Factors Influencing Successful Angiographic Occlusion of Aneurysms Treated by Coil Embolization

Limitations of the Study

The study population was derived from patients with aneurysms who were referred for endovascular therapy at our unit between 1992 and 1995. Referral was based on anticipated surgical difficulty or surgical failure, and the aneurysms in the group of patients studied were generally representative of those aneurysms that were then treated by the endovascular route (1, 2, 4). The greater number of posterior circulation aneurysms and large and giant aneurysms in this study reflects the increased surgical morbidity in these patients (9, 18, 19). The study, therefore, is not representative of the population with aneurysms treated surgically (7, 20, 21), and caution should be exercised in extrapolating data derived from it to the general population. A second caveat is derived from the selection criteria that required preservation of the parent artery, thus excluding those aneurysms treated by endovascular parent artery occlusion. Finally, the aneurysms treated represented the early experience of the authors, and current success rates would be expected to be greater with more recently treated aneurysms because of improved selection criteria, treatment strategies, and the availability of a greater range of coil sizes and types. The higher success rates at treatment and at follow-up as well as the lower rate of occurrence of compaction in the second group of 32 aneurysms treated is illustrative of this point.

A limitation of the study is the lack of any correlation between angiographic success or failure and clinical outcome. At the time of the study, our unit was one of the few centers in Britain providing this treatment, and most referrals were from outside centers. Clinical records, therefore, were difficult to obtain within the period of data analysis and were incomplete as compared with the angiographic data. A decision was made to study only the angiographic data as well as any complete clinical data that were available at the time of referral.

The method used to measure the degree of packing of the aneurysmal lumen is a subjective one and, as such, may be open to criticism. The degree of packing was assessed as a factor in the success of treatment, because experimental studies had shown that compaction occurred after coil placement (13, 14). Graves et al (13) reported in a longitudinal study that compaction occurred in the first week after coil placement, and Mawad et al (14) reported that compaction was a cause of treatment failure. Although we noted a trend for more tightly packed aneurysms to have fewer treatment failures at follow-up (40%, 50%, 77%, and 79% for packing grades 3, 4, 5, and 6, respectively), this trend did not achieve statistical significance (Table 1). In part, this failure to establish significance may be a reflection of the subjective method used to measure the degree of packing.

The reference standard for linear measurements was derived from Zubillaga et al (15). While this decision may be criticized on the grounds that we have studied a different population, it does at least provide a “norm,” so that the same error, if any, applies to all cases. Second, this reference value cannot be applied in cases in which vasospasm is present (15). In those patients, a different method was used (see Methods). It may be argued that because seven of the 10 patients with vasospasm were in the acutely treated group, use of a different linear reference value in this group introduces a bias that might account for the higher success rate observed in the acutely treated group (84%) versus those treated after 16 days (52%) (Table 2). Exclusion of those patients with vasospasm from analysis, however, actually raises the mean neck size in the acutely treated group to 5.05 mm (4.13 mm for those with delayed treatment) and would only slightly affect the differential success in the two groups (77% for acutely treated patients and 47% for those with delayed treatment), so that use of a different reference value for linear measurements does not appear to be a potentially significant error.

A number of other potential sources of error arise out of the methodology. The first stems from the differences in magnification used when obtaining angiograms. These differences in magnification were corrected for by using a standard lecture room overhead projector to project the angiographic images onto the calibrated paper. By adjusting the distance from the projector to the wall on which the paper was pinned, different degrees of magnification were created that allowed the projected images to be accurately superimposed on one another.

Differences in the angle of projection between the different angiograms might have been more problematic were it not for the fact that the angle of projection that was used for the pretreatment and posttreatment studies was one of the two projections used in coil embolization. This same projection angle was used immediately before deployment of the first coil (the pretreatment angiogram) and at the end of the coiling (the posttreatment study) and was chosen because, among other factors, it best displayed the relationship of the aneurysm and its neck to the parent artery and its branches. A greater difficulty arose when matching these two projections with the follow-up angiogram obtained months later. While it was our practice at the time of follow-up to reproduce as closely as possible the same projection used during embolization, this was not always identical, as shown in Figures 1 and 2, particularly if this projection was not the one that showed compaction or growth to its greatest extent (Fig 1). In the latter situation, we would search the pretreatment and posttreatment angiograms to find the projection that most closely approximated the follow-up study. At all times in the study we attempted to match the anatomy at the neck region above all else in order to calculate changes in the rest as accurately as possible.

We defined a change in the aneurysmal residue between treatment and follow-up as having occurred when the percentage of occlusion changed by greater than 2.5% or when the rest had changed by greater than 0.25 mm. We minimized errors in this calculation by 1) using paper calibrated to 1 mm, 2) always maximally projecting the image of the aneurysm to cover the largest number of whole squares (usually between 200–400), 3) counting partial squares as well as whole ones, and 4) repetitively measuring images of coiled and uncoiled aneurysms before beginning the study until we could achieve an error rate less than the change we were searching for. We accept, however, that errors in our methodology will probably overestimate the number of aneurysms that changed between treatment and follow-up.

Finally, it is axiomatic that aneurysms are 3D structures that cannot be easily reduced to a 2D analysis. It is our hope that a future study that looks at changes in the aneurysm using a 3D analysis will confirm our findings.

In previous reports, the angiographic measurement of the success of endovascular treatment of aneurysms has usually been presented as a subjective percentage of occlusion (1–4, 15, 22) or simply as a descriptive term, such as “incompletely treated” (14). Objective measurements of the percentage of occlusion have the advantage of reproducibility and can be used, as in this study, to assess subtle changes in the residual lumen in longitudinal studies. This method, however, is time-consuming and contains two flaws. First, the same percentage of occlusion in different-sized aneurysms will produce different-sized residual lumens, and, second, the same percentage of occlusion in identically sized aneurysms can leave different lengths of the aneurysmal wall exposed (Fig 3). An alternative is to measure the size of the aneurysmal rest. This measurement of the rest size is quick, reproducible, and widely used in the neurosurgical literature (16, 17, 23–27) (allowing comparison between surgical and endovascular techniques) and establishes a correlation between rest size and risk of rehemorrhage (25). While the minimum size at which a rest is safe from rehemorrhage is controversial (17, 26), one apparently generally held view among various authors is that aneurysmal rests of less than 2 mm carry a basically benign prognosis (17). On this basis, we defined successful treatment as a rest size equal to or less than 2 mm.

Factors Influencing Success Rates

Because of the way we defined success, our results at follow-up will appear to be greater than that of others who defined success as 100% occlusion (4). Applying this same 100% criterion to our aneurysms, however, produces success rates that are almost identical.

By our definition of success, 35% of our aneurysms were failures at follow-up. Even allowing for the increased number of posterior circulation aneurysms in this series, with their higher rate of occurrence of rests (9, 17), this failure rate is significantly higher than the 5% observed in surgical series (16, 17). Even though angiographic follow-up of small surgical rests is not commonly associated with enlargement and rehemorrhage (16), nonetheless an increased risk of such an event exists in incompletely treated aneurysms (11, 23–26, 28–30), and it is important that the factors that influence success be studied to enable better case selection.

Previous studies have reported that luminal size, neck size, and, experimentally, the degree of packing of the aneurysmal lumen are factors that influence successful treatment of aneurysms (2–4, 13–15, 31). Hilal argued the importance of neck size (31), noting that smaller necks would be more likely to be bridged by the mesh of the coil ball, thus excluding the aneurysmal lumen from the parent circulation and allowing the endothelium to cover the neck. A success rate of 85% was achieved for aneurysms with necks 4 mm or less versus 16% for wide-necked aneurysms (15). Our results confirm the importance of neck size as a predictor of success, both at treatment (P = .002) and at follow-up (P = .008). The importance of neck size is reinforced by the finding that both vasospasm (P = .011) and success at the time of treatment (P = .0001) were predictors of success at follow-up. In addition, neck size influences the likelihood of change in the rest, both directly as well as indirectly, through its correlation with luminal diameter, vasospasm, and success at treatment.

Although we have confirmed the trend noted by others for luminal size to influence success (1–3), we do not believe that it is an independent predictor of success, and any significance of luminal size as a factor in this study (P = .07) is likely to be secondary to its direct correlation with neck size (mean neck sizes for small, medium, and large aneurysms were 3.35 mm, 5.34 mm, and 8.2 mm, respectively). We believe that this same argument holds true for the significance of luminal diameter as a factor influencing change in the rest.

The success or failure of treatment at follow-up, according to mode of presentation, is presented in Tables 1 and 2. Our results suggest a further significant differential success rate for the aneurysms treated before and more than 15 days after hemorrhage (84% and 52%, respectively) that cannot be explained on the basis of differences in mean neck size or number of wide-necked aneurysms between these two groups. This observation raises the intriguing possibility that the acutely ruptured aneurysm may be more receptive to thrombosis of the lumen after coil placement than those that are treated after an interval. Evidence that might support this suggestion comes from two sources. Casasco et al (2) treated 71 aneurysms with fiber coils, 67 of them after a hemorrhage, and noted a 100% occlusion rate in 85%. This is a significantly higher success rate than that recorded in this study or elsewhere (4) and may reflect a higher percentage of cases treated after hemorrhage. Part of this success, however, may be explained by the increased thrombogenicity of the fiber coils used (13). Also, endoluminal thrombus may transiently form within an aneurysmal lumen after rupture (32–34), and it is speculated that placement of a coil into the thrombotic milieu of a recently ruptured aneurysm may, with the additional electrothrombotic effect induced by coil detachment (22), encourage aneurysmal thrombosis. Our suggestion that early treatment may be associated with a higher success rate is based on a small number of cases and requires confirmation by a larger study.

Morphologic Changes in the Aneurysmal Residue

Changes in size of the aneurysmal rest between treatment and follow-up have been observed after both coiling and surgical clipping (2, 4, 16) and include decreases in size or obliteration of the residue as well as, less commonly, an increase in its size (16). In this study, changes were observed in 36 (59%) of 61 treated aneurysms. While this suggests that the aneurysmal residues were unstable, the criteria used to decide that a change had occurred were deliberately severe, so as to provide a large enough group for a meaningful analysis of change. In fact, the number of aneurysms that actually changed their status as successes or failures as a result of these changes was much smaller (14 [23%] of 61) and the net deficit of successes was even smaller (four [6.6%] of 61). An example of this is the apparently contradictory finding that while compaction was seen in nine (35%) of 26 vertebrobasilar aneurysms, the difference in success rates at treatment and follow-up was only 14%. The likelihood of change in rest morphologic features appears to depend primarily on neck size, with both growth and compaction occurring more frequently in wide-necked aneurysms. Growth of a preexisting aneurysm occurs from its wall (35), and, to prevent growth after coiling, successful treatment of the aneurysm must isolate the thinner wall of the aneurysm from the hemodynamic forces that contribute to its enlargement (6, 10, 36). Since rehemorrhage after surgical treatment correlates with the presence of a rest (5–11) and this risk increases with increasing rest size (25), it is predictable that aneurysmal growth after endovascular coiling would likewise occur more often in the presence of treatment failure, and particularly those failures that had larger rests. This argument is supported by the finding that five growths occurred in initial treatment failures (n = 18), with a mean rest size of 6.4 mm (range, 4.7–10.4 mm). Wide-necked aneurysms are also more likely to have compaction at follow-up, presumably because hemodynamic forces are able to act on a larger surface area of the coil ball at the neck of the aneurysm. While we did not establish a significant link between the degree of packing and success at treatment, there was a trend toward significance at follow-up (P = .104). Well-packed aneurysms were less likely to change and less likely to show coil-ball compaction than less well-packed lumens, reinforcing experimental suggestions that this is an important factor in the stability of the residue (13, 14).

Strother et al (36) have reported that the flow within experimental aneurysms is predictable and that it varies primarily according to the relationship of the aneurysm to its parent artery. While we did not specifically address this assertion, we did observe a number of findings that lead us to suggest that this is an important variable in the success of endovascular therapy. Aneurysmal growth was most common in sidewall aneurysms, and compaction occurred most frequently in bifurcation and termination types. This may be a reflection of the differences in inflow patterns at the ostia of the aneurysms. Anterior communicating artery aneurysms, which are not easily placed into one of the types described by Strother et al (36), rarely had compaction (one of 11) or growth (one of 11), with nine of 11 having no change or progressive thrombosis. In this case, it is possible that turbulence induced by colliding inflows from the opposing A1 segments acts to encourage the formation of thrombus in the presence of coils within the aneurysmal lumen. Finally, compaction or growth of aneurysms was not seen with aneurysms treated in the presence of vasospasm, and five of eight such aneurysms had progressive thrombosis between treatment and follow-up. While this could be a reflection of the smaller neck size in those aneurysms with vasospasm, alterations in inflow hemodynamics occurring as a consequence of the vasospasm may be a contributing factor.