Patients with end-stage renal disease (ESRD) on chronic hemodialysis generally develop carnitine deficiency, as defined by subnormal plasma and tissue free carnitine and/or elevated acylcarnitine concentrations. These deficiencies may stem from a reduction in renal synthesis, insufficient dietary intake, or elevated intradialytic losses, and may contribute to several abnormalities encountered in dialysis patients including muscle weakness, cardiomyopathy, cramps, and abnormal lipid profile. It has been shown that LC plays an essential role in multiple primary functions including transport of long-chain fatty acids into the mitochondrial matrix, export products of β-oxidation from peroxisomes, and release of mitochondrial coenzyme A (CoA) from acyl-CoA when free CoA supply is limited (1). Most of studies were however restricted to the measurement of free and acylcarnitines, although in some studies the individual acylcarnitines were differentiated. The aim of this study was to apply targeted metabolic fingerprint in order to evaluate the metabolic status of hemodialysis patients. Plasma levels of carnitine and its esters (short-medium- and long-chain) have been quantified in a LC-MS/MS multiplex experimental setup in uremic patients, diabetics and non diabetics, on chronic hemodialytic treatment, and the possible influence on these levels of the hemodialysis session have been addressed by multivariate analysis. In the same cohort of subjects plasma amino acid levels were also quantified.

Materials and Methods

Study design and population

This observational, prospective study, included two periods of time work for a total of twenty-one days: a period of time screening of two weeks in which eligibility of patients was assessed, followed by a week of observation in which two blood samples were withdrawn, before and after the first dialysis treatment of the week. After approval of the protocol by the institutional review board, signed informed consent was obtained from each participating subject. Only clinically stable patients older than 18 years and on chronic hemodialysis since at least 6 months were included in the study. Other exclusion criteria were represented by current or previous treatment (within the last 3 months) with L-carnitine and/or its derivates, severe liver disease, acute infection, haemoglobin concentration (Fig 1).


Whole blood samples obtained after a 12-h fasting period were collected in K 3EDTA vials (3 mL) and centrifuged at 4000 rpm for 20 minutes. Plasma samples were stored at -80°C until analysis. Plasma samples (3.2 μL) were  transferred into 1.5 mL tubes and then extracted with a solution of Methanol/Water (75:25, v:v) and 0.01% oxalic acid (100 μL) containing the stable isotope labeled internal standards. The samples were centrifuged (15,600 rpm at 4°C for 15 minutes), and the supernatant was analyzed by direct infusion mass spectrometry (DIMS). The DIMS analysis for the evaluation of metabolite profile, specifically carnitine species and amino acids, in plasma samples was performed using a Liquid Chromatography-tandem Mass Spectrometry (LC/MS/MS) system consisting of an Alliance HT 2795 HPLC Separation Module coupled to a Quattro Ultima Pt ESI tandem quadrupole mass spectrometer.

Statistical analyses

Baseline characteristics were computed for different groups at baseline and after treatment and presented as mean values and standard deviation as well as median values and 25%-75% percentiles. Between-individuals comparison at baseline among the three groups (controls, non diabetics, and diabetics) have been conducted using non parametric analysis of variance (Kruskall-Wallis test). Pairwise contrasts were corrected for multiple comparison following Bonferroni approach. We first constructed the difference Δ = (measure after intervention) minus (measure at baseline), for each individual and variable, and the relative percentage difference %Δ = 100*Δ/(measure at baseline).


Basal values of plasma carnitine species are reported in Fig 2. Levels were significantly increased in both diabetic and non diabetic patients as compared to controls for the following species: acetylcarnitine (C2), butyrylcarnitine (C4), tiglylcarnitine (C5:1), isovalerylcarnitine (C5), malonylcarnitine/3-hydroxy-butyrylcarnitine (C4OH/C3DC), hexanoylcarnitine (C6), methylmalonylcarnitine/3-hydroxy-isovalerylcarnitine (C5OH/C4DC), glutarylcarnitine/3-hydroxy-hexanoylcarnitine (C5DC/C6OH), octenoylcarnitine (C8:1), octanoylcarnitine (C8), adipylcarnitine (C6DC), decadienoylcarnitine (C10:2), decenoylcarnitine (C10:1). Otherwise, no significant changes were observed for other esters.Basal plasma levels of amino acid proline (Pro), ornithine (Orn), citrulline (Cit) and serine (Ser) were significantly elevated in uraemic patients compared to controls (Fig 3). No differences were observed for the other amino acids tested. Regarding the role of hemodepurative treatment on plasma carnitine species, after hemodialysis it was found a significant reduction as compared to predialysis for short-chain acylcarnitines, medium-chain acylcarnitines and dicarboxylic acylcarnitines (Fig 4).  Levels of long-chain acylcarnitines at variance were significantly modified after dialysis in diabetic patients only. Effect of the hemodialysis session on plasma amino acid profile is reported in Fig 5. A significant difference was only found for short-chain C5:1 carnitine and for dicarboxylic C3DC/C4OH carnitine and for Ser. As can be seen in Fig 6, plasma acetylcarnitne and amino acid profiles discriminate both diabetics and non-diabetics uremic patients from control subjects.

Discussion and Conclusions

Coenzyme A is an essential metabolic cofactor, acting as an acyl group carrier and carbonyl-activating group in crucial biochemical reactions. It is therefore not surprising that specific acyl-CoA esters lie at important crossroads between metabolic pathways, where they may not only exert significant metabolic control through direct modulation of specific key enzymes (i.e., pyruvate carboxylase, pyruvate dehydrogenase, etc), but also affect gene expression, membrane trafficking and ion-channel activity (2). Indeed, due to the activities of the various carnitine acyltransferases, which catalyze a reversibly transfer of an activated acyl unit from CoA to L-carnitine, changes in the availability of L-carnitine in the cell affect acyl-CoA pools rapidly (3). This rapid equilibrium can also serve to mediate the efflux of acylcarnitine esters from different subcellular compartment, since free CoA and its esters are highly compartmentalized and unable to cross biological membranes.  Finally, since some of the metabolites of glutaric aciduria are known to be neurotoxic (i.e., glutaric acid and 2-hydroxyglutaric), it would be interesting to evaluate if carnitine treatment may facilitate their removal. Our study shows that abnormalities in plasma carnitine profile are common in ESRD patients on regular hemodialysis, regardless of the diabetic state. Patient metabolic fingerprint may be a convenient and useful tool to drive supplementation therapies targeted to normalize the altered plasma carnitine composition of patients on the basis of a personalized approach. This in turn might lead to the improvement of several hemodialysis-associated conditions, to the benefit of the patient. Whether this assumption holds true requires further investigation.


  1. Hoppel C. The role of carnitine in normal and altered fatty acid metabolism. Am J Kidney Dis 2003; 41: S4-12.
  2. Arduini A, Bonomini M, Savica V, et al. Carnitine in metabolic disease: potential for pharmacological intervention. Pharmacol Ther 2008; 120: 149-56.
  3. Zammit VA. Carnitine acyltransferases: functional significance of subcellular distribution and membrane topology. Prog Lipid Res 1999; 38: 199-224.