Nephrolithiasis as a Risk Factor for CKD: The Atherosclerosis Risk in Communities Study

Abstract

Background and objectives

Previous studies demonstrated a higher risk of CKD in persons with a history of kidney stones, but these studies examined mostly white populations and did not evaluate important potential interactions such as race and plasma uric acid.

Design, setting, participants, & measurements

In 10,678 Atherosclerosis Risk in Communities (ARIC) study participants free of CKD at baseline (ARIC visit 4 in 1996–1998), we assessed the association between a history of nephrolithiasis (a time-varying variable, defined by a combination of self-report and diagnostic codes) and incident CKD (defined by diagnostic codes from linkage to hospitalizations and US Centers for Medicare and Medicaid Services’ records).

Results

At baseline, 856 participants had a history of nephrolithiasis; 322 developed nephrolithiasis during follow-up. Over a mean follow-up of 12 years, there were 1037 incident CKD events. Nephrolithiasis history was associated with a 29% (hazard ratio [HR], 1.29; 95% confidence interval [95% CI], 1.07 to 1.54) higher risk of CKD in demographic-adjusted analyses, but the association was no longer statistically significant after multivariable adjustment (HR, 1.09; 95% CI, 0.90 to 1.32). The multivariable-adjusted association was stronger among participants with plasma uric acid levels ≤6 mg/dl (HR, 1.34; 95% CI, 1.05 to 1.72) compared with those with levels >6 mg/dl (HR, 0.94; 95% CI, 0.70 to 1.28; P_interaction=0.05). There was no interaction of stone disease and race with incident CKD.

Conclusions

In this community-based cohort, nephrolithiasis was not an independent risk factor for incident CKD overall. However, risk of CKD was unexpectedly elevated in participants with stone disease and lower plasma uric acid levels.

Introduction

Nephrolithiasis is an important clinical condition, which may result in severe flank pain and hematuria. Stones may obstruct urinary outflow, which can lead to ESRD in rare cases when it is bilateral (1). However, other mechanisms for stone-mediated kidney damage have been proposed. Calcium oxalate crystals in particular have been associated interstitial inflammation (2). Infection-related calculi, such as from Proteus mirabilis, may induce papillary necrosis (3). Stone treatment itself may cause kidney damage: In animal models, extracorporeal shock wave lithotripsy disrupts the tubular basement membrane (4). Thus, it is possible that the occurrence of kidney stones may predispose to GFR decline and CKD over the long term.

Nephrolithiasis may also herald certain systemic disorders. Stone disease history has been associated with a greater risk of hypertension, both in cross-sectional studies (5–9) and in a prospective, longitudinal study (10). Nephrolithiasis was also associated with a higher risk of coronary artery disease, as measured by clinical events (11–13) or coronary artery calcification (14). In addition, diabetes and the metabolic syndrome are risk factors for kidney stones (15–19), particularly uric acid nephrolithiasis (16,18).

Recent epidemiologic studies have suggested an association between nephrolithiasis and the development of reduced GFR (CKD stage ≥3). In cross-sectional studies, a history of kidney stones was associated with lower GFR and greater prevalence of CKD (20,21). Two prospective cohort studies from the Mayo Health System and the Alberta Kidney Disease Network demonstrated a higher risk of incident CKD in persons with a history of stone disease (22,23). However, both prospective studies involved universally insured and homogenous populations (96%–99% white) (22,24), thus limiting the generalizability of these findings. In addition, there is evidence that nephrolithiasis may be associated with a greater risk of ESRD (25,26).

We investigated the prospective relationship between a history of nephrolithiasis and incidence of CKD stage ≥3 in the population-based Atherosclerosis Risk in Communities (ARIC) study, with particular attention paid to detecting interactions with race. African Americans appear to be at a disproportionate risk of CKD compared with their white counterparts, much of which is unexplained (27). Although stones are less prevalent in African Americans compared with whites, kidney stones have been found to be a possible risk factor in African Americans with ESRD (28). An additional objective was to evaluate the interaction of plasma uric acid and stone history with incident CKD. Hyperuricemia has been linked to both incident CKD and incident nephrolithiasis (29–31), and we hypothesized that kidney stones might be a stronger risk factor for CKD among those with hyperuricemia.

Materials and Methods

Study Design

The ARIC study is an ongoing prospective population-based cohort study of 15,792 adults aged 45–64 years at enrollment (32). Between 1987 and 1989 (visit 1), approximately 4000 individuals were recruited from each of four participating US communities (Forsyth County, North Carolina; Jackson, Mississippi; suburban Minneapolis, Minnesota; and Washington County, Maryland). Four additional study visits were conducted (visit 2: 1990–1992; visit 3: 1993–1995; visit 4: 1996–1998; and visit 5: 2011–2013). Participants are contacted annually by telephone to provide health status updates. Hospital discharge records and death certificates are collected on a continuous basis. Study participants provided written informed consent at each ARIC study visit. Study procedures followed the ethical standards of the Declaration of Helsinki and the institutional review boards at the participating universities.

Study Population

This analysis uses ARIC visit 4 as its baseline, which included 11,656 participants. This visit was chosen because this was the first time urinary albumin-to-creatinine ratio was obtained. Because of low numbers, we excluded those who reported their race as something other than African American or white (n=31) or who identified themselves as African American from Minneapolis or Washington County, Maryland (n=38). We also excluded persons with ESRD (n=5), those who had an eGFR <60 ml/min per 1.73 m² calculated using the Chronic Kidney Disease Epidemiology Collaboration 2009 creatinine equation (33) (n=739), and those missing data for eGFR (n=96) or self-reported nephrolithiasis at visit 3 (n=69).

Assessment of Kidney Stones

A history of kidney stones was assessed as a combination of self-report and diagnostic codes. As part of the ARIC study visit 3, but not at visit 4, participants were asked whether a doctor had ever told them that they had kidney stones diagnosed by a physician. In addition, inpatient encounters that involved stone disease were identified through ARIC surveillance between 1987 and 2010 by the following International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes on hospital discharge summaries: 592 (calculus of the kidney and ureter), 592.0 (calculus of kidney), 592.1 (calculus of ureter), 592.9 (urinary calculus, unspecified), 594 (calculus of lower urinary tract), and 274.11 (uric acid nephrolithiasis), consistent with codes used in previous cohort studies (22,23). Because these captured only inpatient encounters, we also queried linked data from the US Centers for Medicare and Medicaid Services (CMS) for the same ICD-9-CM codes. With certain exceptions, the CMS data were not collected until participants reached age 65 years; at ARIC visit 4, 62.2% (n=6748) of participants were aged <65 years. We required that a person have one inpatient or two separate outpatient claims to count as having stone disease. Using these criteria, we identified 856 participants with stone disease occurring before baseline (ARIC visit 4): 815 from visit 3 self-reports, 26 from hospital discharge codes, and 15 from outpatient encounters from the CMS data. Between visit 4 and the end of follow-up, we identified an additional 318 incident stones via ARIC hospital and CMS surveillance.

Identification of Incident CKD Stage ≥3

Incident CKD stage ≥3 was captured from visit 4 through December 31, 2010. CKD stage ≥3 was identified via hospitalizations, gathered through ARIC’s hospital surveillance and annual follow-up phone interviews. For each hospitalization, 26 ICD-9-CM diagnosis and procedure codes were abstracted, and CKD was defined using a validated diagnostic code algorithm (34). To better ensure the temporal relation between nephrolithiasis and CKD, we required that incident CKD stage ≥3 occur at least 90 days after the date of the kidney stone diagnostic code. If a diagnosis of CKD occurred within 90 days of stone diagnosis, the stone diagnosis was treated as a censoring event (n=14). In sensitivity analyses, we supplemented the above definition with CKD additionally defined by CMS diagnostic codes.

Measurement of Other Covariables

Most covariables were obtained during ARIC visit 4, except where indicated. Participants were asked to fast overnight. Blood samples were drawn, centrifuged, frozen, and shipped to the ARIC study laboratory, where plasma HDL cholesterol, LDL cholesterol, triglycerides, total cholesterol, high-sensitivity C-reactive protein (hsCRP), uric acid, and glucose were measured following standard ARIC protocols. Plasma creatinine was measured in specimens by the modified kinetic Jaffé method. Urine albumin was measured by a nephelometric method either on a Dade Behring BN100 or Beckman Image nephelometer. Diabetes was defined as a fasting glucose level ≥126 mg/dl, a nonfasting glucose ≥200 mg/dl, current diabetes medication use, or a self-reported physician diagnosis of diabetes. Hypertension was defined as a mean systolic BP of ≥140 mmHg or a diastolic BP of ≥90 mmHg measured at the study visit, or the use of antihypertensive medications. History of coronary heart disease was obtained by self-report at visit 1 and was then supplemented with verified events up to visit 4. Gout history was obtained by self-report at visit 4. Use of allopurinol and diuretics was assessed by inspection of medications during visit 4. Body mass index (BMI) was calculated as weight (in kilograms) divided by height (in square meters). In ARIC visits 1 and 3, the Baecke sports score was obtained as a measure of physical activity and is an ordinal score from 1 to 5 (35).

Statistical Analyses

Our hypothesis was that nephrolithiasis would be associated with CKD incidence. We analyzed nephrolithiasis exposure in two ways: (1) all stones, including baseline and incident stones during follow-up, which we termed time-varying” stones; and (2) history of stones at baseline (visit 4) only. We considered the first of these analyses to be the primary analysis, because it was a more complete history of nephrolithiasis.

Continuous baseline characteristics of participants were compared between those with and without a history of stones before baseline using a t test of the means or the median test for skewed variables, whereas categorical values were evaluated with chi-squared analysis. The hazard of incident CKD stage ≥3 was assessed using Cox proportional hazards regression. Two models were used to adjust for possible confounding variables. Model 1 included age, sex, race, and center, whereas model 2 included the demographic factors in model 1 as well as HDL, hypertension, urine albumin-to-creatinine ratio, eGFR, uric acid (because of a nonlinear relationship, modeled as two linear splines with a knot at 6 mg/dl), smoking status, BMI, diabetes, history of coronary heart disease, diuretic use, and hsCRP. These covariates were chosen based on demonstrated statistical significance in unadjusted and adjusted analyses. Covariates that were evaluated but not included in the final model due to lack of significant association with CKD risk included: total cholesterol, LDL, triglycerides, cholesterol medication use, alcohol consumption, gout status, and visit 1 and visit 3 sports scores.

Two-way interactions between history of nephrolithiasis (yes or no) and sex, race (African American or white), and plasma uric acid level (>6 mg/dl or ≤6 mg/dl) with risk of CKD were tested in the Cox models using cross-product terms. The cutpoint of 6 mg/dl was chosen based on inspection of the continuous association of plasma uric acid and incident CKD. All analyses were conducted using Stata software (MP 13.1).

Results

Baseline Characteristics

In the 10,678 ARIC participants free of CKD at baseline (visit 4), 856 had a history of nephrolithiasis, which equates to a prevalence of 8.0% (12.1% for men, 4.8% for women; 9.2% for whites; 3.5% for African Americans). The average age was 62.5 years; 21.8% of participants were African American and 16.0% had diabetes. The mean plasma uric acid level was 5.5 mg/dl. Individuals reporting kidney stones were more likely to be older, to be white, and to have comorbid conditions such as coronary artery disease, diabetes, and gout; they also had lower levels of total and HDL cholesterol, higher triglycerides, higher plasma uric acid, higher sports physical activity scores, and lower levels of hsCRP (Table 1). Mean BMI, mean BP, history of hypertension, and use of antihypertensive medications did not differ between those with versus without nephrolithiasis.

Table 1.

Characteristics of 10,678 Atherosclerosis Risk in Communities study participants with eGFR ≥60 ml/min per 1.73 m² in stone formers versus nonstone formers at visit 4 (1996–1998)

Variable (Visit 4 Unless Specified)	History of Stones (n=856)	No History of Stones (n=9822)	P Value
Age (yr)	63.1±5.7	62.5±5.6	0.002
Men	66.1	42	<0.001
African-American race	9.7	22.8	<0.001
Body mass index (kg/m2)	28.7±5.2	28.8±5.7	0.63
BP (mmHg)
Systolic	127.2±18.1	127.2±18.8	0.95
Diastolic	70.9±9.9	71.1±10.3	0.72
Hypertension	46.8	45.6	0.56
Use of antihypertensive medications	44.7	41.4	0.06
Diabetes	19.5	15.7	0.003
History of coronary artery disease	11.3	7.5	<0.001
Cholesterol (mg/dl)
Total	195.1±33.3	200.7±37.0	<0.001
HDL	45.8±14.9	50.5±16.5	<0.001
LDL	119.9±30.8	122.4±33.4	0.04
Triglycerides (mg/dl)	130.9 (94.9, 180.8)	120.9 (86.9, 170.8)	<0.001
Use of lipid-lowering medications	18.1	13.5	<0.001
Plasma uric acid (mg/dl)	5.7±1.4	5.5±1.5	0.001
History of gout	7.8	5.2	0.001
Allopurinol use	2.6	1.2	<0.001
Diuretic use (%)	8.9	8.7	0.89
eGFR (ml/min per 1.73 m2)	87.7±12.4	88.7±13.3	0.03
Smoking status
Current	13.2	15.1	<0.001
Former	52.0	42.4
Alcohol use (%)
Current drinker	46.3	50.1	0.006
Former drinker	34.3	29.1
High-sensitivity C-reactive protein (mg/L)	2.1 (0.9, 4.5)	2.4 (1.1, 5.5)	<0.001
Study center (%)			<0.001
Forsyth County, North Carolina	37.4	23.6
Jackson, Mississippi	7.9	20.5
Minneapolis, Minnesota	21.4	29.3
Washington County, Maryland	33.3	26.7
Sports score, visit 3	2.6±0.8	2.5±0.8	0.002
Urine albumin-to-creatinine ratio (μg/mg)	4.0 (1.7, 9.7)	3.7 (1.8, 7.3)	0.16

Continuous variables are presented as means ± SDs, percentages, or medians (interquartile ranges) unless otherwise indicated.

Incidence of CKD

Mean follow-up time after visit 4 was 12 years, and 1037 CKD events were identified. In our primary analysis, using time-dependent modeling, nephrolithiasis was associated with a higher risk of CKD in the demographic-adjusted model (hazard ratio [HR], 1.29; 95% confidence interval [95% CI], 1.07 to 1.54). When including only those with a history of nephrolithiasis before visit 4, we observed a 21% higher incidence of CKD, although this was not statistically significant (HR, 1.21; 95% CI, 0.98 to 1.48) (Table 2).

Table 2.

Incidence rates and adjusted hazard ratios of incident CKD in stone formers versus nonstone formers in 10,678 Atherosclerosis Risk in Communities study participants

Analysis Group	Stone Former	Number in Group	Number of CKD Events	CKD Events per 1000 Person-Years (95% CI)	Model 1, Hazard Ratio (95% CI)	Model 2, Hazard Ratio (95% CI)
Baseline plus time-varying	Yes	1178	138	11.3 (9.6 to 13.4)	1.29 (1.07 to 1.54)a	1.09 (0.90 to 1.32)
	No	9500	899	7.7 (7.3 to 8.3)	1 (reference)	1 (reference)
Baseline (visit 4) only	Yes	856	104	10.4 (8.6 to 12.6)	1.21 (0.98 to 1.48)	0.99 (0.80 to 1.24)
	No	9822	945	8.0 (7.5 to 8.5)	1 (reference)	1 (reference)

Model 1 was adjusted for age, sex, race, and study center. Model 2 was adjusted for the demographic factors in model 1 as well as HDL, hypertension, urine albumin-to-creatinine ratio, eGFR, plasma uric acid (linear spline with knot at 6 mg/dl), diuretic use, smoking status, body mass index, diabetes, history of coronary heart disease, and high-sensitivity C-reactive protein. 95% CI, 95% confidence interval.

^aP=0.01.

After adjustment for additional potentially confounding variables, there was no significant association of CKD with stone presence, overall, in the time-varying analysis (HR, 1.09; 95% CI, 0.90 to 1.32). Using baseline stones only, there also was no evidence of increased CKD risk overall (HR, 0.99; 95% CI, 0.80 to 1.24).

Interaction with Race, Sex, and Uric Acid

Interaction testing showed no significant effect modification of nephrolithiasis and race with risk for CKD (P=0.79) and no significant interaction with sex on risk of CKD (P=0.61). By contrast, there was evidence of an interaction of uric acid and nephrolithiasis with CKD in time-varying analysis (P<0.001) as well as baseline-only analysis (P<0.001) in the multivariable-adjusted model (Table 3). In both cases, the relation between nephrolithiasis and incident CKD was stronger in persons with uric acid ≤6 mg/dl. In time-varying, multivariable-adjusted analysis, stone formers with uric acid levels ≤6 mg/dl had a 1.34-fold greater risk of CKD (95% CI, 1.05 to 1.72) than the reference group (nonstone formers with uric acid levels <6 mg/dl). By contrast, those with uric acid levels >6 mg/dl showed no significant association between a history of nephrolithiasis and incident CKD (HR, 0.94; 95% CI, 0.70 to 1.28).

Table 3.

Incidence rates and adjusted hazard ratios of incident CKD in relation to stone former history (yes or no) and uric acid level (high or low) in 10,608 Atherosclerosis Risk in Communities study participants

Stone Former	Plasma Uric Acid Level	Number of Participants	Number of CKD Events	CKD Events per 1,000 Person-Years (95% CI)	Model 1, Hazard Ratio (95% CI)	Model 2, Hazard Ratio (95% CI)
Time-varying analyses
Yes	>6	452	53	11.2 (8.6 to 14.7)	1.48 (1.11 to 1.98)a	0.94 (0.70 to 1.28)
	≤6	717	85	11.5 (9.3 to 14.2)	1.65 (1.31 to 2.08)a	1.34 (1.05 to 1.72)a
No	>6	3109	409	11.0 (10.0 to 12.1)	1.54 (1.34 to 1.76)a	1.11 (0.96 to 1.28)
	≤6	6330	479	6.1 (5.6 to 6.7)	1 (reference)	1 (reference)
Baseline analyses
Yes	>6	321	37	9.8 (7.1 to 13.6)	1.35 (0.96 to 1.89)	0.83 (0.58 to 1.19)
	≤6	528	67	10.9 (8.5 to 13.8)	1.58 (1.22 to 2.04)a	1.22 (0.93 to 1.61)
No	>6	3240	432	11.4 (10.4 to 12.5)	1.54 (1.34 to 1.76)a	1.10 (0.95 to 1.27)
	≤6	6519	502	6.3 (5.8 to 6.9)	1 (reference)	1 (reference)

Uric acid levels were missing in 70 participants at visit 4. Model 1 was adjusted for age, sex, race, and study center. Model 2 was adjusted for the demographic factors in model 1 as well as HDL, hypertension, urine albumin-to-creatinine ratio, eGFR, plasma uric acid (linear spline with knot at 6 mg/dl), diuretic use, smoking status, body mass index, diabetes, history of coronary heart disease, and high-sensitivity C-reactive protein. 95% CI, 95% confidence interval.

^aP<0.05.

Sensitivity Analyses

When incident CKD was defined by diagnostic codes derived from hospitalizations and CMS data, there were 2071 incident CKD events identified. Results were similar: There was an association between nephrolithiasis history and incident CKD in the demographic-adjusted (HR, 1.21; 95% CI, 1.06 to 1.39) but not in the multivariable-adjusted model (HR, 1.09; 95% CI, 0.90 to 1.32).

Discussion

This large, population-based cohort study of older whites and African Americans found no independent, overall association between nephrolithiasis history and incident CKD. However, there was an unexpected, positive association of stone history with CKD limited to persons with uric acid levels ≤6 mg/dl.

A greater risk of CKD in stone formers, when adjusted for demographic variables only, is consistent in direction with prior case-control and cross-sectional studies (20,21), as well as two large prospective studies of clinical cohorts (22,23). In contrast with the latter two studies, however, our study did not identify kidney stones as an independent risk factor for CKD after multivariable adjustment. There are several potential reasons for this. Definitions of stone history varied between studies, and it is possible that diagnostic codes used in the ARIC surveillance data had lower sensitivity than the previous studies (which incorporated all outpatient codes as well). Furthermore, most CKD in this study was identified through hospitalization records, a definition that has been demonstrated to be only 35.5% sensitive, although >95% specific (34), making it possible that less severe cases eluded our study. Our study population was significantly older than the other two cohorts (63 years versus 44 and 51 years); this raises the possibility that our study missed cases in those individuals who had stone disease but were excluded because of an earlier diagnosis of CKD. We estimate that we had 80% statistical power to detect HRs on the order of 1.15, but our study had fewer patients with kidney stones than the other two large cohort studies, which studied 2969 and 11,609 patients, respectively (22,23).

We hypothesized that race might modify the risk of CKD in relation to stone formation, because a previous case-control study demonstrated increased prevalence of predialysis kidney stones in African Americans on dialysis compared with a dialysis-free African-American cohort, although there was no comparison to whites made (28). Given the disproportionate burden of CKD in African Americans (27) and the lack of representation in previous studies, we investigated a possible effect modification. Interaction analyses did not support differences in the nephrolithiasis-CKD association by race, but our power to test this interaction was limited. The prevalence of stones was much lower (3.5%) in African-American persons than whites (9.2%) at baseline, as previously reported in the ARIC study (36).

Although plasma uric acid does not provide information on urine uric acid, we hypothesized that elevated plasma levels, in combination with stone history, would be synergistic for developing CKD. Uric acid crystals from longstanding hyperuricemia can damage the renal interstitium and have been associated with tubulointerstitial disease (37). Increased plasma uric acid levels have been associated with CKD in large cohort studies (29,30). A recent case-control study found that stone formers who used allopurinol (potentially a surrogate for hyperuricemia) were at high risk of CKD (38). Gout is a risk factor for both CKD and nephrolithiasis (39,40). In this study, there was a significant association between higher plasma uric acid and incident CKD; however, the interaction of uric acid and stone history with CKD was not in the hypothesized direction. Nephrolithiasis was a stronger risk factor in those participants with lower levels of plasma uric acid. Although there was a significant difference between groups in allopurinol use (2.6% versus 1.2%), the limited use of these medications is unlikely to have significantly altered CKD risk. Alternatively, there may be differential nephrotoxicity by stone type, data that are unavailable in the ARIC study (and most clinical scenarios). Although large stone burden of any kind is responsible for 3.2% of ESRD (1) and struvite stones have been associated with CKD (38), episodic calcium stones may confer a different risk for CKD than do sporadic uric acid stones. Previous laboratory research has identified calcium oxalate as potentially damaging to renal tubules and the interstitium, through mechanisms that stimulate gene expression, cause chemoattraction of macrophages and monocytes, activate cell proliferation and inflammatory pathways, generate free radicals, and ultimately lead to tissue fibrosis (41). By contrast, other studies have demonstrated an association between uric acid nephrolithiasis and increased CKD risk (25,42).

Our study has several limitations. First, incident CKD was defined by diagnostic codes, which may be poorly sensitive for milder disease. Second, many stones were captured by self-report only. Reliance on participant recollection may have led to exposure misclassification, which could bias associations toward the null. Third, manual validation of CKD events has been done in a stratified random sample of ARIC participants; however, whether the performance of ICD-9-CM codes differs by stone status was not investigated. It should also be noted that neither the survey question to identify stones nor the ICD-9-CM codes used to identify time-varying stones distinguished between symptomatic, obstructive stones and incidental stones; there may be less of an effect on CKD risk among incidental stone formers. Furthermore, the ARIC study does not have data on recurrent stones or stone burden. Finally, although we stratified CKD risk in nephrolithiasis by plasma uric acid level as a possible surrogate of uric acid stones, we did not have data on urine pH nor information on stone type. Future prospective cohorts studying this issue would benefit from rigorous efforts to collect and store information regarding stone type.

Our prospective, longitudinal study of a large, older, ethnically diverse population demonstrated no independent excess risk of CKD in stone-forming participants overall. Although there did not seem to be racial differences in associations, we did detect an unexpected interaction with plasma uric acid levels, in that stone presence may be a more important risk factor in those with low uric acid levels. Future studies are needed to verify whether this interaction can be replicated.

Disclosures

None.