Rapidly Improving Stroke Symptoms: A Pilot, Prospective Study Clotilde Balucani, MD, PhD,* Riccardo Bianchi, PhD,† Charles Ramkishun, BS,*† Jeremy Weedon, PhD,‡ Susan Law, DO,* Michael Szarek, PhD, MS,‡ Diana Rojas-Soto, MD,* Sara Tariq, BS,* and Steven R. Levine, MD*xk
Background: Rapidly improving stroke symptoms (RISSs) are a controversial exclusion for intravenous recombinant tissue plasminogen activator (rt-PA) for acute ischemic stroke (AIS). We estimated the frequency of 4 prespecified RISS definitions and explored their relationship to clinical outcome. Methods: Pilot, prospective study of AIS patients admitted within 4.5 hours of symptom onset. Serial assessments using National Institute of Health Stroke Scale (NIHSS) were performed every 20 6 5 minutes until a rt-PA treatment decision was made, independent of the study. Improvement was calculated as the difference between baseline NIHSS and treatment decision NIHSS. RISS was defined as a 4-point or greater improvement, 25% or greater, 50% or greater, and according to the previously reported TREAT (The Re-examining Acute Eligibility for Thrombolysis) criteria. Unfavorable outcome was defined as modified Rankin Scale score more than 1 at 90 days after stroke. Logistic regression determined if RISS definition(s) related to the outcome. Results: Fifty patients with AIS were enrolled: mean age 65 years; median baseline NIHSS score 5 (interquartile range, 2-11). RISS frequencies were 10%-22% based on definition. Median treatment decision NIHSS score is 5 (interquartile range, 2-9). Twenty-three (46%) patients received rt-PA. None of the 3 non-TREAT RISS definitions was independently associated with the outcome. Five of fifty (10%) were RISS according to the TREAT criteria, all 5 had good outcome without rt-PA. Conclusions: A Serial NIHSS assessment before treatment decision is feasible and may help determine the frequency and magnitude of RISS. This is the first prospective estimate of RISS frequency and outcome according to various prespecified definitions. The TREAT RISS frequency as a more restrictive definition may better predict good outcome of RISS in future, larger studies. Key Words: Acute stroke—thrombolysis—tissue plasminogen activator—rapidly improving stroke symptoms—minor stroke—National Institutes of Health Stroke Scale. Ó 2015 by National Stroke Association
From the *Department of Neurology and Stroke Center; †Department of Physiology and Pharmacology; ‡Scientific Computing Center; xDepartment of Emergency Medicine, SUNY Downstate Medical Center; and kDepartment of Neurology, Kings County Hospital Center, Brooklyn, New York. Received August 25, 2014; revision received December 26, 2014; accepted January 9, 2015. Study conception was done by C.B. and S.R.L.; statistical analyses were performed by J.W., M.S., and R.B.; data collection was done by C.B., C.R., S.L., D.R.-S., S.T., and S.R.L.; interpretation of results was carried out by C.B., R.B., J.W., S.R.L.; and C.B. and S.R.L. drafted the article and all authors contributed input to the final version. C.B. was supported by an award from the American Heart Association/American Stroke Association—Founders Affiliate and the
American Brain Foundation, ‘‘The Lawrence M. Brass, M.D. Stroke Research Postdoctoral Fellowship Award.’’ S.R.L. was supported by investigator-initiated grant from Genentech, Inc. (IST ML28239); Scientific Advisory Committee for PRISMS, a randomized clinical trial funded by Genentech, Inc. (modest honorarium and travel expenses). R.B., C.R., J.W., S.L., M.S., D.R.-S., and S.T. report no disclosures. Address correspondence to Steven R. Levine, MD, Department of Neurology and Stroke Center, The State University of New York, Downstate Medical Center, 450 Clarkson Avenue, MSC 1213, Brooklyn, NY 11203-2012. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2015.01.017
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2015: pp 1-6
1
C. BALUCANI ET AL.
2
Between 4.4% and 44% of ‘‘timely eligible’’ acute ischemic stroke (AIS) patients who present to the emergency departments (EDs) are not treated with intravenous recombinant tissue plasminogen activator (rt-PA) solely because of rapidly improving stroke symptoms (RISS).1-4 Studies suggest that the outcome of these patients is not invariably benign, bringing into question the decision not to treat them with rt-PA.5-8 Subsequent to the National Institute of Neurological Disorders and Stroke (NINDS) rt-PA Stroke Trials,9 no consensus has been achieved to define RISS.10,11 The rt-PA package insert (http://www.activase.com), American Heart/Stroke Association Guidelines,12 European Stroke Organisation,13 and the National Institutes for Health and Clinical Excellence of the United Kingdom14 do not specifically or quantitatively define RISS. A further attempt toward consensus for defining RISS has been recently developed.8 We obtained pilot, prospective data on frequency and magnitude of RISS, according to 4 prespecified definitions and explored their relationship to clinical outcome.
Methods Study Population This pilot, prospective, observational study evaluated adult (older than age 18 years) AIS patients admitted to SUNY Downstate Medical Center in Brooklyn, New York. Eligibility included sudden onset of focal neurologic deficits suggestive of stroke presenting within 4.5 hours of symptoms; a noncontrast computed tomography scan negative for brain hemorrhage or any other brain lesion rather than ischemic stroke. All subjects/ designee provided informed consent. The protocol was approved by the SUNY Downstate Institutional Review Board. This study did not interfere with and/or delay the standard screening process aimed at identifying patients eligible for rt-PA. Any treatment decision was made independent of this study by the stroke treating physician.
ence between baseline NIHSS (b-NIHSS) and treatment decision NIHSS (td-NIHSS).
RISS and Outcome Definitions Prespecified definitions of RISS were (1) change in NIHSS score with 4-point or greater improvement (IMP) or (2) percent change in NIHSS score: 25% or greater or 50% or greater IMP and change in NIHSS score to nondisabling symptoms according to the previously published, The Re-examining Acute Eligibility for Thrombolysis (‘‘TREAT’’) Task Force Criteria definition.8 The TREAT Task Force comprised members of the original NINDS rt-PA Stroke Trial Steering Committee combined with other leaders in the field of stroke and emergency medicine. This Task Force was formed in an attempt to address the rationale and relevance of individual exclusion criteria to rt-PA and to determine if these criteria require clarifications and more precise definitions that could be implemented in the current clinical practice.8 Specifically, The TREAT Task Force sought to develop a clinically meaningful definition of RISS that may be used to guide stroke treating physicians’ decision to treat or not to treat with rt-PA. Consensus defined RISS as an exclusion criterion for rt-PA as improvement to a mild stroke such that any remaining deficits seem nondisabling.8 To be nondisabling, none of the following deficits should be present: (1) complete hemianopsia ($2 on the NIHSS question 3); (2) severe aphasia ($2 on NIHSS question #9); (3) visual or sensory extinction ($1 on NIHSS question #11); (4) any weakness limiting sustained effort against gravity ($2 on NIHSS question #6 or #7); (5) any deficits that lead to a total NIHSS .5; and (6) any remaining deficit considered potentially disabling in the view of the patient or the treating practitioner.8 Clinical outcome was defined using the modified Rankin Scale (mRS). Unfavorable outcome was defined as mRS greater than 1 at 90 days after stroke and at 7 days after stroke or discharge.
Statistical Analysis Protocol Subjects were administered serial assessments of stroke severity as measured by the National Institutes of Health Stroke Scale (NIHSS)15 performed every 20 6 5 minutes until a treatment decision (rt-PA vs no rt-PA) was made by the treating physician. RISS study team members were notified via a hospital-wide ED ‘‘code stroke’’ paging system. Baseline NIHSS (b-NIHSS) was defined as the first NIHSS score performed at patient arrival at the ED by 1 member of the RISS study team. All examiners were certified in the NIHSS. RISS was defined as a measurable improvement of neurologic deficits between 2 clinical evaluations performed at different times, and specifically, as the differ-
Descriptive statistics were used to evaluate age, sex, race, and other cardiovascular risk factors between those with RISS and no RISS and between those treated and not treated with rt-PA. The Fisher exact test for categorical variables and the Wilcoxon test for numeric data were used for comparison between groups. Exact logistic regression was used to model 90-day mRS score more than 1 from predictors age, b-NIHSS, time to NIHSS, rt-PA use, and (in separate models) 4 different definitions of RISS. P values and 95% confidence limits are based on Mid-P methods.16 A secondary analysis was performed to test the diagnostic utility of specific NIHSS score cut points from 1 to 10 to predict mRS more than 1 at 90 days. We generated
RAPIDLY IMPROVING STROKE SYMPTOMS
3
Table 1. Demographics and clinical characteristics of the study cohort
Patients’ Characteristics Age, y, mean (SD) Male, n (%) Race/ethnicity, n (%) White Black Asian Hispanic Hypertension, n (%) Diabetes, n (%) Hyperlipidemia, n (%) Prior stroke/TIA, n (%) History of heart disease, n (%) Family history of stroke, n (%) Cigarette smoking, n (%) Baseline NIHSS, median (25th-75th) NIHSS at time of treatment decision, median (25th-75th) RISS definitions, n (%) $4-point improvement $25% improvement $50% improvement TREAT-based mRS at discharge/day 7, n (%) 0-1 $2 Missing mRS at day 90, n (%) 0-1 $2 Missing mRS at day 90 with day 7 imputation if missing, n (%) 0-1 $2 Missing
Total sample (n 5 50)
No rt-PA (n 5 27)
Received rt-PA (n 5 23)
65 (15) 28 (56)
68 (13) 17 (63)
61 (16) 11 (48)
7 (14) 39 (78) 0 4 (8) 37 (74) 12 (24) 17 (34) 14 (28) 8 (16) 6 (12) 8 (16) 5 (2-11) 5 (2-9)
4 (15) 22 (81) 0 1 (4) 20 (74) 7 (26) 8 (30) 7 (26) 4 (15) 5 (19) 5 (19) 3 (2-5) 2 (1-4)
3 (13) 17 (74) 0 3 (13) 17 (74) 5 (22) 9 (39) 7 (30) 4 (17) 1 (4) 3 (13) 9 (6-14) 8 (6-13)
5 (10) 11 (22) 6 (12) 5 (10)
3 (11) 6 (22) 5 (18) 5 (18)
2 (9) 5 (22) 1 (4) —
25 (50) 25 (50) 0
17 (63) 10 (37) 0
8 (35) 15 (65) 0
23 (46) 23 (46) 4 (8)
17 (63) 7 (26) 3 (11)
6 (26) 16 (70) 1 (4)
25 (50) 25 (50) 0
18 (67) 9 (33) 0
7 (30) 16 (70) 0
P value .14 .39 .53
1.00 1.00 .56 .76 1.00 .20 .71 .0003 ,.0001 1.00 1.00 .20 — .09
.007
.02
Abbreviations: mRS, modified Rankin Scale; RISS, rapidly improving stroke symptoms; rt-PA, recombinant tissue plasminogen activator; SD, standard deviation; TIA, transient ischemic attack.
a receiver operating characteristic curve and calculated the area under the curve.
treated and not treated with rt-PA with the exception of b-NIHSS and td-NIHSS scores (Table 1).
Frequency of RISS
Results Our study population included 50 AIS patients. Table 1 summarizes the baseline demographics. Median b-NIHSS score was 5 (interquartile range [IQR], 2-11). Before a rt-PA treatment decision was made, at least 2 serial NIHSS scores were obtained in 47 (94%) subjects; treatment decision was based on the b-NIHSS in 3 (6%). Three serial, pretreatment decision NIHSS scores were obtained in 31 (62%) subjects. Median td-NIHSS score was 5 (IQR, 2-9). Twenty-three (46%) patients received rt-PA. There were no differences in baseline variables between those
In the entire cohort of 50 subjects there was a mean overall improvement of 1 6 3 (standard deviation) point on the NIHSS, with a median of 0 (IQR, 0-1) from the b-NIHSS to the td-NIHSS score. Fourteen of fifty (28%) subjects had improvement with a mean improvement in NIHSS of 4 6 3 points (median, 3; IQR, 2-5). Of those, 14 of 50 subjects showing any degree of improvement, 2 subjects improved only 1 point from b-NIHSS scores of 9 and 16, whereas 12 of 50 (24%) subjects improved according to at least 1 of 4 prespecified definitions with a mean of 4 6 3 points (median, 3; IQR, 2-5).
C. BALUCANI ET AL.
4
In 25 of 50 subjects (50%) NIHSS did not change, and in 8 of 50 (16%) subjects worsened by a mean of 2 6 1 points (median, 1; IQR, 1-1). Frequency of RISS varied based on definition used: 10% for a 4-point or greater IMP, 22% for 25% or greater IMP, 12% for 50% or greater IMP, and 10% for the TREAT definition. Twenty-three (46%) patients received rt-PA. For all nonTREAT–based RISS definitions there was no difference in treatment rates between RISS subjects (5 of 12, 42%) versus no RISS (18 of 38, 47%) subjects (P 5 .754).
Frequency of Clinical Outcome In the overall cohort, 42% of patients had good outcome (mRS, 0-1) at 90 days: 17 of 27 (63%) not treated with rt-PA versus 6 of 23 (26%) treated with rt-PA, P 5.02. There was a difference in td-NIHSS between those not given rt-PA (2; IQR, 1-4) and those given rt-PA (8; IQR, 6,13), P , .0001. Frequency of Clinical Outcome Based on RISS Definitions Good outcome was achieved in 1 of 5 (20%) subjects who had a 4-point or greater IMP, 6 of 11 (55%) of subjects with 25% or greater IMP, 5 of 6 (83%) for 50% or greater IMP, and 5 of 5 (100%) of those defined according to the TREAT definition. Those without RISS had a 50% (n 5 19 of 38) probability of good outcome at 90 days. There was no significant difference in frequency of good outcome at 90 days between subjects without RISS and those with RISS of a 4-point or greater IMP (P 5 .348), RISS of 25% or greater IMP (P 5 1.00), and RISS of 50% or greater IMP (P 5 .189). A borderline significant difference was found with the TREAT-based RISS definition (P 5 .050). None of the 4 definitions of RISS was an independent predictor of clinical outcome in the multivariate analysis (Table 2). Figure 1 illustrates the individual NIHSS data points from baseline to treatment decision and subsequent clinical outcomes for subjects with RISS by each of the 4 RISS definitions. Figure 2 shows that the sensitivity and specificity characteristics of NIHSS score cut points progress in a fairly smooth manner in the range of NIHSS score values between 1 and 11. We could not detect an obviously superior specific NIHSS cut point. The area under the curve is .83, indicating moderate predictive utility independent of cut point.
Discussion Our study supports the feasibility of obtaining a serial NIHSS assessment in the ED before treatment decision. We found that over a third of ‘‘timely eligible’’ AIS
Table 2. Probability of unfavorable outcome (mRS .1) at 90 days with any RISS definition
Predictors
Odds ratio (95% confidence intervals) P value
Model 1 (RISS $4 IMP) 1.29 (.07, 52.0) Model 2 (RISS $25% IMP) .61 (.06, 4.89) Model 3 (RISS $50% IMP) .09 (,.01, 2.97)
.746 .819 .165
Abbreviations: IMP, improvement; mRS, modified Rankin Scale; RISS, rapidly improving stroke symptoms; TREAT, The Re-examining Acute Eligibility for Thrombolysis. Odds ratios and P values were not calculated for model 4, which included the TREAT-based RISS definition, due to zero cell frequency. If day 90 mRS missing, imputed with mRS at discharge/ day 7 (n 5 4).
patients appear to have NIHSS changes while in the ED. Current clinical practice and guidelines12-14 do not include serial NIHSS scores to be determined before treatment, which could obscure the identification of these patients. Our pilot data provide, to our knowledge, the first prospective estimates of frequency of RISS according to multiple prespecified definitions. These estimates could help design larger studies to determine treatment effect of rt-PA in these cohorts of varying RISS frequencies. In the time from baseline to td-NIHSS, subjects with any prespecified RISS definition improved approximately 4 (standard deviation, 3) points on average. The frequency of RISS, depending on definition, ranged between 10% and 22%, somewhat lower than the prior retrospective studies of RISS patients excluded from rt-PA.1-4 Although patients with RISS were either treated or not treated with rt-PA in a similar frequency for RISS definitions of 4-point or greater IMP and 25% or greater IMP, only 1 of 6 with 50% or greater IMP was treated, reflecting the clinicians’ bias in not treating those who improved more significantly. We found that the significant 6-point difference in median td-NIHSS scores between the rt-PA and no rt-PA treatment groups likely influenced outcome to a greater degree than rt-PA treatment itself and is a confounder in our results. Our study further suggests that heterogeneity of presenting NIHSS scores and their changes over time are considerable, as can be visualized in Figure 1. This degree of heterogeneity could influence measures of improvement based on absolute or relative NIHSS stroke changes, likely leading to a failure to predict outcome. As all 5 subjects with TREAT-based RISS criteria were not treated with rt-PA as they all improved to a nondisabling state and all had a good outcome, this definition could help predict those AIS patients who are likely to have a good outcome. However, larger cohorts of AIS patients fulfilling TREAT criteria will be required to validate our preliminary results.
RAPIDLY IMPROVING STROKE SYMPTOMS
5
Figure 1. Time course of NIHSS scores and clinical outcomes in subjects with RISS. Plots of serial NIHSS total scores evaluated at the indicated intervals after symptom onset (time 0) in the 12 subjects with RISS. The 4 RISS definitions are illustrated in separate panels: (A) improvement (IMP) in NIHSS score is 4 point or greater; (B) NIHSS IMP of 25% or greater; (C) NIHSS IMP of 50% or greater; (D) TREAT-based RISS. In each plot, the first circle represents NIHSS score at baseline and the green-filled circle indicates the NIHSS score at the treatment decision time (tPA administered 5 red plots; no-tPA administered 5 blue plots). Dashed lines connect the serial NIHSS score plots to corresponding clinical outcomes measured as modified Rankin Scale (mRS, right bars) at 7 days or at hospital discharge (DC) and 90 days after stroke. Abbreviations: NIHSS, National Institutes of Health Stroke Scale; RISS, rapidly improving stroke symptoms; tPA, tissue plasminogen activator; TREAT, The Re-examining Acute Eligibility for Thrombolysis. (Color version of figure is available online.)
Figure 2. ROC curve for baseline NIHSSS as a predictor of modified Rankin Scale more than 1 at 90 days. Abbreviations: NIHSSS, National Institutes of Health Stroke Scale Score; ROC, receiver operating characteristic.
We did not find a specific, individual NIHSS score cut point for optimizing prediction of outcome. This may be because of our sample size or that for each cut point, there is a balanced trade-off between specificity and sensitivity. Limitations to our study include the small sample size, albeit a pilot study, and as such it is hypothesis generating; the potential lack of generalizability because of an urban, predominantly Caribbean black cohort; and the potential clinical bias of rt-PA treatment decision based on stroke severity. Also, these were not consecutive patients as this was a pilot study performed primarily during the day. Those arriving between 7 PM and 7 AM or on weekends were only screened and enrolled if members of the study personnel were available to come in within the time window required. The relatively lower b-NIHSS scores seen in our prospective, consecutive cohort are lower than those in acute stroke treatment trials but do reflect the lower population-based NIHSS of acute stroke patients.17 Finally, patients who were excluded (see Fig. 1) from rt-PA treatment because of their TREAT RISS definition had very low NIHSS scores at presentation. This may significantly limit the utility of this score.
C. BALUCANI ET AL.
6
It is possible that patients had more severe strokes at onset and improved to a mild, nondisabling stroke by the time they were first seen by our study team. Hence, the improvement could have occurred in the prehospital setting before their enrollment. The present study suggests a standardized procedure to identify AIS patients with RISS in the ED before treatment decision. However, the study does not provide information on evaluation of patients in the time between symptoms onset and ED admission. For a more sensitive identification and thorough characterization of RISS patients, additional larger studies are necessary to validate a standard assessment of these patients in the prehospital field, before ED admission. We further need to better define rate of improvement over time. This specific aspect could help differentiate between those patients who improve dramatically very early after symptoms onset (that is, likely to be TIAs and the type of patient who was excluded from the NINDS rt-PA Stroke Trials9), those that improve to a minimally disabling stroke, and those that do not improve over time and present with stable minor deficits. Clinical outcomes of these subgroups can differ substantially and the most appropriate treatment approaches need to be evaluated in controlled clinical trials. Acknowledgment: The authors acknowledge Drs Volodymyr Vulkanov, Sebina Bulic, Priyank Khandelwal, Qing Hao, and Benedict Tan for their assistance in conducting this project.
References 1. Nedeltchev K, Schwegler B, Haefeli T, et al. Outcome of stroke with mild or rapidly improving symptoms. Stroke 2007;38:2531-2535. 2. Barber PA, Zhang J, Demchuk AM, et al. Why are stroke patients excluded from t-PA therapy? An analysis of patient eligibility. Neurology 2001;56:1015-1020. 3. Eissa A, Krass I, Levi C, et al. Understanding the reasons behind the low utilisation of thrombolysis in stroke. Australas Med J 2013;6:152-167. 4. Katzan IL, Hammer MD, Hixson ED, et al. Utilization of intravenous tissue plasminogen activator for acute ischemic stroke. Arch Neurol 2004;61:346-350.
5. Willey JZ, Stillman J, Rivolta JA, et al. Too good to treat? Outcomes in patients not receiving thrombolysis due to mild deficits or rapidly improving symptoms. Int J Stroke 2012;7:202-206. 6. Smith EE, Fonarow GC, Reeves MJ, et al. Outcomes in mild or rapidly improving stroke not treated with intravenous recombinant tissue-type plasminogen activator: findings from Get With The Guidelines-Stroke. Stroke 2011;42:3110-3115. 7. Balucani C, Levine SR. Mild stroke and rapidly improving symptoms: it’s not always a happy ending. Stroke 2011;42:3005-3007. 8. Levine SR, Khatri P, Broderick JP, et al. Review, historical context, and clarifications of the NINDS rt-PA stroke trials exclusion criteria: Part 1: rapidly improving stroke symptoms. Re-examining Acute Eligibility for Thrombolysis (TREAT) Task Force. Stroke 2013;44:2500-2505. 9. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med 1995; 333:1581-1587. 10. Piran P, Balucani C, Ramkishun C, et al. Defining the undefined: rapidly improving stroke symptoms, a systematic review of the literature. Stroke 2014;45:A28. 11. Dirks M, Niessen LW, Koudstaal PJ, et al. Delphi panel on indications and contraindications for intravenous thrombolysis in acute ischaemic stroke. Intravenous thrombolysis in acute ischaemic stroke: from trial exclusion criteria to clinical contraindications. An international Delphi study. J Neurol Neurosurg Psych 2007;78:685-689. 12. Jauch EC, Saver JL, Adams HP Jr, et al. Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2013;44:870-947. 13. European Stroke Organisation (ESO) Executive Committee, ESO Writing Committee. Guidelines for management of ischaemic stroke and transient ischaemic attack 2008. Cerebrovasc Dis 2008;25:457-507. 14. National Institute for Health and Clinical Excellence. Stroke: the diagnosis and initial management of acute stroke and transient ischaemic attack. 2008 (clinical guideline 68) Available at: www.nice.org.uk/CG68. Accessed August 20, 2014. 15. Brott T, Adams HP Jr, Olinger CP, et al. Measurements of acute cerebral infarction: a clinical examination scale. Stroke 1989;20:864-870. 16. Barnard G. On alleged gains in power from lower p-values. Stat Med 1989;8:1469-1477. 17. Reeves M, Khoury J, Alwell K, et al. Distribution of National Institutes of Health stroke scale in the Cincinnati/Northern Kentucky Stroke Study. Stroke 2013; 44:3211-3213.
Background: Rapidly improving stroke symptoms (RISSs) are a controversial exclusion for intravenous recombinant tissue plasminogen activator (rt-PA) for acute ischemic stroke (AIS). We estimated the frequency of 4 prespecified RISS definitions and explored their relationship to clinical outcome. Methods: Pilot, prospective study of AIS patients admitted within 4.5 hours of symptom onset. Serial assessments using National Institute of Health Stroke Scale (NIHSS) were performed every 20 6 5 minutes until a rt-PA treatment decision was made, independent of the study. Improvement was calculated as the difference between baseline NIHSS and treatment decision NIHSS. RISS was defined as a 4-point or greater improvement, 25% or greater, 50% or greater, and according to the previously reported TREAT (The Re-examining Acute Eligibility for Thrombolysis) criteria. Unfavorable outcome was defined as modified Rankin Scale score more than 1 at 90 days after stroke. Logistic regression determined if RISS definition(s) related to the outcome. Results: Fifty patients with AIS were enrolled: mean age 65 years; median baseline NIHSS score 5 (interquartile range, 2-11). RISS frequencies were 10%-22% based on definition. Median treatment decision NIHSS score is 5 (interquartile range, 2-9). Twenty-three (46%) patients received rt-PA. None of the 3 non-TREAT RISS definitions was independently associated with the outcome. Five of fifty (10%) were RISS according to the TREAT criteria, all 5 had good outcome without rt-PA. Conclusions: A Serial NIHSS assessment before treatment decision is feasible and may help determine the frequency and magnitude of RISS. This is the first prospective estimate of RISS frequency and outcome according to various prespecified definitions. The TREAT RISS frequency as a more restrictive definition may better predict good outcome of RISS in future, larger studies. Key Words: Acute stroke—thrombolysis—tissue plasminogen activator—rapidly improving stroke symptoms—minor stroke—National Institutes of Health Stroke Scale. Ó 2015 by National Stroke Association
From the *Department of Neurology and Stroke Center; †Department of Physiology and Pharmacology; ‡Scientific Computing Center; xDepartment of Emergency Medicine, SUNY Downstate Medical Center; and kDepartment of Neurology, Kings County Hospital Center, Brooklyn, New York. Received August 25, 2014; revision received December 26, 2014; accepted January 9, 2015. Study conception was done by C.B. and S.R.L.; statistical analyses were performed by J.W., M.S., and R.B.; data collection was done by C.B., C.R., S.L., D.R.-S., S.T., and S.R.L.; interpretation of results was carried out by C.B., R.B., J.W., S.R.L.; and C.B. and S.R.L. drafted the article and all authors contributed input to the final version. C.B. was supported by an award from the American Heart Association/American Stroke Association—Founders Affiliate and the
American Brain Foundation, ‘‘The Lawrence M. Brass, M.D. Stroke Research Postdoctoral Fellowship Award.’’ S.R.L. was supported by investigator-initiated grant from Genentech, Inc. (IST ML28239); Scientific Advisory Committee for PRISMS, a randomized clinical trial funded by Genentech, Inc. (modest honorarium and travel expenses). R.B., C.R., J.W., S.L., M.S., D.R.-S., and S.T. report no disclosures. Address correspondence to Steven R. Levine, MD, Department of Neurology and Stroke Center, The State University of New York, Downstate Medical Center, 450 Clarkson Avenue, MSC 1213, Brooklyn, NY 11203-2012. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2015.01.017
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2015: pp 1-6
1
C. BALUCANI ET AL.
2
Between 4.4% and 44% of ‘‘timely eligible’’ acute ischemic stroke (AIS) patients who present to the emergency departments (EDs) are not treated with intravenous recombinant tissue plasminogen activator (rt-PA) solely because of rapidly improving stroke symptoms (RISS).1-4 Studies suggest that the outcome of these patients is not invariably benign, bringing into question the decision not to treat them with rt-PA.5-8 Subsequent to the National Institute of Neurological Disorders and Stroke (NINDS) rt-PA Stroke Trials,9 no consensus has been achieved to define RISS.10,11 The rt-PA package insert (http://www.activase.com), American Heart/Stroke Association Guidelines,12 European Stroke Organisation,13 and the National Institutes for Health and Clinical Excellence of the United Kingdom14 do not specifically or quantitatively define RISS. A further attempt toward consensus for defining RISS has been recently developed.8 We obtained pilot, prospective data on frequency and magnitude of RISS, according to 4 prespecified definitions and explored their relationship to clinical outcome.
Methods Study Population This pilot, prospective, observational study evaluated adult (older than age 18 years) AIS patients admitted to SUNY Downstate Medical Center in Brooklyn, New York. Eligibility included sudden onset of focal neurologic deficits suggestive of stroke presenting within 4.5 hours of symptoms; a noncontrast computed tomography scan negative for brain hemorrhage or any other brain lesion rather than ischemic stroke. All subjects/ designee provided informed consent. The protocol was approved by the SUNY Downstate Institutional Review Board. This study did not interfere with and/or delay the standard screening process aimed at identifying patients eligible for rt-PA. Any treatment decision was made independent of this study by the stroke treating physician.
ence between baseline NIHSS (b-NIHSS) and treatment decision NIHSS (td-NIHSS).
RISS and Outcome Definitions Prespecified definitions of RISS were (1) change in NIHSS score with 4-point or greater improvement (IMP) or (2) percent change in NIHSS score: 25% or greater or 50% or greater IMP and change in NIHSS score to nondisabling symptoms according to the previously published, The Re-examining Acute Eligibility for Thrombolysis (‘‘TREAT’’) Task Force Criteria definition.8 The TREAT Task Force comprised members of the original NINDS rt-PA Stroke Trial Steering Committee combined with other leaders in the field of stroke and emergency medicine. This Task Force was formed in an attempt to address the rationale and relevance of individual exclusion criteria to rt-PA and to determine if these criteria require clarifications and more precise definitions that could be implemented in the current clinical practice.8 Specifically, The TREAT Task Force sought to develop a clinically meaningful definition of RISS that may be used to guide stroke treating physicians’ decision to treat or not to treat with rt-PA. Consensus defined RISS as an exclusion criterion for rt-PA as improvement to a mild stroke such that any remaining deficits seem nondisabling.8 To be nondisabling, none of the following deficits should be present: (1) complete hemianopsia ($2 on the NIHSS question 3); (2) severe aphasia ($2 on NIHSS question #9); (3) visual or sensory extinction ($1 on NIHSS question #11); (4) any weakness limiting sustained effort against gravity ($2 on NIHSS question #6 or #7); (5) any deficits that lead to a total NIHSS .5; and (6) any remaining deficit considered potentially disabling in the view of the patient or the treating practitioner.8 Clinical outcome was defined using the modified Rankin Scale (mRS). Unfavorable outcome was defined as mRS greater than 1 at 90 days after stroke and at 7 days after stroke or discharge.
Statistical Analysis Protocol Subjects were administered serial assessments of stroke severity as measured by the National Institutes of Health Stroke Scale (NIHSS)15 performed every 20 6 5 minutes until a treatment decision (rt-PA vs no rt-PA) was made by the treating physician. RISS study team members were notified via a hospital-wide ED ‘‘code stroke’’ paging system. Baseline NIHSS (b-NIHSS) was defined as the first NIHSS score performed at patient arrival at the ED by 1 member of the RISS study team. All examiners were certified in the NIHSS. RISS was defined as a measurable improvement of neurologic deficits between 2 clinical evaluations performed at different times, and specifically, as the differ-
Descriptive statistics were used to evaluate age, sex, race, and other cardiovascular risk factors between those with RISS and no RISS and between those treated and not treated with rt-PA. The Fisher exact test for categorical variables and the Wilcoxon test for numeric data were used for comparison between groups. Exact logistic regression was used to model 90-day mRS score more than 1 from predictors age, b-NIHSS, time to NIHSS, rt-PA use, and (in separate models) 4 different definitions of RISS. P values and 95% confidence limits are based on Mid-P methods.16 A secondary analysis was performed to test the diagnostic utility of specific NIHSS score cut points from 1 to 10 to predict mRS more than 1 at 90 days. We generated
RAPIDLY IMPROVING STROKE SYMPTOMS
3
Table 1. Demographics and clinical characteristics of the study cohort
Patients’ Characteristics Age, y, mean (SD) Male, n (%) Race/ethnicity, n (%) White Black Asian Hispanic Hypertension, n (%) Diabetes, n (%) Hyperlipidemia, n (%) Prior stroke/TIA, n (%) History of heart disease, n (%) Family history of stroke, n (%) Cigarette smoking, n (%) Baseline NIHSS, median (25th-75th) NIHSS at time of treatment decision, median (25th-75th) RISS definitions, n (%) $4-point improvement $25% improvement $50% improvement TREAT-based mRS at discharge/day 7, n (%) 0-1 $2 Missing mRS at day 90, n (%) 0-1 $2 Missing mRS at day 90 with day 7 imputation if missing, n (%) 0-1 $2 Missing
Total sample (n 5 50)
No rt-PA (n 5 27)
Received rt-PA (n 5 23)
65 (15) 28 (56)
68 (13) 17 (63)
61 (16) 11 (48)
7 (14) 39 (78) 0 4 (8) 37 (74) 12 (24) 17 (34) 14 (28) 8 (16) 6 (12) 8 (16) 5 (2-11) 5 (2-9)
4 (15) 22 (81) 0 1 (4) 20 (74) 7 (26) 8 (30) 7 (26) 4 (15) 5 (19) 5 (19) 3 (2-5) 2 (1-4)
3 (13) 17 (74) 0 3 (13) 17 (74) 5 (22) 9 (39) 7 (30) 4 (17) 1 (4) 3 (13) 9 (6-14) 8 (6-13)
5 (10) 11 (22) 6 (12) 5 (10)
3 (11) 6 (22) 5 (18) 5 (18)
2 (9) 5 (22) 1 (4) —
25 (50) 25 (50) 0
17 (63) 10 (37) 0
8 (35) 15 (65) 0
23 (46) 23 (46) 4 (8)
17 (63) 7 (26) 3 (11)
6 (26) 16 (70) 1 (4)
25 (50) 25 (50) 0
18 (67) 9 (33) 0
7 (30) 16 (70) 0
P value .14 .39 .53
1.00 1.00 .56 .76 1.00 .20 .71 .0003 ,.0001 1.00 1.00 .20 — .09
.007
.02
Abbreviations: mRS, modified Rankin Scale; RISS, rapidly improving stroke symptoms; rt-PA, recombinant tissue plasminogen activator; SD, standard deviation; TIA, transient ischemic attack.
a receiver operating characteristic curve and calculated the area under the curve.
treated and not treated with rt-PA with the exception of b-NIHSS and td-NIHSS scores (Table 1).
Frequency of RISS
Results Our study population included 50 AIS patients. Table 1 summarizes the baseline demographics. Median b-NIHSS score was 5 (interquartile range [IQR], 2-11). Before a rt-PA treatment decision was made, at least 2 serial NIHSS scores were obtained in 47 (94%) subjects; treatment decision was based on the b-NIHSS in 3 (6%). Three serial, pretreatment decision NIHSS scores were obtained in 31 (62%) subjects. Median td-NIHSS score was 5 (IQR, 2-9). Twenty-three (46%) patients received rt-PA. There were no differences in baseline variables between those
In the entire cohort of 50 subjects there was a mean overall improvement of 1 6 3 (standard deviation) point on the NIHSS, with a median of 0 (IQR, 0-1) from the b-NIHSS to the td-NIHSS score. Fourteen of fifty (28%) subjects had improvement with a mean improvement in NIHSS of 4 6 3 points (median, 3; IQR, 2-5). Of those, 14 of 50 subjects showing any degree of improvement, 2 subjects improved only 1 point from b-NIHSS scores of 9 and 16, whereas 12 of 50 (24%) subjects improved according to at least 1 of 4 prespecified definitions with a mean of 4 6 3 points (median, 3; IQR, 2-5).
C. BALUCANI ET AL.
4
In 25 of 50 subjects (50%) NIHSS did not change, and in 8 of 50 (16%) subjects worsened by a mean of 2 6 1 points (median, 1; IQR, 1-1). Frequency of RISS varied based on definition used: 10% for a 4-point or greater IMP, 22% for 25% or greater IMP, 12% for 50% or greater IMP, and 10% for the TREAT definition. Twenty-three (46%) patients received rt-PA. For all nonTREAT–based RISS definitions there was no difference in treatment rates between RISS subjects (5 of 12, 42%) versus no RISS (18 of 38, 47%) subjects (P 5 .754).
Frequency of Clinical Outcome In the overall cohort, 42% of patients had good outcome (mRS, 0-1) at 90 days: 17 of 27 (63%) not treated with rt-PA versus 6 of 23 (26%) treated with rt-PA, P 5.02. There was a difference in td-NIHSS between those not given rt-PA (2; IQR, 1-4) and those given rt-PA (8; IQR, 6,13), P , .0001. Frequency of Clinical Outcome Based on RISS Definitions Good outcome was achieved in 1 of 5 (20%) subjects who had a 4-point or greater IMP, 6 of 11 (55%) of subjects with 25% or greater IMP, 5 of 6 (83%) for 50% or greater IMP, and 5 of 5 (100%) of those defined according to the TREAT definition. Those without RISS had a 50% (n 5 19 of 38) probability of good outcome at 90 days. There was no significant difference in frequency of good outcome at 90 days between subjects without RISS and those with RISS of a 4-point or greater IMP (P 5 .348), RISS of 25% or greater IMP (P 5 1.00), and RISS of 50% or greater IMP (P 5 .189). A borderline significant difference was found with the TREAT-based RISS definition (P 5 .050). None of the 4 definitions of RISS was an independent predictor of clinical outcome in the multivariate analysis (Table 2). Figure 1 illustrates the individual NIHSS data points from baseline to treatment decision and subsequent clinical outcomes for subjects with RISS by each of the 4 RISS definitions. Figure 2 shows that the sensitivity and specificity characteristics of NIHSS score cut points progress in a fairly smooth manner in the range of NIHSS score values between 1 and 11. We could not detect an obviously superior specific NIHSS cut point. The area under the curve is .83, indicating moderate predictive utility independent of cut point.
Discussion Our study supports the feasibility of obtaining a serial NIHSS assessment in the ED before treatment decision. We found that over a third of ‘‘timely eligible’’ AIS
Table 2. Probability of unfavorable outcome (mRS .1) at 90 days with any RISS definition
Predictors
Odds ratio (95% confidence intervals) P value
Model 1 (RISS $4 IMP) 1.29 (.07, 52.0) Model 2 (RISS $25% IMP) .61 (.06, 4.89) Model 3 (RISS $50% IMP) .09 (,.01, 2.97)
.746 .819 .165
Abbreviations: IMP, improvement; mRS, modified Rankin Scale; RISS, rapidly improving stroke symptoms; TREAT, The Re-examining Acute Eligibility for Thrombolysis. Odds ratios and P values were not calculated for model 4, which included the TREAT-based RISS definition, due to zero cell frequency. If day 90 mRS missing, imputed with mRS at discharge/ day 7 (n 5 4).
patients appear to have NIHSS changes while in the ED. Current clinical practice and guidelines12-14 do not include serial NIHSS scores to be determined before treatment, which could obscure the identification of these patients. Our pilot data provide, to our knowledge, the first prospective estimates of frequency of RISS according to multiple prespecified definitions. These estimates could help design larger studies to determine treatment effect of rt-PA in these cohorts of varying RISS frequencies. In the time from baseline to td-NIHSS, subjects with any prespecified RISS definition improved approximately 4 (standard deviation, 3) points on average. The frequency of RISS, depending on definition, ranged between 10% and 22%, somewhat lower than the prior retrospective studies of RISS patients excluded from rt-PA.1-4 Although patients with RISS were either treated or not treated with rt-PA in a similar frequency for RISS definitions of 4-point or greater IMP and 25% or greater IMP, only 1 of 6 with 50% or greater IMP was treated, reflecting the clinicians’ bias in not treating those who improved more significantly. We found that the significant 6-point difference in median td-NIHSS scores between the rt-PA and no rt-PA treatment groups likely influenced outcome to a greater degree than rt-PA treatment itself and is a confounder in our results. Our study further suggests that heterogeneity of presenting NIHSS scores and their changes over time are considerable, as can be visualized in Figure 1. This degree of heterogeneity could influence measures of improvement based on absolute or relative NIHSS stroke changes, likely leading to a failure to predict outcome. As all 5 subjects with TREAT-based RISS criteria were not treated with rt-PA as they all improved to a nondisabling state and all had a good outcome, this definition could help predict those AIS patients who are likely to have a good outcome. However, larger cohorts of AIS patients fulfilling TREAT criteria will be required to validate our preliminary results.
RAPIDLY IMPROVING STROKE SYMPTOMS
5
Figure 1. Time course of NIHSS scores and clinical outcomes in subjects with RISS. Plots of serial NIHSS total scores evaluated at the indicated intervals after symptom onset (time 0) in the 12 subjects with RISS. The 4 RISS definitions are illustrated in separate panels: (A) improvement (IMP) in NIHSS score is 4 point or greater; (B) NIHSS IMP of 25% or greater; (C) NIHSS IMP of 50% or greater; (D) TREAT-based RISS. In each plot, the first circle represents NIHSS score at baseline and the green-filled circle indicates the NIHSS score at the treatment decision time (tPA administered 5 red plots; no-tPA administered 5 blue plots). Dashed lines connect the serial NIHSS score plots to corresponding clinical outcomes measured as modified Rankin Scale (mRS, right bars) at 7 days or at hospital discharge (DC) and 90 days after stroke. Abbreviations: NIHSS, National Institutes of Health Stroke Scale; RISS, rapidly improving stroke symptoms; tPA, tissue plasminogen activator; TREAT, The Re-examining Acute Eligibility for Thrombolysis. (Color version of figure is available online.)
Figure 2. ROC curve for baseline NIHSSS as a predictor of modified Rankin Scale more than 1 at 90 days. Abbreviations: NIHSSS, National Institutes of Health Stroke Scale Score; ROC, receiver operating characteristic.
We did not find a specific, individual NIHSS score cut point for optimizing prediction of outcome. This may be because of our sample size or that for each cut point, there is a balanced trade-off between specificity and sensitivity. Limitations to our study include the small sample size, albeit a pilot study, and as such it is hypothesis generating; the potential lack of generalizability because of an urban, predominantly Caribbean black cohort; and the potential clinical bias of rt-PA treatment decision based on stroke severity. Also, these were not consecutive patients as this was a pilot study performed primarily during the day. Those arriving between 7 PM and 7 AM or on weekends were only screened and enrolled if members of the study personnel were available to come in within the time window required. The relatively lower b-NIHSS scores seen in our prospective, consecutive cohort are lower than those in acute stroke treatment trials but do reflect the lower population-based NIHSS of acute stroke patients.17 Finally, patients who were excluded (see Fig. 1) from rt-PA treatment because of their TREAT RISS definition had very low NIHSS scores at presentation. This may significantly limit the utility of this score.
C. BALUCANI ET AL.
6
It is possible that patients had more severe strokes at onset and improved to a mild, nondisabling stroke by the time they were first seen by our study team. Hence, the improvement could have occurred in the prehospital setting before their enrollment. The present study suggests a standardized procedure to identify AIS patients with RISS in the ED before treatment decision. However, the study does not provide information on evaluation of patients in the time between symptoms onset and ED admission. For a more sensitive identification and thorough characterization of RISS patients, additional larger studies are necessary to validate a standard assessment of these patients in the prehospital field, before ED admission. We further need to better define rate of improvement over time. This specific aspect could help differentiate between those patients who improve dramatically very early after symptoms onset (that is, likely to be TIAs and the type of patient who was excluded from the NINDS rt-PA Stroke Trials9), those that improve to a minimally disabling stroke, and those that do not improve over time and present with stable minor deficits. Clinical outcomes of these subgroups can differ substantially and the most appropriate treatment approaches need to be evaluated in controlled clinical trials. Acknowledgment: The authors acknowledge Drs Volodymyr Vulkanov, Sebina Bulic, Priyank Khandelwal, Qing Hao, and Benedict Tan for their assistance in conducting this project.
References 1. Nedeltchev K, Schwegler B, Haefeli T, et al. Outcome of stroke with mild or rapidly improving symptoms. Stroke 2007;38:2531-2535. 2. Barber PA, Zhang J, Demchuk AM, et al. Why are stroke patients excluded from t-PA therapy? An analysis of patient eligibility. Neurology 2001;56:1015-1020. 3. Eissa A, Krass I, Levi C, et al. Understanding the reasons behind the low utilisation of thrombolysis in stroke. Australas Med J 2013;6:152-167. 4. Katzan IL, Hammer MD, Hixson ED, et al. Utilization of intravenous tissue plasminogen activator for acute ischemic stroke. Arch Neurol 2004;61:346-350.
5. Willey JZ, Stillman J, Rivolta JA, et al. Too good to treat? Outcomes in patients not receiving thrombolysis due to mild deficits or rapidly improving symptoms. Int J Stroke 2012;7:202-206. 6. Smith EE, Fonarow GC, Reeves MJ, et al. Outcomes in mild or rapidly improving stroke not treated with intravenous recombinant tissue-type plasminogen activator: findings from Get With The Guidelines-Stroke. Stroke 2011;42:3110-3115. 7. Balucani C, Levine SR. Mild stroke and rapidly improving symptoms: it’s not always a happy ending. Stroke 2011;42:3005-3007. 8. Levine SR, Khatri P, Broderick JP, et al. Review, historical context, and clarifications of the NINDS rt-PA stroke trials exclusion criteria: Part 1: rapidly improving stroke symptoms. Re-examining Acute Eligibility for Thrombolysis (TREAT) Task Force. Stroke 2013;44:2500-2505. 9. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med 1995; 333:1581-1587. 10. Piran P, Balucani C, Ramkishun C, et al. Defining the undefined: rapidly improving stroke symptoms, a systematic review of the literature. Stroke 2014;45:A28. 11. Dirks M, Niessen LW, Koudstaal PJ, et al. Delphi panel on indications and contraindications for intravenous thrombolysis in acute ischaemic stroke. Intravenous thrombolysis in acute ischaemic stroke: from trial exclusion criteria to clinical contraindications. An international Delphi study. J Neurol Neurosurg Psych 2007;78:685-689. 12. Jauch EC, Saver JL, Adams HP Jr, et al. Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2013;44:870-947. 13. European Stroke Organisation (ESO) Executive Committee, ESO Writing Committee. Guidelines for management of ischaemic stroke and transient ischaemic attack 2008. Cerebrovasc Dis 2008;25:457-507. 14. National Institute for Health and Clinical Excellence. Stroke: the diagnosis and initial management of acute stroke and transient ischaemic attack. 2008 (clinical guideline 68) Available at: www.nice.org.uk/CG68. Accessed August 20, 2014. 15. Brott T, Adams HP Jr, Olinger CP, et al. Measurements of acute cerebral infarction: a clinical examination scale. Stroke 1989;20:864-870. 16. Barnard G. On alleged gains in power from lower p-values. Stat Med 1989;8:1469-1477. 17. Reeves M, Khoury J, Alwell K, et al. Distribution of National Institutes of Health stroke scale in the Cincinnati/Northern Kentucky Stroke Study. Stroke 2013; 44:3211-3213.
