- Email: [email protected]
Serum Creatinine Does Not Improve Early Classification of Ischemic Stroke Caused by Small Artery Occlusion
Serum Creatinine Does Not Improve Early Classification of Ischemic Stroke Caused by Small Artery Occlusion Adam B. Cohen,
MD,
Robert A. Taylor, MD, Steven R. Messé, Scott E. Kasner, MD
MD,
and
Background and Purpose: Accurate subtyping of ischemic stroke in the acute setting is a potentially important but very difficult task. Microvascular disease is a systemic disorder, and we hypothesized that impaired renal function, which is most commonly a result of microvascular disease, would correlate with the subtype of small-artery occlusion (SAO) in patients with acute ischemic stroke. Methods: This was a retrospective cohort study of consecutive patients with ischemic stroke. Clinical and laboratory data at admission were analyzed and compared with the final determination of stroke subtype. We determined whether serum creatinine level at admission was independently associated with final determination of the SAO stroke subtype. Results: There was no correlation between elevated baseline creatinine (ⱖ1.5 mg/dL) and SAO (odds ratio ⫽ 0.38, 95% confidence interval 0.01-2.98, P ⫽ .38). The most helpful predictor of a final SAO subtype was lacunar syndrome on initial examination, but only in patients with diabetes (positive predictive value ⫽ 67%, 95% confidence interval 30%-93%). Conclusions: Laboratory evidence of impaired renal function does not aid the early identification of the SAO stroke subtype in patients with acute stroke. In addition, an initial lacunar syndrome only effectively identifies SAO among patients with diabetes, not the general stroke population. Key Words: Stroke classification—microvascular injury— creatinine— renal disease—microvascular injury— creatinine—renal disease. © 2006 by National Stroke Association
Ischemic stroke is a heterogeneous disorder with many potential causes. The Trial of ORG 10172 in Acute Stroke Treatment (TOAST) criteria offer a widely used classification scheme for stroke, which includes 5 causes or subtypes: large-artery atherosclerosis (LAA), cardioembolism, smallartery occlusion (SAO), other/unusual, and cryptogenic.1 Specific stroke subtypes may have unique responses to
From the Department of Neurology, University of Pennsylvania Medical Center, Philadelphia, Pennsylvania. Received October 19, 2005; accepted December 9, 2005 Supported by National Institutes of Heath/National Institute of Neurological Disorders and Stroke K23-NS02147 and the InversoBaglivo Foundation (Dr. Kasner). Address correspondence to Scott E. Kasner, MD, Comprehensive Stroke Center, Department of Neurology, University of Pennsylvania Medical Center, 3400 Spruce St, Philadelphia, PA 19104. E-mail: [email protected]. 1052-3057/$—see front matter © 2006 by National Stroke Association doi:10.1016/j.jstrokecerebrovasdis.2005.12.005
96
acute and preventative therapies.2,3 Unfortunately, stroke subtype diagnoses are often inaccurate during the first 24 hours, and after a thorough and time-consuming clinical investigation, the initial subtype diagnosis is reclassified to another in approximately 40% of cases.4 Therefore, the problem of subtype misclassification precludes early subtype-specific therapy. About 15% to 25% of all ischemic strokes are a result of SAO.5,6 Patients with presumed SAO are often excluded from acute stroke trials, particularly investigations of neuroprotective agents,7 because this type of intervention is believed to offer greater influence on cortical gray matter rather than white matter in animal models of focal ischemia.8 Further, some clinicians choose not to use thrombolysis for patients with presumed SAO.9 Systematic exclusion of patients with presumed SAO is extremely problematic because of the inherent inaccuracy of early diagnosis. Methods to improve early identification of patients with SAO may be critical for the inclusion or exclusion of these patients in future trials.
Journal of Stroke and Cerebrovascular Diseases, Vol. 15, No. 3 (May-June), 2006: pp 96-100
CREATININE AND SMALL ARTERY OCCLUSION STROKE SUBTYPE
Impaired renal function, as evidenced by elevated serum creatinine levels, has been shown to be a marker for an increased risk of certain vascular diseases, including stroke.10,11 In addition, the vascular beds of the kidneys and the brain may have important similarities in regards to their susceptibility to pathology.11-13 The major risk factors for chronic kidney disease, hypertension (HTN) and diabetes mellitus (DM), are also commonly associated with stroke caused by SAO.14-16 Consequently, evidence of renal dysfunction at stroke presentation may be an important clue for early identification of this stroke subtype. Similarly, it is possible that exclusion of patients from acute stroke trials because of elevated creatinine levels at baseline might systematically eliminate strokes caused by SAO from those studies. We hypothesized that impaired renal function, as indicated by an elevated serum creatinine on admission, may be an important early diagnostic indicator of SAO. We also attempted to determine if other combinations of clinical features, risk factors, and laboratory information were predictive in determining SAO.
Methods This was a retrospective cohort study using data obtained from consecutive patients with ischemic stroke who presented to a university hospital during a 5-month period. The protocol was reviewed and approved by the institutional review board. Patients were identified by reviewing neurology admission logs. Patients with transient ischemic attack or primary hemorrhagic stroke were excluded. All records were reviewed for demographic data, baseline patient characteristics, risk factors, laboratory data, and other diagnostic studies, which were abstracted to an anonymous case report form. Each case was assigned to one of the 5 stroke subtypes according to the TOAST criteria.1 The primary rater (A. B. C.) determined stroke subtype while abstracting data from patient records. A second examiner (R. A. T.), who was blinded to the primary reviewer’s subtype, then evaluated the abstracted data forms and made an independent determination of subtype. When the first two examiners differed, a third examiner (S. E. K.), also blinded to the previous decisions, reviewed all documentation and made a final determination of subtype. In our primary analysis, patients with stroke were dichotomized as having impaired renal function if baseline creatinine was 1.5 mg/dL or higher, and normal renal function if baseline creatinine was less than 1.5 mg/dL. The outcome variable was stroke subtype SAO versus non-SAO, analyzed using Chi-square tests. Creatinine was also compared as a continuous variable across all 5 stroke subtypes using analysis of variance and comparing SAO to non-SAO strokes using t tests. We also
97
planned to determine if other clinical variables present at admission (e.g., clinical lacunar syndrome,17 HTN, or DM), using multivariable logistic regression, were associated with particular stroke subtypes. Positive predictive value (PPV) describes the probability that a patient with an elevated creatinine level will obtain a final diagnosis of SAO. If there is no association between creatinine level and SAO, then the PPV of a high creatinine will be only about 15% to 25%, the expected proportion of SAO in the overall stroke population. We expected a PPV of greater than 60%, which would be high enough to be clinically useful in identifying these patients. Sample size calculations depend on the proportion of patients with stroke who have high creatinine levels, which is not well established. Assuming ␣ of 0.05 and power of 80%, a total of 92 patients would be needed if the prevalence of renal dysfunction is 25%, and 230 patients if only 10%. Our original sample size was, therefore, set at 230. A preliminary analysis was performed when data had been collected on 104 patients, and the study was terminated at that time.
Results Data were collected for 104 patients with acute ischemic stroke: 48% were female and 54% African American, 32% Caucasian, 12% other/unknown, and 2% Hispanic. Ischemic stroke subtype was classified as large-artery atherothromboembolism in 16 (15.4%), cardioembolism in 28 (26.9%), SAO in 14 (13.5%), other known mechanism in 14 (13.4%), and cryptogenic in 32 (30.8%). The cryptogenic category included 20 patients with no identified cause and 12 with multiple competing potential causes. The distribution of baseline clinical factors is summarized in Table 1. The distribution of baseline serum creatinine levels among the subtypes is displayed in Fig 1. Among 104 patients with ischemic stroke, 15% had an elevated creatinine (ⱖ1.5 mg/dL) at admission. There was not a statistically significant difference in creatinine level (tested as a continuous variable) among the 5 subtypes (P ⫽ .88) (Table 1 and Fig 1) or when comparing between SAO and all non-SAO stroke subtypes (1.00 v 1.27 mg/ dL, respectively, P ⫽ .38) (Table 2). A diagnosis of SAO tended to be inversely associated with elevated creatinine SAO (odds ratio [OR] ⫽ 0.38, 95% confidence interval [CI] 0.01-2.98). Consequently, the PPV of high creatinine for SAO was only 6% (95% CI 0.1%-30%), which was less than the overall percentage of all patients with high creatinine (15%) (Table 3). Furthermore, high creatinine was not associated specifically with any stroke subtype: LAA compared with all non-LAA subtypes (OR ⫽ 1.33, P ⫽ .69 [95% CI 0.215.86]), cardioembolism versus noncardioembolism (OR ⫽ 1.80, P ⫽ .30 [95% CI 0.48-6.22]), other versus non-other (OR ⫽ 0.38, P ⫽ .36 [95% CI 0.01-2.98]), or cryptogenic
A.B. COHEN ET AL.
98
Table 1. Comparisons of risk factors according to stroke subtype (N ⫽ 104)
Age, y HTN DM Hyperlipidemia Smoking* Prior event† Cr (mg/dL) Cr ⱖ 1.5 mg/dL
Large (16)
Cardio (28)
Small (14)
Other (14)
Crypto (32)
P
61.8 ⫾ 9.8 63% 25% 25% 73% 25% 1.28 ⫾ 0.79 19%
69.4 ⫾ 13.6 82% 36% 29% 45% 36% 1.24 ⫾ 0.66 21%
64.6 ⫾ 11.9 93% 57% 14% 50% 43% 1.00 ⫾ 0.34 7%
53.9 ⫾ 14.2 64% 14% 29% 54% 57% 1.15 ⫾ 0.46 7%
67.4 ⫾ 15.0 66% 25% 28% 45% 44% 1.37 ⫾ 1.79 16%
.009 .18 .12 .87 .44 .46 .88 .67
Cr, Creatinine; DM, diabetes mellitus; HTN, hypertension. *Patients who have ever smoked in lifetime. †Includes stroke, transient ischemic attack, or myocardial infarction. P values reflect simultaneous comparison of all subtypes by 2 test for categorical variables or analysis of variance for continuous variables.
versus noncryptogenic (OR ⫽ 1.03, P ⫽ .96 [95% CI 0.25-3.60]). Similarly, high creatinine was not associated with any combinations, such as LAA and SAO versus other subtypes. There also did not appear to be any threshold level of serum creatinine that specifically correlated with any stroke subtype. In the comparison of risk factors (Table 1), only age significantly differed among the 5 subtypes. Older patients tended to fall more often into the cardioembolic group and younger patients into the other-cause group. However, when comparing SAO to non-SAO (Table 3), DM was significantly associated with SAO. Among patients with SAO, 57% (v 27% of the patients without SAO) had DM (P ⫽ .02). HTN also seemed more prevalent in the SAO group (93% v 70%, P ⫽ .07). As expected, patients with elevated creatinine tended to more often have HTN (OR ⫽ 2.94, 95% CI 0.60-28, P ⫽ .157) or diabetes (OR ⫽ 1.96, 95% CI 0.55-6.6, P ⫽ .221),
but these associations were not significant and CIs were wide, offering limited statistical power to detect such relationships. Clinical presentation with a classic lacunar syndrome was helpful in making an SAO diagnosis, but only in the diabetic population. For patients with DM and a lacunar syndrome at presentation, the PPV of SAO was 67% (95% CI 30%-99%) (Table 3). PPV of SAO did not improve if the patients also had HTN. Inclusion of serum creatinine did not improve identification of a final SAO subtype diagnosis.
Discussion In our study, there was no relationship between reduced kidney function and SAO, nor with any other stroke subtype. Instead, patients with SAO had relatively (but not significantly) lower levels of creatinine than other stroke subtypes. Only 7% of patients with SAO had elevated creatinine. Thus, it would seem unlikely that a larger patient population would demonstrate that elevated creatinine would be useful in the classification of
Table 2. Comparisons of risk factors with small-artery occlusion versus all other subtypes (N ⫽ 104)
Figure 1. Distribution of baseline serum creatinine levels among stroke subtypes. One extreme outlier is not shown: one patient with cryptogenic stroke had baseline serum creatinine of 11.0 mg/dL. Lower and upper boundaries of each box, 25th and 75th percentiles, respectively; central line, median; circles, outliers, defined as those scores that are further away from box by more than 1.5 times interquartile range of scores.
Age, y HTN DM Hyperlipidemia Smoking Prior event Cr (mg/dL) Cr ⱖ 1.5 mg/dL
SAO (14)
Non-SAO (90)
P
64.6 ⫾ 11.9 93% 57% 14% 50% 43% 1.00 ⫾ 0.34 7%
64.9 ⫾ 14.5 70% 27% 28% 52% 40% 1.27 ⫾ 1.18 17%
.93 .07 .02 .28 .90 .84 .38 .36
Cr, Creatinine; DM, diabetes mellitus; HTN, hypertension; SAO, small-artery occlusion.
CREATININE AND SMALL ARTERY OCCLUSION STROKE SUBTYPE
Table 3. Positive predictive value of baseline factors in identifying small-artery occlusion
Risk factor (Overall prevalence of SAO in this cohort) Lacunar syndrome DM HTN High creatinine Lacunar syndrome and DM Lacunar syndrome and HTN Lacunar syndrome and DM and HTN Lacunar syndrome and DM and HTN and high creatinine High creatinine and lacunar syndrome High creatinine and DM High creatinine and HTN DM and HTN
PPV, %
95% Confidence interval
13 37 25 17 6 67
8–22 16–62 11–43 9–27 0.1–30 30–93
44
20–70
67
30–93
50
1–99
25 14 7 29
1–81 0.4–58 0.2–34 13–47
DM, Diabetes mellitus; HTN, hypertension; PPV, positive predictive value; SAO, small artery occlusion. Clinical factors shown in bold font were significantly associated with improved PPV for final diagnosis of the SAO stroke subtype.
stroke caused by SAO, and the study was ended early. Serum creatinine at baseline is not useful for identifying which strokes are ultimately classified as the SAO subtype. Further, our results suggest that excluding patients with high serum creatinine levels from acute stroke trials does not lead to systematic underrepresentation or exclusion of SAO (or any other stroke subtype) in the ultimate study population. In this study, advanced age, diabetes, and HTN aided the early identification of SAO. Clinical presentation with a classic lacunar syndrome was somewhat helpful in diagnosing SAO (PPV ⫽ 37%). However, we found that it was most revealing among patients with diabetes (PPV ⫽ 67%). This may have implications for stroke subtype diagnosis in the acute setting if subtype-specific treatment becomes relevant.2,3 The addition of creatinine level did not improve the PPV of these clinical parameters. A potential drawback of our study was the use of serum creatinine levels rather than a measure of creatinine clearance, such as may be estimated by the wellknown Cockcroft-Gault formula that incorporates age, sex, and weight in addition to serum creatinine level.18 Unfortunately, weights were not recorded in any patients in this cohort while in the emergency department. In practice, actual weight measurement is difficult in patients with acute stroke because of their neurologic defi-
99
cits, the impetus to keep these patients supine, and the scarcity of bed-based scales in emergency departments.19 Further, weight estimation is notoriously inaccurate.19 Consequently, serum creatinine is realistically all that is available to the clinician as a marker of renal function in the acute setting. Similarly, it would be ideal if the cause of the renal impairment is known, but that was not feasible with the current study design. The small sample size is another potential limitation, as we terminated this study because the existing data were contrary to our initial hypothesis, leaving limited power and wide CIs for the secondary analyses. Further, there is the potential for misclassification of stroke subtypes using the TOAST criteria, which could bias our results toward the null, although we believe we minimized this bias through the use of multiple examiners. A large epidemiologic study showed that patients with elevated creatinine, independent of other vascular risk factors such as HTN, were at a significantly increased risk for cerebrovascular disease, with a lesser impact on the risk of coronary disease.11 Impairment of renal function has been proposed as a potent marker of increased risk of vascular disease, and serum creatinine in particular could be a sensitive indicator of injury to the kidney and end organs such as the brain and the heart.11 Breakdown and dysfunction of the cerebral small vessel endothelium, with accompanying leakage of plasma components, contributes to SAO and leukoaraiosis,20 and renal insufficiency appears to contribute to small cerebral vessel dysfunction secondary to permeability irregularities.21 Elevated serum creatinine was shown to be associated with lacunes defined by magnetic resonance imaging (but not clinical stroke) in elderly patients.22 We, therefore, hypothesized that this relationship would be useful for distinguishing SAO from other stroke subtypes. However, we found no relationship between serum creatinine and any specific stroke subtype. Additional research is needed to determine the clinical relevance of the relationships between renal and cerebrovascular disease.
References 1. Adams HP, Bendixen BH, Kappelle LJ, et al, and the TOAST Investigators. Classification of subtype of acute ischemic stroke: Definitions for use in a multicenter clinical trial. Stroke 1993;24:35-41. 2. Adams HP Jr, Bendixen BH, Leira E, et al. Antithrombotic treatment of ischemic stroke among patients with occlusion or severe stenosis of the internal carotid artery: A report of the trial of org 10172 in acute stroke treatment (TOAST). Neurology 1999;53:122-125. 3. Kasner SE, Kimmel SE. Accuracy of initial stroke subtype diagnosis: A decision analysis. Cerebrovasc Dis 2000;10: 18-24. 4. Madden KP, Karanjia PN, Adams HP, et al, and the TOAST Investigators. Accuracy of initial stroke subtype
A.B. COHEN ET AL.
100
5.
6.
7.
8.
9.
10.
11.
12.
diagnosis in the TOAST study. Neurology 1995;45: 1975-1979. Goldstein LB. Accuracy of ICD-9-CM coding for the identification of patients with acute ischemic stroke: Effect of modifier codes. Stroke 1998;29:1602-1604. Petty GW, Brown RD Jr, Whisnant JP, et al. Ischemic stroke subtypes: A population-based study of functional outcome, survival, and recurrence. Stroke 2000;31:10621068. Davalos A, Castillo J, Alvarez-Sabin J, et al. Oral citicoline in acute ischemic stroke: An individual patient data pooling analysis of clinical trials. Stroke 2002;33:2850-2857. Gladstone DJ, Black SE, Hakim AM, Heart, Stroke Foundation of Ontario Centre of Excellence in Stroke R. Toward wisdom from failure: Lessons from neuroprotective stroke trials and new therapeutic directions. Stroke 2002;33:2123-2136. Caplan LR, Mohr JP, Kistler JP, et al. Should thrombolytic therapy be the first-line treatment for acute ischemic stroke? Thrombolysis–not a panacea for ischemic stroke. N Engl J Med 1997;337:1309-1313. O’Hare AM, Glidden DV, Fox CS, et al. High prevalence of peripheral arterial disease in persons with renal insufficiency: Results from the national health and nutrition examination survey 1999-2000. Circulation 2004;109:320-323. Wannamethee SG, Shaper AG, Perry IJ. Serum creatinine concentration and risk of cardiovascular disease: A possible marker for increased risk of stroke. Stroke 1997;28: 557-563. Cooper ME, Bonnet F, Oldfield M, et al. Mechanisms of diabetic vasculopathy: An overview. Am J Hypertens 2001;14:475-486.
13. Endemann DH, Schiffrin EL. Endothelial dysfunction. J Am Soc Nephrol 2004;15:1983-1992. 14. Petty GW, Brown RD Jr, Whisnant JP, et al. Ischemic stroke subtypes: A population-based study of incidence and risk factors. Stroke 1999;30:2513-2516. 15. Arboix A, Morcillo C, Garcia-Eroles L, et al. Different vascular risk factor profiles in ischemic stroke subtypes: A study from the “Sagrat Cor Hospital of Barcelona stroke registry.” Acta Neurol Scand 2000;102:264-270. 16. Kolominsky-Rabas PL, Weber M, Gefeller O, et al. Epidemiology of ischemic stroke subtypes according to TOAST criteria: Incidence, recurrence, and long-term survival in ischemic stroke subtypes, a population-based study. Stroke 2001;32:2735-2740. 17. Fisher CM. Lacunar strokes and infarcts: A review. Neurology 1982;32:871-876. 18. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31-41. 19. Messe SR, Tanne D, Demchuk AM, et al. Dosing errors may impact the risk of rt-pa for stroke: The multicenter rt-pa acute stroke survey. J Stroke Cerebrovasc Dis 2004; 13:35-40. 20. Wardlaw JM, Sandercock PA, Dennis MS, et al. Is breakdown of the blood-brain barrier responsible for lacunar stroke, leukoaraiosis, and dementia? Stroke 2003;34:806812. 21. Lammie GA, Brannan F, Slattery J, et al. Nonhypertensive cerebral small-vessel disease: An autopsy study. Stroke 1997;28:2222-2229. 22. Longstreth WT Jr, Bernick C, Manolio TA, et al. Lacunar infarcts defined by magnetic resonance imaging of 3660 elderly people: The cardiovascular health study. Arch Neurol 1998;55:1217-1225.
MD,
Robert A. Taylor, MD, Steven R. Messé, Scott E. Kasner, MD
MD,
and
Background and Purpose: Accurate subtyping of ischemic stroke in the acute setting is a potentially important but very difficult task. Microvascular disease is a systemic disorder, and we hypothesized that impaired renal function, which is most commonly a result of microvascular disease, would correlate with the subtype of small-artery occlusion (SAO) in patients with acute ischemic stroke. Methods: This was a retrospective cohort study of consecutive patients with ischemic stroke. Clinical and laboratory data at admission were analyzed and compared with the final determination of stroke subtype. We determined whether serum creatinine level at admission was independently associated with final determination of the SAO stroke subtype. Results: There was no correlation between elevated baseline creatinine (ⱖ1.5 mg/dL) and SAO (odds ratio ⫽ 0.38, 95% confidence interval 0.01-2.98, P ⫽ .38). The most helpful predictor of a final SAO subtype was lacunar syndrome on initial examination, but only in patients with diabetes (positive predictive value ⫽ 67%, 95% confidence interval 30%-93%). Conclusions: Laboratory evidence of impaired renal function does not aid the early identification of the SAO stroke subtype in patients with acute stroke. In addition, an initial lacunar syndrome only effectively identifies SAO among patients with diabetes, not the general stroke population. Key Words: Stroke classification—microvascular injury— creatinine— renal disease—microvascular injury— creatinine—renal disease. © 2006 by National Stroke Association
Ischemic stroke is a heterogeneous disorder with many potential causes. The Trial of ORG 10172 in Acute Stroke Treatment (TOAST) criteria offer a widely used classification scheme for stroke, which includes 5 causes or subtypes: large-artery atherosclerosis (LAA), cardioembolism, smallartery occlusion (SAO), other/unusual, and cryptogenic.1 Specific stroke subtypes may have unique responses to
From the Department of Neurology, University of Pennsylvania Medical Center, Philadelphia, Pennsylvania. Received October 19, 2005; accepted December 9, 2005 Supported by National Institutes of Heath/National Institute of Neurological Disorders and Stroke K23-NS02147 and the InversoBaglivo Foundation (Dr. Kasner). Address correspondence to Scott E. Kasner, MD, Comprehensive Stroke Center, Department of Neurology, University of Pennsylvania Medical Center, 3400 Spruce St, Philadelphia, PA 19104. E-mail: [email protected]. 1052-3057/$—see front matter © 2006 by National Stroke Association doi:10.1016/j.jstrokecerebrovasdis.2005.12.005
96
acute and preventative therapies.2,3 Unfortunately, stroke subtype diagnoses are often inaccurate during the first 24 hours, and after a thorough and time-consuming clinical investigation, the initial subtype diagnosis is reclassified to another in approximately 40% of cases.4 Therefore, the problem of subtype misclassification precludes early subtype-specific therapy. About 15% to 25% of all ischemic strokes are a result of SAO.5,6 Patients with presumed SAO are often excluded from acute stroke trials, particularly investigations of neuroprotective agents,7 because this type of intervention is believed to offer greater influence on cortical gray matter rather than white matter in animal models of focal ischemia.8 Further, some clinicians choose not to use thrombolysis for patients with presumed SAO.9 Systematic exclusion of patients with presumed SAO is extremely problematic because of the inherent inaccuracy of early diagnosis. Methods to improve early identification of patients with SAO may be critical for the inclusion or exclusion of these patients in future trials.
Journal of Stroke and Cerebrovascular Diseases, Vol. 15, No. 3 (May-June), 2006: pp 96-100
CREATININE AND SMALL ARTERY OCCLUSION STROKE SUBTYPE
Impaired renal function, as evidenced by elevated serum creatinine levels, has been shown to be a marker for an increased risk of certain vascular diseases, including stroke.10,11 In addition, the vascular beds of the kidneys and the brain may have important similarities in regards to their susceptibility to pathology.11-13 The major risk factors for chronic kidney disease, hypertension (HTN) and diabetes mellitus (DM), are also commonly associated with stroke caused by SAO.14-16 Consequently, evidence of renal dysfunction at stroke presentation may be an important clue for early identification of this stroke subtype. Similarly, it is possible that exclusion of patients from acute stroke trials because of elevated creatinine levels at baseline might systematically eliminate strokes caused by SAO from those studies. We hypothesized that impaired renal function, as indicated by an elevated serum creatinine on admission, may be an important early diagnostic indicator of SAO. We also attempted to determine if other combinations of clinical features, risk factors, and laboratory information were predictive in determining SAO.
Methods This was a retrospective cohort study using data obtained from consecutive patients with ischemic stroke who presented to a university hospital during a 5-month period. The protocol was reviewed and approved by the institutional review board. Patients were identified by reviewing neurology admission logs. Patients with transient ischemic attack or primary hemorrhagic stroke were excluded. All records were reviewed for demographic data, baseline patient characteristics, risk factors, laboratory data, and other diagnostic studies, which were abstracted to an anonymous case report form. Each case was assigned to one of the 5 stroke subtypes according to the TOAST criteria.1 The primary rater (A. B. C.) determined stroke subtype while abstracting data from patient records. A second examiner (R. A. T.), who was blinded to the primary reviewer’s subtype, then evaluated the abstracted data forms and made an independent determination of subtype. When the first two examiners differed, a third examiner (S. E. K.), also blinded to the previous decisions, reviewed all documentation and made a final determination of subtype. In our primary analysis, patients with stroke were dichotomized as having impaired renal function if baseline creatinine was 1.5 mg/dL or higher, and normal renal function if baseline creatinine was less than 1.5 mg/dL. The outcome variable was stroke subtype SAO versus non-SAO, analyzed using Chi-square tests. Creatinine was also compared as a continuous variable across all 5 stroke subtypes using analysis of variance and comparing SAO to non-SAO strokes using t tests. We also
97
planned to determine if other clinical variables present at admission (e.g., clinical lacunar syndrome,17 HTN, or DM), using multivariable logistic regression, were associated with particular stroke subtypes. Positive predictive value (PPV) describes the probability that a patient with an elevated creatinine level will obtain a final diagnosis of SAO. If there is no association between creatinine level and SAO, then the PPV of a high creatinine will be only about 15% to 25%, the expected proportion of SAO in the overall stroke population. We expected a PPV of greater than 60%, which would be high enough to be clinically useful in identifying these patients. Sample size calculations depend on the proportion of patients with stroke who have high creatinine levels, which is not well established. Assuming ␣ of 0.05 and power of 80%, a total of 92 patients would be needed if the prevalence of renal dysfunction is 25%, and 230 patients if only 10%. Our original sample size was, therefore, set at 230. A preliminary analysis was performed when data had been collected on 104 patients, and the study was terminated at that time.
Results Data were collected for 104 patients with acute ischemic stroke: 48% were female and 54% African American, 32% Caucasian, 12% other/unknown, and 2% Hispanic. Ischemic stroke subtype was classified as large-artery atherothromboembolism in 16 (15.4%), cardioembolism in 28 (26.9%), SAO in 14 (13.5%), other known mechanism in 14 (13.4%), and cryptogenic in 32 (30.8%). The cryptogenic category included 20 patients with no identified cause and 12 with multiple competing potential causes. The distribution of baseline clinical factors is summarized in Table 1. The distribution of baseline serum creatinine levels among the subtypes is displayed in Fig 1. Among 104 patients with ischemic stroke, 15% had an elevated creatinine (ⱖ1.5 mg/dL) at admission. There was not a statistically significant difference in creatinine level (tested as a continuous variable) among the 5 subtypes (P ⫽ .88) (Table 1 and Fig 1) or when comparing between SAO and all non-SAO stroke subtypes (1.00 v 1.27 mg/ dL, respectively, P ⫽ .38) (Table 2). A diagnosis of SAO tended to be inversely associated with elevated creatinine SAO (odds ratio [OR] ⫽ 0.38, 95% confidence interval [CI] 0.01-2.98). Consequently, the PPV of high creatinine for SAO was only 6% (95% CI 0.1%-30%), which was less than the overall percentage of all patients with high creatinine (15%) (Table 3). Furthermore, high creatinine was not associated specifically with any stroke subtype: LAA compared with all non-LAA subtypes (OR ⫽ 1.33, P ⫽ .69 [95% CI 0.215.86]), cardioembolism versus noncardioembolism (OR ⫽ 1.80, P ⫽ .30 [95% CI 0.48-6.22]), other versus non-other (OR ⫽ 0.38, P ⫽ .36 [95% CI 0.01-2.98]), or cryptogenic
A.B. COHEN ET AL.
98
Table 1. Comparisons of risk factors according to stroke subtype (N ⫽ 104)
Age, y HTN DM Hyperlipidemia Smoking* Prior event† Cr (mg/dL) Cr ⱖ 1.5 mg/dL
Large (16)
Cardio (28)
Small (14)
Other (14)
Crypto (32)
P
61.8 ⫾ 9.8 63% 25% 25% 73% 25% 1.28 ⫾ 0.79 19%
69.4 ⫾ 13.6 82% 36% 29% 45% 36% 1.24 ⫾ 0.66 21%
64.6 ⫾ 11.9 93% 57% 14% 50% 43% 1.00 ⫾ 0.34 7%
53.9 ⫾ 14.2 64% 14% 29% 54% 57% 1.15 ⫾ 0.46 7%
67.4 ⫾ 15.0 66% 25% 28% 45% 44% 1.37 ⫾ 1.79 16%
.009 .18 .12 .87 .44 .46 .88 .67
Cr, Creatinine; DM, diabetes mellitus; HTN, hypertension. *Patients who have ever smoked in lifetime. †Includes stroke, transient ischemic attack, or myocardial infarction. P values reflect simultaneous comparison of all subtypes by 2 test for categorical variables or analysis of variance for continuous variables.
versus noncryptogenic (OR ⫽ 1.03, P ⫽ .96 [95% CI 0.25-3.60]). Similarly, high creatinine was not associated with any combinations, such as LAA and SAO versus other subtypes. There also did not appear to be any threshold level of serum creatinine that specifically correlated with any stroke subtype. In the comparison of risk factors (Table 1), only age significantly differed among the 5 subtypes. Older patients tended to fall more often into the cardioembolic group and younger patients into the other-cause group. However, when comparing SAO to non-SAO (Table 3), DM was significantly associated with SAO. Among patients with SAO, 57% (v 27% of the patients without SAO) had DM (P ⫽ .02). HTN also seemed more prevalent in the SAO group (93% v 70%, P ⫽ .07). As expected, patients with elevated creatinine tended to more often have HTN (OR ⫽ 2.94, 95% CI 0.60-28, P ⫽ .157) or diabetes (OR ⫽ 1.96, 95% CI 0.55-6.6, P ⫽ .221),
but these associations were not significant and CIs were wide, offering limited statistical power to detect such relationships. Clinical presentation with a classic lacunar syndrome was helpful in making an SAO diagnosis, but only in the diabetic population. For patients with DM and a lacunar syndrome at presentation, the PPV of SAO was 67% (95% CI 30%-99%) (Table 3). PPV of SAO did not improve if the patients also had HTN. Inclusion of serum creatinine did not improve identification of a final SAO subtype diagnosis.
Discussion In our study, there was no relationship between reduced kidney function and SAO, nor with any other stroke subtype. Instead, patients with SAO had relatively (but not significantly) lower levels of creatinine than other stroke subtypes. Only 7% of patients with SAO had elevated creatinine. Thus, it would seem unlikely that a larger patient population would demonstrate that elevated creatinine would be useful in the classification of
Table 2. Comparisons of risk factors with small-artery occlusion versus all other subtypes (N ⫽ 104)
Figure 1. Distribution of baseline serum creatinine levels among stroke subtypes. One extreme outlier is not shown: one patient with cryptogenic stroke had baseline serum creatinine of 11.0 mg/dL. Lower and upper boundaries of each box, 25th and 75th percentiles, respectively; central line, median; circles, outliers, defined as those scores that are further away from box by more than 1.5 times interquartile range of scores.
Age, y HTN DM Hyperlipidemia Smoking Prior event Cr (mg/dL) Cr ⱖ 1.5 mg/dL
SAO (14)
Non-SAO (90)
P
64.6 ⫾ 11.9 93% 57% 14% 50% 43% 1.00 ⫾ 0.34 7%
64.9 ⫾ 14.5 70% 27% 28% 52% 40% 1.27 ⫾ 1.18 17%
.93 .07 .02 .28 .90 .84 .38 .36
Cr, Creatinine; DM, diabetes mellitus; HTN, hypertension; SAO, small-artery occlusion.
CREATININE AND SMALL ARTERY OCCLUSION STROKE SUBTYPE
Table 3. Positive predictive value of baseline factors in identifying small-artery occlusion
Risk factor (Overall prevalence of SAO in this cohort) Lacunar syndrome DM HTN High creatinine Lacunar syndrome and DM Lacunar syndrome and HTN Lacunar syndrome and DM and HTN Lacunar syndrome and DM and HTN and high creatinine High creatinine and lacunar syndrome High creatinine and DM High creatinine and HTN DM and HTN
PPV, %
95% Confidence interval
13 37 25 17 6 67
8–22 16–62 11–43 9–27 0.1–30 30–93
44
20–70
67
30–93
50
1–99
25 14 7 29
1–81 0.4–58 0.2–34 13–47
DM, Diabetes mellitus; HTN, hypertension; PPV, positive predictive value; SAO, small artery occlusion. Clinical factors shown in bold font were significantly associated with improved PPV for final diagnosis of the SAO stroke subtype.
stroke caused by SAO, and the study was ended early. Serum creatinine at baseline is not useful for identifying which strokes are ultimately classified as the SAO subtype. Further, our results suggest that excluding patients with high serum creatinine levels from acute stroke trials does not lead to systematic underrepresentation or exclusion of SAO (or any other stroke subtype) in the ultimate study population. In this study, advanced age, diabetes, and HTN aided the early identification of SAO. Clinical presentation with a classic lacunar syndrome was somewhat helpful in diagnosing SAO (PPV ⫽ 37%). However, we found that it was most revealing among patients with diabetes (PPV ⫽ 67%). This may have implications for stroke subtype diagnosis in the acute setting if subtype-specific treatment becomes relevant.2,3 The addition of creatinine level did not improve the PPV of these clinical parameters. A potential drawback of our study was the use of serum creatinine levels rather than a measure of creatinine clearance, such as may be estimated by the wellknown Cockcroft-Gault formula that incorporates age, sex, and weight in addition to serum creatinine level.18 Unfortunately, weights were not recorded in any patients in this cohort while in the emergency department. In practice, actual weight measurement is difficult in patients with acute stroke because of their neurologic defi-
99
cits, the impetus to keep these patients supine, and the scarcity of bed-based scales in emergency departments.19 Further, weight estimation is notoriously inaccurate.19 Consequently, serum creatinine is realistically all that is available to the clinician as a marker of renal function in the acute setting. Similarly, it would be ideal if the cause of the renal impairment is known, but that was not feasible with the current study design. The small sample size is another potential limitation, as we terminated this study because the existing data were contrary to our initial hypothesis, leaving limited power and wide CIs for the secondary analyses. Further, there is the potential for misclassification of stroke subtypes using the TOAST criteria, which could bias our results toward the null, although we believe we minimized this bias through the use of multiple examiners. A large epidemiologic study showed that patients with elevated creatinine, independent of other vascular risk factors such as HTN, were at a significantly increased risk for cerebrovascular disease, with a lesser impact on the risk of coronary disease.11 Impairment of renal function has been proposed as a potent marker of increased risk of vascular disease, and serum creatinine in particular could be a sensitive indicator of injury to the kidney and end organs such as the brain and the heart.11 Breakdown and dysfunction of the cerebral small vessel endothelium, with accompanying leakage of plasma components, contributes to SAO and leukoaraiosis,20 and renal insufficiency appears to contribute to small cerebral vessel dysfunction secondary to permeability irregularities.21 Elevated serum creatinine was shown to be associated with lacunes defined by magnetic resonance imaging (but not clinical stroke) in elderly patients.22 We, therefore, hypothesized that this relationship would be useful for distinguishing SAO from other stroke subtypes. However, we found no relationship between serum creatinine and any specific stroke subtype. Additional research is needed to determine the clinical relevance of the relationships between renal and cerebrovascular disease.
References 1. Adams HP, Bendixen BH, Kappelle LJ, et al, and the TOAST Investigators. Classification of subtype of acute ischemic stroke: Definitions for use in a multicenter clinical trial. Stroke 1993;24:35-41. 2. Adams HP Jr, Bendixen BH, Leira E, et al. Antithrombotic treatment of ischemic stroke among patients with occlusion or severe stenosis of the internal carotid artery: A report of the trial of org 10172 in acute stroke treatment (TOAST). Neurology 1999;53:122-125. 3. Kasner SE, Kimmel SE. Accuracy of initial stroke subtype diagnosis: A decision analysis. Cerebrovasc Dis 2000;10: 18-24. 4. Madden KP, Karanjia PN, Adams HP, et al, and the TOAST Investigators. Accuracy of initial stroke subtype
A.B. COHEN ET AL.
100
5.
6.
7.
8.
9.
10.
11.
12.
diagnosis in the TOAST study. Neurology 1995;45: 1975-1979. Goldstein LB. Accuracy of ICD-9-CM coding for the identification of patients with acute ischemic stroke: Effect of modifier codes. Stroke 1998;29:1602-1604. Petty GW, Brown RD Jr, Whisnant JP, et al. Ischemic stroke subtypes: A population-based study of functional outcome, survival, and recurrence. Stroke 2000;31:10621068. Davalos A, Castillo J, Alvarez-Sabin J, et al. Oral citicoline in acute ischemic stroke: An individual patient data pooling analysis of clinical trials. Stroke 2002;33:2850-2857. Gladstone DJ, Black SE, Hakim AM, Heart, Stroke Foundation of Ontario Centre of Excellence in Stroke R. Toward wisdom from failure: Lessons from neuroprotective stroke trials and new therapeutic directions. Stroke 2002;33:2123-2136. Caplan LR, Mohr JP, Kistler JP, et al. Should thrombolytic therapy be the first-line treatment for acute ischemic stroke? Thrombolysis–not a panacea for ischemic stroke. N Engl J Med 1997;337:1309-1313. O’Hare AM, Glidden DV, Fox CS, et al. High prevalence of peripheral arterial disease in persons with renal insufficiency: Results from the national health and nutrition examination survey 1999-2000. Circulation 2004;109:320-323. Wannamethee SG, Shaper AG, Perry IJ. Serum creatinine concentration and risk of cardiovascular disease: A possible marker for increased risk of stroke. Stroke 1997;28: 557-563. Cooper ME, Bonnet F, Oldfield M, et al. Mechanisms of diabetic vasculopathy: An overview. Am J Hypertens 2001;14:475-486.
13. Endemann DH, Schiffrin EL. Endothelial dysfunction. J Am Soc Nephrol 2004;15:1983-1992. 14. Petty GW, Brown RD Jr, Whisnant JP, et al. Ischemic stroke subtypes: A population-based study of incidence and risk factors. Stroke 1999;30:2513-2516. 15. Arboix A, Morcillo C, Garcia-Eroles L, et al. Different vascular risk factor profiles in ischemic stroke subtypes: A study from the “Sagrat Cor Hospital of Barcelona stroke registry.” Acta Neurol Scand 2000;102:264-270. 16. Kolominsky-Rabas PL, Weber M, Gefeller O, et al. Epidemiology of ischemic stroke subtypes according to TOAST criteria: Incidence, recurrence, and long-term survival in ischemic stroke subtypes, a population-based study. Stroke 2001;32:2735-2740. 17. Fisher CM. Lacunar strokes and infarcts: A review. Neurology 1982;32:871-876. 18. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31-41. 19. Messe SR, Tanne D, Demchuk AM, et al. Dosing errors may impact the risk of rt-pa for stroke: The multicenter rt-pa acute stroke survey. J Stroke Cerebrovasc Dis 2004; 13:35-40. 20. Wardlaw JM, Sandercock PA, Dennis MS, et al. Is breakdown of the blood-brain barrier responsible for lacunar stroke, leukoaraiosis, and dementia? Stroke 2003;34:806812. 21. Lammie GA, Brannan F, Slattery J, et al. Nonhypertensive cerebral small-vessel disease: An autopsy study. Stroke 1997;28:2222-2229. 22. Longstreth WT Jr, Bernick C, Manolio TA, et al. Lacunar infarcts defined by magnetic resonance imaging of 3660 elderly people: The cardiovascular health study. Arch Neurol 1998;55:1217-1225.