Acute Systemic Inflammatory Response Syndrome in Subarachnoid Hemorrhage
Background and Purpose— Systemic inflammatory response syndrome (SIRS) without infection is a well-known phenomenon that accompanies various acute cerebral insults. We sought to determine whether the initial SIRS score was associated with outcome in subarachnoid hemorrhage (SAH).
Methods— In 103 consecutive patients with SAH, the occurrence of SIRS was assessed according to the presence of ≥2 of the following: temperature of <36°C or >38°C, heart rate of >90 bpm, respiratory rate of >20 breaths/min, and white blood cell count of <4000/mm3 or >12 000/mm3. SIRS criteria and other prognostic parameters were evaluated as predictors of dichotomous Glasgow Outcome Scale score.
Results— SIRS was highly related to poor clinical grade (Hunt and Hess clinical grading scale), a large amount of SAH on CT (Fisher CT group), and high plasma glucose concentration on admission. By univariate analysis, the occurrence of SIRS was associated with higher mortality and morbidity rates than was the nonoccurrence (P<0.001). Among individual SIRS criteria, heart rate (P=0.003), respiration rate (P=0.003), and white blood cell count (P=0.03) were significant outcome predictors. By multivariate logistic regression analysis, the presence of SIRS independently predicted outcome. SIRS carried an increased risk of subsequent intracranial complications such as vasospasm and normal pressure hydrocephalus, as well as systemic complications.
Conclusions— In SAH patients, SIRS on admission reflected the extent of tissue damage at onset and predicted further tissue disruption, producing clinical worsening and, ultimately, a poor outcome.
Systemic inflammatory response without infection is a well-known phenomenon in various types of acute cerebral injury. For instance, hyperthermia in patients with large infarcts was reported to predict the outcome of ischemic stroke in many studies.1–4 Several studies have reported elevations of proinflammatory cytokines in peripheral blood as well as in cerebrospinal fluid in patients with ischemic stroke.4,5 In patients with subarachnoid hemorrhage (SAH), however, this acute-phase response is less well delineated, although fever and leukocytosis can accompany aneurysmal SAH and are associated with an unfavorable outcome.6–8
A defined combination of clinical features, the systemic inflammatory response syndrome (SIRS), can occur in relation to a variety of severe clinical insults. A consensus conference of the American College of Chest Physicians and the Society of Critical Care Medicine recently proposed criteria for SIRS,9 to aid in assessment of the severity of illness in intensive care unit (ICU) patients. SIRS scores already are used prognostically for emergency or critical care patients, and the term SIRS has become incorporated in medical textbooks. The conference defined SIRS according to 4 clinical and laboratory parameters: body temperature, heart rate, respiratory rate, and white blood cell count.9
To our knowledge, no previous studies of SAH have assessed the prognostic implications of SIRS as currently defined. We sought to investigate whether SIRS was linked to the severity of SAH or to other factors, such as aneurysm location, sex, and age. The present study was performed to determine whether the presence of SIRS on admission was associated with poor outcome in SAH patients and whether this inflammatory response predisposed to clinical deterioration.
Subjects and Methods
Our hospital is an academic tertiary care referral center. All patients with SAH were admitted to the neurosurgical unit except for those with cardiopulmonary arrest on arrival. The study was undertaken among patients referred to the neurosurgical unit. At our hospital, 110 patients were admitted within 3 days after SAH between April 1998 and August 2000. Among these patients, 7 were excluded because medical records were not sufficiently complete for all analyses. We retrospectively studied the remaining 103 consecutive patients. Patient characteristics are shown in Table 1. In all patients, SAH was verified by CT on admission. We did not include patients whose clot was confined to the perimesencephalic cisterns, and we restrict the analyses to those with verified or strongly suspected aneurysmal SAH. The study was approved by the Dokkyo University Institutional Review Board.
All patients were treated according to standard intensive care guidelines. The patients showed a wide range of SAH severity. Of 103 patients, 6 (6%) did not undergo cerebral angiography because of grave neurological condition; none of these 6 underwent aneurysm surgery, and all died early in their hospital course. An additional 6 patients (6%) did not undergo treatment for ruptured aneurysm because no aneurysm was verified by repeated cerebral angiography or because their neurological condition was poor. The remaining 91 patients (88%) underwent procedures to treat ruptured aneurysms. Although most patients underwent surgical clipping (88 cases), intravascular coil embolization was selected as the treatment modality in 3 cases. The day of SAH was defined as day 0.
Postoperative management for all patients included mild hypervolemic therapy using intravenous drip infusion of low-molecular-weight dextran as necessary. Dobutamine and nicardipine were administered to most patients during the period of vasospasm. Continuous cisternal, ventricular, or lumbar external drainage was established when necessary to control intracranial pressure. Based on clinical status at 3 months after onsets, we evaluated outcome according to the Glasgow Outcome Scale as representing good recovery, moderate disability, severe disability, or vegetative state.10 If the patient died before 3 months, outcome was recorded as death.
The neurological condition of all patients was recorded throughout the hospital stay by the attending neurosurgeons. To determine the acute response after SAH, the authors thoroughly reviewed the hospital charts of all patients, focusing on clinical and laboratory data at the time of admission. Clinical information and laboratory data were collected from an admission protocol devised by the attending physicians and nursing stuff. SIRS was identified according to the guidelines of the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference9 that were established to identify SIRS as a systemic inflammatory response to a variety of severe clinical insults. SIRS is manifest when ≥2 of the following conditions are met: temperature of <36°C or >38°C, heart rate of >90 bpm, respiratory rate of >20 breaths/min, and white blood cell count of <4000/mm3 or >12 000/mm3. SIRS was diagnosed according to the earliest documentation of relevant variables appearing in the clinical record after arrival at the hospital. In applying these criteria for patients with SAH, we needed to modify the respiratory rate criterion, because many patients with respiratory dysfunction underwent immediate intubation followed by mechanical ventilation. We included these patients with groups who met this respiration criteria.
The following information was recorded for each patient: sex, age, clinical grade on admission (Hunt scale),11 SAH severity by admission CT (Fisher CT group),12 aneurysm location, and plasma glucose concentration on admission. Each variable was tested statistically for the ability to predict outcome.
In addition, occurrence or nonoccurrence of the following events was recorded for each patient: cerebral vasospasm, eventual requirement of a shunt to treat normal-pressure hydrocephalus, and systemic medical complications such as cardiopulmonary dysfunction or multiple organ failure. Cerebral vasospasm was defined as a new focal neurological deficit, significant worsening of existing neurological deficits, or a decrease in level of consciousness between day 4 and day 14 after the exclusion of other causes of deterioration. Occurrence rates of these events were calculated for the 2 groups (SIRS versus non-SIRS) defined by the 4 SIRS criteria.
For statistical analyses, all variables except for aneurysm location were made dichotomous. Associations between SIRS and other dichotomous prognostic parameters were analyzed using the χ2 test. Regarding aneurysm location, because the numbers of patients with vertebrobasilar artery aneurysm or unknown etiology were small, the relationship between SIRS and 3 aneurysm sites (internal carotid, middle cerebral, and anterior cerebral arteries) was assessed using the χ2 test.
To investigate the prognostic value of each variable, including each of the 4 criteria for SIRS, logistic regression analyses were performed using appropriate dummy variables as follows. Patients younger than 60 years were assigned the value of 0, whereas 1 was assigned to those older than 60 years. Plasma glucose concentrations were the basis on which to divide patients into 2 groups: <200 mg/dL (a value of 0) or >200 mg/dL (a value of 1). Hunt scale scores of I, II, or III were assigned a value of 0, and scores of IV or V were assigned a value of 1. Fisher CT group I or II was assigned a value of 0, whereas III or IV was assigned a value of 1. Similarly, patients without SIRS on admission (meeting no criteria or 1 criterion) were assigned a value of 0, whereas SIRS patients (meeting 2 to 4 criteria) were assigned a value of 1. Outcome, dichotomized as favorable (good recovery or moderate disability) or poor (severe disability, vegetative state, or death), was used as the dependent variable. P<0.05 was considered to indicate statistical significance.
Multivariate analysis was performed to control for possible confounding effects between potentially explanatory variables. All 5 dichotomously characterized factors described earlier were considered as potential explanatory variables in multivariate analysis. A forward stepwise method was used to construct a multivariate logistic regression model in relation to poor outcome, with the inclusion criterion of a P value of <0.15.13 All statistical calculations were performed using a statistical software package (SPSS, version 6.1) for a personal computer.
The number of patients meeting each SIRS criterion on admission is shown in Table 1. The number of SIRS criteria met (SIRS score) on admission was 0 in 15 patients, 1 in 32 patients, 2 in 32 patients, 3 in 19 patients, and 4 in 5 patients. Forty-seven patients (46%) meeting no or only 1 SIRS criterion were assigned to the non-SIRS group, whereas the remaining 56 patients (54%) who met ≥2 criteria were assigned to the SIRS group.
Table 2 shows the relationship between SIRS criteria and other selected variables. Female patients were somewhat more likely to be in the SIRS group than were male patients, although this tendency was not statistically significant. Hunt scale, Fisher CT group, and plasma glucose concentration on admission were highly related to SIRS. We found no correlation between SIRS categories and age or aneurysm location.
Table 3 shows the results of univariate logistic analysis concerning relationships of selected variables to poor outcome according to the dichotomous Glasgow Outcome Scale. SIRS criteria, Hunt scale, Fisher CT group, and plasma glucose concentration on admission were significantly related to outcome. Age was not significantly related to outcome by univariate analysis. The results indicated that a higher SIRS score, poor clinical condition, a large amount of SAH on CT, and higher glucose concentration on admission were poor prognostic factors. Other factors such as sex and aneurysm location did not significantly influence outcome. To varying degrees, each SIRS criterion was relate to outcome. Heart rate (P=0.003), respiration rate (P=0.003), and white blood cell count (P=0.03) each were statistically significant outcome predictors in themselves, whereas body temperature fell short of statistical significance (P=0.22).
Stepwise introduction of all 5 variables listed in Table 3 into the multivariate logistic model resulted in the selection of 4 variables: Fisher CT group, age, Hunt scale, and presence or absence of SIRS (Table 4). Fisher CT group was the first variable selected in the model. The OR for poor outcome in patients in Fisher CT group III or IV was 11.6 (95% CI 1.3 to 101.1) relative to those in Fisher CT group I or II. Age was among the variables analyzed for outcome prediction, despite insignificant correlation by univariate analysis; in the multivariate analysis, age older than 60 years indicated a 3.1-fold increase in the odds of a poor outcome (95% CI 1.0 to 9.4) beyond the odds when age was younger than 60 years. Hunt scale and presence of SIRS categories on admission also were retained in the model, with a marginal significance. The OR for poor outcome in patients with Hunt grade IV or V was 3.5 (95% CI 0.9 to 13.3) relative to patients with grade I or II. The OR in patients in the SIRS group for poor outcome was 3.6 (95% CI 0.8 to 15.0) relative to those of the non-SIRS group. Glucose concentration was excluded by the model.
Table 5 shows associated complications for groups defined by SIRS criteria. Patients with SIRS had a higher likelihood of intracranial complications such as vasospasm or subsequent normal-pressure hydrocephalus that required a shunt than did those without SIRS. The difference was statistically significant for both (P=0.04 and P=0.01, respectively). During the hospital stay, significant systemic (nonneurological) complications were observed in 18 patients. The most frequent complication was respiratory dysfunction, manifest in 12 patients; this was followed in frequency by sepsis or disseminated intravascular coagulopathy (n=3), cardiac failure (n=2), and renal failure (n=1). Incidence of systemic complications also was significantly higher in SIRS patients than in non-SIRS patients (P=0.03).
SIRS and SAH Severity
In this retrospective study, the admission SIRS score proved to be associated with severity of SAH indicated by clinical grade (Hunt scale), amount of clot demonstrated by CT (Fisher CT group), and plasma glucose concentration. This implies that SIRS reflects the severity of brain damage caused by SAH. In ischemic stroke, SIRS was reported to occur in patients with large infarcts, usually correlating with the extent of tissue damage.3 Previous studies demonstrated that SAH can also cause leucocytosis.6–8 Activation of early-phase cytokine cascades in conjunction with the biologic effects of complement activation (anaphylatoxin reaction, increased vascular permeability, and vasodilation) may contribute to the development of an SIRS-like condition after SAH.14,15
SIRS and SAH Outcome
A clear relationship between SIRS score and short-term outcome of patients was demonstrated in our univariate analysis. Three of 4 SIRS criteria (heart rate, respiration rate, and white blood cell count) were each significant outcome predictors in themselves. Although SAH has been demonstrated to show a positive association between the degree of posthemorrhagic leukocytosis and unfavorable outcome,8 clinical data regarding other inflammatory parameters have been limited.15,16 In accordance with previous reports,17,18 we found high admission glucose concentration to be associated with poor outcome.
In our multivariate analysis, stepwise introduction of all variables yielded an optimal model constructed with 4 parameters: Fisher CT group, age, Hunt scale, and the presence or absence of SIRS. The high OR and the wide CI for Fisher CT group suggested that the reliability of these values may be relatively low. Although plasma glucose was significantly related to prognosis in the univariate analysis, it did not emerge in the multivariate model because of a close relationship to other parameters. The direct relationship between SIRS and risk of poor outcome remained marginally significant after adjustment for the effects of other prognostic markers, such as Hunt scale, Fisher CT group, and age. These findings indicated that the presence of SIRS on admission was an independent prognostic factor.
SIRS and Subsequent Clinical Deterioration
A remarkable finding in the present study was an association between admission SIRS score and subsequent neurological deterioration due to vasospasm or normal-pressure hydrocephalus during the hospital stay. Although the effects of the acute-phase response after SAH are not well delineated, hyperthermia has been shown in many experimental and clinical studies to exacerbate brain injury in cerebral ischemia.1,3,4 Aggravation of the deleterious effects of cerebral ischemia may be explained by increased metabolism. Many have argued that hyperthermia not only is an indicator of stroke severity but also may worsen the stroke.1,3,4 Conclusions concerning SAH, a different cerebrovascular entity, cannot simply be extrapolated from results derived from clinical and experimental ischemic stroke. However, subsequent ischemia due to vasospasm often occurs after SAH, so the cerebral insult complicating SAH may be subject to similar constraints. Because the presence of circulating immune complexes has been linked to poor outcome, immunological processes involving complement-activating immune complexes may be involved in the pathogenesis of cerebral vasospasm after SAH.15 Although we did not measure concentrations of cytokines, our results may support the view that cytokine-mediated tissue damage participates in mechanisms that may lead to clinical worsening.
We found that subsequent systemic complications developed more frequently in patients with SIRS than in those without SIRS. Medical as well as intracranial complications after SAH contribute significantly to overall mortality rates.19 We speculate that mechanisms of central dysregulation secondary to intracranial aneurysm rupture can contribute to the development and progression of extracerebral organ dysfunction by promoting a state of systemic inflammation that carries the potential for systemic organ dysfunction. In SAH patients, 77% of extracerebral organ system failures reportedly occurred in conjunction with SIRS.16 Although only controlled intervention trials aimed at reversing SIRS can prove whether this relationship is causal, the development of new therapies that modulate cytokine-induced inflammation might be a promising way to prevent neurological and medical complications.
Limitations of the Study
Our study has several limitations. First, 1 factor that limits the applicability of our results is the retrospective nature of the study; as a result, a potential exists for bias that might render any predictive model inapplicable to populations other than that from which it was derived. However, we believe that the study of consecutive SAH patients during a relatively short period can preclude significant selection bias. Second, the release of cytokines is time dependent,14 and we did not control for the confounding effect of time interval from the onset of SAH to admission. In addition, we analyzed only initial clinical and laboratory findings, regardless of whether patients had an infection such as aspiration pneumonia early in their course. Therefore, no conclusions can be drawn regarding further dynamics of the inflammatory response after SAH. Because no association between time course of SIRS and outcome has been reported to date, further studies are needed to elucidate the importance of data in addition to the initial findings and values.
- Received February 5, 2001.
- Revision received May 24, 2001.
- Accepted June 8, 2001.
Castillo J, Davalos A, Marrugat J, Noya M. Timing for fever-related brain damage in acute ischemic stroke. Stroke. 1998; 29: 2455–2460.
Chamorro A, Vila N, Ascaso C, Saiz A, Montalvo J, Alonso P, Tolosa E. Early prediction of stroke severity: role of the erythrocyte sedimentation rate. Stroke. 1995; 26: 573–576.
Vila N, Castillo J, Davalos A, Chamorro A. Proinflammatory cytokines and early neurological worsening in ischemic stroke. Stroke. 2000; 31: 2325–2329.
Neil-Dwyer G, Cruickshank J. The blood leucocyte count and its prognostic significance in subarachnoid hemorrhage. Brain. 1974; 97: 79–86.
Afifi AA, Clark V. Computer-Aided Multivariate Analysis. ed 3. Boca Raton, Fla: Chapman & Hall/CRC; 1996.