Diffusion Corrected Aneurysm Wall Permeability as a Measure of Rupture Risk
Charles G Cantrell1, Parmede Vakil1,2, Sameer A Ansari3, and Timothy J Carroll1,3

1Biomedical Engineering, Northwestern, Chicago, IL, United States, 2College of Medicine, University of Illinois, Chicago, IL, United States, 3Radiology, Northwestern, Chicago, IL, United States

### Synopsis

We report the first evidence of diffusion corrected DCE-MRI modeled contrast permeability in intracranial aneurysms and show the diffusion model more accurately represents physiology than previous methods. Permeability metrics, as measured in 23 patients, demonstrate a statistically significant trend with rupture risk as defined by anatomic imaging and clinical risk factors.

### Introduction

Intracranial aneurysms (IA) affect 2-6% of the population [1]. Rupture of IA carries a mortality rate of 50% and devastates the lives of many otherwise young healthy people who survive rupture. Unruptued IAs are often discovered incidentally and, given the inherent risks associated with surgical clipping or coiling, treatment remains controversial. The search for a “smoking gun” biomarker that predicts IA rupture conducted over the last fifteen years has, to date, been unsuccessful. The purpose of this study was to quantify IA wall permeability (ktrans,vL) to contrast agent utilizing a diffusion corrected Tofts’ model [2]. While previous work has shown ktrans to be associated with wall thinning and IA rupture risk, as defined by various anatomic, imaging and clinical risk factors, it employed a Tofts’ based permeability model [3]. For aneurysmal modelling, a diffusion model [4] more accurately represents physiology and provides better resolution for determining wall thickness and potential bleb formation zones.

### Methods

We employ a finite element model (FEM), to quantify leakage and compare to more traditional Tofts modeling. The FEM model improves on the permeability limited modified Tofts model by allowing contrast to flow from regions of high concentration towards low concentrations: $$\frac{dC(t)}{dt}= k^{trans}C_{p}(t)-k^{trans}\frac{C_{p}(t)}{v_{l}}+\sum_N(\frac{D_i+D_N}{2})\frac{1}{a^2}(\frac{C_N(t)}{v_{lN}}-\frac{C_i(t)}{v_{li}})+v_pC_p(t)$$ Ktrans and vL were evaluated as markers of rupture risk by comparing against established clinical (symptomatic lesions) and anatomic (size, location, morphology, multiplicity) risk metrics and evaluated against values obtained through utilization of the permeability limited Tofts’ Model. One patient opted out of surgical intervention and was imaged 18 months after initial evaluation.

### Patient Studies

Twenty-seven symptomatic, unruptured IAs in 23 patients (M/F, 10/13, <age> = 60.7 ± 12.3) were imaged with DCE-MRI, and wall permeability parameters (ktrans, vL) were calculated in regions adjacent to the aneurysm wall. Ktrans and vL were calculated using both a Tofts’ permeability model as described in [2] and a finite-element diffusion model. Correlation and mean values where compared to test the hypothesis that FEM analysis corrects for parenchymal diffusion of the contrast agent. Statistical significance was defined at the 5% level.

### Results

All IAs had a pronounced increase in wall permeability compared to the paired healthy MCA (p<0.001). Regression analysis demonstrated a significant trend toward increased ktrans with increasing aneurysm size (p<0.05). Diffusion corrected ktrans and vl values correlate strongly with the Tofts’ Model (p<.001) (Figure 1a-b). Diffusion corrected values showed greater permeability distinction between aneurysm wall and healthy vessel (p<.001) (Figure 1d)

### Conclusions

We report the first evidence of diffusion corrected DCE-MRI modeled contrast permeability in IAs. We found contrast agent permeability across the aneurysm wall correlated significantly with both aneurysm size as well as size-independent anatomic risk factors. Moreover, we note the diffusion correction shows greater distinction between permeable regions (i.e. leaky aneurysmal wall) and healthy tissue which can been seen in figure (1d). Additionally we see, in figure (1a-c), distinct regions of faster leakage even within a single aneurysm. Intuitively this makes sense, as we know physiologically aneurysms can be non-uniform. Not all aneurysms show a great discrepancy between model methods, however. Figure 2 shows a 61 year old male who decided to forgo surgical intervention. For this anecdotal case, both methods provided similar results that lined up exceptionally well with future bleb formation.

### Acknowledgements

NIH/NIBIB T32 EB005170, NIH/NHLBI R01 HL088437, AHA 14PRE20380810

### References

[1] Weir, JNS 2002 [2] Vakil, AJNR 2014, [3] Tofts, JMRI, 1999, [4] Fluckiger, Phys Med Bio, 2013

### Figures

Figure 1 (A-B) Pixel-wise comparison between diffusion corrected permeability values (y-axis) with Toft’s derived values (x-axis). (A) Ktrans (B) Vl. The line of unity is also shown. (C) Shows visual comparison between the two models. Notice increased localization when diffusion correction is added. (D) Mean and range of ktrans values surrounding aneurysmal wall and healthy vessel from both models.

Figure 2 The evolution of an untreated IA in a 61 year old male demonstrates that elevated ktrans at baseline correlated with bleb formation. Morphological changes over 1.5 years are observed in CTA (A) and DSA (B); both techniques for calculating ktrans, demonstrated two regions with high ktrans at the time of imaging initial imaging (C – Tofts) (D – Diffusion-corrected)

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4370