1.1 Introduction

The past nearly three decades of endovascular treatment (EVT) of intracranial aneurysms (IA) have demonstrated coiling to be superior to standard surgical treatment for ruptured and unruptured IA. ¹ , ² , ³ Thus globally, EVT is progressively replacing the latter. However, EVT occlusion rates remain highly dependent on aneurysm morphology and location, the coils used, and the operators’ experience and skill. ² , ³ , ⁴ , ⁵ Subsets of large, giant, and wide-neck aneurysms, fusiform and blister aneurysms, as well as aneurysms associated with segmental artery disease remain difficult to treat because of the significant risk of coil herniation from the aneurysm into the parent artery. Using EVT, a subtotal aneurysm occlusion, aneurysm regrowth, and/or recanalization remain a challenge in 20 to 80% of cases due to coil compaction and migration of coils into the aneurysm thrombus. ⁴ , ⁶ , ⁷ Previous Computational Fluid Dynamics (CFD) studies show that complex interaction between parent artery and aneurysm hydrodynamics induces significant forces on the coil mass. ⁸ , ⁹ These calculations demonstrate that a higher coil packing density and a lower permeability of the coil mass at a given packing density could promote faster intra-aneurysmal thrombosis due to increased blood residence time and potentially create a more stable and durable occlusion.

Adjunctive techniques, such as balloon-assisted EVT and intracranial stents, were introduced into the clinical realm to enable embolization of giant and wide-neck aneurysm and reduce recanalization with coils alone. ⁷ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ However, recanalization requiring retreatment remained a challenge in this subset of IA. ¹⁸ , ¹⁹ , ²⁰ In addition, some aneurysms were amenable neither to EVT nor to a safe surgical repair.

To address these challenges in the early days of aneurysm embolization, research was conducted on a new treatment paradigm for IA that later became known as flow diversion. This chapter is a historical review of the development of flow diverters (FDs) from their inception to mature clinical products. The journey spanned nearly two decades of research and development prior to the introduction of a reliable and safe technology in one of the most challenging systems of the human body, the cerebrovasculature.

1.2 In Vivo Observations

Disappointing results with detachable latex and silicone balloons in the 1970s and the early generation of pushable coils in the late 1980s led several researchers to devise other solutions to circumvent existing challenges. Initially, Fedor Serbinenko ²¹ introduced detachable balloons for IA with preservation of the parent artery in 1974. His success and vast experience attracted the interventional neuroradiologist Gerard Debrun to visit and bring Dr. Serbinenko′s techniques to Paris, France. ²² , ²³ , ²⁴ , ²⁵ Other centers in Europe soon followed in his footsteps.

In the late 1970s, Grant Hieshima at the University of California Los Angeles (UCLA) developed silicone balloons. Later, his first fellow and colleague Randy Higashida at University of California San Francisco published their large series of IA successfully treated with detachable balloons. ²⁵ , ²⁶ , ²⁷ , ²⁸ Moret et al developed refinements in the technique to optimize position of the balloon in the aneurysm to achieve a more stable occlusion. ²⁹ , ³⁰ Unfortunately, these results were infrequently hampered by parent artery occlusion due to balloon rupture or migration resulting in thromboembolic ischemic events, as well as leak or rupture of the aneurysm due to a ball-valve mechanism created by movement of the balloon inside the aneurysm.

The next evolution in EVT of IA began more than a decade later. In the late 1980s, Fernando Vinuela at UCLA invited Guido Gugliemli from Italy to work in California. Joined later by Ivan Sepetka, these early pioneers developed the electrically detachable coil system from the bench to the bedside. Their work, however, relied on the development of a microcatheter that could be safely introduced into the cerebrovasculature and shapable tip microwire by Erik Engelson, a bioengineer at Target Therapeutics. This culminated in the first successful reported use of the Tracker microcatheter by Alex Berenstein and In Sup Choi at New York University in 1986. ³¹

Target Therapeutics then began work on the first prototypes of Guglielimi′s coils. In 1990, the first patient with a ruptured superior hypophyseal artery aneurysm was treated with this new technology under compassionate use at UCLA. Based on their experience, the Food and Drug Administration (FDA) approved the commercial sale of the Guglielmi detachable coils (GDCs) in 1995.

Simultaneously in 1989, Ajay Wakhloo was recruited by Martin Schumacher, department chief, to join as a junior attending physician in the Department of Neuroradiology at the University of Freiburg, Germany. It was there that Wakhloo was introduced to Hans Peter Strecker, a body interventional radiologist (IR) and inventor who was working in Karlsruhe, Germany, on a knitted tantalum stent for athero-occlusive disease of the peripheral circulation. Schumacher encouraged Wakhloo and his research team to further develop the stent technology for use in the cerebrovasculature. The tantalum stent, which was manufactured at the time by Boston Scientific Corporation (Watertown, MA) was a knitted tubular-shaped stent. ³² , ³³ Owing to reports of increased intimal hyperplasia and frequent restenosis, tantalum was replaced by nitinol ( Fig. 1.1). ³⁴

In 1990, Wakhloo introduced the stent technology for stent-assisted EVT of IA with coils or balloons. His early observations were presented in 1992 at the 18th Congress of The European Society of Neuroradiology in Stockholm, Sweden ( Fig. 1.2). ³⁶ Further experimental work was conducted in a canine model using bare and venous graft-covered tantalum and nitinol stents for occlusion of carotid aneurysms and arteriovenous fistulas, respectively. To assess the feasibility of the technology, venous pouch side-wall aneurysms were created in the canine common carotid arteries by Joost de Vries, as described previously by Varsos et al. ³⁷ After a maturation period of 2 weeks, the balloon-expandable tantalum stents and self-expanding nitinol stents were placed through a transfemoral approach into the common carotid artery covering the aneurysm. The stent provided a scaffold through which a Tracker microcatheter could be advanced allowing for the deployment of the pushable coils. Due to lack of sophisticated delivery systems, the stents were mounted on a percutaneous transarterial (PTA) balloon (Schneider, Lausanne, Switzerland) for delivery. A crochet technique with nylon threads was used to anchor and deploy the stents.

The balloon-expandable tantalum stents were deployed in the carotid artery at an inflation pressure of 3 to 4 atmospheres and, in most cases, immediate postdeployment angiograms had no evidence of residual aneurysms ( Fig. 1.3). Follow-up angiograms demonstrated a complete and stable aneurysm occlusion up to 9 months after the stent implantation. Over time, the stents became covered by a “thin layer of intimal fibrocellular proliferative tissue” on histologic specimens. ³⁶ Macroscopic inspection also demonstrated aneurysm scarring and shrinkage ( Fig. 1.3). However, Turjman et al, in a porcine side-wall aneurysm model, described instances of incomplete aneurysm obliteration following deployment of the knitted stents, as well as instances of complete parent vessel occlusion. ³⁸ Thus in 1994, Turjman et al recommended the combined use of coils and knitted stents. ³⁹

In 1994, Wakhloo et al proposed that the number of pores of a stent (higher number of pores = smaller pore size) may determine the occlusion rates based on up to 9 months of follow-up studies in a canine aneurysm model, a finding that would become the basis to flow diversion. ³⁴ Nitinol stents with a higher number of pores as compared with knitted tantalum Strecker stents (62.4 pores/cm² and 34.7 pores/cm² at 5 mm diameter, respectively) had less intimal buildup within the implant, but also increased the aneurysm obliteration rate at follow-up. A stent-induced regional flow alteration was suggested as a mechanism for the aneurysm thrombosis which was documented on highframe rate digital subtraction angiography (DSA) at 4 to 6 frames/sec. DSA also showed that stent placement within the parent artery led to the redirection of contrast material toward the distal part of the artery, with decreased inflow into the aneurysm pouch and delayed contrast material washout due to increased circulation time. ³⁴ These findings were described as blood diversion or channeling effect ( Fig. 1.4).

To prevent carotid thrombosis, the research specimens were treated with periprocedural heparin along with aspirin (80 mg/daily) in their diet for the duration of the study. In addition to pore density, the choice of the stent alloy was thought to be important to minimize in-stent intimal buildup. ³⁴ In 1994, Geremia et al published their experience with self-expanding Wallstents (Schneider, Minneapolis, MN) in a canine carotid side-wall aneurysm model ( Fig. 1.1). ³⁵ The cobalt alloy Wallstent, which was designed for use in peripheral athero-occlusive disease, was available premounted on a 7F delivery system. Unlike Turjman et al, the investigators found high aneurysm occlusion rates on follow-up angiography up to 2 months after stent implantation without any in-stent stenosis. ³⁵ Hematoxylin and eosin staining revealed the presence of dense fibrous tissue within the aneurysm pouch ( Fig. 1.4). Immediately after stent placement, not only was delayed filling of the aneurysm observed but a gradual puddling of contrast medium was also noted. Vortex flow patterns as calculated previously by Perktold et al ⁴⁰ in 1984 and observed by Steiger et al in 1987 in aneurysm models ⁴¹ and by Graves et al in a canine side-wall aneurysm model ⁴² was disrupted by placement of the stent. In this study, the carotid arteries remained patent despite lack of anticoagulation most likely due to the high fibrinolytic activity of the canine clotting system.

Besides the initial lack of availability of sophisticated stent delivery systems for the tortuous neurovascular circulation, another major challenge for the clinical use of stents to treat IAs was the preservation of critical perforators or side branches that are often intimately related to the aneurysm. Subsequently, the fate of those small vessels when covered by stents was studied using small muscle branches originating from the vertebral artery in a canine model. ⁴³ Fortunately, no branch occlusions were observed angiographically during an observation period of 9 months after the stent implantation. Nearly a decade later, Kallmes et al and Sadasivan et al used the rabbit lumbar and vertebral arteries, respectively, as substitutes for perforators and side branches to assess mature FDs with high mesh densities in a preclinical setting ( Fig. 1.5). ⁴⁴ , ⁴⁵ The investigators did not report any branch occlusions, even in cases where several overlapping FDs were deployed.