Do plasminogen activator inhibitor (PAI-1) or tissue plasminogen activator PAI-1 complexes predict complications in Type 1 diabetes: the Pittsburgh Epidemiology of Diabetes Complications Study.

AIMS: To examine the predictive power of plasminogen activator inhibitor-1 (PAI-1) and the complexes it forms with tissue plasminogen activator (tPA-PAI-1) for the two major Type 1 diabetes (T1D) complications (coronary artery disease (CAD) and overt nephropathy) in the context of standard risk factors. METHODS: Observational prospective study of 454 participants with childhood onset (< 17 years) T1D, aged 18+ years at baseline. PAI-1 and tPA-PAI-1 were determined using ELISA methodology. Follow-up (6 years) was limited to 382 individuals for CAD and 294 individuals for overt nephropathy, after excluding baseline cases. Total, HDL and LDL-cholesterol, triglycerides, HbA1, blood pressure, body mass index (BMI), waist-hip ratio (WHR), leucocyte count, Beck depression score and fibrinogen were also examined. RESULTS: The 56 incident cases of CAD had marginally lower PAI-1 and higher tPA-PAI-1 levels compared with those free of CAD. However, marginally higher PAI-1 and significantly higher tPA-PAI-1 (P = 0.04) levels were seen in those who developed nephropathy. After controlling for age, both PAI-1 and tPA-PAI-1 showed significant negative correlations with HDL-cholesterol, and positive correlations with triglycerides, WHR, HbA1 and fibrinogen. tPA-PAI-1 was also positively correlated with total and LDL-cholesterol. In multivariate analyses, neither PAI-1 nor tPA-PAI-1 was an independent predictor of CAD or overt nephropathy. CONCLUSIONS: These results suggest little association between PAI-1 and later CAD in patients with T1D. However, tPA-PAI-1 complexes may be involved in the pathogenesis of overt nephropathy.

Physical activity and cardiovascular risk factors in a developing population.

PURPOSE: Noncommunicable diseases are emerging in developing countries. However, few studies have been conducted in those countries to evaluate the role of physical activity in the development of cardiovascular diseases. This study investigated physical activity and its relationship to risk factors for cardiovascular disease in a large population (N = 799) of civil servants from Benin City, Nigeria. METHODS: Physical activity levels were estimated by an interviewer-administered questionnaire, which determined the average hours per week over the past year spent in occupational and leisure activities. Time spent walking or biking to work was assessed as well. Other major measures included body mass index (BMI), waist-hip ratio (WHR), blood pressures, plasma insulin level, lipid profiles, and diet. RESULTS: More of the physical activity was attributed to occupational than to leisure activities. Compared with women, men had a higher activity level. No significant trend was observed across age groups. Male senior staff (a marker of higher socioeconomic status) had a lower physical activity level than male junior staff. Physical activity, especially time walking or biking to work, was inversely correlated with weight, BMI, WHR, blood pressures, insulin, total cholesterol, LDL and HDL cholesterol, and triglycerides in men, while such correlations were not consistent in women. In multivariate analysis in men, blood pressure and insulin were independently associated with BMI but not with walking, while an independent inverse association was seen between walking and BMI. CONCLUSION: Lack of physical activity was associated with adverse risk profiles for cardiovascular disease in this developing population.

Lipid modulation in insulin-dependent diabetes mellitus: effect on microvascular outcomes.

Although hyperlipidemia is associated with the development of diabetes complications, the effect of lipid reduction on microvascular complications is unknown. We initiated a 2-year, randomized, double-blinded placebo-controlled pilot trial of simvastatin/diet vs. diet alone in Type 1 diabetic patients without overt nephropathy. Thirty-nine patients with LDL cholesterol 100-160 mg/dl, >10 year duration of diabetes and an albumin excretion rate (AER) <200 microg/min were recruited for study. The primary end-point was change in AER. Secondary end-points were change in ankle-brachial index, progression of retinopathy status, change in vibratory threshold, and development of new clinical neuropathy. Nineteen patients were treated with simvastatin and twenty with placebo. However, because of the lowering of drug initiation levels by the American Diabetes Association, the trial was terminated early with 2 subjects reaching 2 years, 17 reaching 18 months, 36 reaching 1 year, and all 6 months. Simvastatin significantly reduced total cholesterol (mean on treatment 173.4 vs. 191.4, P=.020) and LDL cholesterol (mean on treatment 105.0 vs. 127.7, P<.001). Simvastatin therapy was associated with a slower rise in AER compared to placebo, though the result was not statistically significant (median rate of change/month 0.004 vs. 0.029). There was a trend towards slower progression of neuropathy as measured by vibratory threshold (median change at 1 year 0.03 simvastatin vs. 0.94, P=.07). There was no difference in change in ankle-brachial index, clinical neuropathy status, or retinopathy status. In conclusion, treatment with simvastatin may have a beneficial effect on early nephropathy and diabetic neuropathy, justifying a fully powered trial. However, this would be difficult under current treatment guidelines.

Lipid and blood pressure treatment goals for type 1 diabetes: 10-year incidence data from the Pittsburgh Epidemiology of Diabetes Complications Study.

OBJECTIVE: Subjects with type 1 diabetes are at high risk for many long-term complications, including early mortality and coronary artery disease (CAD). Few data are available on which to base goal levels for two major risk factors, namely blood pressure and lipid/lipoproteins. The objective of this study was to determine at which levels of LDL and HDL cholesterol, triglycerides, and blood pressure the relative risks of type 1 diabetic complications increase significantly. RESEARCH DESIGN AND METHODS: Observational prospective study of 589 patients with childhood-onset type 1 diabetes (<17 years) aged > or =18 years at baseline; 10-year incidence of mortality, CAD, lower-extremity arterial disease, proliferative retinopathy, distal symmetric polyneuropathy, and overt nephropathy. Relative risks were determined using traditional groupings of blood pressure and lipid/lipoproteins, measured at baseline, using the lowest groupings (<100 mg/dl [2.6 mmol/l] LDL cholesterol, <45 mg/dl [1.1 mmol/l] HDL cholesterol, <100 mg/dl [1.1 mmol/l] triglycerides, <110 mmHg systolic blood pressure, and <80 mmHg diastolic blood pressure) as reference. Adjustments for age, sex, and glycemic control were examined. RESULTS: Driven mainly by strong relationships (RR range 1.8-12.1) with mortality, CAD, and overt nephropathy, suggested goal levels are as follows: LDL cholesterol <100 mg/dl (2.6 mmol/l), HDL cholesterol >45 mg/dl (1.1 mmol/l), triglycerides <150 mg/dl (1.7 mmol/l), systolic blood pressure <120 mmHg, and diastolic blood pressure <80 mmHG: Age, sex, and glycemic control had little influence on these goals. CONCLUSIONS: Although observational in nature, these data strongly support the case for vigorous control of lipid levels and blood pressure in patients with type 1 diabetes.

Are predictors of coronary heart disease and lower-extremity arterial disease in type 1 diabetes the same? A prospective study.

In the Type 1 diabetes population, coronary heart disease (CHD) and lower-extremity arterial disease (LEAD) are the two common macrovascular complications leading to early mortality and morbidity. However, it is not clear if these two complications share the same risk factors. The Pittsburgh Epidemiology of Diabetes Complications (EDC) Study prospectively examined and compared the risk factors for LEAD and CHD (including CHD morbidity and mortality). EDC subjects (332 men and 325 women), all diagnosed at Children's Hospital of Pittsburgh between 1950 and 1980, were first examined at baseline (1986-1988), and then biennially, for diabetes complications and their risk factors. Data used in the current analysis were from the first 6 years of follow-up, 98% provided at least some follow-up data for these analyses. CHD was defined as the presence of angina (diagnosed by the EDC examining physician) or a history of confirmed myocardial infarction or CHD death. An ankle-to-arm ratio of less than 0.9 at rest was considered to be evidence of LEAD. Among 635 subjects without CHD at baseline, 57 developed CHD (1.69/100 person-years), and among 579 without LEAD at baseline, 70 developed LEAD (2.31/100 person-years). CHD incidence rate was slightly higher in males, while LEAD incidence rate was slightly higher in females. Compared to non-incident cases, subjects who developed either complication were older, had a longer diabetes duration, higher LDL and total cholesterol, and were more likely to be hypertensive. In multivariate analyses, hypertension, low HDL cholesterol level, high white cell count, depression, and nephropathy were the independent risk factors for CHD (including morbidity and mortality). For LEAD, higher HbA1 level, higher LDL cholesterol level and smoking were the important contributing factors. In conclusion, the risk factor patterns differ between the two vascular complications. Glycemic control does not predict CHD overall but does predict LEAD, while hypertension and inflammatory markers are more closely related to CHD than to LEAD.

Antibodies to oxidized LDL predict coronary artery disease in type 1 diabetes: a nested case-control study from the Pittsburgh Epidemiology of Diabetes Complications Study.

The pathogenesis of excess cardiovascular risk in type 1 diabetes is unclear. LDL cholesterol is only weakly predictive, and its concentration is often normal in type 1 diabetes. We therefore examined whether markers of LDL oxidation such as antibodies to oxidized LDL (Ab-OxLDL) and LDL-containing immune complexes, rather than LDL concentration, were predictive of coronary artery disease (CAD) in type 1 diabetes. This nested case-control study from an epidemiologic cohort study included 49 incident cases of myocardial infarction (MI), angina, or CAD death and 49 age-, sex-, and duration-matched control subjects. Ab-OxLDL was measured by enzyme immunoassay and the apolipoprotein B (ApoB) content of immune complexes (ApoB-IC) precipitated by polyethylene glycol by immunoelectrophoresis in baseline stored samples. Ab-OxLDL was inversely, and ApoB-IC directly, related to subsequent CAD. In multivariate analyses, Ab-OxLDL remained a significant independent predictor along with previously recognized predictors, hypertension and Beck depression score. In conclusion, oxidation of LDL and the immune response it elicits may play a role in predicting the development of CAD in type 1 diabetes and explain at least some of the enhanced CAD risk in type I diabetes.

Low-density lipoprotein particle size and coronary artery disease in a childhood-onset type 1 diabetes population.

Low-density lipoprotein (LDL) cholesterol has been widely recognized as a strong predictor of coronary artery disease (CAD). Recently, studies have examined the influence of LDL particle size (an integral part of the insulin resistance syndrome) on the development of CAD in the general population. This report examines the correlates of LDL particle size and its association with CAD in a type 1 diabetes population. We evaluated the interrelationships between LDL particle size and the presence of CAD in a cohort of childhood-onset type 1 diabetic subjects using the Pittsburgh Epidemiology of Diabetes Complications (EDC) study. LDL particle size was measured in 337 subjects (mean age, 35.6 years; mean diabetes duration, 27.2 years) who underwent the 8-year follow-up examination. LDL particle size was determined by vertical polyacrylamide gel (2% to 16%) electrophoresis. Subjects with the small dense LDL particle phenotype (<235.5 angstroms) [corrected] had a longer diabetes duration, higher cholesterol, triglyceride, LDL, fibrinogen, waist to hip ratio (WHR), and hemoglobin A1 (HbA1), and lower high-density lipoprotein (HDL) cholesterol compared with subjects with the large LDL particle phenotype (>257 angstroms) [corrected]. Males were also more likely to have an increased body mass index (BMI) and CAD, while females were more likely to have hypertension and a family history of type 2 diabetes (a potential marker of insulin resistance and CAD risk). The odds ratio ([OR] 95% confidence, interval [CI]) using logistic regression analysis for LDL particle size in association with CAD was 0.79 (0.60 to 1.04). Multivariate modeling indicated that the duration of type 1 diabetes, depressive symptomatology, and triglycerides were independently associated with the presence of CAD. We conclude that although small dense LDL particle size is associated with CAD in our type 1 diabetes population, its borderline association can largely be explained by the triglyceride concentration. However, as in the general population, LDL particle size is associated with many elements of the insulin resistance syndrome, including a family history of type 2 diabetes, and is likely an important element in the contribution of insulin resistance to the development of CAD in type 1 diabetes.

Antibodies to oxidized LDL and LDL-containing immune complexes as risk factors for coronary artery disease in diabetes mellitus.

Several groups have published results from clinical studies supporting the involvement of anti-modified LDL antibodies as risk factors for the initiation or progression of cardiovascular disease. However, the data published so far are judged inconclusive because of several contradictory observations concerning the correlation between clinical evidence of arteriosclerosis and the levels of antibodies to oxidized LDL (oxLDL Ab). We have previously reported that oxLDL Ab exist both in free form and as antigen-antibody complexes (LDL-IC) in patients with insulin-dependent diabetes mellitus (IDDM). The presence of LDL-IC in IDDM patients has important implications: it may interfere with the assay of oxLDL antibodies and the levels of LDL-IC may correlate better with the development of arteriosclerosis than the levels of free oxLDL antibodies. To clarify these questions baseline samples collected from 49 IDDM patients, who subsequently developed coronary artery disease (CAD) during an 8-year follow-up period, were compared to baseline samples from 49 age-, sex-, and duration-matched control IDDM subjects who remained free of clinical CAD during an identical follow-up period. The levels of free oxLDL antibody were significantly lower in the patients who developed CAD. The same patients had significantly higher concentrations of total cholesterol, apolipoprotein B, and IgA in immune complex-enriched polyethylene glycol (PEG) precipitates. The concentration of IgG was also higher in PEG precipitates from patients who developed CAD, but did not reach statistical significance. This indicates that patients who develop CAD had higher levels of circulating LDL-IC, a fact that could not be deduced from the measurement of free oxLDL antibody concentrations. A linear regression analysis of the correlation between the concentrations of total cholesterol in PEG precipitates, taken as a surrogate measurement of PEG-precipitated oxLDL-IC, and the concentration of free oxLDL antibody in serum showed a statistically significant negative correlation (r = -0.229, P = 0. 024). Our results support the conclusion that oxLDL-IC may be a risk factor for the development of macrovascular disease in IDDM patients. We also have demonstrated that circulating oxLDL-IC interfere with the assay of free oxLDL antibodies.

Urinary measurement of transforming growth factor-beta and type IV collagen as new markers of renal injury: application in diabetic nephropathy.

Urinary samples were concentrated rapidly and efficiently and were used to develop several protein assays that may be of value in monitoring individuals with progressive renal disorders. Transforming growth factor-beta1 (TGF-11) and retinol binding protein (RBP) were measured with modification of commercially available methods used to assay serum specimens; type 3 collagen (T3C) was measured with a new immunonephelometric assay. The precision characteristics of these assays are comparable with those reported for microalbuminuria. The clinical utility of measuring a panel of these markers was demonstrated in urine samples from 16 control subjects and from 46 individuals with insulin-dependent diabetes mellitus (IDDM) with various albumin excretion rates (AERs). TGF-beta1 and T3C were used as markers of cytokine expression and of the renal fibrogenic process, whereas RBP excretion served as a marker of tubular injury or dysfunction. Compared with controls, T3C excretion was significantly increased in 18 normoalbuminuric and further increased in 13 microalbuminuric (AER 20 < or = 200 microg/min) IDDM subjects. RBP excretion was increased in macroalbuminuric IDDM subjects (AER >200 microg/min, overt nephropathy). Significant correlations were also found between AER and RBP in all but macroalbuminuric individuals, whereas TGF-beta1 correlated with T3C excretion in controls and in normoalbuminuric diabetic subjects. Urinary RBP but not AER was an excellent predictor of diabetic nephropathy as defined by serum creatinine (P = 0.0001). This underscores the importance of an early tubulopathy in the subsequent development of glomerulopathy and overt nephropathy. The data suggest that longitudinal monitoring of a panel of urinary markers such as that used in the current study may better define their relevance in progressive glomerulosclerosis and may also provide greater insight into the mechanisms underlying such process.

Cardiovascular disease in insulin dependent diabetes mellitus: similar rates but different risk factors in the US compared with Europe.

BACKGROUND: Cardiovascular disease (CVD) in insulin dependent diabetes mellitus (IDDM) has been linked to renal disease. However, little is known concerning international variation in the correlations with hyperglycaemia and standard CVD risk factors. METHODS: A cross-sectional comparison was made of prevalence rates and risk factor associations in two large studies of IDDM subjects: the Pittsburgh Epidemiology of Diabetes Complications Study (EDC) and the EURODIAB IDDM Complications Study from 31 centres in Europe. Subgroups of each were chosen to be comparable by age and duration of diabetes. The EDC population comprises 286 men (mean duration 20.1 years) and 281 women (mean duration 19.9 years); EURODIAB 608 men (mean duration 18.1 years) and 607 women (mean duration 18.9 years). The mean age of both populations was 28 years. Cardiovascular disease was defined by a past medical history of myocardial infarction, angina, and/or the Minnesota ECG codes (1.1-1.3, 4.1-4.3, 5.1-5.3, 7.1). RESULTS: Overall prevalence of CVD was similar in the two populations (i.e. men 8.6% versus 8.0%, women 7.4% versus 8.5%, EURODIAB versus EDC respectively), although EDC women had a higher prevalence of angina (3.9% versus 0.5%, P < 0.001). Multivariate modelling suggests that glycaemic control (HbA1c) is not related to CVD in men. Age and high density lipoprotein cholesterol predict CVD in EURODIAB, while triglycerides and hypertension predict CVD in EDC. For women in both populations, age and hypertension (or renal disease) are independent predictors. HbA1c is also an independent predictor-inversely in EURODIAB women (P < 0.008) and positively in EDC women (P = 0.03). Renal disease was more strongly linked to CVD in EDC than in EURODIAB. CONCLUSIONS: Despite a similar prevalence of CVD, risk factor associations appear to differ in the two study populations. Glycaemic control (HbA1c) does not show a consistent or strong relationship to CVD.

Driving patterns and medical conditions in older women.

OBJECTIVES: To describe driving patterns (e.g., driving frequency) in older women drivers and to evaluate the impact of medical conditions and comorbidity on driving patterns. DESIGN: Cross-sectional examination of the association between medical conditions and driving patterns. SETTING: Population-based cohort from the Pittsburgh Center of the Study of Osteoporotic Fractures (SOF). PARTICIPANTS: A total of 1768 women aged 71 years or older. MAIN MEASUREMENTS: Driving information was obtained through a driving questionnaire, including driving status, weekly mileage, longest trip in the past year, etc. Data for demographics, lifestyle behavior, and medical conditions were collected through the SOF study. RESULTS: Among the participants, 1103 (62.3%) were current drivers, 337 (19.1%) had stopped driving, and 329 (18.6%) had never driven in their lifetime. The proportion reporting driving cessation and decline in driving amount increased with age. The prevalence of most medical conditions was higher among former drivers than in current or never drivers. Even after controlling for age and other demographic variables, fractures, heart disease, diabetes, self-reported poor vision or hearing, as well as comorbidity were found to be associated independently with decreased driving amount, including driving cessation, decline in mileage, and avoiding long trips. CONCLUSION: Both individual medical conditions and comorbidity influence driving patterns in older drivers. Because it is common for older people to have several medical conditions simultaneously, comorbidity might be a more comprehensive measure of medical impact on driving.

Cumulative glycemic exposure and microvascular complications in insulin-dependent diabetes mellitus. The glycemic threshold revisited.

BACKGROUND: The development of microvascular insulin-dependent diabetes mellitus (IDDM) complications has been shown to be related to both duration of diabetes and the degree of glycemic exposure. However, controversy exists as to whether there is a threshold of glycemic exposure, below which there is minimal risk. Furthermore, there are few data describing the relationship of total glycemic exposure (duration x degree) to complications rates-a potentially useful research and clinical tool. OBJECTIVES: To determine a cumulative glycemic exposure variable that combines the effect of both degree and duration of hyperglycemia and to evaluate this variable in terms of its relation to microvascular complications. The association between cumulative glycemic exposure and complication risk was also examined to evaluate whether there was a threshold effect. METHODS: A total of 353 patients with IDDM who had completed the first 6 years of follow-up in the Pittsburgh Epidemiology of Diabetes Complications Study were included in this analysis. These subjects had a mean age of 27.9 years, and the mean duration of the disease was 19.4 years. Subjects were examined at baseline (cycle 1) and then biennially (cycle 2, cycle 3, and cycle 4) for diabetes complications. Total glycosylated hemoglobin (HbA1) was measured at each cycle. A cumulative glycemic exposure variable, named A1months, was calculated by multiplying the number of HbA1 units above normal at each cycle by the number of months between the midpoints of the preceding and succeeding cycle intervals. RESULTS: The mean number of A1months experienced at the time of diagnosis of proliferative retinopathy (914), microalbuminuria (952), overt nephropathy (1043), and distal symmetrical polyneuropathy (1043) did not vary by duration of diabetes. Thus, approximately 1000 A1months were needed (on average) for the advanced complications to develop. Although the risk for developing proliferative retinopathy rose gradually as A1months increased, a more abrupt increase in the risk was seen (again at approximately 1000 A1months) for microalbuminuria (odds ratio, 6.9; 95% confidence interval, 2.5-19.1), overt nephropathy (odds ratio, 6.5; 95% confidence interval, 2.0-21.7), and distal symmetrical polyneuropathy (odds ratio, 6.5; 95% confidence interval, 2.4-17.8). Nonetheless, complications developed in the majority of cases at glycemic exposures below 1000 A1months. The cumulative glycemic exposure variable A1months does not predict complications any better than its component variables (duration and HbA1). Furthermore, formal statistical testing failed to show a definitive threshold for any complication. CONCLUSIONS: Although A1months does not enhance prediction of complications, it may be a useful summary measure of glycemic exposure for both patients and physicians. However, although subjects with 1000 A1months or more appear to be at increased risk of developing most microvascular complications, because the majority of complications arise in subjects with less than this exposure, this threshold value should only be considered a minimal goal. For example, our data suggest that for most microvascular complications to develop, it would take, on average, 83 years with an HbA1 unit at 1% above normal, 42 years at 2% above normal, 28 years at 3% above normal. 21 years at 4% above normal, and 18 years at 5% above normal.

Hypertension as a risk factor for diabetic neuropathy: a prospective study.

The pathogeneses of diabetic neuropathy is still unclear. This study prospectively investigated the risk factors for distal symmetrical polyneuropathy (DSP) in a cohort of childhood-onset IDDM patients. Subjects from the Epidemiology of Diabetes Complications (EDC) Study were clinically examined at baseline and then biennially. DSP was diagnosed by a combination of clinical criteria, symptoms and signs (Diabetes Control and Complications Trial [DCCT] exam), and quantitative sensory threshold (QST). Among the 463 (70.4%) subjects who were free of DSP at baseline, 453 (97.8%) participated in at least one biennial reexamination during the first 6 years of follow-up and were included in the current analysis. A total of 68 (15.0%) subjects developed DSP in 6 years, giving a cumulative probability of 0.29. The Cox proportional hazards model shows that longer IDDM duration, hypertension, poor glycemic control, height, and smoking were all independent predictors of the incidence of DSP (all P < 0.0001, except for smoking for which P = 0.03). Hypertension showed the greatest impact on the development of DSP for individuals with either short or long IDDM duration. This study confirms some risk factors for DSP found in cross-sectional studies and suggests a strong relationship between hypertension and DSP. The results indicate that in addition to good glycemic control, avoidance of smoking and good blood pressure control may be helpful in preventing or delaying the onset of DSP in IDDM patients.

Bone mineral density and aortic calcification: the Study of Osteoporotic Fractures.

OBJECTIVE: To investigate the relationship between bone mineral density (in the axial and appendicular skeleton) and calcification of the aorta. DESIGN: Cross-sectional study. SETTING: Community-based study. PARTICIPANTS: A total of 2051 women aged 65 years and older enrolled in the Study of Osteoporotic Fractures. MEASUREMENTS: Bone mineral density (BMD) at the hip, spine, calcaneus, proximal and distal radius; calcification of the aorta (AC); demographic and lifestyle variables; dietary history; functional status; blood pressure; anthropomorphic measures. RESULTS: The prevalence of AC increased with age, ranging from 60% at ages 65 to 69 years to 96% at 85 years and older. BMD in women with calcified arterial plaques was lower (P < .001) when compared with those with no plaques, at all sites measured except the lumbar spine. After adjustment for age, BMD at the hip, spine and calcaneus was not associated with the presence of plaques; only a weak association between BMD and AC remained at the distal and proximal radius. The independent correlates of AC were age, smoking status, systolic blood pressure, coffee drinking, central obesity and a history of diabetes or stroke; current estrogen use was protective. CONCLUSIONS: The results of this study indicate that osteopenia and the deposition of calcific plaques in the wall of the aorta are independent processes that occur as women age. They are probably not causally linked.

The changing course of diabetic nephropathy: low-density lipoprotein cholesterol and blood pressure correlate with regression of proteinuria.

Diabetic nephropathy (DN) as manifested by persistent and clinically evident proteinuria, has long been considered an irreversible process that predicts a rapid decline in renal function. The observation of reversal of DN in several individuals enrolled in a prospective study of the natural course of diabetes complications challenged this view and led to the current investigation into the correlates of such regression of proteinuria. DN was defined as a median albumin excretion rate (AER) over 200 microg/min in two or three urine collections obtained at baseline, and again at 2 and 4 years of follow-up. Among 658 individuals with childhood-onset insulin-dependent diabetes mellitus (IDDM), 146 had DN at baseline. Nine subsequently died without renal failure, and 13 were lost to follow-up. Of the 124 subjects with at least survey follow-up data, 32 (24%) developed renal failure, and 78 of the remaining 92 provided full quantitative data. AER decreased by > or = 10-fold into the microalbuminuric (20 to 200 microg/min) or normal range (<20 microg/min) in 7 of these individuals and are called "regressors of proteinuria." Compared with the remaining 71 subjects, the strongest correlate of regression of proteinuria was an improvement in fasting plasma low-density lipoprotein cholesterol (LDL-C) in the 7 regressors (P < 0.008). Improved glycemic control was not a significant predictor of improved AER. Five of the 7 individuals with improved AER had a baseline median AER below 500 microg/min. When the 7 regressors of proteinuria were combined with an additional 38 individuals who also experienced smaller decreases in median AER, such improvement was associated with a more favorable systolic (or diastolic) blood pressure (BP) change (P < 0.01), and a decrease in plasma LDL-C level (P = 0.01). These data suggest that proteinuria in DN may substantially regress in approximately 6% and improve in at least 34% of individuals with IDDM over a 4-year period, often in association with a decrease in plasma LDL-C concentration or stabilization or improvement in BP. Furthermore, the data suggest that the nonreversibility threshold for diabetic nephropathy may be higher (500 mg/min) than previously reported (200 microg/min).