Impact of L-carnitine on plasma lipoprotein(a) concentrations: A systematic review and meta-analysis of randomized controlled trials

We aimed to assess the impact of L-carnitine on plasma Lp(a) concentrations through systematic review and meta-analysis of available RCTs. The literature search included selected databases up to 31st January 2015. Meta-analysis was performed using fixed-effects or random-effect model according to I2 statistic. Effect sizes were expressed as weighted mean difference (WMD) and 95% confidence interval (CI). The meta-analysis showed a significant reduction of Lp(a) levels following L-carnitine supplementation (WMD: −8.82 mg/dL, 95% CI: −10.09, −7.55, p < 0.001). When the studies were categorized according to the route of administration, a significant reduction in plasma Lp(a) concentration was observed with oral (WMD: −9.00 mg/dL, 95% CI: −10.29, −7.72, p < 0.001) but not intravenous L-carnitine (WMD: −2.91 mg/dL, 95% CI: −10.22, 4.41, p = 0.436). The results of the meta-regression analysis showed that the pooled estimate is independent of L-carnitine dose (slope: −0.30; 95% CI: −4.19, 3.59; p = 0.878) and duration of therapy (slope: 0.18; 95% CI: −0.22, 0.59; p = 0.374). In conclusion, the meta-analysis suggests a significant Lp(a) lowering by oral L-carnitine supplementation. Taking into account the limited number of available Lp(a)-targeted drugs, L-carnitine might be an effective alternative to effectively reduce Lp(a). Prospective outcome trials will be required to fully elucidate the clinical value and safety of oral L-carnitine supplementation.

modulated, and Precocious Coronary Artery Disease (PROCARDIS) study showed that two Lp(a) single nucleotide polymorphisms (SNPs) accounted for 36% of variation in Lp(a) levels 3 . The levels of Lp(a) vary among populations; African descendants have increased Lp(a) levels as compared to Caucasians and Asian descendants 4 . However, different conditions such as diabetes, endocrine disorders, renal and liver failure or the acute-phase response might influence the synthesis or catabolism of Lp(a) 5,6 . The development of immunoassays to precisely measure Lp(a) levels in plasma was impeded by the structural complexity and size heterogeneity of Lp(a) 3 .
Many studies have shown that increased plasma Lp(a) levels are associated with coronary artery disease 7,8 , peripheral arterial disease 9 , cerebrovascular disease 10 , abdominal aortic aneurysm, aortic valve stenosis and calcification 11 , and venous thromboembolism 12 . Moreover, elevated Lp(a) levels have been found to be associated with increased risks for major adverse cardiac events (MACE) 13 . The atherogenic impact influenced by the effective binding of oxidized phospholipids by Lp(a), the expression of adhesion molecules by Lp(a) 14 , the deposit of Lp(a) in the arterial wall and the prothrombotic effect caused by the homology of apo(a) to plasminogen are potential mechanisms by which Lp(a) raises cardiovascular (CV) risk 15,16 . After migration from plasma into the arterial intima, Lp(a) joins to the extracellular matrix through both apo(a) and apoB components, resulting in the growing of the atherosclerotic plaque 17 .
In 2010, the European Atherosclerosis Society (EAS) Consensus Panel recommended a target level for Lp(a) < 50 mg/dL, as a feature of global CV risk 18 . The treatment of choice for the patients with increased Lp(a) levels is niacin or nicotinic acid (1-3 g/day), as recommended by the European Society of Atherosclerosis (EAS) and American Heart Association (AHA)/American Stroke Association (ASA) guidelines 18,19 . However, after negative trials with niacin, statins are worldwide recommended to reduce the global CV risk 20 . Moreover, observational studies suggested that the enhanced CV risk associated with increased Lp(a) levels is no longer observed when LDL-C levels are treated to aggressive levels, suggesting intensification of statin therapy and LDL-C lowering as a potential option in subjects with Lp(a) elevation 13,20 . Niacin therapy produces side-effects via stimulation of prostaglandin D(2) and E(2) synthesis, posing another obstacle to its use for Lp(a) elevation in clinical practice 21 . Therefore, a new therapy to target high Lp(a) levels is considered an important and unmet clinical need 22 . Besides niacin and intensive statin therapy, tibolone has also been investigated in patients with elevated Lp(a) levels 23 . Other drugs currently under development include cholesteryl ester transfer protein (CETP) inhibitors 24 , proprotein convertase subtilisin/kexin type-9 (PCSK9) inhibitory monoclonal antibodies 24,25 , mipomersen, an antisense oligonucleotide (ASO) against apoB mRNA 26 , eprotirome, a thyroid analogue 27 , lomitapide, a microsomal triglyceride transfer protein inhibitor 28 , and ASO therapy (ISIS APO(a) Rx) directly against apo(a) 29,30 . In addition, there has been a surge of interest to screen natural products (nutraceuticals/functional foods) for their effects on Lp(a) concentration 31,32 .
Different studies have indicated that L-carnitine, an amino acid involved in mitochondrial fatty acid oxidation and ATP production, might be associated with reduction in Lp(a) levels 33 . The metabolism of L-carnitine is complex, and involves formation of many short and long chain acylated forms, as well as metabolism to its core structure [31][32][33] . In addition, there are substantial alterations in the metabolism of L-carnitine depending upon whether it is provided as intravenous supplement, such as during hemodialysis, or as oral supplement 34 . With the latter route of supplementation, gut microbes have been shown to significantly impact L-carnitine metabolism, and to convert it into alternative metabolites, linked to increased atherosclerosis risk 35,36 .
Therefore, we aimed to assess the impact of L-carnitine supplementation and type of administration on plasma Lp(a) concentrations through systematic review of literature and meta-analysis of available randomized controlled trials (RCTs).

Methods
Search Strategy. This study was designed according to the guidelines of the 2009 preferred reporting items for systematic reviews and meta-analysis (PRISMA) statement 37 . SCOPUS (http://www.scopus.com) and Medline (http://www.ncbi.nlm.nih.gov/pubmed) databases were searched and the search was limited to the randomized controlled studies (RCTs) carried out up to January 31, 2015 investigating the potential effects of L-carnitine supplementation on Lp(a) concentrations. The databases were searched using the following search terms in titles and abstracts (also in combination with MESH terms): (L-carnitine) AND Lp(a), or (L-carnitine) AND Lipoprotein(a). The wild-card term "*" was used to increase the sensitivity of the search strategy. No language restriction was used in the literature search. The search was limited to studies in human. Selected articles were searched to identify further relevant studies. Two reviewers (CS and AS) evaluated each article separately. Disagreements were resolved by agreement and discussion with a third party (MB).

Study Selection.
Original studies were included if they met the following inclusion criteria: (i) being a randomized controlled trial with either parallel or cross-over design, (ii) investigating the impact of L-carnitine on plasma/serum concentrations of Lp(a), and, (iii) presentation of sufficient information on Lp(a) concentrations at baseline and at the end of follow-up in each group or providing the net change values.
Exclusion criteria were: (i) non-randomized trials, (ii) lack of an appropriate control group in the study design, (iii) observational studies with case-control, cross-sectional or cohort design, (iv) lack of sufficient information on baseline or follow-up Lp(a) concentrations, (v) inability to obtain adequate details of study methodology or results from the article or the investigators, and, (vi) the study was ongoing.

Quality assessment.
A systematic assessment of bias in the included studies was performed using the Cochrane criteria 38 . The items used for the assessment of each study were as follows: adequacy of sequence generation, allocation concealment, blinding, and handling of dropouts (incomplete outcome data), selective outcome reporting, and other potential sources of bias. According to the recommendations of the Cochrane Handbook, a judgment of "yes" indicated low risk of bias, while "no" indicated high risk of bias. Labeling an item as "unclear" indicated an unclear or unknown risk of bias.
Quantitative Data Synthesis. Meta-analysis was conducted using Comprehensive Meta-Analysis (CMA) V2 software (Biostat, NJ) 39 . Net changes in measurements (change scores) were calculated as follows: measure at end of follow-up − measure at baseline. For single-arm cross-over trials, net change in plasma concentrations of L-carnitine was calculated by subtracting the value after control intervention from that reported after treatment. Standard deviations (SDs) of the mean difference were calculated using the following formula: SD = square root [(SD pre-treatment ) 2 + (SD post-treatment ) 2 − (2 R × SD pre-treatment × SD post-treatment )], assuming a correlation coefficient (R) = 0.5. If the outcome measures were reported in median and inter-quartile range, mean and standard SD values were estimated using the method described by Hozo et al. 40 . Where standard error of the mean (SEM) was only reported, standard deviation (SD) was estimated using the following formula: SD = SEM × square root (n), where n is the number of subjects.
Meta-analysis was performed using either a fixed-effects or random-effect model according to the I 2 statistic. I 2 values < 50% and ≥ 50% corresponded with the use of fixed-effects and random-effects model, respectively. The generic inverse variance method was used to weight each individual study included in the meta-analysis. Heterogeneity was quantitatively assessed using the I 2 index. Effect sizes were expressed as weighted mean difference (WMD) and 95% confidence interval (CI). In order to evaluate the influence of each study on the overall effect size, a sensitivity analysis was conducted using the leave-one-out method (i.e., removing one study each time and repeating the analysis).
Publication bias. Potential publication bias was explored using visual inspection of Begg's funnel plot asymmetry, and Begg's rank correlation and Egger's weighted regression tests. Duval & Tweedie's "trim and fill" method was used to adjust the analysis for the effects of publication bias 41 .

Results
Search results and trial flow. The initial screening for possible interpretation eliminated the articles whose titles and/or abstracts were clearly insignificant. After evaluation, 7 studies with 8 treatment arms (intravenous administration: 2 eligible studies with 3 treatment arms, oral administration: 5 eligible studies with 5 treatment arms) met the inclusion criteria and were chosen for the final meta-analysis. A study flow chart is presented on Fig. 1. Characteristics of included studies. In total, 375 participants were randomized, of whom 200 were assigned to the L-carnitine supplementation group and 175 to the control group. The number of participants in these trials ranged from 15 to 46. Studies were published between 2000 and 2014, and were conducted in the Italy, China and Iran. A range of doses from 1 g 3 times per week to 4 g/day of L-carnitine was administered in the included trials. Duration of L-carnitine supplementation ranged from 1 week to 24 weeks. L-carnitine Risk of bias assessment. According to the Cochrane Collaboration, a specific tool for assessing risk of bias in each included study comprises judgment of specific features of the study 42 . This involves evaluating the risk of bias as 'low risk' , 'high risk or 'unclear risk' . The final category implies either lack of information or doubt over the potential for bias. There are seven analyzed domains comprising: sequence generation (selection bias), allocation sequence concealment (selection bias), blinding of participants and personnel (performance bias), blinding  Derosa et al. 63 Galvano et al. 64 Shakeri et al. 65 Shojaei et al. 66 Sirtori et al. 67 Solfrizzi et al. 68 Zhang et al. 69 Year of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective outcome reporting (reporting bias) and other potential sources of bias ( Table 2).

Publication bias.
The funnel plot of the study standard error by effect size (WMD) was asymmetric, suggesting potential publication bias that was also confirmed by the results of Egger's linear regression test (intercept = 1.08, standard error = 0.41; 95% CI = 0.09, 2.09, t = 2.61, df = 6, two-tailed p = 0.04) (Fig. 7). However, there was no sign of bias according to the Begg's rank correlation test (Kendall's Tau with continuity correction = 0.32, Z = 1.11, two-tailed p-value = 0.266). An attempt was made to adjust the effect size by imputing potentially missing studies using "trim and fill" correction. This approach led to the imputation of 3 missing trials on the left side of funnel plot, yielding a corrected effect size of − 9.16 mg/dL (95% CI: − 10.41, − 7.92) (Fig. 7). The "fail-safe N" test showed that 89 studies would be needed to bring the effect size down to a non-significant (p > 0.05) value.

Discussion
To our knowledge, the current systematic review and meta-analysis is the first to analyze evidence from RCTs on the efficacy of supplementation with L-carnitine on plasma Lp(a) concentrations. The results, albeit with small numbers of subjects overall, suggest a significant reduction of Lp(a) levels following oral supplementation with L-carnitine.  L-carnitine, a derivate of the amino acid lysine, is found predominantly in red meat, certain fish, and trace levels observed in dairy and some plant foods, such as avocado 43 . The effects of L-carnitine on lipid metabolism have been investigated in many experimental and clinical studies, but the exact mechanisms responsible for L-carnitine lowering effects on Lp(a) levels are not completely understood. However, the reason for these effects could be that L-carnitine, by stimulating fatty acid break-down in mitochondria, might reduce liver production of Lp(a) 43 . An experimental study has shown that L-carnitine inhibits advanced glycation end products (AGE)-modification of bovine serum albumin more potent than aminoguanidine, a prototype inhibitor of AGE 43 . In hemodialysis patients, L-carnitine supplementation might play a protective role against vascular injury based on reduction in circulating levels of AGEs 44 .
Recent studies have showed that oral ingestion of L-carnitine in both animals and humans results in generation of trimethylamine by intestinal microbiota, which is absorbed and then metabolized to trimethylamine-N-oxide (TMAO) by liver flavin monoxygenases (FMOs) (especially FMO3 and FMO1) 35,45 . Since L-carnitine is contained in red meat, omnivorous humans have higher values of TMAO than vegetarians, explaining why L-carnitine and the microbiota may contribute to the causal link between high consumption of red meat and CV risk 35 . Recent metabolomics and animal studies showed that TMAO, an intestinal microbiota-dependent metabolite formed from dietary trimethylamine-containing nutrients such as phosphatidylcholine (PC), choline, and carnitine, is linked to coronary artery disease pathogenesis, only in the presence of TMA/TMAO generation and intact gut microbiota 35,46 . The mechanisms by which TMAO raises CV risk and accelerates atherosclerosis are not well understood, but studies have shown that TMAO arising from both choline and L-carnitine supplementation in mice may inhibit reverse cholesterol transport 35 and enhance foam cell formation in both arterial and peritoneal cavity 47 . Direct provision of TMAO into diet has been shown to increase the development of atherosclerosis in animal models, confirming its pro-atherogenic properties 47 , and multiple clinical studies have confirmed that an elevated TMAO level is associated with increased risks for incident major adverse CV events 48 .
The levels of TMAO are modulated by hepatic FMO3, which is regulated by a bile acid-activated nuclear receptor named the farnesoid X receptor (FXR) 49 . The FMOs are known to oxidize sulfur-, phosphorus-, and selenium-containing compounds and endogenous amines, including TMA 50 . Antisense oligonucleotide mediated suppression of FMO3 levels have recently been shown to suppress TMAO levels, and to inhibit atherosclerosis in animal models 51 . Furthermore, FMO3 has been shown to serve as a central regulator of tissue cholesterol balance, and to be involved in insulin resistance and dyslipidemic changes in mouse models, suggesting the FMO3 and TMAO pathway as potential mediators of diabetes associated atherosclerosis 52,53 .
In contrast to the potential dangerous effects in the acceleration of atherosclerosis, mixed results have been reported with L-carnitine supplementation in the immediate setting following an acute myocardial infarction (AMI). While the major RCTs examining L-carnitine supplementation in the immediate post-MI setting showed no clinical benefit, a meta-analysis of L-carnitine intervention studies suggested its potential beneficial effect 54 . Other studies have suggested beneficial effects of L-carnitine supplementation on inflammatory parameters 55 , in the secondary prevention of CVD 54,56 , in diabetes mellitus 57 , on serum lipid profile in hemodialysis patients 58 , for adults with end-stage kidney disease on hemodialysis 59 or for patients in maintenance hemodialysis 60 , though detailed analysis of whether the carnitine was provided in oral vs intravenous supplementation, bypassing the gut microbiota, has not been extensively evaluated. In our meta-analysis, L-carnitine significantly reduced total cholesterol and showed borderline significant reduction of LDL-C, while plasma HDL-C and triglycerides were not affected. According to the available data L-carnitine supplementation might also significantly reduce triglycerides levels, but especially when low HDL-C levels are present 61 . However, the real effects of L-carnitine supplementation, taking into account its above-mentioned positive and possibly negative properties (via TMAO), on the process of atherosclerosis and atheroma plaque merit further investigations.
This meta-analysis has several important limitations, and the obtained results should be viewed as suggestive only. Most notable, there were only few eligible RCTs included in this meta-analysis, and most of them had a relatively modest number of participants. Moreover, none of the RCTs included have hard clinical outcome endpoints, and the duration of the interventions was relatively modest. Furthermore, the included studies were heterogeneous concerning the characteristics of patients and study design (2 of studies on L-carnitine supplementation have been performed in the setting of hemodialysis when carnitine metabolism is altered in uremia and the depletion of carnitine is expected 62 , 4 in diabetic/metabolic syndrome patients with dyslipidemia, and 1 in hyperlipidemic patients). There was also the diversity of the co-administered medications (statins or coenzyme Q10). Moreover the duration of the L-carnitine supplementation differs from one study to another, therefore further studies are required to test the time dependence of the observed effect as well as the impact of L-carnitine bioavailability on the Lp(a)-lowering activity. Many characteristics that vary within studies, such as the type of study, the year of publication, the inclusion and exclusion criteria, the sample size, the control group or quality of the studies, could have been the factors of between-study heterogeneity. However, we used "leave-one-out" sensitivity analysis under a conservative random-effects model to assess the possibility of some bias such as unclear double-randomization or large unbalanced dropout. Our results suggest that the significance of estimated pooled effect size is not biased by any single study.
The meta-analysis suggests that L-carnitine supplementation might be associated with significant Lp(a) lowering. However, given the TMAO elevating effect of oral L-carnitine supplementation, whether or not L-carnitine is a reasonable therapy to reduce Lp(a) levels requires further long-term investigations with hard clinical endpoints such as AMI, stroke and mortality risks. Prospective trials are required to fully elucidate the clinical value of L-carnitine supplementation.