New imaging markers in MS and related disorders: Smoldering inflammation and central vein sign

Introduction

Although the past decade has seen remarkable progress in the evaluation and management of multiple sclerosis (MS) and its mimics, many diagnostic, prognostic, and treatment challenges remain. Despite updated McDonald diagnostic criteria in 2017, recent data suggest that around 15% of individuals are misdiagnosed with MS when referred to specialty neurology centers.^1,2 MS misdiagnosis can lead to patient harm in the form of unnecessary immunosuppression morbidity, psychological burden, and healthcare system cost.

Another major clinical gap in the field is the assessment and treatment of MS-associated neurodegeneration. This challenge is now a focus of scientific attention, especially following the tremendous successes of anti-inflammatory immunotherapies developed in the past decade. For example, phase III trials of B-cell depleting antibodies demonstrate greater than 90% reduction in the appearance of clinical or MRI-assessed inflammatory activity compared to platform therapy.³ However, these positive data are mitigated by a ~20% rate of ongoing disability accumulation over the 96-week trial period, even in the setting of successful peripheral immune suppression.⁴ This discordance suggests an ongoing degenerative process contemporaneously referred to as “silent progression”⁵ or “progression independent of inflammatory activity” (PIRA) that occurs in the absence of relapses or new focal MRI lesions. Although still lacking a harmonized definition,⁶ the widespread clinical observance of PIRA challenges current phenotypic paradigms of MS, which emphasize a distinction between relapsing and progressive forms of the disease, and impels the clinical need for development of: (1) prognostic biomarkers to identify patients at highest risk for disability progression; and (2) pharmacodynamic biomarkers to determine the therapeutic efficacy of drug interventions beyond suppression of peripheral immune mechanisms that underlie clinical relapses and new lesions. This pharmaceutical development process is limited by incomplete understanding of the mechanisms leading to clinical progression.

In this chapter, we review recent neuroimaging innovations in MS, primarily focusing on approaches that accentuate MRI-susceptibility phenomena using the T2* signal decay. T2*-weighted sequences differ from conventional (spin-echo) T2-weighted sequences in that they lack a 180° refocusing RF pulse, thus allowing (or emphasizing) the dephasing signal-loss effects of even miniscule diamagnetic or paramagnetic field inhomogeneities within the local spin-lattice environment.⁷ These inhomogeneities — known as magnetic “susceptibility” — can arise from a variety of sources in the brain, most relevant here being forms of paramagnetic heme- and nonheme-iron that include deoxygenated venous blood, subacute/digested blood products, and intracellular iron storage complexes including ferritin. Other relevant paramagnetic signals seen in the brain include those from demyelination (loss of diamagnetic myelin structure) and reactive oxidation products (free radicals).^8,9 Although the magnitude of the T2* signal is sensitive to these effects, complementary and often superior contrast can be obtained from the phase component of the MR signal after application of a homodyne high-pass filter or other phaseunwrapping technique.¹⁰

T2* signals are acquired using gradient-recalled echo (`) sequences, sometimes with echo-planar imaging (EPI) readouts, with tuning effects sensitive to the flip angle and the echo times in particular. Additional postprocessing techniques can be applied to further improve susceptibility contrast (such as susceptibility-weighted imaging,¹⁰ SWI) or reduce artifacts through source estimation to more accurately quantify tissue concentrations of iron or calcium, called quantitative susceptibility mapping (QSM). Methods for producing SWI and QSM are non-standardized. A recent international consensus aims to reduce this variance,¹¹ but QSM remains rarely implemented in clinical routine, and manufacturer protocols for susceptibility-weighted imaging differ based on proprietary decisions. It is the exquisite sensitivity to iron concentration that ultimately allows for a variety of useful contrasts in MS, at submillimeter resolutions. This process allows visualization of multiple pathophysiological features of MS, leading to the recent development of two emerging biomarkers with high translational relevance to clinical practice.

Advances in diagnostic imaging markers in MS and related disorders

The central vein sign (CVS)

The T2* sequence has been exploited in several ways to explore the pathophysiology of MS, revealing useful biomarkers that are highly specific to MS compared to its radiological mimics. The first of these biomarkers capitalizes on the longstanding histological observations of a vein centered within MS white matter lesions.¹² T2* imaging is sensitive to the effects of venous deoxygenated hemoglobin due to the paramagnetic shift of the hemoglobin structure following the unloading of bound molecular oxygen.¹³ This paramagnetic effect accentuates the signal loss within veins on T2* while still capturing the T2 effects from surrounding demyelination and CNS tissue injury. This combination allows for simultaneous visualization of both a hyperintense lesion and a hypointense vein, leading to a naturally high contrast — the “central vein sign” (CVS). Examples of the CVS are depicted in Figure 1A. In 2016, Sati et al. established a consensus definition of the CVS on behalf of the North American Imaging in MS Cooperative (NAIMS) as a small hypointense dot or thin line, visible in at least two different planes and centering a focal MS lesion larger than 3 mm in greatest diameter. At high-resolution 7T MRI, 80–100% of supratentorial white matter lesions exhibit the CVS, impressively, optimized T2*-weighted sequences on 3T MRI scanners can achieve comparable CVS detection rate.^14,15 Lower field-strength scanners, such as 1.5T MRI, demonstrate acceptable sensitivity, with a pooled average of 58% exceeding the recommended cutoff (40%).¹⁶

Figure 1: Central vein sign.

3D FLAIR* images in sagittal (top) and axial (bottom) planes. (A) 54-year-old male with relapsing MS and numerous lesions depicting the CVS. (B) 41-year-old woman referred for concerns of MS who presented with non-specific symptoms. MRI showed non-specific white matter lesions negative for CVS and MS was ruled out. (C) 58-year-old woman with a history of recurrent, bilateral retinal vasculitis and uveitis, and migraine. MRI showed non-specific white matter lesions, most of them negative for the CVS. The patient’s symptoms and MRI were not consistent with MS.

The CVS is a specific diagnostic biomarker of MS pathology and can aid in distinguishing MS from its radiological mimics, such as neuromyelitis optica spectrum disorder (NMOSD),¹⁷ migraine,¹⁸ cerebral small vessel disease,¹⁹ inflammatory vasculopathies,²⁰ and Susac syndrome,²¹ when using a threshold of 40% CVS+ lesions¹⁶ or simplified methods such as counting 3 or 6 CVS+ lesions (see Figure 1B and 1C).^18,22 Many studies have further established the usefulness of the CVS as a diagnostic marker for atypical MS cases. One publication described that in patients with suspected MS, more than 85% of those eventually diagnosed with MS meet CVS criteria, compared to less than 25% of those with alternative diagnoses.²³ Thus, current evidence supports incorporation of the CVS into the MRI criteria for MS to improve diagnostic specificity.

Assessment of the CVS can also have an impact in therapeutic decisions. Kilsdonk et al. showed a significantly lower percentage of CVS+ lesions in the deep white matter in MS patients ≥40 years old compared with younger patients. Another study by Al-Louzi et al. revealed that most new T2 lesions in MS patients exhibit CVS, particularly in younger individuals.²⁴ Absence of CVS in new T2 lesions in older patients with MS and other comorbidities may avoid unnecessary therapeutic escalation.

Paramagnetic rim lesions (PRL)

A subset of chronic MS lesions has been observed to feature a paramagnetic rim on susceptibility-sensitive sequences, including filtered phase, R2* maps, QSM, and SWI; a fraction of these can also be visualized on T2*-weighted magnitude images (see Figure 2). The observed lesion-level prevalence is highly variable (~2–50%) and depends on MRI protocol, field strength, acquired voxel size, and lesion counting protocols, especially in relation to confluent lesion handling.²⁵ Histological examinations of these paramagnetic rims have confirmed the source of this signal to be molecular iron, sequestered intracellularly and bound to ferritin, within pro-inflammatory macrophages and microglia. Whether this iron accumulation is causally involved in the pro-inflammatory process, or rather is an epiphenomenon, is an ongoing area of investigation.^26,27

Figure 2: PRL visualization using different MRI sequences and post-processing techniques.

Multiple PRL examples in a man in his 30s with relapsing MS, as viewed on axial multimodal MRI with primary contrasts between the top row images depicting magnitude-weighting and the bottom row as the high-pass filtered phase. Specifically: (A) T1-weighted, (B) Philips SWIp, 0.6 × 0.6 × 2.0mm (C) Philips EPI-magnitude 0.6 × 0.6 × 0.6mm, (D) GE SWAN, 0.9 × 0.9 × 2.0mm, (E) T2-FLAIR, (F) Philips filtered-phase SWIp, (G) Philips EPI-phase, and (H) GE SWAN filtered phase. The GE scan (D & H) took place several months after the Philips scan. White solid arrows indicate PRL; dotted arrows are central veins. Although all PRL have central veins here, this is not always the case. The PRL depicted in the bottom left has several “red flag” features including an incomplete rim and complex vascularity with multiple traversing veins on this slice. These features reduce confidence; slices more superior to this lesion (not shown) confirmed its status as a definite PRL.

Emerging evidence indicates a high specificity of PRL for MS compared to other neuroinflammatory or MRI mimics. The presence of ≥1 PRL is commonly visualized at rates between 40–60% in MS in the largest studies at 3T,^28–32 although many PRL can be adequately visualized at 1.5T as well (see Figure 3).³³ A recent consensus statement proposed a standardized definition of PRL identification.³⁴ As of the date of writing, eleven adult studies have directly compared the prevalence of PRL between MS and MS-mimics; these are summarized in Table 1.^{21,28,35–42} Despite using heterogenous methods to identify PRL, these independent groups have nonetheless consistently demonstrated a high specificity for MS, ranging between 87–100% (with rare exception in comparison to Susac). One of the largest studies (n=329 pwMS, n=83 mimics) featured a harmonized multisite MRI sequence using isotropic 0.65mm voxels and reported high specificity (93%) with sensitivity of 52%.²⁸ A second independent study examined 254 patients with MS, compared to 308 others with conditions that can mimic the presentation of MS (n=91) and n=217 with small vessel disease.³⁸ The authors found high specificity (99.8%) but low sensitivity (24.0%) for the presence of ≥1 PRL.³⁸ This modest sensitivity may have been in part related to the highly variable acquisition parameters used across different study sites (see supplement from Sinnecker et al.²²), with minimum in-plane resolutions ranging between 0.3 and 0.9 mm and slice thicknesses ranging from 0.5 to 4.0 mm. Only 23% of patients in the total cohort had ≥1 PRL. Examples of white matter lesions lacking paramagnetic rim features are shown in Figure 4.

Figure 3: Comparison of PRL at 3T and 1.5T.

Representative examples of a paramagnetic rim lesion (PRL) with a central vein (CVS, seen as a dot within the lesion) in a man with relapsing MS are presented for 3T and 1.5T MRI. The PRL is clearly visible on unwrapped filtered phase images from different sequences at both 3T and 1.5T. Magnification of the PRL is shown in the insets for each image, accompanied by the image resolution for each MRI sequence.

Table 1:

Summary of studies evaluating PRL as a diagnostic biomarker in MS

Author & year	n (MS)	n (Comparison)		Specificity ≥1 PRL	Sensitivity ≥1 PRL	Field
Hagemeier 2012	135	49	Incidental	98%	22%	3T
Hosseini 2018	17	18		100%	n/r	7T
Kilsdonk 2014	16	16		100%	25%	7T
Wuerfel 2012	10	5	Susac	20%	100%	7T
Chawla 2016	21	21	NMO	100%	10%	7T
Sinnecker 2016	10	10		80–90%	90%	7T
Kim 2022	32	15		87%	75%	3T
Jang 2020	32	21		95%	81%	3T
Clarke 2020	112 (CIS)	35	OIND, NIND, & other mimics	100%	59%	3T
Maggi 2020	329	83		93%	52%	3T
Meaton 2022	254	308	OIND, NIND, leukoaraiosis	99.8%	24%	3T

Figure 4:

Axial MRI showing FLAIR (left) and filtered-phase T2*-weighted (GE “SWAN”, right) scans of four different patients (A-D), all of whom were evaluated for the possibility of MS due to neurological symptoms. Although each patient has white matter injury, none of these lesions have definite paramagnetic rims. Very few of these lesions have central vein features, although these phase sequences do not provide good contrast for evaluation of this feature. (A) A woman in her 50s ultimately diagnosed with glycogen branching enzyme 1-related adult polyglucosan body disease; (B) a woman in her 60s with leukoaraiosis; (C) a man in his 50s diagnosed with small vessel disease; (D) a woman in her 40s diagnosed with migraine-related white matter changes.

There are additional diagnostic clinical considerations related to PRL in early disease, with potential for rapid translation to “bedside” practice. For example, one study showed that in CIS, of patients having ≥1 PRL on a standard SWI sequence, 100% manifested clinical relapsing MS over 4-year follow-up.⁴² Similarly, the number of PRL was found to be a risk factor in the evolution of RIS to relapsing MS over a median 6.3-year longitudinal follow-up.⁴³

These pieces of evidence offer preliminary support for the potential use of PRL as a diagnostic biomarker in early MS. Figure 5 shows a patient from one of the authors’ practices who received an MRI for workup of headaches. This scan revealed periventricular, leukocortical, and infratentorial lesions, many of which featured a definitive paramagnetic rim and CVS. There was little clinical evidence to suggest MS, and a normal exam. Should these PRL change the approach to clinical management? Or increase confidence in offering early treatment? This case highlights the need for ongoing research on the implications of PRL for both diagnosis and prognosis (discussed further in this article) of MS.

Figure 5: Incidental paramagnetic rim lesions in a patient with headache.

Multiple paramagnetic rim lesions in a woman in her 30s presenting with headache and lesions consistent with radiologically isolated syndrome (2009 criteria). The PRL are clearly seen as (A) hyperintense on axial T2-FLAIR with (B) a hypointense rim on filtered-phase T2*-weighted sequence. Images were acquired as part of routine clinical practice on a 3T system (GE Pioneer); this manufacturer sequence is called SWAN (Susceptibility Weighted Angiography). Solid white arrows = definitive PRL, one with a clearly visible CVS (left); Dashed white arrow = diffuse paramagnetic signal at the edge of a PRL, not visible in this slice; Dotted white arrow = possible PRL, with heterogeneous and vascular features as well as an incomplete border; Double-line arrow = probable PRL, with heterogenous edge features.

Subpial lesions

Cortical lesions are observed across a number of demyelinating disorders and are not specific for MS. Extensive cortical involvement has been observed on MS histopathology for decades,⁴⁴ however, reported to be unique to this disease are longitudinally extensive “ribbon-like” areas of demyelination restricted to the subpial (outermost) layer of the cortex.⁴⁵ Noninvasive visualization of this pathology would therefore likely represent a highly specific biomarker for MS. Carrying this idea to fruition, however has been challenging. First, the location of these subpial lesions — within micrometers of the cerebrospinal fluid — creates difficulties in designing high-contrast MRI sequences due to partial volume artifacts. Additionally, subpial cortical lesions are small (usually <2mm thick) and have inherently poor tissue contrast to the surrounding lightly myelinated cortex. For these reasons, detection of cortical lesions remains difficult at both 3T and 7T, although higher field strength is advantageous⁴⁶ due in part to easier acquisition of submillimeter voxels and greater sensitivity for susceptibility effects, including the lesional loss of iron signal.

Visualization of subpial pathology is most sensitive using T2*-weighted^47,48 or combined multiparametric sequences,⁴⁹ though even with the best current approaches, a substantial number of histopathologically proven lesions are missed.⁴⁶ Efforts to optimize the MR contrast for subpial lesions at 3T are ongoing, with one group demonstrating significantly increased sensitivity through the use of a T2*-weighted sequence with nulling of the CSF signal using inversion recovery at 3T.⁵⁰ These are the first steps toward translation into clinical practice.

Advances in the detection of non-relapsing progressive disease

There are currently no “bedside” methods for the immediate detection of non-relapsing progressive MS. Instead, this classification is determined only retrospectively, through careful longitudinal observation of neurological signs. These methods are problematically slow, with current definitions requiring clinical observations separated by at least 6–12 months. Real-world data show that delays of 3 years in the determination of secondary progressive MS are not uncommon.⁵¹ Modest inter-rater exam reliability and unexplained fluctuations in MS disease manifestations add further challenges and “noise” for the clinician or clinical trialist tasked with determining progression. An alternative view of MS is through de-emphasis of phenotype categories, instead framing the disease as continuous spectra of inflammation and slow neurodegeneration, with each interpreted as (partially) independent phenomena requiring different therapeutic approaches. This view of MS highlights an urgent need for reliable and objective biomarkers of progressive and irreversible neurological worsening in the absence of both new clinical relapses and new T2-hyperintense lesions in the CNS. Surrogate biomarkers capturing the various pathological underpinnings of this worsening would greatly facilitate the development and testing of therapeutic interventions aimed at slowing or halting this process.

White matter: imaging chronic active lesions

Chronic active lesions (CAL) are a conceptual term, grounded in histopathology, referring to lesions that exhibit compartmentalized inflammation and ongoing tissue injury behind an intact blood-brain barrier. Between 20–40% of MS lesions are typically observed as CAL,⁵² although this is likely influenced by age and possibly disease duration.⁵³ The underlying mechanisms dictating the transition from acute lesion to CAL remain unknown but are hypothesized to include a failure to engage reparative/anti-inflammatory mechanisms.^8,25 CALs are generally defined histologically by a hypocellular and fully demyelinated lesion core surrounded by a dense rim of CD68+ myeloid-lineage cells (and to a lesser extent astrocytes), often containing iron and/or myelin debris.^54,55 CAL are synonymous with the contemporaneous histological category of “mixed active/inactive” lesions and broadly encompass other terminology as well, including “smoldering” and “slowly expanding” lesions. Imaging correlates of CAL have led to the development of three emerging markers for these lesions: (1) the PRL, (2) the slowly expanding lesion (SEL), and (3) “hot” TSPO+ lesions on positron emission tomography (PET). Each marker captures a complementary but overlapping aspect of CAL pathology.

PRL represent the significant subset of CAL defined by a high density of CD68+, iron-laden myeloid-lineage cells surrounding the edge of the MS plaque. Although the origins and metabolic pathways of iron sequestration remain a matter of investigation, it is confirmed from many MRI-histopathological studies that this iron signal can be reliably detected on susceptibility-sensitive MRI sequences.^{8,25,27,29,56–64} In a small study requiring replication, PRL corresponded very closely to CAL, with 93% of CAL being PRL and 93% of PRL being CAL.⁵⁶ In addition to the diagnostic considerations discussed above, PRL have been observed across all MS phenotypes in roughly 50% of pwMS^28,65 and are strongly associated with worse MS-related disease status and prognosis. Multiple large (n >100 patients) studies across 7 independent sites consistently show pathological associations between PRL and MS severity^28–30,63 or MS disability.^28,29,63,66 No study has shown a protective association, although one failed to find any associations with worse outcomes.³² Six longitudinal studies have been conducted to assess whether PRL have any prognostic significance,^67–71 all of which found that the baseline number of PRL (but not necessarily overall PRL status) predicted disability worsening over follow-up times ranging between 2 and 9 years. Recently, two studies directly linked the presence of baseline PRL to PIRA at follow up.^71–73

Some CAL are presumed to increase in size over time (indeed, one histopathological term for CAL is “slowly expanding lesions”). Slow lesion enlargement can also be detected on MRI as SEL, defined recently as the subset of T2-hyperintense MS lesions that exhibits steady concentric longitudinal enlargement. Formal determination of an SEL requires coregistration of at least 3 separate 3D T1 or T2-weighted longitudinal scans, followed by calculation of the nonlinear deformation of the baseline T1/T2 lesion morphology over 1–2 years follow-up.⁷⁴ The histopathological basis for these lesions has not been studied systematically. Using MRI, several recent comparative studies have shown that PRL represent only 7%⁷⁰ – 17%⁷⁵ of total SEL, which are observed at substantially higher frequency counts compared to PRL. SEL thus likely include a set of lesions with different pathogenic mechanisms, such as iron-independent lesional expansion, evolving periplaque injury, or even within-lesion neurodegeneration. Consistent with this, several studies suggested that a substantial portion of CAL lack iron-laden peripheral rim cells.^25,55

Notwithstanding these differences, SEL also share some MRI characteristics and clinical prognostic implications with PRL. For example, the number of SEL is associated with worse cross-sectional and longitudinal disability outcomes in MS.^76,77 Greater underlying lesional tissue injury is also common to both PRL^64,78–80 and SEL^74,76 compared to T2 lesions without these features: in the one study that examined both PRL and SEL, the greatest underlying tissue injury was seen in lesions that qualified as both PRL and SEL.⁷⁵ Both types of lesion have also been associated with a centrifugal gradient of periplaque injury,^78,79,81 implying ongoing injury outside the lesion. This periplaque involvement has been independently seen on histological analysis and characterized as Wallerian degeneration, axonal loss with spheroids, astrogliosis, and microglia activation.^25,64 A conceptual diagram with the overlap between CAL, PRL, and SEL is shown in Figure 6.

Figure 6: Conceptual Venn diagram relating the overlap between paramagnetic rim lesions and slowly expanding lesions.

Schematic diagram of the spectrum of chronic active or expanding lesions. CAL=chronic active lesions; PRL=paramagnetic rim lesions; SEL=slowly expanding lesions. Percentage overlap between CAL and PRL is based on data from Rahmanzadeh et al. but remains overall uncertain. The indistinct borders of the SEL subset reflect a lack of data regarding the underlying biology of these lesions, with question mark categories connoting speculation on the part of the authors.

Last, in contrast to SEL are longitudinal observations of slowly “eroding” lesions,⁸² referring to the slow ex-vacuo replacement of cavitated lesions as described in a series of studies over 5–10 years.^83,84 The (periventricular) tissue loss of these “atrophied lesions” predicts both short- and long-term disability worsening, highlighting the idea that some chronic (presumably CAL) MS lesions may be undergoing both radial expansion on T2-weighted images while simultaneously losing total volume due to involution and collapse of (central) parenchymal structural tissue. Supporting this are observations that PRL exhibit objectively greater underlying tissue destruction based on quantitative T1 times and diffusion measures that longitudinally worsen.^8,64,85 These data contrast with a PRL-SEL comparative study that did not observe ongoing/worsening tissue destruction in non-expanding PRL.⁷⁵ This dynamic morphology and need for calibrated longitudinal data add complexity in the translational process of SEL and atrophied lesions as bedside clinical tools.

In addition to the use of structural MRI, molecular assessments using PET likely offer complementary information about smoldering pathology. PET imaging uses radioactive tracers conjugated to a targeting molecule to spatially localize areas of interest. In chronic inflammatory diseases, the mitochondrial translocator protein (TSPO) is expressed mostly within myeloid cells, with signal reflecting higher cell density⁸⁶ suggestive of innate immune activation and neuroinflammation. Although lacking spatial precision, combined PET-MRI studies allow for the TSPO-innate immune characterization (overlay) of T2-hyperintense lesions. Like T2*, patterns of chronic lesions have been described, including inactive (no signal), homogeneously TSPO+, or having a TSPO+ rim at the edge or periplaque region. Multiple studies now suggest that TSPO-PET could be used to usefully characterize the molecular characteristics of individual chronic lesions with predictive ability for worsening neurological disability. Methods were heterogeneous, but the periplaque/lesional “rim” often correlated strongly with cross-sectional or longitudinal worsening,^87–90 though another study found that the lesion core carried the highest or most predictive signal.⁹¹ Few studies have compared PET and susceptibility-based metrics, and none have assessed SEL and PET; even fewer have determined histopathological correlates. One interesting study concluded that QSM-defined PRL contained higher TSPO signal compared to non-PRL, which correlated with TSPO expression on postmortem tissue.⁶⁰ The overall feasibility of PET-based studies is limited in clinical practice due, in part, to expense, radiation exposure, and issues around data interpretation, but PET may prove to be a useful surrogate biomarker for clinical trials, pending further histopathological validation and technical standardization.

Neuroimaging pathology of the cortex and meninges

Cortical pathology in MS is extensive and present early⁹² in the disease course. In addition to the subpial (type III) cortical lesions discussed previously, cortical pathology also includes intracortical (type II) and leukocortical (type I) lesions, described based on their location within or across cortical boundaries. A gradient of “outside-in” pathology has been described to correlate with aggregations of perivascular lymphoid cells in the overlying meningeal space, providing significant circumstantial evidence for their role in subpial pathogenesis.^93,94 This gradient of pathology has been visualized on high-field MRI.⁹⁵ Evidence based mostly on cell cultures and animal models suggests that meningeal tertiary lymphoid structures, containing high numbers of CD20+ B cells, secrete a variety of pro-inflammatory soluble cytokines^96,97 and provide a survival niche for the maturation of plasmablasts⁹⁸ and for activating trafficked lymphoid cells.⁹⁹ A recent pathological study showed that in progressive MS, the density of B and T cells in the meninges correlates with the nearby presence of CAL.¹⁰⁰ Although these lymphoid structures have not yet been imaged directly in vivo, meningeal inflammation can share close proximity to inflamed and “leaky” meningeal vessels.¹⁰¹ Focal areas of leptomeningeal enhancement (LME) observed on post-contrast T2-FLAIR MRI are therefore potential indirect markers of meningeal inflammation.¹⁰¹ Although not specific for MS,¹⁰² LME has been correlated with aspects of MRI and clinical pathology,¹⁰³ and its use as a prognostic biomarker is controversial but continues to be explored. A recent study that employed a 3D real-reconstruction inversion recovery protocol showed greatly enhanced sensitivity to detect LME as well as an association with PRL,¹⁰⁴ offering an avenue for improved understanding of the role of LME in MS pathophysiology.

Imaging other neuroimmune spaces: CNS lymphatics and the choroid plexus

Other neuroimmune interfaces in the brain, such as the choroid plexus (see Figure 7) and the cerebral lymphatic system, have recently been explored for their potential value in MS diagnosis and prognosis. The choroid plexus volume,^105–110 gadolinium-enhancement,¹¹¹ PET-TSPO signal,^106,112 and estimated T2 relaxation time (“pseudo-T2”)¹¹⁰ have been shown to discriminate MS from healthy controls^{105,106,111,112} or NMOSD¹⁰⁹ and correlates with MS disease activity,¹⁰⁵ severity,¹⁰⁵ lesion expansion over time,^107,113 PRL and gray matter atrophy,¹⁰⁸ and longitudinal disability worsening.¹¹⁰ It should be noted, however, that enlargement of the choroid is not specific to MS and has also been observed across a range of other vascular and neuropsychiatric pathology.^113–116 One study in healthy controls suggested that choroid enlargement may be due to reduced glymphatic flow,¹¹³ as inferred based on intrathecal gadolinium infusions. A more straightforward way of visualizing dural lymphatic vessels noninvasively was recently demonstrated using post-gadolinium T2-FLAIR and T1-black blood imaging.¹¹⁷ The relative ease and translational potential of this protocol will facilitate future studies on structure and potential (dys)function in MS and other neuroinflammatory diseases.

Figure 7: Choroid plexus enlargement in MS.

Representative example of the prominence of choroid plexus in MS in a man in his 50s. Choroid plexus enlargement can be visualized on T2-FLAIR images (triplanar view is shown in the insets). On susceptibility-based images, punctate paramagnetic signal is seen within the choroid plexus potentially indicating increased infiltration of iron-laden myeloid cells.

Conclusions

Although the MRI markers described here are highly promising for the diagnosis of MS, technical and methodological challenges limit their immediate translation to clinical practice. T2* imaging remains unstandardized in clinical practice, with manufacturers having a wide variety of approaches across scanner types. Manufacturer default in-plane resolutions may limit visualization of smaller PRL, and protocols need to be customized to export the phase images in addition to the processed combination magnitude/phase (“susceptibility-weighted”) protocols. These sequences need to be compared and validated within and between pwMS, as well as at the lower field strengths (1.5T) commonly used in clinical practice. Moreover, questions of histopathological specificity remain, especially as related to SEL and TSPO+ lesions.

The ongoing development of in vivo markers for CAL and cortical/meningeal pathology will inevitably improve therapeutic care through the creation of new surrogate outcome markers sensitive to non-relapsing progression. Indeed, similar to the use of gadolinium-enhancing lesions as an outcome marker, several trials have implemented PRL as an outcome marker in the assessment of next-generation disease-modifying therapies. However, histopathological evidence indicates a variety of other disease processes involved in non-relapsing progression that at present cannot be directly captured using even advanced imaging techniques. Examples of such processes include the loss of synaptic density, mitochondrial dysfunction, and reactive oxidative processes. The development of imaging contrasts and methods to visualize these processes remains an important area for future innovation.

Summary

Susceptibility-sensitive MRI protocols are readily implemented in clinical practice from existing manufacturer protocols and can yield two clinically useful biomarkers, the CVS and PRL. Both MRI markers are specific for MS over other MRI mimics. Emerging MRI biomarkers to detect chronic active lesions include both the PRL and SEL. These markers offer high promise of identifying smoldering inflammation as a novel therapeutic target likely contributing to progressive non-relapsing biology. PRL offer the additional advantages of being readily visualized at a single timepoint. TSPO-PET imaging also offers the possibility of identifying CAL from a single scan but is challenging to implement in clinical practice. Neuroimaging of the cortex and meningeal or lymphatic spaces are promising areas of ongoing innovation and translational development.

Synopsis:

Concepts of multiple sclerosis (MS) biology continue to evolve, with observations such as “progression independent of disease activity” challenging traditional phenotypic categorization. Iron-sensitive, susceptibility-based imaging techniques are emerging as highly translatable MRI sequences that allow for visualization of at least two clinically useful biomarkers: the central vein sign (CVS) and the paramagnetic rim lesion (PRL). Both biomarkers demonstrate high specificity in the discrimination of MS from other mimics and can be seen at 1.5-T and 3-T field strengths. Additionally, PRL represent a subset of chronic active lesions (CAL) engaged in “smoldering” compartmentalized inflammation behind an intact blood-brain barrier. CAL can also be visualized using translocator protein (TSPO)-based positron emission tomography (PET) or with conventional MRI as slowly expanding lesions, although the pathological interpretation is still under debate. Greater numbers of CAL identified in any of these forms are consistently associated with worse clinical status and more rapid disability accumulation. Here we review these emerging imaging biomarkers and their implications for MS clinical care.

Clinics Care Points

• Susceptibility-sensitive (T2*) imaging protocols offer substantial diagnostic benefit in discriminating MS from mimicking conditions.

• The central vein sign and paramagnetic rim lesions, both identified from T2* sequences, are highly specific diagnostic biomarkers of multiple sclerosis that can be readily visualized at clinically accessible field strengths (1.5T and 3T).

• Paramagnetic rim lesions and slowly expanding lesions both indicate chronic active lesion pathology and are consistently associated with worse clinical outcomes in MS.

• The treatment potential of PRL and SEL remains an area of active investigation.

Key Points:

• Susceptibility-sensitive MRI is a clinically important imaging sequence in MS

• The central vein sign and the paramagnetic rim lesion are both specific to discriminate MS from other MRI mimics

• Chronic active lesions can be identified in MS using different imaging MRI modalities and are likely an important source of ongoing injury to the central nervous system