Short-Term Effects of Protein Intake, Blood Pressure, and Antihypertensive Therapy on Glomerular...
-
Upload
independent -
Category
Documents
-
view
0 -
download
0
Transcript of Short-Term Effects of Protein Intake, Blood Pressure, and Antihypertensive Therapy on Glomerular...
Journal of the American Society of Nephrology 2097
Short-Term Effects of Protein Intake, Blood Pressure, andAntihypertensive Therapy on Glomerular Filtration Rate inthe Modification of Diet in Renal Disease Study1’2Modification of Diet in Renal Disease Study Group3’4
From the National Institute of Diabetes. Digestive andKidney Diseases. National Institutes of Health, Be-thesda, MD
Prepared by: Andrew S. Levey. M.D.. Gerald J. Beck.Ph.D. Juan P. Bosch. M.D., Arlene W. Caggiula, Ph.D.,Tom Greene. Ph.D.. Lawrence G. Hunsicker, M.D., andSaulo Klahr, M.D.
(J. Am. Soc. Nephrol. 1996; 7:2097-2109)
ABSTRACTGlomerular filtration rate is often used to assess thelevel of renal function and the progression of renaldisease. However, the short-term effects of dietaryprotein restriction, blood pressure reduction, and spe-cific classes of antihypertensive agents on GFR maybe opposite in direction from their observed long-term beneficial effects on the progression of renaldisease. The purpose of these analyses was to char-acterize these short-term effects and determinewhether they can obscure the relationship betweenrenal structure and function in patients with slowlyprogressive renal disease. The Modification of Diet inRenal Disease Study was a randomized trial of theeffect of dietary protein restriction and strict bloodpressure control on the decline in GFR in 840 patients
with mean (range) baseline GFR of 36. 1 (1 3 to 55)mL/min per 1 .73 m2. In this study, comparisons of therandomized groups and correlational analyses wereused to determine the short-term (4-month) effects onGFR of changes in protein intake, mean arterial pres-sure (MAP), and class of antihypertensive agents(computed as the reduction in GFR associated withstarting versus stopping medications) during the first 4months of follow-up and in subsequent 4-month inter-vals during the first 2 yr offollow-up. Combining resultsover the first 2 yr of follow-up, and controlling forchanges in antihypertensive medications, the inde-pendent effect on GFR of changes in protein intake
1 Received January 31 . 1996. Accepted April 30. 1996.
2 This study was presented In part at the 28th Annual Meeting of the American
Society of Nephrology, San Die9o, November 5-8. 1995.
3 correspondence to Dr. AS. Levey, Division of Nephrology, New EnglandMedical Center. 750 WashIngton Street. Boston. MA 021 1 1.
4 The institutions and Investigators who participated in the MDPD Study are listedin Reference 12.
10.46-6673/0710-2097$03.00/0Journal of the American society of NephrologyCopyright C 1996 by the American Society of Nephrology
and MAP was 1 . 1 mL/min per 0.4 g/kg per day and0.9 mL/min per 10 mm Hg, respectively (P < 0.001).
These effects were observed in patients with increas-ing or decreasing protein intake or MAP, and inpatients with stable or changing antihypertensive reg-imens. Starting treatment with diuretics, p-blockers, orangiotensin-converting enzyme inhibitors was associ-ated with a 4.4-, 3.2-, or 2.2-mL/min greater GFR de-
dine, respectively, than was stopping this treatment(P < 0.001). The effect of changes in protein intake,MAP, and diuretics was greater in patients with higherinitial GFR. After controlling for initial GFR, there wereno significant differences between the short-term ef-
fects observed during the first 4 months of follow-upand the short-term effects during subsequent follow-up. Changes in protein intake, blood pressure, andantihypertensive agents have small but statisticallysignificant short-term effects on GFR. These effects canlead to clinically significant changes in renal functionin patients undergoing multiple interventions and arelarge enough to confound the results of clinical trialsin patients with slowly progressive renal disease. Fu-
ture studies using GFR to assess the progression ofrenal disease should take into account these short-term effects when the length of follow-up is beingplanned.
Key Words: GFR, dietary’ protein. blood pressure, antihyper-
tensive therapy
G bomerular filtration rate is generally considered
the best index of renal function in health and
disease ( 1 ). For this reason, serial estimates of GFR
are frequently used to assess the progression of renal
disease (2). Implicit in this practice is the assumption
that the level of renal function, measured as the GFR,
reflects the severity ofrenab structural injury, and that
the rate of decline in GFR reflects the progression of
injury.
In principle, the GFR is the product of the number of
functioning nephrons and the single-nephron GFR.
Although the number of functioning nephrons may
reflect the severity of renal injury, the single-nephron
GFR may be affected both by the severity of the renal
injury and by hemodynamic changes. Indeed, it is well
known that dietary protein restriction, blood pressure
reduction, and specific classes of antihypertensive
agents can have short-term effects on GFR because of
their direct effects on the hemodynamic determinants
of single-nephron GFR (3,4). Moreover, the short-term
effects of these interventions are opposite in direction
Effects of Diet and Blood Pressure Control on GFR
2098 Volume 7 . Number 10 . 1996
to their observed long-term beneficial effect on the
progression of renal disease (5-8). Therefore, these
short-term fluctuations in GFR could potentially ob-
scure the relationship between renal function and
structure, thereby confounding the interpretation of
clinical trials and limiting the usefulness of these
therapies in clinical practice.
The short-term effects of changes in protein intake,
blood pressure, and antihypertensive agents on the
GFR in patients with renal disease are not well de-
fined. We analyzed data from the Modification of Diet
in Renal Disease (MDRD) Study, a multicenter con-
trobled clinical trial to determine the effect of dietary
protein restriction and strict blood pressure control on
the rate of decline in GFR in patients with chronic
renal disease (9-12), to characterize these short-term
effects. In particular. we determined the short-term
effects on GFR of changes in protein intake, blood
pressure, and the use of antihypertensive agents,
whether these effects were reversible, independent of
each other, and influenced by the initial level of GFR.
In addition, we explored whether the short-term
changes in GFR that occurred during the first 4
months after randomization, when patients were
striving to reach new diet and blood pressure goals,
differed from the short-term changes observed during
the subsequent follow-up period, when patients’ goals
were not changing. The results have implications for
clinical practice and for the design of clinical trials.
METHODS
Study Design
The MDRIJ Study consisted of two randomized, multi-center, two-by-two factorial trials of 840 patients with var)-ous chronic renal diseases (10,12). In 585 patients withbaseline GFR of 25 to 55 mL/min per 1 .73 m2 (Study A), wecompared a usual versus low-protein diet (1.3 versus 0.58g/kg per day) and a usual versus low blood pressure goal(mean arterial pressure � 107 versus �c92 mm Fig). In 255patients with baseline GFR between 13 and 24 mL/min per1 .73 m2 (Study B), we compared a low- versus very-low (0.28g/kg per day)-protein diet, supplemented by a keto acid-amino acid mixture (0.28 g/kg per day), and a usual versuslow blood pressure goal. The keto acid-amino acid supple-ment in Study B contained approximately 0. 18 g/kg per dayamino acids and 0. 10 g/kg per day keto acids. Therefore, theprescribed intake of amino acids from food and supplementin Study B was approximately 0.58 and 0.46 g/kg per day inthe low- and very-low-protein diets, respectively. In bothStudies A and B. the blood pressure goals were 6 mm Hghigher for patients older than 61 yr of age. Glomerularfiltration rate was measured during baseline (before random-ization), after 2 months, after 4 months, and every 4 monthsthereafter. The mean duration of follow-up was 2.2 yr.
Dietitians at each clinical center were trained to implementand monitor the intervention. Patients were informed of thedietary assignment at the first follow-up visit after random-ization (within 1 month). Thereafter, patients were counseledmonthly or more frequently as necessary to attain the pre-scribed levels of intake.
Blood pressure was controlled by a combination of phar-macologic and nonpharmacologic interventions. Nonphar-
macologic therapy consisted primarily of recommendationsfor regular exercise and reductions in weight and in alcoholand sodium intake. Pharmacologic therapy was based on thestepped-care approach defined in the 1988 Report of theJoint National Committee ( 13). All antihypertensive drugswere allowed, but angiotensmn-converting enzyme (ACE) in-hibitors, with or without a diuretic, were encouraged as theagents of first choice. Calcium channel blockers, with orwithout a diuretic, were encouraged as agents of secondchoice. Blood pressure regimens were implemented at thefirst follow-up visit after randomization and were modifiedmonthly or more often as necessary to achieve goal bloodpressure levels.
Measuring Protein Intake, Blood Pressure,Antihypertensive Medications, and GFR
Protein intake (g/day) was calculated as 6.25 X IUUN(g/day) + SBW (kg) x 0.03 1 (g/kg SBW per day)l, where UUNis the urea nitrogen excretion rate and SBW is the standardbody weight ( 1 4, 1 5). For patients in Study B assigned to thevery-low-protein diet, this equation estimates the intake of
protein from food as well as amino acids in the keto acid-amino acid supplement. As reported elsewhere (16), aftercontrolling for intake of protein from food as well as aminoacids in the supplement, there was no difference betweendiet groups in the long-term response of GFR to the dietinterventions in Study B. Hence, for the analyses reportedhere, we calculated changes in protein intake as changes inintake of protein from food and amino acids in the supple-ment. For simplicity, this quantity is referred to as “totalprotein intake.”
Blood pressure was measured using a Hawksley random-zero sphygmomanometer (Copiague, NY) according to estab-lished methodology ( 13). Patients were placed in a quiet roomfor 5 mm before measurement. The sitting blood pressurewas taken three times by the same nurse or technician. Theaverage of the last two measurements was used as the bloodpressure for the visit ( 1 7). Mean arterial pressure (MAP) wascalculated as DBP + #{189}(SBP -DBP). where DBP and SBP arediastolic and systolic blood pressure, respectively.
Glomerular filtration rate was measured as the renal clear-ance of 35 gzCi of ‘2�liothalamate after bolus subcutaneousadministration without concomitant epinephrine (18,19).Measurements were performed in the sitting position, in themorning, after an 8-h fast.
Statistical Methods
Comparison of randomized groups. As reported elsewhere(1 2), the effects of the diet interventions on the rate of GFRdecline in both Studies A and B were similar in the usual andlow blood pressure groups. Likewise, the effects of the bloodpressure intervention on the rate of GFR decline were similarin the usual and low-protein diet groups in Study A and inthe low- and very-low-protein diet groups in Study B. Con-sequently, comparisons of the diet groups are presented forall patients. regardless of blood pressure assignment, andthe blood pressure groups are compared for all patients,regardless of diet assignment.
Effects on GFR were reported previously ( 1 2) and areshown in Figure 1 to illustrate (1) the differences betweenrandomized groups in the initial GFR decline. during the first4 months after randomization, when diet and MAP goals werechanging, and during subsequent follow-up, when goalswere stable; and (2) the greater differences between the initialand subsequent GFR declines in patients with higher base-
A C
C
3
6
9
B3 F4 P12 P20 P28
Modification of Diet in Renal Disease Study Group
Journal of the American Society of Nephrology 2099
StUdyA StudyB
�:t��I12�_�0 � -Low
B3 ft
E
�‘�-E0’””
.E!
�
%,. 12
�F28 F36
��t:z:i:i3
6
9
12
15Protsln(n =2�1)1
12
��L�Prots1n(n=127) �
-VerYLowProtsIn(n=124)I . . .
83 F4 Fe P12 F16 F20 F24
Moidhs
�
�Uw&BP(n=121)1 12
Low BP (n=13U� I
P12 F20
Months
B3 N F8 P12 F16 F20 F24
Months Months
Figure 1 . Estimated mean (SE) GFR decline from baseline to selected follow-up times In Studies A and B. B3 denotes the finalbaseline visit. F denotes monthly follow-up visits. The paffern of GFR deviates significantly from linearity in Study A, but notin StudyB. In Study A, In the first 4 months after baseline, the GFR decline was significantly faster in the low-protein diet group (A) and thelow blood pressure group (B), compared with the usual protein diet group and usual blood pressure group, respectively. After4 months, the GFR decline was 28% slower (P = 0.009) in the low-protein diet group and 29% slower (P = 0.006) In the low bloodpressure group. From baseline to 3 yr offollow-up, the projected decline was 1.2 mI/mm less in the low-protein diet group (P =
0.3) and 1.6 mI/mm less in the low blood pressure group (P = 0.18). In Study B, the GFR decline was 19% less in thevery-low-protein diet group, compared with the low-protein diet group (P = 0.07) and 12% less in the low blood pressure group,compared with the usual blood pressure group (P = 0.28). Figures 1A and lB reproduced with permission from the New EnglandJournal of Medicine 330; 877-884, 1994.
line GFR (Study A compared with Study B). For the analysespresented here, the diet groups and blood pressure groupsare compared from baseline to 4 months of follow-up, with-
out regard to the changes in GFR between the fourth monthof follow-up and the end of the study. This method ofcomputing the GFR decline differs from the previous report( 1 2), but the differences in GFR decline between diet groupsand between blood pressure groups are similar to the previ-ous report.
Correlational analyses. Standard least-squares multipleregression (20) was used to relate the change in GFR in thefirst 4-month interval to changes in achieved protein intake,blood pressure, and in classes of antthypertensive medica-tions. To determine how the short-term responses of GFR tochanges in these factors varied with baseline level ofGFR andwith the duration of follow-up, the successive 4-monthchanges in GFR over the first 2 yr of follow-up were alsorebated to contemporaneous 4-month changes in proteinintake, blood pressure, and in the use of the respectivemedication classes in a combined longitudinal analysis (21).The model included interaction terms of each of these pre-dictor variables with baseline GFR and with the duration of
follow-up. To account for correlations between the successive4-month changes in GFR, a Toeplitz error structure (22) withtwo bands was assumed.
Assessing measurement error. The changes in proteinintake and blood pressure used in the correlational analyseswere derived from measurements at single time points. Thesemeasurements are subject to error (for example, errors in24-h urine collections and laboratory measurements of theurine urea nitrogen levels used to compute protein intake).and do not account for hourly or daily fluctuations. Inprinciple, the effect of measurement error in ascertainingprotein intake and blood pressure would be to weaken theobserved association between changes in these variables and
changes in GFR (23). The size of the biases in these estimatesdepends on the magnitude of the measurement errors, withlarger errors leading to larger biases.
Error in laboratory measurement of urine urea nitrogenwas estimated from 65 masked, split samples of urine ob-tamed as part of the quality control of the study (data notshown). Error in measurement of blood pressure was as-sessed by the variation between the second and third sittingblood pressure measurements (22). The multiple regression
Effects of Diet and Blood Pressure Control on GFR
2100 Volume 7 . Number 10 . 1996
analyses were repeated to control for attenuation (24) asso-
ciated with these estimates of measurement error. We did notattempt to assess the additional biases that might resultfrom patient errors in urine collections or daily fluctuationsin levels of protein intake and blood pressure.
RESULTS
Effects of Protein Intake and Blood Pressure
We examined relationships ofshort-term changes in
protein intake and blood pressure on GFR during two
periods: the first 4 months after randomization, to
assess short-term effects, and from 4 months onward,
at successive 4-month intervals, to assess continua-
tion of short-term effects.
Changes in protein intake and blood pressure from
baseline to 4 months of follow-up. In Study A, mean
(SD) protein intake remained stable at 1 . 1 2 (0. 1 8) and
1 . 1 1 (0.20) g/kg per day from baseline to 4 months in
the usual protein diet group, but declined from 1.12
(0.20) to 0.78 (0. 16) g/kg per day in the low-protein
diet group. In Study B, mean total protein intake (from
food and supplements) declined from 0.86 (0. 18) to
0.73 (0. 14) g/kg per day in the low-protein diet group
and declined from 0.87 (0. 19) to 0.65 (0. 13) g/kg per
day In the very-low-protein diet group. Thus the dif-
ference between randomized groups in mean protein
intake was greater in Study A than in Study B. In
Study A, MAP remained stable at 97.5 ( 10.6) and 97.3
( 10.5) mm Hg from baseline to 4 months in the usual
blood pressure group, but declined from 98. 1 ( 10.7) to
94.8 (10.6) mm Hg in the bow blood pressure group.
Similarly, in Study B, MAP remained stable at 98.0
( 1 1 . 1 ) and 98.6 ( 1 1 . 1 ) mm Hg in the usual blood
pressure group, but declined from 98.4 (10.7) to 95.9
( 10.7) mm Hg in the low blood pressure group.
Changes in GFR during the first 4 months of fol-
bow-up (short-term changes). We compared the GFR
declines between patients assigned to different diets
and blood pressure goals in Studies A and B, irrespec-
tive of achieved levels of protein intake or blood pres-
sure. In addition, we correlated changes in protein
intake and blood pressure and changes in GFR in both
studies, irrespective of assigned diet or blood pres-
sure.
Comparisons of diet groups and blood pressure
groups. Table 1 shows the effect of the bow-protein diet
and low blood pressure goal on changes in GFR during
the first 4 months after randomization in Studies A
and B and in subgroups of patients in Study A defined
by baseline GFR. In Study A, the GFR decline in the
bow-protein diet group was 1 .84 mL/min greater than
in the usual protein diet group (P = 0.00 1 ). There was
a trend (P = 0.07) toward a greater difference between
diet groups among patients with higher baseline GFR.
The mean GFR decline in the bow blood pressure group
was 1 .90 mL/min greater than in the usual blood
pressure group (P = 0.00 1 ), and the difference was
significantly greater among patients with higher base-
line GFR (P = 0.005). In Study B, the mean GFR
decline was similar between the two diet groups and
between the two blood pressure groups.
Correlations of changes in GER with changes in
protein intake and blood pressure. Figure 2 shows
correlations between changes in protein intake or
blood pressure with changes in GFR, irrespective of
diet or blood pressure group. Significant, but weak,
direct correlations were observed for changes in pro-
tein intake in Study A (r = 0. 13, P = 0.007, Figure 2A)
and changes in blood pressure in both Studies A (r =
0.19, P< 0.001, Figure 2B)and B(r = 0.28, P< 0.001,
Figure 2D). The correlation was not significant for
changes in total protein intake in Study B (r = 0.09,
P = 0.21, Figure 2C).
Table 2 shows the relationship (the slope of the
regression line) between changes in protein intake or
blood pressure with changes in GFR Studies A and B
and compares the slopes for various subgroups of
patients within each study. Overall in Study A, there
was a 3. 1 1 -mL/min decline in GFR per 1 .0-g/kg per
day reduction in protein intake (P = 0.007) and a
0.12-mL/min decline in GFR per 1.0-mm Hg reduc-
tion in MAP (P = <0.00 1 ), with stronger relationships
(steeper slopes of the regression lines) in patients with
higher baseline GFR. Consistent with this finding
TABLE 1 . Comparison of change (�) in GFR between diet groups and between blood pressure groups duringthe first 4 months of follow-up, as a function of baseline GFR
Baseline GFR Range(mL/min)
NDiet Interventiona Blood Press ure lnterventlonb
Mean � GFR SE PMean � GFR SE P
Overall Study A 528 + 1,84c 0.58 0.002 + 1 ,90d 0.58 0.00 1Study A GFR > 48.5 172 +2.68 1.13 0.02 +4.43 1.14 <0.001Study A GFR 36.5 to 48.5 181 +2.25 0.96 0.02 +1.13 0.96 0.24Study A GFR <36.5 175 +0.65 0.85 0.45 +0.01 0.84 0.99Overall Study B 231 -0.08 0.48 0.87 +0.02 0.48 0.97
a Study A: Low-protein diet group-usual protein diet group. Study B: Very-low-protein diet group-low-protein diet group.b Studies A and B: Low blood pressure group- usual blood pressure group.C Among patients in Study A. mean change in GFR was 1.84 mL/min greater for the low-protein diet group than for the usual protein diet group.
There is a trend toward a greater difference between the diet groups in patients with higher baseline GFP (P = 0.07).d Among patients in Study A. mean change in GFR was 1 .90 mL/min greater for the low blood pressure group than for the usual blood pressure
group. The greater difference between the blood pressure groups in patients with higher baseline GFR was statistically significant (P = 0.005).
Study Ar=.129
. p=.007
‘E�.� 12
E 8
� 4U.C,C -4
& -8
1-12C.) -16
112
E 8
C, 0
.� -4
R1
0 �
StudyA �r=.194p<.�I #{149} #{149}. �
.
� � #{149}. � � “
:��“
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
change in Protein Intake (Q/k9/d)
C
I’
SludyBr=.087p-.2l � . .
‘E� 10.� 12
E 8� 4
C -4
I -8
j�120 -16
-40-30-28-10 0 10 20 30 40 50
Change in Ms�i Merlel Blood Pressure (mm Hg)
D
j12
E 8
C, 0
.E -4
& -8
1-120 -W
Study Br=278p<.�I
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
chang. ki � Pro�In Intake (g/ICg/d)
-40-30-20-10 0 10 28 30 40 50
change In Me�i Arterial Blood Pressure (mm Hg)
Figure 2. Relationships of changes in protein intake and blood pressure to changes In GFR In Studies A and B. Scatter plots ofindividual patient values for changes In GFR versus changes in protein intake or mean arterial pressure from baseline to thefourth month offollow-up. Extreme values are not Included on the plot (16 extreme values In A and 20 extreme values In B), butare included In the calculation of the regressions and correlations. The figures shows the regression lines (solid) and 95%confidence intervals forthe regression lines (dashed lines). The doffed horizontal line indicates the mean GFR decline during the4-month Interval for each study (-2.79 mL/mln In Study A and -2.13 mI/mm in Study B). Total protein intake in Study B refers tointake of protein from food and amino acids In the keto acid-amino acid supplement.
Journal of the American Society of Nephrology 2101
A
Modification of Diet in Renal Disease Study Group
B
were the weaker relationships (less steep slopes) in
Study B; a l.73-mL/min decline in GFR per l.0-g/kg
per day reduction in total protein intake (P = not
significant) and a 0.094-mL/min decline in GFR per
1 .0-mm Hg reduction in MAP (P < 0.00 1 ). These
relationships were observed whether protein intake or
blood pressure was increasing or decreasing, which
strongly suggests that the reduction in GFR after
decreased protein intake or decreased blood pressure
is the result of physiologic changes in the GFR rather
than the progression of renal disease. Moreover, these
relationships were observed whether the class of pre-
scribed antihypertensive agents was constant or
changing, indicating that the changes in GFR were the
result of changes in protein intake or blood pressure,
per Se, and not the result of hemodynamic effects of
antthypertensive agents.
Changes in GFR during subsequent 4-month inter-
vals (continuation of short-term changes). Although
the mean levels of protein intake and blood pressure
were generally stable after the first 4 months of follow-
up, protein intake, blood pressure, and use of antihy-
pertensive agents fluctuated in individual patients. To
assess whether these short-term changes continued
to affect GFR during follow-up, we repeated the corre-
lation analyses described above in successive
4-month intervals during the first 2 yr of follow-up.
In Study A, the relationship between changes in
protein intake and changes in GFR was significant in
five of the six 4-month intervals throughout the first 2
yr. In both Studies A and B, the relationship between
changes in blood pressure and changes in GFR was
significant during all six 4-month intervals. In Study
B, there was a significant relationship between
changes in total protein intake and changes in GFR
during two of the six 4-month intervals. In separate
analyses, we found no significant effect of the fol-
low-up interval on these relationships in either study
(data not shown); hence, we combined all 4-month
intervals to compute an average slope for the relation-
ship of changes in protein intake or blood pressure to
changes in GFR. Table 3 shows these relationships
(the slopes of the regression lines relating changes in
protein intake and blood pressure with changes in
GFR) in the first four months and over all 4-month
intervals. Because of the barge number of measure-
ments, the estimates of the slopes of the regression
lines over all 4-month intervals are more precise than
Effects of Diet and Blood Pressure Control on GFR
2102 Volume 7 - Number 10 ‘ 1996
TABLE 2. Relationships (slopes of regression lines) of changes (�) in protein intake and mean arterial pressure(MAP) with changes in GFR during the first 4 months of follow-up
Study A (GFR 25 to 55 mL/min per 1.73 m2) Study B (GFR 13 to 24 mI/mm per 1.73 m2)
‘� GFR versus � ProteinIntake
�GFR versus � MAPGFR versus � Total
.
Protein Intakeaz� GFR versus z� MAP
N Slopec SEN Slope’� SE N Slopec SE N Slopeb SE
All Patients 441 3.1 1d ] .14 519 012d 0.026 21 1 1.73 1.38 224 009d 0.022
Baseline GFR(mL/min)
>48.5 147 6.lOd 2.13 171 016d 0.04736.5 to 48.5 146 3.20 2.17 177 0.07 0.048
<36.5 148 0.40 1.60 171 0.09e 0.038Protein Intake
>0 139 104d 3.86 139 014d 0.047 31 -0.36 8.38 31 o.lle 0.051<0 302 47#{176} 2.05 300 010d 0.036 180 2.65 1.79 177 0.1l’� 0.026
MAP>0 197 3.38e 1.52 224 0.11#{176} 0.053 88 1.26 2.01 97 0.05 0.054<0 242 2.68 1.68 295 019d 0.058 120 1.34 1.83 127 017d 0.049
Change in MedsNo 316 3.10#{176} 1.30 376 009d 0.032 145 3.04 1.55 155 008d 0.029Yes 111 3.08 2.50 127 015d 0.046 58 -2.32 2.96 60 0.lOe 0.038
a Total protein intake from food and supplements.b Study A: mean change in GFR (mL/min) per 1.0 g/kg per day change in protein intake. Study B: mean change in GFR (mL/min) per 1.0 g/kg per
day change in total protein intake from food and supplements.C Studies A and B: mean change In GFR (mL/min) per 1.0 mm Hg change in MAP.dp< 0.01.eooi < P< o.os.
the estimates in each 4-month interval, as shown by
the smaller standard errors.
Thus, fluctuations in protein intake and blood pres-
sure were associated with short-term effects on GFR
throughout follow-up. However, because the mean
protein intake and blood pressure were relatively sta-
ble in each the diet and blood pressure groups, the
effect of these fluctuations on GFR would be expected
to cancel. Hence, the GFR mean decline was relatively
constant after 4 months.
Effects of Antihypertensive Medications
The MDRD Study was not designed to assess the
effect of different classes of antihypertensive agents on
the decline GFR. Nonetheless, a variety of antihyper-
tensive agents was used and a large number of pa-
tients changed antihypertensive medications during
follow-up. In Study A, the percent of patients either
starting or stopping the various classes of antihyper-
tensive medications in each 4-month interval ranged
from 2.2 to 1 2.3% in Study A and from 0.7 to 14.7% in
Study B.
For these analyses, we assessed the association of
changes in antihypertensive therapy to short-term
changes in GFR. Thus, we compared GFR changes
between subgroups of patients starting a class of
medications to those stopping the class of medications
during each 4-month interval during the first 2 yr of
follow-up. Specifically, we compared patients who
were not taking a class of medications at the begin-
ning of the interval but who were taking the medica-
tion at the end of the interval to patients who were
taking a class of medications at the beginning of the
interval but not at the end of the interval. In principle,
the difference in GFR change between patients start-
ing and stopping a medication class might be twice as
barge as the difference in GFR change between pa-
tients starting and those remaining off the medication
or between patients stopping and those remaining on
the medication.
These analyses were performed using multiple re-
gression so that the effects of changing the respective
medication classes were assessed after controlling for
the effects of starting or stopping other classes of
medications. We first compared the changes in GFR
between subgroups of patients who started and those
who stopped various classes of antihypertensive
agents in each 4-month interval in Studies A and B
(data not shown). The follow-up interval had no sig-
nificant effect on the association between the change
in any of the medication classes and the change in
GFR in either Study A or B; hence, we combined all
4-month intervals to determine an average associa-
tion over all 4-month intervals (Table 4).
The strongest association in both Studies A and B
was with diuretics (a 5.41- and 2. 18-mL/min greater
GFR decline in patients starting medications versus
patients stopping medications [P < 0.001 and P =
0.0041 in Studies A and B, respectively). Significant
associations were also observed with ACE inhibitors
Modification of Diet in Renal Disease Study Group
Journal of the American Society of Nephrology 2103
TABLE 3. Relationships (slopes of regression lines) of changes (z�) in protein intake and mean arterial pressure(MAP) with changes in GFR over successive 4-month intervals during follow-up
Study A (GFR 25 to 5 5 mI/mm per 1.73 m2) Study B (GFR 13 to 24 mL/min per 1 .73 m2)
Month ofFollow-Up
L� GFR versus � ProteinIntake
GFR versus z� MAPz� GFR versus � Total
Protein IntakeGFR versus z� MAP
N Slope0 SE N Slopeb SE N Slope0 SE N Slopeb SE
Oto4 441 +3.lc 1.1 519 +0.12c .03 211 +1.7 1.4 224 +0.09c 0.020 to 24d +3.0w 0.6 +0.12c .01 +2.7c 0.7 +0.07c 0.01
a Study A: mean change in GFR (mL/min) per 1.0 g/kg per change in protein intake. Study B: mean change in GFR (mL/min) per 1.0 g/kg per daychange in total protein intake from food and supplements.b Studies A and B: mean change in GER (mL/mln) per 1 .0 mm Hg change in MAP.C p< 0.01.d No significant interaction with follow-up interval.
in Studies A and B (a 3.54- and 1.50-mL/min greater
GFR decline [P = <0.001 and P = 0.0351, respectively)
and with 3-bbockers in Studies A and B (a 3. 15- and
2.330-mL/min greater GFR decline [P = 0.016 and
P = 0.0101, respectively). No significant association
was observed for starting or stopping calcium channel
blockers or other classes of antihypertensive medica-
tions.
Independent Effects of Protein Intake, BloodPressure, and Antihypertensive Agents
Multivariable regression analyses. To assess the
independent effects of changes in protein intake,
blood pressure, and medication classes on GFR, we
performed comprehensive multiple regression analy-
ses for all 4-month intervals. In addition, the observed
associations were related to different levels of initial
GFR and to different 4-month intervals during follow-
up. To assess the effects of error in measurement of
TABLE 4. Association of changes (�) in classes of
antihypertensive medications with changes in GFRover all 4-month intervals0
L� Medication z� GFRbc SE P Value
Study A (GFR 25 to 55 mI/mmper 1.73 m2)
ACE Inhibitors -3.54 0.79 <0.001f3Blockers -3.15 1.31 0.016Calcium Channel Blockers -0.67 1.05 0.52Diuretics -5.41 0.85 <0.001Others -2.14 1.22 0.79
Study B (GFR 13 to 24 mL/minper 1.73 m2)
ACE Inhibitors -1.50 0.71 0.035j3 Blockers -2.33 0.90 0.01Calcium Channel Blockers -0.18 0.83 0.83Diuretics -2.18 0.76 0.004Others -0.81 0.90 0.37
a ACE, angiotensin-converting enzyme.b GFR decline in patients starting versus those stopping the class of
antihypertensive agent.C No significant interaction with follow-up Interval.
protein intake and blood pressure, the analyses were
repeated with and without controlling for measure-
ment error in the first 4-month interval.
After controlling for the changes in class of antihy-
pertensive medications, short-term changes in both
protein intake and blood pressure remained indepen-
dently associated with short-term changes in GFR (left
half ofTable 5). A change in protein intake of 1 .0 g/kg
per day was associated with a 2.77-mL/min change in
GFR (P < 0.001). Hence, a reduction in protein intake
of 0.4 g/kg per day would bead to a decline in GFR of
about 1 . 1 mL/min. Likewise, a change in MAP of 1.0
mm Hg was associated with a 0.092-mL/min change
in GFR (P < 0.00 1 ). Thus, a reduction in MAP by 10
mm Hg, without a change in class of antihypertensive
medication, would lead to a decline in GFR of about
0.9 mL/min.
The above relationships were determined in patients
with a range of baseline GFR from 13 to 55 mL/min
per 1.73 m2 (mean GFR = 36.1 mL/min per 1.73 m2).
However, the magnitude ofthe effects ofprotein intake
and MAP on GFR was highly dependent on baseline
GFR, with a larger effect in patients with higher GFR.
For changes in protein intake, the effect is 0.94 mL/
mm per 1 .0 g/kg per day greater for each 10-mL/min
increase in baseline GFR (P = 0.002) (right half of
Table 5). For example, reducing protein intake by 0.4
g/kg per day would bead to a 1 .6-mL/min GFR decline
in a patient with an initial GFR of 50 mL/ mm per 1.73
m2 and a 0.3-mL/min GFR decline in a patient with an
initial GFR of 15 mL/min per 1 .73 m2. For changes in
blood pressure, the decline in GFR is 0.014 mL/min
per 1 .0 mm Hg greater for each 10 mL/min increase in
baseline GFR (P = 0.015). For example, a 10-mm Hg
decline in MAP would be associated with a 1 . 1 -mL/
mm GFR decline in a patient with an initial GFR of 50
mL/min per 1 .73 m2, compared with a 0.6-mL/min
GFR decline in a patient with an initial GFR of 15
mL/min per 1 .73 m2.
Controlling for changes in protein intake and blood
pressure, we found significant effects from changes in
a number of classes of antihypertensive agents (left
half of Table 5). Once again, the largest effect was
Effects of Diet and Blood Pressure Control on GFR
2104 Volume 7 . Number 10 . 1996
TABLE 5. Longitudinal model predicting changes (z�) in GFR from changes in protein intake,a mean arterialpressure (MAP),b and class of antihypertensive medicationsc over the first 2 years of follow�upd
Variable
Overall Interaction
GFR SE P Value GFR x 10 SEP
Value
Protein Intakee +2.77 0.51 <0.001 +0.94 0.30 0.002MAPS +0.09 0.01 <0.001 +0.01 0.01 0.02ACE Inhibitorse -2.18 0.65 <0.001 -0.59 0.42 0.16
.� f3 Blockerse -3.21 1.10 <0.003 -0.90 0.69 0.19
.� Ca Channel Blockerse +0.23 0.85 0.79 +0.80 0.60 0.18Diureticse -4.45 0.67 <0.001 - 1.05 0.45 0.02Otherse -1.71 0.95 0.07 -0.43 0.60 0.48
a Change in GFR (mL/min) per change in protein intake of 1 .0 g/kg per day (from food only in Study A and from food and supplements in Study
B).b Change in GFR (mL/min) per change in MAP of 1 .0 mm Hg.C Difference in GFR associated with starting versus stopping various classes of medications.a Overall coefficients (left half) refer to the change in GFR in a patient with an initial GFR of 36.1 mL/min per 1 .73 m2. Interaction coefficients (right
half) indicate the amount by which the overall coefficients change for every 10-mL/min per 1 .73 m2 difference in initial GFR from the mean of 36.1mL/min per 1 .73 m2. ACE, angiotensin-converting enzyme.e No significant interaction with follow-up interval.
associated with diuretics (a patient starting diuretics
experienced a 4.45-mL/min greater GFR decline than
a patient stopping diuretics [P < 0.001 1). Significant
effects were also associated with changes in n-block-
ers (a 3.21-mL/min greater GFR decline [P = 0.0041)
and ACE inhibitors (a 2. 18-mL/min greater GFR de-
dine [P < 0.00 1 1). There was no apparent association
with changes in calcium channel blockers, but the
association with changes in other classes of antihy-
pertensive medications was nearly significant (a 1.71-
mL/min faster GFR decline [P = 0.0551).
The association of changes in diuretics with
changes in GFR was greater in patients with higher
initial GFR. Compared with an average GFR of 36.1
mL/min, the decline in GFR was 1 .0 mL/min greater
for each 10 mL/min increase in baseline GFR (P =
0.02 1 ) (right half of Table 5). For example, the differ-
ence in GFR decline between patients starting and
stopping diuretics would be 5.9 mL/min if baseline
GFR is 50 mL/min per 1.73 m2, and 2.2 mL/min if
baseline GFR is 15 mL/min per 1.73 m2. The effects of
other antihypertensive agents (other than calcium
channel blockers) was also greater at higher initial
GFR, although they were not statistically significant
(P > 0.15 for each class).
After controlling for the initial level of GFR, we found
no effect for the duration of follow-up on the relation-
ship of changes in protein intake, blood pressure, or
medication class to changes in GFR (data not shown).
Thus the short-term changes in protein intake, blood
pressure, and class of antihypertensive agents af-
fected GFR not only in the first 4 months but also
throughout follow-up.
To determine the effect of measurement error (van-
ability In urine urea nitrogen and blood pressure
measurement) on these analyses, the multiple regres-
slon analyses for the first 4-month interval were per-
formed with and without controlling for measuremeiit
error. In general, the differences in the observed reba-
tionships were small. The effect of controlling for
measurement error was to increase slightly the asso-
ciation of changes in protein intake and blood pres-
sure and to decrease slightly the association of
changes in classes of antihypertensive agents (data
not shown).
Relationship of changes in protein intake, blood
pressure, and class of antihypertensive agents to the
differences in GFR between the diet and blood pres-
sure groups. To assess whether the observed relation-
ships account for the observed differences in GFR
between the diet groups and between the blood pres-
sure groups during the first 4 months in Studies A and
B, we compared the observed differences with those
predicted from the observed changes in protein in-
take, blood pressure and classes of antihypertensive
medications. As discussed earlier, because of mea-
surement error, the predicted differences are generally
expected to be less than the observed differences.
In Study A, the predicted mean difference in GFR
decline between diet groups was 1.34 mL/min, com-
pared with the observed difference of 1 .84 mL/min.
The predicted mean difference between blood pres-
sure groups was 0.59 mL/min, compared with the
observed difference of 1 .90 mL/min. The substantial
underestimation of the difference between the blood
pressure groups may be the result of hourly or daily
fluctuations in blood pressure that are not reflected by
measurement at a single visit every 4 months. It may
also be the result of other factors that were not
accounted for in these analyses, such as changes in
antihypentensive agents in the same class and to
changes in dosage without changes in the agent.
In Study B, both the predicted and observed differ-
ences in the Initial changes in GFR were small. The
predicted mean difference between diet groups was
0.48 mL/min, and the observed difference was -0.08
Modification of Diet in Renal Disease Study Group
Journal of the American Society of Nephrology 2105
mL/min. The predicted mean difference between
blood pressure groups was 0.3 1 mL/min, and the
observed difference was 0.02 mL/min.
DISCUSSION
These analyses show, for the first time, the indepen-
dent effects of changes in protein intake, blood pres-
sure, and different classes of antihypertensive agents
on GFR in a large number of patients with chronic
renal disease. GFR was influenced by increases or
decreases in protein intake and blood pressure,
whether antihypertensive regimens were stable or
changing. Starting diuretics, n-blockers, or ACE in-
hibitors was associated with a greater GFR decline
than was stopping them. The effect of changes in
protein intake, MAP, and diuretics was greater in
patients with higher initial GFR. These short-term
effects on GFR were not confined to the initial fol-
low-up period. Fluctuations in these factors contin-
ued to have short-term effects on GFR throughout
follow-up. Most importantly, these findings are con-
sistent with the hypothesis that these short-term
changes in GFR were caused by alterations in thehemodynamic determinants of GFR, rather than by
permanent changes in nephron structure. Although
the magnitude of these short-term effects is small, as
discussed below, they are potentially important in
clinical practice and in clinical trials.
The MDRD Study is the largest study of the effect of
dietary protein and blood pressure on the decline in
GFR in patients with chronic renal disease. A total of
840 patients with a variety of chronic renal diseases
were included. Baseline GFR ranged from 13 to 55
mL/min per 1.73 m2, with a mean of36. 1 mL/min per
1 .73 m2. During follow-up, GFR was measured seri-
ally using a standard method, the renal clearance of
‘251-iothalamate after subcutaneous injection (18,19).
Thus, we were abbe to assess accurately the relation-
ship of changes in protein intake and blood pressure
to changes in GFR. Moreover, because of the random
assignment to different diets and blood pressure
goals, we can infer cause and effect in the relationship
between these changes.
The MDRD Study was not designed, however, to
assess the effect of different classes of antihyperten-
sive agents on the decline in GFR. Nonetheless, a
variety of antihypertensive medications was used and
a large number of patients changed antihypertensive
medications during follow-up, which permits us to
assess the association between changes in GFR and
changes in classes of antihypertensive medications.
Assignment of cause and effect in these analyses is
more speculative; however, as discussed below, the
effects we observed are generally consistent with the
reported effects of antihypertensive agents.
The analyses correlating changes in GFR with
changes in protein intake and blood pressure can be
affected by errors in the measurement of each van-
able. We found little effect on the strength of the
observed relationships when controlling for measure-
ment error. However, our assessment of the effect of
measurement error did not include the effect of tran-
sient fluctuations (occurring in hours to days), errors
in collection of urine, errors in recording changes in
classes of antihypertensive medications, changes in
medications within the same class, or changes in
dosage of medications. Thus, it is likely that we have
not fully accounted for the effects of measurement
error. The true effects of changes in protein intake and
blood pressure are likely to be greater than our esti-
mates, and the true effects of changes in medication
classes are likely to be less than our estimates.
Our finding of small, but significant, effects on GFR
ofchanges in dietary protein (a 1. 1-mL/min reduction
in GFR after a 0.4-g/kg per day reduction in protein
intake), blood pressure (a 0.9-mL/min decline in GFR
after a 10-mm Hg reduction in MAP), and various
classes of antihypertensive agents (a 4.5-mL/min
greater GFR decline in patients starting versus those
stopping diuretics, a 3.2-mL/min greater GFR decline
in patients starting versus stopping �-bbockers, and a
2.2-mL/min greater GFR decline in patients starting
versus stopping ACE inhibitors) are consistent with
other studies over many years (3,4,25-36). Are these
small changes likely to be clinically important, and
how did they affect the outcome of the MDRD Study?
In clinical practice, changes in GFR of only 2 to 4
mL/min may have little or no consequences. However,
because the effects of changes in diet, blood pressure,
and antihypertensive agents are independent, they
are additive, and in some patients could be dramatic.
For example, in a group of patients with GFR of
36. 1 -mL/min per 1 .73 m2 (the mean GFR of the
MDRD Study population), a 0.4-g/kg per day reduc-
tion in dietary protein, and a 20-mm Hg reduction in
MAP after initiation of an ACE inhibitor, a �-bbocker,
and a diuretic, would bead to a mean decline in GFR by
7.8 mL/min within 4 months to a final value of 28.3
mL/min per 1.73 m2 (a 22% reduction). As stated
earlier, we may have underestimated the size of the
mean effect as a result of measurement error. More-
over, in some patients, the observed effect would be
even banger simply as a result of variability in the
population. The absolute (but not the percentage) GFR
decline would also be expected to be banger in patients
with higher baseline GFR. For example, if the relation-
ships that we observed can be extrapolated to patients
with normal GFR (125 mL/min pen 1.73 m2), a 0.4-
g/kg per day reduction in protein intake would lead to
a 4.4-mL/min GFR decline, a 20-mm Hg reduction in
MAP would bead to a 4.2-mL/min GFR decline, and
starting diuretics would lead to a 6.9-mL/min GFR
decline. However, the effects of starting an ACE inhib-
itor (a 1.1-mL/min GFR decline) and a �-bbocker (a
1.6-mL/min GFR decline) are not affected by baseline
GFR. Therefore, initiating all of these interventions in
a patient with a normal baseline GFR would lead to an
18.2-mL/min (14.6%) decline.
One practical implication of the short-term effects of
Effects of Diet and Blood Pressure Control on GFR
2106 Volume 7 . Number 10 - 1996
dietary protein restriction and antihypertensive then-
apy on GFR is that physicians should anticipate that
these interventions may be followed by a decrease in
renal function. In practice, physicians estimate GFR
from serum creatinine concentration. In principle, the
serum creatinine concentration would tend to rise
more after beginning antihypentensive therapy than
after dietary protein restriction than because of the
concomitant decrease in the urine creatinine excre-
tion rate with protein restriction (37). Furthermore,
the absolute rise in serum creatinine concentration
would be greater in patients with elevated baseline
serum creatinine concentration, because a decline in
GFR provokes a greater absolute increase in serum
creatinine concentration if baseline GFR is reduced.
Thus, a rise in serum creatinine concentration after
starting an antihypertensive agent is not necessarily
an adverse event and should not bead to “automatic”
withdrawal of the agent. Whether the agent should be
continued depends on the level of renal function, for
example, whether metabolic abnormalities on uremic
symptoms arise.
In the MDRD Study, the importance of these small
changes in GFR is best demonstrated by comparing
the hypothesized and observed effects of the interven-
tions. Based largely on studies of patients with dia-
betic nephropathy, the rate of GFR decline in the
MDRD Study was hypothesized to be 6 mL/min per yr
(10). In Study A (baseline GFR, 25 to 55 mL/min per
1 .73 m2), the hypothesized effect of the bow-protein
diet (compared with the usual protein diet) and the low
blood pressure goal (compared with the usual blood
pressure goal) was a 30% on greaten slowing in the rate
of GFR decline, equivalent to a difference between
groups of at least 1 .8 mL/min per yr. During a mean
follow-up interval of 3 yr. this would correspond to a
final difference between groups of at least 5.4 mL/
mm.
Instead, the observed mean rate of GFR decline in
Study A was 1 1 .5 mL/min pen 3 yr. equivalent to
3.8-mL/min peryn(12). At this rate, itwould take 12.6
yr for GFR to decline from 55 mL/min per 1 .73 m2 (the
upper limit of GFR in patients eligible for the MDRD
Study) to 7 mL/min pen 1 .73 m2 (the average GFR at
the onset of renal failure in the MDRD Study I 161).
Thus, the average rate of progression of renal disease
in the MDRD Study was slow. At these slower-than-
expected rates of progression, a 30% slowing in the
rate of GFR decline would correspond to a difference
between diet groups or between blood pressure groups
ofabout only 1 . 1 mL/min per yr or 3.3 mL/min over 3
yr.
In the first 4 months, the difference in GFR between
diet groups and blood pressure groups was 1 .84 and
1 .90 mL/min, respectively, but opposite in direction
to the hypothesized effects (Table 1 ). Correlational
analyses showed that the difference between diet
groups on between blood pressure groups attributable
to the short-term effects of changes in protein intake,
blood pressure, and class of antihypentensive agent
was at least 1.34 and 0.59 mL/min, respectively.
Although these effects were small, they were barge in
proportion to the hypothesized effects of the study
interventions.
The long-term effects (from the fourth month to the
end of follow-up) of the low-protein diet (a 28%-slower
GFR decline) and low blood pressure goal (a 29%-
slower GFR decline) were significant (P = 0.009 and
0.006, respectively) and almost as barge as the hypoth-
esized beneficial effects (Figures 1, A and B) (12).
However, the mean projected GFR decline from base-
line to 3 yr of follow-up was only 1 .2 mL/min less in
the bow-protein diet group (P = 0.3) and only 1.6
mL/min less in the low blood pressure group (P =
0. 18). Hence, considering the entire study population,
we could not detect a beneficial effect of either inter-
vention in Study A on the progression of renal disease.
In principle, given the slow rate of progression of renal
disease and the magnitude of the short-term and
long-term effects of the diet and blood pressure inter-
ventions that we observed, a longer follow-up would
have been required to detect a beneficial effect of the
interventions in Study A. On the other hand, in the
subgroup of patients with protemnuria, the rate of
progression was fasten, and a beneficial effect of the
low blood pressure goal was observed within the fob-
low-up period ( 12).
In Study B (baseline GFR between 13 and 24 mL/
mm pen 1 .73 m2), the correlational analyses also
suggested that the effect of changes in protein intake
and blood pressure on changes in GFR were opposite
in direction to the hypothesized effects, but smaller
than in Study A (Table 3). However, there was no
evidence of a significant deviation from linearity in
GFR decline in any of the diet or blood pressure
groups. Hence, the short-term effects had, at most, a
small effect on the outcome of Study B. Throughout
the entire follow-up period, the mean rate of GFR
decline was 19% less in the very-low-protein diet
group than in the low-protein diet group (P = 0.07)
and 1 2% bess in the low blood pressure group com-
pared with the usual blood pressure group (P = 0.28)
(Figures 1C and D).
Other studies of the progression of renal disease in
patients with baseline GFR at on above the range in
Study A also show short-term effects of the interven-
tions consistent with our observations. The effects are
most apparent in studies of ACE inhibitors on the
progression of diabetic nephropathy. In the study of
Bjorck et al. (38), 40 patients with Type 1 diabetes and
mean baseline GFR of 47 mL/min pen 1.73 m2 were
randomly assigned to treatment with ACE inhibitors
or f3-bbockers. During the first 6 months, patients
treated with ACE inhibitors had an approximately
2-mL/min faster GFR decline than patients treated
with �-bbockers, but a slower GFR decline thereafter.
By the end of the 2-yr mean follow-up interval, the
GFR decline was significantly less in patients treated
with ACE inhibitors. Other studies also report a faster
GFR decline soon after beginning ACE inhibitors,
Modification of Diet in Renal Disease Study Group
Journal of the American Society of Nephrology 2107
compared with �-bbockers, calcium channel blockers,
and other antihypertensive agents in patients with
diabetic nephropathy (39-46).
Consistent with our conclusion that short-term
changes in GFR reflect hemodynamic rather than
structural changes, Hansen et al. (47) and Apperboo et
al. (48) have suggested in diabetics and nondiabetics,
respectively, that the early decline in GFR during
antihypertensive treatment with ACE inhibitors is
reversible after discontinuing therapy months later.
We also reported similar effects after further manipu-
bation of diet and blood pressure during the extended
follow-up of the MDRD Study A cohort (49). On the
basis of these observations, we suggest that the ca-
pacity of the gbomerulus to respond to changes in
protein intake, blood pressure, and antihypertensive
agents by alteration in the determinants of single-
nephron GFR can interfere with the use of GFR to
assess the progression ofnenal disease. Because these
alterations appear to be greater at higher GFR, we
suggest that measuring GFR to assess the effects of
interventions to slow the progression of renal disease
may be less useful in patients with higher baseline
GFR.
In particular, the finding of a greater short-term
effect of ACE inhibitors in patients with higher base-
line GFR may account for the unexpected finding of
Lewis et al. (5) of a greater beneficial effect of ACE
inhibition to slow the decline of renal function in
diabetic nephropathy in patients with higher baseline
serum creatinine concentrations (lower baseline
GFR). Indeed, in subgroup analyses, no beneficial
effect was observed In patients with baseline serum
creatinine concentrations below 1 .5 mg/dL. However,
Mogensen and colleagues (50) have found a beneficial
effect of ACE inhibition on slowing the progression of
microalbuminuria to albuminuria in patients with
normal GFR. Thus, it seems likely that ACE Inhibition
is truly effective in slowing the progression of diabetic
nephropathy In patients with higher GFR. We suggest
that the apparent lack of benefit in patients with
higher GFR in the study ofLewis et al. (5) was, in part,
the result of greater difficulty in detecting the benefi-
dab effect in patients with higher baseline GFR.
In summary, the MDRD Study shows that changes
in protein intake, blood pressure, and antthyperten-
sive therapy have independent short-term effects on
GFR in patients with chronic renal disease. A change
in protein intake of 0.4 g/kg per day was associated
with a change in GFR of 1. 1 mL/min. A change in MAP
of 10 mm Hg was associated with a change in GFR of
0.9 mL/min. Starting diuretics, /3-blockers, or ACE
inhibitors was associated with a 4.4-, 3.2-, and 2.2-
mL/mln greater GFR decline, respectively, than was
stopping them. The effect ofchanges in protein intake,
MAP, and diuretics was greater in patients with higher
initial GFR. These short-term effects are likely caused
by hemodynamic adjustments in GFR rather than
renal structural Injury. Thus, protein intake, blood
pressure, and antihypertensive agents independently
modulate the level of GFR in chronic renal disease.
Dietary protein restriction, blood pressure control,
and specific classes of antihypertensive drugs are
used to slow the progression of chronic renal diseases
in practice and in clinical trials. Although their short-
term effects on GFR are small, they are independent of
one another and are additive; hence, clinically signif-
icant changes in renal function can occur in patients
undergoing multiple interventions. In clinical trials,
these short-term effects are opposite in direction to
their hypothesized beneficial effects on the progres-
sion of renal disease, and large enough to confound
the results of the trial. In principle, confounding is
most likely to occur if the rate of progression of renal
disease is slow, if the duration of follow-up is short, or
if the baseline bevel of GFR is high. Future studies
should focus on specific renal diseases or subgroups
of patients with faster rates of progression, anticipate
the possible short-term effects of the interventions on
GFR, and appropriately extend the duration of follow-
up. For patients with slowly progressing renal dis-
eases, it may be not be practical to study the effects of
these interventions on the rate of decline in GFR.
Instead, it may be necessary to develop other mea-
sures to detect the progression of renal disease.
ACKNOWLEDGMENTS
This study was supported by the National Institute of Diabetes.
Digestive and Kidney Diseases and the Health Care Financing Admin-istration. We are indebted to the patients who participated in theModification of Diet in Renal Disease Study. Marion Merrell Dow,Kansas City, Missouri, donated diltiazem (Cardizem�), and MerckSharp and Dohme, West Point. Pennsylvania. donated enalapril
(Vasotec�) to study participants.
REFERENCES
1 . Smith HW. Diseases of the kidney and urinary tract. In:The Kidney: Structure and Function in Health and Dis-ease. New York: Oxford University Press; 1951:836-887.
2. Levey AS: Assessing the effectiveness of therapy to pre-vent the progression of renal disease. Am J Kidney Dis1993;22:207-214.
3. Premen AJ: Potential mechanisms mediating postpran-dial renal hyperemia and hyperfiltration. FASEB (Fed AmSoc Exp Biol) J 1988;2:131-137.
4. Keane WF, Anderson S, Aurell M, de Zeeuw D, NarinsRG, Povar G: Angiotensin converting enzyme inhibitorsand progressive renal insufficiency: Current experienceand future directions. Ann Intern Med 1989;l 11:503-516.
5. Lewis EJ, Hunsicker LG, Bain RP Rohde 1W, for theCollaborative Study Group: The effect of angiotensmn-converting-enzyme inhibition on diabetic nephropathy.N Engl J Med 1993;329:1456-1462.
6. Peterson JC, Adler 5, Burkant JM, Ct at.: Blood pressurecontrol, proteinunia, and the progression of renal dis-ease: The Modification of Diet in Renal Disease StudyGroup: Ann Intern Med 1995; 123:754 -762.
7. Pedrinl MT. Levey AS, Lau J, Chalmers TC, Wang PH:The effect of dietary protein restriction on the progres-sion of diabetic and non-diabetic renal disease: A meta-analysis. Ann Intern Med 1996; 124:627-632.
8. Maschio G, Abberti D, Janin G, et al., the Angiotensin-Converting-Enzyme Inhibition in Progressive Renal In-sufficiency Study Group: Effect of angiotensmn-convert-ing enzyme inhibitor benazepril on the progression of
Effects of Diet and Blood Pressure Control on GFR
2108 Volume 7 . Number 10 - 1996
chronic renal insufficiency. N Engl J Med 1996;334:939-945.
9. Klahr 5: The Modification of Diet in Renal Disease Study.N Engl J Med 1989:320:864-866.
10. Modification of Diet in Renal Disease Study Group(prepared by Beck GJ, Berg RL, Coggins CH, GassmanJJ, Hunsicker LG, Williams GW): Design and statisticalIssues of the Modification of Diet in Renal Disease Trial.Controlled Clin Trials 199 1 ; 1 2:566 -586.
1 1 . Modification of Diet in Renal Disease Study Group: TheModification of Diet in Renal Disease Study: Design,methods, and results from the Feasibility Study. Am JKidney Dis 1992;20:18-33.
12. Klahr S. Levey AS, Beck GJ, et al., the Modification ofDiet in Renal Disease Study Group: The effects of dietaryprotein restriction and blood pressure control on theprogression of renal disease. N EngI J Med 1994;330:877-884.
13. Joint National Committee: 1988 Report of the JointNational Committee in Detection, Evaluation and Treat-ment ofHigh Blood Pressure. Arch Intern Med 1988; 148:1023-1038.
14. Maroni BJ, Steinman TI, Mitch WE: A method for esti-mating nitrogen intake of patients with chronic renalfailure. Kidney Int 1985;27:58-65.
15. Frisancho AR. Instructions for using the frameter,weight, fat guide and anthropometnic standards. AnnArbor, MI: Health Products; 1989:29-30.
1 6. Modification of Diet in Renal Disease Study Group(prepared by Levey AS, Adler 5, Caggiula AW, et al.):Effects ofdietary protein restriction on the progression ofadvanced renal disease in the Modification of Diet inRenal Disease Study. Am J Kidney Dis 1996;27:652-663.
1 7. Modification of Diet in Renal Disease (MDRD) StudyGroup (presented by Gassman JJ, Berg RL, ByingtonRP, Hebert LA, Lambdin KE, Tanna A): Selecting a pointestimator of blood pressure from consecutive measure-ments lAbstractl. Controlled Clin Trials 1991;12:709-710.
18. Perrone R, Steinman T, Beck GJ, et at., the Modificationof Diet in Renal Disease Study Group: Utility of radioiso-topic filtration markers in chronic renal insufficiency:Simultaneous comparison of ‘251-iothalamate, ‘#{176}#{176}Yb-DTPA, #{176}#{176}mTc-DTPAto inulin. Am J Kidney Dis 1990;16:224 -235.
19. Levey AS, Greene T, Schluchter MD, et at., the MDRDStudy Group, the DCCT Research Group: Glomerularfiltration rate measurements in clinical trials. J Am SocNephrol 1993;4:1 159-1171.
20. Kleinbaum DG, Kupper LL, Muller KE. Applied Regres-sion Analysis and Other Multivanabbe Methods. Boston:PWS-KENT; 1988:102-180.
2 1 . Jennrich RI, Schbuchter MD: Unbalanced repeated-measures models with structured covariance matrices.Biometrics 1 986:42:805- 820.
22. Cressie N. Statistics for Spatial Data. New York: JohnWiley & Sons; 1991.
23. Liu K: Measurement error and its impact on partialcorrelation and multiple linear regression analyses.Am J Epidemiol 1988:127:864-873.
24. Fuller WA. Measurement Error Models. New York: Wiley;1987:103-124.
25. Pullman TN, Alving AS, Dern RJ, Landowne M: Theinfluence of dietary protein intake on the specific renalfunction in normal man. J Lab Clin Med 1954:44:320-332.
26. Woods JW, Blythe WB: Management of malignant hy-pertension complicated by renal insufficiency. N EngIJ Med 1967:277:57-61.
27. Reams GP: Renal effects of antihypertensive drugs. In:Bennett WM, McCarron DA, Stein JH, Eds. Contempo-rary Issues in Nephrology: 28. Pharmacology and Man-agement of Hypertension. New York: Churchill Living-stone; 1994:29-53.
28. Shannon JA, Jobliffe N, Smith HW: The excretion ofurine in the dog: IV. The effect of maintenance diet,feeding, etc. upon the quantity of glomerular filtrate.
Am J Physiol 1932:101:625-638.29. Pitts RF: The effects of infusing glycine and varying the
dietary protein intake on renal hemodynamics in thedog. Am J Physiol 1944:142:355-377.
30. King AJ, Levey AS: Dietary protein and renal function. JAm Soc Nephrol 1993:3:1723-1737.
3 1 . Shipley RI, Study RS: Changes in renal blood flow,extraction of inulin, glomerufar filtration rate, tissuepressure and urine flow with acute alterations in renalartery blood pressure. Am J Physiol 1951:167:676-688.
32. van Brummelen P, Woerlee M, Schalekamp MADH:Long-term versus short-term effects of hydorchobo-rothiazide on renal hemodynamics in essential hyper-tension. Clin Sci 1979:56:463-469.
33. Bauer JH, Reams GP. Beta-adrenergic antagonists andthe kidney. In: Bennett WM, McCarron DA, Brenner BM,Stein JH, Eds. Contemporary Issues in Nephrology: 17.Pharmacotherapy of Renal Disease and Hypertension.New York: Churchill Livingtone; 1987:223-254.
34. Anderson S. Converting enzyme inhibitors. In: BennettWM, McCarron DA, Brenner BM, Stein JH, Eds. Contem-porary Issues in Nephrology: 1 7. Pharmacotherapy ofRenal Disease and Hypertension. New York: ChurchillLivingtone; 1987:277-297.
35. Murphy BF, Whitworth JA, Kincaid-Smith P: Renalinsufficiency with combinations of angiotensmn convert-ing enzyme inhibitors and diuretics. Br Med J 1984;288:844-845.
36. Davidson RC, Bursten SL. Calcium channel blockers inhypertension. In: Bennett WM, McCarron DA, BrennerBM, Stein JH, Eds. Contemporary Issues in Nephrology:1 7. Pharmacotherapy of Renal Disease and Hyperten-sion. New York: Churchill Livingtone; 1987:299-322.
37. Modification of Diet in Renal Disease Study Group(prepared by Levey AS, Bosch JP, Coggins CH, GreeneT, Mitch WE, Schluchter MD): Effects of diet and bloodpressure on creatinine clearance and serum creatinineconcentration in the MDRD Study. J Am Soc Nephrol1996:7:556-565.
38. Bjorck 5, Mubec H, Johnsen SA, Norden G, Aureil M:Renal protective effect of enalapril in diabetic nephrop-athy. Br Med J 1992:304:339-343.
39. Lebovitz HE, Wiegmann TB, Cnaan A, et at: Renalprotective effects of enalapril in hypertensive NIDDM:Role ofbaseline albuminuria. Kidney Int 1994;45ISuppl451:S 150-S 155.
40. Pawing HH, Hommel E, Smidt UM: Protection of kidneyfunction and decrease in albuminuria by captopril ininsulin-dependent diabetics with nephropathy. Br Med J1988;297: 1086-109 1.
4 1 . Romero R, Salinas I, Lucas A, Tebddo J, Audi L,Sanmarti A: Comparative effects of captopril versusnifedipine on proteinuria and renal function of type IIdiabetic patients. Diabetes Res Clin Pract 1 992; 17:191-198.
42. Romero R, Salinas I, Lucas A, et at: Renal functionchanges in microalbuminuric normotensive type II dma-betic patients treated with angiotensmn-converting en-zyme inhibitors. Diabetes Care 1993:16:597-600.
43. Lacourciere Y, Nadeau A, Poirier L, Tancrede G: Capto-pril or conventional therapy in hypertensive type II dia-betics: Three-year analysis. Hypertension (Dallas) 1993:21:786-794.
44. Bauer JH, Reams GP, Hewett J, Klachko D: A random-ized, double-blind, placebo-controlled trial to evaluatethe effect of enalapril in patients with clinical diabeticnephropathy. Am J Kidney Dis 1992:20:443-457.
45. Slataper R, Vicknair N, Sadler R, Bakris GL: Compara-tive effects of different antihypertensive treatments on
progression of diabetic renal disease. Arch Intern Med1993; 153:973-980.
46. Bakris GL, Barnhill BW, Sadler R: Treatment of arterialhypertension in diabetic man: Importance of therapeuticselection. Kidney Int 1992:41:912-919.
47. Hansen HP, Rossing P, Tarnow L, Jielsen PS, Jensen B,Pawing HH: Elevated GFR after withdrawal of chronicantihypertensive treatment in diabetic nephropathy LAb-stractl. J Am Soc Nephrol 1994;5:378A.
Modification of Diet in Renal Disease Study Group
Journal of the American Society of Nephrology 2109
48. Apperboo AJ, de Zeeuw D, van Essen G, de Jong PE: Theinitial GFR decline on antihypertensive treatment inlong-term intervention trials is reversible after with-drawal of treatment [Abstractl. J Am Soc Nephrob 1994:5:324A.
49. ModificatIon of Diet in Renal Disease Study Group(prepared by Levey AS, Beck GJ, Caggiuba AW, GreeneT, Kusek JW, Striker GE, Klahr 5): Trends toward a
beneficial effect of a low protein diet during additionalfollow-up in the Modification of Diet in Renal DiseaseStudy IAbstractl. J Am Soc Nephrol 1994;5:336A.
50. Mogensen CE, European Microalbuminura CaptoprilStudy Group: Effect of captopril on progression to overtclinical proteinuria in patients with renal disease ininsulin-dependent diabetes mellitus. JAMA 1994:271:275-279.