CLINICAL RESEARCH
European Heart Journal (2015) 36, 509–515 doi:10.1093/eurheartj/ehu412
Interventional cardiology
Risk model for estimating the 1-year risk of deferred lesion intervention following deferred revascularization after fractional flow reserve assessment
Division of Cardiology, Department of Medicine, Washington University School of Medicine, 660 South Euclid Avenue, PO Box 8086, St. Louis, MO 63110, USA Received 5 April 2014; revised 22 August 2014; accepted 23 September 2014; online publish-ahead-of-print 21 October 2014
Aims
Although lesions deferred revascularization following fractional flow reserve (FFR) assessment have a low risk of adverse cardiac events, variability in risk for deferred lesion intervention (DLI) has not been previously evaluated. The aim of this study was to develop a prediction model to estimate 1-year risk of DLI for coronary lesions where revascularization was not performed following FFR assessment. ..................................................................................................................................................................................... Methods A prediction model for DLI was developed from a cohort of 721 patients with 882 coronary lesions where revascularand results ization was deferred based on FFR between 10/2002 and 7/2010. Deferred lesion intervention was defined as any revascularization of a lesion previously deferred following FFR. The final DLI model was developed using stepwise Cox regression and validated using bootstrapping techniques. An algorithm was constructed to predict the 1-year risk of DLI. During a mean (+SD) follow-up period of 4.0 + 2.3 years, 18% of lesions deferred after FFR underwent DLI; the 1-year incidence of DLI was 5.3%, while the predicted risk of DLI varied from 1 to 40%. The final Cox model included the FFR value, age, current or former smoking, history of coronary artery disease (CAD) or prior percutaneous coronary intervention, multi-vessel CAD, and serum creatinine. The c statistic for the DLI prediction model was 0.66 (95% confidence interval, CI: 0.61– 0.70). ..................................................................................................................................................................................... Conclusion Patients deferred revascularization based on FFR have variation in their risk for DLI. A clinical prediction model consisting of five clinical variables and the FFR value can help predict the risk of DLI in the first year following FFR assessment.
----------------------------------------------------------------------------------------------------------------------------------------------------------Keywords
Fractional flow reserve † Deferred lesion intervention
Introduction Fractional flow reserve (FFR) is an invasive measure of the physiologic significance of a coronary lesion that is increasingly being used to determine the need for revascularization.1 Fractional flow reserve cut-offs of 0.75 and 0.80 have been empirically validated against inducible ischaemia on non-invasive stress testing1 and for selection of lesions for percutaneous coronary intervention (PCI).2,3
Randomized clinical trials and observational studies have demonstrated that FFR-guided PCI improves clinical outcomes and is cost effective.2 – 6 Deferral of revascularization for lesions judged nonsignificant by FFR is safe and associated with a low risk of adverse cardiac events.2 – 5,7 – 14 Among previous studies, the rate of future intervention for target lesions that were deferred revascularization based on FFR (i.e. deferred lesion intervention, DLI) was highly variable, ranging from 2.5 to 11% within the first year following
* Corresponding author. Tel: +1 314 747 4535, Fax: +1 314 747 1417, Email: [email protected] Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2014. For permissions please email: [email protected].
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Jeremiah P. Depta, Jayendrakumar S. Patel, Eric Novak, Brian F. Gage, Shriti K. Masrani, David Raymer, Gabrielle Facey, Yogesh Patel, Alan Zajarias, John M. Lasala, Amit P. Amin, Howard I. Kurz, Jasvindar Singh*, and Richard G. Bach
510 deferral.2 – 5,7 – 14 Variation in rates of DLI may relate to the reference FFR cut-off (,0.75 vs. ≥0.80) used, lesion-type studied (e.g. left main only), differences in the clinical setting, or in baseline characteristics of the patients studied. Limited evidence exists on the clinical predictors of adverse cardiac events in patients deferred revascularization after FFR.7 – 11,14 To our knowledge, no study has systematically assessed the clinical predictors of DLI following deferral based on FFR assessment. Therefore, our objectives were to determine the clinical risk predictors associated with DLI in a large, contemporary, real-world cohort of patients and to quantify the risk of DLI within the first year following deferred revascularization after FFR assessment.
Methods
Results Baseline patient and lesion characteristics The mean age of the 721 patients with 882 lesions deferred revascularization was 64.5 years and there was a high prevalence of cardiovascular risk factors and history of coronary artery disease (CAD) or PCI (Tables 1 and 2). Approximately half of patients underwent FFR assessment during an ACS, primarily for unstable angina (39%) and less often acute myocardial infarction (12%). Multi-vessel CAD was present in 64% of patients. The mean (+SD) FFR value was 0.87 + 0.05. Intracoronary adenosine was administered in 99% of the lesions (n ¼ 875) assessed with FFR, while the remaining lesions were assessed using intravenous adenosine (n ¼ 7). The mean maximum dose of intracoronary adenosine administered was 140 + 78 mg and no difference in adenosine dose was observed between lesions with and without a DLI event. Of the 882 lesions deferred (Figure 1), for 93% of lesions (n ¼ 816) the FFR value was .0.80, while 7% of lesions (n ¼ 65) had values in the range of 0.75–0.80. One patient was deferred PCI with an FFR value of
Table 1 Baseline patient characteristics of the study population N 5 721 patients
................................................................................ Age (years)
64.5 + 11.2
Male
409 (57%)
................................................................................ Medical history Diabetes mellitus Hypertension
270 (37%) 598 (83%)
Hyperlipidaemia
584 (81%)
Current/former smoker History of CAD or prior PCI
376 (52%) 493 (68%)
Prior CABG
94 (13%)
Peripheral arterial disease Congestive heart failure
81 (11%) 158 (22%)
Creatinine (mg/dL)
1.2 + 1.2
Multi-vessel CADa
465 (64%)
................................................................................ Diagnosis at FFR assessment Acute myocardial infarction STEMI
84 (12%) 7 (1%)
NSTEMI
77 (11%)
Unstable angina Stable angina
280 (39%) 261 (36%)
Asymptomatic/atypical chest pain
96 (13%)
................................................................................ Procedural characteristics PCI of another lesion
195 (27%)
CAD, coronary artery disease; CABG, coronary artery bypass graft; FFR, fractional flow reserve; NSTEMI, non-ST-elevation myocardial infarction; PCI, percutaneous coronary intervention; STEMI, ST-elevation myocardial infarction. a Multi-vessel CAD was defined as two or more significant lesions (angiographic percent stenosis ≥ 50%) at the time of FFR assessment. Values are shown as absolute numbers (percentages) or mean + SD.
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
The study population comprised 721 patients with 882 coronary lesions deferred PCI based on FFR. The primary outcome of the study was DLI, defined as any PCI or coronary artery bypass graft (CABG) of a lesion deferred revascularization based on the index FFR assessment. Patient and lesion-level covariates known at the time of FFR assessment were considered for the prediction model. With the exception of one missing creatinine level, none of the covariates analysed contained missing data. First, associations between potential predictor covariates and DLI were assessed using univariate Cox proportional hazards models with a marginal Cox model approach to account for correlated data in patients having multiple deferred lesions.15,16 To create a parsimonious prediction model, Spearman correlation coefficients were used to examine the linear relationship between covariate pairs. Tolerance values were examined to identify multi-collinearity. Covariate pairs with strong correlations and/or small tolerance values (,0.10) were excluded. The relationship between candidate variables and DLI was examined using a multivariable Cox proportional hazards model reduced using a stepwise variable selection technique employing a threshold P , 0.20 for entry and P ≥ 0.05 for removal. The reduced model was internally validated for discrimination and calibration using bootstrapping techniques to adjust for over-fitting and over-optimistic model performance. Discrimination was evaluated using a survival-based c statistic to describe the ability of the model to distinguish DLI events and non-events during follow-up.17,18 An optimism-corrected c statistic was calculated using 500 bootstrap samples created with replacement. A Cox proportional hazards model was built for each bootstrap sample using the same stepwise variable selection technique. For each bootstrap model, two c statistics were determined, one for the bootstrap sample and one for the original sample. The difference, bootstrap less original, was averaged over all 500 bootstrap samples to identify the optimism present in the c statistic due to internal validation. Calibration was assessed using the mean calibration slope averaged over all bootstrap samples. For each bootstrap model, the predicted value of DLI among the original sample was used as the independent variable in a Cox regression model to examine agreement with observed DLI events. The final model is presented as an algorithm using the regression coefficients to calculate the probability of DLI within 1-year following deferred PCI based on FFR assessment. To additionally examine model fit, a calibration plot was created and 1-year probability of freedom from DLI estimated from the algorithm was compared with the observed 1-year Kaplan –Meier estimates for freedom from DLI. All statistical analyses were performed using SAS 9.3 (SAS Institute, Cary, NC, USA). All comparisons were two tailed. A P-value of ,0.05 was considered statistically significant. Additional study methods are described in Supplementary material online, Appendix.
J.P. Depta et al.
511
The 1-year risk of deferred lesion intervention
Table 2 Lesion characteristics and outcomes of the study population N 5 882 lesions
................................................................................ Lesion and procedural characteristics Lesion location Left main coronary artery
61 (7%)
Proximal left coronary artery (LM/LAD/LCx) Proximal LAD
297 (34%) 129 (15%)
Any LAD lesion
315 (36%)
Diagonal branch Proximal LCx
72 (8%) 107 (12%)
Any LCx
139 (16%)
Obtuse marginal branch Proximal RCA
75 (9%) 64 (7%) 196 (22%)
Coronary artery bypass graft
24 (3%)
Percent stenosis Myocardial jeopardy index score
59 + 10 5.2 + 2.8
................................................................................
Bifurcation lesion
142 (16%)
Ostial lesion Prior PCI of lesion
188 (21%) 80 (9%)
FFR value
0.87 + 0.05
Deferred lesion intervention Overall
155 (18%)
1 year
47 (5%)
Values are shown as absolute numbers (percentages) or mean + SD. All other abbreviations as shown in Table 1. LAD, left anterior descending; LCx, left circumflex; LM, left main; RCA, right coronary artery.
0.74, and did not have a DLI. At the time of FFR assessment, 27% of patients underwent PCI of another lesion. The discharge medications following deferred revascularization after FFR assessment were similar between patients in the DLI and no DLI groups (Table 3).
Incidence, clinical setting and reasons for performing deferred lesion intervention Following deferral of revascularization based on FFR, 155 lesions (18%) underwent DLI during a mean follow-up period of 4.0 + 2.3 years, of which 115 (74%) were revascularized by PCI and the remaining 40 (26%) by CABG. At 1 year, the rate of DLI was 5.3% (n ¼ 47), and 30% of DLIs occurred within the first year following index FFR assessment. During the study period, death occurred in 119 patients (17%), and 44 patients (6%) died within the 12 months after deferred revascularization based on FFR. A total of 79 acute myocardial infarctions (AMIs) (11%) occurred during follow-up in the study population, and the culprit lesion was a lesion previously deferred revascularization based on the index FFR in 38% of the AMIs (n ¼ 30). The rate of AMI due to a previously deferred lesion within the first year after FFR assessment was 0.8%. AMI accounted for 15% of the DLIs performed in the first year following deferral based on FFR.
Deferred lesion intervention risk model and algorithm The univariate associations between potential predictor covariates and DLI are shown in Table 4. Using a stepwise variable selection technique, the final model (Table 5) included the following prediction variables: age, current or former smoking, history of CAD or prior PCI, creatinine, multi-vessel CAD, and FFR value. The c statistic of the final DLI prediction model was 0.66 (95% CI: 0.61–0.70). The optimismcorrected c statistic for the final DLI prediction model was 0.63. The model discriminates DLI in patients who underwent FFR assessment in the setting of ACS vs. no ACS, but model performance is improved in ACS patients compared with non-ACS patients (c statistic 0.71 vs. 0.59, P ¼ 0.02). An algorithm for the final model was created to calculate the probability of freedom from DLI within 1-year following deferred revascularization based on FFR assessment using the b coefficients of the six independent predictors (Table 5). Model calibration, assessment, and the final algorithm are presented in the Supplementary material online, Appendix (to calculate the 1-year probability of DLI following deferral, see online tool at www.DLI.wustl.edu). In Figure 3, the predicted 1-year DLI risk based on the final algorithm for each deferred lesion in the study is shown. The 1-year predicted DLI risk varied from 1 to 40%. The risk for DLI at 1 year was stratified into three categories, low, intermediate, and high risk, based on quintiles of predicted risk.
Discussion In the current study, we examined the incidence and variability of risk of DLI and developed a clinical prediction model to quantify the risk of DLI within the first year following deferral of revascularization based on FFR assessment. To our knowledge, this is the first study to examine the clinical risk factors associated with DLI following FFR in a real-world patient cohort. The prediction model was based on the clinical predictors of DLI known at the time of initial FFR lesion assessment. The predictors of DLI following deferred revascularization were: younger age, a history of current or former smoking, a history of CAD or prior PCI, creatinine level, multi-vessel CAD, and the FFR value for the lesion. Knowledge of a patient’s risk for future revascularization of a lesion deferred based on FFR may provide clinicians and patients with useful information when
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Any RCA lesion
Of the 155 DLIs (Figure 2; Supplementary material online, Appendix), 101 lesions (65%) underwent urgent revascularization: 30 DLIs (19%) were performed for AMI, of which 6 (4%) were for ST-elevation myocardial infarction, and 24 (15%) were for non-STelevation myocardial infarction. The remaining DLI lesions (46%) that underwent urgent revascularization were for unstable angina (n ¼ 71), where 25% (n ¼ 18 of 71) had new ST-segment and/or T-wave electrocardiogram changes. Elective revascularization comprised 34% of the DLIs (n ¼ 52), predominantly for stable angina (n ¼ 50). For two lesions from one patient, the clinical data were insufficient to definitively classify the reasons that led to DLI. Deferred lesion interventions that occurred within the first year following deferred revascularization were associated with a higher proportion of urgent revascularization compared with DLIs performed beyond the first year after deferral (Supplementary material online, Table S1 and Appendix).
512
J.P. Depta et al.
Table 3 Discharge medications following deferred revascularization based on fractional flow reserve assessment Total (n 5 721 patients)
DLI (n 5 134 patients)
No DLI (n 5 587 patients)
P-value
687 (95%) 362 (50%)
126 (94%) 73 (54%)
561 (96%) 289 (49%)
0.50 0.29
................................................................................ Aspirin Clopidogrel ACE/ARB
465 (64%)
84 (63%)
381 (65%)
0.62
Statin Beta blocker
578 (80%) 553 (77%)
107 (80%) 104 (78%)
471 (80%) 449 (76%)
0.90 0.82
Calcium channel blocker
225 (31%)
47 (35%)
178 (30%)
0.30
Nitrates
290 (40%)
62 (46%)
228 (39%)
0.12
Values are shown as absolute numbers (percentages). ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; DLI, deferred lesion intervention.
considering potential strategies to prevent later adverse events leading to revascularization. The cumulative rate of DLI after deferral based on FFR assessment during a mean follow-up period of 4.0 + 2.3 years was 18%, and the 1-year incidence of DLI was 5.3%. In the randomized DEFER study, the rate of DLI in the group deferred PCI (n ¼ 91) was 8.9% at 5 years,4 and in an observational study of 56 patients deferred PCI
after FFR the 5-year rate of DLI was 10.7%.12 However, those studies only included patients referred for elective PCI. In our study, 50% of patients were deferred revascularization in the setting of ACS (12% with AMI) and represent a population at higher risk for DLI compared with elective patients.4,12 The largest previous real-world experience on long-term clinical outcomes in patients following FFR assessment reported on 721 patients who had revascularization deferred based on the FFR results.5 In that study, the rate of DLI over a median follow-up of 48.7 months was 20.6%, consistent with the rate of DLI observed in our study (A. Lerman, personal communication).5 In prior studies, DLI within the first year after deferring revascularization based on FFR ranged from 2.5 to 11%.2 – 5,7 – 14 Two observational studies of deferred PCI following FFR that included both ACS and non-ACS patients reported 1-year rates of DLI of 10–11% (A. Lerman, personal communication).5,14 The observed rate and the rate predicted by the model-derived algorithm identify rates of 1-year DLI that are consistent with prior studies.2 – 7,9 – 16 In our study population, the rate of subsequent MI was 11% over the mean follow-up period of 4.0 + 2.3 years, and the culprit lesion was a previously deferred lesion based on FFR in 38% of the subsequent MIs. The rate of MI at 1 year among deferred lesions was 0.8%, and by the end of follow-up, 3.8% of lesions (n ¼ 30) deferred revascularization by FFR had a later MI. The rate of MI due to lesions deferred revascularization due to FFR assessment has varied in previous reports. In an observational study of left main lesions deferred revascularization based on FFR, the rate of MI related to a deferred wlesion was 5% at a mean follow-up of 11 months.7 In the DEFER
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Figure 1 Distribution of fractional flow reserve values. The frequencies of fractional flow reserve values for the study population are shown as percentages (bars) and counts (# above each bar) for every 0.01 fractional flow reserve increment.
513
The 1-year risk of deferred lesion intervention
vascular ultrasound; FFR, fractional flow reserve; NSTEMI, non-ST-elevation myocardial infarction; STEMI, ST-elevation myocardial infarction.
study, no patients in the deferred group had an MI at 5-year follow-up; however, that study excluded patients with ACS.4 The 2-year follow-up of the FAME trial and other observational studies have reported a deferred lesion-related MI rate of ,1% among patients who underwent FFR assessment after referral for elective PCI.3,8,12 An observational study of 721 deferred lesions reported a rate of MI of 6% at 7 years, but in that study only 4% of the deferred lesions had index FFR assessment performed in the setting of ACS.5 The potentially higher rate of MI in our study may reflect the higher acuity of patients undergoing FFR assessment, where half of the patients deferred revascularization presented with ACS. Of note, the absence of DLI for over 94% of patients at 1 and 82% at an average of 4 years following deferral by FFR in our higher risk population support the safety of FFR-guided decision making for management of intermediate lesions. Our model determined that five clinical variables and the FFR value were able to predict the risk of DLI. Although no prior studies have specifically assessed the clinical predictors of DLI following deferral based on FFR, prior studies have examined variables among patients with deferred lesions to identify predictors of adverse cardiac events, many of which were DLIs and relevant to our study.7 – 11,14 Our study observed an association between higher FFR values and a lower risk for DLI, consistent with previous analyses that identified lower FFR value as a predictor of adverse outcome.10,11 One study analysing
deferral of revascularization for lesions with FFR values ranging between 0.75 and 0.90 observed that for every 0.01 increase in FFR the rate of adverse cardiac events was lower (hazards ratio: 0.87, 95% CI: 0.80– 0.95).11 A second observational study also showed an inverse relationship between FFR value and adverse cardiac events for deferred lesions.10 These data suggest that the index FFR value, even when above the threshold for haemodynamic severity and prompting deferral of revascularization, retains a negative correlation with the risk of subsequent events. A previous study of lesions deferred PCI based on FFR found that a lower dose of intracoronary adenosine for FFR measurement was significantly associated with higher subsequent adverse cardiac events.7 In distinction, our study did not identify adenosine dose as a predictor of DLI, and the mean dose of intracoronary adenosine used to induce maximal hyperaemia was not different between deferred lesions with and without a subsequent DLI event. Whether lesions deferred revascularization based on FFR assessment benefit from any specialized management strategy is unknown. The goal of our study was to develop a risk prediction model to quantify the risk of DLI within the first year following deferred revascularization after FFR assessment with the potential to assist clinicians in management. The appropriate strategy to prevent DLI among highrisk lesions deferred revascularization after initial FFR assessment is
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Figure 2 Clinical setting and reasons for performing deferred lesion intervention following index fractional flow reserve assessment. IVUS, intra-
514
J.P. Depta et al.
not known. While there is no evidence to support pre-emptive early revascularization for such lesions, we speculate that optimizing a patient’s medical management following deferred PCI may offer the best opportunity to prevent subsequent revascularization for lesions at a higher risk for DLI. In our study, current smoking predicted DLI, suggesting aggressive smoking cessation might reduce the risk of DLI. Secondary prevention strategies have not been prospectively tested in patients with deferred lesions following FFR assessment. Our risk prediction model can identify patients at a
Table 4 Univariate relationship between baseline characteristics and deferred lesion intervention HR (95% CI)
P-value
Age (per 1-year increase) Male
0.98 (0.96– 0.99) 1.22 (0.85– 1.74)
,0.001 0.29
Diabetes mellitus
1.12 (0.78– 1.60)
0.55
Hypertension Hyperlipidaemia
1.23 (0.80– 1.89) 1.45 (0.89– 2.35)
0.35 0.13
Current/former smoker
1.60 (1.12– 2.28)
0.01
History of CAD or prior PCI Prior CABG
1.76 (1.15– 2.69) 1.20 (0.73– 1.97)
0.009 0.47
higher risk for revascularization following deferral based on FFR, and thereby identify those who may warrant closer follow-up and derive even greater benefit from a focused intensification of secondary prevention strategies.
Study limitations The risk prediction model was validated internally by bootstrapping techniques. However, external validation of our risk prediction model using an independent sample has not yet been performed. The study is retrospective, observational, and conducted at a large urban tertiary referral centre. It is unknown if the clinical predictors of DLI identified in our study population are generalizable in different populations at hospitals of different sizes, level of acuity, and location (e.g. rural vs. urban). The risk model was designed to use patient and
................................................................................
1.29 (0.78– 2.12)
0.33
Congestive heart failure Creatinine (per 1 mg/dL increase)
1.26 (0.84– 1.89) 1.17 (1.10– 1.25)
0.27 ,0.001
Acute myocardial infarction
1.40 (0.82– 2.39)
0.21
Multi-vessel CAD PCI of another lesion
1.84 (1.21– 2.79) 1.15 (0.79– 1.66)
0.005 0.48
Proximal left coronary artery (LM/LAD/LCx)
1.33 (0.95– 1.87)
0.10
Proximal right coronary artery
0.77 (0.42– 1.42)
0.40
Percent stenosis Myocardial jeopardy index score
1.01 (0.99– 1.02) 1.03 (0.98– 1.09)
0.45 0.29
Bifurcation lesion
0.81 (0.50– 1.31)
0.38
Ostial lesion Prior PCI of lesion
0.93 (0.60– 1.44) 1.29 (0.79– 2.12)
0.74 0.31
FFR value (per 0.05 unit decrease)
1.26 (1.08– 1.46)
0.004
All abbreviations as shown in Tables 1 and 2. HR, hazards ratio.
Table 5
Figure 3 Variability in the predicted 1-year deferred lesion intervention risk using the final algorithm. The frequency of predicted 1-year deferred lesion intervention risk calculated using the final algorithm for each deferred lesion in the study, ranging from 1 to 40%. Deferred lesion intervention risk at 1 year was stratified into three categories based on quintiles of predicted risk: low (,3% ¼ lowest quintile), intermediate (3 – 7% ¼ 3 middle quintiles), and high risk (.7% ¼ highest quintile).
Multivariable predictors and 1-year b regression coefficients for freedom from DLI in the finala model HR (95% CI)
P-value
b coefficients
............................................................................................................................................................................... Age (per 1-year increase) Current/former smoker
0.98 (0.97– 0.99) 1.49 (1.04– 2.14)
0.005 0.03
20.02075 0.39710
History of CAD or prior PCI
1.62 (1.05– 2.49)
0.03
0.48086
Creatinine (per 1 mg/dL increase) Multi-vessel CAD
1.15 (1.08– 1.22) 1.68 (1.09– 2.58)
,0.001 0.02
0.13681 0.51777
FFR value (per 0.05 unit decrease)
1.21 (1.03– 1.42)
0.02
23.81032
a The model was reduced using a stepwise variable selection technique. For prediction purposes, the 1-year baseline estimate of freedom from DLI for a patient with all covariates set to zero or to the reference group is 0.169. All abbreviations as shown in Tables 1 and 2.
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Peripheral arterial disease
The 1-year risk of deferred lesion intervention
lesion characteristics known at the time of FFR assessment. However, patient management can vary over time and potentially alter an individual’s risk for DLI. It is also possible that unmeasured confounders, including variability in practice patterns across the providers and operators in the study, may have contributed to the risk for DLI following deferred revascularization based on FFR.
Conclusions
Supplementary material Supplementary material is available at European Heart Journal online. Conflict of interest: J.P.D., J.S.P., E.N., B.F.G., S.K.M., D.R., G.F., Y.P., A.P.A., H.I.K.: no disclosures; A.Z.: Steering committee of Partner 2 trial; J.M.L.: Boston Scientific Advisory Board.; J.S.: consulting: Abbott Vascular, Boston Scientific, Volcano Corp.; speakers’ Bureaus/Honoraria: Medicines Company, Medtronic Vascular, Volcano Corp., St. Jude Corp; R.G.B.: Research Grants: AstraZeneca, Eli Lilly, Bristol-Myers Squibb, Merck, Schering-Plough (no personal compensation). Consultant (Clinical Event Committee Activity only): Roche (Significant) and Pfizer (Modest).
References 1. Pijls NH, De Bruyne B, Peels K, Van Der Voort PH, Bonnier HJ, Bartunek J, Koolen JJ, Koolen JJ. Measurement of fractional flow reserve to assess the functional severity of coronary-artery stenoses. N Engl J Med 1996;334:1703 –1708. 2. Tonino PA, De Bruyne B, Pijls NH, Siebert U, Ikeno F, van’t Veer M, Klauss V, Manoharan G, Engstrom T, Oldroyd KG, Ver Lee PN, MacCarthy PA, Fearon WF. FAME Study Investigators. Fractional flow reserve versus angiography for guiding percutaneous coronary intervention. N Engl J Med 2009;360:213 –224. 3. De Bruyne B, Pijls NH, Kalesan B, Barbato E, Tonino PA, Piroth Z, Jagic N, Mobius-Winkler S, Rioufol G, Witt N, Kala P, MacCarthy P, Engstrom T, Oldroyd KG, Mavromatis K, Manoharan G, Verlee P, Frobert O, Curzen N, Johnson JB, Juni P, Fearon WF. FAME 2 Trial Investigators. Fractional flow reserveguided PCI versus medical therapy in stable coronary disease. N Engl J Med 2012; 367:991 –1001.
4. Pijls NH, van Schaardenburgh P, Manoharan G, Boersma E, Bech JW, van’t Veer M, Bar F, Hoorntje J, Koolen J, Wijns W, de Bruyne B. Percutaneous coronary intervention of functionally nonsignificant stenosis: 5-year follow-up of the DEFER Study. J Am Coll Cardiol 2007;49:2105 –2111. 5. Li J, Elrashidi MY, Flammer AJ, Lennon RJ, Bell MR, Holmes DR, Bresnahan JF, Rihal CS, Lerman LO, Lerman A. Long-term outcomes of fractional flow reserveguided vs. angiography-guided percutaneous coronary intervention in contemporary practice. Eur Heart J 2013;34:1375 –1383. 6. Fearon WF, Bornschein B, Tonino PA, Gothe RM, Bruyne BD, Pijls NH, Siebert U. Fractional Flow Reserve Versus Angiography for Multivessel Evaluation Study I. Economic evaluation of fractional flow reserve-guided percutaneous coronary intervention in patients with multivessel disease. Circulation 2010;122:2545 –2550. 7. Courtis J, Rodes-Cabau J, Larose E, Potvin JM, Dery JP, Larochelliere RD, Cote M, Cousterousse O, Nguyen CM, Proulx G, Rinfret S, Bertrand OF. Usefulness of coronary fractional flow reserve measurements in guiding clinical decisions in intermediate or equivocal left main coronary stenoses. Am J Cardiol 2009;103: 943 –949. 8. Courtis J, Rodes-Cabau J, Larose E, Dery JP, Nguyen CM, Proulx G, Gleeton O, Roy L, Barbeau G, Noel B, DeLarochelliere R, Bertrand OF. Comparison of medical treatment and coronary revascularization in patients with moderate coronary lesions and borderline fractional flow reserve measurements. Catheter Cardiovasc Interv 2008;71: 541 –548. 9. Hamilos M, Muller O, Cuisset T, Ntalianis A, Chlouverakis G, Sarno G, Nelis O, Bartunek J, Vanderheyden M, Wyffels E, Barbato E, Heyndrickx GR, Wijns W, De Bruyne B. Long-term clinical outcome after fractional flow reserve-guided treatment in patients with angiographically equivocal left main coronary artery stenosis. Circulation 2009;120:1505 –1512. 10. Meuwissen M, Chamuleau SA, Siebes M, de Winter RJ, Koch KT, Dijksman LM, van den Berg AJ, Tijssen JG, Spaan JA, Piek JJ. The prognostic value of combined intracoronary pressure and blood flow velocity measurements after deferral of percutaneous coronary intervention. Catheter Cardiovasc Interv 2008;71:291 –297. 11. Lavi S, Rihal CS, Yang EH, Fassa AA, Elesber A, Lennon RJ, Mathew V, David HR Jr, Lerman A. The effect of drug eluting stents on cardiovascular events in patients with intermediate lesions and borderline fractional flow reserve. Catheter Cardiovasc Interv 2007;70:525 –531. 12. Rieber J, Jung P, Koenig A, Schiele T, Shapiro M, Hoffmann U, Klauss V. Five-year follow-up in patients after therapy stratification based on intracoronary pressure measurement. Am Heart J 2007;153:403–409. 13. Costa MA, Sabate M, Staico R, Alfonso F, Seixas AC, Albertal M, Crossman A, Angiolillo DJ, Zenni M, Sousa JE, Macaya C, Bass TA. Anatomical and physiologic assessments in patients with small coronary artery disease: final results of the Physiologic and Anatomical Evaluation Prior to and After Stent Implantation in Small Coronary Vessels (PHANTOM) trial. Am Heart J 2007;153:296.e1 –e7. 14. Fischer JJ, Wang XQ, Samady H, Sarembock IJ, Powers ER, Gimple LW, Ragosta M. Outcome of patients with acute coronary syndromes and moderate coronary lesions undergoing deferral of revascularization based on fractional flow reserve assessment. Catheter Cardiovasc Interv 2006;68:544 –548. 15. Lee EW, Wei LJ, Amato DA, Leurgans S. Cox-type regression analysis for large numbers of small groups of correlated failure time observations. In: Klein JP, Goel PK (eds), Survival Analysis: State of the Art. Dordrecht, The Netherlands: Kluwer Academic Publishers; 1992, p237 –247. 16. Lin DY, Wei LJ. The robust inference for the proportional hazards model. J Am Stat Assoc 1989;84:1074 – 1078. 17. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361 –387. 18. Pencina MJ, D’Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;23:2109 – 2123.
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
The risk of DLI for coronary lesions deferred revascularization based on FFR was 5.3% within 1 year and 18% within 4 years. We derived a risk prediction model and algorithm that uses five clinical variables known at the time of FFR assessment and the FFR value to predict the risk of DLI within 1 year following deferred revascularization based on FFR. While requiring further study, use of this model for identification of patients at higher predicted risk of DLI may alert clinicians to a need for closer follow-up, optimized medical therapy and intensified risk factor modification.
515
European Heart Journal (2015) 36, 509–515 doi:10.1093/eurheartj/ehu412
Interventional cardiology
Risk model for estimating the 1-year risk of deferred lesion intervention following deferred revascularization after fractional flow reserve assessment
Division of Cardiology, Department of Medicine, Washington University School of Medicine, 660 South Euclid Avenue, PO Box 8086, St. Louis, MO 63110, USA Received 5 April 2014; revised 22 August 2014; accepted 23 September 2014; online publish-ahead-of-print 21 October 2014
Aims
Although lesions deferred revascularization following fractional flow reserve (FFR) assessment have a low risk of adverse cardiac events, variability in risk for deferred lesion intervention (DLI) has not been previously evaluated. The aim of this study was to develop a prediction model to estimate 1-year risk of DLI for coronary lesions where revascularization was not performed following FFR assessment. ..................................................................................................................................................................................... Methods A prediction model for DLI was developed from a cohort of 721 patients with 882 coronary lesions where revascularand results ization was deferred based on FFR between 10/2002 and 7/2010. Deferred lesion intervention was defined as any revascularization of a lesion previously deferred following FFR. The final DLI model was developed using stepwise Cox regression and validated using bootstrapping techniques. An algorithm was constructed to predict the 1-year risk of DLI. During a mean (+SD) follow-up period of 4.0 + 2.3 years, 18% of lesions deferred after FFR underwent DLI; the 1-year incidence of DLI was 5.3%, while the predicted risk of DLI varied from 1 to 40%. The final Cox model included the FFR value, age, current or former smoking, history of coronary artery disease (CAD) or prior percutaneous coronary intervention, multi-vessel CAD, and serum creatinine. The c statistic for the DLI prediction model was 0.66 (95% confidence interval, CI: 0.61– 0.70). ..................................................................................................................................................................................... Conclusion Patients deferred revascularization based on FFR have variation in their risk for DLI. A clinical prediction model consisting of five clinical variables and the FFR value can help predict the risk of DLI in the first year following FFR assessment.
----------------------------------------------------------------------------------------------------------------------------------------------------------Keywords
Fractional flow reserve † Deferred lesion intervention
Introduction Fractional flow reserve (FFR) is an invasive measure of the physiologic significance of a coronary lesion that is increasingly being used to determine the need for revascularization.1 Fractional flow reserve cut-offs of 0.75 and 0.80 have been empirically validated against inducible ischaemia on non-invasive stress testing1 and for selection of lesions for percutaneous coronary intervention (PCI).2,3
Randomized clinical trials and observational studies have demonstrated that FFR-guided PCI improves clinical outcomes and is cost effective.2 – 6 Deferral of revascularization for lesions judged nonsignificant by FFR is safe and associated with a low risk of adverse cardiac events.2 – 5,7 – 14 Among previous studies, the rate of future intervention for target lesions that were deferred revascularization based on FFR (i.e. deferred lesion intervention, DLI) was highly variable, ranging from 2.5 to 11% within the first year following
* Corresponding author. Tel: +1 314 747 4535, Fax: +1 314 747 1417, Email: [email protected] Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2014. For permissions please email: [email protected].
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Jeremiah P. Depta, Jayendrakumar S. Patel, Eric Novak, Brian F. Gage, Shriti K. Masrani, David Raymer, Gabrielle Facey, Yogesh Patel, Alan Zajarias, John M. Lasala, Amit P. Amin, Howard I. Kurz, Jasvindar Singh*, and Richard G. Bach
510 deferral.2 – 5,7 – 14 Variation in rates of DLI may relate to the reference FFR cut-off (,0.75 vs. ≥0.80) used, lesion-type studied (e.g. left main only), differences in the clinical setting, or in baseline characteristics of the patients studied. Limited evidence exists on the clinical predictors of adverse cardiac events in patients deferred revascularization after FFR.7 – 11,14 To our knowledge, no study has systematically assessed the clinical predictors of DLI following deferral based on FFR assessment. Therefore, our objectives were to determine the clinical risk predictors associated with DLI in a large, contemporary, real-world cohort of patients and to quantify the risk of DLI within the first year following deferred revascularization after FFR assessment.
Methods
Results Baseline patient and lesion characteristics The mean age of the 721 patients with 882 lesions deferred revascularization was 64.5 years and there was a high prevalence of cardiovascular risk factors and history of coronary artery disease (CAD) or PCI (Tables 1 and 2). Approximately half of patients underwent FFR assessment during an ACS, primarily for unstable angina (39%) and less often acute myocardial infarction (12%). Multi-vessel CAD was present in 64% of patients. The mean (+SD) FFR value was 0.87 + 0.05. Intracoronary adenosine was administered in 99% of the lesions (n ¼ 875) assessed with FFR, while the remaining lesions were assessed using intravenous adenosine (n ¼ 7). The mean maximum dose of intracoronary adenosine administered was 140 + 78 mg and no difference in adenosine dose was observed between lesions with and without a DLI event. Of the 882 lesions deferred (Figure 1), for 93% of lesions (n ¼ 816) the FFR value was .0.80, while 7% of lesions (n ¼ 65) had values in the range of 0.75–0.80. One patient was deferred PCI with an FFR value of
Table 1 Baseline patient characteristics of the study population N 5 721 patients
................................................................................ Age (years)
64.5 + 11.2
Male
409 (57%)
................................................................................ Medical history Diabetes mellitus Hypertension
270 (37%) 598 (83%)
Hyperlipidaemia
584 (81%)
Current/former smoker History of CAD or prior PCI
376 (52%) 493 (68%)
Prior CABG
94 (13%)
Peripheral arterial disease Congestive heart failure
81 (11%) 158 (22%)
Creatinine (mg/dL)
1.2 + 1.2
Multi-vessel CADa
465 (64%)
................................................................................ Diagnosis at FFR assessment Acute myocardial infarction STEMI
84 (12%) 7 (1%)
NSTEMI
77 (11%)
Unstable angina Stable angina
280 (39%) 261 (36%)
Asymptomatic/atypical chest pain
96 (13%)
................................................................................ Procedural characteristics PCI of another lesion
195 (27%)
CAD, coronary artery disease; CABG, coronary artery bypass graft; FFR, fractional flow reserve; NSTEMI, non-ST-elevation myocardial infarction; PCI, percutaneous coronary intervention; STEMI, ST-elevation myocardial infarction. a Multi-vessel CAD was defined as two or more significant lesions (angiographic percent stenosis ≥ 50%) at the time of FFR assessment. Values are shown as absolute numbers (percentages) or mean + SD.
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
The study population comprised 721 patients with 882 coronary lesions deferred PCI based on FFR. The primary outcome of the study was DLI, defined as any PCI or coronary artery bypass graft (CABG) of a lesion deferred revascularization based on the index FFR assessment. Patient and lesion-level covariates known at the time of FFR assessment were considered for the prediction model. With the exception of one missing creatinine level, none of the covariates analysed contained missing data. First, associations between potential predictor covariates and DLI were assessed using univariate Cox proportional hazards models with a marginal Cox model approach to account for correlated data in patients having multiple deferred lesions.15,16 To create a parsimonious prediction model, Spearman correlation coefficients were used to examine the linear relationship between covariate pairs. Tolerance values were examined to identify multi-collinearity. Covariate pairs with strong correlations and/or small tolerance values (,0.10) were excluded. The relationship between candidate variables and DLI was examined using a multivariable Cox proportional hazards model reduced using a stepwise variable selection technique employing a threshold P , 0.20 for entry and P ≥ 0.05 for removal. The reduced model was internally validated for discrimination and calibration using bootstrapping techniques to adjust for over-fitting and over-optimistic model performance. Discrimination was evaluated using a survival-based c statistic to describe the ability of the model to distinguish DLI events and non-events during follow-up.17,18 An optimism-corrected c statistic was calculated using 500 bootstrap samples created with replacement. A Cox proportional hazards model was built for each bootstrap sample using the same stepwise variable selection technique. For each bootstrap model, two c statistics were determined, one for the bootstrap sample and one for the original sample. The difference, bootstrap less original, was averaged over all 500 bootstrap samples to identify the optimism present in the c statistic due to internal validation. Calibration was assessed using the mean calibration slope averaged over all bootstrap samples. For each bootstrap model, the predicted value of DLI among the original sample was used as the independent variable in a Cox regression model to examine agreement with observed DLI events. The final model is presented as an algorithm using the regression coefficients to calculate the probability of DLI within 1-year following deferred PCI based on FFR assessment. To additionally examine model fit, a calibration plot was created and 1-year probability of freedom from DLI estimated from the algorithm was compared with the observed 1-year Kaplan –Meier estimates for freedom from DLI. All statistical analyses were performed using SAS 9.3 (SAS Institute, Cary, NC, USA). All comparisons were two tailed. A P-value of ,0.05 was considered statistically significant. Additional study methods are described in Supplementary material online, Appendix.
J.P. Depta et al.
511
The 1-year risk of deferred lesion intervention
Table 2 Lesion characteristics and outcomes of the study population N 5 882 lesions
................................................................................ Lesion and procedural characteristics Lesion location Left main coronary artery
61 (7%)
Proximal left coronary artery (LM/LAD/LCx) Proximal LAD
297 (34%) 129 (15%)
Any LAD lesion
315 (36%)
Diagonal branch Proximal LCx
72 (8%) 107 (12%)
Any LCx
139 (16%)
Obtuse marginal branch Proximal RCA
75 (9%) 64 (7%) 196 (22%)
Coronary artery bypass graft
24 (3%)
Percent stenosis Myocardial jeopardy index score
59 + 10 5.2 + 2.8
................................................................................
Bifurcation lesion
142 (16%)
Ostial lesion Prior PCI of lesion
188 (21%) 80 (9%)
FFR value
0.87 + 0.05
Deferred lesion intervention Overall
155 (18%)
1 year
47 (5%)
Values are shown as absolute numbers (percentages) or mean + SD. All other abbreviations as shown in Table 1. LAD, left anterior descending; LCx, left circumflex; LM, left main; RCA, right coronary artery.
0.74, and did not have a DLI. At the time of FFR assessment, 27% of patients underwent PCI of another lesion. The discharge medications following deferred revascularization after FFR assessment were similar between patients in the DLI and no DLI groups (Table 3).
Incidence, clinical setting and reasons for performing deferred lesion intervention Following deferral of revascularization based on FFR, 155 lesions (18%) underwent DLI during a mean follow-up period of 4.0 + 2.3 years, of which 115 (74%) were revascularized by PCI and the remaining 40 (26%) by CABG. At 1 year, the rate of DLI was 5.3% (n ¼ 47), and 30% of DLIs occurred within the first year following index FFR assessment. During the study period, death occurred in 119 patients (17%), and 44 patients (6%) died within the 12 months after deferred revascularization based on FFR. A total of 79 acute myocardial infarctions (AMIs) (11%) occurred during follow-up in the study population, and the culprit lesion was a lesion previously deferred revascularization based on the index FFR in 38% of the AMIs (n ¼ 30). The rate of AMI due to a previously deferred lesion within the first year after FFR assessment was 0.8%. AMI accounted for 15% of the DLIs performed in the first year following deferral based on FFR.
Deferred lesion intervention risk model and algorithm The univariate associations between potential predictor covariates and DLI are shown in Table 4. Using a stepwise variable selection technique, the final model (Table 5) included the following prediction variables: age, current or former smoking, history of CAD or prior PCI, creatinine, multi-vessel CAD, and FFR value. The c statistic of the final DLI prediction model was 0.66 (95% CI: 0.61–0.70). The optimismcorrected c statistic for the final DLI prediction model was 0.63. The model discriminates DLI in patients who underwent FFR assessment in the setting of ACS vs. no ACS, but model performance is improved in ACS patients compared with non-ACS patients (c statistic 0.71 vs. 0.59, P ¼ 0.02). An algorithm for the final model was created to calculate the probability of freedom from DLI within 1-year following deferred revascularization based on FFR assessment using the b coefficients of the six independent predictors (Table 5). Model calibration, assessment, and the final algorithm are presented in the Supplementary material online, Appendix (to calculate the 1-year probability of DLI following deferral, see online tool at www.DLI.wustl.edu). In Figure 3, the predicted 1-year DLI risk based on the final algorithm for each deferred lesion in the study is shown. The 1-year predicted DLI risk varied from 1 to 40%. The risk for DLI at 1 year was stratified into three categories, low, intermediate, and high risk, based on quintiles of predicted risk.
Discussion In the current study, we examined the incidence and variability of risk of DLI and developed a clinical prediction model to quantify the risk of DLI within the first year following deferral of revascularization based on FFR assessment. To our knowledge, this is the first study to examine the clinical risk factors associated with DLI following FFR in a real-world patient cohort. The prediction model was based on the clinical predictors of DLI known at the time of initial FFR lesion assessment. The predictors of DLI following deferred revascularization were: younger age, a history of current or former smoking, a history of CAD or prior PCI, creatinine level, multi-vessel CAD, and the FFR value for the lesion. Knowledge of a patient’s risk for future revascularization of a lesion deferred based on FFR may provide clinicians and patients with useful information when
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Any RCA lesion
Of the 155 DLIs (Figure 2; Supplementary material online, Appendix), 101 lesions (65%) underwent urgent revascularization: 30 DLIs (19%) were performed for AMI, of which 6 (4%) were for ST-elevation myocardial infarction, and 24 (15%) were for non-STelevation myocardial infarction. The remaining DLI lesions (46%) that underwent urgent revascularization were for unstable angina (n ¼ 71), where 25% (n ¼ 18 of 71) had new ST-segment and/or T-wave electrocardiogram changes. Elective revascularization comprised 34% of the DLIs (n ¼ 52), predominantly for stable angina (n ¼ 50). For two lesions from one patient, the clinical data were insufficient to definitively classify the reasons that led to DLI. Deferred lesion interventions that occurred within the first year following deferred revascularization were associated with a higher proportion of urgent revascularization compared with DLIs performed beyond the first year after deferral (Supplementary material online, Table S1 and Appendix).
512
J.P. Depta et al.
Table 3 Discharge medications following deferred revascularization based on fractional flow reserve assessment Total (n 5 721 patients)
DLI (n 5 134 patients)
No DLI (n 5 587 patients)
P-value
687 (95%) 362 (50%)
126 (94%) 73 (54%)
561 (96%) 289 (49%)
0.50 0.29
................................................................................ Aspirin Clopidogrel ACE/ARB
465 (64%)
84 (63%)
381 (65%)
0.62
Statin Beta blocker
578 (80%) 553 (77%)
107 (80%) 104 (78%)
471 (80%) 449 (76%)
0.90 0.82
Calcium channel blocker
225 (31%)
47 (35%)
178 (30%)
0.30
Nitrates
290 (40%)
62 (46%)
228 (39%)
0.12
Values are shown as absolute numbers (percentages). ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; DLI, deferred lesion intervention.
considering potential strategies to prevent later adverse events leading to revascularization. The cumulative rate of DLI after deferral based on FFR assessment during a mean follow-up period of 4.0 + 2.3 years was 18%, and the 1-year incidence of DLI was 5.3%. In the randomized DEFER study, the rate of DLI in the group deferred PCI (n ¼ 91) was 8.9% at 5 years,4 and in an observational study of 56 patients deferred PCI
after FFR the 5-year rate of DLI was 10.7%.12 However, those studies only included patients referred for elective PCI. In our study, 50% of patients were deferred revascularization in the setting of ACS (12% with AMI) and represent a population at higher risk for DLI compared with elective patients.4,12 The largest previous real-world experience on long-term clinical outcomes in patients following FFR assessment reported on 721 patients who had revascularization deferred based on the FFR results.5 In that study, the rate of DLI over a median follow-up of 48.7 months was 20.6%, consistent with the rate of DLI observed in our study (A. Lerman, personal communication).5 In prior studies, DLI within the first year after deferring revascularization based on FFR ranged from 2.5 to 11%.2 – 5,7 – 14 Two observational studies of deferred PCI following FFR that included both ACS and non-ACS patients reported 1-year rates of DLI of 10–11% (A. Lerman, personal communication).5,14 The observed rate and the rate predicted by the model-derived algorithm identify rates of 1-year DLI that are consistent with prior studies.2 – 7,9 – 16 In our study population, the rate of subsequent MI was 11% over the mean follow-up period of 4.0 + 2.3 years, and the culprit lesion was a previously deferred lesion based on FFR in 38% of the subsequent MIs. The rate of MI at 1 year among deferred lesions was 0.8%, and by the end of follow-up, 3.8% of lesions (n ¼ 30) deferred revascularization by FFR had a later MI. The rate of MI due to lesions deferred revascularization due to FFR assessment has varied in previous reports. In an observational study of left main lesions deferred revascularization based on FFR, the rate of MI related to a deferred wlesion was 5% at a mean follow-up of 11 months.7 In the DEFER
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Figure 1 Distribution of fractional flow reserve values. The frequencies of fractional flow reserve values for the study population are shown as percentages (bars) and counts (# above each bar) for every 0.01 fractional flow reserve increment.
513
The 1-year risk of deferred lesion intervention
vascular ultrasound; FFR, fractional flow reserve; NSTEMI, non-ST-elevation myocardial infarction; STEMI, ST-elevation myocardial infarction.
study, no patients in the deferred group had an MI at 5-year follow-up; however, that study excluded patients with ACS.4 The 2-year follow-up of the FAME trial and other observational studies have reported a deferred lesion-related MI rate of ,1% among patients who underwent FFR assessment after referral for elective PCI.3,8,12 An observational study of 721 deferred lesions reported a rate of MI of 6% at 7 years, but in that study only 4% of the deferred lesions had index FFR assessment performed in the setting of ACS.5 The potentially higher rate of MI in our study may reflect the higher acuity of patients undergoing FFR assessment, where half of the patients deferred revascularization presented with ACS. Of note, the absence of DLI for over 94% of patients at 1 and 82% at an average of 4 years following deferral by FFR in our higher risk population support the safety of FFR-guided decision making for management of intermediate lesions. Our model determined that five clinical variables and the FFR value were able to predict the risk of DLI. Although no prior studies have specifically assessed the clinical predictors of DLI following deferral based on FFR, prior studies have examined variables among patients with deferred lesions to identify predictors of adverse cardiac events, many of which were DLIs and relevant to our study.7 – 11,14 Our study observed an association between higher FFR values and a lower risk for DLI, consistent with previous analyses that identified lower FFR value as a predictor of adverse outcome.10,11 One study analysing
deferral of revascularization for lesions with FFR values ranging between 0.75 and 0.90 observed that for every 0.01 increase in FFR the rate of adverse cardiac events was lower (hazards ratio: 0.87, 95% CI: 0.80– 0.95).11 A second observational study also showed an inverse relationship between FFR value and adverse cardiac events for deferred lesions.10 These data suggest that the index FFR value, even when above the threshold for haemodynamic severity and prompting deferral of revascularization, retains a negative correlation with the risk of subsequent events. A previous study of lesions deferred PCI based on FFR found that a lower dose of intracoronary adenosine for FFR measurement was significantly associated with higher subsequent adverse cardiac events.7 In distinction, our study did not identify adenosine dose as a predictor of DLI, and the mean dose of intracoronary adenosine used to induce maximal hyperaemia was not different between deferred lesions with and without a subsequent DLI event. Whether lesions deferred revascularization based on FFR assessment benefit from any specialized management strategy is unknown. The goal of our study was to develop a risk prediction model to quantify the risk of DLI within the first year following deferred revascularization after FFR assessment with the potential to assist clinicians in management. The appropriate strategy to prevent DLI among highrisk lesions deferred revascularization after initial FFR assessment is
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Figure 2 Clinical setting and reasons for performing deferred lesion intervention following index fractional flow reserve assessment. IVUS, intra-
514
J.P. Depta et al.
not known. While there is no evidence to support pre-emptive early revascularization for such lesions, we speculate that optimizing a patient’s medical management following deferred PCI may offer the best opportunity to prevent subsequent revascularization for lesions at a higher risk for DLI. In our study, current smoking predicted DLI, suggesting aggressive smoking cessation might reduce the risk of DLI. Secondary prevention strategies have not been prospectively tested in patients with deferred lesions following FFR assessment. Our risk prediction model can identify patients at a
Table 4 Univariate relationship between baseline characteristics and deferred lesion intervention HR (95% CI)
P-value
Age (per 1-year increase) Male
0.98 (0.96– 0.99) 1.22 (0.85– 1.74)
,0.001 0.29
Diabetes mellitus
1.12 (0.78– 1.60)
0.55
Hypertension Hyperlipidaemia
1.23 (0.80– 1.89) 1.45 (0.89– 2.35)
0.35 0.13
Current/former smoker
1.60 (1.12– 2.28)
0.01
History of CAD or prior PCI Prior CABG
1.76 (1.15– 2.69) 1.20 (0.73– 1.97)
0.009 0.47
higher risk for revascularization following deferral based on FFR, and thereby identify those who may warrant closer follow-up and derive even greater benefit from a focused intensification of secondary prevention strategies.
Study limitations The risk prediction model was validated internally by bootstrapping techniques. However, external validation of our risk prediction model using an independent sample has not yet been performed. The study is retrospective, observational, and conducted at a large urban tertiary referral centre. It is unknown if the clinical predictors of DLI identified in our study population are generalizable in different populations at hospitals of different sizes, level of acuity, and location (e.g. rural vs. urban). The risk model was designed to use patient and
................................................................................
1.29 (0.78– 2.12)
0.33
Congestive heart failure Creatinine (per 1 mg/dL increase)
1.26 (0.84– 1.89) 1.17 (1.10– 1.25)
0.27 ,0.001
Acute myocardial infarction
1.40 (0.82– 2.39)
0.21
Multi-vessel CAD PCI of another lesion
1.84 (1.21– 2.79) 1.15 (0.79– 1.66)
0.005 0.48
Proximal left coronary artery (LM/LAD/LCx)
1.33 (0.95– 1.87)
0.10
Proximal right coronary artery
0.77 (0.42– 1.42)
0.40
Percent stenosis Myocardial jeopardy index score
1.01 (0.99– 1.02) 1.03 (0.98– 1.09)
0.45 0.29
Bifurcation lesion
0.81 (0.50– 1.31)
0.38
Ostial lesion Prior PCI of lesion
0.93 (0.60– 1.44) 1.29 (0.79– 2.12)
0.74 0.31
FFR value (per 0.05 unit decrease)
1.26 (1.08– 1.46)
0.004
All abbreviations as shown in Tables 1 and 2. HR, hazards ratio.
Table 5
Figure 3 Variability in the predicted 1-year deferred lesion intervention risk using the final algorithm. The frequency of predicted 1-year deferred lesion intervention risk calculated using the final algorithm for each deferred lesion in the study, ranging from 1 to 40%. Deferred lesion intervention risk at 1 year was stratified into three categories based on quintiles of predicted risk: low (,3% ¼ lowest quintile), intermediate (3 – 7% ¼ 3 middle quintiles), and high risk (.7% ¼ highest quintile).
Multivariable predictors and 1-year b regression coefficients for freedom from DLI in the finala model HR (95% CI)
P-value
b coefficients
............................................................................................................................................................................... Age (per 1-year increase) Current/former smoker
0.98 (0.97– 0.99) 1.49 (1.04– 2.14)
0.005 0.03
20.02075 0.39710
History of CAD or prior PCI
1.62 (1.05– 2.49)
0.03
0.48086
Creatinine (per 1 mg/dL increase) Multi-vessel CAD
1.15 (1.08– 1.22) 1.68 (1.09– 2.58)
,0.001 0.02
0.13681 0.51777
FFR value (per 0.05 unit decrease)
1.21 (1.03– 1.42)
0.02
23.81032
a The model was reduced using a stepwise variable selection technique. For prediction purposes, the 1-year baseline estimate of freedom from DLI for a patient with all covariates set to zero or to the reference group is 0.169. All abbreviations as shown in Tables 1 and 2.
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
Peripheral arterial disease
The 1-year risk of deferred lesion intervention
lesion characteristics known at the time of FFR assessment. However, patient management can vary over time and potentially alter an individual’s risk for DLI. It is also possible that unmeasured confounders, including variability in practice patterns across the providers and operators in the study, may have contributed to the risk for DLI following deferred revascularization based on FFR.
Conclusions
Supplementary material Supplementary material is available at European Heart Journal online. Conflict of interest: J.P.D., J.S.P., E.N., B.F.G., S.K.M., D.R., G.F., Y.P., A.P.A., H.I.K.: no disclosures; A.Z.: Steering committee of Partner 2 trial; J.M.L.: Boston Scientific Advisory Board.; J.S.: consulting: Abbott Vascular, Boston Scientific, Volcano Corp.; speakers’ Bureaus/Honoraria: Medicines Company, Medtronic Vascular, Volcano Corp., St. Jude Corp; R.G.B.: Research Grants: AstraZeneca, Eli Lilly, Bristol-Myers Squibb, Merck, Schering-Plough (no personal compensation). Consultant (Clinical Event Committee Activity only): Roche (Significant) and Pfizer (Modest).
References 1. Pijls NH, De Bruyne B, Peels K, Van Der Voort PH, Bonnier HJ, Bartunek J, Koolen JJ, Koolen JJ. Measurement of fractional flow reserve to assess the functional severity of coronary-artery stenoses. N Engl J Med 1996;334:1703 –1708. 2. Tonino PA, De Bruyne B, Pijls NH, Siebert U, Ikeno F, van’t Veer M, Klauss V, Manoharan G, Engstrom T, Oldroyd KG, Ver Lee PN, MacCarthy PA, Fearon WF. FAME Study Investigators. Fractional flow reserve versus angiography for guiding percutaneous coronary intervention. N Engl J Med 2009;360:213 –224. 3. De Bruyne B, Pijls NH, Kalesan B, Barbato E, Tonino PA, Piroth Z, Jagic N, Mobius-Winkler S, Rioufol G, Witt N, Kala P, MacCarthy P, Engstrom T, Oldroyd KG, Mavromatis K, Manoharan G, Verlee P, Frobert O, Curzen N, Johnson JB, Juni P, Fearon WF. FAME 2 Trial Investigators. Fractional flow reserveguided PCI versus medical therapy in stable coronary disease. N Engl J Med 2012; 367:991 –1001.
4. Pijls NH, van Schaardenburgh P, Manoharan G, Boersma E, Bech JW, van’t Veer M, Bar F, Hoorntje J, Koolen J, Wijns W, de Bruyne B. Percutaneous coronary intervention of functionally nonsignificant stenosis: 5-year follow-up of the DEFER Study. J Am Coll Cardiol 2007;49:2105 –2111. 5. Li J, Elrashidi MY, Flammer AJ, Lennon RJ, Bell MR, Holmes DR, Bresnahan JF, Rihal CS, Lerman LO, Lerman A. Long-term outcomes of fractional flow reserveguided vs. angiography-guided percutaneous coronary intervention in contemporary practice. Eur Heart J 2013;34:1375 –1383. 6. Fearon WF, Bornschein B, Tonino PA, Gothe RM, Bruyne BD, Pijls NH, Siebert U. Fractional Flow Reserve Versus Angiography for Multivessel Evaluation Study I. Economic evaluation of fractional flow reserve-guided percutaneous coronary intervention in patients with multivessel disease. Circulation 2010;122:2545 –2550. 7. Courtis J, Rodes-Cabau J, Larose E, Potvin JM, Dery JP, Larochelliere RD, Cote M, Cousterousse O, Nguyen CM, Proulx G, Rinfret S, Bertrand OF. Usefulness of coronary fractional flow reserve measurements in guiding clinical decisions in intermediate or equivocal left main coronary stenoses. Am J Cardiol 2009;103: 943 –949. 8. Courtis J, Rodes-Cabau J, Larose E, Dery JP, Nguyen CM, Proulx G, Gleeton O, Roy L, Barbeau G, Noel B, DeLarochelliere R, Bertrand OF. Comparison of medical treatment and coronary revascularization in patients with moderate coronary lesions and borderline fractional flow reserve measurements. Catheter Cardiovasc Interv 2008;71: 541 –548. 9. Hamilos M, Muller O, Cuisset T, Ntalianis A, Chlouverakis G, Sarno G, Nelis O, Bartunek J, Vanderheyden M, Wyffels E, Barbato E, Heyndrickx GR, Wijns W, De Bruyne B. Long-term clinical outcome after fractional flow reserve-guided treatment in patients with angiographically equivocal left main coronary artery stenosis. Circulation 2009;120:1505 –1512. 10. Meuwissen M, Chamuleau SA, Siebes M, de Winter RJ, Koch KT, Dijksman LM, van den Berg AJ, Tijssen JG, Spaan JA, Piek JJ. The prognostic value of combined intracoronary pressure and blood flow velocity measurements after deferral of percutaneous coronary intervention. Catheter Cardiovasc Interv 2008;71:291 –297. 11. Lavi S, Rihal CS, Yang EH, Fassa AA, Elesber A, Lennon RJ, Mathew V, David HR Jr, Lerman A. The effect of drug eluting stents on cardiovascular events in patients with intermediate lesions and borderline fractional flow reserve. Catheter Cardiovasc Interv 2007;70:525 –531. 12. Rieber J, Jung P, Koenig A, Schiele T, Shapiro M, Hoffmann U, Klauss V. Five-year follow-up in patients after therapy stratification based on intracoronary pressure measurement. Am Heart J 2007;153:403–409. 13. Costa MA, Sabate M, Staico R, Alfonso F, Seixas AC, Albertal M, Crossman A, Angiolillo DJ, Zenni M, Sousa JE, Macaya C, Bass TA. Anatomical and physiologic assessments in patients with small coronary artery disease: final results of the Physiologic and Anatomical Evaluation Prior to and After Stent Implantation in Small Coronary Vessels (PHANTOM) trial. Am Heart J 2007;153:296.e1 –e7. 14. Fischer JJ, Wang XQ, Samady H, Sarembock IJ, Powers ER, Gimple LW, Ragosta M. Outcome of patients with acute coronary syndromes and moderate coronary lesions undergoing deferral of revascularization based on fractional flow reserve assessment. Catheter Cardiovasc Interv 2006;68:544 –548. 15. Lee EW, Wei LJ, Amato DA, Leurgans S. Cox-type regression analysis for large numbers of small groups of correlated failure time observations. In: Klein JP, Goel PK (eds), Survival Analysis: State of the Art. Dordrecht, The Netherlands: Kluwer Academic Publishers; 1992, p237 –247. 16. Lin DY, Wei LJ. The robust inference for the proportional hazards model. J Am Stat Assoc 1989;84:1074 – 1078. 17. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361 –387. 18. Pencina MJ, D’Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;23:2109 – 2123.
Downloaded from http://eurheartj.oxfordjournals.org/ by guest on June 6, 2016
The risk of DLI for coronary lesions deferred revascularization based on FFR was 5.3% within 1 year and 18% within 4 years. We derived a risk prediction model and algorithm that uses five clinical variables known at the time of FFR assessment and the FFR value to predict the risk of DLI within 1 year following deferred revascularization based on FFR. While requiring further study, use of this model for identification of patients at higher predicted risk of DLI may alert clinicians to a need for closer follow-up, optimized medical therapy and intensified risk factor modification.
515
