Combination of Angiographic and ClinicalRisk Scores in Left Main Revascularization
Corrado Tamburino, MD, PhDFull Professor of Cardiology, Director of Postgraduate School of Cardiology
Chief Cardiovascular Department, Director Cardiology Division, Interventional Cardiology and Heart Failure Unit, University of Catania, Ferrarotto Hospital, Catania, Italy
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Disclosure Statement of Financial Interest
I, Corrado Tamburino, DO NOT have a financial interest/arrangement or affiliationfinancial interest/arrangement or affiliation with one or more organizations that could be perceived as a real or apparent conflictbe perceived as a real or apparent conflict of interest in the context of the subject of this presentationthis presentation
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
The Ideal Risk Stratification tool
Easy to apply at the bed-side or in the cath-labReproducibleUses data routinely available before the procedurep ocedu eAccurate
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Accuracy is a function of two characteristics
0.00
The The idealidealscore for score for LM PCILM PCI
1.00
0.00sh
ow
ette
r
Discrimination
LM PCILM PCI
2.00
mer
-Lem
es
tion
be
DiscriminationMeasures how much the score can differentiate between
3.00Hos
m
alib
rat differentiate between
poor and good outcomes
0.60 0.65 0.70 0.75 0.80
4.00
Ca
CalibrationMeasures how close the estimates are to a
C-statistic
Discrimination betterreal probability
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Risk Stratification in LM Disease
Angiographic Clinical Combined Functional• SYNTAX score
(Sxscore)• Residual
SYNTAX score
• EuroSCORE• EuroSCORE 2*• STS score*
C
• GRC• CSS• Logistic CSS*
S
• FunctionalSYNTAX score (FSS)*
• Non invasiveSYNTAX score (rSS)*
• ACEF score • NERS • Non-invasive FSS*
*not yet presented or validated in LM PCI
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Risk Stratification in LM Disease
Angiographic Clinical Combined Functional• SYNTAX score
(Sxscore)• Residual
SYNTAX score
• EuroSCORE• EuroSCORE 2*• STS score*
C
• GRC• CSS• Logistic CSS*
S
• FunctionalSYNTAX score (FSS)*
• Non invasiveSYNTAX score (rSS)*
• ACEF score • NERS • Non-invasive FSS*
*not yet presented or validated in LM PCI
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Stand-Alone Clinical Scores (ACEF, EuroSCORE) versus SYNTAX score in LM PCI
EuroSCORE ACEF score SYNTAX scoreP < 0.001 P < 0.001 P < 0.001
EuroSCORE ACEF score SYNTAX score
Δ = 17% Δ = 14% Δ = 12%ΔHIGH-LOW = 17% ΔHIGH-LOW = 14% ΔHIGH-LOW = 12%
Hosmer-Lemeshow: 1.607c-statistic: 0.69
Hosmer-Lemeshow: 0.216c-statistic : 0.69
Hosmer-Lemeshow: 2.448c-statistic : 0.73
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno D et al. JACC Interv 2011;4:287-97
Stand-alone Scores Are Far from Perfection
0.00The The idealidealscore forscore forACEF
1.00
0.00
show
ette
rscore for score for LM PCILM PCI
ACEF
2.00
mer
-Lem
es
tion
be EuroSCORE
SYNTAX Score
3.00Hos
m
alib
rat SYNTAX Score
0.60 0.65 0.70 0.75 0.80
4.00
Ca
C-statistic
Discrimination better
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno D et al. JACC Interv 2011;4:287-97
Combined Risk models in LM PCI: The best of both worlds?
Angiographic• SYNTAX score
(Sxscore)
Clinical• EuroSCORE• ACEF score
Combined• NERS• CSS(Sxscore) • ACEF score • CSS• GRC
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Angiography is not enoughg g p y g
5% Clinical and angiographic1917%
11%5%
ore
Clinical and angiographic scores summarize very different information in patients with unprotected
<19
18%8%
10%
NTA
X sc
o patients with unprotected LM19-27
17%
8%
5%8%
SYN Low Spearman rank
correlation coefficient between SYNTAX score > 27
5% and EuroSCORE(RS=0.204, p = 0.001)
0-2 3-6 > 6EuroSCORE
The frequency of patients for each cross-tabulation cell is shown within a rectangle that is proportional in size to the frequency
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno et al, Am Heart J 2010:159:103-9
The New Risk Classification (NERS) ( )
Based on 17 clinical 4 proceduralBased on 17 clinical, 4 procedural, and 33 angiographic variablesBetter discriminates a broad array of endpoints than SYNTAX scoreof endpoints than SYNTAX score, including MACE, Death, MI, TVR and ST in patients undergoing LM PCI
BackgroundTh t ti l t ib ti f li i lThe potential contributions of clinical, procedural, and angiographic indices in LM patients have not been fully elucidatedelucidated
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Chen SL et al. JACC Interv. 2010;3:632-41
Caveats of NERS
Labor-intensive54 variables
OverfittedOverfitted126 variables tested in 260 patients
No prospective validationUtility in decision-making remains uncertainUtility in decision making remains uncertain
Does include procedural variables, cannot be calculated upfrontcalculated upfront
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Clinical SYNTAX score (CSS)( )
Calculated as SYNTAXCalculated as SYNTAX score*modified ACEF scoreBetter discriminates 5-Year Death and MACE in multivessel CAD thanand MACE in multivessel CAD than SYNTAX scoreBetter discriminates 5-Year Death in All-Comers PCI
BackgroundB i l l b d i hi
All Comers PCI
Being solely based on angiographicvariables, the SYNTAX score cannot account for the variability related to clinical factors which are widelyclinical factors which are widely acknowledged to impact on long-term outcomes
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Garg S et al. Circ Card Interv 2010;3:317-26Girasis C et al. Eur Heart J 2011;32:3115-27
CSS – No improvement in discrimination of MACE vs. SYNTAX score Better discrimination in mortality but limitations remain
Death
score. Better discrimination in mortality, but limitations remain
MACET til t i
P<0.001 P<0.001CSS
17.4±20.5
Tertiles are not in the anticipated order
M d t i iModest gain in calibration vsSYNTAX score (⎟2 8 vs 6)
SYNTAX score P≤0.03 P≤0.03
c-statistic = 0.62 c-statistic = 0.66
11.7±7.3
c-statistic = 0 61 c-statistic = 0 58
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Nam CW et al. J Am Coll Cardiol 2011;58:1211-8
c-statistic = 0.61 c-statistic = 0.58
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of CataniaSerruys PW, ACC 2012
Global Risk Classification (GRC)( )
Developed as an Integration of theDeveloped as an Integration of the SYNTAX score and the EuroSCOREBetter discriminates in-hospital and 2-Year Cardiac Death in LM PCI than2 Year Cardiac Death in LM PCI than SYNTAX score alone
ConclusionsIncorporation of clinical risk factors and comorbidities into existing estimation systems may refine their prognostic ability and guide clinical decisions
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno D et al. Am Heart J 2010;159:103-9
GRC – Prognostic ability in improving bothdiscrimination and calibration vs SYNTAX score
%) 10
th fr
ee s
urvi
val (
% 100
90
80
96.1%
94.6%
78.1%P = 0.004*SYNTAX score
Car
diac
dea
t
70
60SYNTAX score
LOWMIDDLEHIGHL L I0-2
<19 19-27 >27SYNTAX score
E
Time (months)0 12 24
val (
%) 10
098.4%
L L IL L I3-6
uroS
CO
RE
c de
ath
free
sur
viv
90
80
70
84.0%
P < 0.001*I I H> 6Eu
Car
diac 70
60
Time (months)0 12 24
GRC68.6%LOW
MIDDLEHIGH
* log rank test; n = 255 LM patients undergoing PCI
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno D et al. Am Heart J 2010;159:103-9
Cardiac death to 2 Years by stratification ofGRC and CSS in LM PCI (N = 400)
LOW MIDDLE HIGHLOW MIDDLE HIGH
)
40
)
40Global Risk Classification Clinical SYNTAX score
ent R
ate
(%)
30
ent R
ate
(%)
30P < 0.001* P < 0.001*24.2% 25.6%
ΔHIGH LOW = 23 6% ΔHIGH LOW = 23 0%HL = 0.357 c-statistic = 0.74 HL = 3.833 c-statistic = 0.76
mul
ativ
e Ev
e
20
10
mul
ativ
e Ev
e
20
108.7%
ΔHIGH-LOW = 23.6% ΔHIGH-LOW = 23.0%
Cu
0
Ti ( th )0 12 24
Cu
0
Ti ( th )0 12 24
0.6% 2.6%
1.0%
* log rank test; higher HL (Hosmer-Lemeshow) statistic indicates poorer calibration; higher ΔHIGH-LOW (Index of separation) indicates better discrimination
Time (months) Time (months)
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno D et al. JACC Interv 2011;4:287-97
GRC approaches the ideal model for LM PCI
0.00ACEF The The idealideal
score forscore forACEF GRC
1.00
0.00
show
ette
rEuroSCORE
score for score for LM PCILM PCI
ACEF GRC
2.00
mer
-Lem
es
tion
be
SYNTAX Score
EuroSCORE
SYNTAX Score
3.00Hos
m
alib
rat SYNTAX Score
CSS
0.60 0.65 0.70 0.75 0.80
4.00
Ca
C-statistic
Discrimination better
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno D et al. JACC Interv 2011;4:287-97
Lessons on the GRC from the SYNTAX trial
DeathMACCE
Excellent mortality stratificationImproved MACCE stratification
The GRC may help to identify a population at very low risk of events after LM PCI
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Serruys et al., presented at LM Summit 2011
PCI vs CABG in LM Patients with Low GRC in the SYNTAX Trial at 3 Years
HR 0.16, 95% CI HR 0.16, 95% CI 0.030.03--0.70, 0.70, P=0.005P=0.005
HR 0.64, 95% CI 0.39HR 0.64, 95% CI 0.39--1.07, 1.07, P=0.088P=0.088
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Serruys et al. J Am Coll Cardiol 2011;58:87 (abstract)
Decision-making based on the Global RiskClassification
≤22 22-32 ≥32
SYNTAX score The comparative role of PCI and CABG in LM patients in the intermediate
PCI PCI ?0-2
patients in the intermediate GRC risk group (≈40%) is not well defined
PCI PCI ?3-5
roSC
OR
E
? ? CABG≥6
Eur
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Decision-making based on the Global RiskClassification: cardiac death
PCI
≤22 22-32 ≥32
SYNTAX score P=0.65
PCI PCI INT20-2
PCI PCI INT23-5
roSC
OR
E
CABG
P=0 22
INT1 INT1 CABG≥6
Eur P 0.22
INT1 = high clinical/acceptable angiographic riskINT2 = acceptable clinical/high angiographic risk
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno D et al. Int J Cardiol 2011;150:116-7
2 g g g
Decision-making based on the Global RiskClassification
≤22 22-32 ≥32
SYNTAX score Performing CABG in patients with SYNTAX score ≥32 complies with
PCI PCI CABG0-2
score ≥32 complies with guidelines
The efficacy and safety of
PCI PCI CABG3-5
roSC
OR
E The efficacy and safety of performing PCI in patients with SYNTAX score <33 will be addressed by the
PCI /CABG
PCI /CABG CABG≥6
Eur be addressed by the
EXCEL trial
/CABG /CABG
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania
Capodanno D et al. Int J Cardiol 2011;150:116-7
Conclusions
Risk stratification in LM PCI aims to govern gthe unpredictability of random variation Risk estimation and classification are bestRisk estimation and classification are best achieved by integrating clinical, angiographic and functional informationangiographic and functional informationWith multiplication of risk scoring systems
d difi ti f i ti d land modifications from existing models, expert consensus on how to use these tools f d i i ki i ti l t idfor decision-making is essential to avoid a “Tower of Babel” effect
Ferrarotto HospitalUniversity of CataniaFerrarotto HospitalUniversity of Catania