Association Between Early Postoperative Acetaminophen Exposure and Acute Kidney Injury in Pediatric Patients Undergoing Cardiac Surgery

Key Points

Question  Does early postoperative acetaminophen exposure reduce the incidence of acute kidney injury in children who undergo cardiac surgery?

Findings  In this cohort study, acetaminophen exposure in the first postoperative 48 hours was associated with a lower rate of acute kidney injury in pediatric patients who underwent cardiac surgery in primary and validation cohorts from 2 tertiary referral children’s hospitals.

Meaning  Postoperative use of acetaminophen may protect pediatric cardiac surgery patients from acute kidney injury.

Importance  Acute kidney injury (AKI) is a common and serious complication for pediatric cardiac surgery patients associated with increased morbidity, mortality, and length of stay. Current strategies focus on risk reduction and early identification because there are no known preventive or therapeutic agents. Cardiac surgery and cardiopulmonary bypass lyse erythrocytes, releasing free hemoglobin and contributing to oxidative injury. Acetaminophen may prevent AKI by reducing the oxidation state of free hemoglobin.

Objective  To test the hypothesis that early postoperative acetaminophen exposure is associated with reduced risk of AKI in pediatric patients undergoing cardiac surgery.

Design, Setting, and Participants  In this retrospective cohort study, the setting was 2 tertiary referral children’s hospitals. The primary and validation cohorts included children older than 28 days admitted for cardiac surgery between July 1, 2008, and June 1, 2016. Exclusion criteria were postoperative extracorporeal membrane oxygenation and inadequate serum creatinine measurements to determine AKI status.

Exposures  Acetaminophen exposure in the first 48 postoperative hours.

Main Outcomes and Measures  Acute kidney injury based on Kidney Disease: Improving Global Outcomes serum creatinine criteria (increase by ≥0.3 mg/dL from baseline or at least 1.5-fold more than the baseline [to convert to micromoles per liter, multiply by 88.4]) in the first postoperative week.

Results  The primary cohort (n = 666) had a median age of 6.5 (interquartile range [IQR], 3.9-44.7) months, and 341 (51.2%) had AKI. In unadjusted analyses, those with AKI had lower median acetaminophen doses than those without AKI (47 [IQR, 16-88] vs 78 [IQR, 43-104] mg/kg, P < .001). In logistic regression analysis adjusting for age, cardiopulmonary bypass time, red blood cell distribution width, postoperative hypotension, nephrotoxin exposure, and Risk Adjustment for Congenital Heart Surgery score, acetaminophen exposure was protective against postoperative AKI (odds ratio, 0.86 [95% CI, 0.82-0.90] per each additional 10 mg/kg). Findings were replicated in the validation cohort (n = 333), who had a median age of 14.1 (IQR, 3.9-158.2) months, and 162 (48.6%) had AKI. Acetaminophen doses were 60 (95% CI, 40-87) mg/kg in those with AKI vs 70 (95% CI, 45-94) mg/kg in those without AKI (P = .03), with an adjusted odds ratio of 0.91 (95% CI, 0.84-0.99) for each additional 10 mg/kg.

Conclusions and Relevance  These results indicate that early postoperative acetaminophen exposure may be associated with a lower rate of AKI in pediatric patients who undergo cardiac surgery. Further analysis to validate these findings, potentially through a prospective, randomized trial, may establish acetaminophen as a preventive agent for AKI.

Acute kidney injury (AKI) occurs in approximately 25% of children admitted to intensive care units (ICUs).^1-9 Among pediatric ICU patients, AKI is associated with increased length of stay and consistently demonstrates odds ratios (ORs) exceeding 3 for mortality.^1,2,8-12 Among children who have cardiac surgery, rates of AKI are even higher, and AKI is associated with increased morbidity, mortality, and length of stay.^13-26 To date, efforts to reduce the effect of AKI in pediatric patients are focused on early detection and risk reduction because there are no known preventive or therapeutic agents for AKI.

For children who undergo cardiac surgery, the necessary cardiopulmonary bypass (CPB) causes hemolysis, which together with the tissue damage from surgery and administration of packed red blood cells leads to high circulating levels of free hemoglobin in these patients.^27,28 Cell-free plasma hemoglobin has been associated with AKI after cardiac bypass in children.²⁹ Based on data from animal models and adults, it is hypothesized that free hemoglobin contributes to AKI through the oxidation of the central iron of the heme moiety to the highly reactive and nephrotoxic ferryl (Fe⁴⁺) form.^30-32 This pathogenic mechanism for AKI illuminates a potential therapeutic agent. Acetaminophen, a well-established drug used for analgesia and fever reduction, is known to reduce the oxidation of free hemoglobin by preventing the oxidation of iron from Fe³⁺ to Fe⁴⁺.^32,33 The hemoreductant activity of acetaminophen derives from the structural similarity between the heme moiety of hemoglobin and the peroxidase moiety of cyclooxygenase. In recent years, acetaminophen has been increasingly used after surgery in children to reduce opioid exposure.

The primary objective of this study was to test the hypothesis that acetaminophen exposure in the first 48 hours after cardiac surgery is associated with reduced risk of AKI in children. To test this hypothesis, we performed a retrospective observational study of children who had cardiac surgery in a primary cohort from the Monroe Carell Jr Children’s Hospital at Vanderbilt, and we validated our findings in a second independent cohort from Duke Children’s Hospital.

This study was reviewed and approved by the institutional review boards of Vanderbilt University Medical Center, Nashville, Tennessee, and Duke University, Durham, North Carolina. Written informed consent was obtained from parents or guardian, as well as written assent from study participants as appropriate by age. Inclusion criteria for the primary cohort were cardiac surgery at the children’s hospital between July 1, 2008, and June 1, 2016, and age older than 28 days. Exclusion criteria were postoperative extracorporeal membrane oxygenation (because this may include renal replacement therapy and thus affect serum creatinine values) and absence of either the baseline or postoperative serum creatinine measurements to determine AKI status. If individuals had multiple cardiac surgical procedures, data from only the first surgery were included. After the primary cohort was analyzed, a second cohort was assembled to determine if these findings were unique to the institution or if they could be replicated in an independent validation cohort. Inclusion criteria for the validation cohort were the same as for the primary cohort, with the exception that the study inclusion dates were June 25, 2013, to March 29, 2016. All study data, including demographic, laboratory, medication administration, and administrative data, were extracted from electronic health record data sources. Medications administered in the first 48 postoperative hours were categorized as high-risk nephrotoxins (eg, aminoglycosides, ketorolac tromethamine, and tacrolimus), moderate-risk nephrotoxins (eg, loop diuretics, enalapril maleate, and trimethoprim-sulfamethoxazole [eTable 1 in the Supplement]), or nonnephrotoxins. Postoperative hypotension was defined using the systolic blood pressure measurements documented in the first 48 postoperative hours, irrespective of any vasopressor support. Hypotension was defined as one or more systolic blood pressures lower than 70 mm Hg plus 2 times the age in years if 10 years or younger and lower than 90 mm Hg if older than 10 years.³⁴

Exposure was defined as dichotomous exposure to acetaminophen for the initial analysis and then the cumulative weight-adjusted intravenous, enteral, and/or rectal acetaminophen dose (in milligrams per kilogram), alone or in combination with other medications, received in the first 48 hours after surgery. Individuals were also categorized as those receiving no acetaminophen or the lowest (<40 mg/kg), middle (40-80 mg/kg), or highest (>80 mg/kg) doses of acetaminophen. The AKI outcome was defined by change in serum creatinine values. Baseline serum creatinine measurements were defined as the lowest obtained in the 28 days before surgery. Postoperative serum creatinine measurements were defined as those obtained in the 7 days after surgery. Acute kidney injury was defined solely using the creatinine criteria because reliable urine output data are not available for all pediatric inpatients³⁵ and accurate height measurements were not consistently available among pediatric inpatients for estimated glomerular filtration rate. Quiz Ref IDIndividuals were classified as having AKI using Kidney Disease: Improving Global Outcomes (KDIGO)⁷ serum creatinine criteria if any postoperative serum creatinine measurement increased by 0.3 mg/dL or more from baseline (to convert creatinine level to micromoles per liter, multiply by 88.4) or was at least 1.5-fold more than the baseline in the first postoperative week. Those not meeting these criteria were classified as having no AKI. Acute kidney injury stage was determined for each AKI case using serum creatinine–based KDIGO definitions for stage 1 (≥0.3 mg/dL or 1.5-fold increase), stage 2 (2-fold increase), and stage 3 (≥4 mg/dL or 3-fold increase) based on the highest serum creatinine measurement in the 7 days after surgery.

Descriptive statistics were presented as the median (interquartile range [IQR]) or number (percentage) as appropriate. Continuous variables were compared using the Wilcoxon rank sum test and categorical variables using the Pearson χ² test. The Kruskal-Wallis test was used to test the weight-adjusted acetaminophen dose across AKI stages. We analyzed the association between AKI and dichotomous acetaminophen exposure using the Pearson χ² test. We also investigated the association of acetaminophen dose (continuous variable, in milligrams per kilogram) with AKI using logistic regression, after adjusting for prespecified risk factors, such as age, CPB time, red blood cell distribution width (RDW) given prior associations with AKI,^36,37 postoperative hypotension, nephrotoxin exposure, and Risk Adjustment for Congenital Heart Surgery (RACHS) score.³⁸ The nonlinear association of age with AKI was modeled using the restricted cubic spline with 3 knots chosen at the 10th, 50th, and 90th percentiles. The same logistic regression model was used for both the primary and validation cohorts. Subgroup analyses were also done. First, logistic regression was performed for each cohort using the same methods and covariates as in the full cohort, excluding those individuals who had no acetaminophen exposure. Quiz Ref IDIn addition, a subgroup analysis was done by pooling individuals from both cohorts with a RACHS score of 3 or higher as an alternative method to account for variability in clinical status after surgery and likely degree of cell-free hemoglobin. For this analysis, the same logistic regression was performed as above, with the inclusion of clinical site as an additional covariate. Sensitivity analyses of unmeasured confounding were also performed (see the eAppendix in the Supplement). Multiple imputation was performed based on 10 completed data sets by applying predictive mean matching to handle missing data for the variables of CPB time, RDW, and RACHS score.³⁹ All analyses were performed with a statistical programming language (R, version 3.3.0; R Development Core Team). The statistical significance was determined at the level of .05, and all P values were based on 2-sided tests.

The primary cohort included 666 individuals with baseline and postoperative serum creatinine measurements. Half of them were female, and half met criteria for AKI (Table 1). In all, 159 individuals (23.9%) had stage 1 AKI, 131 individuals (19.7%) had stage 2 AKI, and 51 individuals (7.7%) had stage 3 AKI. Those with AKI had more postoperative serum creatinine measurements than those without AKI (8 [IQR, 4-7] vs 5 [IQR, 5-10]). Primary cardiac diagnoses are listed in eTable 2 in the Supplement. The validation cohort comprised 333 individuals, with similar incidence of AKI (Table 1). In the validation cohort, 51 patients (15.3%) had stage 1 AKI, 91 patients (27.3%) had stage 2 AKI, and 20 (6.0%) had stage 3 AKI. Primary cardiac diagnoses for the validation cohort are listed in eTable 3 in the Supplement.

Quiz Ref IDIn the primary cohort, 72 (10.8%) had no acetaminophen exposure in the first 48 hours after surgery. The incidence of AKI was higher among those with no acetaminophen exposure than among those with acetaminophen exposure (Figure 1A). Comparing those with and without AKI, those without AKI had a significantly higher weight-adjusted acetaminophen dose in the first 48 hours after surgery compared with those with AKI (Table 2). Other variables significantly associated with AKI included younger age, longer CPB time, higher RDW, presence of postoperative hypotension, fewer high-risk nephrotoxins, and higher RACHS score. When stratified by AKI stage, those with no AKI had the highest median acetaminophen dose (78 [IQR, 43-104] mg/kg), followed by those with stage 1 AKI (57 [IQR, 17-100] mg/kg), stage 2 AKI (48 [IQR, 28-78] mg/kg), and stage 3 AKI (31 [IQR, 0-70] mg/kg (P < .001 for difference in distribution across groups). Those with no acetaminophen exposure had the highest rates of AKI, with decreasing incidence in the lowest, middle, and highest acetaminophen dose groups (Figure 1B). Quiz Ref IDIn the logistic regression analysis adjusting for age, CPB time, RDW, postoperative hypotension, nephrotoxin exposure, and RACHS score, acetaminophen exposure was protective against postoperative AKI, with an odds ratio (OR) of 0.86 (95% CI, 0.82-0.90) per each additional 10 mg/kg (Figure 2). The association between continuous acetaminophen dose and adjusted odds for AKI is shown in Figure 3. The patients who did not receive acetaminophen were younger and had longer CPB times, higher RDW, and more postoperative hypotension than those who received acetaminophen (eTable 4 in the Supplement). When subset analysis of the primary cohort was performed that excluded those with no acetaminophen exposure, acetaminophen remained protective against AKI (OR, 0.85 [95% CI, 0.80-0.90] per each additional 10 mg/kg; P < .001). We performed analyses to determine the sensitivity of our findings to unmeasured confounding. To make our findings statistically insignificant, an unmeasured confounder must have a minimum strength of association greater than 3.07 for both acetaminophen exposure and AKI (eTable 5 in the Supplement).

For the validation cohort, 18 (5.4%) had no acetaminophen exposure, and there was no significant difference in AKI incidence between those with vs without acetaminophen exposure (Figure 1A). Comparing those with and without AKI, the median early postoperative acetaminophen dose was again higher among those without AKI than among those with AKI (Table 2). Younger age, longer CPB time, higher RDW, fewer high-risk nephrotoxins, and higher RACHS score were also associated with postoperative AKI. When stratified by AKI stage, those with no AKI again had the highest median weight-adjusted acetaminophen dose (70 [IQR, 45-94] mg/kg). Those with stage 1 AKI had the lowest median dose (50 [IQR, 26-76] mg/kg), and acetaminophen doses were similar in those with stage 2 and stage 3 AKI as in those with no AKI (67 [IQR, 43-88] and 66 [IQR, 45-87] mg/kg, respectively; P = .02). Those with no acetaminophen exposure again had the highest rates of AKI, with decreasing incidence in the lowest, middle, and highest acetaminophen dose groups, without statistical significance (Figure 1B). In multivariable analysis adjusting for age, CPB time, RDW, postoperative hypotension, nephrotoxin exposure, and RACHS score, postoperative acetaminophen had a protective association against AKI, with an adjusted OR of 0.91 (95% CI, 0.84-0.99) per each additional 10 mg/kg (Figure 2). The association between continuous acetaminophen dose and adjusted odds for AKI in the validation cohort is shown in Figure 3. When those with no acetaminophen exposure were excluded, acetaminophen exposure remained protective (OR, 0.90 [95% CI, 0.82-0.99] per each additional 10 mg/kg; P = .03).

The pooled subgroup analysis included a total of 310 individuals with a RACHS score of 3 or higher from both clinical sites (237 from the primary cohort and 73 from the validation cohort). After adjusting for clinical site and the covariates included in the adjusted analyses of the full cohorts, acetaminophen dose (in milligrams per kilogram) in this subgroup again showed a protective association against AKI (OR, 0.86 [95% CI, 0.81-0.93] per each additional 10 mg/kg; eFigure in the Supplement).

We present the finding of a protective association of acetaminophen against AKI in children who undergo cardiac surgery. The weight-adjusted total dose of acetaminophen given in the first 48 hours after surgery was inversely correlated with the incidence of AKI in the primary cohort using the dichotomous indicator of acetaminophen exposure and acetaminophen dose as a continuous variable and after adjusting for potential confounders, including age, CPB time, RDW, postoperative hypotension, nephrotoxin exposure, and RACHS score. Given this novel finding, we sought to confirm these observations in an independent data set, in which after adjusting for confounding variables that were not balanced across acetaminophen exposure groups the protective association of acetaminophen was again observed. Subgroup analyses of only those patients who received acetaminophen and those with a RACHS score of 3 or higher, representing more clinically homogeneous patients, yielded consistent results.

The hypothesis that acetaminophen may be protective against AKI in the setting of cardiac surgery stemmed from several observations. In a case-control study³⁰ of adults undergoing cardiac surgery, the 10 individuals with postoperative AKI had higher intraoperative peak free hemoglobin concentrations than risk-matched controls (289.0 vs 104.4 mg/dL, P = .01 [to convert to grams per liter, multiply by 0.01]). This association was replicated in a cohort study²⁹ of 60 children undergoing surgery requiring CPB, where those with AKI had a greater increase in plasma hemoglobin concentration (53.9 vs 38.3 mg/dL, P = .03 [to convert to grams per liter, multiply by 0.01]), and an increase in plasma hemoglobin was an independent risk factor for AKI (OR, 1.02 [95% CI, 1.00-1.03]) per each additional 1 mg/dL). The mechanism whereby free hemoglobin leads to nephrotoxicity is postulated to be through lipid peroxidation, catalyzed by oxidized cell-free hemoglobin, a process that can be inhibited by acetaminophen.^32,40

Cell-free hemoglobin was also associated with death in a retrospective observational study³¹ of critically ill adults with sepsis, and acetaminophen exposure had a protective association with in-hospital mortality. In the subset of patients with no detectable plasma hemoglobin, there was no protective association of acetaminophen. This potential protective association of acetaminophen was evaluated prospectively in the Acetaminophen for the Reduction of Oxidative Injury in Severe Sepsis (ACROSS) trial.³³ That single-center, randomized, placebo-controlled trial of acetaminophen (1 g enterally every 6 hours for 3 days) found no difference in the primary end point, F2-isoprostane levels at day 3 (a measure of oxidative injury), or in hospital mortality, but it demonstrated that acetaminophen treatment affected secondary outcomes. Specifically, acetaminophen lowered F2-isoprostane levels at day 2 (24 vs 36 pg/mL, P = .047) and serum creatinine levels at day 3 (1.0 vs 1.3 mg/dL, P = .039). In another randomized trial⁴¹ that included 700 adult ICU patients with fever, there was no difference in the primary outcome (ICU-free days) and no difference in the peak creatinine level in the first week after randomization. This negative result may indicate that acetaminophen has protective associations only in specific patient populations.

There have been 2 pilot studies of the effect of acetaminophen on lipid peroxidation in patients undergoing CPB. In a prospective double-blind clinical trial,⁴² before initiation of CPB, 60 adult patients were treated with 4 doses of intravenous acetaminophen over 24 hours (1 g for those >50 kg and 15 mg/kg for those <50 kg) or placebo. The primary outcome was oxidative stress, measured by plasma and urine F2-isoprostane and isofuran values. The acetaminophen treatment group had smaller increases in plasma isofuran concentrations compared with the placebo group, although there were no significant differences between groups in plasma F2-isoprostane levels, urinary markers of lipid peroxidation, or AKI. The second pilot study⁴³ randomized 30 children to be treated with 4 doses of intravenous acetaminophen (15 mg/kg for age ≥2 years and 12.5 mg/kg for age <2 years) or placebo, beginning before the start of CPB. As in the adult trial, acetaminophen attenuated increases in plasma isofuran levels, but there were no differences between the intervention and control groups in F2-isoprostane levels, AKI prevalence, AKI severity, or the AKI biomarker neutrophil gelatinase–associated lipocalin (NGAL). There were several limitations to these pilot studies, including limited power because of the small number of participants and administration of 1 or more doses of acetaminophen before and during CPB, when circulating concentration of the drug may not have reached therapeutic levels because of hemodilution. These studies were not definitive in determining the potential role of acetaminophen to protect against AKI in patients with elevated cell-free hemoglobin.

Our results indicate a modest influence of acetaminophen on AKI. Specifically, we saw an approximately 10% reduction in AKI risk for each additional 10 mg/kg administered in the first 48 hours after surgery (one typical dose of acetaminophen). In the subgroup analysis of those with a RACHS score of 3 or higher, we observed a 14% reduction per 10 mg/kg of additional acetaminophen. The ideal dose for renal protection, while avoiding dose-related hepatotoxicity or other adverse events, has not been established. There may be several sources of interindividual variability contributing to differences in the exposure and response to acetaminophen in both adults and children.^44,45 Further delineation of individualized therapeutic concentrations through a precision medicine approach may be required to maximize benefit and minimize risk of acetaminophen use.

Our study has several limitations. This was a retrospective observational study. Therefore, causality between acetaminophen exposure and reduced incidence of AKI cannot be established. The apparent protective association of acetaminophen may be owing to residual confounding. Patients with acetaminophen exposure had fewer risk factors for AKI. Furthermore, these patients may have the most frequent serum creatinine testing, increasing AKI detection. We attempted to control for this potential confounding through multivariable analysis, adjusting for factors like age, CPB time, and RACHS score. Despite inclusion of these variables, acetaminophen remained protective. Our analyses of the sensitivity of the data to unmeasured confounding indicate that our data are robust. Those with AKI had fewer exposures to highly nephrotoxic drugs, which is an observation we have seen in multiple cohorts^3,37 and may be owing to avoidance of these drugs among high-risk patients. Another limitation of our study stems from the definition of AKI. We used the consensus KDIGO definition based on serum creatinine values but did not define AKI based on urine output (which may be inaccurately represented in these retrospective data sets) or biomarkers, such as NGAL or cystatin C.⁴⁶ Our initial finding from one tertiary referral center was validated in independent data from a second tertiary referral center. However, these results may not be generalizable to institutions with a different patient mix, operative procedures, and postoperative care.

In 2 independent cohorts of children undergoing cardiac surgery, we observed a dose-dependent inverse association between postoperative receipt of acetaminophen and development of AKI. Taken together with other retrospective and prospective studies showing protective associations of acetaminophen against AKI in several conditions associated with release of cell-free hemoglobin, these findings provide a compelling rationale to justify prospective evaluation of acetaminophen for prevention of AKI in children undergoing CPB.

Accepted for Publication: February 24, 2018.

Corresponding Author: Sara L. Van Driest, MD, PhD, Department of Pediatrics, Vanderbilt University School of Medicine, 2200 Children’s Way, 8232 Doctor’s Office Tower, Nashville, TN 37232 (sara.van.driest@vanderbilt.edu).

Published Online: May 14, 2018. doi:10.1001/jamapediatrics.2018.0614

Author Contributions: Dr Van Driest had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Van Driest, Jooste, Hill, Roden, Ware.

Acquisition, analysis, or interpretation of data: Van Driest, Jooste, Shi, Choi, Darghosian, Hill, Smith, Kannankeril, Ware.

Drafting of the manuscript: Van Driest, Shi, Choi.

Critical revision of the manuscript for important intellectual content: Van Driest, Jooste, Darghosian, Hill, Smith, Kannankeril, Roden, Ware.

Statistical analysis: Shi, Choi, Darghosian, Hill.

Obtained funding: Van Driest, Ware.

Administrative, technical, or material support: Jooste, Smith.

Study supervision: Hill, Roden, Ware.

Conflict of Interest Disclosures: Dr Van Driest reported receiving an honorarium as an invited speaker to Merck. No other disclosures were reported.

Funding/Support: The project described was supported by the Clinical and Translational Science Award (CTSA) UL1 TR000445 from the National Institutes of Health (NIH) National Center for Advancing Translational Sciences (NCATS) to Vanderbilt University Medical Center, R01 HD084461 from the NIH Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) to Dr Kannankeril, and the Doris Duke Clinical Scientist Development Award to Dr Van Driest. Dr Van Driest has been supported by NIH/NCATS KL2 TR000446. Dr Ware is supported by NIH HL103836 and NIH HL135849.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We thank the Postoperative Arrhythmia in Congenital Heart Surgery study nurses Kim Crum, RN, BS, Carla Hissam, RN, CPN, and Darlene Fountain, RN, BSN (Vanderbilt University Medical Center Department of Pediatrics) and all study participants and families. No compensation was received.