New Insights into the Pathogenesis of Insulin Resistance in Humans Using Magnetic Resonance Spectroscopy


  • Kitt Falk Petersen,

    Corresponding author
    1. Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut
    2. Howard Hughes Medical Institute, Yale University School of Medicine, New Haven, Connecticut
    Search for more papers by this author
  • Gerald I. Shulman

    1. Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut
    2. Department of Cellular & Molecular Physiology, Yale University School of Medicine, New Haven, Connecticut
    Search for more papers by this author

Yale University School of Medicine, The Anlyan Center, Room S263, PO Box 9812, New Haven, CT 06536-8012. E-mail:


Insulin resistance plays a major role in the pathogenesis of type 2 diabetes, yet despite much effort, the underlying factors that are responsible for it are poorly understood. In this review, we focus on some recent advances in our understanding of the pathogenesis of insulin resistance in humans that have been made using magnetic resonance spectroscopy.


Type 2 diabetes is rapidly becoming a worldwide pandemic (1). While the primary cause of this disease is unknown, it is clear that insulin resistance plays a major role in its development. Cross-sectional studies have shown the presence of insulin resistance in virtually all patients with type 2 diabetes, and prospective studies have shown the presence of insulin resistance one to two decades before the onset of the disease (2, 3, 4). In addition, insulin resistance in offspring of parents with type 2 diabetes has been shown to be the best predictor for the later development of the disease (5). Finally, perturbations that reduce insulin resistance prevent the development of diabetes (6). It is, therefore, important to understand the pathogenic mechanisms of insulin resistance to identify novel targets for primary and secondary prevention. In this brief review, we will focus on some recent advances in our understanding of insulin resistance in humans that have been made using magnetic resonance spectroscopy (MRS)1 and propose a unifying hypothesis that may explain its pathogenesis under different pathological conditions.

Insulin Resistance and Muscle Glucose Metabolism

MRS has many advantages over conventional approaches for studying metabolism in humans. For example, it is non-invasive, and it involves no ionizing radiation. It can directly measure concentrations of intracellular metabolites such as glucose, glucose-6-phosphate, triglyceride, and glycogen, which typically cannot be measured accurately or in a repeated manner using conventional biopsy techniques. Using 13C MRS, we were able to non-invasively measure rates of muscle glycogen synthesis in patients with type 2 diabetes and found that, under steady-state plasma concentrations of insulin and glucose that mimic postprandial conditions, muscle glycogen synthesis was ∼50% lower in diabetic subjects than in normal volunteers and accounted for most of the whole body insulin stimulated glucose uptake in both normal and diabetic subjects (7). These studies show that under hyperglycemic, hyperinsulinemic conditions, muscle glycogen synthesis is the major pathway for glucose metabolism in both normal and diabetic individuals and that defective muscle glycogen synthesis plays a major role in causing insulin resistance in patients with type 2 diabetes. In a subsequent study, we combined 13C and 31P MRS studies to simultaneously measure rates of muscle glycogen synthesis along with concentrations of glucose-6-phosphate in patients with type 2 diabetes and age- and weight-matched control subjects (8). Intracellular glucose-6-phosphate is an intermediary metabolite between glucose transport/phosphorylation and glycogen synthesis, so the intracellular concentration of glucose-6-phosophate will respond to the relative activities of these two steps. In the event of decreased activity of glycogen synthase in diabetes, glucose-6-phosphate concentrations would be expected to increase in the diabetic patients relative to that of normal individuals. Glucose-6-phosphate cannot be accurately measured in humans using conventional techniques because its concentration increases because of hydrolysis of glycogen to glucose-6-phosphate during the muscle biopsy procedure. Using 31P MRS to non-invasively measure glucose-6-phosphate, we found that, in type 2 diabetic patients, the insulin-stimulated incremental changes in glucose-6-phosphate were significantly blunted, suggesting either decreased glucose transport activity or decreased glucose phosphorylation activity (through hexokinase II) as the cause of muscle insulin resistance in these subjects.

To examine whether this defect was a primary defect or an acquired defect secondary to other factors, we also studied lean normoglycemic insulin-resistant offspring of parents with type 2 diabetes (insulin-resistant offspring). These individuals have an ∼40% likelihood of developing diabetes later in life. Using a similar protocol, we found that these insulin-resistant offspring had a 50% reduction in the rate of insulin-stimulated whole body glucose metabolism, which could be attributed to a decrease in the rate of muscle glycogen synthesis (9). This reduction in insulin-stimulated muscle glycogen synthesis was associated with a blunted increment of insulin-stimulated intramuscular glucose-6-phosphate concentrations. Taken together, these data suggest that defects in insulin-stimulated muscle glucose transport/phosphorylation activity are very early events in the pathogenesis of type 2 diabetes.

We next examined whether chronic exercise training could reverse this defect in glucose transport/phosphorylation activity (10). Over a period of 6 weeks, a similar cohort exercised 40 minutes four times a week on a Stairmaster (Nautilus, Inc., Vancouver, WA) at 65% of their Vo2max. After this exercise regimen, insulin sensitivity and insulin-stimulated muscle glycogen synthesis normalized in the insulin-resistant offspring, and this could be attributed to correction of their defects in muscle glucose transport/phosphorylation activity. The results of this study strongly suggest that aerobic exercise might be useful in reversing insulin resistance in these pre-diabetic individuals and that it may prevent the development of type 2 diabetes. This hypothesis has recently gained support by the Diabetes Prevention Program study, which showed that the combination of diet and exercise was effective in decreasing the incidence of type 2 diabetes in patients with impaired glucose tolerance (11).

Finally, to delineate whether the activity of glucose transport or phosphorylation through hexokinase II was rate controlling for insulin-stimulated muscle glycogen synthesis in patients with type 2 diabetes, we developed a novel 13C MRS method to non-invasively assess intracellular free glucose concentrations in muscle (12). Intracellular glucose is an intermediary metabolite between glucose transport and glucose phosphorylation, and its concentration reflects the relative activities of these two steps in muscle glucose metabolism. A reduction in hexokinase II activity relative to glucose transport activity in type 2 diabetes should lead to substantial increments in intracellular glucose concentrations, whereas a primary reduction in glucose uptake by glucose transport should lead to proportional changes in intracellular glucose and glucose-6-phosphate concentrations. In patients with type 2 diabetes, we found the intracellular glucose concentrations were far lower than the concentrations expected if hexokinase II was the primary rate-controlling enzyme for glycogen synthesis. These data strongly suggest that defective insulin-stimulated glucose transport activity is the major factor responsible for insulin resistance in patients with type 2 diabetes.

Fatty Acid—induced Muscle Insulin Resistance

Increased plasma free fatty acid concentrations are typically associated with many insulin-resistant states, including obesity and type 2 diabetes (13). In a cross-sectional study of young, normal weight offspring of type 2 diabetic patients, we found an inverse relationship between fasting plasma fatty acid concentrations and insulin sensitivity, consistent with the hypothesis that altered fatty acid metabolism contributes to insulin resistance in patients with type 2 diabetes (14). Furthermore, recent studies measuring intramyocellular triglyceride content by 1H MRS have shown an even stronger relationship between intramyocellular triglyceride content and insulin resistance (14). Randle et al. (15) were the first to show that fatty acids compete with glucose for substrate oxidation in isolated rat heart and diaphragm muscle preparations. They speculated that increased fat oxidation was responsible for the insulin resistance associated with obesity and hypothesized that intracellular fatty acid accumulation would lead to an increase in the intramitochondrial acetyl coenzyme A (CoA)/CoA and NADH/NAD+ ratios, leading to inhibition of pyruvate dehydrogenase and increasing concentrations of intracellular citrate. The citrate accumulation would inhibit phosphofructokinase, a key rate-controlling enzyme in glycolysis, increasing intracellular glucose-6-phosphate concentrations and inhibiting hexokinase II activity. The inhibition of hexokinase II activity would result in an increase in intracellular glucose concentrations and decreased muscle glucose uptake.

A recent series of studies by our group have challenged this proposed mechanism of fat-induced insulin resistance (16). We applied 13C and 31P MRS to measure skeletal muscle glycogen and glucose-6-phosphate concentrations in healthy subjects under conditions of euglycemic hyperinsulinemia with high levels of plasma free fatty acids during an infusion of a lipid/heparin emulsion and in the absence of any exogenous fatty acids during an infusion of glycerol (the other metabolite released by lipolysis). Five hours of maintaining high levels of plasma fatty acid concentrations resulted in a 50% reduction in insulin-stimulated rates of muscle glycogen synthesis and whole body glucose oxidation compared with the control study. In contrast to the prediction by the model proposed by Randle et al., where fat-induced insulin resistance would result in an increase in intramuscular glucose-6-phosphate concentrations, we found that the drop in muscle glycogen synthesis was preceded by a decrement in intramuscular glucose-6-phosphate concentrations. Our findings suggest that increases in plasma fatty acid concentrations initially induce insulin resistance by inhibiting glucose transport and/or phosphorylation activity and that the reduction in muscle glycogen synthesis and glucose oxidation follows. The reduction in insulin-activated glucose transport/phosphorylation activity in normal subjects maintained at high plasma fatty acid levels is similar to the reduction seen in obese individuals (17), patients with type 2 diabetes (8), and lean, normoglycemic insulin-resistant offspring of type 2 diabetic individuals (9). Hence, it seems that accumulation of intramuscular fatty acid metabolites play an important role in the pathogenesis of insulin resistance in obese patients and patients with type 2 diabetes.

To further distinguish between possible effects of fatty acids on glucose transport activity and hexokinase II activity, we measured intracellular concentrations of glucose in muscle using 13C MRS (18). The logic of this experiment was similar to that described above. Because intracellular glucose is an intermediary metabolite between glucose transport and hexokinase II, the concentration reflects the relative activities of these two steps. A decrease in hexokinase II activity should lead to an increment in intracellular glucose concentrations and, if the impairment was at glucose transport, intracellular glucose concentrations should remain stable or decrease. We found that elevated plasma fatty acid concentrations significantly reduced intracellular glucose concentrations, implying that the rate-controlling step for fatty acid—induced insulin resistance in humans is glucose transport. The finding offers further evidence against the Randle mechanism, which predicts an increase in both intracellular glucose-6-phosphate and glucose concentrations.

This reduced glucose transport activity could be the result of fatty acid effects on the glucose transporter-4 (GLUT4) transporter directly—alterations in the trafficking, budding, fusion, or activity of GLUT4—or it could result from fatty acid—induced alterations in upstream insulin signaling events, resulting in decreased GLUT4 translocation to the plasma membrane. To explore the latter possibility, we examined insulin's effects on insulin receptor substrate 1 (IRS-1)–associated phosphatidylinositol 3-kinase (PI3K) activity in muscle biopsy samples under conditions of high or low plasma free fatty acid concentrations using the identical lipid infusion protocol described above. When plasma free fatty acid concentrations were high, the 4-fold increase in insulin-stimulated, IRS-1–associated PI3K activity observed in the glycerol control studies was abolished (18). The effects of intracellular fatty acids (or some fatty acid metabolite) may be a direct reduction of the insulin-stimulated PI3K activity or it may be secondary to alterations in upstream insulin signaling events. Consistent with an indirect effect, we found in rats that the lipid/heparin infusion raising plasma free fatty acids resulted in a reduction of insulin-stimulated IRS-1 tyrosine phosphorylation and activation of protein kinase Cθ, a serine kinase, which can be activated by fatty acid metabolites such as diacylglycerol (19).

Unifying Hypothesis for Insulin Resistance in Skeletal Muscle and Liver

A unifying hypothesis to explain the cause of several forms of insulin resistance in humans holds that increasing intracellular fatty acid metabolites (diacylglycerol, fatty acyl CoAs, etc.) may activate a serine/threonine kinase cascade (19, 20, 21, 22, 23, 24), which may be initiated by protein kinase Cθ in rodents (24) or protein kinase C (β and δ) in humans (25). Subsequent activation of other serine kinases may lead to phosphorylation of critical serine sites (e.g., Ser 307, Ser 612) on IRS-1 (25, 26, 27). Serine phosphorylated forms of IRS-1 fail to associate with and activate PI3K, resulting in decreased activation of glucose transport and other down-stream events. A similar mechanism has been shown to occur in liver involving diacylglycerol activation of PKCε, resulting in decreased insulin-stimulated IRS-2 tyrosine phosphorylation and hepatic insulin resistance (28, 29). If this hypothesis is correct, any perturbation that results in accumulation of intracellular diacylglycerol or other fatty acid metabolites in muscle and liver, through increased delivery from excess caloric intake or alterations in adipocyte fatty acid metabolism and/or through decreased mitochondrial fatty acid oxidation, might be expected to induce insulin resistance in muscle and liver (27). Indeed, all of these possibilities seem to occur in humans.

Increased Energy Intake Leading to Obesity and Insulin Resistance

The most common cause of insulin resistance occurs when energy intake exceeds energy expenditure, leading to obesity. Under these conditions, increased delivery of dietary fat and/or delivery of endogenous fat from increased hepatic lipogenesis results in accumulation of fatty acid metabolites in liver and muscle and subsequently causes insulin resistance through the mechanisms described above. This would mean that obesity invariably leads to insulin resistance; however, it is clear that not all obese individuals are insulin resistant. Vague first described that obesity where the fat is stored centrally (android or apple shape distribution) is typically associated with insulin resistance. In contrast, when obesity is associated with fat stored around the hips (gynoid or pear shape distribution), insulin sensitivity remains normal (30). Although it is widely believed that adipocyte-derived cytokines that are released by the visceral fat such as tumor necrosis factor-α, interleukin-6, adiponectin, and visfatin, may be responsible for the insulin resistance in android-shaped individuals, an alternative explanation is that visceral fat cells do not store fat as effectively as peripheral fat cells and that these individuals have redistributed their fat to intracellular sites in liver and muscle cells, resulting in insulin resistance in these tissues. In contrast, individuals with a gynoid distribution of fat (sometimes referred to as the “fit and fat” individual) store excess fat in peripheral adipocytes, which store fat more effectively and keep fat out of muscle and liver.

Defects in Adipocyte Fatty Acid Metabolism Leading to Insulin Resistance

Evidence for defects in adipocyte metabolism causing insulin resistance comes from recent studies in both transgenic mouse models of lipodystrophy and patients with severe lipodystrophy. Transgenic mice expressing A-ZIP/F-1 are almost totally devoid of fat because the A-ZIP/F-1 protein blocks the function of several classes of transcription factors (31). Interestingly, these fatless mice also manifest severe liver and muscle insulin resistance and develop diabetes (32, 33). Furthermore, they have a 2-fold increase in fatty acyl CoA content in muscle and liver and defects in insulin activation of IRS-1– and IRS-2–associated PI3K in both of these tissues (33). Transplantation of fat tissue from normal littermates into these fatless mice normalized fatty acyl CoA content in muscle and liver and normalized insulin signaling and action in liver and muscle (33).

Similar results have been seen in patients with severe lipodystrophy. These patients were found to have severe insulin resistance in both liver and muscle, which was associated with severe hepatic steatosis, but surprisingly, no increment of intramyocellular lipid, despite severe muscle insulin resistance (34). These data are consistent with recent studies that have shown that intramyocellular triglyceride is not the trigger in mediating muscle insulin resistance but more likely a marker for some other intracellular fatty acid metabolite such as diacylglycerol (24). Before leptin treatment, there was no rational treatment for severe, generalized lipodystrophy. Replacement leptin therapy for 3 to 8 months reversed both hepatic and muscle insulin resistance, and these changes were associated with a reduction in both intrahepatic triglyceride and intramyocellular lipid content (34). In these patients, the complete lack of fat cells combined with hyperphagia likely contributed to the excess fat storage in muscle and liver. The mechanism by which leptin replacement therapy caused this reduction in liver and muscle triglyceride content could most likely be attributed to its ability to reduce the hyperphagia in these patients.

These findings offer further evidence in support of the hypothesis that insulin resistance develops in obesity, type 2 diabetes, and lipodystrophy because of alterations in the partitioning of fat between the adipocyte and muscle and liver. This in turn leads to the accumulation of intracellular fatty acid metabolites (fatty acyl CoAs, diacylglycerol) in muscle and liver, which leads to activation of a serine kinase cascade, leading to defects in insulin signaling and insulin action in these tissues.

This hypothesis might also explain how thiazolidinediones improve insulin sensitivity in muscle and liver tissue: by activating peroxisome proliferator-activated receptor γ (PPARγ) receptors in the adipocyte and promoting adipocyte differentiation, these agents might promote a redistribution of fat from liver and muscle into the adipocyte, much as fat transplantation does in fat-deficient lipodystrophic mice (32, 33). This hypothesis is supported by a recent thiazolidinedione study in patients with type 2 diabetes. Three months of rosiglitazone therapy improved muscle insulin responsiveness and was associated with a marked reduction in intrahepatic fat content, a decrease in the intramyocellular/extracellular lipid content, and increased peripheral adipocyte insulin sensitivity to suppress lipolysis (35). Interestingly, rosiglitazone has also been shown to be effective in reversing skeletal muscle insulin resistance in lipodystrophic A-ZIP/F fatless mice (36). In this case, the liver seems to take over the role of the absent adipocyte in that rosiglitazone treatment results in a redistribution of fat (and fatty acyl CoAs) from muscle to liver, resulting in an increase in intrahepatic fat content (and intrahepatic fatty acyl CoA content) and worsening of hepatic insulin resistance while simultaneously improving muscle insulin sensitivity. Recent studies in muscle-specific PPARγ knock-out mice suggest that there may also be a direct effect of PPARγ in muscle in addition to these indirect effects (37).

Defects in Mitochondrial Function Leading to Insulin Resistance

It might also be expected that any alteration in the ability of muscle and liver to metabolize fatty acids, such as inherited or acquired defects in mitochondrial function, would lead to intracellular accumulation of fatty acid metabolites and subsequent defects in insulin signaling and action (31). Indeed, a recent study by our group using 13C and 31P MRS techniques to assess mitochondrial oxidative-phosphorylation activity showed that insulin resistance in the elderly could be attributed to increased intramyocellular and intrahepatic lipid content, which in turn was linked to a reduction in mitochondrial oxidative-phosphorylation activity (38). The reduction in mitochondrial function and lipid accumulation in muscle and liver can likely be ascribed to an age-associated reduction in mitochondrial content caused by accumulated mutations in mtDNA, which have been described to occur with aging (39). In more recent studies, using the same 31P MRS techniques to assess rates of ATP synthesis in skeletal muscle, we found similar reductions in mitochondrial activity associated with an increase in intramyocellular lipid content in young, lean, insulin-resistant offspring of parents with type 2 diabetes, a group that has a strong tendency to develop diabetes later in life (40). These alterations in mitochondrial function are consistent with studies that have shown reductions in the activity of rotenone-sensitive NADH:O (2) oxidoreductase in isolated muscle mitochondria obtained from type 2 diabetic subjects (41). Taken together, these data suggest that alterations in nuclear encoded genes that regulate mitochondrial biogenesis may form the genetic basis for inheritance of type 2 diabetes (42, 43, 44, 45).

Reversal of Hepatic Steatosis, Hepatic Insulin Resistance, and Hyperglycemia by Moderate Weight Reduction in Patients with Type 2 Diabetes

Previous studies have shown that a relatively modest weight reduction in obese poorly controlled type 2 diabetic subjects can markedly reduce plasma glucose concentrations, although the mechanism responsible for this phenomenon was poorly understood (46). Based on these previous studies, we hypothesized that a relatively small pool of intrahepatic lipid (i.e., hepatic diacylglycerol) might be responsible for the hepatic insulin resistance and fasting hyperglycemia in these individuals. To examine this hypothesis, we measured rates of hepatic glucose production and hepatic insulin sensitivity along with intrahepatic and intramyocellular lipid contents weekly by 1H MRS in eight obese type 2 diabetic subjects with moderate to poor glycemic control at baseline, during 7 weeks of weight loss (1200 kcal/d), and after weight stabilization. The diabetic subjects had increased rates of fasting hepatic glucose production, which was associated with severe hepatic and peripheral insulin resistance. These changes were associated with severe hepatic steatosis and marked increases in intramyocellular lipid content. A weight loss of only ∼8 kg resulted in normalization of fasting plasma glucose concentrations and rates of fasting hepatic glucose production. In addition, hepatic insulin sensitivity normalized, which could be attributed to an almost total reversal of their hepatic steatosis. In contrast, there was no improvement in peripheral insulin resistance or change in their intramyocellular lipid content.

In a subgroup of patients, rates of net hepatic glycogenolysis and gluconeogenesis were assessed before and after weight loss by 13C MRS. Before the weight reduction, the increased rates of glucose production could be entirely accounted for by increased rates of gluconeogenesis. After weight loss, the reduced rates of glucose production could be entirely attributed to a reduction in the rates of gluconeogenesis. These data support the hypothesis that moderate weight loss normalizes fasting hyperglycemia in poorly controlled type 2 diabetic patients by mobilizing a relatively small pool of hepatic lipids, which reverses hepatic insulin resistance and normalizes rates of basal glucose production, independent of any changes in insulin-stimulated peripheral glucose metabolism (47).


Insulin resistance plays a major role in the pathogenesis of type 2 diabetes. In this short review, we discuss some recent MRS studies that potentially shed new light on the pathogenesis of insulin resistance in humans. Specifically, 1) insulin resistance in skeletal muscle can mostly be attributed to defects in insulin-stimulated glucose transport activity; 2) in contrast to the original mechanism proposed by Randle et al. where fatty acids cause insulin resistance by inhibiting pyruvate dehydrogenase activity, we found that fatty acids cause insulin resistance in human skeletal muscle by directly interfering with insulin-stimulated glucose transport activity; 3) reduced insulin activation of glucose transport activity can be attributed to an acquired defect in insulin stimulated IRS-1–associated PI3K activity at the level of IRS-1 tyrosine phosphorylation. This, in turn, can be ascribed to fatty acid activation of a serine kinase cascade, which causes increased serine phosphorylation of IRS-1 at critical sites that interfere and block tyrosine phosphorylation of IRS-1 tyrosine sites that are required to bind and activate PI3K (a similar mechanism seems to occur in the liver involving protein kinase Cε, leading to decreased IRS-2–associated PI3K activity); 4) any perturbation that leads to an increase in intramyocellar fatty acid metabolites (e.g., fatty acyl CoAs, diacylglycerol) content such as acquired (e.g., aging) or inherited (e.g., potential type 2 diabetes genes) defects in mitochondrial fatty acid oxidation or defects in adipocyte fat metabolism (e.g., lipodystrophy), leading to increased fat delivery to liver and muscle, will lead to insulin resistance through this final common pathway. Understanding these key cellular mechanisms of insulin resistance should help elucidate new targets for treating and preventing type 2 diabetes.


This study was supported by grants from the U.S. Public Health Service (R01 AG-23686, R01 DK-40936, and P01 DK-68229) and the American Diabetes Association.


  • 1

    Nonstandard abbreviations: MRS, magnetic resonance spectroscopy: CoA, coenzyme A; GLUT4, glucose transporter-4; IRS-1, insulin receptor substrate 1; P13K, phosphatidylinositol 3-kinase; PPARγ, peroxisome proliferator-activated receptor γ.