performance and validation of a novel biomarker-based...

10.1161/CIRCULATIONAHA.116.022802

1

Performance and Validation of a Novel Biomarker-Based Stroke Risk Score

for Atrial Fibrillation

Running title: Oldgren et al.; Validation of the ABC-stroke score in AF

Jonas Oldgren, MD, PhD1,2; Ziad Hijazi, MD, PhD1,2; Johan Lindbäck, MSc2; John H.

Alexander, MD, MHS3; Stuart J. Connolly, MD4; John W. Eikelboom, MB, BS4; Michael D.

Ezekowitz, MB, ChB, PhD5; Christopher B. Granger, MD3; Elaine M. Hylek, MD, MPH6;

Renato D. Lopes, MD, PhD3; Agneta Siegbahn, MD, PhD2,7; Salim Yusuf, MD, PhD4; Lars

Wallentin, MD, PhD1,2 on behalf of the RE-LY and ARISTOTLE Investigators

1Department of Medical Sciences, Cardiology, Uppsala University, Uppsala, Sweden; 2Uppsala

Clinical Research Center, Uppsala University, Uppsala, Sweden; 3Duke Clinical Research

Institute, Duke Medicine, Durham, NC, USA; 4Population Health Research Institute, Hamilton,

Canada; 5Thomas Jefferson Medical College and the Heart Center, Wynnewood, PA, USA;6Boston University Medical Center, Boston, MA, USA; 7Department of Medical Sciences,

Clinical Chemistry, Uppsala University, Uppsala, Sweden

Address for Correspondence: Jonas Oldgren, M.D., Ph.D. Uppsala Clinical Research CenterUppsala Science ParkDag Hammarskjölds väg 14 BSE-752 37 Uppsala, Sweden Tel: +46 (0) 18 611 2765 Fax: +46 (0) 18 51 55 70 E-mail: [email protected]

Journal Subject Codes: Atrial Fibrillation; Cerebrovascular Disease/Stroke; Risk Factors;Anticoagulants

p g y

Renato D. Lopes, MD, PhD3; Agneta Siegbahn, MD, PhD2,7; Salim Yusuf, MDD, PhPhP DDD444;;; LaLaLars

Wallentin, MD, PhD1,2 on behalf of the RE-LY and ARISTOTLE Investigators

1Depapapartrtrtmemementntnt ooof MeMeM dical Sciences, Cardiology, UpUppsala Universiityt , UpUppsala, Sweden; 2Uppsala

CClC inical Reseaarcch CeCeentntn erer, UpUpUppspspsalaa a UUniversisity,, UpUppspspsalala aa, SSSwewedededenn;; 3DuD keee CCClililinininicacacal ReReResesesearararchchch

Instststitii ute, Dukkke e MeMedicicine, DuDD rham, NCC, USAA; 4PoPopuuulalalatititiononon HHHealtthh Resesearrrcch Instituuutte,, HHamilttoon,

Caaanananadadada;;; 555ThThThomomasas JJefeffersononon MMMedededicici alall CCCololo lelelegegege andnd ttthehehe HHHeaeaeartrtrt CCCenenentetet r,r, WWynynnnenenewowowoododod, , PAPA,, USAA;;;6B t U i it M di l C t B t MA USA 7D t t f M di l S i

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

by guest on M

ay 18, 2018http://circ.ahajournals.org/

Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from

http://circ.ahajournals.org/



























2

Abstract

Background—Atrial fibrillation (AF) is associated with increased but variable risk of stroke.

Our aim was to validate the recently developed biomarker-based ABC-stroke risk score and

compare its performance with the CHA2DS2VASc and ATRIA risk scores.

Methods—ABC-stroke score includes Age, Biomarkers (NT-proBNP and high-sensitivity [hs]

troponin [cTn]), and Clinical history (prior stroke). This validation was based on 8,356 patients,

16,137 person-years of follow-up, and 219 adjudicated stroke or systemic embolic (SE) events in

anticoagulated patients with AF in the RE-LY study. Levels of NT-proBNP, hs-cTnT, and hs-

cTnI were determined in plasma samples obtained at study entry.

Results—The ABC-stroke score was well calibrated with 0.76 stroke/SE events per 100 person-

years in the predefined low (<1%/year) risk group, 1.48 in the medium (1-2%/year) risk group,

and 2.60 in the high (>2%/year) risk group for the ABC-stroke score with hs-cTnT. Hazard ratios

for stroke/SE were 1.95 for medium versus low risk, and 3.44 for high versus low risk groups.

ABC-stroke score achieved C indices of 0.65 with both hs-cTnT and hs-cTnI, as compared with

0.60 for CHA2DS2VASc (p=0.004 for hs-cTnT and p=0.022 hs-cTnI) and 0.61 for ATRIA scores

(p=0.005 hs-cTnT and p=0.034 for hs-cTnI).

Conclusions—The biomarker-based ABC-stroke score was well calibrated and consistently

performed better than both the CHA2DS2VASc and ATRIA stroke scores. The ABC score should

be considered an improved decision support tool in the care of patients with AF.

Clinical Trial Registration—ClinicalTrials.gov identifier: ARISTOTLE; NCT00412984,

ClinicalTrials.gov identifier: RE-LY; NCT00262600

Key words: anticoagulation; atrial fibrillation; stroke prevention; risk model; risk prediction; stroke; oral anticoagulation

years in the predefined low (<1%/year) risk group, 1.48 in the medium (1-2%/year) ) rirr skkk grgrgrouououp,pp

and 2.60 in the high (>2%/year) risk group for the ABC-stroke score with hs-cTnTTT. HaHaH zazazardrdrd rrratatatioioio

for stroke/SE were 1.95 for medium versus low risk, and 3.44 for high versus low risk groups. k

ABC-stroke score achieved C indices of 0.65 with both hs-C cTnT and hs-cTnI, as compared with

0.6000 fffooro CHAAA222DDSD 2VVASc (((p=p 0.004 for hs-cTnT andnd p=0.022 hs-cTcTnI) and 0.61 for ATRIA score

p=0=0=0.005 hs-cTnTT aandd ppp=0.0003334 for hhsh -cTTnI).

Concncnclull sions———ThTThe biomomarkekeker-r baseed d d ABABABC-strororokke scocooreere was wwell ccalibrraatededed and conononsiststently

performed beb tter thah n ff bob th the CHA2DS2VAASc andd ATRIA stroke scores. Thhe ABA C score shhould



ownloaded from



3

Clinical Perspective

What Is New?

The ABC – Age, Biomarkers (hs Troponin and NT-proBNP), and Clinical history of

prior stroke – risk score has now been externally validated in a cohort of 8356 AF

patients randomized to two doses of dabigatran or warfarin.

The biomarker-based score was well calibrated, showed good discriminative ability, and

consistently performed better than the clinical CHA2DS2VASc and ATRIA stroke scores

regardless of type of oral anticoagulation.

What Are the Clinical Implications?

The biomarker-based ABC-stroke score containing only four variables seems useful for

stroke risk prediction in a broad population of anticoagulated patients with AF.

What Are the Clinical Implications?

ThThThe e e biomarker-based ABC-stroke score containing only four variables seems useful for

stroke rrrisissk k k ppredddicicictititionon iiin nn aa brbrbroaoo d d populaatitt onn of ananantitit cococoagagagululaatatedede ppatatientntnts ss wiwiwiththth AF.F.F



ownloaded from



4

Atrial fibrillation (AF) is the most common sustained arrhythmia and is associated with

increased but variable risk for stroke, which is substantially decreased by treatment with oral

anticoagulants.1-3 Risk stratification schemes in AF patients first emerged in the 1990s and have

since been reconstructed, refined, and evaluated in patients not receiving antithrombotic

treatment and in those patients receiving an oral anticoagulant.4-8 Current European9 and US10

AF guidelines recommend a risk-based approach to decisions on anticoagulation treatment in AF

based on the CHA2DS2VASc score, which assigns 1 point each for a history of Congestive heart

failure, Hypertension, Diabetes mellitus, Vascular disease, Age 65-74 years, and Sex category

[female gender], and 2

CHA2DS2VASc score is based solely on clinical variables while the more recent Anticoagulation

and Risk Factors in Atrial Fibrillation Study (ATRIA)8 risk score also includes a measure of

renal function by estimation of glomerular filtration rate.

We previously demonstrated that biomarkers, e.g. N-terminal fragment B-type natriuretic

peptide (NT-proBNP) indicating myocyte stress and high-sensitivity cardiac troponin (hs-cTn)

indicating myocardial injury, provide more prognostic information than most clinical

characteristics in patients with AF.11-14 In accordance with current recommendations for

developing, interpreting, and validating prediction models15-18 we developed a risk score

including the prognostically most important biomarkers and clinical characteristics. This stroke

risk score was derived in a cohort of 14,701 patients with AF and biomarkers measured at entry

in the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation

(ARISTOTLE) trial.19 The score was named after the included variables Age, Biomarkers – NT-

proBNP and hs-cTn – and Clinical history of prior stroke. The ABC-stroke risk score

significantly improved risk prediction compared with the CHA2DS2VASc score.20

CHA2DS2VASc score is based solely on clinical variables while the more recent AAntntnticiccoaoaoagugugulalalatitit on

and Risk Factors in Atrial Fibrillation Study (ATRIA)8 risk score also includes a measure of

enal function by estimation of glomerular filtration rate.

We preeevvviouuusly y y dedeemomom nsnsnstrtrtratatatededed thahat t biommmara keerrs, e.e.e g.g.g NNN---tett rmrmminnnalal ffraragmmmenenent t BBB---tyt peee nnnatatatriririurururete ic

peptptptidiide (NT-pprororoBNNP)) iindicccatatating myyyoccyyte streeesss aannd hhhigigighhh-sensn itivivity carrdiaiaiac tropononniin ((hs-cTn)n)

ndicating myocarddiai l l ini jury, providi e more prognostic information than most clinical



ownloaded from



5

The aim of the present study was to externally validate the ABC-stroke score in a large

cohort of anticoagulated patients with AF from the Randomized Evaluation of Long-Term

Anticoagulation Therapy (RE-LY) trial and to compare the performance and utility of the ABC-

stroke score with the CHA2DS2VASc and ATRIA stroke scores.

Methods

Study populations

Derivation cohort

Details of the ABC-stroke score derivation cohort from the ARISTOTLE trial have been

published previously.20, 21 Briefly, ARISTOTLE was a double blind clinical trial that randomized

patients with AF at increased risk for stroke to treatment with warfarin or apixaban. The median

length of follow-up was 1.9 years for the 14,701 out of 18,201 ARISTOTLE patients with

biomarker samples available at randomization.

Validation cohort

Details of the RE-LY trial have been published previously.22 Briefly, RE-LY was a prospective,

multicenter, randomized trial comparing two blinded doses of dabigatran with open label

warfarin that enrolled 18,113 patients with AF at 951 clinical sites in 44 countries between

December 2005 and March 2009. Inclusion criteria were documented atrial fibrillation and at

least one of the following risk factors for stroke: previous stroke or transient ischemic attack

(TIA); congestive heart failure or reduced left ventricular ejection fraction (<40%); at least 75

years of age; or at least 65 years of age with diabetes mellitus, hypertension, or coronary artery

disease. Exclusion criteria included: severe heart valve disorder; recent stroke; creatinine

clearance less than 30 mL/min; or active liver disease. The median length of follow-up was 1.9

published previously.20, 21 Briefly, ARISTOTLE was a double blind clinical trial tthahaat tt rararandndndomomomizizized

patients with AF at increased risk for stroke to treatment with warfarin or apixaban. The median

ength of follow-up was 1.9 years for the 14,701 out of 18,201 ARISTOTLE patients with

biommmaarker sampmpmples s s avvvaia lalalablblb e atatat rrrananandodd mimizationnn.

Valililidddation cohhhororo t

Details of theh RE-LLY trial have been publb isheh d previously.2222 BrB ieflf y, REE-LYY was a prospectiive,



ownloaded from



6

years for the 8,356 participants with biomarker samples available at randomization.

Endpoint and outcome assessment

Stroke was defined as the sudden onset of a focal neurologic deficit in a location consistent with

the territory of a major cerebral artery and categorized as ischemic, hemorrhagic, or unspecified,

in both the RE-LY22 and ARISTOTLE21 trials. Hemorrhagic transformation of ischemic stroke

was not considered to be hemorrhagic stroke. Systemic embolism (SE) was defined as an acute

vascular occlusion of an extremity or organ, documented by means of imaging, surgery, or

autopsy. Blinded Clinical Events Committees reviewed and centrally adjudicated all suspected

stroke and SE events in both trials. Both trials, including the biomarker programs, were based on

all patients’ written informed consent and approval by institutional review boards or ethics

committees.

Biochemical methods

Blood samples were collected in EDTA tubes at randomization and centrifuged immediately.

Plasma samples were frozen in aliquots, and stored at -70°C until analyzed centrally at the

Uppsala Clinical Research Center (UCR) laboratory, Uppsala, Sweden. Plasma high-sensitivity

cardiac troponin I (hs-cTnI) levels were determined with hs sandwich immunoassays on the

ARCHITECT i1000SR (Abbott Diagnostics) platform according to the instructions of the

manufacturer, NT-proBNP and high-sensitivity cardiac troponin T (hs-cTnT) levels in plasma

were determined with hs sandwich immunoassays on Cobas® Analytics e601 Immunoanalyzer

(Roche Diagnostics, Germany) according to the instructions of the manufacturer and have been

described previously.12-14

Statistical analyses

The derivation and internal validation of the ABC-stroke score has previously been described.20

all patients’ written informed consent and approval by institutional review boards s ororor eeethhhicicicsss

committees.

Biochemical methods

Blooooooddd samplesss wewewererere cccolollelelectctc eddd iiin n n EDEDEDTAA tubes aat raanndomomomizizizatatatioioion n aanand dd cec ntntrifuuugegeged d d imimimmeeedidiiatatatelelely.y.y.

Plaasmsmsma sampleeess wewere frrozennn in aliquots,, and ssts ooreded at t t -7-7-70°0°0°CCC uuntill aanalyzyzededd centrallllyyly aat the

Uppsala Clinical RResearch Center (UCR)R) labboratory, Uppsala, Sweden. Plasma hih gh-sensitivity



ownloaded from



7

The ABC-stroke model included the four variables: age, hs-cTnT (or hs-cTnI), NT-proBNP, and

prior stroke/TIA and is presented as a nomogram in the supplemental Figure S1.

The validation was performed in 8,356 patients from the RE-LY trial. For each patient

the one-year risk for stroke/SE was estimated by applying the model presented in the nomogram

(Supplemental Figure S1). Discrimination was assessed by the C index23 and by comparing

Kaplan-Meier curves and hazard ratios between the predefined risk categories. Risk categories

for the ABC-stroke score were defined as 0-1%, 1-2%, or >2% risk for stroke/SE within one

year.

Calibration was assessed by comparing one-year event rates with predictions from the

derivation model by fitting a Cox-regression model with the estimated one-year event probability

included as a restricted cubic spline. The ABC-stroke score was compared with the

CHA2DS2VASc risk score6, 7 and the ATRIA risk score8, for which calibration were assessed by

comparing observed one-year event rates with the previously published event rates from the

original derivation cohorts (Supplemental Table S1). C indices were compared using 2,000

bootstrap samples. In addition, the ABC model was evaluated in two subgroups: (1) in patients

without prior stroke and (2) in warfarin treated patients at sites with low mean time in therapeutic

range (TTR).

For the evaluation of clinical usefulness and net benefit, the derivation and validation

cohorts were combined to one large dataset, n=23,057. The net benefit of using the ABC-stroke

score as a clinical decision tool was estimated using decision curve analysis24. The net benefit at

a given decision threshold is defined as the difference between the proportion of true positives

and the proportion of false positives where the latter is weighted by the odds of the specific

threshold. The decision curve is then created by calculation of the net benefits for all possible

derivation model by fitting a Cox-regression model with the estimated one-year evevenenentt t prprprobobobabababilillity

ncluded as a restricted cubic spline. The ABC-stroke score was compared with the

CHA2DS2VASc risk score6, 7 and the ATRIA risk score8, for which calibration were assessed by

compmpmparing obseseservrvrvededed ooonenee---yeyey arrr eeeveveventntnt rattese with h h tht ee pprevvvioioiousususlylyly pububbliiishshedd eveeentntnt rratatatesese frooom m m thththeee

origgginininal derivaatatioioi nn ccohohorts (S(S(Supplemenntal Tabblb ee S1S1). CCC iiindndndicicees wweere coompmppaara ed usiingngn 22,,000 CCC

bootstrap samples. InI addition, the ABC moddel was evaluated iin two subgb roups: (1) in patients



ownloaded from



8

thresholds. Different prediction models can be compared in the decision curve analysis. At any

given threshold, the model with the higher net benefit is the preferred model. This is graphically

illustrated with a continuum of potential thresholds for stroke/SE risk (x-axis) and the net benefit

per patient (y-axis) relative to assuming that no patient will have a stroke/SE. The added value of

the ABC-stroke score was further illustrated by plotting Kaplan-Meier curves for the predefined

risk classes by subgroups of the other scores.

These analyses followed the framework for derivation and validation of prediction

models proposed by Harrell, Steyerberg, and Steyerberg and Vergouwe.16, 18, 23 The validation

followed the principles and methods described by Royston and Altman and the reporting

followed the recently published TRIPOD statement.15, 17 All analyses were performed using R

version 3.2 using the packages rms and Hmisc.23, 25 The algorithm for the ABC-stroke risk model

is presented in Supplementary Table S2.

Results

Baseline demographics, biomarker levels, and event rates in the validation cohort

Baseline demographics and median levels for the assessed biomarkers hs-cTnT, hs-cTnI, and

NT-proBNP are presented in Table 1. There were only minor differences in baseline

characteristics and biomarker levels between this validation cohort and the derivation cohort,20

except for a higher median age in the present cohort.

This external validation was based on 16,137 person-years of follow-up, and 219

adjudicated stroke or SE events corresponding to an incidence rate of 1.36 per 100 person-years.

The incidence rates (events per 100 person-years) within each predefined risk class were: 0.76

for low risk, 1.48 for medium risk, and 2.60 for high risk, for the ABC-stroke score with hs-cTnT

followed the recently published TRIPOD statement.15, 17 All analyses were perforrmememed dd usususinininggg R RR

version 3.2 using the packages rms and Hmisc.23, 25 The algorithm for the ABC-stroke risk mode

s presented in Supplementary Table S2.

Resususults

Baseline demographics, biomarker levels, and event rates in thhe validation cohort



ownloaded from



9

(Table 2). The results were similar when using the ABC-stroke score with hs-cTnI (Table 2).

Discrimination and calibration

The relative hazards in the validation cohort were 1.95 for medium vs. low risk, and 3.44 for

high vs. low risk for ABC-stroke score with hs-cTnT (Table 2). The cumulative event rates up to

2.5 years illustrate good calibration and discriminatory ability for the validation and derivation

cohorts (Figure 1). Results for ABC-stroke scores with hs-cTnT and hs-cTnI were similar (Table

2 and Supplemental Figures S2 and S3).

The ABC-stroke score achieved C indices of 0.65 with both hs-cTnT and hs-cTnI in the

validation cohort (Table 3). In the subgroup without prior stroke, the C indices for the ABC score

were 0.63 for hs-cTnT and 0.62 for hs-cTnI. In the TTR<65% subgroup of warfarin treated

patients, the C indices were 0.71 for hs-cTnT and 0.69 for hs-cTnI.

Comparison of the ABC-stroke score with CHA2DS2VASc and ATRIA scores

The ABC-stroke score C index of 0.65 (both hs-cTnT and hs-cTnI) was significantly higher than

the C index of 0.60 for the CHA2DS2VASc stroke risk score, p=0.004 (hs-cTnT) and p=0.022

(hs-cTnI) (Table 3). The C index of 0.61 for the ATRIA stroke score was also significantly lower

than with both hs-cTnT and hs-cTnI-based ABC-stroke scores, p=0.005 and p=0.034,

respectively (Table 3). Similar results were found in the subgroups of patients without prior

stroke and with low TTR.

Calibration of the different scores in the derivation and validation cohorts is displayed in

Figure 2. The calibration of the ABC-stroke score was superior compared with both the

CHA2DS2VASc and ATRIA scores.

Clinical utility

The event rates based on ABC-stroke risk categories within different risk categories based on the

were 0.63 for hs-cTnT and 0.62 for hs-cTnI. In the TTR<65% subgroup of warfarrininn tttrerer atatatededed

patients, the C indices were 0.71 C for hs-cTnT and 0.69 for hs-cTnI.

Comparison of the ABC-stroke score with CHA2DS2VASc and ATRIA scores

Theee AABA C-strokekeke ssscocc reee CCC iiindn exexex ooof f f 0.0.0 65 ((both hhhs-sCC cTTnnT aaandndnd hhhs-s-s-cTTnnnI))I) wwasas sigggnininififif cacacantnn ly hhhigigigheheh r r r thththan

he CCC index of ff 0.00 60 foor the CCCHAHHC 2DS2VAVASc strrrookee rriskskk ssscococ ree, pp=00.0004 (hhs--ccTc nT) ananand p=p=0.0222

hs-cTnI) (TTable 3)3 . ThThe C index of 0.61 for the AATRT IA stroke score was also significantly lowerC



ownloaded from



10

CHA2DS2VASc or ATRIA scores are illustrated in Figures 3 and 4. The ABC-stroke score

identified patients at higher or lower risk within each risk score category of both the

CHA2DS2VASc and ATRIA scores, including patients with ATRIA score less than 6,

CHA2DS2VASc score of 0-1 (where 1516 out of 1560 had CHA2DS2VASc=1), and patients with

only one CHA2DS2VASc point irrespective of sex. The decision curve analysis displayed

consistent positive and larger net benefit of using the ABC-stroke score for decision thresholds

between 1% and 5% one-year stroke/SE risk, as compared with both the CHA2DS2VASc and

ATRIA scores (Figure 5, Supplemental Table S3).

Discussion

The novel biomarker-based ABC-stroke score derived in one large cohort has now been

externally validated in another large cohort of AF patients at risk for stroke receiving oral

anticoagulants. The biomarker-based score was well calibrated, showed good discriminative

ability, and consistently performed better than the CHA2DS2VASc and ATRIA stroke scores.

The ABC-stroke risk score also showed significant net clinical benefit as compared with the

other stroke risk scores across a wide range of stroke risk in decision curve analysis. The new

ABC-stroke risk score should therefore be considered as an improved decision support tool in the

care of patients with AF.

The ABC – Age, Biomarkers, and Clinical history of prior stroke – risk score was

recently derived and internally validated in a large cohort of almost 15,000 patients with AF at

risk for stroke.20 In the development of this biomarker-based score we showed that previously

identified clinical risk factors, such as hypertension, diabetes mellitus, cardiovascular diseases

other than stroke, and female sex, no longer carried important incremental prognostic

Discussion

The novel biomarker-based ABC-stroke score derived in one large cohort has now been

externallyy validated in another large cohort of AF patients at risk for stroke receiving oral

antiiicococoagulantss. ThThThe ee biiiomommarara keeer-r-r-bababasesesed sccoro e waaas s wewell calalalibibibrararateteted,, ssshohohowewed d goododod dddisisiscrcrc iminininatatativivive e e

abilllittty,yy and connnsiss ssteenttlyy perfrfrfooro med bettteer thann n tthee CHCHHAAA222DSDSDS22VVAScSc andd ATATTRRIR A strororokeke sscores.

The ABBC-stroke risk score also shoh wed significant net clinical benefit as compared with the



ownloaded from



11

information when two biomarkers – NT-proBNP and hs-cTn – were included in the risk

prediction model. These two biomarkers are well-established markers of cardiovascular disease

and readily available in many laboratories throughout the world. The ABC-stroke score was well

calibrated; predicted event rates were similar to observed event rates in both the derivation

cohort and in a smaller external validation cohort of more than 1,400 patients, half of the latter

not receiving anticoagulation treatment.20 The previous results have been corroborated by the

present validation in a large independent cohort consisting of more than 8,000 anticoagulated

patients with AF at risk for stroke.

The well-calibrated ABC-stroke score was robust and provided similar results with two

different hs troponin assays. The ABC-stroke score containing only four variables seems

clinically useful for risk prediction of a broad population of AF patients including patients

without prior stroke and patients on sub-therapeutic warfarin treatment. The large number of

events and the relatively straight forward prediction model appear to prevent over-fitting, as

shown by the successful validation in the present cohort and in previously published cohorts of

patients with AF.20 A potential advantage of the ABC score is that three out of the four variables

in the ABC-stroke score are continuous which might improve individual risk prediction

compared with other clinical risk scores based solely on categorical and mainly irreversible risk

factors. The biomarkers used in the ABC score may also change over time, although the clinical

impact of such changes remains to be further elucidated.26, 27

The ABC-stroke score consistently predicted stroke/SE with higher accuracy than the

guideline recommended CHA2DS2VASc risk score,6, 7, 9 and the more recently developed

ATRIA stroke score8. The decision curve analysis, which depicts net benefits of a prediction

model at different thresholds, showed that if annual risks of stroke/SE in the range of 1 to 5 %

different hs troponin assays. The ABC-stroke score containing only four variabless seseseememems s s

clinically useful for risk prediction of a broad population of AF patients including patients

without prp ior stroke and patients on sub-therapeutic warfarin treatment. The large number of

evenenenttst and thee rrelee atatativelele y y y ststs rar igigighththt ffforororwaardrd predidid ctc ioonn momomodeded l l l apapappeeaara tto o prpreve ennntt t ovovovererer-fittiiingngng,, asass d

howowwn by the sssuucu cecessfuful valililiddad tion in ththe presenent cocohohohortrtrt aanddnd inn prreevioususlyyy pppublishedede ccoohorts of

patients with h AF.2202 A potential advantage of f the ABABC score is that three out of theh four variabblel s



ownloaded from



12

would be relevant thresholds to guide clinical decisions, the ABC-stroke score provides better

clinical usefulness compared to the CHA2DS2VASc and ATRIA scores. For instance – at a

threshold of 2% one-year stroke/SE risk the ABC-stroke score will identify 3.0 and 0.8

additional stroke/SE in a population with 13.8 stroke/SE per 1000 person-years compared with

the CHA2DS2VASc and ATRIA scores, respectively, without increasing the number of false

positives.

Clinical utility was further reinforced by the finding that ABC-stroke score improved risk

prediction within different risk classes of both the CHA2DS2VASc and ATRIA stroke scores.

The ABC-stroke score identified patients both at higher or lower risk across a broad range of

CHA2DS2VASc scores, including patients with CHA2DS2VASc score 1 or 2, which are the

thresholds recommended for treatment decisions in international guidelines.9, 10 Thus, the ABC-

stroke score might refine treatment decisions especially in patients identified as low risk based

on current risk scores, e.g. CHA2DS2VASc score equal to 1 or in patients with only one

CHA2DS2VASc point irrespective of sex (Supplemental Figure S4), where benefit of oral

anticoagulation remains debated.28-30 Patients at higher or lower risks were also identified in

patients with low, intermediate, or high stroke risk by the suggested decision thresholds of the

ATRIA scores 0-5, 6, or >6.8

Currently there are several different alternative treatments for stroke prevention in

patients with AF, which have different profiles concerning their effects on the risk of ischemic

stroke and the risk of intracranial and other major bleeding events. There is a need for decision

support models for identification of the treatment strategy with best balance between the

reduction in risk for stroke and the associated risk of bleeding for each patient.31, 32 Our recently

developed biomarker-based ABC-bleeding risk score33 provides a better, more reliable, and

CHA2DS2VASc scores, including patients with CHA2DS2VASc score 1 or 2, whiichchh aaarerere ttthehehe

hresholds recommended for treatment decisions in international guidelines.9, 10 Thus, the ABC-

troke score might refine treatment decisions especially in patients identified as low risk based

on cccuuurrent risk kk sscs ororores,, e.e..g.g.g. CCHAHAHA22DSDSDS2VAVASc scooorer eqqualalal ttto o o 111 oroo iinn n papap tit enentst wwwititith h h onononlylyly onenene

CHHHAAA2DS2VAScScSc ppointt iirrespepepective of sexex (Suppppp lemem ntntntalalal FFFiigurure S4S4), wwhherrer benefittt oof ooral

anticoagulation remains debated.2888-3030 Patients at hih gher or lower risks were alsl o iidentififiei d in



ownloaded from



13

useful tool for risk stratification than the HAS-BLED34 and ORBIT-AF35 bleeding risk scores

which are based only on clinical data and routine laboratory tests. Thus, the simultaneous use of

these two biomarker-based tools, indicating both the risk of stroke and the risk of major

bleeding, might provide a future opportunity for improved tailoring of stroke prevention

treatment in patients with AF. The balancing and discrimination of the risks for stroke and

bleeding should be more accurate with the biomarker-based ABC-stroke and ABC-bleeding risk

scores by inclusion of better targeted, continuous, and fewer overlapping risk factors as

compared with the currently recommended stroke and bleeding scores, e.g. CHA2DS2-VASc and

HAS-BLED.9, 10 The use of biomarkers as part of the ABC scores may also allow monitoring of

changes of risk indicators and alteration of the risk-benefit ratio over time.27 Therefore, the

current findings, in combination with the availability of the recently presented ABC-bleeding

score33, support a transition of risk stratification efforts towards the combination of biomarkers

and clinical information as a next step to personalized “precision” medicine for stroke prevention

in AF.

The clinical implementation of the algorithm used in the ABC-stroke score in daily

practice can either be based on a nomogram (Supplemental Figure S1), or preferably based on an

electronic tool integrated into electronic patient records or as an online tool, please visit

www.ucr.uu.se/en/services/abc-risk-calculators.

Strengths and limitations

Strengths of the development and validation of the ABC-stroke risk score include several large

independent clinical trial cohorts, standardized recording of clinical characteristics, long-term

follow-up, and centrally adjudicated clinical outcome events, while the exclusion of patient with

e.g. severe renal dysfunction or short life expectancy may be a limitation.21, 22, 36 Another

changes of risk indicators and alteration of the risk-benefit ratio over time.27 Therrefeffororore,e,e ttthehehe

current findings, in combination with the availability of the recently presented ABC-bleeding

core33, supppp ort a transition of risk stratification efforts towards the combination of biomarkers

and d d clclclinical infnffooro mamamationonon aaasss a nenenextxtxt ssstetet p toto personono alizized “prprp ecececisisisioon”n”n mmededicicineee fofofor r r stststrorr keee ppprerereveveentntn iond

n AAAFFF.

ThT e clinical implementation of theh algl orithmh used in theh AABC-stroke score in ddaily



ownloaded from



14

strength is that the ABC-stroke score was externally validated and reported according to the

principles and methods described by Royston and Altman with the contemporary consensus

statement on transparent reporting of multivariable prediction models for individual prognosis or

diagnosis (TRIPOD).15, 17

The derivation and the present validation of the ABC-stroke score in cohorts of patients

with AF at risk for stroke and on treatment with oral anticoagulation therapy is a limitation.

However, randomizing patients at increased stroke risk to no anticoagulation would not be

ethical, and better prediction of the risk-benefit balance is still important in anticoagulated

patients, especially those at low to intermediate risk for stroke, where there is uncertainty about

the benefit of oral anticoagulation, i.e. in patients with CHA2DS2-VASc = 1 in men and 1-2 in

women, respectively.26, 28-30

Furthermore, prior studies have shown that patients with TTR below 58-65% derive little

or no net benefit from warfarin.37, 38 Therefore, the present results are strengthened by the

consistent performance of the ABC-stroke score in the derivation and validation subgroups of

warfarin treated patients with TTR <65%, and also supported by the previous validation in a

smaller cohort out of which half of the patients with AF did not receive oral anticoagulation

therapy.20 Further validation of the biomarker-based risk score should be encouraged in broader

populations of AF patients with and if possible without oral anticoagulation, if such cohorts

without appropriate stroke prevention treatment still can be identified, and potentially in cohorts

of AF patients deemed not at risk for stroke by traditional clinical risk scores.

Conclusions

The recently developed biomarker-based ABC-stroke risk score was validated in a large cohort

he benefit of oral anticoagulation, i.e. in patients with CHA2DS2-VASc = 1 in meen n n ananand d d 1-11 222 ininin

women, respectively.26, 28-30

Furthermore, prior studies have shown that patients with TTR below 58-65% derive little

or nnno oo net benefififittt frfrfromomom wwwarararfaf ririin.n.n 3737,7, 383 Thehereforeee, thee presese enenent t t rereresusultltsss arara e e ststrer ngnggthththenennededed by y y ththheee

connssiisi tent perfofoformrmaancece of thhheee ABC-strroke scooro ee inn thehehe ddderere ivivatationn aand vavaliiiddad tion sububu grooups oof

warfarin treated patients with TTRR <65%%, and also supported bby thhe previous validatioi n in a



ownloaded from



15

of anticoagulated patients with atrial fibrillation and was shown to be well calibrated, have good

discriminative ability, and to consistently perform better and provide better utility as decision

support compared to the CHA2DS2VASc and ATRIA risk scores. The ABC-stroke risk score

should therefore be considered for implementation as an improved decision support tool in the

care of patients with AF.

Acknowledgments: Ebba Bergman, Ph.D. and Sanne Carlsson, B.A., B.Sc. at UCR, Sweden,

provided editorial assistance.

Funding Sources: This work was supported by a grant from The Swedish Foundation for

Strategic Research, Stockholm, Sweden. The RE-LY trial was funded by Boehringer Ingelheim,

Ingelheim, Germany. The ARISTOTLE trial was funded by Bristol-Myers Squibb, Co Princeton,

NJ and Pfizer Inc., New York, NY.

Conflict of Interest Disclosure:

J.O.: Consulting and lecture fees from Boehringer Ingelheim, Bayer, Bristol-Myers Squibb,

Pfizer. Z.H.: Institutional research grants from Boehringer Ingelheim, Bristol-Myers

Squibb/Pfizer; lecture fees from Boehringer Ingelheim; consulting fees from Bristol-Myers

Squibb/Pfizer. J.L.: Institutional research grants from Boehringer Ingelheim, Bristol-Myers

Squibb/Pfizer. J.H.A.: Institutional research grants and consulting fee/honoraria from Bristol-

Myers Squibb, Regado Biosciences, Merck; consulting fee/honoraria from Pfizer, AstraZeneca,

Boehringer Ingelheim, Ortho-McNeil-Janssen, Polymedix, Bayer. S.J.C.: Consulting fees,

speaker fees and research grants from Boehringer Ingelheim, Bristol-Myers Squibb, Bayer,

Portola; consulting fees and research grants from Sanofi-Aventis; research grants from Boston

Scientific. J.W.E.: Grants and honoraria from AstraZeneca, Bayer, Boehringer Ingelheim,

Bristol-Myers Squibb/Pfizer, Daiichi-Sankyo, GlaxoSmithKline, Janssen, Sanofi-Aventis;

honoraria from Eli Lilly. M.D.E.: Consulting fees from Boehringer Ingelheim, Pfizer, Sanofi,

Bristol-Myers Squibb, Portola, Bayer, Daiichi-Sankyo, Medtronics, Aegerion, Merck, Johnson &

Strategic Research, Stockholm, Sweden. The RE-LY trial was funded by Boehrinngegeer r r InInIngegegelhlhlheieieimm,m

ngelheim, Germany. The ARISTOTLE trial was funded by Bristol-Myers Squibb, CCoC PPriiinc teton

NJ and Pfizer Inc., New York, NY.

Confnfnflict of Intereesst DDisisisclclclosossururure:

J.O.:: CoCC nsultiiingngng aandn llecctureee fffees fromomom BBBoehrinininggerr IIngegegelheimm,m Bayyeer, BrBristototol-ll Myerrrs ss Sqquuibb,

Pfizer. Z.Z H.: Institutional research grants from BBoehrh inger Ingelheim, BBristoll-Myers



ownloaded from



16

Johnson, Gilead, Janssen Scientific Affairs, Pozen Inc., Amgen, Coherex, Armatheon. C.B.G.:

Grants and consultancy fees from AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb,

GlaxoSmithKline, Pfizer, Sanofi-Aventis, Takeda, The Medicines Company, Daiichi Sankyo,

Janssen, Bayer; grants from Medtronic Foundation, Armetheon; consultancy fees from Hoffman-

La Roche, Salix Pharmaceuticals, Gilead, Medtronic Inc. E.M.H.: Advisory board member and

symposium lecture fees from Bayer, Boehringer Ingelheim, Bristol-Myers Squibb; advisory

board member for Armetheon, Daiichi Sankyo, Janssen, Medtronic, Pfizer, Portola. R.D.L.:

Institutional research grant and consulting fees from Bristol-Myers Squibb; institutional research

grant from GlaxoSmithKline; consulting fees from Bayer, Boehringer Ingelheim, Pfizer, Merck,

Portola. A.S.: Institutional research grants from AstraZeneca, Boehringer Ingelheim, Bristol-

Myers Squibb/Pfizer, GlaxoSmithKline. S.Y.: Consulting fees, lecture fees and grant support

from Boehringer Ingelheim, AstraZeneca, Bristol-Myers Squibb, Sanofi-Aventis, Bayer, Cadila.

L.W.: Institutional research grants, consultancy fees, lecture fees, and travel support from

Bristol-Myers Squibb/Pfizer, AstraZeneca, GlaxoSmithKline, Boehringer Ingelheim;

institutional research grants from Merck & Co, Roche; consultancy fees from Abbott; holds two

patents involving GDF-15.

References

1. Wolf PA, Abbott RD and Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke. 1991;22:983-988. 2. Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV and Singer DE. Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA.2001;285:2370-2375. 3. Hart RG, Pearce LA and Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med. 2007;146:857-867.4. SPAF Investigators. Patients with nonvalvular atrial fibrillation at low risk of stroke during treatment with aspirin: Stroke Prevention in Atrial Fibrillation III Study. JAMA. 1998;279:1273-7.5. Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW and Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA. 2001;285:2864-70.6. Lip GY, Nieuwlaat R, Pisters R, Lane DA and Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-basedapproach: the euro heart survey on atrial fibrillation. Chest. 2010;137:263-72. 7. Lip GY, Frison L, Halperin JL and Lane DA. Identifying patients at high risk for stroke

L.W.: Institutional research grants, consultancy fees, lecture fees, and travel suppoportrtrt fffrororom m m

Bristol-Myers Squibb/Pfizer, AstraZeneca, GlaxoSmithKline, Boehringer Ingelheimim;

nstitutional research grants from Merck & Co, Roche; consultancy fees from Abbott; holds two

patentts s s ininvovovolvlvlving GDGG F-15.

Refefeferer nces

1. Wolf f PA, Abbott RRD D and Kannel WB. Atrial l fiibrb illation as an independent risk factor for



ownloaded from



17

despite anticoagulation: a comparison of contemporary stroke risk stratification schemes in an anticoagulated atrial fibrillation cohort. Stroke. 2010;41:2731-8. 8. Singer DE, Chang Y, Borowsky LH, Fang MC, Pomernacki NK, Udaltsova N, Reynolds K and Go AS. A new risk scheme to predict ischemic stroke and other thromboembolism in atrial fibrillation: the ATRIA study stroke risk score. J Am Heart Assoc. 2013;2:e000250. 9. Camm AJ, Lip GY, De Caterina R, Savelieva I, Atar D, Hohnloser SH, Hindricks G, Kirchhof P and Guidelines ESCCfP. 2012 focused update of the ESC Guidelines for the management of atrial fibrillation: an update of the 2010 ESC Guidelines for the management of atrial fibrillation. Developed with the special contribution of the European Heart Rhythm Association. Eur Heart J. 2012;33:2719-47. 10. January CT, Wann LS, Alpert JS, Calkins H, Cigarroa JE, Cleveland JC, Jr., Conti JB, Ellinor PT, Ezekowitz MD, Field ME, Murray KT, Sacco RL, Stevenson WG, Tchou PJ, Tracy CM, Yancy CW and Members AATF. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines and the Heart Rhythm Society. Circulation.2014;130:e199-267. 11. Hijazi Z, Oldgren J, Andersson U, Connolly SJ, Ezekowitz MD, Hohnloser SH, Reilly PA, Vinereanu D, Siegbahn A, Yusuf S and Wallentin L. Cardiac biomarkers are associated with an increased risk of stroke and death in patients with atrial fibrillation: a Randomized Evaluation of Long-term Anticoagulation Therapy (RE-LY) substudy. Circulation. 2012;125:1605-16. 12. Hijazi Z, Siegbahn A, Andersson U, Granger CB, Alexander JH, Atar D, Gersh BJ, Mohan P, Harjola VP, Horowitz J, Husted S, Hylek EM, Lopes RD, McMurray JJ, Wallentin L and Investigators A. High-sensitivity troponin I for risk assessment in patients with atrial fibrillation: insights from the Apixaban for Reduction in Stroke and other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial. Circulation. 2014;129:625-34.13. Hijazi Z, Wallentin L, Siegbahn A, Andersson U, Alexander JH, Atar D, Gersh BJ, Hanna M, Harjola VP, Horowitz JD, Husted S, Hylek EM, Lopes RD, McMurray JJ, Granger CB and Investigators A. High-sensitivity troponin T and risk stratification in patients with atrial fibrillation during treatment with apixaban or warfarin. J Am Coll Cardiol. 2014;63:52-61. 14. Hijazi Z, Wallentin L, Siegbahn A, Andersson U, Christersson C, Ezekowitz J, Gersh BJ, Hanna M, Hohnloser S, Horowitz J, Huber K, Hylek EM, Lopes RD, McMurray JJ and Granger CB. N-terminal pro-B-type natriuretic peptide for risk assessment in patients with atrial fibrillation: insights from the ARISTOTLE Trial (Apixaban for the Prevention of Stroke in Subjects With Atrial Fibrillation). J Am Coll Cardiol. 2013;61:2274-84. 15. Moons KG, Altman DG, Reitsma JB, Ioannidis JP, Macaskill P, Steyerberg EW, Vickers AJ, Ransohoff DF and Collins GS. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med.2015;162:W1-73. 16. Steyerberg EW. Clinical Prediction Models: A Practical Approach to Development, Validation, and Updating. Springer Verlag, New York. 2009. 17. Royston P and Altman DG. External validation of a Cox prognostic model: principles and methods. BMC Med Res Methodol. 2013;13:33.18. Steyerberg EW and Vergouwe Y. Towards better clinical prediction models: seven steps for development and an ABCD for validation. Eur Heart J. 2014;35:1925-1931.19. Granger CB, Alexander JH, McMurray JJ, Lopes RD, Hylek EM, Hanna M, Al-Khalidi HR, Ansell J, Atar D, Avezum A, Bahit MC, Diaz R, Easton JD, Ezekowitz JA, Flaker G, Garcia D,

Vinereanu D, Siegbahn A, Yusuf S and Wallentin L. Cardiac biomarkers are associatet d with an ncreased risk of stroke and death in patients with atrial fibrillation: a Randomizedd EvEvEvalalluauauatititiononon of

Long-term Anticoagulation Therapy (RE-LY) substudy. Circulation. 2012;125:16060605-5-5 16166..12. Hijazi Z, Siegbahn A, Andersson U, Granger CB, Alexander JH, Atar D, Gersh BJ, Mohan PHarjola VP, Horowitz J, Husted S, Hylek EM, Lopes RD, McMurray JJ, Wallentin L and nvestigators A. High-sensitivity troponin I for risk assessment in patients with atrial fibrillation:nsighghghtststs fffrorom mm thththe ApA ixaban for Reduction in Strokeke and other Thrromo booeembolic Events in Atrial

Fibrrrililillation (ARARARISSSTOTT TLTLTLE)E)E triririalalal.. CiCiCircrr ululata ion. 20202 144;12999:6:6:6252525---34343 .13. HHijazi Z, Walllenntinin LLL, SiSiieege bahnnn AAA,, AAnderssssonn UU, AAAlexandndnder JHH, AAtaar D,DD Gershshsh BJ,J, Hannnana M, HHHarjola VP,P,P HHHororowwiitz JDDD, , Husted SS, Hylekk k EMEM, LoLoLopepepess RDRRD, McMcMuurrrayyy JJ, Grangngngeer CB anndd nvestststigigigataa orrrss s A.AA HHiggh--ssensitttivivivititi y yy trtrtropopoponnnininin T aaanndnd risisk stststrarr tiififificacac titit onnn iiin n pattiennntststs wititthh h aata riiaal

fibrillatiion dduring treatment with apixaban or warfarin. J AmA CoC ll Cardiol. 20014;63:522-61.



ownloaded from



18

Geraldes M, Gersh BJ, Golitsyn S, Goto S, Hermosillo AG, Hohnloser SH, Horowitz J, Mohan P, Jansky P, Lewis BS, Lopez-Sendon JL, Pais P, Parkhomenko A, Verheugt FW, Zhu J, Wallentin L, Committees A and Investigators. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:981-992. 20. Hijazi Z, Lindback J, Alexander JH, Hanna M, Held C, Hylek EM, Lopes RD, Oldgren J, Siegbahn A, Stewart RA, White HD, Granger CB, Wallentin L, Aristotle and Investigators S. The ABC (age, biomarkers, clinical history) stroke risk score: a biomarker-based risk score for predicting stroke in atrial fibrillation. Eur Heart J. 2016;37:1582-1590. 21. Lopes RD, Alexander JH, Al-Khatib SM, Ansell J, Diaz R, Easton JD, Gersh BJ, Granger CB, Hanna M, Horowitz J, Hylek EM, McMurray JJ, Verheugt FW, Wallentin L and Investigators A. Apixaban for reduction in stroke and other ThromboemboLic events in atrial fibrillation (ARISTOTLE) trial: design and rationale. Am Heart J. 2010;159:331-339.22. Ezekowitz MD, Connolly S, Parekh A, Reilly PA, Varrone J, Wang S, Oldgren J, Themeles E, Wallentin L and Yusuf S. Rationale and design of RE-LY: randomized evaluation of long-term anticoagulant therapy, warfarin, compared with dabigatran. Am Heart J. 2009;157:805-810, 810 e1-2. 23. Harrel FE. Regression modeling strategies: with applications to linear models, logistic and ordinal regression, and survival analysis. Springer Verlag, New York. 2015. 24. Vickers AJ, Cronin AM, Elkin EB and Gonen M. Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers. BMC Med Inform Decis Mak. 2008;8:53.25. R Core Team. A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL: http://www.R-project.org/. 2016. 26. Eckman MH, Singer DE, Rosand J and Greenberg SM. Moving the tipping point: the decision to anticoagulate patients with atrial fibrillation. Circ Cardiovasc Qual Outcomes.2011;4:14-21.27. Hijazi Z, Oldgren J, Andersson U, Connolly SJ, Ezekowitz MD, Hohnloser SH, Reilly PA, Siegbahn A, Yusuf S and Wallentin L. Importance of persistent elevation of cardiac biomarkers in atrial fibrillation: a RE-LY substudy. Heart. 2014;100:1193-1200. 28. Friberg L, Skeppholm M and Terent A. Benefit of Anticoagulation Unlikely in Patients With Atrial Fibrillation and a CHA2DS2-VASc Score of 1. J Am Coll Cardiol. 2015;65:225-32. 29. Fauchier L, Lecoq C, Clementy N, Bernard A, Angoulvant D, Ivanes F, Babuty D and Lip GY. Oral anticoagulation and the risk of stroke or death in patients with atrial fibrillation and one additional stroke risk factor: the Loire Valley Atrial Fibrillation Project. Chest. 2016;149:960-968.30. Lip GY, Skjoth F, Nielsen PB and Larsen TB. Non-valvular atrial fibrillation patients with none or one additional risk factor of the CHA2DS2-VASc score. A comprehensive net clinical benefit analysis for warfarin, aspirin, or no therapy. Thromb Haemost. 2015;114:826-834. 31. Hijazi Z, Oldgren J, Siegbahn A, Granger CB and Wallentin L. Biomarkers in atrial fibrillation: a clinical review. Eur Heart J. 2013;34:1475-1480.32. Kirchhof P, Breithardt G, Aliot E, Al Khatib S, Apostolakis S, Auricchio A, Bailleul C, Bax J, Benninger G, Blomstrom-Lundqvist C, Boersma L, Boriani G, Brandes A, Brown H, Brueckmann M, Calkins H, Casadei B, Clemens A, Crijns H, Derwand R, Dobrev D, Ezekowitz M, Fetsch T, Gerth A, Gillis A, Gulizia M, Hack G, Haegeli L, Hatem S, Georg Hausler K, Heidbuchel H, Hernandez-Brichis J, Jais P, Kappenberger L, Kautzner J, Kim S, Kuck KH, Lane D, Leute A, Lewalter T, Meyer R, Mont L, Moses G, Mueller M, Munzel F, Nabauer M, Nielsen

ordinal regression, and survival analysis. Springer Verlag, New York. 2015. 24. Vickers AJ, Cronin AM, Elkin EB and Gonen M. Extensions to decision curveve aaanananalylylysisisis,s,s, aaa novel method for evaluating diagnostic tests, prediction models and molecular maarkrkrkerere s.s. BMBMBMC C CMed Inform Decis Mak. 2008;8:53.25. R Core Team. A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL: http://www.R-project.org/. 2016. 26. EcEcEckmkmkmananan MMMH,HH SSiinger DE, Rosand J and Greenbeberg SM. Movining the e tipping point: the decicicisisision to antitit ccoc agagagulllatateee papap tiienenentststs wwwiti h ata rial fibibi rillaation.n.n. CiCiCircrcrc CCararrdidid ovo asasc QuQuQualala OOOutuu cooomemeesss..20111111;4:14-21.27. HHHijazi Z, OlOlOldgdgd reen J, Andnddeere sson U, CConnollllyy SSJ, EEEzezezekokokowiiw ttz MMDD, HHoohnlnlnloso er SH,H,H, RReeilly PAA, Siegbababahnhnhn AAA,, YuYuYusuuf f S anand WaWaWallllllene tititin n n LL. IIImpmpmporororttat nncee offf pppere sisiistststennent eele eveevata ioion ofofof cccardididiacaca bbioomarkkerersn atrial l fibrili lation: a RER -LY substudy. HeH art. 2014;100:1193-3 12000 .



ownloaded from



19

JC, Oeff M, Oto A, Pieske B, Pisters R, Potpara T, Rasmussen L, Ravens U, Reiffel J, Richard-Lordereau I, Schafer H, Schotten U, Stegink W, Stein K, Steinbeck G, Szumowski L, Tavazzi L, Themistoclakis S, Thomitzek K, Van Gelder IC, von Stritzky B, Vincent A, Werring D, Willems S, Lip GY and Camm AJ. Personalized management of atrial fibrillation: Proceedings from the fourth Atrial Fibrillation competence NETwork/European Heart Rhythm Association consensus conference. Europace. 2013;15:1540-1556. 33. Hijazi Z, Oldgren J, Lindback J, Alexander JH, Connolly SJ, Eikelboom JW, Ezekowitz MD, Held C, Hylek EM, Lopes RD, Siegbahn A, Yusuf S, Granger CB, Wallentin L, Aristotle and RE-LY Investigators R. The novel biomarker-based ABC (age, biomarkers, clinical history)-bleeding risk score for patients with atrial fibrillation: a derivation and validation study. Lancet.2016;387:2302-2311.34. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ and Lip GY. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138:1093-1100. 35. O'Brien EC, Simon DN, Thomas LE, Hylek EM, Gersh BJ, Ansell JE, Kowey PR, Mahaffey KW, Chang P, Fonarow GC, Pencina MJ, Piccini JP and Peterson ED. The ORBIT bleeding score: a simple bedside score to assess bleeding risk in atrial fibrillation. Eur Heart J.2015;36:3258-3264. 36. White H, Held C, Stewart R, Watson D, Harrington R, Budaj A, Steg PG, Cannon CP, Krug-Gourley S, Wittes J, Trivedi T, Tarka E and Wallentin L. Study design and rationale for the clinical outcomes of the STABILITY Trial (STabilization of Atherosclerotic plaque By Initiation of darapLadIb TherapY) comparing darapladib versus placebo in patients with coronary heart disease. Am Heart J. 2010;160:655-661. 37. Hylek EM, Go AS, Chang Y, Jensvold NG, Henault LE, Selby JV and Singer DE. Effect of intensity of oral anticoagulation on stroke severity and mortality in atrial fibrillation. N Engl J Med. 2003;349:1019-1026.38. Connolly SJ, Pogue J, Eikelboom J, Flaker G, Commerford P, Franzosi MG, Healey JS, Yusuf S and Investigators AW. Benefit of oral anticoagulant over antiplatelet therapy in atrial fibrillation depends on the quality of international normalized ratio control achieved by centers and countries as measured by time in therapeutic range. Circulation. 2008;118:2029-37.

2015;36:3258 3264. 36. White H, Held C, Stewart R, Watson D, Harrington R, Budaj A, Steg PG, Cannnnnononon CCCP,P,P, KKKrururug-Gourley S, Wittes J, Trivedi T, Tarka E and Wallentin L. Study design and rationalalale e e fofor r r thththeeeclinical outcomes of the STABILITY Trial (STabilization of Atherosclerotic plaque By Initiationof darapLadIb TherapY) comparing darapladib versus placebo in patients with coronary heart disease. Am Heart HH J. 2010;160:655-661. JJ37. HyHyHylelelekkk EMEMEM, Goo AAS, Chang Y, Jensvold NG, HHennault LE, Selbyby JV aand Singer DE. Effect of nteeensnsnsity of oraaalll annntitt coooagagagululu ata ioioion n n ononon sstrokoke seveeerir tyy aanddd mmmororortatatalil tyy in nn ata riiala fibbbriririlllll atatatioioion. N N N EnEnEnglglgl JJJ

MedededMM . 2003;349:101019-10002226.38. CCConnolly SSSJ,JJ PPoogueue J, EiEiEikkek lboom JJ, Flakerr r G, CComommmemem ffrforord PP, Franznzosssiii MG, HeHeH alleyy JS, Yusuuuf f f SS S and d d InInInvesstiigaatoors AWAWAW.. Beeenenenefitt t ofofo orararal ll aantiticocooagagagulananantt t ovovverrr aaantn ippllateeelelelet tt thherereraapa yy in atriaalfibrillatiion ddependsd on the quality of internationall normalized ratio control achieved bby centers



ownloaded from



20

Table 1. Baseline demographics and biomarker levels

Variable n=8,356Age (years, min – max) 72 (22 – 95)Gender: Female (%) 36.7 (3069)Current smoker (%) 7.4 (660)Diabetes mellitus (%) 22.4 (1868)Hypertension (%) 79.2 (6617)Congestive Heart Failure (%) 28.8 (2407) [1]*

Permanent or persistent AF† (%) 67.2 (5610) [4]*

Prior stroke or TIA‡ (%) 19.4 (1619)Vascular Disease (%) 19.1 (1597)Renal function, CrCl§ (mL/min) 68.2 (53.6 – 86.2) [68]*

Troponin T (ng/L) 12.2 (7.7 – 19.5)Troponin I (ng/L) 6.8 (4.2 – 13.0)NT-proBNP|| (ng/L) 807 (382 – 1447)

For biomarkers the numbers represent median (interquartile range) *Numbers within square brackets indicate number of missing values, if any†AF, atrial fibrillation‡TIA, transient ischemic attack §CrCl, creatinine clearance ||NT-proBNP, N-terminal pro-B-type natriuretic peptide

Table 2. Incidence rates and hazard ratios in ABC-stroke risk classes

Risk class N Events Incidence rate* Hazard ratioABC-stroke score including hs troponin T

Low risk (<1 %) 3287 49 0.76 [0.56, 1.00] 1.00 (ref)Medium risk (1–2 %) 3762 107 1.48 [1.21, 1.78] 1.95 [1.39, 2.74]High risk (>2 %) 1307 63 2.60 [2.00, 3.33] 3.44 [2.37, 4.99]

ABC-stroke score including hs troponin ILow risk (<1 %) 3079 45 0.74 [0.54, 1.00] 1.00 (ref)Medium risk (1–2 %) 3854 105 1.41 [1.15, 1.71] 1.90 [1.34, 2.69]High risk (>2 %) 1423 69 2.61 [2.03, 3.30] 3.52 [2.42, 5.12]

*Per 100 person-years

Numbers within square brackets indicate number of missing values, if anyAF, atrial fibrillationTIA, transient ischemic attack CrCl, creatinine clearance NT-proBNP, N-terminal pro-B-type natriuretic peptide

Tablblb e 2. Incidencee rateteess andd d hhah zarddd rrattioos in AABCBC--stroroke risisiskk k claasses

Riskkk ccclalalassssss NNN EvEvE enntts InInIncicic dedd ncncncee raatte* HaHaHazaarrd ratioABC-stroke score including hs troponin T



ownloaded from



21

Table 3. C-indices [95% confidence interval] for the ABC-stroke, CHA2DS2VASc, and

ATRIA stroke risk scores

Full cohort No prior stroke TTR§ <65%Events/Number of subjects 219/8356 150/6737 30/1109

ABC-stroke* (troponin T) 0.65 [0.61, 0.69] 0.63 [0.58, 0.67] 0.71 [0.62, 0.81]ABC-stroke* (troponin I) 0.65 [0.61, 0.69] 0.62 [0.58, 0.67] 0.69 [0.60, 0.79]CHA2DS2VASc† 0.60 [0.57, 0.64] 0.57 [0.53, 0.62] 0.60 [0.48, 0.72]ATRIA‡ 0.61 [0.58, 0.65] 0.58 [0.54, 0.62] 0.64 [0.52, 0.76]

*ABC-stroke: Age, Biomarkers (cardiac troponin and NT-proBNP), Clinical history (prior stroke/ transient ischemic attack)†CHA2DS2VASc: assigns 1 point each for congestive heart failure, hypertension, diabetes mellitus, vascular disease, age 65-74 years, and sex category [female gender]; and 2 points for a stroke/transient ischemic attack ‡ATRIA: assigns 6 (9 for patients with prior stroke) points for for age 75-84; 3 (7) points for age 65-74; and 0 (8) points for age <65, and 1 point each for female sex, diabetes mellitus, chronic heart failure, hypertension, proteinuria, and estimated glomerular filtration rate <45 ml/min or end-stage renal disease§TTR: Time in therapeutic range (INR 2.0-3.0)

Figure Legends

Figure 1. Cumulative event rates of stroke/systemic embolism stratified by predicted one-year

ABC-stroke risk category. Cumulative event rates of stroke/systemic embolism up to 2.5 years

stratified by predicted one-year (troponin T-based) ABC-stroke risk category (green = 0-1%, blue

= 1-2%, and red >2%) in the validation (solid lines) and the derivation20 (dashed lines) data. The

colored vertical bar indicates the one-year time mark.

Figure 2. Calibration plots of CHA2DS2VASc, ATRIA, and ABC scores, in the derivation20 (red)

and the validation (blue) cohorts. For each discrete point score the figure displays the observed

stroke/systemic embolism event rate per 100 person-years with 95% confidence interval (y axis)

versus the event rate (x axis) previously published from the original derivation cohorts of the

CHA2DS2VASc score, in AF patients without6 or with7 oral anticoagulant treatment, or the

Figure Legends

Figugugure 1. Cumulatitivve evevevent t rararates ofofof ssstrrooke/sysstememicc eembmm olisssmmm straratifieded bybyy ppprediictctctedee onnen -yeararar

ABCCC-stststroke rrrisisiskkk cattegogory. CuCuCumumm laaatititivvev eeevevv nt rrratates oof sttstror kee/s/s/sysystemimm c emembooolililisms uuup p p ttot 2.5 yearsrs

t tifi d b di t d (t i T b d) ABC t k i k t ( 0 1% bl



ownloaded from



22

ATRIA8 score, see supplemental Table S1. For the ABC-stroke score the figure displays the

observed one-year event rate versus the predicted one-year risk of stroke/systemic embolism

using the originally derived model.

Figure 3. Cumulative event rates of stroke/systemic embolism stratified by ABC-stroke risk and

by CHA2DS2VASc risk categories. Cumulative event rates of stroke/systemic embolism stratified

by predicted one-year (troponin T-based) ABC-stroke risk category (colored lines) and by

CHA2DS2VASc risk categories (panels) in the combined derivation and validation cohorts,

n=23,057.


by ATRIA risk categories. Cumulative event rates of stroke/systemic embolism stratified by

predicted one-year (troponin T-based) ABC-stroke risk category (colored lines) and by ATRIA

risk categories (panels) in the combined derivation and validation cohorts, n=23,057.

Figure 5. Decision curve analysis for the ABC-stroke, CHA2DS2VASc, and ATRIA risk

prediction models. The net benefit of using the prediction models to guide clinical decision in

relation to assuming that no one is at risk (all negative) or that all are at risk (all positive) for

stroke/systemic embolism on the y axis, is plotted against relevant decision thresholds on the x

axis, n=23,057.


by ATRIA risk categories. Cumulative event rates of stroke/systemic embolism stratified by

predddicicicted one-yeyeyearrr (((tropopoponononinini TTT-b-b-basasasedee ) ABA C-stttrorokee risk k k cacacatetetegogogoryy (c(c(cololorrede linnneseses) ) ) ananand byyy ATATATRIRIRIAA

iskkk cccategories (p(p(pananelss) in thehehe combinedd derivaationn andndnd vvvalaaliddidatationn ccohoorrts,,, nnn=23,055577.



ownloaded from


Time (years)

Cum

ulat

ive

even

t rat

e (%

)

0

1

2

3

4

5

6

7

0.0 0.5 1.0 1.5 2.0 2.5

DerivationValidation

<1%/year<1%/year

1-2%/year1-2%/year

>2%/yea>2%/yea



ownloaded from




ownloaded from


Agneta Siegbahn, Salim Yusuf and Lars WallentinEikelboom, Michael D. Ezekowitz, Christopher B. Granger, Elaine M. Hylek, Renato D. Lopes,

Jonas Oldgren, Ziad Hijazi, Johan Lindbäck, John H. Alexander, Stuart J. Connolly, John W.Fibrillation

Performance and Validation of a Novel Biomarker-Based Stroke Risk Score for Atrial

Print ISSN: 0009-7322. Online ISSN: 1524-4539 Copyright © 2016 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation published online August 28, 2016;Circulation.

http://circ.ahajournals.org/content/early/2016/08/24/CIRCULATIONAHA.116.022802World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://circ.ahajournals.org/content/suppl/2017/07/10/CIRCULATIONAHA.116.022802.DC2 http://circ.ahajournals.org/content/suppl/2016/08/24/CIRCULATIONAHA.116.022802.DC1

Data Supplement (unedited) at:

http://circ.ahajournals.org//subscriptions/

is online at: Circulation Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer available in the

Permissions in the middle column of the Web page under Services. Further information about this process isOnce the online version of the published article for which permission is being requested is located, click Request

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Circulation Requests for permissions to reproduce figures, tables, or portions of articles originally published inPermissions:



ownloaded from

http://circ.ahajournals.org/content/early/2016/08/24/CIRCULATIONAHA.116.022802

http://circ.ahajournals.org/content/suppl/2016/08/24/CIRCULATIONAHA.116.022802.DC1

http://circ.ahajournals.org/content/suppl/2017/07/10/CIRCULATIONAHA.116.022802.DC2

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://circ.ahajournals.org//subscriptions/


SUPPLEMENTAL MATERIAL

Performance and Validation of a Novel Biomarker-Based Stroke Risk

Score for Atrial Fibrillation

Jonas Oldgren, M.D., Ph.D.1,2; Ziad Hijazi, M.D., Ph.D.1,2; Johan Lindbäck, M.Sc.2; John H.

Alexander, M.D., M.H.S.3; Stuart J. Connolly, M.D.4; John W. Eikelboom, M.B., B.S.4; Michael D.

Ezekowitz, M.B., Ch.B., Ph.D.5; Christopher B. Granger, M.D.3; Elaine M. Hylek, M.D., M.P.H.6;

Renato D. Lopes, M.D., Ph.D.3; Agneta Siegbahn, M.D., Ph.D.2,7; Salim Yusuf, M.D., Ph.D.4; and

Lars Wallentin, M.D., Ph.D.1,2; on behalf of the RE-LY and ARISTOTLE Investigators

1Department of Medical Sciences, Cardiology, Uppsala University, Uppsala, Sweden; 2Uppsala

Clinical Research Center, Uppsala University, Uppsala, Sweden; 3Duke Clinical Research Institute,

Duke Medicine, Durham, NC, USA; 4Population Health Research Institute, Hamilton, Canada; 5Thomas Jefferson Medical College and the Heart Center, Wynnewood, PA, USA; 6Boston

University Medical Center, Boston, MA, USA; 7Department of Medical Sciences, Clinical

Chemistry, Uppsala University, Uppsala, Sweden

Address for correspondence

Jonas Oldgren, M.D., Ph.D.

Uppsala Clinical Research Center

Uppsala Science Park

Dag Hammarskjölds väg 14 B

SE-752 37 Uppsala, Sweden

Phone: +46 (0) 18 611 2765

Fax: +46 (0) 18 51 55 70

E-mail: [email protected]

Content

Table S1. CHA2DS2VASc and ATRIA incidence rates from 2

the original publications

Table S2. The ABC stroke model 3

Table S3. Net benefit of using the ABC, ATRIA, and 4

CHA2DS2VASc scores for identifying stroke/systemic embolism

conditional on different decision thresholds

Figure S1. ABC-stroke score nomogram (with hs troponin T) 5

Figure S2. Cumulative rates of stroke/systemic embolism by 6

predicted one-year (troponin I-based) ABC-stroke risk group

Figure S3. Predicted one-year risk of stroke/systemic embolism 7

using different Troponin methods in the validation cohort

Figure S4. Cumulative rates of stroke/systemic embolism by 8 ABC-stroke risk in patients with one CHA2DS2VASc risk point

1

J Oldgren et al. Validation of the ABC-stroke score in AF

2

Table S1. CHA2DS2VASc and ATRIA incidence rates from the original publications

CHA2DS2VASc CHA2DS2VASc (OAC) ATRIA

Points Person-

years Events Incidence

rate* Person-


rate* Person-


rate*

0 103 0 0.00 2 0 0.00 2652 2 0.08

1 162 1 0.62 653 3 0.46 2819 12 0.43

2 184 3 1.63 1913 15 0.78 1419 14 0.99

3 203 8 3.94 2673 31 1.16 1780 13 0.73

4 208 4 1.92 2665 38 1.43 2960 19 0.64

5 95 3 3.16 1732 42 2.42 3614 36 1.00

6 57 2 3.51 1016 36 3.54 4346 83 1.91

7 25 2 8.00 436 15 3.44 4768 119 2.50

8 9 1 11.11 125 3 2.40 3913 151 3.86

9 1 1 100.00 18 1 5.56 2400 104 4.33

10 1181 75 6.35

11 501 31 6.19

12 183 20 10.93

13 53 4 7.55

14 12 2 16.67

15 7 0 0.00

* Per 100 person-year

Based on

Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Chest. 2010;137(2):263-272;

Lip GY, Frison L, Halperin JL, Lane DA. Stroke. 2010;41(12):2731-2738; and

Singer DE, Chang Y, Borowsky LH, et al. A new risk scheme to predict ischemic stroke and other thromboembolism

in atrial fibrillation: the ATRIA study stroke risk score. J Am Heart Assoc. 2013;2(3):e000250


3

Table S2. The ABC stroke model

The continuous variables are truncated at the 1st and 99th sample percentile in the derivation

cohort. That is, values below the min should be set to the min and values above the max should

be set to the max before calculation. Prior stroke/TIA is either 0 (if subject had no prior

stroke/TIA) or 1 (if subject had a prior stroke/TIA) and “ln(.)” denote the natural logarithm.

Coefficient Min Max

Intercept -3.2864

Prior stroke/TIA 0.8331 0 1

Age [years] 0.0075 44 90

ln(hs-troponin T) [ng/L] 0.2139 ln(3.3) ln(66)

ln(NT-proBNP) [ng/L] 0.2879 ln(25) ln(5900)

The one-year risk of stroke/systemic embolism (SE) is calculated using the following equations:

LP =

-3.2864 +

0.8331 * Prior stroke/TIA +

0.0075 * Age +

0.2139 * ln(hs-cTnT) +

0.2879 * ln(NT-proBNP)

One-year risk of stroke/SE = 1 – 0.9863^exp{LP}


4

Table S3. Net benefit of using the ABC, ATRIA, and CHA2DS2VASc scores for identifying

stroke/systemic embolism conditional on different decision thresholds

Net benefit vs all negative Net benefit vs all positive

Decision

Threshold ABC CHA2DS2VASc ATRIA ABC CHA2DS2VASc ATRIA

(%) No OAC OAC No OAC OAC

1 6.3 5.2 5.8 6.0 1.2 0.1 0.8 0.9

2 2.3 -0.7 1.5 1.5 7.4 4.4 6.6 6.6

3 1.1 -5.9 -0.1 -1.0 16.6 9.5 15.4 14.4

4 0.6 -0.2 0.0 -1.5 26.6 25.9 26.0 24.5

5 0.2 -0.5 0.0 -1.2 37.1 36.3 36.8 35.6

6 0.1 -0.8 0.0 -2.0 48.0 47.0 47.9 45.8

The net benefit was estimated using a decision curve analysis (Vickers AJ, et al. BMC Med Inform Decis Mak.

2008;8:53) on the 23,057 subjects with an event rate of 13.8 stroke/systemic embolism (SE) per 1000 person-years in

the combined data from the ARISTOTLE and RE-LY studies. The three scores are compared with the assumption that

no one is at risk for stroke/SE (all negative) and the assumption that all are at risk for stroke/SE (all positive).

The numbers indicate the number of additional true positives per 1000 person-years the models can identify without

additional false positives.

Event rate per 100 person-years for CHA2DS2VASc and ATRIA are based on the original publications by

Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Chest. 2010;137:263-72; Lip GY, Frison L, Halperin JL, Lane DA. Stroke. 2010;41:2731-8; and

Singer DE, Chang Y, Borowsky LH, et al. Journal of the American Heart Association. 2013;2:e000250


5

Figure S1. ABC-stroke score nomogram (with hs troponin T)

Supplement to Hijazi Z, et al. The ABC (age, biomarkers, clinical history) stroke risk score: a biomarker-based risk

score for predicting stroke in atrial fibrillation. European Heart Journal. 2016;37:1582-90.

6

Figure S2. Cumulative rates of stroke/systemic embolism by predicted one-year (troponin I-

based) ABC-stroke risk group

Cumulative event rates of stroke/systemic embolism up to 2.5 years stratified by predicted one-

year (troponin I-based) ABC-stroke risk category (green = 0-1%, blue = 1-2%, and red >2%) in

the validation (solid lines) and the derivation* (dashed lines) data. The colored vertical bar

indicates the one-year time mark. *Hijazi Z, Lindback J, Alexander JH, et al. European heart journal. 2016;37:1582-90


7

Figure S3. Predicted one-year risk of stroke/systemic embolism using different Troponin

methods in the validation cohort

Dashed lines indicate the pre-defined risk classes 0-1%, 1-2%, and >2%.


8

Figure S4. Cumulative rates of stroke/systemic embolism by ABC-stroke risk in patients

with one CHA2DS2VASc risk point

Cumulative rates of stroke/systemic embolism by ABC-stroke risk (colored lines) in patients with one CHA2DS2VASc risk point irrespective of

sex, in the combined derivation and validation cohorts.

Need Help? mailto:[email protected] Get this transcript in a non‐tabular format

Carolyn Lam: Welcome to Circulation on the Run, your weekly podcast summary and backstage

pass to the journal and its editors. I'm Dr. Carolyn Lam, associate editor from the National Heart Center and Duke National University of Singapore. Our feature discussion today is about the validation of a novel biomarker‐based stroke risk score for atrial fibrillation, the ABC stroke score. But first, here's your summary of this week's journal.

The first paper provides experimental insights into endothelial nitric oxide synthase uncoupling in endothelial dysfunction. In this paper by first author Dr. Lee, corresponding author Dr. Wong and colleagues from Qilu Hospital of Shandong University in China, authors assessed endothelial function in animal models of hyperglycemia, hyperhomocysteinemia, and a dyslipidemia. They demonstrated that GTP cyclohydrolase 1 is the target of the microRNA‐133a and that it's a topic expression and endothelial cells mediates endothelial dysfunction.

Furthermore, Lovastatin up‐regulated GTP cyclohydrolase 1 and tetrahydrobiopterin and re‐coupled endothelial nitric oxide synthase in stress endothelial cells. These actions of Lovastatin were abolished by enforced micro RNA 133A expression and mirrored by a mir‐133a‐antagomir. Finally, the beneficial effect of Lovastatin in mice were abrogated by in vivo mir‐133A over‐expression or by GTP cyclohydrolase 1 knockdown. In summary, this paper offers a mechanistic basis for targeting micro RNA 133A as a therapeutic approach to correct endothelial nitric oxide synthase dysfunction. It also provides further support to the role of statins in combating endothelial dysfunction.

The next study shows us that in hypertrophic cardiomyopathy, calcium mishandling may be the potential link between the primary genetic cause and downstream signaling cascade that leads to hypertrophy and arrythmias. In this study, Dr. Helms and colleagues from University of Michigan analyzed gene expression, protein levels and functional essays for calcium regulatory pathways in 35 human hypertrophic cardiomyopathy surgical samples with and without sarcomere mutations and compared that with 8 control hearts. They found a marked reduction in circa2 abundance, which correlated with reduced circa2 function in hypertrophic cardiomyopathy compared to controlled hearts regardless of the underlying genetic etiology.

However, calcium calmodulin depend protein kinase type 2 or cam2, which is a calcium sensing kinase, was deferentially activated only in the sarcomere gene mutation positive samples. Activation of chem kinase 2 was associated with an increase in phospholamb and phosphorylation in hypertrophic cardiomyopathy. However, neither calcineurin MRNA nor MEF2 activity was increased, suggesting that calcineurin pathway activation was not an upstream cause of increased chem kinase 2 protein abundance or activation.

In summary, this paper demonstrated that calcium mishandling occurs through both genotype specific and common pathways in human hypertrophic

COTR134_22 Page 2 of 6

cardiomyopathy. Post‐translational activation of chem kinase 2 pathway is specific to sarcomere mutation positive hypertrophic cardiomyopathy. While Sarco 2 abundance and sarcoplasmic reticulum calcium uptake are depressed in both sarcomere positive and negative hypertrophic cardiomyopathy. Thus, chem kinase pathway inhibition may improve aberrant calcium cycling in hypertrophic cardiomyopathy. This is discussed further in an accompanying editorial by Dr. Jill Tardiff.

The third study suggests that in patients with a dilated aortic route and trileaflet aortic valve, a ratio of aortic route area to height provides independent and improved stratification for prediction of death. First author Dr. Masry, corresponding author Dr. Desai and colleagues from the Center for Aortic Disease, Heart and Vascular Institute of Cleveland Clinic, studied consecutive patients with a dilated aortic route of greater or equal to 4 centimeters who underwent echocardiography and gated contrast enhanced thoracic aortic computer tomography or magnetic resonance and geography between 2003 and 2007.

A ratio of aortic route area over height was calculated on tomography and a cutoff of 10 squared centimeters per meter of height was chosen as abnormal. In 771 patients with trileaflet aortic valve and concomitant aortopathy, there was incremental prognostic value for indexing aortic route or ascending aortic area to patient height rather than using an unindexed aortic diameter. Incorporation of the ratio significantly and independently reclassified the risk for death and at normal ratio was independently associated with higher long‐term mortality while cardiovascular surgery was associated with improved survival. Importantly, a sizable minority of patients with aortic route diameters between 4.5 and 5.5 centimeters had an abnormal aortic route when indexed to height ratio. 78 percent of deaths in this subgroup occurred in those with an abnormal aortic route area to height ratio. Findings were similar when ascending aortic measurements were considered. The take home message is that an aortic route area to height ratio above 10 squared centimeters per meter of height has significant and independent prognostic utility and may be used to re‐stratify patients with trileaflet aortic valve and a dilated aorta.

The final study provides pre‐clinical data to show that Ticagrelor reduces cardio damage post myocardial infarction to a greater extent than Clopidogrel by an adenosine induced organ protective response. First author Dr. Villaher, corresponding author Dr. Bademan and colleagues from the Cardiovascular Research Center in Barcelona, Spain studied a close‐chest swine model of ischemia reperfusion in which myocardial infarction was induced by 1 hour balloon occlusion of the mid‐left anterior descending coronary artery followed by 24 hours of re‐flow. Prior to occlusion, the animals were randomly assigned to receive either placebo, a loading does of Clopidogrel, a loading does of Ticagrelor or a loading does of Ticagrelor followed by an A1 A2 receptor antagonist. Edema infarc size left ventricular size and left ventricular function were assessed by three T cardiomagnetic resonance imaging. Inhibition of platelet aggregation was the same between the groups receiving a P2Y‐12 inhibitor.


Yet, Ticagrelor reduced infarc size to a significantly greater extent than Clopidogrel,

reducing it by a further 23.5 percent, an effect supported by troponin eye assessment and histopathological analysis. Furthermore, compared to Clopidogrel, Ticagrelor significantly diminished myocardial edema by 24.5 percent, which correlated with infarced mass. Administration of an adenosine A1 A2 antagonist abolished the cardio protective effects of Ticagrelor over Clopidogrel. At a molecular level, aquaporin 4 expression decreased and the expression and activation of AMP kinase cyclin and COX‐2 increased in the ischemic myocardium of Ticagrelor versus Clopidogrel treated animals. In summary, this study shows that Ticagrelor exerts cardio protective effects beyond its anti‐platelet efficacy by adenosine dependent mechanisms, which reduce necrotic injury and edema formation. This is discussed in an accompanying editorial by Drs. Gerbel, Jung and Tantry. That wraps it up for the summaries. Now for our feature discussion.

Today, we are going to be discussing the performance of the ABC score for stroke in atrial fibrillation. And as a reminder for all our listeners there, ABC stands for A for age, B for biomarkers, that's NT‐proBNP and high‐sensitivity troponin, and C for clinical history of prior stroke. And again as a reminder, this risk score was originally derived in the Aristotle trial. However, we have new results about its performance and validation today from first and corresponding author Dr. Jonas Oldgren from Uppsala Clinical Research Center in Sweden. Welcome, Jonas.

Speaker 2: Thank you very much.

Carolyn Lam: We also have today the associate editor who managed this paper, Dr. Sandeep Das from UT Southwestern. Hi Sandeep.

Speaker 3: Hi Carolyn, thanks for having me.

Carolyn Lam: So Jonas, could you start off by telling us why you did this study and what you found?

Speaker 2: We did this study to validate the recently derived ABC stroke risk score. We have had risk scores for predicting stroke in patients with atrial fibrillation derived since the late 1990's and refined later on. But those risk scores have only used clinical markers for risk. We have for several years developed new risk prediction models with biomarkers and now we are combine them in a very simple biomarker based risk score, taking into account age as a clinical variable and the clinical history of prior stroke and only two common used biomarkers. And by that we can predict the risk of stroke with better precision than previous clinical risk scores.

Carolyn Lam: Yeah, I like what you said. I mean it is literally as simple as ABC. So tell us how you validated it and what you found.

Speaker 2: It was derived in a large cohort of patients participating in a clinical trial with new or relapsed coagulant compared to Warfarin and we now validated in almost a full


size group participating it another clinical trial. So we have large data sets of very well described patients where we have good outcome data. Very solid data to rely on. Now we can see that the ABC risk score is now validated but the good precision and good collaboration of the discriminatory abilities is high and better than the previously used clinical risk scores.

Carolyn Lam: Could you give us some numbers behind that that are clinically meaningful? Everyone's going to be wondering compared to the chads‐vasc score for example, how does this ABC score perform in that validation test set?

Speaker 2: We can adjust that by several different aspects. One is of course to calculate the C index which is a statistical method to see how good we can predict risk and the C indices for the ABC stroke score both in the duration and now in the validation cohort is higher than for the chads‐vasc and atrial risk scores. But we can also look at what we have in this paper in circulation ... we can look at predicted outcome rates and observed outcome rates and can see that they clearly overlap both in the duration and validation court. So if you predict a risk that is less than 1 percent per year, it is observed also a risk that is less than 1 percent a year. Does this always ... the thing is when you derive risk or but when you validate it in another cohort, you need to show that it's a similar result.

Carolyn Lam: Yeah, that's true. Sandeep, you are managing this paper. It's very important. How do you think that clinicians should be taking the results?

Speaker 3: I think that clearly using anticoagulation and selected patients at high risk for stroke with atrial fibrillation is one of the best things we do in cardiology. You know in terms of reducing the risk of an important harm to patients. I think there's a fair bit of dissatisfaction out there with currently sort of standard, which is chads‐vasc. Especially in people with a chads‐vasc ... men with a chads‐vasc of 1 or women with a chads‐vasc of 1 to 2 where there's a bit of struggling over how to decide. So I think that one real advantage of this score in addition to the fact that it predicts better by the higher C statistic, which is fantastic and pretty uncommon, right? Lars sort of buried the lead a little bit by not emphasizing that it's relatively rare that we're able to move a c statistic by a point of 5 in the modern era.

But the other thing is that it helps give us an ability to come up with good estimates in people at low risk, which I think has been a challenge and something that people are a little concerned clinically. So I think that this is easily available, biomarkers that we routinely check all the time and it doesn't have the sort of gender challenge with chads‐vasc where you're trying to figure out whether your low risk woman really needs to be on Warfarin or anticoagulation. So I think that it has a lot of clinical utility right out of the box, which is nice.

Carolyn Lam: Actually, Jonas could you let us know is there any sex differences in the performance of the score?

[00:14:46]


Speaker 2: There are no differences in the performance of the score. So we looked ... the advantage of this score is when we derived it in the original model, we looked at all important clinical and biomarker risk factors and we can see that these were the foremost interesting markers. So we only used those. So we can predict much better and as pointed out so nicely by Sandeep, for patients at the lower end of the risk spectrum, we can find patients or have higher or lower risk even within patients with chads‐vasc 1 or chads‐vasc 2. And I think it's also important to see what about patients at higher risk despite proper anticoagulation. We did not know how to treat them but in the future we might perhaps tailor treatment also for those patients with residual high risk of stroke despite proper anticoagulation treatment. For instance, if the left atrial appendage occluded devices are shown in the future to be a good option for those patients, we can find them also by this risk score. So both in the higher and lower end of the risk spectrum.

Carolyn Lam: That's a really good point. On that note, I'm just curious. What do you think is in the future? What more knowledge do we need to address before we put this into practice or are you already using this? Or do you think it should enter guidelines for example? Maybe Sandeep, I could ask for your opinion first.

Speaker 3: We see a lot of biomarkers associated with increased risk kind of studies come out in the literature. You know probably every week you see several of these things come out. So what's really interesting about this is that it's obviously methodologically extremely well done but its been derived and validated in two large cohorts, which is pretty much best practice right? You want to see people validate these risk scores in large and distinct cohorts of patients to build up sort of clinical validity to the reader or consumer. So I think, from my standpoint, this is ready for prime time. I'm really intrigued by the fact that biomarkers, especially troponin, are predicting stroke in this population and there have been some observational reports out there that have showed an association between troponin and increased risk of stroke or worse outcomes after stroke. So I'd be really curious as to what Jonas thinks about why troponin would be predictive of stroke in this population.

Speaker 2: We were extremely intrigued by the finding when we first did those single observations of only troponin as a risk marker because we know that troponin is a very specific protein found in the myocardium. But the clinic predicts risk for stroke also and there are several explanations but they are mainly hypotheses about aging and myocardial function really to identify patients of risk. But the clear cut explanation is still not there.

Carolyn Lam: It's likely that these biomarkers are incorporating aspects that we don't fully understand, which is why they are better predictors isn't it? I mean to your point Sandeep.

Speaker 3: Yeah, no absolutely. And I think that's great.

Carolyn Lam: Exactly. It really opens a lot of other questions that need to be answered in the


meantime. Jonas, any other last words about how you may be applying this clinically in your own patients?

Speaker 2: We have no solid data supporting the use of this clinical risk score and as already pointed out, which I think is very good, all clinic risk scores should of course be in the best world validated as useful decision support truth and really in clinics trial seeing that they improve outcomes. This is to my knowledge never been down with a clinical risk scores. We have never used them prospectively to guide treatment and to improve outcomes. Actually, we are aiming to do that. We hope to start a clinical trial next year with ABC score guided treatment compared to standard of care. But it's a very huge undertake of course to that we can improve treatment by risk or guided management.

Carolyn Lam: That's excellent. So remember everyone, you heard it right here. A new trial that they're engaging. I really congratulate you first for this study, as well as this future efforts which are clearly going to be very important.

Thank you very much both of you for joining us today and thank you listeners for listening. Don't forget to tune in next week.

40 4141Ortega-Gomez et al

October 18, 2016 Circulation. 2016;134:1176–1188. DOI: 10.1161/CIRCULATIONAHA.116.0247901188

25. von Hundelshausen P, Weber KS, Huo Y, Proudfoot AE, Nelson PJ, Ley K, Weber C. RANTES deposition by platelets triggers mono-cyte arrest on inflamed and atherosclerotic endothelium. Circula-tion. 2001;103:1772–1777.

26. Gardiner EE, De Luca M, McNally T, Michelson AD, Andrews RK, Berndt MC. Regulation of P-selectin binding to the neutrophil P-selectin counter-receptor P-selectin glycoprotein ligand-1 by neu-trophil elastase and cathepsin G. Blood. 2001;98:1440–1447.

27. Robledo O, Papaioannou A, Ochietti B, Beauchemin C, Legault D, Cantin A, King PD, Daniel C, Alakhov VY, Potworowski EF, St-Pierre Y. ICAM-1 isoforms: specific activity and sensitivity to cleavage by leukocyte elastase and cathepsin G. Eur J Immunol. 2003;33:1351–1360. doi: 10.1002/eji.200323195.

28. Hermant B, Bibert S, Concord E, Dublet B, Weidenhaupt M, Vernet T, Gulino-Debrac D. Identification of proteases involved in the pro-teolysis of vascular endothelium cadherin during neutrophil trans-migration. J Biol Chem. 2003;278:14002–14012. doi: 10.1074/jbc.M300351200.

29. Kessenbrock K, Dau T, Jenne DE. Tailor-made inflammation: how neutrophil serine proteases modulate the inflammatory response. J Mol Med (Berl). 2011;89:23–28. doi: 10.1007/s00109-010-0677-3.

30. Moreno JA, Ortega-Gomez A, Rubio-Navarro A, Louedec L, Ho-Tin-Noé B, Caligiuri G, Nicoletti A, Levoye A, Plantier L, Meilhac O. High-density lipoproteins potentiate α1-antitrypsin therapy in elastase-induced pulmonary emphysema. Am J Respir Cell Mol Biol. 2014;51:536–549. doi: 10.1165/rcmb.2013-0103OC.

31. Chen HM, Chen JC, Shyr MH, Chen MF, Hwang TL, Fan LL, Chi TY, Chi CP. Neutrophil elastase inhibitor (ONO-5046) attenuates reper-fusion-induced hepatic microcirculatory derangement, energy de-pletion and lipid peroxidation in rats. Shock. 1999;12:462–467.

32. Young RE, Voisin MB, Wang S, Dangerfield J, Nourshargh S. Role of neutrophil elastase in LTB4-induced neutrophil transmigration in vivo assessed with a specific inhibitor and neutrophil elastase de-ficient mice. Br J Pharmacol. 2007;151:628–637. doi: 10.1038/sj.bjp.0707267.

33. Wang J, Sukhova GK, Liu J, Ozaki K, Lesner A, Libby P, Kovanen PT, Shi GP. Cathepsin G deficiency reduces periaortic cal-cium chloride injury-induced abdominal aortic aneurysms in mice. J Vasc Surg. 2015;62:1615–1624. doi: 10.1016/j.jvs.2014.06.004.

34. Rafatian N, Karunakaran D, Rayner KJ, Leenen FH, Milne RW, Whit-man SC. Cathepsin G deficiency decreases complexity of athero-sclerotic lesions in apolipoprotein E-deficient mice. Am J Physiol Heart Circ Physiol. 2013;305:H1141–H1148. doi: 10.1152/ajp-heart.00618.2012.

35. Wang J, Sjöberg S, Tang TT, Oörni K, Wu W, Liu C, Secco B, Tia V, Sukhova GK, Fernandes C, Lesner A, Kovanen PT, Libby P, Cheng X, Shi GP. Cathepsin G activity lowers plasma LDL and reduces atherosclerosis. Biochim Biophys Acta. 2014;1842:2174–2183. doi: 10.1016/j.bbadis.2014.07.026.

36. Chiu JJ, Chien S. Effects of disturbed flow on vascular endothe-lium: pathophysiological basis and clinical perspectives. Physiol Rev. 2011;91:327–387. doi: 10.1152/physrev.00047.2009.

37. Aird WC. Phenotypic heterogeneity of the endothelium: II. Rep-resentative vascular beds. Circ Res. 2007;100:174–190. doi: 10.1161/01.RES.0000255690.03436.ae.

38. Scott DW, Patel RP. Endothelial heterogeneity and adhesion mol-ecules N-glycosylation: implications in leukocyte trafficking in inflammation. Glycobiology. 2013;23:622–633. doi: 10.1093/glycob/cwt014.

39. Scott DW, Vallejo MO, Patel RP. Heterogenic endothelial respons-es to inflammation: role for differential N-glycosylation and vas-cular bed of origin. J Am Heart Assoc. 2013;2:e000263. doi: 10.1161/JAHA.113.000263.

40. Herter J, Zarbock A. Integrin regulation during leukocyte recruit-ment. J Immunol. 2013;190:4451–4457. doi: 10.4049/jimmu-nol.1203179.

41. Iwamoto DV, Calderwood DA. Regulation of integrin-mediated adhesions. Curr Opin Cell Biol. 2015;36:41–47. doi: 10.1016/j.ceb.2015.06.009.

42. Pagano MB, Bartoli MA, Ennis TL, Mao D, Simmons PM, Thompson RW, Pham CT. Critical role of dipeptidyl peptidase I in neutrophil re-cruitment during the development of experimental abdominal aor-tic aneurysms. Proc Natl Acad Sci U S A. 2007;104:2855–2860. doi: 10.1073/pnas.0606091104.

43. Schall TJ, Proudfoot AE. Overcoming hurdles in developing suc-cessful drugs targeting chemokine receptors. Nat Rev Immunol. 2011;11:355–363. doi: 10.1038/nri2972.

44. Rossaint J, Zarbock A. Tissue-specific neutrophil recruitment into the lung, liver, and kidney. J Innate Immun. 2013;5:348–357. doi: 10.1159/000345943.

by SEUN

GM

IN O

H on February 6, 2017


ownloaded from

초록

배경

심방세동은 뇌졸중의 위험도와 연관성이 높지만, 그 정도

는 환자마다 큰 차이가 있다. 본 연구에서는 최근 개발된 생

체표지자(biomarker)를 근거로 한 ABC[age, biomarkers,

and clinical history of prior stroke/TIA(transient ischemic

attack)]-뇌졸중 점수를 이용하여 위험도를 예측하는 방법

의 유용성을 확인하고, 그 효과를 기존의 CHA2DS2VASc

또는 ATRIA(Anticoagulation and Risk Factors in Atrial

Fibrillation) 점수를 이용한 방법과 비교하였다.

방법

ABC-뇌졸중 점수는 연령, 생체표지자[NT-proBNP(N-

terminal fragment B-type natriuretic peptide), hs-cTn(high-

sensitivity cardiac troponin)] 그리고 뇌졸중이나 TIA의 병력

을 점수로 환산하는 방법이다. 이 방법은 RE-LY(Randomized

Evaluation of Long-Term Anticoagulation Therapy) 연구 참여

자를 기반으로 그 유용성을 확인하였는데, 항응고요법으로 치료

받은 8,356명의 환자에서 16,137명-년(person-year)의 추적기간

동안 발생한 219건의 뇌졸중이나 전신색전증을 분석하였다. 생

체표지자인 NT-proBNP, hs-cTnT, hs-cTnI 수치는 연구 시작 시

채취한 환자의 혈장을 이용하여 측정하였다.

결과

hs-cTnT를 이용한 ABC-뇌졸중 점수 측정 시, 기존에 정의된

저위험군(<1%/년)에서는 100명-년당 0.76건의 뇌졸중/전신

색전증 발생을 보였고, 중간 정도의 위험군(1-2%/년)에서는

1.48건 그리고 고위험군(>2%/년)에서는 2.60건의 발생을 보

여 적절한 평가 방법임이 검증되었다. 뇌졸중/전신색전증의 위

험률(hazard ratio)은 저위험군 대비 중간 정도의 위험군에서

1.95였고, 저위험군 대비 고위험군에서는 3.44였다. hs-cTnT와

hs-cTnI를 이용한 ABC-뇌졸중 점수의 C index 수치는 0.65로,

CHA2DS2VASc 점수의 0.60(hs-cTnT에서 P=0.004, hs-cTnI

에서 P=0.022)과 ATRIA 점수의 0.61(hs-cTnT에서 P=0.005,

hs-cTnI에서 P=0.034)에 비해 우수하였다.

결론

생체표지자를 근거로 한 ABC-뇌졸중 점수는 위험도를 적절히

측정하였고, 기존의 CHA2DS2VASc나 ATRIA 점수를 이용한

방법보다 예측도가 우수하였다. 이러한 ABC-뇌졸중 점수는 향

후 심방세동 환자의 진료 시 좀 더 개선된 치료방법의 결정 도구

(decision support tool)로 이용될 수 있겠다.

심방세동에서 생체표지자를 근거로 한 뇌졸중 위험도 예측이 유용하다

최 기 준 교수 서울아산병원 심장내과

Arrhythmia

42 43

Circulation. 2016;134:1697–1707. DOI: 10.1161/CIRCULATIONAHA.116.022802 November 29, 2016

ORIGINAL RESEARCH ARTICLE

1697

Jonas Oldgren, MD, PhDZiad Hijazi, MD, PhDJohan Lindbäck, MScJohn H. Alexander, MD,

MHSStuart J. Connolly, MDJohn W. Eikelboom, MB,

BSMichael D. Ezekowitz, MB,

ChB, PhDChristopher B. Granger,

MDElaine M. Hylek, MD, MPHRenato D. Lopes, MD,

PhDAgneta Siegbahn, MD,

PhDSalim Yusuf, MD, PhDLars Wallentin, MD, PhDOn behalf of the RE-LY and

ARISTOTLE Investigators

Original research article

BACKGROUND: Atrial fibrillation is associated with increased but variable risk of stroke. Our aim was to validate the recently developed biomarker-based ABC (age, biomarkers [high-sensitivity troponin and N-terminal fragment B-type natriuretic peptide], and clinical history of prior stroke/transient ischemic attack)-stroke risk score and compare its performance with the CHA2DS2VASc and ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) risk scores.

METHODS: The ABC-stroke score includes age, biomarkers (N-terminal fragment B-type natriuretic peptide and high-sensitivity cardiac troponin), and clinical history (prior stroke). This validation was based on 8356 patients, 16 137 person-years of follow-up, and 219 adjudicated stroke or systemic embolic events in anticoagulated patients with atrial fibrillation in the RE-LY study (Randomized Evaluation of Long-Term Anticoagulation Therapy). Levels of N-terminal fragment B-type natriuretic peptide, high-sensitivity cardiac troponin T (hs-cTnT), and high-sensitivity cardiac troponin I (hs-cTnI) were determined in plasma samples obtained at study entry.

RESULTS: The ABC-stroke score was well calibrated with 0.76 stroke/systemic embolic events per 100 person-years in the predefined low (<1%/y) risk group, 1.48 in the medium (1%–2%/y) risk group, and 2.60 in the high (>2%/y) risk group for the ABC-stroke score with hs-cTnT. Hazard ratios for stroke/systemic embolic events were 1.95 for medium- versus low-risk groups, and 3.44 for high- versus low-risk groups. ABC-stroke score achieved C indices of 0.65 with both hs-cTnT and hs-cTnI, in comparison with 0.60 for CHA2DS2VASc (P=0.004 for hs-cTnT and P=0.022 hs-cTnI) and 0.61 for ATRIA scores (P=0.005 hs-cTnT and P=0.034 for hs-cTnI).

CONCLUSIONS: The biomarker-based ABC-stroke score was well calibrated and consistently performed better than both the CHA2DS2VASc and ATRIA stroke scores. The ABC score should be considered an improved decision support tool in the care of patients with atrial fibrillation.

CLINICAL TRIAL REGISTRATION: URL: http://www.clinicaltrials.gov. Unique identifiers: ARISTOTLE, NCT00412984; RE-LY, NCT00262600.

Performance and Validation of a novel Biomarker-Based stroke risk score for atrial Fibrillation

© 2016 American Heart Association, Inc.

Key Words: anticoagulation ◼ atrial fibrillation ◼ models, cardiovascular ◼ prevention and control ◼ risk assessment ◼ stroke

Correspondence to: Jonas Oldgren, MD, PhD, Uppsala Clinical Research Center, Uppsala Science Park, Dag Hammarskjölds väg 14 B, SE-752 37 Uppsala, Sweden. E-mail [email protected]

Sources of Funding, see page 1704

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH

on February 6, 2017http://circ.ahajournals.org/

Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from



1697

Jonas Oldgren, MD, PhDZiad Hijazi, MD, PhDJohan Lindbäck, MScJohn H. Alexander, MD,

MHSStuart J. Connolly, MDJohn W. Eikelboom, MB,

BSMichael D. Ezekowitz, MB,

ChB, PhDChristopher B. Granger,

MDElaine M. Hylek, MD, MPHRenato D. Lopes, MD,

PhDAgneta Siegbahn, MD,

PhDSalim Yusuf, MD, PhDLars Wallentin, MD, PhDOn behalf of the RE-LY and

ARISTOTLE Investigators

Original research article

BACKGROUND: Atrial fibrillation is associated with increased but variable risk of stroke. Our aim was to validate the recently developed biomarker-based ABC (age, biomarkers [high-sensitivity troponin and N-terminal fragment B-type natriuretic peptide], and clinical history of prior stroke/transient ischemic attack)-stroke risk score and compare its performance with the CHA2DS2VASc and ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) risk scores.

METHODS: The ABC-stroke score includes age, biomarkers (N-terminal fragment B-type natriuretic peptide and high-sensitivity cardiac troponin), and clinical history (prior stroke). This validation was based on 8356 patients, 16 137 person-years of follow-up, and 219 adjudicated stroke or systemic embolic events in anticoagulated patients with atrial fibrillation in the RE-LY study (Randomized Evaluation of Long-Term Anticoagulation Therapy). Levels of N-terminal fragment B-type natriuretic peptide, high-sensitivity cardiac troponin T (hs-cTnT), and high-sensitivity cardiac troponin I (hs-cTnI) were determined in plasma samples obtained at study entry.

RESULTS: The ABC-stroke score was well calibrated with 0.76 stroke/systemic embolic events per 100 person-years in the predefined low (<1%/y) risk group, 1.48 in the medium (1%–2%/y) risk group, and 2.60 in the high (>2%/y) risk group for the ABC-stroke score with hs-cTnT. Hazard ratios for stroke/systemic embolic events were 1.95 for medium- versus low-risk groups, and 3.44 for high- versus low-risk groups. ABC-stroke score achieved C indices of 0.65 with both hs-cTnT and hs-cTnI, in comparison with 0.60 for CHA2DS2VASc (P=0.004 for hs-cTnT and P=0.022 hs-cTnI) and 0.61 for ATRIA scores (P=0.005 hs-cTnT and P=0.034 for hs-cTnI).

CONCLUSIONS: The biomarker-based ABC-stroke score was well calibrated and consistently performed better than both the CHA2DS2VASc and ATRIA stroke scores. The ABC score should be considered an improved decision support tool in the care of patients with atrial fibrillation.

CLINICAL TRIAL REGISTRATION: URL: http://www.clinicaltrials.gov. Unique identifiers: ARISTOTLE, NCT00412984; RE-LY, NCT00262600.

Performance and Validation of a novel Biomarker-Based stroke risk score for atrial Fibrillation

© 2016 American Heart Association, Inc.

Key Words: anticoagulation ◼ atrial fibrillation ◼ models, cardiovascular ◼ prevention and control ◼ risk assessment ◼ stroke

Correspondence to: Jonas Oldgren, MD, PhD, Uppsala Clinical Research Center, Uppsala Science Park, Dag Hammarskjölds väg 14 B, SE-752 37 Uppsala, Sweden. E-mail [email protected]

Sources of Funding, see page 1704

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

by SEUN

GM

IN O



ownloaded from

by SEU

NG

MIN

OH


Dow

nloaded from

49

Oldgren et al

November 29, 2016 Circulation. 2016;134:1697–1707. DOI: 10.1161/CIRCULATIONAHA.116.0228021698

atrial fibrillation (AF) is the most common sustained arrhythmia and is associated with increased but variable risk for stroke, which is substantially de-

creased by treatment with oral anticoagulants.1–3 Risk stratification schemes in AF patients first emerged in the 1990s and have since been reconstructed, refined, and evaluated in patients not receiving antithrombotic treatment and in those patients receiving an oral anti-coagulant.4–8 Current European9 and US10 AF guidelines recommend a risk-based approach to decisions on anti-coagulation treatment in AF based on the CHA2DS2VASc score, which assigns 1 point each for a history of con-gestive heart failure, hypertension, diabetes mellitus, vascular disease, age 65 to 74 years, and sex category (female sex), and 2 points for age ≥75 years, and prior stroke/transient ischemic attack. The CHA2DS2VASc score is based solely on clinical variables, whereas the more recent ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation)8 risk score also includes a measure of renal function by estimation of glomerular filtration rate.

We previously demonstrated that biomarkers, eg, N-terminal fragment B-type natriuretic peptide (NT-proBNP) indicating myocyte stress and high-sensitivity cardiac tro-ponin (hs-cTn) indicating myocardial injury, provide more prognostic information than most clinical characteristics in patients with AF.11–14 In accordance with current rec-ommendations for developing, interpreting, and validat-ing prediction models,15–18 we developed a risk score including the prognostically most important biomarkers and clinical characteristics. This stroke risk score was derived in a cohort of 14 701 patients with AF and bio-markers measured at entry in the ARISTOTLE trial (Apixa-

ban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation).19 The score was named after the included variables age, biomarkers, NT-proBNP and hs-cTn, and clinical history of prior stroke. The ABC (age, biomarkers [high-sensitivity troponin and N-terminal frag-ment B-type natriuretic peptide], and clinical history of prior stroke/transient ischemic attack)-stroke risk score significantly improved risk prediction in comparison with the CHA2DS2VASc score.20

The aim of the present study was to externally vali-date the ABC-stroke score in a large cohort of anticoagu-lated patients with AF from the RE-LY trial (Randomized Evaluation of Long-Term Anticoagulation Therapy) and to compare the performance and utility of the ABC-stroke score with the CHA2DS2VASc and ATRIA stroke scores.

MethODsstudy PopulationsDerivation CohortDetails of the ABC-stroke score derivation cohort from the ARISTOTLE trial have been published previously.20,21 In brief, ARISTOTLE was a double-blind clinical trial that randomly assigned patients with AF at increased risk for stroke to treat-ment with warfarin or apixaban. The median length of follow-up was 1.9 years for the 14 701 of 18 201 ARISTOTLE patients with biomarker samples available at randomization.

Validation cohortDetails of the RE-LY trial have been published previously.22 In brief, RE-LY was a prospective, multicenter, randomized trial comparing 2 blinded doses of dabigatran with open-label war-farin that enrolled 18 113 patients with AF at 951 clinical sites in 44 countries between December 2005 and March 2009. Inclusion criteria were documented AF and at least one of the following risk factors for stroke: previous stroke or transient ischemic attack; congestive heart failure or reduced left ven-tricular ejection fraction (<40%); at least 75 years of age; or at least 65 years of age with diabetes mellitus, hypertension, or coronary artery disease. Exclusion criteria included severe heart valve disorder, recent stroke, creatinine clearance <30 mL/min, or active liver disease. The median length of follow-up was 1.9 years for the 8356 participants with biomarker samples available at randomization.

end Point and Outcome assessmentStroke was defined as the sudden onset of a focal neurologi-cal deficit in a location consistent with the territory of a major cerebral artery and categorized as ischemic, hemorrhagic, or unspecified, in both the RE-LY22 and ARISTOTLE21 trials. Hemorrhagic transformation of ischemic stroke was not con-sidered to be hemorrhagic stroke. Systemic embolism (SE) was defined as an acute vascular occlusion of an extrem-ity or organ, documented by means of imaging, surgery, or autopsy. Blinded Clinical Events Committees reviewed and centrally adjudicated all suspected stroke and SE events in both trials. Both trials, including the biomarker programs, were based on all patients’ written informed consent and approval by institutional review boards or ethics committees.

clinical Perspective

What is new?• The ABC, age, biomarkers (high sensitivity troponin

and N-terminal fragment B-type natriuretic peptide), and clinical history of prior stroke/transient isch-emic attack, risk score has now been externally validated in a cohort of 8356 patients with atrial fibrillation randomly assigned to 2 doses of dabiga-tran or warfarin.

• The biomarker-based score was well calibrated, showed good discriminative ability, and consistently performed better than the clinical CHA2DS2VASc and ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) stroke scores regardless of type of oral anticoagulation.

What are the clinical implications?• The biomarker-based ABC-stroke score containing

only 4 variables seems useful for stroke risk predic-tion in a broad population of anticoagulated patients with atrial fibrillation.

by SEUN

GM

IN O



ownloaded from

44 45Validation of the ABC-Stroke Score in AF



1699

Biochemical MethodsBlood samples were collected in EDTA tubes at randomiza-tion and centrifuged immediately. Plasma samples were fro-zen in aliquots, and stored at –70°C until analyzed centrally at the Uppsala Clinical Research Center laboratory, Uppsala, Sweden. Plasma high-sensitivity cardiac troponin I (hs-cTnI) levels were determined with hs sandwich immunoassays on the ARCHITECT i1000SR (Abbott Diagnostics) platform accord-ing to the instructions of the manufacturer, NT-proBNP and high-sensitivity cardiac troponin T (hs-cTnT) levels in plasma were determined with hs sandwich immunoassays on Cobas Analytics e601 Immunoanalyzer (Roche Diagnostics, Germany) according to the instructions of the manufacturer and have been described previously.12–14

statistical analysesThe derivation and internal validation of the ABC-stroke score has previously been described.20 The ABC-stroke model included the 4 variables: age, hs-cTnT (or hs-cTnI), NT-proBNP, and prior stroke/transient ischemic attack and is presented as a nomogram in the online-only Data Supplement Figure I.

The validation was performed in 8356 patients from the RE-LY trial. For each patient, the 1-year risk for stroke/SE was estimated by applying the model presented in the nomogram (online-only Data Supplement Figure I). Discrimination was assessed by the C index23 and by comparing Kaplan-Meier curves and hazard ratios between the predefined risk catego-ries. Risk categories for the ABC-stroke score were defined as 0% to 1%, 1% to 2%, or >2% risk for stroke/SE within 1 year.

Calibration was assessed by comparing 1-year event rates with predictions from the derivation model by fitting a Cox regression model with the estimated 1-year event probability included as a restricted cubic spline. The ABC-stroke score was compared with the CHA2DS2VASc risk score6,7 and the ATRIA risk score,8 for which calibrations were assessed by compar-ing observed 1-year event rates with the previously published event rates from the original derivation cohorts (online-only Data Supplement Table I). C indices were compared by using 2000 bootstrap samples. In addition, the ABC model was eval-uated in 2 subgroups: (1) in patients without prior stroke and (2) in warfarin-treated patients at sites with low mean time in the therapeutic range (TTR).

For the evaluation of clinical usefulness and net benefit, the derivation and validation cohorts were combined to 1 large data set, n=23 057. The net benefit of using the ABC-stroke score as a clinical decision tool was estimated by using decision curve analysis.24 The net benefit at a given decision threshold is defined as the difference between the proportion of true positives and the proportion of false positives where the latter is weighted by the odds of the specific threshold. The decision curve is then created by calculation of the net benefits for all possible thresholds. Different prediction mod-els can be compared in the decision curve analysis. At any given threshold, the model with the higher net benefit is the preferred model. This is graphically illustrated with a continuum of potential thresholds for stroke/SE risk (x axis) and the net benefit per patient (y axis) relative to assuming that no patient will have a stroke/SE. The added value of the ABC-stroke score was further illustrated by plotting Kaplan-Meier curves for the predefined risk classes by subgroups of the other scores.

These analyses followed the framework for derivation and validation of prediction models proposed by Harrell, Steyerberg, and Steyerberg and Vergouwe.16,18,23 The valida-tion followed the principles and methods described by Royston and Altman and the reporting followed the recently published TRIPOD statement (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis).15,17 All analyses were performed by using R version 3.2 using the packages rms and Hmisc.23,25 The algorithm for the ABC-stroke risk model is presented in online-only Data Supplement Table II.

resUltsBaseline Demographics, Biomarker levels, and event rates in the Validation cohortBaseline demographics and median levels for the as-sessed biomarkers hs-cTnT, hs-cTnI, and NT-proBNP are presented in Table 1. There were only minor differences in baseline characteristics and biomarker levels between this validation cohort and the derivation cohort,20 with the exception of a higher median age in the present co-hort.

This external validation was based on 16 137 person-years of follow-up, and 219 adjudicated stroke or SE events corresponding to an incidence rate of 1.36 per 100 person-years. The incidence rates (events per 100 person-years) within each predefined risk class were as follows: 0.76 for low risk, 1.48 for medium risk, and 2.60 for high risk, for the ABC-stroke score with hs-cTnT (Table 2). The results were similar when using the ABC-stroke score with hs-cTnI (Table 2).

table 1. Baseline Demographics and Biomarker levels

Variable n=8356

Age, y (min–max) 72 (22–95)

Female sex, % 36.7 (3069)

Current smoker, % 7.4 (660)

Diabetes mellitus, % 22.4 (1868)

Hypertension, % 79.2 (6617)

Congestive heart failure, % 28.8 (2407) [1]*

Permanent or persistent atrial fibrillation, % 67.2 (5610) [4]*

Prior stroke or transient ischemic attack, % 19.4 (1619)

Vascular disease, % 19.1 (1597)

Renal function, creatinine clearance, mL/min

68.2 (53.6–86.2) [68]*

Troponin T, ng/L 12.2 (7.7–19.5)

Troponin I, ng/L 6.8 (4.2–13.0)

NT-proBNP, ng/L 807 (382–1447)

For biomarkers, the numbers represent median (interquartile range). NT-proBNP indicates N-terminal pro-B-type natriuretic peptide.

*Numbers in square brackets indicate number of missing values, if any.

by SEUN

GM

IN O



ownloaded from

Oldgren et al


Discrimination and calibrationThe relative hazards in the validation cohort were 1.95 for medium versus low risk, and 3.44 for high versus low risk for ABC-stroke score with hs-cTnT (Table 2). The cumulative event rates up to 2.5 years illustrate good calibration and discriminatory ability for the validation and derivation cohorts (Figure 1). Results for ABC-stroke scores with hs-cTnT and hs-cTnI were similar (Table 2 and online-only Data Supplement Figures II and III).

The ABC-stroke score achieved C indices of 0.65 with both hs-cTnT and hs-cTnI in the validation cohort (Table 3). In the subgroup without prior stroke, the C indices for the

ABC score were 0.63 for hs-cTnT and 0.62 for hs-cTnI. In the TTR<65% subgroup of warfarin-treated patients, the C indices were 0.71 for hs-cTnT and 0.69 for hs-cTnI.

comparison of the aBc-stroke score With cha2Ds2Vasc and atria scoresThe ABC-stroke score C index of 0.65 (both hs-cTnT and hs-cTnI) was significantly higher than the C index of 0.60 for the CHA2DS2VASc stroke risk score, P=0.004 (hs-cTnT) and P=0.022 (hs-cTnI) (Table 3). The C index of 0.61 for the ATRIA stroke score was also significantly lower than with both hs-cTnT and hs-cTnI-based ABC-

table 2. incidence rates and hazard ratios in aBc-stroke risk classes

risk class n events incidence rate* hazard ratio

ABC-stroke score including high-sensitivity troponin T

Low risk (<1%) 3287 49 0.76 (0.56–1.00) 1.00 (ref)

Medium risk (1%–2%) 3762 107 1.48 (1.21–1.78) 1.95 (1.39–2.74)

High risk (>2%) 1307 63 2.60 (2.00–3.33) 3.44 (2.37–4.99)

ABC-stroke score including high-sensitivity troponin I

Low risk (<1%) 3079 45 0.74 (0.54–1.00) 1.00 (ref)

Medium risk (1%–2%) 3854 105 1.41 (1.15–1.71) 1.90 (1.34–2.69)

High risk (>2%) 1423 69 2.61 (2.03–3.30) 3.52 (2.42–5.12)

ABC indicates age, biomarkers (high-sensitivity troponin and N-terminal fragment B-type natriuretic peptide), and clinical history of prior stroke/transient ischemic attack.

*Per 100 person-years.

Figure 1. cumulative event rates of stroke/systemic embolism stratified by predicted 1-year aBc-stroke risk category. Cumulative event rates of stroke/sys-temic embolism up to 2.5 years stratified by predicted 1-year (troponin T–based) ABC-stroke risk category (green=0%–1%, blue=1%–2%, and red >2%) in the validation (solid lines) and the derivation20 (dashed lines) data. The colored vertical bar indicates the 1-year time mark. ABC indicates age, biomarkers (high sensitivity troponin and N-terminal fragment B-type natriuretic peptide), and clinical history of prior stroke/transient ischemic attack.

by SEUN

GM

IN O



ownloaded from




1701

stroke scores, P=0.005 and P=0.034, respectively (Table 3). Similar results were found in the subgroups of patients without prior stroke and with low TTR.

Calibration of the different scores in the derivation and validation cohorts is displayed in Figure 2. The calibra-tion of the ABC-stroke score was superior in comparison with both the CHA2DS2VASc and ATRIA scores.

clinical UtilityThe event rates based on ABC-stroke risk categories with-in different risk categories based on the CHA2DS2VASc

or ATRIA scores are illustrated in Figures 3 and 4. The ABC-stroke score identified patients at higher or lower risk within each risk score category of both the CHA2DS2VASc and ATRIA scores, including patients with ATRIA score <6, CHA2DS2VASc score of 0 to 1 (where 1516 of 1560 had CHA2DS2VASc=1), and patients with only 1 CHA2DS2VASc point irrespective of sex. The decision curve analysis dis-played consistent positive and larger net benefit of using the ABC-stroke score for decision thresholds between 1% and 5% 1-year stroke/SE risk, in comparison with both the CHA2DS2VASc and ATRIA scores (Figure 5, online-only Data Supplement Table III).

table 3. C indices (95% confidence interval) for the aBc-stroke, cha2Ds2Vasc, and atria stroke risk scores

Full cohort no Prior stroke ttr <65%

Events/number of subjects 219/8356 150/6737 30/1109

ABC-stroke* (troponin T) 0.65 (0.61–0.69) 0.63 (0.58–0.67) 0.71 (0.62–0.81)

ABC-stroke* (troponin I) 0.65 (0.61–0.69) 0.62 (0.58–0.67) 0.69 (0.60–0.79)

CHA2DS

2VASc† 0.60 (0.57–0.64) 0.57 (0.53–0.62) 0.60 (0.48–0.72)

ATRIA‡ 0.61 (0.58–0.65) 0.58 (0.54–0.62) 0.64 (0.52–0.76)

TTR indicates time in therapeutic range (international normalized ratio, 2.0–3.0).*ABC-stroke: age, biomarkers (cardiac troponin and NT-proBNP), clinical history (prior stroke/transient ischemic attack)†CHA

2DS

2VASc: assigns 1 point each for congestive heart failure, hypertension, diabetes mellitus, vascular disease, age 65–74

y, and sex category (female sex); and 2 points for age ≥75 y, and prior stroke/transient ischemic attack.‡ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation): assigns 6 (9 for patients with prior stroke) points for age ≥85 y; 5 (7) for

age 75–84 y; 3 (7) points for age 65–74 y; and 0 (8) points for age <65 y, and 1 point each for female sex, diabetes mellitus, chronic heart failure, hypertension, proteinuria, and estimated glomerular filtration rate <45 mL/min or end-stage renal disease.

Figure 2. calibration plots of cha2Ds2Vasc, atria, and aBc scores, in the derivation20 (red) and the validation (blue) cohorts. For each discrete point score, the figure displays the observed stroke/systemic embolism event rate per 100 person-years with 95% confidence interval (y axis) versus the event rate (x axis) previously published from the original derivation cohorts of the CHA2DS2VASc score, in AF patients without6 or with7 oral anticoagulant treatment, or the ATRIA8 score (see online-only Data Supplement Table I). For the ABC-stroke score the figure displays the observed 1-year event rate versus the predicted 1-year risk of stroke/systemic embolism by using the originally derived model. ABC indicates age, biomarkers (high sensitivity troponin and N-terminal fragment B-type natriuretic peptide), and clinical history of prior stroke/transient ischemic attack; AF, atrial fibrillation; and ATRIA, Anticoagulation and Risk Factors in Atrial Fibrillation.

by SEUN

GM

IN O



ownloaded from

Oldgren et al


DiscUssiOnThe novel biomarker-based ABC-stroke score derived in 1 large cohort has now been externally validated in an-other large cohort of AF patients at risk for stroke receiv-

ing oral anticoagulants. The biomarker-based score was well calibrated, showed good discriminative ability, and consistently performed better than the CHA2DS2VASc and ATRIA stroke scores. The ABC-stroke risk score also showed significant net clinical benefit in comparison

Figure 3. cumulative event rates of stroke/systemic embolism stratified by aBc-stroke risk and by cha2Ds2Vasc risk categories. Cumulative event rates of stroke/systemic embolism stratified by predicted 1-year (troponin T–based) ABC-stroke risk category (colored lines) and by CHA2DS2VASc risk categories (panels) in the combined derivation and validation cohorts, n=23 057. ABC indicates age, biomarkers (high sensitivity troponin and N-terminal fragment B-type natriuretic peptide), and clinical history of prior stroke/transient ischemic attack.

Figure 4. cumulative event rates of stroke/systemic embolism stratified by aBc-stroke risk and by atria risk categories. Cumulative event rates of stroke/systemic embolism stratified by predicted 1-year (troponin T–based) ABC-stroke risk category (colored lines) and by ATRIA risk categories (panels) in the combined derivation and validation cohorts, n=23 057. ABC indicates age, biomarkers (high sensitivity troponin and N-terminal fragment B-type natriuretic peptide), and clinical history of prior stroke/transient ischemic attack; and ATRIA, Anticoagulation and Risk Factors in Atrial Fibrillation.

by SEUN

GM

IN O



ownloaded from




1703

with the other stroke risk scores across a wide range of stroke risk in decision curve analysis. The new ABC-stroke risk score should therefore be considered as an improved decision support tool in the care of patients with AF.

The ABC, age, biomarkers, and clinical history of pri-or stroke, risk score was recently derived and internally validated in a large cohort of almost 15 000 patients with AF at risk for stroke.20 In the development of this biomarker-based score, we showed that previously iden-tified clinical risk factors, such as hypertension, diabe-tes mellitus, cardiovascular diseases other than stroke, and female sex, no longer carried important incremental prognostic information when 2 biomarkers, NT-proBNP and hs-cTn, were included in the risk prediction model. These 2 biomarkers are well-established markers of car-diovascular disease and readily available in many labora-tories throughout the world. The ABC-stroke score was well calibrated; predicted event rates were similar to ob-served event rates in both the derivation cohort and in a smaller external validation cohort of >1400 patients, one-half of the latter not receiving anticoagulation treat-ment.20 The previous results have been corroborated by the present validation in a large independent cohort con-sisting of >8000 anticoagulated patients with AF at risk for stroke.

The well-calibrated ABC-stroke score was robust and provided similar results with 2 different hs troponin as-says. The ABC-stroke score containing only 4 variables seems clinically useful for risk prediction of a broad population of AF patients including patients without prior stroke and patients on subtherapeutic warfarin treat-ment. The large number of events and the relatively straightforward prediction model appear to prevent overfitting, as shown by the successful validation in the present cohort and in previously published cohorts of pa-tients with AF.20 A potential advantage of the ABC score

is that 3 of the 4 variables in the ABC-stroke score are continuous, which might improve individual risk predic-tion in comparison with other clinical risk scores based solely on categorical and mainly irreversible risk factors. The biomarkers used in the ABC score may also change over time, although the clinical impact of such changes remains to be further elucidated.26,27

The ABC-stroke score consistently predicted stroke/SE with higher accuracy than the guideline-recommend-ed CHA2DS2VASc risk score,6,7,9 and the more recently developed ATRIA stroke score.8 The decision curve analysis, which depicts net benefits of a prediction model at different thresholds, showed that, if annual risks of stroke/SE in the range of 1% to 5% would be relevant thresholds to guide clinical decisions, the ABC-stroke score provides better clinical usefulness than the CHA2DS2VASc and ATRIA scores. For instance, at a threshold of 2% 1-year stroke/SE risk, the ABC-stroke score will identify 3.0 and 0.8 additional strokes/SEs in a population with 13.8 strokes/SEs per 1000 person-years in comparison with the CHA2DS2VASc and ATRIA scores, respectively, without increasing the number of false positives.

Clinical utility was further reinforced by the finding that ABC-stroke score improved risk prediction within different risk classes of both the CHA2DS2VASc and ATRIA stroke scores. The ABC-stroke score identified patients both at higher or lower risk across a broad range of CHA2DS2VASc scores, including patients with CHA2DS2VASc score 1 or 2, which are the thresholds recommended for treatment decisions in international guidelines.9,10 Thus, the ABC-stroke score might refine treatment decisions especially in patients identified as low risk based on current risk scores, eg, CHA2DS2VASc score=1 or in patients with only 1 CHA2DS2VASc point irrespective of sex (online-only Data Supplement Figure IV), where benefit of oral anticoagulation remains debat-

Figure 5. Decision curve analysis for the aBc-stroke, cha2Ds2Vasc, and atria risk prediction models. The net benefit of using the prediction models to guide clinical decision in rela-tion to assuming that no one is at risk (all negative) or that all are at risk (all positive) for stroke/systemic embolism on the y axis is plotted against relevant decision thresholds on the x axis, n=23 057. ABC indicates age, biomarkers (high sensitivity troponin and N-terminal fragment B-type natriuretic peptide), and clinical history of prior stroke/transient ischemic attack; and ATRIA, Anticoagulation and Risk Factors in Atrial Fibrillation.

by SEUN

GM

IN O



ownloaded from

Oldgren et al


ed.28–30 Patients at higher or lower risks were also iden-tified in patients with low, intermediate, or high stroke risk by the suggested decision thresholds of the ATRIA scores 0 to 5, 6, or >6.8

Currently, there are several different alternative treat-ments for stroke prevention in patients with AF, which have different profiles concerning their effects on the risk of ischemic stroke and the risk of intracranial and other major bleeding events. There is a need for decision sup-port models for identification of the treatment strategy with best balance between the reduction in risk for stroke and the associated risk of bleeding for each patient.31,32 Our recently developed biomarker-based ABC-bleeding risk score33 provides a better, more reliable, and useful tool for risk stratification than the HAS-BLED (Hyperten-sion, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alco-hol Concomitantly)34 and ORBIT-AF (Outcomes Registry for Better Informed Treatment of Atrial Fibrillation)35 bleeding risk scores that are based only on clinical data and routine laboratory tests. Thus, the simultaneous use of these 2 biomarker-based tools, indicating both the risk of stroke and the risk of major bleeding, might provide a future op-portunity for improved tailoring of stroke prevention treat-ment in patients with AF. The balancing and discrimination of the risks for stroke and bleeding should be more accu-rate with the biomarker-based ABC-stroke and ABC-bleed-ing risk scores by inclusion of better targeted, continuous, and fewer overlapping risk factors in comparison with the currently recommended stroke and bleeding scores, eg, CHA2DS2-VASc and HAS-BLED.9,10 The use of biomarkers as part of the ABC scores may also allow monitoring of changes of risk indicators and alteration of the risk-benefit ratio over time.27 Therefore, the current findings, in combi-nation with the availability of the recently presented ABC-bleeding score,33 support a transition of risk stratification efforts toward the combination of biomarkers and clinical information as a next step to personalized precision medi-cine for stroke prevention in AF.

The clinical implementation of the algorithm used in the ABC-stroke score in daily practice can either be based on a nomogram (online-only Data Supplement Figure I), or preferably based on an electronic tool in-tegrated into electronic patient records or as an online tool, please visit www.ucr.uu.se/en/services/abc-risk-calculators.36

strengths and limitationsStrengths of the development and validation of the ABC-stroke risk score include several large independent clinical trial cohorts, standardized recording of clinical characteristics, long-term follow-up, and centrally adju-dicated clinical outcome events, whereas the exclusion of patient with, eg, severe renal dysfunction or short-life expectancy may be a limitation.21,22,37 Another strength

is that the ABC-stroke score was externally validated and reported according to the principles and methods de-scribed by Royston and Altman with the contemporary consensus statement on transparent reporting of mul-tivariable prediction models for individual prognosis or diagnosis (TRIPOD).15,17

The derivation and the present validation of the ABC-stroke score in cohorts of patients with AF at risk for stroke and on treatment with oral anticoagulation thera-py is a limitation. However, randomly assigning patients at increased stroke risk to no anticoagulation would not be ethical, and better prediction of the risk-benefit bal-ance is still important in anticoagulated patients, espe-cially those at low to intermediate risk for stroke, where there is uncertainty about the benefit of oral anticoagula-tion, ie, in patients with CHA2DS2-VASc=1 in men and 1 to 2 in women, respectively.26,28–30

Furthermore, prior studies have shown that patients with TTR <58% to 65% derive little or no net benefit from warfarin.38,39 Therefore, the present results are strength-ened by the consistent performance of the ABC-stroke score in the derivation and validation subgroups of warfarin-treated patients with TTR <65%, and also sup-ported by the previous validation in a smaller cohort of which half of the patients with AF did not receive oral an-ticoagulation therapy.20 Further validation of the biomark-er-based risk score should be encouraged in broader populations of AF patients with and, if possible, without oral anticoagulation, if such cohorts without appropriate stroke prevention treatment still can be identified, and potentially in cohorts of AF patients deemed not at risk for stroke by traditional clinical risk scores.

cOnclUsiOnsThe recently developed biomarker-based ABC-stroke risk score was validated in a large cohort of anticoagulated patients with AF and was shown to be well calibrated, have good discriminative ability, and to consistently per-form better and provide better utility as decision support in comparison with the CHA2DS2VASc and ATRIA risk scores. The ABC-stroke risk score should therefore be considered for implementation as an improved decision support tool in the care of patients with AF.

acKnOWleDgMentsEbba Bergman, PhD, and Sanne Carlsson, BA, BSc, at Up-psala Clinical Research Center, Sweden, provided editorial assistance.

sOUrces OF FUnDingThis work was supported by a grant from The Swedish Founda-tion for Strategic Research, Stockholm, Sweden. The RE-LY trial was funded by Boehringer Ingelheim, Ingelheim, Germany.

by SEUN

GM

IN O



ownloaded from




1705

The ARISTOTLE trial was funded by Bristol-Myers Squibb, Co, Princeton, NJ, and Pfizer Inc, New York, NY.

DisclOsUresDr Oldgren: Consulting and lecture fees from Boehringer In-gelheim, Bayer, Bristol-Myers Squibb, Pfizer. Dr Hijazi: Institu-tional research grants from Boehringer Ingelheim, Bristol-Myers Squibb/Pfizer; lecture fees from Boehringer Ingelheim; consult-ing fees from Bristol-Myers Squibb/Pfizer. Dr Linbäck: Institu-tional research grants from Boehringer Ingelheim, Bristol-Myers Squibb/Pfizer. Dr Alexander: Institutional research grants and consulting fee/honoraria from Bristol-Myers Squibb, Regado Biosciences, Merck; consulting fee/honoraria from Pfizer, Astra-Zeneca, Boehringer Ingelheim, Ortho-McNeil-Janssen, Polymedix, Bayer. Dr Connolly: Consulting fees, speaker fees and research grants from Boehringer Ingelheim, Bristol-Myers Squibb, Bayer, Portola; consulting fees and research grants from Sanofi-Aventis; research grants from Boston Scientific. Dr Eikelboom: Grants and honoraria from AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb/Pfizer, Daiichi-Sankyo, GlaxoSmithKline, Janssen, Sanofi-Aventis; honoraria from Eli Lilly. Dr Ezekowitz: Consulting fees from Boehringer Ingelheim, Pfizer, Sanofi, Bris-tol-Myers Squibb, Portola, Bayer, Daiichi-Sankyo, Medtronics, Aegerion, Merck, Johnson & Johnson, Gilead, Janssen Scientific Affairs, Pozen Inc., Amgen, Coherex, Armatheon. Dr Granger: Grants and consultancy fees from AstraZeneca, Boehringer Ingel-heim, Bristol-Myers Squibb, GlaxoSmithKline, Pfizer, Sanofi-Aven-tis, Takeda, The Medicines Company, Daiichi Sankyo, Janssen, Bayer; grants from Medtronic Foundation, Armetheon; consultan-cy fees from Hoffman-La Roche, Salix Pharmaceuticals, Gilead, Medtronic Inc. Dr Hylek: Advisory board member and symposium lecture fees from Bayer, Boehringer Ingelheim, Bristol-Myers Squibb; advisory board member for Armetheon, Daiichi San-kyo, Janssen, Medtronic, Pfizer, Portola. Dr Lopes: Institutional research grant and consulting fees from Bristol-Myers Squibb; institutional research grant from GlaxoSmithKline; consulting fees from Bayer, Boehringer Ingelheim, Pfizer, Merck, Portola. Dr Siegbahn: Institutional research grants from AstraZeneca, Boeh-ringer Ingelheim, Bristol-Myers Squibb/Pfizer, GlaxoSmithKline. Dr Yusuf: Consulting fees, lecture fees and grant support from Boehringer Ingelheim, AstraZeneca, Bristol-Myers Squibb, Sanofi-Aventis, Bayer, Cadila. Dr Wallentin: Institutional research grants, consultancy fees, lecture fees, and travel support from Bristol-Myers Squibb/Pfizer, AstraZeneca, GlaxoSmithKline, Boehringer Ingelheim; institutional research grants from Merck & Co, Roche; consultancy fees from Abbott; holds 2 patents involving GDF-15.

aFFiliatiOnsFrom Department of Medical Sciences, Cardiology, Uppsala University, Sweden (J.O., Z.H., L.W.); Uppsala Clinical Research Center, Uppsala University, Sweden (J.O., Z.H., J.L., A.S., L.W.); Duke Clinical Research Institute, Duke Medicine, Dur-ham, NC (J.H.A., C.B.G., R.D.L.); Population Health Research Institute, Hamilton, Canada (S.J.C., J.W.E., S.Y.); Thomas Jef-ferson Medical College and the Heart Center, Wynnewood, PA (M.D.E.); Boston University Medical Center, MA (E.M.H.); and Department of Medical Sciences, Clinical Chemistry, Uppsala University, Sweden (A.S.).

FOOtnOtesReceived April 1, 2016; accepted July 29, 2016.

The online-only Data Supplement, podcast, and transcript are available with this article at http://circ.ahajournals.org/look-up/suppl/doi:10.1161/CIRCULATIONAHA.116.022802/-/DC1.

Circulation is available at http://circ.ahajournals.org.

reFerences 1. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an inde-

pendent risk factor for stroke: the Framingham Study. Stroke. 1991;22:983–988.

2. Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV, Singer DE. Prevalence of diagnosed atrial fibrillation in adults: na-tional implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA. 2001;285:2370–2375.

3. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med. 2007;146:857–867.

4. SPAF Investigators. Patients with nonvalvular atrial fibrillation at low risk of stroke during treatment with aspirin: Stroke Prevention in Atrial Fibrillation III Study. JAMA. 1998;279:1273–7.

5. Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Rad-ford MJ. Validation of clinical classification schemes for predict-ing stroke: results from the National Registry of Atrial Fibrillation. JAMA. 2001;285:2864–2870.

6. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137:263–272. doi: 10.1378/chest.09-1584.

7. Lip GY, Frison L, Halperin JL, Lane DA. Identifying patients at high risk for stroke despite anticoagulation: a comparison of contemporary stroke risk stratification schemes in an anticoagu-lated atrial fibrillation cohort. Stroke. 2010;41:2731–2738. doi: 10.1161/STROKEAHA.110.590257.

8. Singer DE, Chang Y, Borowsky LH, Fang MC, Pomernacki NK, Udaltsova N, Reynolds K, Go AS. A new risk scheme to pre-dict ischemic stroke and other thromboembolism in atrial fibril-lation: the ATRIA study stroke risk score. J Am Heart Assoc. 2013;2:e000250. doi: 10.1161/JAHA.113.000250.

9. Camm AJ, Lip GY, De Caterina R, Savelieva I, Atar D, Hohnloser SH, Hindricks G, Kirchhof P; ESC Committee for Practice Guidelines (CPG). 2012 focused update of the ESC Guidelines for the man-agement of atrial fibrillation: an update of the 2010 ESC Guidelines for the management of atrial fibrillation. Developed with the special contribution of the European Heart Rhythm Association. Eur Heart J. 2012;33:2719–2747. doi: 10.1093/eurheartj/ehs253.

10. January CT, Wann LS, Alpert JS, Calkins H, Cigarroa JE, Cleve-land JC Jr, Conti JB, Ellinor PT, Ezekowitz MD, Field ME, Murray KT, Sacco RL, Stevenson WG, Tchou PJ, Tracy CM, Yancy CW; ACC/AHA Task Force Members. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Associ-ation Task Force on practice guidelines and the Heart Rhythm Society. Circulation. 2014;130:e199–e267. doi: 10.1161/CIR.0000000000000041.

11. Hijazi Z, Oldgren J, Andersson U, Connolly SJ, Ezekowitz MD, Hohnloser SH, Reilly PA, Vinereanu D, Siegbahn A, Yusuf S, Wal-lentin L. Cardiac biomarkers are associated with an increased risk of stroke and death in patients with atrial fibrillation: a Random-ized Evaluation of Long-term Anticoagulation Therapy (RE-LY) sub-study. Circulation. 2012;125:1605–1616. doi: 10.1161/CIRCU-LATIONAHA.111.038729.

by SEUN

GM

IN O



ownloaded from

Oldgren et al


12. Hijazi Z, Siegbahn A, Andersson U, Granger CB, Alexander JH, Atar D, Gersh BJ, Mohan P, Harjola VP, Horowitz J, Husted S, Hylek EM, Lopes RD, McMurray JJ, Wallentin L; ARISTOTLE Inves-tigators. High-sensitivity troponin I for risk assessment in patients with atrial fibrillation: insights from the Apixaban for Reduction in Stroke and other Thromboembolic Events in Atrial Fibrillation (AR-ISTOTLE) trial. Circulation. 2014;129:625–634. doi: 10.1161/CIRCULATIONAHA.113.006286.

13. Hijazi Z, Wallentin L, Siegbahn A, Andersson U, Alexander JH, Atar D, Gersh BJ, Hanna M, Harjola VP, Horowitz JD, Husted S, Hylek EM, Lopes RD, McMurray JJ, Granger CB; ARISTOTLE In-vestigators. High-sensitivity troponin T and risk stratification in patients with atrial fibrillation during treatment with apixaban or warfarin. J Am Coll Cardiol. 2014;63:52–61. doi: 10.1016/j.jacc.2013.07.093.

14. Hijazi Z, Wallentin L, Siegbahn A, Andersson U, Christersson C, Ezekowitz J, Gersh BJ, Hanna M, Hohnloser S, Horowitz J, Huber K, Hylek EM, Lopes RD, McMurray JJ, Granger CB. N-terminal pro-B-type natriuretic peptide for risk assessment in patients with atrial fibrillation: insights from the ARISTOTLE Trial (Apixa-ban for the Prevention of Stroke in Subjects With Atrial Fibrilla-tion). J Am Coll Cardiol. 2013;61:2274–2284. doi: 10.1016/j.jacc.2012.11.082.

15. Moons KG, Altman DG, Reitsma JB, Ioannidis JP, Macaskill P, Steyerberg EW, Vickers AJ, Ransohoff DF, Collins GS. Transpar-ent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med. 2015;162:W1–73. doi: 10.7326/M14-0698.

16. Steyerberg EW. Clinical Prediction Models: A Practical Approach to Development, Validation, and Updating. New York, NY: Springer Verlag; 2009.

17. Royston P, Altman DG. External validation of a Cox prognos-tic model: principles and methods. BMC Med Res Methodol. 2013;13:33. doi: 10.1186/1471-2288-13-33.

18. Steyerberg EW, Vergouwe Y. Towards better clinical prediction models: seven steps for development and an ABCD for valida-tion. Eur Heart J. 2014;35:1925–1931. doi: 10.1093/eurheartj/ehu207.

19. Granger CB, Alexander JH, McMurray JJ, Lopes RD, Hylek EM, Hanna M, Al-Khalidi HR, Ansell J, Atar D, Avezum A, Bahit MC, Diaz R, Easton JD, Ezekowitz JA, Flaker G, Garcia D, Geraldes M, Gersh BJ, Golitsyn S, Goto S, Hermosillo AG, Hohnloser SH, Horowitz J, Mohan P, Jansky P, Lewis BS, Lopez-Sendon JL, Pais P, Parkhomenko A, Verheugt FW, Zhu J, Wallentin L; ARISTOTLE Committees and Investigators. Apixaban versus warfarin in pa-tients with atrial fibrillation. N Engl J Med. 2011;365:981–992. doi: 10.1056/NEJMoa1107039.

20. Hijazi Z, Lindbäck J, Alexander JH, Hanna M, Held C, Hylek EM, Lopes RD, Oldgren J, Siegbahn A, Stewart RA, White HD, Granger CB, Wallentin L; ARISTOTLE and STABILITY Investigators. The ABC (age, biomarkers, clinical history) stroke risk score: a biomarker-based risk score for predicting stroke in atrial fibrillation. Eur Heart J. 2016;37:1582–1590. doi: 10.1093/eurheartj/ehw054.

21. Lopes RD, Alexander JH, Al-Khatib SM, Ansell J, Diaz R, Easton JD, Gersh BJ, Granger CB, Hanna M, Horowitz J, Hylek EM, McMurray JJ, Verheugt FW, Wallentin L; ARISTOTLE Investigators. Apixaban for reduction in stroke and other ThromboemboLic events in atrial fibrillation (ARISTOTLE) trial: design and rationale. Am Heart J. 2010;159:331–339. doi: 10.1016/j.ahj.2009.07.035.

22. Ezekowitz MD, Connolly S, Parekh A, Reilly PA, Varrone J, Wang S, Oldgren J, Themeles E, Wallentin L, Yusuf S. Rationale and design of RE-LY: randomized evaluation of long-term anticoagu-lant therapy, warfarin, compared with dabigatran. Am Heart J. 2009;157:805–10, 810.e1. doi: 10.1016/j.ahj.2009.02.005.

23. Harrel FE. Regression Modeling Strategies: With Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis. New York, NY: Springer Verlag; 2015.

24. Vickers AJ, Cronin AM, Elkin EB, Gonen M. Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers. BMC Med Inform Decis Mak. 2008;8:53. doi: 10.1186/1472-6947-8-53.

25. R Core Team. R: A language and environment for statistical com-puting. R Foundation for Statistical Computing, Vienna, Austria. 2016. https://www.R-project.org/.

26. Eckman MH, Singer DE, Rosand J, Greenberg SM. Moving the tipping point: the decision to anticoagulate patients with atrial fi-brillation. Circ Cardiovasc Qual Outcomes. 2011;4:14–21. doi: 10.1161/CIRCOUTCOMES.110.958108.

27. Hijazi Z, Oldgren J, Andersson U, Connolly SJ, Ezekowitz MD, Hohnloser SH, Reilly PA, Siegbahn A, Yusuf S, Wallentin L. Im-portance of persistent elevation of cardiac biomarkers in atrial fibrillation: a RE-LY substudy. Heart. 2014;100:1193–1200. doi: 10.1136/heartjnl-2013-304872.

28. Friberg L, Skeppholm M, Terént A. Benefit of anticoagulation un-likely in patients with atrial fibrillation and a CHA2DS2-VASc score of 1. J Am Coll Cardiol. 2015;65:225–232. doi: 10.1016/j.jacc.2014.10.052.

29. Fauchier L, Lecoq C, Clementy N, Bernard A, Angoulvant D, Ivanes F, Babuty D, Lip GY. Oral anticoagulation and the risk of stroke or death in patients with atrial fibrillation and one additional stroke risk factor: the Loire Valley Atrial Fibrillation Project. Chest. 2016;149:960–968. doi: 10.1378/chest.15-1622.

30. Lip GY, Skjøth F, Nielsen PB, Larsen TB. Non-valvular atrial fi-brillation patients with none or one additional risk factor of the CHA2DS2-VASc score. A comprehensive net clinical benefit analysis for warfarin, aspirin, or no therapy. Thromb Haemost. 2015;114:826–834. doi: 10.1160/TH15-07-0565.

31. Hijazi Z, Oldgren J, Siegbahn A, Granger CB, Wallentin L. Bio-markers in atrial fibrillation: a clinical review. Eur Heart J. 2013;34:1475–1480. doi: 10.1093/eurheartj/eht024.

32. Kirchhof P, Breithardt G, Aliot E, Al Khatib S, Apostolakis S, Au-ricchio A, Bailleul C, Bax J, Benninger G, Blomstrom-Lundqvist C, Boersma L, Boriani G, Brandes A, Brown H, Brueckmann M, Calkins H, Casadei B, Clemens A, Crijns H, Derwand R, Dobrev D, Ezekowitz M, Fetsch T, Gerth A, Gillis A, Gulizia M, Hack G, Haegeli L, Hatem S, Häusler KG, Heidbüchel H, Hernandez-Brichis J, Jais P, Kappenberger L, Kautzner J, Kim S, Kuck KH, Lane D, Leute A, Lewalter T, Meyer R, Mont L, Moses G, Mueller M, Münzel F, Näbauer M, Nielsen JC, Oeff M, Oto A, Pieske B, Pisters R, Potpara T, Rasmussen L, Ravens U, Reiffel J, Richard-Lordereau I, Schäfer H, Schotten U, Stegink W, Stein K, Stein-beck G, Szumowski L, Tavazzi L, Themistoclakis S, Thomitzek K, Van Gelder IC, von Stritzky B, Vincent A, Werring D, Willems S, Lip GY, Camm AJ. Personalized management of atrial fibrilla-tion: Proceedings from the fourth Atrial Fibrillation competence NETwork/European Heart Rhythm Association consensus con-ference. Europace. 2013;15:1540–1556. doi: 10.1093/euro-pace/eut232.

33. Hijazi Z, Oldgren J, Lindbäck J, Alexander JH, Connolly SJ, Eikel-boom JW, Ezekowitz MD, Held C, Hylek EM, Lopes RD, Siegbahn A, Yusuf S, Granger CB, Wallentin L; ARISTOTLE and RE-LY Investi-gators. The novel biomarker-based ABC (age, biomarkers, clinical history)-bleeding risk score for patients with atrial fibrillation: a derivation and validation study. Lancet. 2016;387:2302–2311. doi: 10.1016/S0140-6736(16)00741-8.

34. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138:1093–1100. doi: 10.1378/chest.10-0134.

35. O’Brien EC, Simon DN, Thomas LE, Hylek EM, Gersh BJ, Ansell JE, Kowey PR, Mahaffey KW, Chang P, Fonarow GC, Pen-cina MJ, Piccini JP, Peterson ED. The ORBIT bleeding score: a simple bedside score to assess bleeding risk in atrial fi-

by SEUN

GM

IN O



ownloaded from




1707

brillation. Eur Heart J. 2015;36:3258–3264. doi: 10.1093/ eurheartj/ehv476.

36. ABC risk calculators. Sweden: UCR, Uppsala Clincial Research Center. http://www.ucr.uu.se/en/services/abc-risk-calculators

37. White H, Held C, Stewart R, Watson D, Harrington R, Budaj A, Steg PG, Cannon CP, Krug-Gourley S, Wittes J, Trivedi T, Tarka E, Wallentin L. Study design and rationale for the clinical outcomes of the STABILITY Trial (STabilization of Atherosclerotic plaque By Initiation of darapLadIb TherapY) comparing darapladib versus placebo in patients with coronary heart disease. Am Heart J. 2010;160:655–661. doi: 10.1016/j.ahj.2010.07.006.

38. Hylek EM, Go AS, Chang Y, Jensvold NG, Henault LE, Selby JV, Singer DE. Effect of intensity of oral anticoagulation on stroke severity and mortality in atrial fibrillation. N Engl J Med. 2003;349:1019–1026. doi: 10.1056/NEJMoa022913.

39. Connolly SJ, Pogue J, Eikelboom J, Flaker G, Commerford P, Franzosi MG, Healey JS, Yusuf S; ACTIVE W Investigators. Benefit of oral anticoagulant over antiplatelet therapy in atrial fibrillation depends on the quality of international normalized ratio control achieved by centers and countries as measured by time in thera-peutic range. Circulation. 2008;118:2029–2037. doi: 10.1161/CIRCULATIONAHA.107.750000.

by SEUN

GM

IN O



ownloaded from

초록

배경

동반질환이 없는 정상 심장의 심방세동(lone atrial fibrillation,

LAF)은 정상 박동으로 회복된 후에도 지속되는 심근병증의 존재

를 반영할 수 있으며, 이는 심방세동의 재발과 연관이 있다. 저자

들은 이 가설을 증명하고자 LAF 환자를 대상으로 전극도자절제

술(radiofrequency ablation)에 의한 정상 박동으로의 회복이 좌

심실 기능과 에너지대사에 미치는 영향을 관찰하였다.

방법

심장판막질환, 조절되지 않는 고혈압과 갑상선질환, 관상동맥질

환, 전신 염증질환, 당뇨병, 폐쇄성 무호흡증 등이 없는 발작성 혹

은 지속성 심방세동 환자에서 전극도자절제술을 시행한 53명의

환자와 정상 박동인 25명의 대조군을 대상으로 연구를 진행하였

다. 자기공명영상(magnetic resonance imaging, MRI)으로 좌심

실 구혈률(left ventricular ejection fraction, LVEF), PSCS(peak

systolic circumferential strain), 좌심방 용적 및 기능을 측정하였

고, 31P-MRS(phosphorus-31 magnetic resonance spectroscopy)

로는 심실의 에너지학적 지표[energetics: phosphocreatine/

ATP(adenosine triphosphate) 비율]를 측정하였다.

결과

전극도자절제술 전의 좌심실 기능과 에너지학적 지표는 심

방세동군에서 대조군에 비해 유의하게 감소되었다[LVEF,

61%(interquartile range[IQR], 52-65%) vs. 71%(IQR, 69-73%),

P<0.001; PSCS, -15%(IQR, -11 to -18%) vs. -18%(IQR, -17 to

-19%), P=0.002; phosphocreatine/ATP 비율, 1.81±0.35 vs. 2.05

±0.29, P=0.004]. 또한, 심방세동군에서 대조군에 비해 좌심방이

커져 있었고, 기능은 저하되어 있었다(모두 P<0.001). 전극도자절

제술 후 초기(1-4일)에 심방세동에서 정상 박동으로 전환된 환자

는 LVEF와 PSCS가 개선되었으나(LVEF, 7.0±10%, P=0.005;

PSCS, -3.5±4.3%, P=0.001), 전극도자절제술 전후 모두 정상 박

동인 환자에서는 두 지표 모두 차이가 없었다(모두 P=NS). 전극

도자절제술 후 6-9개월이 지난 시점에서 심방세동의 발현 비중

(burden)은 유의하게 감소했다[54%(IQR, 1.5-100%) to 0%(IQR

0-0.1%); P<0.001]. 그러나 LVEF와 PSCS는 개선되지 않았으

며(모두 P=NS), 대조군에 비해 저하된 상태로 남아있었다(각

각 P<0.001, P=0.003). 심방 기능도 시술 전에 비해 개선되지 않

았고(P=NS), 대조군과 비교해서 저하되어 있었다(P<0.001).

Phosphocreatine/ATP 비율은 평가하는 동안의 심장 박동 상태

와 시술 전 심방세동의 발현 비중 등에 영향을 받지 않았다(모

두 P=NS). 또한, phosphocreatine/ATP 비율은 시술 후에도 변

화가 없었으며(P=0.57), 동율동으로 회복되거나 심방세동의 재

발이 없음에도 대조군에 비해 감소되어 있었다(각각 P=0.006,

P=0.002).

결론

LAF는 좌심실의 에너지대사 감소와 미묘한(subtle)기능 저하를

동반하였으며, 전극도자절제술 후 정상 박동으로 유지되어도 회

복되지 않았다. 이러한 소견은 심방세동이 심근병증을 유발하는

원인이 아니라, 오히려 시술로 심방세동이 조절된 후에도 회복되

지 않는 잠재적(occult) 심근병증으로 인해 나타난 결과임을 시사

한다.

정상 심장에서 발생한 심방세동은 성공적인 전극도자절제술 후에도 지속되는 좌심실 에너지대사의 저하와 관련이 있다

황 교 승 교수 아주대학교병원 순환기내과

Arrhythmia

performance and validation of a novel biomarker-based...

Documents