Effect of Diabetes on Nitric Oxide Metabolism During Cardiac Surgery

The study was conducted in accordance with the principles expressed in the Declaration of Helsinki, and local ethical committee approval was obtained. Informed written consent was authorized by all participating individuals after the nature of the investigation was explained to them. A total of 170 patients were operated on by one surgeon (M.G.) during a 12-month period. Individuals with valvular disease, ventricular aneurysm, peripheral vascular disease, carotid stenosis, heart failure, and poor left ventricular function were excluded from the study. Only 86 patients, all undergoing elective coronary bypass graft surgery, met the selection criteria, and 40 of these patients (20 with diabetes and 20 without diabetes) consented to take part in the study. The patients were blinded as to the mode of treatment, and data from all of the patients taking part in the study were included in the final analysis. Patients taking part in the study (both with or without diabetes) were assigned numbers, and a computer random number generator was used to create two groups. One group received physiological saline solution, and the other received the NO donor GTN, which was infused at a rate of 1 μg · kg^–1 · min^–1 starting 10 min before the initiation of CPB and then continued for 4 h.

Before surgery, blood glucose levels were well controlled by either insulin (n = 10) or oral antidiabetic therapy (n = 10) in all patients, as shown by the HbA₁ values (range 7.2–9.7% with a mean value of 8.5 ± 0.3% compared with the normal 4–8.5% normal range). Patients with diabetes attended the Diabetes Clinic in the Department of Endocrinology at the University Hospital of Leicester, Leicester General Hospital, for treatment. The disease classification was undertaken in accordance with the revised American Diabetes Association (22) and the World Health Organization criteria (23). During the study, the infusion of insulin to control blood glucose commenced immediately after the termination of CPB in all patients with diabetes and was maintained for 48–72 h. Blood glucose levels were monitored periodically, and they were kept under control by a sliding scale regime for insulin administration using a stock concentration of 1 IU/ml insulin (Actrapid) suspended in normal saline. When blood glucose was 0–4, 4–6, 6–8, 8–10, 10–12, 12–14, and >14 mmol/l, patients were given a corresponding insulin infusion of 0, 2, 4, 6, 8, 10, or 12 ml/h, respectively. Hemoglobin levels were also monitored periodically, and blood transfusion was performed to maintain values >10 mg/dl.

Arterial blood samples were collected at the following time points: before the induction of anesthesia, before the initiation of CPB (time 0), and 0.5, 1, 2, 4, and 8 h thereafter. Samples were collected into sterile EDTA tubes and centrifuged immediately at 1,500g for 12 min at 4°C. The resultant plasma was aliquoted and stored at −80°C until analysis.

Immediately before the induction of anesthesia, a urine catheter (Cobe CML; Cobe Laboratories, Gloucester, U.K.) was inserted into the bladder and connected to a reservoir bag. The initial urine samples were discarded, and samples for analysis were collected at time 0–1, 1–2, 2–4, and 4–8 h after the initiation of CPB and immediately snap frozen and stored at −80°C until analysis.

The administration of anesthesia was in accordance with institutional guidelines. All of the volunteers received morphine (10.0 mg) and prochlorperazine (12.5 mg) administered intramuscularly at least 1 h before surgery. Central venous and radial artery cannulae were inserted under local anesthesia (1% lidocaine) and with midazolam sedation (3.0–4.0 mg i.v.). Anesthesia was induced with an infusion of propofol (8.0 mg · kg^–1 · h^–1 i.v.) and with fentanyl (1.0 mg i.v.) and pancuronium (12.0 mg i.v.). Anesthesia was then maintained by continuous infusion of propofol (4.0 mg · kg^–1 · h^–1 i.v.). Hypotension was controlled by intravenous infusion of fluids or 0.5 mg metaraminol, with increments where appropriate. Heparin (300 IU/kg body wt) was administered in all study patients just before aortic cross-clamping to achieve an activated clotting time of >450 s. A further 5,000 IU heparin were given in the bypass prime. Morphine (5.0 mg i.v.) and diclofenac (100.0 mg i.v.), a nonsteroidal nonnarcotic drug, were given to treat pain when required and if not contraindicated.

After the induction of anesthesia and the systemic application of heparin, CPB was initiated in a standard fashion under conditions of mild body hypothermia (32°C). Briefly, CPB was composed of a roller pump circuit (Stöckert Instrumente, Munich, Germany), a hollow fiber polypropylene oxygenator with an incorporated cardiotomy reservoir (Cobe CML), and plasticized polyvinyl chloride tubing. The pump was primed with 1.4 l Hartmann’s solution and used at a flow rate of 2.4 l · min^–1 · m^–2 body surface area. The coronary bypass grafts were performed using intermittent periods of ischemia without using cardioplegic solutions. Heparin was reversed after bypass by the administration of 3.0 mg/kg protamine.

Plasma and urine samples were assayed for NOx, the stable end product of NO, by using a commercially available kit (Cayman Chemical, Ann Arbor, MI). Briefly, the plasma and urine samples were first filtered through a 0.45-μm Minisart column (Amersham, U.K.). The filtrate was incubated with nitrate reductase and NADH as an enzyme cofactor for 3 h at room temperature to reduce nitrate to nitrite. The total nitrite (NOx) was then analyzed by reacting the samples with Griess reagent and measuring the absorbance of each sample at 540 nm. The amounts of NOx in plasma and urine were calculated from a known concentration range of a standard curve obtained by enzymatic conversion of sodium nitrate to sodium nitrite. To rule out a potential interfering effect of GTN on the measurement of NOx in plasma, pooled plasma obtained from 10 healthy subjects was spiked with 100 μmol/l nitrate. In the presence and absence of 10 μmol/l GTN, the assay detected 116 and 110 μmol/l of NOx in the plasma, respectively, indicating a lack of effect of GTN on the assay. The intra- and interassay coefficient of variation of the NOx assay in plasma was 9.0 and 9.2%, respectively. The detection limit of the NOx assay was 2.5 μmol/l, and regression coefficients for the standard curves ranged between 0.98 and 0.999.

Plasma and urine cGMP were measured by a competitive enzyme immunoassay kit (Cayman Chemical). In plasma, cGMP was at first extracted after precipitation with cold ethanol. After centrifugation at 1,500g for 10 min, the resultant supernatant was collected into clean test tubes. The supernatant was dried under a stream of nitrogen before resuspension in phosphate buffer. Urine samples were diluted 100 times with phosphate buffer and centrifuged at 2,500g at 4°C for 10 min. To enhance the sensitivity of the assay, all of the samples and standards were acetylated before the assay by the addition of 4 mol/l potassium hydroxide followed by acetic anhydride (5:1 ratio).

The assay is based on the competition between free cGMP and a cGMP tracer (cGMP linked to an acetylcholinesterase molecule) for a limited number of cGMP-specific rabbit antiserum binding sites. Thus, the amount of cGMP tracer that is able to bind to the rabbit antiserum is inversely proportional to the concentration of free cGMP in plate wells. A constant tracer concentration was maintained, whereas the concentration of free cGMP (standard and samples) was varied. The resultant rabbit antiserum–cGMP complex (either free or tracer) binds to a mouse monoclonal anti-rabbit antibody that has been previously attached to the plate wells after 18 h incubation at room temperature. After washing to remove any unbound reagents, Ellman’s reagent (containing the substrate to acetylcholinesterase) was added to the wells. The product of this enzymatic reaction has a distinct yellow color and absorbs strongly at 412 nm. The color intensity was proportional to the amount of cGMP tracer bound to the well, which in turn was inversely proportional to the free cGMP present in the well during the incubation. As before, to investigate a potential interfering effect of GTN on the measurement of cGMP in plasma, pooled plasma obtained from 10 healthy subjects was spiked with 100 pmol/ml cGMP. In the extracted supernatants, the assay detected 94 and 92 pmol/ml of cGMP in the presence and absence of 10 μmol/l GTN, respectively, indicating a lack of effect of GTN on the assay. The detection limit of the cGMP assay was 0.09 pmol/ml and the intra- and interassay coefficient of variation was 9 and 9.7%, respectively.

To minimize the influence of changes in renal excretory function, NOx and cGMP values were corrected for creatinine concentration. Creatinine concentration was determined by a colorimetric assay from Sigma Chemicals (Poole, Dorset, U.K.). The assay was based on the reaction between creatinine and alkaline picrate, with a resulting yellow/orange color. Under acid conditions creatinine-picrate color fades faster than interfering chromogens. Therefore, measurement of the difference in color intensity at or near 500 nm before and after acidification is proportional to creatinine concentration. The resultant plasma content and urinary excretion of NOx and cGMP were corrected by creatinine concentration expressed as micromoles per millimole and picomoles per millimoles creatinine, respectively, as described previously (16,17). Creatinine and nitrate clearances were calculated as follows: clearance = urine flow (ml/min) × urinary concentration (μmol/l) ÷ plasma concentration (μmol/l). Pooled plasma with mean creatinine levels of 1.1–6.9 mg/dl assayed by this method on 10 separate occasions produced standard deviations of 0.12–0.25 mg/dl and coefficients of variance of 10.9–3.6%.

Blood glucose was determined by a glucose-oxidase method (Unimate 5 Glucose HK; Roche, Basel, Switzerland) using a Vitros 250 autoanalyzer (Orthoclinical Diagnostics, Buckinghamshire, U.K.). Plasma insulin was determined by an enzyme-linked immunosorbent assay for human insulin (DRG International, Mountainside, NJ). The minimum detectable concentration of insulin estimated by this assay was 0.15 μIU/ml. The intra- and interassay coefficient of variance ranged from 3 to 5.3 and 5.6 to 9.8%, respectively. There was no evidence of cross-reactivity with pro-insulin measurement in this assay.

All of the data are presented as the means ± SE of means in plots of values against time in hours (Figs. 1,2,3,4,5). For the purpose of statistical comparisons, each parameter value plot against time in hours from the start of the study for each subject, and the area under the curve (AUC) was calculated as the primary measure of outcome. There was evidence of positive skewness in some of the raw data-time points, and for this reason logarithm transformation was used to ensure nearly equal variability in the groups being compared. The main analysis is a two-way analysis of variance of the AUC (calculated on the log10 scale) to assess the apparent effect on outcome by GTN, diabetes, and any corresponding interaction. The results were presented as the difference between groups in the mean value of AUC, with the corresponding 95% CIs, and P values ≤5% were considered to be of statistical significance. A secondary analysis was then carried out to validate the statistical significance of these results using the program Proc Mixed in the computer package SAS to fit an autocorrelation for repeated measures. The correlation of any two measurements made on the same subject was assumed to be ρ^d, where d is the intervening period of time and ρ is common to all subjects.

As shown in Table 1, the clinical characteristics and perioperative data were similar in all study groups. Table 2 shows that the medication administered to patients before surgery was also similar in all study groups, with the exception of the antidiabetic treatment. All of the patients with diabetes had the type 2 form of the disease. The results were similar for the patients with diabetes regardless of whether they were on insulin and, because of this, the values are presented together without distinction of the treatment for the control of diabetes.

The results shown in Figs. 1A and B demonstrate that the NOx mean plasma values at baseline (i.e., preanesthesia) were twofold greater in patients with diabetes compared with those without diabetes (P < 0.001). They also show that NOx values were not significantly affected by cardiac surgery (e.g., the first 2 h of sampling) in patients without diabetes, whereas they were significantly (P = 0.02) elevated in patients with diabetes. Interestingly, the infusion of GTN significantly (P = 0.003) reduced plasma NOx only in patients with diabetes.

Figure 2A and B shows that the urinary excretion of NOx was greater in patients with diabetes compared with those without diabetes during the first hour of surgery (P = 0.01), although the overall urinary excretion of NOx over the duration of the study was similar in both study groups (P = 0.83). GTN infusion, however, significantly reduced urinary NOx in both groups (P = 0.005 in the group with diabetes and P = 0.002 in the group without diabetes).

Table 3 shows the results for plasma creatinine clearance, which are presented as ratios of geometric means of AUCs over the entire study period. The results show that plasma creatinine clearance is similar in the two groups (P = 0.5) and that the infusion of GTN had no significant effect on creatinine clearance in both groups of patients (P = 0.9 in the group with diabetes and P = 0.14 in the group without diabetes).

Table 3 also shows the results with regard to plasma NOx clearance. Plasma NOx clearance was found to be influenced by diabetes, with values significantly lower (P = 0.002) in patients with diabetes than in those without diabetes. On the other hand, GTN significantly reduced (P = 0.02) plasma NOx clearance in patients with diabetes but had no significant effect on patients without diabetes (P = 0.3). These results, taken together with the unaffected urinary NOx excretion and plasma creatinine clearance shown above, suggest that the increase in plasma NOx in patients with diabetes represents a real increase in NO production that is not readily eliminated by the kidneys.

Figure 3A and B shows that, in contrast with the higher levels in plasma NOx in patients with diabetes, cGMP plasma mean values were similar in the two patient groups before anesthesia. They also show that as with NOx, the effect of cardiac surgery on cGMP values was not statistically significant in patients without diabetes, but that it caused a significant increase in the early phase of surgery in patients with diabetes (47 ± 6 vs. 102 ± 12 μmol/mmol creatinine, respectively; P = 0.005). However, the ratios of geometric means of AUC for the overall plasma cGMP values were not significantly different (P = 0.3) between the two groups during the entire study period. Importantly, GTN caused a significant increase in the AUC mean values for plasma cGMP in patients without diabetes (P < 0.001), whereas it had no significant (P = 0.89) effect in patients with diabetes.

As a reflection of the results on plasma cGMP, Figs. 4A and B shows that surgery increased urinary cGMP excretion between 1 and 4 h in patients with diabetes, whereas it had no marked effect in the group without diabetes. In addition, ratios of the AUC for the entire study period further indicate that urinary cGMP excretion was significantly greater (P = 0.02) in the group with diabetes (3,854 ± 17) compared with the group without diabetes (2,612 ± 13). These figures also show that the infusion of GTN resulted in an early increase in cGMP excretion in the two study groups (P = 0.001). This was followed by a period of no further significant increase in both study groups (P = 0.5 in the group with diabetes and P = 0.9 in the group without diabetes).

The results shown in Table 3 indicate that the mean AUC for cGMP clearance was significantly greater (P = 0.05) in patients with diabetes compared with those without diabetes. In addition, GTN infusion significantly reduced cGMP clearance in both study groups (P = 0.04 and P = 0.05, respectively), an effect that may be responsible, at least in part, for the plasma cGMP elevation seen in patients without diabetes.

Figure 5A and B shows that cardiac surgery induces a significant increase in plasma insulin in patients without diabetes and a still greater increase in those with diabetes that peaked at 1 h after the initiation of CPB. Interestingly, this rise in plasma insulin was delayed but not suppressed by the infusion of GTN in both groups of patients. There was no further significant increase in insulin levels after the administration of exogenous insulin.

As seen in Table 4, despite the changes in plasma insulin, blood glucose levels remained within the same range throughout the entire study period in both groups of patients; however, values were more elevated in patients with diabetes than in patients without diabetes.

The present study has shown for the first time that cardiac surgery markedly affects the NO metabolism in patients with diabetes. Although plasma NOx values were already significantly elevated before surgery in the group with diabetes as compared with those without diabetes, further increases occurred during surgery in the former but not in the latter, and this was matched by commensurate elevations in plasma cGMP levels. Our study has also demonstrated that the administration of GTN significantly reduces plasma NOx and its urinary elimination in patients with diabetes but has no significant effect on plasma cGMP levels. These results provide new information on the metabolism of NO during cardiac surgery in patients with diabetes, and the scientific and the clinical implications of these findings are further discussed below.

The first major finding of our study is that NO production increased during cardiac surgery in patients with diabetes, as indicated by the increased plasma levels of NOx and cGMP. In addition, this study shows that the NOx and cGMP plasma values were not significantly affected in patients without diabetes, a finding that is supported by other investigators (24,25,26). These results may be explained by the greater NO production in patients with diabetes under basal conditions (e.g., before surgery) seen in the present study and also reported by other investigators (5). However, it is unlikely that the findings are attributable to changes in renal function, because this was unaffected in both groups of patients. Sharma et al. (25) have shown that during cardiac surgery, NO is not elevated in the coronary effluent of patients with diabetes, suggesting that the source of the increase in NO production seen in our study is not the heart.

The reason for the greater NO production in patients with diabetes is unclear, but an increase in the expression of NOS in these patients may play a role. High glucose levels previously shown to increase the expression of constitutive NOS (27) cannot explain the results of the present study because blood glucose values were well controlled at all times (Table 4). It is therefore unlikely that blood glucose levels could have influenced NOS expression or activity in these patients. An alternative explanation for the increase in NO production in the group with diabetes may be the occurrence of hyperinsulinemia. Indeed, there is evidence in the literature for insulin increasing NOS expression (18) and NO production (19,20,21); however, the observed increase in plasma insulin during cardiac surgery in our study was not followed by an increase in NO in patients without diabetes, and the time-course of the NO elevation seen in patients with diabetes was not related to that of insulin (Figs. 1 and 5). In addition, the infusion of GTN, which blunted NO production in the group with diabetes, delayed but did not suppress the elevation of plasma insulin, which further exposes a lack of association between NO production and hyperinsulinemia during cardiac surgery. It is of interest that the levels of HbA₁ before surgery in patients with diabetes were within the normal range, confirming that the metabolic control in this group was adequate. It is also unlikely that inflammatory factors produced during CPB mediated the increased NO production, because the time-course of factors such as proinflammatory cytokines has no temporal relationship with the time-course of NO (28). The early increase of NO production suggests an involvement of the constitutive NOS rather than the inducible NOS, whose expression and synthesis requires several hours and generates large amounts of NO over long periods of time (7).

The effect of the increase in NO production during CPB on the well being of patients was not under investigation in this study; however, this is still the subject of controversy in the literature. Several investigators have shown that NO may be protective to the ischemic myocardium (29,30,31), whereas others (32,33,34,35,36) have reported the contrary. In recent studies, our laboratory has demonstrated that NO had no detrimental effects on the human myocardium during ischemia/reperfusion (37), but other studies have shown that the interaction of NO with other free radical species promotes oxidative stress (32,33), nitration of proteins (34,35), and modification of enzyme activity (36) that can be detrimental. As for NO, the consequences of increases in cGMP also remain unresolved. Evidence from experimental heart transplantation (29) has suggested that elevated cGMP improves myocardial preservation and graft blood flow. In contrast, other investigators have reported that activation of the cGMP-dependent protein kinase G pathway can induce DNA fragmentation and apoptotic death of cardiac myocytes (33). It is clear that further work is required to define the role of NO metabolism for patients with diabetes-associated cardiovascular diseases.

The second major finding of this study is that GTN influences NO metabolism during cardiac surgery. Thus, infusion of GTN significantly reduced plasma NOx levels in patients with diabetes. Our study has not revealed the mechanism of this GTN effect; however, it is possible that GTN downregulates constitutive NOS activity, a thesis that finds support in the observation that NO acts as a negative feedback modulator of all three NOS isoforms (38,39,40). Another possible mechanism is GTN reducing in the cellular uptake of l-arginine, the precursor of NO. In this connection, there is recent evidence that GTN suppresses the uptake of l-arginine by cultured bovine endothelial cells after 1–4 h of treatment (41). The above two mechanisms are not mutually exclusive and may both contribute to the GTN-induced reduction of plasma NOx in patients with diabetes (diagram in Fig. 6). It is worth noting that although the plasma NOx mean values were not significantly affected by GTN in patients without diabetes, the observed decrease in urinary NOx excretion suggests that GTN also reduces NO production in this group of patients.

The absence of significant changes in plasma cGMP during cardiac surgery in patients without diabetes and the increase in plasma values in patients with diabetes were expected because plasma NOx values were unaffected in the former and elevated in the latter. However, the observed increase in plasma cGMP induced by GTN in patients without diabetes, in whom plasma NOx was unaffected by GTN, and the consistent elevation of plasma cGMP in patients with diabetes, in whom GTN decreased plasma NOx, is at first sight difficult to interpret. This dissociation between plasma NOx and cGMP caused by the administration of GTN in both groups of patients may find an explanation in the kinetics and metabolism of GTN. The mechanism by which GTN affects NO metabolism remains undefined (42,43), but it is possible to speculate that GTN acts directly or indirectly on the enzyme guanylate cyclase by a mechanism that is NO-independent (Fig. 6). A recent study using isolated rabbit aorta and electron paramagnetic resonance with an NO spin-trapping agent has shown that NO is not generated from GTN (44), further suggesting that the action of GTN on guanylate cyclase and the generation of cGMP are not mediated by NO. A recent report by Piatti et al. (45) has shown that in insulin-resistant nondiabetic subjects, plasma NO is elevated, whereas cGMP is reduced. This report also lends support to the thesis that under certain conditions cGMP may be regulated by an NO-independent mechanism.

The mechanism by which GTN reduces cGMP clearance in patients with and without diabetes and NOx clearance in those with diabetes is unknown, but it is possible that some renal mechanism is central to the effects. The reduced clearance of cGMP resulted in a significant increase in plasma cGMP and, as discussed above, this may have important consequences, such as modulation of the vascular tone and blood flow of tissues, including the kidneys.

The effect of GTN on the regulation of NO and cGMP formation seen during cardiac surgery in this study may have important clinical implications for cardiac surgery and other clinical conditions. It may be particularly relevant in disease states such as diabetes, where basal NO formation is increased. The mechanism(s) of the GTN effects remains unclear, as does the clinical consequences of its actions, and both require further investigation before NO donor agents are considered as a tool for the modulation of NO and cGMP production.

This study was partially supported by grants from the Glenfield Hospital NHS Trust, the Royal College of Surgeons of U.K., and Link-Up and Heart Link Trusts. We would like to thank Dr. Melanie Davies (Department of Endocrinology, Leicester General Hospital) for useful advice on patient selection. We acknowledge Drs. David Duthie and J. Swanevelder (Department of Anesthesia, Glenfield Hospital) for administering anesthesia to the patients involved in this study. We also acknowledge Nick Taub (Department of Epidemiology and Public Health, University of Leicester) for the advice on statistical analyses. Lastly, we acknowledge the contribution of the late Professor David de Bono (Professor of Medicine, Division of Cardiology, University of Leicester) for support during this study.