Proinflammatory Cytokines, Markers of Cardiovascular Risks, Oxidative Stress, and Lipid Peroxidation in Patients With Hyperglycemic Crises

A total of 58 patients admitted to the Regional Medical Center (Memphis, TN) for DKA and severe hyperglycemia were treated on a low-dose insulin protocol using intravenous infusion of insulin with the established rate of 0.1 unit · kg body wt⁻¹ · h⁻¹ (29). Patients with DKA had an admission blood glucose >250 mg/dl (13.9 mmol/l), pH <7.3, bicarbonate <18 mmol/l, anion gap >15 mmol/l, and positive ketonemia and ketonuria. Patients with NKH were admitted with blood glucose >400 mg/dl (22.4 mmol/l), pH >7.3, and serum bicarbonate >18 mmol/l. There was no apparent infection or other known precipitating illness for DKA and/or hyperglycemia in any of the study subjects. We excluded patients with gastrointestinal bleeding, fever, obvious endocrine disorders, history of myocardial infarction, pregnancy, congestive heart failure, history of cardiovascular disease, chronic obstructive pulmonary disease, or chronic renal failure. The criteria for resolution was defined as blood glucose <250 mg/dl (13.9 mmol/l), HCO₃ >18 mmol/l, pH >7.3, normal anion gap, and normal mental status. Blood was drawn on admission for diagnosis and clinical management of DKA and NKH and at 4-h intervals until total resolution of hyperglycemic crisis, which was at 20–24 h after initiation of insulin therapy. The patients were on intravenous fluids and received no food by mouth for the entire study. The total amount of insulin per kilogram of body weight was similar in all hyperglycemic patients (∼90 units until resolution of hyperglycemia/ketoacidosis). Laboratory tests included a complete metabolic profile, cell blood count, arterial pH, blood glucose, blood cultures, and other routine chemistries, which were performed in the hospital laboratory. The other specified assays in this report were performed in the endocrinology laboratory at the University of Tennessee Health Science Center (UTHSC). Blood specimens for these assays were drawn in citrated tubes on admission and at resolution of hyperglycemic crisis at 20–24 h. The specimens were centrifuged at 4°C, and plasma was separated and stored at −70°C until assayed. The consent forms for the protocol, which was approved by the institutional review board at UTHSC, were obtained from the patient or next of kin.

Levels of proinflammatory cytokines (TNF-α and IL-1B, -6, and -8), markers of cardiovascular risks (high-sensitivity CRP and homocysteine), and metabolic hormones (growth hormone [GH], cortisol, and C-peptide) were measured in the plasma using a solid-phase, two-site sequential chemiluminescent immunometric assay on an Immulite analyzer (Diagnostic Products, Los Angeles). The coefficients of variation of the assays were all <5%. The instrument calibrations for the assays were performed as recommended by the manufacturers and were within the specifications.

Assays for markers of oxidative stress and lipid peroxidation were determined by TBA assay (30), and ROS were determined by DCF assay (31). FFA (32) and β-hydroxybutyrate levels (33) were determined by the methods established in this laboratory, as previously described. PAI-1 was assayed using the Zymutest PAI-1 activity enzyme-linked immunosorbent assay (Hyphen BioMed, Andresy, France), which measures only active PAI-1 (34–37). The normal range for fasting normal subjects for this assay was established to be <5 ng/ml. Absorbances of the enzyme-linked immunosorbent assay were determined on an HTS 7000 Plus microplate reader (Perkin-Elmer, Norwalk, CT) and the HTS data analysis software.

Normal fasting values of lean control subjects established in this laboratory for the cytokine assays are as follows: TNF-α <5.0 pg/ml, IL-1β <3.0 pg/ml, IL-6 <5.0 pg/ml, and IL-8 <10 pg/ml. Normal fasting values for the other assays measured in this laboratory are listed in Tables 1, 2, and 3.. Two levels of assay controls were determined with each assay for each analyte, and all control values were within the established ranges.

The mean ± SE were calculated for all continuous variables. Baseline demographics and clinical characteristics between groups were compared using ANOVA and Scheffe’s method for continuous variables, with log transformations when necessary. χ² analyses were performed for comparison of categorical variables. A two-tailed P value of <0.05 was considered statistically significant. StatView version 5.0.1 (SAS Institute, Cary, NC) was the statistical software used for the analysis.

Table 1 shows the clinical characteristics of study subjects. The groups consisted of 28 obese DKA subjects, 20 lean DKA subjects, and 10 obese subjects with NKH. In addition, 12 obese and 12 lean nondiabetic subjects, matched for age, BMI, and ethnicity, served as control subjects. All subjects were African American. None of the subjects had elevated temperature or a white blood cell count >16 × 10⁶ cell/ml or a recognized precipitating cause of DKA. The mean HbA_1c level on admission was 12.1%.

Table 2 shows the admission laboratory values before insulin therapy and at resolution of DKA or NKH, as well as baseline values in lean and obese control subjects. As previously shown (38), patients with DKA had lower levels of C-peptide than those with NKH. With insulin therapy, levels of counterregulatory hormones were significantly decreased at resolution of DKA and/or hyperglycemia.

Figure 1A–D shows levels of inflammatory cytokines (TNF-α and IL-8, -6, and -1β) on admission and at resolution of the hyperglycemic state. The values for lean and obese control subjects are also included. All values (in pg/ml) are means ± SE. Figure 1A shows the relationship of IL-8 at admission and resolution for the three groups of patients and the control subjects. The admission and resolution levels, respectively, for the groups are lean DKA: 29.3 ± 3.4 and 10.6 ± 2.3 pg/ml; obese DKA: 27.4 ± 3.8 and 11.9 ± 2.8 pg/ml; obese hyperglycemic: 25.8 ± 3.4 and 9.3 ± 2.8 pg/ml; lean control subjects: 4.9 ± 1.4 pg/ml; and obese control subjects: 5.5 ± 1.7 pg/ml. Figure 1B shows the relationship between the admission and resolution levels, respectively, of IL-6 of the groups: lean DKA: 14.9 ± 2.6 and 3.9 ± 1.1 pg/ml; obese DKA: 12.6 ± 2.1 and 4.3 ± 0.6 pg/ml; obese hyperglycemic: 10.2 ± 1.7 and 3.3 ± 0.7 pg/ml; lean control subjects: 1.8 ± 0.2 pg/ml; and obese control subjects: 2.1 ± 0.3 pg/ml. Similarly, Fig. 1C shows this relationship for TNF-α in these groups: lean DKA: 22.7 ± 3.6 and 4.6 ± 0.9 pg/ml; obese DKA: 28.3 ± 2.8 and 5.9 ± 0.7 pg/ml; obese hyperglycemic: 24.3 ± 3.1 and 5.1 ± 1.3 pg/ml; lean control subjects: 1.7 ± 0.2 pg/ml; and obese control subjects: 3.9 ± 0.6 pg/ml. This relationship of IL-1β levels can also be seen for each group at admission and resolution in Fig. 1D: lean DKA: 9.8 ± 2.3 and 2.1 ± 0.2 pg/ml; obese DKA: 13.7 ± 2.1 and 2.4 ± 0.3 pg/ml; obese hyperglycemic: 11.4 ± 0.8 and 3.1 ± 0.4 pg/ml; lean control subjects: 1.3 ± 0.2 pg/ml; and obese control subjects: 1.9 ± 0.3 pg/ml. In all patients, levels of cytokines on admission were significantly higher than at resolution or with matched control subjects. TNF-α levels were significantly higher in the obese DKA than the lean DKA subjects on admission and were correlated significantly with their BMI (r = 0.81, P < 0.05). The other cytokines did not demonstrate significant correlation with BMI (data not shown). The cytokine levels at resolution of hyperglycemia with insulin therapy were reduced to levels not significantly different from the lean or obese control subjects. Admission levels of TNF-α and IL-8 were greater than those of IL-1β and -6.

Figure 2 shows serum levels of CRP, homocysteine, PAI-1, TBA, DCF, and FFAs on admission and at resolution of DKA and NKH. As can be seen, admission levels were significantly higher in all groups than after insulin therapy and resolution of hyperglycemia. The actual values for these parameters are presented in Table 3. After resolution, levels of DCF, TBA, PAI-1, and FFAs were reduced to near control values, whereas reductions in circulating levels of CRP and homocysteine were not as robust and remained above control values. It is of interest that homocysteine was not decreased significantly with resolution of DKA/hyperglycemia and insulin treatment.

Hyperglycemia in subjects with or without a history of diabetes is a common finding in hospitalized patients admitted to critical care settings (39,40) and is associated with greater morbidity and mortality compared with those diagnosed with diabetes (39). The reason for greater mortality may be related to the delay in diagnosis of diabetes and/or its proper therapy (39). On the other hand, numerous prospective studies in diabetic and nondiabetic subjects admitted to critical care units have clearly shown a salutary effect of insulin, improving clinical outcome through correction of hyperglycemia and inflammation (41–45). Some of these studies have shown direct correlation between high blood glucose levels and increased mortality (45). A more recent study in such patients has demonstrated the salutary effect of insulin to be attributable to its anti-inflammatory properties (27).

Although previous studies have suggested an elevation of IL-6 and TNF-α in uncontrolled diabetes (46), elevation of IL-1B and -8 along with an increase in counterregulatory hormones and cardiovascular markers, to our knowledge, have not been recorded before. Of interest is our finding that these significant elevations of proinflammatory cytokines as well as cortisol, GH, TBA, PAI-1, DCF, and FFAs are all reduced to normal levels promptly in response to insulin therapy and normalization of blood glucose. The times for resolution of DKA and NKH were remarkably close, with both reaching total resolution within ≤24 h. Increased levels of these markers occurred in both ketotic and nonketotic hyperglycemic patients with similar blood glucose (i.e., >600 mg/dl.).

We must therefore conclude that hyperglycemia and ketoacidosis independently induce changes in proinflammatory cytokines, oxidative stress, and cardiovascular markers without synergistic effects of one or the other. It is of note that lean DKA patients exhibited as much of an increase in cardiovascular risk markers, oxidative stress, counterregulatory hormones, and cytokines as obese ketoacidotic patients. The only exception to this statement was the level of TNF-α, which was significantly greater in obese DKA than either lean DKA or NKH subjects. Although TNF-α values exhibited high correlation with BMI (r = 0.81, P < 0.05) in all three groups, other cytokines did not demonstrate such a correlation.

To our knowledge, this is the first report demonstrating increased levels of GH, cortisol, cytokines, cardiovascular risk markers, and oxidative stress in obese and lean patients with DKA (in the absence of any history or evidence of cardiovascular events, trauma, or severe infection) and their prompt suppression in response to intravenous insulin. Although previous studies support the association of DKA with oxidative stress and have studied possible mechanisms for the generation of ROS (47–49), it is not clear whether there is a causal relationship between markers of oxidative stress and acute diabetes complications, since the majority of these markers have been demonstrated in plasma.

Of interest is our finding in regard to homocysteine concentration and the fact that elevated levels of this risk factor did not respond to insulin as robustly as FFA, PAI-1, TBA, and DCF, all of which returned to control levels with resolution of DKA and hyperglycemia by insulin treatment. Homocysteine is regulated by many factors, and its levels are affected by various drugs (50), but its lack of response may be similar to the observations of Fonseca et al. (51), in that homocysteine levels in diabetic patients, unlike nondiabetic subjects, do not respond to insulin because these patients are insulin resistant. Our study confirms this phenomenon and extends these findings to patients in hyperglycemic crises.

In a more systematic study in type 1 diabetes, a variety of markers of oxidative stress were measured, including TBA, organoperoxide, vitamin E, vitamin C, glutathione, and glutathione peroxidase. The results suggest that most markers are not associated with long-term complications of diabetes (52,53) and that intracellular determination of these markers may be a better method of assessment of oxidative stress.

Our study confirms the well-known phenomenon of leukocytosis in hyperglycemic crises, without obvious infection, on febrile events. Although the mechanism of this finding is not known, the proinflammatory state demonstrated here and elsewhere certainly could bring about such an event, as well as stimulation of sympathetic nervous system, secondary to stress of hyperglycemia. It is well known that in normal subjects, increased sympathetic activity results in leukocytosis and elevation of TNF-α and IL-6 (54). The latter cytokines, along with IL-1β, may regulate production of acute-phase protein by raising body temperature (55), and the organism may be undergoing a compensatory mechanism in the immune and hypothalamic-pituitary-adrenal axis (56). It is of interest that despite leukocytosis and elevation of cytokines and other signs of oxidative stress, the body temperature in DKA remains hypothermic in the absence of florid infection (23). After insulin treatment, however, the temperature returns to normal, concomitant with a reduction of leukocytosis. The mechanisms of this interesting phenomenon are not fully understood, but coupled with the present demonstration of a dramatic reduction of cytokines and mediators of oxidative stress with insulin, they deserve further investigation.

Our recent studies in patients with DKA demonstrated in vivo activation of T-cells, which led us to hypothesize that hyperglycemia and/or ketosis through production of ROS and generation of proinflammatory cytokines may result in de novo emergence of growth factor receptor insulin, IGF-1, and IL-B (24). In the present study, we report elevated levels of proinflammatory cytokines and ROS as markers of oxidative stress in conjunction with elevated levels of FFAs and glucose in our patients with DKA as well as NKH (Table 3).

Based on our present and previous studies as well as other work from other laboratories (10,17,57), we may expand our hypothesis by proposing that the initial activating event for T-cells may be the presence of high levels of both FFAs and glucose in patients with hyperglycemic crises, which may result in the generation of ROS through diacylglycerol/PKC-activated NAD(P)H, supporting earlier findings on the role of FFAs on human muscle cells (10). However, further studies are needed to establish the presence of several hypothetical intermediates, including diacylglycerol, protein kinase C, and activated NAD(P)H in activated T-cells in hyperglycemic patients. To our knowledge, however, the presence of these biochemical steps and intermediary metabolites have not been demonstrated in T-cells of DKA or NKH patients.

Multiple studies have now demonstrated not only the anti-inflammatory effect of insulin in vitro and in vivo, but also its profibrinolytic activity (25); its suppressive effects on endothelial growth factor, metaloproteinase-9 (26), plasma tissue factor, and PAI-1 (25); and its effect on activation of endothelial cells (58) and polymorphonuclear leukocytes (59). To our knowledge, however, studies on the effect of insulin on subpopulations of polymorphonuclear leukocytes such as CD-4 and CD-8 T-cells, where their in situ activation was demonstrated in DKA (23), has not been reported.

In conclusion, our study clearly demonstrates a hitherto undescribed phenomenon of anti-inflammatory effect of insulin in hyperglycemic crises concomitant with the reduction of multiple cytokines, markers of cardiovascular risk and oxidative stress, and counterregulatory hormones. Our findings thus extend the previous observation on the robust and prompt anti-inflammatory effect of insulin and other conditions not associated with DKA and NKH (25–28,58,59).

The present study, however, does not permit us to draw any definitive conclusion regarding cause and effect of these events because time-related assessment of these factors and isolation of the intermediate products were not planned in this protocol.

This work was partially supported by a grant from the National Institutes of Health, Division of Research Resources (GCRC RR00211), and a grant from the American Diabetes Association (to G.E.U.).

We thank John Crisler for assay of chemicals and hormones and our patients who participated in this study.