- Email: [email protected]
Validation of the DRAGON Score in a Chinese Population to Predict Functional Outcome of Intravenous Thrombolysis–Treated Stroke Patients
Validation of the DRAGON Score in a Chinese Population to Predict Functional Outcome of Intravenous Thrombolysis–Treated Stroke Patients Xinmiao Zhang, MD, Xiaoling Liao, MD, Chunjuan Wang, MD, Liping Liu, MD, Chunxue Wang, MD, Xingquan Zhao, MD, Yuesong Pan, MD, Yilong Wang, MD, and Yongjun Wang, MD
Background: The DRAGON score predicts functional outcome of ischemic stroke patients treated with intravenous thrombolysis. Our aim was to evaluate its utility in a Chinese stroke population. Methods: Patients with acute ischemic stroke treated with intravenous thrombolysis were prospectively registered in the Thrombolysis Implementation and Monitor of acute ischemic Stroke in China. We excluded patients with basilar artery occlusion and missing data, leaving 970 eligible patients. We calculated the DRAGON score, and the clinical outcome was measured by the modified Rankin Scale at 3 months. Model discrimination was quantified by calculating the C statistic. Calibration was assessed using Pearson correlation coefficient. Results: The C statistic was .73 (.70-.76) for good outcome and .75 (.70-.79) for miserable outcome. Proportions of patients with good outcome were 94%, 83%, 70%, and 0% for 0 to 1, 2, 3, and 8 to 10 score points, respectively. Proportions of patients with miserable outcome were 0%, 3%, 9%, and 50% for 0 to 1, 2, 3, and 8 to 10 points, respectively. There was high correlation between predicted and observed probability of 3-month favorable and miserable outcome in the external validation cohort (Pearson correlation coefficient, .98 and .98, respectively, both P , .0001). Conclusions: The DRAGON score showed good performance to predict functional outcome after tissue-type plasminogen activator treatment in the Chinese population. This study demonstrated the accuracy and usability of the DRAGON score in the Chinese population in daily practice. Key Words: Ischemic stroke—intravenous thrombolysis—prognosis—prediction model. Ó 2015 by National Stroke Association
The National Institute of Neurological Disorders and Stroke demonstrated that tissue-type plasminogen activator (t-PA) improved outcomes after acute ischemic stroke within 3 hours,1 which led to the Food and Drug Administration approval in 1996. Then, the benefit of t-PA treatment between 3 and 4.5 hours after symptom
onset was directly confirmed in the European Cooperative Acute Stroke Study III randomized controlled trial.2 As the only approved effective medical therapy for acute ischemic stroke, an increasing number of hospitals have adopted t-PA treatment as one of the most important conventional therapies to treat acute ischemic stroke patients.
From the Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China. Received October 23, 2014; revision received January 16, 2015; accepted March 18, 2015. The authors did not receive any grant support. Address correspondence to Yongjun Wang, MD, Department of Neurology, Beijing Tiantan Hospital, Capital Medical University,
No. 6 Tiantanxili, Dongcheng District, Beijing 100050, China. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2015.03.046
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2015: pp 1-6
1
X. ZHANG ET AL.
2
However, the t-PA treatment of stroke patients could also bring a risk of serious hemorrhage and complications, which could result in severe disabilities and deaths.3 So it has become essential for neurologists and physicians in emergency departments to have a useful tool to help them make the quick decision whether to administer alteplase (Actilyse, Boehringer Ingelheim, Ingelheim, Germany) and explain to patients and families about the likelihood of a favorable outcome if t-PA is given. And a prediction model could be helpful for early estimation of long-term prognosis and early identification of high-risk patients. The DRAGON score4 (Table 1) was exactly developed to predict 3-month clinical outcome after intravenous thrombolysis of patients with acute ischemic stroke. The DRAGON score derived from the Helsinki cohort consists of parameters that are available soon after patients’ admission. The authors recommended further validation studies to ensure general validity. Several studies have shown differences between regions and races in stroke incidence and outcome.5 Although the DRAGON score has been validated in Western populations,6-8 its performance has not been examined in a large Eastern stroke population. To assess the accuracy and usability of the DRAGON score in a Chinese population in daily practice, we investigated the external validity in a multicenter cohort of Chinese patients with acute ischemic stroke who received t-PA treatment. Table 1. DRAGON score (0-10 points) Category Hyperdense cerebral artery sign or early infarct signs on admission CT head scan None Either of them Both mRS score .1, prestroke No Yes Age ,65 y 65-79 y $80 y Glucose level on admission #8 mmol/L (144 mg/dL) .8 mmol/L (144 mg/dL) Onset to treatment time #90 min .90 min NIHSS score on admission 0-4 5-9 10-15 .15
Points
0 1 2 0 1 0 1 2 0 1 0 1 0 1 2 3
Abbreviations: CT, computed tomography; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale.
Methods Multicenter Validation Cohort Data were abstracted from the ongoing project Thrombolysis Implementation and Monitor of acute ischemic Stroke in China (TIMS-China).9 TIMS-China was a welldesigned, national, prospective stroke registry of thrombolytic therapy using intravenous t-PA in patients with acute ischemic stroke. Sixty-seven major stroke centers nationwide participated in this stroke registry. The steering committee selected hospitals from different regions to ensure the representation. The study protocol was approved by the Ethics Committee of Beijing Tiantan Hospital. Information was collected through face-to-face interview and medical records by trained research coordinators. The registry was regularly monitored independently by the quality monitoring committee of TIMS-China and the Contract Research Organization. All patients or their designated relatives were informed about study participation, and informed written consent was obtained. In this study, the multicenter validation cohort included ischemic stroke patients treated with intravenous alteplase (.9 mg/kg body weight) within 4.5 hours from symptom onset and registered in TIMS-China. The ischemic stroke was diagnosed by a neurologist. Noncontrast computed tomography was performed on arrival for all patients. Patients were followed up through face-toface/phone interview after 3 months. Following the rules of the DRAGON score,4 we excluded patients with basilar artery occlusion, missing baseline parameters, and lack of necessary follow-up data. The main outcome of interest was 3-month functional outcome. Good 3-month outcome was defined as the modified Rankin Scale (mRS) score of 0 to 2 (namely no symptoms, no significant disability, and slight disability), whereas miserable outcome as mRS 5 to 6 (severe disability and dead).
Statistical Analysis The baseline characteristics of the original Helsinki cohort and our TIMS-China cohort were presented. Continuous variables were presented as medians and interquartile ranges (IQR), and categorical variables as absolute values and percentages. Comparisons were performed using the c2 test, and Fisher exact test or Yates correction was used when needed. Patients’ DRAGON score was calculated in a manner identical to that reported in the original article. The sensitivity and specificity was calculated for each individual value of the score using a 2 3 2 table. The performance of the DRAGON score in the validation samples was determined by generating C statistic. A C statistic of 1.0 indicates perfect prediction, whereas a C statistic of .5 indicates no better than random prediction. The curves of receiver operating characteristic were also
VALIDATION OF THE DRAGON SCORE IN CHINA
3
Table 2. Demographics, baseline characteristics, and proportions of the 3-month outcome Characteristics
Derivation cohort (n 5 1319)
TIMS-China cohort (n 5 970)
Age, y, median (IQR) Sex, n (%) Male Onset to treatment time, min, median (IQR) Baseline NIHSS score, points, median (IQR) Hyperdense cerebral artery sign, n (%) Early infarct signs, n (%) Prestroke mRS score, n (%) .1 Glucose, mmol/L, median (IQR) Medical history, n (%) Hypertension Diabetes mellitus Previous stroke Hyperlipidemia Atrial fibrillation Proportions of 3-month outcome mRS score, n (%) 0-2 3-4 5-6
69 (60-77)
65 (56-73)
729 (55.3) 118 (88-158) 9 (5-14) 233 (17.7) 406 (30.6)
586 (60.41) 170 (140-195) 11 (7-16) 72 (7.42) 55 (5.67)
82 (6.2) 6.6 (5.7-7.8)
42 (4.33) 6.8 (5.9-8.4)
787 (59.7) 191(14.5) 168 (12.7) 516 (39.1) 365 (27.7)
571 (58.87) 172 (17.73) 178 (18.35) 67 (6.91) 176 (18.14)
798 (60.5) 339 (25.7) 182 (13.8)
563 (58.04) 273 (28.14) 134 (13.81)
Abbreviations: IQR, interquartile range; mRS, modified Rankin Scale; NIHSS5National Institutes of Health Stroke Scale; TIMS-China, Thrombolysis Implementation and Monitor of acute ischemic Stroke in China.
drawn. Reliability was investigated with calibration curves, which were assessed using Hosmer–Lemeshow goodness-of-fit c2 test. However, as Hosmer–Lemeshow test was known to be oversensitive to small deviations from good fit in large samples, we also compared predicted versus observed rate of the good and the miserable outcome using Pearson correlation coefficient. The absolute error for calibration was calculated by subtracting the predicted probabilities from the observed ones. The mean absolute error less than .4 was defined as acceptable. All analyses were conducted using SAS statistical software (version 9.2, SAS Institute Inc., Cary, NC).
P , .01) and atrial fibrillation (18.14% versus 27.7%, P , .01). Distribution of 3-month outcome of each point for the DRAGON score in the TIMS-China cohort is shown in Figure 1. An extreme DRAGON score of 9 and 10 did not occur; the most frequent score was 4, in 260 (26.8%) patients. Proportions of patients with good outcome were 94%, 83%, 70%, and 0% for 0 to 1, 2, 3, and 8 to 10 score
Results From May 2007 to April 2012, a total of 1440 patients who received intravenous t-PA treatment were registered in the TIMS-China. After excluding the cases thrombolysed outside 0- to 4.5-hour window (n 5 307), with basilar artery occlusion (n 5 54), with missing data (ie, time from symptom onset to thrombolysis, or baseline National Institutes of Health Stroke Scale; n 5 87), and with lack of follow-ups (n 5 22), 970 patients were included in the present study. The baseline characteristics of patients are shown in Table 2. In the TIMS-China cohort, the median age was 65 years (IQR, 56-73), and 60.41% were men. The cohort had a lower rate of hyperlipidemia (6.91% versus 39.1%,
Figure 1. Three-month outcome per increasing points of the DRAGON score in the TIMS-China cohort (n 5 970). Good 3-month outcome (gray bars) indicates modified Rankin scale (mRS) score of 0 to 2, and miserable outcome (black bars) indicates mRS 5 to 6. Abbreviations: TIMS-China, Thrombolysis Implementation and Monitor of acute ischemic Stroke in China.
X. ZHANG ET AL.
4
Table 3. Sensitivity, specificity, and positive likelihood ratio for good and miserable 3-month outcome for each individual DRAGON score Good outcome (mRS score 0-2) DRAGON 0-1 2 3 4 5 6 7 8-10
Miserable outcome (mRS score 5-6)
Sensitivity, %
Specificity, %
Positive LR
Sensitivity, %
Specificity, %
Positive LR
9.1 (7.0-11.8) 21.4 (18.3-25.1) 27.7 (24.2-31.6) 25.2 (21.8-29.0) 10.1 (7.9-12.9) 5.7 (4.1-7.9) .0 (.0-.0) .0 (.0-.0)
99.8 (98.7-100) 93.9 (91.1-95.8) 83.1 (79.1-86.4) 71.0 (66.4-75.2) 70.0 (65.4-74.3) 87.2 (83.6-90.1) 95.6 (93.1-97.2) 99.5 (98.2-99.9)
38.0 (5.3-273.4) 3.5 (2.3-5.3) 1.6 (1.3-2.1) .9 (.7-1.1) .3 (.3-.5) .4 (.3-.7) .1 (.0-.4) —
.0 (.0-.0) 3.7 (1.6-8.4) 11.9 (7.5-18.5) 20.1 (14.2-27.7) 35.8 (28.2-44.2) 20.9 (14.9-28.5) 6.7 (3.6-12.3) .0 (.0-.0)
93.7 (91.8-95.1) 83.1 (80.4-85.5) 75.0 (72.0-77.8) 72.1 (69.0-75.1) 84.3 (81.7-86.6) 93.9 (91.4-94.8) 98.6 (97.5-99.2) 99.9 (99.3-100.0)
— .2 (.1-.5) .5 (.3-.8) .7 (.5-1.0) 2.3 (1.7-3.0) 3.1 (2.1-4.7) 4.7 (2.0-10.9) 6.2 (.4-99.1)
Abbreviations: LR, likelihood ratio; mRS, modified Rankin Scale.
points, respectively. Proportions of patients with miserable outcome were 0%, 3%, 9%, and 50% for 0 to 1, 2, 3, and 8 to 10 points, respectively. The sensitivity, specificity, and the positive likelihood ratios of the prediction for each score are also shown (Table 3). At both extremes of the score in predicting good and miserable outcome, the specificity performed excellent of almost reaching up to 100%. The likelihood ratios were also correlatively high. The external validation results are shown in Figure 2. The C statistics were .73 (95% CI, .70-.76) for good outcome prediction and .75 (95% CI, .70-.79) for miserable outcome prediction. Moreover, the score showed good calibration. The calibration curves are shown in Figure 3. Comparing the diminished calibration for 3-month good outcome (Hosmer–Lemeshow test; P , .05) with 3-month miserable outcome (Hosmer– Lemeshow test; P 5 .259), we also plotted predicted versus observed functional outcome in the validation samples (Fig 3, A for good outcome and Fig 3, B for miser-
able outcome). There was a high correlation between the predicted and observed rates (Pearson correlation coefficient, .98 for 3-month good outcome and .98 for 3-month miserable outcome; both P , .0001). The mean absolute errors were .05 and .05 for the model predicting good and poor outcome, respectively.
Discussion The DRAGON score was recently developed as a simple and quick tool to predict 3-month functional outcome in ischemic stroke patients treated with intravenous thrombolysis. The accuracy and usability of the score, which was confirmed in the present study, emphasized the potential of the DRAGON score as a useful risk instrument that can be applied to clinical daily practice and stroke research. This report represented the first external validation of the DRAGON score in the Chinese population.
Figure 2. Area under the receiver operating characteristic curves (AUC-ROC) of the DRAGON score in the TIMS-China cohort for 3-month good (A) and miserable (B) outcome. The AUC-ROC was .73 (.70-.76) for good outcome and .75 (.70-.79) for miserable outcome. Abbreviations: TIMS-China, Thrombolysis Implementation and Monitor of acute ischemic Stroke in China.
VALIDATION OF THE DRAGON SCORE IN CHINA
5
Figure 3. Predicted and observed probabilities of 3-month good outcome (A) and miserable outcome (B) in the TIMS-China cohort (Pearson correlation coefficient, .98 for 3-month good outcome and .98 for 3-month miserable outcome; both P , .0001). Abbreviations: TIMS-China, Thrombolysis Implementation and Monitor of acute ischemic Stroke in China.
It is critical to notice that the DRAGON score was not only a tool to help clinicians in immediately estimating outcome based on a number of variables when discussing prognosis of intravenous thrombolysis with stroke patients and their families, but also a useful instrument to support the decision-making process when t-PA treatment alone did not offer high likelihood of favorable outcome. With a score of 8, 9, or 10, patients were highly likely to have a miserable functional outcome even if t-PA was given, and possible add-on treatment (ie, endovascular strategy) should be considered. The longer the onset-to-needle time of intravenous thrombolysis, the higher the likelihood of acute ischemic stroke patients to achieve the miserable outcome at 3 months. What we could not ignore is the fact that the time-window of intravenous thrombolysis could be one of the confounding factors in our study. So the onset-toneedle time and the door-to-needle time in each group stratified by the DRAGON score are also shown (Fig 4).We could tell from Figure 4 that each group of different DRAGON score has a similar onset-to-needle time and door-to-needle time (P . .05), so the confounding influence of time-window of treatment could be excluded. In the TIMS-China cohort, the DRAGON score showed a graded effect: the higher the score, the lower the risk of a favorable outcome. When the score reached 8 point, the proportion of good functional outcome came to 0%. An excellent performance of specificity of the score was seen at both the extreme poles, which could decrease the possibility of incorrect prediction of dismal prognosis to the greatest degree, and thus had more significance when considering invasive add-on rescue strategies, which was costly and had a risk of severe complications. The study also demonstrated that the (hyper) Dense cerebral artery sign/early infarct signs on admission CT scan, prestroke modified Rankin Scale (mRS) score,
Age, Glucose level at baseline, Onset-to-treatment time, and baseline National Institutes of Health Stroke Scale score (DRAGON) score had an acceptable external validity for the prediction of 3-month functional outcome after intravenous t-PA for ischemic stroke patients, despite the differences between the original and the validation cohorts. The C statistics for both favorable and miserable outcome were good, similar to that in the validation of the Third International Stroke Trial (IST-3) cohort (.78[.76-.81]).8 Moreover, it is higher than most of the existing prediction models: the iScore (.67),10,11 the Glucose at presentation, Race, Age, Sex, systolic blood Pressure at presentation, Severity of stroke at presentation (GRASPS) score (.71 [.68-.73]),12 the Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS) score (.70),13 the Stroke Prognostication using Age and NIH Stroke Scale (SPAN)-100 score (.64),14 the IST-3 score (.71[.68-.74]),8 and so forth.15 In the meantime, the predicted and observed rates of 3-month outcome in the TIMS-China cohort were in close agreement across the entire spectrum of risk based on the DRAGON score. And the mean absolute errors
Figure 4. The onset-to-needle time and the door-to-needle time in each group stratified by the DRAGON score. Dots represent the median time in each group, and vertical lines represent the interquartile ranges (IQR).
X. ZHANG ET AL.
6
were less than .4 in both outcomes, which has previously been used to define excellent calibration.16 All mentioned above suggested that the DRAGON score was a more powerful instrument to predict clinical outcome of stroke patients treated with t-PA. There has been a rapid proliferation of scores for acute ischemic stroke treated with intravenous thrombolysis,8,11-15,17-20 and the DRAGON score has its obvious superiority. Besides the higher C statistics, the DRAGON score consists of 6 variables which are available shortly after patient admission. So compared with the higher request of obtaining variables and the complex logistic regression equation, the DRAGON score has a greater significance to direct physicians to make the quick and right decision before alteplase administration. Our study has limitations. There could be a selection bias. First, there were only 2 patients with a DRAGON score of 8 and no patients with the score of 9-10 in the TIMS-China cohort. The lack of patients with DRAGON score of 8-10 may be related to the relatively small sample of the TIMS-China cohort because of the comparatively low rate of intravenous thrombolysis in China.21 What’s more, the indication and contraindication of intravenous thrombolysis to acute stroke patients were strictly followed by the hospitals which have the ability to administer alteplase. Thus, many off-labeled patients, such as 80 years or older and National Institutes of Health Stroke Scale score more than 25, which count 1 point separately in the DRAGON score, were excluded to receive the intravenous thrombolysis for the sake of patients’ safety. And this may also lead to the lack of higher DRAGON score in TIMS-China cohort. Second, considering the selection of participating hospitals, although covering all areas of China, was mostly from urban regions. The applicability of the DRAGON score to all hospitals in China with a larger population needs further investigation. In conclusion, the DRAGON score is a valid and easyto-obtain clinical instrument to predict favorable and miserable outcome at 3 months after intravenous thrombolysis. This study demonstrated the accuracy and usability of the DRAGON score in the Chinese population in daily practice.
References 1. Tissue plasminogen activator for acute ischemic stroke. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. N Engl J Med 1995; 333:1581-1587. 2. Hacke W, Kaste M, Bluhmki E, et al. Thrombolysis with alteplase 3 to 4.5 hours after acute ischemic stroke. N Engl J Med 2008;359:1317-1329. 3. Saver JL. Hemorrhage after thrombolytic therapy for stroke: the clinically relevant number needed to harm. Stroke 2007;38:2279-2283. 4. Strbian D, Meretoja A, Ahlhelm FJ, et al. Predicting outcome of IV thrombolysis-treated ischemic stroke
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
patients: the DRAGON score. Neurology 2012;78: 427-432. Stegmayr B, Vinogradova T, Malyutina S, et al. Widening gap of stroke between east and west. Eight-year trends in occurrence and risk factors in Russia and Sweden. Stroke 2000;31:2-8. Strbian D, Seiffge DJ, Breuer L, et al. Validation of the DRAGON score in 12 stroke centers in anterior and posterior circulation. Stroke 2013;44:2718-2721. Seiffge DJ, Karagiannis A, Strbian D, et al. Simple variables predict miserable outcome after intravenous thrombolysis. Eur J Neurol 2014;21:185-191. Whiteley WN, Thompson D, Murray G, et al. Targeting recombinant tissue-type plasminogen activator in acute ischemic stroke based on risk of intracranial hemorrhage or poor functional outcome: an analysis of the third international stroke trial. Stroke 2014;45:1000-1006. Liao XL, Wang CX, Wang YL, et al. Implementation and outcome of thrombolysis with alteplase 3 to 4.5 h after acute stroke in Chinese patients. CNS Neurosci Ther 2013;19:43-47. Saposnik G, Kapral MK, Liu Y, et al. IScore: a risk score to predict death early after hospitalization for an acute ischemic stroke. Circulation 2011;123:739-749. Saposnik G, Demchuk A, Tu JV, et al. The iScore predicts efficacy and risk of bleeding in the National Institute of Neurological Disorders and Stroke Tissue Plasminogen Activator Stroke Trial. J Stroke Cerebrovasc Dis 2013; 22:876-882. Menon BK, Saver JL, Prabhakaran S, et al. Risk score for intracranial hemorrhage in patients with acute ischemic stroke treated with intravenous tissue-type plasminogen activator. Stroke 2012;43:2293-2299. Mazya M, Egido JA, Ford GA, et al. Predicting the risk of symptomatic intracerebral hemorrhage in ischemic stroke treated with intravenous alteplase: safe Implementation of Treatments in Stroke (SITS) symptomatic intracerebral hemorrhage risk score. Stroke 2012;43:1524-1531. Saposnik G, Guzik AK, Reeves M, et al. Stroke prognostication using age and NIH Stroke Scale: SPAN-100. Neurology 2013;80:21-28. Cucchiara B, Tanne D, Levine SR, et al. A risk score to predict intracranial hemorrhage after recombinant tissue plasminogen activator for acute ischemic stroke. J Stroke Cerebrovasc Dis 2008;17:331-333. Johnston KC, Wagner DP, Wang XQ, et al. Validation of an acute ischemic stroke model: does diffusion-weighted imaging lesion volume offer a clinically significant improvement in prediction of outcome? Stroke 2007; 38:1820-1825. Lou M, Safdar A, Mehdiratta M, et al. The HAT Score: a simple grading scale for predicting hemorrhage after thrombolysis. Neurology 2008;71:1417-1423. Kent DM, Selker HP, Ruthazer R, et al. The strokethrombolytic predictive instrument: a predictive instrument for intravenous thrombolysis in acute ischemic stroke. Stroke 2006;37:2957-2962. Strbian D, Engelter S, Michel P, et al. Symptomatic intracranial hemorrhage after stroke thrombolysis: the SEDAN score. Ann Neurol 2012;71:634-641. Flint AC, Faigeles BS, Cullen SP, et al. THRIVE score predicts ischemic stroke outcomes and thrombolytic hemorrhage risk in VISTA. Stroke 2013;44:3365-3369. Wang Y, Liao X, Zhao X, et al. Using recombinant tissue plasminogen activator to treat acute ischemic stroke in China: analysis of the results from the Chinese National Stroke Registry (CNSR). Stroke 2011;42:1658-1664.
Background: The DRAGON score predicts functional outcome of ischemic stroke patients treated with intravenous thrombolysis. Our aim was to evaluate its utility in a Chinese stroke population. Methods: Patients with acute ischemic stroke treated with intravenous thrombolysis were prospectively registered in the Thrombolysis Implementation and Monitor of acute ischemic Stroke in China. We excluded patients with basilar artery occlusion and missing data, leaving 970 eligible patients. We calculated the DRAGON score, and the clinical outcome was measured by the modified Rankin Scale at 3 months. Model discrimination was quantified by calculating the C statistic. Calibration was assessed using Pearson correlation coefficient. Results: The C statistic was .73 (.70-.76) for good outcome and .75 (.70-.79) for miserable outcome. Proportions of patients with good outcome were 94%, 83%, 70%, and 0% for 0 to 1, 2, 3, and 8 to 10 score points, respectively. Proportions of patients with miserable outcome were 0%, 3%, 9%, and 50% for 0 to 1, 2, 3, and 8 to 10 points, respectively. There was high correlation between predicted and observed probability of 3-month favorable and miserable outcome in the external validation cohort (Pearson correlation coefficient, .98 and .98, respectively, both P , .0001). Conclusions: The DRAGON score showed good performance to predict functional outcome after tissue-type plasminogen activator treatment in the Chinese population. This study demonstrated the accuracy and usability of the DRAGON score in the Chinese population in daily practice. Key Words: Ischemic stroke—intravenous thrombolysis—prognosis—prediction model. Ó 2015 by National Stroke Association
The National Institute of Neurological Disorders and Stroke demonstrated that tissue-type plasminogen activator (t-PA) improved outcomes after acute ischemic stroke within 3 hours,1 which led to the Food and Drug Administration approval in 1996. Then, the benefit of t-PA treatment between 3 and 4.5 hours after symptom
onset was directly confirmed in the European Cooperative Acute Stroke Study III randomized controlled trial.2 As the only approved effective medical therapy for acute ischemic stroke, an increasing number of hospitals have adopted t-PA treatment as one of the most important conventional therapies to treat acute ischemic stroke patients.
From the Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China. Received October 23, 2014; revision received January 16, 2015; accepted March 18, 2015. The authors did not receive any grant support. Address correspondence to Yongjun Wang, MD, Department of Neurology, Beijing Tiantan Hospital, Capital Medical University,
No. 6 Tiantanxili, Dongcheng District, Beijing 100050, China. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2015.03.046
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2015: pp 1-6
1
X. ZHANG ET AL.
2
However, the t-PA treatment of stroke patients could also bring a risk of serious hemorrhage and complications, which could result in severe disabilities and deaths.3 So it has become essential for neurologists and physicians in emergency departments to have a useful tool to help them make the quick decision whether to administer alteplase (Actilyse, Boehringer Ingelheim, Ingelheim, Germany) and explain to patients and families about the likelihood of a favorable outcome if t-PA is given. And a prediction model could be helpful for early estimation of long-term prognosis and early identification of high-risk patients. The DRAGON score4 (Table 1) was exactly developed to predict 3-month clinical outcome after intravenous thrombolysis of patients with acute ischemic stroke. The DRAGON score derived from the Helsinki cohort consists of parameters that are available soon after patients’ admission. The authors recommended further validation studies to ensure general validity. Several studies have shown differences between regions and races in stroke incidence and outcome.5 Although the DRAGON score has been validated in Western populations,6-8 its performance has not been examined in a large Eastern stroke population. To assess the accuracy and usability of the DRAGON score in a Chinese population in daily practice, we investigated the external validity in a multicenter cohort of Chinese patients with acute ischemic stroke who received t-PA treatment. Table 1. DRAGON score (0-10 points) Category Hyperdense cerebral artery sign or early infarct signs on admission CT head scan None Either of them Both mRS score .1, prestroke No Yes Age ,65 y 65-79 y $80 y Glucose level on admission #8 mmol/L (144 mg/dL) .8 mmol/L (144 mg/dL) Onset to treatment time #90 min .90 min NIHSS score on admission 0-4 5-9 10-15 .15
Points
0 1 2 0 1 0 1 2 0 1 0 1 0 1 2 3
Abbreviations: CT, computed tomography; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale.
Methods Multicenter Validation Cohort Data were abstracted from the ongoing project Thrombolysis Implementation and Monitor of acute ischemic Stroke in China (TIMS-China).9 TIMS-China was a welldesigned, national, prospective stroke registry of thrombolytic therapy using intravenous t-PA in patients with acute ischemic stroke. Sixty-seven major stroke centers nationwide participated in this stroke registry. The steering committee selected hospitals from different regions to ensure the representation. The study protocol was approved by the Ethics Committee of Beijing Tiantan Hospital. Information was collected through face-to-face interview and medical records by trained research coordinators. The registry was regularly monitored independently by the quality monitoring committee of TIMS-China and the Contract Research Organization. All patients or their designated relatives were informed about study participation, and informed written consent was obtained. In this study, the multicenter validation cohort included ischemic stroke patients treated with intravenous alteplase (.9 mg/kg body weight) within 4.5 hours from symptom onset and registered in TIMS-China. The ischemic stroke was diagnosed by a neurologist. Noncontrast computed tomography was performed on arrival for all patients. Patients were followed up through face-toface/phone interview after 3 months. Following the rules of the DRAGON score,4 we excluded patients with basilar artery occlusion, missing baseline parameters, and lack of necessary follow-up data. The main outcome of interest was 3-month functional outcome. Good 3-month outcome was defined as the modified Rankin Scale (mRS) score of 0 to 2 (namely no symptoms, no significant disability, and slight disability), whereas miserable outcome as mRS 5 to 6 (severe disability and dead).
Statistical Analysis The baseline characteristics of the original Helsinki cohort and our TIMS-China cohort were presented. Continuous variables were presented as medians and interquartile ranges (IQR), and categorical variables as absolute values and percentages. Comparisons were performed using the c2 test, and Fisher exact test or Yates correction was used when needed. Patients’ DRAGON score was calculated in a manner identical to that reported in the original article. The sensitivity and specificity was calculated for each individual value of the score using a 2 3 2 table. The performance of the DRAGON score in the validation samples was determined by generating C statistic. A C statistic of 1.0 indicates perfect prediction, whereas a C statistic of .5 indicates no better than random prediction. The curves of receiver operating characteristic were also
VALIDATION OF THE DRAGON SCORE IN CHINA
3
Table 2. Demographics, baseline characteristics, and proportions of the 3-month outcome Characteristics
Derivation cohort (n 5 1319)
TIMS-China cohort (n 5 970)
Age, y, median (IQR) Sex, n (%) Male Onset to treatment time, min, median (IQR) Baseline NIHSS score, points, median (IQR) Hyperdense cerebral artery sign, n (%) Early infarct signs, n (%) Prestroke mRS score, n (%) .1 Glucose, mmol/L, median (IQR) Medical history, n (%) Hypertension Diabetes mellitus Previous stroke Hyperlipidemia Atrial fibrillation Proportions of 3-month outcome mRS score, n (%) 0-2 3-4 5-6
69 (60-77)
65 (56-73)
729 (55.3) 118 (88-158) 9 (5-14) 233 (17.7) 406 (30.6)
586 (60.41) 170 (140-195) 11 (7-16) 72 (7.42) 55 (5.67)
82 (6.2) 6.6 (5.7-7.8)
42 (4.33) 6.8 (5.9-8.4)
787 (59.7) 191(14.5) 168 (12.7) 516 (39.1) 365 (27.7)
571 (58.87) 172 (17.73) 178 (18.35) 67 (6.91) 176 (18.14)
798 (60.5) 339 (25.7) 182 (13.8)
563 (58.04) 273 (28.14) 134 (13.81)
Abbreviations: IQR, interquartile range; mRS, modified Rankin Scale; NIHSS5National Institutes of Health Stroke Scale; TIMS-China, Thrombolysis Implementation and Monitor of acute ischemic Stroke in China.
drawn. Reliability was investigated with calibration curves, which were assessed using Hosmer–Lemeshow goodness-of-fit c2 test. However, as Hosmer–Lemeshow test was known to be oversensitive to small deviations from good fit in large samples, we also compared predicted versus observed rate of the good and the miserable outcome using Pearson correlation coefficient. The absolute error for calibration was calculated by subtracting the predicted probabilities from the observed ones. The mean absolute error less than .4 was defined as acceptable. All analyses were conducted using SAS statistical software (version 9.2, SAS Institute Inc., Cary, NC).
P , .01) and atrial fibrillation (18.14% versus 27.7%, P , .01). Distribution of 3-month outcome of each point for the DRAGON score in the TIMS-China cohort is shown in Figure 1. An extreme DRAGON score of 9 and 10 did not occur; the most frequent score was 4, in 260 (26.8%) patients. Proportions of patients with good outcome were 94%, 83%, 70%, and 0% for 0 to 1, 2, 3, and 8 to 10 score
Results From May 2007 to April 2012, a total of 1440 patients who received intravenous t-PA treatment were registered in the TIMS-China. After excluding the cases thrombolysed outside 0- to 4.5-hour window (n 5 307), with basilar artery occlusion (n 5 54), with missing data (ie, time from symptom onset to thrombolysis, or baseline National Institutes of Health Stroke Scale; n 5 87), and with lack of follow-ups (n 5 22), 970 patients were included in the present study. The baseline characteristics of patients are shown in Table 2. In the TIMS-China cohort, the median age was 65 years (IQR, 56-73), and 60.41% were men. The cohort had a lower rate of hyperlipidemia (6.91% versus 39.1%,
Figure 1. Three-month outcome per increasing points of the DRAGON score in the TIMS-China cohort (n 5 970). Good 3-month outcome (gray bars) indicates modified Rankin scale (mRS) score of 0 to 2, and miserable outcome (black bars) indicates mRS 5 to 6. Abbreviations: TIMS-China, Thrombolysis Implementation and Monitor of acute ischemic Stroke in China.
X. ZHANG ET AL.
4
Table 3. Sensitivity, specificity, and positive likelihood ratio for good and miserable 3-month outcome for each individual DRAGON score Good outcome (mRS score 0-2) DRAGON 0-1 2 3 4 5 6 7 8-10
Miserable outcome (mRS score 5-6)
Sensitivity, %
Specificity, %
Positive LR
Sensitivity, %
Specificity, %
Positive LR
9.1 (7.0-11.8) 21.4 (18.3-25.1) 27.7 (24.2-31.6) 25.2 (21.8-29.0) 10.1 (7.9-12.9) 5.7 (4.1-7.9) .0 (.0-.0) .0 (.0-.0)
99.8 (98.7-100) 93.9 (91.1-95.8) 83.1 (79.1-86.4) 71.0 (66.4-75.2) 70.0 (65.4-74.3) 87.2 (83.6-90.1) 95.6 (93.1-97.2) 99.5 (98.2-99.9)
38.0 (5.3-273.4) 3.5 (2.3-5.3) 1.6 (1.3-2.1) .9 (.7-1.1) .3 (.3-.5) .4 (.3-.7) .1 (.0-.4) —
.0 (.0-.0) 3.7 (1.6-8.4) 11.9 (7.5-18.5) 20.1 (14.2-27.7) 35.8 (28.2-44.2) 20.9 (14.9-28.5) 6.7 (3.6-12.3) .0 (.0-.0)
93.7 (91.8-95.1) 83.1 (80.4-85.5) 75.0 (72.0-77.8) 72.1 (69.0-75.1) 84.3 (81.7-86.6) 93.9 (91.4-94.8) 98.6 (97.5-99.2) 99.9 (99.3-100.0)
— .2 (.1-.5) .5 (.3-.8) .7 (.5-1.0) 2.3 (1.7-3.0) 3.1 (2.1-4.7) 4.7 (2.0-10.9) 6.2 (.4-99.1)
Abbreviations: LR, likelihood ratio; mRS, modified Rankin Scale.
points, respectively. Proportions of patients with miserable outcome were 0%, 3%, 9%, and 50% for 0 to 1, 2, 3, and 8 to 10 points, respectively. The sensitivity, specificity, and the positive likelihood ratios of the prediction for each score are also shown (Table 3). At both extremes of the score in predicting good and miserable outcome, the specificity performed excellent of almost reaching up to 100%. The likelihood ratios were also correlatively high. The external validation results are shown in Figure 2. The C statistics were .73 (95% CI, .70-.76) for good outcome prediction and .75 (95% CI, .70-.79) for miserable outcome prediction. Moreover, the score showed good calibration. The calibration curves are shown in Figure 3. Comparing the diminished calibration for 3-month good outcome (Hosmer–Lemeshow test; P , .05) with 3-month miserable outcome (Hosmer– Lemeshow test; P 5 .259), we also plotted predicted versus observed functional outcome in the validation samples (Fig 3, A for good outcome and Fig 3, B for miser-
able outcome). There was a high correlation between the predicted and observed rates (Pearson correlation coefficient, .98 for 3-month good outcome and .98 for 3-month miserable outcome; both P , .0001). The mean absolute errors were .05 and .05 for the model predicting good and poor outcome, respectively.
Discussion The DRAGON score was recently developed as a simple and quick tool to predict 3-month functional outcome in ischemic stroke patients treated with intravenous thrombolysis. The accuracy and usability of the score, which was confirmed in the present study, emphasized the potential of the DRAGON score as a useful risk instrument that can be applied to clinical daily practice and stroke research. This report represented the first external validation of the DRAGON score in the Chinese population.
Figure 2. Area under the receiver operating characteristic curves (AUC-ROC) of the DRAGON score in the TIMS-China cohort for 3-month good (A) and miserable (B) outcome. The AUC-ROC was .73 (.70-.76) for good outcome and .75 (.70-.79) for miserable outcome. Abbreviations: TIMS-China, Thrombolysis Implementation and Monitor of acute ischemic Stroke in China.
VALIDATION OF THE DRAGON SCORE IN CHINA
5
Figure 3. Predicted and observed probabilities of 3-month good outcome (A) and miserable outcome (B) in the TIMS-China cohort (Pearson correlation coefficient, .98 for 3-month good outcome and .98 for 3-month miserable outcome; both P , .0001). Abbreviations: TIMS-China, Thrombolysis Implementation and Monitor of acute ischemic Stroke in China.
It is critical to notice that the DRAGON score was not only a tool to help clinicians in immediately estimating outcome based on a number of variables when discussing prognosis of intravenous thrombolysis with stroke patients and their families, but also a useful instrument to support the decision-making process when t-PA treatment alone did not offer high likelihood of favorable outcome. With a score of 8, 9, or 10, patients were highly likely to have a miserable functional outcome even if t-PA was given, and possible add-on treatment (ie, endovascular strategy) should be considered. The longer the onset-to-needle time of intravenous thrombolysis, the higher the likelihood of acute ischemic stroke patients to achieve the miserable outcome at 3 months. What we could not ignore is the fact that the time-window of intravenous thrombolysis could be one of the confounding factors in our study. So the onset-toneedle time and the door-to-needle time in each group stratified by the DRAGON score are also shown (Fig 4).We could tell from Figure 4 that each group of different DRAGON score has a similar onset-to-needle time and door-to-needle time (P . .05), so the confounding influence of time-window of treatment could be excluded. In the TIMS-China cohort, the DRAGON score showed a graded effect: the higher the score, the lower the risk of a favorable outcome. When the score reached 8 point, the proportion of good functional outcome came to 0%. An excellent performance of specificity of the score was seen at both the extreme poles, which could decrease the possibility of incorrect prediction of dismal prognosis to the greatest degree, and thus had more significance when considering invasive add-on rescue strategies, which was costly and had a risk of severe complications. The study also demonstrated that the (hyper) Dense cerebral artery sign/early infarct signs on admission CT scan, prestroke modified Rankin Scale (mRS) score,
Age, Glucose level at baseline, Onset-to-treatment time, and baseline National Institutes of Health Stroke Scale score (DRAGON) score had an acceptable external validity for the prediction of 3-month functional outcome after intravenous t-PA for ischemic stroke patients, despite the differences between the original and the validation cohorts. The C statistics for both favorable and miserable outcome were good, similar to that in the validation of the Third International Stroke Trial (IST-3) cohort (.78[.76-.81]).8 Moreover, it is higher than most of the existing prediction models: the iScore (.67),10,11 the Glucose at presentation, Race, Age, Sex, systolic blood Pressure at presentation, Severity of stroke at presentation (GRASPS) score (.71 [.68-.73]),12 the Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS) score (.70),13 the Stroke Prognostication using Age and NIH Stroke Scale (SPAN)-100 score (.64),14 the IST-3 score (.71[.68-.74]),8 and so forth.15 In the meantime, the predicted and observed rates of 3-month outcome in the TIMS-China cohort were in close agreement across the entire spectrum of risk based on the DRAGON score. And the mean absolute errors
Figure 4. The onset-to-needle time and the door-to-needle time in each group stratified by the DRAGON score. Dots represent the median time in each group, and vertical lines represent the interquartile ranges (IQR).
X. ZHANG ET AL.
6
were less than .4 in both outcomes, which has previously been used to define excellent calibration.16 All mentioned above suggested that the DRAGON score was a more powerful instrument to predict clinical outcome of stroke patients treated with t-PA. There has been a rapid proliferation of scores for acute ischemic stroke treated with intravenous thrombolysis,8,11-15,17-20 and the DRAGON score has its obvious superiority. Besides the higher C statistics, the DRAGON score consists of 6 variables which are available shortly after patient admission. So compared with the higher request of obtaining variables and the complex logistic regression equation, the DRAGON score has a greater significance to direct physicians to make the quick and right decision before alteplase administration. Our study has limitations. There could be a selection bias. First, there were only 2 patients with a DRAGON score of 8 and no patients with the score of 9-10 in the TIMS-China cohort. The lack of patients with DRAGON score of 8-10 may be related to the relatively small sample of the TIMS-China cohort because of the comparatively low rate of intravenous thrombolysis in China.21 What’s more, the indication and contraindication of intravenous thrombolysis to acute stroke patients were strictly followed by the hospitals which have the ability to administer alteplase. Thus, many off-labeled patients, such as 80 years or older and National Institutes of Health Stroke Scale score more than 25, which count 1 point separately in the DRAGON score, were excluded to receive the intravenous thrombolysis for the sake of patients’ safety. And this may also lead to the lack of higher DRAGON score in TIMS-China cohort. Second, considering the selection of participating hospitals, although covering all areas of China, was mostly from urban regions. The applicability of the DRAGON score to all hospitals in China with a larger population needs further investigation. In conclusion, the DRAGON score is a valid and easyto-obtain clinical instrument to predict favorable and miserable outcome at 3 months after intravenous thrombolysis. This study demonstrated the accuracy and usability of the DRAGON score in the Chinese population in daily practice.
References 1. Tissue plasminogen activator for acute ischemic stroke. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. N Engl J Med 1995; 333:1581-1587. 2. Hacke W, Kaste M, Bluhmki E, et al. Thrombolysis with alteplase 3 to 4.5 hours after acute ischemic stroke. N Engl J Med 2008;359:1317-1329. 3. Saver JL. Hemorrhage after thrombolytic therapy for stroke: the clinically relevant number needed to harm. Stroke 2007;38:2279-2283. 4. Strbian D, Meretoja A, Ahlhelm FJ, et al. Predicting outcome of IV thrombolysis-treated ischemic stroke
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
patients: the DRAGON score. Neurology 2012;78: 427-432. Stegmayr B, Vinogradova T, Malyutina S, et al. Widening gap of stroke between east and west. Eight-year trends in occurrence and risk factors in Russia and Sweden. Stroke 2000;31:2-8. Strbian D, Seiffge DJ, Breuer L, et al. Validation of the DRAGON score in 12 stroke centers in anterior and posterior circulation. Stroke 2013;44:2718-2721. Seiffge DJ, Karagiannis A, Strbian D, et al. Simple variables predict miserable outcome after intravenous thrombolysis. Eur J Neurol 2014;21:185-191. Whiteley WN, Thompson D, Murray G, et al. Targeting recombinant tissue-type plasminogen activator in acute ischemic stroke based on risk of intracranial hemorrhage or poor functional outcome: an analysis of the third international stroke trial. Stroke 2014;45:1000-1006. Liao XL, Wang CX, Wang YL, et al. Implementation and outcome of thrombolysis with alteplase 3 to 4.5 h after acute stroke in Chinese patients. CNS Neurosci Ther 2013;19:43-47. Saposnik G, Kapral MK, Liu Y, et al. IScore: a risk score to predict death early after hospitalization for an acute ischemic stroke. Circulation 2011;123:739-749. Saposnik G, Demchuk A, Tu JV, et al. The iScore predicts efficacy and risk of bleeding in the National Institute of Neurological Disorders and Stroke Tissue Plasminogen Activator Stroke Trial. J Stroke Cerebrovasc Dis 2013; 22:876-882. Menon BK, Saver JL, Prabhakaran S, et al. Risk score for intracranial hemorrhage in patients with acute ischemic stroke treated with intravenous tissue-type plasminogen activator. Stroke 2012;43:2293-2299. Mazya M, Egido JA, Ford GA, et al. Predicting the risk of symptomatic intracerebral hemorrhage in ischemic stroke treated with intravenous alteplase: safe Implementation of Treatments in Stroke (SITS) symptomatic intracerebral hemorrhage risk score. Stroke 2012;43:1524-1531. Saposnik G, Guzik AK, Reeves M, et al. Stroke prognostication using age and NIH Stroke Scale: SPAN-100. Neurology 2013;80:21-28. Cucchiara B, Tanne D, Levine SR, et al. A risk score to predict intracranial hemorrhage after recombinant tissue plasminogen activator for acute ischemic stroke. J Stroke Cerebrovasc Dis 2008;17:331-333. Johnston KC, Wagner DP, Wang XQ, et al. Validation of an acute ischemic stroke model: does diffusion-weighted imaging lesion volume offer a clinically significant improvement in prediction of outcome? Stroke 2007; 38:1820-1825. Lou M, Safdar A, Mehdiratta M, et al. The HAT Score: a simple grading scale for predicting hemorrhage after thrombolysis. Neurology 2008;71:1417-1423. Kent DM, Selker HP, Ruthazer R, et al. The strokethrombolytic predictive instrument: a predictive instrument for intravenous thrombolysis in acute ischemic stroke. Stroke 2006;37:2957-2962. Strbian D, Engelter S, Michel P, et al. Symptomatic intracranial hemorrhage after stroke thrombolysis: the SEDAN score. Ann Neurol 2012;71:634-641. Flint AC, Faigeles BS, Cullen SP, et al. THRIVE score predicts ischemic stroke outcomes and thrombolytic hemorrhage risk in VISTA. Stroke 2013;44:3365-3369. Wang Y, Liao X, Zhao X, et al. Using recombinant tissue plasminogen activator to treat acute ischemic stroke in China: analysis of the results from the Chinese National Stroke Registry (CNSR). Stroke 2011;42:1658-1664.