Author + information
- Received January 26, 2005
- Revision received May 18, 2005
- Accepted May 31, 2005
- Published online October 18, 2005.
- Ivan P. Casserly, MB, BCh⁎,
- Alex Abou-Chebl, MD†,
- Robert B. Fathi, MD, PhD†,
- David S. Lee, MD†,
- Jacqueline Saw, MD‡,
- Jose E. Exaire, MD§,
- Samir R. Kapadia, MD†,
- Christopher T. Bajzer, MD† and
- Jay S. Yadav, MD†,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Jay S. Yadav, Department of Cardiovascular Medicine, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Desk F25, Cleveland, Ohio, 44195
Objectives The purpose of this research was to define the predictors of the “slow-reflow” phenomenon during carotid artery intervention with filter-type embolic protection devices (EPDs) and to determine its prognostic significance.
Background During carotid artery intervention using filter-type EPDs, we have observed cases in which there is angiographic evidence of a significant reduction in antegrade flow in the internal carotid artery proximal to the filter device, termed “slow-flow.” The predictors of this phenomenon and its prognostic significance are unknown.
Methods Using a single-center prospective carotid intervention registry, patients with slow-flow were compared to patients with normal flow during carotid intervention with respect to clinical, procedural, and lesion characteristics, and the 30-day incidence of death and stroke.
Results A total of 414 patients underwent 453 carotid artery interventions using EPDs. Slow-flow occurred in 42 patients (10.1%) undergoing 42 carotid interventions (9.3%), and most commonly occurred after post-stent balloon dilatation (71.4%). Multivariate logistic regression analysis identified the following predictors of slow-flow: recent history (<6 months) of stroke or transient ischemic attack (odds ratio [OR] 2.8, 95% confidence interval [CI] 1.4 to 5.6, p = 0.004), increased stent diameter (OR 1.4, 95% CI 1.02 to 1.94, p = 0.044), and increased patient age (OR 1.05, 95% CI 1.01 to 1.09, p = 0.025). Among patients with slow-flow, the 30-day incidence of stroke or death was 9.5% compared to 2.9% in patients with normal flow (chi-square = 4.73, p = 0.03). This difference was driven by the disparity in the 30-day incidence of stroke (9.5% vs. 1.7%).
Conclusions Slow-flow during carotid intervention with EPDs is a frequent event that is associated with an excess risk of periprocedural stroke. The association of the phenomenon with clinically symptomatic carotid lesions and use of larger stent diameters suggests that embolization of vulnerable plaque elements may play a pathogenic role.
During carotid artery intervention using filter-type embolic protection devices (EPDs), we have observed cases in which there is angiographic evidence of a significant reduction in antegrade flow in the internal carotid artery (ICA) proximal to the filter device (Fig. 1),termed “slow-flow” (1,2). The angiographic appearance is reminiscent of the no-reflow phenomenon initially described during percutaneous coronary intervention in the absence of EPDs (3), although these two processes are clearly pathologically distinct. At this time, the mechanism of slow-flow during carotid intervention is unknown. Our working hypothesis is that particulate debris containing plaque elements results in blockage of the filter pores, impeding antegrade flow. The reduction in antegrade flow leads to a stagnant debris-containing column of blood in the artery proximal to the filter. As a result, our practice has been to aspirate the column of blood proximal to the filter in cases in which significant slow-flow is observed, in an effort to minimize the risk of distal embolization after filter collapse and retrieval (Fig. 2).
In a large single-center experience, we prospectively identified all patients with slow-flow during carotid intervention with filter-type EPDs. In an attempt to gain insights into the mechanism and prognostic significance of the phenomenon, we sought to identify clinical, procedural, or angiographic predictors of slow-flow, and to compare the 30-day clinical outcomes of patients with slow-flow versus those with normal flow during the procedure.
All patients undergoing percutaneous carotid interventions at the Cleveland Clinic Foundation are entered in a prospective carotid intervention database, in which baseline clinical characteristics, procedural and angiographic data, and clinical outcomes are collected. From this database, we identified all patients undergoing carotid intervention using filter-type EPDs from the date of their initial use (February 2000) until March 2004. Procedures in which slow-flow occurred were prospectively recorded.
Regarding angiographic data, the degree of carotid stenosis was visually estimated. The length of carotid lesions was visually estimated using the diameter of the guide catheter or sheath, or the length of interventional equipment (i.e., angioplasty balloon or stent) as a reference dimension. The following carotid plaque morphological characteristics were assessed qualitatively: the presence of an ulcerated plaque surface, eccentric distribution of plaque in vessel lumen, and prominent calcification of the plaque.
The primary clinical outcome measure of this analysis was prospectively defined as the incidence of stroke and/or death at 30 days. Stroke was defined as a new neurologic deficit that persisted beyond 24 h.
Carotid stent technique
All patients were treated with pre-procedural aspirin and clopidogrel (for at least 48 h) and were continued for a minimum of four weeks after the intervention. Intravenous heparin was administered intraprocedurally to achieve a target activated clotting time (ACT) of 300 s. After delivery of an 8-F guide catheter or 6-F sheath in the common carotid artery, the lesion was crossed with a filter-type EPD. Most lesions were pre-dilated with a coronary-type 4-mm diameter balloon, followed by deployment of a self-expanding nitinol stent, and post-dilation of the stent with a 5.0- to 5.5-mm diameter balloon. Carotid flow was angiographically assessed after each step. When slow-flow was observed, multiple aspirations of the column of blood proximal to the filter (for a total of 40 to 60 cc) were performed. This was achieved using a multipurpose catheter or the Export aspiration catheter (Medtronic Inc., Minneapolis, Minnesota). At the completion of the carotid intervention, the filter device was retrieved, and final assessment of carotid and cerebral flow was performed. Patients were discharged one day after an uncomplicated procedure after evaluation by a neurologist. Outpatient follow-up with neurological assessment by a neurologist was scheduled at 30 days.
Continuous variables appear as mean values with SDs and categorical variables as frequencies with percentages. Using chi-square and Fisher exact (when any one cell of a comparison table had less than five items in it) tests for discrete variables, and independent-samples ttest for continuous variables, the baseline clinical characteristics, and procedural and angiographic variables were compared between patients who experienced slow-flow during the procedure with those who had normal flow. Binary logistic and multivariate logistic regression analyses were performed to identify univariate and multivariate predictors of slow-flow, respectively. For this analysis, only the index procedure was included for patients with multiple procedures, and only data for the ICA lesion was included for patients with multiple lesions treated during the procedure. The frequency of death and/or stroke at 30 days was calculated for the slow-flow and normal flow groups. Kaplan-Meier event-free curves at 30 days were constructed, and chi-square test performed to assess for significance. Statistical analysis was performed using SPSS 9.0 statistical software (SPSS Inc., Chicago, Illinois) and JMP 5.1 statistical software (SAS Institute, Cary, North Carolina).
A total of 414 patients underwent 453 carotid artery interventions using filter-type EPDs during the study period. Slow-flow in the ICA occurred in 42 patients (10.1%) during 42 procedures (9.3%). Among patients with slow-flow, complete absence of antegrade flow was observed in 10 patients (23.8%), while reduced antegrade flow was observed in the remaining 32 patients to varying degrees (76.2%). Slow-flow was detected at the following stages of the procedure: after pre-dilation in one patient (2.4%), after stent deployment in 11 patients (26.2%), and after post-stent balloon dilatation in the remaining 30 patients (71.4%). Slow-flow resolved in all patients after retrieval of the filter.
Prediction of slow-flow
The clinical characteristics of the 42 patients with slow-flow and the remaining 372 patients with normal flow are shown in Table 1.Patients with slow-flow were older and more likely to have a history of an anterior circulation transient ischemic attack (TIA) or stroke within the previous six months.
Table 2lists the procedural variables for the slow-flow and normal flow groups. All of the 42 patients with slow-flow underwent a single carotid intervention during the study period. These events were distributed evenly throughout the study period. The remaining 372 patients underwent a total of 411 carotid interventions during the same period. The mean peak procedural ACT and mean heparin dose/kg were equivalent in both groups, as was the mean fluoroscopic time for the procedure. Primary stenting was performed in ∼10% of procedures, whereas balloon angioplasty or cutting balloon angioplasty alone was performed in 2% of patients in the normal flow group only. The most common stent type used was the Precise/Smart stent (Cordis, Johnson & Johnson, Warren, New Jersey). During the study period, four different filter-type EPDs were utilized: Angioguard (Cordis, Miami, Florida), FilterWire EX/FilterWire EZ (Boston Scientific, Natick, Massachusetts), AccuNet (Guidant Corp., Indianapolis, Indiana), and NeuroShield (Abbott, Galway, Ireland). The Angioguard was the dominant filter employed, being used in 64% of all procedures.
A comparison of the lesion characteristics in the slow-flow and normal flow groups is shown in Table 3.A total of 44 lesions were treated during the 42 procedures in the slow-flow group. A total of 420 lesions were treated during the 411 procedures in the normal flow group. Plaque ulceration (45% vs. 28%, p = 0.021), larger stent diameter (7.9 ± 0.7 mm vs. 7.4 ± 1.4 mm, p = 0.012), and smaller post-stent balloon dilation diameter (5.5 ± 1 mm vs. 5.8 ± 0.5 mm, p = 0.026) were more commonly associated with slow-flow.
Using binary logistic regression, statistically significant (i.e., p ≤ 0.05) clinical, procedural, and lesion-related univariate predictors of slow-flow were determined and included: increased age, history of TIA/stroke within six months, history of congestive heart failure, lesion percent stenosis at baseline, lesion ulceration, stent diameter, and post-stent balloon dilation diameter. A formal interaction analysis confirmed the absence of a significant interaction between these variables. The univariate predictors were entered into a multivariate logistic regression analysis. Increased patient age, a history of TIA/stroke within six months, and larger stent diameter were independently predictive of the slow-flow phenomenon (Table 4).The C statistic for the model was 0.89.
The 30-day incidence of stroke, death, and myocardial infarction among patients with slow-flow was 9.5% (n = 4), 0%, and 2.3% (n = 1), respectively (Fig. 3).All four strokes in this group occurred on the day of the procedure. Neurological symptoms were first noted after retrieval of the filter in all four cases. Of the four strokes, two were classified as minor, with a normal neurological exam at 30-day follow-up (National Institutes of Health Stroke Scale = 0). Among patients with normal flow during carotid intervention, the 30-day incidence of stroke, death, and myocardial infarction was 1.7% (n = 7), 1.5% (n = 6), and 1.2% (n = 6) (Fig. 4).The pre-defined primary clinical outcome of stroke and/or death at 30 days occurred in 9.5% (n = 4) of patients with slow-flow and 2.9% (n = 9) of patients with normal flow (chi-square = 4.73, p = 0.03) (Fig. 4).
This is the first report that defines the predictors and prognostic significance of patients experiencing slow-flow during carotid artery intervention using filter-type EPDs. In this large single-center experience (n = 414), slow-flow occurred in 10.1% of patients, and 9.3% of all carotid procedures. Significant multivariate predictors of the event included a recent history of TIA/stroke, increased patient age, and increased stent diameter. Slow-flow was associated with an adverse clinical outcome, with a 30-day incidence of stroke and/or death of 9.5%, compared to 2.9% among patients with normal flow (p = 0.03).
Despite the high frequency of slow-flow in this series, the phenomenon has not been well described in the carotid intervention literature. In a recent multicenter registry of 753 patients, slow-flow was noted in 7.9% of patients in whom an EPD was utilized (2). Smaller single-center studies reported an incidence of between 7.2% and 22% (1,4). Together with the current study, these data support that the phenomenon is real and that, using the current generation of filter-type EPDs, the true incidence is likely to be at least in the 8% to 10% range.
Previous studies have provided little further descriptive detail about the phenomenon of slow-flow. In this series, slow-flow occurred exclusively in patients who had a stent placed, although it should be accepted that only eight patients were treated with angioplasty alone. The event occurred predominantly after post-stent balloon dilatation, although over a quarter occurred after deployment of the stent alone. No-flow represents the most severe manifestation of the phenomenon, where antegrade flow completely ceases and occurred in nearly a quarter of the patients. The remaining patients exhibited reduced antegrade flow, which encompasses a broad spectrum of flow impairment. In its mildest form, the ICA fills completely but marginally slower than the external carotid artery. In its most severe form, the contrast fills only part of the ICA lumen and moves very slowly beyond the filter.
The underlying pathogenesis of slow-flow during carotid intervention with EPD is uncertain. One indisputable fact is the observation that normal antegrade flow is restored in all patients with slow-flow after retrieval of the filter EPD. This clearly isolates the filter as the culprit and strongly suggests that occlusion of the pores in the filter membrane is responsible for the impairment in antegrade flow. Thus, the filter-type EPD is converted into a partial or completely occlusive device. The lack of antegrade flow would then allow stagnation of the blood column proximal to the filter and the accumulation of microemboli and debris in this column. Histopathological analysis of the debris released during carotid artery stenting and of the debris retrieved from the membranes of filter-type EPDs reveals the presence of amorphous material derived from the core of atherosclerotic plaque, principally lipid-laden macrophages, cholesterol clefts, and fibrin material (1,4–6). A rationale hypothesis is that the phenomenon of slow-flow during carotid intervention with EPD occurs due to an increased extrusion of these plaque elements during carotid intervention. This is supported by the close temporal relationship of slow-flow to stent deployment and post-stent balloon dilatation, which are primarily responsible for procedure-related plaque embolization (7). Our finding that symptomatic carotid lesions and increased stent diameter are independent predictors of the phenomenon is also consistent with this hypothesis. Symptomatic plaques are more likely to contain the pathological elements that contribute to the particulate debris retrieved from filter-type EPDs after carotid intervention (1,8). It also seems reasonable that plaque extrusion may be increased with larger stent diameter sizes. Slow-flow may actually serve as an important marker of vulnerable unstable plaque that is prone to embolization, even among asymptomatic patients.
An alternative potential explanation for complete or partial occlusion of the filter during slow-flow is the in-situ formation of thrombus in the filter device. Our data argues against this explanation. Various measures of the degree of anticoagulation in both groups (peak procedural ACT, heparin dose/kg) were no different between patients with slow-flow and normal flow. All patients received the same standardized pre-procedural regimen of aspirin and clopidogrel. Finally, total fluoroscopy times for the procedure were no different between the groups, making it unlikely that slow-flow occurred in patients with prolonged procedures and waning levels of anticoagulation.
The importance of slow-flow is clearly underscored by the disparity in clinical outcome between patients who experience the phenomenon and those who do not. Procedure-related stroke represents the major hazard observed in patients with slow-flow, occurring in 9.5% of patients, compared to 1.7% of patients with normal flow. Magnetic resonance imaging from one of the patients with slow-flow and stroke in this study is shown in Figure 5.The appearance is that of multifocal punctate infarcts in the distribution of the anterior circulation, which is consistent with an embolic mechanism. It is unclear whether these emboli represent a primary failure of the filter to capture extruded debris, or reflect the embolization of debris that accumulated in the column of blood proximal to the filter under the conditions of slow-flow and remained despite the aspiration techniques we employed in this cohort. We suspect that the latter is the more likely explanation, because the patients developed neurological symptoms only after the filter was collapsed. In the first patient in which one of the authors (J.Y.) observed this phenomenon at another institution, aspiration was not performed before capturing the filter, and the patient suffered a large stroke. This resulted in our practice of aspirating proximal to the filter after the recognition of slow-flow in an attempt to remove this debris and prevent its distal embolization after retrieval of the filter. The practice is supported by occasional histological analysis of the aspirate obtained during cases of slow-flow that demonstrate the presence of acellular and amorphous debris containing lipid-rich macrophages and cholesterol crystals (Fig. 6).We strongly advise that this technique be employed in all patients with angiographic evidence of slow-flow. Although we accept that strokes still occurred with increased frequency in patients with slow-flow in this study, without aspiration, it is our belief that the stroke rate would have been even higher, and that more of the slow-flow patients would have suffered major strokes.
The patient cohort in this observational study represents a “high-risk” group with multiple comorbidities, studied at a single institution, who were largely treated using the Angioguard EPD and Precise nitinol stent. Further examination of frequency, predictors, and outcomes of slow-flow should be performed in “high-risk” groups at other centers, in “lower-risk” cohorts, and using other types of EPDs and stents in order to confirm our findings. It is important to emphasize that the clinical outcome of patients in this study with slow-flow during carotid intervention reflects the outcome achieved using a strategy of aspiration before filter retrieval. It is accepted that this strategy is not supported by rigorous clinical investigation (i.e., randomized study), but seems entirely reasonable based on our current understanding of the phenomenon of slow-flow. It is possible that the difference in outcome between patients with and without slow-flow would be even greater if this strategy was not employed.
- Abbreviations and Acronyms
- activated clotting time
- embolic protection device
- internal carotid artery
- transient ischemic attack
- Received January 26, 2005.
- Revision received May 18, 2005.
- Accepted May 31, 2005.
- American College of Cardiology Foundation
- Angelini A.,
- Reimers B.,
- Della Barbera M.,
- et al.
- Rezkalla S.H.,
- Kloner R.A.
- Reimers B.,
- Corvaja N.,
- Moshiri S.,
- et al.
- Manninen H.I.,
- Rasanen H.T.,
- Vanninen R.L.,
- Vainio P.,
- Hippelainen M.,
- Kosma V.M.
- Golledge J.,
- Greenhalgh R.M.,
- Davies A.H.