Introduction

Diabetes continues to be a major public health problem, currently affecting over 425 million people worldwide [1]. This number is expected to increase due to the growing rates of obesity, which contributes to a rising incidence of type 2 diabetes and a projected increase in the prevalence of type 1 diabetes globally [2, 3]. Despite the different aetiologies of type 1 and type 2 diabetes, both are associated with a number of complications affecting the cardiovascular system, kidneys, eyes and nerves [4]. Cardiovascular disorders, heart failure, atherosclerosis and cerebrovascular events primarily result from damage to the macrovasculature. The other major complications have been long considered as ‘microvascular’ injuries and manifest as a ‘diabetic triopathy’ of diabetic kidney disease, diabetic retinopathy and diabetic neuropathy [4]. These complications account for the majority of morbidity and mortality risks associated with diabetes. While glucose and glucose metabolism have been the focus of research to understand the pathophysiology of these complications for decades, altered lipid metabolism has recently received increasing attention as a key player in disease pathology. In this review, we address the recent progress made using preclinical models and human-based approaches to shed new light on the relative roles of glucose and lipid metabolism in the onset and progression of diabetic microvascular complications.

Clinical overview of diabetic complications

Diabetic kidney disease

Diabetic kidney disease affects 30–40% of all individuals with diabetes and is the leading cause of end-stage renal disease in the USA, accounting for significant morbidity and mortality risks in individuals with type 1 and type 2 diabetes [5]. Clinically, diabetic kidney disease is characterised by increased urinary albumin output and a decline in the glomerular filtration rate, both reflecting a progressive deterioration of renal function [6]. This loss of function coincides with histopathological changes, including glomerular and tubular hypertrophy with gradual thickening of the glomerular and tubular basement membranes, podocyte effacement and eventually glomerulosclerosis and tubulointerstitial fibrosis [7].

Diabetic retinopathy

Diabetic retinopathy is the primary cause of blindness among the working-age population worldwide. After 20 years of diabetes, virtually all individuals with type 1 diabetes and more than 60% of those with type 2 diabetes eventually develop signs of diabetic retinopathy [8]. After years of clinically silent intraretinal changes, vascular tortuosity, retinal haemorrhage, microaneurysm, cotton-wool spots and lipid exudates become evident as features of non-proliferative diabetic retinopathy [9]. Some affected individuals develop pathological retinal neovascularisation, recognised as proliferative diabetic retinopathy. Fluid accumulation within the central neural retina, referred to as diabetic macula oedema (DME), manifests as abnormal retinal thickening and cystoid formation and is the most common cause of visual loss in individuals with diabetic retinopathy [10, 11].

Diabetic neuropathy

Diabetic neuropathy is a highly prevalent complication of both type 1 and type 2 diabetes, affecting at least 50% of individuals with diabetes [12]. Various types of peripheral nerve disorders can develop, the most common being distal symmetric polyneuropathy, affecting nerves of the extremities in a bilateral, symmetric pattern and progressing in a distal-to-proximal manner. Distal symmetric polyneuropathy, referred to as diabetic neuropathy from this point onwards, is a highly debilitating complication associated with an increased susceptibility to ulcerations and infections and may eventually lead to lower-limb amputations. In addition to the well-characterised axonal injury [13], diabetes also targets Schwann cells and the vascular endothelium as demonstrated by segmental demyelination and endoneurial microangiopathy [14].

Mechanisms of onset and progression of diabetic complications

The importance of intensive glycaemic control in reducing the progression of complications in type 1 diabetes was confirmed in the Diabetes Control and Complications Trial (DCCT) and Epidemiology of Diabetes Interventions and Complications (EDIC) trials [15]. Based on these studies, patient care has been focused on intensifying glucose control, which improved but did not prevent the development and progression of all three diabetic complications in type 1 diabetes. In type 2 diabetes, intensive glycaemic control has a modest effect on these complications [16]. The UK Prospective Diabetes Study [17] and the Steno-2 trial [18] demonstrated amelioration of diabetic kidney disease and diabetic retinopathy following 10 years of intensive glucose-lowering therapy in individuals newly diagnosed with type 2 diabetes but no effect on diabetic neuropathy was found. Other large intervention trials failed to report clear beneficial effects of near normalisation of blood glucose levels on all three complications associated with type 2 diabetes. The Action to Control Cardiovascular Risk in Diabetes (ACCORD), Action in Diabetes and Vascular Disease (ADVANCE) and Veteran Affairs Diabetes Trial (VADT) [19] trials showed modest improvements in the early manifestations of diabetic kidney disease, while no improvements in diabetic retinopathy or diabetic neuropathy were observed between the intensive and the conventional therapy groups [20, 21]. Collectively, these studies suggest that glucose alone is not responsible for diabetic complications, especially in individuals with type 2 diabetes. Rather, responsibility lies with a cluster of factors, including obesity, hypertension, dyslipidaemia, inflammation and insulin resistance, which have an impact on the adipose tissue fatty acid metabolism that underlies the onset and progression of diabetic kidney disease, retinopathy and neuropathy [22]. Thus, this review will focus on our increased understanding of the basic physiology underlying the global and tissue-specific changes in cellular metabolism that contribute to the pathogenesis of diabetic complications.

Role of hyperglycaemia in tissue-specific injury

The mechanisms underlying hyperglycaemia-mediated cellular damage include formation of advanced glycation end-products, increased oxidative stress, mitochondrial dysfunction and activation of the polyol and hexosamine pathways [23]. While preclinical studies have supported these hypotheses, clinical trials that targeted each of these pathways failed to confirm a direct pathogenic role for hyperglycaemia-induced metabolic imbalance in diabetic individuals [24, 25]. The disappointing outcomes of these clinical interventions led our group to ask a fundamental question: do complication-prone tissues (i.e. kidney, eye and nerve) metabolise glucose in a similar or tissue-specific manner under diabetic conditions? We combined metabolomics with transcriptomics to examine changes in glucose metabolism in the kidney, eye and nerve of a murine model of type 2 diabetes, the leptin receptor knockout BKS db/db mouse model. We discovered that glycolytic genes were uniformly upregulated in the kidney and peripheral nerve, while glycolytic metabolites were increased in the kidney and retina but decreased in the peripheral nerve [26]. Furthermore, we systemically administered substrates (glucose, pyruvate or palmitate) labelled with the stable isotope 13C to investigate how the metabolism of each of these substrates differed in the kidney, eye and nerve between diabetic and control mice. Analysis of metabolic flux, or tracing of each substrate through downstream metabolites in glycolysis, the pentose phosphate pathway, β-oxidation (acylcarnitines) and the tricarboxylic acid (TCA) cycle, identified tissue-specific differences in substrate utilisation in this type 2 diabetic mouse model. Particularly, glucose flux increased in the kidney, moderately increased in the retina and remained unchanged in the nerve (Table 1), highlighting differences in glucose metabolism in diabetes complications [26]. To our knowledge, no other study has examined alterations in glucose metabolism across complications within the same animal model of diabetes. Besides this study, transcriptomic studies, in combination with metabolomic and proteomic analysis of tissues from both animal models of diabetes and human biopsies from diabetic individuals, have implicated both glucose-dependent and independent mechanisms [26,27,28,29,30,31,32,33]. Using metabolomics analysis, Priyadarsini et al found no effect of glucose-derived metabolite flux in human corneal stroma of individuals with type 1 and type 2 diabetes [33]. On the other hand, Freeman et al showed an increase in glucose and polyol pathway intermediates in sciatic nerves of a streptozotocin (STZ)-induced rat model of type 1 diabetes using an integrated proteomics and metabolomics approach [32]. Importantly, all these studies support the involvement of key factors other than glucose in the pathogenesis of diabetic kidney disease, retinopathy and neuropathy, including a central role for disturbances in lipid metabolism [26,27,28,29,30,31,32,33,34].

Table 1 Pathway summary of in vivo metabolic flux analysis in kidney cortex, sciatic nerve and retina from 24-week-old diabetic mice compared with control mice

In the following sections, we will discuss these alternative mechanisms and highlight the utility of systems biology approaches for improving our understanding of the differential roles of glucose and lipid metabolism in tissue-specific complications pathophysiology.

Rise of the role of lipid metabolism in diabetic complications

The fact that the risk of diabetic complications persists despite improved glycaemic control in type 2 diabetes suggests that other components of the metabolic syndrome may play a role in the onset and progression of these complications. Our group has focused on understanding how dyslipidaemia and tissue-specific lipid metabolism contribute to diabetic kidney disease, retinopathy and neuropathy, and the intersection of lipid metabolism with hyperglycaemia [26] (Fig. 1).

Fig. 1
figure 1

Systemic and tissue-specific alterations in lipid metabolism specifically associated with the microvascular complications: diabetic kidney disease, diabetic retinopathy and diabetic neuropathy. FA, fatty acid; LDL-c, LDL-cholesterol. This figure is available as part of a downloadable slideset

Lipogenesis and dyslipidaemia

Dyslipidaemia is defined as elevated levels of plasma triacylglycerols and cholesteryl esters, which are components of low-density lipoprotein (LDL)-cholesterol and high-density lipoprotein (HDL)-cholesterol. Increased hepatic lipogenesis and adipose tissue fatty acid metabolism in response to elevated blood glucose and insulin resistance contribute to dyslipidaemia in both type 1 and type 2 diabetes [35, 36]. Multiple preclinical and clinical studies have proposed dyslipidaemia as a major contributor to the pathophysiology of diabetic kidney disease (reviewed in [37]). In the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study, treatment with fenofibrate, a lipid-lowering drug, was found to reduce albuminuria and delay the decline of renal function in individuals with type 2 diabetes [38]. However, the effect of statins (antihyperlipidemic drugs) on diabetic kidney disease progression remains controversial. While several preclinical and clinical studies have shown a beneficial effect of statin treatment on kidney function [39, 40], other clinical trials failed to show any improvement in estimated glomerular filtration rate in individuals with type 2 diabetes [41, 42]. Additionally, it remains unclear whether the beneficial effects of statin therapy are the result of a pleiotropic action of the drug on renal function, independent of its lipid-lowering effect [43].

Controversy also exists regarding the exact role of dyslipidaemia in diabetic retinopathy [44,45,46]. Several small uncontrolled studies examining the effects of statin use on diabetic retinopathy risk found a beneficial effect [47, 48] but were contradicted by larger studies [49, 50]. On the other hand, a recent well-conducted large study in Taiwan, of more than 18,000 individuals with type 2 diabetes with age and sex-matched controls, found that statin use significantly reduced retinopathy risk, diminished the incidence of macular oedema and reduced the need for further interventions [51]. Moreover, the FIELD and ACCORD Eye studies found that fenofibrate use significantly lowered the risk of diabetic retinopathy progression and/or the need for laser intervention in diabetic individuals [52, 53]. However, meta-analysis of studies examining the contribution of serum lipids to the risk of diabetic retinopathy confirmed that while fenofibrate treatment is beneficial, only a slight increase in systemic LDL-cholesterol was associated with diabetic retinopathy risk [54]. Recent studies in diabetic rodents suggest that fenofibrate improves retinal pathology by exerting direct effects on the tissue itself as it improves neuronal survival and function, promotes neuroprotective mechanisms and reduces local inflammation [55,56,57,58].

Emerging evidence indicates that dyslipidaemia is a major player in the pathogenesis of diabetic neuropathy [59, 60]. In these studies, increased plasma triacylglycerols correlated with neuropathy progression in type 2 diabetes while obesity and systemic dyslipidaemia were independently associated with an increased risk of developing neuropathy. In the FIELD study, lipid-lowering therapy was associated with a reduced risk of lower-limb amputations, suggesting that lipid lowering may have beneficial effects on diabetic neuropathy [61]. In the aforementioned study of more than 18,000 individuals with type 2 diabetes from Taiwan, statin therapy significantly reduced the risk of new-onset diabetic neuropathy and foot ulcers [51]. While the mechanisms involved are not fully understood, oxidised LDL-cholesterol induced oxidative stress has been shown to mediate nerve damage in a mouse model of dyslipidaemia-induced neuropathy [62] and likely plays a similar role in neural injury in diabetic neuropathy [13].

Collectively, these results suggest that circulating lipids may have distinct effects on ‘microvascular’ complications. While increasing evidence suggests an important role for lipid metabolism in diabetic complications, future mechanistic studies are needed to provide more insight into the contribution of dyslipidaemia to diabetes-induced kidney, eye and nerve damage.

Distinct patterns of lipid metabolism in diabetes complications

Studies examining the role of dyslipidaemia in the onset or progression of diabetic complications typically measure plasma levels of LDL, HDL and total triacylglycerols and cholesterol. However, a limitation of measuring the levels of these lipids alone is that this does not effectively reflect the aberrant tissue-specific lipid metabolism associated with diabetes. The complex relationships between plasma lipid levels and diabetic kidney disease, retinopathy and neuropathy suggest that the gross level of these lipid levels in plasma may be inadequate as a biomarker for the prediction of risk of progression. Moreover, glucose is a single molecule but lipids represent a complex mixture of molecular species and, while numerous studies have examined the effects of diabetes on glucose metabolism, few studies have explored how diabetes alters lipid content and metabolism within complication-prone tissues. To address this knowledge gap, our group has used integrated approaches to link systemic and tissue-specific lipid metabolism to diabetic complications pathologies [26, 31]. We have measured levels of multiple lipid species in plasma, kidney, retina and nerve in the BKS db/db mouse [30]. In doing so, we uncovered tissue-specific alterations in several lipid classes, both in their level and directionality, which were defined by class, carbon side-chain length and saturation levels (Fig. 2). While alterations in individual lipids were mostly tissue-dependent, the kidneys and nerves from diabetic mouse models exhibited an overall trend towards increased levels of numerous lipid species. In contrast, overall lipid content in the retinas of diabetic mice was generally decreased. In line with these findings, extensive lipid accumulation has been reported within kidney glomeruli and tubulointerstitium in individuals with type 2 diabetes who had diabetic kidney disease [63]. Moreover, using post-mortem human retina samples, we recently obtained data suggesting alterations in lipid metabolism in retinas from donors with diabetic retinopathy (P. E. Fort, unpublished data), consistent with the reduction observed in BKS db/db mice [30]. The importance of these findings is highlighted by the observation that genetic deletion of the fatty acid synthase gene in the neuroretina leads to progressive neurodegeneration, with synaptic dysfunction, neuronal cell apoptosis and visual impairment [64]. As these pathological changes are also observed in diabetic retinopathy, the possibility arises that impaired lipid synthesis may contribute to the neurodegenerative effects of diabetes on the retina. Similarly, transcriptomic analysis showed altered lipid metabolism in human sural nerve biopsies obtained from diabetic individuals with progressive vs stable diabetic neuropathy [34, 65]. Taken together, these findings suggest a complicated balance and interaction between systemic and tissue-specific lipid levels and warrant further analysis of the relative importance of dyslipidaemia and tissue-specific lipid metabolism in the pathogenesis of diabetic complications.

Fig. 2
figure 2

Diabetes is associated with tissue-specific lipid alterations. The Venn diagram represents significantly altered lipid features observed in the BKS db/db mouse model of diabetes compared with control. Only 15 lipid features were significantly altered across complication-prone tissues, including diacylglycerol, phospholipids and cardiolipins. The direction of change in these shared features varied across different tissues with only four of the 15 common to all three tissues changing in the same direction (number of features with common direction is shown in parentheses), further demonstrating the tissue-specific nature of the lipid alterations associated with diabetes. Data derived from [30]. This figure is available as part of a downloadable slideset

Fatty acid regulation

Insulin directly regulates circulating fatty acids by controlling the conversion of non-esterified fatty acids to triacylglycerol and its subsequent secretion from the liver. As such, diabetes, low insulin levels and insulin resistance disrupt plasma fatty acid and complex lipid composition [66]. Insulin resistance is linked to changes in levels of several lipid classes, including triacylglycerols, while diabetes risk correlates with lower fatty acid carbon number and lower degree of fatty acid saturation [67]. Recently, lipidomic analysis was used to associate plasma fatty acid and triacylglycerol composition with progression of kidney disease in diabetes [27, 29]. Of note, as diabetic kidney disease progresses and the glomerular filtration barrier deteriorates, the kidney is likely to encounter higher levels of albumin-bound fatty acids in the filtrate, potentially increasing the impact of dyslipidaemia on tubular cells.

In the retina, highly enriched polyunsaturated fatty acids (PUFAs) have been intensely studied in the context of their role in regulating vascular function and angiogenesis [68]. In models associated with late stages of diabetic retinopathy, lipidomic analyses identified a reduction in some of the largest n-3 PUFAs [68], while transcriptomics revealed reduced expression of the enzymes involved in PUFA synthesis [69]. Interestingly, dietary manipulations focusing on PUFAs caused a positive impact on pathological retinal neovascularisation in an animal model of oxygen-induced retinopathy. This effect may be related to the role of PUFAs in inflammation [70] or in the synthesis of very-long-chain ceramides, which seem to maintain vascular integrity in experimental diabetes [71].

Diabetic neuropathy, of the three complications, most closely associates with dyslipidaemia [16]. Obesity induced by a diet rich in saturated fatty acids is associated with the development of diabetic neuropathy in individuals with type 2 diabetes [72] and, as with diabetic retinopathy, PUFA supplementation ameliorates symptoms of diabetic neuropathy [73]. This clinical finding is strongly supported by the preclinical work of Yorek and colleagues (reviewed in [74]), that shows clinical benefit of PUFA treatment in multiple rodent models of diabetic neuropathy. Moreover, transcriptomic analysis reveals an upregulation in fatty acid metabolism in the nerves of BKS db/db mice with diabetic neuropathy [75]. In cellular studies, high concentrations of the saturated fatty acid palmitate impair sensory neuron mitochondrial trafficking and alter mitochondrial energy production [76]. However, supplementation of cultured neurons with monounsaturated fatty acids such as oleate prevents palmitate-induced mitochondrial abnormalities through lipid droplet formation [77]. We also found that cultured Schwann cells increase the expression of medium-to-long-chain acylcarnitines in response to long-chain fatty acid treatment. Accumulation of these acylcarnitines has been associated with neurotoxicity and insulin resistance [78, 79].

Collectively, these preclinical and clinical studies suggest an important role for fatty acid dysregulation in all three diabetic complications. Overall, saturated fatty acids are damaging to complication-prone tissues, while supplementation with PUFAs may restore normal function and provide a treatment modality. While more research is needed, the concept of ‘bad’ (saturated) and ‘good’ (unsaturated) fatty acids holds true at the present time.

Lipid β-oxidation

Peroxisome proliferator-activated receptors (PPARs) and PPAR, gamma, coactivator 1, alpha (PGC1-α) play a key role in the regulation of fatty acid oxidation as they regulate the expression of proteins responsible for fatty acid uptake, such as the multifunctional fatty acid translocase CD36. In animal models and humans, PPAR signalling is highly dysregulated in diabetes [34, 75, 80, 81]. CD36 is upregulated in kidneys of diabetic animals and individuals with diabetes [63, 82], and is involved in mediating the apoptotic and oxidative effects of oxidised LDL immune complexes in metabolically challenged human retinal pericytes [83]. Similarly, CD36 expression is increased in peripheral nerves of BKS db/db mice and individuals with progressive diabetic neuropathy [34, 75].

Fatty acid profiles in tissues not only depend on plasma levels and uptake capacity, as discussed above, but also on the local lipid metabolism in these tissues. We hypothesised that, like glucose, there would be tissue-specific alterations in lipid β-oxidation with diabetes. Indeed, using a combination of transcriptomic and metabolomic approaches to assess the impact of type 2 diabetes on β-oxidation, we demonstrated a tissue-specific response of this pathway in BKS db/db mice [26]. Whereas transcriptomic analyses revealed increased expression of genes for fatty acid oxidation enzymes in the diabetic kidney and nerve, diabetes increased levels of the metabolites associated with fatty acid metabolism and β-oxidation in the kidney and retina and decreased these metabolites in peripheral nerves. The use of palmitic acid labelled with the stable isotope 13C to analyse the effects of type 2 diabetes on the metabolic flux through fatty acid catabolism showed increased palmitate uptake in all three tissues. Moreover, the analysis confirmed a progressive increase in β-oxidation in the kidney only, while suggesting incomplete fatty acid catabolism in retina and peripheral nerves. Consistent with these findings, Schwann cells exposed to high levels of long-chain fatty acids undergo incomplete fatty acid catabolism. This leads to altered acylcarnitine accumulation, which is known to have neurotoxic properties [84], as discussed earlier in this review.

We observed increased kidney β-oxidation in a preclinical murine model of diabetic kidney disease, while transcriptomic analysis using kidney tissue from individuals with chronic kidney disease revealed decreased expression of fatty acid oxidation enzymes in the presence of fibrosis [85]. This dichotomy suggests that lipid metabolism may be differentially regulated throughout disease progression, something that has not been thoroughly examined in each of the complication-prone tissues. On a tissue-specific level, lipid β-oxidation may regulate tissue function by inducing epigenetic changes [86] and metabolic intermediates of lipid β-oxidation, such as succinate and fumarate, have been shown to stimulate pathogenic pathways [87, 88]. Additionally, lipid β-oxidation (especially of very-long-chain fatty acids) may mediate diabetes-induced increases in oxidative stress, potentially to a greater degree than increased glucose catabolism, thus driving diabetic complications [89, 90].

Lipids and mitochondrial dysfunction

Given their role in energy metabolism, mitochondria are presumably a key interaction point between glucose metabolism, lipid metabolism and diabetic complications. In addition to undergoing β-oxidation in mitochondria (as discussed above), lipids also directly affect mitochondrial function (reviewed in [91]). Cardiolipins (inner mitochondrial membrane lipids) are important for mitochondrial structure and mitochondrial metabolism. We recently identified an increase in immature cardiolipins in the diabetic kidney [30], which is associated with reduced bioenergetics [92], a hallmark of diabetic kidney disease [93]. Interestingly, preserving mitochondrial structure with a cardiolipin-binding peptide maintains mitochondrial function in the kidney [94, 95], reverses visual decline in diabetic mice [96] and improves bioenergetics and mitochondrial trafficking in neurons [97]. Additionally, exposure of Schwann cells to a mixture of long-chain fatty acids similar to that observed under diabetic conditions impairs mitochondrial function and induces oxidative stress [84]. As discussed above, saturated fatty acids damage sensory axons by depolarising neuronal mitochondria and impairing mitochondrial trafficking [76]. Supplementation with monounsaturated fatty acids prevents mitochondrial dysfunction through lipid droplet formation, highlighting again a complex role for lipid products and metabolism in diabetes complications [77].

Lipids in inflammation

Integration of omics studies have revealed correlations, if not causal relationships, between altered lipid metabolism and inflammation. PUFAs, especially n-3 PUFAs, can regulate inflammation by inhibition of NF-κB activation and nuclear translocation [98, 99]. In contrast, n-6 PUFAs are precursors of bioactive lipids such as arachidonic acid, which would itself promote a proinflammatory environment and development of diabetic complications. Elevated arachidonic acid-derived eicosanoids in serum independently predict diabetic kidney disease progression [29], while altered plasma fatty acid composition is associated with retinal and systemic inflammation [69]. Multiple studies suggest that dietary manipulations that increase the ratio of n-3 to n-6 PUFAs, decrease the risk of diabetic complications in adult participants with type 2 diabetes [100,101,102]. Lipid levels are dependent on multiple mechanisms of regulation including dietary intake and de-novo synthesis, and also on the exchange and recycling systems. Among those regulatory mechanisms, reverse cholesterol transport and its regulation by specific receptors, namely the farnesoid X receptor (FXR) and the liver X receptor (LXR), has been receiving increasing attention in relation to inflammation, especially in the pathogenesis of diabetic kidney disease [103, 104] and diabetic retinopathy [105]. LXR activation was also shown to improve peripheral nerve function in rodents with obesity and diabetes-induced neuropathy, specifically by reducing endoplasmic reticulum stress or restoring myelin lipid composition; however, little is known about the anti-inflammatory effects of LXR in the setting of neuropathy [106, 107]. These studies are intriguing and certainly warrant further exploration of the implication of these mechanisms in the onset and progression of diabetic complications.

Other aspects of lipid metabolism have been suggested to impact systemic as well as local inflammation. Multiple studies using unbiased approaches analysing the transcriptome, proteome or metabolome of various tissues affected by diabetes have reported lipid metabolism and inflammation as being among the top pathways affected [26, 85, 108,109,110,111]. These studies in animal models of diabetes are now complemented and supported by studies in individuals with diabetes. For example, we recently used transcriptomic analysis to demonstrate that abnormal lipid metabolism and inflammation in peripheral nerves are conserved across multiple mouse models of type 2 diabetes and in individuals with progressive diabetic neuropathy [65]. Moreover, several studies found positive associations between serum markers of inflammation, lipid metabolism and onset and progression of complications in individuals with type 1 and type 2 diabetes [29, 112, 113].

Treatment of diabetic complications: one size does not fit all

Tissues with distinct structures and functions are affected by diabetes, suggesting the involvement of tissue-specific mechanisms in the pathogenesis of the disease. In the kidney, metabolism is associated with haemodynamic factors and diminishing filtration, which allow waste products to accumulate in blood [114]. In the retina, metabolic alterations have been proposed to be responsible for dysregulation of the glutamine–glutamate cycle in retinal Müller glial cells, an effect believed to be associated with glutamate excitotoxicity and cell death [115]. In the peripheral nerves, metabolic irregularities in Schwann cells are thought to induce mitochondrial deficits, accumulation of neurotoxic species and transfer of lipotoxic species to axons, a series of injurious events that may lead to axonal degeneration [24, 79]. Altogether, these studies, while not being exhaustive, highlight once more some of the ‘tissue/cell’ specificity of the impact of diabetes and emphasise the need for global analysis to assess the tissue-specific mechanisms at play in order to develop targeted interventions.

Conclusions

While it is tempting to combine ‘microvascular’ complications under one mechanism, it is clear that diabetes dysregulates metabolism in a tissue-specific manner. A role for generalised mechanisms cannot be fully excluded but the current body of knowledge strongly suggests that tissue-specific metabolic alterations, rather than a single unifying metabolic mechanism, are driving the onset and progression of diabetic complications. Metabolic dysregulation includes not only hyperglycaemia but also alterations in systemic and local lipid metabolism. Therapeutic development for diabetic complications will require a better understanding of the crosstalk between glucose and lipid metabolism in individual tissues, if not individual cell types, which most likely differs in each tissue despite the common underlying condition. Additionally, while the large amount of information gathered from numerous animal models of diabetes relative to the pathophysiology of these complications has been critical, difficulties associated with translation still highlight the need for human-based studies. Discovery approaches, especially those performed in human tissues across the gene–transcript–protein–metabolite continuum, have already demonstrated a potential for identifying tissue-specific adaptations and will continue to be the starting point in unveiling the mechanisms of diabetic complications and identifying novel therapeutic targets. Ultimately, a better integration and communication between researchers working on the respective complications, combined with a broader use of human samples and clinical data, is necessary to perform integrated analysis of complications pathophysiology, characterise their respective mechanisms and define new treatment options.