Box plot showing multivariate adjusted median diffusion-weighted magnetic resonance imaging (DWI) infarct volumes (solid bar), interquartile range (bar width), and minimum and maximum values (whiskers) across the range of serum calcium level quartiles. *Diffusion-weighted magnetic resonance imaging volume of quartile 1 greater than that of quartiles 2, 3, and 4 (P ≤ .05).
Buck BH, Liebeskind DS, Saver JL, Bang OY, Starkman S, Ali LK, Kim D, Villablanca JP, Salamon N, Yun SW, Razinia T, Ovbiagele B. Association of Higher Serum Calcium Levels With Smaller Infarct Volumes in Acute Ischemic Stroke. Arch Neurol. 2007;64(9):1287-1291. doi:10.1001/archneur.64.9.1287
Elevated serum calcium levels at admission in patients with stroke have been associated with less severe clinical deficits and with better outcomes; however, the relationship between serum calcium levels and volumetric measurement of cerebral infarct size on neuroimaging has not been studied, to our knowledge.
To assess the relationship between serum calcium levels at admission and initial diffusion-weighted magnetic resonance imaging (DWI) infarct volumes among patients with acute ischemic stroke.
Secondary analysis of prospectively collected hospital quality improvement data.
Tertiary university hospital.
One hundred seventy-three consecutive patients with acute ischemic stroke initially seen within 24 hours of the last known well time.
Main Outcome Measures
Total serum calcium levels were measured on admission and were collapsed into quartiles. The DWI lesions were outlined using a semiautomated threshold technique. The relationship between serum calcium level quartiles and DWI infarct volumes was examined using multivariate quartile regression analysis.
One hundred seventy-three patients (mean age, 70.3 years [age range, 24-100 years]; median National Institutes of Health Stroke Scale score, 4 [range, 0-38]) met the study criteria. The median DWI infarct volumes for the serum calcium level quartiles (lowest to highest quartile) were 9.42, 2.11, 1.03, and 3.68 mL. The median DWI infarct volume in the lowest serum calcium level quartile was larger than that in the other 3 quartiles (P < .005). After multivariate analysis, the median adjusted DWI infarct volumes for the serum calcium level quartiles (lowest to highest) were 8.9, 5.8, 4.5, and 3.8 mL. The median adjusted DWI infarct volume in the lowest serum calcium level quartile was statistically significantly larger than that in the other 3 quartiles (P < .05).
Higher serum calcium levels at admission are associated with smaller cerebral infarct volumes among patients with acute ischemic stroke. These results suggest that serum calcium level may serve as a clinical prognosticator following stroke and may be a potential therapeutic target for improving stroke outcome.
In ischemic stroke, excessive intracellular serum calcium accumulation triggers a cascade of cytotoxic events that lead to the activation of enzymes involved in cell death.1 In preclinical models, low extracellular serum calcium levels paradoxically enhance this overloading of intracellular serum calcium and potentiate cell death.2 Whether serum calcium levels affect serum calcium level–dependent excitotoxic pathways in the setting of human acute cerebral ischemia remains unclear, but mounting data indicate that higher serum calcium levels at admission are associated with better clinical outcomes after ischemic stroke.3,4
To our knowledge, no prior study has examined the relationship between serum calcium levels and actual volumetric measurements of cerebral infarct size on brain imaging. Infarct size is a key determinant of clinical outcome from stroke.5 In this study, we hypothesized that elevated serum calcium levels at admission would be independently associated with smaller diffusion-weighted magnetic resonance (MR) imaging (DWI) lesions, indicating advanced tissue bioenergetic compromise.
This is a secondary analysis of data that were collected prospectively during 18 months beginning January 1, 2004, on consecutive ischemic stroke admissions to a university hospital stroke program as part of an ongoing hospital quality improvement project.6 The inclusion criteria for enrollment into the study were age 18 years or older, diagnosis of acute ischemic stroke, serum total calcium level collected on admission, and multimodal MR imaging performed within 24 hours of the last known well time. Patients with contraindications to MR imaging were excluded, as were those who were initially seen or underwent imaging more than 24 hours from the last known well time. The study was approved by the hospital institutional review board.
Patients underwent imaging before receiving any reperfusion therapy using a 1.5-T scanner (Siemens Vision; Siemens Medical Systems, Erlangen, Germany) based on a protocol detailed previously7 that included DWI, perfusion-weighted imaging, gradient-recalled echo examination, and fluid-attenuated inversion recovery imaging. The DWI was performed using 2 levels of diffusion sensitization (b value, 0 and 1000 s/mm2) with the following variables: 5-mm section thickness, no gap, and 17 to 20 sections. The DWI infarct volumes were measured using available software (Medical Image Processing, Analysis and Visualization, version 3.0; National Institutes of Health, Bethesda, Maryland). Image raters (B.H.B. and O.Y.B.) outlined regions of acute diffusion abnormality on the image at 1000 s/mm2, consulting apparent diffusion coefficient and fluid-attenuated inversion recovery imaging sequences to distinguish acute from nonacute diffusion change. Acute diffusion lesions were defined on a section-by-section basis using a semiautomatic threshold approach by one of us (B.H.B.) blinded to all clinical information. Infarct volumes were calculated by multiplying section thickness by the total lesion area. To assess interrater reliability, infarct volumes were measured by a second rater (O.Y.B.) on a randomly selected subset of 19 patients.
Serum calcium levels were collapsed into quartiles. Baseline demographics and clinical characteristics were compared across quartiles using χ2 test for percentages and Kruskal-Wallis rank sum test for medians. Interrater reliability for DWI measurements was assessed by calculation of the intraclass correlation coefficient, with greater than 0.80 set as the threshold for good agreement. The relationship between median DWI infarct volume and serum calcium level quartiles was evaluated using a semiparametric approach with univariate and multivariate quartile regression analysis.8 For the multivariate models, the median DWI infarct volumes were corrected for potential confounding variables. Categorical variables (stroke subtype, previous stroke, history of diabetes mellitus, and history of atrial fibrillation or statin use) and continuous variables (age, temperature, blood glucose level, time to MR imaging, and systolic blood pressure) were used as covariates based on prior literature.9,10
One hundred seventy-three of 322 patients met the study criteria. Reasons for exclusion were initial examination outside of 24 hours (85 patients) and unavailability of serum calcium level or MR imaging results (64 patients). Among 173 study patients, the mean age was 70.3 years (age range, 24-100 years), and 101 (58.4%) were women. The racial/ethnic distribution was 66.0% white, 12.0% black, 13.0% Asian, 8.4% Hispanic, and 0.6% other. The median National Institutes of Health Stroke Scale score at initial examination was 4 (range, 0-38), and the median time between the last known well time and the MR imaging was 7.1 hours (range, 0.7-24 hours). Stroke subtypes using the modified Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification system11 were 41.0% cardioembolic, 25.4% small-vessel occlusion, 17.3% large-vessel atherothromboembolic, 10.5% stroke of undetermined cause, and 5.8% other cause. On discharge, the median National Institutes of Health Stroke Scale score was 3 (range, 0-30), and 94 patients (54.3%) had a discharge modified Rankin score of at least 2.
Calcium levels were obtained at a mean of 9.7 hours (range, 0.5-23.8 hours) after the last known well time. The mean ± SD serum calcium level was 8.95 ± 0.50 mg/dL (to convert to millimoles per liter, multiply by 0.25) (range, 7.60-10.40 mg/dL). Table 1 gives the demographic characteristics and frequency of clinical variables by quartile of serum calcium level. Patients in the lowest serum calcium level quartile had a higher median National Institutes of Health Stroke Scale score at admission compared with patients in the highest quartiles (P = .045). There were no statistically significant differences between serum calcium level quartiles in time to MR imaging or in the interval between MR imaging and serum calcium collection. There were no other differences in baseline clinical characteristics between serum calcium level quartiles.
There was good interrater reliability in the measurement of DWI infarct volumes (interclass correlation coefficient, 0.99). Across all patients, the median DWI infarct volume was 2.9 mL (interquartile range, 19.75-0.52 mL [full range, 0-279.0 mL]). The median DWI infarct volumes for the serum calcium level quartiles (lowest to highest) were 9.42, 2.11, 1.03, and 3.68 mL. The distribution of DWI infarct volumes was positively skewed, reflecting a larger proportion of patients with smaller infarct volumes (skewness, 3.4; kurtosis, 13.2). Nonparametric Spearman rank correlation revealed a statistically significant negative correlation between serum calcium level and DWI infarct volume (r = −0.199; P = .009).
The results of the univariate and multivariate analyses of the relationship between serum calcium level at admission and pretreatment DWI infarct volume are given in Table 2. Univariate analysis indicated that the overall quartile regression model for the relationship between median DWI infarct volume and serum calcium level quartile was statistically significant, with a negative slope (P ≤ .001). The median DWI infarct volume in the lowest serum calcium level quartile was statistically significantly larger than those in the upper 3 quartiles (P < .05). The multivariate analysis indicated again a statistically significant overall trend toward increased DWI infarct volumes across calcium quartiles (P ≤ .001). The median adjusted DWI infarct volumes, interquartile range, and full range for each of the serum calcium level quartiles are shown in the Figure and (lowest to highest) were 8.9, 5.8, 4.5, and 3.8 mL. Again, the median adjusted DWI infarct volume in the lowest serum calcium level quartile was larger than that in the upper 3 quartiles (P < .05).
In patients with acute ischemic stroke evaluated within 24 hours of symptom onset, our results show that total serum calcium levels at admission are inversely associated with the volume of DWI abnormality. Patients with serum calcium levels in the lowest quartiles had DWI infarct volumes almost twice as large as those in the upper 3 quartiles, and this difference persisted after adjusting for potential confounding factors.
These findings are consistent with previous studies that examined serum calcium levels in patients with acute ischemic stroke. Prior studies found that lower serum calcium levels were associated with worse clinical outcomes. Specifically, patients with stroke having lower serum calcium levels had more severe strokes as indexed by National Institutes of Health Stroke Scale score at admission and worse functional outcomes at discharge (modified Rankin score, ≥ 2)4 and were more likely to die during hospitalization.3 Although the better outcomes seen among patients with higher serum calcium levels in these prior studies were postulated to be caused by smaller infarct volumes, this hypothesis could not be directly evaluated because neither study included neuroimaging.
We used initial DWI lesions as a measure of infarct volume. Early in ischemic stroke, DWI demarcates brain tissue in which there has been a failure of cellular energy-dependent processes due to reduced cerebral perfusion. Initial DWI lesions correlate strongly with final infarct volumes.12
Several potential mechanisms may explain why serum calcium levels are related to the extent of bioenergetically compromised tissue in patients with acute ischemic stroke. These mechanisms can be divided into 2 broad and nonmutually exclusive categories. First, raised serum calcium levels directly or indirectly attenuate the volume of ischemic tissue; second, serum calcium levels may alternately drop in response in the presence of tissue ischemia.
Elevated serum calcium levels may reduce tissue ischemic injury by affecting excitotoxic pathways and ischemic preconditioning. Intracellular serum calcium level plays a central role in the mechanism of excitotoxicity in ischemic stroke. With tissue ischemia, there is excessive release of endogenous glutamate, which in turn leads to an uncontrolled accumulation of intracellular serum calcium level through N-methyl-D-aspartate receptors and voltage-dependent calcium channels.2 Brain ischemia is associated with at least a temporary depletion of cerebral tissue extracellular calcium,13 although it is unclear whether the magnitude and time course of this decrease is sufficient to account for the lower total serum calcium levels observed in patients with larger infarcts. Most important, decreased extracellular serum calcium levels are an important factor in the positive feedback mechanism that potentiates the inward serum calcium level currents following ischemic injury.2 During ischemia, decreased extracellular serum calcium levels lead to disinhibition of calcium-sensing nonselective channel currents and to further membrane depolarization and additional influx of calcium.
If low extracellular and serum calcium levels might potentiate ischemic injury, then it would be predicted that raising serum calcium levels might have neuroprotective effects. There is limited evidence to support this hypothesis. In a rodent model of ischemic stroke, stroke-prone animals fed a diet high in calcium had decreased stroke lesion volumes.14 There are also epidemiological studies15,16 that have linked increased dietary calcium intake to decreased mortality from ischemic and hemorrhagic stroke; however, neither of these studies measured serum calcium levels. Because dietary calcium intake is only 1 of the factors affecting calcium homeostasis, the extent to which these epidemiological observations reflect variations in serum calcium levels remains unclear.
This study is limited by its cross-sectional design and retrospective analysis. Other limitations are that we measured total serum calcium levels rather than ionized serum calcium levels, the physiologically active component. Also, only a single serum calcium level measurement was collected, so it is impossible to determine whether serum calcium levels at admission remained stable over time. Of particular interest in trying to better determine the potential cause and effect relationship would be to see if serum calcium levels change over time in relationship to DWI lesion growth. Initial DWI infarct volume was analyzed, not final T2-weighted infarct volume, as late T2-weighted studies were not obtained routinely in these patients. However, early DWI infarct volumes correlate well with final T2-weighted infarct volumes.17
In conclusion, elevated serum calcium levels in acute ischemic stroke are an early predictor of smaller volumes of ischemic tissue and better clinical outcomes. This effect was seen across a range of stroke subtypes and persisted after adjustment for other factors that could potentially affect infarct volume. Further investigations will be required to elucidate the mechanism of this effect and to assess the role of serum calcium level as a prognostic variable and of calcium modulation as part of a potential neuroprotective strategy.
Correspondence: Brian H. Buck, MD, FRCPC, Division of Neurology, Department of Medicine, Sunnybrook Health Sciences Centre, University of Toronto, 2075 Bayview Ave, Room A421, Toronto, ON M4N 3M5, Canada (email@example.com).
Accepted for Publication: March 12, 2007.
Author Contributions:Study concept and design: Buck, Liebeskind, Saver, Ali, and Ovbiagele. Acquisition of data: Buck, Liebeskind, Young Bang, Starkman, Kim, Yun, Razinia, and Ovbiagele. Analysis and interpretation of data: Buck, Liebeskind, Young Bang, Villablanca, Salamon, and Ovbiagele. Drafting of the manuscript: Buck, Liebeskind, Saver, Young Bang, and Ovbiagele. Critical revision of the manuscript for important intellectual content: Buck, Liebeskind, Saver, Starkman, Ali, Kim, Villablanca, Salamon, Yun, Razinia, and Ovbiagele. Statistical analysis: Buck. Obtained funding: Buck, Saver, and Ovbiagele. Administrative, technical, and material support: Liebeskind, Saver, Starkman, Ali, Villablanca, Salamon, Yun, Razinia, and Ovbiagele. Study supervision: Liebeskind, Saver, Kim, and Ovbiagele.
Financial Disclosure: None reported.
Funding/Support: This study was supported in part by a Heart and Stroke Foundation of Canada fellowship award (Dr Buck) and by grant P50 NS044378 from the National Institute of Neurological Disorders and Stroke, National Institutes of Health (Drs Saver and Ali).