The metabolites for muscle and osteoclast activity are indicators of femoral neck osteoporosis

Abstract

Femoral neck osteoporosis is a critical risk factor for a bedbound state, which significantly decreases overall life expectancy, if a patient falls and suffers a fracture. We performed a comprehensive blood metabolomic analysis of elderly individuals with and without femoral neck osteoporosis. Among the 129 metabolites that were identified, the abundance of six metabolites was significantly affected by femoral neck osteoporosis; decreases in phosphocreatine, malate, succinate, and histidine levels and increases in N1-methyladenosine and S-adenosylmethionine levels. The decline in muscle metabolite levels suggests that leg muscle activity is connected to the maintenance of bone mass; consistent with this, we also observed that TCA metabolite levels and leg muscle mass were decreased in patients with femoral neck osteoporosis. Interestingly, as creatine kinase is also activated in osteoclasts, the observed change in blood phosphocreatine levels reflects a change in bone homeostasis. On the other hand, as S-adenosyl-methionine serves as a donor in methylation reactions important for several biological processes, including osteoclast differentiation, increases in the levels of two methylation-related compounds (N1-methyladenosine and S-adenosyl-methionine) might indicate a link to osteoclast activity. These indicators have revealed novel aspects of the metabolic link between bone and muscle metabolism and methylation in femoral neck osteoporosis.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-36570-7.

Introduction

With the rapid acceleration of global ageing, the significant gap between healthy life expectancy and life expectancy is observed. Among several reasons why elderly people require nursing care, falls and fractures account for 12–24% of cases of disability, including a chairbound or bedridden state^1,2. Osteoporosis is one of aging-relevant diseases³, defined as a condition with decreased bone mineral density and bone strength. Consequently, the elderly patients with osteoporosis suffer increased bone fragility and face with a greater risk of fracture^3,4. Approximately 200 million individuals are estimated to suffer from osteoporosis worldwide^5,6. Bone mass and strength are deeply affected by the activities of component cells; bone formation by osteoblasts, bone resorption by osteoclasts, and mechanosensory function by osteocytes^7,8. About 90 percents of osteoporotic patients are females, as the decline in sex hormones, one of the critical etiological factors, over-activates osteoclasts. Although osteoporotic changes can occur in various bones throughout the body, femoral neck fractures are especially concerning due to their impact on disability in walking⁹, quality of life, and overall life expectancy¹⁰.

The other critical etiological factor for bone fractures is the falling. The ageing-related conditions, including frailty and sarcopenia, contribute to a greater risk of falls^11,12. Frailty is a condition characterized by declines in physical and cognitive function, leading to increased risks of disability and hospitalization, and is strongly correlated with mortality^11,13. Sarcopenia is defined as a loss of skeletal muscle mass and muscle strength in elderly individuals due to muscle ageing¹². Thus, there is increasing concern about the growing number of frail or sarcopenic individuals at risk of becoming bedridden or requiring long-term care⁵.

Recent metabolomic studies have shown that the essential information on metabolic alterations can provide keys to elucidate the pathophysiology of the diseases and also to search for diagnostic- or bio-markers that reflect the quantitative or qualitative changes in the diseased states of individual patients^14,15. Since about one-third of blood metabolites are labile, we previously established whole-blood metabolomics approach. Distinct from the other studies using serum or plasma as samples, analysing whole blood has allowed us to identify various biomarkers and obtain more comprehensive information and understanding via blood metabolomics¹⁶. For example, a comparison of the whole-blood metabolomes of young and elderly people revealed 14 markers of ageing, indicating a decline in antioxidant metabolites among elderly individuals¹⁷. Metabolomic analysis of individuals in a 3-day fasted state revealed an unexpected increase in metabolic activity, as the abundance of 44 metabolites, including antioxidant metabolites, was upregulated during fasting¹⁸. Moreover, decreased levels of antioxidant metabolites have been observed in frail individuals¹⁹. In sharp contrast to frailty, we observed a significant decrease in seven mitochondrial metabolites associated with sarcopenia²⁰.

In addition to hormonal changes, nutritional factors, including calcium and some vitamins, are deeply involved in the pathogenesis of osteoporosis²¹. Several metabolome studies on osteoporosis have been reported; however, the findings vary much among the results of these^22–26. We designed the present study focusing mainly on femoral neck osteoporosis, because fractures in the femoral neck significantly limit not only mobility, but also ADLs (activity of daily life) and increase mortality in aged individuals. Here, a comprehensive whole-blood metabolomics analysis was performed in elderly individuals with and without femoral neck osteoporosis to investigate the metabolic aspects of osteoporosis. Our study also includes the individual information on the sarcopenia and frailty status, which was combined to the analysis of osteoporosis metabolites. We showed that among the six metabolites identified as markers for osteoporosis at the femoral neck, four metabolites related to muscle metabolism were decreased, and two metabolites related to methylation were increased in abundance.

Results

Muscle mass was correlated with bone mineral density

To investigate the metabolic pathways and metabolites in femoral neck osteoporosis, clinical data, including the results of a frailty assessment, were collected and analysed. Frailty and sarcopenia were assessed to evaluate the association of physical and physiological changes with femoral neck osteoporosis. We focused on elderly individuals who were able to attend university hospital outpatient clinics, excluding those who were bedridden or had significantly reduced ADLs. Seventeen elderly participants, all community-dwelling females, participated in this study. These participants lived independently in their own homes and did not reside in a medical or assisted living facility. The average age of the individuals was 80.1 ± 7.8 years, and their ages ranged from 66 to 90 years. Osteoporosis was defined as a Bone Mineral Density(BMD) at or below 70% of the young adult mean (YAM), where the %YAM (young adult mean) was defined as the relative value of the patients’ BMD after normalization to the average BMD of young (20–29 years of age) controls, which was taken from the literature²⁷. Dual-energy X-ray absorptiometry (DEXA) was used to detect BMD at the femoral neck. Ten participants displayed femoral neck osteoporosis (82.5 ± 5.8 years), whereas the seven remaining participants (control group; 76.6 ± 9.4 years) presented values above 70% (Fig. 1A and Table S1). No participants in the control group had a history of fracture, while four participants in the femoral neck osteoporosis group had a history of fractures. The study participants were on medications for osteoporosis, such as bisphosphonates, vitamin D preparations, female hormone derivatives, and calcium supplements (Fig. S1). In addition to an evaluation of general health status (body mass index (BMI) and blood pressure) and blood tests, participants were examined for frailty using the Edmonton Frail Scale (EFS)²⁸ and Timed Up and Go Test (TUG) (Fig. 1A). They were also assessed for sarcopenia using the skeletal muscle mass index (SMI), leg muscle mass, and handgrip strength values. To evaluate cognitive function, the Japanese version of the Montreal Cognitive Assessment (MoCA-J) was used. Table S1 shows the clinical characteristics of the study participants. The BMD (%YAM) at the femoral neck was significantly lower in the femoral osteoporotic group than in the control group (83.1 ± 9.0% vs. 63.4 ± 5.0%, p = 0.001) (Fig. 1B). Blood tests were conducted to exclude factors other than femoral neck osteoporosis, ensuring that systemic conditions did not influence the study results. Blood tests and CBC data were comparable between the two groups (Table S1). Other clinical parameters, such as the EFS, TUG score, grip strength, score on the MoCA-J, and BMI, were not significantly different between the two groups (Fig. 1B and Table S1). Notably, the SMI and leg muscle mass were also significantly lower in patients with femoral neck osteoporosis (6.30 ± 0.77 vs. 5.18 ± 0.75, p = 0.01; 11.7 ± 1.2 kg vs. 8.6 ± 2.0 kg, p = 0.001) (Fig. 1B and Table S1). Consistently, BMD (%YAM) at the femoral neck was positively correlated with the SMI and leg muscle mass (R = 0.63; 95% confidence interval (CI), 0.22–0.85; p = 0.007 and R = 0.81; 95% CI, 0.54–0.92; p = 0.00001, respectively) (Fig. 1C and D). Thus, BMD and muscle mass are correlated in this study.

Fig. 1.

Elderly individuals with femoral neck osteoporosis exhibit lower leg muscle mass and a lower SMI than do controls. (A) Overview of the study protocol. Seventeen participants underwent clinical assessments and whole-blood metabolome analysis. (B) Dot plot comparing BMD (%YAM) at the femoral neck, the SMI, leg muscle mass, Moca-J, and EFS between the femoral neck osteoporosis group and the control group and among all participants. BMD (%YAM) was determined via DEXA. The SMI and leg muscle mass were determined via bioelectrical impedance analysis (BIA). * p < 0.05. ** p < 0.01.; statistical significance was determined via a t test. The data are shown as the mean values ± standard deviations (SDs). (C) Dot plot of participants for pearson’s correlation was used to determine the best-fit linear relationship between the BMD (%YAM) at the femoral neck and the SMI. ** p < 0.01. (D) Dot plot of participants for pearson’s correlation showing the best-fit linear relationship between the BMD (%YAM) at the femoral neck and leg muscle mass. ** p < 0.01. BMD (%YAM), the percentage value of BMD relative to the YAM. YAM, young adult mean BMD. The YAM at the proximal femur represents the average BMD of healthy individuals aged 20–29 years.

Four metabolites related to muscle metabolism were decreased in abundance in patients with femoral neck osteoporosis

To identify metabolites with differential abundance between the control and femoral neck osteoporosis groups and to analyse the physiological pathways involved, we performed nontargeted comprehensive metabolomics of 129 metabolites in whole-blood samples collected from the participants. In this analysis, the metabolites of interest were assigned by standard compounds or MS/MS data via LC‒MS (Data S1)^19,29. We then performed a correlation analysis to gain deeper insights into the associations between metabolic pathways and physiological properties and to evaluate the correlations between different metabolites. Using LC-MS, we found that the levels of 6 metabolites significantly differed between the femoral neck osteoporosis group and the control group (Fig. 2A and B). Among the six metabolites, the abundance of four (phosphocreatine, malate, succinate, and histidine) was significantly lower (p = 0.03, p = 0.005, p = 0.049, and p = 0.046, respectively) in the femoral neck osteoporosis group (Fig. 2A). Moreover, phosphocreatine was significantly correlated with the BMD (%YAM) at the femoral neck, with a Pearson correlation coefficient of 0.70 (95% CI, 0.32–0.88; p = 0.002) (Fig. 3A). Phosphocreatine serves as a form of energy storage in cellular and subcellular compartments³⁰. Phosphocreatine, which plays an important role in ATP production, is converted from creatine by creatine kinase, CK³⁰. Malate and succinate are metabolites generated by the TCA cycle. Succinate and histidine were previously reported to be human sarcopenia markers^20,31. Notably, the levels of other TCA metabolites, including citrate and 2-oxaloglutarate, were comparable between the groups (Data S1). Histidine has antioxidant properties due to its imidazole group and is utilized as a precursor for antioxidants such as carnosine and ergothioneine^32–34; however, the levels of other antioxidant metabolites, including 7 frailty markers, did not differ between the femoral neck osteoporosis and the control groups (Fig. S2). Furthermore, the levels of 15 frailty markers identified in a previous whole-blood metabolomics study¹⁹ did not significantly differ between the femoral neck osteoporosis group and the control group (Fig. S3). Among six metabolites associated with femoral neck osteoporosis, only phosphocreatine was significantly decreased in the frailty group (Fig. S4). There were no significant differences in these metabolites between the sarcopenia and control groups (Fig. S4). To further assess frailty, we analyzed the Timed Up and Go (TUG) test³⁵. We evaluated TUG performance using a cutoff value of 13.5 s, which effectively distinguishes individuals at risk of falls³⁶. Among these metabolites, phosphocreatine was significantly decreased in the group with poor TUG performance (Fig. S4), while no significant differences were observed between poor handgrip strength and control. Among the four osteoporotic markers identified in this study, the levels of three metabolites related to muscle metabolism (phosphocreatine, malate, and succinate) were correlated with each other (Fig. 3B) as follows: malate and succinate levels were correlated (R = 0.80; 95% CI, 0.52–0.92; p = 0.0001), phosphocreatine and malate levels were correlated (R = 0.68; 95% CI, 0.30–0.87; p = 0.003), and phosphocreatine and succinate levels were correlated (R = 0.52; 95% CI, 0.05–0.80; p = 0.03). Overall, these findings indicate that the metabolic markers whose abundance was significantly decreased in individuals with femoral neck osteoporosis (phosphocreatine, malate, succinate, and histidine) are involved in muscle metabolism.

Fig. 2.

Comprehensive blood metabolomics identified compounds related to muscle metabolism and methylation as markers of femoral neck osteoporosis. (A) Dot plot comparing the peak areas of phosphocreatine, malate, succinate, and histidine for the femoral neck osteoporosis group, the control group, and all participants. The peak areas of the metabolites were detected via LC‒MS and quantified with MZ-mine2 software. Single asterisks indicate differences for which * p < 0.05. The data are shown as the means ± SDs. (B) Dot plot comparing the peak areas of N1-methyladenosine and S-adenosyl-methionine (SAM) in the femoral neck osteoporosis group, the control group, and all participants. Single asterisks indicate significant differences; * p < 0.05. Double asterisks indicate differences for which ** p < 0.01. The data are shown as the mean values ± SDs.

Fig. 3.

Metabolites related to muscle metabolism were negatively correlated with metabolites related to methylation. (A) Dot plot for pearson’s correlation showing the best-fit linear relationship between BMD (%YAM) at the femoral neck and two femoral neck osteoporosis-related metabolites (phosphocreatine: left panel, N1-methyladenosine: right panel). ** p < 0.01. * p < 0.05. (B) Two-way correlations among the 6 osteoporosis markers (phosphocreatine, malate, succinate, histidine, N1-methyladenosine, and S-adenosylmethionine). Positive and negative correlations are shown in red and blue, respectively.

The abundance of two methylation-related metabolites was increased in patients with femoral neck osteoporosis

In addition, we identified two metabolites related to methylation (N1-methyladnosine³⁷ and S-adenosyl-methionine (SAM)³⁸ that were significantly increased in abundance in the femoral neck osteoporosis group (p = 0.02 and p = 0.01, respectively) (Fig. 2B). N1-methyladenosine is involved in the methylation of RNAs, including tRNAs, mRNAs, rRNAs, and mtRNAs^37,39, whereas SAM functions as a ubiquitous methyl donor during the methylation of proteins, nucleotides, or histones by methyltransferases^38,40,41. The level of S-adenosyl-homocysteine, which is generated from SAM by a methyltransferase, was comparable between the femoral neck osteoporosis and the control groups (Data S1). N1-methyladenosine level was significantly correlated with the BMD (%YAM) at the femoral neck, with a Pearson correlation coefficient of − 0.49 (95% CI, − 0.78 to − 0.01; p = 0.049) (Fig. 3A). N1-methyladenosine and SAM levels were positively correlated with each other (R = 0.71; 95% CI, 0.34–0.89; p = 0.001) (Fig. 3B). Furthermore, a group of three metabolites associated with femoral neck osteoporosis whose abundance was decreased (phosphocreatine, malate, and succinate) and a group of two metabolites related to methylation whose abundance was increased (N1-methyladenosine and SAM) were negatively correlated (Fig. 3B). Overall, these findings indicate that both of the metabolic markers whose abundance was significantly increased in individuals with femoral neck osteoporosis (N1-methyladenosine and SAM) are involved in methylation metabolism.

PCA with osteoporosis markers distinguished the group with femoral neck osteoporosis from the control group

Our study revealed that metabolites related to muscle and methylation are associated with femoral neck osteoporosis. Finally, we investigated whether these markers could differentiate the participants with femoral neck osteoporosis from those in the control group. We performed principal component analysis (PCA) of the 6 metabolites associated with femoral neck osteoporosis (phosphocreatine, malate, succinate, histidine, N1-methyladenosine, and SAM). The first two principal components (PCs) were effective in distinguishing participants with femoral neck osteoporosis from control individuals (Fig. 4A). The PCA results suggest that these six metabolites could serve as markers of femoral neck osteoporosis.

Fig. 4.

PCA distinguished the femoral neck osteoporosis group from the control group. (A) PCA plots based on 6 markers of femoral neck osteoporosis (phosphocreatine, malate, succinate, histidine, N1-methyladenosine, and S-adenosylmethionine) were generated to discriminate elderly subjects with and without femoral neck osteoporosis. (B) Summary of the 6 metabolites related to femoral neck osteoporosis. The four metabolites whose abundance was decreased in the femoral neck osteoporosis group are related to muscle metabolism or sarcopenia, whereas two metabolites related to methylation were increased in abundance in the femoral neck osteoporosis group.

Discussion

The present study focused on the metabolic aspects of femoral neck osteoporosis by utilizing whole blood and a nontargeted metabolomics approach. Our comprehensive analysis identified six of 129 blood metabolites as indicators of femoral neck osteoporosis. Among them, four muscle-related metabolites (phosphocreatine, malate, succinate, and histidine) were decreased in abundance in femoral neck osteoporosis patients, supported by our finding that femoral neck osteoporosis is significantly correlated with decreased muscle mass of lower limbs. On the other hand, the abundance of two methylation-relevant metabolites (N1-methyladnosine and SAM) were increased in these patients (Fig. 4B). The PCA results demonstrated the potential of these metabolites for diagnostic purposes.

Our metabolomic study on femoral neck osteoporosis added the new dimensions to the field as follows. First, the majority of previous studies on osteoporosis evaluated the metabolites in serum or plasma, and reported altered profiles in several amino acid levels^22–26. Our analysis identified several RBC-enriched metabolites (phosphocreatine, malate, and SAM) as the markers, as our whole blood approach covers both cellular and non-cellular compartments²⁹. In clinics, the patients with osteoporosis are diagnosed and characterised as the decline of BMD either in lumbar spine, femoral neck, or radius. In the present study, we mainly focused on femoral neck osteoporosis, because the fractures of femoral neck interfere more severely the daily life and health status of the patients, compared to the others⁴². Such study design enabled us to disclose the intriguing links of metabolites between femoral neck osteoporosis and muscle mass, through the identification of several muscle metabolites as the markers.

It has been known that bone mass is maintained by proper mechanical stretching or gravity stress^43–46. Mechanical stress also influences the activities of osteoclasts, osteoblasts, osteocytes, and skeletal muscle cells^47–49, partly via the modulation of their metabolisms by soluble factors, such as osteokines and myokines⁵⁰. Previous studies have reported that physical exercise or activity influences the levels of these metabolites, including phosphocreatine and tricarboxylic acid (TCA) cycle intermediates, reflecting muscle energy metabolism^51,52. However, it has been unclear which types of muscle metabolites are involved in femoral osteoporotic states. We observed the reduction both of leg muscle mass and of muscle-related metabolites (phosphocreatine, malate, succinate, and histidine) in femoral neck osteoporosis, suggesting that these metabolites could serve as biomarkers for the mechanical stress experienced by the femoral bone or as metabolic links between the muscle and bone. Moreover, these findings suggest that a lack of exercise is a contributing factor in osteoporosis and highlight the importance of promoting physical activity in the management of osteoporosis. Phosphocreatine was decreased in both the frailty and group with poor TUG performance. Frailty and poor TUG performance are known risk factors for falls^11,36, suggesting that phosphocreatine may serve as a potential marker for high-risk femoral neck fractures.

Second, these muscle-related metabolites also reflect aberrant bone metabolisms. Phosphocreatine is generated by the enzymatic activity of creatine kinase (CK), which is mainly expressed in the muscle and brain in forms known as muscle-type and brain-type CK (CK-M and CK-B, respectively). Interestingly, recent studies suggested that osteoclasts are also equipped with CK-B^53,54. The activated osteoclasts display increased CK-B activity, whereas inhibiting CK-B in osteoclasts reduces their bone resorption ability⁵³. Experimental osteoporosis is ameliorated in CK-B knockout mice, while patients with bone fractures or osteoporosis present elevated CK-B levels in blood^55–57. Collectively, CK-B activity in osteoclasts is deeply involved in the pathogenesis of osteoporosis⁵³. It is possible that the increased utilization of phosphocreatine by osteoclasts results in decreased blood levels of phosphocreatine in femoral neck osteoporosis. Histidine is utilized to generate histamine, a mediator of allergic reactions⁵⁸, which is also involved in osteoclast generation⁵⁹. Although the application of anti-histamine drugs against osteoporosis patients has been controversial⁶⁰, the knockout mice of histidine decarboxylase, generating histamine from histidine, displayed the mitigation of osteoporosis^59,61,62. In this context, a decreased abundance of histidine in femoral neck osteoporosis might indicate the involvement of histamine in osteoclast function in human patients. The lack of correlation between histidine and the other four metabolites related to femoral osteoporosis may reflect its dependence on a distinct metabolic pathway in addition to muscle metabolism.

Thirdly, our identification of increased SAM and N1-methyladenosine suggests the impact of methylation in femoral neck osteoporosis. Noteworthy, in mice models, osteoporosis is alleviated by the inhibition of methylation in DNA or histone^63–65, while SAM, a doner for methylation, increased in activated osteoclasts. Concerted elevation of SAM and N1-methyladenosine would be linked to the activation of osteoclasts with their modified epigenetic states. Previous studies in genetically-modified mice in DNA methylation and RNA methylation-related enzymes support the notions that the methylation of DNA and RNA also contributes to the pathogenesis of osteoporosis^64,66.

Collectively, our whole-blood metabolome analysis revealed unexpected aspects of femoral neck osteoporosis, decreased muscle metabolites and increased methylation. These markers would indicate either decreased muscle activity or increased osteoclast activity or both. These findings deepen our understanding of femoral neck osteoporosis and open avenues for exploring the impact of metabolites related to bone, muscle, and methylation in elderly individuals. Thus, our study provides a basis of clinical applications for developing future preventative approaches against femoral neck osteoporosis or a bedbound state after falls.

Study limitations

By focusing on the specific study population in our study, the evaluation of participants’ background information in terms of sex and health conditions was limited. These population characteristics represent inherent limitations of our study design.

In metabolomic studies, many studies succeeded in reaching valid conclusions through a well-designed study with clearly defined scope and objectives in small-scale settings⁵². However, increasing the sample size with detailed statistical analyses or conducting prospective cohort could further enhance the robustness of the present findings.

To minimize the potential influence of diet and malnutrition on the metabolomic profiles, we performed blood sampling after overnight fasting and conducted dietary interviews. However, information on dietary habits was not collected through the questionnaire. The lack of these data represents another limitation of our study.

Figure S 1 suggested the information on medication use in both the femoral neck osteoporosis and control groups. No significant differences in these six metabolites for femoral neck osteoporosis were observed between the bisphosphonate-treated and untreated groups. Although other osteoporosis medications or concomitant drugs may have influenced the metabolomic profiles, a comprehensive evaluation of all medications remains to be performed.

In our study focusing on blood metabolomic profiling, osteoclast activity was not directly assessed through biological assay. Although previous studies have reported associations between osteoclast function and specific metabolic pathways involving creatine kinase B, histamine decarboxylase, and S-adenosylmethionine^53,59,64, further studies integrating metabolomic data with experimental biological analyses will be important.

This study focused solely on human blood metabolomics and did not include direct assessments of bone metabolism or osteoclast activity. Therefore, the causal link between blood phosphocreatine and osteoclast activity remains to be clarified.

In this study, histamine was not included as a standard compound in our LC–MS setting, and therefore could not be evaluated. Little is known about the association between blood histamine levels and femoral neck osteoporosis. Hence, evaluating blood histamine dynamics in femoral neck osteoporosis will be an important subject for future studies.

Incorporating such improvements and experimental approaches in future research will help to further elucidate the sequence of metabolic processes involved in osteoporosis.

Materials and methods

Clinical assessment

All clinical data were recorded at Kyoto University Hospital in 2017 and 2018. Seventeen elderly female participants aged 65 years and over were subjected to clinical interviews, physical examinations, and blood tests. Patients who were bedridden or had kidney failure (serum creatinine over 2.0 mg/dL) or liver disease (serum AST and ALT over 50 U/L) were excluded from the study. BMD was measured via DEXA with a Hologic Discovery A DXA system (MA, US), and Patients with a BMD (%YAM) at or below 70% of the YAM at the femoral neck were diagnosed with femoral neck osteoporosis. Details regarding the BMD measurement, diagnostic criteria, and physical examinations are provided in the Supplementary Methods. The SMI and leg muscle mass were measured with an Inbody 720 instrument (Seoul, South Korea). Handgrip strength was measured with a Smedley-type hand-held dynamometer (Matsumiya Ika Seiki Seisakusho Co., Ltd., Tokyo, Japan).

Preparation of blood samples for metabolomic analysis

Blood samples were collected and prepared for metabolomic analysis as previously reported^17–20. Blood was collected once at the laboratory of Kyoto University Hospital in the morning between 09:30 and 11:00, with separate containers used simultaneously to collect blood for metabolomic analysis and various clinical blood tests. All participants were asked to fast overnight for at least twelve hours until blood sampling. The participants were advised to drink calorie-free beverages and to otherwise spend their time as normal before the blood test. Because some metabolites are easily degraded, the blood samples were rapidly quenched at − 40°C in methanol for quick sample processing. Moreover, 10 nmol of 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) and 10 nmol of piperazine-N, N′-bis (2-ethanesulfonic acid) (PIPES) were added to each sample to serve as internal standards.

LC‒MS conditions

Untargeted, comprehensive analysis by LC‒MS was conducted as previously reported^17–20. LC‒MS data were obtained using an Ultimate 3000 DGP-3600RS liquid chromatograph (Thermo Fisher Scientific, Waltham, MA, USA) combined with an LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). LC separation was performed via a ZIC-pHILIC column (Merck SeQuant, Umeå, Sweden; 150 mm × 2.1 mm, 5 μm particle size). The mobile phase was composed of ammonium carbonate buffer (10 mM, pH 9.3) and acetonitrile. Gradient elution from 80 to 20% acetonitrile was performed over 30 min at a flow rate of 100 µL/min. An electrospray ionization (ESI) source was utilized for MS detection. One microlitre of each sample was injected twice, once in positive ionization mode and once in negative ionization mode. The spray voltage was set to 4.0 or 2.8 kV (positive or negative ESI, respectively), while the capillary temperature was kept at 350–300 °C. Nitrogen gas was used as a carrier. The mass spectrometer was used in full scanning mode with a range of 100–1000 m/z and MS/MS fragmentation scanning in an automatic data-dependent manner.

LC‒MS data processing and analysis

LC‒MS data processing and analysis were conducted using MZmine 2 (version 2.29) software (mzmine.github.io) for peak area measurement and metabolite identification. Detailed methods are described in the Supplementary Methods section. The lists of identified metabolites are provided in Data S1.

Statistical analysis

Statistical analysis was carried out via Student’s t test to confirm significant differences between groups (statistical significance was set at p < 0.05) and the 95% CI value of the difference, the ordinary least squares method was carried out to confirm the suitability of linear regression, and Pearson’s correlation was carried out to analyse relationships between metabolites and clinical data (assuming p < 0.05). Principal component analysis was performed with R software to visualize the metabolomic model.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

MY and HK designed the research. MK analysed the data. MK, HK, and MY performed the research. MK, MY, and HK wrote the manuscript.

Data availability

Raw LC‒MS data are accessible from the MetaboLights repository (http://www.ebi.ac.uk/metabolights/MTBLS11650). Detailed information is provided in Supplementary Methods section.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

All participants signed informed consent forms before examination, in accordance with the Declaration of Helsinki. The experiments were performed in accordance with the relevant rules and official guidelines in Japan. The study protocols were approved by both the Human Research Ethics Committee of Kyoto University and the Review Committee on Human Subjects Research at OIST.