Patient characteristics, treatment patterns, and adherence to lipid-lowering therapies following an acute coronary syndrome

Endocrinologie, Hospital Pitie-Salpetriere, 75013, Paris, France Health Economics, Amgen SAS, 92100, Boulogne-Billancourt, France Pharmacoepidemiology, Certara, Evidence & Access (previously Analytica Laser), EC1N 6SN, London, United Kingdom Medical Affairs, Amgen SAS, 92100, Boulogne-Billancourt, France Real World Evidence, Certara, Evidence & Access (previously Analytica Laser), 75009, Paris, France Health Economics and Outcomes Research, Amgen GmbH, 6343, Rotkreuz, Switzerland


Introduction
Cardiovascular disease (CVD) is a major cause of global morbidity and mortality, with age-standardized CVD mortality rates in France of 275.2 and 174.1 per 100,000 people, for men and women, respectively (Townsend et al., 2016). Acute coronary syndrome (ACS), which includes unstable angina (UA) and both non-ST-elevation myocardial infarction (MI) and ST-elevation MI, accounts for half of all deaths due to cardiovascular disease (Kolansky, 2009), with observed 1-year case-fatality rates following ACS of 11.1% (Gabet et al., 2019). Treating ACS is costly: in France, the estimated yearly cost burden (including drug, procedures and hospitalizations), in 2018 Euro, is over € 1.5 billion, resulting in a yearly cost per patient of over € 10,000 (Taylor et al., 2007). Further, lost productivity costs of patients with ACS and their caregivers has been estimated to be € 13,953 per case (Kotseva et al., 2019).
Given the established relationship between reduction in lowdensity lipoprotein cholesterol (LDL-C) and decrease in the risk of suffering major vascular events (Cholesterol Treatment Trialists' , CTT), lipid-lowering therapies (LLT) are recommended in patients following an ACS by the European Society of Cardiology (ESC) and the European Atherosclerosis Society (EAS) Mach et al., 2020;Piepoli et al., 2016). Moreover, ACS patients are recommended to start treatment with high-intensity statins as soon as possible (Class I, Level A recommendation), regardless of LDL-C levels (Ibanez et al., 2018;Roffi et al., 2016).
Despite the availability of effective therapies and clinical practice guidelines for the management of dyslipidaemias, many patients, including those with ACS, do not reach LDL-C targets Reiner et al., 2016;Schiele et al., 2018). This may be due to either a lack of awareness or compliance to guidelines around the use of more intense therapies or sub-optimal patient adherence (Deshpande et al., 2017;Kotseva et al., 2016;Laufs et al., 2016). This study describes the baseline demographics, clinical characteristics and treatment patterns (intensity and adherence) of patients treated with LLT following an ACS event in France.

Design and study population
We conducted a retrospective observational cohort study using the Pharmacoepidemiologic General Research eXtension program (PGRx)-ACS registry. Patients were recruited in the PGRx-ACS registry in a prospective and consecutive manner by participating cardiology centres between 2013 and 2016 (Grimaldi-Bensouda et al., 2010). The registry includes data collected both prospectively (via physicians and patient interviews) and retrospectively (via prescription records). 60 cardiology centres participated in the PGRx-ACS registry. These centres are widely distributed across France and represent a balanced range of health care settings where ACS and cardiac death patients are seen. This methodology has been used for several other studies across numerous therapeutic areas, such as oncology, psychiatry, neurology, rheumatology, internal medicine, endocrinology, and has been validated through extensive studies (Grimaldi-Bensouda et al., 2011, 2012a,b, 2010. Patients were finally included in our study if they were ≥ 18 years old at time of index ACS (unstable angina with myocardial revascularization or MI), and if they were treated with LLT at index ACS upon hospital discharge or within following 92 days (Fig.  S1). LLTs included any of the following: nicotinic acid derivatives, fibrates, bile acid sequestrants, lipid regulating drugs (ezetimibe), statins and their combinations. Patients included in the PGRx-ACS registry were followed from index ACS for approximately 12 months until the scheduled follow-up interview, new CV event, death or lost to follow-up. If death occurred within 30 days of a new CV event or a hospitalisation for CVD event, followup was censored at date of death. If a same CV event occurred in a patient within 30 days of the index ACS, this was not considered a new CV event and follow-up was not censored.
LDL-C data was not registered during follow-up, so it was not possible to assess treatment effectiveness. However, there were patients in the study who had previous CVD documented before suffering the index ACS. These patients should have already been on LLT before suffering the index ACS to lower their LDL-C below the recommended levels. Therefore, and in the absence of follow-up LDL-C for the overall sample, baseline LDL-C values from this subgroup of patients (248) were used as a proxy to provide information on LDL-C goal achievement among patients diagnosed with CVD (1.8 mmol/L according to ESC/EAS guidelines in force at that time) (Reiner et al., 2011).
It was not an objective of the study to look at cardiovascular outcomes due to the limitations of the dataset and limited followup time frame.

Data collection
Data collection in the PGRx-ACS registry was done prospectively via a physician-completed ACS/MI electronic case report form (eCRF) and patient interviews, and retrospectively via prescription data. The ACS/MI eCRF was completed for drug use with data obtained from hospital reports (e.g. discharge letters including hospital prescriptions); treating cardiologist reports and prescriptions; and treating physician or pharmacist reports (prescriptions or lists of drugs dispensed since the index ACS). The eCRF included details on the index ACS diagnosis, ACS treatment and interventions, baseline LDL-C, and history of the following medical conditions: diabetes (type 1 and 2), hypertension, ischaemic stroke (IS), transient ischemic attack, MI, UA, chronic ischaemic heart disease, heart failure, atrial fibrillation, peripheral arterial disease (PAD), carotid artery disease, abdominal aortic aneurism, chronic obstructive pulmonary disease, and chronic kidney disease. Prescription and patient interview data provided information on drug utilization, including specific name, dose and duration, as well as BMI, alcohol consumption, smoking status and physical exercise.

Outcomes
Outcomes of interest included treatment intensity and adherence to LLT. Treatment intensity was classified as low, moderate, and high based on the American College of Cardiology and American Heart Association (ACC/AHA) guidelines (Table S1) (Stone et al., 2014). If intensity changed during the observation period, the intensity assigned was the one with the longest duration. Adherence was measured using the proportion of days covered (PDC) based on physicians' prescriptions (Andrade et al., 2006;Anghel et al., 2019;Raebel et al., 2013). Patients with a PDC ≥ 80% were considered as adherent and patients with a PDC < 80% were defined as non-adherent (Deshpande et al., 2017;Nau, 2011;Stone et al., 2014).

Statistical methods
Categorical variables were summarized by calculating the number of patients and percentages, and continuous variables were summarized by mean plus standard deviation. Demographic data, baseline clinical characteristics, treatment intensity and adherence are presented using descriptive statistics. In order to represent baseline LDL-C, the value closest to index ACS was selected for those patients with more than one measure available. LLT treatment patterns were described for the overall population.

Baseline characteristics
The PGRx-ACS registry included 3,122 patients with an ACS that may or may not have been their first ACS event. Of these patients, 2,695 met the study inclusion criteria. Patients were predominantly male (77%) with a mean age of 63.1 (SD: 12.8) years (Table 1). The most common diagnosis for the index ACS event was MI (72.7%), and most patients underwent a percutaneous transluminal coronary angioplasty (81.3%) as an ACS intervention (Table 1). Patients had a mean BMI of 27.0 (SD: 4.4); among those reporting data on physical activity, nearly 60% (n = 954/1,696) reported doing more than 30 minutes of physical activity per day. Patients had a mean Charlson Comorbidity Index (CCI) of 2.7 (SD: 2.3) ( Table 1).

Baseline LDL-C
LDL-C values at baseline were available for 77% of patients (2,076/2,695), with a mean LDL-C of 2.9 mmol/L (SD: 1.2) (Supplementary material Table S2). Among the study population, there was a trend of lower baseline LDL-C in patients with older age at time of index ACS (Fig. 1). A Pearson correlation method was used to evaluate the relationship between age and baseline LDL-C. The correlation coefficient (95% CI) between age and LDL-C was equal to -0.20 (-0.24; -0.16) suggesting that there was a small negative linear relationship between age and LDL-C (i.e. the older the age of the patient, the lower was the baseline LDL-C).
There was a sizeable amount of patients presenting with very high LDL-C values, with 4.6% of patients with LDL-C ≥ 5 mmol/L, 7.6% of patients with LDL-C ≥ 4.5 mmol/L and 15.5% of patients with LDL ≥ 4.0. The proportion of patients with high LDL-C was higher among younger patients (Table S2). This proportion of patients with high LDL-C values may indicate a relatively high proportion of patients with familial hypercholesterolemia (FH) within those presenting with an ACS. We used the Dutch Lipid Clinic Network Score (DLCNS) to identify patients with "possible'' familial hypercholesterolemia. According to the DLCNS, patients with a baseline LDL-C ≥ 5 mmol/L are qualified as "possible'' familial hypercholesterolemia (FH) irrespective of presence of other risk factors (WHO Human Genetics Programme, 1999). Additionally, patients presenting with premature coronary artery disease (premature is defined as younger than 55 in men and younger than 60 in women) and with a baseline LDL-C ≥ 4.0 are also classified as possible FH case. In our study, around 61% (61.5%, 139/226) of patients with an LDL-C between 4.0 and 5.0 were younger than 60 years old (Table S2). With this, we estimated that in our study around 11% of patients presenting with an ACS were possible FH patients. This is probably an underestimate since most of the criteria included in the DLCNS to identify FH could not be verified with an information available in the dataset and some of the patients were already treated with LLTs. The majority of patients with a baseline LDL-C ≥ 5 mmol/L (71.6%) were not receiving any previous LLT treatment at the time of suffering the ACS (Table S3).

Treatment intensity and adherence
The majority of patients were started on a high (56.4%, 1,520/2,695) or moderate (36.0%, 971/2,695) intensity statin, whereas only 3.2% of patients (86/2,695) were on low intensity statins. A further 1.7% (46/2,695) were on statin/ezetimibe combination, 1.6% (42/2,695) on other LLT and 1.1% (30/2,695) on ezetimibe monotherapy alone. Age and baseline LDL-C were drivers of treatment intensity (Fig. 2). The use of high intensity statins decreased with age: 69.2% of patients aged < 50 and 36.2% of patients ≥ 80 were on a high-intensity statin. Conversely, the use of high intensity statins increased with baseline LDL-C: 51.1% among those with an LDL-C of < 1.8 mmol/L, and 75.3% among those with an LDL-C of ≥ 4.5 mmol/L were on a high intensity statin. Few patients were on a statin plus ezetimibe or ezetimibe alone (Fig. 2). The relationship between age and baseline LDL-C with the use of high intensity statins was assessed with the Pear-

Fig. 2. Representation of the proportion of patients treated with each LLT category following an ACS, according to age (A) and according to baseline LDL-C (mmol/L) (B).
son Correlation method. The correlation coefficient (95% CI) between age and treatment intensity was equal to -0.19 (-0.23, -0.16) suggesting that there was a very weak negative linear relationship between age and treatment intensity (lower use of high intensity therapies in older patients). The correlation coefficient (95% CI) between baseline LDL-C and treatment intensity was equal to 0.10 (0.05; 0.14) suggesting that there was a very weak positive linear relationship between baseline LDL-C and treatment intensity (higher use of high intensity therapies in patients presenting with higher baseline LDL-C).
Around 70% of patients were considered adherent (defined as PDC ≥ 80%) to LLT within one year of follow-up. A higher proportion of adherent patients was observed within those treated with moderate intensity statins (75.6%) as compared to those on low (62.5%) or high intensity statins (66.5%); the Pearson correlation coefficient was -0.052 (-0.090; -0015) showing a negligible correlation between treatment intensity and adherence. A relationship between age and treatment adherence was not that clear, although patients older than 80 years old show the lowest proportion of adherent patients (55.6%); the Pearson correlation coefficient was -0.052 (-0.034; 0.042) showing that correlation was not significant (Table S4).

Subgroup of patients with a previous CVD event
Prior to index ACS, a total of 310 patients had a documented CVD event (237 with previous MI, 49 with a previous IS, 15 with a previous MI and IS and 9 with documented PAD); of these, 248 had baseline LDL-C values available. At time of index ACS, 67.3% (167/248) of patients were already being treated with LLT (prevalent LLT users); and the remaining 32.7% (81/248) initiated LLT treatment for the first time following the index ACS event (incident LLT users). Just over 31% of patients (79/248) had reached a LDL-C level < 1.8 mmol/L (Table 2).

Discussion
In this study, we found that a substantial proportion of patients in France were not treated with high intensity statins after suffering an ACS (even fewer patients were on ezetimibe/statin combination) and only 70% of patients were defined as adherent within the follow-up period. Conversely, very few patients (around 3%) were on low intensity statins following an ACS. It's also noteworthy that in patients who had been previously diagnosed with CVD, almost one third of them were not receiving statin therapy at the time of suffering the index ACS.
ESC/EAS guidelines recommend first-line therapy with statin LLTs to reduce the risk of CVD outcomes and to start highintensity statins as soon as possible for patents hospitalized with Table 2. Treatment intensity and baseline LDL-C among patients with previous CVD history*. ACS, irrespective of their LDL-C level Mach et al., 2020;Piepoli et al., 2016;Schiele et al., 2018). At the time of this study, recommendations stated target levels of LDL-C of less than 1.8 mmol/L (Reiner et al., 2011), and 32% of patients met this target. More recent guidelines have lowered the LDL-C target to 1.4 mmol/L; based on this updated recommendation, only 15% (38/248) would have met the target levels (Mach et al., 2020). This low utilization of high intensity statins leaves patients with an ACS at considerable risk of subsequent cardiovascular events (Boklage et al., 2018). In France, a previous real-world effectiveness study done with the PGRx registry demonstrated that statin use was associated with decreased risk of first non-fatal ACS (adjusted odds ratio, 0.67) (Grimaldi-Bensouda et al., 2013).

LDL-C (mmol/L) at LLT Treatment Users baseline, N (%) All Incident users Prevalent users
Patient adherence in our study was sub-optimal, with 70% of patients defined as adherent within one year of follow-up. Previous studies have found that treatment adherence has as much of an impact on reducing the risk of CVD as treatment intensity (Khunti et al., 2018). Strategies to increase treatment adherence among patients should be explored to increase the effectiveness of LLT.
In the present study, we identified a large proportion of patients potentially representing patients with FH. Despite the increased risk of suffering from CVD in this group of patients, most of these patients remained untreated. These findings are consistent with results from another study that found that over one-quarter of FH patients in France may not be receiving LLT and only 13% of those treated with LLTs received high intensity statins (Berard et al., 2019). FH is associated with a 2-fold increased risk of coronary event recurrence at one year following an ACS event (Nanchen et al., 2016). Efforts should be undertaken to identify and adequately control young patients with potentially high LDL-C, as they are at higher risk of excess morbidity and mortality related to dyslipidaemias (Schiele et al., 2018).
Several studies have examined treatment patterns of LLT in a French population. Results from a multicentre European survey, including centres in France, found that there are patients with very-high risk who remain untreated: 1 out of every 10 patients discharged from the hospital after a coronary event did not receive any statin treatment . Ferrières et al. examined LLT utilization and lipid goal attainment in a 2015 population of French patients with atherosclerotic CVD disease with or without diabetes, including patients recently diagnosed with an ACS . Similar to our study, they found that LLT usage and LDL-C goal achievement were suboptimal in recent ACS patients with only 75% prescribed an LLT therapy and 29% of patients achieving LDL-C levels < 1.8 mmol/L . Though the prescribing environment described in Ferrières et al. was a general practice setting, differing from our database informed by cardiologists, the results are consistent in highlighting that patients with ACS remain undertreated.
Several hypotheses have been proposed to explain the discordance between guideline recommendations and what is observed in practice. It may be possible that clinicians are either not aware of guideline recommendations or are avoiding higher-intensity regimens due to concerns about possible side effects (Laufs et al., 2016). Patients' attitude towards statin therapy, as influenced by the media, and thus the decision to discontinue or continue statin therapy, may result in observed lower prevalence of LLT therapy (Nielsen and Nordestgaard, 2016). Despite the unfavourable context in France for statins for cardiovascular prevention during the study period (Bezin et al., 2014), we found overall adherence to be 70% of patients, similar to what was observed in other studies (Danese et al., 2017;Naderi et al., 2012).
This study's key strength is that it uses a real-world dataset, the PGRx-ACS registry, which reflects cardiologist practice conditions in France. Previous studies using these networks have shown that cases and referents are very representative of the population observed in clinical practice (Grimaldi-Bensouda et al., 2010). Further, demographic and clinical characteristics from the current study are also similar to those reported in the EURHOBOP study, a benchmarking study of ACS management in Europe, including France (André et al., 2014). Despite using patient-reported data, the PGRx database has also shown good concordance between physician and patient report of CVD drug utilization (Grimaldi-Bensouda et al., 2010).
Our study also has limitations. The study may not be generalizable to all French patients with ACS since patients in PGRx are recruited by cardiologists and not from a general practice setting; however, in clinical practice, the majority of ACS patients are treated by cardiologists in the first months following the event (?). Missing data are a reality in real-world secondary databases; in this study nearly a quarter of patients had missing LDL-C values at baseline. The proportion of missing data, however, did not appear to differ according to age. Finally, we only had access in our study to limited information (mainly baseline LDL-C) as compared to all the clinical data available to the physician when confronted to the prescription decision. Hence, clinicians may be guided in clinical practice by other sources of data such as imaging at the time of the ACS (multivessel disease vs single lesions) or other forms such as OCT (optical coherence tomography) of other non-culprit lesions. The information provided by these techniques may lead to the clinician to place a patient at an even higher level of risk and to strive to an even lower LDL-C goal with the use of additional and more intensive lipid lowering therapies.
Managing ACS correctly is critical as adherence to guidelines is associated with better patient prognosis in both the acute phase of the syndrome and in longer term secondary prevention (Sabouret et al., 2010). Based on the data observed in this study, there is potential to improve management of patients with ACS, thereby reducing mortality and higher case-fatality rates. Strategies to better identify at-risk patients along with improving com-pliance with dyslipidemia guidelines, treatment adherence and optimization of therapy, may lead to better LDL-C goal attainment and improved CVD outcomes among patients hospitalized for ACS in France, particularly for those with possible FH and those with a prior history of CVD.