Benecial Effects of Fixed-Dose Combination of Amlodipine and Atorvastatin in Patients with Concomitant Hypertension and Hypercholesterolemia: A Multi-Institutional Cohort Study

Blood pressure (BP) and low-density lipoprotein cholesterol (LDL-C) are important risk factors for cardiovascular (CV) diseases and treating these factors simultaneously is recommended by current guidelines but only short-term clinical results were available. The objective of this study was to examine the longer-term ecacy and safety of xed-dose combination (FDC) versus free combination of amlodipine and atorvastatin in patients with concomitant hypertension and hypercholesterolemia. signicantly changed by this aggressive treatment strategy. MASD, maximum absolute standardized difference; ACEi, angiotensin converting enzyme inhibitors; ARBs, angiotensin receptor blockers; HTN, hypertension; GLP-1 RA, glucagon-like peptide-1 receptor agonist; SGLT2i, sodium-glucose co-transporter 2 inhibitor; LDL, low density lipoprotein cholesterol; HDL, density lipoprotein density lipoprotein This multi-institutional retrospective study is the rst study to evaluate the mid-term and FDC versus free combination of amlodipine and atorvastatin in patients with concomitant hypertension and hypercholesterolemia. During the 30-month follow-up period, we found no signicant difference in the composite CV outcome among the three study groups. FDC of amlodipine 5 and atorvastatin 20 mg resulted in a greater reduction in LDL-C than the other two regimens but the HbA1c level was not signicantly different.


Introduction
Hypertension and hypercholesterolemia are two important modi able risk factors for cardiovascular disease (CVD). They have been reported to coexist in up to 30% of patients with CVD [1,2], and their synergistic effect on cardiovascular mortality is greater than each condition alone [3]. Therefore, current clinical guidelines recommended treating these risk factors simultaneously rather than in isolation [4,5].
However, the increased pill burden when prescribing antihypertensive and lipid-lowering therapy concomitantly may have a negative impact on drug adherence [6], which may then attenuate the bene cial effects of the simultaneous treatment strategy.
Relatively few studies have compared the e cacy, adherence and interaction between FDC and free combination strategies in patients with two different diseases, such as hypertension and hypercholesterolemia. Among these studies, the follow-up periods ranged from only 6 weeks to 6 months [13][14][15][16][17][18][19][20][21][22]. In our previous work, we demonstrated improved clinical outcomes with the use of FDC of amlodipine and atorvastatin in patients with concomitant hypertension and dyslipidemia compared to a free-equivalent combination (FEC), including major adverse cardiovascular events, hospitalization for coronary artery disease and newly initiating hemodialysis [23]. However, these results were generated from the National Health Insurance Research Database (NHIRD) of Taiwan, which is a large administrative database that does not contain personal data such as smoking, body weight, blood pressure (BP) records or laboratory data. Therefore, the e cacy of lowering BP and cholesterol, and the safety pro les such as blood sugar, renal and liver function could not be estimated.
In the present study, we aimed to analyze the longer-term e cacy and safety of FDC versus free combination of amlodipine and atorvastatin in patients with concomitant hypertension and hypercholesterolemia registered in a real-world, multi-institutional, electronic medical record (EMR) database.

Data source
The data used in this study were retrospectively obtained from the Chang Gung Research Database (CGRD), which is a multi-institutional, de-identi ed standardized EMR database maintained by the Chang Gung Memorial Hospital (CGMH) organization, and also the largest such database in Taiwan [24,25].
The CGMH organization is currently the largest medical system in Taiwan, comprising two medical centers, two regional hospitals and three district hospitals, with a total of 10,070 beds, more than 280,000 admissions, 8,500,000 outpatient visits and 500,000 emergency department visits a year [25].
The CGRD contains more clinical details than administrative claims databases, including pathological reports, laboratory results, procedure reports, smoking habit, vital sign records and body mass index (BMI). Diseases were recorded using International Classi cation of Diseases, Ninth Revision, Clinical Modi cation (ICD-9-CM) codes before 2016, and ICD-10-CM codes thereafter. The research was performed in accordance with the Declaration of Helsinki in 1964. It was approved and the need for informed consent was waived by the ethics committee of Institutional Review Board of CGMH, Linkou, Taiwan (committee's approval number: 202100864B0) because of the retrospective design of the study and anonymized clinical data.

Study cohort and design
We identi ed patients diagnosed with hypertension in the CGRD from September 1st, 2016 to September 30th, 2019 who had prescriptions of FDC or free combinations of amlodipine with atorvastatin ( Figure 1).
The only available dosages of amlodipine/atorvastatin FDC in Taiwan are amlodipine 5 mg with atorvastatin 10 or 20 mg, and both dosages of FDC were available at CGMH from September 2016. To avoid the potential confounding of previously prescribed medications, we only included patients with a rst prescription of either FDC or free combination of amlodipine and atorvastatin. The date of the rst prescription of the studied medication was de ned as the index date.
Patients who received any form of dihydropyridine calcium-channel blockers (DCCB) or statins before the index date were excluded from this study. To evaluate the long-term e cacy, we excluded patients who developed cardiovascular (CV) outcomes within 3 months after the index date or whose follow-up period was less than 90 days. To ascertain the long-term use of the studied drugs, we also excluded patients who switched drugs or received the treatment medication for less than 60 days within 3 months after the index date. Other exclusion criteria were an age less than 18 years, a diagnosis of liver cirrhosis, those undergoing dialysis and those with heart failure before the index date. After exclusion, three study cohorts were generated. The rst cohort consisted of patients who received FDC of amlodipine 5 mg and atorvastatin 10 mg (Fixed 5/10 group), the second received FDC of amlodipine 5 mg and atorvastatin 20 mg (Fixed 5/20 group), and the third received free combination of amlodipine 5 mg plus atorvastatin 10 mg (Free 5/10 group). The baseline characteristics and clinical outcomes of these three cohorts were compared.

Covariates
Covariates were obtained from the CGRD including age, sex, BMI, smoking status, CVD (including coronary artery disease, peripheral artery disease, acute coronary syndrome or stroke), comorbidities, Charlson's Comorbidity Index (CCI) score, concomitant medications, vital signs (o ce BP and heart rate) and laboratory data. Comorbidities included diabetes mellitus, chronic kidney disease, atrial brillation, malignancy and chronic obstructive pulmonary disease. The presence of CVD and comorbidities was con rmed if the patients had at least one inpatient or two outpatient diagnoses before the index date.
Concomitant medications, BMI, vital signs and laboratory data were extracted from the EMRs within 3 months before or after the index date.

Outcomes
The primary outcome was composite CV outcome, including all-cause death, coronary intervention, acute myocardial infarction (MI) and stroke. Coronary interventions were identi ed by inpatient procedure codes of percutaneous coronary intervention or coronary arterial bypass grafting. Acute MI was de ned as having a principal inpatient diagnosis of MI with an elevated cardiac troponin level above the 99th percentile upper reference limit during hospitalization. Stroke was de ned as having a principal inpatient diagnosis and an image (computed tomography or magnetic resonance imaging) showing stroke.
Information on deaths was obtained from the sub-database of death certi cates in the CGRD.
The secondary outcomes were renal, safety and laboratory/BP outcomes. The renal outcomes included a decline in eGFR of more than 40%, newly initiating dialysis, the composite of both outcomes, and allcause death. The safety outcomes included new-onset diabetes mellitus (NODM) and abnormal liver function. NODM was identi ed as having newly diagnosed diabetes mellitus and an HbA1c level greater than 6.5% during follow-up. Abnormal liver function was de ned as an elevation in ALT level of more than three times the upper reference limit, namely greater than 105 U/L. Laboratory outcomes were long-term LDL-C and HbA1c levels during follow-up.
Medication adherence was assessed by using the proportion of days covered (PDC) according to the EMRs, which was de ned as the total number of days covered by the study drugs divided by the total number of follow-up days [7,9,13,23,26]. The follow-up period started from the index date of the rst prescription of the study drug until the date of an outcome, death, the date of switching among the studies drugs, the last visit date in the CGRD, or the end of the study period (September 30th, 2019), whichever occurred rst.

Statistical analyses
The distribution of baseline characteristics among the three study groups (Fixed 5/10 vs. Fixed 5/20 vs. Free 5/10) was balanced by using generalized boosted modeling-inverse probability of treatment weighting (GBM-IPTW) based on propensity scores with 10,000 trees [27]. The propensity scores were calculated based on all of the baseline characteristics, except that the follow-up year was replaced with the index date. Baseline characteristic data that were missing were imputed using a single expectationmaximization algorithm before conducting GBM-IPTW. The balance among the three study groups before and after GBM-IPTW was assessed by using the maximum absolute standardized difference (MASD), and an MASD less than 0.2 was considered to indicate good balance among the groups [27].
The risk of fatal outcomes (i.e., all-cause death, composite CV outcome) among the three study groups was compared by using a Cox proportional hazard model. The incidence of non-fatal outcomes (i.e., decline in renal function) among the three study groups was compared using a Fine and Gray subdistribution hazards model, which considered all-cause death during follow-up as a competing risk. The study groups were the only explanatory variables in the aforementioned survival analyses. A subgroup analysis of primary CV outcomes was further strati ed by prior CVD. Changes in laboratory data and BP from baseline to long-term follow-up among the three study groups were compared by using a generalized estimating equation which contained the intercept, main effects of the study groups and time (treated as a continuous variable) and an interaction effect of the study groups by time. Changes between groups were considered to be signi cantly different when the interaction was statistically signi cant.
A two-sided P value less than 0.05 was considered to be statistically signi cant. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC), including the "PHREG" procedure for conducting the survival analysis and the "TWANG" macro for estimating GBM-IPTW [27].

Baseline characteristics
A total of 16,156 hypertensive patients with prescriptions of the studied drugs were identi ed in the CGRD during the study period. After exclusion, 1,788 patients were eligible for further analyses, including 787 patients in the Fixed 5/10 group, 458 patients in the Fixed 5/20 group, and 543 patients in the Free 5/10 group (Fig. 1). The mean age of all patients was 60 ± 12.2 years, and 53.9% were male.
Medication adherence rates assessed by PDC after IPTW were 58 ± 35% in the Fixed 5/10 group, 62.6 ± 39.4% in the Fixed 5/20 group and 58.8 ± 45.9% in the Free 5/10 group, which were not signi cantly different (MASD 0.087; data not shown). Compared to the patients in the other two groups, those in the Free 5/10 group were older, had lower BMI, higher prevalence of stroke and higher CCI score (MASD >0.2; Table 1). Regarding the baseline concomitant medications, patients in the Free 5/10 group received more anti-platelet agents but fewer SGLT2is (MASD 0.272 and 0.238, respectively) than those in the other two groups, whereas the Fixed 5/20 group received more ARBs (43.9%, MASD 0.277). Both baseline systolic and diastolic BP were signi cantly higher in the Fixed 5/20 group than in the other groups. The Fixed 5/20 group also had higher LDL-C, non-HDL-C, total cholesterol and ALT levels. After imputation and GBM-IPTW, all covariates at baseline were well-balanced with no signi cant differences among the three study groups (all MASD values <0.2; Supplemental Table S1). The maximum follow-up period in this study was 30 months, and the mean follow-up periods were 1.7 ± 0.9 years in the Fixed 5/10 group, 1.6 ± 0.8 years in the Fixed 5/20 group, and 1.7 ± 0.9 years in the Free 5/10 group, respectively (MASD 0.133).

Clinical outcomes
The number of events in each study group in the original cohort before GBM-IPTW is listed in Supplemental Table S2. After imputation and GBM-IPTW, the risk of clinical outcomes was compared among three study groups. The risk of composite CV outcome was not signi cantly different among the three groups (6.1% in the Fixed 5/10 group, 6.3% in the Fixed 5/20 group and 6.0% in the Free 5/10 group) as shown in Table 2 and Fig. 2. The results showed that the risks of each component of the composite CV outcome did not differ among groups. The incidence of composite renal outcome was also comparable among the groups, including eGFR decline >40%, newly initiating dialysis or all-cause mortality (10.4% in the Fixed 5/10 group, 11.1% in the Fixed 5/20 group and 11.5% in the Free 5/10 group). We further analyzed the composite CV outcome among the three study groups in patients with or without previously established CVD as primary and secondary prevention, which disclosed comparable results (P for interaction = 0.332) (Fig. 3).

Laboratories and BP changes
We further evaluated the longitudinal changes of LDL-C, HbA1c and BP among the three groups. The mean levels and changes from baseline in the laboratory data and BP at different time points are provided in Supplemental Table S3.  (Fig. 4a).
We also evaluated HbA1c changes among the three groups and no signi cant effects were observed (Fig. 4d).

Discussion
This multi-institutional retrospective study is the rst study to evaluate the mid-term e cacy and safety of FDC versus free combination of amlodipine and atorvastatin in patients with concomitant hypertension and hypercholesterolemia. During the 30-month follow-up period, we found no signi cant difference in the composite CV outcome among the three study groups. FDC of amlodipine 5 mg and atorvastatin 20 mg resulted in a greater reduction in LDL-C than the other two regimens but the HbA1c level was not signi cantly different.
In our previous study, we demonstrated that FDC regimen of amlodipine and atorvastatin improved composite CV outcomes compared to FEC of the same medications in patients with newly diagnosed hypertension and dyslipidemia during a 5-year follow-up period [23]. In that study, the medication adherence as assessed by PDC was better in the FDC than in the FEC group (0.49 ± 0.26 versus 0.32 ± 0.3, P < 0.001), and this may explain the results. However, based on the nature of the large administrative NHIRD, the major limitations of the study were a lack of possible confounding variables and e cacy parameters including BP and LDL-C changes. In the current study, we analyzed the e cacy and safety of FDC of amlodipine and atorvastatin by using data from the CGRD, a real-world, multi-institutional, standardized EMR database. In all previous studies, drug compliance with FDC is always better than free combination, which is the main explanation for the bene cial clinical effects of FDC [7, 9-11, 13, 14, 23]. However, the medication adherence as assessed by PDC in the current study was not signi cantly different among the three study groups, which may be related to the stricter study criteria compared to previous studies that we excluded patients who switched drugs or received the study medications for less than 2 months within 3 months after the index date. Subsequently, the remaining patients, especially those in the free combination group, may have had better drug compliance and tolerability which may explain the comparable medication adherence between the FDC and free combination groups in the present study. Moreover, the follow-up period and patient number were relatively limited that may have further resulted in the similar composite CV outcome among the three study groups.  -30], and it is thus reasonable to rst prescribe or early up-titrate to Fixed 5/20 rather than Fixed 5/10 or Free 5/10 in order to achieve better BP and LDL-C control concomitantly.
The risk of NODM with statin therapy has been shown to be positively correlated with the strength of the statins [31], with a reported overall risk of 9% [32]. In the current study, we also found that the risk of reported NODM was highest in the Fixed 5/20 group and lowest in the Free 5/10 group, and this may be explained by the different strengths of the statins. However, there were no signi cant differences among the three study groups with regards to HbA1c level during follow-up. The reason for this discrepancy may be multifactorial, such as the different diagnostic criteria for diabetes among physicians, not routinely checking HbA1c level in all patients or by chance. On the other hand, statin-associated liver toxicity is well established, however there are very few reports of liver failure directly attributed to statins [33,34]. In the current study, we also found an increased risk of ALT elevation in the higher strength statin (Fixed 5/20) group. However, it should be emphasized that the bene cial effects of statins on CVD outweigh the risk of NODM development or mild liver function abnormalities, and therefore adequate dosages of statin should be prescribed if indicated to improve CV outcomes. Myopathy, myalgia and fatigue are also possible adverse effects of statin but it was not routinely examined or reported in our database.
Adequate BP and LDL-C control are recommended by clinical guidelines both in primary and secondary CVD prevention [29,35,36]. In our previous study, we only demonstrated the bene cial effect of reducing major adverse cardiovascular events with FDC of amlodipine/atorvastatin in the primary prevention setting [23]. In the present study, we further analyzed the differences in e cacy among three study groups with regards to primary and secondary CVD prevention, and the results were comparable in composite CV outcome. The major adverse cardiovascular event rates were not signi cantly different among the three study groups both in primary and secondary prevention, which may be due to relatively homogenous drug compliance among the groups, shorter follow-up period, different population and study design.

Study limitations
This study was based on a multi-institutional standardized EMR database and has several limitations. First, although the CGRD is the largest EMR database in Taiwan and covers almost 10% of the entire population, we could not collect the clinical events that developed in hospitals that were not involved in the CGRD, which may have led to underestimation of the actual event rates. Meanwhile, we used IPTW to balance the confounding medications but any additional drugs from other institutes could not be obtained. However, this should be balanced among the three study groups, and the between-group comparisons should still be reasonable. Second, the BP values used in the present analysis were based on o ce BP records, which may not represented home BP and ambulatory BP monitoring. Therefore, we could not rule out the possibility of white-coat or masked hypertension. Third, we used PDC as a surrogate marker of medication adherence but we could not ensure that the patients consumed the medications accordingly, and therefore drug compliance may have been overestimated. Finally, this is a retrospective, non-randomized study, and the results may be confounded by other unmeasured factors.
Therefore, the results should be interpreted with caution.

Conclusions
In this retrospective, EMR-based study, FDC of amlodipine and atorvastatin, especially the regimen with a higher dosage of statins, signi cantly reduced the mid-term LDL-C level compared to free combination in patients with concomitant hypertension and hypercholesterolemia. HbA1c control during the follow-up period was not compromised by this aggressive treatment strategy.  The IPTW-adjusted cumulative incidence of composite cardiovascular outcomes in the FDC of amlodipine 5 mg/atorvastatin 10 mg (green line), FDC of amlodipine 5 mg/atorvastatin 20 mg (blue line), and free combination of amlodipine 5 mg/atorvastatin 10 mg (red line) groups. The cumulative incidence was derived from Kaplan-Meier estimate. IPTW, inverse probability of treatment weighting; FDC, xeddose combination; HR, hazard ratio; CI, con dence interval. Figure 3