Percutaneous Closure of Left Atrial Appendage significantly affects Lipidome Metabolism

Patients with non-valvular atrial fibrillation (AF) and a high risk for oral anticoagulation can be treated by percutaneous implantation of left atrial appendage occlusion devices (LAAC) to reduce the risk of cardio-embolic stroke. This study evaluates whether LAAC may influence lipid metabolism, which has never been investigated before. Patients with successful LAAC were included consecutively. Venous peripheral blood samples of patients were collected immediately before (T0, baseline) and 6 months after (T1, mid-term) LAAC. A targeted metabolomics approach based on electrospray ionization liquid chromatography–mass spectrometry (ESI-LC-MS/MS) and MS/MS measurements was performed. A total of 34 lipids revealed a significant change from baseline to mid-term follow-up after successful LAAC. Subgroup analysis revealed confounding influence by gender, age, diabetes mellitus type II, body mass index, left ventricular ejection fraction, creatinine and NT-proBNP. After multivariable adjustment within logistic regression models, these 34 lipids were still significantly altered after LAAC. Successful percutaneous LAAC may affect lipid metabolism and thereby may potentially affect pro-atherogenic and cardio-toxic effects.

LAAC and blood sampling. LAAC was performed using either the Watchman (Boston Scientific, Marlbrough, MA, USA) or Amplatzer Amulet (St. Jude Medical, St. Paul, MN, USA) device. Blood samples were taken by venous puncture within 24 hours prior to cardiac intervention (T0). Secondary blood samples were taken at least 6 months later (i.e. mid-term) (T1). Successful LAAC was confirmed by TEE during index procedure, as well as at mid-term follow-up by TEE and cardiac computed tomography angiography (CCTA).
Venous blood samples were taken from each patient and collected into serum monovettes ® and EDTA monovettes ® and centrifuged at 2500 × g for 10 minutes at 20 °C. The aliquoted samples were cooled down with liquid nitrogen before being stored at −80 °C until analysis. The whole processing took part within two hours after blood extraction.

Metabolite Analysis.
A targeted metabolomics approach based on electrospray ionization liquid chromatography-mass spectrometry (ESI-LC-MS/MS) and MS/MS measurements was performed using the AbsoluteIDQ ™ p180 Kit (BIOCRATES Life Sciences AG, Innsbruck, Austria). The assay allows simultaneous quantification of in total 188 metabolites out of 10 µL plasma samples, including amino acids, biogenic amines, glycerophospholipids, sphingolipids and the sum of hexoses. Analyses were carried out on a QTRAP 4000 System (Sciex Deutschland GmbH, Darmstadt, Germany) and a Thermo TSQ (ThermoFisher Scientific Waltham USA).
For the evaluation of metabolite concentrations, internal standards served as a reference. BIOCRATES MetIDQ ™ software was used for the processing and technical validation of the metabolite data.

Statistical analysis.
To exclude metabolites of which concentration values are below LOD, a general cleaning of the data set based on an 80% rule was performed. Remaining values below LOD in the data set were then imputed applying a logspline imputation method and the resulting data set was log2 transformed 13,14 .
Principal Component Analysis (PCA), Partial Least Squares Discrimination Analysis (PLS-DA) and Hierarchical Cluster Analysis (HCA) were used as supervised and unsupervised multivariate approaches 15 . To compare significant differences, data were subjected to a student's t-test or repeated measures ANOVA (rANOVA). To control the False-Discovery-Rate (FDR) during multiple comparisons an adjusted p-value (Benjamini-Hochberg correction) was additionally calculated 16 .
A regression analysis based on a linear mixed effect model was applied for the evaluation of significant metabolites dependent on all seven subgroups (gender, age, diabetes mellitus type II (DM), body mass index (BMI), left ventricular ejection fraction (LVEF), creatinine, aminoterminal pro-B type natriuretic peptide (NT-proBNP)). The median within the study population or internal standards were used to detect the cut-off points for subgroup analysis. Statistical analysis was performed using R-Studio 17 . A p value of <0.05 was considered significant for all statistical analyses.

Study population.
A total of 44 patients with successful interventional LAAC were included into the present study. Median CHA2DS2-VASc score and HAS-BLED score was 4 (interquartile range (IQR) 3-5 and 3-4.3, respectively). Baseline characteristics of the patients are outlined in Table 1. Patients' medication influencing volume status, electrolytes, sympathetic activation, blood sugar or lipid status, such as diuretics, beta-blockers, ACE/aldosterone antagonists, insulin, biguanides or statins, did not change in type or dosage in between baseline and follow up. Clinical parameters like body weight were in steady state during follow up.
Hierarchical cluster and lipid-subclass alterations. Six lipids did not pass the 80% LOD rule (PC aa C26:0, PC aa C30:2, PC aa C40:1, PC ae C42:0, Lyso PC a 14:0, SM C22:3). Resuming 99 metabolites of the lipidome were measured at time T0 and T1. The hierarchical cluster analysis illustrates the results for each tested metabolite subdivided in three lipid subclasses PCs, lysoPCs and SMs (Fig. 1a-c). Several metabolites of each class either in-or decreased at mid-term follow-up.
Principle component analysis (Fig. 2a) and partial least squares discriminant analysis (Fig. 2b) showed discrimination of plasma metabolome between T0 and T1. Figure 3 shows overall changes during mid-term follow-up based on lipid subclasses as mean log2 concentration. All three subclasses LysoPCs, PCs and SMs showed increases over time.
Overall changes of lipidome metabolites before and at mid-term follow-up after successful LAAC intervention. A total of 29 PCs, 4 SMs, and 1 LysoPC significantly changed over mid-term follow-up after successful LAAC (bold typed, Table 2).  Multivariable regression model to evaluate the influence of LAAC on lipidome. The predefined subgroups including gender, BMI, age, DM, LVEF, creatinine and NT-proBNP were adjusted within a multivariable regression model. Even after adjustment, only marginal differences in significant expression of lipidome metabolites were demonstrated (Table 3). In multivariable regression models, only follow-up time T0 vs. T1 revealed to be a significant influencing factor on lipid metabolism, whereas none of the subgroups themselves revealed any significant influence after adjustment.    Fig. 4. Overall 33 metabolites increased significantly, whereas only one metabolite decreased significantly over time (i.e. SM.C18.1). Figure 5 illustrates the intracellular lipid pathways implementing the final significant changes.

Discussion
The present study demonstrates for the first time that successful LAAC significantly impacts lipid metabolism within mid-term follow-up. It was demonstrated that only time after successful LAAC (i.e. mid-term follow-up period) revealed to be associated significantly on the alterations of lipid metabolites, which might indirectly reflect the effect of successful LAAC, as the main cause for these differences. The subgroups within our open-label study cohort did not reveal a significant influence, which makes biases by patient characteristics unlikely.  General hypotheses. Interactions between cardiovascular diseases and the plasma lipidome were shown to reveal both cardio-protective and -toxic effects [18][19][20] . On the other hand, the tissue of the LAA, respectively LAA cardiomyocytes, are supposed to reveal different physiological and histological characteristics compared to left atrial tissue itself. The LAA cardiomyocytes form comb-like trabeculae within the appendage, which combined with relative blood stasis in AF predestinates the LAA for thrombus formation and increases the risk for stroke in patients with AF 21 . Furthermore, in AF the LAA is undergoing remodeling processes affecting its role as an endocrine organ 7,8,22 . Lipid metabolites represent a main source of energy supply for human cells. Depending on their biochemical properties, lipids are subdivided traditionally into the main sub-classes like free fatty acids (FFA), triglycerides (TG), phospholipids (PL) or sphingolipids (SL) 23 . Furthermore, subsequent underclasses consist of saturated or unsaturated lipids. Their lipophilic structure necessitates a protein-linked transport in blood plasma. Pathological alterations of the reference plasma lipoproteins such as high density (HDL), low density lipoproteins (LDL) or chylomicrons contribute as established cardiovascular risk factors causing hyper-or dyslipidemias and are associated with an adverse cardiovascular outcome 24 .  Some lipid sub-classes like PLs being measured in this study are essential structural components of the cell membrane. PLs are commonly defined as phosphorylated lipids and mainly consist of glycerophospholipid (gPL) and SL species. Structurally, SLs are composed of a long-chain sphingoid base, an amide-linked fatty acid, and a polar head group at the 1-position. Phosphorylated SLs are called SMs and therefore part of the PL subclass 25 . gPLs on the other hand consist of a glycerol based structure esterified with two varying organic fatty acids and one molecule of phosphoric acid. gPL species like phosphatidylcholine (PC), phosphatidylethanolamine (PE) or phosphatidylserine (PS) can be synthetized through attached various alcohols 26,27 . Lipase enzymes in turn can hydrolyze gPLs and form corresponding lyso-phospholipids like lysoPCs 28 . Effects on intra-cellular lipid metabolism. After undergoing intra-cellular lipolysis, PLs and TGs develop so called free fatty acids (FFAs), which serve as carrier molecules for acetyl-CoA, the main molecule within the citric acid cycle (TCA). Intermediate metabolites from TCA in turn are mainly used within the oxidative phosphorylation pathway at the inner mitochondrial membrane finally generating adenosine triphosphate (ATP) -the main cellular source of energy. On the other hand, TCA may deliver acyl-CoA, which can be used to as a substrate in fatty acid synthesis. Fatty acids can further be processed to PLs, SM and LysoPLs, as being outlined in the present analysis (Fig. 5). The present data show an increase of serum lipid subclasses lysoPCs, PCs and partially SMs, which might be interpreted as a hint for a developing imbalance of lipid metabolism with a predominating synthesis of fatty acids and subsequent lipid-subclasses after successful LAAC. Reduced fatty acid oxidation has already been associated with the development of cardiac hypertrophy 29 . increase, whereas SM.C18.1 was solely decreased significantly. Within the SLs several metabolites are known to be bioactive molecules regulating signal transduction pathways and influence on cell-cell interactions 30 . Focusing on cardiovascular diseases, ceramide -the basic SL -was shown to be involved in cardiac ischemia and reperfusion injury. The exact pathophysiology is not yet clarified, but ceramide seems to mediate apoptosis via activation of CD95 and tumor necrosis factor receptors resulting in cardio-toxic effects 18,31,32 . Besides a denovo synthesis pathway ceramide can be produced directly through SMs 30 . Decreased levels of SM.C18.1 and increased levels of the other SMs might potentially reflect an imbalance in ceramide synthesis. Altered forms of bioactive SL were shown in cardiovascular diseases such as myocardial infarction 33,34 . Effects on phosphatidylcholines and lysophosphatidylcholines. A total of 29 PCs with attaching mostly higher carbon numbers fatty acids were significantly increased after successful LAAC.
As known, hydrolyzed PCs do form corresponding lysoPCs 28 , and several lysoPCs were measured in the present study, revealing LysoPC.a.C16.1 being increased over time. Furthermore, increased levels of several PCs constitute a higher amount of substrates for the later synthesis of lysoPCs. Balances between both sub-classes represent important intra-cellular pathways, because lysoPCs are known to reveal pro-inflammatory and atherogenic effects 19 . While there are no direct cardio-modulating effects of PCs described so far, increased levels of PCs from gut bacteria such as trimethylamine N-oxide (TMAO) and choline may be associated with an increased cardiovascular risk profile [35][36][37] . Choline was shown to be associated with the future development of adverse cardiac events in addition to troponins 38 .

Conclusions
Several metabolites from lipid classes of PCs, SMs and lysoPCs were altered after successful LAAC treatment in patients with non-valvular AF. It may be speculated whether these alterations might reflect potential pro-atherogenic and pro-inflammatory changes at mid-term follow-up after LAAC. Therefore, a more differentiated clinical follow-up of these patients may be widened to metabolomic alterations, and future research may show-up the pathophysiological relevance of the present findings.

Study limitations.
The present study is the first evaluating the influence of successful LAAC on lipid metabolism and is therefore descriptive and hypothesis generating. The main limitations are the small sample size, lack of a control group without LAAC, lack of a control group according to drug treatment (especially statins therapy) and the pre-defined follow-up time of mid-term 6 months, disregarding short-term or long-term effects after LAAC. A systemic increase of potential atherogenic and proinflammatory metabolites being measured in plasma must not represent necessarily local pathological processes of cardiac tissue.