Article Text

Original research
Dexmedetomidine: a safe alternative to general anesthesia for endovascular stroke treatment
  1. Matthew K Whalin1,
  2. Susan Lopian2,3,
  3. Katleen Wyatt3,
  4. Chung-Huan J Sun2,
  5. Raul G Nogueira2,3,
  6. Brenda A Glenn3,
  7. Raphael Y Gershon1,
  8. Rishi Gupta2,3
  1. 1Department of Anesthesiology, Emory University, Atlanta, Georgia, USA
  2. 2Department of Neurology, Emory University, Atlanta, Georgia, USA
  3. 3Marcus Stroke and Neuroscience Center, Grady Memorial Hospital, Atlanta, Georgia, USA
  1. Correspondence to Dr Rishi Gupta, Department of Neurology, Emory University, 49 Jesse Hill Jr Drive, SE, Faculty Office Bldg #393, Atlanta, GA 30303, USA; Rishi.gupta{at}emory.edu

Abstract

Background and purpose There have been reports that general anesthesia (GA) is associated with worse clinical outcomes during intra-arterial treatment (IAT) for stroke. Since traditional sedatives carry the risk of respiratory depression, this retrospective study was designed to compare sedation with the α2 adrenergic agonist dexmedetomidine (DEX) and with GA for IAT procedures.

Methods We reviewed our institutional endovascular database of 216 consecutive patients who received DEX or GA for IAT of anterior circulation strokes between September 2010 and July 2012. The demographic, radiographic and angiographic variables between the GA and DEX groups were compared, as well as hemodynamic changes during the procedure. Binary logistic regression models were generated to determine the independent predictors of a favorable outcome (defined as a modified Rankin Score at 90 days of 0–2).

Results 83 patients had IAT performed under DEX sedation. Their demographic characteristics were similar to those given GA except that they were older and had less severe strokes. The GA group experienced greater variations in blood pressure, more hypotension with induction (54% vs 28%, p<0.001) and greater use of vasopressors (79% vs 58%, p<0.001). In our regression models, independent predictors of a good outcome included age, NIH Stroke Scale (NIHSS) score, Alberta Stroke Program Early CT score (ASPECTS), successful reperfusion, lower baseline systolic blood pressure and higher blood pressures during the procedure. DEX was associated with a good outcome when models included NIHSS as the sole measure of stroke severity but was equivalent to GA when ASPECTS was added to the analysis.

Conclusions DEX can be safely administered in patients undergoing endovascular reperfusion therapies. Further study is required to determine if outcomes are different among sedatives used during such procedures.

  • Stroke
  • Thrombectomy

Statistics from Altmetric.com

Request Permissions

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.

Introduction

Institutional preferences for type of sedation vary tremendously in endovascular reperfusion therapies for acute ischemic stroke. The use of general anesthesia (GA) may allow for protection of the airway in patients with large strokes and may increase safety by decreasing patient movement. On the other hand, hypotension during the induction or maintenance of GA may have an adverse effect on outcome.1 Recent retrospective studies have shown that conscious sedation appears to be safe for such procedures, particularly with regard to vessel perforation and intracranial hemorrhagic complications.2 ,3 Concerns with conscious sedation practices include respiratory depression from traditional intravenous sedatives and the potential need for emergency airway protection during the procedure when catheters are in position. This may lead to a potentially dangerous scenario for the patient.

Dexmedetomidine (DEX) is an α2 adrenergic agonist which may provide patient sedation and comfort with little or no respiratory depression. It is approved for sedation in procedures with monitored anesthetic care, and randomized controlled trials have shown advantages over benzodiazepine sedation in the ICU.4 ,5 The lack of respiratory depression frequently allows DEX infusions to continue during breathing trials and after extubation in the critical care setting.6 Not only is respiratory drive preserved, but patients sedated with DEX can be easily aroused to cooperate with breath holds or neurologic testing.7 We began to use DEX sedation for endovascular reperfusion therapy with our anesthesia department in 2010. The purpose of this study was to assess if the use of DEX is a safe alternative to GA and if there are outcome differences based on the type of sedation employed during reperfusion therapy procedures.

Methods

As part of our continuous quality improvement process we maintain a prospective database of patients treated in our neurointerventional suite. After approval from our institutional review board, we retrospectively reviewed the medical records of patients from this database treated with endovascular reperfusion therapy for acute ischemic stroke between September 2010 and July 2012. Patients underwent angiography for definitive reperfusion after assessment of imaging and the clinical condition of the patient. The treatment decision was determined jointly by a vascular neurologist and neurointerventionalist. A total of 216 consecutive patients with an anterior circulation stroke treated with endovascular reperfusion therapy with GA or DEX were included in the analysis. Patients with a posterior circulation stroke (n=21) were excluded from the analysis as all of these patients were treated with GA.

Baseline characteristics were retrospectively collected from the medical and anesthesia records and included age, gender, vascular risk factors (diabetes mellitus, hypertension, atrial fibrillation, hyperlipidemia, smoking, ejection fraction, HgbA1c), NIH Stroke Scale (NIHSS) score at presentation, Alberta Stroke Program Early CT scores (ASPECTS) on initial CT imaging,8 time to intravenous thrombolysis or time to groin puncture and location of the thrombus. Angiographic variables including the rate of successful recanalization (defined as Thrombolysis in Cerebral Infarction 2b or 3)9 and use of intravenous tissue plasminogen activator prior to the intervention were also collected. Procedure times were calculated as the interval between initial groin puncture and successful reperfusion. If successful reperfusion could not be achieved, the time of the final device pass was used to determine the overall procedure time. Hemodynamic parameters were collected before, during induction and throughout the procedure based on the anesthesia record. Almost all patients began with non-invasive blood pressure measurements but many had arterial catheters placed during the procedure. Post-sedation blood pressures were recorded as mean arterial pressures (MAPs) since this offers better agreement between non-invasive and invasive measurements than systolic or diastolic. Hypotension was defined as a 20% fall in blood pressure and bradycardia as a 20% fall in heart rate. Additionally, use of medications such as vasopressor agents or antimuscarinic agents (atropine or glycopyrrolate) was captured along with the maximum and minimum blood pressure and heart rate. DEX administration varied by anesthesia provider, with typical infusion rates starting at 0.5 μg/kg/h and ranging from 0.3 to 1.0 μg/kg/h. Not all providers gave loading doses, but our recommendation is to provide 0.5 μg/kg over 5 min which may be repeated if the patient remains hemodynamically stable. The number of patients converted from DEX to GA was also captured along with the reason for conversion. Blinded neurological outcome was obtained at 90 days using the modified Rankin Score (mRS) by a certified healthcare provider either via telephone interview or clinic visits. Patients with mRS of 0–2 were classified as having a good neurological outcome.

Medical complications

Patients were assessed for the need for a tracheostomy or percutaneous endoscopic gastrostomy (PEG) tube after the procedure. Nosocomial pneumonias were identified as any new or evolving pulmonary infiltrate on radiographic chest imaging occurring at least 48 h after admission with a combination of two of the following features of infectious origin: fever ≥38°C, leukopenia or leukocytosis and/or purulent secretions consistent with positive cultures.10

Radiographic measurements

CT images were reviewed 24–36 h after the procedure to assess for hemorrhagic conversion using the European Cooperative Acute Stroke Study definition. Symptomatic hemorrhages were defined as a parenchymal hematoma types 1 or 2.11 All included patients were imaged 24–48 h from time of admission by MRI or >48 h with CT if MRI was contraindicated. The final infarct volume (FIV) was calculated by measuring the area of the infarct on each slice and summating individual slice thicknesses of all outlined areas. The same methodology was employed to measure the FIV on CT of the brain at >48 h.

Statistical analysis

Data are expressed as mean (SD) unless otherwise noted and analysis was by intention-to-treat. A univariate analysis was performed to compare the GA and DEX groups across demographic, hemodynamic, angiographic and outcome variables. A Student t test was used for continuous variables, Mann–Whitney U test for ordinal data and a χ2 test for categorical variables. Similarly, a univariate analysis was performed for predictors of good neurological outcome to assess whether the sedation type may have an impact on the outcome. Multivariable binary logistic regression models were constructed using variables with a p value<0.10 to determine the independent variables associated with a good neurological outcome.

Results

A total of 216 consecutive patients were analyzed (mean±SD age 65±14.3 years, median (IQR) NIHSS score 19 (15–23)). Fifteen (11%) of the 133 patients in the GA group were intubated prior to arrival in the neurointerventional suite, either in the emergency department or at outside hospitals. Table 1 summarizes the baseline characteristics of patients treated with GA compared with those treated with DEX. Of note, patients treated with DEX were older and presented with lower NIHSS scores. ASPECT scores were available for 186 patients and were, on average, higher in patients treated with DEX. The DEX group had a significantly lower probability of developing hypotension, particularly during initiation of sedation, but had a higher probability of developing bradycardia (although this was seldom hemodynamically significant). Patients treated with DEX had lower rates of pneumonia, less need for PEG or tracheostomy and lower FIV, translating to better neurological outcomes without a higher rate of asymptomatic or symptomatic hemorrhages.

Table 1

Patient characteristics and outcomes by sedation type

Stent retrievers were less commonly used in patients with conscious sedation than in those given GA (19% vs 29%, p=0.125) whereas thromboaspiration was more frequent in the DEX group (37% vs 30%, p=0.268), although the difference was not statistically significant. However, the percentage of patients treated with either stent retrievers or thromboaspiration was equivalent between the two anesthetic strategies (DEX 57% vs GA 58%, p=0.770).

Online supplementary table S1 summarizes the predictors of a good neurological outcome based on univariate analysis. As shown in other studies, successful reperfusion, younger age, lower baseline NIHSS score and lower FIV are predictors of a good neurological outcome. Pre-procedure lower baseline blood pressure, absence of hypotension and use of DEX were also associated with better neurological outcomes. Gender, clot location, hypertension and a history of atrial fibrillation were not independent predictors when evaluated in binary regression models.

When all patients were placed in the regression model the use of DEX was independently associated with better neurological outcomes after controlling for age, NIHSS score, symptomatic hemorrhage and reperfusion status (p=0.027, OR=2.13, 95% CI 1.09 to 4.18). Figure 1A demonstrates the outcome difference between DEX and GA for the subgroup of patients with NIHSS score ≤15. However, when ASPECT scores were included in the model, DEX was not associated with a good neurological outcome relative to GA in the 186 patients for whom ASPECTS data were available (table 2).

Table 2

Binary logistic regression model identifying factors associated with good functional outcome following endovascular therapy for acute ischemic stroke

Figure 1

Rates of good neurological outcome by subgroups. (A) General anesthesia (GA) and dexmedetomidine  (DEX) had similar outcomes for patients with NIH Stroke Scale (NIHSS) score >15. (B) Lower systolic blood pressures at baseline were associated with improved outcomes but, if mean arterial pressures (MAP) fell <70 mm Hg during the case, many patients fared worse.

One challenge noted with DEX in our cohort is that 12 of 83 patients in the DEX group (14%) required conversion to GA. A learning curve among our anesthesia providers may have played a role: seven of the conversions occurred in the first 22 DEX cases but only three of the last 42 DEX patients converted to GA. Although no firm conclusions are possible, conversion was not associated with a worse outcome in this small subgroup (9 of 12 had good outcomes). Among GA patients, the subgroup that arrived intubated did not appear to fare worse than those intubated for the procedure (rates of good outcome were 33% and 30%, respectively).

Hemodynamic parameters were associated with neurological outcome. Figure 1B shows rates of good outcome by baseline systolic blood pressures (SBP) and lowest recorded MAP. Lower baseline SBP appeared to correlate with a good outcome, especially for patients with a starting SBP <140 mm Hg. In contrast, for the lowest recorded MAPs during the procedure, higher values correlated with better outcomes except in the patients who presented with severe systolic hypertension. Examining lowest MAPs by quartile revealed better outcomes for those whose MAP stayed above 70 mm Hg during the procedure. These trends held when stratifying both by type of anesthesia and NIHSS and in a regression model of the full dataset (data not shown). Table 3 shows a second regression model incorporating baseline SBPs and lowest recorded intraprocedure MAPs among the 186 patients with ASPECTS scores. DEX was not independently associated with outcome in this model, which is consistent with the decreased hypotension seen in the DEX group, whereas lowest MAPs ≤70 mm Hg and baseline systolic BP ≥140 mm Hg were both independent predictors of a poor outcome.

Table 3

Binary logistic regression model identifying hemodynamic factors associated with good functional outcome following endovascular therapy for acute stroke

Discussion

DEX is a novel sedative with potential neuroprotective properties and a growing role in neuroanesthesia.12 Early concerns about uncoupling of cerebral blood flow and cerebral metabolism seen in animals have not been observed in small studies of human volunteers13 or those undergoing operations for neurovascular malformations.14 DEX has been used to facilitate ongoing neurological evaluations during uncomfortable procedures such as awake craniotomy7 and deep brain stimulator implantation.15 It has also been used to limit movement and preserve respiration during pediatric MRI.16 DEX represents an attractive sedative for endovascular stroke therapy but has not been evaluated in that setting until now.

The current study demonstrates that DEX is a reasonable alternative to GA when treating patients with endovascular stroke reperfusion therapy. In keeping with prior studies, we saw no increase in hemorrhagic or procedural complications in our sedation cohort. Our good outcome (OR 2.13) in the full dataset is consistent with previous retrospective studies of conscious sedation for endovascular therapy which used the NIHSS score alone for stroke severity. In the largest of these studies, Abou-Chebl and colleagues analyzed 980 patients from 12 centers and found that GA had an OR for poor neurological outcome of 2.33 in a regression model including age, NIHSS, Thrombolysis In Myocardial Infarction recanalization and hemorrhage.2 A subset of 126 of these patients treated at the University of Pittsburgh for M1 occlusions were analyzed separately by Jumaa and colleagues.3 In their multivariate model, which controlled for age, NIHSS and recanalization, conscious sedation had an OR for good neurological outcome of 3.06. A third study of 96 patients by Davis et al1 found that conscious sedation had a RR of 3.2 for good neurological outcome after controlling for NIHSS and blood glucose concentrations.

None of these previous models incorporated ASPECTS scores as an imaging measure of stroke severity. In the dataset of 126 patients studied by Jumaa et al, ASPECTS as an ordinal variable had a p value of 0.082 in the univariate analysis whereas in our population the binary variable of ASPECTS >7 had a p<0.001. When we added ASPECTS to our regression model, the OR for DEX dropped to 1.48 and was no longer statistically significant (table 2). Although it is possible that our sample size of 186 was too small to detect a true difference between DEX and GA, our results suggest that using NIHSS scores alone in multivariate models may not capture some important aspects of stroke severity. Neither our patients nor those in previous studies were randomized to GA or sedation and, in the cohorts reported to date, patients given GA had higher NIHSS scores. Future studies should include ASPECT scores and, until randomized trials of sedation type are performed, differences in baseline severity will remain a potential confounder.

Although previous studies may have overstated the adverse effects of GA, hypotension during GA could compromise collateral flow to the ischemic penumbra and affect neurological recovery. Davis and colleagues examined hypotension as a potential contributor to poor neurological outcome and one possible mechanism for the worse outcomes they observed in GA patients. They constructed a second regression model substituting lowest SBP <140 mm Hg for anesthesia type and found that the RR of good outcome was 0.59.1 We too observed a correlation between the lowest recorded blood pressure (MAP ≤70 mm Hg) and outcome (figure 1B, table 3).

In contrast to this correlation between lowest recorded blood pressures and poor outcome, our patients whose SBP was <140 mm Hg at baseline appeared to have better outcomes (figure 1B). These somewhat contradictory results probably reflect complex interactions between blood pressure and outcome and highlight the difficulty in studying hemodynamics in predictive models. Previous studies have reported a U-shaped relationship between baseline SBP and patient outcomes following acute strokes, in which higher blood pressures were associated with higher incidences of death from cerebral edema whereas lower blood pressures correlated with adverse events relating to coronary artery disease.17 ,18 A prospective trial has also shown that SBP reduction within 36 h of symptom onset improves post-stroke outcomes at 90 days in a population of patients who do not receive thrombolytic agents.19 Whether such observations and management can be similarly applied to the acute intraprocedural setting among patients treated with endovascular therapies remains to be seen. A prospective trial of tight blood pressure control would be necessary to discern whether active blood pressure support with vasopressors can improve outcomes. In the face of this uncertainty, the smooth hemodynamic profile of DEX appears to simplify patient management during the case and represents an advantage over GA.

cIn addition to the retrospective nature of our study, another limitation of this work is the absence of a non-DEX sedation group. Although our OR for DEX versus GA fits with previous studies using NIHSS but not ASPECTS, it is not possible to compare DEX with traditional sedatives based on this study. One measure of the success of a sedation protocol is the rate of conversion to GA. The data collection used in the study by Abou-Chebl et al precluded measurement of this rate, but Jumaa and colleagues reported that 2 of 73 patients (2.7%) converted to GA.2 ,3 We too analyzed by intention-to-treat and had a conversion rate of 14% overall, but this fell to 7% as we gained experience with DEX in this population. It may be that we continue to dose too conservatively, since a study of pediatric MRI found it necessary to increase the loading dose to 3 μg/kg and the maintenance dose to 2 μg/kg/h in order to successfully complete scans without supplemental opioids or benzodiazepines.16 Nevertheless, given the comorbidities of our stroke patients, we prefer to keep our dosing within the drug manufacturer's guidelines and supplement with midazolam or fentanyl as needed. No matter what sedation protocol is used, it is important to have an anesthesia provider present in case conversion is required or the patient's respiratory status deteriorates.

In conclusion, DEX may be a viable alternative to GA during endovascular stroke reperfusion procedures. We found no adverse effects on neurological outcomes or bleeding, and our univariate analysis showed lower rates of complications such as pneumonia. There is a learning curve that may exist to gain comfort with DEX during such intra-arterial treatment, but hemodynamic variability requiring vasopressor support is less frequent than with GA. Further prospective studies are required to determine if outcomes can be improved based on sedation practices.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors MKW, RG and C-HJS wrote the manuscript. MKW, RG, RGN and RYG conceived the research. MKW, RG and C-HJS performed the statistical analysis. RGN and RYG revised the manuscript. SL, KW, MKW, C-HJS and BAG collected the data. RG and MW had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

  • Competing interests RGN serves on the scientific advisory board for Stryker Neurovascular, Covidien and CoAxia. He also serves on the Data Safety Monitoring Board for Rapid Medical and Imaging Core Lab for Covidien and Reverse Medical. He is an Editor of Interventional Neurology. RG serves on the scientific advisory board for Stryker Neurovascular, Covidien and CoAxia. He is also a member of the Data Safety Monitoring Board for Reverse Medical and Rapid Medical and is an Associate Editor for the Journal of Neuroimaging. All other authors have no financial disclosures to report.

  • Ethics approval Ethical approval was granted by the Institutional Review Board at Emory University prior to the start of the study.

  • Data sharing statement All inquiries pertaining to the access of data presented in the study may be addressed directly to RG.