Article Text

Download PDFPDF

Original article
Effect of partial inhibition of fatty acid oxidation by trimetazidine on whole body energy metabolism in patients with chronic heart failure
  1. Gabriele Fragasso1,2,
  2. Anna Salerno1,2,
  3. Guido Lattuada1,
  4. Amarild Cuko1,2,
  5. Giliola Calori1,
  6. Antonella Scollo1,
  7. Francesca Ragogna1,
  8. Francesco Arioli1,2,
  9. Giorgio Bassanelli1,2,
  10. Roberto Spoladore1,2,
  11. Livio Luzi1,3,
  12. Alberto Margonato1,2,
  13. Gianluca Perseghin1,3
  1. 1Division of Metabolic and Cardiovascular Sciences, Istituto Scientifico H San Raffaele, Milano, Italy
  2. 2Clinical Cardiology, Heart Failure Clinic, Istituto Scientifico H San Raffaele, Milano, Italy
  3. 3Department of Sport Sciences, Nutrition and Health, Università degli Studi di Milano, Italy
  1. Correspondence to Gabriele Fragasso, Division of Metabolic and Cardiovascular Sciences, Istituto Scientifico San Raffaele, Via Olgettina 60, 20132 Milano, Italy; gabriele.fragasso{at}


Objective Trimetazidine may have beneficial effects on left ventricular (LV) function in patients with systolic heart failure. The authors assessed whether long-term addition of trimetazidine to conventional treatment could improve, along with LV function, resting whole body energy metabolism in patients with chronic systolic heart failure.

Design Single blind randomised study.

Setting University Hospital.

Patients 44 patients with systolic heart failure receiving full medical treatment.

Interventions Indirect calorimetry and two-dimensional echocardiography at baseline and after 3 months.

Main outcome measures Whole body resting energy expenditure (REE), percentage of predicted REE, LV ejection fraction (EF), NYHA class, quality of life.

Results Trimetazidine increased EF compared with conventional therapy alone (from 35±8% to 42±11% vs from 35±7% to 36±6%; p=0.02, analysis of variance for repeated measures). NYHA class and quality of life also improved compared with conventional therapy (p<0.0001). REE (from 1677±264 to 1580±263 kcal/day) and percentage of predicted REE (based on the Harris–Benedict equation: from 114±10% to 108±9%) decreased in the trimetazidine group, but not in the control group (REE from 1679±304 to 1690±337 kcal/day and percentage of predicted REE from 113±12% to 115±14%). The variation was different between groups (p=0.03 and 0.023, respectively).

Conclusions In patients with systolic heart failure, improvement in functional class and LV function induced by middle-term trimetazidine therapy is paralleled by a reduction in whole body REE. The beneficial cardiac effects of trimetazidine may be also mediated by a peripheral metabolic effect.

  • Systolic-dysfunction heart failure
  • left ventricular function
  • metabolic therapy
  • energy expenditure
  • heart failure

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


In chronic heart failure (HF), therapeutic strategies have traditionally focused on the modification of haemodynamic alterations that occur in the failing heart. However, in addition to haemodynamic alterations, HF causes large changes in both systemic and cardiac metabolic milieu. Trimetazidine (1-(2,3,4-trimethoxybenzyl)piperazine dihydrochloride) has been reported to have anti-ischaemic properties without affecting myocardial oxygen consumption and blood supply.1–6 The beneficial effect of this agent has been attributed to preservation of intracellular concentrations of phosphocreatine and ATP7 8 and reduction of cell acidosis,9 10 calcium overload10 and free-radical-induced injury caused by ischaemia.11 More importantly, trimetazidine affects myocardial substrate utilisation by inhibiting oxidative phosphorylation and shifting energy production from free fatty acids (FFAs) to glucose oxidation.12 13 This effect appears to be predominantly caused by a selective block of long-chain 3-ketoacyl-CoA thiolase activity, the last enzyme involved in ß-oxidation.14 A few studies performed in small groups of patients with post-ischaemic left ventricular dysfunction have shown that trimetazidine may be beneficial in terms of preservation of left ventricular function and control of symptoms.15–20 On this basis, it has been shown that this pharmacological approach may also be useful in the treatment of patients with HF of various aetiology.21–24 In patients with HF, a higher resting metabolic rate has been observed,25–27 and this factor probably contributes to progressive worsening of the disease. Therefore intervention strategies aimed at optimising global and cardiac metabolism may be useful for interrupting the vicious cycle of reduced function at greater metabolic expense in different cardiac conditions.28

In this context, we assessed whether the addition of trimetazidine to standard current treatment of patients with systolic HF, apart from the effects on symptoms and left ventricular function, could also modulate whole body energy expenditure.


Study subjects

We recruited 44 consecutive patients with systolic HF from the heart failure clinic of our institution (six female, 38 male; mean±SD age 70±10 years old, range 40–84 years; 15 with type 2 diabetes). All patients were in class II–IV of the New York Heart Association (NYHA) classification.

Thyroid and renal function were assessed at study entry. As routine practice in our outpatient clinic, all patients undergo stress/rest cardiac imaging testing and, in the case of positive results for either fixed and/or reversible defects, coronary angiography. HF was secondary to ischaemic heart disease in 29 patients, to hypertensive heart disease in six, to dilated cardiomyopathy in four, and to unknown causes in five. All patients were receiving standard treatment with ACE inhibitors and/or angiotensin receptor blockers, β-blockers, long-acting nitrates, digoxin, diuretics, aldosterone antagonists, statins, polyunsaturated fats and antiplatelet drugs, as required. Patients were considered for the study in the presence of: (1) persistent symptoms (exertional dyspnoea, fatigue, orthopnoea, paroxysmal nocturnal dyspnoea or peripheral oedema) despite optimised treatment of HF for at least 12 weeks and optimised up-titration of ACE inhibitors and β-blockers, with stable doses for the last 4 weeks; (2) an ejection fraction (EF) <45% by 2D-echocardiography. Patients were excluded in the presence of an acute myocardial infarction or unstable angina pectoris within 3 months, primary valvular disease, history of any alcohol misuse within 6 months, high-grade arrythmias, significant renal insufficiency (serum creatinine >2.2 mg/dl), coronary lesions suitable for revascularisation, left ventricular aneurysm or known active neoplastic process. After informed consent had been obtained, patients were randomly allocated to either conventional therapy plus trimetazidine (20 mg three times a day; 25 patients, 16 post-ischaemic) or conventional therapy alone (19 patients, 13 post-ischaemic). The randomisation was provided by individual sealed envelopes prepared in advance at the investigation site according to a computer-generated random list. The enrolment period lasted 36 months, and the trial was scheduled to last 3 months. Diabetes was present in nine patients randomised to trimetazidine and in six randomised to conventional therapy alone. Investigators were allowed to modify the dose of conventional treatment and to lower or discontinue the dose of trimetazidine, where necessary; in this case, patients would have been excluded from the study. All patients received nutritional counselling in order to maintain an isocaloric dietary regimen.

Experimental protocol

At the end of the follow-up period, all patients underwent functional evaluation; the physicians performing the tests were blinded with regard to the arm to which patients had been assigned. The following examinations were performed.

  • – Collection of medical history and physical examination. Symptoms relative to HF were classified according to the NYHA classification. Overall quality of life (QoL) was evaluated on a visual analogue scale (range 0–100).

  • – Echocardiography. All studies were performed with a Sonos 5500 (Philips Medical Systems, Bothell, Washington, USA) with broad-band transducers capable of second harmonic imaging (Hewlett–Packard S4 with 1.8/3.6 MHz transducer) and recorded on half-inch VHS tape for later review. Fractional shortening was calculated by M-mode measurements in the short-axis view. Measurements of left ventricular end-diastolic (EDV) and left ventricular end-systolic (ESV) volumes were obtained from the apical four-chamber view using the biplane Simpson's rule from which left ventricular EF was calculated as (EDV–ESV)/EDV×100.

Indirect calorimetry

After patients had been lying quietly for 30 min, rate of energy expenditure (REE) was measured by continuous indirect calorimetry with a ventilated hood system (Sensor Medics 2900 Metabolic Measurement Cart, Yorba Linda, California, USA) performed for 45 min as previously described.29 REE was calculated using Weir's standard equation30 from the rates of oxygen consumption and carbon dioxide production measured by indirect calorimetry (excluding the first 10 min of data acquisition), and from urinary nitrogen excretion. Predicted REE was calculated using the Harris–Benedict equations.31 Glucose, lipid and protein oxidation were determined as previously described.29

Statistical analysis

Values are given as mean±1SD or as percentages where appropriate. Unpaired Student t test or χ2 test as appropriate was performed to assess differences in pre-therapy values. Variables not showing a Gaussian distribution (assessed by Kolmogorov–Smirnov test) were compared by Wilcoxon test (NYHA class, QoL, plasma glucose). A two-way analysis of variance (ANOVA) for repeated measures on one factor (time: basal vs follow-up) was used to assess the treatment effect on the variables of interest (thyroid-stimulating hormone, creatinine, respiratory quotient (RQ), REE, percentage of predicted REE, body mass index (BMI) and EF). The time effect, the treatment effect and the interaction time×treatment were obtained (as summarised in the text or in tables 1 and 2). All calculated p values are two-tailed and considered significant when <0.05. Analysis was performed with SPSS V.13 software.

Table 1

Anthropometric and metabolic features in patients receiving conventional therapy and trimetazidine (+TMZ; n=25) or conventional therapy alone (−TMZ; n=19)

Table 2

Functional and echocardiographic results in patients receiving conventional therapy and trimetazidine (+ TMZ; n=25) or conventional therapy alone (−TMZ; n=19)


One patient in the conventional therapy group (post-ischaemic HF) was excluded from the protocol because of angina relapse (with positive myocardial perfusion tomoscintigraphy and scheduled coronary angiography). No differences in age (70±9 vs 69±12 years; p=0.61, independent t test between groups) and distribution of sex and diabetes (p=0.66 and p=0.74, respectively; Pearson χ2) were detected. Follow-up visits were attended by all recruited 44 patients, 25 receiving conventional therapy + trimetazidine (16 post-ischaemic, 64%) and 19 receiving conventional therapy alone (13 post-ischaemic, 68%). Apart from trimetazidine, all patients were receiving the same cardiological treatment at baseline and at follow-up, apart from adjustments of loop diuretics and aldosterone antagonists. None of the study subjects discontinued trimetazidine due to side effects, and the dose was not modified during the study. Compared with baseline values, trimetazidine affected neither blood pressure and heart rate nor QT interval (data not shown).

Anthropometry and laboratory data

BMI did not differ between the groups at baseline (table 1; p=0.11). It fell during the study in the whole population of patients (table 1, ANOVA for repeated measures within subjects), but no differences were detected between groups, and the same trend with time was observed in the two treatment groups (interaction: F=1.215; p=0.28). The same findings were obtained for thyroid function, as evaluated by thyroid-stimulating hormone activity (from 0.86±0.50 to 0.92±0.75 μU/ml in those receiving trimetazidine and from 1.15±1.52 to 0.98±1.27 μU/ml in those receiving conventional therapy; F=0.837; p=0.37; ANOVA for repeated measures for interaction) and in kidney function, as assessed by serum creatinine concentration (from 1.32±0.56 to 1.35±0.57 mg/dl in those receiving trimetazidine and from 1.46±1.02 to 1.59±0.89 mg/dl in those receiving conventional therapy; F=0.62; p=0.44; ANOVA for repeated measures for interaction). Fasting plasma glucose measured in the morning of the study also did not differ between groups, and no treatment effect was observed (from 116±51 to 122±57 mg/dl in those receiving trimetazidine and from 107±42 to 109±32 mg/dl in those receiving conventional therapy; p=0.57; F=0.21; p=65, ANOVA for repeated measures for interaction). No differences were detected either in total cholesterol, high-density lipoprotein-cholesterol, triglycerides and serum FFAs, although the latter showed a trend towards reduction (table 1). Similarly, sub-grouping patients according to the presence of diabetes did not reveal any significant difference in terms of glucose concentration.

Indirect calorimetry results

No differences were found in REE and percentage of predicted REE at baseline (table 1). At follow-up, a difference between treatment groups was detected for both variables of interest. A significant difference between treatments was found in both REE (table 1: F=4.635; p=0.038; ANOVA for repeated measures for interaction) and percentage of predicted REE (table 1: F=5.602; p=0.023; ANOVA for repeated measures for interaction). No detectable changes were found in oxidative substrate disposal as reflected by the RQ: from 0.81±0.06 to 0.81±0.05 in the group of patients receiving conventional therapy and from 0.82±0.08 to 0.84±0.04 in the group of patients receiving trimetazidine (F=0.585; p=0.45; ANOVA for repeated measures for interaction).

Specifically, no difference in whole body glucose oxidation was detected in the group of patients receiving conventional therapy (from 1.21±0.82 to 1.22±0.72 mg/kg/min) and those receiving trimetazidine (from 1.36±1.03 to 1.48±0.55 mg/kg/min) (F=0.32; p=0.58; ANOVA for repeated measures for interaction). However, a trend for a difference in whole body lipid oxidation was found: from 0.85±0.38 to 0.85±0.31 mg/kg/min in those receiving conventional therapy and from 0.79±0.44 to 0.66±0.24 mg/kg/min in those receiving trimetazidine (F=1.88; p=0.18; ANOVA for repeated measures for interaction). Diabetic and non-diabetic subgroups did not show any significant differences in terms of REE, RQ and glucose oxidation.

Functional status

In the group receiving trimetazidine, NYHA class decreased by one grade in 12 patients, and remained stable in 13 patients, while, in the group receiving conventional therapy, it decreased by one grade in two patients, increased by one grade in three patients, and remained stable in 14 patients (table 2; p<0.0001; ANOVA for repeated measures for interaction). Accordingly, the QoL score showed a parallel response (table 2; p<0.0001; ANOVA for repeated measures for interaction).

Echocardiographic results

Baseline functional features were similar at baseline in the two groups (table 2). In patients receiving trimetazidine, ESV significantly decreased (table 2: F=5.534; p=0.024; ANOVA for repeated measures for interaction), while EF significantly increased (table 2: F=5.845; p=0.02; ANOVA for repeated measures for interaction).


The main finding of this study is that trimetazidine, a specific partial inhibitor of FFA oxidation, when added to the usual treatment consistently reduces whole body REE as well as improving functional class, QoL and left ventricular function in patients with systolic HF, regardless of its aetiology and diabetic status. The beneficial effect of this agent has been previously attributed to preservation of cellular energy reserve,7 8 which has been shown to be a significant predictor of mortality.32 A recent meta-analysis has confirmed that trimetazidine has a significant protective effect against all-cause mortality and cardiovascular events and hospitalisation in patients with HF.33

Whole body energy metabolism in HF

Energy consumption at rest appears higher in patients with HF than in healthy subjects.34–36 It has been shown that increased REE is related to increased serum FFA oxidation and that both energy expenditure and serum FFA oxidation correlate inversely with left ventricular EF and positively with growth hormone, epinephrine and norepinephrine concentrations.34 Norepinephrine increases whole body oxygen consumption, circulating FFA concentrations, and FFA oxidation.35 These changes have been attributed to stimulation of hormone-sensitive lipase in adipose tissue and to stimulation of oxygen consumption independently of lipolysis by norepinephrine.36 These data, together with close correlations between plasma norepinephrine concentrations, energy expenditure at rest and FFA oxidation, make increased sympathetic activity the most likely explanation for alterations in fuel homoeostasis in patients with HF.36 In the present study, we report for the first time that the beneficial effect of trimetazidine on left ventricular function is also paralleled by a reduction in whole body REE compared with conventional treatment (table 1 and figure 1), underlying the possibility that the drug effect may be mediated through a reduction in metabolic demand at the level of the peripheral tissues and, in turn, in some sort of central (cardiac) relief.

Figure 1

Rate of energy expenditure at baseline and 3 months follow-up in patients receiving conventional therapy alone (CT, left histograms) or conventional therapy and trimetazidine (CT+TMZ, right histograms). *p=0.038 (analysis of variance for repeated measures for time×treatment interaction).

Effects of trimetazidine on cardiac metabolism

When circulating FFA concentrations are high, as in HF,37 oxidation of glucose and pyruvate is decreased. As a result, pyruvate is redirected towards lactate production and released from the heart. This produces protons, which the heart must also clear, a process that requires energy, and results in redirecting ATP away from contractile function, which can decrease cardiac efficiency.38 Conversely, decreasing plasma FFA concentrations, or directly inhibiting their oxidation, increases pyruvate oxidation and cardiac efficiency.39 In fact, a major factor in the development and progression of HF is already a reduced availability of ATP, determining a metabolic state that has been defined as ‘energy starvation’.40 Previous studies have suggested that trimetazidine may inhibit the utilisation of fatty acid substrates and shift metabolism from FFA to glucose oxidation, by selectively blocking the activity of 3-ketoacyl-CoA thiolase, the last enzyme of the oxidative chain,14 but this issue is still controversial.41 42 By inhibiting FFA oxidation, trimetazidine stimulates total glucose utilisation, including both glycolysis and glucose oxidation. Because utilisation of fatty acid oxidation is less efficient than glucose oxidation, it may be possible to improve myocardial contractile function by reducing FFA oxidation and increasing the flux through pyruvate dehydrogenase.43 In this context, recent studies have shown that energy deficiency in HF may result from increased uncoupling proteins (ie, less efficient ATP synthesis)44 and depleted glucose transporter protein (ie, reduced glucose uptake).45 On these grounds, the use of metabolic therapies (such as trimetazidine) aimed at interrupting the vicious metabolic cycle in HF has been advocated.46 Our data may suggest that the metabolic effect of trimetazidine may also take place in other organs and tissues. In fact, apart from a reduction in whole body energy demand, we also detected a trend in a reduction in whole body lipid oxidation (see the Results section) and fasting plasma FFA concentration (table 1). Reduction of whole body energy demand may be the mechanism by which trimetazidine improves symptoms and left ventricular function in patients with HF, as shown in the present and previous studies. If food intake in these patients were hypothetically kept constant, it would be interesting to evaluate whether, on the basis of energy expenditure restriction, a longer treatment period would be associated with weight gain. In fact, it is noteworthy that, in spite of the observed reduction in energy expenditure, we did not observe any effect on BMI. We believe that this finding may be related to the short duration of the intervention and eventually to a potential change in the usual physical activity of these patients. If food intake in these patients were hypothetically kept constant over a longer treatment period, we might be able to detect an association between treatment and weight gain. On the other hand, it is also possible that, owing to the observed HF improvement, these patients also improved their level of habitual physical activity, counteracting the potential effect of drug treatment on weight gain. Finally, another potential mechanism involved in the beneficial effect of the drug treatment may not be metabolically mediated as hypothesised by us, but may result from a secondary effect of the improvement in HF on the low-grade inflammatory state typically present in these patients. In fact, it has been previously shown that trimetazidine may also improve the inflammatory state.47 48

Study limitations

One limitation of this study is the lack of a placebo arm, which would have strengthened the study, even though the blindness to treatment by the involved physicians presumably ensured good study performance. In addition, although reduced peripheral (skeletal muscle) fatty acid metabolism may be the main mechanism beyond the beneficial effects of trimetazidine, this cannot definitely be ascertained by the present study. In fact, although there was a trend to a reduction in indirectly estimated lipid oxidation, there was no hint of a difference in RQ. Furthermore, a separate mechanism due to improvement in the HF syndrome with secondary correction of peripheral skeletal muscle effects cannot be excluded.


The results of this study support the concept that, in patients with systolic HF, trimetazidine, a partial fatty acid oxidation inhibitor, apart from being an effective adjunctive treatment in terms of symptoms, QoL and improvement in left ventricular function, is also associated with a reduction in whole body energy metabolism, suggesting that its beneficial cardiac effects may possibly also be mediated via a peripheral metabolic effect. The question of whether these effects could translate into decreased morbidity and mortality needs further investigation.


The financial support of the European Association for the Study of Diabetes (EASD) is gratefully acknowledged.



  • Competing interests GF has received travel reimbursement and given remunerated lectures for Servier, the manufacturer of trimetazidine.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Istituto Scientifico San Raffaele, Milano.

  • Provenance and peer review Not commissioned; externally peer reviewed.