https://orcid.org/0000-0001-5735-8482
Abstract
Despite intensive lipid-lowering therapies (LLTs), most patients with homozygous familial hypercholesterolaemia (HoFH) do not achieve guideline recommended low-density lipoprotein cholesterol (LDL-C) targets and are at increased risk of premature cardiovascular death. This analysis aimed to predict the impact of evinacumab and standard-of-care LLTs on life expectancy in an HoFH population using mathematical modelling.
Mathematical models were developed using efficacy data for evinacumab from the phase 3 ELIPSE HoFH trial plus efficacy data for standard-of-care LLTs from peer-reviewed publications. Treatment strategies evaluated included (i) untreated, (ii) high-intensity statin (HIS) only, (iii) HIS plus ezetimibe, (iv) HIS plus ezetimibe plus proprotein convertase subtilisin/kexin type 9 inhibitor (PCSK9i), and (v) HIS plus ezetimibe plus PCSK9i plus evinacumab. Markov analyses were used to assess differences in survival probability for different LLT strategies. The median survival for untreated HoFH patients was only 33–43 years, depending on different assumptions on baseline untreated LDL-C levels. In the most robust model, we estimated that HIS increased median survival by 9 years and ezetimibe further increased median survival by an additional 9 years. When PCSK9i was added on top of HIS plus ezetimibe, median survival was further improved by 14 years. Finally, the addition of evinacumab to standard-of-care LLTs was estimated to increase median survival by ∼12 years.
In this mathematical modelling analysis, evinacumab treatment could potentially increase long-term survival vs. standard-of-care LLTs for patients with HoFH.
Lay Summary
This mathematical modelling analysis demonstrated that evinacumab in addition to standard-of-care lipid-lowering treatments (LLTs; high-intensity statin plus ezetimibe plus proprotein convertase subtilisin/kexin type 9 inhibitor) could increase long-term survival to a median of 77 years vs. the 65 years achieved with only standard-of-care LLTs in patients with homozygous familial hypercholesterolaemia.
Introduction
Homozygous familial hypercholesterolaemia (HoFH) is a rare genetic disorder with an estimated prevalence of 1:250 000.1–4 The condition is characterized by markedly elevated levels of low-density lipoprotein cholesterol (LDL-C) from birth.2–4 Over 90% of HoFH cases are caused by genetic mutations causing loss-of-function variants of the low-density lipoprotein receptor (LDLR);5,6 the remaining cases are caused by mutations in apolipoprotein B (APOB), proprotein convertase subtilisin/kexin type 9 (PCSK9) genes or, rarely, the LDLR adaptor protein gene (LDLRAP1), resulting in severely impaired LDL-C clearance from the circulation.7,8 Such genetic variants can cause a complete absence of LDLR expression (null–null variants) or partially reduced LDLR activity from two non-null alleles or one null and one non-null allele (non-null variants).8 Homozygous familial hypercholesterolaemia remains difficult to treat and leads to early, accelerated-onset atherosclerotic cardiovascular disease (ASCVD) despite the use of currently available standard-of-care lipid-lowering therapies (LLTs).3
Untreated, the average age of death can be as low as 18 years, with many patients with HoFH not expected to live into their third decade.9,10 Chances of survival depend mainly on the degree of serum LDL-C control, hence why treatment to reduce LDL-C levels as early and aggressively as possible is critical.10 As a consequence of the extremely elevated levels of LDL-C (usually >400 mg/dL if untreated), patients with HoFH have a high risk of ASCVD and cardiovascular-related events at a young age, mainly myocardial infarction but also stroke and the need for coronary/vascular procedures such as angioplasty, coronary stent insertion, aortic valve replacement or repair, and coronary artery bypass surgery; they may also experience early death due to a cardiovascular cause.11,12
Low-density lipoprotein cholesterol targets for patients with HoFH are <100 mg/dL for children and <70 mg/dL in adults with clinical ASCVD.3,4 In newer guidelines, the 2019 European Atherosclerosis Society/European Society of Cardiology lipid management guidelines recommend LDL-C targets of <55, <70, and <100 mg/dL in patients with very-high, high, or moderate risk for ASCVD, respectively;1 the 2018 American Heart Association/American College of Cardiology guidelines recommend LDL-C goals of <70 and <100 mg/dL in the presence and absence of coronary artery disease or other major risk factors, respectively.13 Current options to lower LDL-C involve dietary and lifestyle modifications alongside several LLTs, with varying levels of effectiveness and uptake in the real world.1,7,13–15 High-intensity statin (HIS) therapy is recommended as a first-line treatment option for HoFH; ezetimibe should be added to statin therapy if LDL-C treatment targets are not achieved.1 For secondary prevention in patients at very-high ASCVD risk (including HoFH), the addition of a PCSK9 inhibitor (PCSK9i) may be considered.1 Further treatment options for HoFH include evinacumab, lomitapide, and lipoprotein apheresis.7,16
Although multiple treatments for HoFH are available, options that rely on up-regulating LDLR function (such as statins and PCSK9i) have limited-to-no effect in patients with null–null LDLR variants. In addition, there are known tolerability and feasibility issues with treatments such as lomitapide, HIS, and lipoprotein apheresis.3,13,17,18 Consequently, despite available LLTs, guideline-recommended LDL-C treatment targets are rarely achieved, representing a great unmet need for patients with HoFH.17
Evinacumab is a monoclonal antibody that inhibits angiopoietin-like 3.19,20 Evinacumab leads to a reduction in circulating LDL-C independent of the presence of the LDLR, by promoting very LDL processing and clearance upstream of LDL formation.20–22 The phase 3 ELIPSE HoFH trial randomized 65 patients with HoFH to evinacumab 15 mg/kg body weight every 4 weeks (Q4W) or placebo while receiving stable LLT.21 At Week 24, patients who received evinacumab had a 49.0% placebo-corrected reduction in LDL-C from baseline. Evinacumab is generally well tolerated.22 In February 2021, evinacumab was approved by the Food and Drug Administration (FDA) as an adjunct to other LDL-C–lowering therapies for the treatment of adult and paediatric patients aged ≥12 years with HoFH and was approved by the European Medicines Agency (EMA) in June 2021.22,23
Due to the rarity of HoFH, it is not possible to determine the effect of LLTs on survival outcomes via clinical trials. In instances where disease prevalence is low, mathematical modelling approaches are often undertaken to ascertain the impact of treatment on survival. Thus, the primary objective of this research was to use a mathematical model to estimate the potential impact of evinacumab on life expectancy in an HoFH population. The model was also used to predict the impact of other LLTs on life expectancy. As the ELIPSE HoFH trial did not record any cardiovascular events,21 data on baseline cardiovascular risk for patients with HoFH were taken from a previous retrospective cohort study.12
Methods
Target population
The initial age of the base case population was established per the approved FDA and EMA labelled authorization of evinacumab—i.e. patients aged ≥12 years with a diagnosis of HoFH (Table 1). In the model, the mean age of patients was set to 42.0 years, which was equivalent to the mean age of patients in the ELIPSE HoFH trial (41.7 years).21 An equal proportion of male and female patients was assumed. Moreover, the model assumed that 20% of patients had null–null LDLR variants and that 80% of patients had non-null LDLR variants (Table 1).6,21
Default model inputs for cohort characteristics and efficacies
| Parameter | Default inputs |
|---|---|
| Proportion of males, % | 50 |
| Initial age, years | 12 |
| Patients with null–null mutation, %6,21 | 20.0 |
| Hazard ratio for cardiovascular death12 | |
| Model 1 | 0.77 |
| Model 2 | 0.88 |
| Reduction in LDL-C due to high-intensity statin, %24,25 | 20.8 |
| Reduction in LDL-C due to ezetimibe 10 mg, %26 | 14.0 |
| Reduction in LDL-C due to PCSK9i, %27–29 | 30.8 |
| Reduction in LDL-C due to evinacumab 15 mg/kg, %21 | 49.0 |
| Untreated LDL-C, mmol/L (mg/dL)12 | |
| Model 1 | 15.9 (615) |
| Model 2 | 10.4 (400) |
| Parameter | Default inputs |
|---|---|
| Proportion of males, % | 50 |
| Initial age, years | 12 |
| Patients with null–null mutation, %6,21 | 20.0 |
| Hazard ratio for cardiovascular death12 | |
| Model 1 | 0.77 |
| Model 2 | 0.88 |
| Reduction in LDL-C due to high-intensity statin, %24,25 | 20.8 |
| Reduction in LDL-C due to ezetimibe 10 mg, %26 | 14.0 |
| Reduction in LDL-C due to PCSK9i, %27–29 | 30.8 |
| Reduction in LDL-C due to evinacumab 15 mg/kg, %21 | 49.0 |
| Untreated LDL-C, mmol/L (mg/dL)12 | |
| Model 1 | 15.9 (615) |
| Model 2 | 10.4 (400) |
LDL-C, low-density lipoprotein cholesterol; PCSK9i, proprotein convertase subtilisin/kexin type 9 inhibitor.
Default model inputs for cohort characteristics and efficacies
| Parameter | Default inputs |
|---|---|
| Proportion of males, % | 50 |
| Initial age, years | 12 |
| Patients with null–null mutation, %6,21 | 20.0 |
| Hazard ratio for cardiovascular death12 | |
| Model 1 | 0.77 |
| Model 2 | 0.88 |
| Reduction in LDL-C due to high-intensity statin, %24,25 | 20.8 |
| Reduction in LDL-C due to ezetimibe 10 mg, %26 | 14.0 |
| Reduction in LDL-C due to PCSK9i, %27–29 | 30.8 |
| Reduction in LDL-C due to evinacumab 15 mg/kg, %21 | 49.0 |
| Untreated LDL-C, mmol/L (mg/dL)12 | |
| Model 1 | 15.9 (615) |
| Model 2 | 10.4 (400) |
| Parameter | Default inputs |
|---|---|
| Proportion of males, % | 50 |
| Initial age, years | 12 |
| Patients with null–null mutation, %6,21 | 20.0 |
| Hazard ratio for cardiovascular death12 | |
| Model 1 | 0.77 |
| Model 2 | 0.88 |
| Reduction in LDL-C due to high-intensity statin, %24,25 | 20.8 |
| Reduction in LDL-C due to ezetimibe 10 mg, %26 | 14.0 |
| Reduction in LDL-C due to PCSK9i, %27–29 | 30.8 |
| Reduction in LDL-C due to evinacumab 15 mg/kg, %21 | 49.0 |
| Untreated LDL-C, mmol/L (mg/dL)12 | |
| Model 1 | 15.9 (615) |
| Model 2 | 10.4 (400) |
LDL-C, low-density lipoprotein cholesterol; PCSK9i, proprotein convertase subtilisin/kexin type 9 inhibitor.
Model structure and health states
The structure of the model is depicted in Figure 1. The Markov model population is categorized into three health states: HoFH; death due to an ischaemic or non-ischaemic cardiovascular event; and death due to a non-cardiovascular cause.
Markov model structure with cardiovascular and non-cardiovascular death. HoFH, homozygous familial hypercholesterolaemia.
Efficacy
Comparators
To assess the effectiveness of evinacumab 15 mg/kg Q4W as an adjunct to standard-of-care LLTs, the following strategies for the treatment of HoFH were modelled: (i) no treatment with LLTs, (ii) HIS (atorvastatin or rosuvastatin) only, (iii) HIS plus ezetimibe, (iv) HIS plus ezetimibe plus PCSK9i (evolocumab or alirocumab), and (v) HIS plus ezetimibe plus PCSK9i plus evinacumab. Of note, in the ELIPSE HoFH trial, 63.1% of patients were receiving at least three LLTs, with 44.1% receiving a statin plus ezetimibe plus PCSK9i.21 The model simulated identical cohorts for each treatment strategy, and all patients were assumed to be receiving LLT for lifetime duration.
Low-density lipoprotein cholesterol reduction
Efficacy data for percentage LDL-C lowering with evinacumab 15 mg/kg Q4W were obtained from the ELIPSE HoFH trial.21 For the standard-of-care LLTs, efficacy data for percentage LDL-C lowering specific to HoFH were determined based on clinical trials on HoFH patients via a literature search. If more than one LDL-C–lowering percentage was available for the standard-of-care LLTs, the mean value was used.24,26–29 Percentages of LDL-C reductions assumed with each LLT are shown in Table 1. Of note, evinacumab achieved a 52% reduction in LDL-C among those on statin, ezetimibe, and PCSK9i in the ELIPSE HoFH trial,30 which is similar to the 49% reduction for all trial patients. Post hoc analysis on ELIPSE HoFH trial data showed that efficacy of evinacumab did not differ substantially with different baseline LLTs.30 Therefore, we used 49% LDL-C reduction for evinacumab in the model. Starting from a baseline untreated LDL-C level, the assumed absolute reduction in LDL-C after the stepwise addition of each standard-of-care LLT is additive. Given that statins and PCSK9i are ineffective in patients with null–null LDLR variants, LDL-C reductions with HIS and PCSK9i were only applied to patients with non-null LDLR variants. By comparison, as ezetimibe is equally effective in both null–null and non-null LDLR variants,26 the same LDL-C reductions were applied to both groups.
Survival analysis
As there were no cardiovascular events recorded during the ELIPSE HoFH trial, baseline untreated cardiovascular risk for the model was derived from a retrospective cohort study that comprised 149 patients with HoFH from two specialized lipid clinics in South Africa.12 In this retrospective study, the number of individual cardiovascular events was reported, and a Cox proportional hazard model with time-varying exposure was used to estimate the risk of all-cause mortality. The probability of survival for untreated patients with HoFH was estimated from a digitization of the reported all-cause mortality curve, and the proportion of cardiovascular deaths was calculated from the reported causes of death.
The probabilities for mortality due to a non-cardiovascular cause were obtained from US life tables.31 Final survival probabilities were determined by adding the probabilities for mortality due to a non-cardiovascular cause (obtained from US life tables) to the calculated probabilities of death due to a cardiovascular cause derived from Raal et al.12
The probability of cardiovascular mortality post-treatment was calculated using baseline mortality probabilities, baseline LDL-C, and efficacy of the LLT being analysed. Depending on the source for baseline untreated LDL-C and hazard ratio (HR) of cardiovascular mortality due to LDL-C reduction, we constructed two main models to test the prediction of survival benefit resulting from lowering LDL-C by multiple therapies. Model 1 relied on the South African retrospective cohort study on HoFH patients, with mean baseline LDL-C for untreated patients of 15.9 mmol/L (615 mg/dL).12 Data for the relationship between cardiovascular events and LDL-C reduction were also taken from the retrospective study12 and entailed comparison of the treated and untreated cohorts. Using these data, the LDL-C reduction with each LLT could be related to the reduction in cardiovascular event probabilities (HR normalized per 1 mmol/L or 38.7 mg/dL of LDL-C). These data were also used to adjust the baseline cardiovascular event probability based on the initial untreated LDL-C. Based on the reported HR of 0.34 for all-cause mortality with an LDL-C change of 4.2 mmol/L and the majority (77%) of deaths being cardiovascular causes, we assumed that the HR of 0.34 applies to cardiovascular death and normalized it to an HR per 1 mmol/L. The calculated HR for cardiovascular mortality was 0.77 per 1 mmol/L (40 mg/dL) of LDL-C reduction.
Model 2 adopted a baseline LDL-C level that matches multiple HoFH clinical trials and an HR of 0.88 for cardiovascular mortality, as reported by the Cholesterol Treatment Trialists’ (CTT) Collaborators for those with vascular disease. This is a robust meta-analysis based on all statin trials involving a quite large sample. However, the population is not HoFH specific. In terms of LDL-C levels from trials, a pivotal trial for ezetimibe reported an LDL-C level of 341 mg/dL for those on HIS, and the ODYSSEY HoFH trial with alirocumab reported an LDL-C level of 262 mg/dL for those on HIS plus ezetimibe. Back calculation assuming the efficacies obtained from trials listed in Table 1 resulted in an untreated baseline LDL-C level of 10.5 (409 mg/dL) and 9.4 mmol/L (365 mg/dL), respectively. Due to the general consensus that patients with HoFH usually have an untreated LDL-C level over 400 mg/dL, we adopted a baseline LDL-C level of 400 mg/dL in Model 2,8 which resulted in an LDL-C level of 8.6 mmol/L (333 mg/L) for those on HIS, 7.4 mmol/L (286 mg/dL) for those on HIS plus ezetimibe, 5.6 mmol/L (216 mg/dL) for those on HIS plus ezetimibe plus PCSK9i, and 2.8 mmol/L (110 mg/dL) for those on HIS plus ezetimibe plus PCSK9i plus evinacumab. These numbers match well with LDL-C levels reported in trials and may better represent the LDL-C level of contemporary HoFH patients compared with the South African cohort.26,28
The probability of cardiovascular mortality post-treatment for any given point of time was compared with the probability of cardiovascular death from life tables to ensure that the higher of the two values was used in the model. Annual survival was calculated using the annual cardiovascular death probability plus annual non-cardiovascular death from US life tables.31
Sensitivity analyses
Probabilistic sensitivity analyses were conducted for the two main models to calculate the 95% confidence intervals around the reported results, in which cohort characteristics like proportion of males, proportion of patients with null–null mutations, baseline LDL-C, treatment efficacies, and rate ratio for cardiovascular death were probabilistically sampled from distributions around their respective means. We also conducted several scenario analyses, including (i) a scenario analysis for Model 1 changing the HR of 0.77 for cardiovascular mortality obtained from Raal et al. to 0.88 reported by the CTT Collaborators;32 and (ii) a scenario analysis for Model 2 changing baseline LDL-C to the level that matches the ELIPSE HoFH trial population. We used the LDL-C level of 271.1 mg/dL for those on statin, ezetimibe, and PCSK9i reported by a post hoc analysis of the ELIPSE HoFH trial data and back calculated the untreated LDL-C levels assuming the same efficacy of HIS, ezetimibe, and PCSK9i as well as the per cent of null–null mutation used in the main analysis; and (iii) separate analyses for null-null and non-null patients in Model 1 and Model 2.
Safety
No safety information was included in the model, as adverse events occurred at similar rates in the evinacumab and placebo groups during the treatment period of the ELIPSE HoFH trial.21 Moreover, no patients discontinued evinacumab treatment because of an adverse event, and no deaths were observed.21
Assumptions
Patients were assumed to be on treatment continuously and indefinitely over the treatment duration defined in the model. In addition, patients were assumed to be adherent to treatment for the duration of the analysis period. Low-density lipoprotein cholesterol is assumed to be reduced instantaneously with LLTs and then to remain constant over time where no change in treatment occurs, as detailed in the literature previously.3,26,29 In addition, we assumed that every patient on evinacumab had background therapies of HIS, and ezetimibe plus PCSK9i the reduction in LDL-C after the stepwise addition of each LLT was therefore assumed to be additive. Benefits observed in LDL-C reduction were assumed to last the full treatment duration, and it was assumed that any decreases in cardiovascular events due to lowering of LDL-C were immediate for a treated patient.
Results
When untreated, patients with HoFH had significantly shorter life expectancy than the general US population. Compared with the median survival of 79 years for the general US population, the median survival for untreated HoFH patients was only 33 years in Model 1 and 43 years in Model 2, depending on the different baseline untreated LDL-C levels used. This equates to a mean decrease in life expectancy of 36–46 years in untreated patients with HoFH vs. the general US population.
Markov analyses were used to assess the differences in survival probability between cohorts of patients on different LLT strategies. Table 2 summarizes the increased life years with different LLTs. In Model 1, HIS increased median survival by 9 years, and ezetimibe further increased median survival by an additional 9 years. When PCSK9i was added on top of HIS plus ezetimibe, median survival was further improved by 14 years. Finally, the addition of evinacumab to standard-of-care LLTs was estimated to increase median survival by ∼12 years. In Model 2, median survival was increased by 5 years with HIS, 3 years with add-on ezetimibe, 5 years with add-on PCSK9i, and 7 years with add-on evinacumab (Table 2). Survival curves for patients taking different LLT combinations in Model 1 and Model 2 are shown in Figure 2A and B, respectively.
Comparison of treatment strategies and the impact on survival probability for (A) Model 1 and (B) Model 2. HIS, high-intensity statin; PCSK9i, proprotein convertase subtilisin/kexin type 9 inhibitor.
Estimated increased life years by model
| Model 1: South African data, HR = 0.77 | SA1: Change HR to 0.88 | Model 2: Baseline LDL-C matches ezetimibe and alirocumab trial data, HR = 0.88 | SA2: Change baseline LDL-C to match evinacumab trial | |
|---|---|---|---|---|
| Untreated median survival | 33 | 33 | 43 | 38 |
| Increased life years by HIS | 9 | 4 | 5 | 4 |
| Increased life years by add-on ezetimibe | 9 | 4 | 3 | 4 |
| Increased life years by add-on PCSK9i | 14 | 6 | 5 | 6 |
| Increased life years by add-on evinacumab | 12 | 11 | 7 | 9 |
| Model 1: South African data, HR = 0.77 | SA1: Change HR to 0.88 | Model 2: Baseline LDL-C matches ezetimibe and alirocumab trial data, HR = 0.88 | SA2: Change baseline LDL-C to match evinacumab trial | |
|---|---|---|---|---|
| Untreated median survival | 33 | 33 | 43 | 38 |
| Increased life years by HIS | 9 | 4 | 5 | 4 |
| Increased life years by add-on ezetimibe | 9 | 4 | 3 | 4 |
| Increased life years by add-on PCSK9i | 14 | 6 | 5 | 6 |
| Increased life years by add-on evinacumab | 12 | 11 | 7 | 9 |
HIS, high-intensity statin; HR, hazard ratio; LDL-C, low-density lipoprotein cholesterol; PCSK9i, proprotein convertase subtilisin/kexin type 9 inhibitor; SA, sensitivity analysis.
Estimated increased life years by model
| Model 1: South African data, HR = 0.77 | SA1: Change HR to 0.88 | Model 2: Baseline LDL-C matches ezetimibe and alirocumab trial data, HR = 0.88 | SA2: Change baseline LDL-C to match evinacumab trial | |
|---|---|---|---|---|
| Untreated median survival | 33 | 33 | 43 | 38 |
| Increased life years by HIS | 9 | 4 | 5 | 4 |
| Increased life years by add-on ezetimibe | 9 | 4 | 3 | 4 |
| Increased life years by add-on PCSK9i | 14 | 6 | 5 | 6 |
| Increased life years by add-on evinacumab | 12 | 11 | 7 | 9 |
| Model 1: South African data, HR = 0.77 | SA1: Change HR to 0.88 | Model 2: Baseline LDL-C matches ezetimibe and alirocumab trial data, HR = 0.88 | SA2: Change baseline LDL-C to match evinacumab trial | |
|---|---|---|---|---|
| Untreated median survival | 33 | 33 | 43 | 38 |
| Increased life years by HIS | 9 | 4 | 5 | 4 |
| Increased life years by add-on ezetimibe | 9 | 4 | 3 | 4 |
| Increased life years by add-on PCSK9i | 14 | 6 | 5 | 6 |
| Increased life years by add-on evinacumab | 12 | 11 | 7 | 9 |
HIS, high-intensity statin; HR, hazard ratio; LDL-C, low-density lipoprotein cholesterol; PCSK9i, proprotein convertase subtilisin/kexin type 9 inhibitor; SA, sensitivity analysis.
Several sensitivity analyses were conducted to test the robustness of the model. First, for Model 1, we changed the HR of 0.77 for cardiovascular mortality based on Raal et al. to HR of 0.88 reported in the CTT Collaborators’ analysis.32 Median survival was increased by 4 years with HIS, 4 years with add-on ezetimibe, 6 years with add-on PCSK9i, and 11 years with add-on evinacumab (Table 2). The resulting survival pattern with different combinations of LLTs is shown in Supplementary material online, Figure S1. Second, for Model 2, a scenario analysis was conducted by changing baseline untreated LDL-C to the level that matches ELIPSE HoFH trial patients instead of that matches ezetimibe and alirocumab trials. Adding evinacumab to HIS plus ezetimibe plus PCSK9i increased median survival by 9 years (Table 2 and Supplementary material online, Figure S2). In addition, results of separate analyses for those with non-null mutation and those with null–null mutation in both Model 1 and Model 2 were shown in Supplementary material online, Table S1, and probabilistic sensitivity analysis result was shown in Supplementary material online, Figure S3.
Discussion
As HoFH is an ultrarare disease, there is limited published literature pertaining to the long-term survival of this patient population. Moreover, the low prevalence precludes any assessment of cardiovascular outcomes in patients who are at high risk of ASCVD. Consequently, mathematical modelling exercises are a practical way to gain an understanding of the likely long-term effects of therapeutic interventions, other than LDL apheresis.
Current treatment options to reduce LDL-C in patients with HoFH have either limited efficacy or known tolerability and feasibility issues. Consequently, guideline-recommended LDL-C treatment targets are rarely met with maximally tolerated combinations of LLTs, and patients remain at risk of cardiovascular events and premature mortality.1,13 Thus, a persistent unmet need remains for an effective treatment that can be started early in life and can reduce cardiovascular morbidity and mortality.3
In our analyses, we estimated that adding evinacumab on top of standard-of-care LLTs (HIS plus ezetimibe plus PCSK9i) substantially increased median survival of HoFH patients by 7–12 years, depending on assumptions on baseline LDL-C levels and HR for mortality due to LDL-C reduction. Therefore, the availability of evinacumab greatly helped to address the unmet needs of the HoFH population, and improved their life expectancy to a level closer to the US general population. These analyses demonstrate that, for patients with HoFH, effective LDL-C lowering is a plausible means to achieve long-term survival benefit.
Baseline LDL-C and HR of mortality resulting from LDL-C reduction are key assumptions of our modelling. Due to the rarity of the disease and lack of optimal evidence for HoFH population, we built two main models based on different data sources for these key model assumptions. In general, we consider results from Model 1 based on the South African cohort study to be more plausible as it is based on risk reductions derived from an actual HoFH population, which were also adopted by past similar studies.33 Also, this is the only source that provided information on how reduction of LDL-C related to survival exclusively among patients with HoFH. However, we do acknowledge the limitations of this data source. This study included historical data from the 1970s, and all patients were from South Africa. Therefore, the extremely high baseline untreated LDL-C level may not be applicable to current HoFH patients in the USA. Consequently, the reduced LDL-C level with evinacumab treatment predicted in the model was higher than that observed in multiple HoFH trials. In addition, the HR of 0.77 may be an overestimate on the effect of LDL-C reduction due to the high baseline LDL-C level. As a result, it is possible that Model 1 is over-predicting survival as it is unlikely that the current US patients with lower baseline LDL-C levels would get the same absolute reduction in LDL-C compared with the South African cohort. To try to address such concerns, Model 2 utilized an HR of 0.88 derived from CTT Collaborators and baseline LDL-C that matches a more recent HoFH trial population. Therefore, the population in Model 2 may be better representative of current HoFH patients in the USA. However, the downside of Model 2 is that the HR of 0.88 was based on statin trials conducted among the general population at risk of cardiovascular disease and is thus less likely to be applicable to HoFH patients, as HoFH patients have significantly higher cardiovascular risk and therefore may get a larger benefit from LDL-C reduction compared with the general population. In fact, we found the treated LDL-C level with add-on evinacumab was at 2.8 mmol (110 mg/dL) in Model 2, which was very close to the average LDL-C level of the general US population (111.7 mg/dL).34 However, patients achieved shorter life expectancy in Model 2 compared with the general population (63 vs. 79 years). This finding suggests that our assumption of an HR of 0.88 is likely to be an underestimate of the effect of LDL-C reduction on mortality for HoFH patients, and resulting survival gains are therefore conservative estimates.
Of note, lomitapide was not included as a treatment option in the current model because of its known dose-limiting issues, such as hepatic steatosis and gastrointestinal side effects,35–38 especially following long-term use.39 A previous publication on a modelling analysis of the effect of lomitapide on cardiovascular outcomes in patients with HoFH reported an increase in life expectancy due to additional LDL-C lowering by lomitapide. However, a limitation of this previous modelling analysis is the omission of the effect of ezetimibe and PCSK9is.33
In addition to the limitation mentioned above that a few key assumptions relied on the South African study, several other limitations of this study should be noted. First, in this model, we assumed that every patient taking evinacumab concurrently took HIS, ezetimibe, and PCSK9is, based on the recommendation that evinacumab is to be added to maximally tolerated LLTs including statins, ezetimibe, and PCSK9is. Therefore, the calculated reduction in LDL-C after the stepwise addition of each LLT was additive. However, in the ELIPSE HoFH trial, while patients were being treated with maximally tolerated doses of currently available LLTs, only 44.1% of patients had background treatment of statins, ezetimibe, and PCSK9is; it is also unlikely that all patients will receive all three LLTs before evinacumab in the real world. In addition, the HR of mortality per 1 mmol/L used in our model was derived from studies on the effect of statins.12,32 We applied this HR to evinacumab, which has a different mechanism of action to statins. It is not proved that treatment by LLTs with different mechanism of actions, especially ANGTPL3 inhibition, will lead to a survival benefit similar to statins.
Conclusion
In conclusion, in the context of limited long-term data on HoFH, we developed a mathematical model that indicates that evinacumab treatment may offer a potential long-term survival benefit vs. standard-of-care LLTs for patients with HoFH. Future research should focus on developing real-world registries for HoFH patients to facilitate the knowledge of this ultrarare disease.
Supplementary material
Supplementary material is available at European Journal of Preventive Cardiology.
Acknowledgements
The authors would like to thank the patients, their families, and all investigators involved in the parent study. Medical writing assistance and editorial support, under the direction of the authors, was provided by Rebecca Mottram, PhD, and Atif Riaz, PhD, of Prime, Knutsford, UK, funded by Regeneron Pharmaceuticals, Inc. according Good Publication Practice guidelines (link). The sponsors were involved in the study design and collection, analysis, and interpretation of data, as well as data checking of information provided in the manuscript. The authors had unrestricted access to study data, were responsible for all content and editorial decisions, and received no honoraria related to the development of this publication.
Author contributions
Study conception and design: J.G. and A.K.; model development: A.C. and T.S.S.; interpretation of results: J.G., A.K., P.Q., and F.J.R.; manuscript writing: J.G., A.K., P.Q., and F.J.R.; and all authors reviewed results and approved the final version of the manuscript.
Funding
This study was funded by Regeneron Pharmaceuticals, Inc.
Data availability
All data analysed during this study are reported in the previously published articles cited in the methods section. All analysis results of this study are presented in the manuscript and in the supplementary material.
References
Author notes
Conflict of interest: J.G., A.K., and P.Q. are employees of and stockholders in Regeneron Pharmaceuticals, Inc. A.C. and T.S.S. are employees of Axtria. F.J.R. has received research grants, honoraria, or consulting fees for professional input from, and/or delivered lectures for, Sanofi, Regeneron, Amgen, Novartis, and LIB Therapeutics.
Comments