We established a state-transition Markov model to assess the economic value of alirocumab within established MI population (Figure 1). After a discharge of MI, patients entered the model as an initial event-free state, and the entire cohort was redistributed across states in every 1-year cycle. The model included mutually exclusive health states (non-fatal MI, non-fatal ischemic stroke [IS] and death). The major side effect (local injection-site reaction) was incorporated into the model. The time frame was 25 years in basic analysis to take the majority of Chinese old patients into consideration. Half-cycle correction was used for all events in every year.

Alirocumab plus statin therapy (high-intensity or maximally tolerated) vs. statin monotherapy were projected through accrued simulation. The incremental cost-effectiveness ratio (ICER), a primary outcome, was calculated by incremental cost per quality-adjusted life-years (QALYs) gained. We assumed the cost-effectiveness threshold as 212,676 Chinese Yuan (CNY, 1 CNY = 0.1450 USD in 2019 (National Bureau of Statistics of China, 2019a)), which was three times of per capita gross domestic product (GDP) in 2019 (National Bureau of Statistics of China, 2019a), based on the China Guidelines for Pharmacoeconomic Evaluations (Wu et al., 2015).

A representative real-world cohort with established MI was modeled to reflect the Chinese high-risk CVD population. Patients were selected from the China PEACE Prospective AMI study (China Patient-centered Evaluative Assessment of Cardiac Events Prospective Study of Acute Myocardial Infarction) (Dreyer et al., 2019), the first large clinical outcomes study in MI population in China. A total of 3,415 patients with AMI were recruited from 53 hospitals located in 21 of 31 provinces. During the one-year follow-up period, main clinical outcomes contained non-fatal MI (1.7%), non-fatal IS (0.9%), cardiovascular death (2.2%) and all-cause death (3.1%). More detailed characteristics of this target population are showed in Supplementary Table S1.

The effect of alirocumab has been much-discussed in previous research literatures. We aimed to use two approaches to incorporate the efficacy of alirocumab in this health economic model (Table 1). One was applying the hazard ratios (HRs) observed in ODYSSEY OUTCOMES (Schwartz et al., 2018) to all population directly, taking no account of LDL-C levels, which was preferred assumption in evidence-based intervention and most economic evaluations. To model the effect of alirocumab on non-fatal IS, we used outcome data from ODYSSEY OUTCOMES for an end point defined as “fatal or non-fatal IS.” We assumed that the reduction of all-cause mortality was mediated through the reduction in the cardiovascular death risk. The other was integrating more specific lipid-lowering efficacy of the PCSK9 inhibitor in Chinese population observed in ODYSSEY EAST study (Han et al., 2020) with its estimated effect on major adverse cardiovascular events (MACEs) based on a Cholesterol Treatment Trialists’ Collaboration (CTTC) meta-analysis of statins trials (Baigent et al., 2010), which was also suitable for other medications such as ezetimibe and PCSK9 inhibitors.

We assumed that the initial conversion probabilities above were applied to the first year and increased in every succedent 5 years in line with the growth rates of real-time natural mortalities in all age groups in China Health Statistics Yearbook (National Health Commission of the People’s Republic of China, 2019). All 25-years event rates in the basic analysis are listed in Supplementary Table S2.

The setting of this economic evaluation was based on Chinese healthcare sector. The direct costs associated with MACEs and subsequent chronic treatment, including initial event-free state, were obtained from previous literatures (Wang et al., 2015; Yin et al., 2018; Fu et al., 2020) and adjusted for inflation to 2019 (National Bureau of Statistics of China, 2019b) (Table 1).

Currently, public and private medical insurances do not cover PCSK9 inhibitors. Following the usual therapeutic dose and frequency (75 mg/2 weeks), the annual official price of alirocumab was estimated to be 51,532 CNY per person. As the PCSK9 inhibitor manufacturers had offered an average rebate of 33% since April 2020, main base case analysis was performed for discounted net price of alirocumab in 34,355 CNY annually. We also calculated the average bidding price of ezetimibe (10 mg), equally 2827 CNY per year, for scenario analyses.

Quality-of-life (QOL) estimates for health states were derived from a time tradeoff study (Matza et al., 2015). The discrepancy of mild injection-site adverse reaction (3.8% in alirocumab vs. 2.1% in placebo) was taken into account with a small penalty in QOL (Khazeni et al., 2009) and there was no increase in costs or therapy discontinuation. Future costs and QOL were discounted at 5.0% per year (Wu et al., 2015).

In one-way sensitivity analyses, we varied single input through plausible range (e.g., 95% confidence interval [CI]) (Table 1), while holding others at their base case values to search for key determinants to the ICERs. In probabilistic sensitivity analyses, multiple input parameters were randomly selected according to their pre-specified statistical distributions and generated 10,000 individual outputs surrounding the mean point estimate. The percentages of simulations falling below various willingness-to-pay (WTP) thresholds were drawn as cost-effectiveness acceptability curves.

Based on the ODYSSEY OUTCOMES trial, we made different hypothesis about effect of alirocumab. The efficacy of alirocumab on all-cause death (HR: 0.85; 95% CI: 0.73–0.98) was applied instead of the efficacy of alirocumab just on cardiovascular death in the base case. We modeled ezetimibe vs. alirocumab based on statin therapy. The efficacy of ezetimibe was modeled according to the IMPROVE-IT trial (Cannon et al., 2015) and the decrease of LDL-C (Baigent et al., 2010).

A range of scenario analyses were performed in high-risk subgroups to identify preferred candidates for the PCSK9 inhibitor. Women in China had higher all-cause/cardiovascular mortalities vs. men at 1-year post-MI (Dreyer et al., 2019), so we explored the value of alirocumab in female population. We involved poorly controlled FH patients after MI and assumed that the risk of MACEs was 2.3 times than those without FH (Wang et al., 2019). Meanwhile, a maximal reduction in risk of MACEs was modeled by the absolute change of LDL-C (Baigent et al., 2010). We also focused on patients with polyvascular disease in 3 vascular beds (MI with both peripheral artery disease and cerebrovascular disease) (Subherwal et al., 2012). In the ODYSSEY OUTCOMES research, a more pronounced absolute risk reduction was observed in patients with polyvascular disease than monovascular disease (Jukema et al., 2019). As the traditional risk factors for adverse events, cost-effectiveness of alirocumab were calculated in pre-existing diabetes mellitus (DM) (Li et al., 2016; Zhuo et al., 2019) and hypertension population (Chen et al., 2009; De Luca et al., 2014) with MI, respectively. Other scenario analyses were also performed, including initiating treatment among patients in different ages, varying the duration of alirocumab use between 5 and 30 years and applying various discount rates to the model.

We conducted all these analyses in disparate prices of alirocumab and simulation ways, if feasible, and key parameters are showed in Supplementary Table S3.

The model was performed by using TreeAge Pro 2020 R1.1 software program (TreeAge Software, Williamstown, MA). We followed the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) guidelines (Husereau et al., 2013) and the reporting checklist for economic evaluation of health interventions is showed in Supplementary Table S4.

In the main result applying the HRs observed in ODYSSEY OUTCOMES to the model, mean costs would be 871,321 CNY with alirocumab added to statin therapy and 400,705 CNY with mere statin therapy. The clinical benefit of alirocumab were 0.29 gained QALYs. Compared with statin monotherapy, alirocumab at its present price (34,355 CNY, 33% rebate) was estimated to produce an ICER of 1,613,997 CNY per QALY gained (Table 2). If, instead, a full list price of alirocumab was used (51,532 CNY), the ICER would rise to 2,465,017 CNY per QALY. The conventional WTP threshold would be achieved when the annual cost of alirocumab was reduced by 88% from the full official price to 6071 CNY. Another indirect use of relative risks (RRs) hypothesis of LDL-C reduction with alirocumab yielded more optimistic ICER as 805,795 CNY per QALY based on discounted price of alirocumab accompanied with a net value-based price of 11,861 CNY per year.

In one-way sensitivity analyses, the main drivers on the ICER of alirocumab therapy at its present price based on clinical follow-up efficacy were the utility of initial event-free and post-IS (Figure 2). Additional tornado diagrams based on different evaluation ways and price points of alirocumab are presented in Supplementary Figure S1. The uncertain results for a wide range of assumptive inputs were still far beyond the WTP threshold, demonstrating the robustness and consistency of model outcomes.

Probabilistic sensitivity analyses were demonstrated in Monte Carlo simulation scatters plots (Supplementary Figure S2) and cost-effectiveness acceptability curves (Figure 3A). Assuming the presupposed WTP threshold, the probability of alirocumab being cost-effective in MI population was 0.7% (0.3% for full list price) with clinical follow-up efficacy and 1.7% (1.1% for full list price) with LDL-C reduction hypothesis.

Separate scenarios were undertaken to evaluate the treatment benefit (Table 3). An all-cause death benefit yielded an ICER of 1,032,482 CNY and a value-based price of 8513 CNY annually. With the combination of ezetimibe and statin therapy, the ICER increased to 6,564,056 CNY and the cost of alirocumab must be reduced to 4232 CNY per year, which was unprecedented among biological medications in China. We also performed cost-effectiveness acceptability curves in three treatment ways (Figure 3B) and the main result showed that at the basic WTP threshold, the optimal treatment strategy was statin alone in 4.6%, statin plus ezetimibe in 95.2% and statin plus alirocumab in 0.2% of the simulation.

Similar or slightly favorable economic value of the PCSK9 inhibitor was observed in several subgroups, including female, DM or hypertension with MI. For poorly controlled FH patients after MI, the ICER was 254,945 CNY per QALY gained and the price was expected to be 30,299 CNY, which was extremely close to the present acquisition price. In patients with polyvascular disease in 3 vascular beds, the ICER for alirocumab plus statin vs. statin alone was 111,750 CNY, which achieved the accepted WTP threshold. Meanwhile, the valued-based price was 50,734 CNY per year, just a small reduction from the full list price. Regarding to different initial treatment ages, the ICERs based on clinical follow-up efficacy declined from 1,412,372 CNY in 65 years old to 842,649 CNY in 80 years old. Long-term use of alirocumab (over 20 years) brought about relatively better economic benefit in general MI population. The diverse discount rates of 3.5% and 6% produced an annual expected price of 6325 CNY and 5900 CNY, respectively.

This study had several limitations. Our cohort selected from PEACE study had only one-year follow-up duration, which may overestimate the incidence of adverse events after MI and the ICERs would be even higher. In addition, the PEACE researchers did not describe the initial average LDL-C level of their cohort. To overcome the shortcomings, we used the absolute LDL-C reduction from the ODYSSEY EAST study involving Chinese population to simulate the effect of alirocumab. And we did not perform subgroup analyses in patients with baseline LDL-C level over or below 100 mg/dL for the above reason, just as a significant difference of therapeutic effect reported in ODYSSEY OUTCOMES trial. These limitations could not affect the conclusion that alirocumab was also not cost-effective in patients with initial LDL-C > 100 mg/dL accounted for the lack of economic value in MI patients with FH defined as higher LDL-C level over 200 mg/dL. In real-world clinical practice, the adherence of patients may decline based on over optimism of the condition or exorbitant costs. From an economic standpoint, there is no increase in cost or benefit when patients stop taking the medicine and therefore do not change the economic outcomes. If the legacy effect of PCSK9 inhibitors is proved, cost-effectiveness will likely rise. Lastly, our analysis is not generalizable to primary prevention or patients with stable CVD.