Abstract
Total serum homocysteine (tHcy) has been shown to predict de novo and recurrent cardiovascular events in many studies. However, results in diabetic populations with minimal nephropathy are mixed. The independent relationship between tHcy and arteriosclerotic outcomes and congestive heart failure (CHF) events in a population with high cardiovascular risk and diabetic nephropathy was examined. A total of 1575 individuals were enrolled in the international Irbesartan Diabetic Nephropathy Trial (IDNT) and followed for 2.6 yr. All participants had baseline diabetic nephropathy, overt proteinuria, and hypertension and were recruited between 1996 and 1999. A total of 492 total arteriosclerotic outcomes (primary outcome) and 317 CHF events (secondary outcome) were tallied. Established cardiovascular risk factors were highly prevalent, as were high tHcy levels (quintiles [μM]: first 4.5 to 11; second >11 to 13; third >13 to 15; fourth >15 to 19; fifth >19). No association between tHcy and arteriosclerotic outcomes was observed in a univariate model or after adjustment for study randomization and established cardiovascular risk factors. The strongest outcome predictor was the presence of baseline cardiovascular disease, followed by an inverse relationship to diastolic BP. The significant univariate association between tHcy and CHF events disappeared when serum creatinine alone was added to the model. These findings question the utility of tHcy in predicting de novo or recurrent cardiovascular events in individuals with diabetic nephropathy. Further studies are needed to confirm whether these negative results apply to other populations with heavy cardiovascular risk burdens. Previous positive findings can potentially be explained through tHcy’s role as a sensitive surrogate marker for kidney disease, itself a recognized cardiovascular risk factor.
The sulfur amino acid homocysteine has been investigated intensively in recent years as a possible risk factor for arteriosclerotic outcomes. Total serum homocysteine (tHcy) levels have been found in many but not all prospective studies to be positively associated with de novo and recurrent cardiovascular events in the general population independent of traditional risk factors (1,2).
Only a handful of observational studies have examined the relationship between tHcy and cardiovascular outcomes in individuals with type 2 diabetes, and results are mixed (3–6). The largest and most recent such report concluded that tHcy was, in fact, a strong and independent predictor of cardiovascular disease (CVD) (3). However, these studies included few individuals with frank diabetic nephropathy, a subgroup at extremely high risk for cardiovascular morbidity and death (7,8).
Therefore, we endeavored to study the hypothesis that baseline concentrations of tHcy predict development of incident or recurrent arteriosclerotic outcomes among high-risk individuals with type 2 diabetic nephropathy, independent of established cardiovascular risk factors. All study participants were enrolled in the multicenter, international Irbesartan Diabetic Nephropathy Trial (IDNT). A secondary analysis was also performed to confirm a recent report in the population-based Framingham cohort linking baseline tHcy levels with incident congestive heart failure (CHF) (9).
Materials and Methods
Population
The IDNT was an investigator-initiated, prospective, three-arm, randomized, double-masked study in 1715 patients with type 2 diabetic nephropathy that compared treatment with irbesartan, amlodipine, or placebo. Randomization occurred between March 21, 1996, and February 25, 1999, in >200 clinical centers in North and South America, Europe, and Asia. Details of the baseline patient characteristics and renal and cardiovascular outcomes of this study have been published (8,10,11). The institutional review board or ethics committee of each center approved the protocol, and all patients gave written informed consent after reviewing a written summary of the study plan. In brief, eligibility criteria included the following: Age between 30 and 70 yr, type 2 diabetes and overt nephropathy as manifested by a 24-h urinary protein excretion rate ≥900 mg, a serum creatinine between 1.0 and 3.0 mg/dl in women or 1.2 and 3.0 mg/dl in men, and baseline BP >135/85 mmHg or treatment with antihypertensive agents. The primary outcome of the IDNT was time to occurrence of a composite renal outcome of doubling of entry serum creatinine, ESRD, or all-cause cardiovascular mortality. Our analysis was an ancillary study of the IDNT and included the 92% of participants (n = 1575) who had serum samples available for determination of baseline tHcy.
Measurements and Laboratory Tests
Body mass index (BMI) was defined as weight (kg) divided by height (m)2—i.e., BMI=W/(H2). BP were taken in the seated position using standardized sphygmomanometers. Baseline samples were drawn between March 21, 1996, and February 25, 1999, and stored at −70°C. tHcy levels were measured in the Vitamin Metabolism and Aging Laboratory (USDA-Human Nutrition Research Center, Boston, MA) by HPLC with fluorescence detection (12). Serum creatinine (using the modified Jaffe method), hemoglobin A1c (HbA1c), lipid levels, and 24-h urinary protein levels were measured in one of four IDNT regional laboratories using standard automated clinical chemistry techniques. Proteinuria measurements were based on the urinary protein/creatinine ratio derived from the 24-h urine collection.
Definition of Cardiovascular Outcomes
The total arteriosclerotic outcome, the primary study outcome in this analysis, included all of the following (10): Cardiovascular mortality, myocardial infarction documented by clinical data such as enzyme and electrocardiogram changes, stroke documented by brain imaging or clinical deficits persisting >24 h and requiring hospitalization, lower extremity amputations, and unplanned (at time of randomization) cardiac or peripheral revascularization procedures. All outcomes were ultimately adjudicated by an IDNT Outcomes Committee (8). A separate secondary analysis defined CHF outcomes as CHF requiring hospitalization or, alternatively, treatment with renin-angiotensin-aldosterone blockade.
Statistical Analyses
Because the cardiovascular outcomes that we modeled could occur more than once, we determined relative risks using the Anderson-Gill formulation of the proportional hazards model (13) in which patients are considered at risk for the first event from randomization to the time of the first event, at risk for the second event from the day after the first event to the time of the second event, and so forth, permitting use of all of the data. In accordance with the method of Lee et al. (14), we used a robust variance estimate that takes into account the possibility of correlation of risk for multiple events within each patient.
Models were developed that included the study randomization group and the following traditional baseline cardiovascular characteristics: Age, gender, smoking history, history of previous cardiovascular disease, BMI, seated systolic and diastolic BP, HbA1c, the inverse of serum creatinine, and the total/HDL cholesterol ratio. Analyses of time to first arteriosclerotic and CHF events gave results that were essentially the same as those for time to total events. tHcy was transformed logarithmically for the analyses. Quintiles of tHcy were analyzed as ordered factors and were the following (μM): first, 4.5 to 11; second, >11 to 13; third, >13 to 15; fourth, >15 to 19; fifth, >19. The P value of the first order (linear) component of the factor was used to test for a linear trend across the quintiles. Twenty-four-hour urine protein was logarithmically transformed when measured at baseline. Statistical analyses were performed and graphics were generated using S-Plus for Windows Version 6.2 (Insightful Corp., Seattle, WA).
Results
Of the 1715 patients who were originally enrolled in the IDNT, 1575 had samples available for baseline tHcy measurements. Aside from minor clinical but statistically significant differences in BMI (31.0 versus 29.3; P = 0.001) and protein/creatinine ratio (3.27 versus 2.53; P = 0.02), the relevant characteristics were equivalent between groups. Traditional cardiovascular risk factors were highly prevalent in this population (Table 1). Nearly half the participants had baseline CVD, and most had a history of smoking, obesity, and dyslipidemia. In addition, the majority of individuals had abnormally elevated tHcy levels (>12 μM) as defined in societies, such as the United States, that routinely supplement foodstuffs with folate (15). Total arteriosclerotic outcomes were very common in this high-risk population as well, with 492 events occurring in 364 patients over a 2.6-yr mean follow-up period. There were 317 secondary outcome CHF events in 206 patients during this period.
Baseline characteristicsa
Table 2 confirms the well-recognized strong inverse relationship between tHcy and the GFR, as reflected by serum creatinine (16). In addition, much weaker positive and negative correlations, respectively, were found with age and HbA1c. Baseline tHcy levels were higher in individuals with pre-existing CVD (14.7 versus 14.0; P = 0.002) but were not statistically different in men versus women or in smokers versus nonsmokers.
Correlations between tHcy and continuous baseline characteristics
Univariate analyses revealed no significant relationship between tHcy and total arteriosclerotic outcomes (Table 3). As seen in Table 3 and Figure 1, the lack of association remained after adjustment for study randomization and established cardiovascular risk factors. The most powerful predictor of outcomes in the fully adjusted model was the presence of baseline CVD (relative [95% confidence interval] 1.77 [1.41 to 2.21]; P < 0.0001), followed by an inverse relationship to diastolic BP (per 10-mmHg increase; 0.85 [0.77 to 0.94]; P = 0.0015).
tHcy and risk for total arteriosclerotic events adjusted for randomization group and the following recognized cardiovascular risk factors: Age, gender, systolic BP, diastolic BP, body mass index (BMI), renal function, cardiovascular history, smoking, lipids, and hemoglobin A1c (HbA1c).
tHcy and cardiovascular event risk
It is interesting that the secondary analysis found a significant univariate association between tHcy and CHF events, as seen in Table 3. However, significance was lost when serum creatinine alone was added to the model (data not shown, although full model in Table 3). Quintiles of tHcy and CHF outcomes are seen in Figure 2. Similar to the primary outcome, a history of CVD was the most important predictor of CHF events (2.09 [1.56 to 2.80]). The only significant predictor in the multivariate CHF model was inverse serum creatinine. For example, with an increase in serum creatinine from 1 to 2 mg/dl, the relative risk was 2.02 (1.24 to 3.28).
tHcy and risk for congestive heart failure events adjusted for randomization group and the following recognized cardiovascular risk factors: Age, gender, systolic BP, diastolic BP, BMI, renal function, cardiovascular history, smoking, lipids, and HbA1c.
Discussion
This is the first study to determine the predictive value of the putative atherothrombotic risk factor tHcy on CVD outcomes in the high-risk population of individuals with overt diabetic nephropathy. This large, multicenter trial found no association between baseline tHcy levels and subsequent arteriosclerotic events independent of traditional cardiovascular risk factors. Because diabetes is a growing societal threat and individuals with diabetic nephropathy compose an important subgroup of the chronically ill in the United States (including nearly half of people who ultimately require dialysis therapy [17]), these findings have important clinical implications.
Most smaller prospective cohort studies in populations with type 2 diabetes have found tHcy to be marginally predictive of future cardiovascular events (4,6). In contrast, the most recent and by far the largest such study (n = 830) found a much stronger predictive effect (relative cardiovascular mortality risk for tHcy >15 μM [versus <15 μM] was 2.94 [1.72 to 5.01]) (3). Our study distinguishes itself from these other trials in a number of ways. First, as an international, multicenter cohort larger than all previous studies combined, it has arguably more external validity. Second, our study population is composed exclusively of individuals with overt diabetic nephropathy. This is in stark contrast to the other studies, which included few such patients. The existence of frank nephropathy reflects a more chronic or severe exposure to type 2 diabetes, as well as perhaps to other recognized cardiovascular risk factors, such as obesity, hypertension, and smoking. One plausible explanation as to why tHcy was not a useful predictive biomarker in our population is that the heavy “traditional” CVD risk burden accounted for the overwhelming portion of cardiovascular events. This raises the important question of whether tHcy can predict future CVD in populations with similar elevated baseline cardiovascular risk. Of note, any association between tHcy and cardiovascular outcomes may be strongest in the short term, so the relatively short follow-up period should not necessarily explain the study’s negative findings (18). In addition, the IDNT adjudication process in confirming CVD end points was rigorous, so misclassifications were unlikely to occur. Another plausible explanation for our negative findings is that tHcy is simply not a true atherothrombotic risk factor. tHcy has been positively and independently associated with increased cardiovascular risk in many case-control, retrospective, and nonrandomized prospective trials (1,2,19,20). An association has also been observed in small hemodialysis cohorts, although the tHcy levels associated with increased risk were high enough to suggest possible confounding by an underlying B-vitamin deficiency (21,22). Possible disease mechanisms include promotion of endothelial cell injury, upregulation of oxidative or pro-coagulant pathways, and enhanced smooth muscle proliferation (23). In contrast, recent randomized, controlled trials have not found that lowering tHcy offers cardiovascular protection (24–26). Our results are most consistent with the negative findings. However, our study is limited by its nonrandomized study design. Results from the ongoing randomized, placebo-controlled National Institutes of Health cooperative FAVORIT (Folic Acid for Vascular Outcome Reduction in Transplantation) trial will help to resolve this issue more definitively. Although we are not certain why a higher diastolic pressure was associated with fewer cardiovascular events, a lower diastolic BP may simply be a marker of cardiovascular dysfunction (and subsequent events), as it seems to be in chronic hemodialysis patients (27).
The univariate association between tHcy and incident CHF is an interesting finding that underscores the great importance of accounting for kidney function, itself a cardiovascular risk factor (28), when studying the link between hyperhomocysteinemia and CVD. The strong inverse relationship between tHcy and the GFR is very well established (16,29). That the association between tHcy and CHF loses significance when serum creatinine is added to the model suggests that a more biologically plausible relationship is between CHF and kidney function, rather than with tHcy (because tHcy does not seem to induce progression of kidney disease independently [30]). After all, a reduction in glomerular filtration results both in higher tHcy levels and perturbations in salt/water homeostasis that could very well predispose to CHF in susceptible individuals. In this circumstance, tHcy would simply act as a surrogate marker for kidney function—in effect, an “expensive creatinine” (31). This theory is supported by recent randomized studies that found a positive association between tHcy and adverse CVD outcomes despite that reductions in tHcy levels have no salutary effect (25,26). Serum creatinine is a relatively crude marker for glomerular filtration, particularly in the “normal” range (32), and simply presuming that kidney function is normal because serum creatinine levels are not elevated (3,9) does not exclude the real possibility of confounding from reduced glomerular filtration.
In conclusion, tHcy does not predict arteriosclerotic events in patients with type 2 diabetes and nephropathy independent of traditional cardiovascular risk factors. These findings question the utility of tHcy in predicting de novo or recurrent cardiovascular events in this and possibly other populations with high pre-existing cardiovascular risk loads. Previous positive findings can potentially be explained through tHcy’s role as a sensitive surrogate marker for kidney disease, itself a recognized cardiovascular risk factor.
Acknowledgments
Support was provided by the United States Department of Agriculture, under agreement 581950-9-001. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of the United States Department of Agriculture. The Irbesartan Diabetic Nephropathy Trial was supported by grants from Bristol-Myers Squibb and Sanofi-Synthelabo. The main funding for this research effort was provided by RO1 HL 67695-02, awarded to A.G.B.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
Access to UpToDate on-line is available for additional clinical information at http://www.jasn.org/
- © 2005 American Society of Nephrology