ARIA Education

Incidence and Risk Factors

ARIA occurs in 10–40% of patients treated with anti‑amyloid therapies, with rates varying by antibody and individual genetic risk profile. New ARIA‑H commonly co‑occurs with ARIA‑E. Each individual case should be assessed for risk factors, which could require more frequent MRI monitoring.

• Genetics: Apolipoprotein E4 (APOE ε4) genotype is one of the strongest predictors, with homozygous individuals carriers at highest risk
• Dose dependence: The risk of ARIA‑E increases with higher antibody doses
• Amyloid burden: Greater baseline amyloid‑β pathology appears to predispose to ARIA‑E
• Antithrombotic use: Some studies showed that antithrombotic medications are a significant risk factor for ARIA-H.
• CAA markers on MRI: Features of cerebral amyloid angiopathy (CAA)—such as microhemorrhages, cortical siderosis, enlarged perivascular spaces, and white‑matter hyperintensities—also predict ARIA. In clinical trials, patients with superficial siderosis or five or more microhemorrhages were excluded
  • CAA itself is caused by amyloid‑β accumulation in cerebral vessels and/or leptomeninges (the two innermost layers of the brain’s protective membranes), and may occur in AD or independently
  • CAA‑related inflammation (CAA‑ri) can lead to spontaneous, steroid‑responsive ARIA in patients with extensive vessel amyloid pathology.

Considerations for patients with existing cerebrovascular disease

ARIA detection

For detection, pre-treatment and on-treatment scans should be compared side-by-side, for example, using PACS viewing environments with automatic linking by slice position. It is recommended that reading is done by an experienced reader, preferably a (neuro)radiologist with expertise in reading ARIA safety scans. Standardized reporting and timely communication of ARIA findings are essential for adequate management of ARIA, which may involve treatment interruption or suspension.

Imaging protocols

The core sequences for ARIA detection are T2-FLAIR and GRE. 3T scanners are recommended, as they detect abnormalities with higher sensitivity, although 1.5T scanners are also acceptable. It is of note that classification of ARIA-H is based on 2D GRE data, as these sequences have primarily been used in ARIA clinical trials. Efforts should be made to follow each patient with the same acquisition protocol, scanner, and coil. Detailed acquisition protocols for each MRI vendor can be found here.

Imaging indications

1. Baseline assessment:
   • To identify any brain MRI findings that could contraindicate treatment or require more frequent monitoring
   • To identify T2-FLAIR hyperintensities and microhemorrhages to serve as a reference for comparison with safety monitoring scans
     
2. Asymptomatic ARIA monitoring:
   • Depending on the anti-amyloid drug being prescribed, a schedule of monitoring scans linked to the scheme of infusions should be initiated
   • This is particularly relevant in the first few months of treatment, since this is the period when the majority of ARIA cases occur
   • Continuing treatment in an asymptomatic patient with ARIA depends on the level of severity
     
3. Symptomatic patient under anti-amyloid treatment
   • See Differential diagnosis section

Proposed protocol ARIA monitoring

BASELINE	ASYMPTOMATIC	SYMPTOMATIC
2D T2-FLAIR (or 3D)1	2D T2-FLAIR (or 3D)1	2D T2-FLAIR (or 3D)1
T2* GRE	T2* GRE	T2* GRE
DWI	DWI	DWI
T2 Optional: SWI	Optional: SWI	+ Additional sequences as needed

¹Even though 2D FLAIR was the sequence used during trials, 3D T2-FLAIR is preferred over 2D for superior spatial resolution and uniform cerebrospinal fluid (CSF) suppression

DWI = Diffusion-weighted imaging

Core imaging features

ARIA – E (Edema/Effusion)

• Imaging appearance:
  • T2‑FLAIR: Hyperintense (bright) signals from interstitial edema and/or sulcal effusions
  • Conventional T2‑weighted: Parenchymal edema visible, but sulcal effusions resemble CSF
  • DWI/ADC: May show increased ADC signal (vasogenic edema) without restricted diffusion
    
• Morphology & distribution:
  • Lesions are typically ill‑defined and can be monofocal or multifocal
  • Most commonly affect posterior subcortical white matter (parietal/occipital), but may extend into deeper or posterior fossa structures
  • Mass effect frequently present; severe cases can appear pan‑hemispheric with midline shift
    
• Technical considerations:
  • Sulcal effusions are best detected on T2‑FLAIR—protein‑rich fluid remains hyperintense against suppressed CSF
  • 3D T2‑FLAIR is preferred over 2D for superior spatial resolution and uniform CSF suppression
  • In very mild cases, the only clue may be sulcal effacement (narrowed cortical sulci from partial CSF signal retention)
    

• Common interpretive errors and artifacts:
  • ARIA‑E can be difficult to differentiate from pre‑existing white matter hyperintensities—common in normal aging and cerebrovascular disease—since both appear as parenchymal hyperintensities on T2‑FLAIR images.
  • Care should be taken to avoid mislabeling bright CSF signals in the orbitofrontal region and around the brainstem, where CSF flow artifacts can be bright depending on vendor-specific implementations

ARIA‑H (Hemosiderin)

• Imaging appearance:
  • Susceptibility‑sensitive sequences (T2* GRE or SWI) reveal new hypointense (dark) foci
  • Microhemorrhages: Round lesions 3–10 mm in diameter, typically lobar (cortico‑subcortical) and often clustered when multiple
  • Superficial siderosis: Linear or curvilinear hypointensities along the pial surface of one or more sulci, producing a “tram‑track” pattern
    
• Distribution:
  • Microhemorrhages: Predominantly near the cortical ribbon, mimicking the pattern seen in CAA
  • Siderosis: Usually confined to one sulcus but can be multifocal
    
• Common interpretative errors and artifacts:
  • Flow voids: Take care not to mistake normal vessel flow‑voids for microhemorrhages—correlate with T2 or time‑of‑flight sequences
  • Susceptibility artifact: Beware of dark areas near the skull base or from patient motion
  • Venous structures: Subdural or subarachnoid veins track continuously to the dura and remain unchanged on serial scans; true ARIA‑H lesions are new and non‑anatomical

Differential diagnosis of ARIA