Background The Raymond–Roy Occlusion Classification (RROC) qualitatively assesses intracranial aneurysm occlusion following endovascular coil embolization. The Modified Raymond–Roy Classification (MRRC) was developed as a refinement of this classification scheme, and dichotomizes RROC III occlusions into IIIa (opacification within the interstices of the coil mass) and IIIb (opacification between the coil mass and aneurysm wall) closures.
Methods To demonstrate in an external cohort the predictive accuracy of the MRRC, the records of 326 patients with 345 intracranial aneurysms treated with endovascular coil embolization from January 2007 to December 2013 were retrospectively analyzed.
Results Within this cohort, 84 (24.3%) and 83 aneurysms (24.1%) had MRRC IIIa and IIIb closures, respectively, during initial coil embolization. Progression to complete occlusion was more likely with IIIa than IIIb closures (53.6% vs 19.2%, p≤0.01), while recanalization was more likely with IIIb than IIIa closures (65.1% vs 27.4%, p<0.01). Kaplan–Meier estimates demonstrated a significant difference in the test of equality for progression to complete occlusion (p=0.02) and recurrence (p<0.01) between class IIIa and IIIb distributions. For the entire cohort, male gender (p<0.01), ruptured aneurysm (p=0.04), intraluminal thrombus (p<0.01), and MRRC IIIb closure (p<0.01) were identified as predictors of recanalization. For aneurysms with an initial RROC III occlusion, MRRC IIIa closure was found to be an independent predictor of progression to complete occlusion (p=0.02).
Conclusions This study confirms that the MRRC enhances the predictive accuracy of the RROC.
Statistics from Altmetric.com
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.
Following the introduction of Guglielmi Detachable Coils (GDCs) to clinical neuroendovascular practice in 1995, endovascular coil embolization has evolved to become the dominant treatment modality for intracranial aneurysms.1–4 Occurring in approximately 30% of cases, aneurysm recanalization following coil embolization remains a significant drawback of this minimally invasive treatment approach. Aneurysm recanalization necessitates rigorous long-term follow-up angiography, possible repeat coil embolization, and potentially places patients at risk of aneurysm rupture.5–8 Numerous clinical studies have examined risk factors and predictors for aneurysm recanalization following coil embolization, and initial aneurysm occlusion status consistently emerges as a strong predictor.5 ,6 ,8–12 Historically, the Raymond–Roy Occlusion Classification (RROC) has been the scale by which intracranial aneurysms treated with coil embolization are qualitatively assessed.5 ,13 Recognizing that not all RROC III occlusions behave similarly, Mascitelli et al8 developed the Modified Raymond–Roy Classification (MRRC) in 2014 to supplement the original RROC. The MRRC dichotomizes RROC III (residual aneurysm opacification) occlusions into IIIa (residual aneurysm opacification within the interstices of the coil mass; figure 1) and IIIb (residual aneurysm opacification between the coil mass and the aneurysm wall; figure 2) occlusions, with IIIa closures portending a more favorable angiographic prognosis. Although this modified classification system has expanded our understanding of the angiographic outcome of coiled aneurysms, it was developed through retrospective analysis of data from a single institution. Validation of the MRRC using an external cohort of intracranial aneurysms treated with coil embolization from a high-volume neurovascular center may encourage its adoption into general neuroendovascular practice. In this report we analyze 326 consecutive patients with 345 intracranial aneurysms treated via endovascular coil embolization and demonstrate the ability of the MRRC to reliably predict eventual progression to aneurysm occlusion or recurrence.
Patients and data collection
We performed an observational, retrospective, single-center study to assess the durability of aneurysm occlusion by coil embolization. Following institutional review board approval at Massachusetts General Hospital (MGH), the records of 484 patients with 517 intracranial saccular aneurysms treated with endovascular coil embolization between January 2007 and December 2013 were retrospectively reviewed. Of the 517 aneurysms reviewed, 172 had inadequate angiographic follow-up (due to patient death, relocation, and general loss to follow-up) and were excluded from further analysis. Of the 345 aneurysms included in the analysis, 182 underwent both short-term and long-term angiographic follow-up. The study sample was collected by reviewing the MGH cerebrovascular surgery and radiology databases within the study period. Infectious, dissecting, and fusiform aneurysms were not included in the study sample. All historical, clinical, radiographic, and follow-up information was obtained from the electronic medical record in accordance with the Health Insurance Portability and Accountability Act (HIPAA).
Detailed historical, clinical, radiographic, and follow-up data were collected for each patient. For the purpose of this study, progression to complete aneurysm occlusion was defined as any non-MRRC I closure that progressed to a MRRC I closure at the time of last angiographic follow-up. Recanalization, or recurrence, was defined as an increase in neck or aneurysm dome opacification at the time of first and/or last angiographic follow-up compared with the initial post-treatment cerebral angiogram.
All initial and follow-up cerebral angiograms were reviewed by the staff neuroendovascular specialist at the time of study acquisition. While RROC designations were not universally provided, study reports reliably included specific descriptions of the degree of aneurysm occlusion, including residual neck or aneurysm opacification, if present. For the purpose of this study, all initial and follow-up cerebral angiograms were retrospectively reviewed by CJS, and RROC and MRRC designations were assigned, with initial studies reviewed prior to any follow-up studies in order to reduce bias. These designations were subsequently compared with the written descriptions provided in the study reports by the staff neuroendovascular specialist. When there was disagreement between the designation provided by CJS and the staff neuroendovascular specialist, CMT acted as an independent arbiter. MRRC designations were applied in accordance with the stipulations detailed in the original publication.8
Descriptive statistics were calculated for clinical and radiographic factors, using the median as a measure of central tendency. A univariate analysis of clinical characteristics and outcomes was performed. Comparisons of continuous variables with non-normal distributions were made using the non-parametric Mann–Whitney U test. Contingency statistics on categorical variables were performed with the Fisher exact test. To determine predictors of a MRRC IIIb closure, aneurysm recanalization, and progression to complete occlusion following initial coil embolization, multivariate regression analyses were performed by employing the forward stepwise entry method using the likelihood ratio statistic to create the most parsimonious model using factors found to be statistically significant on screening univariate analyses. Kaplan–Meier estimates of aneurysm occlusion and recurrence were performed for MRRC IIIa and IIIb occlusions; a log rank (Mantel–Cox) test was used to determine the equality for complete occlusion or recurrence distributions. All statistical tests were two-sided and p<0.05 was prospectively determined to establish statistical significance. All analyses were performed using SPSS Statistics V.21 (IBM, Armonk, New York, USA).
Patient, aneurysm, and procedural data
Three hundred and twenty-six patients with 345 untreated saccular intracranial aneurysms were analyzed. Table 1 summarizes the patient, aneurysm, and procedural details of this cohort. The median age was 56.9 years and 83.1% of the cohort were women. A history of smoking and hypertension was confirmed in 184 (56.4%) and 160 (49.1%) patients, respectively. A family history notable for intracranial aneurysms was confirmed in 61 (18.7%) patients. One hundred and four patients (31.9%) harbored more than one intracranial aneurysm.
The median aneurysm size was 6.2 mm, with 283 (82.0%) aneurysms measuring less than 10 mm. The median aneurysm volume was 72.8 mm3 and the median neck size was 3.1 mm. The most common aneurysm locations were the ophthalmic artery (17.1%), anterior communicating artery (17.1%), superior hypophyseal artery (16.5%), and the basilar artery apex (13.3%). One hundred and nineteen (34.5%) aneurysms harbored blebs or dome irregularities and 31 (9.0%) aneurysms had intrasaccular thrombus. Of the 345 coiled aneurysms, 97 (28.1%) were ruptured.
Stand-alone coiling was performed on 257 (74.5%) aneurysms, 43 (12.5%) aneurysms were treated with stent assistance, and 47 (13.6%) were treated with balloon assistance. The median packing density following initial coil embolization was 32.3%. Eight (2.3%) patients suffered an intraprocedural aneurysm rupture. Of the 84 aneurysms with an initial IIIa closure, eight (9.5%) required retreatment for recurrence while 34 (41.0%) of the 83 IIIb occlusions required repeat coil embolization (p<0.01). In total, of the 127 aneurysms that recanalized, six (4.7%) ruptured following treatment, all of which had initial MRRC IIIb occlusions (p<0.01).
Of the 345 aneurysms studied, short-term angiographic follow-up was available on all aneurysms and long-term follow-up was available on 182 (55.8%). For all aneurysms, the median time of the first angiographic follow-up was 6.6 months and the median time of the last angiographic follow-up (prior to any aneurysm retreatment) was 11.9 months. For the 182 aneurysms with both short- and long-term follow-up, the median time of the last angiographic follow-up (prior to any aneurysm retreatment) was 37.6 months. Table 2 outlines comparative patient, aneurysm, and procedural data for all MRRC IIIa and IIIb closures. Factors associated with an initial IIIb closure included cavernous internal carotid artery location (p<0.01), increasing aneurysm size (p<0.01), increasing aneurysm volume (p<0.01), increasing aneurysm neck size (p<0.01), increasing aspect ratio (p=0.02), presence of blebs or dome irregularities (p=0.02), presence of intraluminal thrombus (p<0.01), ruptured aneurysm (p<0.01), and decreasing coil packing density (p<0.01). Multivariate logistic regression analysis identified large aneurysm size (p<0.01; OR 3.89, 95% CI 1.51 to 9.91), ruptured aneurysm (p=0.04; OR 2.46, 95% CI 1.04 to 5.79), and decreasing coil packing density (p<0.01; OR 0.95, 95% CI 0.92 to 0.98) as predictors of an initial MRRC IIIb closure.
Table 3 lists MRRC designations for all 345 aneurysms at the time of initial occlusion and at first and last angiographic follow-up, where available. Progression to complete aneurysm occlusion at the time of last follow-up was more likely to occur in MRRC IIIa than IIIb closures (53.6% vs 19.2%, p<0.01), while aneurysm recanalization was more likely to occur in MRRC IIIb than IIIa closures (65.1% vs 27.4%, p<0.01). Kaplan–Meier estimates demonstrated a statistically significant difference in the test of equality for progression to complete occlusion (figure 3; p=0.02) and recurrence (figure 4; p<0.01) between class IIIa and IIIb distributions. For the entire cohort, multivariate logistic regression analysis identified male gender (p<0.01; OR 3.18, 95% CI 1.45 to 6.97), ruptured aneurysm (p=0.04; OR 2.08, 95% CI 1.04 to 4.16), intraluminal thrombus (p<0.01; OR 8.65, 95% CI 2.83 to 26.4), and MRRC IIIb closure (p<0.01; OR 6.32, 95% CI 2.43 to 16.42) as predictors of aneurysm recanalization (table 4). For aneurysms with an initial RROC III closure, MRRC IIIa occlusion was an independent predictor of eventual progression to occlusion (p=0.02; OR 2.43, 95% CI 1.13 to 5.24).
Over the past two decades the use of endovascular coil embolization for the treatment of both unruptured and ruptured intracranial aneurysms has increased significantly.1–4 Despite this paradigm shift, the efficacy of long-term complete aneurysm occlusion differs greatly between aneurysms treated with coil embolization and clip occlusion. Aneurysm recanalization following coil embolization can be as high as 30%, while recurrence following microsurgical clip occlusion can be as low as 1.5%.5–8 ,14 ,15 Although a variety of patient and aneurysm specific factors may influence the angiographic outcome of an aneurysm treated with coil embolization, several studies have demonstrated that the degree of initial aneurysm occlusion is a robust predictor.5 ,6 ,8–11 By elucidating the factors that may predict aneurysm recanalization, more informed decisions regarding the optimal treatment strategy for a particular aneurysm can be made. Initially designed simply as a means of categorizing aneurysm closures following coil embolization, the RROC has been widely adopted into general neuroendovascular practice and is used as the metric for qualitatively assessing the degree of aneurysm occlusion. The widespread applicability of the RROC is derived from its simplicity and ease of application. While aneurysms with RROC I and II occlusions behave in largely predictable patterns,5 ,6 ,8 ,16 RROC III closures are a heterogeneous group, some of which progress to occlusion while others either remain stable or recanalize, requiring retreatment.8 The MRRC accounts for these disparate angiographic trajectories and seeks to correct this shortcoming of the RROC by dichotomizing RROC III closures into IIIa and IIIb occlusions. The present study serves to validate the MRRC for intracranial aneurysms treated with coil embolization in an external cohort of 345 aneurysms.
The hallmark of the MRRC is the distinction between aneurysm occlusions in which there is residual opacification only within the interstices of the coil mass (IIIa) and those in which there is opacification between the coil mass and the aneurysm wall (IIIb). As Mascitelli et al observed in their original publication, it is likely that aneurysm thrombosis occurs more effectively in IIIa closures given that the region of residual opacification is contained within a prothrombotic coil mass and in a region with less hemodynamic inpact.17–19 Alternatively, the prothrombotic environment of the coil mass may be less effective at precipitating thrombus formation when there is residual aneurysm filling between the coil mass and aneurysm wall. In addition, with a IIIb closure, there remains a portion of the aneurysm sac which may be subject to high hemodynamic forces as a result of a persistent aneurysm inflow jet.20 ,21 The presence of this physical space between the coil mass and aneurysm wall may allow for coil compaction and aneurysm regrowth over time, increasing the risk of aneurysm recanalization, rupture, and the need for retreatment. The advent of computational fluid dynamics may allow for the development of computer models that simulate the hemodynamic forces and wall stresses present in MRRC IIIa and IIIb closures.22 Furthermore, the dichotomization of RROC closures into MRRC IIIa and IIIb occlusions has implications beyond an aneurysm's angiographic prognosis. In the present series, mirroring the findings by Mascitelli et al, IIIb closures were 4.3-fold more likely to require retreatment than IIIa occlusions. In addition, of the 127 aneurysms that recanalized following treatment, six ruptured, all of which had initial IIIb occlusions.
Not surprisingly, this study found several important patient and aneurysm specific differences between aneurysms with MRRC IIIa and IIIb occlusions. Comparative data demonstrated cavernous internal carotid artery location, increasing aneurysm size, increasing aneurysm volume, increasing aneurysm neck size, increasing aspect ratio, presence of blebs or dome irregularities, presence of intraluminal thrombus, ruptured aneurysm, and decreasing coil packing density to be associated with a class IIIb closure. Incorporating these parameters into a multivariate logistic regression model, we further identified large aneurysm size, ruptured aneurysm, and decreasing coil packing density to be the strongest predictors of an initial MRRC IIIb closure. Independently and in aggregate, these factors increase the difficulty of achieving complete aneurysm occlusion with coil embolization.9 Furthermore, predictors of aneurysm recanalization included male gender, ruptured aneurysm, intraluminal thrombus, and MRRC IIIb closure. Ruptured aneurysms, aneurysms with intraluminal thrombus, and incompletely coiled aneurysms are well-documented risk factors for recurrence following coil embolization.10 ,11 The finding of male gender as an additional risk factor has also been previously reported.9 ,23
Unlike MRRC IIIb occlusions, which tend to recanalize and require retreatment, IIIa closures tend to progress to complete occlusion, behaving similarly to RROC I closures, as shown on multivariate regression analysis in this study. In fact, when creating the RROC, Roy et al recognized that some aneurysms with a small degree of opacification within the coil interstices progressed to complete occlusion at the time of first follow-up. According their discussion, when it was believed that this “minimal residual flow within the aneurysmal sac [was] artificially maintained by systemic heparinization, the first angiographic follow-up result served as the ‘immediate result’.”5 This was done in order to temper the number of occlusions showing ‘improved results’ from the initial embolization to the time of first follow-up.5 As such, the MRRC adds transparency to the original RROC. Despite the effect of antiplatelet medications or heparinization, aneurysms with residual opacification after coil embolization are assigned a IIIa closure if interstitial opacification is seen within the coil mass. In doing so, the MRRC IIIa designation captures aneurysms that would have naturally progressed to complete closure, as observed by Roy et al.
An additional point of consideration with respect to the MRRC—and IIIa occlusions specifically—is the issue of coil packing density. RROC I and MRRC IIIa closures have similar angiographic prognoses and may exist on the same spectrum of coil occlusion, with a densely packed coil mass extending to the aneurysm neck (RROC I) on the one extreme and a more loosely packed coil frame (MRRC IIIa) on the other. The aneurysm with a densely packed coil mass is unlikely to recanalize, whereas the aneurysm with a loosely packed coil frame, even if no opacification is seen between the coil mass and the aneurysm wall, is very likely to recur. In the middle of these two extremes, though, lie the majority of RROC I and MRRC IIIa closures. While RROC I occlusions remain the ideal angiographic outcome, the pursuit of this result is sometimes associated with an increased procedural risk.24 MRRC IIIa occlusions as a whole portend a favorable angiographic prognosis, and developing a more precise understanding of the specific coil packing density and distribution required to achieve this outcome will undoubtedly increase both the safety and efficacy profile of these endovascular coil embolizations.
Predicting the likelihood of aneurysm recanalization after endovascular intervention is critically important; however, it is estimated that less than 50% of recurrences actually warrant repeat treatment.6 The Aneurysm Recanalization Stratification Scale (ARSS) was developed with aneurysm retreatment, not recurrence, as an endpoint in order to fully capture the extent of recanalization as well as the cost, anxiety, and patient enthusiasm associated with an additional embolization procedure.10 ,11 This scale utilizes aneurysm size, presence of intraluminal thrombus, rupture status, use of procedural assist devices, and degree of initial aneurysm occlusion as assessed by the RROC to determine the probability of a given aneurysm recurring and requiring retreatment. The scale was created through retrospective evaluation of a single institution's data, but was subsequently validated in a multicenter cohort of five high-volume neurovascular centers. Given the robust predictive accuracy of the ARSS in its present form, incorporation of the MRRC may allow for an even more predictive assessment of the probability of recanalization and retreatment following initial coil embolization.
Our study has the limitations inherent in any single-institution retrospective series. Notably, assigning RROC and MRRC grades was done in a retrospective rather than a prospective fashion. In addition, the assignment of a RROC or MRRC grade is naturally subjective and disagreement between raters may occur. We attempted to limit this through independent adjudication by two image readers (CJS and CMT). Computerized models that rate the degree and location of residual opacification, while also incorporating factors such as contrast stasis and wall stress that may influence the trajectory of an aneurysm, may in the future be more objective and accurate.
The MRRC for intracranial aneurysms treated with coil embolization is a useful tool for the qualitative assessment of aneurysm occlusion. Expanding upon the RROC, the MRRC dichotomizes RROC III occlusions into IIIa (residual aneurysm opacification within the interstices of the coil mass) and IIIb (residual aneurysm opacification between the coil mass and the aneurysm wall) occlusions, and the present study validates the observation that IIIa occlusions portend a more favorable angiographic prognosis. We recommend that this updated intracranial aneurysm occlusion scale be adopted into widespread neuroendovascular practice.
Contributors All authors had integral participation in the study. CJS, CMT, BPW, JRM, TML-M, CSO, and ABP conceived the project idea. CJS, BPW, and CMT performed the data collection. CJS, CMT, JDR, TML-M, JAH, AJY, and CSO performed the angiographic review. CJS performed the statistical analysis and figure/table presentations. All authors were involved in the manuscript preparation and final approval.
Competing interests None declared.
Ethics approval MGH IRB 2014P001039.
Provenance and peer review Not commissioned; externally peer reviewed.