Influence of Type 2 Diabetes on Functional and Structural Properties of Coronary Artery Bypass Conduits

Demographics of the 160 consecutive patients (aged 46–78 years; mean age 64 ± 7 years) included in the study are shown in Table 1. Patients who had a history of chronic heart or renal failure, had overt peripheral vascular disease, or were undergoing emergency surgery were excluded from the study. Patients on combined oral hypoglycemic agents and insulin treatment (n = 3) were also excluded. Of the 160 subjects, 90 had type 2 diabetes; of those 90, 48 were on oral hypoglycemic therapy (sulfonylureas alone or in combination with metformin) and 42 were on insulin therapy. None of the diabetic patients were on thiazolidinediones or metformin alone. The nondiabetic (control) group comprised 70 patients who had no preoperative history of diabetes. All patients were on conventional antianginal therapy (vasodilators, β-blockers, or calcium antagonists). Additional risk factors were present in the majority of the enrolled patients (Table 1). Patients previously diagnosed as dyslipidemic (total cholesterol level >200 mg/dl) (13) or hypertensive (systolic blood pressure >140 mmHg, diastolic blood pressure >90 mmHg) (14) were on appropriate lipid-lowering or hypotensive agent treatments.

Fasting blood glucose (gluco-oxidase, Beckman Glucose Analyzer II), HbA_1c (chromatographic method; Biodata, Roma, Italy; normal range 3–6%), and fasting C-peptide (radioimmunoassay method; normal range 0.9–2.5 ng/ml) were measured in all patients on the day they were admitted to the hospital. Urinary sodium, potassium, and creatinine concentrations were measured by flame photometry. All patients also underwent 24-h urine collection for detection and quantification of proteinuria (immunonephelometric method; nephelometer BNA 100, Behring Scoppito, Italy; intra-assay coefficient of variation [CV] 4.3%, inter-assay CV 4.4%, sensitivity limit of the assay 1.76 mg/l).

The excess segments of left internal thoracic arteries (LITAs) and SVs collected during routine CABG were used for the study. All vascular segments were harvested from the legs using the open method, and care was taken not to clamp any of the conduit segment freed for functional or histological assessment. The specimens were immediately immersed in a modified Krebs-Henseleit solution gassed with 95% O₂ and 5% CO₂ and with the following composition (in mmol/l): NaCl, 113; KCl, 4; CaCl, 1.9; MgSO₄, 1.2; NaHCO₃, 25; KH₂PO₄, 1.2; and glucose, 11.5. This solution was prepared to achieve a pH of 7.4. The vascular wall was separated from the surrounding tissue and spirally cut to obtain strips 2 cm long and 3 mm wide. The vascular strips were set up in an isolated organ bath (5 ml) containing a Krebs-Henseleit solution at 37°C (pH 7.4). Vessel preparations were suspended under an isometric passive stretch of 1 g and left to equilibrate for 120 min to achieve complete fiber relaxation. Tension was measured with an isometric transducer connected to a pen-writing recorder. The isometric transducer was calibrated before each experiment. Contractile and vasodilative responses were assessed by infusion of KCl (90 mmol/l), norepinephrine (10⁻⁸ to 10⁻⁵ mol/l), and serotonin (10⁻⁹ to 10⁻⁴ mol/l). The vasoconstrictive effect was evaluated as the vertical force acting on the isometric transducer (grams). When a contractile plateau was maintained for 10–15 min, it was antagonized by endothelial-independent vasodilative agents (nifedipine, captopril, isosorbide dinitrate, and trinitroglycerine) at increasing dosages from 10⁻⁹ to 10⁻⁵ mol/l. The extent of vasodilation was expressed as an inhibitory percentage of the ongoing vasoconstriction. Endothelium-dependent vascular reactivity was also assessed by evaluating the antagonism induced by acetylcholine (10⁻⁸ to 10⁻⁵ mol/l) during norepinephrine-induced specimen contraction (10⁻⁵ mol/l).

Histological assessment was carried out on formalin-fixed specimens (vessel length 0.3–1.8 cm), processed for light microscopy. All vascular tissues for histology were immediately fixed intraoperatively. Hematoxylin-eosin, orcein (elastic fiber specific), and factor VIII (endothelial specific, with factor VIII antibodies) staining were used to analyze arterial and venous segments. Intimal involvement was judged according to the morphometric classification reported by Thiene et al. (15), as follows: grade 1, focal intimal proliferation with minimal narrowing; grade 2, concentric proliferation with mild narrowing; grade 3, concentric proliferation with moderate narrowing; and grade 4, concentric proliferation with marked narrowing. Pathologists (V.V., R.S.) were blinded in relation to the patient subgroups of the analyzed tissue specimen.

All data are presented as means ± SE. Functional data are presented as the mean of five or six experiments in relation to each pharmacological test. Comparisons between groups were performed using the t test for unpaired data. We used a limit of significance of 0.05. Univariate and multivariate analyses (Cox regression model) were performed to analyze predictors of structural impairment in relation to preoperative variables such as age, sex, type of diabetes treatment, duration of diabetes, glucose control, and micro- and macroalbuminuria (Table 1). Albuminuria and GHb levels were also analyzed as separate classes (<30, ≥30 and ≤300 mg, and >300 mg/24 h for albuminuria, and ≤6 or >6% for GHb) in the regression model in relation to the pattern of intimal morphometry.

Patient profiles did not differ between the diabetic and nondiabetic groups, nor between insulin-treated and non-insulin-treated diabetic patients (Table 1). The prevalence of type 2 diabetic patients as compared with normal subjects in our series can be explained not only by epidemiological reasons (diabetes as major risk factor for coronary artery disease) (16), but also by the active cooperation of our Endocrine Unit as well as the fact that complicated patients are referred to our center. The similar historical prevalence of dyslipidemia and hypertension (defined by World Health Organization criteria) among the diabetic and nondiabetic patients was likely attributable to the high-risk atherosclerotic population included in the study. Alternatively, it could have resulted from the fact that in the present study blood lipid and blood pressure were not specifically assessed (e.g., on and off treatment).

Metabolic characteristics of the patients are shown in Table 2. Diabetic patients showed significant increase in urinary albumin excretion. Macroalbuminuria was present only in diabetic patients. The treatment modality of diabetic subjects did not apparently influence glycometabolic control as no significant differences were found in terms of HbA_1c levels between diabetic patients treated with insulin and those treated with oral hypoglycemic agents.

The contractile response of vascular segments induced by vasoconstrictor agents is shown in Fig. 1. KCl induced a stable contraction of the arterial graft at a concentration of 90 mmol/l; the response did not increase at higher dosages. Norepinephrine and serotonin induced a significant dosage-dependent increase in vessel tension starting from concentrations of 10⁻⁸ and 10⁻⁹ mmol/l, respectively. The different effect of vasoconstrictors on arterial and venous grafts of diabetic and nondiabetic subjects are shown in Fig. 1. The vascular reactivity to spastic agents of LITA and SV conduits was somewhat increased in diabetic patients compared with nondiabetic patients, and in both groups the vein graft response was more active as compared with arterial conduits, but these differences were not significant. The vasodilatory response of CABG grafts to endothelial-independent compounds did not differ between the groups at all dosages tested (Fig. 2). Plateau vasoconstriction as well as reversal with acetylcholine were similar with all three vasoconstrictive agents (KCl, norepinephrine, and serotonin). The endothelium-dependent vasodilation was well preserved in arterial conduits of both groups of patients. On the contrary, SV grafts coming from diabetic patients proved to be significantly less reactive to acetylcholine (10⁻⁸ to 10⁻⁵ mol/l) than vein segments harvested from nondiabetic patients (Fig. 3).

The histological studies of arterial conduits showed preserved structure in both groups, with scattered areas of intimal hyperplasia in some specimens. In contrast, SV grafts of diabetic patients showed more profound abnormalities of the vessel layers as compared with the control group, particularly at the intima level (Fig. 4). Severe intimal hyperplasia (“arterialized vein”) was a common finding in SV grafts of diabetic patients, but milder intimal degeneration was also found in nondiabetic subjects. Calcified and diffuse atherosclerotic plaques were found in the venous conduits of some diabetic patients (Fig. 5), and were not documented in any of the normal subjects. The relation between histological findings and endothelium-dependent vasodilatory response is shown in Fig. 6. Noticeably, venous grafts harvested from diabetic patients proved to be more dysfunctional than correspondent conduits of nondiabetic patients with similar lesions; this finding suggests that additional factors, other than the structural derangement, may contribute to the impairment of the endothelium-mediated vasodilation observed in diabetic subjects.

On univariate analysis (Table 3), preoperative hyperglycemia, micro- and macroalbuminuria, and high plasma levels of GHb and C-peptide were predictors of significant (grade 3 and 4) structural SV graft abnormalities. On multivariate analysis, however, microalbuminuria >300 mg/day and high levels of GHb were found to be independent predictors of vessel wall degeneration of venous conduits in diabetic subjects (Table 4). Conversely, on both univariate and multivariate analyses, age, sex, and type of diabetes treatment were not found to be predictive of intraoperative graft abnormalities.

Several clinical studies have reported that diabetes is a major determinant of prognosis in CAD patients and exerts a negative impact on short- and long-term survival and freedom from myocardial ischemia-related events, particularly after surgical revascularization (1–7). The mechanisms by which diabetes influences the early and late outcome of CABG subjects remain, however, controversial. Micro- and macroangiopathy are well-known complications of long-lasting or poorly controlled diabetes. The extent of vascular compromise in diabetes is particularly relevant in CAD patients, as it may account for the more severe compromise of the coronary arteries and may also impair the biological integrity of bypass grafts (4,5,9). Preserved functional and structural properties of CABG conduits are essential for a successful myocardial revascularization (8,9). Abad et al. (17) showed a low incidence of atherosclerotic changes in LITA grafts of patients undergoing CABG. Likewise, macro- and microscopic signs of atherosclerosis were found only in 12% of 314 LITA grafts analyzed by Sisto and Isola (18). However, the patients in these studies were relatively young and few had diabetes. In our series, we included many elderly and diabetic patients; nonetheless, no abnormal structural characteristics of the arterial grafts could be documented despite the concomitance of additional risk factors in many cases.

Vasoactive response of the bypass conduits may also influence postoperative outcome, particularly during the early phases. Our findings showed normal vasomotor properties of the LITA grafts in diabetic patients and concurred with the data of Wendler et al. (19), who recently showed preserved vasoreactivity of arterial conduits (internal thoracic and radial arteries) in diabetic patients who underwent myocardial revascularization. Therefore, arterial grafts of diabetic patients appear to maintain their biological integrity, even in the presence of poorly compensated glucose metabolism. These concepts may explain the long-lasting and excellent patency rate of the LITA graft, even in the presence of pathologies potentially affecting vessel biology (5,10).

Unfortunately, the intraoperative condition and postoperative fate of the vein grafts appeared different. Surgical manipulation may affect SV graft integrity and may promote neointima formation (20–22). Furthermore, the postoperative maladaptive changes of the SVs exposed to systemic arterial pressure (“arterialization response”) are well known (9). In contrast, the effects of diabetes on the venous structure and function are poorly defined. Our data indicate that diabetes does deeply influence the biological properties of the systemic venous system. Histology of venous bypass conduits in diabetic patients has showed signs of the “arterialization process” before coronary implantation. However, intimal derangement was found, although less frequently and to a limited extent, in the control group, suggesting that additional factors may contribute to the compromise of the vein structure. This is in accordance with the findings of Thiene et al. (15), who showed abnormalities of the vein intima and endothelial derangement in 70% of patients undergoing CABG procedure. The endothelium is the major component of local vascular control and, therefore, of the antiatherogenic pathway. Endothelial cells of the vessels have been shown to act as a paracrine-autocrine system that effectively balances local vasomotor response, but concomitantly controls endothelium-platelet interaction, smooth muscle cell proliferation, and leukocyte adhesion, all important initiating factors in the development of atherosclerotic plaque. Accordingly, the endothelial response and integrity in transplanted vein conduits play even more important roles in the new functional demands of a SV conduit in CABG. In accordance with the findings of Karasu et al. (23), our study showed preserved endothelium-independent vasodilation, whereas the endothelium-dependent relaxation was greatly impaired in SV grafts in diabetic subjects. Rodriguez-Manas et al. (12) showed that endothelium-related release of NO was reduced in veins from diabetic rats. Reduced stimulatory activity of human venous endothelial cells on prostacyclin production, which may be related to reduced NO secretion, reduced NO-dependent vasodilation, and greater predisposition to vascular wall atherosclerosis, was shown in diabetic subjects (24,25). Finally, Aaltomaa et al. (26) showed that SV endothelial cells of diabetic patients have a greater tendency to proliferate compared with those cells harvested from normal subjects, suggesting an abnormal intrinsic regulation of endothelial functions. These findings suggest that diabetes may exacerbate the postoperative maladaptive changes of venous conduits (27,28). Diabetes, however, does not invariably alter vascular integrity, particularly at the venous district. We found that the extent of structural compromise of the SV intima was directly related to the metabolic control of the endocrine disorder. Microalbuminuria, which is a well-established marker of early-stage nephropathy and an index of poor control of diabetes, has been shown to be a negative predictor of cardiovascular-related survival in diabetic subjects (29,30). The mechanism linking diabetes-induced microalbuminuria to increased mortality in CAD patients is, at present, not well understood. Several hypotheses have been proposed, but microalbuminuria does seem to be a marker of vascular damage in diabetes, particularly of impaired endothelial function (11). High levels of GHb, a well-established marker of long-term glycometabolic control, also correlated with the degree of SV morphological degeneration in our series. This finding highlights the importance of careful diabetes monitoring and management in CAD patients, particularly if they have undergone the CABG procedure. Therefore, a simple assessment of preoperative glucose metabolism−related profile in diabetic CABG candidates might provide some clues regarding venous graft quality and prompt adjustment in surgical strategy, taking into consideration that a more liberal use of internal mammary arteries has been shown to not increase postoperative complications in diabetic patients (31). Arterial grafts, which in our study showed preserved functional and structural integrity, appeared, therefore, to be more suitable for CABG procedure, particularly in the presence of poorly compensated diabetes or when signs of peripheral diabetes-related complications have already occurred. More aggressive or patient-tailored pharmacological therapy could be also advisable in diabetic subjects (32–36), especially those with poorly controlled disease, in an attempt to reduce the impact of other risk factors on the fate of venous conduits, although this approach has given conflicting results.

With regard to the limitations of this study, it did not include patients with frank peripheral vascular or severe renal disease. Vascular complications may therefore have been underestimated, especially for diabetic patients. The relatively small number of patients may also have affected the study results. Functional and morphometric analyses were performed in diabetic and nondiabetic subjects who exhibited concomitant pathologies (hypertension, dyslipidemia) that might variably contribute to vascular derangement. Ideally, we should have included only diabetic patients with no other confounding risk factors for the biological properties of bypass grafts, but routine CABG candidates affected by only one risk factor are extremely rare. The influence of associated pathologies on the vascular effects of diabetes must therefore be taken into account.

In conclusion, our study findings indicate that LITA grafts maintain their biological integrity in diabetes. In contrast, abnormal intimal-related morphometry and endothelium-dependent dysfunction of vein conduits are likely in diabetic subjects undergoing surgical myocardial revascularization, particularly in the presence of suboptimal glucose control. Based on these findings, the surgical and postoperative management of diabetic patients submitted to CABG procedures requires special therapeutic strategies. A more liberal use of arterial grafts and closer postoperative monitoring of the efficacy of hypoglycemic treatment are highly advisable to ensure a less complicated perioperative course and to favor longer benefit from myocardial revascularization. Further investigations, however, are warranted to show conclusively whether preoperative graft features induced by diabetes may significantly affect early and long-term patency rate and, ultimately, patient outcome after CABG procedure.

We gratefully acknowledge the Associazione “CUORE e Ricerca” for providing invaluable support.