SEARCH

SEARCH BY CITATION

Keywords:

  • acute stroke therapy;
  • cerebral infarction;
  • hypertension;
  • intervention;
  • ischaemic stroke;
  • risk factors

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References

Background

We recently developed the Totaled Health Risks In Vascular Events (THRIVE) score to predict outcomes after endovascular stroke treatment. The THRIVE score, which incorporates age, National Institutes of Health Stroke Scale score, and three medical comorbidities (hypertension, diabetes mellitus, and atrial fibrillation), was developed using data from the Mechanical Embolus Removal in Cerebral Ischemia (MERCI) and Multi MERCI trials.

Aims

We set out to perform external validation of the THRIVE score using data from the largest registry of endovascular stroke treatment performed to date, the Merci Registry.

Methods

We compared the performance of the THRIVE score in two different data sets: the development cohort (the MERCI and Multi MERCI trials, n = 305) and a validation cohort (the Merci Registry, a prospective multicenter registry of patients undergoing endovascular stroke treatment, n = 1000). We examined the predictive utility of the THRIVE score across the range of clinical outcomes and used receiver–operator characteristics curve analysis to compare score performance in the two data sets.

Results

The THRIVE score predicted good outcome, death, and the full range of the modified Rankin Scale in a similar fashion between the MERCI trials and the Merci Registry. Receiver–operator characteristics curve comparisons showed no statistically significant difference in the performance of the THRIVE score between the two data sets: for good outcome, the receiver–operator characteristics area under the curve was 0·293 for the MERCI trials and 0·266 for the Merci Registry (P = 0·47) and for death, the receiver–operator characteristics area under the curve was 0·692 for the MERCI trials and 0·717 for the Merci Registry (P = 0·48). The THRIVE score and vessel recanalization were also found to be independent and unrelated predictors of clinical outcome.

Conclusions

The THRIVE score reliably predicts outcomes after endovascular stroke treatment and may be useful as a clinical prognostic tool and to perform severity adjustments in stroke clinical research.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References

Endovascular stroke treatment has been associated with improved neurological outcomes in a randomized controlled trial of intra-arterial prourokinase [1] and several single-arm studies of mechanical clot retrieval [2-5]. However, the overall outcomes of ischemic stroke patients receiving endovascular therapy remain poor, with 24% dying in the hospital and 51% being discharged to long-term care [6]. In addition, endovascular stroke treatment carries significant risks, and up to 10% of patients undergoing this emergency procedure may suffer symptomatic intracranial hemorrhage (ICH) [7]. Therefore, careful patient selection is important to maximize the odds of good outcomes while minimizing the risks of iatrogenic harm and inappropriate use of invasive therapies that may be of limited benefit in particular patients.

Published studies of factors associated with outcomes after endovascular therapy for stroke have shown that recovery correlates with age, baseline stroke severity, thrombus burden, and successful vessel recanalization [8, 9]. In addition, medical comorbidities are known to impact stroke outcomes [10]. While procedural factors such as recanalization and symptomatic hemorrhage are associated with clinical outcomes [1-3, 11], these factors cannot be predicted at the time of presentation with stroke. A clinical prediction tool that incorporates factors that are readily available at the time of presentation, such as age, stroke severity, and medical comorbidities, would be useful in caring for patients with acute ischemic stroke in the emergency setting.

Aims

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References

We have recently developed a clinical prediction tool, the Totaled Health Risks In Vascular Events (THRIVE) score, which predicts the long-term outcomes of patients undergoing endovascular stroke treatment and confirms that chronic medical comorbidities are an important determinant of outcome [12]. To establish the external validity of the THRIVE score, we now compare the performance of the THRIVE score in two independent data sets: the MERCI trials (the development cohort) and the Merci Registry (the validation cohort – the largest prospective clinical series to date of patients undergoing emergency endovascular treatment of acute ischemic stroke) [11].

Predictive scores such as the ABCD2 score [13] for risk stratification of Transient Ischemic Attack (TIA) patients in the emergency department (ED) have improved the ability of physicians involved in acute stroke care to optimize safe patient disposition. With this experience as a model, our goal is to establish the THRIVE score as a reliable and easy to use outcome prediction score for potential endovascular stroke treatment patients to aid in decision making regarding treatment options.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References

Data sources

Data included in the present analyses came from two clinical trials of the Merci retriever (Concentric Medical, Mountain View, CA), the MERCI trial [2] and the Multi MERCI trial [3] (with a total of 305 patients in the two trials) and a large registry of 1000 patients treated with the Merci retriever, the Merci Registry [11]. The trial design for MERCI and Multi MERCI was similar and has been described in detail elsewhere [2, 3]. The Merci Registry is a phase IV, postmarket registry of the use of the Merci Retriever system [14], and recanalization rates, safety data, and outcomes from the complete cohort have been presented [11].

Participants and interventions

MERCI and Multi MERCI were single-arm multicenter trials, and patient selection criteria have been described elsewhere [2, 3]. The major difference between MERCI and Multi MERCI was that MERCI did not allow inclusion of patients who had been treated with intravenous tissue plasminogen activator (tPA) [2], while Multi MERCI allowed inclusion of intravenous tPA-treated patients [3]. Details of the devices and procedures used in MERCI and Multi MERCI have been previously published [2, 3]. Successful recanalization was defined as achieving a Thrombolysis in Myocardial Infarction grade 2 or 3 flow in all treatable vessels [2, 3]. Data on age, sex, preprocedure National Institutes of Health Stroke Scale (NIHSS) score, and medical comorbidities were recorded. Primary outcomes in the original MERCI and Multi MERCI trials were vessel recanalization and device-related complications. Secondary outcomes in the original MERCI and Multi MERCI trials included the modified Rankin Scale (mRS) and NIHSS scores at 30 and 90 days. The prespecified definition of good neurologic outcome was mRS 0–2. A total of 305 patients were enrolled in MERCI and Multi MERCI combined. Of these, 285 patients in MERCI and Multi MERCI had complete data needed to determine both THRIVE score and 90 days mRS.

The Merci Registry was a postmarket clinical registry of the Merci family of stroke thrombectomy devices and involved 1000 patients across 38 centers [11, 15]. As the Merci Registry was not a clinical trial, there were no formal inclusion/exclusion criteria. As was done for the MERCI trials, data on age, sex, preprocedure NIHSS score, and medical comorbidities were recorded in the Merci Registry. The primary and secondary clinical outcomes from MERCI and Multi MERCI, including vessel recanalization, device-related complications, and mRS at 90 days, were also recorded in the Merci Registry. In analysis of the Merci Registry, successful recanalization was defined as a Thrombolysis in Cerebral Infarction grade of 2a or higher [11].

Measurements

The development of the THRIVE score has been described in detail elsewhere [12]. In brief, multivariable modeling was used to determine the relative impact of NIHSS score, age, and chronic medical conditions on outcomes after endovascular stroke treatment [12]. Three chronic medical conditions (hypertension, diabetes mellitus, and atrial fibrillation) were found to independently predict outcome to an approximately equal extent, and these three conditions were therefore combined in a three-point Chronic Disease Scale (CDS) subcomponent of the overall THRIVE score. The resultant score ranges from zero to nine points, with possible contributions of zero to four points from NIHSS score, zero to two points for age, and zero to three points for the CDS (Table 1). In the THRIVE score, cut points used for the NIHSS were chosen to produce reasonable categories of mild (≤10), moderate (11–20), and severe (≥21) strokes, with rounding of the cut point to the nearest decade of NIHSS to facilitate easy recall, and cut points for age (≤59, 60–79, and ≥80 years) were chosen to approximate the tertiles of age in the MERCI + Multi MERCI combined data set, with rounding to the nearest decade of age to facilitate easy recall [12].

Table 1. The Totaled Health Risks In Vascular Events (THRIVE) score
 Points
  1. An online calculator for the THRIVE score is available at http://www.thrivescore.org.

  2. HTN, hypertension; DM, diabetes mellitus; AFib, atrial fibrillation.

National Institutes of Health Stroke Scale
≤100
11–202
≥204
Age 
≤590
60–791
≥802
Chronic Disease Scale (one point each for HTN, DM, AFib)
00
11
22
33
THRIVE score = ____ (0–9)

Data collection

All data in the MERCI trial, Multi MERCI trial, and Merci Registry were collected and stored centrally (Concentric Medical, Mountain View, CA) as previously described [2, 3, 11]. For the present study, relevant data fields for all subjects in the three studies were conveyed to the investigators in de-identified form for analysis. The analyses in the present study were conceived and performed by the investigators.

Data analysis

Categorical data in contingency tables were analyzed by Fisher's exact test. For contingency tables with an ordinal predictor variable and a dichotomous outcome variable, the Mantel–Haenszel chi-squared test for trend of odds was used. Multivariable regression was performed using logistic regression for dichotomous outcomes (e.g. dichotomized mRS 0–2 vs. 3–6) and ordinal logistic regression for ordinal outcomes (e.g. the ordinal range of the mRS from 0 to 6). All multivariable models in the present analysis used simultaneous variable entry and used an alpha of 0·05. For ordinal logistic regression, all models satisfied the proportional odds/parallel lines assumption, tested at an alpha of 0·05.

Comparative receiver–operator characteristics (ROC) curve analysis is a robust way to compare the performance of an ordinal scoring system across two separate data sets. Specifically, ROC curve comparison affords a direct statistical test of whether there is a difference in the range of sensitivities and specificities at varying cut points across the range of scores when the scoring system is used on two different data sets [16]. ROC curves were constructed to examine the relationship between THRIVE score and clinical outcome and compare this relationship in two different data sets. The area under the curve (AUC) for ROC curves was compared with a chi-square test with an alpha of 0·05. All statistical analyses were performed using Stata SE, version 10·0 (StataCorp, College Station, TX).

The MERCI and Multi MERCI trials received institutional review board approvals from the institutions involved in the conduct of these clinical trials. The Merci Registry similarly received institutional review board approvals from the institutions involved in the registry. The analyses presented here were performed using a de-identified data set – no protected health information was transmitted from Concentric Medical to the authors. The authors are solely responsible for the design and conduct of the present analyses.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References

Patient characteristics

Patients in the MERCI trials and the Merci Registry had similar characteristics (Table 2). There were no significant differences in age, sex, hypertension, diabetes mellitus, atrial fibrillation, or site of vessel occlusion. While there was a significant but slight difference in baseline NIHSS, the mean baseline THRIVE score was not different (Table 2). Rates of symptomatic ICH and clinical outcomes at 90 days were not different between the two datasets (Table 2).

Table 2. Patient characteristics – the MERCI trials and the Merci Registry
 MERCI trials (%)Merci registry (%) P
  1. P-values for comparison of continuous or ordinal measures are from Student's t-test and P-values for comparison of categorical measures are from Fisher's exact test.

  2. MCA, middle cerebral artery; ICA, internal cerebral artery; VB, vertebrobasilar; ICH, intracerebral hemorrhage; mRS, modified Rankin Scale; good outcome, mRS of 0 to 2; THRIVE, totaled health risks in vascular events.

Age (mean)67·667·30·75
Female(52·1)(49·5)0·43
Hypertension(72·0)(76·0)0·17
Diabetes mellitus(20·2)(25·1)0·09
Atrial fibrillation(42·4)(42·9)0·89
National Institutes of Health Stroke Scale (mean)19·617·8<0·001
THRIVE score (mean)4·94·70·04
Time to groin puncture4·4 h6·4 h<0·001
Vessel occluded   
MCA(58·4)(61·8)0·50
ICA(32·5)(30·4)
VB(9·2)(7·8)
Vessel Recanalization(65·6)(80·0)<0·001
Symptomatic ICH(8·6)(7·0)0·36
mRS at 90 days3·73·80·49
Good outcome at 90 days(32·4)(29·5)0·34
Mortality at 90 days(38·1)(36·1)0·54

Relationship between THRIVE score and good outcome or death at 90 days

In our score development cohort (pooled data from the MERCI and Multi MERCI trials), higher THRIVE scores were associated with worsened chances of a good outcome (mRS = 0 to 2) at 90 days poststroke [12]. We therefore sought to validate the relationship between THRIVE and good outcome in an external data set, the Merci Registry. The association between trichotomized THRIVE score (0 to 2 vs. 3 to 5 vs. 6 to 9) and good outcome showed a similar trend for MERCI trials data (Fig. 1a) and Merci Registry data (Fig. 1b) (Mantel–Haenszel chi-square test for trend for each, P < 0·001). Higher THRIVE scores also predicted increased risk of death by 90 days in both the MERCI trials data set (Fig. 1c) and the Merci Registry data set (Fig. 1d) (Mantel–Haenszel chi-square test for trend for each, P < 0·001).

figure

Figure 1. Totaled health risks in vascular events (THRIVE) score prediction of good outcome and death in the MERCI trials and the Merci Registry. (a) Percentage with good outcome at three-months in the MERCI trials at three levels of the trichotomized THRIVE score (0 to 2, 3 to 5, and 6 to 9). (b) Percentage with good outcome at three-months in the Merci Registry at three levels of the trichotomized THRIVE score (0 to 2, 3 to 5, and 6 to 9). (c) Percentage dead at three-months in the MERCI trials at three levels of the trichotomized THRIVE score (0 to 2, 3 to 5, and 6 to 9). (d) Percentage dead at three-months in the Merci Registry at three levels of the trichotomized THRIVE score (0 to 2, 3 to 5, and 6 to 9).

Download figure to PowerPoint

In the original development cohort, the THRIVE score was trichotomized as shown in Fig. 1 to demonstrate the relationship between THRIVE and outcomes because there was an insufficient number of patients to provide a meaningful estimate of the outcome at each level of the THRIVE score (n = 285 with THRIVE score and good outcome assessment; n = 294 with THRIVE score and mortality assessment). When the MERCI trial, Multi MERCI trial, and Merci Registry data sets are combined (n = 1192 with THRIVE score and good outcome assessment; n = 1202 with THRIVE score and mortality assessment), the relationship between the full range of THRIVE and outcome can be seen for both good outcome by 90 days (Fig. 2a) and death by 90 days (Fig. 2b). In addition, individual outcome estimation (without trichotomizing a patient's age and NIHSS) was determined by applying the logistic function, f(z) = 1 / (1 + e–z), to the complete data set, where for good outcome z = [3·957 + (–0·03607 × age) + (–0·12867 × NIHSS) + (–0·12843 × CDS)] and for death z = [–5·5107 + (0·0448 × age) + (0·08633 × NIHSS) + (0·14437 × CDS)] (calculator available online at http://www.thrivescore.org).

figure

Figure 2. Prediction of outcomes across the full range of the totaled health risks in vascular events (THRIVE) score. (a) Increasing THRIVE score predicts a worsened chance of good outcome at three-months poststroke in the combined (MERCI trials + Merci Registry) data set. Open circles show percentage with good outcome at each level of the THRIVE score. Solid curve represents a second-order polynomial fit to the data and dotted lines are 95% confidence intervals for the best-fit estimation (goodness of fit R 2 = 0·96). (b) Increasing THRIVE score predicts an increased chance of death by three-months in the combined data set. Filled circles show percentage dead at each level of the THRIVE score. Solid curve represents a second-order polynomial fit to the data and dotted lines are 95% confidence intervals for the best-fit estimation (goodness of fit R 2 = 0·94).

Download figure to PowerPoint

THRIVE score and full range of the mRS at 90 days

In the MERCI trials data sets, the THRIVE score predicts outcomes after endovascular stroke treatment across the full range of the mRS, with higher THRIVE scores associated with a shift toward worsened outcomes on the mRS [12]. We thus compared the distribution of mRS from 0 to 6 according to trichotomized THRIVE score in the two data sets. The overall shift toward worsened mRS scores is seen with higher THRIVE scores in both the MERCI trials data set (Fig. 3a) and the Merci Registry data set (Fig. 3b). In ordinal logistic regression modeling the full range of the mRS (0 to 6), the trichotomized THRIVE score similarly predicted a worsened mRS in the MERCI trials [odds ratio (OR) = 3·51, 95% confidence interval (CI) 2·51 to 4·91, P < 0·001] and the Merci Registry (OR = 3·69, 95% CI 3·05 to 4·45, P < 0·001). Using the full range of the THRIVE score (0 to 9) as a predictor of mRS (0 to 6), similar results were obtained for the MERCI trials (OR = 1·47, 95% CI 1·31 to 1·65, P < 0·001) and the Merci Registry (OR = 1·61, 95% CI 1·51 to 1·72, P < 0·001).

figure

Figure 3. Totaled health risks in vascular events (THRIVE) score prediction of the full range of the modified Rankin Scale (mRS) at three-months. (a) Graphical representation of the full range of mRS outcomes (0 to 6) at each level of the trichotomized THRIVE score in the MERCI trials. Increasing shades of gray tone from white to black indicate increasing levels of the mRS. (b) Graphical representation of the full range of mRS outcomes (0 to 6) at each level of the trichotomized THRIVE score in the Merci Registry.

Download figure to PowerPoint

Similarly, each component of the THRIVE score [trichotomized age, trichotomized NIHSS, and CDS] independently predicted the full range of the mRS in ordinal logistic regression (Table 3).

Table 3. Components of the THRIVE score: ordinal logistic regression models
PredictorMERCI trialsMerci registry
Odds ratio for increase in mRS95% CI P Odds ratio for increase in mRS95% CI P
  1. Trichotomized NIHSS, NIHSS as included in the THRIVE score (10 or lower vs. 11–20 vs. 21 or higher); trichotomized age, age as included in THRIVE score (59 and younger vs. 60–79 vs. 80 and older); CDS, Chronic Disease Scale (one point each of history of hypertension, diabetes, and atrial fibrillation); NIHSS, National Institutes of Health Stroke Scale; THRIVE, totaled health risks in vascular events; mRS, modified Rankin Scale; CI, confidence interval.

Trichotomized NIHSS1·611·16–2·230·0042·311·93–2·78<0·001
Trichotomized age1·541·27–1·89<0·0011·631·48–1·79<0·001
CDS1·291·00–1·680·0471·221·06–1·410·006

ROC curve comparison

ROC curve analysis showed that the THRIVE score predicted good outcome over a comparable range of sensitivity and specificity for both the MERCI trial data set (n = 285) and the Merci Registry data set (n = 907) (Fig. 4a). Because THRIVE score is inversely related to good outcome, the ROC curves lie below the line of unity for sensitivity = (1 – specificity) (Fig. 4a), and therefore, the AUC values must lie between 0·5 (indicating no predictive value) and 0 (indicating perfect predictive value). The AUC for the ROC curves for good outcome was not significantly different between MERCI trial patients (AUC = 0·293, 95% CI = 0·231 to 0·355) and Merci Registry patients (AUC = 0·266, 95% CI = 0·230 to 0·302) (P = 0·47).

figure

Figure 4. Receiver–operator characteristics (ROC) curve comparison of totaled health risks in vascular events (THRIVE) score outcome prediction in the MERCI trials and the Merci Registry. (a) ROC curves for the relationship between THRIVE score and good outcome in the MERCI trials (filled circles) and the Merci Registry (open circles). Because THRIVE score is inversely related to good outcome, the ROC curves lie below the line of unity for sensitivity = (1 – specificity) (solid line). The area under the curve (AUC) for good outcome was not different between MERCI trial patients (AUC = 0·293) and Merci Registry patients (AUC = 0·266) (P = 0·47). (b) ROC curves for the relationship between THRIVE score and death in the MERCI trials (filled circles) and the Merci Registry (open circles). Because THRIVE score is directly related to risk of death, the ROC curves lie above the line of unity for sensitivity = (1 – specificity) (solid line). The area under the curve (AUC) for death was not different between MERCI trial patients (AUC = 0·692) and Merci Registry patients (AUC = 0·717) (P = 0·48).

Download figure to PowerPoint

ROC curve analysis also showed that the THRIVE score predicted death in a similar fashion for the MERCI trial data set (n = 294) and the Merci Registry data set (n = 908) patients (Fig. 4b). In this case, the THRIVE score is directly related to risk of death so the ROC curves lie above the line of unity for sensitivity = (1 – specificity) (Fig. 4b), and therefore, the AUC values must lie between 0·5 (indicating no predictive value) and 1 (indicating perfect predictive value). The AUC for the ROC curves for death was not significantly different between MERCI trial patients (AUC = 0·692, 95% CI = 0·632 to 0·752) and Merci Registry patients (AUC = 0·717, 95% CI = 0·683 to 0·750) (P = 0·48).

THRIVE score and recanalization

The THRIVE score is based on patient factors available at the time of initial assessment, prior to the endovascular stroke procedure. We found in the development cohort that the THRIVE score predicted outcome independent of whether vessel recanalization was achieved during the procedure [12]. Similar independence of THRIVE score and vessel recanalization in predicting outcomes were seen using Merci Registry data. In ordinal logistic regression of THRIVE score modeling the full range of the mRS, odds ratio estimates and 95% CIs were not substantially altered by the addition of vessel recanalization status to the models (Table 4). In logistic regression, there was no relationship between THRIVE score and the chances of successful recanalization (OR = 0·97 for combined data set, 95% CI 0·90–1·03, P = 0·33; similar nonsignificant results for each data set analyzed separately).

Table 4. Independence of THRIVE score and vessel recanalization – ordinal logistic regression models
ModelMERCI trialsMerci registry
Odds ratio for increase in mRS95% CIOdds ratio for increase in mRS95% CI
  1. THRIVE is the model in which an increase in the THRIVE score predicts an increase in the modified Rankin Scale at 90 days (mRS). THRIVE* is the model in which an increase in THRIVE score predicts an increase in mRS, after addition of vessel recanalization to the model. Vessel recanalization was added to the THRIVE* models as a dichotomous variable representing final vessel recanalization status at end of procedure (for MERCI trials, TIMI grade II or II flow, for Merci Registry, TICI grade of 2a or higher). All odds ratios presented are for a one level increase in the mRS. All predictors in all models were significant at a P-value of <0·001.

  2. mRS, modified Rankin Scale; CI, confidence interval; TIMI, Thrombolysis in Myocardial Infarction; TICI, Thrombolysis in Cerebral Infarction; THRIVE, totaled health risks in vascular events.

THRIVE1·471·30–1·651·611·51–1·72
THRIVE*1·521·35–1·721·641·53–1·76

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References

In a large prospective registry of patients undergoing mechanical thrombectomy after ischemic stroke, we have validated the THRIVE score as a simple clinical rule that predicts long-term functional recovery. Importantly, the THRIVE score reliably predicts outcomes regardless of the success of vessel recanalization. The availability of a simple prediction rule based on clinical factors that are available at the time of stroke presentation will aid clinicians, patients, and patients' surrogates when balancing the invasiveness and risk of endovascular therapy with its potential to improve outcomes if the occluded vessel can be recanalized.

Our study design benefits from the use of data from two clinical trials and prospectively collected data from a multicenter registry. In addition, the THRIVE score incorporates simple clinical variables that have already been shown in prior studies to correlate with recovery from stroke [8, 10], which lends credibility to its performance in a patient population separate from the original derivation cohort.

Our study has limitations. First, this analysis was not prespecified before data collection in the MERCI registry. However, the definitions of variables used in the THRIVE score were standardized and should not be susceptible to significant bias. Second, our study design may be susceptible to bias if patients with comorbidities underwent endovascular therapy only if they had less severe strokes compared with those without comorbidities, in other words, if endovascular therapy was likely to be deferred in patients with severe strokes and multiple comorbidities. However, this would be expected to reduce the impact of comorbidities on long-term outcomes in this cohort of patients undergoing endovascular therapy and would therefore buttress our findings. Third, we did not have data on radiographic characteristics and certain clinical variables such as dementia. The lack of imaging data in the form of advanced MRI-based (Magnetic Resonance Imaging) neuroimaging that may also help in predicting clinical outcomes after stroke [17] is a limitation. Despite this, the simplicity and reliability of the variables included in this score ensure that clinicians caring for potential endovascular stroke treatment candidates will have all the required data at hand to reliably and quickly estimate prognosis for recovery shortly after a patient's presentation with stroke, thereby aiding timely clinical decision making in the acute stages of a stroke. While the THRIVE score can be easily determined in any ED, advanced neuroimaging techniques such as MRI perfusion may not be available in all centers. Lastly, our analysis is limited to a single form of endovascular stroke treatment, mechanical thrombectomy with the MERCI Retriever (with or without adjuvant thrombolytic administration), and therefore cannot necessarily be generalized to other types of endovascular therapy.

We found that THRIVE score and vessel recanalization were independent predictors of outcome, and we also found that THRIVE score did not predict the chances of successful recanalization. The THRIVE score can be thus be thought of as a set of nonmodifiable predictors of clinical outcome among stroke patients, while vessel recanalization is a modifiable predictor of outcome that is distinct from the nonmodifiable factors of age, stroke severity, and comorbidities represented by the THRIVE score. Given the independence of the THRIVE score from vessel recanalization in predicting clinical outcomes, the THRIVE score may perform well in a similar patient population undergoing endovascular stroke treatment with a different type of device (such as the Penumbra system or the Trevo system) and might predict outcomes in other stroke patients (such as those receiving tPA and those not eligible for tPA), but further confirmation with separate data sets will be required to explore these possibilities.

Our results show that the THRIVE score is a reliable tool for predicting the likelihood of neurological recovery after undergoing endovascular stroke treatment. The THRIVE score may help physicians and families with decisions about whether to proceed with an invasive and potentially risky therapy that simultaneously offers hope of significant benefit. Given the wide variation in patient and surrogate preferences regarding aggressive care for neurological emergencies and the fact that these decisions depend crucially on likely long-term functional outcomes [18], a simple and reliable tool such as the THRIVE score promises to help with time-sensitive and high-stakes decisions regarding acute stroke care.

Author contributions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References

A. C. F., V. A. R., and S. P. C. conceived of the study. A. C. F., H. K., B. S. F., and W. S. S. contributed to study design. A. C. F., H. K., and B. S. F. analyzed the data. A. C. F. and H. K. drafted the manuscript, and all authors contributed substantially to its revision. W. S. S. was a principal investigator for the MERCI trial, Multi MERCI trial, and the Merci Registry. A. C. F. takes responsibility for the paper as a whole.

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References

W. S. S. is on the Scientific Advisory Board for Concentric Medical Inc. W. S. S. is an employee of the University of California, which holds the patent on the Merci retriever. W. S. S. has stock ownership in Concentric Medical. W. S. S. was the principal investigator in the MERCI and Multi MERCI trials. W. S. S. was a principal investigator for the Merci Registry.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Aims
  5. Methods
  6. Results
  7. Discussion
  8. Author contributions
  9. Disclosures
  10. References
  • 1
    Furlan A, Higashida R, Wechsler L et al. Intra-arterial prourokinase for acute ischemic stroke. The PROACT II study: a randomized controlled trial. Prolyse in Acute Cerebral Thromboembolism. JAMA 1999; 282:20032011.
  • 2
    Smith WS, Sung G, Starkman S et al. Safety and efficacy of mechanical embolectomy in acute ischemic stroke: results of the MERCI trial. Stroke 2005; 36:14321438.
  • 3
    Smith WS, Sung G, Saver J et al. Mechanical thrombectomy for acute ischemic stroke: final results of the Multi MERCI trial. Stroke 2008; 39:12051212.
  • 4
    Flint AC, Duckwiler GR, Budzik RF, Liebeskind DS, Smith WS. Mechanical thrombectomy of intracranial internal carotid occlusion: pooled results of the MERCI and Multi MERCI Part I trials. Stroke 2007; 38:12741280.
  • 5
    The Penumba Pivotal Stroke Trial Investigators. The penumbra pivotal stroke trial: safety and effectiveness of a new generation of mechanical devices for clot removal in intracranial large vessel occlusive disease. Stroke 2009; 40:27612768.
  • 6
    Brinjikji W, Rabinstein AA, Kallmes DF, Cloft HJ. Patient outcomes with endovascular embolectomy therapy for acute ischemic stroke: a study of the national inpatient sample: 2006 to 2008. Stroke 2011; 42:16481652.
  • 7
    Baker WL, Colby JA, Tongbram V et al. Neurothrombectomy devices for the treatment of acute ischemic stroke: state of the evidence. Ann Intern Med 2011; 154:243252.
  • 8
    Nogueira RG, Liebeskind DS, Sung G, Duckwiler G, Smith WS. Predictors of good clinical outcomes, mortality, and successful revascularization in patients with acute ischemic stroke undergoing thrombectomy: pooled analysis of the Mechanical Embolus Removal in Cerebral Ischemia (MERCI) and Multi MERCI trials. Stroke 2009; 40:37773783.
  • 9
    Barreto AD, Albright KC, Hallevi H et al. Thrombus burden is associated with clinical outcome after intra-arterial therapy for acute ischemic stroke. Stroke 2008; 39:32313235.
  • 10
    Tessier A, Finch L, Daskalopoulou SS, Mayo NE. Validation of the Charlson Comorbidity Index for predicting functional outcome of stroke. Arch Phys Med Rehabil 2008; 89:12761283.
  • 11
    Rymer MM, Budzik RF, Devlin TG, Lutsep HL, Smith WS. Interventional acute stroke therapy with the Merci Retriever embolectomy device: results from 1000 patients in an open label prospective multicenter registry. In Late-Breaking Abstracts. Los Angeles, California, 2011. Available at http://my.americanheart.org/professional/Sessions/InternationalStrokeConference/ScienceNews/Late-Breaking-Abstracts_UCM_426023_Article.jsp (accessed 25 August 2011).
  • 12
    Flint AC, Cullen SP, Faigeles BS, Rao VA. Predicting long-term outcome after endovascular stroke treatment: the totaled health risks in vascular events score. AJNR Am J Neuroradiol 2010; 31:11921196.
  • 13
    Johnston SC, Rothwell PM, Nguyen-Huynh MN et al. Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack. Lancet 2007; 369:283292.
  • 14
    Merci Registry – real world use of the merci retrieval system in acute ischemic stroke. ClinicalTrials.gov Identifier NCT00478478. Available at http://clinicaltrials.gov/ct2/show/NCT00478478?term=NCT00478478&rank=1 (accessed 25 August 2011).
  • 15
    Merci Registry. Available at http://www.strokecenter.org/trials/TrialDetail.aspx?tid=758 (accessed 25 August 2011).
  • 16
    Clarke JL, Johnston SC, Farrant M, Bernstein R, Tong D, Hemphill JC 3rd. External validation of the ICH score. Neurocrit Care 2004; 1:5360.
  • 17
    Grigoryan M, Tung CE, Albers GW. Role of diffusion and perfusion MRI in selecting patients for reperfusion therapies. Neuroimaging Clin N Am 2011; 21:247257, ix–x.
  • 18
    Kiphuth IC, Köhrmann M, Kuramatsu JB et al. Retrospective agreement and consent to neurocritical care is influenced by functional outcome. Crit Care 2010; 14:R144.