical worsening
both transcriptomic phenotypes demonstrated strong evidence for an IFN-γ response signature,
without evidence for relapse pathology.
suggesting that a lymphocyte-mediated autoimmune process caused this chronic demyelinating Definitions vary but one common
tissue injury [44]. Because CALs correlate quantitatively with higher meningeal lymphocytic in- description is the defined degree of
flammation [48], it is plausible that meningeal lymphoid aggregates and perivascular cuff lympho- worsening on the Expanded Disability
Status Scale (clinical rating system for
cytes [55] (compartmentalized in anatomic continuity with meningeal aggregates; [56]) are
MS states), confirmed after 3 or 6
principal sources of IFN-γ and other macrophage-stimulatory cytokines. months.
Progressive MS pathology: com-
We propose that, during the RR-MS phase of disease, PRLs (CALs) arise from a small proportion prises clinical and radiographic features.
PIRA captures the clinical features of
of new GdE lesions (Figure 1), which may occur 5–15 times per year, more than tenfold more
progressive pathology. On MRI, pro-
frequently than clinical attacks, as revealed by seminal monthly MRI studies in the early 1990s gressive pathology is the loss of CNS
[57]. PRLs cause persistent and ever-expanding injury both to myelin and nerve fibers, contributing parenchymal tissue, commonly
to PIRA early in the disease course. Other contributors to PIRA include degeneration of detected by quantification of brain or
spinal cord atrophy.
persistently-demyelinated neurites in inactive lesions [14] and subpial demyelinating lesions Relapse pathology: denotes clinical
found adjacent to foci of meningeal inflammation [19]. Neuritic injury also occurs as a consequence relapses (attacks or exacerbations), or
of relapse pathology, as indicated by landmark autopsy tissue studies [16,58] from the late 1990s. appearance of new MS lesions on MRI.
These reports used novel immunohistochemistry reagents [16] and confocal microscopy [58] to The lesions present either as GdE acute
lesions or as new or enlarging T2-bright
document abundant neurite and axon injury, including transection, in acute lesions of MS. This lesions.
type of damage had been suspected previously, based on classic silver-staining neuropathology, T1-dark (T1-hypointense) lesion: T1-
but had not been convincingly demonstrated. Recently, elevations of neurofilament light chain weighted brain MRI emphasizes fea-
tures of fatty tissue components; chron-
(NfL) were reported to be associated with active neuroinflammatory disease activity in both relapsing
ically T1-hypointense MS lesions (more
and progressive phases of MS [59]. Because NfL is derived from nerve fibers, these observations than three months after appearance)
conclusively established that nerve fiber injury is concomitant with relapsing pathology in MS. represent severe tissue injury with loss of
Reflecting the severity of neurite injury in PRLs, serum NfL concentrations have been elevated in myelin and nerve fibers; also termed T1
black holes; considered to have equiva-
patients with PRLs, compared to those without PRLs [60]. However, serum NfL was not elevated
lent prognostic implications as whole
relative to clinical manifestations of PIRA [61]. These findings suggest that PRLs may identify a brain atrophy. T1-dark lesions are also
population of MS patients with ongoing inflammatory damage to nerve fibers. In inactive P-MS, T2-bright.
serum NfL concentrations were not elevated in the absence of relapse pathology and did not predict T2-bright (hyperintense) lesion: T2-
weighted brain MRI emphasizes fea-
clinical worsening, suggesting that a substantial portion of nerve fiber injury occurs at the time of lesion
tures of tissue water. T2-hyperintense
formation associated with relapse pathology [46]. However, PRLs were not assessed in this study, foci can represent a wide spectrum of
lessening the statistical power to relate clinical outcomes to underlying CALs. However, our interpre- tissue change. In MS, T2-bright lesions
tation is that MS patients with PRLs uniquely show elevated serum NfL in the absence of explicit can be subdivided into those which are
also T1-dark and represent areas of
relapse pathology, illustrating ongoing neurite injury in these lesions. Hence, the overall hypothesis severe tissue injury, unlike lesions which
is consistent with the observation that PRL resolution is not accelerated by DMTs – which also fail are T1-isointense. These latter lesions
to reduce the clinical impact of PIRA. Overall, this concept highlights unresolved issues that remain are relatively nonspecific, can represent
to be addressed to bring to fruition a comprehensive approach to the treatment of MS. gliosis, loss of myelin, or other less-well-
characterized tissue changes.

Relapsing pathology recedes over time, either due to beneficial effects of DMTs or through
immune senescence [11], which may be a plausible explanation for the reduced frequency of

270 Trends in Immunology, April 2023, Vol. 44, No. 4

Trends in Immunology
OPEN ACCESS

Trends in Immunology

Figure 1. Evolution of human gadolinium-enhanced lesions of MS via MRI. For illustrative purposes, MRI brain scans of a patient with MS are shown at baseline (0 months)
and at 14 months’ follow-up. At 0 months, the T1-weighted scan with gadolinium enhancement (upper right) shows two foci of enhancement, indicated by red and blue circles. T1-
weighted scan without gadolinium (upper middle) shows that both foci of enhancement are T1-hypointense. The T2/FLAIR scans (upper left) show involvement of the periventricular
white matter with T2-hyperintense signal abnormality, characteristic of MS. The extent of abnormal T2 signal does not allow for localization of the enhancing foci on that pulse
sequence. At 14 months, on treatment, the T1/with gadolinium scan (lower, second from right) no longer shows enhancement. The T1-weighted scan without gadolinium
(lower, third from right) shows that one lesion (red circle) is persistently T1-hypointense indicating severe tissue damage. The other lesion (blue circle) is T1-isointense indicating
mild tissue damage. The T2/FLAIR scan is virtually unchanged from month 0. The phase image (lower right) shows that the T1-hypointense lesion (red circle) has a
paramagnetic rim, indicating ongoing radially expansive demyelinating tissue injury. Abbreviations: FLAIR, fluid -attenuated inversion recovery; MRI, magnetic resonance imaging;
MS, multiple sclerosis.

relapses and GdE lesions during natural history of MS. In its wake, PIRA increasingly dominates
the clinical phenotype. Here, we propose that, in patients with worsening symptoms, PIRA is
associated with PRLs: some PRLs represent unresolved lesions arising during the RR-MS
phase of disease [34,43]. Other PRLs derive from GdE lesions occurring with active progressive
MS [34,43]. Inflammatory demyelination at the edges of CALs/PRLs contributes to neurite degen-
eration [14], as well as demyelination-related functional impairment. As observed in RR-MS, PIRA
occurring in P-MS is unaffected by DMTs. This view of the significance and pathogenesis of PRLs
in relation to PIRA needs to be established by longitudinal clinical–MRI correlations, which may be
enabled by existing cohorts [34]. Notably, PRLs are not proposed as the sole underpinning

Trends in Immunology, April 2023, Vol. 44, No. 4 271

 Trends in Immunology
OPEN ACCESS

Key figure
Meningeal lymphoid aggregates are implicated in relapse and progressive MS pathology

Trends in Immunology

Figure 2. The figure shows distinct functional roles of meningeal lymphoid aggregates in both progressive (at left) and relapsing (right) MS pathology. Meningeal lymphoid
aggregates, composed of T and B cells, and follicular dendritic cells [62,63] are shown as clusters of magenta-colored cells in the cortical sulci. As a key component of
progressive MS pathology, meningeal lymphoid aggregates are associated with severe disease course [62], and are topographically related to subpial demyelinating
foci, shown as grey regions of demyelination with phagocytic macrophages (in purple) at the lesion edge [19,49]. It has been proposed [55] that soluble factors (pink
arrows) produced by meningeal aggregates promote this demyelinating tissue injury. Meningeal inflammation is also associated with white matter CALs [48]; shown as
a grey demyelinated circle with macrophages (purple) at lesion edges. A subset of these macrophages contains iron, enabling detection of CALs as paramagnetic rim
lesions [34,40]. CALs show topographic proximity to foci of perivascular inflammation (cluster of magenta T and B cells around vessel). Macrophages at edges of CALs
show a single cell RNA-Seq transcriptomic profile consistent with response to IFN-γ [44]. This finding is consistent with their dependence on signals from both
perivascular and meningeal lymphocytes (red arrows). Meningeal lymphoid aggregates are present early in MS disease course [49] and participate in relapsing
pathology (at right). T cells that traffic through the meninges (patrolling T cells in light blue) carry out immune surveillance of the central nervous system [64]. The
relevant APCs are meningeal macrophages [65] or B lymphocytes associated with lymphoid aggregates [66]. Autoimmune T cells (glowing green) can be found within
the patrolling T cell population and can interact with APCs (here shown as magenta B cell within meningeal aggregate), generating a local cytokine flux (red arrow) as a
result of this stimulatory interaction (yellow sunburst [65]). White matter vessels are readily activated (halo around white matter blood vessel) by cytokines which diffuse
efficiently through fiber tracts [67]. Inflamed vessels briskly recruit circulating leukocytes and transition (blue arrowheads) to vessels with perivascular inflammatory-cell
accumulations [68]. With sufficient compromise of the blood–brain barrier and macrophage recruitment [69–71], an acute demyelinating lesion can result from this
inflammatory cascade (blue arrowheads). Abbreviations: APC, antigen-presenting cell; CAL, chronic active lesion; MS, multiple sclerosis.

272 Trends in Immunology, April 2023, Vol. 44, No. 4

Trends in Immunology
OPEN ACCESS

characteristic of PIRA. Subpial cortical demyelination and neurite degeneration in ‘burnt out’ Outstanding questions
lesions can also contribute. Rather, PRLs are both contributors to PIRA and beacons, At the tissue level, PRLs, which are
showing that meningeal inflammation [48] and attendant cortical demyelination [19] are likely associated with worse clinical course,
represent CALs. In autopsy studies,
to be present.
increased numbers of CALs
correspond to worse clinical course
Lastly, why do DMTs affect the likelihood, timing, and severity of P-MS? DMTs reduce the fre- during life. What is the relationship of
quency of GdE lesions by as much as 80–95%, depending on the agent [7]. As GdE lesions PRLs to PIRA?
are the source of PRLs (Figure 1), their reduction or elimination may lower the accumulation of
During clinical trials lasting 2 years,
PRLs and associated PIRA. In turn, we posit that P-MS would be delayed or even prevented. PIRA was observed even in active-
We suggest that meningeal lymphoid aggregates play a central role in causing tissue injury that arm patients receiving high-efficacy
drives PIRA (Figure 2). This view is congruent with the recent focus on mechanisms underlying DMTs. How will high-efficacy DMTs af-
fect incidence and severity of PIRA
symptom worsening, rather than descriptive clinical categories [21].
over longer time frames?

Concluding remarks In MS patients with PIRA, serum NfL

This Opinion highlights PIRA as a crucial determinant of the outcome of MS, and focuses on its has not shown a relationship with
long-term neurological impairment;
tissue determinants: subpial demyelination, neurite degeneration, and acute neurite injury, as rather, it seems to report relapse pa-
well as the presence of CALs, which feature both demyelination and neurite injury. Mechanisti- thology, which damages neurites. In
cally, subpial demyelination and CALs are related to meningeal lymphoid aggregates (Figure 2) MS patients with PRLs, serum NfL is
elevated. How will biomarkers includ-
[19,48]. Preliminary evidence suggests that this compartmentalized CNS inflammatory process,
ing serum NfL relate to PIRA in MS pa-
reflected by the presence of PRLs (Figure 1), is stubbornly resistant to DMTs, because the pro- tients with PRLs?
portion of patients with PRLs in a cohort survey has been unrelated to the treatment status of
the patients [34]. Thus, an evaluation of current technology for monitoring processes related to Meningeal lymphoid aggregates are
the principal source of cerebrospinal
PIRA motivates us to suggest that PRLs indicate the presence of both tissue injury and meningeal
fluid (CSF) OCBs. Additionally,
lymphoid aggregates. The near-universal presence of meningeal aggregates in MS is supported CXCL13, elevated in MS CSF, is likely
by the high frequency (>90%) of OCBs in MS cerebrospinal fluid [51] (see Outstanding questions). related to the biology of these aggre-
PRLs in turn, will likely correlate to PIRA in individual patients. Thus, clinical trials to ameliorate gates. What is the appropriate bio-
marker strategy to assess meningeal
PIRA may be improved by determining the presence and evolution/resolution of PRLs. Where lymphoid aggregate biology in MS clin-
possible, evaluation of meningeal inflammation, as signaled by OCBs and other analytes [47], ical trials?
may indicate the potential for clinical benefit.
Traditional categorical clinical classifiers
such as RR-MS and chronic P-MS are
Acknowledgments not mechanistically informative. What
We thank Daniel Reich (NINDS/NIH) for providing anonymized MRI images shown in Figure 1. Additionally, we gratefully is the appropriate terminology for the
acknowledge his high-level input understanding the status of MRI research into MS pathogenesis. We thank Andy Fukutome pathological processes that underlie
for superior support in generating Figure 2. No funding was used in the preparation of this Opinion piece. MS outcomes? How can this terminol-
ogy be implemented to communicate
with investigators, stakeholders, and
Declaration of interests
regulatory agencies?
R.R. is interim Chief Medical Officer and scientific Co-Founder of Abata Therapeutics, which is developing treatments for au-
toimmune conditions including progressive MS.

Resources
i
https://clinicaltrials.gov/ct2/show/NCT01247324
ii
https://clinicaltrials.gov/ct2/show/NCT01412333

References
1. Walton, C. et al. (2020) Rising prevalence of multiple sclerosis 4. Zalc, B. (2018) One hundred and fifty years ago Charcot reported
worldwide: insights from the Atlas of MS, third edition. Mult. multiple sclerosis as a new neurological disease. Brain 141,
Scler. 26, 1816–1821 3482–3488
2. Sawcer, S. et al. (2011) Genetic risk and a primary role for cell- 5. Kremenchutzky, M. et al. (2006) The natural history of multiple
mediated immune mechanisms in multiple sclerosis. Nature 476, sclerosis: a geographically based study 9: observations on the
214–219 progressive phase of the disease. Brain 129, 584–594
3. International Multiple Sclerosis Genetics Consortium (2019) Mul- 6. Brown, J.W.L. et al. (2019) Association of initial disease-
tiple sclerosis genomic map implicates peripheral immune cells modifying therapy with later conversion to secondary progressive
and microglia in susceptibility. Science 365, eaav7188 multiple sclerosis. Jama 321, 175–187

Trends in Immunology, April 2023, Vol. 44, No. 4 273

 Trends in Immunology
OPEN ACCESS

7. Freeman, L. et al. (2022) High-efficacy therapies for treatment- 33. Arnold, D.L. (2007) The place of MRI in monitoring the individual
naïve individuals with relapsing-remitting multiple sclerosis. CNS MS patient. J. Neurol. Sci. 259, 123–127
Drugs 36, 1285–1299 34. Absinta, M. et al. (2019) Association of chronic active multiple scle-
8. Confavreux, C. and Compston, A. (2006) The natural history of rosis lesions with disability in vivo. JAMA Neurol. 76, 1474–1483
multiple sclerosis. In McAlpine’s Multiple Sclerosis, pp. 183–272, 35. Martire, M.S. et al. (2022) What is the potential of paramagnetic
Elsevier rim lesions as diagnostic indicators in multiple sclerosis? Expert.
9. Schwehr, N.A. et al. (2020) Age-related decreases in relapses Rev. Neurother. 22, 829–837
among adults with relapsing-onset multiple sclerosis. Mult. 36. Kuhlmann, T. et al. (2017) An updated histological classification
Scler. 26, 1510–1518 system for multiple sclerosis lesions. Acta Neuropathol. 133,
10. Filippi, M. et al. (2001) Enhancement frequency decreases with in- 13–24
creasing age in relapsing-remitting multiple sclerosis. Neurology 37. Ng Kee Kwong, K.C. et al. (2021) The prevalence of paramag-
56, 422–423 netic rim lesions in multiple sclerosis: A systematic review and
11. Bolton, C. and Smith, P.A. (2018) The influence and impact of meta-analysis. PLoS One 16, e0256845
ageing and immunosenescence (ISC) on adaptive immunity dur- 38. Luchetti, S. et al. (2018) Progressive multiple sclerosis patients
ing multiple sclerosis (MS) and the animal counterpart experi- show substantial lesion activity that correlates with clinical dis-
mental autoimmune encephalomyelitis (EAE). Ageing Res. Rev. ease severity and sex: a retrospective autopsy cohort analysis.
41, 64–81 Acta Neuropathol. 135, 511–528
12. Tutuncu, M. et al. (2013) Onset of progressive phase is an age- 39. Kolb, H. et al. (2022) From pathology to MRI and back: clinically
dependent clinical milestone in multiple sclerosis. Mult. Scler. relevant biomarkers of multiple sclerosis lesions. Neuroimage
19, 188–198 Clin. 36, 103194
13. Bjartmar, C. et al. (2003) Axonal loss in the pathology of MS: 40. Maggi, P. et al. (2020) Paramagnetic rim lesions are specific to
consequences for understanding the progressive phase of the multiple sclerosis: an international multicenter 3T MRI study.
disease. J. Neurol. Sci. 206, 165–171 Ann. Neurol. 88, 1034–1042
14. Kornek, B. et al. (2000) Multiple sclerosis and chronic autoimmune 41. Absinta, M. et al. (2016) Persistent 7-tesla phase rim predicts
encephalomyelitis: a comparative quantitative study of axonal injury poor outcome in new multiple sclerosis patient lesions. J. Clin. In-
in active, inactive, and remyelinated lesions. Am. J. Pathol. 157, vest. 126, 2597–2609
267–276 42. Rocca, M.A. et al. (2017) The role of T1-weighted derived mea-
15. Lewén, A. et al. (1995) Traumatic brain injury in rat produces sures of neurodegeneration for assessing disability progression
changes of beta-amyloid precursor protein immunoreactivity. in multiple sclerosis. Front. Neurol. 8, 433
Neuroreport 6, 357–360 43. Dal-Bianco, A. et al. (2017) Slow expansion of multiple sclerosis
16. Ferguson, B. et al. (1997) Axonal damage in acute multiple scle- iron rim lesions: pathology and 7 T magnetic resonance imaging.
rosis lesions. Brain 120, 393–399 Acta Neuropathol. 133, 25–42
17. Moll, N.M. et al. (2008) Cortical demyelination in PML and MS: 44. Absinta, M. et al. (2021) A lymphocyte-microglia-astrocyte axis in
similarities and differences. Neurology 70, 336–343 chronic active multiple sclerosis. Nature 597, 709–714
18. Junker, A. et al. (2020) Extensive subpial cortical demyelination is 45. Maggi, P. (2022) Chronic active multiple sclerosis lesions are poorly
specific to multiple sclerosis. Brain Pathol. 30, 641–652 responsive to anti-CD20 antibody treatment. Abstract O118,
19. Howell, O.W. et al. (2011) Meningeal inflammation is widespread ECTRIMS 2022, 26–28 October, Amsterdam, Netherlands
and linked to cortical pathology in multiple sclerosis. Brain 134, 46. Barro, C. et al. (2023) Serum GFAP and NfL levels differentiate sub-
2755–2771 sequent progression and disease activity in patients with progres-
20. Steenwijk, M.D. et al. (2016) Cortical atrophy patterns in multiple scle- sive multiple sclerosis. Neurol. Neuroimmunol. Neuroinflamm. 10,
rosis are non-random and clinically relevant. Brain 139, 115–126 e200052
21. Kuhlmann, T. et al. (2023) Multiple sclerosis progression: time for 47. Kee, R. et al. (2022) A review of compartmentalised inflammation
a new mechanism-driven framework. Lancet Neurol. 22, 78–88 and tertiary lymphoid structures in the pathophysiology of multi-
22. Granziera, C. et al. (2022) Time to change the current clinical ple sclerosis. Biomedicines 10, 2604
classification of multiple sclerosis? JAMA Neurol. 80, 128–130 48. Ahmed, S.M. et al. (2022) Accumulation of meningeal lympho-
23. Cree, B.A.C. et al. (2019) Silent progression in disease activity- cytes correlates with white matter lesion activity in progressive
free relapsing multiple sclerosis. Ann. Neurol. 85, 653–666 multiple sclerosis. JCI Insight 7, e151683
24. Cagol, A. et al. (2022) Association of brain atrophy with disease 49. Lucchinetti, C.F. et al. (2011) Inflammatory cortical demyelination
progression independent of relapse activity in patients with re- in early multiple sclerosis. N. Engl. J. Med. 365, 2188–2197
lapsing multiple sclerosis. JAMA Neurol. 79, 682–692 50. Peterson, J.W. et al. (2001) Transected neurites, apoptotic
25. Kappos, L. et al. (2018) Greater sensitivity to multiple sclerosis neurons, and reduced inflammation in cortical multiple sclerosis
disability worsening and progression events using a roving ver- lesions. Ann. Neurol. 50, 389–400
sus a fixed reference value in a prospective cohort study. Mult. 51. Arrambide, G. et al. (2018) The value of oligoclonal bands in the
Scler. 24, 963–973 multiple sclerosis diagnostic criteria. Brain 141, 1075–1084
26. Kappos, L. et al. (2020) Contribution of relapse-independent pro- 52. Miller, D. et al. (2005) Clinically isolated syndromes suggestive of
gression vs relapse-associated worsening to overall confirmed multiple sclerosis, part I: natural history, pathogenesis, diagnosis,
disability accumulation in typical relapsing multiple sclerosis in a and prognosis. Lancet Neurol. 4, 281–288
pooled analysis of 2 randomized clinical trials. JAMA Neurol. 53. Obermeier, B. et al. (2008) Matching of oligoclonal immunoglobulin
77, 1132–1140 transcriptomes and proteomes of cerebrospinal fluid in multiple
27. Tur, C. et al. (2022) Association of early progression independent of sclerosis. Nat. Med. 14, 688–693
relapse activity with long-term disability after a first demyelinating 54. Lovato, L. et al. (2011) Related B cell clones populate the menin-
event in multiple sclerosis. JAMA Neurol. 80, 151–160 ges and parenchyma of patients with multiple sclerosis. Brain
28. Hauser, S.L. et al. (2017) Ocrelizumab versus interferon beta-1a 134, 534–541
in relapsing multiple sclerosis. N. Engl. J. Med. 376, 221–234 55. Lassmann, H. (2018) Pathogenic mechanisms associated with
29. Lublin, F.D. et al. (2022) How patients with multiple sclerosis ac- different clinical courses of multiple sclerosis. Front. Immunol. 9,
quire disability. Brain 145, 3147–3161 3116
30. Lublin, F.D. et al. (2014) Defining the clinical course of multiple 56. Mapunda, J.A. et al. (2022) How does the immune system enter
sclerosis: the 2013 revisions. Neurology 83, 278–286 the brain? Front. Immunol. 13, 805657
31. Capanna, M. et al. (2022) Is the effect of drugs in progressive MS 57. Harris, J.O. et al. (1991) Serial gadolinium-enhanced magnetic
only due to an effect on inflammation? A subgroup meta-analysis resonance imaging scans in patients with early, relapsing-
of randomised trials. Mult. Scler. 28, 1744–1751 remitting multiple sclerosis: implications for clinical trials and nat-
32. Ontaneda, D. et al. (2017) Progressive multiple sclerosis: pros- ural history. Ann. Neurol. 29, 548–555
pects for disease therapy, repair, and restoration of function. 58. Trapp, B.D. et al. (1998) Axonal transection in the lesions of mul-
Lancet 389, 1357–1366 tiple sclerosis. N. Engl. J. Med. 338, 278–285

274 Trends in Immunology, April 2023, Vol. 44, No. 4

Trends in Immunology
OPEN ACCESS

59. Williams, T. et al. (2021) Neurofilaments in progressive multiple 66. Molnarfi, N. et al. (2013) MHC class II-dependent B cell APC
sclerosis: a systematic review. J. Neurol. 268, 3212–3222 function is required for induction of CNS autoimmunity indepen-
60. Maggi, P. et al. (2021) Chronic white matter inflammation and dent of myelin-specific antibodies. J. Exp. Med. 210, 2921–2937
serum neurofilament levels in multiple sclerosis. Neurology 97, 67. Brown, D.A. and Sawchenko, P.E. (2007) Time course and distri-
e543–e553 bution of inflammatory and neurodegenerative events suggest
61. Masanneck, L. et al. (2022) Detecting ongoing disease activity in structural bases for the pathogenesis of experimental autoim-
mildly affected multiple sclerosis patients under first-line therapies. mune encephalomyelitis. J. Comp. Neurol. 502, 236–260
Mult. Scler. Relat. Disord. 63, 103927 68. Ransohoff, R.M. (1999) Mechanisms of inflammation in MS
62. Magliozzi, R. et al. (2007) Meningeal B-cell follicles in tissue: adhesion molecules and chemokines. J. Neuroimmunol.
secondary progressive multiple sclerosis associate with early 98, 57–68
onset of disease and severe cortical pathology. Brain 130, 69. Huang, D. et al. (2002) Pertussis toxin-induced reversible en-
1089–1104 cephalopathy dependent on monocyte chemoattractant
63. Serafini, B. et al. (2004) Detection of ectopic B-cell follicles with protein-1 overexpression in mice. J. Neurosci. 22, 10633–10642
germinal centers in the meninges of patients with secondary pro- 70. Huang, D. et al. (2005) Chronic expression of monocyte chemo-
gressive multiple sclerosis. Brain Pathol. 14, 164–174 attractant protein-1 in the central nervous system causes de-
64. Ransohoff, R.M. and Engelhardt, B. (2012) The anatomical and layed encephalopathy and impaired microglial function in mice.
cellular basis of immune surveillance in the central nervous sys- FASEB J. 19, 761–772
tem. Nat. Rev. Immunol. 12, 623–635 71. Huang, D.R. et al. (2001) Absence of monocyte chemoattractant
65. Kivisäkk, P. et al. (2009) Localizing central nervous system im- protein 1 in mice leads to decreased local macrophage recruit-
mune surveillance: meningeal antigen-presenting cells activate ment and antigen-specific T helper cell type 1 immune response
T cells during experimental autoimmune encephalomyelitis. in experimental autoimmune encephalomyelitis. J. Exp. Med.
Ann. Neurol. 65, 457–469 193, 713–726

Trends in Immunology, April 2023, Vol. 44, No. 4 275