Purpose To develop a reproducible technique for selective ophthalmic artery infusion chemotherapy (SOAIC) that is technically efficacious in children with unfavorable patterns of ophthalmic artery (OA) flow.
Materials and methods Initially, all SOAIC was performed with intention to treat using a standard selective OA (microcatheter) infusion technique (sSOAIC). Temporary balloon occlusion (TBO) of the external carotid artery (ECA), a balloon-assisted SOAIC (bSOAIC) technique, was performed only if OA angiography did not show robust and sustained anterograde OA flow. In our more recent experience, all SOAIC was performed with intention to treat by bSOAIC. Verapamil infusion into the OA and internal carotid artery was performed in selected cases. Technical success was defined as delivery of chemotherapeutic agent(s) into the OA with robust and sustained anterograde perfusion. sSOAIC was considered to have failed if converted to bSOAIC.
Results 19 eyes were treated in 17 patients (age 5 months to 16 years) between December 2008 and May 2013. Eighty-three procedures were undertaken and the OA was successfully catheterized in all. Technical success was achieved in 35/41 (85%) sSOAIC cases and 42/42 (100%) bSOAIC cases. TBO of the ECA augmented anterograde OA flow and converted all cases of retrograde OA flow to anterograde. Verapamil further augmented anterograde ocular perfusion during SOAIC. There were no access site complications, strokes, or deaths.
Conclusions TBO of the ECA is a safe, effective, and reproducible method for optimizing ocular hemodynamics during SOAIC regardless of baseline OA flow pattern. Verapamil infusion may further favorably modify OA flow.
Trial registration number NCT01466855.
- Blood Flow
Statistics from Altmetric.com
Selective ophthalmic artery infusion chemotherapy (SOAIC) offers the hope of eye salvage and vision to children with retinoblastoma who have intermediate- to high-grade tumors.1–3 The rationale for the development of SOAIC stems from its pharmacologic advantages. Since the administered dose of chemotherapy is not diluted into the systemic circulation before arriving in the tumor, the concentration of chemotherapeutic drug achieved in tumor tissue is 250 times greater than that achieved by systemic chemotherapy.4 This enables enhanced tumoricidal potency, even though the total dose of chemotherapeutic given to the patient is about 10% of the dose typically administered during systemic therapy.5 ,6 This substantial dose reduction results in fewer bone marrow and other systemic toxicities.
The concept of intra-arterial chemotherapy for retinoblastoma was born in Japan more than 50 years ago. The first approaches were not catheter-based interventions, but direct puncture-infusion methods.7 ,8 The first catheter-based approaches were pioneered in Japan, circa 1988.9 The earliest methods involved temporary balloon occlusion (TBO) of the supraclinoid internal carotid artery (ICA), and non-selective infusion of melphalan through a base catheter in the cervical ICA.9
Advances in microcatheter technology and interventional technique led to the evolution of super-selective ophthalmic artery (OA) infusion methods. SOAIC for retinoblastoma was first described in the Japanese literature in 1993.9 More than10 years later, Abramson et al6 reported the results of a phase I/II study in 10 children with intraocular retinoblastoma. The remarkable responses demonstrated in this small group of patients fueled widespread enthusiasm for SOAIC, and paved the way for a much larger clinical study.10 Retrospective studies have shown that angiographic features consistent with strong anterograde ocular perfusion predict therapeutic success.11
A significant technical challenge pertinent to this principle involves hemodynamic shifts that occur when the ostium of the OA is obturated by a microcatheter. Even the smallest commercially available 1.2 French microcatheters occupy a substantial portion of the cross-sectional area of the OA in an infant or toddler. The hemodynamic effect produced by this artificial stenosis, combined with the effects of catheter-induced spasm, may allow backflow from external carotid artery (ECA) anastomoses to overcome anterograde OA flow, leading to OA flow reversal and ICA reflux. Consequently, the microcatheter infusate can never reach its tumor target. One strategy used to manage backflow is to administer intranasal vasconstrictors. This strategy has limited efficacy since it targets a minor group of ethmoidal artery anastomoses between the sphenopalatine artery and the OA. Another strategy used to manage backflow is to administer chemotherapy through ECA collaterals that arise from the middle meningeal artery (MMA) or the superficial temporal artery (STA). Chemoinfusion through MMA anastomoses is potentially efficacious if robust flow to the OA via the superficial recurrent meningeal artery and/or the meningolacrimal artery can be demonstrated angiographically.12 ,13 Likewise, chemoinfusion through the zygomatico-orbital branch of the STA may be efficacious if there is a strong anastomosis with the OA.14 The problem with chemoinfusions delivered through ECA collaterals is that the microcatheter infusate is countered by anterograde OA flow from the ICA, and thwarted by this counter-flow from reaching its tumor target. Although it may be possible to overcome this competitive flow with higher injection pressures, such an approach can lead to excessive reflux of chemotherapy into the cerebral circulation.
An alternative strategy that we developed to optimize ocular hemodynamics involves TBO of the ECA (figure 1). The large pressure drop in the ECA, caused by TBO, reliably promotes a strong anterograde flow stream in the OA augmenting ocular perfusion throughout the entire OA territory (figure 2). An additional modification involving infusion of the vasodilator verapamil, selectively into the ICA, and superselectively into the OA can further augment anterograde OA flow and may have a secondary benefit related to tumor biology. This report summarizes our initial experience using these adjunctive techniques to optimize ocular hemodynamics.
Materials and methods
The patients comprised a consecutive series of children referred with unilateral or bilateral, intermediate to advanced intraocular retinoblastoma (International Classification System groups C, D, and E) treated with SOAIC. Children were offered SOAIC if they showed a potential for functional vision with the affected eye, regardless of treatment history or age.
Intra-arterial chemotherapy treatment protocols
SOAIC was given in cycles at 3–4 week intervals using single drug, melphalan or topotecan, or multidrug therapy (melphalan, topotecan and carboplatin). The clinical response to treatment was assessed 3–4 weeks after each treatment by ophthalmic examination under anesthesia with fundus mapping and high-resolution retinal photos using RetCam 3. Treatment cycles were repeated until there was no clinical evidence of viable tumor found by ophthalmic examination under anesthesia. Adjunctive topical treatments, including laser thermal treatment, cryotherapy, and subconjunctival carboplatin, were administered based on the extent of tumor response and vitreal seeding. Patients were treated with prednisolone acetate ophthalmic suspension (Allergan Inc, Irvine, California, USA) for 3 days after SOAIC to ameliorate treatment-related ocular inflammation.
Basic interventional methods
All SOAIC was performed under general endotracheal anesthesia by a single operator in an Allura FD20/20 biplane neuroangiography suite (Philips Healthcare, Andover, Massachusetts, USA). Low-dose fluoroscopy and digital subtraction angiography (DSA) frame rates of 1–2 frames per second were employed. All patients were systemically anticoagulated with intermittent boluses of heparin. Adjunctive albuterol (180 μg to endotracheal tube by metered dose inhaler) and decadron (0.1 mg/kg intravenously) were given at the onset of each procedure. Adjunctive intranasal oxymetazoline (2 puffs on the side of OA chemoinfusion) was also intraoperatively administered to patients in our earlier practice (16/42 balloon-assisted SOAIC (bSOAIC) procedures and 19/41 standard SOAIC (sSOAIC) procedures). More recently, intranasal oxymetazoline has been avoided because TBO of the ECA was adopted as a more effective means of optimizing ocular hemodynamics and the small but not insignificant risk was no longer considered justified.15
All SOAIC was initiated by placing a four French sheath in the femoral artery with ultrasound guidance. Indwelling arterial catheters were perfused continuously with heparinized saline (2 units/mL). Selective catheterization of the ICA on the side of the tumor was performed with a four French Tempo catheter (Cordis-Johnson and Johnson, Miami, Florida, USA) and 0.035 in Bentson wire (Cook Medical, Bloomington, Indiana, USA). In our more recent procedures, a dose of verapamil (20 μg/kg) was infused directly into the ICA through the diagnostic catheter to pre-emptively counter catheter-induced spasm. A 1.2–1.5 French microcatheter (Magic, Balt Extrusion, Montmorency, France or Marathon, ev3-Covidien, Plymouth, Minnesota, USA) was introduced into the Tempo catheter and used to perform OA catheterization using a combination of over-the-wire and flow-directed technique. An 0.008 in Mirage microguidewire (ev3-Covidien) was used for over-the-wire maneuvers. ICA and OA angiograms were obtained to evaluate the anatomy and flow dynamics. When verapamil was administered into the ICA, a second dose of verapamil (80 μg/kg) was infused through the microcatheter, directly into the OA, to pre-empt catheter-induced spasm and augment ocular perfusion.
Adjunctive TBO of the ECA
bSOAIC was performed through a three French transfemoral arterial sheath placed contralateral to the sheath used for introduction of the guide catheter–microcatheter system. A 4 mm balloon occlusion catheter (Hyperglide, ev3-Covidien, Irvine, California, USA or Scepter C, Microvention-Terumo, Tustin, California, USA) was introduced through the sheath and advanced over a microguidewire to its final position in the ECA ipsilateral to the index tumor. The X-pedion 10 microguidewire (ev3-Covidien) was used in conjunction with the Hyperglide balloon occlusion catheter, and the Transcend EX floppy microguidewire (Stryker Neurovascular, Fremont, California, USA) was used in conjunction with the Scepter C balloon occlusion catheter. In our early procedures, the balloon was positioned in the internal maxillary artery (11/42 bSOAIC procedures), but we found that we could improve the ocular hemodynamics when the balloon was positioned more proximally to cover the origin of the facial artery (31/42 bSOAIC procedures). Anterograde flow arrest in the ECA was documented by DSA.
Intention-to-treat strategies and definitions of technical failure
During our early procedures (‘standard period’), all patients were managed by intention to treat with the sSOAIC technique. More recently (‘modification period’), all patients were managed by intention to treat with the bSOAIC technique.
Any SOAIC case in which the OA was not catheterized, or the full scheduled dose of chemotherapeutic agent/s was not delivered was considered a technical failure. If the full scheduled dose of chemotherapeutic agents was not successfully delivered during any given cycle of sSOAIC, the subsequent treatment cycle was undertaken as intention to treat using bSOAIC technique. Any case, which was initiated as an intention to treat by sSOAIC and was converted to bSOAIC was counted as a technical failure for sSOAIC. Once the treatment of any eye was converted from sSOAIC to bSOAIC, all subsequent treatment cycles were initiated as intention to treat with bSOAIC and were not counted as technical failures of for sSOAIC.
Chemotherapy and infusion technique
All chemotherapy was prepared in 30–45 mL volumes and administered directly into the OA through microcatheters. All chemoinfusions were manually administered using a pulse synchronous staccato technique over a mean period of 30.3 min (range 18–45 min). An attempt was made to synchronize each injection pulse of the infusion with the systolic phase of the corresponding cardiac cycle by listening to pulse oximetry and pedal Doppler pulses.
The optimal chemoinfusion pulse volume and rate rise was determined by empirical analysis of microcatheter contrast injections monitored fluoroscopically as the operator varied the rate rise and pulse volume to achieve a robust uniform anterograde filling of the OA without ICA reflux or microcatheter kickback.
The microcatheter position in the OA was monitored by intermittent fluoroscopy to check the position of the microcatheter tip in relation to bony landmarks. In most cases, the chemoinfusion was interrupted after half of the total chemotherapy dose had been administered to evaluate the stability of OA hemodynamics, either by microcatheter DSA or by fluoroscopic monitoring of contrast injected through the microcatheter.
Nineteen eyes (13 in the standard period, 6 in the modification period) in 17 patients (11 in the standard period, 6 in the modification period) were treated by a single operator, between December 2008 and May 2013. The standard period extended from December 2008 to November 2012. The modification period extended from November 2012 to June 2013.
Eighty-three procedures were undertaken with 94% technical success. The OA was successfully catheterized in all 83 cases. Technical success was achieved in 35/41 (85%) sSOAIC cases and 42/42 (100%) bSOAIC cases. An average of 4.4 treatment cycles per eye was used. During our initial experience (standard period), 37 cases were initiated as intention to treat using sSOAIC technique. Five (13.5%) of these sSOAIC cases were technical failures because of incomplete dose delivery and conversion to the bSOAIC technique in the subsequent treatment cycle. One (2.7%) sSOAIC case was a technical failure because of sustained retrograde OA flow and intraprocedural conversion to bSOAIC technique.
Reasons for incomplete dose delivery resulting in conversion from the sSOAIC to the bSOAIC technique in the subsequent treatment cycle were intermittent retrograde OA flow during sSOAIC (one case), weak anterograde OA flow during sSOAIC (one case), pressure-dependent OA backflow during sSOAIC (two cases), and weak angiographic blushing of intraocular structures during sSOAIC (one case).
In all instances of conversion from the sSOAIC technique to the bSOAIC technique, TBO of the ECA transformed OA flow into a strongly anterograde pattern that was sustained for the duration of the treatment.
Characteristics of patients treated with the bSOAIC technique
Forty-two bSOAIC procedures were used to treat 13 eyes in 13 patients (7 during the standard period (27 treatments) and 6 during the modification period (15 treatments)). Patients were aged 5 months to 16 years (median 3 years) and weighed 6–48 kg. In one patient (a 29-month old girl with bilateral group D tumors), six cycles of bSOAIC were administered to both eyes concurrently.
The dose–area product (DAP) for bSOAIC procedures (average 50 Gy/cm2) was higher than that for sSOAIC procedures (average 29 Gy/cm2). The difference is statistically significant based on a χ2 analysis with a two-tailed p value of 0.0005.
Intra-arterial verapamil infusions
None of the 41 sSOAIC procedures and nine of the 42 bSOAIC procedures were performed with intra-arterial verapamil. Verapamil infusions were associated with a subjective angiographic increase in OA diameter, anterograde OA flow, and intensity of the retinal choroidal, lacrimal, and ciliary body blushes.
Treatment history and enucleation rates
Thirteen of 19 eyes were pretreated with systemic chemotherapy. Six of 19 eyes were enucleated, all for tumor progression. Four eyes in four patients continued to be treated actively at the time of this retrospective study.
Direct retinal examination during chemoinfusion
In one case, RetCam 3 assessment of the fundus was performed during chemoinfusion. This single observation was performed owing to reports of drug emboli during chemoinfusion. Fortunately, no evidence of drug crystals or thromboemboli was found, though an immediate constriction in tumor size associated with drug infusion was noted and progressive tumor pallor was evident.
Bronchospasm requiring treatment occurred in five procedures (6%). In three of these procedures a balloon assist technique was used. One patient had a procedure-related anaphylactic reaction caused by carboplatin hypersensitivity and required resuscitation, including the intravenous administration of epinephrine. There were no contrast media reactions.
Four cases of symptomatic ophthalmitis required systemic steroid therapy. bSOAIC was used in three of these cases.
We observed one case of reversible cerebral vasoconstriction syndrome during one of four sSOAIC procedures performed in a 20-month-old girl. In this case, the process was self-limited, and without neurologic complications or evidence of stroke on brain MRI.14
No access site complications were seen during the study. There were no instances of clinically significant intraprocedural hypotension or bradycardia related to administration of verapamil. No instances of bone marrow suppression requiring transfusion of blood products or hospitalization occurred.
Our results confirm that SOAIC is a safe and effective treatment for children with intraocular retinoblastoma. Others have reported comparable outcomes.10 ,16 ,17 In this technical case series of patients with retinoblastoma treated with SOAIC, we employed two new modifications to augment anterograde ocular blood flow, reduce reflux into the ICA, and maximize tumor chemoperfusion. Those modifications were TBO of the ECA, and intra-arterial verapamil infusion to the ICA and OA.
TBO of the ECA consistently and strongly converted retrograde OA flow to anterograde. Since blood flows from a higher pressure to a lower pressure, reducing the pressure in the ECA by TBO strongly promotes anterograde flow in the OA. Unlike chemoinfusions administered through ECA collaterals such as the MMA or the STA, the bSOAIC method establishes the microcatheter-bearing vessel through which the chemoinfusion is administered as the dominant and nearly exclusive source of blood flow to the ocular circulatory bed. This minimizes backflow, reflux, errant non-target delivery, and dilution factors which can diminish technical efficacy. If the occlusion balloon is positioned to cover the facial artery, any competitive ECA collateral flow to the OA via reconstituted flow in the facial, internal maxillary, or superficial temporal arteries must cross at least two anastomotic watershed barriers in series. In contrast, when chemoinfusions are administered through ECA branches, competitive flow enters the OA directly or after crossing only one anastomotic barrier.
One weakness of our study lies in the retrospective nature of our data collection and analysis. Such methodology is well known to be associated with patient selection bias and information bias. Although we believe that selection bias was minimized by including all patients treated with SOAIC at our institution, information bias is much more difficult to exclude.
One disadvantage of the bSOAIC method is the requirement for bilateral femoral artery access. Although we found no access site complications during the study, we acknowledge that the requirement for repeated bilateral femoral artery access may increase the number of potentially significant access site complications in the long run, and for individual patients undergoing multiple treatment cycles the risk may increase over time as a result of additive vessel injury. In infants and toddlers with small femoral arteries, the risks are higher. In such cases, the risks and benefits of the bSOAIC method should be carefully weighed against those of alternative strategies. We believe that the risk of access site complications can be minimized by the use of ultrasound guidance and the implementation of short (≤5.5 cm), low-profile access devices (3–4 French arterial introducer sheaths). An additional measure we have taken to minimize the exposure of vessels to cumulative injury is to alternate the side of the larger four French sheath between consecutive treatment cycles.
The measured DAP was higher in the bSOAIC group than in the sSOAIC group, indicating a slightly higher radiation exposure for the bSOAIC procedure. The difference is statistically significant and of uncertain clinical significance. Although threshold doses for deterministic radiation effects were not reached for any case in this series, stochastic effects are much harder to predict. In one study, the mean total DAP in individual pediatric patients undergoing single or multiple neuroangiography procedures for stroke, trauma, or tumor was 280.5 Gy/cm2.18 The authors estimated a 1% increase in the lifetime excess risk of cancer attributable to this exposure. The results of that study suggest that the lifetime excess risk of cancer attributable to treatment-related radiation exposure in a patient completing SOAIC treatment for a unilateral retinoblastoma (4.4 cycles in this series) is <1%, regardless of what technique is used. Unfortunately, this crude estimate does not consider risk modifiers that enhance the carcinogenic effects of radiation in patients with retinoblastoma. Even if the risks of stochastic effects are low, minimizing treatment-related radiation exposure is a major consideration in children with retinoblastoma, who may have a significant lifetime risk of second cancers due to the presence of ubiquitous tumor suppressor gene mutations.19 We strongly advocate that any increase in treatment-related radiation exposure must be justified, and that the increased risk of stochastic effect is outweighed by the therapeutic advantage.
Except for patients with highly unfavorable OA flow patterns that preclude effective drug delivery using the sSOAIC technique, the therapeutic advantage of the bSOAIC technique remains theoretical. Increasing experience and refinements in the bSOAIC technique may further reduce patient radiation exposure. Advances in imaging technology have already led to major improvements in radiation exposure during SOAIC. Cooke et al20 recently showed that the radiation dose from SOAIC procedures can be significantly reduced by altering the fluoroscopy settings to decrease the dose per pulse. The use of low-dose pediatric protocols that include a higher level of spectral filtration have also been associated with decreased patient exposure.21 Other advances in imaging technology promise to decrease radiation doses by >60%.22 Patient exposure to radiation can be further reduced by operator attention to detail and the use of strict interventional techniques with judicious application of low frame rate fluoroscopy, minimal magnification, tight collimation, and single-plane imaging, when appropriate.
In theory, adjunctive TBO of the ECA may also improve the efficacy of SOAIC in patients who have anterograde OA flow, because it augments anterograde ocular perfusion with chemotherapeutic agents, and therefore increases exposure of the tumor to these agents. Theoretically, this might reduce the overall number of intra-arterial treatments necessary to achieve tumor-free remission. Owing to the small number of patients in our study, and the use of both techniques (sSOAIC and bSOAIC) to treat individual tumors, we cannot deal with this question.
Only a minority of patients in the bSOAIC group, and none of the patients undergoing sSOAIC, received intra-arterial verapamil. Such treatment is often used in interventional practice to manage vasospasm, and can be used to increase blood flow above basal levels even in the absence of vasospasm.23 In this series, intra-arterial verapamil was administered in two parts. One-quarter of the total dose was administered directly into the ICA before selective catheterization of this vessel to suppress vasospasm. The remaining three-quarters of the total dose was administered directly into the OA via a microcatheter. After verapamil infusion, a progressive angiographic increase in anterograde OA blood flow was subjectively assessed in all cases. When combined with the hemodynamic effect of temporary ECA occlusion, the effect of verapamil seems to be additive. Such assessments were possible because direct microcatheter infusions of verapamil into the OA were given after TBO of the ECA had been established, and control OA angiograms were obtained before and after verapamil infusion.
One theoretical advantage of intra-arterial verapamil when used in conjunction with SOAIC is that it may sensitize retinoblastoma cells to chemotherapy by inhibiting multidrug resistance mediated by membrane-associated glycoproteins.24–26 This phenomenon has been demonstrated in the test tube and is being explored in other clinical models of intra-arterial chemotherapy.24 ,27–29 Although the chemosensitizing effects of verapamil have been known for some time, it has not been frequently used in conventional chemotherapy because the tissue concentration of verapamil necessary to achieve the relevant effects cannot be safely achieved by oral or intravenous administration. The unique pharmacokinetic profile afforded by intra-arterial administration makes it possible to take advantage of the concentration-dependent chemosensitizing effect of verapamil.27
Reversible cerebral vasoconstriction syndrome without evidence of cerebral ischemia developed in one patient. Careful study of this case led us to conclude that the event was triggered by simultaneous co-administration of intranasal oxymetazoline and phenylephrine containing mydriatic eye drops.15 We now avoid intranasal oxymetazoline, especially in younger patients, and particularly in the setting of retrograde OA flow. Theoretically, retrograde OA flow could allow intranasally administered oxymetazoline to enter the cerebral circulation. Since we adopted adjunctive TBO of the ECA, intranasal oxymetazoline is considered only in the management of patients who strongly reconstitute anterograde flow in the internal maxillary artery despite TBO of the proximal ECA. In such cases, the ICA collateralizes the ECA through the artery of the foramen rotundum. The reconstituted sphenopalatine artery raises pressure in the ethmoidal arteries, and diminishes the pressure gradient favoring anterograde OA flow. Intranasal oxymetazoline can constrict the sphenopalatine artery and restore a favorable pressure gradient.
Procedure-related bronchospasm has been reported in up to one-quarter of patients undergoing SOAIC.10 This occurred in 6% of our cases and required treatment despite prophylactic pretreatment with nebulized albuterol. The most widely accepted explanation for this phenomenon is that microcatheter navigation through the cavernous ICA triggers an autonomic reflex that is particularly active in infants and toddlers owing to the sensitivity of vagal and trigeminal afferents in this age group.
Others have reported some form of symptomatic ophthalmitis in 12% of SOAIC procedures.10 In our series 4.8% of SOAIC procedures were complicated by symptomatic ophthalmitis, most of which were in children who had been treated with the bSOAIC technique. Although the low event rate in our series does not enable us to determine if TBO of the ECA increases the risk of symptomatic ophthalmitis, it is possible that the enhanced ocular chemoperfusion associated with the method aggravates non-target tissue toxicity in the orbit. It is unlikely, however, that the minor and self-limited symptoms associated with augmented ocular chemoperfusion would offset the potential benefits of increased tumoral chemoperfusion.
Adjunctive TBO of the ECA is safe and effective for converting retrograde OA flow to anterograde and for optimizing ocular hemodynamics in favor of tumor chemoperfusion. Adjunctive intra-arterial verapamil infusions administered into the ICA and the OA further optimize ocular hemodynamics in favor of tumor chemoperfusion and may have additional chemosensitizing effects. Further study of the clinical benefits afforded by these new technical modifications is warranted.
Contributors TAA conceived the work, analyzed the data, drafted and revised the paper. He is the guarantor. JIG revised the draft paper. DAK collected and analyzed the data, co-drafted and revised the paper. SM collected and analyzed the data and revised the draft paper. ZMC revised the draft paper. KC collected data and revised the draft paper. JJA revised the draft paper.
Funding This work was partially funded by a CancerFree Kids research grant from Cincinnati Children's Hospital, Cincinnati, OH, USA.
Competing interests TAA receives material support and consultant fees from Microvention-Terumo, Tustin, California, USA, and Philips Healthcare, Andover, Massachusetts, USA.
Ethics approval Cincinnati Children's Hospital Medical Center.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data relevant to this case series are presented within the body of this paper.
More info Aspects of this work have been presented at the annual meetings of the Society of Pediatric Interventional Radiology (27 October 2012), the American Society of Neuroradiology (20 May 2013), and the World Federation of Therapeutic Interventional Neuroradiology (11 November 2013).
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.