Please Wait...

Validation of a simple severity scale for assessing ARIA-E

L. Bracoud, J. Fiebach, D. Purcell, E. Gaensler, A. Gass, C. Lindan, J. Barakos and J. Suhy


Vasogenic edema and sulcal effusions, resulting from leakage of proteinaceous fluid in the brain parenchyma and leptomeningeal spaces, became a topic of interest several years ago when it became apparent that it was a somewhat common adverse event (AE) in amyloid-lowering therapies, where it was later hypothesized to be caused by the removal of amyloid plaque within the brain. It was even given a dedicated name, ARIA-E [Sperling, Lancet Neurol 2012], and began to be systematically monitored in all such subsequent clinical trials. While those findings can be subtle and require thorough training to be detected, one also needs to standardize the way their severity is reported, since those may lead to suspension/withdrawal from treatment depending on severity and potential clinical symptoms. This work assessed the robustness of two variations of a simple scale, which would be easy to implement while still providing enough granularity for proper AE management.


MRI scans (baseline and follow-up FLAIR data) from 39 subjects with no, mild, moderate or severe ARIA-E were reviewed twice each, at 5-month intervals, by 6 blinded neuroradiologists with experience in ARIA-E monitoring.

A 3-point severity scale was defined by assessing the extent of ARIA-E findings, e.g. parenchymal and/or sulcal hyperintensities with or without gyral swelling and sulcal effacement, affecting an area of <5 cm (single=mild, multiple=moderate), 5-10 cm (moderate) or >10 cm (severe) in single greatest dimension. 2 sub-levels, mild+ and moderate+, were added by scoring whether findings were mono- or multi-focal (see Table 1).

Intra- and inter-reader agreements were calculated using ICC, and Cohen’s/Fleiss’ Kappa respectively.


A high overall inter-reader agreement was observed both for the 3-level (ICC=0.94, 95%CI [0.91,0.96]) and 5-level (0.94 [0.92,0.97]) scales. Intra-reader agreement was equally high. Kappa statistics confirmed substantial/almost perfect agreement. See Table 2 for detailed results.


Both proposed scales provide a simple severity rating, based on a single overall assessment of ARIA-E extent, with a high degree of agreement among readers. Such rating is sufficient for proper treatment management in subjects experiencing ARIA-E. Those scales are clinically relevant, reliable, valid and easy to use, which are key aspects for future applications.

Table 1 – Scale definition

ARIA-E Extent ARIA-E Focality 3-point scale 5-point scale
No ARIA-E N/A 0 0
< 5 cm Monofocal 1 (Mild) 1 (Mild)
Multifocal 2 (Moderate) 2 (Mild+)
5-10 cm Monofocal 3 (Moderate)
Multifocal 4 (Moderate+)
> 10 cm Monofocal 3 (Severe) 5 (Severe)

Table 2 – Inter- and Intra-reader ICC and Cohen's Kappa (inter-reader) and Fleiss' Kappa (intra-reader) [95% CI] results

Scale Inter 1 Inter 2 Intra 1 Intra 2 Intra 3 Intra 4 Intra 5 Intra 6
3-level ICC = 0.93 [0.90,0.96]
ΚF = 0.95 [0.88, 1.00]
ICC = 0.93 [0.90,0.96]
ΚF = 0.84 [0.78, 0.90]
ICC = 0.98 [0.97,0.99]
ΚC = 0.92 [0.82, 1.00]
ICC = 1.00 [1.00,1.00]
ΚC = 1.00 [1.00, 1.00]
ICC = 0.98 [0.96,0.99]
ΚC = 0.95 [0.87, 1.00]
ICC = 1.00 [1.00,1.00]
ΚC = 0.85 [0.70, 1.00]
ICC = 0.95 [0.91,0.97]
ΚC = 1.00 [1.00, 1.00]
ICC = 0.96 [0.93,0.98]
ΚC = 0.91 [0.80, 1.00]
5-level ICC = 0.93 [0.89,0.96]
ΚF = 0.72 [0.65, 0.80]
ICC = 0.96 [0.93,0.97]
ΚF = 0.72 [0.65, 0.80]
ICC = 0.98 [0.97,0.99]
ΚC = 0.90 [0.79, 1.00]
ICC = 0.98 [0.95,0.99]
ΚC = 0.93 [0.85, 1.00]
ICC = 0.98 [0.96,0.99]
ΚC = 0.87 [0.76, 0.99]
ICC = 0.99 [0.98,0.99]
ΚC = 0.81 [0.67, 0.94]
ICC = 0.94 [0.89,0.97]
ΚC = 0.97 [0.90, 1.00]
ICC = 0.97 [0.95,0.99]
ΚC = 0.81 [0.67, 0.94]


Leader in Clinical Trial
Management Solutions

Successful clinical trials require the ability to see key details and uncover hidden insights. Bioclinica utilizes science and technology to bring clarity to clinical trials, helping companies to develop new life-improving therapies more efficiently and safely.

Think a #CTMS implementation takes too much time and too many resources? It doesn’t have to. Learn how a configurab…
Bioclinica (2 days ago)
We think you should have access to the full suite of #ClinicalDataManagement services to ensure high-quality, relia…
Bioclinica (2 weeks ago)
A mid-study change to a new #IRT vendor can be required for a number of reasons and can introduce a certain amount…
Bioclinica (2 weeks ago)
A clinical trial rescue or transition presents a unique set of challenges for study teams. Join us on July 21 for a…
Bioclinica (3 weeks ago)
Rapid #CTMS implementation shouldn’t mean sacrificing features and functionality. With a modular approach, a full-f…
Bioclinica (4 weeks ago)
How prepared are you for a #clinicaltrial transition? With proper planning, the impact to sites and patients can be…
Bioclinica (5 weeks ago)