Evaluation of Two Highly Effective Lipid-Lowering Therapies in Subjects With Acute Myocardial Infarction: A Lipidomic Analysis

Introduction For cardiovascular disease prevention, statins alone or combined with ezetimibe have been recommended to achieve low density lipoprotein - cholesterol targets, but their effects on other lipids are less reported. Objective To examine lipid changes in subjects with ST-segment elevation myocardial infarction (STEMI) after two highly effective lipid-lowering therapies. Methods Twenty patients with STEMI were randomized to be treated with rosuvastatin 20 mg qd or simvastatin 40 mg combined with ezetimibe 10 mg qd during 30 days. Fasting blood samples were collected in the rst day (D1) and after 30 days (D30). Lipidomic analysis was performed using the Lipidyzer platform. Selected reaction monitoring (SRM) was used in positive and negative ionization modes with and without differential mobility spectrometry (DMS). Univariate and multivariate analyses were performed. Results Comparable classic lipid prole was observed in both groups of lipid-lowering therapies at D1 and after treatments. However, differences in other lipids were observed between groups at these time points. After treatments, main differences between groups were for lysophosphatidylcholine and triacylglycerides. Conclusion Despite similar changes in the classic lipid prole, differences in lipidomic analysis were found before and after exposure to highly effective lipid-lowering therapies in subjects after STEMI. Trial registration: registered 28/09/2014.


Introduction
Therapies aiming to reduce LDL-C substantially changed the natural history of cardiovascular disease (CVD), especially coronary heart disease (CHD) 1 . However, despite the achievement of very low levels of LDL-C, recurrent events after acute coronary syndromes are still observed suggesting that other lipid components, including FFAs may contribute to the atherothrombotic disease 2,3,4 . Interestingly, a prospective cohort of individuals with peripheral artery disease showed that the occurrence of myocardial infarction was inversely related to baseline serum concentrations of phosphatidylcholine (PC) species 5 .
In addition, decreased plasma levels of plasmalogens were reported in subjects with CHD 6 . In fact, their role in atherosclerosis is still poorly understood, involving possibly antioxidant properties that impact the movement of molecules in and out of the cells, as plasmalogens are components of cell membranes 7 .
High free fat acids (FFA) concentrations can induce activation of NLRP3 in ammasome, triggering a proin ammatory response related to atherosclerosis, and lipid-lowering therapies combinations, such as statin combined with the inhibitor of intestinal cholesterol absorption (simvastatin plus ezetimibe), can partially revert many in ammatory biomarkers 8,9,10 . Following rosuvastatin treatment, a lipidomic study revealed signi cant decrease in sphingomyelin (SM), triglycerides (Tg), phosphatidylinositol (PI) and phosphatidylethanolamines (PE) levels, but for lysophosphatidylcholines (LPC) and phosphatidyl cholines (PC), no signi cant changes were reported 11 . Despite effective achievement of LDL-C and non HDL-C targets by the use of less potent statin combined with ezetimibe, their effects in other lipids are less reported. Therefore, this study aimed to compare lipid composition in the plasma of subjects with very high cardiovascular risk with STEMI at baseline and after 30 days of exposure to two highly effective lipid-lowering therapies (rosuvastatin alone or simvastatin combined with ezetimibe).

Study design
This prospective, randomized, open label study was delineated to evaluate differences in the composition of lipids in patients with STEMI at D1 and at D30 after implementing the two lipid-lowering therapies (rosuvastatin 20 mg [Crestor®, AstraZeneca] or simvastatin 40 mg combined with ezetimibe10 mg [Vytorin, MSD]). Patients were randomized using a central computerized system. Plasma of patients were categorized in four groups: patients in the rosuvastatin group at D1 (G1) and at D30 (G2); patients treated with simvastatin plus ezetimibe at D1 (G3) and at D30 (G4). Investigators involved in the biochemical and lipidomic analyses were unaware of the lipid-lowering treatment.

Cohort
The study included mainly middle-aged males, approximately half of them with type 2 diabetes. Table S1 shows main characteristics of study population. The cohort is part of the B And T Types of Lymphocytes Evaluation in Acute Myocardial Infarction (BATTLE-AMI) study 39 (ClinicalTrials.gov, NCT02428374, registered on 28/09/2014). They had no prior MI and were naive for lipid-lowering treatment. All patients were submitted to pharmacological thrombolysis with tenecteplase in the rst 6 h of STEMI, followed by coronary angiogram and percutaneous intervention when needed in the rst 24 h of STEMI (pharmacoinvasive strategy). Key exclusion criteria included hemodynamic instability, autoimune disease, known malignancy, pregnancy and signs of active infections. The same sample size of the BATTLE-AMI trial was previsouly published 39 and, for this sub analysis, a convenient sample was used. The study protocol was approved by the local ethics committee (Escola Paulista de Medicina -UNIFESP IRB 0297/2014; CAAE: 71652417.3.0000.5505), which follows the Declaration of Helsinki, and written informed consent was provided by all subjects before their inclusion. Fasting blood samples were collected, in the morning at D1 and at D30 after lipid-lowering therapy, in tubes containing EDTA, followed by centrifugation at 1300 g for 15 min, at room temperature and storage at -80 ºC before analysis.

Clinical measurements
All samples for general biochemical tests, including the classic lipid pro le were performed in the Central Laboratory of the University Hospital and the LDL-C was estimated by the Friedewald equation.

Sample preparation
Exactly 100 μL of plasma were transferred to a borosilicate glass culture tube (16 × 100 mm). Next, 900 μL water, 2 mL methanol, and 900 μL dichloromethane were added to all samples and the mixture was vortexed for 5 s. Samples were left to incubate at room temperature for 30 min. Next, another 1 mL water and 900 μL dichloromethane were added to the tube, followed by gentle vortexing for 5 s, and centrifugation at 2500 g at 15 °C for 10 min. The bottom organic layer was transferred to a new tube and 1.8 mL dichloromethane were added to the original tube for a second extraction. The combined extracts were concentrated under nitrogen. Exactly 100 μL of the isotope labeled internal standards mixture were added to the dried extract and another 30 min incubation was allowed until equilibrium is reached.

Sample analysis
Quantitative lipidomics was performed with the Sciex Lipidyzer platform con gured by an ExionLC AD instrument (Sciex) coupled to a QTRAP® 5500 mass spectrometer (Sciex) equipped with SelexION® for differential mobility spectrometry (DMS). The solvent 1-propanol was used as the chemical modi er for the DMS. Samples were introduced to the mass spectrometer by ow injection at 8 µL min -1 . Each sample was injected twice, with the DMS on (PC/PE/LPC/LPE/SM) and off (CE/CER/DAG/DCER/FFA/HCER/LCER/TAG). Lipid molecular species were analysed by selected reaction monitoring (SRM) in both electrospray ionization modes using positive/negative polarity switching.
Positive ion mode was used to detect the lipid classes SM/DAG/CE/CER/DCER/HCER/DCER/TAG and the negative ion mode to detect the lipid classes LPE/LPC/PC/PE/FFA. All data obtained from the Lipidyzer Platform were automatically processed in the Lipidomics Work ow Manager (LWM). Signals of all lipids and fatty acids obtained for each sample were quanti ed with their appropriate internal standards using the Lipidyzer platform, which allowed for automated data acquisition, data processing (S2). Data acquisition and quanti cation of lipids was performed using the Lipidyzer™ platform. A detailed description of the quantitation process can be found in previous references 40,41,42,43,44 .
Quality control (QC) consisted of a standard plasma sample was obtained from the Lipydizer kit. The reconstituted lyophilized plasma was extracted following the procedure described previously. The QC sample was injected ve times at the beginning of the randomized sample batch, every 10 injections and, at the end of the sample batch.
From the original generated table compiling the lipids and FFAs identi ed, only the ones that had presented concentrations (µmol L -1 ) different of zero were used for data treatment (S2). The discriminant lipids of the four groups were obtained from multivariate analysis by OPLS-DA loadings plot (UV-scaling and CV-ANOVA evaluation), by variable importance in projection (VIP) evaluation, and/or univariate analysis by t-test and q-value (FPR) 45,46 with 95% con dence interval.

Classic lipid parameters
Prior to the lipidomic study, a classic lipid pro le was obtained. Fig. 1 shows the box plots of measurements of total cholesterol (TC), high-density lipoprotein -cholesterol (HDL-C), low-density lipoprotein -cholesterol (LDL-C) and triglycerides (Tg) in each group of samples. As expected mainly changes were observed for TC and LDL-C.  (Fig. 2b). Comparing the mean values in each group, it is possible to observe that all metabolites were decreased in both groups after the exposure to the treatments at D30, except for LPC, LPE and TAG.

Correlation studies
Levels of LDL-C, HDL-C, TC, and Tg were correlated with levels of lipid classes (CE, CER, FFA, LPC, LPE, PC, PE, SM, and TAG) for all studied patient groups (Fig. 3). Fig. 3a presents the correlations between the two therapies at D1 of STEMI. Opposite correlations between the clinical parameters and some lipids (PE, TAG, LPE, and CE) were observed.
Considering rosuvastatin administration, PE was negatively correlated with CT, HDL-C and LDL-C and positively correlated with Tg; TAG was negatively correlated with HDL-C, but positively correlated with Tg; LPE was negatively correlated with LDL-C and positively correlated with HDL-C; and CE was positively correlated with Tg. Fig. 3b depicts correlations between the two therapies at D30. Opposite correlations between the clinical parameters and some lipid classes were also observed. However, besides PE, TAG, LPE, and CE, PC, SM, and CER also presented contrasting correlations with the clinical parameters at D30.
Considering simvastatin plus ezetimibe administration, PE was negatively correlated with CT and HDL-C; TAG was negatively correlated with TC and HDL-C, but positively correlated with Tg; LPE was negatively correlated with HDL-C; CE was positively correlated with TC and LDL-C and Tg, but positively correlated with Tg; PC was negatively correlated with CT and HDL-C; SM was negatively correlated with TC, HDL-C and LDL-C; and CER was positively correlated with Tg. Figure 3 shows different behaviors among some clinical parameters with lipids at D30. For instance, alterations in LDL-C and Tg levels related to PE were not observed at D30 in both lipid-lowering therapies and the same was observed for LDL-C levels regarding LPE.
With rosuvastatin, HDL-C levels related to LPE, which were positively correlated at D1, became negatively correlated at D30. In addition, TC and HDL-C levels related to PC and SM became negatively correlated with CE, as well as LDL-C levels related to SM, whereas, Tg levels related to CT, LDL-C, and LPEbecame positively correlated. With simvastatin plus ezetimibe at D30, Tg levels related to TC, LDL-C, LPE, PC and SM became positively correlated and with rosuvastatin at D30.

Univariate and multivariate analyses
Data quality was carried out by inspecting the repeatability of lipids and FFAs in the QC plasma sample, analysed throughout data acquisition. More than 80% of the quanti ed metabolites in the QCs have acceptable coe cient of variation porcentages (% CV) for peak areas; in this work, < 20% CV was used as a criterion to retain that particular component in the dataset for further evaluation, which were in agreement to the recomendation 12,13 .

Discussion
This study revealed differences in lipid composition in subjects with STEMI despite similar classic lipid pro les. Even after highly effective lipid-lowering therapies, examining the effects of rosuvastatin or simvastatin plus ezetimibe, promoting similar changes in the classic lipid pro le, and substantial alterations in the lipid composition, differences in the some lipid were still observed between these lipidlowering therapies. These differences were specially noted for LPC, TAG, and LPE. These ndings seems relevant, due to the involvement reported for free fat acids (FFA) in crucial mechanisms of atherosclerosis, such as in ammation and oxidative stress 14 .
A recent review article addressing CVD with lipidomic analysis showed that CER and PCs containing saturated (SAFA) and monounsaturated (MUFA) fatty acids were related to CVD events, while those containing polyunsaturated fatty acids (PUFA) were inversely associated with cardiovascular outcomes 15 . This observation also suggests a potential bene t, once the increase in FFA concentrations are related to the activation of in ammatory pathways and insulin resistance 16 . In ammation has been considered as a part of the pathophysiology of acute coronary syndromes and their recurrences 17,18 . Different FFA can be metabolized into different pro-and anti-in ammatory signaling molecules, in particular, a few n-6 FFA ( rst and foremost arachidonic acid) are precursors to pro-in ammatory molecules (primarily prostaglandins) 16 . SAFAs act as major inducers of in ammation through several mechanisms, one of these mechanisms involves the activation of toll-like receptor-4 (TLR4), which activates the production of in ammatory cytokines (IL-1beta; IL-6) 19 . FFA and their esters are the major sources of energy for the heart muscle. However, an excess of FFA has profound effects on the heart causing an enhanced susceptibility to oxidative stress and ischemic damage ( brosis and hypertrophy). Endothelial dysfunction due to the NF-kappa B activation are also induced by SAFA resulting in increased superoxide production, while NLRP3 in ammation activation increases endothelial permeability 20,21 .
In our study, in the two arms of therapy, higher concentrations of SAFA, MUFA, PUFA, and PC levels were observed at D1. However, PC and FFA levels were decreased after treatment, in G2 and G4. These results suggest a potential bene cial effect on CVD. Interestingly, our ndings demonstrated that FFAs were positively correlated to PC in all groups, but less positively correlated in G2. However, the reasons for different implications of PCs according to the types of FA (MUFA, SAFA or PUFA) on CVD are still poorly understood, the same for their effects on the metabolism of lipoproteins.
The literature reports that LPC, formed by hydrolysis of PC, is related to atherosclerosis development 22,23 . In our study, LPC(20:4/0:0) and PC(18:1/20:4) were signi cantly altered after treatment (G2 and G4). Choi et al. reported that LPC and PC presented large differences in drug response. The hydrolysis of PC to LPC is promoted by phospholipase A2 and rosuvastatin may affect the A2 activity. Thus, the levels of PC and LPC seems related to drug response 11 . The same work reports that LPC, ether-linked, and plasmalogens species of PC were negatively associated with stable coronary heart disease. In patients with stable coronary heart disease, rosuvastatin treatment promoted an increase of PCs, while atorvastatin a decrease 24 .
In our study, interesting results were related to TAG ndings. The most discriminant metabolites in G4 were TAGs (S2). Simvastatin plus ezetimibe treatment resulted in an increment of TAG species enriched with long chain FFAs, mainly MUFA and PUFA. These can be due to targeted effects of statin action on lipid metabolism 25 . Differences between rosuvastatin and simvastatin effects on TAG can be explained by the shorter half-life of simvastatin. When TAGs are cleaved, the resulting FFAs are not always directly taken up by nearby cells. In such cases, these spill-over FFAs bind to serum albumin and are transported through the circulation to other cells 26 . Another source of plasma FFA is their endogenous synthesis from excess of carbohydrates uptake, in a process termed lipogenesis. After synthesis, FFA are converted in TAG and released by the liver as VLDL particles.
In addition, our study showed that both lipid-lowering therapies decreased SM concentrations at D30. Sphingomyelin was measured in the large Multi-Ethnic Study of Atherosclerosis (MESA), and the authors reported a modest negative association with incident CVDs, after adjustment for lipoproteins and full adjustment for other risk factors 27 . Higher decrease with rosuvastatin compared with atorvastatin treatments was found in the SM/SM+PC ratio 28  Considering the correlation between lipids, FFA, and clinical parameters, a previous study reported that LPC are predominantly found in HDL, CER in LDL, and PC are present in both lipoproteins (HDL and LDL) 15 . Thus, FFA composition may also be important for the lipoprotein function and metabolism. Studies examining CVD with lipidomics found that after adjusting for HDL-C and LDL-C levels, only PC remained associated with cardiovascular events 15,30,31,32,33,34,35 . Kolovou et al. observed a correlation of SM, PC, PE, phosphatidylinositol, CEs and Tg in subjects with obesity, and a positive correlation between SM and LDL-C, and between LPC and VLDL-C, among subjects with familial hypercholesterolemia. In addition, the same lipidomic study demonstrated that every lipoprotein class was associated with a particular arrangement of lipids (LDL-C with CER and SM; HDL-C with PC, PE, and PE based plasmalogens) 24 . Some ndings in this work were in agreement with previous reports; PC and LPC were positively correlated to HDL-C at for both lipid-lowering therapies, respectively, at D1 and D30. SM and CER were positively correlated to LDL-C, but CER only at D1 in G1. FFA was positively correlated to HDL-C, LDL-C and CT for both lowering therapies. Depending on the therapy, the correlations between PC and SM with HDL-C were different at D30. Same results were found for CE, PC, SM and TAG with CT.
Finally, regarding to the changes in the classic lipid pro le, our results are in agreement with previous studies 36,29,37,11,38 . There are no direct comparisons between rosuvastatin and simvastatin plus ezetimibe in cardiovascular outcomes.

Limitations
This study compared two highly effective lipid-lowering therapies in the acute phase of myocardial infarction. However, we are unable to estimate the effects of the acute myocardial infarction per se on lipids and FA composition. It is expected some decrease in lipids due to the healing process involving the necrotic and ischemic myocardium, but these changes have been reported as insigni cant in the following days after the acute coronary event. Our sample size is relatively small, but the patients included in the study were all submitted to same treatment strategy (pharmacoinvasive) with similar characteristics at baseline. All subjects included in this study were fully compliant to the study drug and no harm related to the lipid-lowering theraphy was observed. There were no follow-up losses.

Conclusions
In spite of comparable classic lipid pro le at baseline and after the exposure to the treatments, signi cant differences in the lipid composition were found following between two highly effective lipid-lowering therapies. The main differences were higher levels of TAG with simvastatin plus ezetimibe therapy and increased levels of LPC following rosuvastatin therapy.

Declarations
Acknowledgements The authors would like to thank Dr. Adair Rangel de Oliveira Junior with the support with software Minitab and, to Sciex for data acquisition, in special to Helio Martins, Mackenzie Pearson and Santosh Kapil Kumar Gorti. Author contributions AK conducted data treatment, prepared and revised the manuscript; AS analyzed data and revised the manuscript; ECSC conducted correlation data treatment; FAHF designed study, prepared and revised the manuscript and, acquired funding for It; MCI prepared and revised the manuscript; AMFN helped with the research design, revised the manuscript and provided nancial support; MFMT analyzed data and revised the manuscript; PBMCD designed study ,conducted experiments and revised the manuscript; CESF designed study; RTB helps with data treatment; SCF helps with data treatment; ATF helps with data treatment; ASL analyzed data and revised the manuscript.
Competing Interests