Volume 54, Issue 6 , Pages 1052-1061, December 2009
Effect of a Low- Versus Moderate-Protein Diet on Progression of CKD: Follow-up of a Randomized Controlled Trial
Article Outline
Background
Whether low-protein-diet (LPD) as opposed to moderate-protein-diet (MPD) regimens improve the long-term survival of patients with chronic kidney disease (CKD) or induce protein-caloric malnutrition is unknown.
Study Design
Intention-to-treat analysis of follow-up data from a randomized controlled trial.
Setting & Participants
423 patients with CKD (stages 4-5) were randomly assigned between January 1999 and January 2003 and followed up until December 2006 or death. The first phase of follow up was from January 1999 to June 2004; additional follow-up was from July 2004 to December 2006.
Intervention
LPD versus MPD (protein intake, 0.55 vs 0.80 g/kg/d).
Outcomes
Protein-caloric malnutrition (defined as the occurrence of 1 of the following: loss of body weight > 5% in 1 month or 7.5% in 3 months or body mass index < 20 kg/m2 with serum albumin level < 3.2 g/dL and normal C-reactive protein level [<0.5 mg/dL]), dialysis, death, or the composite outcome of dialysis and death.
Results
Baseline mean age was 61 years, estimated glomerular filtration rate was 16 mL/min/1.73 m2, proteinuria had protein excretion of 1.67 g/d, body mass index was 27.1 kg/m2, protein intake was 0.95 g/kg/d, and there were 57% men. Duration of follow-up was 32 months (median, 30 months; 25th-75th percentiles, 21-39). Average protein intakes were 0.73 ± 0.04 g/kg/d for the LPD and 0.9 ± 0.06 g/kg/d for the MPD. 3 patients (0.7%) met criteria for protein-caloric malnutrition. 48 patients died (11%), 83 initiated dialysis therapy (20%), and 113 (27%) reached the composite outcome. In unadjusted Cox survival analyses, effects of the LPD on these outcomes were 1.01 (95% CI, 0.57-1.79), 0.96 (95% CI, 0.62-1.48), and 0.98 (95% CI, 0.68-1.42), respectively.
Limitations
Low event rates for dialysis therapy initiation and death.
Conclusions
Most patients, who were regularly followed up in CKD clinics, were acceptably adherent to the prescribed dietary protein intake restrictions; the LPD and MPD did not lead to protein wasting; and the LPD did not decrease the risk of death or dialysis therapy initiation compared with the MPD.
Index Words: Chronic kidney disease, low-protein diet, outcomes
The effect of dietary protein restriction on the survival of patients with chronic kidney disease (CKD) is unknown. The rationale for such an effect is based on the ability of protein restriction to control several mechanisms of disease that are believed to be responsible for the high morbidity and mortality of patients with failing kidneys. Decreased protein, sodium, and phosphate intake is able to reduce the accumulation of nitrogenous compounds; optimize serum levels of bicarbonate, potassium, and phosphate1, 2, 3, 4; prevent the development of severe secondary hyperparathyroidism5, 6 and resistant hypertension7, 8; and reduce proteinuria9, 10 and the degree of anemia.11
A large body of evidence indicates that advanced stages of CKD, including end-stage renal disease requiring renal replacement therapies, are characterized by high rates of adverse outcomes.12 However, it still is uncertain whether achievement of the clinical and metabolic goals indicated by international guidelines with dietary interventions decreases the risk of dialysis therapy and death in patients with CKD.
Only recently, the Modification of Diet in Renal Disease (MDRD) Study Group has investigated the effects of a low-protein diet (LPD) on the onset of kidney failure and mortality through extended follow-up after trial completion.13 This study explored whether assignment to an LPD (0.58 g/kg/d) compared with the usual-protein diet (1.3 g/kg/d) influenced the risk of reaching end-stage renal disease and the composite outcome of kidney failure requiring dialysis therapy and all-cause mortality. However, the major limitation of this study is the absence of clinical follow-up and dietary information after trial completion.
We have recently reported data from a randomized controlled trial of patients with CKD stages 4-5 (estimated glomerular filtration rate [eGFR] < 30 mL/min/1.73 m2) of the metabolic effects of 2 diet regimens based on different protein content: 0.55 and 0.80 g/kg/d.4 In the present study, we report results of a 48-month follow-up phase of the initial trial. This study sought to determine: (1) whether the risk of malnutrition, the major drawback of an LPD, is considerable and increases as protein intake decreases; (2) the extent to which patients adhere to the prescribed diet regimen over time; and (3) whether patient outcomes are affected by diet prescriptions and nutritional status through improved metabolic control and achievement of the National Kidney Foundation's Kidney Disease Outcomes Quality Initiative (KDOQI) target goals for 7 main cardiovascular risk factors.14, 15, 16, 17 The design of the study allowed us to test whether dietary prescription had a carryover effect after dialysis therapy was started.
Methods
Study Design
The study was conducted at the CKD Clinic of the University Federico II of Naples, Italy, where 753 consecutive patients (CKD stages 2-5) were screened from January 1999 to January 2003. Details of the study have been described previously.4 The study was approved by the local medical ethics committee. Briefly, the following enrollment criteria were used: aged 18 years and stable kidney function with basal eGFR < 30 mL/min/1.73 m2. After a monthly check of eGFR for 3 months (run-in), patients with stable kidney function (eGFR variability < 15%) were deemed eligible for the study. During this period, patients maintained their previous diet. Renal function was expressed as eGFR calculated using the 6-variable MDRD Study equation.18
At baseline, 423 patients were enrolled in the study (Fig 1) and randomly assigned to 1 of the 2 test diets: 0.55 g of protein per kilogram of ideal body weight per day (LPD) or 0.80 g of protein per kilogram of ideal body weight per day (moderate-protein diet [MPD]). A simple randomization list was generated using a computer and kept concealed using numbered, opaque, sealed envelopes opened in sequence by administrative staff personnel not involved in patient care. Amounts of protein in the 2 diets were chosen on the basis of the minimum protein requirement (0.55 g/kg/d) for the general population as recommended by the World Health Organization19 and the protein requirement of 97.5% of the adult population (0.75 g/kg/d), to which 0.05 g/kg/d was added.20 All patients received multivitamin and calcium supplementation, but no keto analogues were prescribed. Iron was given if necessary, and sodium intake was restricted to 2.5 g/d (sodium chloride, 5 g/d). Details of the prescribed diets are available upon request.
Figure 1.
Patient flow diagram shows selection and discontinuation according to groups. The 1st period (January 1999-June 2004) corresponds to the duration of the randomized controlled trial, and the 2nd period (July 2004-December 2006) corresponds to long-term follow-up.
Demographic and clinical data were obtained at baseline. Primary kidney disease was classified according to European Renal Association codes. Coexisting comorbidity was indexed for all patients using the Charlson method.21
After randomization, patients were instructed to modify their intake of protein, sodium, phosphorus, and, if necessary, calories to achieve the goals of the assigned diet. Dietary instructions and verification of adherence to the prescribed diet were accomplished by an expert renal dietitian who followed up all patients included in the study at each CKD clinical visit. Patients in the first trial (recruited from January 1999-January 2003) were followed up for a maximum of 18 months.4
Protein Intake Estimation and Malnutrition
Dietary protein intake was estimated under stable noncatabolic conditions using daily urinary excretion of urea nitrogen.22 Adherence to calorie prescription was verified at each visit by the dietitian and by careful monitoring of body weight variation. Twenty-four–hour urinary creatinine excretion was measured as an indicator of muscle mass loss.23
Protein-calorie malnutrition was defined as the occurrence of 1 of the following: loss of body weight > 5% in 1 month or 7.5% in 3 months or body mass index < 20 kg/m2 with serum albumin level < 3.2 g/dL and normal C-reactive protein level (<0.5 mg/dL).
Long-term Follow-up
The 423 patients were followed up at the same institution from randomization until December 31, 2006, or death. Per protocol, study participants entered a 30-month extension phase of the initial 18-month trial.4 Laboratory chemistry tests were planned at randomization and every 6 months thereafter, including the extension phase, until dialysis therapy start or at the study end date. Assays included serum urea nitrogen, creatinine, phosphate, calcium, intact parathyroid hormone, total cholesterol, high- and low-density lipoprotein cholesterol, triglycerides, hemoglobin, albumin, and transferrin. Samples for urea nitrogen, creatinine, phosphate, sodium, potassium, and proteins were obtained from 24-hour urine collections. Standard laboratory procedures were used for blood and urinary measurements.4 Trained study staff were responsible for laboratory testing and follow-up visits. To avoid missing information, staff responsible for follow-up contacted participants twice before each visit. New appointments were arranged within 2 weeks for those who failed to show up.
Participants followed the assigned diet and received the full clinical and dietary assessment of the previous study until renal replacement therapy became necessary. Malnutrition was monitored at each visit. Pharmacologic and nonpharmacologic therapies were used to achieve the desired therapeutic targets for hemoglobin (>11 g/dL), parathyroid hormone (70-110 pg/mL for CKD stage 4; 150-300 pg/mL for CKD stage 5), calcium (8.4-10.2 mg/dL), phosphate (3.5-5.5 mg/dL), arterial pressure (<130 mm Hg systolic and <80 mm Hg diastolic), and low-density lipoprotein cholesterol (<100 mg/dL).14, 15, 16, 17
The need to start dialysis therapy was not strictly defined. However, because this was a single-center study, criteria were those in use in our center and were homogeneous for both groups studied: eGFR < 10 mL/min/1.73 m2, hyperkalemia, intractable fluid overload or hypertension, weight loss, or other evidence of malnutrition.24 Of patients who needed renal replacement therapy, none received a preemptive transplant, 3%-4% from both groups underwent peritoneal dialysis therapy, and the remaining received standard bicarbonate hemodialysis.
Statistical Analysis
Descriptive StatisticsData are expressed as mean ± standard deviation, median and quartile (Q) 1 to Q3, or frequencies.
Linear ModelsLinear mixed models (with random intercepts and slopes specific to each participant) were used to study the relationship between diet program and repeated longitudinal measurements of eGFR, serum urea nitrogen, protein intake, and proteinuria. The validity of such models was verified using formal and graphical testing.
Time-to-Event ModelsWe considered both dialysis therapy initiation and death as outcomes and the possibility that they may be correlated within the same study participant, fitting a marginal competing-risk model25 in addition to standard Cox regression to the combined end point, and a Cox mortality model with dialysis status as the time-varying covariate (absent before dialysis therapy initiation and present thereafter). In both separate single-event models and the competing-risk model, covariates considered for adjustment included time-invariant clinical characteristics and Charlson comorbidity score at randomization and time-varying values for biochemical and urinary data during follow-up. For these last covariates, effects were estimated for current values, previous-visit values, and change from previous values. The largest possible meaningful model considered included clinical and biochemical variables on the basis of their plausible univariable relation to the outcome, considering the overall model fit and hazard proportionality. Variables to be dropped as nonconfounders were eliminated manually, monitoring variations of the exposure regression coefficient and giving validity precedence over precision.26 The survival effect of variables that may be (at least in part) in the causal chain between kidney function decrease and the outcome (eg, phosphate level) was studied treating eGFR as a covariate or stratifying variable. The contribution of covariates to explain the dependent variable was assessed using a 2-tailed Wald test, with P < 0.05 considered significant. Model specification, proportionality assumption, and overall fit were checked using re-estimation, formal and graphical tests based on residuals, and testing the interaction with time of the variables in the model.
Power AnalysisAssuming an average multiple-event rate per year of 10% in controls (dialysis therapy start and death), we estimated that a 4-year follow-up study of 400 participants would have power to 90% to detect as significant at a 2-sided P < 0.05 a relative risk for dialysis therapy or death of at least 0.5 (LPD vs MPD) if such an effect existed. Analyses were performed using Stata 9.2 SE (Stata Corp, College Station, TX).
Results
Baseline Characteristics
Table 1 lists baseline characteristics of the study population. Mean age of the cohort was 61 ± 17 years, 57% were men, and 12% had a history of type 2 diabetes. Mean baseline eGFR was 16 ± 7 mL/min/1.73 m2, protein intake was on average 0.95 ± 0.11 g/kg/d, and mean urinary protein excretion was 1.67 ± 2.34 g/d (median, 0.74 g/d; Q1 to Q3, 0.42-1.95). There was no difference between the 2 groups for the distribution of CKD stages, renal function, age, sex, protein intake, and diabetes. Similar baseline values were found for the main clinical and biochemical parameters (serum albumin, hemoglobin, serum phosphate, body mass index, and systolic and diastolic blood pressure). No difference was observed in the use of erythropoiesis-stimulating agents, angiotensin-converting enzyme inhibitors, and angiotensin receptor blockers.
Table 1. Baseline Patient Characteristics
| All | Low-Protein Diet | Moderate-Protein Diet | |
|---|---|---|---|
| No. of patients | 423 | 212 | 211 |
| Age (y) | 61 ±17 | 61±16 | 62±18 |
| Men | 240(57) | 122(58) | 118(56) |
| Weight (kg) | |||
| Men | 78±17 | 77±17 | 78±16 |
| Women | 66±14 | 68±14 | 64±15 |
| Body mass index (kg/m2) | 27.1±4.9 | 27.0±4.7 | 27.3±5.0 |
| Chronic kidney disease stage | |||
| 4 | 214(51) | 104(49) | 110(52) |
| 5 | 209(49) | 108(51) | 101(48) |
| eGFR (mL/min/1.73 m2) | 16±7 | 16±6 | 17±8 |
| Diabetes | 51(12) | 24(11) | 27(13) |
| Erythropoiesis-stimulating agent | 47(11) | 24(11) | 23(11) |
| ACE inhibitor and/or ARB | 188(44) | 97(46) | 91(43) |
| Protein intake (g/kg/d) | 0.95±0.11 | 0.96±0.13 | 0.94±0.09 |
| Serum albumin (g/dL) | 3.9±0.6 | 4.0±0.6 | 3.9±0.7 |
| Urinary protein excretion (g/d) | 1.67±2.34 | 1.53±2.22 | 1.79±2.44 |
| Systolic arterial pressure (mm Hg) | 138±27 | 140±25 | 136±30 |
| Diastolic arterial pressure (mm Hg) | 77±19 | 74±21 | 81±16 |
| Hemoglobin (g/dL) | 11.8±1.5 | 11.8±1.4 | 11.7±1.6 |
| Serum phosphate (mg/dL) | 4.3±0.9 | 4.5±1.0 | 4.0±0.7 |
Adherence to the Dietary Assignment and Progression of Renal Disease
Considering the entire period of follow-up for each patient, including the first trial, the 2 groups of patients maintained significantly different protein intakes (LPD, 0.73 ± 0.04 g/kg/d; MPD, 0.90 ± 0.06 g/kg/d; P < 0.05), with a difference between the 2 groups of 0.17 ± 0.05 g/d, which lasted from month 6 until the study end date (Fig 2A). The percentage of patients assigned to the LPD who had protein intake ≤ 0.7 g/kg/d in all follow-up visits varied from 25%-30%. Similarly, serum urea nitrogen levels, which represented the primary outcome of the first trial, increased in both groups during follow-up, but were significantly lower with the LPD, with a mean difference of 7.2 ± 2.0 mg/dL starting from month 6 (Fig 2B). Conversely, mixed models of eGFR and proteinuria failed to detect a difference in slope by group assignment (Fig 2C and D). Of note, monthly decreases in eGFRs were 0.19 ± 0.48 mL/min/1.73 m2 in the LPD group and 0.18 ± 0.46 mL/min/1.73 m2 in the MPD group.
Figure 2.
Comparison of protein intake and laboratory values in the low-protein diet (LPD; solid line) and moderate-protein diet (MPD; dashed line) groups during follow-up; time (T) in months. (A) Protein intake (g/kg/d), (B) serum urea nitrogen level (mg/dL), (C) glomerular filtration rate (GFR; mL/min/1.73 m2), and (D) urinary protein excretion (g/d) during the entire follow-up period. Numbers adjacent to each line show participants in that group who were followed up at the time. **P < 0.01 for the effect of the LPD on the protein intake trend. *P < 0.05 for the effect of the LPD on the SUN level trend. Conversion factors for units: serum urea nitrogen in mg/dL to mmol/L, ×0.357; GFR in mL/min/1.73 m2 to mL/s/1.73 m2, ×0.01667.
Nutritional Parameters During Follow-up
Table 2 lists data concerning nutritional parameters of the 2 groups collected during follow-up. Both groups maintained body weight and 24-hour urinary creatinine excretion similar to the basal value during the entire observation period. No differences were observed for serum albumin and transferrin values between groups, and their values did not change during follow-up. Low-density lipoprotein cholesterol values decreased significantly in the LPD group, but not the MPD group (Table 2). During the entire observation period of the study, only 3 of 423 patients met the predefined criteria for protein-calorie malnutrition: 1 patient (assigned to MPD) had weight loss > 5% in 1 month and 2 patients (assigned to LPD) reached a body mass index < 20 kg/m2 with a serum albumin level < 3.2 g/dL.
Table 2. Main Nutritional Parameters of Patients in the 2 Diet Groups
| Body Weight (% of baseline) | 24-h Urinary Creatinine (% of baseline) | Serum Albumin (g/dL) | LDL Cholesterol (mg/dL) | Transferrin (mg/dL) | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| LPD | MPD | LPD | MPD | LPD | MPD | LPD | MPD | LPD | MPD | |
| Baseline (423) | — | — | — | — | 4.0±0.6 | 3.9±0.7 | 125±43 | 124±40 | 242±42 | 241±51 |
| Month 6 (404) | 99.7±6.6 | 100.3±4.5 | 97.6±37.7 | 101.8±29.8 | 4.1±0.5 | 4.0±0.6 | 118±42 | 122±41 | 235±46 | 238±47 |
| Month 12 (390) | 99.8±7.7 | 100.6±7.0 | 98.6±27.8 | 98.5±27.9 | 4.0±0.4 | 4.0±0.5 | 118±37 | 120±37 | 234±49 | 240±50 |
| Month 18 (376) | 99.8±8.0 | 100.9±6.3 | 94.7±29.3 | 97.7±33.1 | 4.1±0.4 | 4.0±0.4 | 116±33 | 115±33 | 235±40 | 241±42 |
| Month 24 (359) | 100.1±8.1 | 100.9±5.7 | 96.5±27.0 | 94.6±26.1 | 4.2±0.4 | 4.1±0.4 | 111±32a | 123±36 | 232±37 | 234±48 |
| Month 30 (331) | 99.8±8.0 | 100.8±5.3 | 95.3±26.9 | 93.9±28.1 | 4.2±0.4 | 4.1±0.4 | 110±31a | 121±36 | 230±36 | 233±41 |
| Month 36 (293) | 100.0±8.0 | 101.2±6.2 | 90.1±28.1 | 93.2±26.6 | 4.1±0.4 | 4.2±0.4 | 118±27 | 124±34 | 228±04 | 241±48 |
| Month 42 (271) | 100.0±8.2 | 101.6±6.8 | 89.3±27.9 | 92.8±27.4 | 4.2±0.4 | 4.2±0.4 | 112±31a | 121±34 | 226±42 | 236±47 |
| Month 48 (261) | 99.9±8.4 | 101.8±7.3 | 90.1±24.4 | 91.9±26.5 | 4.2±0.4 | 4.1±0.4 | 113±29a,b | 111±37 | 215±46 | 225±52 |
aP < 0.05, statistically different from baseline. |
bP < 0.05, trend of LPD statistically different from MPD. |
Renal and Patient Survival
After an average follow-up of 32 months (median, 30 months; Q1 to Q3, 21-39), 48 patients died: 23 (11%) in the LPD group and 25 (12%) in the MPD group, with a median time to death of 27 months (Q1 to Q3, 18-37). Thirty patients died before the beginning of dialysis therapy. During follow-up, 83 participants required dialysis therapy: 41 (19%) patients in the LPD group and 42 (20%) patients in the MPD group. Average survival on dialysis therapy was 12 ± 10 months (median, 13 months; Q1 to Q3, 4-20). Average eGFR at the start of dialysis therapy was 7 ± 2 mL/min/1.73 m2 (median, 6 mL/min/1.73 m2; Q1 to Q3, 5-10) for the LPD group and 7 ± 3 mL/min/1.73 m2 (median, 7 mL/min/1.73 m2; Q1 to Q3, 5-11) for the MPD group. Cumulative incidences of death and dialysis therapy start were unaffected by the diet regimen, with 113 patients reaching the composite outcome: 56 (26%) patients in the LPD group and 57 (27%) patients in the MPD group.
Figure 1 shows the event number during the metabolic study (4 events; January 1999-June 2004) and follow-up study (July 2004-December 2006). Only 11% of participants died during the study, with 66% and 64% remaining dialysis free or death free in the LPD and MPD groups, respectively.
Kaplan-Meier curves by randomized groups are shown in Fig 3. No difference was found in terms of survival probabilities from the analysis of all-cause mortality, dialysis therapy initiation, or the composite outcome (dialysis therapy or death). Baseline covariate adjustment including time-varying covariates did not modify the effect of diet assignment on risk of mortality, dialysis therapy initiation, and composite outcome observed in unadjusted analyses (Table 3). The start of dialysis therapy significantly increased the risk of death (hazard ratio [HR], 2.81; 95% confidence interval [CI], 1.56-5.05). However, in this time-dependent Cox model, the effect of the diet regimen on patient death remained the same. Similar findings were obtained from the competing-risk model of dialysis therapy initiation and death, in which participants remained at risk of death after the first event (dialysis therapy initiation) occurred. In this multiple-event model, comorbidity index (HR, 1.15; 95% CI, 1.01-1.33) and age (HR, 1.06; 95% CI, 1.03-1.09) were associated with death only, whereas serum phosphate level (HR, 1.32; 95% CI, 1.06-1.66), eGFR (HR, 0.94; 95% CI, 0.91-0.97), erythropoiesis-stimulating agent therapy (HR, 2.14; 95% CI, 1.39-3.29), and use of either angiotensin-converting enzyme inhibitors or angiotensin receptor blockers (HR, 0.56; 95% CI, 0.35-0.90) were associated with both the risk of dialysis therapy initiation and death. However, the effect of group assignment was the same in the presence and absence of all these covariates, and none appeared to confound or modify effects of the main exposure on either outcome. Importantly, point estimates of the estimated HR of an LPD versus an MPD from all these models were close to 1 (0.95-1.12) and very imprecise (ie, with wide 95% confidence bounds from 0.57-1.99).
Figure 3.
Kaplan-Meier survival curves for (A) all-cause mortality, (B) dialysis therapy, and (C) the composite outcome (death + dialysis therapy). The low- (LPD) and moderate-protein diets (MPD) refer to 0.55 and 0.80 g/kg/d prescribed protein intake.
Table 3. Effect of LPD Versus MPD on Outcomes
| HR (95% CI) | |
|---|---|
| Progression to dialysis | |
| Unadjusted | 0.96(0.62-1.48) |
| Adjusted for baseline covariates | 0.98(0.64-1.51) |
| Adjusted for baseline covariates + TVC | 1.00(0.65-1.55) |
| Mortalitya | |
| Unadjusted | 1.01(0.57-1.79) |
| Adjusted for baseline covariates | 1.04(0.59-1.83) |
| Adjusted for baseline covariates + TVC | 1.12(0.64-1.99) |
| Death with dialysis as TVC | |
| Adjusted for dialysis | 1.01(0.57-1.79) |
| Adjusted for dialysis + baseline covariates | 1.04(0.59-1.83) |
| Adjusted for dialysis + baseline covariates + TVC | 1.07(0.61-1.86) |
| Composite end point (death or dialysis) | |
| Unadjusted | 0.98(0.68-1.42) |
| Adjusted for baseline covariates | 0.99(0.68-1.43) |
| Adjusted for baseline covariates + TVC | 1.03(0.71-1.50) |
| Competitive risk set | |
| Unadjusted | 0.95(0.68-1.34) |
| Adjusted for baseline covariates | 0.95(0.66-1.37) |
| Adjusted for baseline covariates + TVC | 1.01(0.70-1.46) |
aIncludes deaths before and after dialysis. |
Discussion
To our knowledge, the present study represents the first attempt to investigate within a randomized controlled trial the long-term effects (48 months) of an LPD on the survival of patients with CKD stages 4-5. Our study shows that: (1) prescription of an LPD does not lead to protein wasting, (2) most patients regularly followed up in CKD clinics adhere acceptably to the prescribed dietary protein intake restrictions, and (3) an LPD does not seem to impact on patient outcomes. These findings have implications of clinical relevance.
One major concern about the use of reduced protein intake in patients with renal failure is the risk of protein wasting and malnutrition with a possible impact on mortality, especially after dialysis therapy is started. Our data do not give support to such a fear. The 2 most relevant clinical markers of nutritional status that were monitored, body weight and urinary creatinine excretion, respectively, reflecting changes in caloric intake and lean body mass, remained stable for the duration of the study with both diets. Similar behavior showed that serum albumin and serum transferrin levels, both markers of visceral protein synthesis, are sensitive to nutritional and inflammatory status. Therefore, patients advised to follow an LPD of 0.55 g/kg/d of protein or a moderately restricted regimen of 0.80 g/kg/d of protein and carefully followed up by a renal team that included a physician and a dietitian have a very low risk of developing alterations in nutritional status.
A second interesting result of the present study is that protein intake can be decreased and maintained restricted for a long period. This is true even if full adherence to an LPD (0.55 g/kg/d) is attainable in only a fraction of patients (25%-30%). Our data show that although not all patients strictly adhered to the prescribed diet, the mean separation in protein intake between the 2 groups was 0.17 g/kg/d. Although smaller than prescribed, this degree of separation was maintained throughout the 48-month study period. Similar results have been shown by the MDRD long-term study, for which the difference in protein intake between the LPD and usual-protein diet groups was 0.19 g/kg/d. However, in the MDRD extension study, data were available only after 9 months from the end of the primary study.13
A third finding of the present study concerns the risk of progression to dialysis therapy and death. As shown previously by our group,4 protein intake of 0.55 compared with 0.80 g/kg/d guarantees better metabolic control and decreased need of drugs. To determine whether the lower protein diet would also prolong survival compared with the moderate protein restriction was the second main purpose of this study. Our results show that the risk of reaching dialysis therapy and the risk of mortality overall were low in this study cohort and unaffected by the type of diet regimen. The only data available in the literature concerning the impact of an LPD on the mortality of patients with CKD are those recently obtained from the MDRD Study Group.13 These authors found no benefit of an LPD in terms of the composite outcome of kidney failure and all-cause mortality. However, it is notable that these patients were not followed up from the original centers participating in the study after trial completion.
Another important difference with our trial is that in the MDRD Study, an LPD (0.58 g/kg/d) was compared with a normal-protein diet (1.3 g/kg/d). In our study, both tested diets contained a decreased amount of protein, 0.55 or 0.80 g/kg/d, which represent the 2 dietary protein levels most commonly prescribed to patients with CKD.27, 28, 29 Our data and those from the MDRD Study apparently suggest that an LPD does not have a significant role in ameliorating the survival of patients with CKD versus either a standard regimen13 or a moderate restriction (the present study). However, it is worth emphasizing that the mortality rate of the present cohort was very low (3.8%/y). This may be explained by several factors, including either or both the improved metabolic control associated with decreased protein intake from either protein restriction scheme (low or moderate) and the efforts in place to achieve the KDOQI target goals for the main (7) cardiovascular risk factors. Alternatively, the low mortality was unaffected by the diet and may be related to the type of care received in tertiary-care centers and participation in the clinical trial (Hawthorn effect). A low mortality rate has been similarly reported in different cohorts of patients with CKD followed up in a nephrology setting, such as the MDRD Study cohort in the United States30 and the CKD stage 4 cohort in British Columbia, Canada.31 In both studies, the average mortality rate was close to ours at 2.3% per year and 3.5% per year, respectively.
The negative results of the present trial are in contrast to those from other nonrandomized studies and laboratory data.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 However, data from observational studies and metabolic investigations need to be confirmed by clinical trials to inform clinical practice and add knowledge. In this respect, it is necessary to note that the mechanisms through which dietary restrictions may influence survival are multiple, complex, and only partially known. These include the concurrent decrease in many nutrients (proteins, sodium, phosphate, and calcium), the different response of the original disease (adult polycystic kidney disease or interstitial nephritis), and not least, the degree of patient adherence to the prescribed diet. Finally, treatment of CKD today often involves several “nephroprotective” drugs,32 and the interactions between these therapies and the LPD are still unknown. All these factors, together with the need for long follow-up, make it very difficult to design and carry out a proper study aimed at showing a significant impact of dietary recommendations on the survival of patients with CKD.
Our study has limitations. The most important is related to the secondary nature of the analysis. The study hypothesis of differential survival experience according to diet regimen was tested using data from a randomized trial powered to test another hypothesis related to metabolic and laboratory outcome measures.4 The follow-up study was powered extending observation time and assuming a basal risk that was observed. However, a larger effect was hypothesized (ie, risk reduction by 50% with the LPD). Thus, the negative results of the present study can be interpreted as false negative if the study failed to detect such a large effect and the effect exists. This is possible, although the present study findings are consistent with those from the MDRD Study.13 Another possibility is that the effect of an LPD is smaller than that originally hypothesized and closer to 1. In this case, previous data from the MDRD Study are of little help because of the different comparisons made. It is possible that an LPD changes the risk of hard end points (dialysis therapy initiation and death), and this change may occur in either direction. The point estimate of the relative risk close to 1 and the large CIs (from 0.5 to almost 2 for the risk of death) suggest that a much larger and expensive study is necessary to detect such an effect if it is small (<10% decrease), possibly clinically irrelevant (eg, relative risk, 0.95), or nonexistent.
This study also has strengths. This is the first prospective study exploring the effects of protein intake on renal and patient survival in a cohort regularly followed up in a nephrology tertiary-care clinic. In addition, a very low incidence of loss to follow-up was registered in this study (16 of 423 patients; 3.78%). Finally, use of time-varying covariates makes analysis of data highly reliable.
In conclusion, follow-up data from our clinical trial of different LPD diet regimens have major clinical and research implications. Patients prescribed a lower protein diet (0.55-0.80 g/kg/d) decrease their protein intake and remain adherent to the diet for years, although the protein intake achieved is higher than the amount prescribed. The decreased protein intakes tested in the present study do not increase the risk of protein-calorie malnutrition when the follow-up team includes a nephrologist and dietitian. Finally, prescription of an LPD (0.55 g/kg/d) does not seem to confer a survival advantage compared with an MPD (0.80 g/kg/d).
Acknowledgements
Part of this study has been presented at the following meetings: European Renal Association–European Dialysis and XLV Transplant Association, Stockholm, Sweden, May 10-13, 2008; and 41st annual American Society of Nephrology, Philadelphia, PA, November 4-8, 2008. The authors thank the nurses (Wanda Marchese and Maddalena Emmaus), renal dietitian (Patrizia Lombardi), and patients who made this trial possible.
Support: This study, which was investigator designed (by Drs Cianciaruso, Sabbatini, and Ravani) and independently initiated, was partially funded by a grant from the University of Naples Federico II.
Financial Disclosure: None.
References
- . Metabolic studies of low protein diets in uremia (I. Nitrogen and potassium). Medicine. 1973;52:593–595
- . Diets for patients with chronic kidney disease, still worth prescribing. J Am Soc Nephrol. 2004;15:234–237
- . Beneficial responses to modified diets in treating patients with chronic kidney disease. Kidney Int Suppl. 2005;94:S133–S135
- Metabolic effects of two low protein diets in chronic kidney disease stage 4-5—a randomized controlled trial. Nephrol Dial Transplant. 2008;23(2):636–644
- Low-protein diet and renal osteodystrophy. Nephron. 1991;58:250–252
- . The role of dietary phosphorus restriction in the conservative management of chronic renal disease. J Ren Nutr. 2005;15(1):189–192
- Renal adaptation to dietary sodium restriction in moderate renal failure resulting from chronic glomerular disease. J Am Soc Nephrol. 1996;7:306–313
- Very low protein diet supplemented with ketoanalogs improves blood pressure control in chronic kidney disease. Kidney Int. 2007;71(3):245–251
- . Effects of moderate changes in protein intake on urinary albumin excretion in type I diabetic patients. Nutrition. 1995;11:345–349
- . Additive antiproteinuric effect of ACE inhibition and a low-protein diet in human renal disease. Nephrol Dial Transplant. 1995;10:497–504
- Supplemented very low protein diet ameliorates responsiveness to erythropoietin in chronic renal failure. Kidney Int. 2003;64(5):1822–1828
- Chronic kidney disease and risk of death, cardiovascular events and hospitalization. N Engl J Med. 2004;23:1296–1305
- Effect of dietary protein restriction on the progression of kidney disease: long-term follow-up of the Modification of Diet in Renal Disease (MDRD) Study. Am J Kidney Dis. 2006;48:879–888
- . Guideline 1: goals of antihypertensive therapy in CKD. In: K/DOQI Clinical Practice Guidelines on Hypertension and Antihypertensive Agents in Chronic Kidney Disease. Am J Kidney Dis. 2004;43(suppl 1):S65–S230
- . K/DOQI Clinical Practice Guidelines for Bone Metabolism and Disease in Chronic Kidney Disease. Am J Kidney Dis. 2003;42(suppl 3):S1–S201
- . K/DOQI Clinical Practice Guidelines for Managing Dyslipidemias in Chronic Kidney Disease. Am J Kidney Dis. 2003;41(suppl 3):
- . K/DOQI Clinical Practice Guidelines for Anemia of Chronic Kidney Disease: Update 2000. Am J Kidney Dis. 2001;37:182–238
- . A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999;130:461–470
- . Energy and protein requirements (Report of a joint expert consultation). World Health Organ Tech Rep Ser. 1985;724:1–206
- . Dietary reference intakes for energy, carbohydrate, fiber, fat, fatty acids, cholesterol, protein, and amino acids (macronutrients). In: Dietary Reference Intakes. Washington DC: National Academy Press; 2002;p. 1–8
- . A new method of classifyng prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383
- . A method for estimating nitrogen intake of patients with chronic renal failure. Kidney Int. 1985;27:58–65
- Measurement of muscle mass in humans: validity of the 24-hour urinary creatinine method. Am J Clin Nutr. 1983;37:478–494
- . K/DOQI Clinical Practice Guidelines for Hemodialysis Adequacy, 2000. Am J Kidney Dis. 2001;37(suppl 1):S7–S64
- . Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. J Am Stat Assoc. 1989;84:1065–1073
- . Modelling strategy guidelines and modelling strategy for assessing interaction and confounding. In: Logistic Regression: A Self-Learning Test. New York, NY: Springer-Verlag; 2002;p. 130–226
- . K/DOQI Clinical Practice Guidelines for Nutrition in Chronic Renal Failure. Am J Kidney Dis. 2000;35(6 suppl 2):
- . Clinical Practice Guidelines of the Canadian Society of Nephrology for the treatment of patients with chronic renal failure: a re-examination. Contrib Nephrol. 2003;140:163–169
- . Italian Society of Nephrology (Conservative therapy guidelines for chronic renal failure). G Ital Nefrol. 2003;20(suppl 24):S48–S60
- Long-term outcomes in nondiabetic chronic kidney disease. Kidney Int. 2008;73(11):1310–1315
- . Variability and risk factors for kidney disease progression and death following attainment of stage 4 CKD in a referred cohort. Am J Kidney Dis. 2008;52(4):661–671
- . Renal function and requirement for dialysis in chronic nephropathy patients on long-term ramipril: REIN follow-up trial. Lancet. 1998;352:1252–1256
Originally published online as doi:10.1053/j.ajkd.2009.07.021 on October 5, 2009.
Trial registration: www.controlled-trials.com; study number: ISRCTN58881100.
PII: S0272-6386(09)01078-6
doi:10.1053/j.ajkd.2009.07.021
© 2009 National Kidney Foundation, Inc. Published by Elsevier Inc All rights reserved.
Volume 54, Issue 6 , Pages 1052-1061, December 2009
