Relationship between uric acid and kidney function in adults at risk for tumor lysis syndrome

Abstract

Uric acid drives acute kidney injury in tumor lysis syndrome (TLS). This study investigated the relationship between uric acid and changes in estimated glomerular filtration rate (eGFR) in adults at risk for TLS. Linear regression was used to evaluate the relationship between uric acid area under the curve (AUC) and percent change in eGFR from baseline at hospital dismissal, 1 and 3 months. In 210 included participants, each 100 mg*hour/dL increase in 24 h AUC was associated with an average decline in eGFR at hospital dismissal of 9% (95%CI 3, 15) in univariate analysis. Each 100 mg*hour/dL increase in 24 h AUC was independently associated with an average decline in eGFR of 8% (95%CI 2, 13) at 1 month after dismissal. Additional research is needed to confirm these findings and determine whether treatments that reduce overall uric acid exposure improve kidney outcomes. Preserving kidney health could favorably impact cancer treatment eligibility, tolerability, and outcomes.

Introduction

Acute kidney injury (AKI) is a devastating disease- or treatment-related complication for patients diagnosed with non-Hodgkin lymphoma (NHL) and other hematologic cancers. Consequences of AKI are severe and include a 2-fold increased risk for in-hospital mortality, a tripling of inpatient costs, long-term decline in kidney function, limited access to preferred cancer treatment options and, ultimately, decreased disease response and 6-month survival [1–5].

While the cause of AKI is often multifactorial, as much as 20–48% of adults with NHL will experience AKI as a result of tumor lysis syndrome (TLS). TLS is a condition where cellular contents, such as electrolytes and nucleic acids, are released into systemic circulation and cause numerous toxicities [2,6,7]. Serum uric acid (UA), a byproduct of nucleic acid breakdown, has been proposed as a key driver of the pathophysiology of AKI in this setting. Uric acid levels can increase precipitously and cause kidney damage through crystal deposition in the renal tubules, oxidative stress, local inflammation, and renal vasoconstriction, which leads to ischemic injury [8,9]. Elevated serum UA concentration has been repeatedly linked with increased risk for TLS-associated AKI. Rising UA concentration has been correlated with decreasing estimated glomerular filtration rate (eGFR), the most commonly used marker of kidney function [6,7,10–13]. Use of medications to control or lower UA is a corner-stone of supportive care for patients with newly diagnosed NHL about to receive chemotherapy. Both allopurinol and rasburicase are commonly prescribed for this purpose. Rasburicase is known to reduce UA levels much more rapidly than allopurinol, as it solubilizes existent UA rather than only modifying its future production [14,15]. The clinical implications of this accelerated UA reduction remain unknown and there is limited data evaluating the impact of longitudinal UA exposure in the context of TLS on kidney function outcomes. Such knowledge would advance our understanding of the importance and role of UA-lowering medications and inform decisions regarding the ideal candidate for and timing of these therapies.

The primary purpose of this study was to investigate the relationship between UA exposure and changes in kidney function in adults with newly diagnosed NHL and at risk for TLS and AKI. We hypothesized that increased UA exposure, as measured by the area under the curve (AUC), would be associated with a decrease in eGFR from baseline at hospital dismissal and within the 3 months after hospitalization.

Materials and methods

Study design and sample

A population-based cohort study was performed from 2005 to 2017 using the Rochester Epidemiology Project, a medical records linkage system that unifies records from multiple medical care providers in Olmsted County, Minnesota and the surrounding 27 counties. The Rochester Epidemiology Project includes demographic data and comprehensive information related to diagnoses, hospital admissions and outpatient follow-up care [16]. Residents have similar age- and sex-specific mortality rates to those found in Minnesota and the United States [17]. Adult participants (age >18) were identified using the first documented International Classification of Diseases code (9th and 10th editions) for NHL (Supplemental Table S1) within the study timeframe that corresponded with pathology or hematopathology studies performed within 60 days. Those diagnosed in 2005 (first year of the study timeframe) were evaluated for a malignancy diagnosis assigned in the preceding year in order ensure no prevalent cases were included. Participants were included only once, at the first hospitalization within the study timeframe. Those with end-stage kidney disease requiring dialysis at the time of study entry were excluded along with those who lacked adequate UA data for AUC calculation. Participants must have had at least one UA measurement within 24 h of hospital admission and per 24 h period for which the AUC was calculated. Demographic data collected included age at diagnosis, NHL subtype, and a history of cardiovascular or chronic kidney disease, diabetes, or hypertension. Comorbidities were identified using International Classification of Diseases codes. Diagnosis with one of these comorbidities required 2 codes separated by at least 30 days within 3 years of hospitalization. A pre-admission serum creatinine (SCr) was established for each member of the cohort using the median outpatient SCr values from the period of 6 months up to 7 days prior to hospitalization or, if unavailable, estimated using the MDRD formula [18]. This SCr was also used to determine pre-admission eGFR, referred to as a participant’s ‘baseline eGFR’, calculated using the Chronic Kidney Disease Epidemiology Collaborative equation. Routine laboratory data collected from the hospital encounter included SCr measurements, which were used to identify and stage AKI per the KDIGO criterion [19]. AKI was classified as community- or hospital-acquired based on time of onset (within 48 h of hospitalization for community-acquired AKI, >48 h for hospital-acquired AKI) [20].

Allopurinol is the standard of practice for TLS prevention at the included practice sites and is prescribed in all patients with NHL undergoing chemotherapy, including this study sample, due to the moderate or high risk for TLS [6,21]. Additional data was collected on the administration and timing of rasburicase and dialysis.

Determination of uric acid exposure

The AUC was calculated in mg*hour/dL by the linear trapezoidal method for the first 24, 48 and 72 h of hospitalization. At Mayo Clinic, where the majority of participants were hospitalized, it is the standard of care to assess UA levels every 8 h in adults who are recognized to be at risk for TLS and are about to receive chemotherapy. The last value carried forward method was used to estimate missing data.

Outcome assessment

We assessed the relationship between the AUC for the first 24 h of admission (24 h AUC) and percent change in eGFR from baseline at hospital dismissal and 1 (± 14 days) and 3 months (± 30 days) following hospitalization. Serially collected outpatient SCr values available in the electronic health record were used to calculate eGFR at these intervals or last known follow-up within this timeframe. The percent change in eGFR was determined by comparison to participants’ baseline values. Additional, exploratory analyses were performed using the AUC for the first 48 and 72 h of admission.

Statistical analysis

Descriptive statistics, including median and interquartile range (IQR) or counts and percentages, summarize baseline variables. Pearson’s correlation coefficient was used to describe the association between select baseline variables and 24 h AUC. The t-test for the difference in means was used to compare 24 h AUC between those with and without AKI during hospitalization. We examined the association between AUC and percent change in eGFR from baseline, the primary endpoint of interest, using linear regression. The following clinically relevant variables were included in multivariable linear regression: age, NHL subtype, baseline and admission eGFR, rasburicase use, AKI during hospitalization, and hospital readmission (for the 1- and 3-month models). Models were optimized using backward stepwise selection to avoid over-fitting. The partial F-test was used as the test for significance when evaluating NHL subtype in multivariable models. To determine if AKI status or rasburicase use modified the effect of 24 h AUC on eGFR change from baseline at hospital dismissal, linear regression was used to test for an interaction between AKI and 24 h AUC and rasburicase and 24 h AUC. An alpha level of 0.05 was set as the threshold for statistical significance and we report 95% confidence intervals.

Results

Participants

There were 574 individuals with newly diagnosed NHL and free from end-stage kidney disease identified within the study timeframe (Figure 1). Of these, 210 (37%) had a UA measured within the first 24 h of hospitalization. Table 1 describes characteristics of included and excluded individuals. Excluded individuals were older, on average, and more often received a cancer diagnosis during the hospitalization.

Figure 1.

Participant flowchart.

Table 1.

Cohort characteristics and uric acid area-under-the-curve.

	Included participants (n = 210)	Excluded individuals (n = 364)
Baseline Characteristics
Age, years	65 (54, 74)	71 (58, 82)
Male sex	136 (65)	197 (54)
Race
White	195 (93)	344 (95)
Asian	3 (1)	3 (1)
Black	2 (1)	4 (<1)
Other	6 (3)	8 (2)
Unknown/Refused	4 (2)	5 (<1)
Body mass index, kg/m2	28 (25, 32)	27 (24, 32)
Cardiovascular disease	28 (13)	66 (18)
Diabetes	32 (15)	65 (18)
Hypertension	63 (30)	140 (38)
Non-Hodgkin lymphoma subtype
Diffuse large B-cell lymphoma	74 (35)	87 (24)
Burkitt lymphoma	20 (10)	3 (<1)
Follicular lymphoma	19 (9)	34 (9)
Lymphoblastic lymphoma	8 (4)	–
T-cell lymphoma	7 (3)	16 (4)
Other	82 (39)	196 (54)
Lymphoma diagnosis during hospitalization	74 (35)	231 (63)
In-hospital death	8 (4)	21 (6)
Length of hospital stay, days	5 (3, 9)	–
Baseline SCr, mg/dL	0.9 (0.8, 1.1)	–
Baseline eGFR, ml/min/1.73m2	80 (62, 94)	–
Baseline eGFR < 60 ml/min/1.73m2	49 (23)	–
Admission SCr, mg/dL	1.2 (0.8, 1.3)	–
Admission eGFR, ml/min/1.73m2	74 (51,92)	–
eGFR at hospital dismissal, ml/min/1.73m2	87 (66, 101)	–
Acute kidney injury during hospitalization	57 (27)	–
Community-acquired acute kidney injury	43 (75)	–
Received rasburicase during hospitalization	36 (17)	–
Admission UA, mg/dL	6.4 (4.5, 8.5)	–
Admission UA > 8 mg/dL	62 (30)	–
UA 24-h AUC, mg*hour/dL	138 (99, 182)	–
UA 48-h AUC, mg*hour/dLa	258 (171,345)	–
UA 72-h AUC, mg*hour/dLb	359 (247,513)	–

Continuous data reported as median (IQR); Count data reported as n, (%).

UA, uric acid; SCr serum creatinine; eGFR, estimated glomerular filtration rate; AUC, area-under-the-curve.

^aAssessed in n=120.

^bAssessed in n=83.

Hospital course description

Fifty-seven (27%) participants experienced an episode of AKI while hospitalized. The maximal stage was most often stage 1 (n = 33) followed by stages 2 (n = 12) and 3 (n = 12). The mean 24 h AUC was 84 (95% CI 58, 109) mg*hour/dL higher in participants with AKI compared to those free from AKI (p < 0.001). Six participants received dialysis during hospitalization at a median of 4 (2, 11) days. Additional characteristics described in Supplemental Table S2. Rasburicase was administered to 19 participants with AKI and 17 participants free from AKI. It was given within 24 h of hospitalization in 29 of the 36 recipients. Amongst those treated with rasburicase, median admission SCr and eGFR was 1.4 (1.1, 1.7) mg/dL and 48 (36, 78) ml/min/1.73m², respectively. Additional details and AUC results are described in Table 1.

Baseline eGFR was not correlated with 24 h AUC (r = −0.09), however admission eGFR was moderately correlated with 24 h AUC (r = −0.42). Admission UA was highly correlated with UA 24 h AUC (r = 0.89), 48 h AUC (r = 0.70), and 72 h AUC (r = 0.59).

Outcomes

All 210 participants had an available eGFR at hospital dismissal and 149 and 124 had an available eGFR at 1 and 3 months following hospitalization, respectively. Twenty-four participants (11%) died in the 90-day follow-up period.

Results from univariate linear regression are shown in Table 2. Increasing AUC at 24, 48, and 72 h was associated with a decrease in eGFR from baseline at hospital dismissal (p < 0.05 for all models). Each 100 mg*hour/dL increase in 24 h AUC was associated with an average decline in eGFR at hospital dismissal of 9% (95% CI 3, 15). The 24 h AUC was not significantly associated with change in eGFR at 1 or 3 months following hospital dismissal in univariate models (p = 0.13 and p = 0.98, respectively). NHL subtype was not associated with a change in eGFR at any time point (p > 0.05 for all models).

Table 2.

Percent change in in eGFR from baseline at hospital dismissal and 1 and 3 months in univariate linear regression.

	Hospital dismissal		1 month		3 months
	Estimate (95% CI)	p Value	Estimate (95% CI)	p Value	Estimate (95% CI)	p Value
24 h AUC (per 100 mg*hour/dL increase)	−8.8 (−14.9, −2.6))	0.005	−5.7 (−13.0, 1.6)	0.13	0.08 (−7.4, 7.5)	0.98
48 h AUC (per 100 mg*hour/dL increase)	−6.4 (−11.7, −1.2)	0.017	−1.7 (−7.6,4.3)	0.57	−1.8 (−7.8,4.2)	0.54
72 h AUC (per 100 mg*hour/dL increase)	−5.3 (−10.4, −0.2)	0.042	−2.4 (−7.8, 3.0)	0.37	−0.3 (−5.4, 4.8)	0.91
Age (per 10-year increase)	−1.7 (−4.5, 1.1)	0.22	1.3 (−1.9,4.5)	0.42	1.7 (−1.7, 5.1)	0.33
Baseline eGFR (per 10 ml/min/m2 increase)	−4.3 (−6.0, −2.5)	<0.001	−6.7 (−8.5, −5.0)	<0.001	−6.3 (−8.1, −4.5)	<0.001
Admission eGFR (per 10 ml/min/m2 increase)	0.1 (−1.5, 1.8)	0.25	−3.2 (−5.0, −1.4)	<0.001	−3.9 (−5.7, −2.1)	<0.001
Acute kidney injury	−28.5 (−38.4, −18.5)	<0.001	−15.7 (−27.9, −3.5)	0.01	−7.5 (−20.4, 5.4)	0.25
Rasburicase	13.0 (0.6,25.5)	0.04	19.7 (5.6, 33.7)	0.006	12.0 (−2.6,26.6)	0.11
Hospital readmission	–	–	6.3 (−4.5, 17.0)	0.25	1.4 (−10.5, 13.3)	0.80

There was no significant interaction between AKI status (p = 0.92 for the interaction term) or rasburicase use (p = 0.18 for the interaction term) and 24 h AUC, when considering change in eGFR from baseline at hospital dismissal as the outcome variable.

Multivariable models are reported in Table 3. After adjustment for age, AKI, baseline eGFR, and rasburicase administration, 24 h AUC was not significantly associated with a change in eGFR at hospital dismissal (adjusted R² 0.38, p = 0.71 for 24 h AUC). However, increasing 24 h AUC was significantly associated with greater decline in eGFR at 1 month, after adjusting for age, baseline eGFR, and hospital readmission (adjusted R² 0.45, p = 0.007 for 24 h AUC). Each 100 mg*hour/dL increase in 24 h AUC was associated with an average decline in eGFR at 1 month of 8% (95% CI 2, 13). Rasburicase use and AKI were not significant predictors of eGFR change at 1 month. Only age and baseline eGFR remained significant predictors of eGFR change at 3 months.

Table 3.

Percent change in in eGFR from baseline at hospital dismissal and 1 and 3 months in multivariable linear regression using backward stepwise regression.

	Hospital dismissal		1 month		3 months
Adjusted R2	0.38		0.45		0.40
	Estimate (95% CI)	p Value	Estimate (95% CI)	p Value	Estimate (95% CI)	p Value
UA 24 h AUC (per 100 mg*hour/dL increase)	−1.1 (−6.9, 4.7)	0.71	−7.7 (−13.3, −2.2)	0.007	−0.8 (−6.6, 4.9)	0.77
Age (per 10 year increase)	−9.1 (−12.0, −6.2)	<0.001	−8.0 (−11.1, −4.9)	<0.001	−8.8 (−12.2, −5.3)	<0.001
Baseline eGFR (per 10 ml/min/m2 increase)	−7.6 (−9.5, −5.7)	<0.001	−10.5 (−12.5, −8.6)	<0.001	−9.9 (−12.0, −7.8)	<0.001
Acute kidney injury	−29.7 (−39.5, −19.8)	<0.001	–	–	–	–
Rasburicase	17.5 (7.1,27.9)	0.001	–	–	–	–
Hospital readmission	–	–	10.5 (2.4, 18.5)	0.012	–	–

Data not reported for variables that did not retain statistical significance in the models, except for the primary predictor of interest, UA 24 h AUC.

Discussion

Results showed that increasing UA exposure, as measured by AUC, was not an independent predictor of eGFR at hospital dismissal, after adjusting for confounding variables including AKI and rasburicase use. However, increasing 24 h AUC was independently associated with a decline in eGFR from baseline at 1 month following hospital dismissal in multivariable modeling.

There has been much debate over the relationship between UA levels and kidney function, as measured by eGFR. Elevated serum UA is hypothesized to cause eGFR decline through crystal deposition in the renal tubules, oxidative stress, and local inflammation leading to renal vasoconstriction and ischemic injury [8,9,22,23]. It can also be an indirect marker of kidney function. In the setting of chronic kidney disease, serum UA increases due to enhanced reabsorption and decreased secretion by the renal tubules and reduced glomerular filtration [13,24]. Outside of the context of TLS, increasing UA levels are associated with albuminuria, arteriolar damage, onset and progression of early-stage chronic kidney disease, and increased risk for cardiovascular and all-cause mortality [22–27]. A prior study showed decreasing serum UA levels correlated with increases in eGFR over a 4-day period in adults with acute myeloid leukemia [12]. Authors proposed that the use of UA-lowering therapy as the primary mechanism of UA elimination and a weak association between UA level and eGFR at day 0 suggest the observed relationship was not due to changes in GFR altering UA levels. While that study examined the relationship between concurrently measured UA and GFR, we sought to determine if UA exposure could predict future changes in GFR (e.g. hospital discharge). Increased renal vasoconstriction and crystal deposition in the context of higher AUC represents a biologically plausible mechanism for persistent kidney damage evidenced by eGFR decline [8,9]. Our data also indicate that eGFR declines with increasing UA exposure, suggesting the effect of UA exposure on eGFR may be durable. AKI and rasburicase use also significantly impacted eGFR in the short term. Their effect on eGFR at hospital dismissal highlight the complexity of the relationship between UA management, AKI, and kidney function outcomes. Further study in a controlled setting is warranted to determine if UA exposure has a clinically significant effect on eGFR in the setting of newly diagnosed cancer.

Understanding the relationship between UA exposure and kidney function outcomes has important implications for UA-lowering therapy prescribed to prevent or treat TLS and associated AKI. Approximately 70% of AKI episodes in adults with newly diagnosed NHL and at risk for TLS are community-acquired, indicating underlying kidney damage is likely ongoing at the time of hospitalization [6]. This may partially account for the lack of evidence to suggest rasburicase use favorably impacts risk for important kidney function outcomes, such as AKI prevention, in the context of TLS [6,28]. Even if unable to prevent AKI compared to the standard of care (allopurinol and hydration), rasburicase drastically lowers UA levels in a short time frame. Rapid reduction in UA exposure through use of drugs or dialysis would have significant positive impact for patients if it improved clinical outcomes. Despite an established correlation between UA and kidney outcomes, several randomized controlled trials and meta-analyses demonstrated that UA-lowering therapy with allopurinol or febuxostat did not significantly slow eGFR decline in patients with moderate chronic kidney disease [29–31]. Conversely, a study of 40 elderly adults with hyperuricemia (outside the context of TLS) showed a single dose of rasburicase resulted in lower serum urate and creatinine concentrations and increased creatinine clearance at 1- and 2-month follow-up compared to placebo[32]. Inconsistent findings may be related to heterogeneity of study designs [33]. Additionally, these results may not be fully generalizable to the population in the present study, patients with newly diagnosed cancer and at risk for TLS and AKI. A prior report of children at high risk for TLS showed rasburicase significantly reduced UA 24 h AUC when compared to allopurinol alone, however the study was too small to determine if this resulted in a difference in kidney function outcomes [28]. It is possible rasburicase more effectively dissolves UA crystals in the renal tubules, which contribute to structural kidney damage, than other UA-lowering therapies and may therefore have different effects on kidney function outcomes than have been previously observed with allopurinol or febuxostat. This, or other mechanistic differences, may partially explain why administration of rasburicase was significantly associated with an increase in eGFR from baseline at hospital dismissal in our study, despite the fact AUC was not a significant predictor of eGFR change at this time point. This finding should be considered hypothesis-generating. Additionally, relatively few participants received rasburicase and sample size may have limited our ability to detect an important interaction between rasburicase and AUC, which may have affected the relationship between AUC and eGFR. More study is needed to determine if rapid, substantial reductions in UA, by rasburicase or other methods, portend favorable outcomes in adults at risk for TLS. If such therapies prove effective in mitigation of kidney injury and preservation of GFR, this would have a positive impact on cancer treatment eligibility and tolerability and improve patients’ quality of life and health outcomes [2,34–36].

This study has several limitations. Only 37% of those with newly diagnosed NHL had an available UA measurement within the first 24 h of hospital admission. This was likely due, in part, to a higher percentage of the excluded individuals receiving an NHL diagnosis during the hospitalization. These patients would be less likely to receive up-front monitoring of UA. Additionally, though the standard of care is to routinely monitor UA in patients at risk for TLS, individual providers may not have placed these orders promptly upon hospitalization. Patients lacking the necessary UA data could also have been hospitalized for reasons other than imminent chemotherapy or risk for TLS. The sample size and retrospective study design limited our ability to comprehensively evaluate risk factors for the outcome of eGFR decline and additional confounders or effect modifiers of the relationship between UA and eGFR may have been unaccounted for in multivariable models. We were able to account for many important characteristics however, including comorbidities, NHL subtype, UA-lowering therapy, and kidney function at baseline and during hospitalization. A significant association between AUC and eGFR change from baseline was not consistent in all analyses and additional studies are needed to validate these results or establish a causal relationship. There is concern about the validity of serum creatinine-based methods for estimating GFR in acutely ill patients. Despite this limitation, use of these methods represents real-world clinical practice due to their wide-spread availability and familiarity and lack of a feasible alternative, in many cases.

This study examines the relationship between UA and changes in kidney function in 210 adults with recently diagnosed NHL and at risk for TLS and AKI. Increasing UA exposure within the first 24 h of hospitalization was associated with greater decline in eGFR at 1 month after hospital dismissal, after adjusting for other clinically significant factors. Due to the limitations of this study and variable findings, additional research is critical to validate findings and further explore whether significant reductions in UA exposure, by rasburicase or other therapies, portend favorable kidney function outcomes. These data would have significant implications for determining the role of UA-lowering therapies in optimizing clinically important outcomes. Identification of methods to preserve kidney function is of the utmost importance for adults with cancer, as it greatly impacts treatment eligibility and tolerability and health outcomes.