Introduction
Exercise provides health benefits for individuals with type 1 diabetes, through reducing CVD, improving skeletal muscle health and increasing fitness [
1,
2]. Despite the benefits of exercise, barriers to its use as a diabetes therapy remain. Exercise tolerance is impaired in type 1 diabetes [
3‐
10] and the metabolic response to exercise can be acutely disrupted, with high inter-individual variability, leading to dysglycaemia [
11‐
14]. This is despite high intra-individual reproducibility under repeated laboratory conditions, suggesting the existence of distinct subpopulations within the type 1 diabetes population exhibiting perturbed metabolic responses to exercise [
15].
The variability in metabolic response to exercise in people with type 1 diabetes is not fully explained, is poorly classified and cannot be predicted [
15]. A number of physiological contributors to exercise response are disrupted in people with type 1 diabetes, possibly contributing to inter-individual variation [
15]. These include perturbations to fuel selection [
16] in the form of carbohydrates (glycolysis) and lipids (NEFA, triacylglycerols, acylcarnitines), and anaerobic processes (lactate) and central carbon metabolism (tricarboxylic acid [TCA] cycle) [
16].
The phenotype and population heterogeneity of aerobic capacity in exercise and metabolic dysfunction in type 1 diabetes makes effective screening, diagnosis and management a substantial challenge, contributing to lower activity levels within the type 1 diabetes population [
16‐
18]. Omic approaches are a potential solution. Metabolomics has advanced understanding of metabolic diseases and is ripe for application to the heterogeneous pathological metabolic phenotype induced by exercise in people with type 1 diabetes, having become a core tool in personalised medicine [
19‐
21]. Metabolomics is effective in defining ‘silent’ phenotypes [
22] (that manifest under specific stressors) and identifying disease biomarkers preceding overt pathology [
20,
21]. Therefore, metabolomics is highly appropriate for studying the metabolic perturbation in type 1 diabetes that acutely manifests under stress (increased metabolic rate, carbohydrate oxidation, and insulin sensitivity) in exercise. The study of the metabolome alongside exercise indices, such as maximal aerobic capacity (
\(\dot{V}{\text{O}}_{\text{2peak}}\)), may provide novel information regarding an individual’s metabolic phenotype and distinguish those with lower or higher aerobic capacity.
Another unexplored mechanism possibly contributing to the metabolic response to exercise is residual beta cell function (i.e. C-peptide secretion) in people with type 1 diabetes of long duration. Between 35% and 80% of people with type 1 diabetes have detectable beta cell function at >5 years post diagnosis [
23,
24], with 5–16% at the threshold for clinical benefits (peak C-peptide 200 pmol/l) found in the DCCT. As beta cell function declines in individuals with type 1 diabetes, glycaemic control deteriorates [
25,
26]. Individuals with high stimulated C-peptide (>400 pmol/l) spend greater time in euglycaemia at rest compared with those with undetected, low (17–200 pmol/l) and intermediate (200–400 pmol/l) levels, suggesting an influence on basal metabolism [
27]. Individuals with higher beta cell function (>200 pmol/l C-peptide) display lower glycaemic variability than individuals with low and undetectable C-peptide following aerobic exercise [
28]. This suggests that residual beta cell function may contribute to variation in acute blood glucose levels following exercise in type 1 diabetes. Since C-peptide status influences metabolism (glycaemic control) post exercise, it may also contribute to the metabolic response to exercise. Application of metabolomics in people with type 1 diabetes during aerobic exercise, alongside characterisation of C-peptide status, may provide valuable insight into the impact of residual beta cell function on metabolic phenotype.
We aimed to use targeted metabolomics to characterise the metabolic signature of type 1 diabetes at rest, exercise and recovery. We explored the contribution made by residual beta cell function to metabolic response to exercise in individuals with type 1 diabetes and sought to identify circulating metabolite markers diagnostic of maximal aerobic capacity to facilitate the personalised medicine approach to diabetes.
Discussion
We used targeted metabolomics to define the metabolic baseline phenotype and response to exercise in serum of individuals with type 1 diabetes and control individuals. We showed that residual beta cell function contributes to the metabolic phenotype of individuals with type 1 diabetes both at rest and in response to acute aerobic exercise. During aerobic exercise, NEFAs are released from triacylglycerol in adipose tissue via lipolysis and enter the circulation to provide fuel for contracting muscle, where they are converted to ACs to enter the mitochondria and undergo β-oxidation [
42‐
44]. This response is reflected in our data by the increased serum NEFAs common to control and diabetic participants in aerobic exercise. Unique to people with type 1 diabetes was an increase in serum ACs in aerobic exercise, a feature that also distinguishes the metabolic effect of lower residual beta cell function. We identified that the serum metabolome of both control and diabetic participants at rest was related to maximal aerobic capacity. In healthy volunteers, this relationship was characterised by an inverse correlation between medium/long-chain NEFAs and
\(\dot{V}{\text{O}}_{\text{2peak}}\). In diabetic participants this relationship was distinctly described by an inverse correlation with TCA cycle metabolites malic acid and
cis-aconitate and a positive correlation with TCA cycle substrate pyruvate. A key driver for reduced physical activity in people with type 1 diabetes is fear of hypoglycaemia [
18], although this is multifactorial, with exercise intolerance [
3‐
10] and high inter-individual variation in the metabolic response to exercise [
16,
17] contributing. Such barriers limit the use of exercise as a therapy and reduce population-wide uptake, with low maximal aerobic capacity contributing to worse outcomes and risk of diabetes complications [
15]. Therefore, we used ROC curves to identify the diagnostic potential of metabolites for maximal aerobic capacity. The malic acid/pyruvate ratio was revealed to be a putative diagnostic marker for differentiating maximal aerobic capacity specifically in individuals with type 1 diabetes. Recently, altered mitochondrial function has been described in the skeletal muscle in type 1 diabetes [
2,
45,
46]. Moreover, blood ACs are a marker of mitochondrial (dys)function [
44,
47]. Therefore, perturbed serum ACs and TCA cycle intermediates, mitochondrial metabolite species, during exercise may reflect dysfunction of skeletal muscle mitochondria in type 1 diabetes.
Study limitations include participant recruitment from a single cohort and centre. Most of the diabetic participants had good glycaemic control and exercise testing was conducted under laboratory conditions with acute exercise intervention and close monitoring of blood glucose levels. Although our experimental groups did not differ for sex and therefore our findings may be generalisable across sexes in the type 1 diabetes population, we did not directly investigate interacting variables (e.g. age, sex and duration of diabetes, which were not significantly different between groups in our study) that may contribute to both the metabolome and maximal aerobic capacity in the diabetic population.
Exogenous insulin may blunt the metabolic response to exercise in type 1 diabetes [
48]. We did not observe this phenomenon, possibly because our volunteers with type 1 diabetes arrived fasted, maintaining only basal exogenous insulin delivery. In addition, although C-peptide is a robust equimolar insulin secretion marker and indicator of beta cell function in response to an MMTT [
49], other factors such as insulin resistance and renal impairment can affect serum C-peptide concentration [
50]. However, these increase C-peptide concentrations [
50] and given our volunteers have normal kidney function and did not exhibit between-group difference for HbA
1c, are unlikely to have affected our study. Additionally, C-peptide has direct insulin-independent metabolic effects on several tissues (e.g. sensory nerves, vasculature) [
31]. The C-peptide-associated metabolic responses to exercise may be influenced by these direct effects and not residual beta cell insulin secretion alone. This may be pertinent, as the high-C-peptide group had a slightly shorter duration of diabetes than comparator groups. Nevertheless, four volunteers within this group had diabetes for >15 years, and recruitment of volunteers ≥3 years post diagnoses reduces the impact of the early honeymoon phase.
To our knowledge, we are the first to suggest a circulating metabolome-derived marker of maximal aerobic capacity in people with type 1 diabetes. This approach may facilitate future studies validating the serum malic acid/pyruvate ratio as a biomarker of maximal aerobic capacity in the wider type 1 diabetes population. Our approach may expedite identification of patients with type 1 diabetes and lower maximal aerobic capacity to target structured support for exercise. Moreover, future, higher-powered, studies will be important to investigate the influence of other interacting variables (e.g. age of diabetes onset, diabetes duration) and could focus on whether metabolomic signatures associate with increased hypoglycaemic risk during and post exercise. This approach could enable the identification of individuals at highest risk and generate personalised medicine strategies to reduce exercise-induced hypoglycaemia in those most in need. It will be important to establish the metabolomic response to exercise in fasting, fed and high and low exogenous insulin conditions, while carefully avoiding hypo- and hyperglycaemia, to determine the contribution of exogenous insulin to the metabolic response to aerobic exercise in type 1 diabetes. Studies of the interaction between the metabolome and exercise capacity in differing exercise modalities such as resistance and high-intensity interval training, with physiological markers distinct from \(\dot{V}{\text{O}}_{\text{2peak}}\) and over longer durations, are needed. The analysis of blood ACs and TCA cycle intermediates in people with type 1 diabetes, with concomitant assessment of muscle mitochondrial function, may inform personalised medicine approaches to target aerobic exercise programmes to improve muscle health.
In summary, people with type 1 diabetes exhibit a unique metabolic phenotype in response to aerobic exercise, characterised by increased circulating ACs. We identify a metabolic signature of perturbed serum TCA cycle intermediates at rest, correlating with maximal aerobic capacity in people with type 1 diabetes. The serum malic acid/pyruvate ratio may have diagnostic potential in determining individuals with type 1 diabetes and lower maximal aerobic capacity. The putative metabolite markers may inform identification, management and therapy of lower maximal aerobic capacity in exercise in people with type 1 diabetes.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.