Angiographic assessment of the efficacy of flow diverter treatment for cerebral aneurysms

Abstract

Background: The recent growth of neuro-endovascular treatment has rekindled interest in the use of angiographic techniques for flow assessment. Aneurysm treatment with flow diverters is particularly amenable to such analysis. We analyze contrast time-density curves - recorded within aneurysms before (pre) and immediately after (post) flow diverter implantation to estimate six-month treatment outcomes.

Methods: Fifty-six patients with 65 aneurysms were treated with flow diverters at two institutions. A region of interest was drawn around the aneurysm perimeter in image sequences taken both pre and post angiography, and the temporal variation in grayscale intensity within the aneurysm (time-density curve) was recorded. Eleven parameters were quantified from each time-density curve. Aneurysm occlusion status was recorded six months post treatment. The change in parameters from pre to post treatment was statistically evaluated between aneurysm occluded and non-occluded groups.

Results: Of the 11 parameters, eight were significantly different before and immediately after flow diversion. Considering the entire data set, none of the parameters was statistically different between the occluded and non-occluded groups. However, subgroup analyses showed that four variables were significantly different between the aneurysm occluded and non-occluded groups. The sensitivity of these variables to predict aneurysm occlusion at six months ranged from 60% to 89%, while the specificity ranged from 55% to 70%.

Conclusions: Device-induced intra-aneurysmal flow alterations quantified by simple aneurysmal time-density curves can potentially be used to predict long-term outcomes of flow diversion. Large multi-center studies will be required to confirm these findings. Patient-to-patient variability in coagulation may need to be incorporated for clinically relevant predictive values.

Keywords: Concentration–time curve; contrast washout; time–density curve.

Figure 1.

(a) Time–density curve (TDC) recorded from region of interest (ROI; red boundary) drawn around aneurysm on angiographic sequence. delta: amplitude of curve. (b) Basic parameters measured on normalized curve. MTT: mean transit time or time coordinate of centroid of curve; TTP: time to peak intensity; AUC: area under the curve; WIS: wash-in slope; WOS: washout slope. (c) Curve fit with equation (inset) parameters: σ, μ, convection time constant (τ_conv), diffusion time constant (τ_diff), and convection amplitude (ρ_conv). A total of 11 parameters (gray boxes) were thus quantified for each curve (pre and post for 62 aneurysms = 124 curves).

Figure 2.

Several parameters were significantly different before (pre) and immediately after (post) flow diversion. ⁺Plot normalized by pre value for scaling. ***p < 0.0001, n = 62 for all cases; error bars represent standard error of the mean; y-axis units vary with variable.

Figure 3.

Examples of TDC analysis from two right carotid-ophthalmic segment aneurysms. Example #1 treated with two Pipeline embolization devices (PED) that showed persistent aneurysm filling at follow-up (a), and Example #2 treated with one PED that showed complete aneurysm occlusion at follow-up (b). Immediately after device deployment, the TDC in Example #2 shows a greater contrast washout delay compared to Example #1. This is reflected in the parameters obtained from TDC quantification (c). The ratios of post-deployment parameters to pre-deployment parameters are generally greater in Example #2 compared to Example #1, showing better flow diversion was achieved in this case.

Figure 4.

Subgroup analyses suggest four variables are statistically different between aneurysm occluded and non-occluded groups; subgroup numbers are explained in the text. Numbers within bars represent sample sizes; error bars represent standard deviation. The sensitivity (Sens) and specificity (Spec) of each variable to predict aneurysm occlusion at six months is noted; y-axis units vary with variable.