How do you grade the severity of Acute Reversible Idiopathic Axial (ARIA) Edema (E) and Hemorrhage (H) on Magnetic Resonance Imaging (MRI)?

Grading Severity of ARIA-E and ARIA-H on MRI

ARIA-E and ARIA-H are graded as mild, moderate, or severe based on standardized MRI scoring systems that assess the extent and distribution of imaging abnormalities, with management decisions directly tied to these severity grades. 1

ARIA-E Severity Grading

Validated Scoring Systems

The most widely validated approach uses either a 3-point or 5-point severity scale (SSAE-3 or SSAE-5) based on the largest linear measurement of the lesion and its spatial distribution. 2 These scales have demonstrated excellent inter-reader reliability with Cohen/Fleiss kappa scores of 0.79 for SSAE-3 and 0.73 for SSAE-5, and have been used in most major Alzheimer's disease trials to date. 2

An alternative comprehensive scoring system uses a 6-point regional score for each of the 6 brain regions bilaterally (range 0-60), with separate assessments for:

• Parenchymal hyperintensity (increased T2 signal with potential mass effect) 3
• Sulcal hyperintensity (non-suppression of CSF signal on T2 FLAIR indicating sulcal effusions) 3
• Combined scores showing high inter-rater agreement (ICC = 0.89) 3

Clinical Severity Categories

Mild ARIA-E: Represents the majority of cases (50-73.8% depending on dose and APOE status), typically asymptomatic or minimally symptomatic 4

Moderate ARIA-E: Intermediate extent of parenchymal edema and/or sulcal effusions with variable symptoms 1

Severe ARIA-E: Extensive parenchymal involvement, significant mass effect, and typically symptomatic presentation requiring treatment modification 1

Key Imaging Features to Assess

• Anatomic location and distribution across brain regions 1
• Extent of parenchymal T2 FLAIR hyperintensity indicating vasogenic edema 1
• Presence and extent of sulcal FLAIR hyperintensity indicating effusions 1
• Associated mass effect on surrounding structures 1

Important caveat: Gyral swelling scores have lower inter-rater reliability (ICC = 0.54) and are not currently recommended for inclusion in severity grading. 3

ARIA-H Severity Grading

Classification Components

ARIA-H encompasses two distinct findings that should be assessed separately: 1

1. Microhemorrhages (mH): Focal, round, very low intensity lesions ≤10 mm diameter on T2* GRE or SWI sequences 1, 5
2. Superficial siderosis: Curvilinear low intensities on T2* sequences adjacent to brain surface 1, 5

Severity Grading Criteria

Mild ARIA-H:

• 1-4 new microhemorrhages 5, 6
• Minimal or no superficial siderosis 1

Moderate ARIA-H:

• 5-9 new microhemorrhages 5
• Limited superficial siderosis 1

Severe ARIA-H:

• ≥10 new microhemorrhages or presence of macrohemorrhages >10 mm diameter 5, 6
• Extensive superficial siderosis 1, 5

Critical threshold: The presence of ≥4 microhemorrhages at baseline or macrohemorrhages >10 mm are exclusionary findings for initiating anti-amyloid therapy due to significantly increased bleeding risk. 5, 6

Technical Requirements for Accurate Grading

Mandatory MRI Sequences

All three sequences are required for comprehensive ARIA assessment: 1

• T2 FLAIR: For detecting ARIA-E (parenchymal edema and sulcal effusions) 1
• T2 GRE or SWI:* For detecting ARIA-H (microhemorrhages and superficial siderosis) 1
• DWI: For excluding acute ischemic events and assessing restricted diffusion 1

3T MRI provides superior sensitivity compared to 1.5T for detecting microhemorrhages and should be used when available. 5, 6

Reading Standards

Central reading by trained neuroradiologists with specific ARIA expertise is strongly preferred over local reads to ensure accuracy and inter-rater reliability. 1 When local reads are necessary, radiologists must receive specific education about ARIA manifestations with multiple case examples, and a detailed reporting form with checkboxes should be provided. 1

Diagnostic certainty levels should be specified: 1

• "Definite" findings should guide clinical management decisions
• "Possible" findings should not be used as exclusionary criteria and may require serial imaging for clarification 1

Clinical Correlation and Management Implications

The relationship between ARIA-E and ARIA-H is clinically significant: In 49% of ARIA-E cases, there is associated appearance of ARIA-H, suggesting a common pathophysiologic mechanism related to vascular amyloid clearance and increased vascular permeability. 7 In patients without ARIA-E, the risk for incident blood products is only 4%. 7

Management is determined by both MRI severity grade and clinical symptoms: 1

• Detection of ARIA may require temporary or permanent cessation of anti-amyloid therapy 1
• Symptomatic moderate-to-severe ARIA-E may require corticosteroid treatment 8
• Severe ARIA-H or macrohemorrhages mandate permanent treatment discontinuation 5

Common pitfall: ARIA-E can present with subtle MR signal alterations that may be missed on routine reads, while fulminant ARIA-E can be misidentified as subarachnoid hemorrhage or venous infarction. 1 Serial comparison with baseline and prior scans is essential for accurate detection and grading. 1