Clinical Considerations in the Diagnosis and Management of Primary CNS Vasculitis

Primary central nervous system vasculitis (PCNSV) remains one of the more diagnostically complex conditions in autoimmune neurology, often presenting with heterogeneous clinical features that can mimic a range of vascular and inflammatory disorders. While advances in imaging and diagnostic tools have improved evaluation strategies, confirming the diagnosis and selecting appropriate treatment—particularly in hospitalized patients—continues to challenge clinicians in real-world practice.

At the 2026 American Academy of Neurology (AAN) Annual Meeting, held April 18-22 in Chicago, Illinois, Edith Graham, MD, presented data from a retrospective cohort of 47 hospitalized patients with PCNSV, evaluating clinical characteristics and outcomes associated with first-line induction therapies. In this cohort, cyclophosphamide was used in the majority of patients, while a smaller subset received rituximab or combination therapy. The study highlighted variability in diagnostic findings, including multifocal lesions in most patients and a notable proportion without CSF pleocytosis, underscoring the limitations of relying on a single diagnostic modality.

In this Q&A, Graham, an assistant Professor of Neurology at Northwestern Feinberg School of Medicine, discusses the early clinical features that should raise suspicion for CNS vasculitis, the challenges of distinguishing it from mimicking conditions, and the role of angiography, biopsy, and CSF analysis in diagnosis. She also outlines real-world considerations for treatment selection, including differences between cyclophosphamide and rituximab, timing of therapy initiation, and the need for safer, more effective long-term treatment strategies supported by prospective clinical trials.

Transcript edited for clarity.

NeurologyLive: When evaluating a hospitalized patient with suspected primary CNS vasculitis, what early clinical features should raise concern?

Edith Graham, MD: Thank you for highlighting this research that was done here in Chicago at Northwestern. I wanted to point out that this was a local retrospective study where we looked back at 20 years of hospitalized patients with CNS vasculitis. What we found was that the characteristics of patients who present with CNS vasculitis are similar to what has been published previously in other cohorts.

Instead of seeing acute stroke-like presentations that you would think of with typical ischemic strokes, we are seeing more of a subacute presentation over days, weeks, even months. Patients are having cognitive decline, focal weakness, numbness. What we found was that patients presenting with seizures were actually more of the minority, only about 10%, and fevers were also a very low percentage of patients. But more cognitive symptoms and even psychiatric presentations tended to predominate.

When we looked at imaging, the majority of patients had multifocal white matter lesions, but that was not the only finding. We had patients with leptomeningeal disease, patients with microhemorrhages and macrohemorrhages, and some with tumor-like lesions. So it is really a heterogeneous population with CNS vasculitis.

How challenging is it to accurately diagnose PCNSV given its heterogeneity and overlap with mimickers?

Absolutely, it is very difficult to diagnose CNS vasculitis. In our population, about 54% of patients underwent biopsy, and the yield in those patients was actually quite high, around 77%. But there are many steps before getting to biopsy, and that is not the first diagnostic approach.

All patients had some form of angiography. Noninvasive imaging like CTA or MRA was done in every patient, but those only detected vasculitis findings in about 32%, so they are not as sensitive as we would like. Diagnostic angiography was about twice as sensitive, with 64% of patients showing positive findings, so I strongly encourage providers to pursue diagnostic angiography when there is suspicion.

We now also have vessel wall imaging, which has been talked about a lot. We did not have many patients in our cohort with this modality due to the 20-year timeframe, but in other studies it can be helpful. However, there are important cautions regarding false positives, including enhancement from normal vasovasorum, leptomeningeal inflammation, infarcts, or hemorrhage. Vessel wall imaging can also be positive in atherosclerosis, which is one of the main mimickers we see now, particularly in younger patients.

Other mimickers include fibromuscular dysplasia and moyamoya. So it is essential to keep an open mind, as the diagnosis can evolve.

One very important point is that CSF analysis is a must. Typically, we expect pleocytosis, but we were surprised to find that several patients had completely normal CSF profiles yet still had biopsy-proven vasculitis. This may occur in patients with more focal disease rather than widespread involvement. So, no single test is sufficient—clinicians need to use a comprehensive approach.

How do you approach treatment selection and escalation in these patients?

In this study, we looked at 47 hospitalized patients with PCNSV. Our primary question was whether there was a difference in outcomes between cyclophosphamide and rituximab as induction therapy. All patients received steroids, which remain the standard initial treatment.

Cyclophosphamide has been the mainstay of treatment since the 1980s, first described in systemic vasculitis by Dr. Anthony Fauci, MD. Not much has changed in the last 40 years, which is a key issue.

In our study, we wanted to know whether rituximab could perform similarly. However, interpretation was limited because only 8 patients received rituximab monotherapy. Additionally, those patients were generally less severe at baseline and less likely to have biopsy-confirmed disease.

Cyclophosphamide works quickly, with lymphocyte depletion occurring within 10 to 14 days. However, it is more toxic. Using CTCAE grading, we found that severe adverse events, such as grade 4 neutropenia or sepsis, were more common with cyclophosphamide, whereas no rituximab patients experienced events above grade 3.

At 6 months, patients treated with cyclophosphamide had greater improvement in mRS compared with rituximab. At 12 months, the difference was no longer present. Since cyclophosphamide is typically given for about 6 months, this aligns with its expected benefit window.

The main takeaway is that we still lack optimal therapies for CNS vasculitis. We need randomized controlled trials and safer long-term treatment options, particularly those that do not carry risks like infertility or secondary malignancies.

Has the treatment paradigm evolved at all in recent years?

Yes, I think there have been some improvements. For example, IV cyclophosphamide is generally preferred over oral administration because it results in lower cumulative exposure. We also consider combining therapies early to achieve longer remission. Agents like mycophenolate are used as steroid-sparing or cyclophosphamide-sparing options, often for two to five years after initial presentation.

Another key point is dosing. In our study, doses of cyclophosphamide below 3 grams were not very effective. A single dose is unlikely to have a meaningful impact. Dosing strategies similar to rheumatologic protocols, such as 750 mg/m² monthly, are more effective at inducing remission.

So, if clinicians are going to use cyclophosphamide, it should be used at the appropriate dose—but not indefinitely.

What are the next steps from a research perspective?

The next step is multicenter collaboration to develop randomized clinical trials. We need to evaluate therapies beyond steroids and cyclophosphamide.

In rheumatology, complement inhibitors have been used in ANCA-associated vasculitis, and it is worth exploring whether similar approaches could be effective in CNS vasculitis. We also need CNS-penetrant therapies that can directly target inflammation within the brain, particularly since PCNSV is not associated with systemic disease.

Another important area is women’s health. We need therapies that are safer for fertility, as cyclophosphamide can impact fertility in both women and men. This is a major consideration when selecting treatment.

What are the challenges of conducting randomized trials in this population?

One of the biggest challenges is that these patients are often very sick, making randomization difficult. There is also no FDA-approved standard therapy, so designing a control arm can be ethically challenging. Any trial would likely need to include early rescue options. For example, if a patient does not respond after a defined period, such as four weeks, they should be able to receive cyclophosphamide.

This is similar to challenges seen in other rare neurologic diseases, where withholding treatment is not feasible. So trial design must carefully balance scientific rigor with patient safety.

Click here for more AAN 2026 coverage.