Subsequent urinary stone events are predicted by the magnitude of urinary oxalate excretion in enteric hyperoxaluria

Abstract

Data directly demonstrating the relationship between urinary oxalate (UOx) excretion and stone events in those with enteric hyperoxaluria (EH) are limited. Therefore, we assessed the relationship between UOx excretion and risk of kidney stone events in a retrospective population-based EH cohort. In all, 297 patients from Olmsted County, Minnesota were identified with EH based upon having a 24-h UOx ≥40 mg/24 h preceded by a diagnosis or procedure associated with malabsorption. Diagnostic codes and urologic procedures consistent with kidney stones during follow-up after baseline UOx were considered a new stone event. Logistic regression and accelerated failure time modeling were performed as a function of UOx excretion to predict the probability of new stone event and the annual rate of stone events, respectively, with adjustment for urine calcium and citrate. Mean ± standard deviation age was 51.4 ± 11.4 years and 68% were female. Median (interquartile range) UOx was 55.4 (46.6–73.0) mg/24 h and 81 patients had one or more stone event during a median follow-up time of 4.9 (2.8–7.8) years. Higher UOx was associated with a higher probability of developing a stone event (P < 0.01) and predicted an increased annual risk of kidney stones (P = 0.001). Estimates derived from these analyses suggest that a 20% decrease in UOx is associated with 25% reduction in the annual odds of a future stone event. Thus, these data demonstrate an association between baseline UOx and stone events in EH patients and highlight the potential benefit of strategies to reduce UOx in this patient group.

Background

Data directly demonstrating the relationship between urinary oxalate (UOx) excretion and stone events in those with enteric hyperoxaluria (EH) are limited.

Methods

We assessed the relationship between UOx excretion and risk of kidney stone events in a retrospective population-based EH cohort. In all, 297 patients from Olmsted County, Minnesota were identified with EH based upon having a 24-h UOx ≥40 mg/24 h preceded by a diagnosis or procedure associated with malabsorption. Diagnostic codes and urologic procedures consistent with kidney stones during follow-up after baseline UOx were considered a new stone event. Logistic regression and accelerated failure time modeling were performed as a function of UOx excretion to predict the probability of new stone event and the annual rate of stone events, respectively, with adjustment for urine calcium and citrate.

Results

Mean ± SD age was 51.4 ± 11.4 years and 68% were female. Median (interquartile range) UOx was 55.4 (46.6–73.0) mg/24 h and 81 patients had ≥1 stone event during a median follow-up time of 4.9 (2.8–7.8) years. Higher UOx was associated with a higher probability of developing a stone event (P < 0.01) and predicted an increased annual risk of kidney stones (P = 0.001). Estimates derived from these analyses suggest that a 20% decrease in UOx is associated with 25% reduction in the annual odds of a future stone event.

Conclusions

These data demonstrate an association between baseline UOx and stone events in EH patients and highlight the potential benefit of strategies to reduce UOx in this patient group.

KEY LEARNING POINTS

What is already known about this subject?

• enteric hyperoxaluria (EH), a disorder associated with fat malabsorption, is a well-recognized cause of nephrolithiasis; elevated urine oxalate (UOx) is felt to be a primary contributor of stone formation, but there are limited data directly establishing this association.

What this study adds?

• increasing UOx was associated with a higher probability of the development of a stone event and an increased annual risk of kidney stones.

What impact this may have on practice or policy?

• demonstrating evidence of the association of UOx with nephrolithiasis in EH would provide objective risk stratification of stone formation for patients and providers as well as support studies of new therapies being developed to prevent kidney stones and other disorders associated with EH.

INTRODUCTION

Enteric hyperoxaluria (EH) can be caused by any gastrointestinal condition that is associated with fat malabsorption. The current assumption is that increased delivery of fat to the large intestine results in greater complexation of calcium with fatty acids, leading to less calcium available to form an insoluble complex with oxalate, resulting in an increase in free oxalate that can then be absorbed [1]. Increased delivery of bile acids to the colon in those patients with diseases of the distal ileum may also increase intestinal permeability and contribute to increased oxalate absorption. EH causes include malabsorptive bariatric surgical procedures [2–8], inflammatory bowel disease (especially including those after ileocecal resection) [9], exocrine pancreatic insufficiency and other disorders associated with intestinal malabsorption [10–13]. Given the high prevalence of morbid obesity [14] and resulting number of bariatric surgeries performed (roughly 200 000/year) [15], patients who have had bariatric surgical procedures are a particularly large subset of EH patients in the USA. While these surgeries offer significant benefits including resolution of diabetes and improved overall survival rates [16–18], the development of hyperoxaluria, kidney stones, nephrocalcinosis and renal failure are known complications [2,3,5–8,19].

It is assumed that hyperoxaluria is the primary lithogenic factor in patients with EH, although the underlying gastrointestinal malabsorption can cause additional urinary changes that favor stone formation, including a more concentrated urine due to fluid losses and low pH and citrate due to bicarbonate loss [1,5]. Interventions to treat EH are often aimed at all of these factors simultaneously [20]. Aside from a low-fat, low oxalate diet and use of generous calcium supplements with meals to bind oxalate, there are currently no additional targeted therapeutic options. To fill this gap, novel new agents under development have reported promising early results for reducing urinary oxalate (UOx) [21,22]. In this study, we utilized a large clinical database of EH patients and the infrastructure of the Rochester Epidemiology Project (REP) [23] in order to explore and quantify the relationship between the degree of hyperoxaluria and urinary stone disease risk.

MATERIALS AND METHODS

Study population

This retrospective cohort study was approved by the Mayo Clinic Institutional Review Board. We queried a database of all Mayo Clinic Rochester patients who had completed 24-h urine supersaturation studies between 1 January 2000 and 24 January 2017 (n = 13 092). EH cases were defined as adults ≥18 years old with a prior history of bariatric surgery or other malabsorptive enteric disorder as well as a 24-h UOx excretion above the upper limit of normal for Mayo Clinic Renal Testing Laboratory (≥40 mg/day) obtained after the enteric diagnosis date. Disorders were identified by international classification of diseases (ICD-9/ICD-10) code associated with malabsorption or Current procedural terminology (CPT) code for bariatric surgery (Supplementary data, Table S1) and validated with manual chart review. Patients with primary hyperoxaluria and patients with a known primary kidney stone composition other than calcium oxalate monohydrate or calcium oxalate dihydrate were excluded since these other stone types are less likely to be associated with EH.Figure 1 contains a flow diagram of patients included in the study after the exclusion criteria were applied, including the subset of patients from Olmsted County, ascertained with resources of the REP [23]. Subjects in the REP are part of a unique linked records system with reliable capture of clinical data that provides the infrastructure to perform population-based studies on health outcomes. Thus, only EH patients who resided in Olmsted County (n = 297) were used for the outcome study (kidney stone events), while the remainder were used as a larger descriptive and comparative group (n = 845). The baseline date (or date of EH diagnosis) was the date of the first documented 24-h UOx excretion ≥40 mg that followed the date of the first diagnosis of a malabsorptive enteric disorder (date of enteric disorder diagnosis). Last follow-up date was the last date a patient was seen for care at Mayo Clinic (up to 30 November 2017).

FIGURE 1.

Study flow diagram.

Urine chemistries

Urine studies were performed for one of two reasons: (i) as part of a metabolic workup for kidney stone risk factors (patients found to have had at least one prior kidney stone episode;n = 69 in the Olmsted County cohort) or (ii) as part of routine follow-up care after bariatric surgery as a screen for hyperoxaluria and/or hypercalciuria (described previously;n = 228 in the Olmsted County cohort) [5]; this group was without a stone event between the surgery and 24-h urine collection although the majority had a previous stone event prior to their bariatric surgery [5]. The stone events after an EH diagnosis were a mixture of presumed incident and known recurrent stones as 23% of the Olmsted County portion of the cohort had a prior kidney stone history (Table 1). The supersaturation panel was analyzed in the Mayo Clinic Renal Testing Laboratory and included urine volume, pH, citrate, oxalate, calcium and all necessary components to calculate calcium oxalate supersaturation by EQUIL2 [24]. Samples with a total urine volume <500 mL and urine creatinine <600 mg were rejected as implausible. We included mean urine calcium and mean urine citrate as covariates in statistical analysis.

Table 1.

Baseline characteristics of patients with EH

	Olmsted Co. cohort (n = 297)	Non-Olmsted Co. cohort (n = 845)	P-value
Demographics
Age, years	51.4 ± 11.4	55.0 ± 11.8	<0.001
BMI, kg/m2	34.3 ± 7.4	33.5 ± 8.9	0.13
Gender: female	201 (67.7)	536 (63.4)	0.19
Caucasian	286 (96.9)	790 (93.5)	0.10
History of kidney stone event prior to EH diagnosis	69 (23.2)	280 (33.1)	<0.001
Hypertension	183 (61.6)	510 (60.4)	0.75
Diabetes	114 (38.4)	327 (38.7)	0.98
Follow-up, years	4.9 (2.8–7.8)	2.9 (0.7–5.9)	<0.001
EH etiology			<0.001
Roux-en-Y gastric bypass	172 (57.9)	323 (38.2)	–
Gastric band	3 (1.0)	3 (0.4)	–
Sleeve gastrectomy	7 (2.4)	12 (1.4)	–
Other bariatric surgerya	16 (5.4)	72 (8.5)	–
History of bariatric surgeryb	76 (25.6)	274 (32.4)	–
Inflammatory bowel diseasec	11 (3.7)	67 (7.9)	–
Cystic fibrosis	1 (0.3)	2 (0.2)	–
Chronic pancreatitis	4 (1.3)	16 (1.9)	–
Other intestinal malabsorption	7 (2.4)	76 (9.0)	–
Urine studiesd
Baseline UOx	55.4 (46.6–73.0)	54.6 (44.9–71.3)	0.25
Baseline UOx ≥60	120 (40.4)	335 (39.6)	0.82
Baseline urine citrate	490 (230.0–761.0)	448 (195–755)	0.26
Baseline urine calcium	115 (67.0–181.0)	101 (54–164)	0.013
Renal function
Baseline eGFRe	89.2 ± 28.3	79.1 ± 31.8	<0.001

Values are expressed in mean ± SD or median (IQR) if not normally distributed, orn (%).

^aOther bariatric surgery types as outlined inSupplementary data, Table S1 including various combination restrictive/malabsorptive procedures.

^bPrior history of bariatric surgery at another facility.

^cIncluding Crohn’s disease and ulcerative colitis.

^dValues are in mg/24 h.

^eeGFR is in mL/min/1.73 m².

BMI, body mass index; eGFR, estimated glomerular filtration rate.

Kidney stones events

Kidney stone events were defined as any new kidney or ureteral stone ICD-9/ICD-10 diagnostic code(s) with or without an associated stone procedure CPT code(s) (Supplementary data, Table S2) occurring ≥30 days after the qualifying 24-h urine value. A prior history of kidney stones was determined by the presence of one or more kidney stone events prior to the diagnosis date of EH. In the event of multiple diagnoses or procedure codes for a subject, events were considered unique during follow-up if separated by ≥30 days. Outcomes assessed after EH diagnosis included time to first stone event for time to event analysis and any stone event (yes/no) for logistic regression analysis during follow-up. Analyses for stone events after the diagnosis of EH were limited to the patients who were residents of Olmsted County.

Statistical analysis

For baseline characteristics, continuous variables were reported as mean ± standard deviation (SD) or median with interquartile ranges (IQRs) for non-normally distributed variables. Categorical variables were expressed asn, %. Comparison of continuous variables was performed using two-sample Student’st-test or Wilcoxon rank-sum test while comparison of proportions between groups was made using the Chi-square test. The study period for each patient was defined by the baseline date of a qualifying UOx measurement until date of last follow-up. Patients were censored at time of last follow-up or end of study period if no events occurred.

We determined the association between stone events and baseline UOx using both logistic regression and accelerated failure time (AFT) models after adjustment for urine calcium and citrate due to their previously known association with stone formation [25–27]. Using logistic regression, we assessed the relationship between baseline UOx, urine calcium and urine citrate (all on the log scale) and for the occurrence of any stone event during follow-up. The output is$\mathrm{Logit}\left({\mathrm{p}}_{\mathrm{event}}\right) = {\widehat{\beta }}_0+{U\mathrm{Ox}·\widehat{\beta }}_{\mathrm{UOx}}+{\mathrm{Ca}·\widehat{\beta }}_{\mathrm{Ca}}+{\mathrm{Citrate}·\widehat{\beta }}_{\mathrm{Cit}}$ with variance–covariance matrix for the$\widehat{\beta }$ parameter estimates (Table 2 andSupplementary data, Table S4) with resulting plot of the probability of an event during follow-up (median 4.9 years) versus baseline UOx level with a confidence envelope. The confidence envelope was estimated via Taylor’s expansion using the Delta method and the parameter variance–covariance matrix. This association was used to estimate the reduction in odds of any stone event during follow-up given a decrease in UOx.

Table 2.

Parameter estimates from logistic regression modeling any stone event after EH diagnosis

Variable	Estimate (SE)	P-value
Log(baseline UOx)	1.29 (0.44)	0.004
Log(baseline urine calcium)	0.35 (0.23)	0.12
Log(baseline urine citrate)	−0.09 (0.18)	0.64
$\mathrm{P}\left(\mathrm{stone}\mathrm{ }\mathrm{event}\right)=\frac{{\mathrm{e}}^{-7.42+1.29\mathrm{*}\log \left(\mathrm{oxalate}\right)+0.35\mathrm{*}\log \left(\mathrm{calcium}\right)-0.09\mathrm{*}\log \left(\mathrm{citrate}\right)}}{1+{\mathrm{e}}^{-7.42+1.29\mathrm{*}\log \left(\mathrm{oxalate}\right)+0.35\mathrm{*}\log \left(\mathrm{calcium}\right)-0.09\mathrm{*}\log \left(\mathrm{citrate}\right)}}$

We used AFT models to examine the relationship between baseline UOx and the risk of having a stone event over time in order to predict the annual risk of a stone event. The AFT model included log time to first stone event as the dependent variable and covariates for baseline UOx, urine calcium and urine citrate [28]. The output is$\mathrm{Log}\left(\mathrm{time}\right)={\widehat{\beta }}_0+{\mathrm{UOx}·\widehat{\beta }}_{\mathrm{UOx}}+{\mathrm{Ca}·\widehat{\beta }}_{\mathrm{Ca}}+{\mathrm{Citrate}·\widehat{\beta }}_{\mathrm{Cit}}$ with variance–covariance matrix for the$\widehat{\beta }$ parameter estimates (Table 3 andSupplementary data, Table S5).

Table 3.

Parameter estimates for AFT analysis modeling any stone event after EH diagnosis

Variable	Estimate (SE)	P-value
Log(baseline UOx)	−2.60 (0.84)	0.002
Log(baseline urine calcium)	−0.64 (0.43)	0.14
Log(baseline urine citrate)	0.15 (0.34)	0.66
$\mathrm{Annual}\mathrm{ }\mathrm{Event}\mathrm{ }\mathrm{Rate}={\mathrm{e}}^{\left\{-\left(16.932-2.604\log \left(\mathrm{oxalate}\right)-0.637\log \left(\mathrm{calcium}\right)+0.149\log \left(\mathrm{citrate}\right)\right){\mathrm{e}}^{-0.814}\right\}}$

RESULTS

Baseline characteristics

Baseline characteristics of the Olmsted County EH cohort (n = 297) and the comparator EH group seen at Mayo Clinic that resided outside of Olmsted County (n = 845) are outlined inTable 1. The majority of patients had prior history of bariatric surgery and prior history of stone disease. Patients with EH in Olmsted County had a mean age of 51 ± 11 years and the majority were White (96.9%) and female (67.7%). Median follow-up was 4.9 (2.8–7.8) years. The subset of patients with EH residing in Olmsted County were more likely to be younger, have prior history of stone disease, have longer follow-up, have history of bariatric surgery and have higher baseline glomerular filtration rate. The 24-h urine studies including UOx and citrate were similar between groups; however, urine calcium was higher in the Olmsted County cohort (P = 0.013).

Kidney stone episodes in the Olmsted County cohort

Overall, 81 patients (27.3%) had at least one stone episode during follow-up (Table 4) and 43 patients (14.5%) had four or more stone episodes. Over half of the patients (n = 55) with a stone episode required a urologic procedure. The median time to a stone episode after the EH diagnosis was 0.35 (0.19–1.27) years.Supplementary data, Table S3 compares the baseline characteristics between stone-formers and nonstone-formers during follow-up. Of note, stone-formers had longer follow-up and higher baseline UOx, and were less likely to have undergone bariatric surgery.

Table 4.

Kidney stone events during followup after EH diagnosis in the Olmsted County cohort

Stone episodes	N = 297
Patients with stone episodes,n (%)	81 (27.3)
Stone episodes/patient,n (%)
0	216 (73)
1	15 (5.1)
2	14 (4.7)
3	9 (3.0)
4+	43 (14.5)
Time to first stone event (years)	0.35 (0.19–1.27)

Values are expressed inn (%) or median (IQR).

Association between UOx and kidney stones during follow-up

By logistic regression (Table 2 andSupplementary data, Table S4), the probability of a stone event occuring during follow-up was positively associated with baseline UOx (P < 0.01) after adjusting for baseline urine calcium (P = 0.12) and citrate (P = 0.64). During a median of 4.9 years of follow-up, a baseline UOx of 65 mg/24 h predicted a 30% probability of a stone episode, and a baseline UOx of 90 mg/24 h predicted a 40% probability of a stone episode (Figure 2).

FIGURE 2.

The probability of any stone event as a function of baseline UOx.

UOx as a predictor of annual risk of kidney stone events

We performed AFT modeling to predict the absolute risk of a kidney stone event in terms of an annual event rate (Table 3 andSupplementary data, Table S5). Consistent with our logistic regression analysis, there was a significant association of baseline UOx (P = 0.001) after adjusting for baseline urine calcium (P = 0.14) and citrate (P = 0.66), with kidney stone events. As seen inFigure 3, the annual event rate increased with increasing baseline UOx. For example, an EH patient with baseline UOx of 70 mg/24 h would have a predicted 20% annual risk of a kidney stone episode. Other AFT approaches, such as Weibull, LogLogistic and logNormal, gave rise to the same conclusions (Supplementary data, Table S6).

FIGURE 3.

Annual event rate of stone development as a function of baseline UOx.

Estimate of UOx reduction and subsequent risk of kidney stone events

While UOx reduction was not directly studied in this protocol, it is possible to estimate the effect of reducing UOx on subsequent kidney stone risk can by using estimates derived from the logistic regression model parameters.Table 5 provides examples of the decrease in odds of having a stone event that were estimated from a given shift in baseline UOx. For example, a 20% lower baseline UOx was associated with a 25.1% reduction in the odds of having a subsequent stone event (odds ratio = 0.75, 95% CI 0.62–0.91).

Table 5.

Estimated reduction in odds of a stone event after EH diagnosis for a given decrease in urine oxalate derived from the logistic regression analysis

% decrease in UOx	Odds ratio of stone (95% CI)	% reduction in odds (95% CI)
50	0.41 (0.22–0.75)	59.2 (25.5–77.6)
40	0.52 (0.33–0.81)	48.3 (19.5–66.9)
30	0.63 (0.46–0.86)	36.9 (14.0–53.7)
20	0.75 (0.62–0.91)	25.1 (9.0–38.3)
10	0.87 (0.80–0.96)	12.7 (4.4–20.4)

DISCUSSION

This study demonstrates that progressively higher 24-h UOx excretion predicts progressive increase in the risk of subsequent kidney stone events in patients with EH. Any cause of fat malabsorption can lead to EH, with the caveat that an intact colon is necessary since that is the site of oxalate absorption in this disorder. The most common cause in our cohort was malabsorptive bariatric surgical procedures. An earlier study of Olmsted County patients suggested that the risk of kidney stones is approximately doubled in patients after a standard Roux-en-Y gastric bypass compared with obese non-operated controls [5]. Other enteric conditions associated with fat malabsorption and EH include inflammatory bowel disease with or without small intestinal resections, celiac sprue, small bowel syndrome from any cause and exocrine pancreatic insufficiency. In addition to increased UOx excretion, other urinary factors can also change when gastrointestinal conditions cause malabsorption, potentially amplifying the propensity for calcium oxalate crystallization and kidney stone formation. Examples include decreased urine volume, urine citrate and pH. Although it has long been appreciated that hyperoxaluria is the major underlying factor associated with increased risk of calcium oxalate kidney stones in patients with fat malabsorptive syndromes, the relationship between UOx and kidney stone risk has not been quantified. Therefore, this study was conducted to provide a framework to define the quantitative relationship between the severity of hyperoxaluria and kidney stone risk in this population.

It was previously reported that progressively higher UOx is associated with a higher relative risk of kidney stones [25]. However, studies evaluating predictors of kidney stone recurrence have not shown a direct relationship between UOx and subsequent kidney stone events [26,29,30]. A plausible explanation for the lack of an association between UOx and stone recurrence is that these analyses were conducted on a general stone-forming population, and not in a population specifically at risk for hyperoxaluria. In fact, in these studies, the mean UOx was within the normal range [26,29,30]. One study did, however, reveal a strong association between calcium oxalate supersaturation and stone recurrence [30]. While many urinary ions used by EQUIL2 to calculate calcium oxalate supersaturation [24], UOx is particularly a significant contributor [31]. Thus one would predict that the substantial hyperoxaluria observed in EH could increase kidney stone risk. Other limitations with existing studies regarding UOx and kidney stone risk are related to the lack of reliable 24-h urine chemistries themselves, due to inconsistency in the collection of samples, assay variability, clinical interpretation of results and the effect of dietary changes [32]. In addition, there is significant biological variation in UOx [33]. Despite these limitations, 24-h urine collections remain the mainstay for evaluating risk for kidney stone formation and recurrence, and to monitor response to interventions to ameliorate identified risk factors.

In this study, we evaluated patients with enteric disorders known to be associated with malabsorption, and limited the analysis to patients who had UOx ≥40 mg/24 h, since EH by definition requires hyperoxaluria. Overall, 27.3% had at least one stone episode during follow-up, and over half of the stone-formers had four or more episodes. A large proportion of nonstone-formers were bariatric surgery patients. This is possible because many of these patients underwent screening 24-h urine collections [5] that were not as part of metabolic evaluation for history kidney stone(s) and performed as part of a standardized follow-up after the surgery. Furthermore, stone-formers with prior history of symptomatic stone(s) episodes are more likely to develop kidney stone episodes during follow-up [26].

Our analysis demonstrated that a higher baseline UOx predicted a greater likelihood of experiencing a subsequent stone episode after adjusting for calcium and citrate (P < 0.01). Overall, there was a 30% probability of having a stone event with a baseline UOx 65 mg/24 h and 40% if the baseline UOx was 90 mg/24 h. In addition, the AFT model revealed that a higher UOx also predicts a higher rate of annual stone events. Lastly, the results of this analysis enable an estimate of the magnitude of the potential effect of reducing UOx on subsequent kidney stone events. Even when the baseline UOx was only 20% lower, the estimated risk of a subsequent stone event fell by 25%, which would be clinically meaningful for patients who suffer frequent kidney stone recurrences. These findings should allow providers to discuss the potential benefit of interventions to reduce UOx on future stone episodes.

Typical management recommendations for EH include a strict low-fat, low oxalate diet and a generous use of calcium supplements; the latter can bind oxalate to form an insoluble complex that is then eliminated in the stool [34]. In addition, oral citrate supplements may improve urinary citrate excretion [35], and bile acid resins can be considered for patients with conditions that predispose toward bile acid malabsorption [36]. Nevertheless, these measures are not entirely effective in reducing kidney stone recurrence [20]. Thus new strategies are needed to prevent kidney stones, and lowering UOx may protect patients from chronic kidney disease associated with recurrent kidney stones and/or oxalate nephropathy. Other ancillary therapies that have been studied include the use of probiotics, oxalate-degrading bacteria or oxalate-degrading enzymes, with the goal to reduce free oxalate available for absorption [22,37,38]. This study provides some estimates about the magnitude of effect a given reduction in UOx might have on the risk of kidney stones. It is not uncommon for EH patients to have UOx excretions that ∼100 mg/day. The results of this study suggest that even incremental reductions in UOx could have a meaningful effect on the risk of subsequent kidney stone events.

This study has certain limitations. This was a retrospective study performed at a single center, which relied on electronic medical records and billing information to identify patients with enteric disorders associated with malabsorption and define kidney stone events. The majority of this population was White, had previously undergone a bariatric procedure and had prior history of kidney stone disease. Thus this cohort may not be generalizable to the broader group of patients with EH. Additionally, the majority of patients were female, and kidney stones of other causes typically occur more commonly in males. However, women have disproportionately opted for bariatric surgery over the last decade and this historic pattern may be reversing. In addition, the cohort included many patients who had 24-h urine collections as part of an observational study rather than for recurrent kidney stones, and the risk of kidney stones is increased in patients with prior history of kidney stones. The bariatric surgery cohort also included a small number of patients who previously underwent a primarily restrictive procedure, which have not been typically associated with an increased risk of hyperoxaluria or kidneys stones. However, many of these patients did evidence of hyperoxaluria with a UOx >40 mg/24 h (required for inclusion in the analysis database), thus they may have other factors contributing to hyperoxaluria. Given the inconsistent availability of 24-h urine chemistries during follow-up, we did not assess the association between kidney stone events and the change in UOx over time. Given the retrospective study design, we did not control the diets of patients included in this study and did not have reliable information whether or not they received education on a low oxalate diet. Thus it is unclear whether measures to reduce hyperoxaluria influenced the rates of stone formation we observed. We did not have sufficient power to stratify by EH etiology or by history of stone disease prior to EH diagnosis. However, our main focus was a modifiable and measurable risk factor (24-h UOx excretion), rather than the measures used to try and influence it. Finally, we assessed clinical outcomes related to kidney stone events, and no radiographic changes and/or the development or growth of asymptomatic stones, which might have identified additional events. However, those kidney stone events that resulted in seeking medical attention are presumably the most relevant and patient-centered target of any therapeutic measure to treat patients with EH. Despite these limitations, this defined population provided a relatively detailed and complete follow-up to enable estimating the potential benefit to be derived from reducing UOx.

In the future, prospective studies that employ dietary or other measures to reduce UOx excretion are warranted in EH to determine whether they reduce the risk of clinical kidney stone events. These data would further support ongoing clinical trials aimed at lowering oxalate in stone-formers with hyperoxaluria.

SUPPLEMENTARY DATA

Supplementary data are available at ndt online.

FUNDING

Investigators on this project were supported by thenational institutes of health (NIH) (grantU54-DK100227 from theO’Brien Urology Research Center and the Rare Kidney Stone Consortium grantU54DK083908-01, part of theRare Diseases Clinical Research Network, an initiative of office of rare diseases research (ORDR), national center for advancing translational sciences (NCATS) and national institute of diabetes and digestive and kidney diseases (NIDDK)). Funding for this project for statistical analysis was provided by a grant from Allena Pharmaceuticals. A.T.K. reports being a salaried employee of Allena Pharmaceuticals during the conduct of the study. J.C.L. reports personal fees from American Board of Internal Medicine, grants from Dicerna, grants from Alnylam, grants from Retrophin and grants from OxThera, all of which were unrelated to this study.

CONFLICT OF INTEREST STATEMENT

The results presented in this article have not been published previously in whole or part, except in abstract format. M.R.D., A.T.K., K.J.C., J.P.I., F.T.E., J.C.L, K.C.M. and R.A.M. have nothing to disclose.