Neurocrit Care (2008) 8:301–307 DOI 10.1007/s12028-007-9019-x

REVIEW ARTICLE

S100 as a Marker of Acute Brain Ischemia: A Systematic Review David L. Nash Æ M. Fernanda Bellolio Æ Latha G. Stead

Published online: 30 October 2007
Humana Press Inc. 2007

Abstract Background Studies show S100 as a possible acute ischemic stroke (AIS) marker. Objectives Determine (1) whether S100 serum concentrations correlate with stroke symptom onset, infarction volume, stroke severity, functional outcome, or length of hospital stay; (2) whether S100 serial measurements are useful markers for ongoing brain ischemia, and (3) whether S100 levels at various time intervals are higher in AIS patients than controls. Methods Literature was searched using OVID and MEDLINE from January 1950 to February 2007, and all relevant reports were included. Results Eighteen studies (1,643 patients) satisfied entry criteria. S100 peaks from symptom onset between 24 and 120 h with significantly raised values measured from 0 to 120 h. Higher S100 values indicated significantly larger infarction volumes, more severe strokes, and worse functional outcome. There was a significant difference in S100 levels between AIS patients and controls. Conclusion Peak values after stroke onset varied. S100 was significantly increased after stroke onset, and correlates with infarct volume, stroke severity, and functional outcome, and was a possible marker for ongoing ischemia. Its serum concentration during acute stroke is a useful marker of infarct size and long-term clinical outcome. Keywords Serum markers
S100
Acute brain ischemia
Systematic review

D. L. Nash
M. F. Bellolio
L. G. Stead (&) Mayo Clinic College of Medicine, Rochester, MN, USA e-mail: [email protected]

Introduction The glial protein S100 belongs to a family of calciumbinding proteins found as homo- or hetero-dimers of two different subunits (alpha and beta). Different combinations of the subunits make up the heterodimeric forms alpha– alpha, alpha–beta and beta–beta; types alpha–beta and beta–beta are described as S100B protein and are shown to be highly specific for nervous tissue. When structural damage such as infarction occurs in the cytosol of glial and Schwann cells, S100 is released into the cerebrospinal fluid and the blood [1]. S100 has been reported as a marker of blood–brain barrier dysfunction, and its concentration in cerebrospinal fluid is 40 times higher than in the serum [2]. It is not affected by hemolysis and remains stable for several hours without the need for immediate analysis. Its short half-life makes measurements crucial in the emergency and intensive care settings [3]. Given these characteristics, S100 has shown promise regarding its use as a possible serum marker for acute ischemic stroke. To date, most of the work delineating S100 as a marker for stroke has not investigated enough aspects of S100 to be able to make any concrete conclusions. We undertook the present systematic review to help overcome this limitation. Ideally, markers for acute ischemic stroke (AIS) would peak and accumulate early in ischemic cascade, have rapid diffusion through ischemic tissue into bloodstream, have a half-life of at least a few hours, and are specific for ischemic neural tissue. Our goal was to ascertain whether S100 fit the preceding criteria and whether concentrations correlate with the presence of AIS. Specifically, our objectives were the following: (1) to determine whether serum concentrations of S100 correlate

302

with time of onset of stroke symptoms, volume of the infarcted tissue, stroke severity, functional outcome, or length of hospital stay; (2) to determine whether serial measurements of S100 levels are useful as a marker for ongoing brain ischemia; and (3) to determine whether S100 levels at various time intervals are significantly higher in patients with stroke than in controls.

Methods Criteria for Studies All randomized controlled trials, quasi-randomized controlled trials, and case-control studies were considered. Participants were adult patients (18 years or older) with acute ischemic stroke. We evaluated S100 concentrations in terms of the following types of outcome measures: (1) time of onset of stroke symptoms as ascertained with history and computed tomographic findings; (2) volume of the infarcted tissue as ascertained with computed tomography or magnetic resonance imaging; (3) stroke severity as ascertained with the National Institutes of Health Stroke Scale (NIHSS) at hospital admission; (4) functional outcome at hospital dismissal as ascertained with the modified Rankin scale (mRs), activities of daily living scale, and NIHSS; (5) functional outcome at follow-up as ascertained with the Glasgow Outcome Scale, Barthel Index, mRs, and Lindley score; (6) length of hospital stay; (7) serial S100 levels as markers for ongoing brain ischemia as determined with computed tomography; and (8) determination of whether S100 levels were significantly higher in patients with stroke than in patients without stroke at various time intervals.

Neurocrit Care (2008) 8:301–307

Description of Studies No ongoing studies were identified or investigated. Eleven studies were excluded. Two were reviews, in which only the abstracts were available [4, 5]. Two involved patients with carotid artery stenosis [6, 7]. One exclusively involved patients who had central nervous system lymphoma [2]. One examined adverse reactions after cardiac surgery including stroke, but not isolating it for examination [8]. One examined patients with conventional cardiopulmonary bypass [9]. One was an animal study [10]. One was an editorial. Two were letters to the editor [11, 12]. Eighteen studies, which included 1,159 ischemic stroke patients and 484 controls, satisfied entry criteria. Characteristics of included studies are summarized in Table 1. The average ages and male–female ratios varied across studies and populations. Studies with controls either used age matched or age–sex matched. All patients were entered into the studies within 72 h of the onset of stroke; 12 included patients who were presented within 24 h. Three of these studies, performed by the same author, have overlapping cohorts, but different primary outcomes.

Methodologic Quality of Included Studies Nearly all studies included consecutive patients presenting with stroke. Although S100 levels were objective findings, no studies mentioned whether the values were concealed from the physicians treating the patients. This could potentially introduce bias when outcome scores were calculated.

Results Search Strategy for Identification of Studies All literature was searched using OVID and MEDLINE from January 1950 to February 2007. Reference lists of published narrative and systematic reviews were reviewed. No language restrictions were applied. Unpublished data were not sought directly from the authors.

Methods of the Review All abstracts and published full reports identified as potentially relevant through literature searches were independently assessed for inclusion by each reviewer. Data was transcribed onto pre-designed data abstraction forms. The forms were qualitatively compared, and disagreements were resolved with discussion.

Onset of Stoke Symptoms Versus Earliest Detectable S100 Levels Reynolds et al. [13] showed S100 levels to be significant enough that their use along with a panel of makers was significantly correlated to AIS from 0 to 12 h and still good predictions from 0 to 6 h. Lynch et al. [14] found S100 levels to be significantly higher in stroke patients than controls when measured between both 0–6 h and 6–24 h. Wunderlich et al. [15] found significantly higher S100 levels as early as 6 h after stroke onset, and Foerch et al. [16] showed S100 levels to be significantly higher at 12 h with peak values ranging from 24 to 120 h. Starting with the 12-h value, S100 serum concentrations were significantly higher in malignant MCA stroke patients compared to patients with nonmalignant infarction [17]. Foerch et al.

Patients

Acute cardioembolic stroke

Within 48 h of stroke onset

Patients from National Institute of Neurological Disorders and Stroke (NINDS) presented to the ED. Patients received tPA

Acute non-lacunar MCA infarctions less than 6 h after symptom onset

Within 6 h after stroke onset

Proximal MCA occlusion, within 6 h of stroke onset

Within 24 h of symptom onset

Within 24 h of symptom onset

MCA/M1 occlusion

Reference

Mizukoshi et al. [21]

Abraha et al. [1]

Jauch et al. [22]

Foerch et al. [16]

Wunderlich et al. [15]

Foerch et al. [17]

Lynch et al. [14]

Reynolds et al. [13]

Foerch et al. [23]

Table 1 Description of studies

Germany; 23 patients (15 M) mean age 70.2 years

US; 275 patients 214 age matched controls

US; 65 patients, 157 age– sex matched controls

Germany; 51 patients (24 F) mean age 69 years

Germany; 32 patients (21 M) mean age 63 years

Germany; 39 patients mean age, 69 years

US; 359 patients

UK; 81 patients (68 ischemic) 51 age–sex matched controls

Japan; 7 patients (4 F) mean age 73 years

Participants

Monoclonal 2-site immunoluminometric assay and a fully automatic LIA-mat system

ELISA

Alkaline phosphatase-conjugated secondary antibodies, Altophos substrate

Monoclonal 2-site immunoluminometric assay in a fully automated LIA-mat system

Immunoradiometric sandwich assays using directly coated magnetic microparticles (LIASION)

Monoclonal 2-site immunoluminometric assay and a fully automatic LIA-mat system

ELISA

Immunoradiometric assay—IRMA Sangtec, Bromma, Sweden

ELISA

Assay

MRI, CT, NIHSS, mRS

Unknown

CT, MRI

NIHSS, MR angiography in 39 patients, IA in 4 and TCD in 8

Infarct volume, NHISS, mRs

mRs, infarct volume

NIHSS, mRs, and infarct size on CT at 24 h and the effect of fibrinolytic therapy

Barthel Index, Rankin, and Lindley score

MRI, infarct volume, NIHSS scores, urinary content

Outcomes

Correlations between final volume and single S100 values obtained between 24 and 96 h after stroke onset. A single S100 value 0.2 lg/l

Higher S100B associated with worse mRS. S100B levels associated with final infarct volume

Higher peaks associated with higher baseline NIHSS and larger CT volumes. Patients with favorable outcomes had smaller changes in S100 levels over first 24 h

S100 was significantly higher in infarct than in the control groups. Serum S100 was related to infarct size, and higher levels at admission related with poor outcome at 3 months.

S100B peaks during 3–5 days after onset

Comment

Neurocrit Care (2008) 8:301–307 303

Within 4 h of symptoms onset

Stroke after cardiac surgery

Within 24 h of symptom onset Acute MCA infarction with measures at 0–10 days after stroke

Bertsch et al. [24]

Jo¨nsson et al. [26]

Hill et al. [29]

Within 72 h of symptom onset

Within 4 h of onset of symptoms

Patients with acute stroke 5 cc exhibited significantly increased S100. S100 correlated with outcome at 10, 24, and 72 h

Peak blood levels of S100 on day 2.5. Peak S100 values correlated with infarct volume and with clinical outcome

Correlates with outcome and NIHSS S100 elevated in MCA patients but not in controls. S100 levels peaked at 2–3 days after stroke and concentrations were higher in patients with more severe deficits

S100 level at 48 h correlated with size of infarct. High S100 48 h after surgery has a negative predictive value for median term survival

Lesions larger than 5 cm3 exhibited significantly increased serum levels of S100 at 10, 24, and 72 h. Association between S100 and neurological outcome

Comment

304 Neurocrit Care (2008) 8:301–307

Neurocrit Care (2008) 8:301–307

305

[18] measured S100 within 6 h from symptoms onset in 275 patients, finding serum S100 levels over the median in 79.3% of the patients. Bu¨ttner et al. [3], Missler et al. [19], and Kim et al. [20] measured peak S100 levels between 2 and 3 days; while Mizukoshi et al. [21] found S100 peak 3–5 days after stroke onset, though this study was limited to seven patients.

greater baseline NIHSS values (rs = 0.263, P < 0.0001). Hill et al. [29] found peak levels of S100 to be significantly correlated with admission NIHSS scores (P < 0.05).

Volume of Infarcted Tissue

The studies examined in this review used several different scales to measure function outcome. These included the Barthel Index, mRs, Lindley score, Glasgow Outcome Scale, Scandinavian Stroke Scale, and activities of daily living scale. Ten of the studies reviewed examined the relationship between functional outcome and S100. Abraha et al. [1] used three of these scores for functional outcome, and S100 significantly correlated to clinical outcome at 3 months after the stroke with the modified Barthel index (rs = -0.285, P = 0.01), mRs (rs = 0.313, P = 0.004), and Lindley score (rs = 0.262, P = 0.018) where higher S100 values related to poorer functional outcome. Jauch et al. [22] found that patients with smaller changes in S100 concentrations over the first 24 h after stroke had a more favorable outcome at 3 months. Foerch et al. [16] found the correlation between higher S100 values and poorer functional outcome at 6 months. Measurements taken between baseline and the 6th day from stroke, all reported P < 0.001 indicating a high level of significance. Bertsch et al. [24] found that as S100 level increases, neurological outcome decreases with measurements taken at 10 h (r = -0.56, P < 0.05), 24 h (r = -0.8, P < 0.005), and 72 h (r = -0.78, P < 0.005). Hill et al. [29] also correlated significantly with functional outcome at discharge using the mRs scale (P < 0.05). Missler et al. [19] found S100 peaks to correlate with worse clinical outcomes at discharge (r = 0.44, P < 0.01) using the Activities of Daily Living Scale and at the long-term 6 months followup (r = 0.49, P < 0.01) using the Glasgow Outcome Scale. Using the Scandinavian Stroke Scale during the 2nd or 3rd week after the stroke, Fassbender et al. [25] found neurological outcome to inversely correlate with serum levels of S100 at 10 h (r = -0.56, P < 0.05), 24 h (r = 0.80, P < 0.005), and 72 h (r = -0.78, P < 0.005). Using the simplified activities of daily living test, Aurell et al. [27] found S100 levels to be correlated to clinical outcome (P < 0.001). In another study, S100 serum concentrations were associated with the functional outcome at 3 months from 6 h on and with a maximum correlation obtained for S100 at 48 h (P > 0.01) [15]. One study found no significant (P > 0.05) correlations between S100 values and the functional prognosis at all time points. However, when the mRS was dichotomized as minor stroke (0–1) and major stroke (2–5), S100 levels for

Ten studies showed a correlation between S100 levels and volume of infarction. Abraha et al. [1] showed S100 concentrations to correlate with size of stroke based on four locations with corresponding measurements: Total anterior circulation infarcts (0.45 lg/l, r2 = 0.22); Partial anterior circulation infarcts (0.27 lg/l, r2 = 0.07); Posterior circulation infarcts (0.25 lg/l, r2 = 0.25); and Lacunar infarcts (0.20 lg/l, r2 = 0.06) (P = 0.0005). Jauch et al. [22] associated higher peak S100 concentrations with larger lesion volumes on 24 h CT scans (rs = 0.238, P < 0.0001). In a study, Foerch et al. [16] significantly correlated S100 values with infarct volume at 24, 48, 72, 96, 120, and 144 h (P < 0.001) and at baseline (P = 0.004). In another study, Foerch et al. [17] found significant correlations between S100 and infarct volume from 12 h after symptom onset (12 h, r = 0.525; 16 h, r = 0.539; 20 h, r = 0.681; 24 h, r = 0.723; all P < 0.001). A third study by Foerch et al. [23] demonstrated significant positive correlations between final lesion volume and each single S100 value obtained between 24 and 96 h after stroke onset (P < 0.0056). Bertsch et al. [24] and Fassbender et al. [25] found patients with detectable S100 had larger lesion volumes (P < 0.05), and when patients were dichotomized into large (>0.05 cm3) and smaller (