PCSK9 inhibitor reduces first and subsequent nonfatal CV events and death after ACS
Alirocumab Reduces Total Nonfatal Cardiovascular and Fatal Events - The ODYSSEY OUTCOMES Trial
Literature - Szarek M, White HD, Schwartz GG et al., - J Am Coll Cardiol. 2019; 73(4) DOI: 10.1016/j.jacc.2018.10.039Introduction and methods
CV outcome trials generally assess the effect of an intervention on delaying the time to first occurrence of an event. Analysis of only the first event may not capture the whole clinical impact of an intervention or the burden of disease for the patient.
In trials with patients with CVD, the majority of patients are censored because they survive the follow-up period. Another reason for censoring, when evaluating the effect on nonfatal endpoints, is death. In statistical methods used to analyze total events, it is typically assumed that censoring relates to non-informative data. This makes sense if the risks of nonfatal events and death are unrelated to one another, and the censoring due to death is similarly non-informative to censoring due to completion of follow-up. However, If the risk of nonfatal events is positively associated with the risk of death, occurrence of death may violate this assumption. This may yield erroneous estimates of nonfatal event risk.
The ODYSSEY OUTCOMES trial demonstrated that the PCSK9 inhibitor alirocumab reduced the first occurrence of the primary composite outcome of MACE (15%) and was associated with fewer events of the secondary endpoint of all-cause death in a post-ACS population, as compared with placebo, in addition to maximum tolerated statin therapy [1].
The current pre-specified analysis used a novel approach to examine total events in ODYSSEY OUTCOMES (n=18.924), to jointly model total nonfatal CV and fatal events, while allowing for the possibility that patients may experience multiple related nonfatal events. The method formally quantifies the association between nonfatal events and death, and it accounts for competing deaths that prevent follow-up for nonfatal events.
The primary analysis included all-cause death and total nonfatal CV events (MI, ischemic stroke, unstable angina requiring hospitalization, hemorrhagic stroke, HF requiring hospitalization and ischemia-driven coronary revascularization). A joint semiparametric model (also known as: frailty model) was applied, which allows for multiple nonfatal CV events within an individual, while simultaneously assessing and adjusting for possible informative censoring of the nonfatal event process by death. The model gives separate hazard functions for nonfatal and fatal events, linked by a shared frailty [2]. It has been demonstrated to provide accurate relative estimates of nonfatal and fatal event risk if patients at greater risk of nonfatal events are also at increased risk for death [3].
Association parameters are calculated to quantify the relationship between the nonfatal and fatal events. A value equal to 0 indicates that death is noninformative for nonfatal events, whereas a value >0 indicates that patients at greater risk of nonfatal events are also at greater risk for death. When the parameter is >1, death is informative for the nonfatal CV event rate. This implies that patients at the highest risk of death were also at elevated risk for nonfatal events, but their death removes them from the risk set for nonfatal events.
All analyses were intention-to-treat. Patients were followed for survival for a median of 2.8 years (IQR: 2.3-3.4). Patients (all but four patients in the alirocumab and 3 in the placebo group) continued randomized treatment after their first nonfatal CV events.
Main results
- 5425 Total deaths or nonfatal CV events were recorded, of which 77% were not first events (n=3064). 1260 patients experienced >1 event.
- Death occurred as a first event in 2.2% and 2.5% in the alirocumab and placebo groups, respectively. After a first nonfatal CV event, death as a second event occurred in 5.7% and 5.0% of patients, respectively. Death as a third event occurred in 6.2% and 6.6% of patients treated with alirocumab and placebo, respectively.
- The joint model confirms the qualitative observation that each successive prior nonfatal CV event is associated with an increased subsequent risk for death. An association parameter of 2.04 (95%CI: 1.78-2.29) links the risks of nonfatal CV events and death. When limiting the analysis to the association between only MI, stroke or unstable angina requiring hospitalization and death, the association parameter was 3.29 (95%CI: 2.86-3.72).
- In the alirocumab group, 385 fewer total events were seen (2905 with placebo, 2520 with alirocumab); 190 fewer first events (1627 and 1437, respectively) and 195 fewer additional events in patients who had suffered a first nonfatal CV event.
- When normalized to duration of follow-up, 7.2 first events and 14.6 total events were avoided with alirocumab treatment per 1000 patient-years of assigned treatment.
- Treatment with alirocumab was associated with a reduction of total nonfatal events (HR: 0.87, 95%CI: 0.82-0.93) and death (HR: 0.83, 95%CI: 0.71-0.97).
- Alirocumab was associated with 255 fewer total fatal and nonfatal events among 5629 patients with LDL-c ≥100 mg/dL, as compared with placebo. Among 13.295 patients with LDL-c <100 mg/dL, 130 fewer events were seen with alirocumab vs. placebo.
Conclusion
This pre-specified total events analysis of ODYSSEY OUTCOMES shows a greater treatment effect when considering total nonfatal CV events and death, as compared with the original analysis that considered only first events. The data suggested that having a nonfatal event was associated with an increased risk of subsequent death. This analysis therefore illustrates the high burden of ongoing disease in the study population and how alirocumab can reduce that burden.
Editorial comment
Genest and Nguyen [4] call the total events analysis ‘atypical, and perhaps controversial’, but also ‘in the present case, very informative’. The analysis takes into account not only the fact that patients may have multiple nonfatal events, but also that death may censor nonfatal outcomes. Since a positive association was found between nonfatal events and death, Szarek et al. concluded that there was indeed some censoring of nonfatal events by death. They acknowledged the limitation that there may have been baseline characteristics that partly drove the association, but were not included in the model.
Genest and Nguyen point out the importance of acknowledging that not all CV events are alike. An attempt to assess the impact of disease for a patient, researchers [5] have assigned a weight to events on a scale of 1 to 10, with death getting a value of 10, based on patient’s perception of the severity of an event. That type of analysis would be interesting on the ODYSSEY data. Szarek et al. did, however, perform an analysis that excluded ischemia-driven coronary revascularization and hospitalization for congestive HF, and found that these events had minimal impact on the results.
Overall, Genest and Nguyen are positive about the analysis and think that it should be included in economic analysis. They conclude that the study ‘reinforces the role of PCSK9 inhibitors in the treatment of patients post–acute coronary syndrome who have high LDL-C despite maximal statin therapy.’
References
1. Schwartz GG, Steg PG, Szarek M, et al., for the ODYSSEY OUTCOMES Committees and Investigators. Alirocumab and cardiovascular outcomes after acute coronary syndrome. N Engl J Med 2018;379:2097–107.
2. Liu L, Wolfe RA, Huang X. Shared frailty models for recurrent events and a terminal event. Biometrics 2004;60:747–56.
3. Rondeau V, Mathoulin-Pelissier S, Jacqmin-Gadda H et al. Joint frailty models for recurring events and death using maximum penalized likelihood estimation: application on cancer events. Biostatistics 2007;8:708–21
4. Genest J & Nguyen D. The Cat Has 9 Lives, Until it Dies. JACC 2019; 73(4). DOI: 10.1016/j.jacc.2018.11.025
5. Stafinski T, Menon D, Nardelli A, et al. Incorporating patient preferences into clinical trial design: results of the opinions of patients on treatment implications of new studies (OPTIONS) project. Am Heart J 2015;169:122–31.e22.