RCT indicates randomized clinical trial.
aThese are approximate data.
Absolute differences are the estimated differences in proportions between the atorvastatin and placebo groups derived from model transformations. Quasi-Poisson log-linear regression was used to calculate the estimates and 95% CIs. The estimates should be interpreted as the relative risk of treatment for the primary end point of AKI. The Pearson χ2 test was used to calculate the P values.
eTable 1. Propensity score-adjusted analyses for primary and secondary outcomes among all patients that completed the study (intention to treat cohort)
eTable 2. Unadjusted and propensity score-adjusted analyses of primary and secondary outcomes among all patients that completed the study per protocol
Data and Safety Monitoring Board Charter
Timeline of Events Related to Second Interim Analysis
Billings FT, Hendricks PA, Schildcrout JS, Shi Y, Petracek MR, Byrne JG, Brown NJ. High-Dose Perioperative Atorvastatin and Acute Kidney Injury Following Cardiac SurgeryA Randomized Clinical Trial. JAMA. 2016;315(9):877-888. doi:10.1001/jama.2016.0548
Statins affect several mechanisms underlying acute kidney injury (AKI).
To test the hypothesis that short-term high-dose perioperative atorvastatin would reduce AKI following cardiac surgery.
Design, Setting, and Participants
Double-blinded, placebo-controlled, randomized clinical trial of adult cardiac surgery patients conducted from November 2009 to October 2014 at Vanderbilt University Medical Center.
Patients naive to statin treatment (n = 199) were randomly assigned 80 mg of atorvastatin the day before surgery, 40 mg of atorvastatin the morning of surgery, and 40 mg of atorvastatin daily following surgery (n = 102) or matching placebo (n = 97). Patients already taking a statin prior to study enrollment (n = 416) continued taking the preenrollment statin until the day of surgery, were randomly assigned 80 mg of atorvastatin the morning of surgery and 40 mg of atorvastatin the morning after (n = 206) or matching placebo (n = 210), and resumed taking the previously prescribed statin on postoperative day 2.
Main Outcomes and Measures
Acute kidney injury defined as an increase of 0.3 mg/dL in serum creatinine concentration within 48 hours of surgery (Acute Kidney Injury Network criteria).
The data and safety monitoring board recommended stopping the group naive to statin treatment due to increased AKI among these participants with chronic kidney disease (estimated glomerular filtration rate <60 mL/min/1.73 m2) receiving atorvastatin. The board later recommended stopping for futility after 615 participants (median age, 67 years; 188 [30.6%] were women; 202 [32.8%] had diabetes) completed the study. Among all participants (n = 615), AKI occurred in 64 of 308 (20.8%) in the atorvastatin group vs 60 of 307 (19.5%) in the placebo group (relative risk [RR], 1.06 [95% CI, 0.78 to 1.46]; P = .75). Among patients naive to statin treatment (n = 199), AKI occurred in 22 of 102 (21.6%) in the atorvastatin group vs 13 of 97 (13.4%) in the placebo group (RR, 1.61 [0.86 to 3.01]; P = .15) and serum creatinine concentration increased by a median of 0.11 mg/dL (10th-90th percentile, −0.11 to 0.56 mg/dL) in the atorvastatin group vs by a median of 0.05 mg/dL (10th-90th percentile, −0.12 to 0.33 mg/dL) in the placebo group (mean difference, 0.08 mg/dL [95% CI, 0.01 to 0.15 mg/dL]; P = .007). Among patients already taking a statin (n = 416), AKI occurred in 42 of 206 (20.4%) in the atorvastatin group vs 47 of 210 (22.4%) in the placebo group (RR, 0.91 [0.63 to 1.32]; P = .63).
Conclusions and Relevance
Among patients undergoing cardiac surgery, high-dose perioperative atorvastatin treatment compared with placebo did not reduce the risk of AKI overall, among patients naive to treatment with statins, or in patients already taking a statin. These results do not support the initiation of statin therapy to prevent AKI following cardiac surgery.
clinicaltrials.gov Identifier: NCT00791648
Acute kidney injury (AKI) complicates recovery from cardiac surgery in up to 30% of patients.1 Surgical, anesthetic, and critical care advancements have reduced perioperative mortality but have not reduced the incidence of AKI, and some studies have reported increasing rates of postoperative dialysis.2- 4 Recent large cohort studies confirm that even mild postoperative AKI is independently associated with a 5-fold increase in death while in the hospital.5 A diagnosis of AKI following cardiac surgery is accompanied by higher rates of postoperative arrhythmias, respiratory failure, systemic infection, and myocardial infarction.1,5- 7 Treatments are needed to reduce this debilitating postoperative outcome.
Statins affect several mechanisms underlying postoperative AKI, and statin treatment may affect AKI in patients following cardiac surgery. In a recent observational study of 17 000 patients who underwent cardiac surgery, investigators reported a 22% relative risk (RR) reduction in AKI associated with preoperative statin treatment in adjusted analyses,8 whereas investigators of a similar study of 11 000 patients observed no effect.9 These observational studies that compared patients who are taking statins with those who are not taking statins did not adequately test the hypothesis that preoperative, intraoperative, and early postoperative (perioperative) statin treatment reduces AKI. It is both unclear if de novo initiation of perioperative statin treatment in patients naive to statins affects AKI, or if statin continuation or perioperative withdrawal in patients already taking statin treatment affects AKI.
We performed the Statin AKI Cardiac Surgery [randomized clinical trial] RCT to test the hypothesis that short-term high-dose perioperative atorvastatin reduces AKI following cardiac surgery.
The Statin AKI Cardiac Surgery RCT was an investigator-initiated, double-blinded, placebo-controlled study conducted to test the hypothesis that short-term, high-dose perioperative atorvastatin treatment reduces AKI following cardiac surgery (the Trial Protocol appears in Supplement 1). Adult patients undergoing elective coronary artery bypass grafting, valvular heart surgery, or ascending aortic surgery at Vanderbilt University Medical Center were eligible for study participation (Figure 1).
Patients were excluded if they had (1) prior statin intolerance; (2) acute coronary syndrome (defined as ST or non-ST elevation myocardial infarction with elevated serum troponin concentrations); (3) liver dysfunction (defined as serum transaminase concentrations >3 times the upper limit of normal [120 U/L], a bilirubin concentration >3 mg/dL, or a diagnosis of cirrhosis); (4) current use of potent CYP3A4 inhibitors, including azole antifungals, protease inhibitors, and macrolide antibiotics; (5) current use of cyclosporine; (6) current use of renal replacement therapy; (7) a history of kidney transplant; (8) required emergency or urgent surgery; or (9) were pregnant.
The study was approved by the Vanderbilt University Institutional Review Board for Research on Human Subjects and conducted according to the Declaration of Helsinki. All patients provided written informed consent. Anesthetic, surgical, and postoperative management were conducted according to institutional protocols (eMethods in Supplement 2).
Patients naive to statin treatment were randomized to receive 80 mg of atorvastatin the day prior to surgery, 40 mg of atorvastatin the morning of surgery (at least 3 hours prior to surgery), and 40 mg of atorvastatin daily at 10 am following surgery for the duration of hospitalization or to a matching placebo regimen. Patients already taking a statin were randomized to receive the study drug only on days they otherwise would have not been prescribed a statin under standard of care because it was deemed unethical to randomize these patients to placebo throughout the perioperative period based on prior observational studies and clinical trials in similar patient populations demonstrating that statin withdrawal may increase AKI.10- 13 Therefore, patients already taking a statin continued taking their preenrollment statin until the day of surgery and resumed taking their previously prescribed statin on postoperative day 2. On the day of surgery, patients were randomly assigned to 80 mg of atorvastatin at least 3 hours prior to surgery and 40 mg of atorvastatin the day after surgery at 10 am or to a matching placebo regimen.
Statin administration subsequent to study intervention (starting at hospital discharge in patients naive to statin treatment and starting on postoperative day 2 in patients already taking a statin) was at the discretion of the treating physician. The intervention both in patients naive to statin treatment and in those already taking a statin was designed to be pragmatic, lending itself to maximum generalizability for subsequent patients undergoing cardiac surgery. We chose atorvastatin because short-term periprocedural atorvastatin treatment has been reported in studies to reduce arrhythmia following cardiac surgery and myocardial injury following percutaneous coronary intervention.14,15 The loading dose of 80 mg and maintenance dose of 40 mg of atorvastatin were chosen to maximize potential benefit, to limit toxicity, and to fall within recommended daily dosing.
Participants were randomized in permuted block sizes of 2 or 4 to receive oral atorvastatin or matching placebo within strata defined by prior statin use, by the presence of chronic kidney disease (CKD; defined as a glomerular filtration rate of <60 mL/min/1.73 m2, estimated using the modification of diet in renal disease formula), and by a history of diabetes. The Vanderbilt University Medical Center Investigation Drug Service compounded and maintained atorvastatin and placebo capsules identical in size, color, texture, weight, and taste. Using a randomization schedule provided by the study statistician, the Investigation Drug Service assigned the intervention to recruited patients, dispensed atorvastatin or matching placebo, and logged treatment assignment in private records. Investigators, clinicians, patients, and research personnel were blinded to treatment for the duration of the study.
Outpatient adherence to the intervention was monitored by directly questioning patients and family members on the morning of surgery, and adherence while patients were in the hospital was monitored by reviewing the electronic medication administration record. For intubated patients unable to swallow pills, the study drug was mixed with water and administered via oral gastric tube.
The primary end point was the diagnosis of AKI according to criteria from the Acute Kidney Injury Network (AKIN),16 specifically an increase of 0.3 mg/dL in serum creatinine concentration (to convert to μmol/L, multiply by 88.4) or the initiation of renal replacement therapy within 48 hours of surgery. Stage 1 AKI was defined as an increase of 0.3 mg/dL in serum creatinine or 50%; stage 2, a 100% increase; and stage 3, a 200% increase or initiation of renal replacement therapy. Baseline creatinine concentration was defined as the most recent creatinine measurement prior to surgery, and serum creatinine concentration was measured in all inpatients on the morning of surgery and no more than 7 days prior to surgery in outpatients. Postoperatively, serum creatinine concentration was measured at 2 am daily throughout the duration of hospitalization. All creatinine measurements were performed in the Vanderbilt University Medical Center clinical laboratory.
Secondary end points included the maximum increase in creatinine concentration from baseline to 48 hours following surgery (postoperative day 2), the incidence and duration of delirium while in the intensive care unit (ICU) (incorrectly noted as a co-primary end point during the editing of the study registration), the degree of myocardial injury, and the incidence of postoperative atrial fibrillation, pneumonia, or stroke. Study personnel assessed for delirium once in the morning and once in the afternoon while patients were in the ICU using the Richmond agitation and sedation scale and the confusion assessment method for the ICU.17
We quantified myocardial injury by measuring serum creatine kinase MB (CK-MB) concentrations on postoperative day 1 in the clinical laboratory. Atrial fibrillation was diagnosed by the patient’s physician and confirmed by research personnel using electrocardiograms and rhythm strips. We defined pneumonia as a positive sputum culture or postoperative pulmonary infiltrate with systemic signs of infection (temperature >38.6°C or white blood cell count >12.0 × 109/L) and the use of parenteral antibiotics or documentation of the diagnosis by the patient’s physician. Stroke was defined as any new deficit during the neurological examination and confirmed with radiological evidence. Additional secondary end points were duration of postoperative mechanical ventilation, duration of ICU stay, and incidence of death during hospitalization.
Safety outcomes included acute liver toxicity assessed by measuring serum aspartate aminotransferase concentration on postoperative day 1, acute muscle toxicity assessed by questioning participants daily for myalgias and measuring serum creatine kinase concentration on postoperative day 1, and adverse events. Adverse events were defined as any untoward medical occurrence in a participant and included postoperative fever, tachycardia, hypotension, vasoplegia, decreased cardiac output, anemia, leukocytosis, and lactic acidosis (specific diagnostic criteria appear in the eMethods in Supplement 2).
The data and safety monitoring board (DSMB) reviewed patient recruitment practices, safety reporting, and data quality after 30 patients completed the study; performed an interim analysis after 277 patients (the first third of planned enrollment) had completed the study to assess safety of the intervention; and performed a second interim analysis after 546 patients (the second third of planned enrollment) had completed the study to assess the safety, efficacy, and futility of the intervention. The DSMB made recommendations based on qualitative assessments of the safety, efficacy, and futility of the intervention (details appear in Supplement 2).
In October 2012, the DSMB recommended study continuation after the first planned interim analysis. In August 2014, following the second interim analysis, the DSMB recommended discontinuation of the recruitment of patients naive to statin treatment due to evidence of increased risk of AKI in patients randomized to atorvastatin, particularly in patients with preexisting CKD. Following a conditional power analysis for patients already taking a statin that assumed the true treatment effect would be consistent with what had been observed in those who had completed the study, the DSMB recommended in October 2014 discontinuation of the recruitment of patients already taking a statin due to futility on the primary end point.
Demographic, medical history, hemodynamic, anesthetic, surgical, cardiopulmonary bypass, transfusion, medication exposure, ICU, and laboratory data were collected on all patients. Race and ethnicity data were determined by patient self-reporting and were collected because African American ancestry is associated with AKI. In addition to safety and clinical care laboratory measurements, a lipid panel was assessed prior to surgery. To assess baseline mental competence, all participants were administered a Mini-Mental State Examination and a Trails B test prior to surgery by research personnel.
The analysis was completed according to the statistical analysis plan (details appear in Supplement 2). We planned to study 820 patients to detect a 30% RR reduction of AKI between treatment groups with an assumed AKI incidence of 27.6% in the placebo group,12 a type I error probability of 0.05, and 80% power. We summarized demographics, patient characteristics, and safety and efficacy end points with the 50th (10th-90th) percentiles for continuous variables and with percentages for categorical variables. We used the Pearson χ2 test and the Wilcoxon rank sum test to compare the randomized treatment groups regarding categorical and continuous outcomes, respectively.
To estimate treatment effects, we used Quasi-Poisson log-linear regression for binary outcomes with event rates greater than 10% due to ease of interpretation of RRs and because we designed the study to detect a RR reduction, logistic regression for rare binary outcomes due to model stability with event rates less than 10% and close approximation of the odds ratio with the RR for rare outcomes,18,19 proportional odds regression for ordered categorical and highly skewed outcomes (eg, duration of delirium and ICU stay), and linear regression for continuous outcomes.
The CK-MB, time to extubation, creatine kinase, and aspartate aminotransferase outcomes were log-transformed to improve model fit. In the Tables, we include absolute differences between treatment groups derived from model transformations. For dichotomous outcomes, the absolute differences are the estimated differences in proportions between the statin and placebo groups. For continuous outcomes, the absolute differences are the estimated differences in the geometric means between the statin and placebo groups. For the duration outcomes of delirium in the ICU and ICU stay, the absolute differences are the differences in the proportions of those in the statin group greater than the median value in the placebo group minus 0.5 (ie, the proportion of those in the placebo greater than the median value in the placebo group).
Several subgroup analyses were planned prior to initiating the study because treatment effects could vary based on preexisting statin use and baseline kidney disease. These groups included patients (1) who were naive to statin treatment, (2) who were already taking a statin, (3) who had CKD, (4) who were naive to statin treatment and had CKD, and (5) who were already taking a statin and had CKD. The primary unadjusted analyses are presented herein. The propensity score–adjusted analyses and treatment-received analyses appear in the eResults in Supplement 2.
The analyses were performed using R version 3.1.2 (R Foundation for Statistical Computing), and a 2-sided significance level of .05 was required to achieve statistical significance.
Between November 2009 and October 2014, 4466 adult patients presented for coronary bypass, valve, or ascending aorta surgery (Figure 1). Ten percent of these patients had acute coronary syndrome, 3.7% end stage renal disease, 3.1% liver dysfunction, 2.6% were allergic to statins, and 5.5% required emergent or urgent surgery. Of those approached for study participation, approximately 30% did not provide consent. We recruited 653 patients. Twenty-two were subsequently excluded from study and 14 withdrew consent prior to randomization.
One patient withdrew consent following randomization prior to receiving treatment, and 1 patient was withdrawn from the study on the day of surgery by his intensivist. Therefore, 615 patients (199 naive to statin treatment and 416 already taking a statin) were included in the primary analysis; 308 were randomized to atorvastatin and 307 to placebo. Patient characteristics were well balanced between the treatment groups (Table 1). The median patient age was 67 years, 30.6% were women, and 32.8% had diabetes. Coronary artery bypass grafting was performed in half of the patients, and two-thirds of patients received valve surgery. Compared with patients in the placebo group, mean serum low-density lipoprotein cholesterol concentrations at induction of anesthesia (following treatment initiation but before surgery) were lower by 8.1 mg/dL (95% CI, 2.2-14.1 mg/dL) in the atorvastatin group (P = .01).
Treatment adherence (ie, received all protocol-directed doses) was achieved in 2356 of 2400 (98.2%) total doses of study drug. Of 199 patients naive to statin treatment, 180 (90.5%) received all protocol-directed doses. Of 416 patients already taking a statin, 391 (94.0%) received all protocol-directed doses (Supplement 2).
Among all patients (n = 615), the primary end point of AKI (by AKIN criteria) occurred in 64 of 308 (20.8%) in the atorvastatin group compared with 60 of 307 (19.5%) in the placebo group (P = .75 by Pearson χ2). Perioperative atorvastatin treatment did not affect risk for AKI in the total cohort (RR, 1.06 [95% CI, 0.78 to 1.46]; Figure 2) or perioperative serum creatinine concentrations. Median serum creatinine concentrations increased by 0.07 mg/dL (10th-90th percentile, −0.13 to 0.51 mg/dL) within the first 48 postoperative hours among participants in the atorvastatin group compared with 0.07 mg/dL (10th-90th percentile, −0.12 to 0.52 mg/dL) in the placebo group (mean difference, −0.01 mg/dL [95% CI, −0.06 to 0.04 mg/dL]; P = .89).
Among patients naive to statin treatment (n = 199), AKI occurred in 22 of 102 (21.6%) in the atorvastatin group compared with 13 of 97 (13.4%) in the placebo group (RR, 1.61 [95% CI, 0.86 to 3.01]; P = .15) and median serum creatinine concentrations within the first 48 postoperative hours increased by 0.11 mg/dL (10th-90th percentile, −0.11 to 0.56 mg/dL) in the atorvastatin group compared with 0.05 mg/dL (10th-90th percentile, −0.12 to 0.33 mg/dL) in the placebo group (mean difference, 0.08 mg/dL [95% CI, 0.01 to 0.15 mg/dL]; P = .007).
Among patients already taking a statin (n = 416), high-dose perioperative atorvastatin or short-term statin withdrawal did not affect AKI risk. Acute kidney injury occurred in 42 of 206 patients already taking a statin (20.4%) in the atorvastatin group compared with 47 of 210 patients (22.4%) in the placebo group (RR, 0.91 [95% CI, 0.63-1.32]; P = .63); postoperative changes in serum creatinine concentration were not different between groups.
In the baseline CKD subgroup (179 of 615 patients), the incidence of AKI was also similar between treatment groups. Thirty of the 84 participants with CKD in the atorvastatin group (35.7%) developed AKI and 31 of the 95 participants with CKD in the placebo group (32.6%) developed AKI (RR, 1.09 [95% CI, 0.73 to 1.65]; P = .76).
In the subgroup of patients with CKD and naive to statin treatment (n = 36), AKI occurred in 9 of 17 (52.9%) in the atorvastatin group compared with 3 of 19 (15.8%) in the placebo group (RR, 3.35 [95% CI, 1.12 to 10.05]; P = .03). In this subgroup, postoperative serum creatinine concentrations increased by a median of 0.26 mg/dL (10th-90th percentile, −0.22 to 0.94 mg/dL) within the first 48 postoperative hours in patients in the atorvastatin group compared with a decrease of 0.06 mg/dL (10th-90th percentile, decrease of 0.16 mg/dL to increase of 0.41 mg/dL) in the placebo group (mean difference, 0.28 mg/dL [95% CI, 0.02 to 0.54 mg/dL]; P = .04).
Among the subgroup of patients with CKD who were already taking a statin (n = 143), AKI occurred in 21 of 67 (31.3%) in the atorvastatin group compared with 28 of 76 (36.8%) in the placebo group (RR, 0.85 [95% CI, 0.54 to 1.35]; P = .59). In this subgroup, postoperative changes in serum creatinine concentration were not different between treatment groups.
In the total cohort, AKI required dialysis in 5 of the 308 patients (1.6%) in the atorvastatin group compared with 3 of the 307 (1.0%) patients in the placebo group (P = .71). Of the 124 cases of AKI in the study, 106 (85.4%) were stage 1; 6 (4.8%), stage 2; and 12 (9.7%), stage 3. There was no observed treatment effect on moderate or severe (stages 2 or 3) AKI, and treatment effects on moderate or severe AKI were similar to treatment effects on AKI stage 1, 2, or 3 in the entire cohort and in the subgroups.
Perioperative high-dose atorvastatin treatment did not affect the secondary end points of delirium, duration of delirium in the ICU, atrial fibrillation, myocardial injury, or stroke (Table 2 and Table 3). The incidence of postoperative pneumonia was numerically lower but not significantly different in patients randomized to atorvastatin. This result should be interpreted with caution based on multiple comparisons for secondary end points and risk of type I error.
All but 1 of the 17 strokes was ischemic in nature rather than hemorrhagic. Length of stay in the ICU and all-cause in-hospital mortality were not significantly different between treatment groups.
Propensity-adjusted analyses yielded results that were largely consistent with the intention-to-treat unadjusted analyses, although the evidence of harm from atorvastatin treatment in the patients naive to statin treatment with preexisting CKD was attenuated (eTable 1 in Supplement 2).
There was no evidence to suggest that perioperative high-dose atorvastatin treatment increased skeletal muscle or hepatic markers of statin toxicity. Six patients (1.9%) in the atorvastatin group reported proximal muscle myalgias on postoperative day 1, 2, or 3 compared with 5 patients (1.6%) in the placebo group (P = .77, Table 4). Postoperative day 1 concentrations of creatine kinase and aspartate aminotransferase were also not increased by high-dose short-term perioperative atorvastatin treatment, nor were adverse events (details of adverse events appear in Supplement 2).
This double-blinded, placebo-controlled RCT found no evidence that high-dose perioperative atorvastatin reduces the incidence or severity of AKI following cardiac surgery. Among patients naive to statin treatment, high-dose perioperative atorvastatin increased serum concentrations of creatinine, and there was some evidence that statin treatment may increase AKI among patients naive to statin treatment with preexisting CKD. Among patients already taking a statin, there was no evidence that perioperative statin continuation or withdrawal affected postoperative AKI. The results from the prespecified secondary end points of delirium, myocardial injury, atrial fibrillation, and stroke were similar between randomized treatment groups.
Based on preclinical studies of statin mechanisms and observational studies of cardiac surgery cohorts, statin treatment is an attractive therapy to reduce AKI following cardiac surgery. Short-term statin treatment has been shown to limit vascular superoxide generation,20,21 reduce endothelial dysfunction by restoring endothelial-derived nitric oxide synthase activity during hypoxia,22 and attenuate lymphocyte activation.23 These effects decrease inflammation, endothelial dysfunction, and oxidant stress, mechanisms implicated in AKI development.24 In a study of patients undergoing cardiac surgery, 3 weeks of treatment with a 20-mg/d dose of atorvastatin reduced peak plasma IL-6 and IL-8 concentrations and neutrophil adhesion.25
Prior studies of statins to reduce AKI following cardiac surgery are limited to a pilot study and numerous observational reports. Prowle et al26 randomized 100 patients who underwent cardiac surgery to 4 days of atorvastatin (40 mg) or placebo starting the day prior to surgery and compared increases in serum creatinine concentrations from baseline to postoperative day 5. This study26 found no difference in creatinine concentration and a marginal but insignificant reduction in AKI rates defined by risk, injury, failure, loss of kidney function, and end-stage kidney disease (RIFLE criteria27) in patients randomized to atorvastatin.
The majority of observational studies compare patients who are taking statins prior to surgery with patients not taking statins prior to surgery. The association between preoperative statin use and postoperative AKI is inconsistent, possibly due to selection bias for statin use, variable effects of treatment, and disparate patient populations. Some of these studies reported a lower incidence of AKI among patients taking statins preoperatively,8,28- 30 whereas others reported no differences between patients taking statins and those not taking statins.9,31- 34 Few studies have reported statin exposure during the entire perioperative period (ie, the day prior to surgery, day of surgery, and the first 2 days after), a period when acute inflammation, endothelial dysfunction, and oxidant injury surge, and therefore a potential therapeutic window for preventing AKI.
Contrary to our hypothesis, de novo initiation of daily perioperative atorvastatin treatment did not reduce the incidence of AKI or reduce the increase in serum creatinine concentration associated with cardiac surgery. On the contrary, initiation of high-dose atorvastatin in patients naive to statin treatment increased serum creatinine concentration more from baseline to postoperative day 2 than initiation of placebo. Among patients with CKD and naive to statin treatment, there was evidence that statin treatment may increase AKI. Limitations to this finding are (1) the small number of patients in the subgroup with CKD and naive to statin treatment, (2) the wide confidence interval, and (3) the loss of statistical significance in the propensity score–adjusted model.
The mechanism of any deleterious effects is unclear. The kidneys eliminate less than 1% of atorvastatin,35 so any renal toxic effect is unlikely to be related to increased circulating atorvastatin in patients with CKD. Because patients with CKD are at high risk for AKI,2 any injurious mechanism of de novo atorvastatin initiation may be magnified in these patients. A harmful effect of de novo statin treatment on renal function has been previously reported in a different critically ill patient population. The Acute Respiratory Distress Syndrome Network reported that patients randomized to daily rosuvastatin after being admitted to the ICU with acute respiratory distress syndrome experienced increased renal failure compared with patients randomized to placebo.36 Eighty-five percent of these patients were naive to statin treatment.
Most patients presenting for cardiac surgery, however, are already taking statins, and in the current study there was little evidence that continuation or withdrawal from statin treatment on the day of surgery and postoperative day 1 affects AKI. The 2-day duration of treatment for patients already taking a statin could be a limitation to this finding, but this treatment duration is consistent with the opportunity for intervention in current clinical practice. We deemed a longer period of statin withdrawal unethical based on reports that perioperative statin withdrawal for periods as short as 1 to 2 days correlates with increased risk of AKI following cardiac or vascular surgery, the observation that 40% of patients already taking a statin resume statin use within 1 day of cardiac surgery, and a clinical trial in patients already taking a statin and undergoing percutaneous coronary intervention that demonstrated placebo administration on the day of intervention increases myocardial injury compared with atorvastatin (statin continuation) treatment.10- 13 Therefore, to limit risk in patients already taking a statin randomized to placebo, we did not reduce statin exposure compared with usual care. Also in keeping with the pragmatic nature of the trial, we did not initiate a preoperative treatment in patients naive to statin treatment multiple days or weeks before surgery because many patients receive surgery soon after consultation with their surgeon. We deemed it impractical to require patients to return 1 year after surgery; therefore, the effect of treatment on long-term renal function is unknown.
We acknowledge other limitations. This was not a multicenter study. We diagnosed AKI using AKIN criteria,16 but urine output AKIN criteria were not used, and AKIN criteria are insensitive to AKI that develops late during the postoperative period. The statistical significance for the deleterious effect of treatment on the primary end point in the subgroup with CKD and naive to statin treatment was lost in the secondary propensity score–adjusted analysis, possibly as a result of type I error or lack of power in this subgroup, and the cohort as a whole was at low risk for more severe clinical outcomes such as stage 2 or 3 AKI, which have been more closely associated with long-term outcomes.
Among patients undergoing cardiac surgery, high-dose perioperative atorvastatin treatment compared with placebo did not reduce the risk of AKI overall, among patients naive to treatment with statins, or in patients already taking a statin. These results do not support the initiation of statin therapy to prevent AKI following cardiac surgery.
Corresponding Author: Frederic T. Billings IV, MD, MSc, Departments of Anesthesiology and Medicine, Vanderbilt University School of Medicine, 1211 21st Ave S, Ste 526, Nashville, TN 37212 (firstname.lastname@example.org).
Published Online: February 23, 2016. doi:10.1001/jama.2016.0548.
Author Contributions: Drs Billings and Schildcrout had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Billings, Brown.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Billings, Schildcrout, Shi.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Schildcrout, Shi.
Obtained funding: Billings, Brown.
Administrative, technical, or material support: Hendricks, Petracek, Byrne, Brown.
Study supervision: Billings, Brown.
Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Brown reported receiving grants from Shire Pharmaceuticals and New Haven Pharmaceuticals; and personal fees from Novartis Pharmaceuticals and Alnylam Pharmaceuticals. No other disclosures were reported.
Funding/Support: This work was supported by grants K23GM102676, K12ES015855, R01GM112871, and UL1RR024975 from the National Institutes of Health and by the Vanderbilt University Medical Center Department of Anesthesiology.
Role of the Funder/Sponsor: The National Institutes of Health and the Vanderbilt University Medical Center Department of Anesthesiology 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 members of the data and safety monitoring board: Mias Pretorius, MD, MSc (chair), Julia B. Lewis, MD, Stephen K. Ball, MD, and Leena Choi, PhD. We also thank Anthony DeMatteo, BS, for technical assistance, Jennifer Morse, MS, and Misty Hale for data management assistance, and Will Hardeman, BA, Cleo Carter, BA, Kiersten Card, Jennifer Morse, MS, and Damon Michaels, BS (all with the Vanderbilt Department of Anesthesiology’s Perioperative Clinical Research Institute) for assisting in data collection. The persons listed were not compensated for their contributions.