Impaired Renal Function Is N

ot Associated with IncreasedVolume of Intracerebral Hemorrhage

Shawna Cutting, MD,* Chloe Castro, BS,* Vivien H. Lee, MD,*

and Shyam Prabhakaran, MD, MS†

From *Department o

Medical Center; and †

University, Chicago, Illin

Received August 3,

accepted September 19, 2

Address corresponden

sity Medical Center, De

W Harrison St, Ste 1121,

rush.edu.

1052-3057/$ - see front

� 2014 by National Str

http://dx.doi.org/10.1

86

Background: Patients with low glomerular filtration rate (GFR) are at risk for hemor-

rhagic stroke and experience poor long-term outcomes after stroke. These associa-

tions may be mediated by hematoma volume. We investigated the relationship

between impaired renal function (GFR ,60 mL/min/1.73 m2) and hematoma

size. Methods: Consecutive patients admitted with spontaneous intracerebral hem-

orrhage (sICH) between August 2006 and January 2010 were evaluated; patients

with acute kidney injury or end stage renal disease were excluded. Data on patient

demographics, admission creatinine, size and location of bleed, and disposition at

discharge were collected. Hematoma size and location and discharge outcomes

were compared among those with normal and impaired renal function. Results:Among 573 patients admitted with sICH, 411 met our study criteria (mean age

61.4 years; 52.3% female; median ICH volume 11.2 mL). Mean GFR was 73.8 mL/

min/1.73 m2, and 99 patients (24.1%) had a GFR ,60 mL/min/1.73 m2. There

was no correlation between admission GFR and ICH volume (rs 5 0.014; P 5 .77);

those with GFR ,60 versus $60 mL/min/1.73 m2 also had similar ICH volumes

(median 10.8 v 11.4 mL; P5 .54). There was no association between in-hospital mor-

tality and admission GFR, although those with GFR ,60 mL/min/1.73 m2 were

more likely to die or be discharged to nursing homes (adjusted odds ratio 1.9;

P 5 .03). Conclusions: In a large sICH cohort, impaired renal function was not

associated with final hematoma volume. Additional study should focus on

the mechanism by which renal function impacts functional outcomes after

sICH.KeyWords: Cerebral hematoma—glomerular filtration rate—kidney disease.

� 2014 by National Stroke Association

There is increasing interest in the relationship between

abnormal kidney function and spontaneous intracerebral

hemorrhage (sICH). Patients with low glomerular filtra-

tion rate (GFR) are known to have endothelial and platelet

dysfunction and are at risk for both thrombotic and hem-

orrhagic events.1 Those with low GFR are at greater risk

f Neurological Sciences, Rush University

Department of Neurology, Northwestern

ois.

2012; revision received August 29, 2012;

012.

ce to Shawna Cutting, MD, Rush Univer-

partment of Neurological Sciences, 1725

Chicago, IL 60612. E-mail: shawna_cutting@

matter

oke Association

016/j.jstrokecerebrovasdis.2012.09.010

Journal of Stroke and Cer

for hemorrhagic than ischemic stroke and experience

poorer long-term outcomes after stroke.2 Previous studies

have suggested that patients with chronic kidney disease

have larger lobar hemorrhages and increased mortality.3,4

We sought to validate whether kidney function was asso-

ciated with hematoma volume and discharge outcomes in

a large cohort of patients with sICH.

Methods

With institutional review board approval, all patients

hospitalized at our institution with sICH between August

2006 and January 2010 were prospectively entered into

an acute stroke registry. Exclusion criteria included:

(1) patients ,18 years of age; (2) patients with vascular

anomalies, such as aneurysm, arteriovenous malforma-

tion, fistula, or venous sinus thrombosis; (3) patients

ebrovascular Diseases, Vol. 23, No. 1 (January), 2014: pp 86-90

RENAL FUNCTION AND INTRACEREBRAL HEMORRHAGE 87

taking anticoagulant medications with heparin or warfa-

rin at time of presentation; (4) patients who were trans-

ferred from community hospitals with only an initial

scan and death within 24 hours (because of the inability

to calculate final hematoma volume); and (5) patients

who were later determined to have an underlying mass

on follow-up contrast-enhanced imaging.

Data collected included age, sex, race, current use of

antiplatelet medications, hypertension, initial Glasgow

Coma Scale (GCS) score, blood urea nitrogen and creati-

nine levels at admission, current smoking status, length

of hospital stay, and disposition at discharge. Data were

not consistently available regarding duration of antiplate-

let drug use before presentation or dose of aspirin. ICH

score was calculated using the previously published scor-

ing system.5 We defined poor outcome as discharge to

skilled nursing facility, hospice care, or in-hospital death.

Patients with end stage renal disease were excluded be-

cause of the potential confounder of heparin use during

dialysis. Patients with acute kidney injury were also ex-

cluded from our analysis because accurate measurements

of renal function were difficult to determine in this set-

ting. We defined acute kidney injury as a rise in creatinine

of 0.3 mg/dL over baseline, a definition based on ac-

cepted definitions for acute kidney injury.6 We used dis-

charge creatinine as a surrogate marker for baseline in

the absence of other data points, such as previous preho-

spitalization values. GFR was calculated using the Modi-

fication of Diet in Renal Disease (MDRD) equation.7

Impaired renal function was defined as ,60 mL/min/

1.73 m2.

The 24-hour computed tomography (CT) scan obtained

at our institution was reviewed by 1 study neurologist

(Dr. Lee), and final hematoma volume was calculated us-

ing the ABC/2 method as previously described.8 Hemor-

rhages were grouped into 4 locations: basal ganglia or

thalamus, lobar, infratentorial, or other (including multi-

ple locations or intraventricular hemorrhage). For our pri-

mary hypothesis, we performed correlational statistics

using the Spearman coefficient for the relationship be-

tween GFR and final intracerebral hemorrhage (ICH) vol-

ume and compared median hematoma volumes in

patients with normal versus impaired renal function us-

ing theMann–Whitney test. The relationships between re-

nal function and clinical factors, hematoma size, and

outcomes were assessed using the Student t or Mann–

Whitney test (for continuous variables) and the Chi-

square or Fisher exact test (for categorical variables).

Multivariable logistic regression analyses assessedwhether

kidney function was a predictor of poor discharge out-

come independent of known predictors, including the

ICH score and its individual elements. Adjusted odds ra-

tios (ORs) and 95% confidence intervals (CIs) were esti-

mated from the final model.

Data analysis was performed using SPSS software (ver-

sion 16.0; SPSS Inc, Chicago, IL). P , .05 was considered

statistically significant in all analyses. With our sample

size, we had .90% power to detect hematoma volume

differences of at least 5 mL between those with normal

and impaired renal function and 80% power to detect

an increased OR of early mortality of at least 1.72.

Results

Five hundred seventy-three patients were identified

with sICH. We excluded a total of 162 patients because

of end-stage renal disease (n 5 36), acute kidney injury

(n 5 104), missing data (n 5 11), or death before the final

CT scan (n 5 11). The remaining 411 patients ranged in

age from 25 to 96 years (mean 61.4 6 14.7 years); 52.3 %

were female. The median volume of hemorrhage was

11.2 mL (interquartile ratio 3.5-33.9 mL); 117 patients

(28.5%) had a large (.30 mL) hemorrhage volume. Two

hundred eight (50.6%) hemorrhages were located in the

basal ganglia or thalamus; 64 (15.6%) were infratentorial;

and 139 (33.8%) were lobar. Intraventricular extension

was observed in 179 (43.6%) patients. Mean GFR was

73.8 6 23.6 mL/min/1.73 m2, and 99 patients (24.1%)

had a GFR ,60 mL/min/1.73 m2. Table 1 summarizes

the patient demographic, clinical, and hemorrhage char-

acteristics.

On univariable analysis, patients with impaired renal

function were older (mean age 66.1 vs 59.9 years;

P , .001) and more likely to have hypertension (83.8%

vs 66.3%; P , .01) and diabetes (28.3% vs 18.3%). There

was no linear correlation between admission GFR and

volume of hemorrhage (r2 5 0.014; P 5 .77). Those with

GFR ,60 versus $60 mL/min/1.73 m2 also had similar

median ICH volumes (11.4 vs 10.8 mL; P 5 .54) and had

similar proportions with large (.30 mL) volume (31.3%

vs 27.6%; P 5 .47). In stratified analyses, GFR was lower

in patients with infratentorial hemorrhages (66.8 vs 75.0

mL/min/1.73 m2; P 5 .01) compared to other sites. GFR

did not differ in the presence or absence of intraventricu-

lar hemorrhage (74.1 vs 73.5 mL/min/1.73 m2; P 5 .80).

There was no association between in-hospital mortality

and GFR analyzed as a continuous variable (74.2 vs 72.1

mL/min/1.73 m2; P 5 .54) or when GFR was dichoto-

mized (,60 mL/min/1.73 m2 [24.2%] vs $60 mL/min/

1.73 m2 [17.3%]; P 5 .125) or trichotomized ($60 mL/

min/1.73 m2 [17.3%] vs 45-60 mL/min/1.73 m2 [20.3%]

vs ,45 mL/min/1.73 m2 [30.0%]; P 5 .150). Those with

low GFR were less likely to be discharged to home or to

acute rehabilitation facilities (45.5% vs 62.8%; P 5 .002;

Fig 1).

Using multivariable logistic regression to determine

predictors of poor discharge outcome by adjusting for

ICH score (model 1—all elements of ICH score included

separately; model 2—ICH score dichotomized as ,2

and $2), GFR ,60 mL/min/1.73 m2 was independently

associated with greater odds of poor discharge outcome

(Table 2).

Table 1. Patient demographics and hemorrhage characteristics

All patients (n 5 411)

GFR (mL/min/1.73 m2)

P value$60 (n 5 312) ,60 (n 5 99)

Baseline characteristics

Mean age, y (SD) 61.4 (14.7) 59.9 (14.0) 66.1 (15.9) ,.001

Female, n (%) 215 (52.3) 157 (50.3) 58 (58.6) .151

Race, n (%) .272

Black 239 (58.2) 183 (58.7) 56 (56.6)

White 123 (29.9) 93 (29.8) 30 (30.3)

Hispanic 41 (10.0) 28 (9.0) 13 (13.1)

Asian 8 (1.9) 8 (2.6) 0

Diabetes mellitus, n (%) 85 (20.7) 57 (18.3) 28 (28.3) .032

Hypertension, n (%) 290 (70.6) 207 (66.3) 83 (83.8) .001

Current tobacco use, n (%) 104 (25.3) 86 (27.6) 18 (18.2) .061

Current antiplatelet use, n (%) 101 (24.6) 74 (23.7) 27 (27.3) .474

Clinical data

Median GCS on admission (IQR) 14 (7-15) 14 (7-15) 13 (6-15) .137

GCS by severity, n (%) .092

13-15 242 (58.9) 185 (59.3) 57 (57.6)

5-12 104 (25.3) 84 (26.9) 20 (20.2)

3-4 65 (15.8) 43 (13.8) 22 (22.2)

Median length of stay, days (IQR) 5 (3-11) 6 (3-11) 5 (3-10) .055

Admission laboratory values, median (IQR)

Platelets 224 (178-261) 223 (173-261) 230 (187-266) .450

Prothrombin time 12.9 (12.2-13.6) 12.9 (12.2-13.6) 12.8 (12.2-13.5) .729

Partial thromboplastin time 26 (24-29) 26.0 (24-28.6) 26.4 (23-30) .198

Location, n (%) .498

Basal ganglia or thalamus 208 (50.6) 164 (39.9) 44 (44.4)

Lobar 139 (33.8) 103 (33.0) 36 (36.3)

Posterior fossa 60 (14.6) 42 (13.5) 18 (18.2)

Undefined/multiple 4 (1.0) 3 (1.0) 1 (1.0)

Intraventricular hemorrhage, n (%) 179 (43.6) 135 (43.3) 44 (44.4) .837

Median hematoma volume, mL (IQR) 11.2 (30.4) 11.4 (29.4) 10.8 (32.6) .537

ICH score, n (%) .052

0-1 297 (72.3) 233 (74.7) 64 (64.6)

$2 114 (27.7) 79 (25.3) 35 (35.4)

Mean GFR on admission, mL/min/1.73 m2 (SD) 73.8 (23.6) 83.0 (18.5) 45.1 (12.0) ,.001

Abbreviations: GCS, Glasgow Coma Scale; GFR, glomerular filtration rate; ICH, intracerebral hemorrhage; IQR, interquartile range.

S. CUTTING ET AL.88

Discussion

Among patients with sICH, we did not find any rela-

tionship between impaired renal function and final hema-

toma volume or in-hospital mortality. We did find greater

odds for poor outcome (defined as discharge to a skilled

nursing facility, hospice, or death) in patients with im-

paired renal function. This relationship was independent

of other known predictors of poor outcome.5,9 Impaired

renal function was not associated with a higher likelihood

of lobar hemorrhage, contrary to a previous report,3 but

was associated with infratentorial location.

There is an increasing body of literature on sICH in pa-

tients with impaired renal function, although the proxi-

mate effect of renal dysfunction on ICH characteristics

(i.e., size, ventricular extension, and location) is not clear.

Given the size of our cohort, power to detect differences

in hematoma size, and the low likelihood of type I and

type II errors in the primary analysis, the lack of associa-

tion between GFR and hematoma volume we observed is

noteworthy. We conclude that it is unlikely that hemor-

rhage volume mediates the relationship between renal

dysfunction and outcomes after ICH. There was also no

relationship between GFR and intraventricular extension,

although we did not quantify intraventricular hemor-

rhage using conventional methods such as the Graeb

score.10 While we observed a relationship between GFR

and hematoma location, we are unable to posit a reason-

able mechanism by which impaired renal function prefer-

entially results in infratentorial hemorrhage location. We

caution the overinterpretation of this finding given that

Figure 1. Histogram of patient discharge disposition.

RENAL FUNCTION AND INTRACEREBRAL HEMORRHAGE 89

it was derived from stratified analyses andmay be subject

to type I errors. The causal link between renal disease,

platelet dysfunction, and hematoma characteristics re-

quires additional supporting evidence.

Decreased GFR is a risk factor for the development of

hemorrhagic stroke11 and has been linked to higher

odds of in-hospital mortality after ischemic and hemor-

rhagic stroke, particularly when severe disease (GFR

,15 mL/min/1.73 m2) or end-stage renal disease is pres-

ent.12,13 In a study of acute stroke patients, GFR,45 mL/

min/1.73 m2 was associated with an increased OR for

death at 1 month and 1 year, particularly after hemor-

rhagic stroke.2 However, the subset with hemorrhage

was small (112 patients), and patients with end-stage re-

nal disease were included. Proteinuria on admission

may be associated with poorer discharge functional activ-

ity and a lower likelihood of being discharged home di-

rectly for patients with ischemic stroke.14 A more recent

study involving 94 patients with ICH found that those pa-

tients discharged home were less likely to have a reduced

GFR or proteinuria on admission, although with multi-

variable analysis these associations were not statistically

significant.15 We confirm the impact of renal function

on discharge outcomes, although the relationship with

Table 2. Multivariate logistic regression analysis for poor

outcome (defined as skilled nursing, hospice, or death)

Adjusted OR (95% CI) P value

Model 1

GFR ,60 1.954 (1.074-3.555) .028

Age .80 y 2.743 (1.323-5.687) .007

IVH 2.972 (1.757-5.029) ,.001

ICH .30 mL 3.790 (2.145-6.695) ,.001

GCS ,12 3.458 (1.921-6.224) ,.001

Model 2

GFR ,60 1.802 (1.044-3.110) .034

ICH score $2 15.575 (8.979-27.017) ,.001

Abbreviations: CI, confidence interval; GCS, Glasgow Coma

Scale; GFR, glomerular filtration rate; ICH, intracerebral hemor-

rhage; IVH, intraventricular hemorrhage; OR, odds ratio.

short-term mortality was not present. Our study, despite

its larger size, focused on moderate renal disease, and

therefore the effect of severe dysfunction on volume

could not be robustly assessed.

We speculate that other factors, such as medical comor-

bidities and secondary insults (i.e., ischemic injury, sei-

zures, and infections), may indirectly drive outcomes in

patients with renal dysfunction. Moderate to severe renal

disease is included in the Charlson comorbidity index,

which independently influences outcome after ICH.16

Long-term outcomes may also be influenced by elevated

risk of recurrent ischemic and hemorrhagic stroke in

this high-risk population. Therefore, impaired renal func-

tion likely represents an important, albeit indirect, marker

of long-term outcomes after ICH, but its proximate effect

on immediate or short-term outcomes is unclear.

Our study has some notable limitations. CT scans at

standardized intervals (i.e., 6 hours) were not routinely

obtained, and so hematoma expansion could not be con-

sistently evaluated. We therefore evaluated the relation-

ship between GFR and final ICH volume, which may be

a more important surrogate of the effect of renal function

on platelet function than hematoma growth. Functional

outcome measures (i.e., modified Rankin Scale scores)

were also not available at baseline, discharge, or at long-

term follow-up. We were unable to establish with cer-

tainty baseline renal function or duration of chronic

kidney disease before hospitalization and using discharge

renal function as a surrogate measure of baseline renal

function may be inaccurate.17

We could not show a relationship between GFR and he-

matoma volume, implying that other mechanisms may

explain the impact of renal function on outcome after

ICH. Given previous reports implicating the role of renal

function on risk of ICH, early and late mortality, and func-

tional outcomes,3,8 additional studies should investigate

whether these are mediated by acute hematoma expan-

sion, location of hemorrhage, risk of recurrent hemor-

rhage, medical complications, and long-term functional

outcomes.

References

1. Jalal DI, Chonchol M, Targher G. Disorders of hemostasisassociated with chronic kidney disease. Semin ThrombHemost 2010;36:34-40.

2. YahalomG, Schwartz R, Schwammenthal Y, et al. Chronickidney disease and clinical outcome in patients withacute stroke. Stroke 2009;40:1296-1303.

3. Molshatzki N, Orion D, Tsabari R, et al. Chronic kidneydisease in patients with acute intracerebral hemorrhage:Association with large hematoma volume and poor out-come. Cerebrovasc Dis 2011;31:271-277.

4. Rhoney DH, Parker D Jr, Millis SR, et al. Kidney dysfunc-tion at the time of intracerebral hemorrhage is associatedwith increased in-hospital mortality: A retrospective ob-servational cohort study. Neurol Res 2012;34:518-521.

S. CUTTING ET AL.90

5. Hemphill JC, Bonovich DC, Besmertis L, et al. The ICHscore: A simple, reliable grading scale for intracerebralhemorrhage. Stroke 2001;32:891-897.

6. Mehta RL, Kellum JK, Shah SV, et al. Acute Kidney InjuryNetwork: Report of an initiative to improve outcomes inacute kidney injury. Crit Care 2007;11:1-8.

7. Nephron Information Center web site. Fadem SZ,Rosenthal B. CKD EPI and MDRD GFR Calculator: 4variable MDRD CKD EPI equation (with SI units) usingstandardized serum creatinine, age, race, gender. Avail-able at http://www.nephron.com/MDRD_GFR.cgi. Ac-cessed April 3, 2012.

8. Kothari RU, Brott T, Broderick JP, et al. The ABCs of mea-suring intracerebral hemorrhage volumes. Stroke 1996;27:1304-1305.

9. Chanderraj R, Schwab K, FitzMaurice E, et al. Predictionof functional outcome in patients with primary intracere-bral hemorrhage: The FUNC score. Stroke 2008;39:2304-2309.

10. Graeb DA, Robertson WD, Lapointe JS, et al. Com-puted tomographic diagnosis of intraventricular hem-orrhage. Etiology and prognosis. Radiology 1982;143:91-96.

11. Bos MJ, Koudstaal PJ, Hofman A, et al. Decreased glo-merular filtration rate is a risk factor for hemorrhagicbut not for ischemic stroke: The Rotterdam study. Stroke2007;38:3127-3132.

12. Hojs Fabjan T, Hojs R, Tetickovic E, et al. Ischaemicstroke—Impact of renal dysfunction on in-hospital mor-tality. Eur J Neurol 2007;14:1351-1356.

13. Ovbiagele B. Chronic kidney disease and risk of deathduring hospitalization for stroke. J Neurol Sci 2011;301:46-50.

14. Ovbiagele B, Sanossian N, Liebeskind DS, et al. Indices ofkidney dysfunction and discharge outcomes in hospital-ized stroke patients without known renal disease. Cere-brovasc Dis 2009;28:582-588.

15. Ovbiagele B, Pineda S, Saver JL. Renal dysfunction anddischarge destination in patients with intracerebral hem-orrhage. J Stroke Cerebrovasc Dis 2011;20:145-149.

16. Bar B, Hemphill JB. Charlson comorbidity index ad-justment in intracerebral hemorrhage. Stroke 2011;42:2944-2946.

17. Haider DJ, Ferrari J, Mittermayer F. A transient improve-ment in renal function occurs after ischemic stroke. RenFail 2012;34:7-12.