Fatty Acids and Insulin Modulate Myocardial Substrate Metabolism in Humans With Type 1 Diabetes

We studied 34 type 1 diabetic subjects and 12 nondiabetic control subjects. We chose to study patients with type 1 diabetes to avoid the possible confounding effects of obesity and hypertension that often accompany type 2 diabetes (14,15). Type 1 diabetes classification was based on the need for supplemental insulin during the first year, history of ketoacidosis, and plasma C-peptide level <0.50 μmol/ml. No subject had active retinopathy, clinically significant autonomic neuropathy, or serum creatinine >1.5 mg/dl. Sedentary subjects were chosen to minimize confounding effects of training-induced adaptations on substrate metabolism (16). All were nonsmokers and normotensive. Cardiac disease was excluded by a normal physical exam and normal rest/exercise echocardiograms. Subjects were not taking any vasoactive medications at the time of the study and did not have other systemic illnesses. Type 1 diabetic subjects were randomly assigned to one of three groups. Twelve were studied under resting euglycemic conditions, 10 during an Intralipid infusion, and 12 during an HIEG clamp. The nondiabetic control subjects were age, sex, and BMI matched with the euglycemic type 1 diabetic subjects. The study was approved by the Human Studies Committee and the Radioactive Drug Research Committee at the Washington University School of Medicine. Written informed consent was obtained from all subjects before enrollment in the study.

All studies were performed on a Siemens tomograph (ECAT 962 HR+; Siemens Medical Systems, Iselin, NJ). All subjects were admitted overnight to the General Clinical Research Center and fasted for 12 h before the study. Two intravenous catheters were placed: one for infusion and one for blood sampling. All were studied at 8:00 a.m. to avoid circadian variations in metabolism (17). In the euglycemia group, the substrate environment was standardized using a low-dose insulin infusion with or without D5W infusion to maintain blood glucose level in the 80- to 120-mg/dl range. The Intralipid subjects were given a heparin bolus intravenously (7.0 units/kg heparin) followed by a continuous infusion of heparin/Intralipid mixture (5, 000 units heparin in 500 cc Intralipid) at 0.7 ml · kg⁻¹ · h⁻¹. They also received a regular insulin infusion with or without D5W infusion to maintain blood glucose in the 80- to 120-mg/dl range. Subjects in the HIEG clamp group were started on the clamp 2 h before the PET imaging session (18). Nondiabetic control subjects were studied after the same fasting period as that undergone by the type 1 diabetic subjects. All subjects were on telemetry and had blood pressures obtained throughout the study. Rate-pressure product was calculated as systolic blood pressure × heart rate.

Myocardial oxygen consumption (MVo₂), blood flow, fatty acid, and glucose metabolism were measured after injections of ¹⁵O-water, ¹¹C-acetate, 1-¹¹C-palmitate, and 1-¹¹C-glucose, respectively. During the study, plasma substrates and insulin levels were measured. ¹¹CO₂ and ¹¹C-lactate activity were also measured to correct the PET data as required for compartmental modeling of the myocardial kinetics of the metabolic tracers (19,20).

Myocardial ¹⁵O-water, ¹¹C-acetate, 1-¹¹C-palmitate, and 1-¹¹C-glucose images were generated. Blood and myocardial time-activity curves were generated as reported previously (19–22). The curves were used in conjunction with well-established kinetic models to quantify myocardial blood flow, MVo₂, fatty acid, and glucose extraction fractions (20,22). Myocardial fatty acid/glucose utilization was calculated as the product of myocardial fatty acid/glucose extraction fraction × myocardial blood flow × (plasma FFA/glucose). Myocardial fatty acid extraction fraction was also divided into the portion of extracted fatty acids oxidized and the portion that entered slow turnover pools or esterification. Myocardial fatty acid oxidation was calculated as the fraction of the oxidized myocardial fatty acids × myocardial blood flow × (plasma FFAs). Myocardial fatty acid esterification was calculated as the fraction of the myocardial fatty acids that entered slow-turnover pools × myocardial blood flow × (plasma FFAs).

Plasma insulin levels were measured by radioimmunoassay. Plasma glucose and lactate levels were measured using a glucose-lactate analyzer (Yellow Springs Instruments, Yellow Springs, OH). Plasma FFA level was determined by capillary gas chromatography and high-performance liquid chromatography.

During the PET study, subjects had a complete two-dimensional and Doppler echocardiographic examination using a Sequoia-C256 (Acuson-Siemens, Mountain View, CA). Left ventricular mass index, ejection fraction, relative wall thickness, and Em were obtained as previously described (14,23).

SAS software (SAS Institute) was used for statistical analyses. Data are expressed as means ± SD. Comparisons of continuous and categorical variables between nondiabetic control subjects and the euglycemic type 1 diabetes group were made using unpaired Student's t and χ² tests, respectively. Comparisons among the three type 1 diabetic groups were made using ANOVA for continuous variables and the χ² test for categorical variables. Pearson correlations were performed to determine the univariate relationships between the independent variables (age, BMI, diabetes duration, A1C, rate-pressure product, and plasma insulin level) and the predetermined dependent variables (myocardial glucose extraction fraction and utilization; myocardial fatty acid utilization, oxidation, and esterification; percentage of myocardial fatty acid utilization oxidized; and percentage of myocardial fatty acid utilization esterified). A Pearson correlation was also used to determine the relationship between plasma insulin and plasma FFAs and the relationship between plasma FFA level (during the 1-C¹¹-glucose imaging) and myocardial glucose extraction fraction and utilization. (The relationship between plasma FFA and the dependent variables regarding myocardial fatty acid utilization and metabolism was not analyzed because plasma FFA is used in their calculation.) For entry into multivariate analyses, an independent variable was required to have a significant relationship with the dependent variables at the P < 0.10 level. Type III sum of squares multivariate analyses were used to determine the independent predictors of dependent variables.

Patients in the euglycemic type 1 diabetic group and nondiabetic control subjects were not different in terms of age, sex, race, metabolic profiles, left ventricular structure, or function. Only diastolic blood pressure was higher, and rate-pressure product trended toward being higher, in the type 1 diabetic subjects (Table 1 ). As shown in Table 2, the three type 1 diabetic groups were also well matched in demographics, metabolic characteristics, hemodynamics, and left ventricular structure and function.

Myocardial blood flow between the euglycemic type 1 diabetic group and the nondiabetic control subjects did not differ. However, MVo₂ was higher in the type 1 diabetic group, and their MVo₂ per rate-pressure product was trending toward being higher than that in nondiabetic control subjects (Table 3). There were no differences in myocardial blood flow among the euglycemia (1.16 ± 0.28 ml · g⁻¹ · min⁻¹), Intralipid (1.10 ± 0.25 ml · g⁻¹ · min⁻¹), and HIEG clamp (1.06 ± 0.23 ml · g⁻¹ · min⁻¹) groups (P = 0.83). There were no differences in MVo₂ among the same groups (5.34 ± 1.27 [Intralipid] and 6.09 ± 1.29 μmol · g⁻¹ · min⁻¹ [HIEG clamp], P = 0.47).

The type 1 diabetic euglycemic group had higher plasma glucose levels and higher insulin levels compared with those of control subjects. However, plasma FFA levels did not differ between the two groups. Myocardial glucose extraction fraction and utilization were not different between the two groups, but myocardial glucose utilization/plasma insulin level was lower in type 1 diabetic compared with control subjects (Table 3). There were no differences in plasma glucose concentrations at the time of the PET 1-¹¹C-glucose imaging among the three groups (5.75 ± 0.73, 5.34 ± 1.27, and 6.09 ± 1.29 mmol · l⁻¹ · l⁻¹ for euglycemia, Intralipid, and HIEG clamp groups, respectively; P = 0.67). There were marked differences in plasma insulin levels (Fig. 1A), with the HIEG clamp group having higher levels than the other two groups. Myocardial glucose extraction fraction and utilization paralleled the changes in insulin, with the HIEG clamp group having the highest myocardial glucose extraction fraction and utilization levels (Fig. 1B and C). Furthermore, the higher the plasma insulin level, the higher the myocardial glucose extraction fraction (r = 0.53, P < 0.005) and glucose utilization (r = 0.48, P < 0.005). Conversely, the higher the plasma FFA levels, the lower the myocardial glucose extraction fraction and utilization (Fig. 2). Interestingly, in the multivariate analyses only plasma FFA level was an independent predictor of myocardial glucose extraction fraction and utilization (P < 0.05 and P < 0.01; correlation between plasma FFAs and insulin levels was r = −0.65, P < 0.0001).

Table 3 shows that neither plasma FFA nor any of the measures of myocardial fatty acid uptake or metabolism was different between the euglycemic type 1 diabetic subjects and fasting nondiabetic control subjects. Plasma FFA levels were markedly different among the three type 1 diabetic subject groups, with the HIEG clamp group having the lowest levels (Fig. 3A). Plasma FFA levels were higher in the Intralipid group than in the euglycemia group. Of note, myocardial fatty acid extraction fraction was fairly constant among the three groups (Fig. 3B). In contrast, myocardial fatty acid utilization was markedly lower in the HIEG clamp group compared with that in either the euglycemia or Intralipid group (Fig. 3C), paralleling the differences in plasma FFA levels. Myocardial fatty acid utilization was strongly and negatively correlated with plasma insulin level (r = −0.60, P < 0.0005), and it was the only independent predictor of myocardial fatty acid utilization (P < 0.0005).

The change in fatty acid oxidation also mirrored the differences in plasma FFA levels and myocardial fatty acid utilization, with a difference among the three groups and significant differences between the HIEG clamp group and the other two (Fig. 4A). Myocardial fatty acid oxidation, like fatty acid utilization, inversely correlated with insulin level (r = −0.57, P < 0.001), and it was the only independent predictor of fatty acid oxidation in a multivariate model (P < 0.001).

To evaluate the differences in fractional myocardial fatty acid oxidation among the groups apart from the influence of plasma FFA levels, we also evaluated the differences in percentage of oxidation among the three groups. The percentage of the total myocardial fatty acid utilization accounted for by oxidation was highest in the euglycemic group and lowest in the HIEG clamp group (Fig. 4B).

There were also significant differences in myocardial fatty acid esterification among the three groups. In this comparison, the Intralipid group had the highest level compared with the other two groups (Fig. 5A). As with myocardial fatty acid utilization, esterification was inversely related to plasma insulin levels (Fig. 6); myocardial fatty acid esterification also correlated significantly but positively with age (r = 0.39, P < 0.05) and duration of diabetes (r = 0.38, P < 0.05). Multivariate analysis demonstrated that both plasma insulin and age were independent predictors of myocardial fatty acid esterification (P < 0.05 and P = 0.01, respectively).

Since the percentage of myocardial fatty acid utilizationaccounted for by myocardial fatty acid esterification is by definition the percentage of fatty acid not oxidized, the HIEG clamp group had the highest percentage of myocardial fatty acid utilization going to esterification and the euglycemic group had the lowest (Fig. 5B). Since in the calculation of the percentage of oxidation or esterification the concentration of plasma FFAs drops out, this suggests that the differences in percentage of myocardial fatty acid oxidation and esterification cannot be explained solely by differences in plasma FFA levels. Thus, although overall myocardial esterification correlated inversely with plasma insulin levels, the percentage of myocardial utilization accounted for by esterification correlated directly with plasma insulin level (r = 0.39, P < 0.05), and it was the only independent predictor of the percentage of myocardial esterification (P < 0.05).

Results of prior studies have shown that the myocardium in humans with type 1 diabetes relies more heavily on myocardial fatty acid (as opposed to glucose) metabolism than that in nondiabetic subjects (7,24,25). The results of our study further extend this concept. First, our results show that when type 1 diabetic subjects are euglycemic and fasting and have similar plasma FFA levels, they have myocardial fatty acid metabolism similar to that in nondiabetic fasting control subjects. This occurs despite type 1 diabetic subjects exhibiting higher plasma insulin levels. In our previous study of type 1 diabetes, we sought to match insulin and glucose (rather than FFA) levels between nondiabetic and type 1 diabetic subjects, so the control subjects were fed. In that study, myocardial fatty acid utilization increased in the type 1 diabetic subjects primarily as a result of the increase in plasma FFA levels (7). Taken together, these results further highlight the importance of increased plasma FFA delivery in determining the myocardial metabolic pattern in type 1 diabetes. Furthermore, the presence of similar rates of myocardial glucose use and fatty acid oxidation in type 1 diabetic compared with nondiabetic subjects despite higher plasma insulin levels suggests that myocardial insulin resistance is present in type 1 diabetes under these conditions. Second, our results show that although the myocardium in type 1 diabetes is overly dependent on fatty acid metabolism, both myocardial glucose and fatty acid uptake can be manipulated by altering plasma hormonal and substrate milieu. Last, we have shown that alterations in plasma FFA and insulin levels can change the fate of FFA extracted by the heart in humans with type 1 diabetes.

Our findings that myocardial fatty acid utilization may be manipulated by increasing/decreasing FFA delivery extend the findings of R.J. Bing and others, who showed that substrate delivery to the heart is an important determinant of myocardial substrate utilization in humans without type 1 diabetes, to those with type 1 diabetes (26,27). Thus, although the myocardium in type 1 diabetes relies on fatty acid utilization for generation of ATP for contractile function, it can still be manipulated.

Our data also show that increasing plasma free FFA levels increase the rates of myocardial fatty acid oxidation above the high baseline levels seen in euglycemic type 1 diabetic subjects. Thus, the euglycemic level of fatty acid oxidation in type 1 diabetes is not at its maximal oxidative capacity and may be further increased. This may have detrimental consequences. Results of studies of animal models show that high fatty acid oxidation occurs early in diabetes before marked changes are seen in cardiac function (28). Increased production of reactive oxygen species with increased oxidation may impair efficient calcium handling (10,29,30). Our finding that increased plasma FFA (and hence increased myocardial fatty acid oxidation) was associated with a decrease in myocardial glucose utilization would also be detrimental during ischemia (when the myocardium prefers glucose use). These observations also extend those of Nuutila et al. (31), who demonstrated Randle cycle operation in nondiabetic human myocardium, to type 1 diabetes myocardium.

Despite the increase in myocardial fatty acid oxidation with the increase in plasma FFAs, it appeared that the oxidative capacity of the myocardium can be overwhelmed, as evidenced by a trend toward a decrease in the percentage oxidized and an increase in the percentage esterified in the Intralipid group. Although increase in esterification rate does not necessarily translate to increase in chronic lipid deposition, it agrees with findings of studies in animal models of diabetes and human autopsies, which demonstrated increased cardiac triglyceride content in diabetic compared with control subjects (13,32–34). This increase in esterification may be due to fatty acid uptake in excess of oxidation and/or an increase in the amount or activity of triglyceride synthesis enzymes in response to increased fatty acid availability (35). This fatty acid esterification increase in humans with type 1 diabetes may be detrimental, based on the results of studies in animals and humans suggesting that excessive myocardial fatty acid storage may result in apoptosis, oxidative stress, abnormal energy metabolism, fibrosis, and contractile dysfunction (12,36–40). Future imaging and/or pathological sample studies of lipid accumulation and function are necessary to confirm this hypothesis in humans.

Increasing plasma insulin in type 1 diabetes has a more complicated effect on myocardial glucose and fatty acid utilization and on the metabolic fate of fatty acids within the myocardium. Our results are consistent with those of Monti et al. (41), who found that increased plasma insulin resulted in increased myocardial glucose extraction fraction and utilization in patients with type 1 diabetes. We also demonstrated for the first time in humans with type 1 diabetes that increasing plasma insulin levels decreases rates of myocardial fatty acid utilization. This decrease is likely due to insulin's inhibition of peripheral lipolysis and the resultant decrease in plasma FFA, since myocardial fatty acid extraction fraction was not different in the HIEG clamp group compared with the euglycemic group. In addition, we showed in humans with type 1 diabetes that increase in plasma insulin and myocardial glucose utilization impaired myocardial fatty acid oxidation proportionately more than other fatty acid processes, such as myocardial fatty acid extraction and esterification (Figs. 3B, 4B, and 5B). This decrease in myocardial fatty acid oxidation with an increase in glucose metabolism, previously demonstrated in animal models, has not heretofore been demonstrated in humans with type 1 diabetes (42,43). The net result of the HIEG clamp is that plasma FFA decreases to such a degree that myocardial fatty acid esterification rates decrease, but insulin's anabolic effect on FFAs that do enter the myocardium encourages a high proportion to enter esterification processes in lieu of oxidation.

Because our data demonstrate that it is possible to manipulate myocardial glucose and fatty acid metabolism by manipulating plasma substrate and hormone levels in humans with type 1 diabetes and because animal data demonstrate deleterious effects of excessive fatty acid oxidation and/or storage on the myocardium, it may be desirable to decrease excessive delivery of FFAs to the myocardium in humans with type 1 diabetes, particularly during ischemia, when the myocardium needs to switch fuel sources and utilize predominantly glucose. Thus, our data, although not obtained in patients with ischemia, indirectly support the utility of glucose-insulin-potassium therapy or other treatments for decreasing fatty acid and increasing glucose utilization in patients with type 1 diabetes and ischemia. Exogenous insulin therapy, clearly necessary in subjects with type 1 diabetes for myocardial glucose metabolism, may also have a beneficial effect on decreasing plasma FFA levels, thereby ameliorating the tendency for excessive fatty acid utilization. The myocardium in type 1 diabetes may be especially prone to excessive dependence on fatty acid metabolism and its potential deleterious effects (particularly during ischemia) because it appears to be somewhat insulin resistant (although it still responds to high doses of insulin during the HIEG clamp) (12,44). It appears that insulin resistance based on our data showing that the myocardial glucose utilization/plasma insulin was lower in subjects with euglycemic type 1 diabetes than in normal control subjects. Although requiring further study, these findings may be very applicable to patients with type 2 diabetes, in whom insulin resistance predominates.

Our results may not be extrapolated to subjects who do not fit our inclusion/exclusion criteria. We did not study nondiabetic control subjects under all of the same conditions as those for the type 1 diabetic subjects, although how the nondiabetic myocardium's substrate choice is modified by substrates, insulin, and condition is known (45,46). Myocardial metabolism of other substrates (e.g., lactate) were not measured; however, since subjects were not ketotic, ischemic, or exercising at the time of the study, these should be minor contributors to metabolism. Myocardial glucose oxidation and myocardial metabolism or deposition of endogenous substrates were not quantified. Based on previous studies, endogenous triglycerides would be expected to contribute less to fatty acid oxidation when plasma FFA levels are high, and glycogen should contribute more during adrenergic stimulation (not an intervention in our study) (47,48). Also, Intralipid infusion and an HEIG clamp are not physiological conditions; however, high levels of FFAs (as evoked by Intralipid in our study) may be seen in obese subjects or those with poorly controlled type 1 diabetes, and very low levels (e.g., ∼100 nmol/ml) may be seen after a high-carbohydrate meal (plus insulin) in type 1 diabetic subjects. Thus, our interventions mimicked these physiologic conditions without altering the fasting status or glucose and ketone levels of our subjects. Moreover, although a HIEG clamp is not a physiologic state, it is being clinically tested as a therapy for patients with ischemia, including those with type 1 diabetes, and therefore is a relevant intervention.

Humans with type 1 diabetes have higher MVo₂ than nondiabetic subjects, which is mostly but perhaps not completely accounted for by increased cardiac work. In humans with type 1 diabetes, myocardial utilization and the metabolic fate of substrates can be manipulated by altering plasma FFA and insulin levels, although some degree of myocardial insulin resistance is present. Alterations in myocardial glucose and fatty acid metabolism affected by increasing plasma FFA levels conform to Randle's hypothesis, with increasing fatty acid utilization and oxidation decreasing myocardial glucose extraction fraction and utilization. Furthermore, short-term increases in plasma FFA can increase myocardial fatty acid oxidation rates but also overwhelm the already overtaxed myocardial oxidative capacity and lead to increased myocardial fatty acid esterification rates. Insulin therapy increases myocardial glucose extraction and utilization; decreases myocardial fatty acid utilization, oxidation, and esterification; and yet increases the percentage of the myocardial fatty acid extraction fraction directed to esterification. These findings fine-tune our notions of myocardial metabolism in humans with type 1 diabetes and support the theory that metabolic manipulation of the myocardium is feasible and may have benefit in humans with type 1 diabetes.

This work was supported by National Institutes of Health grants (PO1-HL13581, MO1-RR00036, RO1-HL073120, and P60-DK020579) and a grant from the Barnes-Jewish Hospital Foundation.

Part of this work was presented in abstract form in J Nucl Cardiol 10:S3–S14, 2003.

We extend special thanks to the participants in this study and to the staff of the General Clinical Research Center and our laboratory for their help with data collection and technical assistance. We thank Jean E. Schaffer, MD, for critical reading of the manuscript and Ava Ysaguirre for secretarial assistance.