Most often, diabetic ketoacidosis (DKA) in adults results from insufficient insulin administration and acute infection. DKA is assumed to release proinflammatory cytokines and stress hormones that stimulate lipolysis and ketogenesis. We tested whether this perception of DKA can be reproduced in an experimental human model by using combined insulin deficiency and acute inflammation and tested which intracellular mediators of lipolysis are affected in adipose tissue. Nine subjects with type 1 diabetes were studied twice: 1) insulin-controlled euglycemia and 2) insulin deprivation and endotoxin administration (KET). During KET, serum tumor necrosis factor-α, cortisol, glucagon, and growth hormone levels increased, and free fatty acids and 3-hydroxybutyrate concentrations and the rate of lipolysis rose markedly. Serum bicarbonate and pH decreased. Adipose tissue mRNA contents of comparative gene identification-58 (CGI-58) increased and G0/G1 switch 2 gene (G0S2) mRNA decreased robustly. Neither protein levels of adipose triglyceride lipase (ATGL) nor phosphorylations of hormone-sensitive lipase were altered. The clinical picture of incipient DKA in adults can be reproduced by combined insulin deficiency and endotoxin-induced acute inflammation. The precipitating steps involve the release of proinflammatory cytokines and stress hormones, increased lipolysis, and decreased G0S2 and increased CGI-58 mRNA contents in adipose tissue, compatible with latent ATGL stimulation.
Diabetic ketoacidosis (DKA) remains one of the most common, serious, and demanding medical emergencies within the field of diabetology. In Western societies, the annual frequency of DKA is ∼5% in patients with type 1 diabetes (1,2). A Scottish survey estimated that death from DKA is associated with the single largest percentage of the loss of life expectancy occurring before the age of 50 years, contributing to 25% of the excess mortality in this group of patients (3). The magnitude of these figures and the associated clinical challenges likely are much higher in developing areas with less advanced health care (4).
The most common precipitating factor of DKA in adults is lack of insulin and infection, leading to a vicious cycle with release of proinflammatory cytokines and counterregulatory hormones, increased lipolysis and ketogenesis, hyperosmolarity, and dehydration (5). This concept to some extent is based on observational studies showing increased levels of cytokines, stress hormones, and free fatty acids (FFAs) in patients with full-blown DKA; these changes possibly reflect rather than cause the metabolic disarrangement (6). Experimental studies primarily have used insulin withdrawal to assess the precipitating events in type 1 diabetic DKA. These studies have shown increased lipolysis/FFA availability and increased levels of glucagon to be important triggers of ketogenesis (7–9), whereas proinflammatory cytokines in general have not been measured, and levels of cortisol, epinephrine, and growth hormone (GH), when measured, have not been increased. Studies in subjects without diabetes have showed that administration of tumor necrosis factor-α (TNF-α) or endotoxin (lipopolysaccharide [LPS]), as a model for acute inflammation, increases circulating levels of stress hormones and stimulates lipolysis (10–13).
Uncontrolled lipolysis plays a pivotal role in the pathogenesis of DKA by increasing the amount of FFAs reaching the liver and acting as precursors for ketogenesis. Lipolysis is controlled by adipose triglyceride lipase (ATGL), hormone-sensitive lipase (HSL), and monoacylglycerol lipase, which hydrolyze triglycerides to FFAs and glycerol (14). ATGL activity is stimulated by comparative gene identification-58 (CGI-58) and inhibited by counterregulatory G0/G1 switch gene 2 (G0S2) (14).
The current study tested whether an experimental clinical model of DKA in subjects with type 1 diabetes is possible by combining exposure to acute inflammation (endotoxin/LPS) and insulin deficiency and whether this model would reproduce the whole spectrum of metabolic events observed in the clinical setting. In addition, we aimed to define precipitating intracellular adipocyte signaling events participating in this scenario, particularly modulation of ATGL.
Research Design and Methods
Study Design and Protocol
We used a randomized controlled crossover design comprising two arms separated by at least 3 weeks: 1) euglycemic control condition (CTR) and 2) hyperglycemic ketotic condition (KET) induced by endotoxin administration plus insulin deprivation. Long-acting insulin was paused 24 h before each study day and replaced by rapid-acting insulin. Patients were hospitalized at 10:00 p.m., and subcutaneous insulin administration was changed to continuous intravenous administration (Actrapid; Novo Nordisk, Copenhagen, Denmark). An additional intravenous catheter was placed in a dorsal hand vein for blood glucose and ketone measurements overnight. During CTR, the subjects were given a variable amount of insulin, and blood glucose was kept at 5–7 mmol/L. During KET, the dose of insulin was reduced to 15% of the basal insulin requirements from 10:00 p.m. onward. A physician measured blood glucose every hour and ketone bodies (FreeStyle Precision; Abbott Diabetes Care) every second hour during the night. Studies were commenced at 0700 h (t = 0 min), and subjects were examined for 5 h. The dorsal hand vein was heated to arterialize the blood. Nine young male volunteers with type 1 diabetes were included after informed consent was obtained. The following inclusion criteria were used: type 1 diabetes (C-peptide negative), male sex, 20–40 years of age, no medication other than insulin, BMI 19–26 kg/m2, and no chronic diseases or diabetes complications. A medical examination and routine blood check were conducted; 13 subjects were screened and 9 were found eligible for inclusion.
The project was approved by the Danish Mid-Regional Ethics Committee (1-10-72-98-14) in accordance with the Declaration of Helsinki.
Escherichia coli endotoxin (LPS) (10,000 USP Endotoxin, lot HOK354; U.S. Pharmacopeial Convention, Rockville, MD) was diluted in isotonic saline. A bolus of 1 ng/kg (10 units/kg) body weight was given intravenously at t = 0 min (0700 h).
Outcomes and Analytical Methods
Indirect calorimetry (Oxycon Pro; Intramedic, Gentofte, Denmark) was performed at t = 150 min to measure substrate oxidation rates as described by Ferrannini (15). Albumin-bound [9,10-3H]-palmitate (GE Healthcare, Brøndby, Denmark) was infused (0.3 μCi/min) at t = 200–260 min and analyzed in triplicate as previously described (9).
Abdominal subcutaneous adipose tissue biopsy specimens were obtained at t = 0 and t = 270 min and were immediately cleansed and snap-frozen in liquid nitrogen. Western blot analysis was performed using standard protocols and commercially available materials and antibodies (Supplementary Data).
Serum concentrations of FFA, cortisol, glucagon, GH, lactate, and C-peptide and plasma concentrations of TNF-α, interleukin (IL)-10, and IL-6 were quantified by routine assays as described previously (16). Serum concentrations of β-hydroxybutyrate (3-OHB) were measured using hydrophilic interaction liquid chromatography–tandem mass spectrometry (17).
mRNA was isolated using TRIzol (Gibco BRL; Life Technologies, Roskilde, Denmark), and quantitative PCR was performed in a LightCycler 480 (Roche). G0S2, ATGL, CGI-58, and cell death–inducing DFFA-like effector C and A (CIDEC/CIDEA) genes were quantified using the housekeeping gene GAPDH. GAPDH was tested and found to be equal in the two groups (Supplementary Data).
Results are presented as mean ± SEM, unless otherwise specified. Normal distribution of data was ensured by inspection of Q–Q plots. Data were logarithmically transformed if the distribution was unequal. If data remained unequally distributed, a rank sum test was performed to test for differences between the two study days. Statistical association of the two study days was tested using a paired t test and two-way repeated-measures ANOVA when relevant. P < 0.05 was considered significant.
The nine male volunteers reported variable degrees of chills, nausea, discomfort, and headache during KET. Symptoms peaked approximately at t = 120 min and persisted throughout the day. No serious adverse events occurred (Table 1).
Inflammation, Clinical Picture, and Stress Hormones
Blood concentrations of TNF-α, IL-6, IL-10, cortisol, glucagon, and GH were elevated during KET compared with CTR (P < 0.05). Respiratory rate, temperature, and heart rate were all significantly higher in KET (P < 0.02). Serum bicarbonate and pH decreased modestly (P < 0.001) (Table 1).
Indirect Calorimetry and Palmitate Tracer
Energy expenditure increased 30% (560 kcal/day) and lipid oxidation rates increased by 50% (500 kcal/day) during KET (P < 0.001). The palmitate rate of appearance and serum palmitate concentrations were 2.5-fold higher during KET than during CTR (P < 0.001) (Fig. 2).
Protein and mRNA
Adipose tissue content of G0S2 mRNA was decreased fivefold (P = 0.002) and CGI-58 mRNA increased twofold (P = 0.01) during KET compared with CTR. The intracellular mRNA content of ATGL did not differ between the 2 days. CIDEC mRNA content was 40% lower (P = 0.03) during KET than during CTR (Fig. 3).
Western blot analysis of the adipose tissue biopsy specimens did not reveal significant differences in expression of ATGL, G0S2, CGI-58, and CIDEC (data not shown). In addition, phosphorylated HSL Ser552, 554, 650, PKA-substrate/PLIN1, phosphorylated Akt Ser473, Akt, phosphorylated AS160 Thr642, AS160, and CIDEA remained unaltered.
This study was designed to create an experimental clinical model of DKA in subjects with type 1 diabetes by combining endotoxin/LPS exposure (infection/inflammation) and insulin deficiency and to assess whether such a model replicates the entire metabolic disarray observed in the clinical setting of DKA. This approach succeeded in replicating both the high levels of cytokines, stress hormones, FFA, and 3-OHB observed clinically and the increased rate of lipolysis observed when experimental insulin deprivation is used as a DKA model, confirming that all these factors are involved in the pathogenesis of DKA. In addition, we found decreased G0S2 and increased CGI-58 adipose tissue contents of mRNA, suggesting incipient ATGL activation, which together with decreased CIDEC mRNA expression, may lead to increased lipolysis (14).
These findings cement the prevailing notion that in general, DKA in adults is caused by insulin deficiency and the presence of acute inflammation, triggering the release of proinflammatory cytokines and stress hormones, all of which modulate intracellular adipose tissue signaling and eventually lead to uncontrolled lipolysis and ketogenesis (5). As outlined previously (6), proinflammatory cytokines likely play a key role either as primary agents in the case of acute inflammation by infection or as secondary agents in the absence of obvious infection. At some stage, this will activate the hypothalamo-pituitary axis and release stress hormones (10). In human studies, TNF-α (11), IL-6 (18), epinephrine (19), cortisol, and GH (20) coherently increase lipolysis, and glucagon augments ketogenesis when the FFA concentration is increased (8). One study suggested that the lipolytic response to TNF-α is markedly dampened in patients with hypopituitarism without cortisol and GH responses, underlining the importance of intact pituitary function and stress hormone release for maximum stimulation of lipolysis to occur (13).
In subjects without diabetes, the dose of endotoxin used in the current study has previously been shown to cause increments in inflammatory cytokines and stress hormones and a less pronounced peak increase in 3-OHB (170 vs. 1,500 μmol/L in the current study) and FFA (650 vs. 1,200 μmol/L) (16). A previous study using interruption of continuous subcutaneus insulin infusion for 6 h in subjects with type 1 diabetes reported slightly lower peak values of 3-OHB of around 900–1,000 μmol/L (21). Although we observed marked increments in 3-OHB and FFA, together with modest decreases in pH and plasma bicarbonate, patients admitted for full-blown DKA generally have at least two- to threefold higher 3-OHB levels and are more acidotic.
We did not observe detectable alterations of the phosphorylation of HSL or the protein expression of ATGL, G0S2, and CGI-58. The lack of protein alterations could be due to a negative feedback mechanism involving 3-OHB—and perhaps also lactate—inhibition of lipases and lipolysis (22,23). Human studies have shown an ∼40% reduction of lipolysis after 3-OHB infusion (24). On the other hand, the current findings of increased CGI-58 mRNA and decreased and unaltered CGI-58 protein and decreased intracellular mRNA expression of G0S2 are compatible with latent ATGL stimulation. In addition, the decreased CIDEC mRNA contents in adipose tissue suggests that decreased CIDEC activity may participate in stimulation of lipolysis (14). Of note, we observed a minor decrease in blood glucose 1–2 h after endotoxin administration; endotoxin has been reported to acutely increase glucose disposal, perhaps related to TNF-α and IL-6 actions (25).
The current study design has limitations. Adipose biopsy specimens were obtained from subcutaneous abdominal depots 5 h after endotoxin exposure, and the results may have been different if the biopsy specimens had been taken at other time points and/or from other locations. We only observed relatively modest increments in 3-OHB levels, implying that the findings only apply to the initial precipitating events triggering DKA.
In conclusion, we show that a combination of insulin deficiency and endotoxin-induced acute inflammation can be used as a model to mimic the clinical picture of incipient DKA and that the precipitating events involve release of proinflammatory cytokines and stress hormones. The increased lipolysis is associated with decreased G0S2 and increased CGI-58 mRNA contents in adipose tissue, suggesting latent ATGL activation.
Acknowledgments. The authors thank Annette Mengel, Karen Mathiassen, Eva Schriver, Lenette Pedersen, and Helle Zibrandtsen from the Department of Clinical Medicine, Aarhus University Hospital, Aarhus, Denmark, for technical assistance and consistent work throughout the study. The authors also thank the nine subjects for participation and completion of the study days.
Funding. This work was funded by the Danish Council for Strategic Research (grant no. 0603-00479B).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. M.S. contributed to the research design and conduct, data analysis, and writing and final approval of the manuscript. U.K. and P.L.P. contributed to the research design and final approval of the manuscript. T.V. and N.R. contributed to the research design and conduct and final approval of the manuscript. S.B.P., M.J., T.S.N., and N.J. contributed to the data analysis and final approval of the manuscript. N.M. contributed to the research design, data analysis, and writing and final approval of the manuscript. N.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Clinical trial reg. no. NCT02157155, clinicaltrials.gov.
This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-1645/-/DC1.
- Received December 2, 2015.
- Accepted February 9, 2016.
- © 2016 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.