Cardiovascular disease (CVD) is the leading cause of premature death and disability in Europe.1 A recent meta-analysis showed that a greater reduction in low-density lipoprotein cholesterol (LDL-C) was associated with greater cardiovascular benefit.2 Only 25% of patients with CVD3 and 3.4% of those with familial hypercholesterolemia (FH)4 achieve their lipid targets. Furthermore, around 5% to 20% of patients treated in real-world clinical practice and around 2% in randomized clinical trials have statin intolerance.5 Proprotein convertase subtilisin/kexin type 9 inhibitors (PCSK9i) are drugs that can reduce LDL-C by up to 65% and have a good safety profile.6 The FOURIER study7 demonstrated that adding evolocumab to statin treatment reduced the relative risk of the primary outcome of the study by 15%. In the coming months, the ODYSSEY Outcomes trial8 will help clarify the role of alirocumab in patients with coronary artery disease.

Nonetheless, the cost of PCSK9 inhibitors remains a considerable limitation to their more widespread use. Such factors have led the Ministry for Health, Social Services and Equality,9,10 the SEA (Spanish Society of Arteriosclerosis),11 the SEC (Spanish Society of Cardiology),12 NICE (National Institute for Health and Care Excellence),13,14 and the European Society of Cardiology/European Atherosclerosis Society (ESC/EAS) Task Force15 to establish indications for use.

The aim of this study was to estimate the number and type of patients who are candidates for receiving PCSK9 inhibitors in this country, according to different indication criteria, using a population database of 2 500 907 patients.

This was an observational study that used the Information System for the Development of Research in Primary Care (SIDIAP, by its initials in Spanish), a clinical database of anonymous longitudinal registries of 6 177 972 patients between 2006 and 2014. SIDIAP includes information on the clinical activity of 3414 physicians from 274 primary care centers of the Catalan Institute for Health, a public entity providing health care cover to 85% of the population in Catalonia.16 It includes diagnoses (International Classification of Disease [ICD-10]), information on hospital discharge (ICD-9/10), laboratory findings, and medications dispensed by pharmacies. The validity of SIDIAP data for studying CVD epidemiology has been documented.17 The Clinical Research and Ethics Committee of the Institut Universitari d’Investigació en Atenció Primària (University Institute for Research in Primary Care) approved the study.

The study included all SIDIAP entries for individuals alive and aged ≥ 18 years in December 2014 with at least 1 LDL-C measurement between 2006 and 2014. Patients with a history of hypothyroidism, nephrotic syndrome, or baseline triglycerides ≥ 400 mg/dL were excluded.

There were 2 764 917 people who had at least 1 recorded LDL-C measurement. Of those, 2 500 907 patients met all the inclusion criteria. Figure 1 shows a flow diagram of the study. We identified 14 458 patients with an FH phenotype and 207 411 with CVD.

Figure 1.

Summary diagram of study participant selection. LDL-C, low-density lipoprotein cholesterol; SIDIAP, Information System for the Development of Research in Primary Care.

(0.33MB).

In the whole study population, 43.3% of the individuals on treatment did not have pretreatment LDL-C values available and therefore imputed values were used. The mean LDL-C before treatment was 166mg/dL and the mean observed LDL-C (treated and untreated) was 110mg/dL. The mean intensity of lipid-lowering therapy used was 32.6% and the calculated treatment adherence showed a medication possession ratio of 71.3% (Table 2).

The number of candidates, according to the different criteria, is shown by patient subgroup and by treatment in Table 3.

Table 4 shows the number of candidate patients in an optimized treatment scenario. Approximately half of those who were PCSK9i candidates had an FH phenotype according to the SEA11 and NICE13,14 criteria. According to the ESC/EAS Task Force15 and to a greater extent the SNS9,10 and SEC12 criteria, more than 75% of the candidates were patients in secondary prevention.

To estimate the number of PCSK9i candidates in the Spanish population aged ≥ 18 years, the prevalence of CVD observed in the assigned population in the whole SIDIAP data set was 6.2%. The prevalence of FH was 0.53%. In an optimized treatment scenario, the SNS9,10 criteria were those that indicated most PCSK9i candidates, at 1.7% of the Spanish population aged 18 years or older. The NICE13,14 criteria were the most restrictive, at 0.1% (Table 5).

Optimization of lipid-lowering therapy resulted in a very substantial reduction in the number of candidates for treatment, by as much as half of the candidates for most of the criteria, being highest in the case of the SNS9,10 criteria (which went from 3.26% to 1.67% of the population) and lowest in the case of the NICE13,14 criteria (which went from 0.60% to 0.11% of the population) (Table 5).

In the hypothetical situation of a 50% reduction in LDL-C from baseline, the group of patients who are candidates for PCSK9 inhibitors was reduced considerably compared with the scenario based on LDL-C reductions observed in real-life clinical practice, between 0.89% according to SNS and 0.04% for NICE (Appendix 3 of the supplementary material).

Figure 2 shows the distribution of the population according to the LDL-C values in conditions of real-life clinical practice. LDL-C values > 130mg/dL were present in 77.4% of patients with FH but not CVD, in 44.9% of patients with FH and CVD, and in 17.2% of patients with CVD but not FH. Figure 3 shows the same distribution, but includes only patients who received optimized treatment. In this situation, LDL-C values > 130mg/dL were present in 51.4% of patients with FH but not CVD, 25.2% of patients with FH and CVD, and 3.3% of patients with CVD but not FH. In conditions of real-life clinical practice, in patients with CVD but not FH, 44.5% were above 100mg/dL and 81.7% were above 70mg/dL.

Figure 2.

Population distribution according to LDL-C values in conditions of real-life clinical practice. CVD, cardiovascular disease; FH, familial hypercholesterolemia; LDL-C, low-density lipoprotein cholesterol.

(0.11MB).

Figure 3.

Population distribution according to LDL-C values in patients receiving optimized treatment. CVD, cardiovascular disease; FH, familial hypercholesterolemia; LDL-C, low-density lipoprotein cholesterol.

(0.1MB).

The sensitivity analysis of the results with 20 imputations showed no significant differences in the outcomes, population definitions, LDL-C values, or the percentage of PCSK9i candidates (Appendix 4 of the supplementary material). The analysis with complete data showed lower LDL-C values and a lower percentage of PCSK9i candidates (Appendix 5 of the supplementary material).

This study used a database from real-life clinical practice in Spain to analyze the number of patients who were candidates for receiving PCSK9 inhibitors when different indication criteria were applied. Considerable differences were found in the type and number of candidates among the different criteria analyzed, ranging from 0.1% to 1.7% of the population aged ≥ 18 years. The required LDL-C target in each set of criteria largely explains these differences. Optimization of lipid-lowering therapy would considerably reduce the number of candidates. If optimization achieved a 50% LDL-C reduction from baseline, this reduction in candidate number would be very substantial.

In the subgroup of patients with FH phenotype in primary prevention, the number of patients who would be candidates after optimizing treatment was around 9% for NICE13,14 and 86%for SNS9,10; for the subgroup with FH in secondary prevention, it would be between 21% for NICE13,14 and 97% for SEC,12 and in patients with CVD but not FH, between 0.8% for NICE13,14 and 20% for SNS.9,10 The percentage of candidate patients in the population with FH according to the ESC/EAS Task Force15 criteria obtained in our study (12% in primary prevention and 23% in secondary prevention) is in line with the published figures from the FH registry from the SEA.20 A large part of the differences between criteria, especially in patients with FH in primary prevention, is due to the LDL-C cutoff for the treatment indication, and there are considerable increases if this is moved from 130 to 100mg/dL or 100 to 70mg/dL. 63.8% of patients with CVD but not FH have LDL-C values between 70 and 130mg/dL. In our population, moving from an LDL-C cutoff of 100 to 70mg/dL meant doubling the number of patients who are candidates (Figures 2 and 3). In the FOURIER study, 67% of patients on treatment with evolocumab achieved LDL-C levels < 40mg/dL and 42% achieved < 25mg/dL, and the absolute risk of CVD decreased from 11.3% to 9.8%.7 It should be noted that, for the same LDL-C value, the risk of CVD is up to 4 times higher in patients with FH than in those without FH.21 It is essential to determine the optimal LDL-C cutoff and the criteria for poor prognosis for PCSK9i indication in each patient subgroup so that such recommendations can be publicized. The SAFEHEART investigators have developed the first equation that can help improve prediction of CV risk in FH.22 Recent, diffuse, coronary artery disease that is progressive or cannot be revascularized indicates a poor prognosis in patients in secondary prevention.23

The type of patients also varies between the different criteria. Most patients with FH would be treated, if applying the SNS,9,10 SEA,11 and especially the SEC12 criteria, which focus on patients with FH and coronary disease. The criteria that relate more to patients with CVD without FH are the SEC12 and ESC/EAS Task Force15 criteria.

Another consideration is that the impact of cholesterol differs between coronary disease and other manifestations of CVD.24 The recommendations of the Spanish Society of Cardiology12 are exclusively for patients with coronary disease.

Another aspect to consider is the lack of studies on the cardiovascular benefit associated with dramatic reductions in LDL-C for patients over 75 years old in primary prevention.25 None of the guidelines mention potential age limits.

A recent meta-analysis put lipid control in the Spanish population at around 15% in secondary prevention and up to 65% in primary prevention, and detected–as in our study–underuse of high-intensity treatment and of combined use with ezetimibe.26 One of the biggest challenges in optimizing the rational use of PCSK9 inhibitors will be to avoid the over-diagnosis of statin intolerance.5 The use of electronic systems in decision-making could improve the number of patients with good lipid control.27 In the present study, when we simulated a scenario with optimized lipid treatment, the percentage of patients who were candidates for PCSK9 inhibitors according to the SEC12 criteria was reduced 1.7-fold.

If LDL-C levels were reduced by 50% from baseline in all patients, the percentage of candidates according to the SEC criteria would be reduced 2.7-fold (Appendix 3 of the supplementary material).

In a recent study, Cannon et al.28 created a simulation model for PCSK9i candidates in a USA population with CVD but not FH, based on successive adjustments in the dose and intensity of lipid-lowering therapy and with a target LDL-C < 70mg/dL. They estimated that 14% would be candidates, lower than the 24.5% estimated in our population according to the SEC13 criteria, which also set a target of LDL-C < 70mg/dL. These differences could be mostly explained by the simulation method used. In the same article they discuss different simulated scenarios and the percentage of PCSK9i candidates varies from 6% to 28%. The simulation scenario is based on the effects observed in real-life clinical practice in optimized patients, and in our opinion, this scenario is more realistic than assuming that the effect would be optimal and linear in all patients. In our alternative scenario with a 50% LDL-C reduction (Appendix 3 of the supplementary material), applying the same criteria, the percentage of candidates would be 10.9%.

Given that the prevalence of the FH phenotype in our setting is close to 0.5%29 and that it requires long-term treatment and has an onset at a young age, the direct impact of PCSK9 inhibitors on the public health system could be significant. To date, cost-effectiveness studies have had mixed results. A recent study found that the 5-year number needed to treat with PCSK9 inhibitors to prevent 1 cardiovascular event in patients at very high risk with an LDL-C target < 70mg/dL would be around 50.30 Another study showed that adding evolocumab to treatment with statins and ezetimibe in patients with FH may be a cost-effective measure.31 A recent study in the United States concluded that treatment with PCSK9 inhibitors would need to be around 70% cheaper to be cost-effective.32 A Norwegian study estimated that PCSK9 inhibitors would be cost-effective only as secondary prevention in older patients at very high risk.33 The only study in a Spanish population published to date showed that evolocumab may be cost-effective in patients with FH and patients in secondary prevention, with an incremental cost-effectiveness ratio of 30 893 euros and within the limits set as cost-effective in the Spanish population.34 However, we will have to wait for future cost-effectiveness studies on the results of the FOURIER7 and ODYSSEY Outcomes8 trials. Future studies to stratify risk within each subgroup will also be necessary.

The number of potential candidates for PCSK9 inhibitors in conditions of real-life clinical practice is very high and varies substantially depending on the recommendations of the different scientific societies, at between 0.1% and 1.7% of the Spanish population aged ≥ 18 years. These differences are largely explained bytThe LDL-C target required in each set of criteria. The type of patients also varies between the different criteria. The subgroup with the highest percentage of potential candidates is composed of patients with FH and CVD, although in absolute terms the group with most candidates consists of patients in secondary prevention without FH. An intensive lipid-lowering treatment, with high-dose statins and combination therapy with ezetimibe, would considerably reduce the number of PCSK9i candidates.

Ministry of Economy, via the Instituto de Salud Carlos III (Cardiovascular Research Network-HERACLES Program RD12/0042 and redIAPP [Red de Investigación en Actividades Preventivas y Promoción de la Salud] RD12/0007) and the European Regional Development Fund, CIBER de Enfermedades Cardiovasculares. The government of Catalonia via the Agència de Gestió Ajuts Universitaris de Recerca (2014 SGR 240) and (2014 SGR 902) and via the Agència de Qualitat i Avaluació Sanitàries de Catalunya, Pla Estratègic de Recerca i Innovació en Salut (SLT002/16/00145).

N. Plana has given presentations funded by Alexion, Amgen, Ferrer, MSD, and Sanofi and attended scientific meetings funded by Amgen and Rovi. L. Masana has given presentations funded by Amgen, MSD, and Sanofi. À. Vila has given presentations funded by Ferrer, Sanofi, and Esteve and attended scientific meetings funded by Amgen and Ferrer.