mopdf.com

ACCEPTED MANUSCRIPTEPTEDMANUSCRIPTAbstractBACKGROUND In time-to-first-event analyses, icosapent ethyl significantly reduced the riskof ischemic events, including cardiovascular death, among patients with elevated triglyceridesreceiving statins. These patients are at risk for not only first but also subsequent ischemic events.OBJECTIVES Pre-specified analyses determined the extent to which icosapent ethyl reducedtotal ischemic events.METHODS The Reduction of Cardiovascular Events with EPA-Intervention Trial (REDUCEIT) randomized 8,179 statin-treated patients with triglycerides 135 and 500 mg/dL (medianbaseline of 216 mg/dL) and LDL-cholesterol 40 and 100 mg/dL (median baseline of 75mg/dL), and a history of atherosclerosis (71% patients) or diabetes (29% patients) to icosapentethyl 4g/day or placebo. The main outcomes were total (first and subsequent) primary compositeendpoint events (cardiovascular death, nonfatal myocardial infarction, nonfatal stroke, coronaryrevascularization, or hospitalization for unstable angina) and total key secondary compositeendpoint events (cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke). As apre-specified statistical method, we determined differences in total events using negativebinomial regression. We also determined differences in total events using other statisticalmodels, including Andersen-Gill, Wei-Lin-Weissfeld (Li and Lagakos modification), both prespecified, and a post hoc joint-frailty analysis.RESULTS In 8,179 patients, followed for a median of 4.9 years, 1,606 (55.2%) first primaryendpoint events and 1,303 (44.8%) subsequent primary endpoint events occurred (whichincluded 762 second events, and 541 third or more events). Overall, icosapent ethyl reduced totalprimary endpoint events (61 versus 89 per 1000 patient years for icosapent ethyl versus placebo,respectively; RR 0.70, 95% CI 0.62-0.78, P 0.0001). Icosapent ethyl also reduced eachcomponent of the primary composite endpoint, as well as the total key secondary endpoint events(32 versus 44 per 1000 patient years for icosapent ethyl versus placebo, respectively, RR 0.72,95% CI 0.63-0.82, P 0.0001).CONCLUSIONS Among statin-treated patients with elevated triglycerides and cardiovasculardisease or diabetes, multiple statistical models demonstrate that icosapent ethyl substantiallyreduces the burden of first, subsequent, and total ischemic events.TRIAL REGISTRATION clinicaltrials.gov identifier: NCT01492361Keywords: Icosapent ethyl, eicosapentaenoic acidACCCondensed Abstract: The results of analyses by multiple statistical models presented here forREDUCE-IT (median follow-up of 4.9 years) demonstrate that icosapent ethyl 4 grams dailysignificantly reduced the rate of total primary endpoint events (RR 0.70, 95% CI 0.62-0.78,P 0.0001), each primary endpoint component, including cardiovascular death, and total keysecondary endpoint events in statin-treated patients with elevated triglycerides and establishedcardiovascular disease or diabetes at risk for not only first but also subsequent ischemic events.AbbreviationsCEC Clinical Endpoint CommitteeCI confidence intervalCRP C-reactive proteinEPA eicosapentaenoic acidHR hazard ratio4

ACCEPTED MANUSCRIPTACCEPTEDMANUSCRIPTLDL low density lipoproteinMI myocardial infarctionREDUCE-IT Reduction of Cardiovascular Events with EPA - Intervention TrialTG Triglyceride5

ACCEPTED MANUSCRIPTDespite the tremendous advance of statin therapy in secondary and primary prevention,ischemic events continue to occur in patients with cardiovascular risk factors such as elevatedtriglycerides, atherosclerosis, or diabetes (1-4). In addition to their initial events, such patientsRIPTare at substantial risk for recurrent, potentially fatal events. Assessment of these recurrent eventsprovides a perspective on the total atherosclerotic event burden these patients face (5-11). From apatient’s perspective (and also for physicians and payors), it is not only first events that areSCimportant, but subsequent events as well.One marker of this residual cardiovascular risk that predisposes patients to initial andMANUrecurrent ischemic events is elevated triglyceride levels (12,13). Multiple epidemiologic andgenetic analyses have demonstrated an independent association with increased cardiovascularrisk (14). Among several properties, icosapent ethyl reduces triglyceride levels and other lipidsand lipoproteins without increasing LDL-cholesterol when compared with placebo and has alsoTEDbeen reported to have anti-inflammatory and plaque stabilizing properties, as well as stabilizingeffects on cell membranes (15-19). Recently, icosapent ethyl has been demonstrated to reducethe first occurrence of the primary composite endpoint of cardiovascular death, nonfatalEPmyocardial infarction, nonfatal stroke, coronary revascularization, or hospitalization for unstableangina in the Reduction of Cardiovascular Events with Icosapent Ethyl–Intervention TrialACC(REDUCE-IT), with a 25% relative risk reduction and a 4.8% absolute risk reduction (numberneeded to treat [NNT] of 21) (20). The time to first occurrence of the key secondary compositeendpoint of cardiovascular death, nonfatal myocardial infarction, and nonfatal stroke was alsoreduced with icosapent ethyl, with a 26% relative risk reduction and a 3.6% absolute riskreduction (NNTof 28). The results were also consistent across each of the primary and key6

ACCEPTED MANUSCRIPTsecondary endpoint components and appear to be applicable to a substantial proportion ofpatients in clinical practice (21).We sought to determine the impact of icosapent ethyl on total ischemic events (first andRIPTsubsequent events) to characterize better the totality of the ischemic event burden across theoverall study population.MethodsSCStudy design and participantsThe details of the REDUCE-IT design have been previously published (22). Briefly, patientsMANUwere randomized in a double-blind manner to icosapent ethyl 4 g/day (2 grams twice daily withmeals) or placebo (Online Figure 1, Online Figure 2). Approximately 1,612 events wereprojected necessary for 90% power to detect a 15% relative risk reduction after accounting fortwo protocol pre-specifed interim analyses (final two-sided alpha level 0.0437). This resultedTEDin a target patient population of approximately 7,990 patients. Among all randomized patients,70.7% were enrolled on the basis of secondary prevention and 29.3% for primary prevention.Patients were randomized to one of two treatment arms on a 1:1 ratio using a computer-EPgenerated randomization schema. Study medication and placebo capsules were similar in sizeand appearance to maintain blinding. Randomization was stratified according to cardiovascularACCrisk cohort (secondary or primary prevention), use of ezetimibe (yes/no), and by geographicalregion (Westernized, Eastern European, and Asia Pacific countries). There were 473 sites in 11countries randomizing patients from 2011 to 2016. The protocol was submitted to and approvedby appropriate health authorities, ethics committees, and institutional review boards. Trialcompletion occurred after achieving the approximate number of pre-specified necessary events.7

ACCEPTED MANUSCRIPTTo be eligible, patients were required to be either 45 years of age with establishedcardiovascular disease (secondary prevention stratum) or 50 years old with type 2 or type 1cardiovascular risk factor (primary prevention stratum) (21,22).RIPTdiabetes mellitus requiring treatment with medication, and to have at least one additionalPatients had fasting triglycerides of 135 mg/dL and 500 mg/dL and LDL-cholesterol 40 mg/dL and 100 mg/dL. The initial version of the protocol permitted a 10% variance in theSClower qualifying triglyceride levels of 150 mg/dL, therefore patients with triglycerides 135mg/dL were randomized. After approximately 60% of the patients were enrolled, an amendmentMANUincreased the lower limit of permissible triglyceride levels to 200 mg/dL with no variabilityallowance. The study included 841 (10.3%) patients with baseline triglyceride levels 150mg/dL. Patients were required to be on stable statin therapy for four weeks with wellcontrolled LDL-C to investigate the potential benefit of icosapent ethyl 4g/day beyond theTEDcurrent standard of care. Additional inclusion and exclusion criteria published previously (22)are provided in the online appendix.After randomization, follow-up visits continued at 4 months, 12 months, and annuallyEPthereafter in this event-driven trial until approximately 1,612 primary efficacy endpoint eventsoccurred, after which patients made a final end-of-study visit.ACCThe original projected annual primary endpoint event rate for the REDUCE-IT placebogroup was 5.9%; this was derived prior to study initiation (and therefore prior to the two interimanalyses conducted by the data monitoring committee) and was based on data available fromcardiovascular outcome trials with similar high-risk statin-treated patients and reported endpointcomponents similar to the primary endpoint in REDUCE-IT (23-29). The observed annualizedprimary endpoint event rate for placebo patients in REDUCE-IT was 5.74%, which holds8

ACCEPTED MANUSCRIPTconsistent with cardiovascular outcome studies, including those published since the design ofREDUCE-IT, with comparable patient populations and expanded or hard major adversecardiovascular events (MACE) (4,8,9,30-44).RIPTFor the present pre-specified analysis, the primary outcome was the total of first plussubsequent ischemic events consisting of the composite of cardiovascular death, nonfatalmyocardial infarction, nonfatal stroke, coronary revascularization, or hospitalization for unstableSCangina. Protocol Amendment 2 (July 2016) designated the composite of hard MACE(cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke) as the “key secondaryMANUendpoint” per suggestions from the Food and Drug Administration and with REDUCE-ITSteering Committee concordance. Exploratory analyses of the total of first and subsequent eventswere also performed for the key secondary composite endpoint.Baseline characteristics were compared between treatment groups using the chi-squaredTEDtest for categorical variables and the Wilcoxon rank sum test for continuous variables. Theanalysis of total cardiovascular events was pre-specified in the study protocol. There are severalmethods for analyzing first and subsequent (recurrent) event data. As a pre-specified statisticalEPmethod, we used the negative binomial regression model to calculate rates and rate ratios fortotal cardiovascular events, which accounts for the variability in each patient’s risk of events (45-ACC47). As pre-specified supportive analyses, we used the modified Wei-Lin-Weissfeld method (Liand Lagakos modification) to calculate hazard ratios for the time to the first event, second event,or third event (48-49). An additional pre-specified analysis, the Andersen-Gill model using aCox proportional-hazard with the counting-process formulation was performed to model the totalevents (50,51). In addition, to account for informative censoring due to cardiovascular death, wecalculated the hazard ratio for total nonfatal events using a joint frailty model (52). The joint9

ACCEPTED MANUSCRIPTfrailty model simultaneously estimates hazard functions for nonfatal and fatal cardiovascularevents and takes into account the fact that patients who are prone to have nonfatal events have anelevated risk of a cardiovascular death. Our application of the joint frailty model used a gammaRIPTdistribution for the frailty term.To improve the performance and validity of our statistical models, a bundling approachwas employed, whereby nonfatal events occurring on the same day as a cardiovascular deathSCwere excluded, and at most, one nonfatal event was counted on any given day (e.g., for coronaryrevascularization occurring after an MI which eventually resulted in the patient’s death, only theMANUdeath would be included). Statistical analyses using the full adjudicated endpoint events datasetwithout exclusions for this bundling approach are also included in the online supplementarymaterials.All efficacy analyses were conducted in accordance with the intention-to-treat principle.TEDAll tests were based on a 2-sided nominal significance level of 5% with no adjustments formultiple comparisons, consistent with pre-specified plans for such endpoints. All statisticalanalyses were conducted using SAS version 9.4 software (Cary, North Carolina). All analyses ofand Pocock.ACCResultsEPfirst, subsequent, and total events were independently generated and validated by Drs. GregsonA total of 8,179 patients were randomized and followed for a median of 4.9 years. Thebaseline characteristics were well matched across the icosapent ethyl and placebo groups (OnlineTable 1). At baseline, median triglyceride levels were 216 mg/dL, with median LDL-C levels of75 mg/dL. Additional baseline characteristics across treatment groups and for patients with no10

ACCEPTED MANUSCRIPTevents, a single event, and multiple subsequent events are shown in Online Tables 1 and 2,respectively.Total events for the primary efficacy endpointRIPTAcross 8,179 randomized patients, there were 1,606 (55.2%) first primary endpointevents and 1,303 (44.8%) additional primary endpoint events, for a total of 2,909 endpoint events(Table 1, Online Figures 3, 4, and 5). The proportions of first and subsequent primary endpointSCevents, overall and by component type, are depicted in Figure 1. There were 762 second events,272 third events, and 269 fourth or more events. Overall, total (first and subsequent) primaryMANUendpoint event rates were reduced to 61 from 89 per 1000 patient years for icosapent ethyl versusplacebo, respectively, rate ratio (RR) 0.70, 95% CI 0.62-0.78, P 0.0001 (Central Illustration,Figure 2a). Using the Wei-Lin-Weissfeld model, the first occurrence of a primary compositeendpoint was reduced with icosapent ethyl versus placebo (HR 0.75, 95% CI 0.68-0.83, PTED 0.0001) as was the second occurrence (hazard ratio [HR] 0.68, 95% CI 0.60-0.78, P 0.0001).There was a 30% relative risk reduction in the total (first and subsequent) ischemic events for theprimary composite endpoint with icosapent ethyl. First events were reduced by 25%, secondEPevents by 32%, third events by 31%, and fourth or more events by 48%. The cumulative eventsover time are shown in Figure 2. Total key secondary endpoint event rates were significantlyACCreduced to 32 from 44 per 1000 patient years for icosapent ethyl versus placebo, respectively(RR 0.72, 95% CI 0.63-0.82, P 0.0001) (Figure 2b). The times to first occurrence, secondoccurrence, third occurrence, or fourth occurrence of the primary composite endpoint wereconsistently reduced (Figure 3) with icosapent ethyl. There were similar results for the modelsirrespective of whether bundling and/or single event accounting was employed (Online Tables 3,11

ACCEPTED MANUSCRIPT4, and 5). Total events for each component of the primary endpoint were also significantlyreduced (Figure 4, Online Figure 3).The risk differences for every 1000 patients treated for five years with icosapent ethyl forRIPTthe five components of the composite primary endpoint are shown in Figure 5; approximately159 total primary endpoint events could be prevented within that timeframe: 12 cardiovasculardeaths, 42 myocardial infarctions, 14 strokes, 76 coronary revascularizations, and 16 episodes ofSChospitalization for unstable angina.We explored study drug adherence in patients with recurrent events. At the time of a firstMANUprimary endpoint event (fatal or nonfatal), 81.3% (573/705) of icosapent ethyl and 81.8%(737/901) of placebo patients with a first primary endpoint event were receiving randomizedstudy drug. At the time of subsequent primary endpoint events (fatal or nonfatal), 79.7%(188/236) and 79.5% (299/376) of patients with a second event, 68.1% (49/72) and 74.1%TED(106/143) of patients with a third event, and 68.0% (17/25) and 71.6% (48/67) of patients with afourth event were receiving randomized study drug in the icosapent ethyl and placebo groups,respectively. Therefore, the majority of the first, second, third, and fourth events occurred whileEPpatients were on randomized study treatment. Numerical differences in study drug adherenceamong patients with recurrent events were not statistically significant between treatment groups.ACCDiscussionWe found large and significant reductions in total ischemic events with icosapent ethylversus placebo in these total event analyses of REDUCE-IT. Three pre-specified and one posthoc analyses with various statistical methodologies demonstrated consistent effects on totalischemic events, with substantial relative and absolute risk reductions. There was a 30% relativerisk reduction in the total (first and subsequent) ischemic events for the primary composite12

ACCEPTED MANUSCRIPTendpoint with icosapent ethyl. For every 1,000 patients treated with icosapent ethyl for fiveyears, approximately 159 total primary endpoint events could be prevented. Total events for thehard MACE key secondary endpoint also demonstrated large and clinically meaningfulRIPTreductions, which further corroborated the significant reduction in important ischemic eventsseen with the primary endpoint.There were significant reductions in the first, subsequent, and total ischemic events forSCeach individual component of the composite primary endpoint. This benefit of icosapent ethylacross a variety of different ischemic endpoints (e.g., coronary, cerebral, fatal and nonfatalMANUevents, and revascularizations) indicates that the drug benefit is not likely to be explained bytriglyceride lowering alone and suggests strongly that there are multiple mechanisms of action ofthe drug beyond triglyceride lowering that may work together to achieve the observed benefits.Preclinical mechanistic investigations and smaller clinical studies support this contentionTED(12,18,19,53-57).Icosapent ethyl was well tolerated with no significant differences in rates of seriousadverse events versus placebo (20). Although overall rates were low in both treatment groups,EPand none of the events were fatal, with icosapent ethyl there was a trend towards increasedserious bleeding albeit with no significant increases in adjudicated hemorrhagic stroke, seriousACCcentral nervous system bleeding, or gastrointestinal bleeding. There was a small but statisticallysignificant increase in hospitalization for atrial fibrillation or flutter endpoints noted inREDUCE-IT (20). Nevertheless, the large number of important ischemic events averted with thedrug, including a significant reduction in fatal and nonfatal stroke (28%), cardiac arrest (48%),sudden death (31%), and cardiovascular death (20%), is indicative of a very favorable riskbenefit profile (20).13

ACCEPTED MANUSCRIPTStudy drug adherence in patients with recurrent events was strong in both treatmentgroups at the time of their first primary endpoint event, decreasing somewhat across bothtreatment groups from the occurrence of the first to the fourth event. For example, at the time ofRIPTa first occurrence of a fatal or nonfatal primary endpoint event, 81.3% of icosapent ethyl and81.8% of placebo patients with a first primary endpoint event were on study drug; these ratesdecreased to 68.0% and 71.6% for patients with a fourth primary endpoint event.SCThe REDUCE-IT primary study results (20) and the recurrent and total endpoint eventfindings discussed herein stand in stark contrast to cardiovascular outcome studies with otherMANUagents that lower triglyceride levels and with low-dose omega-3 fatty acid mixtures, wherecardiovascular outcome benefit has not been consistently observed in statin-treated patients (13).However, the REDUCE-IT results are aligned with the JELIS study results (17). The distinctionof the cardiovascular benefits observed in REDUCE-IT and JELIS from the lack ofTEDcardiovascular benefits observed in statin-treated populations with add-on omega-3 fatty acidmixtures is likely due specifically to the high EPA levels. EPA has unique lipid and lipoprotein,anti-inflammatory, anti-platelet, anti-thrombotic, and cellular modifying effects, all of whichEPmay contribute to benefits in atherosclerotic processes such as reduced development, slowedprogression, and increased stabilization of atherosclerotic plaque (19, 54-56). The aggregateACCcontribution of thes

(32 versus 44 per 1000 patient years for icosapent ethyl versus placebo, respectively, RR 0.72, 95% CI 0.63-0.82, P 0.0001). CONCLUSIONS Among statin-treated patients with elevated triglycerides and cardiovascular disease or diabetes, multiple statisti