This article requires a subscription to view the full text. If you have a subscription you may use the login form below to view the article. Access to this article can also be purchased.
Abstract
Background and objectives Protein and potassium intake and the resulting diet–dependent net acid load may affect kidney stone formation. It is not known whether protein type or net acid load is associated with risk of kidney stones.
Design, setting, participants, & measurements We prospectively examined intakes of protein (dairy, nondairy animal, and vegetable), potassium, and animal protein-to-potassium ratio (an estimate of net acid load) and risk of incident kidney stones in the Health Professionals Follow-Up Study (n=42,919), the Nurses’ Health Study I (n=60,128), and the Nurses’ Health Study II (n=90,629). Multivariable models were adjusted for age, body mass index, diet, and other factors. We also analyzed cross-sectional associations with 24-hour urine (n=6129).
Results During 3,108,264 person-years of follow-up, there were 6308 incident kidney stones. Dairy protein was associated with lower risk in the Nurses’ Health Study II (hazard ratio for highest versus lowest quintile, 0.84; 95% confidence interval, 0.73 to 0.96; P value for trend <0.01). The hazard ratios for nondairy animal protein were 1.15 (95% confidence interval, 0.97 to 1.36; P value for trend =0.04) in the Health Professionals Follow-Up Study and 1.20 (95% confidence interval, 0.99 to 1.46; P value for trend =0.06) in the Nurses’ Health Study I. Potassium intake was associated with lower risk in all three cohorts (hazard ratios from 0.44 [95% confidence interval, 0.36 to 0.53] to 0.67 [95% confidence interval, 0.57 to 0.78]; P values for trend <0.001). Animal protein-to-potassium ratio was associated with higher risk (P value for trend =0.004), even after adjustment for animal protein and potassium. Higher dietary potassium was associated with higher urine citrate, pH, and volume (P values for trend <0.002).
Conclusions Kidney stone risk may vary by protein type. Diets high in potassium or with a relative abundance of potassium compared with animal protein could represent a means of stone prevention.
- acid load
- nutrition
- potassium
- protein
- urolithiasis
- NEAP
- kidney stones
- Body Mass Index
- Citrates
- Citric Acid
- Cross-Sectional Studies
- Diet
- Dietary Proteins
- Epidemiologic Studies
- Follow-Up Studies
- Kidney Calculi
- Potassium, Dietary
- Risk
- Vegetables
Introduction
Although protein intake is thought to be an important risk factor for kidney stone formation, evidence is limited. Higher animal protein intake may result in increased urinary excretion of calcium and uric acid and reduced citrate and urine pH, thus potentially favoring formation of calcium oxalate and uric acid stones (1,2). However, relatively few studies have examined associations between animal protein intake and risk of actual kidney stones. In one randomized trial, the risk of recurrence in calcium oxalate stone formers with lower animal protein intake was not significantly different from the control group with higher animal protein intake (3), but the study was limited by the high dropout rate (46%). Observational studies also have reported conflicting results; in particular, previous analyses of the association between animal protein and risk of stones in the Nurses’ Health Study I (NHS I) and the Nurses’ Health Study II (NHS II) cohorts showed no association (4,5), and a positive association in the Health Professionals Follow-Up Study (HPFS) was only in men with a body mass index (BMI) <25 kg/m2 (6). To date, no data are available on the risk of stones associated with intakes of different types of animal protein (i.e., dairy versus nondairy) or vegetable protein.
Data on the association between intake of potassium, a marker of dietary intake of organic anion, and risk of stones also are inconclusive. Higher dietary potassium intake would decrease urinary excretion of calcium, thus potentially protecting against stone formation (7). To date, no trial has examined the effect of a diet rich in fruits and vegetables (a major dietary source of potassium) on the risk of stones; however, potassium citrate has been shown to be effective in reducing recurrence in stone formers (8). Higher potassium intake has been associated with lower risk of kidney stones in the NHS I and the HPFS but not in the NHS II (4–6).
The balance between protein and potassium intake from diet also might be important with regard to risk of forming kidney stones. These two factors are major determinants of net endogenous acid production (NEAP), with acid production from sulfur–containing amino acids from protein counterbalanced by bicarbonate generation from alkali–containing potassium salts (9). Higher estimated NEAP values have been associated with higher urinary calcium and lower citrate excretion (10,11), and cross-sectional studies have described higher dietary acid load among stone formers (12); however, the association between dietary net acid load and incident kidney stones has not been analyzed in longitudinal prospective studies.
We analyzed prospective associations between intake of different types of protein (vegetable, dairy, and nondairy animal), intake of potassium, and the interaction between animal protein and potassium intake and risk of incident kidney stones in three large cohort studies.
Materials and Methods
Study Populations
The HPFS was established in 1986 with the enrollment of 51,529 men who were health professionals (dentists, optometrists, osteopaths, pharmacists, podiatrists, and veterinarians) between the ages of 40 and 75 years old. The NHS I was established in 1976 with the enrollment of 121,700 women who were registered nurses between the ages of 30 and 55 years old, and the NHS II was established in 1989 with the enrollment of 116,430 women who were registered nurses between the ages of 25 and 42 years old. At enrollment, participants from each cohort completed a questionnaire with detailed information on diet, medical history, and medications. Participants subsequently were followed with biennial questionnaires including updated information and newly diagnosed diseases.
For this analysis, follow-up was started in 1986 for the HPFS and the NHS I and 1991 for the NHS II (the dates of the first detailed food frequency questionnaire [FFQ]) and went until the latest available information on confirmed incident stones: January of 2012 for the HPFS, May of 2006 for the NHS I, and May of 2011 for the NHS II. We only included participants without a history of kidney stones at baseline; furthermore, participants with a history of malignancy (except for nonmelanoma skin cancer) at baseline were excluded, and those who developed malignancies during follow-up were censored. For the NHS I, only participants who answered questionnaires in 1992 (the year of the first lifetime kidney stone history inquiry) or later were included.
Assessment of Diet
In 1986 (the HPFS and the NHS I) and 1991 (the NHS II), participants completed an FFQ asking about the average annual intake of >130 foods and 22 beverages; subsequently, the information was updated through FFQs every 4 years. Frequencies of consumption of each unit of food were used to compute intakes of nutrients using data from the US Department of Agriculture, except for oxalate content, which was measured by capillary electrophoresis in the majority of foods from the FFQ (13,14). All nutrients were energy adjusted to account for total amount of food eaten. FFQs were previously found to be reproducible and valid in the HPFS and the NHS I (15,16).
Dietary–dependent net acid load was estimated as the animal protein-to-potassium ratio (APKR), which previously was shown to be highly correlated with renal net acid excretion (r=0.90) (9). We also estimated NEAP from total protein and potassium intake using the equation in the work by Frassetto et al. (9), which is also highly correlated with renal net acid excretion (r=0.84).
Assessment of Kidney Stones
Participants who reported the occurrence of a kidney stone on the biennial questionnaire were asked to complete a supplementary questionnaire to establish the date of occurrence and accompanying symptoms. For this analysis, only kidney stones accompanied by pain and/or hematuria were considered; those participants reporting a new stone event without accompanying pain or hematuria were censored. Evaluation of medical records among 582 participants from the HPFS, 194 participants from the NHS I, and 858 participants from the NHS II who reported a kidney stone confirmed the diagnosis in 95%, 96%, and 98% of patients, respectively.
Assessment of Other Covariates
Information on age, BMI, history of diabetes, history of hypertension, use of thiazides, and amount of calcium supplements was obtained from the biennial questionnaire. Self-reported weight, used to calculate BMI, was validated in the HPFS and the NHS I cohorts (17).
Assessment of Urinary Composition
Twenty-four–hour urine samples were collected in three cycles. In the first cycle, participants were eligible if they were ≤70 years of age in the HPFS or ≤65 years of age in the NHS I and had no history of cancer or cardiovascular disease. In the second cycle, participants were eligible if they were ≤75 years of age and had no history of cancer (other than nonmelanoma skin cancer). The third cycle was performed only by the participants in the NHS II, with the following criteria: age ≤55 years old, white race, and no history of high BP, coronary heart disease, or cancer. Urine samples collected in the first two cycles were analyzed with the system provided by Mission Pharmacal (San Antonio, TX), whereas the samples collected in the third cycle were analyzed by the Litholink Corporation (Chicago, IL). Participants with a history of kidney stones were oversampled in the first two cycles.
Participants with possible over- or undercollections (defined as urinary creatinine excretion in the top or bottom 1% of the nonstone-formers distribution) were removed from the analysis. For participants who provided more than one collection, the first sample was analyzed.
Statistical Analyses
The analysis of kidney stone risk was prospective; dietary intakes were assessed before the incident kidney stone. For each cohort, participants were divided into quintiles of intakes, and time at risk was calculated as time from start of follow-up to an incident kidney stone, loss to follow-up, death, or censoring (whichever happened first). Cox proportional hazards regression models generated hazard ratios (HRs) and 95% confidence intervals (95% CIs) adjusted for age (continuous), BMI (13 categories), history of diabetes (yes versus no), history of hypertension (yes versus no), use of thiazides (yes versus no), supplemental calcium intake (four categories), and dietary intakes of fluid, calcium, sodium, fructose, oxalate, phytate (all quintiles), and alcohol (four categories). Vegetable, nondairy animal, and dairy protein models were simultaneously adjusted for the other types of protein and potassium intake (all quintiles) but not for dietary calcium (because of high correlation between dietary calcium and dairy protein); APKR and NEAP models were further adjusted for their components: animal protein intake (APKR), all types of protein (NEAP), and potassium intake (APKR and NEAP). With the components included in the models, estimates for APKR and NEAP are interpreted as interaction terms. Information on exposures and covariates was updated every 4 years. Tests for linear trends were constructed by analyzing the median value for each quintile as a continuous variable. Effect modification by age (≤50 versus >50 years old), BMI (≤25 versus >25 kg/m2), and use of thiazides was investigated through interaction terms for each type of protein, dietary potassium intake, and APKR.
Differences in urinary components across quintiles of exposure were analyzed with linear regression models using each urinary component as the dependent variable; adjusted for age, BMI, history of kidney stones, history of diabetes, history of hypertension, use of thiazides, supplemental calcium, study cohort, intake of fluid, urine creatinine, and sodium, and expressed as adjusted means, differences, and 95% CIs. When dietary potassium was used as the exposure, models were further adjusted for all types of protein; when protein intake and APKR were used as the exposure, models were also further adjusted for dietary potassium. For these analyses, we used information from the most recent questionnaire before the urine collection.
For each analysis, we evaluated between-cohort heterogeneity; pooled estimates obtained by random effects meta–analysis are presented whenever the P value for heterogeneity was >0.05.
Results
In total, 193,676 participants were included in the analysis of incident kidney stones (42,919 from the HPFS, 60,128 from the NHS I, and 90,629 from the NHS II). Baseline characteristics of the study participants are shown in Table 1 and Supplemental Table 1.
Baseline characteristics of the study cohorts
Association between Type of Protein and Risk of Stones
Over 3,108,264 person-years of follow-up, 6308 incident kidney stones were confirmed. The association between the type of protein and risk of kidney stones is presented in Tables 2 and 3 and Supplemental Table 2.
Quintiles of dairy protein intake and risk of kidney stones
Quintiles of nondairy animal protein intake and risk of kidney stones
There was a significant inverse association between dairy protein intake and risk of stones in the NHS II (the multivariate-adjusted HR for the highest quintile compared with the lowest was 0.84; 95% CI, 0.73 to 0.96; P value for trend <0.01) but not in the HPFS (HR, 0.93; 95% CI, 0.78 to 1.10; P value for trend =0.59) or the NHS I (HR, 0.95; 95% CI, 0.76 to 1.17; P value for trend =0.89). The pooled analysis showed that the HR was 0.89 (95% CI, 0.81 to 0.98; P value for trend =0.06).
The HRs for nondairy animal protein intake were 1.15 in the HPFS (95% CI, 0.97 to 1.36; P value for trend =0.04), 1.20 in the NHS I (95% CI, 0.99 to 1.46; P value for trend =0.06), and 1.02 in the NHS II (95% CI, 0.90 to 1.17; P value for trend =0.64). The pooled analysis showed an HR of 1.10 (95% CI, 1.00 to 1.21; P value for trend =0.20). Because the higher risk was manifest predominantly in the highest quintile of nondairy animal protein, we also analyzed participants with intakes higher than those in the top quintile median. In this subanalysis, the HRs were 1.20 (95% CI, 0.99 to 1.46; P value for trend =0.03) for the HPFS, 1.28 (95% CI, 1.02 to 1.61; P value for trend =0.02) for the NHS I, and 0.97 (95% CI, 0.83 to 1.14; P value for trend =0.66) for the NHS II.
There was no significant association between intake of vegetable protein and risk of stones. There was no significant effect modification by age, BMI, or use of thiazides for any source of protein.
Association between Intake of Potassium and Risk of Stones
The association between dietary potassium intake and risk of kidney stones is presented in Table 4. After multivariable adjustment, there was a strong inverse association between potassium intake and risk of stones: the HR was 0.44 (95% CI, 0.36 to 0.53) for the HPFS, 0.57 (95% CI, 0.45 to 0.72) for the NHS I, and 0.67 (95% CI, 0.57 to 0.78) for the NHS II (all P values for trend <0.001). There was no significant effect modification by age, BMI, or use of thiazides.
Quintiles of dietary potassium intake and risk of kidney stones
Association between Dietary Net Acid Load and Risk of Stones
The association between estimated dietary net acid load and risk of kidney stones is presented in Table 5 and Supplemental Table 3. After multivariable adjustment including animal protein and potassium intake, there was a significant positive association between APKR and risk of stones: the pooled HR was 1.41 (95% CI, 1.18 to 1.68; P value for trend =0.004). A similar association was seen with NEAP (pooled HR, 1.22; 95% CI, 1.05 to 1.41; P value for trend <0.01). There was no significant effect modification by age, BMI, or use of thiazides.
Quintiles of animal protein-to-potassium ratio and risk of kidney stones
Association between Type of Protein and Urinary Composition
The association between the type of protein and urinary composition is reported in Supplemental Tables 4 and 5. After multivariable adjustment, compared with those in the lowest quintile, those in the highest quintile of dairy protein intake had 24 (95% CI, 16 to 32) mg/d higher urine calcium, 1.4 (95% CI, 0.5 to 2.4) mg/d lower urine oxalate, 36 (95% CI, 11 to 60) mg/d higher urine citrate, and higher relative supersaturation for calcium phosphate. Those in the highest quintile of nondairy animal protein intake had 54 (95% CI, 30 to 78) mg/d lower urine citrate, 24 (95% CI, 13 to 35) mg/d higher urine uric acid, 0.13 (95% CI, 0.09 to 0.17) lower urine pH, and higher relative supersaturation for uric acid.
Association between Potassium Intake and Urinary Composition
The association between potassium intake and urinary composition is reported in Table 6. After multivariable adjustment including total fluid intake, those in the highest quintile had 1.6 (95% CI, 0.6 to 2.5) mg/d higher urine oxalate, 39 (95% CI, 13 to 65) mg/d higher urine citrate, 26 (95% CI, 14 to 38) mg/d higher urine uric acid, 0.15 (95% CI, 0.10 to 0.19) higher urine pH, 237 (95% CI, 173 to 299) ml/d higher urine volume, and significantly lower relative supersaturations for calcium oxalate and uric acid. There was no association between potassium intake and urinary excretion of calcium.
Pooled adjusted mean values and 95% confidence intervals of 24-hour urine components by quintile of dietary potassium intake
Association between Dietary Net Acid Load and Urinary Composition
The association between APKR and urinary composition is reported in Supplemental Table 6. After multivariable adjustment including main components of APKR, those in the highest quintile had 70 (95% CI, 22 to 118) mg/d lower urine citrate, 0.16 (95% CI, 0.07 to 0.23) lower urine pH, and 140 (95% CI, 23 to 256) ml/d lower urine volume.
Associations between dietary intakes and urine composition were similar among those participants with and without a history of kidney stones.
Discussion
We found that associations between vegetable, dairy, and nondairy animal protein intake and kidney stone risk were either null or modest and of borderline statistical significance. However, different types of protein may affect the risk of kidney stones differently: in particular, vegetable protein was not associated with risk of stones, dairy protein was associated with reduced risk in younger women, and nondairy animal protein may have been associated with modestly higher risk in men and older women. In contrast, higher dietary potassium was associated with a statistically significant and large reduction in kidney stone risk in all cohorts. APKR was associated with higher risk of stones, independent of the components of APKR (animal protein and potassium intake). The 24-hour urine composition also varied by type of protein. Higher dairy protein intake was associated with higher excretion of calcium and citrate but lower oxalate and uric acid. Higher nondairy animal protein, however, was associated with significantly lower citrate, higher uric acid, and a more acidic urine. Taken together, these findings suggest that dairy protein intake might have a mixed effect on urinary lithogenic factors. The higher urinary calcium might be because of the effect of casein on intestinal calcium absorption (18) and/or the higher content of calcium present in dairy products. The latter possibility may explain the lower excretion of oxalate with higher dairy protein intake, because higher calcium in the gut can bind with oxalate, thus reducing intestinal absorption of oxalate (19). However, nondairy animal protein may have a more marked prolithogenic effect on urine composition, likely to favor calcium stone formation (because of lower citrate) and uric acid stone formation (because of lower pH and higher uric acid). Another potential mechanism through which large intakes of animal protein could affect stone risk is the increased synthesis of oxalate secondary to metabolism of amino acids, such as hydroxyproline. However, studies with controlled dietary intakes of oxalate did not find any association between intake of animal protein and urinary oxalate excretion (20). Similarly, in our study, urinary oxalate was not higher among participants with larger intakes of animal protein.
A major finding in our study is the strong inverse association between higher dietary potassium intake, a marker of dietary intake of organic anion, and risk of stones: there was a 33%–56% lower risk of stones for those participants in the highest quintile of potassium intake. Previous analyses of these cohorts reported associations of smaller magnitude for the HPFS (6) and the NHS I (4) and no association for the NHS II (5). Our updated findings are likely caused by increased statistical power. Participants in the highest quintile of potassium intake had higher urinary levels of citrate, larger urine volumes, and higher urine pH, resulting in reduced relative supersaturations for calcium oxalate and uric acid, despite higher urinary excretion of oxalate. The larger urine volume, even after adjustment for fluid intake from beverages, might be explained by the high water content in fruits and vegetables.
We also found an association between higher levels of dietary net acid load, estimated by APKR and NEAP, and a higher risk of kidney stones. Such associations were significant, even after including the respective components of the ratios in the model. An imbalance between intakes of acid-generating and alkali-generating foods, reflected in higher APKR and NEAP values, respectively, could lead to a series of metabolic derangements, including higher urinary excretion of calcium (plausibly through bone buffering because of subclinical systemic acidosis [10]), reduced excretion of citrate (11), and reduced urine pH (which would increase the risk of uric acid stones). Of note, we found significantly lower excretion of citrate and lower urine pH for higher quintiles of APKR and a higher relative supersaturation for uric acid. The APKR results may partially account for our previous finding that a dietary approaches to stop hypertension–style diet, which includes higher intakes of fruits and vegetables and lower intakes of red and processed meats, is associated with a lower risk of kidney stones in the same cohorts (21).
Our study has limitations. We did not have information on stone composition for all participants and thus, could not determine whether type of protein or estimated dietary net acid load was associated with risk of selectively forming specific stone types. Also, our study participants are predominantly white, and our results may not apply to other populations.
In conclusion, the risk of kidney stone formation associated with protein intake may vary by protein type. In contrast with vegetable and dairy protein, nondairy animal protein may be associated with a slightly higher risk of kidney stones among men and older women. Higher potassium intake is strongly associated with a lower risk of stones, and a higher estimated net dietary acid load is associated with higher risk independent of potassium intake. Our data suggest that diets rich in fruits and vegetables as well as diets with a relative abundance of fruits and vegetables compared with animal protein may represent effective interventions to prevent kidney stone formation.
Disclosures
P.M.F. discloses consultant’s fees from BioHealth Italia (Rivoli, Turin, Italy) (2014–2015). G.C.C. is a consultant for Allena Pharmaceuticals (Newton, MA), AstraZeneca Pharmaceuticals (Wilmington, DE), and Exponent (Los Angeles, CA); receives royalties from UpToDate (author and Section Editor); and receives honorarium from the American Society of Nephrology (Editor-in-Chief of the Clinical Journal of the American Society of Nephrology). All other authors have nothing to disclose.
Acknowledgments
Because G.C.C. is the Editor-in-Chief of the Clinical Journal of the American Society of Nephrology, he was not involved in the peer review process for this manuscript. Another editor oversaw the peer review and decision-making process for this manuscript.
This work was supported by American Kidney Fund Clinical Scientist in Nephrology Program and National Institutes of Health research grants DK094910, DK91417, CA186107, CA176726, and CA167552.
The content of this manuscript was presented in abstract form at the 2015 American Society of Nephrology Kidney Week Meeting (San Diego, CA; November 3–8, 2015).
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
This article contains supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.01520216/-/DCSupplemental.
- Received February 10, 2016.
- Accepted June 20, 2016.
- Copyright © 2016 by the American Society of Nephrology
References
In this issue
