ORIGINAL ARTICLE

---

Although it has been less well characterized in nondiabetic individuals, the association of MA with other CV risk factors (3,6) suggests that its determinants and CV significance may be similar in both diabetic and nondiabetic individuals.

The Heart Outcomes Prevention Evaluation (HOPE) Study is an international randomized double-blind placebo-controlled trial of 9,541 patients with an estimated annual risk for CV disease of ~4–5%. Because 38% of the participants have diabetes, a key preplanned analysis is to determine if vitamin E and/or ramipril are preventive therapies in this subgroup. Urinary albumin was measured in both diabetic and nondiabetic participants as part of the MICRO-HOPE (MA, CV, and Renal Outcomes in the HOPE Study) Study. The purpose of this substudy is to determine if ACE inhibitors and/or vitamin E prevent the development of diabetic nephropathy in diabetic patients with MA and prevent the development of MA in normoalbuminuric diabetic patients. This substudy will also examine the relationship of MA to other renal and CV risk factors and CV outcomes in both diabetic and nondiabetic participants and will determine the impact of MA on the response to the randomized interventions.

This article describes baseline characteristics of diabetic and nondiabetic individuals in the HOPE Study and the distribution of MA in both of these subgroups. It also identifies key variables associated with MA in both of these subgroups and provides estimates of the magnitude of the relationship between these determinants and MA.

RESEARCH DESIGN AND METHODS — The design of the HOPE Study has been described in detail elsewhere (7). In brief, diabetic and nondiabetic individuals aged 55 years or older with a history of previous CV disease (either coronary artery disease, stroke, or peripheral vascular disease) or with a history of diabetes plus at least one other CV risk factor (total cholesterol >5.2 mmol/l, HDL cholesterol <0.9 mmol/l, hypertension, known MA, or current smoking) were recruited. Key exclusion criteria included dipstick-positive proteinuria or established diabetic nephropathy, other significant renal disease, hyperkalemia, congestive heart failure, low cardiac ejection fraction, or hypersensitivity to vitamin E or ACE inhibitors (7). Participants were randomized using a factorial design to receive the ACE inhibitor ramipril (titrated up to 10 mg/day) or placebo, in addition to natural source vitamin E (400 U/day) or placebo. Study follow-up was completed in 1999. The primary end point of the study is the development of either myocardial infarction, stroke, or CV death.

All participants were asked to provide the first morning urine specimen for the determination of MA at baseline after symptomatic urinary tract infections were ruled out. Urinary albumin and urinary creatinine were both measured centrally (1); albuminuria was expressed as the albumin-to-creatinine ratio (ACR). MA was defined as an ACR >2 mg/mmol. All participants were also asked if they have diabetes and the responses were recorded; biochemical testing was not done to identify individuals with undiagnosed diabetes. People were considered to have type 1 diabetes if they developed diabetes before age 30 years and were taking insulin. Other demographic and clinical variables were determined at the baseline assessment on the basis of the history and physical examination. The ankle-arm index (AAI) was calculated as the ratio of the arm to ankle systolic pressure measured by a sphygmomanometer or Doppler (if available). GHb was determined for diabetic patients only by each study center's local laboratory; the upper limit of normal for the assay used was recorded and results were expressed as the percentage above the upper limit of normal. Serum creatinine was similarly measured locally in all participants, and the results were recorded; where necessary, results were converted to Système International (SI) units. Coronary artery disease was defined as a history of myocardial infarction, angina, coronary angioplasty, or bypass surgery; peripheral vascular disease was defined as a history of angioplasty, claudication, or amputation; and left ventricular hypertrophy (LVH) was determined by the investigator's local interpretation of the baseline electrocardiogram.

Statistical analysis
Baseline data for all diabetic and nondiabetic participants according to the presence or absence of MA were summarized. Continuous and categorical data were compared using Student's t tests and Mantel-Haenszel ² tests, respectively. The association of categorical variables with MA was expressed as an odds ratio and 95% CI. Multiple logistic regression models with backward elimination of nonsignificant variables were used to identify the independent determinants of MA in all participants, diabetic participants, and nondiabetic participants. The following variables were tested in the regression models: age; sex; diabetes duration; previous hypertension; dyslipidemia (defined as a documented total cholesterol >5.2 mmol/l or HDL <0.9 mmol/l); laser therapy of one or both eyes; coronary artery disease; stroke or endarterectomy; peripheral vascular disease; smoking status (current or past smoker); oral glucose-lowering agent use; insulin use; aspirin use; BMI; waist-to-hip ratio (in 0.05 increments); systolic and diastolic blood pressure; AAI; LVH; creatinine; and GHb (in 15% increments, which is equivalent to increments of 0.9% if the upper limit of normal of the GHb assay was 6%). Diabetes-related variables were omitted from the model that included the subgroup of nondiabetic participants. All statistical analyses were done using SAS version 6.11.

RESULTS — The HOPE Study includes 9,541 participants; 38.3% had a history of diabetes at randomization. Baseline urinary albumin determinations were available in 3,574 (97.8%) diabetic participants (98% classified as having type 2 diabetes) and 5,708 (97.0%) nondiabetic participants. An ACR >2 mg/mmol was detected in 1,151 (32.3%) diabetic participants and 837 (14.7%) nondiabetic participants.

Table 1 describes the baseline clinical characteristics analyzed in both the diabetic and nondiabetic participants in whom there was a baseline determination of ACR.

When compared with diabetic participants with normoalbuminuria, diabetic participants with MA were older, had a longer duration of diabetes, and had a higher waist-to-hip ratio, systolic and diastolic blood pressure, GHb, AAI, and serum creatinine (Tables 2 and 3). They also were more likely to have a history of retinal laser therapy, hypertension, cerebral vascular disease, and peripheral vascular disease, and were more likely to be smokers, require insulin therapy, and have electrocardiogram criteria for LVH than normoalbuminuric diabetic patients. Similar findings were noted when nondiabetic participants were analyzed according to the presence or absence of MA (Tables 2 and 3). In addition, in nondiabetic patients, females were more likely to have MA than males (Table 3), and individuals taking aspirin were less likely to have MA than those not taking aspirin.

The independent determinants of MA in all participants, diabetic participants, and nondiabetic participants were identified using separate logistic regression models and backwards elimination of nonsignificant variables. The adjusted odds ratios for those determinants remaining in the model for each group are listed in Table 4. For all participants, older age, abdominal obesity, diabetes, smoking, hypertension, vascular disease, and LVH were significantly associated with MA. Sex, history of dyslipidemia, use of aspirin, history of coronary artery disease, BMI, AAI, and creatinine were not independent determinants of MA after adjustment for other factors.

Similar determinants were noted for both the diabetic and nondiabetic subgroups. In addition, diabetic participants with MA were 1.3 times more likely to be on insulin than those without MA and were 1.16 times more likely to have MA for every 10.4 years of diabetes duration. Glycemic control remained an important determinant of MA, even after adjustment for all other variables. For a GHb assay in which the upper limit of the nondiabetic range is 6%, the likelihood of MA increased 8% (95% CI 4–13) for every 0.9% increase in GHb (Table 4).

CONCLUSIONS — The HOPE Study recruited diabetic and nondiabetic patients at high risk for CV events, and the MICRO-HOPE substudy determined the baseline ACR in 97% of all participants. At randomization, 32.3% of diabetic participants and 14.7% of nondiabetic participants had an ACR >2 mg/mmol. Despite the fact that for diabetic patients MA was a possible eligibility criteria for inclusion in the study (and hence diabetic patients with MA were preferentially included), the prevalence in diabetic participants is consistent with observations made in other studies of middle-aged patients with diabetes (1,8). Conversely, because MA was not an inclusion criteria for nondiabetic participants, the 15% prevalence observed in the nondiabetic participants is likely to reflect the prevalence of MA in similar middle-aged individuals with a history of vascular disease and no known history of diabetes.

MA was clearly and independently associated with several risk factors for CV disease, including age, hypertension, current smoking, previous CV disease, LVH, abdominal obesity, and (in diabetic participants) glycemic control. These observations are consistent with other studies in both diabetic and nondiabetic (2–4,6,9) individuals and support the view that albuminuria is an easily measured marker of underlying vascular disease. They also support the possibility that interventions (such as ACE inhibitors) that reduce or prevent MA may prevent both clinically significant renal disease as well as cardiovascular disease—a hypothesis that is being tested in the current study.

These data have several limitations. First, ACR was determined on the basis of a single urine collection that participants were asked to collect on the morning of their visit. However, the large number of samples collected and the high diagnostic accuracy of a single urine collection (10–13) should minimize the uncertainty associated with day-to-day differences in the degree of albumin excretion (14). Second, no objective test was done to identify participants with undiagnosed diabetes. Recent population studies suggest that as many as 35% of all patients with diabetes may be undiagnosed (15). It is therefore possible that a sizeable portion of the "nondiabetic" microalbuminuric patients in the study may have undiagnosed diabetes. Other studies in nondiabetic individuals report that high glucose levels are a risk factor for both MA (9,16) and CV disease (17). These studies support the possibility that there is a high prevalence of undiagnosed diabetes and nondiabetic hyperglycemia in these individuals. Third, both GHb and serum creatinine determinations were done locally and not centrally—a fact that decreases the sensitivity to detect differences between groups. However, the variability introduced because of the way in which these variables were measured (i.e., ACR, diabetes status, and GHb) will lead to an underestimate of any association. Thus, the true associations between MA and other variables are likely to be even stronger than those observed. Fourth, the same ACR cutoffs were used for both men and women to define MA. These cutoffs may overestimate the prevalence of MA in women, in whom the currently recommended diagnostic cutoff for MA is 2.8 mg/mmol. The fact that some nonmicroalbuminuric women would have been classified as having MA, would have also had the effect of diluting the strength of association of any of the determinants with MA. Thus, the observed odds ratios may moderately underestimate the odds ratios for the relationship between the measured determinants and MA.

Finally, this article reports a cross-sectional analysis of high-risk individuals who consented to participate in a long randomized trial. As such, it is not a random sample of diabetic and nondiabetic individuals. As noted in the methods section, individuals with diabetes were included if they had previous CVD or if they had one other CV risk factor in the absence of previous CVD. By contrast, all of the nondiabetic individuals needed to have a history of CVD to participate. These different inclusion criteria partially account for some of the differences noted in Table 1 and highlight the importance of interpreting differences in the baseline characteristics of the diabetic and nondiabetic participants very cautiously. They may also account for the differences in the size of the univariate odds ratios for some of the risk factors (Table 3).

These data therefore provide further support for the conclusion that MA identifies individuals at high risk for CV disease. They also suggest that participants with MA are likely to be at higher risk for CV disease than those with a normal ACR and may therefore be expected to have a higher CV event rate during the course of the HOPE Study. Furthermore, the large number of microalbuminuric participants should allow the detection of any differential effect of ACE inhibitors and/or vitamin E on the risk of CV disease in HOPE Study participants with MA compared with those participants without baseline MA.

---