The Modification of Diet in Renal Disease (MDRD) Study compared the rates of decline in glomerular filtration rate in 840 patients with a diverse array of renal diseases who were randomly assigned to either a usual or a low blood pressure goal [10, 11]. In study A (baseline glomerular filtration rate, 25 to 55 mL/min·1.73 m²), the intent-to-treat analysis included all patients and showed that the mean decline in glomerular filtration rate was faster in the first 4 months of follow-up and slower thereafter in the low than in the usual blood pressure group [10]. These results suggest that the low blood pressure goal has a long-term benefit. However, the duration of follow-up (0 to 3.7 years) was insufficient to show a difference between the two blood pressure groups in the decline of glomerular filtration rate at the end of the study. In study B (baseline glomerular filtration rate, 13 to 24 mL/min·1.73 m²), the decline in glomerular filtration rate was linear and did not differ significantly between the two blood pressure groups.

In studies A and B, subgroup analyses showed that baseline proteinuria was a strong predictor of subsequent decline in glomerular filtration rate and that assignment to the low blood pressure goal produced a significantly greater benefit on the rate of decline in glomerular filtration rate in patients with higher baseline proteinuria [10, 12].

We examined the relation of prescribed and achieved blood pressure with decline in glomerular filtration rate, placing particular emphasis on how this relation is affected by proteinuria at baseline. We also evaluated the effect of blood pressure on changes in proteinuria and the relation between changes in proteinuria during follow-up and subsequent decline in glomerular filtration rate.

Methods

Top Methods Results Discussion Author & Article Info References

Study Design

The MDRD Study consisted of two randomized clinical trials done in a total of 840 patients who had chronic renal diseases of diverse cause [10, 11]. The study protocol was approved by the review boards of the 15 participating institutions. Eligibility criteria included the following: age of 18 to 70 years; serum creatinine level of 1.2 to 7.0 mg/dL for women and 1.4 to 7.0 mg/dL for men or a creatinine clearance less than 70 mL/min·1.73 m²; and mean arterial pressure of 125 mm Hg or less [13]. Patients were excluded if they had diabetes requiring insulin, proteinuria of 10 g/d or more, or body weight less than 80% or more than 160% of standard body weight [14]. After a 3-month baseline period, 585 patients who had a glomerular filtration rate of 25 to 55 mL/min·1.73 m², dietary protein intake of more than 0.9 g/kg·d, and mean arterial pressure of 125 mm Hg or less were entered into study A. Two hundred fifty-five patients who had a glomerular filtration rate of 13 to 24 mL/min·1.73 m² and a mean arterial pressure of 125 mm Hg or less were entered into study B.

Patients in both studies were randomly assigned to either a usual or a low blood pressure goal. The usual blood pressure goal was a mean arterial pressure of 107 mm Hg or less (for patients less than equals 60 years of age) or 113 mm Hg or less (for patients more than equals 61 years of age). The low blood pressure goal was 15 mm Hg lower: It was a mean arterial pressure of 92 mm Hg or less (for patients less than equals 60 years of age) or 98 mm Hg or less (for patients more than equals 61 years of age).

Patients in study A were also randomly assigned to receive either a usual protein (1.3 g/kg·d) or low-protein (0.58 g/kg·d) diet. Patients in study B were assigned to receive either the low-protein or a very-low-protein (0.28 g/kg·d) diet supplemented with a mixture of ketoacids and amino acids (0.28 g/kg·d; Ross Laboratories, Columbus. Ohio). Results of the dietary interventions have been reported elsewhere [10, 15].

Randomization, done at the Data Coordinating Center, was stratified according to clinical center and average mean arterial pressure at baseline (both studies A and B) and according to the rate of change in serum creatinine levels before study entry (study A only). Additional details about the baseline period have been reported elsewhere [13, 16].

The mean duration of follow-up was 2.2 years. In study A, 553 patients had glomerular filtration rate measurements extending to at least 1 year of follow-up; 381 patients had glomerular filtration rate measurements extending to at least 2 years of follow-up; and 143 patients had glomerular filtration rate measurements extending to at least 3 years of follow-up. In study B, 219 patients had glomerular filtration rate measurements extending to at least 1 year of follow-up; 137 patients had glomerular filtration rate measurements extending to at least 2 years of follow-up; and 62 patients had glomerular filtration rate measurements extending to at least 3 years of follow-up.

Antihypertensive Regimens

Both nonpharmacologic and pharmacologic interventions were implemented. During the baseline period, antihypertensive regimens were prescribed to achieve the usual blood pressure goal. After randomization, the regimens were modified to achieve either the low or the usual blood pressure goal. Nonpharmacologic therapy included recommendations for regular exercise and for reductions in body weight and intake of alcohol and sodium. Pharmacologic therapy was based on the stepped-care approach defined in the 1988 Report of the Joint National Committee [17]. Use of all antihypertensive drugs was allowed, but angiotensin-converting enzyme inhibitors, with or without a diuretic, were encouraged as the agents of first choice. Calcium channel blockers, with or without a diuretic, were encouraged as the agents of second choice. Blood pressure was measured monthly using a standardized Hawksley random zero sphygmomanometer (Lancing, United Kingdom). Blood pressure regimens were modified monthly; they were modified more often as was necessary to achieve the blood pressure goals.

Measurements and Definitions of Variables

The primary outcome variable was decline in glomerular filtration rate. This rate was measured as the renal clearance of ¹²⁵I-iothalamate [18, 19] at the beginning and end of the baseline period and at 2 months, at 4 months, and at every 4 months thereafter during follow-up. Protein intake was monitored by monthly 24-hour urinary urea nitrogen excretion tests [20]. Blood for hematologic and serum tests was obtained at the beginning and end of the baseline period and every 2 months during follow-up.

Renal diagnoses were made using medical records and review of available historical information [13]. Each patient was classified as having one of nine renal diagnoses. If patients had baseline proteinuria of less than 3.0 g/d; had presumptive (rather than established) diagnoses of hypertensive nephrosclerosis, tubulointerstitial diseases, or other diseases; and had not had renal biopsy, they were placed in the "other or unknown" category. Patients with proteinuria of more than 3.0 g/d who had not had renal biopsy and who had the presumptive diagnoses listed above were placed in the glomerular diseases category.

Mean baseline blood pressure was defined as the average of the two mean arterial pressure measurements made at the end of the first and second months of the baseline period. Mean follow-up blood pressure was defined as the average of all mean arterial pressure measurements obtained at nonglomerular filtration rate visits beginning at the third monthly follow-up visit (mean arterial pressure during glomerular filtration rate visits was consistently higher than during other visits). Proteinuria was measured monthly during the baseline period and every 2 months during the follow-up period in 24-hour urine samples by using the trichloroacetic acid (Ponceau) technique [21]. Baseline proteinuria was the average of four measurements. In some analyses, patients were classified into subgroups according to whether their baseline proteinuria was 0 to 0.25 g/d; 0.25 to 1.0 g/d; 1.0 to 3.0 g/d; or 3.0 g/d or more.

For the analyses reported here, patients were classified as taking a class of antihypertensive agents if they reported taking agents of that class for more than 50% of follow-up visits. This definition was selected because most patients reported taking agents of any particular class for either less than 25% or more than 75% of follow-up visits.

Statistical Analysis

Hypothesis tests were considered statistically significant if P < equals 0.05, two-sided. No adjustments were made for multiple comparisons. To eliminate positive skewness, proteinuria was log transformed, and changes in proteinuria were expressed as the percentage change from baseline. Baseline characteristics were compared between groups using t-tests, analysis of variance, or chi-square tests, as appropriate.

Comparisons of Randomized Groups

Mean protein intake was similar in the usual and low blood pressure groups in both study A and study B. The effect of the blood pressure intervention was similar in the usual and low-protein diet groups in study A and in the low-protein and very-low-protein diet groups in study B. The effect of the dietary intervention was not influenced by baseline proteinuria. Consequently, comparisons of the blood pressure groups included all patients, regardless of dietary assignment.

In study A, the decline in glomerular filtration rate in the blood pressure groups was compared using a 2-slope model in which each patient was assumed to have an initial rate of decline in glomerular filtration rate during the first 4 months of follow-up and a possibly different slope thereafter [10]. We used a mixed-effects model that allowed different rates of decline for each patient [22]. In study B, we compared the decline in glomerular filtration rate in the blood pressure groups by using a 1-slope informative censoring model with log-normally distributed times to renal failure stop points [23]. These analyses included terms for the randomization stratification factors.

To assess the uniformity of the dietary and blood pressure interventions, the 1-slope and 2-slope models were used to compare subgroups that were defined by five factors: age, sex, renal diagnosis, baseline glomerular filtration rate, and baseline proteinuria [10]. The relation between baseline proteinuria and the effect of the blood pressure intervention is illustrated here by plots of the mean glomerular filtration rate slope for each blood pressure group on the basis of separate analyses in the four baseline proteinuria subgroups specified above. These plots are similar to (but not exactly the same as) plots presented previously [10], in which the mean glomerular filtration rate slope within each subgroup was estimated using joint analyses of all patients in studies A or B. The use of separate analyses has the advantage of accounting for a higher interpatient variability of glomerular filtration rate slopes in patients with higher baseline proteinuria [12].

To account for different lengths of follow-up, plots of the pattern of change in glomerular filtration rate and in proteinuria were based on spline models with break points at 2 months, at 4 months, and at every 4 months thereafter. Estimated mean changes to later follow-up times took into account early measurements from patients with shorter follow-up times. Because many patients in study B had short follow-up periods because of renal failure or death, plots of follow-up glomerular filtration rate and urine protein level are given only for patients in study A.

Correlational Analyses

The 2-slope mixed model (study A) or the 1-slope informative censoring model (study B) was used to relate decline in glomerular filtration rate to mean follow-up blood pressure after controlling for the baseline covariates (see below) and mean follow-up protein intake. Multiple regression models were used to relate mean follow-up blood pressure to changes in proteinuria and to relate changes in proteinuria during the first 4 months of follow-up to subsequent decline in glomerular filtration rate. These latter analyses were done only in patients who had been followed for at least 8 months.

Because these analyses are not based on the direct comparison of randomized groups, it was necessary to control for possible confounding variables that might have been jointly related to blood pressure level, proteinuria, and rate of decline of glomerular filtration rate. We initially screened 41 potential baseline predictors of the rate of decline of glomerular filtration rate by using a backward selection procedure [12]. Six factors (log proteinuria, diagnosis of polycystic kidney disease, mean arterial pressure, black race, serum transferrin level, and serum high-density lipoprotein [HDL] cholesterol level) were identified as independent predictors of the rate of loss of glomerular filtration in either study A or study B. To these covariates, we also added sex, age, history of hypertension, mean baseline serum cholesterol level, and mean baseline protein intake, because of the clinical relevance of these factors to the relations under investigation. The resulting 11 baseline variables are hereafter referred to as the relevant baseline covariates. We calculated the percentage of variance of glomerular filtration rate slopes accounted for by this and other sets of variables from the decrease in the between-patient variance component of the glomerular filtration rate slopes that resulted from adding the set of variables to the model. The percentages of variance in glomerular filtration rate slopes over 3 years accounted for by the 11 relevant baseline covariates and mean follow-up protein intake was 34.8% in study A and 35.5% in study B.

To compare the strength of the association between follow-up blood pressure and glomerular filtration rate slope between different blood pressure levels, we tested the significance of a quadratic term in mean follow-up mean arterial pressure after controlling for a linear term in mean follow-up mean arterial pressure and the baseline covariates. Given a significant quadratic effect, linear spline models with break points at 92 and 98 mm Hg in study A and at 90, 95, 100, and 105 mm Hg in study B were used to illustrate the relation between different levels of blood pressure and glomerular filtration rate slope.

Results

Top Methods Results Discussion Author & Article Info References

Baseline Characteristics

Most patients had proteinuria of less than 1.0 g/d (Table 1). This reflects, in part, the many patients with polycystic kidney disease (who generally had low proteinuria levels) and the exclusion before randomization of patients with proteinuria of 10 g/d or more or serum albumin levels of less than 3.0 g/dL. Proteinuria was similar in patients randomly assigned to the usual and low blood pressure goals in study A (0.90 and 0.97 g/d, respectively) and study B (1.49 and 1.39 g/d, respectively).

Baseline proteinuria was greater in patients with glomerular diseases, diabetes, and hereditary nephritis than in patients with other diseases (Table 2). It was also significantly greater in men than in women, in black persons than in white persons, in patients with a history of hypertension than in patients without such a history (in study A only), and in persons younger than 55 years of age. In study A, baseline proteinuria correlated positively with mean arterial pressure (r = 0.20) and serum total cholesterol level (r = 0.21), and it correlated negatively with glomerular filtration rate (r = –0.17) and serum albumin level (r = –0.44) (P < 0.01 for all correlations). In study B (data not shown), proteinuria correlated with mean arterial pressure and serum levels of cholesterol and albumin levels but not with glomerular filtration rate (r = –0.04; P = 0.53).

Achieved Blood Pressure and Use of Antihypertensive Medication during Follow-Up

Patients with greater proteinuria at baseline had higher mean arterial pressure during follow-up (P < 0.001). Using the combined data from studies A and B, we found that mean arterial pressure in the usual blood pressure group increased from 99.6 mm Hg at baseline to 100.1 mm Hg during follow-up in patients with baseline proteinuria of 1.0 g/d or more and that it increased from 95.2 mm Hg at baseline to 96.5 mm Hg during follow-up in patients with baseline proteinuria of less than 1.0 g/d. In the low blood pressure group, mean arterial pressure declined from 98.1 mm Hg at baseline to 94.1 mm Hg during follow-up in patients with baseline proteinuria of 1.0 g/d or more, and it declined from 95.4 mm Hg at baseline to 92.5 mm Hg during follow-up in patients with baseline proteinuria of less than 1.0 g/d.

In study A, 80% of patients in the usual blood pressure group and 90% of patients in the low blood pressure group took antihypertensive (including diuretic) medications for more than 50% of follow-up visits. In study B, 85% of patients in the usual blood pressure group and 98% of patients in the low blood pressure group took such medications. In study A, among patients with baseline proteinuria of less than 1.0 g/d, angiotensin-converting enzyme inhibitors were used alone or in combination with other agents for more than 50% of follow-up visits by 32% of patients in the usual blood pressure group and 54% of patients in the low blood pressure group. Among patients with baseline proteinuria of 1.0 g/d or more, angiotensin-converting enzyme inhibitors were used alone or in combination with other agents for more than 50% of follow-up visits by 39% of patients in the usual blood pressure group and 54% of patients in the low blood pressure group. In study B, angiotensin-converting enzyme inhibitors were used alone or in combination with other agents for more than 50% of follow-up visits by 26% of patients in the usual blood pressure group and 41% of patients in the low blood pressure group who had baseline proteinuria of less than 1.0 g/d, and by 28% of patients in the usual blood pressure group and 48% of patients in the low blood pressure group who had baseline proteinuria of 1.0 g/d or more. Calcium channel blockers and ß –or -adrenergic agents were used less frequently (data not shown).

Decline in Glomerular Filtration Rate

Relation with Baseline Proteinuria and Blood Pressure Intervention

In study A, patients with baseline proteinuria of more than 0.25 g/d had faster mean rates of decline in glomerular filtration rate and a greater beneficial effect from the low blood pressure goal Figure 1, left]. In study B, similar results were seen in patients with baseline proteinuria greater than 1.0 g/d Figure 1, right). These results were not substantially altered after we controlled jointly for the 10 relevant baseline covariates described above (excluding baseline proteinuria) and the interactions between the baseline covariates and the blood pressure intervention (data not shown). Thus, the relations of baseline proteinuria with decline in glomerular filtration rate and with the beneficial effect of the low blood pressure intervention are independent of these covariates.

View larger version (46K): [in this window] [in a new window]   Figure 1. Baseline proteinuria and decline in glomerular filtration rate (GFR). For study A, estimated mean (± SE) rates of decline in glomerular filtration rate from baseline to 3 years based on the 2-slope model are shown. Mean (± SE) rates of decline in glomerular filtration rate estimated from the 1-slope informative censoring model are shown for study B. Closed circles designate the usual blood pressure group; open circles designate the low blood pressure group. The number in parentheses in each column is the total number of patients in both blood pressure groups who had at least one follow-up glomerular filtration rate measurement. Eight patients in study A and 24 patients in study B had no follow-up glomerular filtration rate measurements. Greater baseline proteinuria is associated with steeper mean glomerular filtration rate decline and with a greater benefit from the low blood pressure goal (P = 0.02 in study A; P = 0.01 in study B).

Early and Late Periods in Study A

Glomerular filtration rate declined faster in the first 4 months of follow-up and slower thereafter in the low blood pressure group than in the usual blood pressure group [10]. Figure 2 shows changes in glomerular filtration rate in patients assigned to the usual or low blood pressure goal within subgroups defined by baseline proteinuria. Figure 3 compares changes in glomerular filtration rate in the usual and the low blood pressure groups during the first 4 months Figure 3, left] and from 4 months on Figure 3, right). During the first 4 months, the mean decline in glomerular filtration rate was consistently steeper in patients in the low blood pressure group, but the difference between the blood pressure groups did not vary significantly among subgroups defined by baseline proteinuria. From 4 months on, the mean decline in glomerular filtration rate was significantly less rapid in the low than in the usual blood pressure group in patients with greater baseline proteinuria (P = 0.006 for the interaction of baseline proteinuria with blood pressure group).

View larger version (51K): [in this window] [in a new window]   Figure 2. Estimated mean decline in glomerular filtration rate (GFR) from baseline to selected follow-up times in study A. Estimated mean (±SE) declines in glomerular filtration rate (mL/min) from baseline to follow-up points coinciding with glomerular filtration rate measurements for different levels of baseline proteinuria. The usual (dashed line) and low (solid line) blood pressure groups are compared. B3 equals the third monthly baseline visit (before randomization); F equals follow-up visits at each given number of months. Three hundred five patients had baseline proteinuria of 0 to 0.25 g/d (mean, 0.08 g/d); 120 had baseline proteinuria of 0.25 to 1.0 g/d (mean, 0.58 g/d); 105 had baseline proteinuria of 1.0 to 3.0 g/d (mean, 1.8 g/day); and 55 had baseline proteinuria of 3 g/d or more (mean, 4.8 g/d).

View larger version (51K): [in this window] [in a new window]   Figure 3. Baseline proteinuria and the initial and final decline in glomerular filtration rate (GFR) in study A. Mean (± SE) initial (baseline to 4 months, left) and final (4 months to end of follow-up, right) rates of decline in glomerular filtration rate based on the 2-slope model. Closed circles designate the usual blood pressure group; open circles designate the low blood pressure group. Greater baseline protein excretion is associated with steeper initial and final mean rates of decline in glomerular filtration rate. Mean initial rates of decline do not differ significantly between blood pressure groups for patients with different levels of protein excretion. Greater baseline protein excretion is associated with a greater beneficial effect of the low blood pressure goal on the final mean rate of decline (P = 0.006).

Correlation with Achieved Blood Pressure

Without adjustment for the baseline covariates, a higher mean follow-up blood pressure was associated with a faster decline in glomerular filtration rate in both study A (over 3 years, r = –0.20; P < 0.001) and study B (r = –0.34; P < 0.001). These relations remained significant (P < 0.001) after we controlled for the 11 relevant baseline covariates and mean follow-up protein intake. The relations were stronger in patients with greater baseline proteinuria in both study A (P < 0.001) and study B (P = 0.008). In study A Figure 4, the relation between follow-up blood pressure and decline in glomerular filtration rate over 3 years was significantly nonlinear (P = 0.011), and the strength of the association between decline in glomerular filtration rate and mean follow-up mean arterial pressure increased at higher mean arterial pressure levels. The rate of decline of glomerular filtration rate was unrelated to follow-up blood pressure in patients with baseline proteinuria of less than 0.25 g/d. However, in patients with baseline proteinuria of 0.25 to 3.0 g/d, the association of higher blood pressure with faster decline in glomerular filtration rate was apparent beginning at about 98 mm Hg. In patients with baseline proteinuria of 3.0 g/d or more, a greater decline in glomerular filtration rate was seen at mean arterial pressures greater than about 92 mm Hg. In study B Figure 5, the relations appeared to be linear. The rate of decline in glomerular filtration rate was unrelated to follow-up blood pressure in patients with baseline proteinuria of less than 1.0 g/d. However, in patients with baseline proteinuria of 1.0 g/d or more, higher follow-up blood pressure was associated with faster decline in glomerular filtration rate at all blood pressure levels.

View larger version (42K): [in this window] [in a new window]   Figure 4. Mean glomerular filtration rate (GFR) decline and achieved follow-up blood pressure in study A. Regression lines relating the estimated mean glomerular filtration rate decline over 3 years to mean follow-up mean arterial pressure (MAP) for groups of patients defined according to baseline proteinuria. Within each group, a 3-slope model was used with break points at 92 and 98 mm Hg. Increasing mean follow-up mean arterial pressure is significantly related to steeper decline in glomerular filtration rate for mean arterial pressure greater than 98 mm Hg in patients with baseline proteinuria 0.25 to 3.0 g/d and for mean arterial pressure greater than 92 mm Hg for patients with baseline proteinuria of 3 g/d or more.

View larger version (39K): [in this window] [in a new window]   Figure 5. Mean glomerular filtration rate (GFR) decline and achieved follow-up blood pressure in study B. Regression lines relating mean decline in glomerular filtration rate to mean follow-up mean arterial pressure (MAP) for groups of patients defined according to baseline proteinuria. Within both groups, a 5-slope model was used with break points at 90, 95, 100, and 105 mm Hg. Decline in glomerular filtration rate is inversely related to follow-up blood pressure for patients with baseline proteinuria of 1 g/d or more but not for patients with baseline proteinuria of less than 1 g/d.

The rate of decline in glomerular filtration rate also correlated significantly with follow-up systolic and diastolic blood pressures (P < 0.001 in both study A and study B) after we controlled for the relevant covariates, and the strength of these correlations increased at higher baseline proteinuria levels. However, follow-up systolic blood pressure and mean arterial pressure were more predictive of the rate of decline in glomerular filtration rate than was diastolic blood pressure. In study A, follow-up mean arterial pressure and its interaction with baseline proteinuria accounted for an additional 7.2% of the variance of the glomerular filtration rate slopes after we controlled for the covariates. This Figure compares with an additional 9.5% of variance accounted for by follow-up systolic blood pressure and 2.0% of variance accounted for by follow-up diastolic blood pressure. In study B, follow-up mean arterial pressure accounted for 6.6% of the variance of the glomerular filtration rate slopes, and systolic and diastolic blood pressures accounted for 7.5% and 3.7% of the variance, respectively.

Change in Proteinuria: Relation with Blood Pressure Control

Mean protein excretion was significantly reduced in the low blood pressure group during years 1, 2, and 3 of follow-up in studies A and B (P < 0.05) (Figure 6). The effect of the low blood pressure goal on the percentage change in proteinuria during follow-up was similar for all levels of baseline proteinuria. This indicates that the low blood pressure goal can delay the increase in proteinuria, even in patients with low baseline proteinuria. However, in patients with baseline proteinuria of less than 1.0 g/d, the difference in percentage change in proteinuria between the blood pressure groups corresponds to a minimal difference in the magnitude of proteinuria. Similar results were obtained when proteinuria was factored by glomerular filtration rate (data not shown).

View larger version (58K): [in this window] [in a new window]   Figure 6. Changes in urine protein from baseline to selected follow-up times in study A. Comparison of changes from baseline in urine protein between patients in the usual (dotted line) and the low (solid line) blood pressure groups within subgroups defined according to baseline proteinuria. Proteinuria was log transformed. Changes in proteinuria are expressed as percentage changes. Three hundred five patients had baseline proteinuria of 0 to 0.25 g/d (mean, 0.08 g/d); 120 had baseline proteinuria of 0.25 to 1.0 g/d (mean, 0.58 g/d); 105 had baseline proteinuria of 1.0 to 3.0 g/d (mean, 1.8 g/d); and 55 had baseline proteinuria of 3.0 g/d or more (mean, 4.8 g/d).

The change in proteinuria also correlated with mean follow-up blood pressure. After we controlled for the relevant baseline covariates, partial correlation coefficients for the rate of change in urine protein (expressed on the log scale) and follow-up blood pressure were 0.28 (P < 0.001) in study A and 0.21 (P = 0.002) in study B.

Change in Proteinuria: Relation with Subsequent Decline in Glomerular Filtration Rate

To assess the effects of changes in proteinuria on the subsequent progression of renal disease, the glomerular filtration rate slope after 4 months of follow-up was regressed on the change in proteinuria during the first 4 months. An initial reduction in proteinuria of 1.0 g/d was associated with a slower mean decrease in glomerular filtration rate in both study A (by 0.92 ± 0.31 mL/min·y; P = 0.003) and study B (by 1.32 ± 0.46 mL/min·y; P = 0.005). This analysis controlled for the 11 relevant baseline covariates and the initial changes between baseline and 4 months in mean arterial pressure, protein intake, and serum levels of total and HDL cholesterol and transferrin. The results were similar when proteinuria was factored by glomerular filtration rate (data not shown).

Discussion

Top Methods Results Discussion Author & Article Info References

Proteinuria may identify patients with renal disease who would benefit from blood pressure control stricter than that currently recommended. As reported previously [10], patients with greater baseline proteinuria had faster declines in glomerular filtration rate and greater benefits from assignment to the low blood pressure goal. For example, patients with baseline proteinuria of more than 1.0 g/d had mean rates of decline in glomerular filtration rate of 6.6 mL/min · y over 3 years in study A and 5.7 mL/min · y in study B. In these patients, assignment to the low blood pressure goal substantially slowed the mean decline in glomerular filtration rate [10]. Current data show a similar trend in patients with baseline proteinuria between 0.25 and 1.0 g/d (Figure 1). In contrast, in patients with proteinuria of less than 0.25 g/d, glomerular filtration rate declined relatively slowly, and assignment to the low blood pressure goal did not affect the decline of glomerular filtration rate.

Correlational analyses confirmed a slower decline in glomerular filtration rate in patients with lower follow-up blood pressure, especially in those subgroups defined by greater baseline proteinuria (Figures 4 and 5). In study A, these analyses showed a nonlinear relation between glomerular filtration rate decline and mean arterial pressure; the association between higher blood pressure and steeper decline in glomerular filtration rate began to become apparent at about 92 mm Hg if proteinuria was more than 3.0 g/d and at about 98 mm Hg or less if proteinuria was between 0.25 and 3.0 g/d. We found no evidence to show that the low blood pressure goal was beneficial in patients with proteinuria of less than 0.25 g/d. However, patients with proteinuria of less than 0.25 g/d had the lowest mean rate of decline in glomerular filtration rate during the MDRD Study. Thus, the risk for developing renal failure is also low in these patients.

Like the results of all correlational analyses, our results relating decline in glomerular filtration rate to achieved blood pressure are limited by potential confounding effects from other variables. We cannot rule out the possibility that a variable we did not measure jointly influenced proteinuria, blood pressure, and decline in glomerular filtration rate to induce the observed relations among these variables. However, we did find that both the association of achieved blood pressure control with decline in glomerular filtration rate and the dependence of this association on baseline proteinuria remained significant after we controlled for baseline covariates previously shown [12] to be jointly predictive of decline in glomerular filtration rate. These correlational analyses might also be affected by the effects of the presumed dependent variable (glomerular filtration rate decline) on the presumed independent variable (blood pressure). Potentially, patients with more severe renal disease may have both a faster decline in glomerular filtration rate and higher blood pressure. However, two reasons make it unlikely that this would fully account for the observed relations. First, blood pressure was controlled and was not allowed to vary freely. Second, similar relations among blood pressure control, proteinuria, and decline in glomerular filtration rate were seen in the comparisons of randomized groups, where causal relations can be more reliably inferred. Thus, the beneficial effects of reduced achieved blood pressure on decline in glomerular filtration rate appear to be robust.

Another potential confounding effect is the differential use of antihypertensive agents among patients with varying levels of follow-up blood pressure. The MDRD Study was not designed to determine which classes of antihypertensive agents are beneficial in slowing the progression of renal disease independent of their effects on blood pressure. However, the recommended regimens were similar for hypertensive patients in both blood pressure groups, regardless of blood pressure level, and they are commonly used in practice. Our conclusions apply to patients whose blood pressure is controlled principally with angiotensin-converting enzyme inhibitors, calcium channel blockers, or agents affecting the ß –and -adrenergic nervous system, with or without diuretics.

In its most recent recommendations, the Joint National Committee for the Detection, Evaluation, and Treatment of High Blood Pressure suggested a target blood pressure for patients with renal disease of 130/85 mm Hg or less, equivalent to a mean arterial pressure of 100 mm Hg or less [14]. We suggest that proteinuria level should be considered when blood pressure goals are defined for patients with chronic renal disease. In patients with proteinuria of 1 g/d or more, we recommend a mean arterial pressure goal of 92 mm Hg or less (equivalent to a blood pressure of 125/75 mm Hg or less). In patients with proteinuria of 0.25 to 1 g/d, a mean arterial pressure of 98 mm Hg or less (approximately equivalent to a blood pressure of 130/80 mm Hg or less) may be an appropriate goal. In patients with proteinuria of less than 0.25 g/d, a goal lower than that recommended by the Joint National Committee had no apparent benefit. Of course, a critical component of such recommendations is the safety of the lower blood pressure goals. Preliminary results from the MDRD Study suggest that a mean arterial pressure of 92 mm Hg or less is safe and well tolerated for as long as 3 years [24].

Patients with renal disease caused by type 1 diabetes and type 2 diabetes who required insulin were excluded from the MDRD Study. These patients characteristically have proteinuria of 1 g/d or more. A low blood pressure goal has been reported to have had a beneficial effect in diabetic nephropathy [25, 26]. These findings are consistent with our finding that the low blood pressure goal benefits patients with proteinuria.

Previous studies [27-31] postulated that greater proteinuria is associated with more rapid progression of renal disease. However, these studies were limited by their relatively small numbers of patients, their relatively short follow-up periods, and their use of the reciprocal of serum creatinine or creatinine clearance as an index of glomerular filtration rate. Although they were not obtained from a comparison between randomized groups, our results are based on a larger number of patients followed for a longer period of time and on more accurate measurements of glomerular filtration rate [32]. They show a strong association of greater baseline proteinuria with faster decline in follow-up glomerular filtration rate. They also suggest that the reduction of proteinuria, independent of the reduction of blood pressure, is associated with a subsequent beneficial effect on the progression of renal disease. These findings support previous suggestions that proteinuria is not only a risk factor for more rapid decline in glomerular filtration rate but may also contribute pathogenetically to the progression of renal disease [3, 9]. Moreover, if these results are verified in prospective clinical trials, they would suggest that the blood pressure goal and classes of antihypertensive drugs should be selected to minimize proteinuria.

Our study provides evidence to show that, in humans with renal disease, greater levels of proteinuria may predict a more rapid decline in glomerular filtration rate, and that lowering blood pressure below the goal recommended by the Joint National Committee retards the progression of renal disease in patients with proteinuria.

Dr. Adler: Harbor-University of California, Los Angeles, Medical Center, Division of Nephrology and Hypertension, 1124 West Carson Street, C-1 Annex, Torrance, CA 90502.

Dr. Burkart: Bowman-Gray School of Medicine, Section of Nephrology, Medical Center Boulevard, Winston-Salem, NC 27157-1053.

Dr. Greene: Department of Biostatistics and Epidemiology, Cleveland Clinic Foundation, 9500 Euclid Avenue, P88, Cleveland, OH 44195.

Dr. Hebert: Ohio State University, Department of Internal Medicine, Nephrology, N210 Means Hall, 1654 Upham Drive, Columbus, OH 43210.

Dr. Hunsicker: University of Iowa Hospitals and Clinics, E300-F General Hospital, Department of Internal Medicine, 200 Hawkins Drive, Iowa City, IA 52242.

Dr. King: New England Medical Center Hospital, Division of Nephrology, Box 148, 750 Washington Street, Boston, MA 02111.

Dr. Klahr: Department of Medicine, The Jewish Hospital of St. Louis, Washington University School of Medicine, 216 South Kingshighway Boulevard, St. Louis, MO 63110.

Dr. Massry: University of Southern California School of Medicine, LACUSC Medical Center, 1200 North State Street, Room 4004, Los Angeles, CA 90033.

Dr. Seifter: Renal Division, Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115.