Skip to main content
NCBI home page
GeroScience logoLink to GeroScience
. 2024 Nov 27;47(2):2567–2580. doi: 10.1007/s11357-024-01453-0

Plasma triacylglycerol length and saturation level mark healthy aging groups in humans

Weisha Li 1,2, Bauke V Schomakers 1,3, Michel van Weeghel 1,2,3, Lotte Grevendonk 4, Frédéric M Vaz 1,2,3, Gajja S Salomons 1,2,3, Patrick Schrauwen 5,7, Joris Hoeks 4, Arwen W Gao 1,2, Riekelt H Houtkooper 1,2,6, Georges E Janssens 1,2,✉
PMCID: PMC11979014  PMID: 39601998

Abstract

Complex lipids, essential components in biological processes, exhibit conserved age-related changes that alter membrane properties and cellular functions and are implicated as biomarkers and contributors to longevity and age-related diseases. While physical activity alleviates age-related comorbidities and physical impairments, comprehensive exploration of the underlying biological mechanisms, particularly at the level of complex lipids, remains limited. However, clinical studies suggest that physical activity may counteract these age-related lipidomic changes, presenting a promising avenue for intervention. We performed lipidomic profiling of plasma from an extensively characterized cohort of young and aged individuals. Annotating 1446 unique lipid species across 24 lipid classes, we found the most prominent difference in older adults was an accumulation of triacylglycerols (TGs), with lower physical activity levels associated with higher TG levels in plasma and reduced physical functionality. Remarkably, lipid species in the TG class did not accumulate uniformly. Rather, our study unveiled a negative correlation between higher physical activity levels and TGs with shorter chain lengths and more double bonds in this demographic. Overall, our research highlights that plasma TG length and saturation level can help mark healthy aging groups in humans. These findings deepen our understanding of how aging affects complex lipids and the influence of physical activity on this process.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-024-01453-0.

Keywords: Aging, Healthy aging, Lipidomics, Human plasma, Triacylglycerols, Physical activity

Introduction

Aging is an inherent biological phenomenon characterized by progressive changes in various physiological processes that ultimately result in a decline in general health and functionality. Multiple key hallmarks have been recognized and delineated as contributing to the aging process. These 12 hallmarks include, among others, mitochondrial dysfunction, impaired immune function, reduced autophagy, epigenetic alterations, aberrant intercellular communication, loss of telomeres, altered nutrient sensing, and impaired protein homeostasis [1–6]. Understanding the drivers of different rates of biological aging could enable the development of more tailored and effective preventive and therapeutic strategies aimed at maximizing health span [7]. This understanding could also enhance diagnostics of aging, including more accurate biological clocks and providing a clearer assessment of aging [8, 9]. The determinants of biological aging remain largely unknown, yet lipid metabolism has been suggested to be involved [10–12].

Lipids are very diverse and complex biological compounds that play crucial roles in a variety of biological processes [13]. Indeed, age-related changes in complex lipids are linked to longevity and age-related diseases, serving as both biomarkers and contributors [12, 14–16]. Therefore, plasma lipid profiles hold promise as biomarkers for age-related changes and associated disease risks [15, 17, 18]. Certain lipid species in human plasma, such as ceramides (CERs), sphingomyelins (SMs), lysophosphatidylcholines (LPCs), and cholesteryl esters (CEs), tend to increase with age [15, 19]. Studies have also indicated higher levels of certain sphingomyelin (SM) species, such as SM(d18:1/18:2) and SM(d18:1/17:0), positively impact healthy aging [20]. Moreover, the concentration of plasma long-chain free fatty acids is inversely correlated with longevity across various mammalian species, suggesting their potential role as a biomarker for longevity [18]. Changes in plasma triacylglycerol (TG) levels and composition also occur with age [21]; four long-chain and highly polyunsaturated TG species, TG(54:7), TG(54:6), TG(56:7), and TG(56:6), are positively correlated to human longevity in the plasma lipidome [22, 23]. Additionally, factors related to lipid changes have been shown to regulate longevity [12, 18, 23]. For example, different types of intermittent fasting and caloric restriction can significantly improve the serum lipid profile by reducing TGs, total cholesterol, and low-density lipoprotein (LDL) cholesterol concentrations, which are associated with age-related diseases [24–26].

Furthermore, regular physical activity is associated with improved plasma lipid profiles in the elderly, including lower levels of TGs, total cholesterol, and LDL cholesterol, as well as higher levels of high-density lipoprotein (HDL) cholesterol [27–30]. Physical activity is therefore also associated with reduced cardiovascular risks and mortality in older people [27–29]. However, the understanding of how lipidomic patterns in the elderly are influenced by different physical activity statuses remains unclear. Understanding inter-individual differences in lipid characteristics and the effect of physical activity levels is a promising avenue of research.

We performed lipidomic profiling of plasma from an extensively characterized cohort of young and aged individuals [30, 31]. This study aimed to deepen our understanding of aging processes at the lipidomic level and explore the complex relationship between physical activity and lipid metabolism during aging.

Results

The lipidome of aging human plasma

To better understand which lipids are associated with aging and how physical active levels relate to plasma lipid aging, we performed semi-targeted lipidomic profiling of plasma from an extensively characterized cohort of young and older individuals (Fig. 1A, Table S1; NCT03666013, clinicaltrails.gov) [30, 31]. The specific cohort analyzed here included 14 young individuals (young adults) whereas the group of older participants consisted of exercise-trained older adults (trained older adults, n = 19), older adults with normal physical activity levels (normal older adults, n = 16), and physically impaired older adults (impaired older adults, n = 3)) [32]. The comparison between the group characteristics showed that trained old adults had a significantly higher movement level than the other groups (Supplementary Fig. 1). Meanwhile, there were notable differences in the age groups compared to young when assessing blood sugar level, body mass index (BMI), heart rate in beats per minute, systolic and diastolic blood pressure, and glucose infusion rate, while no differences in muscle mass (Supplementary Fig. 1). Of note, “impaired” older adults, while less able to perform physically, were not technically physically disabled. All participants were free of any diseases.

Fig. 1.

Fig. 1

Open in a new tab

The lipidome of aging human plasma. A Flowchart depicting the inclusion of study participants categorized as either young or older adults, with the subsequent acquisition of plasma samples for mass spectrometry-based lipidomic analysis. B Pie chart showing the number of lipids of each lipid class as a percentage of the total number of detected lipids in plasma. C Partial least squares discriminant analysis (PLS-DA) of lipidomes of young and older individuals possessing different lipids levels (young versus normal older adults. Young individuals, n = 14; normal older adults, n = 16). D The heatmap illustrates a comparative analysis of lipid profiles between young adults and normal older adults, showcasing variations in lipid abundance. Elevated levels are represented in red, while reduced levels are indicated in blue

We conducted a comprehensive analysis of plasma samples obtained from both young and older adults of the three different physical fitness levels (Fig. 1A, Table S2). We annotated 1446 unique lipid species distributed across 24 lipid classes (Fig. 1B, Table S2). To better understand the lipid profile changes that occur with aging, we performed a partial least squares discriminant analysis (PLS-DA) on all groups, which showed a general separation between young individuals and normal older adults (Fig. 1C), suggesting aging produces significant changes on plasma lipids. To explore the biological differences between young and older groups, we compared the lipid abundance between young individuals and older adults. Overall, we found a general increase in the levels of multiple lipids in older adults (Fig. 1D). The detection of this more general global increase is technically feasible on our lipidomics platform, as quantification is performed relative to an internal standard and not necessarily normalized to the total amounts measured. Together, these plasma lipidomic findings illustrate the distinction between young individuals and older adults, highlighting an age-associated increase in most lipids.

Prevalent accumulation of triacylglycerol (TG) species among older adults versus young individuals

We next analyzed the lipidome in the plasma of the three groups of older adults with varying physical activity levels by comparing the lipids that are more or less abundant. Here we found an age-associated increase in lipids in all three groups of older adults (Fig. 2A–C). Since only a few lipids were significantly depleted in the aged groups compared to young adults, we shifted our focus to those that increased. Within this set of significantly increased lipids, 54 lipids were significantly higher in all older adults (Fig. 2D). Of these, 48 were identified as TG species, five diacylglycerol (DG) species, and one phosphatidylinositol (PI) species (Fig. 2D, E).

Fig. 2.

Fig. 2

Open in a new tab

Prevalent accumulation of TG species among elderly versus young individuals. A–C Volcano plots of fold change (x-axis, log2 scale) versus adjusted p-value (y-axis, − log10 scale) for trained older adults (A), normal older adults (B), and impaired older adults (C) compared to young individuals. Values on x-axis greater than zero indicate accumulation with age, and less than zero indicates depletion with age. The horizontal line indicates significance (adjusted p-value < 0.05), and the vertical line shows absolute log2 fold change > 1. Significance was determined using an empirical Bayes-moderated t-test. p values were corrected for multiple hypotheses testing using the Benjamini–Hochberg method. D Venn diagram of the overlap of significantly higher abundant lipids in each aged group (trained older adults, normal older adults, and impaired older adults) compared to young individuals (p < 0.05 and log2 fold change > 1). E Heatmap of 54 lipids that accumulated in trained, normal, and impaired older adults compared to young adults. F Abundance of TGs in the four groups (young individuals, trained older adults, normal older adults, and impaired older adults). Global significance was determined using a two-way ANOVA. Sample sizes: young individuals, n = 14; trained older adults, n = 19; normal older adults, n = 16; impaired older adults, n = 3. *p < 0.05, **p < 0.01, ***p < 0.001

When investigating the total levels of lipid classes, we found that TG was not only one of the most accumulated lipid classes in healthy older adults when compared to young individuals but also displayed an association with activity levels, with more severe TG accumulation in impaired older adults (Fig. 2F). These findings are in line with numerous studies that have consistently reported plasma TG levels to be higher in older adults versus younger adults [21, 33–35]. Therefore, our lipidomics offers a lens to not only focus on TG abundance but also on chain length and saturation levels.

TG carbon chain length and saturation levels have different distributions in older adults compared to young individuals

To further investigate how subsets of TG species change with age, we investigated their composition further within the lipid space of TG chain length and saturation. Carbon chain length and the number of double bonds in those chains displayed divergent patterns (Fig. 3A–C). For instance, TG in trained older adults showed markedly higher levels of highly unsaturated lipids (> 4 double bonds) when compared to young individuals (Fig. 3A). Similarly, changes in the TG composition of normal older versus young also showed a significant increase of highly unsaturated TG (> 4 double bonds) (Fig. 3B). Conversely, impaired older adults exhibited a significant increase in lipids representing a broader range of double bonds, with increases in polyunsaturated fatty acid (PUFA)-containing species and also in species with fewer double bonds (≤ 4 double bonds) (Fig. 3C). Indeed, impaired older adults exhibited a high level of TGs with shorter carbon-chain lengths (< 56) in comparison to both normal older adults and trained older adults (Fig. 3D, E). The composition of aged groups compared to each other showed that both normal and trained older adults differed from impaired older adults in a highly similar way (Fig. 3D, E). Together, our data suggest that TGs with fewer double bonds (≤ 4 double bonds) and shorter carbon-chain lengths (< 56) accumulated more in impaired older adults, whereas trained older adults have more extremely unsaturated TG species.

Fig. 3.

Fig. 3

Open in a new tab

TG carbon chain length and saturation levels have different distributions in older adults compared to young adults. A–C Changes in the TG composition of trained (A), normal (B) and impaired (C) older adults versus young. The log2 fold change and adjusted p-value were calculated by comparing impaired older adults with young individuals. D–E Changes in the TG composition of trained older adults (D) and normal older adults (E) versus impaired older adults. The log2 fold change and adjusted p-value were calculated by comparing trained (D) and normal (E) older adults to impaired older adults. The red color indicates lipids accumulated more. Sample sizes are n = 14 for young individuals, n = 19 for trained older adults, n = 16 for normal older adults, and n = 3 for impaired older adults

Physical activity level is negatively correlated with carbon chain length and saturation levels in older adults

To further explore the relations between TG composition and physical activity statuses, we classified the TGs into 6 subgroups: (a) “shorter-saturated,” (b) “shorter-less-unsaturated,” (c) “shorter-more-unsaturated,” (d) “longer-saturated,” (e) “longer-less-unsaturated,” (f) “longer-more-unsaturated” (Fig. 4A).

Fig. 4.

Fig. 4

Open in a new tab

Physical activity level is negatively correlated with carbon chain length and saturation levels in older adults. A Classification of TGs into six categories based on carbon chain length and double bonds level, segregated at thresholds of 56 to distinguish between longer chain length and shorter chain length, and at 0 and 4 to delineate saturated and more or less unsaturated. B–D Abundance of shorter-less-unsaturated TG (B), shorter-more-unsaturated TG (C), and longer-more-unsaturated TG (D) in the four groups (young individuals, trained older adults, normal older adults, and impaired older adults). E, F Abundance of TG(69:6) (E) and TG(50:1) (F) in the four groups (young individuals, trained older adults, normal older adults, and impaired older adults). For TG subgroups (B–D), global significance was determined using a two-way ANOVA, while for lipid species (E, F), statistics were determined using an empirical Bayes moderated t-test. Sample sizes are n = 14 for young individuals, n = 19 for trained older adults, n = 16 for normal older adults, and n = 3 for impaired older adults. NS, not significant; *p < 0.05, **p < 0.01, ***p < 0.001

Then, we conducted a comparative analysis of TG subgroup abundance levels within the different groups (Fig. 4B–D). Our findings revealed that impaired older adults exhibited higher levels of shorter-less-unsaturated TGs (Fig. 4B), reflecting the previous pattern we identified of total TGs (Fig. 2F). Furthermore, the three groups of older adults all displayed a higher level of shorter-more-unsaturated TGs (Fig. 4C). Meanwhile, longer-more-unsaturated TGs were more accumulated in the trained older adults (Fig. 4D). These data demonstrate that the level of physical activity in older adults relates to levels of TG chain length and double bonds, where impaired older adults have a greater abundance of shorter-less-unsaturated TGs and trained older adults possess a greater abundance of longer-more-unsaturated TGs. Specific lipid species display similar trends in differentiating between older adults with varying physical activity levels. For example, the abundant TG(69:6), a longer-more-unsaturated TG, was most abundant in the trained older adults (Fig. 4E), while in contrast, TG(50:1), a shorter-less-unsaturated TG, was highest in the impaired older adults (Fig. 4F). Additionally, physical activity appeared also to be related to unsaturated TG levels (Supplementary Fig. 2A, B), but had no clear relation with longer-less-unsaturated TG (Supplementary Fig. 2C).

Association between physiological phenotypes and TG composition in older adults

To further delineate how the physiological phenotypes of our study participants were related to the lipidome, we performed a cross-correlation of all lipid classes to physical and physiological parameters in older individuals (Supplementary Fig. 3). Among various significant correlations here, we found that for TGs, BMI showed a positive correlation (Supplementary Fig. 3). Following this, we narrowed our search focused on TGs and analyzed the six categories of TGs (Fig. 4A) to physical parameters (Fig. 5A); particularly, shorter-less-unsaturated TGs, which accumulate more in physically impaired older adults, showed a positive correlation with BMI. However, there was no significant correlation between BMI and more-unsaturated TGs, nor was there a significant correlation between BMI and physical activity levels (movement level of adults) (Fig. 5A). However, we found that movement level (steps per day) showed a distinct negative correlation with the shorter-more-unsaturated TGs (R = 0.39, Spearman’s p = 0.017) (Fig. 5A, B), while correlations with other TG subsets were not significant. Similarly, whole-body insulin sensitivity (measured as glucose infusion rate during a glucose clamp) had a strong negative correlation with the shorter-more-unsaturated TGs (R = − 0.38, Spearman’s p = 0.022) (Fig. 5A, C). Furthermore, we found a positive correlation between diastolic and systolic blood pressure and shorter-more-unsaturated TGs (R = 0.39, Spearman’s p = 0.014 and R = 0.32, Spearman’s p = 0.05, respectively) (Fig. 5A, D, and E). We also found a positive correlation between diastolic blood pressure and longer-less-unsaturated TGs (R = 0.45, Spearman’s p = 0.005) (Supplementary Fig. 4A). Additionally, blood sugar levels were positively associated with both longer-less-unsaturated TGs and longer-saturated TGs (R = 0.33, Spearman’s p = 0.045 and R = 0.34, Spearman’s p = 0.034, respectively) (Supplementary Fig. 4B, C).

Fig. 5.

Fig. 5

Open in a new tab

Association between physiological phenotypes and TG composition in older adults. A Spearman’s correlation coefficient matrix illustrates associations between TG levels and physical conditions in a cohort of old groups (n = 38). Color scale: blue (negative correlation) to white (no correlation) to red (positive correlation). Data is available as source data from Grevendonk et al. (2021), as presented in Table 2 therein. B–E Scatterplots illustrating specific correlations between B average daily steps and shorter-more-unsaturated TG, C glucose infusion rate and shorter-more-unsaturated TG, D diastolic blood pressure and shorter-more-unsaturated TG, and E systolic blood pressure and shorter-more-unsaturated TG among the elderly (n = 38). Statistical significance was assessed using Spearman’s rank correlation coefficient. *p < 0.05, **p < 0.01, ***p < 0.001

Taken together, we here present a view of the strong relationship between the shorter-more-unsaturated TGs and movement levels in older adults, and our study also revealed a relationship between shorter-more-unsaturated TGs and systolic and diastolic blood pressure and glucose infusion rate.

Discussion

In the current study, we investigated age-related plasma lipidomic changes by comparing young and older individuals. To identify how plasma TG differences in older adults relate to physical activity, we evaluated three groups of older adults ranging in physical fitness. Our investigation revealed that (1) TGs accumulated with age regardless of more or less physical activity, though (2) older adults with lower physical activity levels show much higher TG levels, and finally (3) plasma triglyceride length and saturation level can help distinguish healthy aging groups in humans.

Although age is clearly a risk factor for metabolic disease, the mechanisms that contribute to this age-related risk have not been fully elucidated. One major contributing factor to metabolic disease is aberrant TG metabolism [21, 36]. Individuals with metabolic syndrome or type 2 diabetes (T2D) often exhibit elevated TG content in the heart and liver [36]. Furthermore, altered plasma TG metabolism plays a crucial role in the development of metabolic diseases [21]. Additionally, high levels of plasma TG serve as a risk factor for cardiovascular disease [37]. Many studies have confirmed that plasma TG levels are higher in older adults versus younger adults [21, 33, 35]. However, accumulation of TGs can also occur in athletes [38, 39]; therefore, understanding how TG lipid composition relates to health is of paramount importance. Considering that double bonds and carbon chain length are crucial features distinguishing lipids from general metabolites, our research delves deeply into the relationship between TG length, saturation level, and aging. A more comprehensive understanding of plasma TG’s role should facilitate the development of markers for risk prediction, diagnosis, prognosis, and response to therapies, potentially guiding new directions for therapeutic interventions. In line with this, our study demonstrated an association between shorter-more-unsaturated TGs and systolic and diastolic blood pressure, as well as insulin sensitivity.

According to the TG chain length and double bond distribution, we defined six TG subgroups and found a negative relationship between the shorter-more-unsaturated TGs and physical fitness levels in older individuals. Higher blood TG is an indication of cardiovascular disease, which is among the leading causes of death worldwide [40, 41]. Considerable evidence supports physical activity as an effective adjunctive intervention for abnormal lipid levels, with studies revealing that regular exercise can help lower TG levels in the blood [42–44]. Our study demonstrates that physical activity correlates with plasma TGs in older adults, particularly those with specific chain lengths and saturation levels, which are commonly elevated in older adults. Previous research has reported that longer lifespans in various species were correlated with a higher average fatty acid chain length in the plasma [45]. Conversely, shorter-less-unsaturated TG content was reported to be associated with an increased risk of metabolic disorders like type 2 diabetes, implying their accumulation during decreased metabolic activity [46]. This aligns with our study, which found that impaired older adults accumulate more shorter-less-unsaturated TGs. This suggests that carbon chain length and the number of double bonds also play a major role in healthy aging.

Studies have observed a moderate positive association between physical activity and circulating levels of behenic acid (22:0) and lignoceric acid (24:0), which are considered to be very long-chain fatty acids (VLCFAs) [47, 48]. This association suggests that physical activity might enhance the synthesis or availability of these fatty acids, although the precise mechanism remains unclear [47]. Physical activity has been shown to affect the expression and activity of enzymes involved in fatty acid metabolism [49, 50]. For instance, physical activity can reduce the activity of stearoyl-CoA desaturase (SCD1) in adipose tissue, which is involved in the synthesis of monounsaturated fatty acids [50]. This modulation could potentially explain the increased longer-more-unsaturated TGs in the trained older adults.

Several limitations exist in our study. For example, while the SPPB remains the gold standard for assessing physical ability [51], it would be possible to apply other approaches to define physical ability in our cohort. Additionally, a limitation stems from the small sample size of impaired older adults in our study (n = 3, compared to n = 14, 19, or 16, for young adults, trained older adults, or normal older adults, respectively). No statistical analysis was conducted to determine the appropriate sample size for this study (since this cohort was originally designed to study mitochondrial fitness in muscle aging [30]). This suggests that we may have missed potential other differences existing between the aging groups and the small sample size also makes it nearly impossible to stratify the results by gender. The small sample size for impaired older adults raises the possibility of potential bias that must therefore be reproduced in larger sample sizes before the current findings are considered conclusive. Nonetheless, our findings offer important trends and patterns that can guide further research. Finally, our investigation into the correlation between physical activity and TG levels is not founded on data collected through intentional exercise interventions imposed on subjects. Rather, it relies on analyzing participants’ health statuses and daily physical activities. Consequently, to delineate the precise influence of aerobic and anaerobic exercise on TG levels and composition with the health trend, a clinical trial with exercise would be warranted.

Some of our reasons for conducting plasma lipidomics were to identify biomarkers, candidate causal influencers in aging, and provide direction for future research. Our study provided evidence that aging and physical activity levels relate to TG carbon chain length and the number of carbon bonds. Shorter-more-unsaturated TGs are more prevalent in older adults, potentially serving as markers of aging. Specifically, TG species such as TG(54:7), TG(54:6), TG(56:7), and TG(56:6), previously identified as markers of aging [23], are also included in this group. Moreover, shorter-more-unsaturated TGs appear to be associated with healthier aging, whereas shorter-less-unsaturated TGs are more prevalent among less-healthy older adults. This suggests the clinical utility of assessing TG characteristics in marking health status among older individuals. Exploring the effects of different types of physical exercises on lipid profiles within these demographic groups will yield valuable insights. Furthermore, expanding the sample size to include individuals with diverse health statuses would enhance comprehension of the changes in TG carbon chain length and carbon bond numbers in response to health trends.

Recent studies have demonstrated that in regression models using “-omic” platforms to predict chronological age, the residual variation in predicted age correlates with health outcomes [52–54]. These findings suggest that “omic clocks” can serve as measures of biological age. Key metabolites, such as kynurenine, phenylalanine, and those involved in steroid hormone biosynthesis, have been identified for their high predictive accuracy in predicting age [53]. Additionally, consistent age–related changes in metabolites like pyruvate, glutathione, ATP, and NAD+ highlight their potential as global aging markers [32, 55]. Integrating metabolic aging clocks with DNA methylation and transcriptomic clocks could provide a more comprehensive measure of biological age [56–58]. Similarly, bis(monoacylglycerol)phosphate (BMP) and sphingomyelins, phosphocholines, ceramides, and fatty acids have been recognized as useful components of lipidomic aging clocks [12, 59, 60]. Lipidomic markers are advantageous as they reflect functional aging processes, are relatively accessible to measure, and are dynamically modulated by interventions targeting aging [61]. However, challenges including standardizing lipidomics platforms, accounting for confounders, validating findings across diverse populations, and identifying causal roles of specific lipids still remain. Overall, our findings implicate TG species subsets in healthy aging biology, and future work integrating these lipid-based markers into comprehensive aging clocks may serve to enhance precision measurements of healthy aging in the clinic.

Methods

Study participants

Fifty-two participants, including 14 young (7 male and 7 female) and 38 older (21 male and 17 female) individuals, were recruited in the community of Maastricht and its surroundings through advertisements at Maastricht University, local newspapers, supermarkets, and sports clubs. The study protocol was approved by the institutional Medical Ethical Committee and conducted in accordance with the declaration of Helsinki. All participants provided written informed consent and the study was registered at clinicaltrials.gov with identifier NCT03666013.

Human subjects and procedures

Participants were categorized into the following study groups: young individuals with normal physical activity (20–30 years), older adults with normal physical activity (65–80 years), trained older adults (65–80 years), and physically impaired older adults (65–80 years). Participants were considered normal and physically active if they completed no more than one structured exercise session per week. Participants were considered trained if they engaged in at least three structured exercise sessions of at least 1 h each per week for an uninterrupted period of at least 1 year (following a self-reported physical activity questionnaire). Participants were classified as older adults with impaired physical function in case of an SPPB score of ≤ 9. The SPPB score was calculated according to the cutoff points determined previously [62]. Following classification into groups, activity level (steps per day) and high-intensity activity times were objectively measured using an ActivPAL monitor Wearable device (PAL Technologies) that was worn for a consecutive period of 5 days. High-intensity activities (HPA) were defined based on the amount of steps taken in 1 min. Specifically, > 110 steps/min was considered HPA [30]. Young and normal older adults took an average of ~ 10,000 steps per day and spent < 1 h of their active time in high-intensity activities (Table S1). The trained older adults were more active (~ 13,000 steps per day on average) and had more time spent in high-intensity activities (> 3 h), whereas the physically impaired were less active (~ 6000 steps per day on average) and had less time spent in high-intensity activities (< 1 h) (Table S1). Subject characteristics are summarized in Table S1, with additional details available in the source data from Grevendonk et al. [30], as presented in Table 2 of the Nature Communications paper.

Plasma collection

At 9 AM, following an overnight fast starting at 10 PM the previous evening, blood was drawn from participants in a fasted state. A standardized meal was provided the night before, and plasma samples were immediately stored at − 80 °C for later analysis.

Lipidomics

In a 2-mL tube, the following amounts of internal standards dissolved in 1.5 mL 1:1 (v/v) methanol:chloroform were added to each sample: BMP(14:0)2 (0.2 nmol), CE(16:0)-D7 (2.5 nmol), CL(14:0)4 (0.1 nmol), DAG(14:0)2 (0.5 nmol), LPA(14:0) (0.1 nmol), LPC(14:0) (0.5 nmol), LPE(14:0) (0.1 nmol), LPG(14:0) (0.02 nmol), PA(14:0)2 (0.5 nmol), PC(14:0)2 (2 nmol), PE(14:0)2 (0.5 nmol), PG(14:0)2 (0.1 nmol), PI(8:0)2 (0.5 nmol), PS(14:0)2 (5 nmol), SM(d18:1/12:0) (2 nmol), and TAG(14:0)3 (0.5 nmol). Samples were thoroughly mixed, followed by centrifugation for 10 min at 20,000 g. The supernatant was transferred to a clean 1.5-mL tube and evaporated under a stream of nitrogen at 40 °C.

The residue was dissolved in 100 µL of 1:1 (v/v) methanol:chloroform. Lipids were analyzed using a Thermo Scientific Ultimate 3000 binary HPLC coupled to a Q Exactive Plus Orbitrap mass spectrometer. For normal phase separation, 2 µL of each sample was injected into a Phenomenex® LUNA silica (250 × 2 mm, 5 µm 100 Å). Column temperature was held at 25 °C. Mobile phase consisted of (A) 85:15 (v/v) methanol:water containing 0.0125% formic acid and 3.35 mmol/L ammonia and (B) 97:3 (v/v) chloroform:methanol containing 0.0125% formic acid. Using a flow rate of 0.3 mL/min, the LC gradient consisted of the following: Dwell at 10% A 0–1 min, ramp to 20% A at 4 min, ramp to 85% A at 12 min, ramp to 100% A at 12.1 min, dwell at 100% A 12.1–14 min, ramp to 10% A at 14.1 min, dwell at 10% A for 14.1–15 min. For reversed phase separation, 5 µL of each sample was injected into a Waters HSS T3 column (150 × 2.1 mm, 1.8 µm particle size). Column temperature was held at 60 °C. The mobile phase consisted of (A) 4:6 (v/v) methanol:water and (B) 1:9 (v/v) methanol:isopropanol, both containing 0.1% formic acid and 10 mmol/L ammonia. Using a flow rate of 0.4 mL/min, the LC gradient consisted of the following: Dwell at 100% A at 0 min, ramp to 80% A at 1 min, ramp to 0% A at 16 min, dwell at 0% A for 16–20 min, ramp to 100% A at 20.1 min, dwell at 100% A for 20.1–21 min. MS data were acquired using negative and positive ionization using continuous scanning over the range of m/z 150 to m/z 2000. Data were analyzed using an in-house developed lipidomics pipeline written in the R programming language (http://ww.r-project.org). All reported lipids were normalized to corresponding internal standards according to lipid class. Lipid identification has been based on a combination of accurate mass, (relative) retention times, previously obtained fragmentation spectra, analysis of samples with known metabolic defects, and the injection of relevant standards[63]. Lipidomics data from these experiments can be found in Table S3.

TG classification

We classified the TGs into 6 subgroups based on the saturation level and carbon chain length as follows: (a) “shorter-saturated” had the criteria of TG carbon chain length < 56 and double bonds = 0; (b) “shorter-less-unsaturated” had the criteria of carbon chain length < 56 and double bonds ≤ 4; (c) “shorter-more-unsaturated” had the criteria of carbon chain length < 56 and double bonds > 4; (d) “longer-saturated” had the criteria of carbon chain length ≥ 56 and double bonds = 0; (e) “longer-less-unsaturated” had the criteria of carbon chain length ≥ 56 and double bonds ≤ 4; (f) “longer-more-unsaturated” had the criteria of carbon chain length ≥ 56 and double bonds > 4.

Data analyses

Data were processed and analyses were performed with R version 4.2.2 [64]. Data were processed in part with the R package dplyr version 1.1.4 [65] and tidyverse version 2.0.0 [66].

Data statistics

Significant differences for lipid species between groups were assessed using an empirical Bayes moderated t-test on log2-transformed data within Limma’s linear model framework, taking participant groups into account. p values of significant differences between groups for lipid species were adjusted for multiple hypotheses testing using the Benjamini–Hochberg method (used e.g., in volcano plots). Global comparisons of groups were performed using an ANOVA (e.g., for global differences between TG lipid classes). Cross-correlations of parameters were performed in R using Spearman’s test. Comparisons between two groups are either empirical Bayes moderated t-test or Student’s t-test. p values are noted as asterisk in plots: *p < 0.05, **p < 0.01, ***p < 0.001. Statistical analyses were performed in the R programming language version 4.3.2 [67].

Data visualization

Visualization of data was performed in R version 4.3.2 [67] using ggplot2 version 3.4.4 [68], pheatmap version 1.0.12 [69], ggpubr version 0.6.0, and VennDiagram version 1.7.3 [70].

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 1140 KB) (1.1MB, docx)
Supplementary file2 (DOCX 19 KB) (19.1KB, docx)

Author contribution

Conceptualization: W.L., G.E.J., R.H.H.; human study design and collection of plasma: L.G., P.S., J.H.; formal analysis: W.L., B.V.S., G.E.J.; investigation: W.L., A.W.G.; data curation: B.V.S., G.E.J.; writing—original draft: W.L., A.W.G. G.E.J., R.H.H.; writing—review and editing: W.L., A.W.G. G.E.J., F.V., G.S.S., L.G., P.S., J.H., R.H.H.; supervision: G.E.J., R.H.H.; funding acquisition: G.E.J., R.H.H.

Funding

Work in the Janssens lab is supported by a VENI grant from ZonMw and AGEM Talent and Development grants. The project was supported by a Longevity Impetus Grant from Norn Group (to G.E.J. and R.H.H.). R.H.H is supported by the Velux Stiftung (no. 1063) and an NWO-Middelgroot grant (no. 91118006) from Netherlands Organization for Scientific Research (NWO). AWG is supported by an Amsterdam UMC Postdoc Career Bridging Grant, a grant from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement, and AGEM Talent and Development grant. Human interventions were further financed by the TIFN research program Mitochondrial Health (ALWTF.2015.5) and NWO. W.L. is financially supported by the CSC (China Scholarship Council).

Data availability

Lipidomics data are available as supplementary materials accompanying this manuscript as processed abundance values. All other data supporting the findings of this study are either also available as supplementary materials accompanying this manuscript or are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1.Tenchov R, Sasso JM, Wang X, Zhou QA. Aging hallmarks and progression and age-related diseases: a landscape view of research advancement. ACS Chem Neurosci. 2023 10.1021/ACSCHEMNEURO.3C00531/ASSET/IMAGES/LARGE/CN3C00531_0013.JPEG [DOI] [PMC free article] [PubMed]
  • 2.Maldonado E, Morales-Pison S, Urbina F, Solari A. Aging hallmarks and the role of oxidative stress. Antioxidants (Basel) 2023;12. 10.3390/ANTIOX12030651 [DOI] [PMC free article] [PubMed]
  • 3.López-Otín C, Blasco MA, Partridge L, et al. The hallmarks of aging. Cell. 2013;153:1194. 10.1016/j.cell.2013.05.039. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Singh PP, Demmitt BA, Nath RD, Brunet A. The genetics of aging: a vertebrate perspective. Cell. 2019;177:200–20. 10.1016/J.CELL.2019.02.038. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.van der Rijt S, Molenaars M, McIntyre RL, et al. Integrating the hallmarks of aging throughout the tree of life: a focus on mitochondrial dysfunction. Front Cell Dev Biol 2020;8. 10.3389/FCELL.2020.594416 [DOI] [PMC free article] [PubMed]
  • 6.López-Otín C, Blasco MA, Partridge L, et al. Hallmarks of aging: an expanding universe. Cell. 2023;186:243–78. 10.1016/J.CELL.2022.11.001. [DOI] [PubMed] [Google Scholar]
  • 7.Liu D, Aziz NA, Landstra EN, Breteler MMB. The lipidomic correlates of epigenetic aging across the adult lifespan: a population-based study. Aging Cell 2023;22. 10.1111/ACEL.13934 [DOI] [PMC free article] [PubMed]
  • 8.Ferrucci L, Gonzalez-Freire M, Fabbri E, et al. Measuring biological aging in humans: a quest. Aging Cell 2020; 19. 10.1111/ACEL.13080 [DOI] [PMC free article] [PubMed]
  • 9.Belsky DW, Caspi A, Houts R, et al. Quantification of biological aging in young adults. Proc Natl Acad Sci U S A. 2015;112:E4104–10. 10.1073/PNAS.1506264112/SUPPL_FILE/PNAS.1506264112.SAPP.PDF. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Avisar H, Guardia-Laguarta C, Area-Gomez E, et al. Lipidomics prediction of Parkinson’s disease severity: a machine-learning analysis. J Parkinsons Dis. 2021;11:1141–55. 10.3233/JPD-202476. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Afshinnia F, Rajendiran TM, He C, et al. Circulating free fatty acid and phospholipid signature predicts early rapid kidney function decline in patients with type 1 diabetes. Diabetes Care. 2021;44:2098–106. 10.2337/DC21-0737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Janssens GE, Molenaars M, Herzog K, et al. A conserved complex lipid signature marks human muscle aging and responds to short-term exercise. Nat Aging. 2024. 10.1038/S43587-024-00595-2. [DOI] [PubMed] [Google Scholar]
  • 13.Gille B, Galuska CE, Fuchs B, Peleg S. Recent advances in studying age-associated lipids alterations and dietary interventions in mammals. Front Aging. 2021;2:773795. 10.3389/FRAGI.2021.773795/BIBTEX. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Skowronska-Krawczyk D, Narayan P, Tessarz P. Editorial: molecular role of lipids in aging. Front Aging. 2022;3:946884. 10.3389/FRAGI.2022.946884/BIBTEX. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Johnson AA, Stolzing A. The role of lipid metabolism in aging, lifespan regulation, and age-related disease. Aging Cell 2019; 18. 10.1111/ACEL.13048 [DOI] [PMC free article] [PubMed]
  • 16.Schomakers BV, Hermans J, Jaspers YRJ, et al. Polar metabolomics in human muscle biopsies using a liquid-liquid extraction and full-scan LC-MS. STAR Protoc. 2022;3:101302. 10.1016/J.XPRO.2022.101302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Mc Auley MT, Mooney KM. Computationally modeling lipid metabolism and aging: a mini-review. Comput Struct Biotechnol J. 2015;13:38–46. 10.1016/J.CSBJ.2014.11.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Jové M, Naudí A, Aledo JC, et al. Plasma long-chain free fatty acids predict mammalian longevity. Sci Rep 2013; 3. 10.1038/SREP03346 [DOI] [PMC free article] [PubMed]
  • 19.Slade E, Irvin MR, Xie K, et al. Age and sex are associated with the plasma lipidome: findings from the GOLDN study. Lipids Health Dis. 2021; 20. 10.1186/S12944-021-01456-2 [DOI] [PMC free article] [PubMed]
  • 20.Jové M, Naudí A, Gambini J, et al. A stress-resistant lipidomic signature confers extreme longevity to humans. J Gerontol Ser A. 2017;72:30–7. 10.1093/GERONA/GLW048. [DOI] [PubMed] [Google Scholar]
  • 21.Spitler KM, Davies BSJ. Aging and plasma triglyceride metabolism. J Lipid Res. 2020;61:1161–7. 10.1194/JLR.R120000922. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Schoenmaker M, de Craen AJM, de Meijer PHEM, et al. Evidence of genetic enrichment for exceptional survival using a family approach: the Leiden Longevity Study. Eur J Hum Genet. 2006;14:79–84. 10.1038/SJ.EJHG.5201508. [DOI] [PubMed] [Google Scholar]
  • 23.Gonzalez-Covarrubias V. Lipidomics in longevity and healthy aging. Biogerontology. 2013;14:663–72. 10.1007/S10522-013-9450-7. [DOI] [PubMed] [Google Scholar]
  • 24.Almeida I, Magalhães S, Nunes A. Lipids: biomarkers of healthy aging. Biogerontology. 2021;22:273–95. 10.1007/S10522-021-09921-2. [DOI] [PubMed] [Google Scholar]
  • 25.Wadhera RK, Steen DL, Khan I, et al. A review of low-density lipoprotein cholesterol, treatment strategies, and its impact on cardiovascular disease morbidity and mortality. J Clin Lipidol. 2016;10:472–89. 10.1016/J.JACL.2015.11.010. [DOI] [PubMed] [Google Scholar]
  • 26.Gille B, Galuska CE, Fuchs B, Peleg S. Recent advances in studying age-associated lipids alterations and dietary interventions in mammals. Front Aging 2021;2:. 10.3389/FRAGI.2021.773795 [DOI] [PMC free article] [PubMed]
  • 27.Warburton DER, Nicol CW, Bredin SSD. Health benefits of physical activity: the evidence. CMAJ. 2006;174:801–9. 10.1503/CMAJ.051351. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Taylor AH, Cable NT, Faulkner G, et al. Physical activity and older adults: a review of health benefits and the effectiveness of interventions. J Sports Sci. 2004;22:703–25. 10.1080/02640410410001712421. [DOI] [PubMed] [Google Scholar]
  • 29.Ou SM, Chen YT, Shih CJ, Tarng DC. Impact of physical activity on the association between lipid profiles and mortality among older people. Sci Rep. 2017;2017(7):1–9. 10.1038/s41598-017-07857-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Grevendonk L, Connell NJ, McCrum C, et al. Impact of aging and exercise on skeletal muscle mitochondrial capacity, energy metabolism, and physical function. Nat Commun. 2021;12:1–17. 10.1038/s41467-021-24956-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Janssens GE, Grevendonk L, Perez RZ, et al. Healthy aging and muscle function are positively associated with NAD+ abundance in humans. Nat Aging. 2022;2:254–63. 10.1038/s43587-022-00174-3. [DOI] [PubMed] [Google Scholar]
  • 32.Janssens GE, Grevendonk L, Perez RZ, et al. (2022) Healthy aging and muscle function are positively associated with NAD+ abundance in humans. Nat Aging. 2022;2:254–63. 10.1038/s43587-022-00174-3. [DOI] [PubMed] [Google Scholar]
  • 33.Cassader M, Gambino R, Ruiu G, et al. Postprandial triglyceride-rich lipoprotein changes in elderly and young subjects. Aging (Milano). 1996;8:421–8. 10.1007/BF03339605. [DOI] [PubMed] [Google Scholar]
  • 34.Cohn JS, McNamara JR, Cohn SD, et al. Postprandial plasma lipoprotein changes in human subjects of different ages. J Lipid Res. 1988;29:469–79. 10.1016/s0022-2275(20)38520-5. [PubMed] [Google Scholar]
  • 35.Herzstein J, Wang CI, Adlersberg D. Fat-loading studies in relation to age. AMA Arch Intern Med. 1953;92:265–72. 10.1001/ARCHINTE.1953.00240200115014. [DOI] [PubMed] [Google Scholar]
  • 36.Unger RH, Clark GO, Scherer PE, Orci L. Lipid homeostasis, lipotoxicity and the metabolic syndrome. Biochim Biophys Acta. 2010;1801:209–14. 10.1016/J.BBALIP.2009.10.006. [DOI] [PubMed] [Google Scholar]
  • 37.Jansen GR, Kendall PA, Jansen CM. Diet evaluation: a guide to planning a healthy diet. Academic Press, 1990. p 279.
  • 38.Bergman BC, Perreault L, Strauss A, et al. Intramuscular triglyceride synthesis: importance in muscle lipid partitioning in humans. Am J Physiol Endocrinol Metab. 2018;314:E152. 10.1152/AJPENDO.00142.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Johnson NA, Stannard SR, Thompson MW. Muscle triglyceride and glycogen in endurance exercise: implications for performance. Sports Med. 2004;34:151–64. 10.2165/00007256-200434030-00002/METRICS. [DOI] [PubMed] [Google Scholar]
  • 40.Muscella A, Stefàno E, Marsigliante S. The effects of exercise training on lipid metabolism and coronary heart disease. Am J Physiol Heart Circ Physiol. 2020;319:H76–88. 10.1152/AJPHEART.00708.2019. [DOI] [PubMed] [Google Scholar]
  • 41.Buzdagli Y, Tekin A, Eyipinar CD, et al. The effect of different types of exercise on blood lipid profiles: a meta-analysis of randomized controlled studies. Sci Sports. 2022;37:675–87. 10.1016/J.SCISPO.2022.07.005. [Google Scholar]
  • 42.Trejo-Gutierrez JF, Fletcher G. Impact of exercise on blood lipids and lipoproteins. J Clin Lipidol. 2007;1:175–81. 10.1016/J.JACL.2007.05.006. [DOI] [PubMed] [Google Scholar]
  • 43.Hypertriglyceridemia: causes, risk factors & treatment. https://my.clevelandclinic.org/health/diseases/23942-hypertriglyceridemia. Accessed 22 Apr 2024
  • 44.Loeser RF. Age-related changes in the musculoskeletal system and the development of osteoarthritis. Clin Geriatr Med. 2010;26:371. 10.1016/J.CGER.2010.03.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Jové M, Naudí A, Gambini J, et al. A stress-resistant lipidomic signature confers extreme longevity to humans. J Gerontol A Biol Sci Med Sci. 2017;72:30–7. 10.1093/GERONA/GLW048. [DOI] [PubMed] [Google Scholar]
  • 46.Elingaard-Larsen LO, Villumsen SO, Justesen L, et al. Circulating metabolomic and lipidomic signatures identify a type 2 diabetes risk profile in low-birth-weight men with non-alcoholic fatty liver disease. Nutrients 2023;15. 10.3390/NU15071590/S1 [DOI] [PMC free article] [PubMed]
  • 47.Lai KZH, Yehia NA, Semnani-Azad Z, et al. Lifestyle factors associated with circulating very long-chain saturated fatty acids in humans: a systematic review of observational studies. Adv Nutr. 2023;14:99. 10.1016/J.ADVNUT.2022.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Imierska M, Kurianiuk A, Błachnio-Zabielska A. The influence of physical activity on the bioactive lipids metabolism in obesity-induced muscle insulin resistance. Biomolecules. 2020;10:1–20. 10.3390/BIOM10121665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Shanmugamprema D, Muthuswamy K, Ponnusamy V, et al. Exercise modifies fatty acid perception and metabolism. Acta Physiol. 2023;238:e13968. 10.1111/APHA.13968. [DOI] [PubMed] [Google Scholar]
  • 50.Mika A, Macaluso F, Barone R, et al. Effect of exercise on fatty acid metabolism and adipokine secretion in adipose tissue. Front Physiol. 2019;10:431145. 10.3389/FPHYS.2019.00026/BIBTEX. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Kameniar K, MacKintosh S, Van Kessel G, Kumar S. The psychometric properties of the short physical performance battery to assess physical performance in older adults: a systematic review. J Geriat Phys Ther. 2024;47:43–54. 10.1519/JPT.0000000000000337. [DOI] [PubMed] [Google Scholar]
  • 52.Hwangbo N, Zhang X, Raftery D, et al. A metabolomic aging clock using human cerebrospinal fluid. J Gerontol A Biol Sci Med Sci. 2022;77:744. 10.1093/GERONA/GLAB212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Zhang L, Mo S, Zhu X, et al. Global metabolomics revealed deviations from the metabolic aging clock in colorectal cancer patients. Theranostics. 2024;14:1602–14. 10.7150/THNO.87303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Han JDJ. The ticking of aging clocks. Trends Endocrinol Metab. 2024;35:11–22. 10.1016/J.TEM.2023.09.007. [DOI] [PubMed] [Google Scholar]
  • 55.Fang W, Chen S, Jin X, et al. Metabolomics in aging research: aging markers from organs. Front Cell Dev Biol. 2023;11:1198794. 10.3389/FCELL.2023.1198794/BIBTEX. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Hartmann A, Hartmann C, Secci R, et al. Ranking biomarkers of aging by citation profiling and effort scoring. Front Genet 2021; 12. 10.3389/FGENE.2021.686320/FULL [DOI] [PMC free article] [PubMed]
  • 57.Epigenetic Tests, human microbiome clocks & telomere length. Which aging biomarkers are best for calculating biological age? https://novoslabs.com/best-aging-biomarkers-for-biological-age/. Accessed 24 May 2024
  • 58.Jansen R, Han LKM, Verhoeven JE, et al. An integrative study of five biological clocks in somatic and mental health. Elife. 2021;10:1–20. 10.7554/ELIFE.59479. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Rahman SA, Gathungu RM, Marur VR, et al. Age-related changes in circadian regulation of the human plasma lipidome. Commun Biol. 2023;6:1–14. 10.1038/s42003-023-05102-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Unfried M, Ng LF, Cazenave-Gassiot A, et al. LipidClock: a lipid-based predictor of biological age. Front Aging 2022;3. 10.3389/FRAGI.2022.828239/FULL [DOI] [PMC free article] [PubMed]
  • 61.Ekroos K, Lavrynenko O, Titz B, et al. Lipid-based biomarkers for CVD, COPD, and aging - a translational perspective. Prog Lipid Res 2020;78. 10.1016/J.PLIPRES.2020.101030 [DOI] [PubMed]
  • 62.Guralnik JM, Simonsick EM, Ferrucci L, et al. A short physical performance battery assessing lower extremity function: association with self-reported disability and prediction of mortality and nursing home admission. J Gerontol 1994;49. 10.1093/GERONJ/49.2.M85 [DOI] [PubMed]
  • 63.Molenaars M, Schomakers B V., Elfrink HL, et al. Metabolomics and lipidomics in Caenorhabditis elegans using a single-sample preparation. DMM Disease Models and Mechanisms 2021;14. 10.1242/DMM.047746/237744 [DOI] [PMC free article] [PubMed]
  • 64.R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2013. https://www.R-project.org/.
  • 65.Hadley Wickham, Romain François, Lionel Henry, et al. dplyr: a grammar of data manipulation. In: R package version 1.1.4, 2023, https://github.com/tidyverse/dplyr. https://dplyr.tidyverse.org. Accessed 7 Feb 2024
  • 66.Function reference • tidyverse. https://tidyverse.tidyverse.org/reference/index.html. Accessed 12 Jun 2024
  • 67.R: The R project for statistical computing. https://www.r-project.org/. Accessed 12 Jun 2024
  • 68.Wickham Hadley. ggplot2: elegant graphics for data analysis. New York: Springer-Verlag; 2016. [Google Scholar]
  • 69.CRAN: package pheatmap. https://cran.r-project.org/web/packages/pheatmap/index.html. Accessed 12 Jun 2024
  • 70.CRAN: package VennDiagram. https://cran.r-project.org/web/packages/VennDiagram/index.html. Accessed 12 Jun 2024

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary file1 (DOCX 1140 KB) (1.1MB, docx)
Supplementary file2 (DOCX 19 KB) (19.1KB, docx)

Data Availability Statement

Lipidomics data are available as supplementary materials accompanying this manuscript as processed abundance values. All other data supporting the findings of this study are either also available as supplementary materials accompanying this manuscript or are available from the corresponding author upon reasonable request.

Articles from GeroScience are provided here courtesy of Springer

RESOURCES