UBARACHNOID hemorrhage following aneurysm rup-ture is a common clinical entity that carries high mor-bidity and mortality rates. Clinical outcome in this

patient population has improved over the past few years,mostly because of advances in neurosurgical and neurocrit-ical care (that is, early aneurysm clipping, calcium channelblocker administration, and neuromonitoring).11,15 Nev-ertheless, cerebral ischemia secondary to vasospasm re-mains an important cause of morbidity and death.20 Cerebralsymptomatic vasospasm is most likely to occur between 4and 14 days after SAH and if left untreated can lead to per-manent neurological deficits or death.29,35 Patients with SAHfrequently become hypovolemic and hemodynamically im-paired a few days after symptom onset.23,29 This finding hasbeen implicated as a risk factor for development of symp-tomatic vasospasm and poor clinical outcome. Because ofthis, postoperative hypervolemic therapy is widely prac-

ticed in the US. Fluid infusions in patients with SAH haveconsisted of isotonic crystalloids (0.9% saline), 5 or 25%human albumin solutions, and more recently, hypertoniccrystalloids (3% saline).23,29,36

Normal saline and human albumin, administered eitherconcomitantly or in alternating fashion, are the solutionsmost frequently used to maintain normovolemia or hyper-volemia in patients after SAH. In all studies reported in theliterature in which investigators have commented on thetreatment of symptomatic vasospasm after SAH, human al-bumin has been used as a way of achieving hypervolemia.Therefore, we cannot conclude at this point whether al-bumin administration exerts any beneficial or deleteriouseffect in this patient population. Nevertheless, analysis ofrecent data obtained in animals has shown that human albu-min is neuroprotective in acute ischemic stroke models im-proving neurological function and reducing infarct volumeand brain edema.2,4,5,16 Because cerebral vasospasm leads tocerebral ischemia and infarction, we speculate that humanalbumin may also exert some neuroprotective effects in pa-tients with SAH.

J. Neurosurg. / Volume 100 / April, 2004

J Neurosurg 100:585–590, 2004

Effect of human albumin administration on clinical outcomeand hospital cost in patients with subarachnoid hemorrhage

JOSE I. SUAREZ, M.D., LARRY SHANNON, B.SC., OSAMA O. ZAIDAT, M.D.,MUHAMMAD F. SURI, M.D., GRWANT SINGH, M.D., GWENDOLYN LYNCH, M.D.,AND WARREN R. SELMAN, M.D.

Neurosciences Critical Care, Cerebrovascular Center, and Departments of Neurology andNeurosurgery, University Hospitals of Cleveland, Case Western Reserve University, Cleveland, Ohio

Object. Human albumin is used to induce hypervolemia (central venous pressure [CVP] � 8 mm Hg) after sub-arachnoid hemorrhage (SAH). Unfortunately, human albumin may increase the mortality rate in critically ill patients;because of this, its use became restricted in the authors’ hospital in May 1999. The goal of this study was to determinethe effect of human albumin on outcome and cost in patients with SAH before and after this restriction was put intoplace.

Methods. All patients with aneurysmal SAH who were admitted to the authors’ institution between May 1998 andMay 2000 were studied. Basic demographic information, dosage of human albumin given, length of stay, and the inci-dence of in-hospital deaths and complications were collected. The authors obtained Glasgow Outcome Scale (GOS)scores at 3 months after SAH (good outcome, GOS � 4). Data were analyzed using t-test and chi-square analysis. Lo-gistic regression was used to identify independent associations between use of human albumin and outcome.

The authors studied 140 patients: 63 who were admitted between May 1998 and May 1999 (Group 1) and 77 treat-ed between June 1999 and May 2000 (Group 2). Two subgroups of patients were further analyzed. Group 1 patientswho received human albumin (albumin subgroup, 37 patients) and Group 2 patients who would have received albu-min under the old protocol (that is, those who failed to achieve CVP � 8 mm Hg after normal saline administration;nonalbumin subgroup, 47 patients). Patients in the nonalbumin subgroup were more likely to be male (38% comparedwith 16%), to experience hypertension (55% compared with 30%), to suffer from hypomagnesemia (49% comparedwith 5.4%), and to have hydrocephalus (47% compared with 27%). There was a trend for these patients to have morevasospasm (28% compared with 19%, p = 0.2). Patients in the albumin subgroup were more likely to have a good out-come at 3 months.

Conclusions. Administration of human albumin after SAH may improve clinical outcome and reduce hospital cost.

KEY WORDS • albumin • subarachnoid hemorrhage • outcome

S

585

See the Editorial and the Response in this issue, pp 581–582.

Abbreviations used in this paper: CVP = central venous pressure;GCS = Glasgow Coma Scale; GOS = Glasgow Outcome Scale;NSU = Neurosciences Critical Care Unit; SAH = subarachnoid hem-orrhage; SD = standard deviation.

Reports in the critical care literature indicate an associ-ation between human albumin administration and excessmortality.8,30,39 Based on this body of evidence, on costs (500ml of 0.9% saline costs $0.54 compared with $38 for 250ml of 5% human albumin), and because of a national short-age of human albumin, its use in our NSU became restrict-ed in May 1999. Patients with SAH were treated with nor-mal saline to achieve prophylactic hypervolemia. In the oldprotocol, if bolus administration of normal saline (500 ml)failed to achieve CVP greater than 8 mm Hg, patients re-ceived human albumin (5 or 25%, 12.5 g). Since May 1999,patients with SAH have been treated only with crystalloidsat our institution. All other aspects of care were unchanged;this has allowed us to compare clinical data before and afterimplementation of the new policy. Our goal was to deter-mine the effect of human albumin administration on clini-cal outcome in patients after SAH. We also wanted to deter-mine the effect of this therapy on the duration of hospitalstay and costs.

Clinical Material and Methods

Patient Selection

All medical records of adult patients admitted to ourNSU with a diagnosis of SAH between May 1998 and May2000 were retrospectively screened. Only patients withconfirmed aneurysmal rupture and subsequent clip occlu-sion or endovascular treatment were considered for thestudy because they were the patients treated with hypervo-lemia (defined as CVP 8–12 mm Hg). We extracted pa-tients’ demographic data, medical comorbidities, baselinelaboratory values, Hunt and Hess grade, Fisher scale score(on head computerized tomography scans), and GCS37

score on admission. We also collected baseline CVP valuesand the total dose of human albumin administered, alongwith the duration of stay in the hospital and NSU, in-hospi-tal deaths, and total hospital and radiological costs. Med-ical complications studied included symptomatic vaso-spasm, communicating hydrocephalus (as determined by

the treating physicians and confirmed by the neuroradiolo-gy report), incidence of seizures, congestive heart failureor pneumonia (confirmed on chest x-ray films), electrolyteimbalances, and prolonged respiratory failure (� 96 hoursof mechanical ventilation).

Management Protocol for Patients With SAH

All patients underwent early (� 48 hours of admission)aneurysm clip placement or coil occlusion, and received ni-modipine for prophylaxis of symptomatic vasospasm andphenytoin for prophylaxis of seizures. All patients weretreated with hypervolemia (CVP � 8 mm Hg) immediate-ly after ruptured aneurysms were secured. Initial fluid man-agement consisted of intravenous bolus administration ofnormal saline (500 ml) to achieve hypervolemia. If patientsfailed to achieve hypervolemia after the initial treatment,they then received human albumin (12.5 g) followed byalternating boluses of normal saline and human albumin.After May 1999, patients were treated only with normalsaline to achieve hypervolemia. Elevation of blood pressure(mean arterial pressure 110–130 mm Hg) was instituted inpatients with symptomatic vasospasm. In patients with ahistory of heart disease or pulmonary edema, catheters wereinserted to maintain pulmonary artery occlusion pressuresof 12 to 18 mm Hg. Endovascular procedures (angioplastyand/or intraarterial papaverine injections) were performedat the discretion of the treating physicians if patients failedto improve clinically.

Diagnosis of Symptomatic Vasospasm

All records were carefully reviewed to identify episodesof neurological deterioration as documented by physiciansin the NSU. A diagnosis of symptomatic vasospasm wasbased on the following previously published criteria:10,20 1)clinical deterioration in the patient’s neurological conditionbetween 3 and 14 days after SAH (including insidious onsetof confusion, disorientation, or decline in the level of con-sciousness) or focal deficits that may fluctuate in severity;2) exclusion of structural causes of neurological worseningby appropriate investigations, including computerized to-mography scanning of the head; and 3) absence of otheridentifiable causes of neurological worsening such as serumelectrolyte or glucose alterations, hypoxia, hypercapnia, orseizures confirmed by clinical or electroencephalographic

J. I. Suarez, et al.

586 J. Neurosurg. / Volume 100 / April, 2004

TABLE 1Clinical features in 140 patients with SAH*

Demographic Data† Value

age (yrs) 56 � 13male sex (%) 49 (35)caucasian (%) 88 (63)medical history (%)

hypertension 53 (38)active smoker 34 (24)former smoker 7 (5)alcohol use 10 (7)CAD 12 (8.5)DM 6 (4)

examinationHunt & Hess grade �III (%) 98 (70)Fisher Grade 1–2 (%) 43 (31)GCS score �8 (%) 38 (27.1)baseline CVP (mm Hg) 5.1 � 2Hct 34.0 � 4.3

* CAD = coronary artery disease; DM = diabetes mellitus; Hct = hemat-ocrit.

† Age, baseline CVP, and hematocrit are given as the mean � SD.

TABLE 2Length of hospital stay and medical complications

experienced by 140 patients with SAH

Characteristic* Value

days in hospital 12.7 � 8.4days in NSU 10.0 � 7.3in-hospital deaths (%) 31 (22)complications (%)

hydrocephalus 49 (35)vasospasm 33 (23.5)hypomagnesemia 36 (26)respiratory failure 38 (27)seizures 15 (11)congestive heart failure 15 (11)pneumonia 11 (8)

* Days in hospital and days in NSU are given as the mean � SD.

findings. Clinical diagnosis of symptomatic vasospasm wasconfirmed using transcranial Doppler ultrasonography andcerebral angiography in all patients.

Outcome Measures

We obtained the GOS17 score from outpatient records orvia telephone interview to correspond to 3 months after di-agnosis of SAH. The GOS has been validated in numerousclinical trials in patients with various neurological and neu-rosurgical conditions and assessments are easily obtainablevia telephone or personal interview. The GOS scores rangefrom 1 to 5 and include the following categories: GOS 5, noor minimal disability; GOS 4, moderate disability; GOS3, severe disability; GOS 2, vegetative state; and GOS 1,death. For purposes of statistical analysis the GOS was di-chotomized as good outcome (Scores 4–5) or poor outcome(Scores 1–3).

This study was reviewed and approved by the Institu-tional Review Board at our institution, and all proceduresdescribed in this study were in accordance with Boardguidelines. All patients or their legal representatives gavetheir consent before we obtained GOS scores. Also, pa-tients’ primary care physicians or physicians of record atthe time of admission were contacted before we reviewedthe medical records.

Statistical Analysis

A Kolmogorov–Smirnov test for normality and an equalvariances test were completed before further statistical anal-ysis of these data were performed. Continuous variableswere compared using two-tailed t-tests, and proportionswere compared using the chi-square or Fisher exact test be-tween groups. Independent risk factors for good outcome at3 months based on the GOS were determined using logis-tic regression analysis. We considered a probability valueof less than 0.05 significant. All data are expressed as themean � SD or as the median.

Results

We identified 164 consecutive patients with SAH whowere treated during the study period. Of these, 24 were ex-cluded for the following reasons: in 19 there was no evi-dence of cerebral aneurysm on cerebral angiography stud-ies, one had an arteriovenous fistula, and four others hadarteriovenous malformations. Patients were divided intotwo groups: those treated between May 1998 and May 1999(Group 1, 63 patients), and those treated between June 1999and May 2000 (Group 2, 77 patients). A total of 11 patientswere lost to follow up, and they were assigned the worstpossible score for the outcome measure studied (five pa-tients in Group 1 and six in Group 2).

Patient Characteristics

Most of the patients studied were female and caucasian,and hypertension was the most common medical comor-bidity (Table 1). Physical examination revealed that the ma-jority of patients presented with low Hunt and Hess gradesbut with a significant amount of intracranial blood as deter-mined by Fisher grades. Patients stayed in the hospital amean of 12 days and 22% of them died during their admis-

sion (Table 2). The most common complications, in order offrequency, were communicating hydrocephalus, respiratoryfailure, hypomagnesemia, and symptomatic vasospasm.

Group Analysis

We identified and studied 63 patients in Group 1 and 77in Group 2 (Table 3). The proportion of male patients wassignificantly higher in Group 2. Total hospital cost was alsosignificantly higher in Group 2 compared with Group 1;this difference was mostly driven by higher radiology costsin Group 2. Although the difference was not statisticallysignificant, patients in Group 2 had an increased proportionof symptomatic vasospasm and GCS scores less than 8. Lo-gistic regression analysis revealed that after adjusting forage, sex, race, and GCS score less than 8, the administrationof human albumin was not a predictor of in-hospital deathor poor clinical outcome at 3 months after discharge.

Subgroup Analysis

Because not all patients in Group 1 received human albu-min and not all those in Group 2 received extra boluses ofnormal saline, we further analyzed each group of patientsand classified them as albumin and nonalbumin subgroups.The albumin subgroup was made up of 37 patients (58.7%)in Group 1 who received human albumin. The nonalbuminsubgroup consisted of 47 patients (61%) in Group 2 whoreceived extra boluses of normal saline to achieve hyper-volemia. The latter subgroup of patients would have re-ceived human albumin if the policy for administering thisagent had not changed. The total dosage of human albuminranged from 1.5 to 6 g/kg.

In the univariate analysis patients in the nonalbumin sub-group were more likely to be male, to experience hyper-tension, and to suffer from hypomagnesemia and hydro-cephalus (Table 4). There was a trend for this group toexperience more symptomatic vasospasm and to have ahigher in-hospital mortality rate. There was no difference,however, in the incidence of cardiorespiratory complica-tions between the subgroups. The timing of presentation

J. Neurosurg. / Volume 100 / April, 2004

Human albumin in patients with subarachnoid hemorrhage

587

TABLE 3Comparison of patients’ main characteristics by group*

Features Group 1 Group 2 p Value†

no. of patients 63 77age (yrs) 55 � 14 58 � 16 0.3male sex (%) 16 (25) 33 (43) 0.03caucasian (%) 34 (54) 54 (70) 0.16GCS score �8 (%) 14 (22) 24 (31) 0.2baseline CVP (mm Hg) 5.2 � 2 4.8 � 2 0.6symp vaso (%) 8 (13) 14 (18) 0.3days in hospital 12.4 � 10.6 13.0 � 7.1 0.7days in NSU 9.6 � 7.5 10.4 � 7.1 0.5cost data (US$ � 1000)

total hospital 62.0 � 39.0 81.0 � 49.0 0.02laboratory 3.7 � 2.9 4.4 � 3.5 0.3radiology 15.0 � 12.0 23.0 � 16.0 �0.01

* Group 1 included patients treated between May 1998 and May 1999;Group 2 patients were treated between June 1999 and May 2000. Cost dataare corrected for inflation. Unless otherwise noted, values are given as themean � SD. Abbreviation: symp vaso = symptomatic vasospasm.

† According to the t-test or chi-square test, as appropriate.

of symptomatic vasospasm was similar between the sub-groups (albumin subgroup, 5.8 � 3 days; nonalbumin sub-group, 6.1 � 3 days). The proportion of patients with goodoutcome at 3 months was significantly higher in the al-bumin compared with the nonalbumin subgroup (Table 5).After adjusting for age, sex, race, and GCS score less than8, a logistic regression showed that human albumin admin-istration was an independent predictor of better outcome 3months after SAH. Nevertheless, such an effect was not evi-dent at 6 and 12 months post-SAH. Administration of hu-man albumin was not an independent predictor of in-hospi-tal death.

Discussion

We have presented our experience with the use of humanalbumin to achieve hypervolemia in patients after SAH.Our data indicate that human albumin is not harmful for thispatient population. In fact, it may be neuroprotective, asshown by improved clinical outcome after discharge andfewer in-hospital neurological complications.

Use of Human Albumin in Critically Ill Patients

Albumin is a 67-kD protein that is responsible for 80%of the colloid osmotic pressure of plasma and supplies mostof the acid/base buffering action of the plasma proteins.33 Itis a very important component of the body because it alsoserves as a vehicle for metabolites, is a factor in lipid metab-olism, and performs many other functions that are still be-ing elucidated.

Human albumin has been used in critically ill patientsfor more than 50 years.8,30,39 This is partly based on the no-tion that serum hypoalbuminemia is associated with an in-creased mortality rate in this patient population.14 Humanalbumin has been used mostly for treatment of hypovole-

mic shock, burns, and in clinical situations in which hypo-albuminemia is thought to be important. This practice hasbeen challenged because of findings associating human al-bumin administration with excess mortality.8,30,39 This hasled many physicians to change their practice and reconsid-er the use of this solution. Nevertheless, a recent metaanal-ysis has indicated that human albumin does not increasemortality rates in patients with various conditions such asa postoperative status, trauma, burns, hypoalbuminemia,and ascites.40 The authors of the aforementioned analysisstressed the lack of well-designed, powered clinical studiesin this area.

Use of Albumin in the NSU Population

Human albumin has been used mostly for maintenanceof hypervolemia or normovolemia in patients with SAH asdelineated in the preceding paragraph. Otherwise its use islimited to situations of hypovolemic shock, as is the case inother critical care units. Nevertheless, we cannot concludeat this point whether its administration exerts any beneficialor deleterious effect in this patient population because of thelack of randomized studies comparing human albumin withother intravenous fluid therapies. Our study represents aninitial attempt to investigate the effect of human albuminadministration in patients with significant neurological in-jury such as SAH. Our data indicate that human albuminadministration is not associated with increased healthcareexpenditures in patients with SAH when total hospital costis used as a surrogate measure. In fact, our findings indicatethat despite the cost associated with human albumin treat-ment (� $1500/day), its beneficial effects may offset theexpense. For example, we used more radiological resourcesin patients who did not receive human albumin.

Use of Human Albumin as a Neuroprotective Agent

Analysis of recent data obtained in animals has shownthat human albumin is neuroprotective in models of acuteischemic stroke, improving neurological function and re-ducing infarct volume and brain edema.2–5,16 Possible mech-anisms advanced to explain this neuroprotective effect ofhuman albumin in cerebral ischemia include the following:1) a direct protective effect on both parenchymal and vas-cular elements of the brain; 2) hemodilutional properties; 3)antioxidant effect; 4) maintenance of normal microvascularpermeability and inhibition of endothelial cell apoptosis; 5)cerebral dehydration; and 6) stimulation of glial scar for-mation and sustaining neuronal metabolism by increasingthe export of pyruvate to neurons.3 Because cerebral vaso-

J. I. Suarez, et al.

588 J. Neurosurg. / Volume 100 / April, 2004

TABLE 4Comparison of patients’ main characteristics by subgroup

Subgroup

Characteristic* Albumin Nonalbumin p Value†

no. of patients 37 47age (yrs) 53.0 � 15.0 54.9 � 14.2 0.5male sex (%) 6 (16) 18 (38) 0.01caucasian (%) 19 (51) 31 (66)history of 11 (30) 26 (55) �0.01

hypertension (%)admission Hct 34.8 � 5.6 33.3 � 3.9 0.2admission GCS 6 (16) 13 (28) 0.1

score �8 (%)days in hospital 15.8 � 11.3 14.5 � 6.3 0.5days in NSU 12.5 � 7.5 12.0 � 6.4 0.7in-hospital deaths (%) 2 (5.4) 9 (19) 0.07hypomagnesemia after 2 (5.4) 23 (49) �0.001

admission (%)congestive heart 6 (16) 9 (19) 0.7

failure (%)respiratory failure (%) 10 (27) 18 (38) 0.3hydrocephalus (%) 10 (27) 22 (47) 0.03symp vaso (%) 7 (19) 13 (28) 0.2

* Values for age, hematocrit, and days in the hospital and NSU are givenas the mean � SD.

† According to the t-test or chi-square test, as appropriate.

TABLE 5Proportion of patients with good outcome after SAH*

Subgroup

Albumin NonalbuminOutcome (37 patients) (47 patients) OR (CI)

good at 3 mos 68% 39% 3.2 (1.1–11.0)in-hospital deaths 5% 21% 0.2 (0.03–1.1)

* Human albumin administration as a predictor of good outcome (definedas GOS scores � 4) was analyzed using logistic regression (after adjustingfor age, sex, race, and GCS score � 8). Abbreviations: CI = confidence in-terval; OR = odds ratio.

spasm leads to cerebral ischemia and infarction, human al-bumin administration may be appropriate to ameliorate theresulting neurological disturbances in patients after SAH.

Despite the fact that symptomatic vasospasm is associ-ated with significant rates of morbidity and mortality af-ter SAH, the mechanisms leading to it remain to be fullyunraveled.9 Blood products, particularly oxyhemoglobin,most likely contribute to cerebral vasospasm by an as yetunknown pathway. Nevertheless, various potential conse-quences of SAH that may be directly linked to the patho-physiological features of symptomatic vasospasm havebeen revealed in several studies.7,9,13,18,19,21,22,24–28,31,34,38,43 Theseconsequences include the following: 1) decreased endothe-lium-dependent relaxation through nitric oxide scavenging,or inactivation of guanylate cyclase; 2) increased reactiveO2 species and oxidative stress; 3) enhanced production ofendothelin-1 and big endothelin-1, which are potent vaso-constrictors; 4) lipid peroxidation; 5) increased activity ofthe protein kinase C pathway, leading to augmented intra-cellular calcium concentration; 6) migration of inflammato-ry cells across the endothelial surface; 7) increased produc-tion of endothelial growth factor; and 8) induction of genessuch as heme oxygenase-1 that may help counteract vaso-spasm.

Human albumin has been shown to interact with the en-dothelial cell surfaces, which influences its properties.42

Based on this interaction and its consequences, severalmechanisms of action for the neuroprotective effect of hu-man albumin in patients with SAH have been proposed,1,6,12,

32,41,42 including the following: 1) modulation of intracellularcalcium concentrations; 2) modulation of arachidonic acidrelease and membrane fluidity; 3) inhibition of apoptosis; 4)modulation of nitric oxide metabolism; 5) modulation ofcellular antioxidant redox signal by increasing glutathioneconcentration and decreasing tumor necrosis factor �–me-diated nuclear factor-� B activation; and 6) inhibition of ad-hesion molecule expression. The presence of a predominantmechanism is unlikely. In fact, it may be more likely that allor many of the mechanisms mentioned are simultaneous-ly exerting their effect. The end result of human albumin–cerebral endothelium interaction would be an ameliorationor prevention of symptomatic vasospasm. Because of thecomplex processes leading to symptomatic vasospasm, theadministration of a multifunctional compound with neuro-protective effects (human albumin) to patients with SAHwould be logical.

Study Limitations

The main limitation of our study is its design; its retro-spective, nonrandomized nature and the small sample size(limited power) may have introduced biases regarding pa-tient and data collection. Nevertheless, the fact that patientcare protocols except for human albumin administrationwere not changed during the study period gives reassurancethat the results obtained may represent a possible beneficialeffect of human albumin on clinical outcome. The patientsincluded represent the cohort of all individuals who weresequentially admitted to our institution during the study pe-riod. Some patients were lost to follow up; they were as-signed the worst possible outcome. The proportion of pa-tients with such outcomes was slightly higher for Subgroup1, giving more weight to the possible neuroprotective effectof human albumin.

Conclusions

We have presented a retrospective study of the possibleneuroprotective effect of human albumin in patients withSAH. Our preliminary data indicate that human albuminadministration may be associated with a decreased inci-dence of symptomatic vasospasm and an increased propor-tion of good clinical outcome after SAH. Further validationof our data with appropriate prospective studies is neededbefore human albumin administration can be recommendedas a neuroprotective therapy for patients with SAH.

References

1. Beck R, Bertolino S, Abbot SE, et al: Modulation of arachidonicacid release and membrane fluidity by albumin in vascular smoothmuscle and endothelial cells. Circ Res 83:923–931, 1988

2. Belayev L, Busto R, Zhao W, et al: Effect of delayed albuminhemodilution on infarction volume and brain edema after tran-sient middle cerebral artery occlusion in rats. J Neurosurg 187:595–601, 1997

3. Belayev L, Liu Y, Zhao W, et al: Human albumin therapy of acuteischemic stroke: marked neuroprotective efficacy at moderatedoses and with a broad therapeutic window. Stroke 32:553–560,2001

4. Belayev L, Zhao W, Pattany PM, et al: Diffusion-weighted mag-netic resonance imaging confirms marked neuroprotective effi-cacy of albumin therapy in focal cerebral ischemia. Stroke 29:2587–2599, 1998

5. Berger A: Why albumin may not work. Br Med J 317:240, 19986. Cantin AM, Paquette B, Richter M, et al: Albumin-mediated reg-

ulation of cellular glutathione and nuclear factor kappa B activa-tion. Am J Respir Crit Care Med 162:1539–1546, 2000

7. Clatterbuck RE, Oshiro EM, Hoffman PA, et al: Inhibition of va-sospasm with lymphocyte function-associated antigen–1 mono-clonal antibody in a femoral artery model in rats. J Neurosurg 97:676–682, 2002

8. Cochrane Injuries Group Albumin Reviewers: Human albuminadministration in critically ill patients: systematic review of ran-domised controlled trials. Br Med J 317:235–240, 1998

9. Dietrich HH, Dacey RG Jr: Molecular keys to the problems of ce-rebral vasospasm. Neurosurgery 46:517–530, 2000

10. Elliott JP, Newell DW, Lam DJ, et al: Comparison of balloon an-gioplasty and papaverine infusion for the treatment of vasospasmfollowing aneurysmal subarachnoid hemorrhage. J Neurosurg88:277–284, 1998

11. Feigin VL, Rinkel GJE, Algra A, et al: Calcium antagonists in pa-tients with aneurysmal subarachnoid hemorrhage: a systematic re-view. Neurology 50:876–883, 1998

12. Fuentes E, Nadal A, Jacob R, et al: Actions of serum and plasmaalbumin on intracellular Ca2+ in human endothelial cells. J Phys-iol (Lond) 504:315–326, 1997

13. Gabikian P, Clatterbuck RE, Eberhart CG, et al: Prevention of ex-perimental cerebral vasospasm by intracranial delivery of a nitricoxide donor from a controlled-release polymer: toxicity and effi-cacy studies in rabbits and rats. Stroke 33:2681–2686, 2002

14. Goldwasser P, Feldman J: Association of serum albumin and mor-tality risk. J Clin Epidemiol 50:693–703, 1997

15. Hop JW, Rinkel GJE, Algra A, et al: Case-fatality rates and func-tional outcome after subarachnoid hemorrhage: a systematic re-view. Stroke 28:660–664, 1997

16. Huh PW, Belayev L, Zhao W, et al: The effect of high-dose albu-min therapy on local cerebral perfusion after transient focal cere-bral ischemia in rats. Brain Res 804:105–1113, 1998

17. Jennett B, Bond M: Assessment of outcome after severe braindamage. A practical scale. Lancet 1:480–484, 1975

18. Juvela S: Plasma endothelin and big endothelin concentrationsand serum endothelin-converting enzyme activity following aneu-

J. Neurosurg. / Volume 100 / April, 2004

Human albumin in patients with subarachnoid hemorrhage

589

590 J. Neurosurg. / Volume 100 / April, 2004

rysmal subarachnoid hemorrhage. J Neurosurg 97:1287–1293,2002

19. Kamezaki T, Yanaka K, Nagase S, et al: Increased levels of lipidperoxides as predictive of symptomatic vasospasm and poor out-come after aneurysmal subarachnoid hemorrhage. J Neurosurg97:1302–1305, 2002

20. Kassell NF, Torner JC, Haley EC Jr, et al: The International Coop-erative Study on the Timing of Aneurysm Surgery. Part 1: Over-all management results. J Neurosurg 73:18–36, 1990

21. Kim DE, Sun YS, Lee MS, et al: Vascular NAD(P)H oxidase trig-gers delayed cerebral vasospasm after subarachnoid hemorrhagein rats. Stroke 33:2687–2691, 2002

22. Koide M, Nishizawa S, Ohta S, et al: Chronological changes of thecontractile mechanism in prolonged vasospasm after subarach-noid hemorrhage: from protein kinase C to protein tyrosine ki-nase. Neurosurgery 51:1468–1476, 2002

23. Lennihan L, Mayer SA, Fink ME, et al: Effect of hypervolemictherapy on cerebral blood flow after subarachnoid hemorrhage: arandomized controlled trial. Stroke 31:383–391, 2000

24. Macdonald RL: Pathophysiology and molecular genetics of vaso-spasm. Acta Neurochir Suppl 77:7–11, 2001

25. McGirt MJ, Lynch JR, Blessing R, et al: Serum von Willebrandfactor, matrix metalloproteinase-9, and vascular endothelial-growth factor levels predict the onset of cerebral vasospasm af-ter aneurysmal subarachnoid hemorrhage. Neurosurgery 51:1128–1135, 2002

26. McGirt MJ, Lynch JR, Parra A, et al: Simvastatin increases endo-thelial nitric oxide synthase and ameliorates cerebral vasospasmresulting from subarachnoid hemorrhage. Stroke 33:2950–2956,2002

27. McGirt MJ, Parra A, Sheng H, et al: Attenuation of cerebral vaso-spasm after subarachnoid hemorrhage in mice overexpressing ex-tracellular superoxide dismutase. Stroke 33:2317–2323, 2002

28. Mocco J, Mack WJ, Kim GH, et al: Rise in serum soluble inter-cellular adhesion molecule-1 levels with vasospasm following an-eurysmal subarachnoid hemorrhage. J Neurosurg 97:537–541,2002

29. Mori K, Arai H, Nakajima K, et al: Hemorheological and he-modynamic analysis of hypervolemic hemodilution therapy forcerebral vasospasm after aneurysmal subarachnoid hemorrhage.Stroke 26:1620–1626, 1995

30. Offringa M: Excess mortality after human albumin administrationin critically ill patients. Clinical and pathophysiological evidencesuggests albumin is harmful. Br Med J 317:223–224, 1998 (Edi-torial)

31. Ono S, Komuro T, Macdonald RL: Heme oxygenase-1 genetherapy for prevention of vasospasm in rats. J Neurosurg 96:1094–1102, 2002

32. Osterloh K, Ewert U, Pries AR: Interaction of albumin with theendothelial cell surface. Am J Physiol Heart Circ Physiol 283:H398–H405, 2002

33. Peters T Jr: All About Albumin: Biochemistry, Genetics, andMedical Applications. San Diego: Academic Press, 1996, pp188–250

34. Pilitsis JG, Coplin WM, O’Regan MH, et al: Free fatty acids in hu-man cerebrospinal fluid following subarachnoid hemorrhage andtheir potential role in vasospasm: a preliminary observation. JNeurosurg 97:272–279, 2002

35. Saito I, Ueda Y, Sano K: Significance of vasospasm in the treat-ment of ruptured intracranial aneurysms. J Neurosurg 47:412–429, 1977

36. Suarez JI, Qureshi AI, Parekh PD, et al: Administration of hyper-tonic (3%) sodium chloride/acetate in hyponatremic patients withsymptomatic vasospasm following subarachnoid hemorrhage. JNeurosurg Anesthesiol 11:178–184, 1999

37. Teasdale G, Jennett B: Assessment of coma and impaired con-sciousness. A practical scale. Lancet 2:81–84, 1974

38. Vatter H, Mursch K, Zimmerman M, et al: Endothelin-convertingenzyme activity in human cerebral circulation. Neurosurgery 51:445–452, 2002

39. Vermeulen LC Jr, Ratko TA, Erstad BL, et al: A paradigm for con-sensus: the University Hospital Consortium guidelines for the useof albumin, nonprotein colloid, and crystalloid solutions. ArchIntern Med 155:373–379, 1995

40. Wilkes MM, Navickis RJ: Patient survival after human albuminadministration. A meta-analysis of randomized, controlled trials.Ann Intern Med 135:149–164, 2001

41. Zhang WJ, Frei B: Albumin selectively inhibits TNF�-inducedexpression of vascular cell adhesion molecule-1 in human aorticendothelial cells. Cardiovasc Res 55:820–829, 2002

42. Zoellner H, Hofler M, Beckmann R, et al: Serum albumin is a spe-cific inhibitor of apoptosis in human endothelial cells. J Cell Sci109:2571–2580, 1996

43. Zuccarello M: Endothelin: the “prime suspect” in cerebral vaso-spasm. Acta Neurochir Suppl 77:61–65, 2001

Manuscript received August 13, 2003.Accepted in final form December 15, 2003.This study was supported by a grant from the Plasma Protein

Therapeutics Association to Dr. Suarez.Address reprint requests to: Jose I. Suarez, M.D., Neurosciences

Critical Care, University Hospitals of Cleveland, Hanna 5, 11100Euclid Avenue, Cleveland, Ohio 44106. email: Jose.Suarez@UHHS.com.

J. I. Suarez, et al.