Relationship Between C-Reactive Protein, Systemic Oxygen Consumption, and Delayed Cerebral Ischemia After Aneurysmal Subarachnoid Hemorrhage
Background and Purpose—Subarachnoid hemorrhage (SAH) is known to result in elevated systemic oxygen consumption (Vo2) and increases in high-sensitivity C-reactive protein (hsCRP), although the relationship among hsCRP, Vo2, and delayed cerebral ischemia (DCI) after SAH remains unknown. We hypothesized that hsCRP is directly associated with Vo2 and that elevated Vo2 is a predictor of DCI after SAH.
Methods—Prospective serial assessments of Vo2 and hsCRP over 4 prespecified time periods during the first 14 days after bleed in consecutive SAH patients admitted to a single academic medical center for a 2-year period.
Results—One hundred ten SAH patients met study criteria (mean age, 55±16 years; 62% women), with a median admission Hunt Hess grade of 3 (interquartile range, 2–4). In multivariate generalized estimating equation model of the first 14 days after bleed, Vo2 was associated with younger age (P=0.01), male gender (P=0.01), and hsCRP levels (P=0.03). Twenty-four (22%) patients had DCI develop, with a median onset on day 7 after bleed (interquartile range, 5–11). The mean Vo2 (291±65 mL/min versus 226±55 mL/min; P=0.003) was higher in DCI patients. In a multivariable Cox proportional hazards model, younger age (hazard ratio, 1.2 per 5 years; 95% CI, 1.1–1.3), a higher modified Fisher scale score (hazard ratio, 3.4 per 1-point increase; 95% CI, 1.7–6.9), and higher Vo2 (HR, 1.2 per 50-mL/min increase; 95% CI, 1.1–1.3) were predictive of DCI.
Conclusions—Systemic oxygen consumption is associated with hsCRP levels in the first 14 days after SAH and is an independent predictor of DCI.
Aneurysmal subarachnoid hemorrhage (SAH) affects >25 000 people annually,1 and advances in the medical and surgical management of SAH have resulted in a dramatic decline in hospital mortality.2 Nevertheless, SAH remains a major cause of premature mortality, accounting for 27% of all stroke-related potential years of life lost before age 65.3 Cerebrovascular vasospasm, which occurs most commonly between 4 and 14 days after hemorrhage, results in delayed cerebral ischemia (DCI) in ≈21% of SAH patients and is a leading cause of long-term morbidity.4,5
Activation of the systemic immune response, manifested by elevated levels of circulating cytokines, is believed to play a significant role in the pathogenesis of cerebrovascular vasospasm.6–8 High-sensitivity C-reactive protein (hsCRP) is an acute-phase reactant that, aside from its role as a marker of infection or inflammation, has a wide variety of biological properties and functions that impact on vascular disease.9 Analyses of hsCRP after SAH have found higher levels in patients with ischemic complications from vasospasm, even after correcting for the presence of infectious complications.10,11
The effect of the inflammatory response on energy expenditure and systemic oxygen consumption is well-described after traumatic brain injuries and sepsis.12,13 Of the few studies that have assessed the metabolic response after SAH, the average oxygen consumption and energy expenditure values have been found to be similar to that of traumatic brain injuries and sepsis patients, with patients with vasospasm demonstrating the highest elevations in energy expenditure.14,15 Inflammation-mediated elevations in oxygen consumption may be an important precursor to the development of DCI; however, no previous study has analyzed the interaction between any marker of inflammation and metabolism after SAH. In the present study, we sought to define the interaction between hsCRP and systemic oxygen consumption. We hypothesized that hsCRP would be directly related to systemic oxygen consumption levels in the first 14 days after SAH, and further that elevations in systemic oxygen consumption (Vo2) would be associated with the development of DCI after SAH.
Patients and Methods
Patient Selection and Data Collection
This is a prospective observational study of aneurysmal SAH patients admitted to the neurological intensive care unit at Columbia University Medical Center between May 2008 and June 2010. The clinical care for SAH patients at Columbia University Medical Center has been described previously16 and conforms to guidelines set forth by the American Heart Association.3 All patients undergoing aneurysmal clipping were treated with intravenous steroids (dexamethasone 10 mg) intraoperatively and were continued on a scheduled taper for the first 5 postoperative days.
We sought to understand the longitudinal relationship between hsCRP and Vo2 in only those aneurysmal SAH patients who were critically injured but would survive for at least 1 week after injury. Therefore, we established the following criteria for inclusion in this study: (1) SAH attributable to a ruptured aneurysm as detected by digital subtraction angiography; (2) age 18 years or older; and (3) admission to the neurological intensive care unit within 48 hours of hemorrhage. The exclusion criteria were: (1) death from withdrawal of care or brain death expected within 72 hours of hemorrhage; (2) intensive care unit length of stay expected to be ≤72 hours; (3) unable to perform indirect calorimetry (IDC) within 72 hours of hemorrhage because of patient refusal, agitation, or high fraction of inspired oxygen (Fio2) requirement (≥0.6); and (4) inability to assess urine urea nitrogen levels because of anuria.
All study patients underwent serial assessments of hsCRP as well as metabolic parameters during the first 14 days after SAH. Each assessment was conducted once during 4 predefined time periods or phases: days 0 to 3 after bleed; days 4 to 7 after bleed; days 8 to 10 after bleed; and days 11 to 14 after bleed. All hsCRP and metabolic parameters were measured during the same time within each phase. Data collection was considered complete in instances when patients died or were discharged from the hospital before completion of the 4 phases. The clinical intensive care unit team was blinded to all IDC results and hsCRP measurements.
SAH Data Collection
This study was conducted in parallel to our Subarachnoid Hemorrhage Outcomes Project, which has been previously described in detail.17,18 Briefly, Subarachnoid Hemorrhage Outcomes Project is a prospective outcomes database that since July 1996 has collected data regarding admission and in-hospital characteristics, as well as long-term global outcome in all SAH patients admitted to the neurological intensive care unit at Columbia University Medical Center. The development of the following infectious complications was tracked daily and classified according to established guidelines17: pneumonia; urinary tract; and blood stream, cerebrospinal fluid, and Clostridium difficile stool infections. DCI was defined as either the presence of symptomatic vasospasm or the presence of an infarction on CT scan attributable to vasospasm.4 Symptomatic vasospasm was defined as clinical deterioration (ie, a new focal deficit, decrease in level of consciousness, or both) in the presence of confirmed vasospasm determined by CT angiography or cerebral angiography. Decreased level of consciousness was defined as a 2-point decrease in the Glasgow Coma Scale score in a 24-hour period. All patients who experience clinical deterioration underwent CT angiography to determine the presence of vasospasm and rule out other causes of deterioration (eg, fever, hydrocephalus, rebleeding, cerebral edema), followed by medical or interventional therapy or both as indicated. All end points are classified according to a priori criteria and adjudicated weekly at a Subarachnoid Hemorrhage Outcomes Project database meeting. The adjudication process involves a consensus agreement of each end point by neurocritical care faculty (N.B., K.L., J.C., S.A.M.) after a complete review of source documentation, imaging, and laboratory tests.
Data from Subarachnoid Hemorrhage Outcomes Project were linked with data collected from this study at the time of analysis, enabling the assessment of SAH-specific end points such as day after bleed of symptomatic vasospasm and DCI, as well as global outcome measures such as modified Rankin Scale score at 14 days and 3 months after bleed.
Serum samples were assayed for hsCRP using an enzyme-linked immunoassay (BioCheck; normal range, <3.0 mg/L). White blood cell (WBC) counts were measured daily as part of routine laboratory testing and recorded as part of the assessment for inflammation and infectious disease status. Prealbumin levels (mg/dL) were measured from serum samples using an immunoturbidmetric assay (Cobas-Integra 400 Plus Chemistry Analyzer; normal range, 17–40 mg/dL).
Oxygen Consumption Measurements
IDC studies were performed once during each phase using a Vmax Spectra (Sensormedic) to measure inspired and expired concentrations of oxygen (O2) and carbon dioxide (CO2). In mechanically ventilated patients, the circuit system was connected to the oxygen delivery and exhaust systems of the ventilator. In nonmechanically ventilated patients, the circuit was connected to an air-tight canopy that covered the patient's head and neck and delivered a measured constant flow of air (21% O2). Both methods allowed for continuous measurements of oxygen and carbon dioxide concentration in the inspired and expired air, allowing calculation of oxygen consumption (Vo2, mL/min) and carbon dioxide production (Vco2, mL/min).
Regular device calibration was performed according to manufacturer guidelines to ensure accuracy of the oxygen and carbon dioxide sensory equipment. Each IDC assessment was based on a steady state, which was defined as a 20- to 30-minute interval during which average minute oxygen consumption (Vo2) and carbon dioxide production (Vco2) changed by <5% and <10%, respectively. Continuous measurements were averaged every 60 seconds throughout the entire IDC session. IDC studies were not performed in patients who required Fio2 ≥60% or who were known to be seizing. The absence of seizure in obtunded or comatose patients was confirmed by continuous video EEG at the time of IDC. Each of the aforementioned conditions previously has been shown to alter the reliability of IDC measurements.19
Continuous variables were assessed for normality. Normally distributed data were reported as a mean and standard deviation. Nonparametric data were reported and analyzed as a median with 25% and 75% percentile values. Categorical variables were reported as count and proportions in each group. Generalized estimating equation analyses were performed with an identity link function and an exchangeable within-group correlation structure to determine the relationship between hsCRP and Vo2, as well as factors that may predict either hsCRP or Vo2 measurements. Temperature was tracked by continuous reading from a bladder catheter temperature probe that was time-locked with the IDC device for later retrieval and analysis. Temperature burden (C*min) was calculated for each minute of IDC testing by the following equation: Tburden= temperaturebladder−37°C. Multivariate models were built by entering in those factors found to have P≤0.1 on univariate analysis. Among similar variables that were highly intercorrelated, only the variable with the largest Wald χ2 value and smallest probability value in the generalized estimating equation analysis was used as a candidate variable in the final multivariate model. The occurrence of DCI and symptomatic vasospasm was treated as a censored event by day after bleed and corresponding study phase. Baseline characteristics that were found on univariate analysis to be associated (P<0.1) with DCI along with Vo2 were entered into a Cox proportional hazards model to calculate hazards ratios (HR) and corresponding 95% CI for developing DCI. For all tests, significance was set at P<0.05. All analyses were performed with SPSS version 18.0 (Chicago, Illinois).
Given the observational design, critical illness of participants, and utilization of residual blood and urine for laboratory assessments, this study was conducted with a waiver of consent. Data were linked with the Subarachnoid Hemorrhage Outcomes Project database, which utilizes a tiered consent process, whereby consent was obtained from those patients who were able to provide consent at the time of injury. In neurologically impaired patients, family members were approached for participation in the study. In cases in which capacity was regained, patients were directly approached for consent. This process of consent and the conduct of both studies were approved by the Institutional Review Board and were consistent with guiding principles for research involving humans.20
There were 160 aneurysmal SAH patients admitted during the study period, with 110 qualifying for inclusion in this study (Table 1). The reasons for exclusion included death or withdrawal of care within 72 hours of ictus (n=18), inability to perform calorimetry (n=15), and admission ≥72 hours after ictus (n=17).
Admission characteristics for the study population are shown in Table 1. The average intensive care unit length of stay was 14±7 days and the hospital length of stay was 21±11 days. Ninety percent (n=100) of patients survived or remained in the hospital long enough to complete 3 phases of study. The 110 SAH patients in this study underwent 383 serial Vo2 and hsCRP measurements.
Factors associated with Vo2 after SAH are shown in Table 2. Oxygen consumption was predicted by male gender (Wald=13.1; P=0.005) and younger age (Wald=7.9; P=0.005), but not by any clinical or radiographic measures of severity.
Pressors, Sedation, and Mechanical Ventilation
The majority of IDC assessments were performed when patients were not using intravenous vasoactive pressors (n=269/383; 70%), mechanically ventilated (n=249/383; 65%), and not sedated (n=280/ 383; 73%). The most common pressor utilized was norepinephrine (n=94; median dose, 0.1 μg/kg/min; range, 0.01 μg/kg/min–0.76 μg/kg/min), followed by neosynepherine (n=37; median dose, 0.3 μg/kg/min; range, 0.02 μg/kg/min–6.1 μg/kg/min), vasopressin (n=15; median dose, 0.3 U/min; range, 0.1–0.6 U/min), and dopamine (n=6; median dose, 5 μg/kg/min; range, 3–12 μg/kg/min). The most common sedative utilized was propofol (n=40/103; median dose, 20 μg/kg/min; range, 10–75 μg/kg/min), followed by fentanyl (n=40/103; median dose, 50 μg/hr; range, 12.5–100 μg/hr), and dexmedetomidine (n=23/103; median dose, 0.5 μg/kg/hr; range, 0.2–1.5 μg/kg/hr). There was no association between mechanical ventilation (Wald=2.2; P=0.1), sedation (Wald=1.0; P=0.3), or pressor use (Wald=2.6; P=0.1) and Vo2 measurements.
Temperature Modulation and Shivering
The mean maximal body temperature on the day of IDC was 37.6°C±0.4°C. The temperature burden during each IDC was 0.2±0.3°C*min and was not associated with Vo2 (Wald=0.6; P=0.5). There were 38 patients who had 82 IDC measurements performed during therapeutic normothermia (goal temperature, 37°C). The distribution of the bedside shivering assessment scale (BSAS)21 scores during therapeutic normothermia were as follows: BSAS 0 (none), 17 (21%); BSAS 1 (mild), 43 (52%); BSAS 2 (moderate), 17 (21%); and BSAS 3 (severe), 5 (6%). Overall, the application of therapeutic normothermia was not associated with Vo2 (Wald=1.3; P=0.3), although the degree of shivering assessed by the BSAS score during normothermia was associated with the Vo2 (Wald=6.5; P<0.01).
Fifty-two patients (47%) had a total of 60 infectious complications develop during the study period. Pneumonia was the most common type of infection (n=28; 47% of infections), followed by urinary tract infection (n=14; 23%), ventriculitis (n=10; 17%), blood stream infection (n=3; 5%), C. difficile infections (n=3; 5%), and wound infections (n=2; 3%). Both the presence of an infection (Wald=−17.2; P<0.001) and a higher WBC count (Wald=−8.3; P=0.004) were associated with higher hsCRP levels. The presence of an infection, administration of dexamethasone, and WBC counts were not associated with Vo2. When analyzing only patients who underwent aneurysm coiling (n=34), there was a trend between WBC and Vo2 (Wald=2.4; P=0.07).
The univariate and multivariate generalized estimating equation analyses are shown in Table 3. The factors independently associated with hsCRP throughout the entire 14-day period were Vo2 (Wald=−9.9; P=0.002), prealbumin level (Wald=−28.3; P<0.001), and presence of infection (Wald=−4.1; P=0.04).
Predictors of Vo2 After SAH
Younger age (Wald=−8.3; P=0.01), male gender (Wald=−8.4; P=0.002), and higher hsCRP levels (Wald=−4.2; P=0.04) were independent predictors of higher Vo2 in a multivariate generalized estimating equation model that adjusted for admission Acute Physiology and Chronic Health Evaluation II score, vasopressor use, and mechanical ventilation.
There were 24 (22%) patients who had DCI develop on median day 7 after bleed (range, 3–14). Univariate analysis of predictors of DCI on admission is shown in Table 1. The changes in both Vo2 and hsCRP over the first 14 days after SAH were significantly different for patients who had DCI develop (β=0.03; Wald=−7.7; P=0.05; Figure). The median time from hsCRP and Vo2 measurement to diagnosis of DCI was 24 hours (range, 12–48 hours). Younger age (adjusted HR for every 5 years of age, 1.2; 95% CI, 1.1–1.3), a higher modified Fisher score (adjusted HR for every 1-point increase, 3.4; 95% CI, 1.7–6.9), and higher oxygen consumption (adjusted HR for every 50 mL/min, 1.2; 95% CI, 1.1–1.3) were found to be independent factors predicting the occurrence of DCI in a backward stepwise (Wald) Cox proportional hazards model (Table 4). Mean hsCRP, WBC count, and female gender were not found to be independent predictors in the final model predicting time to DCI.
The median modified Rankin Scale score was 3 (interquartile range, 1–5), with 6 (5%) patients who died by day 14 after bleed. A higher 14-day mean Vo2 was associated with a higher 14-day modified Rankin Scale score (ANOVA: F=−2.4; P=0.03). At 3 months after bleed, the median modified Rankin Scale score was 3 (interquartile range, 1–4), with 15 (14%) patients who had died. A higher 14-day mean Vo2 was not associated with a higher 3-month modified Rankin Scale score (ANOVA: F=−0.9; P=0.5). A higher proportion of patients who had DCI develop had a poor outcome at 3 months after SAH (58% versus 35%; P=0.04) development of DCI was predictive of poor outcome (modified Rankin Scale score ≥4) at 3 months.
We found that systemic oxygen consumption (Vo2) levels in the first 2 weeks after SAH were correlated with hsCRP, even after adjusting for age, sex, vasopressor use, and mechanical ventilation at the time of measurement. Furthermore, Vo2 levels were an independent predictor of DCI in a model that adjusted for age, sex, and radiographic severity of bleeding.
The cascade of inflammatory responses seen after SAH and its relationship to cerebrovascular vasospasm and ischemia have been very well-described.8,22 In particular, higher hsCRP levels10,23,24 have been related to the severity of injury and implicated in the pathogenesis of vasospasm. By contrast, a hypermetabolic response, characterized by increases in systemic oxygen consumption, has not been well-described after SAH. One previous study found an association between resting energy expenditure and clinical severity with the occurrence of vasospasm;15 however, this analysis was limited to nonintubated patients and did not analyze Vo2 in terms of time to DCI. In the present study, we provide a more comprehensive assessment of metabolism by performing serial Vo2 measurements in a wider spectrum of both intubated and nonintubated patients in the first 2 weeks after injury. As a result, we found a dynamic Vo2 response after SAH that was related to hsCRP and development of DCI in the first 2 weeks after SAH. Neither of these associations has been reported previously.
The mechanisms involved with the high oxygen consumption observed in the present study cannot be fully identified; however, a similar dynamic interrelationship between oxygen consumption and hsCRP has been observed in other critical illnesses, with sequelae such as protein catabolism and lipid peroxidation.25–29 Both negative nitrogen balance and free fatty acid production have been reported to be associated with vasospasm after SAH.15,30 Our analysis found an additional finding of a strong inverse relationship between prealbumin, a nonspecific marker of protein catabolism, and hsCRP. A more comprehensive measurement of both protein catabolism and lipid peroxidation with assessments of nitrogen balance and free fatty acid production will lead to a better understanding of the impact of hypermetabolism on DCI.
There were unexpected results from our analyses. Based on previous studies,15,31 we expected that the severity of injury would have a direct influence on the systemic oxygen consumption level. One possible explanation for why we did not find such a relationship may center on the clinical care of SAH patients, such as the use of sedative agents. This may have masked the impact of severity on Vo2. Another possible explanation is that the impact of clinical severity on Vo2 may be mediated through activation of the inflammatory cascade, as demonstrated by the fact that hsCRP was associated with measures of clinical and radiographic severity and Vo2. We also expected there to be an association between core body temperature and Vo2; however, our practice of therapeutic normothermia (temperature, 37°C) eliminated measurements during febrile periods. As a result, this study does not address the impact fever may have on Vo2. As we have previously reported, shivering, as measured by the BSAS, was associated with Vo2.21 Finally, we anticipated that the use of vasopressors would be an independent predictor of Vo2, but concurrent use of sedative infusions during the majority of IDC measurements made during vasopressor usage (n=87/114; 76%) undoubtedly influenced the relationship between sedatives, vasopressors, and Vo2. Finally, we did not find an association between WBC, another nonspecific marker of inflammation, and Vo2, which may have been influenced by the use of dexamethasone in our clipped patients, as noted by the trend in association between WBC and Vo2 in our coiled patients.
Our study has several strengths. First, all measurements and data collection were conducted prospectively, with the clinical team blinded to results from IDC testing and hsCRP assessments. This eliminated any influence the knowledge of metabolic or inflammatory measurements may have had on diagnostic or therapeutic management, especially as it pertained to DCI. We carefully recorded and analyzed all pharmacological and physiological parameters that may have confounded the relationship between Vo2 and hsCRP. Finally, we used a strict validated definition, prospectively documented on the day of occurrence, for our primary end point, delayed cerebral ischemia,4 which has been previously validated as a strong predictor of outcome after SAH.
There are limitations to this study, however. Our inclusion criteria limited enrollment to those who were expected to survive the first 7 days after hemorrhage, diminishing our ability to comment on the relationship between the severity of injury and Vo2. We did not measure hsCRP and Vo2 measurements daily, potentially biasing our analyses regarding their inter-relationship and time to DCI. The median time difference between measurement and diagnosis was brief; therefore, the effect, if any, of a time delay between measurement and DCI diagnosis should be minimal. The hsCRP is a sensitive but nonspecific measure of inflammation and, despite our efforts to account for documented infections, levels of this marker may be confounded by infectious complications. Our findings, however, are similar to those of previous studies of hsCRP after SAH.10,32 Finally, we did not actually measure sympathetic activation, we only measured the impact of vasopressors. Therefore, we do not know the contribution of endogenous sympathetic activation on the relationship between Vo2 and DCI.
An association between hsCRP and oxygen consumption has potential implications for the nutritional management of SAH patients. Unlike sepsis patients,33 there are no current data to suggest that specialized enteral and parenteral nutritional formulations alter Vo2 by modifying the immune response after SAH. Further validation of our findings with more thorough assessments of inflammation in addition to measures of protein catabolism and lipid peroxidation may indicate a role for immunonutrition formulations in the management of DCI.
Sources of Funding
The project described was supported by grant UL1 RR024156 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH) and NIH Roadmap for Medical Research, and its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. Information on NCRR is available on the NCRR Web site. Information on re-engineering the clinical research enterprise can be obtained from NIH Roadmap website.
- Received January 20, 2011.
- Accepted April 4, 2011.
- © 2011 American Heart Association, Inc.
- King JT Jr.
- van Gijn J,
- Rinkel GJ
- Bederson JB,
- Connolly ES Jr.,
- Batjer HH,
- Dacey RG,
- Dion JE,
- Diringer MN,
- et al
- Frontera JA,
- Fernandez A,
- Schmidt JM,
- Claassen J,
- Wartenberg KE,
- Badjatia N,
- et al
- Devaraj S,
- Singh U,
- Jialal I
- Kubo Y,
- Ogasawara K,
- Kakino S,
- Kashimura H,
- Tomitsuka N,
- Sugawara A,
- et al
- Frankenfield D
- Esper DH,
- Coplin WM,
- Carhuapoma JR
- Badjatia N,
- Strongilis E,
- Gordon E,
- Prescutti M,
- Fernandez L,
- Fernandez A,
- et al
- Kacira T,
- Kemerdere R,
- Atukeren P,
- Hanimoglu H,
- Sanus GZ,
- Kucur M,
- et al