Graphical abstract graphic file with name fx1.jpg [63]Open in a new tab Highlights * • HFpEF and HFrEF patients have different carnitine metabolism profiles * • Increasing plasma carnitine levels raises cardiovascular risk in HFrEF patients * • Both low and high plasma carnitine levels are linked to higher cardiovascular risk in HFpEF __________________________________________________________________ Health sciences; Cardiovascular medicine; Human physiology; Metabolomics Introduction Heart failure (HF) is categorized based on the left ventricular ejection fraction (LVEF) into HF with reduced LVEF (<40%; HFrEF) and HF with preserved LVEF (≥50%; HFpEF). HF is a complex clinical syndrome.[64]^1 HFpEF affects nearly half of all HF patients, impacting over 13 million adults worldwide. Patients with HFpEF exhibit symptoms similar to those with HFrEF, and they also experience high rates of hospitalization, morbidity, and mortality.[65]^2 Neurohormone-targeted drug therapy has shown beneficial effects on HFrEF outcomes.[66]^3 However, the complexity of HFpEF, including its clinical presentation and associated comorbidities, complicates the development of effective treatments.[67]^4 A significant obstacle in treatment development is the inadequate understanding of the pathogenesis and pathophysiology of HFpEF. Therefore, studying the metabolic profiles of HFrEF and HFpEF is beneficial for understanding the differences between the two, and further elucidating the molecular mechanisms of HFpEF and identifying new therapeutic targets are crucial gaps in cardiovascular medicine.[68]^5 The heart has exceptionally high energy demands, with approximately 95% of its energy derived from mitochondrial oxidative metabolism and the remaining 5% from glycolysis. The primary substrates for mitochondrial oxidative metabolism are fatty acids and glucose, with about 70% of the energy generated through free fatty acid metabolism.[69]^6 Currently, HF is considered a systemic multi-organ syndrome fundamentally driven by metabolic failure. The failing heart is often described as a “fuel-depleted engine”.[70]^7^,[71]^8 These metabolic alterations include a reduction in fatty acid oxidation rates, partially compensated by an increase in glucose utilization, contributing to the progression of myocardial dysfunction. Recently, there has been growing interest in the overall metabolic landscape, including oxidative stress, inflammation, and mitochondrial dysfunction. For instance, a study have shown that, compared to HFrEF(n = 30), HFpEF (n = 38) myocardial tissues exhibit lower levels of fatty acid metabolites, tricarboxylic acid cycle intermediates, and branched-chain amino acid metabolites, differences that are not observed in plasma.[72]^9 Zhao et al. found that supplementing with carnitine, a crucial metabolite in fatty acid metabolism, can reverse N,N,N-trimethyl-5-aminovaleric acid (TMAVA) - induced myocardial hypertrophy.[73]^10 However, perhaps due to limitations in sample size and other factors, there may be inconsistencies in the metabolic profile research results of HF patients.[74]^9^,[75]^10^,[76]^11^,[77]^12 Therefore, conducting metabolic profile studies in large queues can help to more accurately understand the metabolic status of two types of HF and provide a basis for further mechanism research. Metabolomics analysis techniques can simultaneously quantify various molecular intermediates from multiple major bioenergetic pathways, making them highly suitable for studying metabolic profiles in HF patients.[78]^13^,[79]^14^,[80]^15 As an alternative to untargeted methods, a novel strategy called pseudo-targeted metabolomics offers high sensitivity, specificity, and excellent quantification. This approach can monitor hundreds to thousands of metabolites using dynamic multiple reaction monitoring (MRM) and ensures high data quality,[81]^16 and the pseudo-targeted metabolic profiling, using UHPLC-HRMS (ultrahigh-performance liquid chromatography - high-resolution mass spectrometry), provides abundant metabolic information and ensures comprehensive coverage of the metabolome. So, the pseudo-targeted method is suitable for large-scale sample analyses. To address the utility of this approach in assessing HF, we evaluated the metabolic spectrum differences between HFpEF and HFrEF using an improved pseudo-targeted method ([82]Figure 1A). Our primary aim is to identify metabolic abnormalities in HFpEF and to determine the differential pathways altered between HFpEF and HFrEF. Figure 1. [83]Figure 1 [84]Open in a new tab Metabolic differences of carnitine between heart failure and control groups were explored based on an improved pseudotargeted metabolomics technology (A) The improved pseudotargeted metabolomics technology research process. (B) Principal component analysis of the control, HFpEF, and HFrEF groups. (C and D) Supervised orthogonal partial least squares discriminant analyses (OPLS-DA) of HFpEF and HFrEF in positive and negative ion modes separately. (E) Pathway enrichment analysis between HFpEF and HFrEF. Results Baseline characteristics of the participants We used two cohorts in discovery and validation, including control, HFpEF, and HFrEF groups, respectively. The discovery population (n = 514) for modified pseudo-targeted metabolomics included 203 control samples, 155 patients with HFpEF, and 156 patients with HFrEF ([85]Table 1). The median (25th to 75th percentile) age was 61 years. Compared with control subjects, individuals with HF had a greater prevalence of hypertension, diabetes, CHD, and hyperlipidemia (all p < 0.001). Additionally, patients with HF exhibited higher levels of BUN and creatinine than control subjects, but lower HDL (all p < 0.001). However, compared with the control group, TC and TG levels showed an increasing trend in HFpEF, while they showed a decreasing trend in HFrEF (all p ≤ 0.01). Table 1. Demographic and clinical characteristics of control and HF groups in learning population All (n = 514) Control (n = 203) HFpEF (n = 155) HFrEF (n = 156) P-value Demographic characteristics __________________________________________________________________ Age (years) 61 [52; 68] 61 [52; 68] 61 [52; 68] 62 [51; 69] 0.903 Sex: Female 28.0% 30.5% 29.0% 23.7% 0.341 Smoking 38.5% 36.0% 39.4% 41.0% 0.600 Systolic pressure (mmHg) 129 [117; 145] 130 [120; 143] 134 [120; 150] 124 [111; 140] 0.002 Diastolic pressure (mmHg) 79 [70; 88] 77 [69; 86] 80 [70; 88] 80 [70; 89] 0.162 Heart Rate (bpm) 76 [68; 86] 75 [65; 80] 72 [65; 82] 86 [75; 100] <0.001 __________________________________________________________________ Medical history __________________________________________________________________ History of hypertension 67.1% 33.5% 86.5% 91.7% <0.001 History of diabetes 22.8% 5.91% 32.3% 35.3% <0.001 History of CHD 32.3% 2.96% 58.7% 44.2% <0.001 History of hyperlipidemia 18.3% 5.91% 34.8% 17.9% <0.001 __________________________________________________________________ Laboratory measurements and echocardiography __________________________________________________________________ TC (mmol/L) 3.88 [3.25; 4.59] 3.92 [3.30; 4.47] 4.05 [3.44; 4.76] 3.66 [3.06; 4.53] 0.010 TG (mmol/L) 1.12 [0.78; 1.69] 1.15 [0.80; 1.68] 1.27 [0.84; 1.91] 1.00 [0.72; 1.38] 0.002 HDL (mmol/L) 0.98 [0.83; 1.17] 1.06 [0.90; 1.25] 0.93 [0.82; 1.11] 0.91 [0.77; 1.14] <0.001 LDL (mmol/L) 2.29 [1.78; 2.85] 2.24 [1.84; 2.74] 2.37 [1.79; 3.05] 2.25 [1.64; 2.94] 0.347 BUN (mmol/L) 6.07 [4.64; 7.56] 5.52 [4.36; 6.79] 6.08 [4.69; 7.67] 7.18 [5.77; 9.26] <0.001 Creatinine (μmol/L) 80 [65; 96] 74 [64; 84] 78 [65; 100] 91 [75; 114] <0.001 LVEF (%) 60 [38; 66] 65 [62; 69] 62 [57; 66] 33 [27; 38] <0.001 NT-proBNP (pg/mL) – – 307 [136; 1195] 2766 [1560; 7835] – [86]Open in a new tab Note: Variables are expressed as percentages or medians (interquartile range). Abbreviations: CHD, coronary heart disease; TC, total cholesterol; TG, triglyceride; HDL, high-density lipoprotein; LDL, low-density lipoprotein; BUN, urea nitrogen; LVEF, left ventricle ejection fraction. An expanded cohort was used as a validation cohort (n = 3368), consisting of 1000 controls, 955 HFrEF patients, and 1413 HFpEF patients ([87]Table 2). The median (25th to 75th percentile) age was 59 years. Compared with control subjects, individuals with HF had a greater prevalence of smoking, hypertension, diabetes, CHD, and hyperlipidemia (all p < 0.001). Additionally, patients with HF exhibited higher levels of BUN and creatinine than control subjects, but lower HDL (all p < 0.001). Although there were differences in TC, TG, and LDL levels among the control, HFpEF, and HFrEF groups, it was observed that the changing trends in HFpEF and HFrEF compared with the control group were different. Consistent with the discovery cohort, these results suggest that lipid metabolism pathways differ between HFpEF and HFrEF. Table 2. Demographic and clinical characteristics of control and HF groups in validation cohort All (n = 3368) Control (n = 1000) HFpEF (n = 1413) HFrEF (n = 955) P-value Demographic characteristics __________________________________________________________________ Age (years) 59 [51; 67] 53 [48; 61] 63 [55; 72] 59 [48; 67] <0.001 Sex: Female 41.8% 55.6% 40.0% 30.1% <0.001 Smoking 32.8% 22.6% 34.3% 41.4% <0.001 Systolic pressure (mmHg) 128 [115; 144] 126 [116; 140] 133 [118; 151] 122 [110; 138] <0.001 Diastolic pressure (mmHg) 79 [70; 88] 78 [70; 87] 78 [70; 88] 80 [69; 90] 0.395 Heart Rate (bpm) 78 [68; 90] 75 [67; 82] 76 [65; 85] 86 [75; 102] <0.001 __________________________________________________________________ Medical history __________________________________________________________________ History of hypertension 65.6% 29.8% 83.3% 77.4% <0.001 History of diabetes 23.1% 6.30% 31.4% 28.7% <0.001 History of CHD 31.1% 1.70% 52.7% 29.5% <0.001 History of hyperlipidemia 17.8% 13.0% 21.7% 16.7% <0.001 __________________________________________________________________ Laboratory measurements and echocardiography __________________________________________________________________ TC (mmol/L) 3.87 [3.25; 4.61] 4.06 [3.49; 4.77] 3.77 [3.18; 4.52] 3.74 [3.13; 4.52] <0.001 TG (mmol/L) 1.19 [0.85; 1.79] 1.27 [0.87; 1.91] 1.23 [0.86; 1.85] 1.08 [0.79; 1.55] <0.001 HDL (mmol/L) 0.99 [0.82; 1.20] 1.12 [0.96; 1.33] 0.96 [0.82; 1.14] 0.88 [0.71; 1.10] <0.001 LDL (mmol/L) 2.30 [1.79; 2.88] 2.39 [1.87; 2.92] 2.19 [1.70; 2.82] 2.37 [1.87; 2.95] <0.001 BUN (mmol/L) 5.96 [4.67; 7.74] 5.16 [4.24; 6.40] 6.18 [4.82; 7.94] 7.40 [5.70; 10.0] <0.001 Creatinine (μmol/L) 77 [63; 96] 68 [57; 81] 79 [65; 100] 89 [71; 111] <0.001 LVEF (%) 60 [38; 66] 65 [61; 69] 63 [57; 67] 31 [25; 36] <0.001 NT-proBNP (pg/mL) – – 535 [137; 2046] 3804 (1745; 8584) – [88]Open in a new tab Note: Variables are expressed as percentages or medians (interquartile range). Abbreviations: CHD, coronary heart disease; TC, total cholesterol; TG, triglyceride; HDL, high-density lipoprotein; LDL, low-density lipoprotein; BUN, urea nitrogen; LVEF, left ventricle ejection fraction. Metabolic spectrum analysis of heart failure based on pseudo-targeted metabolomics technology In the learning population, a quality control sample containing an equal mix of the samples to be analyzed was prepared to gather comprehensive metabolite information via UHPLC-HRMS. A total of 2359 MRM transitions were selected from the untargeted metabolic profiles in both the positive and negative ion modes. The extracted ion chromatograms of these MRM transitions are shown in [89]Figure S1, with 1178 transitions accurately detected in the samples. A multivariate data analysis was conducted to compare the HF and control group metabolomes. An unsupervised principal component analysis revealed a separation trend among the three groups ([90]Figure 1B). Furthermore, a supervised orthogonal partial least squares discriminant analysis improved the clustering segregation between HFpEF and HFrEF, achieving distinct separation ([91]Figures 1C and 1D). The differentiation between HFpEF and HFrEF prompted the identification of potential metabolite biomarkers that contributed to metabolomic diversity. To determine the disrupted metabolic pathways in HFpEF, a metabolomic pathway analysis was based on the Kyoto Encyclopedia of Genes and Genomes Pathway Database. [92]Figure 1E lists all disrupted metabolic pathways in HFpEF and HFrEF, with the top potential pathways being arachidonic acid metabolism and fatty acid metabolism. A recent study found significantly lower levels of epoxyeicosatrienoic acid in patients with HFpEF than in controls, indicating a dysfunctional arachidonic acid metabolic pathway.[93]^17 Consequently, we focused on changes in fatty acid metabolism.[94]^10 In the enriched beta oxidation of very long chain fatty acids metabolic pathway, the metabolite L-carnitine was involved and played a significant role. Methodological validation of LC-MS-based quantitative detection of carnitine L-carnitine plays a crucial role in energy metabolism and has garnered increasing attention for its potential in the prevention and treatment of cardiovascular diseases, although it remains controversial.[95]^18 Therefore, this study aimed to further quantify carnitine levels in the plasma of a large cohort, focusing on HF patients, especially those with HFpEF and HFrEF. We developed an LC-MS-based quantitative detection method, and the MRM chromatograms of carnitine and its internal standard in plasma are shown in [96]Figure S2. To ensure the stability and reliability of the quantitative detection method, we conducted rigorous methodological validation.[97]^19 Validation of the analytical method was performed by assessing linearity, sensitivity, precision, stability, and recovery. A detailed validation summary, including acceptance criteria, as shown below. Linearity and sensitivity Establish calibration curves using calibration standards of different concentrations ranging from 0 to 200 μmol/L. The corresponding parameters of the regression curve are shown in [98]Table S1. The obtained linearity is sufficient, with a correlation coefficient value higher than or equal to 0.9996. The limit of quantification (LOQ) is 31.25 pmol/L ([99]Figure S3). Recovery [100]Table S2 summarizes the extraction recovery rates of carnitine at three different concentrations. The extraction recovery rate is 99∼101.27%, with an average recovery rate of 100.2%. Therefore, the extraction recovery rate of our method meets the acceptance criteria. Precision Instrument precision As shown in [101]Table S3, the RSD of carnitine concentration for the same sample after six consecutive injection tests is 0.94%. Method precision As shown in [102]Table S3, six independent samples were tested and analyzed, and the intra day and inter day precision of carnitine concentration were 2.47% and 3.44%, respectively. These results indicate that our approach is repeatable and reliable in the study. Stability We found that carnitine was very stable after three freeze-thaw cycles, with a relative standard deviation of 5.37% in its quantitative results. The detailed results are shown in [103]Table S1, indicating that carnitine has sufficient freeze-thaw stability. The results above demonstrated that our quantitative method is stable and accurate, suitable for the quantification of carnitine in large clinical cohorts. Analysis of different metabolic levels of L-carnitine in HFpEF and HFrEF Previous research disclosed a pronounced decrease in both medium-chain and long-chain acylcarnitines, integral components of fatty acid metabolism, within myocardial tissues of HFpEF and HFrEF patients compared to controls. Interestingly, plasma levels showed either no significant discrepancies or instances of elevation.[104]^9 However, acknowledging limitations in the earlier study, including a relatively modest sample size and a lack of investigation into changes in bioactive free carnitine levels, our research revealed that carnitine levels in HFpEF patients were significantly lower than those in the control group (p < 0.05), with no significant difference compared to HFrEF patients in the learning population (n = 514) ([105]Figure 2A). Compared to the control group, carnitine levels were significantly reduced in the overall HF population (HFpEF + HFrEF) (p < 0.05) ([106]Figure 2B). Figure 2. [107]Figure 2 [108]Open in a new tab Metabolic differences of carnitine between different groups in learning and validation populations (A and B) Distribution of carnitine in different groups (learning population). (C, D) Verification of the absolute quantitative analysis of carnitine in different groups (validation population). Data are represented as medians (interquartile range). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Subsequent studies in an independent validation population (n = 3368), consisting of 1000 controls, 955 HFrEF patients, and 1413 HFpEF patients, aimed to investigate carnitine changes in HF. In the second population, carnitine levels in HFpEF patients were significantly lower than those in the control (p < 0.001) and HFrEF groups (p < 0.001) ([109]Figure 2C). Compared to the control group, carnitine levels in the overall HF population (HFpEF + HFrEF) showed a decreasing trend ([110]Figure 2D). Thus, by using a larger population, we accurately assessed plasma carnitine levels and found that carnitine may serve as a differential metabolic target for HFpEF and HFrEF. This suggests that the two types of HF have distinct lipid metabolism pathways. Nonlinear analysis A Cox regression with restricted cubic spline (RCS) analysis revealed a non-linear and U-shaped ([111]Figure 3A) or J-shaped ([112]Figure 3B) association between carnitine levels and the composite risk of cardiovascular death or heart transplantation in HFpEF and HFrEF patients, adjusted for sex, age, smoking status, SBP, CHD, diabetes, HDL, LDL, Cr, and NT-proBNP. RCS is a widely used method for analyzing non-linear relationships between variables and outcomes.[113]^20^,[114]^21^,[115]^22 It essentially acts as a piecewise polynomial but requires continuity and smoothness in the first and second derivatives at each knot, ensuring a good model fit for most scenarios. Figure 3. [116]Figure 3 [117]Open in a new tab Survival and nonlinear analysis of carnitine between heart failure with preserved ejection fraction (HFpEF) and heart failure with reduced ejection fraction (HFrEF) (A and B) The changing trend of carnitine in the HFpEF and HFrEF groups was characterized by a non-linear statistical analysis. (C and D) Prognostic analysis of HFpEF and HFrEF. These patients were divided into two groups according to the inflection point. HFpEF patients with low carnitine levels (≤40.18 μmol/L), but not those with high carnitine levels (>40.18 μmol/L), were significantly associated with poorer survival. This U-shaped association was robust in the analyses ([118]Figure 3A). In other words, in HFpEF patients, there was no significant increase in cardiac risk associated with plasma carnitine concentrations above 40.18 μmol/L compared to those with low carnitine levels (≤40.18 μmol/L). However, higher carnitine levels were significantly associated with poorer survival in patients with HFrEF. Conversely, those with carnitine levels <35.67 μmol/L were not at an increased composite risk of cardiovascular death or heart transplantation. This J-shaped association was robust in the analyses ([119]Figure 3B). In other words, lower carnitine levels (<35.67 μmol/L) in HFrEF patients might be beneficial, being significantly associated with better survival rates. Survival analysis During a median 24-month follow-up, 434 patients (18%) experienced cardiovascular death or underwent heart transplantation (the endpoint). Patients with HF were divided into four groups based on free carnitine quartiles. Survival curves for HFpEF patients showed that those with plasma free carnitine levels in the lowest (<36.5 μmol/L) and highest (>58.4 μmol/L) quartiles faced a higher risk of cardiac events ([120]Figure 3C). This suggests that carnitine levels within the range of 36.5–58.4 μmol/L are associated with a lower risk of cardiac events in HFpEF patients. Kaplan-Meier analyses indicated an increased risk of cardiovascular death or heart transplantation with higher carnitine levels in HFrEF patients ([121]Figure 3D). This means that lower carnitine levels are linked to a reduced risk of cardiac events in HFrEF patients. The survival analysis results align with the previous non-linear analysis findings. They recommend specific plasma free carnitine levels for HFpEF and HFrEF patients: (1) HFpEF Patients: A carnitine level range of 36.5–58.4 μmol/L is associated with a lower risk of cardiac events and better survival rates. (2) HFrEF Patients: Lower carnitine levels (<35.67 μmol/L) are linked to a lower risk of cardiac events and better survival rates. This suggests different optimal carnitine levels for improving survival outcomes in HFpEF and HFrEF patients. Discussion This study, utilizing a substantial population, revealed a non-linear and U-shaped ([122]Figure 3A) or J-shaped ([123]Figure 3B) association between circulating free carnitine levels and the composite risk of cardiac events. Previous research has commonly reported linear associations between exposure factors and the composite risk of cardiac events.[124]^23^,[125]^24 A notable strength of this study is that we did not assume a linear relationship between plasma free carnitine levels and the composite risk of cardiac events. Instead, we investigated non-linear associations, adjusting for other influencing factors (such as sex, age, smoking status, SBP, CHD, diabetes, HDL, LDL, Cr, and NT-proBNP) that may affect the observed associations. Therefore, when conducting clinical trials for HF, it is crucial to consider both the HF subtypes (HFpEF or HFrEF) and the levels of carnitine (low or high). Carnitine plays a crucial role in the process of fatty acid entry into the mitochondria, and its imbalance can affect fatty acid oxidation.[126]^25 Under normal physiological conditions, the primary source of energy for the heart is oxidative phosphorylation of fatty acids in the mitochondria. However, in HF, there is a shift in the heart’s energy metabolism from fatty acid oxidation to glucose.[127]^26^,[128]^27 Carnitine is emerging as an important factor in cardiovascular diseases, However, previous studies have shown conflicting results regarding plasma levels of L-carnitine in patients with chronic HF. Some studies with small sample sizes (n < 50) reported elevated or normal levels of plasma L-carnitine,[129]^28^,[130]^29 while larger studies (n = 168) indicated decreased myocardial carnitine levels with elevated plasma carnitine levels during HF.[131]^23 In our study, we examined a relatively large cohort of HF patients with different etiologies and disease severities. In the discovery population (n = 514), plasma carnitine levels were significantly reduced in HF patients compared to controls ([132]Figure 2B). However, in a larger validation population (n = 3368), although there was a decreasing trend in plasma carnitine levels in HF patients compared to controls, the difference was not statistically significant ([133]Figure 2D). These findings emphasize the importance of having a larger sample size to ensure more precise estimates. In our validation population, we observed a downward trend in plasma carnitine levels in HF patients compared to the control group, though the difference was not statistically significant. This suggests that there might be different lipid metabolic pathways in the two subtypes of heart failure (HFpEF and HFrEF), necessitating separate analyses. Several studies have investigated serum free carnitine levels in patients with HFpEF, HFrEF, and non-HF individuals, but the evidence is also contradictory. Some studies report that carnitine levels decrease in HFpEF patients compared to non-HF patients, while others indicate an increase in carnitine levels.[134]^11^,[135]^30^,[136]^31^,[137]^32^,[138]^33 Zhao et al. utilized liquid chromatography-tandem mass spectrometry (LC-MS/MS) to measure carnitine levels in serum of HFpEF, HFrEF, HFmrEF, and non-HF patients. They found that HFrEF patients had significantly higher carnitine levels (p < 0.001), compared to HFpEF, and all HF patients had higher carnitine levels compared to non-HF individuals.[139]^32 Hage et al. studied 46 HFpEF patients (LVEF ≥50%) and 75 HFrEF patients (LVEF <40%) and assessed serum carnitine levels using LC-MS/MS. They observed that HFpEF and HFrEF exhibit distinct metabolic characteristics. However, serum carnitine levels did not differ significantly between HFpEF and HFrEF.[140]^33 In our study, in the discovery population (n = 514), both HFpEF and HFrEF patients showed significant reductions in plasma free carnitine levels compared to controls, whereas there was no significant difference in plasma carnitine levels between HFpEF and HFrEF ([141]Figure 2A). In the validation population (n = 3368), HFpEF patients demonstrated significantly lower plasma free carnitine levels compared to controls and HFrEF, while HFrEF patients exhibited significantly higher plasma free carnitine levels compared to HFpEF and no difference compared to controls ([142]Figure 2C). This further confirms our hypothesis that different subtypes of heart failure (HFpEF and HFrEF) involve distinct metabolic pathways that warrant differentiated analysis. Analyzing the conflicting results of existing studies, variations in sample sizes—from relatively small to large—appear to be a contributing factor. This underscores the importance of larger sample sizes for ensuring more accurate estimates, which is a strength of our study. Currently, studies on carnitine in HF patients not only dispute its levels as mentioned earlier but also debate its effects on the heart. Some research suggests that carnitine has a protective effect on the heart.,[143]^34 while others associate carnitine with coronary heart disease and HF risks.[144]^18 Junko Naito et al. reported that carnitine deficiency can lead to cardiac dysfunction and that carnitine significantly improves HFpEF, as well as left ventricular (LV) systolic function and reduces LV hypertrophy in hemodialysis (HD) patients.[145]^35 Amin Mirrafiei et al., through systematic review and meta-analysis, found that supplementing L-carnitine slightly improves cardiovascular risk factors in adults with type 2 diabetes.[146]^36 However, Yoriko Heianza et al. found a long-term elevation of L-carnitine levels associated with subsequent coronary heart disease incidence, particularly among women with higher red meat intake.[147]^24 Currently, the contradictory understanding of carnitine’s role compels us to reanalyze its origin and metabolic pathway. Endogenous carnitine biosynthesis involves the rate-limiting step of γ-butyrobetaine (γ-BB) hydroxylation by γ-butyrobetaine hydroxylase (BBOX), leading to carnitine production in both eukaryotes and prokaryotes.[148]^37^,[149]^38 On the other hand, carnitine can be metabolized into γ-BB exclusively by gut microbiota, which can further break down into trimethylamine (TMA). These represent two distinct metabolic pathways: the former supporting energy metabolism with associated health benefits, and the latter contributing to trimethylamine-N-oxide (TMAO) biosynthesis in the gut, implicated in inflammatory processes underlying diseases like cardiovascular disease (CVD) and metabolic syndrome. Studies suggesting beneficial effects of carnitine have found that treatment with N,N,N-trimethyl-5-aminovaleric acid (TMAVA) significantly reduces plasma and cardiac carnitine levels, indicating the inhibition of fatty acid β-oxidation. Conversely, supplementing with exogenous carnitine alleviates TMAVA-induced cardiac hypertrophy. TMAVA competes with γ-butyrobetaine (γ-BB) for binding to the enzyme BBOX, thereby inhibiting carnitine synthesis. Additionally, TMAVA effectively inhibits the uptake of carnitine by cardiac myocytes through organic cation transporter 2 (OCTN2). These findings highlight the essential role of carnitine in cardiac metabolism.[150]^10 However, another research suggests that carnitine has negative effects. It emphasizes that the metabolism of dietary carnitine by gut microbiota produces trimethyllysine (TML) and trimethylamine (TMA), which in turn enhances the formation of trimethylamine N-oxide (TMAO), especially in individuals with high TMAO levels. Studies have shown that omnivores produce higher levels of TMAO compared to vegetarians or vegans.[151]^39 In our study, we identified a non-linear relationship between plasma free carnitine levels and the composite risk of cardiac events. Moreover, the role of carnitine in the heart should be discussed based on specific subtypes of heart disease (such as HFpEF and HFrEF, as in this study). For HFpEF patients, lower risk of cardiac events was associated with relatively higher carnitine levels (36.5–58.4 μmol/L) ([152]Figure 3A), correlating significantly with better survival rates ([153]Figure 3C). Conversely, for HFrEF patients, lower carnitine levels (<35.67 μmol/L) were linked to reduced risk of cardiac events ([154]Figure 3B) and better survival rates ([155]Figure 3D). The different effects of carnitine in HFpEF and HFrEF seem to suggest that the two types of HF have different patterns of carnitine production and metabolism. However, the specific mechanisms require further in-depth research in the future. Therefore, it is not definitive whether carnitine is universally beneficial or detrimental in the heart; rather, its effects depend on specific disease subtypes and the non-linear risk thresholds observed in different conditions. This study provides recommended ranges of plasma free carnitine levels for HFpEF and HFrEF patients, offering another type of biological insights particularly for HFpEF, and laying a foundation for precise therapeutic guidance development. Limitations of the study This study still has some limitations that need to be addressed. Study population: The participants in our study were limited to the Han Chinese population in China. It is a single-center prospective cohort study approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China. However, the relatively large study population provides a certain level of accuracy in interpreting the results. Subsequent studies are also ongoing to further validate the findings through multi-ethnic, multi-center cohort studies. Research findings: This study primarily elucidated the significant role of carnitine metabolism in cardiac energy metabolism, especially in HF diseases, particularly in HFpEF, where effective treatments are currently lacking. Subsequent research will further explore changes in amino acid, lipid, and other metabolite profiles occurring in HF. Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, L.Z. (zhengl@bjmu.edu.cn). Materials availability Materials generated in this study are available from the [156]lead contact upon request. Data and code availability * • Data: the raw metabolomics MS data generated in this study have been deposited in the MetaboLights database under accession code MTBLS10993. * • Code: this paper does not report original code. * • All other items: any additional information required to reanalyze the data reported in this paper is available from the [157]lead contact upon request. Acknowledgments