Increased Urinary Albumin Excretion, Insulin Resistance, and Related Cardiovascular Risk Factors in Patients With Type 2 Diabetes: Evidence of a sex-specific association

This study was conducted in Caucasian patients with type 2 diabetes who were residents of the Gargano region and surrounding areas, i.e., the center east coast of Italy. A total of 712 type 2 diabetic patients (363 men and 349 women) who met the following criteria were consecutively recruited at the Endocrine Unit of the CSS Scientific Institute in San Giovanni Rotondo: diabetes diagnosed after age 30 years, absence of ketones at diagnosis, and absence of clinically evident autoimmune disease. These criteria were chosen to minimize the risk of including late-onset type 1 diabetic patients. In addition, anti-GAD antibodies were determined in 200 randomly selected patients, and only 3 of 200 patients turned out to be anti–GAD antibody positive (>0.9 units/ml), thus indicating that the risk of misdiagnosis is, in fact, trivial in our sample. All patients were interviewed regarding the age of type 2 diabetes diagnosis and ongoing antidiabetic treatments. Duration of diabetes was calculated from the calendar year of data collection minus the calendar year of diabetes diagnosis. Clinical features of patients recruited and subgrouped according to sex and presence or absence of micro- and macroalbuminuria are summarized in Table 1. Among 403 patients who were on antihypertensive treatment, 291 (72%) patients were on either ACE inhibitors or angiotensin II receptor 1 blockers (ARBs); among these, dihydropiridinic calcium antagonists were added in 76 (26%), whereas diuretics and α- or β-blockers were added in 183 (63%) patients. All subjects enrolled in the study underwent physical examination, including measurements of height, weight, waist circumference, and blood pressure (i.e., two measurements rounded to the nearest 2 mmHg in the sitting position after at least 5 min rest, using an appropriate-sized cuff; diastolic blood pressure was recorded at the disappearance of Korotkoff sound, phase V).

Fasting venous blood was sampled from an antecubital vein from all patients for the measurement of serum insulin, total cholesterol, HDL cholesterol, serum triglycerides, and HbA_1c, as previously described (17). Urinary albumin and creatinine concentration were determined the same morning of the clinical examination on an early morning first void sterile urine sample by the nephelometric method (Behring Nephelometer Analyzer) and the Jaffè reaction-rate method (Hitachi 737 Autoanalyzer), respectively. The urinary albumin-to-creatinine ratio (ACR) was then calculated. Microalbuminuria was diagnosed if the ACR was ≥2.5 mg/mmol in men and 3.5 mg/mmol in women but <30 mg/mmol. Macroalbuminuria was defined as an ACR >30 mg/mmol, a level that approximates an albumin excretion of 300 mg/24 h, considered as the upper limit of microalbuminuria (18). A total of 179 patients (27%) had microalbuminuria and 37 (5%) had macroalbuminuria. The exclusion of patients with macroalbuminuria from the analysis did not change the results obtained, which were, therefore, presented for the entire cohort. The group of patients with micro- or macroalbuminuria was named the MA group.

Creatinine clearance was calculated with the Cockcroft-Gault formula (19) and was used as a measure of glomerular filtration rate. The homeostasis model assessment of insulin resistance (HOMA_IR) index was calculated as fasting serum insulin (mU/ml) × fasting plasma glucose (mmol/l)/22.5 (ref. 20).

Patients were considered to have arterial hypertension if systolic blood pressure (SBP) was ≥130 mmHg and diastolic blood pressure (DBP) ≥85 mmHg or were currently receiving antihypertensive treatment. Patients were considered to have dyslipidemia if they were currently receiving lipid-lowering treatment or had total cholesterol ≥200 mg/dl, HDL cholesterol ≤40 mg/dl in men and 50 mg/dl in women, and triglycerides ≥150 mg/dl (21). Presence of retinopathy was defined on funduscopy examination or as a history of laser therapy. Information on smoking habits was obtained by a questionnaire.

The study was performed according to the Helsinki Declaration, and the protocol was approved by the local ethics committee. All subjects provided written informed consent.

Data are reported as the mean ± SD or median (range). Mean differences were compared by unpaired Student’s t test or Mann-Whitney U tests, as appropriate. A P value <0.05 was considered to be significant.

Univariate and multivariate analyses were performed to correlate independent variables, such as HOMA_IR, HbA_1c, SBP, DBP, waist, duration of disease, sex, age, smoking, lipids, antidiabetic therapy, and use of ACE inhibitors or ARBs with the dependent variable as ACR. Then, in a multivariate logistic regression analysis, we explored the effect of independent variables (i.e., HOMA_IR) on the binary dependent variable (i.e., micro- or macroalbuminuria: yes or no) and expressed as odds ratio (OR) (95% CI). For these analysis, skewed distributed variables (i.e., HOMA_IR, ACR, and triglycerides) were logarithmically transformed.

Interaction between HOMA and sex in determining UAE was tested with GLM Univariate analysis, statistical packaged SPSS version 11.5 (SPSS, Chicago, IL). The same statistical package was also used for all other statistical analysis.

In the entire cohort, ACR was significantly correlated with HOMA_IR (r = 0.15, P = 0.0001), age (r = 0.12, P = 0.001), duration of diabetes (r = 0.13, P = 0.0005), SBP (r = 0.15, P = 0.0009), HbA_1c (r = 0.16, P = 0.00001), triglycerides (r = 0.19, P < 0.00001), and waist circumference (r = 0.10, P = 0.02). HOMA_IR was also correlated with triglycerides (r = 0.20, P < 0.00001) and HbA_1c (r = 0.24, P < 0.00001). A similar correlation was observed when smoking was also added to this multivariate model (P = 0.02).

When the relative contribution of all the above-mentioned variables singly related to ACR was analyzed by multiple regression analysis, HOMA_IR (P = 0.03), age (P = 0.009), SBP (P = 0.0001), HbA_1c (P = 0.005), and triglycerides (P = 0.0001) but not duration of diabetes (P = 0.1) and waist circumference (P = 0.6) were still significantly associated.

The correlation between HOMA_IR and ACR was still significant when antidiabetic (P = 0.02) or antihypertensive therapy (P = 0.001) or only therapy with ACE inhibitors and/or ARBs (P = 0.03) were considered as independent variables. To further adjust for the possible confounding effects of antidiabetic and antihypertensive treatments on both ACR and insulin resistance, data were reanalyzed in the subgroup of normotensive patients (n = 206) or in the subgroup of patients treated with diet alone (n = 82). In both subgroups, the correlation between HOMA_IR and ACR was still significant (r = 0.14, P = 0.049 and r = 0.3, P = 0.009, respectively).

When the relationship between HOMA_IR and ACR was investigated separately in women and men, although a similar trend was observed in the two sexes, a significant correlation was observed in men (n = 363; r = 0.21, P = 0.0001) but not in women (n = 349; r = 0.08, P = 0.14). In fact, a significant interaction (P = 0.04) between HOMA_IR and sex in determining ACR was observed. The correlation between HOMA_IR and ACR in men was still significant (P = 0.001) in a multiple regression analysis model after adjusting for age, duration of disease, SBP, HbA_1c, triglycerides, waist circumference, and smoking.

ACR measurement reproducibility in our study was calculated in a representative sample of 329 patients (46% of the entire cohort) who collected a second urine sample in a similar fashion. The κ coefficient of ACRs in this subgroup was 0.64, thus indicating a good reproducibility of our ACR measurement. When the second ACR measurement was considered in this subgroup of patients, a significant correlation with HOMA_IR was observed in men (r = 0.18, P = 0.024) but not women (r = 0.02, P = 0.8).

When patients were stratified according to quartiles of HOMA_IR value, there was a progressively increased risk of belonging to the MA group, which became significant for patients of the fourth quartile (i.e., the most insulin resistant) (OR 2.3 [95% CI 1.4–3.7]) (Table 2). Similar to that observed on the linear correlation between the two variables, the association between insulin resistance and micro- and macroalbuminuria was significant among men but not women (Table 2).

We then explored the relationship between UAE and the three specific cardiovascular risk factors that are strictly related to insulin resistance–dependent metabolic syndrome, i.e., waist circumference (>102 cm for men and >88 cm for women), increased blood pressure, and dyslipidemia, as previously defined. Accordingly, patients were subdivided in two groups: group A with fewer than two cardiovascular risk factors (n = 90) and group B (n = 622) with two or more cardiovascular risk factors. ACR was significantly higher in group B (median 1.33 [range 0.1–1367.6]) than in group A (0.91 [0.2–157.5]) (P = 0.01 after adjusting for age and sex). This association was also significant in men (0.90 [0.2–115.1] vs. 1.56 [0.1–1367.6], P = 0.005) but not women (1.18 [0.3–157.5] vs. 1.22 [0.2–134.0], P = 0.9) patients.

The relationship between insulin resistance and UAE in type 2 diabetic patients has been a matter of debate with both positive (7–10) and negative (4–6) results obtained in samples of small to moderate size ranging from 20 (4) to 155 (6) patients; this might have increased the risk of both false-positive and false-negative results. Our present data, obtained in a large sample, clearly indicate that in patients with type 2 diabetes, there is a highly significant association between insulin resistance and UAE. Similarly, insulin resistance was also associated with the presence of micro- and macroalbuminuria. In line with the associations we describe here is a recent report (22) indicating that insulin sensitizer drugs may decrease UAE in type 2 diabetic patients. The association between insulin resistance and UAE is independent of possible confounders affecting both variables, including age, smoking, duration of disease, triglycerides, waist circumference, smoking, metabolic control, blood pressure, and related treatments.

A novel finding of our study is that the correlation between insulin resistance and UAE was significant in male but not female type 2 diabetic patients, although a similar trend was also observed in the latter group. Worth noting, the two groups were of similar magnitude; whether a statistical significance would be reached also in women with a much larger sample size is a possibility that deserves further studies. With the present sample size, sex and insulin resistance seem to interact in modulating UAE in our cohort. It is a matter of debate whether a similar interaction also occurs in the general population (14–16). Why the association between insulin resistance and UAE seems to be stronger in men than women is unknown. Differences in hormonal assessment in the two sexes might be relevant. In our cohort, the vast majority (87%) of women were >50 years of age and, therefore, presumably in the postmenopausal state. Therefore, although it cannot be excluded that hormonal exposition in previous years might be important, estrogens are unlikely to have a major role in this context. At variance, androgens that are known to activate the renin angiotensin system (23) could have played a role in modulating the detrimental effects of insulin resistance and related abnormalities on endothelial cells. Differences in lifestyle between the two sexes, including smoking, diet, and physical activity, might also be implicated in the apparently different results obtained.

Also, the biology of the association between UAE and insulin resistance is not clearly defined. There are several possible explanations. First, it might be based on compensatory hyperinsulinemia characterizing insulin resistance, which may stimulate renal sodium reabsorption, leading to volume expansion, increased adrenergic activity, and hypertension (24–25). Worth noting, insulin resistance could contribute to greater salt sensitivity, increased glomerular pressure, and increased UAE, as recently described in type 2 diabetic patients with microalbuminuria put on a high-salt diet (26). Thus, although in our present series a correlation between insulin resistance and blood pressure was not detectable, likely because of the ongoing antihypertensive treatment, one could speculate that the difference in blood pressure level before treatment was started may have played a significant role in determining the association between UAE and insulin resistance. Second, several reports have suggested that high triglyceride levels, a typical feature of insulin resistance, represents an independent predictor of renal damage in both type 2 diabetes (27) and type 1 diabetes (28). In fact, in our sample, HOMA_IR values were significantly correlated with serum triglycerides concentration. Increased formation of small dense LDL, which have increased susceptibility to oxidation, is a likely mechanism through which triglycerides could initiate endothelial damage and eventually induce renal disease (27). Third, patients with insulin resistance are likely to have significantly worse metabolic control (i.e., higher HbA_1c levels), which in fact turned out to be the case in the cohort we studied. This difference may have also played a role in determining increased UAE.

It is worth noting that UAE was also associated with the presence of multiple cardiovascular risk factors strictly related to insulin resistance, namely increased waist circumference, arterial hypertension, and dyslipidemia. This association may partly explain the well-established association between increased UAE and cardiovascular morbidity and mortality (29–30) shown in type 2 diabetes.

We acknowledge that the HOMA_IR index used in our present study for measuring insulin resistance is a surrogate of the euglycemic-hyperinsulinemic clamp, the gold standard technique for measuring insulin sensitivity. However, the clamp technique is not applicable to several hundreds of patients. In addition, recent reports have shown a strong correlation between HOMA_IR and euglycemic-hyperinsulinemic clamp in both nondiabetic (31) and type 2 diabetic individuals over wide ranges of fasting plasma glucose (31–32), thus validating the use of HOMA_IR for large epidemiological studies.

It should be considered that our study was conducted on referred patients; therefore, results obtained cannot be applied to the whole diabetic population. A similar limitation also affects previous results on this issue (4–5,7). In conclusion, because of the large sample analyzed here, we believe our present results add conclusive evidence that in type 2 diabetic patients, UAE is strictly associated with insulin resistance and related cardiovascular risk factors. This association seems to be stronger in men than in women. Because of the possibility to treat insulin resistance with several drugs (11–12) and given the efficacy of this treatment in reducing cardiovascular mortality (13) in diabetic patients, our study suggests that type 2 diabetic patients with increased UAE could particularly benefit from treatment for insulin resistance. In line with this hypothesis, Bakris et al. (22) have recently shown that rosiglitazone, an insulin sensitizer, was effective in reducing UAE in patients with type 2 diabetes. Further prospective studies are needed to better clarify the causative role of insulin resistance on UAE and the possible clinical relevance of this association.

This research was supported from a grant of the Italian Minister of Health.

We thank Professor Giancarlo Viberti (London, U.K.) for reading the manuscript and providing useful suggestions.