Cerebrovascular Reactivity Assessment with Breath-Hold Functional MRI in Patients with Moyamoya Angiopathy: Which Time Period to Analyze?

TL;DR

This study finds that using short time periods during breath-hold fMRI improves the accuracy of cerebrovascular reactivity assessments in Moyamoya Angiopathy patients.

Contribution

The study identifies optimal time periods for analyzing breath-hold fMRI data to improve agreement with PET scans in Moyamoya Angiopathy.

Findings

Short time periods (±0s and ±1s) showed the highest agreement between bh-fMRI and [15O]water PET.

Agreement was significantly higher for short time periods compared to longer ones (≥ ±2s).

Inter-rater agreement was almost perfect for most time periods, except TPcereb.max ±1s.

Abstract

Background/Objectives: Quantifying cerebrovascular reactivity (CVR) is essential for stroke risk assessment in patients with Moyamoya Angiopathy (MMA). Breath-hold functional MRI (bh-fMRI) is an easily implementable method to assess CVR. Determining the optimal time period of the BOLD signal for analyzing the best bh-fMRI data quality remains an open question. Methods: A retrospective analysis of 46 bh-fMRI data sets of MMA patients was conducted. The percentage BOLD signal changes were evaluated at different time periods (time point of the maximum cerebellar signal peak (TPcereb. max) ± 0 s, TPcereb. max ± 1 s, TPcereb. max ± 2 s, TPcereb. max ± 3 s, TPcereb. max ± 4 s, TPcereb. max ± 5 s). The agreement between the bh-fMRI maps and [15O]water PET maps was independently and consensually rated on a 4-point Likert scale (1 = poor, 2 = moderate, 3 = good, 4 = excellent) and compared with the Friedman test. The inter-rater agreement was calculated separately for each time period using quadratic weighted Cohen’s kappa κw. Results: The selected time period had a significant impact on the agreement between bh-fMRI and [15O]water PET (χ2(5) = 79.448, p < 0.001, W = 0.345). Short time periods of TPcereb.max ± 0 s or TPcereb.max ±1 s demonstrated the highest level of concordance between bh-fMRI and [15O]water PET (median = 3.5 for TPcereb.max ± 0 s; median = 3 for TPcereb.max ± 1 s, modus = 4 in both cases). The agreement between bh-fMRI and [15O]water PET was significantly higher when evaluating time periods of TPcereb.max ± 0 s than when evaluating all time periods ≥ TPcereb. max ± 2 s. The inter-rater agreement was almost perfect for all time periods except one (TPcereb. max ± 1 s). Conclusions: Short time periods should be selected when evaluating CVR with bh-fMRI, as this study suggests a high level of validity in comparison to [15O]water PET.

Figures