DOI: https://doi.org/10.21203/rs.3.rs-575728/v1
Background
High-protein (HP) diets have been recommended for weight loss including obese persons. However, the potential effects of HP regimens on kidney health for persons without chronic kidney disease (CKD) are still controversial.
Methods
To investigate the effects of HP diets versus standard protein/low protein (SP/LP) ones on renal function in individuals without CKD, we conducted this meta-analysis.
Results
Thirty-nine RCTs including 3400 participants were considered in this meta-analysis. HP diets resulted in an increased GFR (standardized mean difference [SMD] = 0.64, 95% confidence interval [CI]: 0.03, 1.26) and concentrations of serum urea (MD = 1.05, 95% CI: 0.66, 1.44), creatinine (MD = 2.94, 95% CI: 1.30, 4.58), and uric acid (MD = 19.89, 95% CI: 12.35, 27.43) in obese subjects when compared with SP/LP diets. The results in T2D and health participants did not show a notable detrimental effect on renal outcomes. Subgroup analysis showed that an increase in GFR was presented in obese subjects following an intervention shorter than 6 months. No significant differences were found in the urinary albumin excretion between the HP and SP/LP diets in obese and T2D populations, except for the healthy participants which was reported by only one study.
Conclusions
This meta-analysis showed that HP diets were associated with increased GFR, serum urea, creatinine, and uric acid in obese adults. Future studies are warranted to examine whether resulted glomerular hyperfiltration from HP diet can cause kidney damage in obese individuals.
Since a growing population can be classified as overweight or obese, a low-carbohydrate, high-protein (HP) diet has recently emerged to be popular for weight loss [1]. Even in patients with Type 2 diabetics (T2D), HP diets have been demonstrated to induce reduction in appetite and improvement in outcomes and have been recommended as a medical nutrition therapy in some guidelines [2]. The International Society of Sports Nutrition states that the daily intake of 1.4–2.0g/kg protein for physical activity may improve the adaptability of training to exercise training, and provides new evidence that higher protein intake (>3.0g/kg/d) may promote the loss of fat mass and the building of muscle mass in resistance-trained individuals [3]. Therefore, such diets have recently gained increasing attention. However, the potential effects of HP regimes on the parameters of kidney health for individuals without chronic kidney disease (CKD; glomerular filtration rate [GFR] < 60mL/min) are still controversial.
The recommended daily amount of protein for the general population is 0.83g/kg, which is sufficient to avoid negative nitrogen balance and meet 97% − 98% of the population’s requirements. Although the consensus definition of an HP diet remains unclear, most definitions had set thresholds of 1.2 g/kg to 2.0 g/kg per day [1]. High intake of dietary protein, including animal and plant protein, may accelerate the loss of kidney function for patients with CKD or at risk of CKD (e.g., patients with diabetes or obese patients with microalbuminuria). A body of evidence has suggested that an HP regime leads to an increase in GFR and triggers renal glomerular hyperfiltration, which results in renal damage [4]. In animal studies, HP diets have been associated with increased blood flow to the kidneys, a deterioration of renal injury, increased albuminuria, and histopathological changes in the kidneys [5]. Human data have also revealed hyperfiltration with HP consumption. The largest short-term (< 6 months) randomized controlled trial (RCT) conducted by Juraschek et al. [6] reported that a diet rich in protein (25% of kilocalories from protein) significantly increased the eGFR by 4 mL/min/1.73 m2 after six weeks in contrast to a lower protein diet (LP, 15% of kilocalories from protein). A community-based prospective study enrolling a total of 9,226 subjects showed that long-term HP consumption increased the risk of renal hyperfiltration and reduced the renal function in the general population compared with a stand-protein diet (SP) after a 4-year follow-up [7]. By contrast, several RCTs have reported no detrimental effects of HP diets on renal outcomes, including GFR, albuminuria, or serum laboratory values [8-10].
With the increased interest for HP diets in obese, T2D, and healthy population, as well as given aforementioned controversies, concerns about the renal effects of an HP protocol have been raised. Therefore, this meta-analysis was focused on the comparison of HP diet and SP/low-protein (LP) diet patterns on the renal function parameters in adult individuals.
Search strategy
Literature search was performed using PubMed, Cochrane Central, and Embase database to identify RCTs that evaluated the effects of HP diet on the renal function in individuals without CKD until 29th December 2020. The following search terms were used: (“high protein diet” OR “protein overconsumption” OR “protein overuse” OR “high meat diet”) and (“kidney” OR “renal”). Moreover, references from previous reviews and retrieved original articles were scanned to further find relevant studies. Only English and Chinese languages were imposed in the study selection.
Inclusion and exclusion criteria
Inclusion criteria were as follows: (1) a RCTs or crossover design; (2) intervention period was more than 1 week; (3) study population was ≥ 18 years old; (4) comparison of HP diet and SP / low protein diet on renal function (According to a previous review conducted by Santesso et al. in 2012 [11], the authors considered an increase of 5% difference in the total energy intake to distinguish between HP and SP/LP diets); (5) trial participants were obese, overweight, T2D, and healthy adults; (6) the study assessed one or more outcomes (i.e., GFR, serum urea, serum creatinine, serum uric acid, urinary albumin, and kidney volume) from baseline to the end of the intervention. Participants with CKD (GFR < 60 mL/min/1.73 m2), albuminuria beyond normal range, and type 1 diabetes were excluded.
Study selection and data extraction
Two reviewers (Y Zhu and ZJ Xie) conducted the study selection by screening the title and abstract of all articles independently. The final recruited literatures were determined by reading the full text of the articles. Inconsistencies were resolved by consensus with a third reviewer. The following information was extracted from each study: the first author’s last name, publication year, study design, duration period, baseline characteristics of the participants (i.e., age, sex, number, BMI, and race), sample size, protein intake between two groups (% of total energy content), and outcomes. If no significant difference was found between the two groups about the baseline data or the change from baseline values were not available, post-intervention mean values were used.
Quality assessment
The Cochrane Risk of Bias tool was applied to evaluate the quality of included RCTs. Risk of bias consisted of the following bias domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete data outcome, selective reporting, and other biases. The studies were assessed as having high, low, or unclear risk for each domain. Two authors independently reviewed and judged the assessment.
Statistical analysis
In this meta-analysis, mean difference (MD) with 95% confidence interval (CI) was analyzed for continuous data. If the unit was different across various studies, standardized mean difference (SMD) with 95% CI was used. When the mean or standard deviation (SD) values were not available, the values were estimated on the basis of other relevant measurements. The heterogeneity was assessed across included studies by I2 statistics and Cochran’s Q-test. The study with an I2 > 50% or a P < 0.1 in the Cochran’s Q-test was considered to have substantial heterogeneity. The results were pooled using a random effect model in the presence of considerable heterogeneity across all studies. Otherwise, a fixed effect model was selected. In addition, subgroup analyses (based on the duration of intervention) and sensitivity analysis were performed to explore the substantial causes of heterogeneity. Publication bias was evaluated using Egger’s test, and a funnel plot was constructed for the GFR values. All analyses were conducted using RevMan Manager software (version 5.3), and P < 0.05 was considered to be statistically significant.
A total of 1398 relevant articles were screened in the initial search. Among these reports, 1359 were excluded and 39 eligible RCTs with 3216 participants were finally included for this meta-analysis [6, 8–10, 12–46]. The PRISMA flowchart is shown in Fig. 1. Twenty-six studies were parallel RCTs, and thirteen had a crossover study design. Among the included studies, participants were either obese, overweight, T2D, or healthy adults. The main characteristics of the participants and the protein diet protocols were summarized in Table 1. The mean age was between 24.1 years and 75.4 years, the median BMI was 33.0 (range: 22.3 to 43.6), and the sample size of the included studies ranged from 6 to 419. The protein content ranged from 20–40% of the total energy intake in the HP diets and from 10–24% in the SP/LP diets. The duration of HP diet intervention was < 6 months in 23 trials, and ≥ 6 months in 16 trials. In addition, protein intakes in two groups were mostly of animal origin, except for one trial, in which vegetable protein (specifically, wheat gluten protein) was used.
| Study | Design | Race | Protein intake | BMI (kg/m2) | Age (y) | Patients n (% of male | Baseline level of GFR | Treatment During | participants |
|---|---|---|---|---|---|---|---|---|---|
| Brinkworth 2004 | RCT | - | C:15% H:30% | C: 33.6 ± 0.8 H: 34.6 ± 0.9 | C: 51.5 ± 1.6 H: 52.0 ± 2.6 | C: 22 (31.8) H: 21 (23.8) | C: 101.8 ± 7.1 H: 100.2 ± 11.8 | 68W | Obese adults |
| Brinkworth 2010 | RCT | Asian | C:24% H:35% | C: 33.5 ± 3.8 H: 33.6 ± 4.0 | C: 51.3 ± 6.5 H: 51.2 ± 7.7 | C: 33 (33.3) H: 35 (40) | C: 83.8 ± 13.8 H: 90.0 ± 17.0 | 52W | Obese adults |
| Campos 2017 | RCT | - | C: 0.8g/kg/d H: 1.34g/kg/d | C:33.3 ± 5.0 H: 31.5 ± 4.7 | C:41.1 ± 10.6 H :47.4 ± 11.5 | C: 59 (30.5) H: 59 (44.1) | - | 6M | Obese adults |
| Friedman 2010 | Cross over | White, Black | C:50g/d H:140g/d | 43.6 ± 9.2 | 45 ± 10 | 17 (12) | - | 1W | Obese adults |
| Friedman 2012 | RCT | White, Black, Asian, other | C: 15% H: unlimited protein consumption | C: 36.1 ± 3.5 H: 36.1 ± 3.6 | C: 44.9 ± 10.2 H: 46.2 ± 9.2 | C: 154 (31.8) H: 153 (32.7) | C: 133 ± 41.8 H: 135 ± 35.3 | 24M | Obese adults |
| Gulati 2017 | RCT | Asian Indians | C:15% H:29% | C:30.3 ± 4.4 H:30.4 ± 3.5 | C:35.2 ± 8.6 H:39.8 ± 9.9 | C: 60 (41.7) H: 62 (43.5) | - | 3M | Obese adults |
| Leidy 2007 | RCT | - | C:18% H:30% | C:30.5 ± 0.6 H:30.7 ± 0.9 | C:53 ± 3 H:46 ± 2 | 46 (0) | C: 74 ± 3 H: 86 ± 2 | 12W | Obese adults |
| Li 2010 | RCT | Asian, Black, White, others | C:15% H:30% | C:34.3 ± 10.3 H:34.7 ± 6.8 | C:49.7 ± 9.1 H:48.9 ± 11.8 | C: 41 (36.6) H: 44 (18.2) | C: 116.9 ± 44.4 H: 129.8 ± 60.1 | 12M | Obese adults |
| Liu 2013 | RCT | Asian | C:17.7% H:26% | C:26.9 ± 0.4 H:26.6 ± 0.5 | 47.9 ± 0.9 | 50 (0) | - | 12W | Obese adults |
| Luscombe 2005 | RCT | - | C:20% H:40% | C:34.6 ± 1.0 H:33.4 ± 0.9 | C:48.9 ± 3 H:51.7 ± 2 | C: 30 (43.3) H: 27 (44.4) | C: 117 ± 13 H: 121 ± 10 | 16W | Obese adults |
| Moller 2018 | RCT | - | C:15% H:25% | 33.2 ± 4.6 | 61.4 ± 4.5 | 309(57.9) | 76.3 ± 13.5 | 12M | Obese adults |
| Study | Design | Race | Protein intake | BMI (kg/m2) | Age (y) | Patients n (% of male | Baseline level of GFR | Treatment During | participants |
| Noakes 2005 | RCT | - | C:17% H:34% | C:33 ± 4 H:32 ± 6 | C:49 ± 9 H:50 ± 10 | C: 48 (0) H: 52 (0) | C: 82.3 ± 3.3 H: 81.9 ± 3.3 | 12W | Obese adults |
| Skov 1999 | RCT | White, Asian | C:12% H:25% | C:30.8 ± 0.4 H:30.0 ± 0.4 | C:39.4 ± 2.0 H:39.8 ± 1.9 | C: 25 (76) H: 25 (76) | C: 114.3 ± 3.8 H: 105.5 ± 2.9 | 6M | Obese adults |
| Stern 2004 | RCT | White, Black, Hispanic | C: 67g/d H: 84g/d | C:42.9 ± 7.7 H:42.9 ± 6.6 | C:54 ± 9 H:53 ± 9 | C: 68 (85) H: 64 (80) | - | 12M | Obese adults |
| Wycherley 2012 | RCT | - | C:17% H:35% | 33.0 ± 3.9 | C:50.2 ± 9.3 H:51.3 ± 9.4 | 68 (100) | C: 103.1 ± 23.1 H: 106.4 ± 24.9 | 52W | Obese adults |
| Chen 2016 | RCT | Asian | C:15% H:30% | C: 28.8 ± 2.4 H: 28.7 ± 2.5 | C: 42.1 ± 10.7 H: 42.0 ± 10.1 | C: 50 (36) H: 51 (51) | - | 12W | T2D |
| Gannon 2003 | Cross over | - | C:15% H:30% | 31 (22, 37) | 61 (39, 79) | 12 (83.3) | - | 5W | T2D |
| Krebs 2012 | RCT | White, Maori, Pacific, other | C:15% H:30% | C:36.7 ± 6.4 H:36.6 ± 6.7 | C: 58.0 ± 9.2 H: 57.7 ± 9.9 | C: 212 (34) H: 207 (46) | - | 24M | T2D |
| Carsen 2011 | RCT | - | C:15% H:30% | - | C: 58.8 (55.8, 61.7) H: 59.6 (57.5, 61.8) | C: 46 (39) H: 53 (57) | C: 72.6 (68.5, 76.7) H: 70.2 (67.0, 73.4) | 12M | T2D |
| luger 2013 | RCT | - | C:15% H:30% | C:33.6 ± 5.3 H:33.0 ± 4.2 | C:63.7 ± 5.2 H:61.0 ± 5.7 | C: 22(27.3) H: 22(63.6) | C: 70.8 ± 15.0 H: 65.5 ± 14.5 | 12W | T2D |
| Nuttall 2003 | RCT | - | C:15% H:30% | - | - | 12 (83.3) | - | 5W | T2D |
| Nuttall 2006 | RCT | - | C:15% H:30% | 31 (27, 36) | 63 (51, 82) | 8 (100) | C: 143 ± 15 H: 144 ± 17 | 5W | T2D |
| Sargrad 2005 | RCT | Black, White, Hispanic | C:19% H:27% | C:36 ± 3 H:33 ± 2 | C:47 ± 1.9 H:48 ± 5.5 | C: 6 (33.3) H: 6 (16.7) | - | 8W | T2D |
| Study | Design | Race | Protein intake | BMI (kg/m2) | Age (y) | Patients n (% of male | Baseline level of GFR | Treatment During | participants |
| Tay 2015 | RCT | Asian | C:17% H:28% | C:35.1 ± 4.1 H:34.2 ± 4.5 | C:58 ± 7 H:58 ± 7 | C: 58(64) H: 57(51) | C: 91 ± 1.8 H: 96 ± 1.5 | 12M | T2D |
| Tay 2018 | RCT | Asian | C:17% H:28% | C:34.2 ± 1.63 H:35.1 ± 1.63 | C:58 ± 2.96 H :58 ± 2.96 | C: 58(64) H: 57(51) | C: 92 ± 1.5 H: 96 ± 1.5 | 24M | T2D |
| Antonio-1 2016 | Cross over | White, Black, Pacific | C:2.51g/kg/d H: 3.32g/kg/d | - | 26.3 ± 3.9 | 14(100) | 95 ± 19 | 6M | Healthy adults |
| Antonio-2 2016 | Cross over | - | C: 2.6g/kg/d H: 3.3g/kg/d | - | 25.9 ± 3.7 | 12(100) | 96 ± 20 | 8W | Healthy adults |
| Cao 2011 | Cross over | - | C: 10% H: 20% | 26.8 ± 3.1 | 56.9 ± 3.2 | 16 (0) | - | 7W | Healthy adults |
| Ferrara 2006 | RCT | - | C: 15% H: 22% | C: 23.7 ± 2 H: 23.4 ± 2.1 | 26.4 ± 5 | 15 (100) | - | 6M | Healthy adults |
| Frank 2009 | Cross over | White | C: 13% H: 27% | 22.3 ± 2.0 | 24.1 ± 2 | 24 (100) | - | 1W | Healthy adults |
| Hegsted 1981 | Cross over | White | C: 46g/d H:123g/d | - | 23–28 | 6(0) | - | 2M | Healthy adults |
| Jenkins 2001 | Cross over | - | C:16% H:27% | 26.0 ± 0.7 | 55.6 ± 1.9 | 20 (75) | - | 4W | Healthy adults |
| Johnston 2004 | RCT | - | C:15% H:32% | C:28.7 ± 2.0 H:29.1 ± 2.6 | C:36.4 ± 4.2 H:40.1 ± 3.6 | C: 7 (14.3) H: 9 (11.1) | C: 79 ± 6 H: 101 ± 8 | 6W | Healthy adults |
| Juraschek 2013 | Cross over | Black, White | C:15% H:25% | 30.2 ± 6.1 | 53.5 ± 10.8 | 164 (55) | 92.0 ± 16.3 | 6W | Healthy adults |
| Juraschek 2016 | Cross over | White, Black, Asian | C:16% H:23% | 32.3 ± 5.5 | 52.1 ± 11.4 | 159 (48) | 100.1 ± 16.4 | 5W | Healthy adults |
| Study | Design | Race | Protein intake | BMI (kg/m2) | Age (y) | Patients n (% of male | Baseline level of GFR | Treatment During | participants |
| Kerstetter 1997 | Cross over | White, Asian | C: 0.7g/kg/d H:2.1g/kg/d | 22.3 ± 0.6 | 26.7 ± 1.3 | 16(0) | C: 90 ± 5 H: 94 ± 5 | 2W | Healthy adults |
| Kerstetter 2015 | RCT | Caucasian | C: 72.7g/d H: 90.7g/d | C:26.4 ± 4.0 H:26.1 ± 3.4 | C:75.4 ± 15.4 H:74.4 ± 8.8 | C: 102(12.7) H: 106(16.0) | C: 74.5 ± 16.0 H: 72.3 ± 13.3 | 18M | Healthy adults |
| Roughead 2003 | Cross over | White, Asian | C:12% H:20% | 26.5 ± 4.0 | 60.5 ± 7.8 | 15(0) | - | 8W | Healthy adults |
| Wagner 2007 | Cross over | White, Black, Asian | C: 0.5g/kg/d H:2.0g/kg/d | Y:25.1 ± 5.7 O:25.8 ± 2.7 | Y:30.8 ± 4.0 O:60.2 ± 4.2 | Y: 12 (33) O: 10(30) | - | 6W | Healthy adults |
| C: control group, including standard protein diet and low protein diet; H: high protein diet group. T2D: type 2 diabetics. | |||||||||
Figure 2 showed the quality assessment of the included studies. Overall, all of the studies described a random sequence generation. Only 10 trials (25.6%) reported appropriate allocation concealment methods. Double-blind design was conducted in two studies, and whereas single-blind were used in four studies. Since all results in the present study were laboratory parameters, which were less likely to be affected by an open-label design. Thirty-five trials were considered to have low risk of bias on the domain of incomplete outcome data. The GFR levels were used to assess the publication bias, and according to the result of Egger’s test (p = 0.142, t = 1.51), no evidence of publication bias was found in the included studies.
Twenty-seven studies provided the effect of HP diets on the GFR level and were pooled in our analysis (Fig. 3). HP diets were associated with significantly increased GFR in comparison to the SP/LP diets in obese subjects (SMD = 0.64, 95% CI 0.03 to 1.26, P = 0.04) with significant heterogeneity among studies (I2 = 93%). In T2D and healthy subjects, HP diets did not show any change in the GFR level compared with SP/LP diets (T2D: SMD = − 0.54, 95% CI = − 1.36 to 0.28; P = 0.20, I2 = 92%; Healthy: SMD = 0.78; 95% CI = − 0.43 to 1.94; P = 0.21, I2 = 99%). The results from the subgroup analyses showed that the HP diets increased the GFR in a long-term duration (< 6 months) in obese subjects, but no significant difference was observed after a short-term intervention (> 6 months). Detailed results of subgroup analyses were shown in Table 2.
| Subset | Number of studies | Number of participants | MD or SMD (95% CI) | P-value | Heterogeneity analysis (P Value I2, %) | |
|---|---|---|---|---|---|---|
| GFR Obesity <6 months ≧6 months T2D <6 months ≧6 months Healthy < 6 months ≧ 6 months | 4 6 3 3 9 2 | 237 433 74 329 1144 149 | 0.16 (− 1.23 to 1.55) 0.93 (0.25 to 1.60) −0.06 (− 0.52 to 0.40) −1.03 (− 2.26 to 0.20) 0.79 (− 0.72 to 2.29) 0.59 (− 0.84 to 2.01) | P = 0.82 P = 0.007 P = 0.79 P = 0.10 P = 0.31 P = 0.42 | P < 0.00001 96% P < 0.00001 91% P = 0.38 0% P < 0.00001 96% P < 0.00001 99% P = 0.0007 91% | |
| Urinary albumin excretion Obesity <6 months ≧6 months T2D <6 months ≧6 months | 3 1 4 3 | 206 85 82 311 | −0.12 (− 0.76 to 0.52) −0.30 (− 0.73 to 0.13) −0.32 (− 0.76 to 0.12) −0.17 (− 0.28 to 0.62) | P = 0.71 P = 0.17 P = 0.16 P = 0.45 | P = 0.71 77% - P = 0.16 0% P = 0.45 74% | |
| Serum urea Obesity <6 months ≧6 months | 3 5 | 180 633 | 1.20 (0.90 to 1.50) 0.85 (0.28 to 1.43) | P < 0.00001 P = 0.0004 | P = 0.003 83% P < 0.00001 99% | |
| Subset | Number of studies | Number of participants | MD or SMD (95% CI) | P-value | Heterogeneity analysis (P Value I2, %) | |
| Serum urea Healthy <6 months ≧6 months | 6 1 | 219 28 | 2.09 (1.52 to 2.66) 0.85 (0.28 to 1.43) | P < 0.00001 P = 1.00 | P < 0.00001 96% - | |
| Serum creatinine Obesity < 6 months ≧ 6 months T2D < 6 months ≧ 6 months Healthy <6 months ≧ 6 months | 5 6 5 3 9 2 | 276 701 175 524 1199 43 | 4.09 (1.75 to 6.43) 1.39 (0.09 to 2.68) 3.14 (− 0.47 to 6.76) 1.36 (− 0.50 to 3.23) 0.68 (− 0.43 to 1.78) 0.00 (− 6.78 to 6.78) | P = 0.0006 P = 0.04 P = 0.09 P = 0.15 P = 1.00 P = 0.23 | P < 0.00001 92% P = 0.11 47% P = 0.26 24% P < 0.00001 96% P < 0.00001 98% P = 1.00 0% | |
| Serum uric acid Obesity < 6 months ≧ 6 months | 2 1 | 57 174 | 18.55 (10.48 to 26.63) 29.00 (7.90 to 50.10) | P < 0.00001 P < 0.00001 | P = 0.94 0% - | |
Twelve studies with 732 participants reported the effects of HP diets on the urinary albumin excretion (UAE, Fig. 4). No significant difference was found in the UAE between HP diets and SP/LP diets in obese participants (SMD = − 0.16, 95% CI = − 0.61 to 0.28; P = 0.47, I2 = 70%) and T2D (SMD = − 0.03, 95% CI = − 0.38 to 0.31; P = 0.85, I2 = 57%). Only one study with 48 participants compared HP diets with SP/LP diets in healthy young adults in a randomized crossover study. Urinary albumin measured in 24-h urine samples was significantly elevated in the HP diets compared with the SP diets (SMD = 1.35, 95% CI = 0.72 to 1.98; P < 0.0001). Subgroup analyses showed that the effect of HP diets in obese and T2D individuals on the UAE was independent of the duration (Table 2).
The pooled results in the meta-analysis showed that HP diets were associated with a significant increase in serum urea (MD = 1.05 mmol/L, 95% CI = 0.66 to 1.44, P < 0.00001; I2 = 98%), creatinine (MD = 2.94 µmol/L, 95% CI = 1.30 to 4.58, P = 0.0005; I2 = 91%), and uric acid (MD = 19.89 µmol/L, 95% CI = 12.35 to 27.43, P < 0.00001; I2 = 0%) in obese adults compared with the SP/LP protocols. In T2D participants, no changes were found in the serum urea (MD = − 0.99 mmol/L, 95% CI = − 4.49 to 2.52, P = 0.58; I2 = 95%) and uric acid (MD = 10.21 µmol/L, 95% CI = − 4.48 to 24.89, P = 0.17; I2 = 17%) between two groups. Compared with the SP/LP group, although a significant elevation in the serum creatinine (MD = 1.80 µmol/L, 95% CI = 0.17 to 3.43, P = 0.03; I2 = 88%) was observed in the HP diet group, subgroup analyses showed no difference in this parameter between two groups, whether under a short-term or a long-term HP diet intervention. For healthy adults, no significant differences were observed in the serum creatinine (MD = 0.66 µmol/L, 95% CI = − 0.43 to 1.75, P = 0.24; I2 = 98%) and uric acid (MD = 17.90 µmol/L, 95% CI = − 36.37 to 72.17) when the HP diets were compared with the SP/LP diets. The pooled results also indicated that a protein-enriched diet increased the serum urea (MD = 1.90 mmol/L, 95% CI = 1.36 to 2.44, P < 0.00001; I2 = 95%) in healthy adults, as shown in Supplemental Figure S1, S2, and S3. But subgroup analyses revealed that long-term HP diet intervention had no effect on the serum urea (Table 2).
In this meta-analysis of 39 RCTs, which is an update on this topic, the effects of HP and SP/LP diets on the renal function parameters in general population without CKD were compared. The evidences suggested that HP diets probably led to a significantly increased GFR and an increased concentration of serum urea, uric acid, and creatinine in obese subjects. However, the results in T2D patients and healthy participants did not show a notable detrimental effect on the renal outcomes. When the participants were sub grouped according to the duration of intervention, an increase in GRF was only presented in the obese subjects following a long-term intervention. As for albuminuria, no significant difference was observed in those individuals after an HP diet, except for healthy participants which was reported by only one study.
Increased protein consumption has been considered as a potential regulator of renal function as a result of increasing renal plasma flow and GFR [1]. In a mouse model, a 4-week HP diet led to a significant increase in the GFR by 33.6%, renal blood flow by 38.0%, and kidney weight by 25.7% compared with an LP diet [47]. A randomized study had demonstrated that a high dietary protein intake over six months in overweight subjects induced an increase in GFR by approximately 10%, and similar changes were observed in renal size [31]. In this study, a significant difference in GFR was found in obese adults between the HP diet and SP/LP diet groups after a follow-up over six months, but no elevated GRF was detected following a short-term HP intervention. Increased glomerular pressure and hyperfiltration occurs as a rise in GFR, which may result in the loss of kidney function over a long term. In addition, a meta-analysis including 17 studies showed that HP diets were associated with increased all-cause mortality [48]. However, some researchers recognize that a rise in GFR may probably be a normal adaptation for kidney due to an elevated urea filtration following an ingestion of HP diets [49].
In healthy individuals, an increase in protein intake up to a maximum of 3.32 g/kg/d over 4 months had no adverse effects on the marks of kidney function [34]. The present study resulted in the same finding in healthy and T2D subjects independent of the intervention duration, which was also consistent with findings from several RCTs among T2D individuals [8, 17]. In a 2-year investigation of 115 adults with T2D, comparison of the HP (28% total energy) diet with LP (17% total energy) diets showed that both diets achieved no adverse renal effects, including estimated GFR or albuminuria [8]. Similarly, a short-term trial by Luger et al. showed that a 5-week HP diet did not increase creatinine clearance among T2D subjects [17].
UAE is another marker for kidney damage and a predictor of end-stage renal disease in the general population and in individuals with diabetes mellitus [50]. Increased UAE has recently been reported to be associated with increased mortality and morbidity of cardiovascular risk in healthy adults and patients with diabetes [50, 51]. Although HP diet has been demonstrated to increase urinary albumin creatinine ratio in animal models [5] and short-term clinical trials [52], this meta-analysis showed that HP diet did not alter UAE in subjects without CKD. Additionally, a clinical intervention study reported no difference between plant and animal protein meals on the urine albumin in T2D subjects without microalbuminuria [53].
Protein-enriched meals can also result in the proportional elevation in urea and the generation of other nitrogenous waste products which due to increased protein metabolism. An experimental study has shown that high serum urea concentration, namely azotemia, is a common indicator for uremia, which may lead to oxidative stress, inflammation, endothelial dysfunction, and cardiovascular disease by increasing protein carbamylation and generating reactive oxygen species [54]. In the present meta-analysis, a rise in serum urea concentration was found in obese and healthy adults following an HP diet. Other nitrogenous waste product, such as creatinine, also increased in T2D and obesity subjects after an HP diet intervention. Of note, no difference was found between two groups when the T2D subjects were divided into subgroups according to the HP intervention duration. Increased serum urea and GFR among subjects without CKD was only an adaptation to a high protein load, because nitrogenous waste products may result in osmotic diuresis, and no indication of kidney damage was found [9].
A recent meta-analysis conducted by Schwingshackl et al. in 2014 [55] showed that HP diets were associated with increased GFR, serum urea, serum uric acid, and urinary calcium excretion in healthy individuals or individuals with T2D and obesity. Moreover, the study enrolled participants with normal and microalbuminuria, and subgroup analysis was not performed. Considering that most of the RCTs included in the previous meta-analysis recruited obese individuals, the authors infer that HP diets will increase the risk of kidney dysfunction for this population and this inference was consistent with this update study.
Several study limitations exist in this meta-analysis. First, the main outcome parameter GFR was evaluated by creatinine-based method. Clearance of endogenous creatinine, a previously widely used method to measure GFR, overestimates GFR by probably 10% because of the reabsorption and secretion of creatinine in renal tubules. In addition, HP intake, body size, and muscle mass are important sources of error in creatinine-based GFR [56], especially in those undergoing remarkable weight loss. Thus, a conservative attitude should be considered toward the GFR results reported in this meta-analysis. A randomized crossover trial by Juraschek et al. showed that the GFR changes during the HP diet measured using a cystatin C-based method differed from that using a creatinine-based method [6]. Second, compliance with HP diet intervention reported poor and higher dropout rates in most long-term trials, and enough protein are difficult to take to maintain the inclusion criteria for participants, especially to those with HP diet intervention. Third, several outcomes presented high heterogeneity even if subgroup analysis based on the intervention duration and participants have been performed. Some reasons for the significant heterogeneity may include different study designs (RCTs and crossover studies), study populations (age, sex, and race), and definition of HP and SP/LP diets. Fortunately, the outcomes were all laboratory test results, that is, the data were quantitative, not qualitative.
Although the renal-related contraindications to an HP diet were not clear in populations without CKD, and the elevation of GFR and serum nitrogenous waste products may be a normal adaptation of the kidney. The potential risk demonstrated in the experimental study should be carefully considered, especially to obese adults. Furthermore, HP diets may accelerate the progression of CKD and should be avoided in the dietary behaviors of these individuals. Based on a clinical trial in apparently healthy young US adults, only half of the participants presented an optimal range of UAE [51]. Considering the latent feature of CKD, all individuals should measure their serum creatinine concentration and urine specimen test for proteinuria before they take a protein-enriched meal.
HP: high-protein; T2D: Type 2 diabetics (T2D); CKD: chronic kidney disease; GFR: glomerular filtration rate; RCT: randomized controlled trial; SP: standard-protein diet; LP: low-protein diet; MD: mean difference; CI: confidence interval; SMD: standardized mean difference; SD: standard deviation; UAE：urinary albumin excretion
Ethics approval and consent to participant: Not applicable.
Consent for publication: Each author meets the criteria for authorship and assumes the corresponding responsibility.
Availability of data and materials: Not applicable.
Competing interests: The authors declared that they had no conflict of interest.
Funding: The authors received no funding for this work.
Author’s contributions: Z.Y. research idea and study design. Z.Y. and X.Z.J. searched for and selected the included studies. X.Z.J. and O.J.H. acquired, extracted, and analyzed the data. X.Z.J. and O.J.H. drafted the manuscript. X.C. and Z.Y. critically revised the manuscript.