Inside the Heterogeneity of Primary CNS Vasculitis

Abstract

Background and Objectives

Primary CNS vasculitis (PCNSV) is a heterogeneous condition. This study examines a large cohort with long-term follow-up to identify potential disease subsets.

Methods

We retrospectively analyzed 216 patients with PCNSV (Mayo Clinic, 1983–2023), using standardized diagnostic criteria, classifying by vessel size, histopathology, and outcomes. Subsets and predictors of functional and therapeutic outcomes were evaluated.

Results

Diagnosis was based on cerebral angiography in 142 patients and histologically confirmed in 74. Isolated small vessel involvement was positively associated with mass-lesion presentation (odds ratio [OR] 19.38, p = 0.02), meningeal-enhancing lesions (OR 39.10, p < 0.0001), elevated CSF protein (OR 4.04, p = 0.03), and β-amyloid vascular deposits (OR 23.43, p = 0.0001), but negatively with focal manifestations (OR 0.32, p = 0.04) and cerebral infarcts (OR 0.22, p = 0.003). Lymphocytic vasculitis was linked to younger age at diagnosis (p = 0.006), longer symptom-to-diagnosis interval (p = 0.05), more seizures (p = 0.04), and lower disability (p = 0.003) and mortality (p = 0.008). Necrotizing vasculitis was associated with intracranial hemorrhage (p = 0.008). Two or more relapses occurred in 12.7%, associated with histologic diagnosis (OR 3.15, p = 0.009) and inversely with gadolinium-enhanced lesions (OR 0.33, p = 0.01). Therapy response occurred in 82.9%, long-term remission in 23.6%. Cerebral infarcts, especially multiple, were associated with poor therapy response (OR 0.11, p = 0.03). Histologic diagnosis was inversely associated with long-term remission (OR 0.44, p = 0.03), whereas aspirin use was positively associated (OR 2.8, p = 0.002). A rapidly progressive course occurred in 13.4% of patients and was linked to increasing age (OR 1.34/10 years, p = 0.04), cognitive dysfunction (OR 5.59, p = 0.02), cerebral infarctions (OR 5.02, p = 0.004), and large vessel involvement (OR 3.51, p = 0.02). Gadolinium-enhanced lesions (OR 0.36, p = 0.04) and aspirin (OR 0.42, p = 0.08) were protective. Mortality (21.3%) was associated with older age (HR 1.42, p = 0.002), cognitive dysfunction (HR 3.93, p = 0.006), and cerebral infarctions (HR 1.94, p = 0.03).

Discussion

PCNSV heterogeneity, driven by vessel size and histology, affects presentation and outcomes; our findings offer insights to improve diagnosis and treatment.

Introduction

Primary CNS vasculitis (PCNSV) is a rare and heterogeneous disease confined to the brain and spinal cord, with diverse clinical, radiologic, and histopathologic features and variable outcomes.^1-13 Although traditionally considered a single condition, PCNSV likely comprises distinct clinical subsets.^2,14-18 Biopsy is the diagnostic gold standard but is often avoided because of risk, while cerebral angiography, although less specific, is commonly used.² Reliance on angiography increases misdiagnosis, and mimicking conditions add to PCNSV's heterogeneity and diagnostic complexity. Variability in study design, expertise, and diagnostic protocols, particularly in retrospective, multicenter studies, along with small sample sizes and short follow-up, further limits diagnostic accuracy.

We analyzed a Mayo Clinic cohort of 216 patients with PCNSV identified over a 40-year period, defined by uniform radiologic and pathologic criteria, with detailed clinical data and long-term follow-up. This enabled evaluation of disease subsets by vessel size, histopathology, and outcomes, providing insights into PCNSV variability to improve diagnosis and guide treatment.

Methods

Identification of Patients

The study included an updated cohort of 216 consecutive patients with PCNSV evaluated at the Mayo Clinic over a 40-year period from 1983 to 2022. All patients were identified using the same predefined diagnostic criteria, which included^1,11,19 (1) a recent history or presence of an acquired neurologic deficit unexplained by other causes, (2) evidence of vasculitis in a CNS biopsy specimen, or (3) a cerebral angiogram with changes characteristic of vasculitis.

Conditions that can mimic PCNSV, such as reversible cerebral vasoconstriction syndrome, hypercoagulable states, monogenic disorders causing cerebrovascular disease, varicella zoster infection, other systemic or infectious vasculitides, and related disorders, were rigorously excluded. For cases with diagnostic uncertainty, 2 rheumatologists (C.S., G.G.H.) and a neurologist (R.D.B.) reviewed the medical records and reached a consensus.

Diagnostic histopathologic features of vasculitis included transmural vascular inflammation with damage to the vessel wall (with or without fibrinoid mural necrosis) involving leptomeningeal and/or parenchymal vessels.^9,20

Angiographic changes indicating a high probability of vasculitis included areas of smooth-wall segmental narrowing or dilation and occlusions affecting multiple cerebral arteries without the proximal vessel changes characteristic of atherosclerosis or other causes.¹

A neuropathologist (C.G.) reviewed all biopsy specimens, and a neuroradiologist (J.H.) reviewed all cerebral angiograms and other imaging studies. This comprehensive approach ensured a rigorous and uniform diagnostic process, enhancing the reliability of the study's findings.

Clinical Data Collection

Detailed information was collected from comprehensive Mayo Clinic medical records at diagnosis and during follow-up, including clinical manifestations, comorbidities, laboratory results, radiologic imaging, CNS biopsy or autopsy findings, therapies, relapses, functional status, and cause of death.^1,11

Data and causes of death were documented by reviewing clinical information and correspondence in the electronic medical records, including data from external centers accessed through Mayo Clinic's integrated Epic EHR system, which unifies patient information across all locations and affiliated centers.

Definitions

Patients with an abnormal cerebral angiogram were classified as having medium/large vessel involvement. Angiograms were further categorized into 2 groups^1,21: large artery involvement (intracranial internal carotid artery, basilar or distal vertebral arteries, and proximal segments of the anterior, middle, and posterior cerebral arteries) and medium artery involvement (intracranial second division branches and subsequent branches detectable by angiography).

Involvement of small cerebral cortical/leptomeningeal vessels, which are typically undetectable by conventional cerebral angiography, was considered in patients who either had a negative angiogram or in whom an angiogram was not performed, provided brain biopsy findings were positive.

Patients with angiogram consistent with PCNSV and negative biopsy were characterized as having isolated involvement of medium/large cerebral vessels, while patients with abnormal angiogram and positive biopsy were characterized as involvement of both small and medium/large cerebral vessels.

Three overarching patterns were identified: granulomatous (with or without β-A4 amyloid deposition), purely lymphocytic, and acute necrotizing.^9,20

Relapse was defined as the recurrence or worsening of symptoms or the progression of existing or new lesions on MRI while the patient was untreated or on a stable dose of medication. Diagnosis of relapse required a preceding period of clinical stability and escalating the current therapy or initiating new treatment. Patients with at least 2 relapses were classified as having a multiple relapsing course.

Long-term remission was defined as the absence of active PCNSV manifestations for at least 1 year after stopping therapy.

Response to therapy was based on the treating physician's global opinion, derived from a detailed review of medical records.

The degree of disability at admission and the last visit was determined through review of the medical record and categorized using the modified Rankin Scale.²²

The modified Rankin Scale was used to identify rapidly progressive disease, defined as Rankin 5 (severe disability) or 6 (death) because of PCNSV at diagnosis or at last follow-up. Patients with a Rankin score <5 at diagnosis were classified as rapidly progressive if they developed a score of 5 or 6 within 6 months of diagnosis.

Subjects were followed until the death or last follow-up visit. The median duration of follow-up of the entire cohort was 22.2 months (range 0–337 months).

The manuscript was prepared following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Statistical Analysis

We used a 2-sided, 2-sample t test to compare numerical parameters, or a Wilcoxon rank-sum test when the distributions were skewed. For categorical parameters, Fisher exact test was used. Rankin scores were dichotomized into 0–3 and 4–6, as the difference between these 2 categories is clinically relevant.²²

Logistic regression models were used to identify diagnostic characteristics that increased the likelihood of a poor outcome, experiencing at least 2 relapses, treatment response, long-term remission, and rapidly progressive disease course. In addition, these models were used to assess potential associations between diagnostic characteristics and the size of vessel involvement: comparing patients with isolated small vessel involvement with those with medium/large vessel involvement (with or without small vessel involvement), and comparing patients with small vessel involvement (with or without medium/large vessel involvement) with those with isolated medium/large vessel involvement. We also explored the relationships between diagnostic characteristics, outcomes, and histopathologic patterns. Both univariate and age-adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were reported.

We used the Cox proportional hazards model to assess the relationships between demographic, clinical, laboratory, radiologic, pathologic, and therapeutic parameters at diagnosis and patient survival. Univariate and age-adjusted hazard ratios (HRs) along with 95% CIs were calculated. When age was significantly associated with the outcome, results were reported as age-adjusted. All p values were two-sided, with statistical significance defined as p < 0.05. The analyses were performed using SAS version 9 (SAS Institute Inc, Cary, NC).

Standard Protocol Approvals, Registrations, and Patient Consents

The Mayo Clinic Institutional Review Board approved the study, deemed it minimal risk, and waived informed consent. Patients without research authorization for use of their medical records were excluded.

Data Availability

Data supporting the specific analyses reported herein are available on request from one of the authors (R.D.B.). The data are not publicly available because they contain information that could compromise the privacy of research participants.

Results

Between 1983 and 2022, 216 patients (120 female and 96 male) treated at Mayo Clinic (Rochester, MN) fulfilled the diagnostic criteria for PCNSV (Table 1). The median age at diagnosis was 49 years (range, 17–85 years). The median time between symptom onset and diagnosis was 0.1 year.

Table 1.

Findings at Diagnosis in the 216 Patients With PCNSV^a

	All patients (n = 216)	Patients diagnosed by biopsy (n = 74)	Patients diagnosed by cerebral angiography (n = 142)
Male	96 (44.4)	41 (55.4)$	55 (38.7)
Age at diagnosis, median (range), y	49 (17–85)	59 (17–84)	48 (17–85)
Interval from symptom onset to diagnosis, median (range), y	0.1 (0.0–10.1)	0.2 (0.0–10.1)	0.1 (0.0–5.2)
Clinical manifestations at presentation
Headache	122 (56.5)	40 (54.1)	82 (57.7)
Cognitive dysfunction	113 (52.3)	48 (64.9)^^	65 (45.8)
Persistent neurologic deficit or stroke	97 (44.9)	15 (20.3)&	82 (57.7)
Seizures	42 (19.4)	18 (24.3)	24 (16.9)
Intracranial hemorrhage	18 (8.3)	5 (6.8)	13 (9.2)
Systemic manifestationsb	18 (8.3)	8 (10.8)	10 (7.0)
Fever	17 (7.9)	9 (12.2)	8 (5.6)
CSF abnormalityc
White blood cells, median (range), cells/mm3,d	7 (0–615)	16 (0–535)&	5 (0–615)
Protein, median (range), mg/dLd	67.5 (15–1,034)	97.0 (21–1,034)&	57 (15–242)
White blood cells >10/mm3,d	74 (43.5)	35 (57.4)#	39 (35.8)
Protein >70 mg/dld	83 (48.8)	43 (70.5)&	40 (36.7)
Protein >45 mg/dL or WBC count >5 cells/mm3	141 (82.0)	56 (91.8)$	85 (76.6)
Protein >70 mg/dL or WBC count >10 cells/mm3	107 (62.9)	48 (78.7)^^^	59 (54.1)
ESR, median (range) mm/h	10.0 (0–124)	8 (0–110)	10 (0–124)
Initial MRI findingse
Cerebral infarction	112 (56.3)	21 (30.4)&	91 (70.0)
Gadolinium-enhanced lesions (parenchymal or meningeal)	76 (38.2)	51 (73.9)&	25 (19.2)
Parenchymal gadolinium-enhanced lesions	42 (21.2)	24 (35.3)&&	18 (13.8)
Meningeal gadolinium-enhanced lesions	42 (21.1)	32 (46.4)&	10 (7.7)
Angiographic findingsf
Large/proximal vessel vasculitis	100 (65.8)	5 (50.0)	95 (66.9)
Medium/distal vessel vasculitis	138 (90.8)	8 (80.0)	130 (91.5)

Abbreviations: ESR = erythrocyte sedimentation rate; PCNSV = primary CNS vasculitis; WBC = white blood cell.

^aExcepted where indicated otherwise, values are the number (%) of patients.

^bDefined as the presence of at least one of the following: fatigue, anorexia, weight loss, and arthralgia.

^cCSF data were available for 170 patients for protein >70 mg/dL, 172 patients for protein >45 mg/dL or white blood cell (WBC) count >5 cells/mm³, and 170 patients for protein >70 mg/dL or WBC count >10 cells/mm³.

^dNormal reference ranges for spinal fluid values were defined as 0–5 cells/mm³ for white blood cells and 14–45 mg/dL for protein.

^eMRI was performed in 199 patients.

^fCerebral angiography was performed in 172 patients, and information on vasculitis type (large vessel and medium vessel) was available for 152 patients. Significant comparisons between patients diagnosed by biopsy and those diagnosed by angiography: $p = 0.02, #p = 0.01, ^^p = 0.009, &p = 0.0001, ^^^p = 0.001, &&p = 0.0009.

Table 2 summarizes the diagnostic tests for PCNSV in the 216 patients. Diagnosis was based on cerebral angiography in 142 patients, including 111 without biopsy and 31 with negative biopsy results. Histologic examination confirmed vasculitis in 74 patients (72 throughout CNS biopsy and 2 at autopsy). Among these, 30 underwent cerebral angiography, showing abnormalities in only 9. In 21 cases, biopsy confirmed PCNSV despite a normal angiogram, while in 31 cases, the angiogram was positive, but biopsy was negative.

Table 2.

Results of Tests Used for the Diagnosis of PCNSV in 216 Patients

	Total (N = 216)
Test, n (%)
Angiogram positive, brain biopsy not done	111 (51.4)
Angiogram positive, brain biopsy negative	31 (14.4)
Brain biopsy positive, angiogram positive	9 (4.2)
Brain biopsy positive, angiogram negative	21 (9.7)
Brain biopsy positive, angiogram not done	44 (20.4)

Information on the type of vasculitis (medium and large vessels), based on cerebral angiography, was available for the 151 angiography-positive patients. Among them, 100 had large vessel involvement, with 13 showing only large vessel involvement, while 51 had exclusively medium vessel involvement.

Patients Diagnosed Histologically Compared With Those Diagnosed by Angiogram

Table 1 compares the clinical findings of 74 patients diagnosed histologically with those of 142 patients diagnosed by angiography. Biopsy-diagnosed patients more often showed cognitive dysfunction (p = 0.009), gadolinium-enhanced parenchymal (p = 0.0009) and meningeal (p = 0.0001) MRI lesions, and elevated CSF protein (p = 0.0001) and white blood cell counts (p = 0.0001). Furthermore, the prevalence of patients with noninflammatory CSF at diagnosis was significantly lower in histologically diagnosed patients compared with angiography-diagnosed patients (8.2% vs 23.4%, p = 0.002).

In contrast, persistent neurologic deficit and MRI evidence of infarcts were more frequent in angiography-diagnosed patients (p = 0.0001 for both). Univariate logistic analysis showed that a histologically proven diagnosis was significantly associated with a relapsing course (OR: 3.15, 95% CI 1.33–7.43, p = 0.009), while it was inversely associated with long-term remission (OR: 0.44, 95% CI 0.21–0.93, p = 0.03) (Table 3).

Table 3.

Characteristics Associated With Outcomes Related to Treatment Response

Characteristics	OR	95% CI	Univariate P
Relapsing course (at least 2 relapses) vs no relapsesa
Method of diagnosis (biopsy vs angiography)	3.15	1.33–7.43	0.009
Gadolinium-enhanced lesions (parenchymal or meningeal) on MRI	0.33	0.14–0.80	0.01
Gadolinium-enhanced meningeal lesions on MRI	0.23	0.08–1.35	0.08
Response to therapyb
Cerebral infarction on MRI	0.27	0.11–0.65	0.004
Multiple infarctions on MRI	0.11	0.01–0.83	0.03
Large vessel involvement on angiogram	0.43	0.18–1.06	0.07
Long-term remissionc
Method of diagnosis (biopsy vs angiography)	0.44	0.21–0.93	0.03
Aspirin as initial treatment	2.82	1.46–5.43	0.002
Rapidly progressive coursed
Age, per 10-y difference	1.34	1.02–1.77	0.04
Cognitive dysfunction	5.59	1.24–25.2	0.02
Cerebral infarctions on MRI	5.02	1.65–15.27	0.004
Gadolinium-enhanced lesions (parenchymal or meningeal)	0.36	0.13–0.98	0.04
Large vessel involvement on angiogram	3.51	1.22–10.10	0.02
Aspirin as initial treatment	0.42	0.16–1.11	0.08

Abbreviation: OR = odds ratio.

Univariate logistic regression models were used for age-adjusted analysis.

^aTwenty-six patients experienced at least 2 relapses during the follow-up period, while 139 patients did not have any relapses.

^bOne hundred sixty-five patients responded to the therapy, while 35 patients did not (data were not evaluable for 17 patients).

^cFifty-one patients had long-term remission defined as absence of manifestations of active PCNSV after therapy discontinuation for at least 1 y.

^dTwenty-nine patients had a rapidly progressive course defined by the presence of Rankin 5 (severe disability) or 6 (death) at diagnosis, or Rankin 5 or 6 at last follow-up (patients with Rankin <5 at diagnosis must develop a score of 5 or 6 within 6 mo after the diagnosis).

We also evaluated the frequency of aspirin use: aspirin treatment was significantly more frequent in patients diagnosed by angiography than in those diagnosed by biopsy (59/142, 41.5% vs 9/74, 12.2%; p = 0.0001).

Comparisons of Patients According to the Size of the Vessel Involved

Table 4 compares 3 groups of patients with different cerebral vessel involvement: 31 with isolated medium/large vessel involvement (angiogram+/biopsy-), 21 with isolated small vessel involvement (biopsy+/angiogram-), and 9 with involvement of both vessel types (biopsy+/angiogram+). Patients with isolated small vessel involvement showed the following significant differences compared with the other groups: less frequent persistent neurologic deficits (19.0% vs 44.4% vs 54.8%, p = 0.03) and lower likelihood of cerebral infarction on MRI (38.1% vs 88.9% vs 69.0%, p = 0.01) at diagnosis, lower mortality at last follow-up (14.3% vs 55.6% vs 22.6%, p = 0.05), and more frequent MRI evidence of meningeal involvement (52.4% vs 33.3% vs 6.9%, p = 0.002) and lower ESR levels (5.5 mm/h vs 26 mm/h vs 8 mm/h, p = 0.008) at diagnosis. These patients also demonstrated better therapeutic response (90% vs 42.9% vs 72.4%, p = 0.04). A trend toward older age at diagnosis (p = 0.08) and less rapid disease progression (p = 0.08) was also noted.

Table 4.

Findings at Diagnosis and Outcomes Related to the Size of the Vessel Involved (Medium/Large Vessels vs Small/Medium/Large Vessels vs Small Vessels)^a

	Angiogram+, biopsy- (medium/large vessels) (n = 31)	Biopsy+, angiogram+ (small/medium/large vessels) (n = 9)	Biopsy+, angiogram- (small vessels) (n = 21)	p Value
Male	16 (51.6)	6 (66.7)	10 (47.6)	0.62
Age at diagnosis, median (range) y	46 (21–74)	44 (26–66)	61 (20–78)	0.08
Interval from symptom onset to diagnosis, median (range) mo	1.5 (0.1–63)	2.5 (0.8–37)	2.4 (0.1–121)	0.36
Clinical manifestations at presentation
Headache	22 (71.0)	6 (66.7)	13 (61.9)	0.79
Cognitive dysfunction	14 (45.2)	6 (66.7)	14 (66.7)	0.24
Persistent neurologic deficit or stroke	17 (54.8)	4 (44.4)	4 (19.0)	0.03
Seizures	8 (25.8)	0	3 (14.3)	0.17
Intracranial hemorrhage	2 (6.5)	0	1 (4.8)	0.73
Systemic manifestationsb	4 (12.9)	2 (22.2)	6 (28.6)	0.37
CSF abnormality
Protein >45 mg/dL or WBC count >5 cells/mm3,c	21/27 (77.8)	8/8 (100.0)	17/19 (89.5)	0.24
ESR, median (range) mm/h	8 (2–107)	26 (8–35)	5.5 (1–10)	0.008
Initial MRI findings
Infarcts	20/29 (69.0)	8/9 (88.9)	8/21 (38.1)	0.01
Meningeal gadolinium-enhanced lesions	2/29 (6.9)	3/9 (33.3)	11/21 (52.4)	0.002
Rankin score at diagnosis
0–2	11 (35.5)	4 (44.4)	11 (52.4)
3	9 (29.0)	2 (22.2)	3 (14.3)	0.72
4–5	11 (35.5)	3 (33.3)	7 (33.3)
Outcomes
At least one relapse	13/30 (43.3)	2/8 (25.0)	7/20 (35.0)	0.60
At least 2 relapses	4/30 (13.3)	2/8 (25.0)	4/20 (20.0)	0.70
Response to therapy	21/29 (72.4)	3/7 (42.9)	18/20 (90.0)	0.04
Long-term remissiond	6/31 (19.4)	1/9 (11.1)	4/21 (19.0)	0.84
Rapidly progressive coursee	2/31 (6.5)	2/9 (22.2)	0/21	0.08
Rankin score 4–6 at last follow-up	10/31 (32.2)	6/9 (66.7)	5/21(23.8)	0.17
Death	7/31 (22.6)	5/9 (55.6)	3/21 (14.3)	0.05

Abbreviations: ESR = erythrocyte sedimentation rate; WBC = white blood cell.

^aExcept where indicated otherwise, values are the number (%) of patients.

^bDefined as the presence of at least one of the following: fatigue, anorexia, weight loss, and arthralgia.

^cThe normal range of protein is 14–45 mg/dL.

^dLong-term remission defined as at least 1 year of inactive disease without therapy.

^eRapidly progressive course defined by the presence of Rankin 5 (severe disability) or 6 (death) at diagnosis, or Rankin 5 or 6 at last follow-up (patients with Rankin <5 at diagnosis must develop a score of 5 or 6 within 6 mo after the diagnosis).

Patients with both medium/large and small vessel involvement tend to have a more severe and inflammatory disease, with higher ESR values (median 26 mm/h, range 8–35 mm/h) and greater likelihood of cerebral infarction on MRI (88.9%) at diagnosis, less frequent therapy response (42.9%), more likely to have rapid progression (22.2%), and increased disability (Rankin score 4–6) and mortality at last follow-up (66.7% and 55.6%, respectively) (Table 4).

At univariate analysis adjusted for age, patients with isolated small vessel involvement exhibited notable differences at diagnosis compared with those with medium/large vessel involvement (Table 5). Features positively associated with isolated small vessel involvement included mass lesion presentation (OR 19.4, p = 0.02), gadolinium-enhanced lesions (parenchymal or meningeal) (OR 21.2, p < 0.0001) and meningeal enhancing lesions (OR 39.1, p < 0.0001) on MRI, elevated CSF protein concentration (>70 mg/dL) (OR 4.0, p = 0.03), and histologic evidence of vascular deposits of β-amyloid peptide (OR 23.4, p = 0.0001). Conversely, focal manifestations (OR 0.32, p = 0.04) and cerebral infarctions on MRI (OR 0.22, p = 0.003) were negatively associated.

Table 5.

Characteristics of the Patients at Diagnosis Related to the Size of the Vessel Involved: Isolated Small Vessel Involvement vs Medium/Large Vessel Involvement and Small Vessel Involvement vs Isolated Medium/Large Vessel Involvement

Characteristics	OR	95% CI	Univariate P
Histologic +/angiogram - patients vs angiogram + patientsa (isolated small vessels vs medium/large vessels)
Focal manifestationsb	0.32	0.11–0.97	0.04
Mass lesion presentation	19.38	1.56–240.15	0.02
Cerebral infarctions on MRI	0.22	0.08–0.59	0.003
Gadolinium-enhanced lesions (parenchymal or meningeal) on MRI	21.16	5.82–76.99	<0.0001
Meningeal enhancing lesions on MRI	39.10	4.26–359.17	<0.0001
Total protein concentration in CSF >70 mg/dL	4.04	1.10–14.81	0.03
Histologic evidence of vascular deposits of β-amyloid peptide	23.43	4.99–110.08	0.0001
Histologic + patients vs histologic -/angiogram+ patientsc (small vessels vs isolated medium/large vessels)
Age at diagnosis, per 10-y difference	1.34	1.02–1.75	0.03
Cerebral infarctions at MRI	0.17	0.06–0.46	0.0005
Gadolinium-enhanced lesions (parenchymal or meningeal) at MRI	14.39	4.64–44.63	<0.0001
Meningeal enhancing lesions at MRI	11.98	1.96–73.10	0.007
Large vessel involvement at angiogram	2.80	1.04–7.52	0.04
Histologic evidence of vascular deposits of β-amyloid peptide	12.24	1.44–104.2	0.02
Cognitive dysfunction	3.93	1.49–10.3	0.006

Abbreviation: OR = odds ratio.

Univariate logistic regression models were used for age-adjusted analysis.

^aTwenty-one patients had histologic evidence of vasculitis with negative cerebral angiogram, while 151 patients had positive cerebral angiogram.

^bFocal manifestations included paraparesis or quadriparesis, ataxia, seizure, focal symptoms suggesting TIA, stroke (cerebral infarction or intracerebral hemorrhage), hemiparesis, visual field deficit, diplopia, monocular visual symptoms or amaurosis fugax, blurred vision or decreased visual acuity, parkinsonism or extrapyramidal signs, and aphasia.

^cSeventy-four patients had histologic evidence of vasculitis, while 31 patients had positive cerebral angiogram with absence of vasculitis on histologic examination.

Univariate analysis adjusted for age was also used to compare patients with small vessel disease at diagnosis, including those with concurrent medium/large vessel involvement, with those with isolated medium/large vessel involvement (Table 5). Small vessel vasculitis was positively associated with increasing age (calculated per 10-year increment) (OR 1.34, p = 0.03), cognitive dysfunction (OR 3.93, p = 0.006), gadolinium-enhancing lesions (parenchymal or meningeal) (OR 14.39, p < 0.0001) and meningeal enhancing lesions (OR 11.98, p = 0.007) on MRI, angiographic evidence of large vs medium vessel involvement (OR 2.8, p = 0.04), and histologic evidence of vascular β-amyloid deposits (OR 12.2, p = 0.02). Conversely, cerebral infarctions on MRI were negatively associated (OR 0.17, p = 0.0005).

Histopathologic Patterns and Patient Characteristics at Diagnosis and Outcomes

Histologic examination confirmed vasculitis in 74 patients. However, in 3 of these cases, the evidence of vasculitis was based on the original pathology report, as the histologic materials were not available for reexamination, unlike the remaining 71 cases. Among the 71 patients, granulomatous vasculitis was observed in 43, necrotizing vasculitis in 10, lymphocytic vasculitis in 17, and granulomatous and necrotizing vasculitis in one. Of these, 24 patients had Aβ-related angiitis (ABRA), which included 22 with granulomatous vasculitis and 2 with necrotizing vasculitis.

Table 6 includes all patients, including those with ABRA, comparing findings at diagnosis and outcomes across the 3 histopathologic patterns. We conducted the same analyses excluding patients with ABRA to compare findings at diagnosis and outcomes specifically among patients without this subtype.

Table 6.

Findings at Diagnosis and Outcomes Related to Histopathologic Patterns (Granulomatous vs Necrotizing vs Lymphocytic Vasculitis) Including Patients With Aβ-Related Angiitis (ABRA)^a

	Granulomatous vasculitis (n = 43)	Necrotizing vasculitis (n = 10)	Lymphocytic vasculitis (n = 17)	p Value
Male	28 (65.1)	4 (40)	9 (52.9)	0.30
Age at diagnosis, median (range), y	63 (26–84)	52 (30–73)	34 (17–79)	0.006
Interval from symptom onset to diagnosis, median (range), mo	2.2 (0.1–37)	1.8 (0.0–12)	11.2 (0.1–121)	0.05
Clinical manifestations at presentation
Headache	25 (58.1)	6 (60.0)	6 (35.3)	0.24
Cognitive dysfunction	31 (72.1)	7 (70.0)	8 (47.1)	0.17
Persistent neurologic deficit or stroke	8 (18.6)	3 (30.0)	4 (23.5)	0.71
Seizures	7 (16.3)	3 (30.0)	8 (47.1)	0.04
Intracranial hemorrhage	2 (4.7)	3 (30.0)	0	0.008
Systemic manifestationsb	8 (18.6)	4 (40.0)	4 (23.5)	0.34
CSF abnormality
Protein >45 mg/dL or WBC count >5 cells/mm3,c	35/37 (94.6)	8/8 (100.0)	9/12 (75.0)	0.07
ESR, median (range) mm/h	10 (1–110)	10 (1–89)	4 (1–41)	0.08
Initial MRI findings
Cerebral infarction(s)	16/43 (37.2)	3/9 (33.3)	2/17 (11.8)	0.15
Meningeal gadolinium-enhanced lesions	21/43(48.8)	5/9 (55.6)	3/17 (17.6)	0.06
Angiographic findingsd
Large/proximal vessel vasculitis	4/7 (57.1)	0/1	0/1	0.35
Medium/distal vessel vasculitis	5/7 (71.4)	1/1 (100.0)	1/1 (100.0)	0.69
Rankin score at diagnosis
0–2	20 (46.5)	5 (50)	12 (70.6)
3	5 (11.6)	0	2 (11.8)	0.29
4–5	18 (41.9)	5 (50)	3 (17.6)
Outcomes
At least one relapse	15/41 (36.6)	4/9 (44.4)	7/15 (46.7)	0.75
At least 2 relapses	9/41 (21.9)	2/9 (22.2)	4/15 (26.7)	0.91
Response to therapy	35/41 (85.4)	8/8 (100)	13/15 (86.7)	0.51
Long-term remissione	9/43 (20.9)	2/10 (20.0)	0/17	0.12
Rapidly progressive coursef	1/43 (2.3)	1/10 (10)	0/17	0.30
Rankin score 4–6 at last follow-up	17/43 (39.5)	3/10 (30)	0/17	0.003
Death	14/43 (32.6)	1/10 (10)	0/17	0.008

Abbreviations: ESR = erythrocyte sedimentation rate; WBC = white blood cell.

^aExcepted where indicated otherwise, values are the number (%) of patients.

^bDefined as the presence of at least one of the following: fatigue, anorexia, weight loss, and arthralgia.

^cThe normal range of protein is 14–45 mg/dL.

^dCerebral angiography was performed in 9 patients.

^eLong-term remission defined as at least 1 year of inactive disease without therapy.

^fRapidly progressive course defined by the presence of Rankin 5 (severe disability) or 6 (death) at diagnosis, or Rankin 5 or 6 at last follow-up (patients with Rankin <5 at diagnosis must develop a score of 5 or 6 within 6 mo after the diagnosis).

We also evaluated the frequency of aspirin treatment in histologically diagnosed patients, comparing the proportion receiving aspirin between those with and without CAA. No significant differences were observed (5/24, 20.8% vs 3/47, 6.4%; p = 0.109).

Analysis Including Patients With ABRA

When comparing histopathologic subtypes, patients with lymphocytic vasculitis had a significantly younger age at diagnosis (p = 0.006), experienced a longer interval from symptom onset to diagnosis (p = 0.05), and a higher frequency of seizures (p = 0.04) at diagnosis (Table 6). They also had significantly lower rates of high disability scores (Rankin score of 4–6) (p = 0.003) and mortality (p = 0.008) at last follow-up. Patients with necrotizing vasculitis had a significantly higher frequency of intracranial hemorrhage at diagnosis (p = 0.008), a presentation not observed in those with lymphocytic vasculitis.

Univariate logistic analysis adjusted for age comparing characteristics at diagnosis and outcomes of patients with granulomatous and necrotizing vasculitis with those with lymphocytic vasculitis showed a positive association with poor outcome (Rankin 4-6 at last follow-up) (OR 5.47, p = 0.02) and amyloid angiopathy (OR 8.62, p = 0.01) with granulomatous and necrotizing vasculitis, while treatment with glucocorticoids (GCs) alone was negatively associated (OR 0.33, p = 0.04) (Table 7). Although not statistically significant, there was a trend toward a positive association of male sex (OR 2.7, p = 0.06) and a negative association of gadolinium-enhanced lesions (parenchymal or meningeal) on MRI (OR 0.27, p = 0.06) with granulomatous and necrotizing vasculitis.

Table 7.

Characteristics of the Patients at Diagnosis and Outcomes Related to the Histopathologic Patterns

Characteristics	OR	95% CI	Univariate p
Granulomatous and necrotizing vasculitis vs lymphocytic vasculitis, including patients with ABRAa
Rankin score 4–6 at last follow-up	5.47	1.32–22.71	0.02
Histologic evidence of vascular deposits of β-amyloid peptide	8.62	1.53–48.49	0.01
Isolated glucocorticoid therapy	0.33	0.11–0.95	0.04
Gadolinium-enhanced lesions (parenchymal or meningeal) at MRI	0.27	0.07–1.09	0.06
Male sex	2.7	0.94–8.0	0.06
Granulomatous and necrotizing vasculitis vs lymphocytic vasculitis, excluding patients with ABRAb
Focal manifestationsc	0.23	0.06–0.87	0.03
Cerebral infarction(s) on MRI	3.68	0.97–13.91	0.05
CSF protein >45 mg/dL or WBC count >5 cells/mm3	5.59	0.96–32.53	0.05
Isolated glucocorticoid therapy	0.28	0.09–0.93	0.03
Rankin score 4–6 at last follow-up	4.73	1.02–22.04	0.04

Abbreviation: OR = odds ratio.

Univariate logistic regression models were used for age-adjusted analysis.

^aForty-three patients had histopathologic evidence of granulomatous vasculitis, 17 of lymphocytic vasculitis, 10 of necrotizing vasculitis, and 24 patients associated cerebral amyloid angiopathy (22 with granulomatous vasculitis and 2 with necrotizing vasculitis).

^bThe analysis excluded the 24 patients with Aβ-related angiitis (ABRA).

^cFocal clinical manifestations included paraparesis or quadriparesis, ataxia, seizure, focal symptoms suggesting TIA, stroke (cerebral infarction or intracerebral hemorrhage), hemiparesis, visual field deficit, diplopia, monocular visual symptoms or amaurosis fugax, blurred vision or decreased visual acuity, parkinsonism or extrapyramidal signs, and aphasia.

Analysis Excluding Patients With ABRA

Excluding patients with ABRA largely confirmed the overall findings (eTable 1). In cases of lymphocytic vasculitis, the previously observed younger age at diagnosis and longer interval from symptom onset to diagnosis were no longer statistically significant. However, the median age at diagnosis for patients with lymphocytic vasculitis (34 years; range 17–79) remained lower than that for patients with granulomatous (46 years; range 26–73) and necrotizing vasculitis (45 years; range 30–73), although this difference was not significant (p = 0.21). Notably, a higher frequency of seizures at diagnosis (p = 0.01), lower rates of severe disability (Rankin score 4–6, p = 0.01), and reduced mortality at last follow-up (p = 0.04) remained statistically significant. Similarly, patients with necrotizing vasculitis continued to show a higher frequency of intracranial hemorrhage at diagnosis (p = 0.007). Cerebral infarctions on MRI at diagnosis were observed more frequently in patients with granulomatous and necrotizing vasculitis (p = 0.05). Univariate analysis confirmed previous significant associations with granulomatous and necrotizing vasculitis and revealed a new negative association with focal manifestations at diagnosis (OR 0.23, p = 0.03) (Table 7). A positive association was observed between granulomatous and necrotizing vasculitis and both abnormal CSF findings (OR 5.59, p = 0.05) and cerebral infarctions on MRI (OR 3.68, p = 0.05).

Outcomes Related to Treatment Response

Relapsing Course

During follow-up, 65/204 (31.9%) patients experienced at least one relapse, while 26/204 (12.7%) patients experienced 2 or more relapses. No significant associations were found at univariate analysis adjusted for age between baseline characteristics and the occurrence of at least one relapse during follow-up. However, a multiple relapsing course defined by at least 2 relapses was significantly positively associated with the method of diagnosis using biopsy vs angiography (OR 3.15, p = 0.009) and negatively associated with gadolinium-enhanced lesions (parenchymal or meningeal) (OR 0.33, p = 0.01) (Table 3).

Response to Therapy and Long-Term Remission

Adequate information to judge the response to initial therapy was available in 199 patients. Overall, a favorable response was observed in 165 (82.9%). Long-term remission was observed in 51 (23.6%) patients, with no flares reported in this group. Oral cyclophosphamide (CYC) was used more frequently in patients with long-term remission than in those without (43.1% vs 29.1%, p = 0.161), although the difference was not statistically significant. However, patients in long-term remission had a significantly longer median duration of oral CYC therapy (11.5 months vs 6.0 months, p = 0.002). Univariate logistic analysis adjusted for age revealed that the presence of cerebral infarction on MRI at diagnosis (OR 0.27, p = 0.004), particularly when multiple (OR 0.11, p = 0.03), was associated with poor therapy response. Histologic diagnosis was associated with a lower probability of long-term remission compared with angiographic diagnosis (OR 0.44, p = 0.03), whereas aspirin therapy was positively associated (OR 2.82, p = 0.002) (Table 3). Accordingly, patients diagnosed by angiography had a higher likelihood of long-term remission (OR 2.22, 95% CI 1.06–4.65, p = 0.03).

Rapidly Progressive Course

A rapidly progressive course was observed in 29 (13.4%) patients. Features at diagnosis associated with a higher likelihood of rapidly progressive course included increasing age (calculated per 10-year increment) (OR 1.34, p = 0.04), cognitive dysfunction (OR 5.59, p = 0.02), cerebral infarctions on MRI (OR 5.02, p = 0.004), and large vessel involvement on angiogram (OR 3.51, p = 0.02), while gadolinium-enhanced lesions (parenchymal or meningeal) on MRI (OR 0.36, p = 0.04) and aspirin therapy (OR 0.42, p = 0.08) were associated with a lower risk (Table 3).

Status at Last Follow-up

eTable 2 provides a comparison of Rankin scores at presentation and last follow-up, categorizing disability into 3 groups at each time point. At diagnosis, scores were grouped as low (0–2), intermediate (3), and severe (4–5), while at last follow-up, scores were grouped as low or intermediate (0–3), severe (4–5), and death (6). Follow-up duration was divided into 3 intervals. Among the 141 patients with low or intermediate disability at diagnosis (score 0–3), 111 (78.7%) maintained similar scores at last follow-up. Of the 75 patients with severe disability at diagnosis (score 4–5), 43 (57.3%) showed improvement, achieving low or intermediate scores (0–3) at follow-up. During follow-up, 46 patients (21.3% of 216) died, with deaths equally distributed across all follow-up intervals.

Univariate logistic regression analysis adjusted for age assessed associations between specific findings at diagnosis and Rankin score outcomes (eTable 3). High disability scores (Rankin scores, 4–6) at last follow-up were significantly associated with increasing age (calculated per 10-year increments) (OR 1.41; p = 0.001), cognitive dysfunction (OR 3.0, p = 0.04), and cerebral infarctions on MRI (OR 1.89, p = 0.05) at diagnosis. The univariate Cox proportional hazards model adjusted for age analyzed associations between increased mortality and findings observed at diagnosis (eTable 3). Increased mortality was significantly associated with increasing age (calculated per 10-year increments) (HR 1.42, p = 0.002), cognitive dysfunction (HR 3.93, p = 0.006), and cerebral infarctions on MRI at diagnosis (HR 1.94, p = 0.03). No significant differences in high disability scores or mortality were observed when patients were stratified by treatment type (prednisone alone vs prednisone and cyclophosphamide).

Discussion

In 2010, we debated whether PCNSV represented a single disease or a syndrome with subsets.²³ Over the past 15 years, larger series have confirmed its heterogeneity, with subgroups showing different treatment responses and outcomes.

Retrospective studies link presentation, diagnosis, and outcomes to vessel size.^5-8,24-28 Small vessel disease, usually biopsy-diagnosed, presents with cognitive dysfunction, seizures, MRI-enhancing lesions, fewer cerebral infarctions, and CSF abnormalities but rarely vessel wall enhancement. These patients often experience a relapsing course.^5,8,24 Medium/large vessel disease, usually imaging-diagnosed, presents with persistent neurologic deficit, multiple cerebral infarctions, stenoses/occlusions, and vessel wall enhancement, and some patients follow a progressive, treatment-refractory course.^12,13 Noninflammatory CSF may be present in approximately 20% of patients with medium/large vessel involvement.

We analyzed 216 unselected patients with PCNSV seen over 40 years at a single institution, identified using uniform radiologic and pathologic criteria. An experienced neuropathologist and neuroradiologist reviewed all specimens and cerebral angiograms to confirm diagnosis and subtype classification.^9,20

With detailed clinical data and long-term follow-ups, we aimed to clarify PCNSV's complexity and identify homogeneous subsets. Histologic and angiographic diagnoses differed, reinforcing distinctions between small and medium/large vessel disease.^5-8,24-28 However, most prior studies lacked both evaluations, limiting accurate vessel size classification.

In our series, 61 patients had both histopathologic and angiographic evaluations at diagnosis, enabling precise comparison of isolated medium/large vessel disease (angiogram+/biopsy-), isolated small vessel disease (biopsy+/angiogram-), and combined involvement (biopsy+/angiogram+). Patients with isolated small vessel involvement form a distinct subset with characteristic presentation, CSF findings, MRI features, and outcomes. Compared with medium/large vessel disease, they had fewer persistent neurologic deficits and focal manifestations, more often presented with mass lesions and meningeal gadolinium-enhanced lesions, fewer cerebral infarctions, more CSF inflammatory findings, better treatment response, lower mortality, and a less frequent rapidly progressive course. Histologic β-amyloid peptide deposits in the vessel wall were strongly associated with this subset.

We compared all small vessel (biopsy+) patients, regardless of medium/large vessels involvement, with isolated medium/large vessel cases (biopsy-/angiogram+). Biopsy-positive patients were older (per 10-year increment), more often had cognitive dysfunction at presentation, more gadolinium-enhancing lesions, and fewer cerebral infarctions. Cerebral amyloid angiopathy, affecting small vessels, was common in this group. Thus, precise vessel size stratification confirmed small vessel involvement as a distinct subset of PCNSV.

We compared patients with both medium/large and small vessel involvement (biopsy+ and angiography+) with those with isolated small or isolated medium/large vessel disease. Patients with combined involvement had more severe and inflammatory disease, with higher ESR levels and greater likelihood of cerebral infarctions on MRI at diagnosis, lower therapeutic response, faster progression, and increased disability and mortality at last follow-up.

Of interest, patients with small vessel involvement (biopsy+) with concurrent medium/large vessel involvement showed more frequent involvement of large rather than medium cerebral vessels on angiogram than those with isolated medium/large vessel involvement (angiogram+/biopsy-), although this finding is limited by the small sample. Whether this represents a distinct, more severe vasculitis subset with a stronger inflammatory response requires further studies.

Our study highlights distinct clinical and prognostic differences among histopathologic subtypes. Lymphocytic vasculitis was associated with a more favorable prognosis, with no patients experiencing severe neurologic disability (Rankin 4 or 5) or death at last follow-up, whereas about one-third of patients with granulomatous or necrotizing vasculitis had severe disability or died. Patients with lymphocytic vasculitis were younger at diagnosis, experienced a longer diagnostic delay, and more often presented with seizures, while intracranial hemorrhage, associated with necrotizing vasculitis, was absent in this group. Previous studies link necrotizing vasculitis to intracranial hemorrhage²⁹ and spinal cord involvement¹⁰ and granulomatous vasculitis to Hodgkin lymphoma¹⁶ and CAA.^15,30

We recently reported our experience with spinal cord involvement in 10 patients with PCNSV.¹⁰ All of these patients were included in the present cohort and were analyzed in this study. By excluding patients with ABRA, we achieved a more accurate comparison of histopathologic patterns. This exclusion eliminated the statistical significance of younger age at diagnosis for lymphocytic vasculitis, although these patients remained younger than those with granulomatous or necrotizing vasculitis, whose ages were similar. This likely reflects the older age and granulomatous pattern of patients with ABRA,^30,31 as well as the smaller sample size.

We performed a univariate analysis comparing the more homogeneous group of granulomatous and necrotizing vasculitis patients with those having lymphocytic vasculitis, both including and excluding ABRA cases. Poor outcomes (Rankin scores 4–6) were associated with granulomatous and necrotizing vasculitis. These patients were less often treated with GC monotherapy and more frequently received combination therapy with immunosuppressants, typically cyclophosphamide (CYC). Excluding patients with ABRA, focal neurologic manifestations were negatively associated, whereas cerebral infarctions on MRI and abnormal CSF findings were positively associated with granulomatous and necrotizing vasculitis.

We compared ABRA with CAA without inflammation and PCNSV without CAA.¹⁵ Our findings, including the response to GCs alone or with CYC, suggest that ABRA may represent a distinct subset of PCNSV. In the short term, vascular inflammation, rather than Aβ deposition alone, drives disease manifestations. Subsequently, the progressive course of underlying CAA often results in severe disability and high mortality.^11,15 However, ABRA's classification remains uncertain.^11,15,29,30 Its similarities with perivascular CAA-related inflammation (CAA-RI) and overlap with CAA suggest that ABRA and CAA-RI may represent early, inflammation-dominant stages of the CAA disease spectrum.

Primary CNS vasculitis is a rare, heterogeneous disease with few studies on outcome predictors.^{5,8,11,24,25,32,33} Identifying patient subsets at diagnosis is crucial for tailoring treatment to disease severity. Although most patients respond well to GC alone or combined with CYC or mycophenolate mofetil, our study found considerable long-term morbidity and mortality. At last follow-up, one-fifth of patients had died, and nearly half with severe disability showed little improvement. About 10% exhibited a rapidly progressive course,^12,13 facing higher risks of disability, mortality, and poor therapy response. This subgroup requires aggressive initial immunosuppression, including GC pulses with high-dose oral GCs and CYC. Key features of this high-risk subgroup include older age, predominant large vessel involvement on angiography, cognitive dysfunction as the initial symptom, multiple cerebral infarctions on MRI, fewer gadolinium-enhanced lesions, and no initial aspirin use with immunosuppressive therapy. Early recognition is crucial for promptly initiating intensive therapy and improving outcomes.

Our study highlights the importance of initiating aspirin with immunosuppressive therapy in initial PCNSV treatment. Aspirin may reduce the risk of a rapidly progressive disease course and support long-term remission without increasing the risk of intracranial hemorrhage.¹¹ Prior studies show that aspirin decreases ischemic events in large vessel vasculitis and improves survival in giant cell arteritis.^34-36 Weyand et al.³⁷ found that aspirin inhibits cytokine transcription in inflamed arteries and enhances GC anti-inflammatory effects. Aspirin's antithrombotic and anti-inflammatory effects, combined with its synergy with GCs, likely explain its efficacy in PCNSV, supporting its use to improve long-term outcomes. Patients with medium/large vessel involvement were treated with aspirin 3.5 times more frequently than those with small vessel involvement. Among biopsy-positive patients, no differences in aspirin use were observed between those with and without CAA. In accordance with the European Stroke Organisation guidelines on PCNSV, we suggest that aspirin should be used primarily in patients with medium/large vessel involvement.³⁸

In our study, 32% of patients experienced at least one relapse, 13% had 2 or more relapses, and 24% maintained remission for at least 12 months after therapy discontinuation. Reported relapse rates in major series range from 12% to 59%,^{8,11,25,32,33,39} with the French cohort reporting long-term remission in 66% of patients.³² Variations may reflect differing definition of long-term remission. Few studies have assessed relapse and remission predictors at diagnosis. De Boysson et al. found that gadolinium-enhanced lesions were associated with a higher relapse risk and lower prolonged remission, while maintenance therapy favored sustained remission. Some studies associated relapse with small vessel involvement,^5,8,24 although the Cleveland Clinic group observed more frequent relapses in medium/large vessel disease.²⁵

In clinical practice, identifying patients with 2 or more relapses is essential for accurately defining relapsing disease and guiding management. We found a multiple-relapsing course was positively associated with small vessel vasculitis (histologic diagnosis) and negatively with gadolinium-enhanced lesions at diagnosis. Angiographic diagnosis and initial low-dose aspirin were linked to higher sustained remission. Patients achieving long-term remission had no flares and received longer initial oral CYC therapy, suggesting extended CYC exposure may better suppress inflammation. The higher likelihood of sustained remission in angiography-diagnosed patients may also reflect physicians' perception of more severe disease, leading to extended CYC use.

eTable 4 summarizes our PCNSV experience, one of the largest clinical series, highlighting subset characteristics by vessel size and histopathology, along with outcomes that underscore the disease's heterogeneity.

Our study has limitations but also notable strengths. As with all retrospective studies, incomplete datasets may have influenced findings. Referral bias is possible; the higher mortality rate in our series (21.3% vs 8% in French registry)³² may reflect Mayo Clinic's role as a tertiary referral center, where more severe and complex cases are seen. In contrast, the French PCNSV registry includes patients from multiple hospitals. Although most of our diagnoses were angiographic, the findings strongly suggested vasculitis, and long-term follow-up confirmed PCNSV. Variability in treatment regimens was another limitation. Strengths include a large, consecutive cohort defined by uniform radiologic and pathologic criteria, detailed clinical imaging and pathology data, and comprehensive long-term follow-up.

In conclusion, adult PCNSV is a heterogeneous disease, in which vessel size and histopathology influence presentation and outcome. Early recognition of high-risk patients is crucial for timely aggressive treatment and better long-term outcomes.

Glossary

ABRA

Aβ-related angiitis

CYC

cyclophosphamide

GC

glucocorticoid

HR

hazard ratio

OR

odds ratio

PCNSV

primary CNS vasculitis

Author Contributions

C. Salvarani: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. G.G. Hunder: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data. T. Christianson: analysis or interpretation of data. John Huston III: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data. C. Giannini: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data. R.D. Brown: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data.

Study Funding

The authors report no targeted funding.

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/NN for full disclosures.