Entry remnants in flow-diverted aneurysms: Does branch geometry influence aneurysm closure?

Abstract

Objective

Numerous studies have suggested a relationship between delayed occlusion of intracranial aneurysms treated with the Pipeline Embolization Device (PED) and the presence of an incorporated branch. However, in some cases, flow diversion may still be the preferred treatment option. This study sought to determine whether geometric factors pertaining to relative size and angulation of branch vessel(s) can be measured in a reliable fashion and whether they are related to occlusion rates.

Methods

Eighty aneurysms treated at a single neurovascular center from November 2008 to June 2014 were identified. Two blinded raters prospectively reviewed the imaging performed at the time of the procedure and measured the following geometric variables: inflow jet/incorporated branch direction angle and branch artery/ parent artery ratio. Delayed occlusion was defined as the absence of complete aneurysmal occlusion at one year. Analysis was performed using logistic regression and intra-class correlation co-efficient (ICC).

Results

Twenty-four (30%) aneurysms with 28 incorporated branches were identified. A trend toward higher inflow jet/incorporated branch direction angle was found in the group of aneurysms demonstrating delayed occlusion when compared to the group with complete occlusion. ICC revealed high correlation. Overall lower one-year occlusion rates of 53% versus 73% for aneurysms with and without incorporated branches, respectively, were also noted.

Conclusions

The presence of an incorporated branch conferred a 20% absolute risk increase for delayed aneurysmal occlusion. Incorporated branches with a larger angle between the inflow jet and the incorporated branch direction exhibited a trend toward lower occlusion rates. This might be further investigated using a multicenter approach in conjunction with other potentially relevant clinical and angiographic variables.

Introduction

Numerous studies have suggested a relationship between delayed occlusion of intracranial aneurysms treated with the Pipeline Embolization Device (PED; Covidien-Medtronic, Irvine, CA) and the presence of an incorporated branch, regardless of aneurysmal location.^1–5 In some cases, however, flow diversion may still represent the preferred treatment modality, such as in cases of difficult surgical access or branch vessel incorporation into the aneurysm precluding standard or adjunctive coiling. In situations of delayed aneurysm occlusion with branch incorporation, careful inspection of the catheter angiograms often demonstrates the appearance of abnormal flow of contrast through the aneurysm body “entry remnant” and then into the incorporated branch.

The pattern of flow into the aneurysm entry remnant may be a risk factor for treatment failure and may be influenced by geometric and hemodynamic considerations related to the outflow branch(es) and their relationship to the parent artery. Although this can be modeled with computational fluid dynamics (CFD), in practice, it is not readily available to clinicians, and different modeling strategies can produce varying results. This study sought to determine whether geometric factors that can be easily derived from the catheter angiogram pertaining to relative size and geometry of branch vessel(s) can be measured in a reliable fashion and whether they are related to occlusion rates at one year following treatment. It was hypothesized that a larger relative size of the outflow branch(es) and a larger angle between the inflow jet and the aneurysm branch may be risk factors for treatment failure.

Methods

Patient population

Institutional research ethics board approval was obtained. Eighty aneurysms treated with the PED at a single neurovascular center from November 2008 to June 2014 were reviewed. Consecutive aneurysms that harbored an incorporated branch were analyzed retrospectively from a prospectively maintained database.

Data collection and analysis

Data on aneurysmal diameter and location, acute versus elective treatment, number and type of PEDs, presence of previous coils or use of adjunctive coils, incorporated branches, and duration of antiplatelet therapy were recorded. Intra-procedural digital subtraction angiography (DSA), as well as available follow-up DSA, computed tomography angiography (CTA), and magnetic resonance angiography (MRA) were analyzed.

Aneurysm and incorporated branch geometry

All DSA runs performed for the flow-diversion procedure were reviewed by the two blinded raters (A.H.C. and B.J.D.) on the hospital Picture Acquisition and Communication System (PACS). A representative pre-PED angiographic image was chosen to show the inflow jet vector (IJ) and the direction of flow in the incorporated branch direction (BD). The angle between the IJ and BD was then recorded. The branch artery (BA) and parent artery (PA) diameter were measured using calibers as close to the aneurysm as possible on the same image and a BA/PA ratio was generated. The number and relation of the incorporated branch with respect to the aneurysm were also recorded: whether it arose from the branch at neck (BAN) or the branch at fundus (BAF). Definitions and examples are illustrated in Figures 1 and 2.

Figure 1.

(a) Schematic depicting definitions. BAF: arising from branch at fundus; BAN: arising branch at neck; PA: parent artery. Gray area delineates planned landing of Pipeline Embolization Device (PED). (b) Branch artery diameter (BA) shown with white arrowheads, measured as close to the aneurysm as possible. PA diameter shown with black arrowheads. Branch artery/parent artery diameter ratio: BA/PA. Angle between the inflow jet vector (IJ) and the incorporated branch direction (BD): IJ–BD angle, shown with gray arrows. (c) Right ICA catheter angiogram in RAO projection demonstrates a giant posterior communicating aneurysm with an incorporated branch (passing inferiorly). The parent ICA designated for PED deployment is marked PA. Measurement of the IJ - BD angle is demonstrated (white arrows). (D) Measurements of the BA diameter (white arrowheads) and PA diameter (black arrowheads) are demonstrated.

Figure 2.

(A) Lateral projection from a right internal carotid artery (ICA) planning rotational angiogram demonstrates a 23 mm right A1 blowout aneurysm in a 38-year-old male who presented with visual impairment (patient 17). He was treated with a single PED (4.25 mm × 18 mm) uneventfully. His aneurysm remained open beyond 12 months, despite clopidogrel cessation at this time. He was well at last follow-up, 65 months post PED placement. (b) Schematic demonstrating the fundus origin to his incorporated branch (the right ACA origin; BAF). The MCA forms the distal PA, and is projected end-on laterally. (c) Postero-anterior projection from a left ICA injection demonstrates a 27 mm ophthalmic aneurysm in a 61-year-old female (patient 60). She underwent placement of two PEDs (4.75 mm × 20 mm and 5 mm × 20 mm) uneventfully, and the aneurysm was found to be closed at four months. Clopidogrel was stopped at seven months, and she remains well. (d) Schematic demonstrating the neck origin to the small incorporated ophthalmic artery (BAN). The small IJ–BD angle is demonstrated by gray arrows. (e) Left vertebral injection, postero-anterior projection in a 70-year-old female (patient 75) demonstrates an incidental aneurysm after stroke from which two small duplicated superior cerebellar artery branches arise. (f) The larger arises from the neck (BAN), the smaller arises from the fundus on rotational angiography (not shown). Note is made of the unfavorable IJ–BD angle (gray arrows); the aneurysm was unchanged 15 months post insertion of two PEDs (4.5 mm × 20 mm and 4.25 mm × 18 mm). Clopidogrel was stopped at seven months, and the patient was well at 15 months.

Outcomes measures

Late occlusion was defined as the absence of complete occlusion at one year. Occlusion was defined as no residual filling (Raymond–Roy class 1), whether by computed tomography (CT), magnetic resonance imaging (MRI), or DSA. Radiological outcome of aneurysms without incorporated branches was also reviewed. Patients were followed clinically and with both time-of-flight and gadolinium-enhanced MRA and CTA at two months, six months, and one year following PED placement. Follow-up after this period was left to operator discretion. Aneurysms with available follow-up imaging of less than a year were excluded. Follow-up performed at external hospitals was not accepted if complete CT, MRI, or angiographic studies could not be obtained for viewing on PACS or were deemed suboptimal for analysis by a neuroradiologist at the authors’ institution. For aneurysms that had repeat PED placement, the first treatment was considered “lost to follow-up,” as occlusion related to the first treatment could not be assessed.

Anticoagulation and antiplatelet regimen

Dual antiplatelet therapy consisting of 325 mg of aspirin daily and 75 mg of clopidogrel daily five days prior to PED placement was employed for unruptured aneurysms. For acute treatment, patients were loaded with 650 mg of aspirin and 600 mg of clopidogrel two hours prior to the procedure. Patients were heparinized to a target activated clotting time (ACT) of ≥300, and this was not continued after the procedure, but occasionally it was reversed at the operator’s discretion. Dual antiplatelet therapy was continued for six months. All patients continued to receive aspirin after cessation of clopidogrel. The standard protocol did not differ for aneurysms harboring a branch vessel. At the clinician’s discretion, dual antiplatelet therapy could have been prolonged for various reasons such as in-stent stenosis or a cerebral ischemic event.

Data analysis

Logistic regression was performed using IBM SPSS Statistics for Windows v23 (IBM Corp., Armonk, NY). The correlation between measurements by the two authors was analyzed with intra-class correlation co-efficient (ICC).

Results

Baseline characteristics

Baseline aneurysms and treatment characteristics are outlined in Table 1. Of 80 consecutive aneurysms in 23 patients, 24 (30%) possessed incorporated inflow branches, with 28 separate discrete branches identified. No follow-up could be obtained for four aneurysms (two deaths), leaving 19 aneurysms and 23 inflow branches for analysis. One aneurysm was treated twice, and both treatments were considered “different” for the purpose of hemodynamic analysis (total of 20 procedures analyzed). Of those without incorporated branches, 19 aneurysms were excluded due to retreatment or death, leaving 37 for comparison. Four aneurysms harbored more than one incorporated branch.

Table 1.

Baseline data on aneurysms with incorporated branches and flow diversion treatment.

Characteristic	n
Location
Cavernous	1
Intradural ICA	11
ACA	3
MCA	1
Posterior circulation	8
Total	24
Location of incorporated branch
Neck	17
Fundus	6
Ruptured	5
Size, mean	15 mm (range 3–29 mm)
Morphopathology
Saccular	18
Fusiform	6
Number of PED used
>1	13
*mean two devices, range 1–8
Adjunctive coiling	3

ICA: internal carotid artery; ACA: anterior carotid artery; MCA: middle cerebral artery; PED: Pipeline Embolization Device.

Outcome measures

The occlusion rate at one year between the presence and absence of a branch was 10/19 (52.6%) compared with 27/37 (73.0%), respectively. This yields a p-value of 0.128 and an odds ratio (OR) of 0.41 (0.13–1.31; p = 0.13), both of which trended toward significance. No statistically significant difference in duration of dual antiplatelet therapy was noted, with a median of six months overall, in aneurysms that closed before one year and aneurysms that demonstrated delayed closure, as well as patients with aneurysms that had no incorporated branch.

The average IJ–BD angle for all branches was 101° (range 20–163°). A higher mean (110° vs. 89°), median (104° vs. 85°), and interquartile range (87–140° vs. 63–114°) were identified for aneurysms that demonstrated delayed occlusion (Figure 3), which trended toward significance (p = 0.21). The average BA/PA ratio for all incorporated branches was 0.38 (range 0.13–0.80). For aneurysms closed at one year, this was 0.42 (range 0.13–0.80), and for aneurysms not closed, it was 0.35 (range 0.19–0.58), with no statistically significant difference (p = 0.39). Seventeen (74%) branches arose from the aneurysmal neck, and six (26%) arose from the fundus. No significant difference was noted in occlusion rates between aneurysms harboring BAF and BAN (p = 0.71). The ICC was excellent for both IJ–BD angle and BA/PA diameter ratio (0.83 and 0.94, respectively).

Figure 3.

Relationship between IJ–BD angle in degrees with aneurysmal sac closure.

Discussion

There is mounting evidence showing favorable technical success and safety profile of flow diversion, as well as emerging long-term data suggesting efficacy.^1,6–14 However, the presence of a pre-existing stent⁶ or an incorporated branch arising from the aneurysm^1–4 have been identified as potential factors for delayed or failed aneurysm occlusion. This study investigated branch artery relative size and geometry as risk factors for delayed aneurysmal occlusion. A lower occlusion rate was observed for aneurysms incorporating a branch artery treated by PED. The occlusion rate at one year for aneurysms with an incorporated branch in the registry was 53%, conferring a 20% absolute risk increase in delayed occlusion. To explore the cause for this phenomenon further, a validated metric was developed pertaining to the size and geometry of incorporated branches, and it was shown that they could be reliably measured off treatment angiograms with satisfactory ICCs. Based on contemporary experience with flow diversion, the focus was on three variables related to the incorporated branch that could potentially affect aneurysm occlusion.

First, the anatomical relation of the incorporated branch with respect to the aneurysm was considered (BAN [neck] vs. BAF [fundus]). It was hypothesized that a branch arising from the aneurysmal fundus would require blood flow through the body of the aneurysm (perhaps more likely to promote aneurysm filling), rather than only through its neck or the flow diverter. Only 23 aneurysms incorporating a branch treated by flow diversion were identified, of which only six were arising at the fundus. This reflects a selection process at the authors’ institution where branch incorporation especially from the aneurysm body is avoided because of the perceived risk of treatment failure. The occlusion rate between the presence and absence of a branch was 52.6% compared to 73.0%, respectively (p = 0.128). This yielded a p-value of 0.128 and an OR of 0.41 (0.13–1.31; p = 0.13), which both trended toward significance. This ultimately demands a larger sample study for the basis for which this pilot establishes.

Second, it was hypothesized that a larger incorporated branch would presumably demand more flow and would therefore be more likely to promote residual flow from the parent artery, through the flow diverter and into the aneurysmal component harboring the branch. No statistically significant association was found between the BA/PA ratio and the rate of occlusion. This could be explained either by the lack of power of the study or by the fact that vessel size is not an independent factor. The focus was on vessel size as a surrogate for flow. However, different anatomical branches may require different demand for flow, independently from their caliber, such as balanced pcom-P1 systems where there may be a low flow condition, despite large vessel size. Future work should compare the most common anatomical branches involved. CFD studies may also be helpful for further exploration of this feature. The philosophy of the current study was to remain pragmatic and generalizable, since the use of CFD models remains largely experimental and, to our knowledge, limited data have currently been published on the association of these models and radiographic outcome.^15–18

Third, a large angle between the parent and branch arteries would suggest greater intra-aneurysmal vorticity (“swirling” in the entry remnant, seen on catheter angiograms) before the blood is able to exit the aneurysm into its incorporated branch. A trend was found toward a higher IJ–BD angle in aneurysms that demonstrated delayed occlusion. While it is recognized that the method of single-image measurement analysis is oversimplified and that analysis through CFD might be more accurate, the goal was to determine whether any practical intra-procedural “rules of thumb” could be established for the neurointerventionalist to predict the likelihood of occlusion in the contemporary setting. Consideration toward this potential factor for aneurysmal non-occlusion might be entertained with a multicenter approach on a much larger cohort of patients.

In a recent multicentric multivariate analysis of 171 aneurysms treated by flow diversion, Park et al. identified the following risk factors for residual filling following flow diversion: aneurysm size >15 mm, Raymond score of 2 or 3, age >60 years, lack of coils, and incorporated branch.¹⁹ Although this study reported a mixed population of fusiform and saccular aneurysms, the analysis was focused on aneurysmal geometry and configuration of the branch artery as a causative factor for occlusion rather than on underlying physiopathologic mechanisms. Conflicting data on differential occlusion rates for fusiform aneurysms following flow diversion are currently available.^20,21 This question needs to be studied further in the future, conditional to the appropriate sample size. Continued search for a subgroup of aneurysms with an incorporated branch that can be treated efficaciously with the PED should be encouraged.

Identifying risk factors for flow diverter treatment failure versus success is important because once a flow diverter is placed, endovascular access to the sac is usually lost and surgical treatment may be more difficult. Additional flow diverters can be placed with additional thromboembolic risk. Therefore, other options such as primary aneurysm coiling, test occlusion and sacrifice with or without bypass, and adjunctive coiling with flow diversion might be considered. Earlier conversion to aspirin monotherapy to promote occlusion may also be considered, but the risk of thromboembolic complication needs to be factored in.

Limitations

A lack of power is the main weakness of this study. A decision was made to perform this retrospective study on a single-center population due to the availability of all images for blinded review by two authors who were not involved in any of the procedures. While a single-center study maintains integrity of analysis, this yielded a smaller population of aneurysms eligible for analysis. Given this sample size, other reported factors that may play a role in aneurysm occlusion were not emphasized (potentially introducing bias in the group comparison) such as patient age, aneurysmal size, the number of overlapped flow diverters, caliber mismatch between the proximal and distal parent vessel, adjunctive coiling, and modification of intra-aneurysmal and parent artery flow dynamics. It is also recognized that vascular geometry can be altered following intracranial stenting.^22–24 Although this study emphasized assessment of angiographic imaging pre-implantation of a flow diverter, future work should also look at the modification of IJ–BD angles following FD and its potential effect on the rate of occlusion. As this study is retrospective, the acquisition protocols might have differed from one patient to another. The standard aneurysm treatment DSA acquisition frame rate is two or three frames per second. It is recognized that the accuracy of measuring some geometric variables may be affected by the imaging data acquisition protocol. For instance, the inflow jet might only be picked up in a very early arterial phase, while the peak branch vessel might fill later, or vice versa. It is also acknowledged that measurement of these geometric variables can be operator dependent and potentially generate high variability, although an acceptable ICC was obtained between two blinded raters. A high acquisition frame rate and eventually automated software calculations could be considered in future research protocols on this matter.

Conclusion

Aneurysms harboring incorporated branches show a lower rate of occlusion, presumed to be at least in part due to the demand flow through the aneurysm entry remnant caused by the persistently filling branch. This study confirmed a higher rate of delayed occlusion in flow-diverted aneurysms with branch incorporation. This pilot study explored the geometric relation and relative size of the incorporated branch as possible risk factors affecting occlusion rates. A trend toward a lower rate of occlusion was found for aneurysms demonstrating a higher IJ–BD angle. A lower rate of occlusion was not found in patients with larger relative size of the branch compared to the parent artery, or branch incorporation in the fundus versus the neck. The study showed the feasibility of measuring these geometric factors with satisfactory ICCs. Further evaluation of these geometric factors in a multicenter study is warranted.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.