Diagnostic performance of central vein sign and paramagnetic rim lesion integration for multiple sclerosis

Abstract

Background:

The central vein sign (CVS) and the paramagnetic rim lesion (PRL) are neuroimaging biomarkers of multiple sclerosis (MS).

Objectives:

To determine the diagnostic performance of CVS and PRL integration in people presenting for initial diagnostic evaluation of MS.

Methods:

Adults with clinical/radiological suspicion of MS, aged 18–65, with CVS and PRL assessment from the CentrAl Vein Sign in MS (CAVS-MS) pilot were included. Diagnostic performance of CVS and PRL combinations was evaluated, with 2017 McDonald criteria as the reference standard.

Results:

Seventy-eight participants were included (71% female, 86% white, 37 with MS). The combination of ⩾1 CVS and ⩾1 PRL demonstrated sensitivity/specificity of 0.76 (95% CI, 0.59, 0.88) and 0.93 (95% CI, 0.80, 0.98). The combination of ⩾6 CVS and ⩾1 PRL demonstrated sensitivity/specificity of 0.57 (95% CI, 0.39, 0.73) and 1.00 (95% CI, 0.91, 1.00). In those with cerebrospinal fluid testing (n = 46, 24 with MS), ⩾6 CVS and ⩾1 PRL diagnostic specificity was 1.00 (95% CI, 0.85, 1.00), similar to that of oligoclonal bands and dissemination in space by magnetic resonance imaging (0.82 [95% CI, 0.60, 0.95]).

Discussion:

Integrating CVS and PRL represents potentially advantageous MS diagnostic biomarkers, with increased specificity and without substantial reduction in sensitivity.

Keywords

Biomarker central vein sign diagnostic imaging multiple sclerosis paramagnetic rim lesion

Introduction

The central vein sign (CVS) and the paramagnetic rim lesion (PRL) are radiological hallmarks of lesions in multiple sclerosis (MS).¹ To minimize misdiagnosis, CVS and PRL are treated as diagnostic neuroimaging biomarkers for MS.² A meta-analysis demonstrated 92% specificity and 95% sensitivity with 40% CVS+ threshold for MS.³ Methods based on ⩾3 or ⩾6 CVS positive lesions (CVS+), termed Select 3 and Select 6, were explored as less time-consuming approaches, with the latter demonstrating comparable specificity to the 40% threshold.^4
–6 Although less prevalent in MS than CVS, PRL is rarely detected in non-MS conditions, with ⩾90% specificity.^7
–9 In prior studies, CVS and PRL were studied in combination with other MS findings, such as cortical lesions,^6,8 but CVS and PRL, which can be assessed with a single, clinically feasible, susceptibility-sensitive MRI scan,^1,10 remain scarcely explored as an integrated diagnostic biomarker. As a step toward improving the diagnostic process with the use of reliable neuroimaging biomarkers, the 2024 McDonald revisions enable the use of CVS and PRL to respectively secure a diagnosis of MS in combination with other clinical and/or imaging features.² However, the “Select 6” and “⩾1 PRL” rating methods, which are incorporated into the diagnostic framework, have not been studied in integration for diagnostic purposes in MS. In addition, studies assessing the diagnostic performance of CVS or PRL are often conducted in well-established and long-standing cohorts with MS or alternative diagnoses, which may not be representative of real-world performance in those presenting for the evaluation of MS.^6
–8,11

The CentrAl Vein Sign in MS (CAVS-MS) pilot was a multi-center cross-sectional study enrolling participants with clinical/radiological suspicion of MS, allowing the assessment of CVS and PRL in a setting similar to usual clinical referral.^5,12 In this cohort, Select 6 demonstrated a specificity of 98%,⁵ and ⩾1 PRL a specificity of 90%.¹² The integration of CVS and PRL as specific neuroimaging biomarkers for MS may further enhance their diagnostic performance and minimize false-positive findings. With the existing gap in literature and due to potential utility of integrated CVS and PRL to improve diagnostic specificity or require less lesions for assessment with a single scan, we explored the performance of these neuroimaging biomarker combinations for a diagnosis of MS in the CAVS-MS pilot.

Methods

CAVS-MS pilot

The cross-sectional CAVS-MS pilot study was approved by the institutional review board at all included sites, and consenting participants were enrolled. Participating institutions included: Cedars-Sinai Medical Center, Cleveland Clinic, Johns Hopkins University, University of California San Francisco, University of Southern California, University of Texas Health Science Center at Houston, University of Toronto–St. Michael’s Hospital, University of Vermont, and Yale University. Inclusion criteria were: referral to a participating academic site for a new clinical/radiological suspicion of MS, focal white matter MRI T2-weighted hyperintensities, and age 18–65 years (inclusive). Exclusion criteria were: contraindications or inability to tolerate MRI of gadolinium-based contrast agents, exposure to disease-modifying therapies, or exposure to systemic corticosteroids within 4 weeks before enrollment.

Current study design

The study recruited 97 participants, with five exclusions due to suboptimal imaging quality. The diagnosis of MS was adjudicated by three investigators (neurologists and a neurologist/neuroradiologist) according to the 2017 McDonald criteria, based on available clinical, imaging, and cerebrospinal fluid (CSF) testing results at the time of enrollment. Of 92 participants, 78 with MS or an alternative diagnosis were included for the main analysis, while 14 participants with a diagnosis of clinically isolated syndrome (CIS) or radiologically isolated syndrome (RIS) were excluded (Supplementary Figure 1). To assess the robustness of the main analysis findings, additional pre-planned sensitivity analyses were conducted by including the participants diagnosed with RIS or CIS, considering these as MS, and subsequently, as non-MS. A separate subgroup analysis was based on those with available CSF testing, which enabled comparisons of integrated CVS and PRL to oligoclonal bands (OCB) as a CSF biomarker. Finally, to assess for potential diagnostic performance differences from the standpoint of age, which may be associated with an increased number of white matter lesions (vascular or non-specific) or reduced utility of abbreviated CVS methods due to confluent lesions, biomarker combinations were additionally assessed in those younger or older than 50 years of age, respectively.

Lesion ratings

In addition to clinical, CSF, and imaging (brain MRI and/or spinal cord MRI) data available from diagnostic evaluation of each participant before enrollment, study assessments included MS disability measures and a 3-tesla MRI with a standardized protocol,^5,10,12,13 consisting of 3D sagittal acquisition of the entire brain for all sequences: precontrast T2-weighted FLAIR, precontrast and postcontrast T2*-weighted echo-planar imaging (3D-EPI), precontrast and postcontrast T1-weighted imaging. FLAIR* images were generated as previously published.^13,14 Postcontrast images were obtained after administration of a single-dose macrocyclic gadolinium-based contrast agent (0.1 mmol/kg) according to each study-site routine practice (gadoterate, gadoteridol, or gadobutrol).¹⁵ The CVS assessment was conducted on post-gadolinium FLAIR* images by site investigators and a central rater, and PRLs were assessed by three central raters and an adjudicator on phase images derived from T2*-weighted 3D-EPI (Figures 1 and 2). Central raters were blinded to the diagnostic adjudication, and a cloud-based centralized method was used to evaluate lesions systematically. Lesions were rated according to North American Imaging in MS (NAIMS) criteria for CVS and PRL^16,17 and as previously published.^5,12,15 Central ratings were used for the determination of CVS+ percentage at the scan level, while site CVS ratings were used for abbreviated CVS methods.

Details regarding participating institutions, study inclusion/exclusion criteria, participants’ diagnoses, and imaging protocol were also previously published.^5,12,13,18 Deidentified data not published within this article may be shared at the request of a qualified investigator.

Biomarker combinations and statistical analysis

Clinical and imaging features of interest were tabulated and stratified by diagnosis status. Continuous data were assessed with Mann–Whitney U test or t-test, and categorical with Fisher’s exact test. Combinations were generated using CVS methods (⩾1 CVS+ as Select 1, ⩾3 CVS+ as Select 3, ⩾6 CVS+ as Select 6, ⩾40% CVS+) and PRL methods (⩾1 PRL or ⩾2 PRLs). These combinations were motivated by previous results showing utility of such CVS and PRL approaches, in isolation, for diagnostic purposes in MS.^3,5,12,19 Abbreviated CVS methods (Select 1, Select 3, and Select 6) were rated uniformly across scans regardless of number of lesions per respective scan. Performance of combinations was compared to dissemination in space (DIS) on baseline MRI according to 2017 McDonald criteria, as well as to stand-alone methods. In those with available CSF testing, performance was compared to CSF-specific OCB. Diagnostic performance was described with sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy (proportion agreement), following calculations based on contingency tables. McNemar’s test was used for statistical comparisons. The chosen comparisons were focused and motivated by a pragmatic rationale employing CVS and PRL approaches previously determined to be useful as stand-alone biomarkers in MS. p-values were reported to a third decimal, and a correction according to the Benjamini-Hochberg (B-H) procedure was applied (α = 0.05). R Studio with packages tableone, DTComPair, and epiR was used for the analysis (version 4.4.2, R Foundation for Statistical Computing, Vienna, Austria).

Results

Of 37 participants with MS, 33 (89%) had ⩾1 CVS+, while 32 (87%) had ⩾1 PRL (Table 1). Of 41 participants with an alternative non-MS diagnosis (Supplementary Table 1), 27 (66%) had ⩾1 CVS+, while 4 (10%) had ⩾1 PRL. The latter four participants were diagnosed with non-specific white matter changes. There were no statistically meaningful differences for specificity between CVS methods (Select 1, Select 3, Select 6, ⩾40% CVS+) in combination with ⩾1 PRL (p-values ⩾0.05). Abbreviated CVS methods in combination with ⩾1 PRL demonstrated higher specificity and PPV than DIS by MRI (Table 2). The Select 6 and ⩾1 PRL combination demonstrated the highest point estimate for specificity (1.00), which increased that of stand-alone PRL (0.90, p = 0.046, non-significant after B-H correction) (Supplementary Table 2). Sensitivities between specified combinations with ⩾1 PRL (Select 1, Select 3, Select 6, ⩾40% CVS+, or DIS by MRI) were similar (non-significant p-values after B-H correction).

Table 1.

Characteristics of included participants, stratified by multiple sclerosis diagnosis status.

	MS (n = 37)	Non-MS (n = 41)	p-value
Age (median, IQR)	38 [33, 48]	52 [45, 57]	0.001^a
Female sex, n (%)	21 (56.8)	34 (82.9)	0.022^b
Race, n (%)			0.346^b
American Indian/Alaskan Native	1 (2.7)	0 (0.0)
Asian	0 (0.0)	1 (2.4)
Black or African American	4 (10.8)	2 (4.9)
Unknown	0 (0.0)	3 (7.3)
White	32 (86.5)	35 (85.4)
LatinX, n (%)	4 (10.8)	2 (4.9)	0.578^b
Weeks since first symptom onset, median [IQR]	53 [14, 204]	108 [42, 260]	0.263^a
Timed 25-Foot Walk, mean time in s (SD)	4.6 (1.1)	5.7 (1.4)	<0.001^c
Dominant 9-HPT, mean time in s (SD)	20.5 (5.1)	21.0 (4.0)	0.617^c
Binocular LCVA at 2.5%, mean n correct (SD)	36.0 (8.6)	36.9 (9.2)	0.638^c
Binocular LCVA at 1.25%, mean n correct (SD)	24.3 (11.0)	25.3 (12.8)	0.722^c
SDMT, mean n correct (SD)	51.9 (10.9)	50.3 (11.4)	0.540^c
EDSS, median [IQR]	1.50 [0.8, 2.0]	1.50 [1.0, 2.0]	0.553^a
CSF-specific oligoclonal bands present, n (%)	18/24 (75.0)	5/22 (22.7)	0.001^b
Number of T2 lesions, median [IQR]	37 [23, 74]	56 [20, 122]	0.286^a
DIS by MRI, n (%)	34 (91.9)	23 (56.1)	0.001^b
DIT by MRI, n (%)	26 (70.3)	6 (14.6)	<0.001^b
DIS and DIT by MRI	24 (64.9)	5 (12.2)	<0.001^b
Number of CVS+ lesions, median [IQR]	11 [3, 20]	2.00 [0, 4]	<0.001^a
Percentage of CVS+ lesions, median [IQR]	78.6 [62.5, 100.0]	7.7 [0.0, 20.0]	<0.001^a
⩾40% CVS+, n (%)	34 (91.9)	10 (24.4)	<0.001^b
Select 1, n (%)	33 (89.2)	27 (65.9)	0.017^b
Select 3, n (%)	30 (81.1)	13 (35.1)	<0.001^b
Select 6, n (%)	24 (64.9)	1 (2.4)	<0.001^b
⩾1 PRL, n (%)	32 (86.5)	4 (9.8)	<0.001^b
⩾2 PRL, n (%)	22 (59.5)	2 (4.9)	<0.001^b
Select 1 and ⩾1 PRL, n (%)	28 (75.7)	3 (7.3)	<0.001^b
Select 3 and ⩾1 PRL, n (%)	26 (70.3)	2 (4.9)	<0.001^b
Select 6 and ⩾1 PRL, n (%)	21 (56.8)	0 (0.0)	<0.001^b
⩾40% CVS+ and ⩾1 PRL, n (%)	29 (78.4)	2 (4.9)	<0.001^b
⩾40% CVS+ and ⩾2 PRL, n (%)	21 (56.8)	2 (4.9)	<0.001^b
DIS by MRI and ⩾1 PRL, n (%)	30 (81.1)	4 (9.8)	<0.001^b
DIS by MRI and ⩾2 PRL, n (%)	21 (56.8)	2 (4.9)	<0.001^b
⩾40% CVS+ and DIS by MRI, n (%)	31 (83.8)	4 (9.8)	<0.001^b

9-HPT: 9-Hole Peg Test; CSF: cerebrospinal fluid; CVS: central vein sign; DIS by MRI: dissemination in space by magnetic resonance imaging (according to 2017 McDonald criteria); DIT by MRI—dissemination in time by magnetic resonance imaging (according to 2017 McDonald criteria); DIS and DIT by MRI: dissemination in space and time by magnetic resonance imaging (according to 2017 McDonald criteria); EDSS: Expanded Disability Status Scale; IQR: interquartile range; LCVA: low-contrast visual acuity; MS: multiple sclerosis; PRL: paramagnetic rim lesion; SD: standard deviation; SDMT: Symbol Digit Modalities Test.

Mann-Whitney U test.

Fisher’s exact test.

t-test.

Table 2.

Diagnostic performance of biomarker combinations for multiple sclerosis (MS), in descending order by accuracy (n = 78, 37 with MS).

	Sensitivity	Specificity	PPV	NPV	Accuracy
⩾40% CVS+ and ⩾1 PRL	0.78 (0.62, 0.90)	0.95 (0.83, 0.99)	0.94 (0.79, 0.99)	0.83 (0.69, 0.92)	0.87 (0.77, 0.94)
⩾40% CVS+ and DIS by MRI	0.84 (0.68, 0.94)	0.90 (0.77, 0.97)	0.89 (0.73, 0.97)	0.86 (0.72, 0.95)	0.87 (0.78, 0.94)
DIS by MRI and ⩾1 PRL	0.81 (0.65, 0.92)	0.90 (0.77, 0.97)	0.88 (0.73, 0.97)	0.84 (0.70, 0.93)	0.86 (0.76, 0.93)
Select 1 and ⩾1 PRL	0.76 (0.59, 0.88)	0.93 (0.80, 0.98)	0.90 (0.74, 0.98)	0.81 (0.67, 0.91)	0.85 (0.75, 0.92)
Select 3 and ⩾1 PRL	0.70 (0.53, 0.84)	0.95 (0.83, 0.99)	0.93 (0.76, 0.99)	0.78 (0.64, 0.88)	0.83 (0.73, 0.91)
Select 6 and ⩾1 PRL	0.57 (0.39, 0.73)	1.00 (0.91, 1.00)	1.00 (0.84, 1.00)	0.72 (0.58, 0.83)	0.79 (0.69, 0.88)
⩾40% CVS+ and ⩾2 PRL	0.57 (0.39, 0.73)	0.95 (0.83, 0.99)	0.91 (0.72, 0.99)	0.71 (0.57, 0.82)	0.77 (0.66, 0.86)
DIS by MRI and ⩾2 PRL	0.57 (0.39, 0.73)	0.95 (0.83, 0.99)	0.91 (0.72, 0.99)	0.71 (0.57, 0.82)	0.77 (0.66, 0.86)
DIS and DIT by MRI	0.62 (0.45, 0.78)	0.88 (0.74, 0.96)	0.82 (0.63, 0.94)	0.72 (0.58, 0.84)	0.76 (0.65, 0.85)
DIS by MRI	0.92 (0.78, 0.98)	0.44 (0.28, 0.60)	0.60 (0.46, 0.72)	0.86 (0.64, 0.97)	0.67 (0.55, 0.77)
p-values for comparisons (McNemar’s test)
Select 1 and ⩾1 PRL vs. DIS by MRI	0.014	<0.001	<0.001	0.438
Select 3 and ⩾1 PRL vs. DIS by MRI	0.005	<0.001	<0.001	0.245
Select 6 and ⩾1 PRL vs. DIS by MRI	<0.001	<0.001	<0.001	0.057

CVS: central vein sign; DIS by MRI: dissemination in space by magnetic resonance imaging (according to 2017 McDonald criteria); DIS and DIT by MRI: dissemination in space and time by magnetic resonance imaging (according to 2017 McDonald criteria); NPV: negative predictive value; PPV: positive predictive value; PRL: paramagnetic rim lesion. Estimates provided as proportions with 95% confidence intervals. Bolded p-values indicate significance after correction for multiple comparisons.

Of the 14 participants with RIS or CIS included for the sensitivity analysis (Supplementary Table 3), 12 had ⩾1 CVS+ lesion, and 5 demonstrated ⩾1 PRL. When the RIS/CIS participants were included as MS, most combinations retained similar specificity point estimates (Supplementary Table 4). When the RIS/CIS participants were treated as non-MS, point estimates of specificities decreased but remained highest for Select 6 and ⩾1 PRL at 0.98 (Supplementary Table 5).

In those with available CSF testing (n = 46, 24 with MS), 23 were OCB+, Select 6 was present in 17 (37%), and 26 (57%) had ⩾1 PRL. In that subgroup, the Select 1 and ⩾1 PRL combination had similar sensitivity/specificity compared to OCB+ (Table 3); however, Select 6 and ⩾1 PRL maintained the highest point estimate for specificity (0.77 vs 1.00, p = 0.025, non-significant after B-H correction). The former combination had a similar sensitivity when compared to OCB+ and DIS by MRI (0.88 vs. 0.71, p = 0.045, non-significant after B-H correction). Following sensitivity analyses, the specificity of Select 6 and ⩾1 PRL remained perfect, whether additional participants were diagnostically considered as MS or non-MS (Supplementary Tables 6 and 7).

Table 3.

Diagnostic performance of biomarkers and combinations for multiple sclerosis (MS) in those with available cerebrospinal fluid testing results, in descending order by accuracy (n = 46, 24 with MS).

	Sensitivity	Specificity	PPV	NPV	Accuracy
⩾40% CVS+ and ⩾1 PRL	0.88 (0.68, 0.97)	0.95 (0.77, 1.00)	0.95 (0.77, 1.00)	0.88 (0.68, 0.97)	0.91 (0.79, 0.98)
DIS by MRI and ⩾1 PRL	0.96 (0.79, 1.00)	0.86 (0.65, 0.97)	0.88 (0.70, 0.98)	0.95 (0.75, 1.00)	0.91 (0.79, 0.98)
Select 1 and ⩾1 PRL	0.88 (0.68, 0.97)	0.91 (0.71, 0.99)	0.91 (0.72, 0.99)	0.87 (0.66, 0.97)	0.89 (0.76, 0.96)
⩾40% CVS+ and DIS by MRI	0.88 (0.68, 0.97)	0.91 (0.71, 0.99)	0.91 (0.72, 0.99)	0.87 (0.66, 0.97)	0.89 (0.76, 0.96)
Select 3 and ⩾1 PRL	0.83 (0.63, 0.95)	0.91 (0.71, 0.99)	0.91 (0.71, 0.99)	0.83 (0.63, 0.95)	0.87 (0.74, 0.95)
Select 6 and ⩾1 PRL	0.67 (0.45, 0.84)	1.00 (0.85, 1.00)	1.00 (0.79, 1.00)	0.73 (0.54, 0.88)	0.83 (0.69, 0.92)
⩾40% CVS+ and OCB+	0.71 (0.49, 0.87)	0.91 (0.71, 0.99)	0.89 (0.67, 0.99)	0.74 (0.54, 0.89)	0.80 (0.66, 0.91)
DIS by MRI and ⩾2 PRL	0.62 (0.41, 0.81)	0.95 (0.77, 1.00)	0.94 (0.70, 1.00)	0.70 (0.51, 0.85)	0.78 (0.64, 0.89)
OCB+	0.75 (0.53, 0.90)	0.77 (0.55, 0.92)	0.78 (0.56, 0.93)	0.74 (0.52, 0.90)	0.76 (0.61, 0.87)
OCB+ and DIS by MRI	0.71 (0.49, 0.87)	0.82 (0.60, 0.95)	0.81 (0.58, 0.95)	0.72 (0.51, 0.88)	0.76 (0.61, 0.87)
p-values for comparisons (McNemar’s test)
Select 1 and ⩾1 PRL vs. OCB+	0.083	0.083	0.044	0.036
Select 6 and ⩾1 PRL vs. OCB+	0.157	0.025	0.011	0.899
Select 1 and ⩾1 PRL vs. OCB+ and DIS by MRI	0.045	0.157	0.082	0.021
Select 6 and ⩾1 PRL vs. OCB+ and DIS by MRI	0.317	0.046	0.026	0.693

CVS: central vein sign; DIS by MRI: dissemination in space by magnetic resonance imaging (according to 2017 McDonald criteria); NPV: negative predictive value; OCB: cerebrospinal fluid-specific oligoclonal bands; PPV: positive predictive value; PRL: paramagnetic rim lesion. Estimates provided as proportions with 95% confidence intervals. Bolded p-values indicate significance after correction for multiple comparisons.

In those younger than 50 (n = 48, 28 with MS), performance of diagnostic biomarker combinations showed similar trends as the main analysis (Supplementary Table 8), while in those aged ⩾50 years (n = 30, 9 with MS), specificity was generally similar or with a lower point estimate (Supplementary Table 9), except Select 6 and ⩾1 PRL, which also demonstrated the highest accuracy (90%).

Discussion

Combinations of CVS and PRL increased diagnostic specificity for MS, particularly Select 6 and ⩾1 PRL, which had 100% specificity. This finding was consistent in the subgroup of those with available CSF results. Moreover, Select 6 and ⩾1 PRL was the most accurate (90%) of the assessed combinations in those ⩾50 years of age. Our findings support the integration of abbreviated counting methods for MS diagnosis.

Besides expanding on a previous retrospective study which reported 99% specificity and 59% sensitivity of the ⩾40% CVS+ and ⩾1 PRL combination for a diagnosis of MS,⁷ the current study was based on a different population, similar to usual clinical referral. This is particularly valuable from the standpoint of sampling validity, since a recent systematic review on abbreviated CVS counting methods and PRL as diagnostic biomarkers in MS pointed out considerable concerns for selection bias in most of the included studies.¹⁹ Although the PRL prevalence in those with a diagnosis of MS was higher than previously reported (86% vs 50%),⁷ our findings are similar to a recent Chinese MS cohort in which 104/120 demonstrated a PRL (87%) when assessed with 7-tesla MRI.²⁰ A systematic review and meta-analysis on patient-level PRL prevalence in MS, published in 2021, reported a pooled estimate of 40%, with a meta-regression as part of the same analysis showing a decrease in lesion-level PRL prevalence with increasing age or longer disease duration.²¹ These pooled results are based on studies conducted prior to the availability of a universal radiological definition of PRL, now published by NAIMS,¹⁷ which was observed in the current study. In retrospective cohorts and with the use of 3-tesla MRI, ⩾1 PRL prevalence in MS has been variable, for example, 23%²² vs. 52%,⁷ likely reflecting the impact of population factors and imaging assessment (use of susceptibility-weighted imaging vs. optimized T2*-weighted 3D-EPI). While a 1.5-tesla field strength may be adequate, the use of 3-tesla MRI with optimized susceptibility-based sequences (3D-EPI) as part of the CAVS-MS pilot protocol likely further increased the sensitivity for detecting a PRL. Current PRL ratings were completed by trained raters with the availability of an expert adjudicator, which improved the reliability of assessments but may have additionally contributed to a higher detection rate. Results also support that the average number of lesions per scan was adequate, whereas a low number of eligible lesions on an individual scan could decrease the likelihood of detecting a PRL or a specified number of CVS+ lesions. Automated tools for CVS and PRL detection are being developed, and their implementation may lead to more uniform lesion assessment.^23
–25

Study limitations include the cross-sectional design, a relatively small cohort size, and the possibility that some of those with CIS or RIS would be eventually diagnosed with MS or vice versa. In addition, CSF was not available from all participants, and associations with other measures of intrathecal inflammation such as kappa-free light chains were not investigated (not available at the time of the study). With the inclusion of adult population, it is unclear if these findings are generalizable to pediatric populations. Based on published case series and smaller pediatric MS cohorts, PRL was detected in 60%–70%,^26,27 while the CVS+ status based on 40% threshold or Select 6 was established in a majority (50%–100%).^27
–29

Although CVS and PRL were shown to be specific as stand-alone biomarkers, the combination of Select 6 and ⩾1 PRL demonstrated no false positives in the current study. Despite the sensitivity for this combination being lower than its specificity, the improved point estimate of sensitivity and overall accuracy in the subgroup of those ⩾50 years of age suggests additional relevance in presentations with later onset, which may be more challenging to diagnose. The proportion of CVS+ lesions in MS may decrease with older age and presence of vascular comorbidities,³⁰ which is also suggested by our results with the sensitivity of ⩾40% CVS+ in combination with ⩾1 PRL showing a lower point estimate in those aged 50 years or older than in younger patients. Conversely, the combination of Select 6 with ⩾1 PRL demonstrated a lower point estimate for sensitivity in those <50 years of age than in those aged 50 years or older. A different threshold with lower number of CVS+ lesions in combination with a PRL may be preferred for those with earlier presentations.

The combination of Select 6 and ⩾1 PRL may be diagnostically relevant for non-MS mimics where PRLs, but not CVS, have been detected, such as Susac syndrome, and for those that may be associated with a variable proportion of CVS+ lesions and potentially very large lesion counts, but not PRL, such as sarcoidosis, neuromyelitis optica spectrum disorder, or myelin oligodendrocyte glycoprotein antibody-associated disease.^6
–8,11,12 This combination, comprised of CVS and PRL methods already included in the 2024 McDonald revisions, demonstrated a perfect specificity across multiple analyses and may represent a pathway for reliable neuroimaging-based diagnosis of MS regardless of clinical features. The ongoing longitudinal CAVS-MS study with 400 participants is poised to provide more insights regarding diagnostic performance of combined CVS and PRL.³¹

With the evolution of MS diagnostic criteria, there has been a growing integration of imaging and CSF biomarkers as paraclinical evidence to support a diagnosis. The inclusion of CVS and PRL into the diagnostic algorithm as specific neuroimaging biomarkers for MS is expected to improve the diagnostic process.² Further exploration of using these biomarkers in combination is of potential value for future refinements.

Supplemental Material

sj-docx-1-msj-10.1177_13524585261434218 – Supplemental material for Diagnostic performance of central vein sign and paramagnetic rim lesion integration for multiple sclerosis

Supplemental material, sj-docx-1-msj-10.1177_13524585261434218 for Diagnostic performance of central vein sign and paramagnetic rim lesion integration for multiple sclerosis by Karlo Toljan, Brian Renner, Lynn Daboul, Melissa L Martin, Quy Cao, Carly M O’Donnell, Paulo Rodrigues, John Derbyshire, Christina J Azevedo, Amit Bar-Or, Eduardo Caverzasi, Peter A Calabresi, Bruce A C Cree, Leorah Freeman, Roland G Henry, Erin E Longbrake, Kunio Nakamura, Jiwon Oh, Nico Papinutto, Daniel Pelletier, Rohini D Samudralwar, Matthew K Schindler, Elias S Sotirchos, Nancy L Sicotte, Andrew J Solomon, Russell T Shinohara, Daniel S Reich, Pascal Sati and Daniel Ontaneda in Multiple Sclerosis Journal

Footnotes

Acknowledgements

This research was supported in part by the Intramural Research Program of the National Institutes of Health (NIH). The contributions of the NIH authors are considered works of the United States Government. The findings and conclusions presented in this paper are those of the author(s) and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services.

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Karlo Toljan is a training grant for National MS Society. Lynn Daboul is part of the NIH Medical Research Scholars Program.

Brian Renner: Nothing to disclose. Paulo Rodrigues is employed by and holds options of QMENTA. Christina J Azevedo is consulting for Genentech, EMD Serono, Alexion Pharmaceuticals, and Sanofi. Amit Bar-Or is consulting for Accure, Atara Biotherapeutics, Biogen, Bristol Myers Squibb, GlaxoSmithKline, Gossamer, Janssen, AstraZeneca, EMD Serono, Novartis, Genentech, and Sanofi, and receives research support from Biogen, Genentech, EMD Serono, and Novartis. Peter A Calabresi receives research support from Roche and is in the advisory board of Lilly, Biogen, Idorsia, Nervgen, and Vaccitech. Bruce A C Cree is consulting for Alexion, Atara, Biogen, EMD Serono, Novartis, Sanofi, and TG Therapeutics. Leorah Freeman is in the advisory board of Genentech, Novartis, and Celgene; is consulting for EMD Serono, Celgene, and Biogen; receives program sponsorship from Biogen and EMD Serono. Roland G Henry is consulting for Neurona, Roche, Novartis, Sanofi, QIA, Celgene/BMS, Atara, Medday, and Boston Pharma and receives research support from Roche and Atara. Erin E Longbrake is consulting for Genentech, Sanofi, Alexion, EMD Serono, Bristol Myers Squibb, Novartis, and TG Therapeutics and receives research support from Genentech and Biogen. Kunio Nakamura receives research support from the Department of Defense, Patient Centered Outcomes Research Institute, National MS Society, NINDS, NIAMS, Biogen, Bristol Myers Squibb, and Synaptogenix. Jiwon Oh receives research support from Biogen, Roche, and EMD Serono and is consulting for EMD Serono, Sanofi, Biogen, Roche, Celgene, and Novartis. Nico Papinutto receives research support from Race to Erase MS Foundation. Daniel Pelletier is consulting for Sanofi, Roche, and Novartis. Rohini D Samudralwar is in the advisory board of Biogen, EMD Serono, and Sanofi and is consulting for EMD Serono and Biogen. Elias S Sotirchos is consulting for Alexion, Viela Bio, Horizon Therapeutics, Genentech, and Ad Scientiam; receives honoraria from Alexion, Viela Bio, and Biogen. Nancy L Sicotte receives research support from NIH, National MS Society, Patient Centered Outcomes Research Institute, Race to Erase MS Foundation, and Biogen. Andrew J Solomon is in the advisory board of Genentech, Biogen, Alexion, Celgene, Greenwich Biosciences, and TG Therapeutics; is consulting for Octave Bioscience; does non-promotional speaking for EMD Serono; receives research support from Bristol Myers Squibb, Sanofi, Biogen, Novartis, Janssen, and Genentech; and receives trainee funding from Biogen. Russell T Shinohara is consulting for Octave Bioscience and receives research support from NIH and the National MS Society. Daniel S Reich is supported by Intramural Research Program of NINDS and receives research support from Abata and Sanofi. Pascal Sati is part of the National MS Society. Daniel Ontaneda receives research support from NIH, National MS Society, Patient Centered Outcomes Research Institute, Race to Erase MS Foundation, Genentech, Sanofi, and Novartis and is consulting for Biogen, Genentech, Sanofi, Janssen, Novartis, and Merck. All other authors have nothing to disclose.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the Race to Erase Multiple Sclerosis Foundation. Additional effort was funded by the National Institute of Neurological Disorders and Stroke Intramural Research Program, and the National Institutes of Health Medical Research Scholars Program (R01NS112274 and R01MH123550).

Ethical Considerations

The CAVS-MS pilot study was approved by the institutional review board at all included sites, and consenting participants were enrolled.

ORCID iDs

Karlo Toljan

Quy Cao

Christina J Azevedo

Amit Bar-Or

Eduardo Caverzasi

Peter A Calabresi

Bruce A C Cree

Kunio Nakamura

Jiwon Oh

Elias S Sotirchos

Andrew J Solomon

Daniel S Reich

Pascal Sati

Daniel Ontaneda

Data Availability Statement

Deidentified data not published within this article may be shared at the request of a qualified investigator.

Supplemental Material

Supplemental material for this article is available online.

References

Hemond

Gaitán

Absinta

, et al. New imaging markers in multiple sclerosis and related disorders. Neuroimaging Clin N Am 2024; 34(3): 359–373.

Montalban

Lebrun-Frénay

, et al. Diagnosis of multiple sclerosis: 2024 revisions of the McDonald criteria. Lancet Neurol 2025; 24: 850–865.

Castellaro

Tamanti

Pisani

, et al. The use of the central vein sign in the diagnosis of multiple sclerosis: A systematic review and meta-analysis. Diagnostics 2020; 10: 1025.

Solomon

Watts

Ontaneda

, et al. Diagnostic performance of central vein sign for multiple sclerosis with a simplified three-lesion algorithm. Mult Scler 2018; 24(6): 750–757.

Daboul

O’Donnell

Amin

, et al. A multicenter pilot study evaluating simplified central vein assessment for the diagnosis of multiple sclerosis. Mult Scler 2024; 30(1): 25–34.

Cagol

Cortese

Barakovic

, et al. Diagnostic performance of cortical lesions and the central vein sign in multiple sclerosis. JAMA Neurol 2024; 81: 143.

Maggi

Sati

Nair

, et al. Paramagnetic rim lesions are specific to multiple sclerosis: An international multicenter 3T MRI study. Ann Neurol 2020; 88(5): 1034–1042.

Borrelli

Martire

Stölting

, et al. Central vein sign, cortical lesions, and paramagnetic rim lesions for the diagnostic and prognostic workup of multiple sclerosis. Neurol Neuroimmunol Neuroinflamm 2024; 11(4): e200253.

Clarke

Pareto

Pessini-Ferreira

, et al. Value of 3T susceptibility-weighted imaging in the diagnosis of multiple sclerosis. AJNR: Am J Neuroradiol 2020; 41(6): 1001–1008.

10.

Sati

Thomasson

, et al. Rapid, high-resolution, whole-brain, susceptibility-based MRI of multiple sclerosis. Mult Scler 2014; 20(11): 1464–1470.

11.

Martire

Moiola

Rocca

, et al. What is the potential of paramagnetic rim lesions as diagnostic indicators in multiple sclerosis. Expert Rev Neurother 2022; 22(10): 829–837.

12.

Renner

Verter

Absinta

, et al. Frequency and diagnostic implications of paramagnetic rim lesions in people presenting for diagnosis to a multiple sclerosis clinic. Neurology 2025; 105: e213912.

13.

Martin

Cao

Luskin

, et al. Reliability of central vein sign imaging with 3T FLAIR* in a multicenter study. J Neuroimaging 2025; 35(1): e70011.

14.

Sati

George

Shea

, et al. FLAIR*: A combined MR contrast technique for visualizing white matter lesions and parenchymal veins. Radiology 2012; 265(3): 926–932.

15.

Daboul

O’Donnell

Cao

, et al. Effect of GBCA use on detection and diagnostic performance of the central vein sign: evaluation using a 3-T FLAIR* sequence in patients with suspected multiple sclerosis. AJR: Am J Roentgenol 2023; 220(1): 115–125.

16.

Sati

Constable

, et al. The central vein sign and its clinical evaluation for the diagnosis of multiple sclerosis: A consensus statement from the North American Imaging in Multiple Sclerosis Cooperative. Nat Rev Neurol 2016; 12(12): 714–722.

17.

Bagnato

Sati

Hemond

, et al. Imaging chronic active lesions in multiple sclerosis: A consensus statement. Brain 2024; 147: 2913–2933.

18.

Toljan

Daboul

Raza

, et al. Diagnostic performance of central vein sign versus oligoclonal bands for multiple sclerosis. Mult Scler 2024; 30(10): 1268–1277.

19.

Toljan

Amin

Ontaneda

Central vein sign and paramagnetic rim lesion abbreviated counting methods for diagnosis of multiple sclerosis: A systematic review and meta-analysis. Clin Radiol 2025; 88: 106984.

20.

Zhang

Gao

, et al. Brain lesion characteristics in Chinese multiple sclerosis patients: A 7-T MRI cohort study. Ann Clin Transl Neurol 2025; 12(2): 300–310.

21.

Ng Kee Kwong

Mollison

Meijboom

, et al. The prevalence of paramagnetic rim lesions in multiple sclerosis: A systematic review and meta-analysis. PLoS ONE 2021; 16: e0256845.

22.

Meaton

Altokhis

Allen

, et al. Paramagnetic rims are a promising diagnostic imaging biomarker in multiple sclerosis. Mult Scler 2022; 28(14): 2212–2220.

23.

Maggi

Fartaria

Jorge

, et al. CVSnet: A machine learning approach for automated central vein sign assessment in multiple sclerosis. NMR Biomed 2020; 33(5): e4283.

24.

Barquero

La Rosa

Kebiri

, et al. RimNet: A deep 3D multimodal MRI architecture for paramagnetic rim lesion assessment in multiple sclerosis. Neuroimage Clin 2020; 28: 102412.

25.

Ren

Chen

, et al. Automated segmentation of multiple sclerosis lesions, paramagnetic rims, and central vein sign on MRI provides reliable diagnostic biomarkers. Imaging Neurosci (Camb) 2025; 3: 932.

26.

Micheletti

Maldonado

Watal

, et al. Utility of paramagnetic rim lesions on 1.5-T susceptibility phase imaging for the diagnosis of pediatric multiple sclerosis. Pediatr Radiol 2022; 52(1): 97–103.

27.

Margoni

Preziosa

Pagani

, et al. Assessment of central vein sign and paramagnetic rim lesions in pediatric multiple sclerosis. Ann Clin Transl Neurol 2024; 11(11): 3031–3036.

28.

Boccia

Lapucci

Cellerino

, et al. Evaluating the central vein sign in paediatric-onset multiple sclerosis: A case series study. Mult Scler 2023; 29(3): 475–478.

29.

Kuchling

Koukou

Bartels

, et al. Prevalence of the central vein sign on susceptibility-weighted imaging in people with pediatric-onset multiple sclerosis. Neurology Open Access 2025; 1: e000031.

30.

Guisset

Lolli

Bugli

, et al. The central vein sign in multiple sclerosis patients with vascular comorbidities. Mult Scler 2021; 27(7): 1057–1065.

31.

Ontaneda

Sati

Raza

, et al. Central vein sign: A diagnostic biomarker in multiple sclerosis (CAVS-MS) study protocol for a prospective multicenter trial. Neuroimage Clin 2021; 32: 102834.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.19 MB