The Impact of Protein Intake on Renal Function Decline in Women with Normal Renal Function or Mild Renal Insufficiency

Context Although dietary protein restriction appears to slow the decline in renal function among patients with moderate renal insufficiency, its effect on normal and mildly decreased renal function is unknown. Contribution Among women from the Nurses' Health Study with normal renal function, protein intake was not associated with decline in glomerular filtration rate (GFR). In women with mild renal insufficiency, high protein intake, particularly of nondairy animal origin, was associated with more rapid than expected decline in GFR. Implications High protein intake is associated with declining GFR among women with mild renal insufficiency. A causal connection has not been demonstrated. Additional studies are needed to show that reducing protein intake protects the kidney. The Editors The potential effects of dietary protein consumption on renal function in persons with normal renal function or mild renal insufficiency have important public health implications given the prevalence of high-protein diets and protein supplementation (1-3). The American Heart Association's most recent revised guidelines suggest that a sustained high-protein diet may have adverse effects on renal function (4), but no data support this claim in people with normal renal function or mild renal insufficiency. However, there are theoretical reasons for such concern, including the fact that a high-protein diet may acutely increase the glomerular filtration rate (GFR) (5, 6) and possibly cause intraglomerular hypertension, which may lead to progressive loss of renal function (7). Many clinical studies have demonstrated that protein restriction may slow renal function decline compared with usual protein intake in patients with moderate renal insufficiency (8-11). However, these results remain controversial because the largest study of protein restriction in patients with moderate renal insufficiency found no significant benefit (12). A recent meta-analysis estimated that among patients with moderate renal insufficiency, GFR decreases by 0.53 mL/min less per year in those who follow a low-protein diet compared with those who do not (13). The authors of this meta-analysis suggested that benefits of a low-protein diet might be more apparent with longer follow-up. Some experimental evidence also suggests that animal proteins may play a greater role in the progression of renal disease than vegetable proteins (14-16), but not all studies have confirmed these results (17). In experiments in humans, meat protein acutely increases GFR compared with dairy protein (18). The primary purpose of our study was to examine the association between total protein intake and renal function decline over an 11-year period in women with normal renal function or mild renal insufficiency. We also examined the association between intake of different types of protein and renal function decline. Methods Study Sample The Nurses' Health Study (NHS) began in 1976, when 121 700 female nurses 30 to 55 years of age completed a detailed questionnaire regarding health-related information (19). Since then, questionnaires have been sent to participants biennially. Information on lifestyle factors and new medical diagnoses is collected every 2 years, and a detailed dietary questionnaire is mailed every 4 years (20). The creatinine measurements used to estimate renal function were initially obtained as part of a study designed to assess the association between analgesic use and change in renal function. We limited our study sample to the 32 826 participants who provided a blood sample in 1989. Of these, we identified women who reported using acetaminophen, aspirin, or nonsteroidal anti-inflammatory drugs 15 or more days per month on both the 1992 and the 1998 biennial questionnaires. For comparison, we also included a group of women who reported no use of any of the three analgesics on the 1990, 1992, and 1998 biennial questionnaires. After we excluded women with a history of cancer (except nonmelanoma skin cancer) or cardiovascular disease (myocardial infarction, angina, stroke, or transient ischemic attack), 4238 women were eligible for inclusion. A supplementary questionnaire was mailed in 1999 to collect detailed information on current and lifetime use of analgesics from these 4238 women, and 3876 women (91%) returned it. A second blood sample was collected in 2000 from women who originally provided a specimen in 1989. From those who returned the supplementary questionnaire and provided a second blood sample, we selected all women with lifetime consumption of at least 1501 tablets of one of the analgesics and a random sample of women who had taken fewer than 1501 tablets. Of the 1769 women selected, 20 were missing a creatinine value from either 1989 or 2000, 4 were excluded because they had a creatinine concentration less than 0.4 mg/dL (<35 mol/L) in 1989, 33 were excluded because they reported a history of abnormal kidney function, 21 were excluded because they had an estimated baseline GFR less than or equal to 55 mL/min per 1.73 m2, and 67 were excluded because data were missing for other covariates. Therefore, a total of 1624 women were included in the current study. Of these, 98% were white and 13 (1%) were African American. Assessment of Dietary Protein Intake and Other Nutritional Variables In 1990, participants were asked to complete a semi-quantitative food-frequency questionnaire that inquired about the average intake of specified foods and beverages during the previous year (coinciding with the first blood specimen). In 1994, the questionnaire was repeated. The reproducibility and validity of this questionnaire have been described in detail elsewhere (20). Nutrient intake, including protein intake, was computed from the reported frequency of consumption of each specified unit of food or beverage by using published data on the nutrient content of the specified portions (20). Using this information, we were able to estimate protein consumption from different sources, as well as intake of phosphorus and animal fat. We examined total protein intake continuously (per 10-g increment) and in quintiles and examined nondairy animal, dairy, and vegetable protein continuously. The Pearson correlation coefficient (r) between total protein intake in 1990 and 1994 was 0.51 (P < 0.001). Total protein intake also correlated with phosphorus intake (r = 0.64; P < 0.001) and animal fat intake (r = 0.32; P < 0.001). Nutrient values were adjusted for total energy intake by regressing total caloric intake on absolute nutrient intake (21, 22). Because total energy intake for a given person tends to be fixed within a narrow range, variations in nutrient intake are largely attributable to changes in composition of diet, not the total amount of food consumed. Energy-adjusted values reflect the nutrient composition of the diet independent of the total amount of food consumed. In addition, adjustment for energy reduces any variation introduced by questionnaire responses that underreported or overreported intake, thus improving the accuracy of nutrient measurements (21, 22). Ascertainment of Other Factors Age, weight, height, diabetes, hypertension, smoking, alcohol use, hypercholesterolemia, analgesic medication use, and antihypertensive medication use were examined as potentially important confounders. We used 1989 weight to calculate the estimated 1989 creatinine clearance and used 1998 weight to calculate the estimated 2000 creatinine clearance, since a weight from the 2000 questionnaire was not yet available. Diabetes, hypertension, and hypercholesterolemia were recorded if a woman reported any of these diagnoses from 1976 to 1996. Smoking status, alcohol consumption, and analgesic use were obtained from the 1990 questionnaire. Smoking was classified as current smoker, past smoker, or nonsmoker; alcohol intake was classified as none, 0.1 to 14.9 g/d, or at least 15 g/d; and acetaminophen, aspirin, and nonsteroidal anti-inflammatory use was classified according to days of use per month. We used information on use of antihypertensive medications, including angiotensin-converting enzyme inhibitors, from the 1994 and 1996 questionnaires. Estimation of Renal Function Renal function was estimated by using creatinine values from blood samples that had been drawn in 1989 and 2000 and stored at 130 C, as well as self-reported measurements of height and weight. Creatinine values for both years were determined simultaneously at Boston Children's Hospital laboratory in 2001 by using a modified Jaffe method. The coefficient of variation was 10% for the 371 masked samples included with the study sample. We used two formulas to estimate renal function (23, 24). Our primary estimate of GFR was based on data from the Modification of Diet in Renal Disease (MDRD) Study (24). This formula, 186 creatinine concentration 1.154 age 0.203 0.742, was empirically derived from 1070 patients with renal insufficiency by using iothalamate GFR measurements and was subsequently validated in 558 patients in the same study (25). Creatinine is measured in mg/dL, and age is measured in years. Results are multiplied by a factor of 1.212 for African-American women. The second formula was a modification of the CockcroftGault formula for estimating creatinine clearance (26), which Salazar and Corcoran (23) developed to estimate creatinine clearance on the basis of fat-free body mass. The modified formula has the advantage of attenuating the overestimation of creatinine clearance in obese persons that occurs with the CockcroftGault formula and providing similar results in average-weight women. The formula for women is (146 age) [(0.287 weight) + (9.74 height 2)]/(60 creatinine concentration), where age is measured in years, weight in kilograms, height in meters, and creatinine in mg/dL. This formula has been validated by comparison with actual measurements of creatinine clearance (23). Statistical Analyses For continuous variab