symptomatic intracranial hemorrhage after stroke thrombolysis: the sedan score

Symptomatic intracranial hemorrhage

after stroke thrombolysis:

the SEDAN Score

Daniel Strbian, MD, PhD

Department of Neurology, Helsinki University Hospital

Helsinki, Finland

Atte Meretoja

Satu Mustanoja

Tiina Sairanen

Nina Forss

Markku Kaste

Turgut Tatlisumak

Basel, Switzerland

Stefan Engelter

Frank J Ahlhelm

Philippe Lyrer

Lausanne, Switzerland

Patrik Michel

Maria Cordier

Geneva, Switzerland

Lucka Sekoranja

Igor Kuzmanovic

tPA for ischemic stroke

tPA license: 1996 (USA), 1999 (Canada), 2002 (EU): but only a few percent of acute ischemic stroke patients are treated

~ 17% in Helsinki (Kaste M, Stroke 2007)

logistic problem: time window

fear of the most serious complication: symptomatic intracranial hemorrhage (sICH)

How many patients will have sICH?

Number Needed to Harm: 36.5 (fatal outcome) / 29.7 to 40.1 (any worsening, mRS)

Saver JL, Stroke 2007

But which patients?

prediction scores


Developing the score

Based on easily assessable variables

Applicable on / shortly after admission, prior to administration of iv tPA

Helsinki University Central Hospital

1104 consecutive ischemic stroke patients treated with iv tPA (1995-2008)

excluded: patients with basilar artery occlusion (n=119): very different protocol

the score was tested separately in this cohort

excluded: patients that underwent endovascular treatment (n=11): beyond the focus

final cohort: 974 patients


Developing the score – Imaging & Criteria

Baseline CT head scan: all patients

Follow-up imaging

routinely at 24 hours post tPA

whenever hemorrhage was suspected

CT: 940 / MRI: 20 / both: 14

Neuroradiologist: radiological ICH subtypes (HI-1, HI-2, PH-1, PH-2)

Stroke neurologist (without knowledge of patients’ 3-month outcome) classified sICH according to the ECASS-II criteria:

any hemorrhage (HI-1, HI-2, PH-1, PH-2)

clinical deterioration causing an increase in the NIHSS score of ≥ 4 points

time window: 7 days

the hemorrhage was likely to be the cause of the clinical deterioration

in case of doubt regarding whether edema or hemorrhage was the leading pathology, an association of the hemorrhage with the deterioration was assumed


Developing the score – Statistics

Univariate comparison of patients with and without sICH

the Mann-Whitney Rank Sum test or the Pearson χ2 test

Variables were entered into the logistic regression model based on univariate

associations (p<0.10)

Backward logistic regression with a stepwise removal of p<0.10

Various combinations of cut-off values of continuous baseline parameters

improvement of the model? (area under ROC curve)

Calibration of the model: the Hosmer-Lemeshow test


Developing the score – Univariate results

Univariate associations

age (p<0.01)

baseline NIH Stroke scale score (p<0.001)

hyperdense cerebral artery sign and early infarct signs on admission head CT scan (p<0.001)

baseline glucose level (p<0.01)

No univariate associations

gender, onset-to-treatment time, blood pressure prior to tPA, platelet count

medical history of hypertension, DM, A. fib, hyperlipidemia, CHF, previous stroke

prior medication: antihypertensive, OAC, any antithrombotic


Blood pressure prior to thrombolysis

Sare GM et al. Stroke 2009

Meretoja A et al. Stroke 2010

Kellert L et al. Stroke 2011


Blood pressure following thrombolysis Yong M & Kaste M. Stroke 2008

Ahmed N et al. Stroke 2009

unsignificant

significant Δ 0-24t

unsignificant

significant Δ 0-24t, Δ 0-7d


Developing the score – Final model

Characteristic Category Points Regression coefficient

(95% CI) p-value

Sugar (glucose) level on admission ≤ 8 mmol/L

(≤144 mg/dL) 0 Reference

8.1-12mmol/L

(145-216 mg/dL) 1 0.80 <0.01

> 12 mmol/L

(>216 mg/dL) 2 1.70 <0.001

Early infarct signs on admission CT head scan

no 0 Reference

yes 1 0.68 0.03

(hyper)Dense cerebral artery sign on admission CT

no 0 Reference

yes 1 0.89 <0.01

Age ≤ 75 years 0 Reference

> 75 years 1 0.65 0.02

NIHSS score on admission 0-9 points 0 Reference

10- points 1 0.71 0.02


SEDAN in the derivation cohort

Helsinki, n=974; sICH: 7.0% (5.5-8.8%)

Likelihood ratios

0: 0.19 (0.06-0.59)

1: 0.40 (0.21-0.76)

2: 1.23 (0.85-1.79)

3: 1.85 (1.26-2.72)

4: 3.68 (2.10-6.45)

5: 5.66 (2.40-13.40)

The Hosmer-Lemeshow test: 2 =1 .7, df=7, p=0.98

SEDAN in patients with BA occlusion (n=119)

0

10

20

30

40

50

60

0 1 2 3 4 5


Internal cross-validation

Accuracy of the final regression model (not of the score)

1000 bootstrap replicates (R-project 2.10.1)

93.1% (91.4-94.5)


SEDAN in the validation cohort


Basel, Lausanne, Geneva, n=828

sICH: 6.5% (5.0-8.4%)

Helsinki vs. Switzerland

very heterogeneous cohorts

Similar likelihood ratios

c-statistics=0.77 (0.71-0.83)

Comparison with other scores

0

20

40

60

0 1 2 3 4-5

HAT score_original % sICH, n=302, c=0.68

0

5

10

15

0 1 2 3 4-5

HAT score_validation % sICH, n=965, c=0.59

0

10

20

0 1 2 3-

MSS score_original % sICH n=481, c=0.68

0

10

20

30

40

0 1 2 3 4 5

SEDAN_Helsinki % sICH, n=972, c=0.77

0

10

20

30

0 1 2 3 4 5

SEDAN_validation % sICH, n=828, c=0.77

SEDAN: Strbian et al. Ann Neurol 2012 (in press)

Sugar 0 / 1 / 2 (<8-12 mmol/L>)

Early infarct signs 0 / 1

Dense artery sign 0 / 1

Age 0 / 1 (>75 y)

NIHSS 0 / 1 (≥10 p)

HAT: Lou et al. Neurology 2008

Validation (HAT&MSS): Cucchiara et al. Int J Stroke 2011

Glucose or DM 0 / 1 (>11.1 mmol/L)

Early infarct signs 0 / 1 / 2 (No; <1/3 MCA>)

NIHSS 0 / 1 / 2 (<15-20>)

MSS: Cucchiara et al. J Stroke Cerebrovasc Dis 2008

Glucose 0 / 1 (>8.3 mmol/L)

Age 0 / 1 (>60 y)

NIHSS 0 / 1 (>10 p)

Platelet count 0 / 1 (<150000/mm3) 0

5

10

0 1 2 3 4

MSS score_validation % sICH n=965, c=0.59


Limitations of the study

Additional baseline parameters may have impact on sICH

leukoaraiosis?

detailed information on the status of vessel occlusion, infarct core, and salvageable tissue

based on CT imaging only

(“Time is brain”; door-to-needle-time in Helsinki: median 22 minutes)

Interpretation of early infarct signs requires radiological expertise

stroke physician + radiologist in Helsinki

telemedicine consultation systems

Summary and Implications

The SEDAN score is simple, fast to perform, and reliably assesses risk for sICH

The score is based on 5 easily accessible baseline variables

Internal cross-validation showed high accuracy of the final model

The score was validated externally with similar results, AUC=0.77

Heterogeneity between the derivation and validation cohorts supports generalization

No single cut-off above which tPA should be withheld

The score scrutinizes the sICH risk to create a single number – helps in decision-making process

High-risk patients intensive post-tPA blood pressure & hyperglycemia monitoring and management

recruitment into RCTs: “add-on” trials studying reduction of tPA-related hemorrhage

to be used together with scores predicting the final functional outcome after iv tPA informed consent

? candidates for endovascular procedures?

The SEDAN score must be evaluated prospectively


symptomatic intracranial hemorrhage after stroke thrombolysis: the sedan score

Documents