Acute kidney injury during pediatric sickle cell vaso-occlusive pain crisis

Abstract

Background

Patients who develop sickle cell disease (SCD) nephropathy are at high risk for mortality. The pathophysiology of vaso-occlusive pain crisis may contribute to acute kidney injury (AKI). Non-steroidal anti-inflammatory drugs, known inducers of AKI, are used to treat pain crises. Multiple gaps exist in knowledge about the impact of AKI in SCD.

Methods

We conducted a two year retrospective review of AKI events in patients admitted for vaso-occlusive crisis. AKI was defined by an increase by ≥0.3 mg/dl or 50 % increase in serum creatinine from baseline. Laboratory values and ketorolac administration by days and dose (mg/kg) were identified from hospital records. Generalized mixed effects model for binary outcomes evaluated AKI based on laboratory variables and ketorolac administration. Generalized mixed Poisson effects model analyzed the association of AKI with hospital length of stay.

Results

Thirty three (17 %) of 197 admissions for vaso-occlusive pain crisis were associated with AKI. Fifty two percent of the cases presented to the Emergency Room (ER) with AKI. Every one unit drop in hemoglobin from baseline to admission increased the risk of AKI by 49 %. Among patients who received ketorolac for pain, both total days and doses of ketorolac were associated with AKI. Finally, patients with pain and AKI required longer hospitalizations than patients without AKI.

Conclusion

AKI during sickle cell pain crisis is common and may be an important modifiable risk factor for developing chronic kidney disease (CKD). Further studies are needed to determine the impact of nephrotoxic medications on progressive SCD nephropathy.

Introduction

Patients with Sickle Cell Disease (SCD) are at high risk for developing chronic kidney disease (CKD) during their lifetime [1–5]. It is important to prevent progression to end-stage renal disease (ESRD) as SCD patients have markedly higher one and five year mortality rates compared to ESRD patients without SCD [5–7]. It is possible that this progressive loss of kidney function is due to chronic complications from anemia, hemolysis, or inflammation; however the potential impact of repeated acute kidney injury (AKI) episodes during SCD crisis on progression to SCD nephropathy has not been explored in depth [8, 9]. Recent studies in animals, children and adults without SCD have established a clear link between AKI episodes and progression to CKD [10–14]. We know that 10–60 % of survivors from pediatric intensive care units with AKI develop CKD over the next few years [15–17]. Furthermore, one study identified that 60 % of non-critically ill pediatric patients who developed AKI due to nephrotoxic medications have CKD or are at risk for CKD six months after discharge [18]. Therefore, additional evaluations that better delineate the incidence and risk factors associated with AKI during admissions for SCD are warranted.

Despite an emphasis on conducting clinical trials of novel therapies for SCD pain crisis, the NIH expert panel on SCD management recommends treating pain events with opioids and, in the absence of contraindications, non-steroidal anti-inflammatory drugs (NSAIDs) [19, 20]. The recommendation for NSAIDs during pain crisis is based on low quality evidence and does not provide any specific guidance on contraindications to NSAIDs [19]. The frequent administration of NSAIDs during prolonged pain crises likely places patients at increased risk for nephrotoxicity and AKI due to the underlying relative dehydration from hyposthenuria that SCD patients develop over time [21–23].

The potential impact of repeated AKI events and on progression to CKD is poorly understood in the SCD literature; however, in other diseases, a clear link has been established between AKI events and the development of CKD [14]. To begin to address this critical knowledge gap in SCD, we conducted a retrospective review to better understand the incidence and risk factors associated with development of AKI. Then, we evaluated the impact of ketorolac on the development of AKI. Finally, we evaluated the impact of AKI on hospitalization length.

Methods

Human subjects and study variables

The institutional review board at the University of Alabama at Birmingham approved this retrospective review of AKI during vaso-occlusive pain crisis (VOC). From January 1, 2014 to December 31, 2015, we reviewed all ICD-10 defined VOC admissions from Children’s of Alabama Pediatric Emergency Department (ED). We identified 234 admissions for pain during this time. The primary outcome for this study was the development of AKI during VOC events. AKI was defined by the Kidney Disease Improving Global Outcomes (KDIGO) definition of an increase in serum creatinine (SCr) by ≥0.3 mg/dl or 50 % increase in serum creatinine from baseline [24]. Baseline serum creatinine was recorded from the most recent “well-child” outpatient visit obtained within 12 months from admission. Thirty seven of the 237 admissions were excluded as we could not determine a baseline serum creatinine level within 12 months. The Emergency Room (ER) and daily hospital records were reviewed to identify the maximum serum creatinine. This maximum serum creatinine was recorded and compared to baseline serum creatinine to define AKI.

At our center, patients begin intravenous fluids in the ER based on individual attending assessment of hydration. Patients admitted to the hospital receive intravenous fluids at maintenance rate until oral hydration status is deemed appropriate by the Pediatric Hematologist. Patients with pain crisis are started on IV opioids in the ER. Admitted patients remain on opioids based on the individual Hematology attending’s determination of the dose and frequency of opioid administration required for analgesia. Finally, the decision to initiate ketorolac is also determined by the ER attending or Pediatric Hematologist. The dose of ketorolac was 0.5 mg/kg IV every 6 hours, not exceeding 30 mg per dose and continued for a maximum of 5 days.

We recorded clinical variables for each VOC event that included age, SCD genotype (categorized as severe: HbSS and SB0 thalassemia or mild: HbSC and SB+ thalassemia), SCD modifying therapy (transfusion, hydroxyurea, no SCD modifying therapy), total hospital days, and ketorolac administration by days and dose (mg/kg). Laboratory variables were obtained from baseline (within 12 months from admission during “well” visit) and admission, including: hemoglobin (Hb) value, white blood cell count (WBC), absolute neutrophil count (ANC), platelet count, and absolute reticulocyte count (ARC). Change in laboratory values from baseline to admission were recorded and used in the statistical analysis.

Statistical analysis

Participants were dichotomized as either having developed AKI or not. Comparisons of the demographic, clinical, and laboratory characteristics of those with and without AKI were analyzed using the t-test for continuous variables and Fisher’s exact test or chi-squared test, where applicable, for categorical variables. To assess the associations between clinical variables and AKI, generalized linear mixed effects models for binary outcome with random intercept were fitted to accommodate repeated hospitalizations for the same subject. To examine the effect of the use of ketorolac in the development of AKI, a generalized linear mixed model for binary outcome was fitted to accommodate repeated hospitalizations for the same subject. Similarly, a generalized linear mixed Poisson model with random intercept was fitted to determine if AKI was independently associated with hospital length of stay.

Results

Among 197 admissions for VOC from 97 patients over a two year period, 33 (17 %) were identified with AKI. The mean and median number of admissions per patient was 2.1 and 1, respectively. Seventeen (52 %) of the thirty three AKI events were first identified from their ER SCr level; sixteen AKI events were identified during the pain hospitalization (mean: 1.8 days from admission). The mean age of hospitalized patients with AKI (13.2±4.9 yrs) was not statistically different from those without AKI (12.7±5.1 yrs, p=0.6). The frequency of AKI in females was 11 (11 %) out of 96 events as compared to 22 (22 %) out of 101 events in males (p=0.06). AKI events occurred in 28 (16 %) out of 175 patients with HbSS or SB0 thalassemia and 5 (23 %) out of 22 patients with HbSC or SB+ thalassemia (p=0.4). Only one (4 %) of 25 AKI events occurred in patients on transfusion therapy, 24 (18 %) out of 135 on hydroxyurea and 8 (22 %) out of 37 on no SCD modifying therapy (p=0.2).

Pediatric SCD patients with AKI had a larger drop in Hb from baseline to admission than patients without AKI (1.21 vs 0.50 g/dL, p=0.008) (Table 1). We identified no differences in AKI events vs. no AKI events by age (p=0.6) or change from baseline to admission for WBC (p=0.4), platelets (0.08) or absolute reticulocyte count (p=0.2) or need for IV fluid bolus in the Emergency department (p=0.6). For every 1 unit drop in Hb from baseline to admission Hb, the odds of AKI increased by 49 % (OR= 1.49, 95% CI 1.1–2.0). In multivariate analysis accounting for change from baseline to admission values for Hb and platelets, a drop in Hb remained the only significant variable (OR= 1.49, 95% CI 1.1–2.0).

Table 1.

Laboratory and clinical variables among patients with and without acute kidney injury (AKI)

Patient characteristics	Patient events with AKI (n=33)	Patient events without AKI (n=164)
Continuous variables	Mean (SD)	Mean (SD)	p-value
Age (years)	13.2 (4.9)	12.7 (5.2)	0.6
Delta Hemoglobin (g/dL)	−1.2 (1.1)	−0.5 (1.2)	0.008*
Delta WBC × 109/L	4.7 (4.7)	3.8 (5.7)	0.4
Delta platelet count × 109/L	−66 (147)	−11 (160)	0.08
Delta ARC × 109/L	0.03 (0.19)	−0.02 (0.18)	0.16
Nominal Variables
SCD Phenotype			0.4
Severe SCD (n=175)	28 (16%)	147 (84%)
Mild SCD (n=22)	5 (23%)	17 (77%)
Gender
Male (n=101)	22 (22%)	79 (78%)	0.06
Female (n=96)	11 (11%)	85 (89%)
SCD Modifying Therapy
No therapy (n=37)	8 (22%)	29 (78%)	0.2
Hydroxyurea (n=135)	24 (18%)	111 (82%)
Transfusion (n=25)	1 (4%)	24 (96%)

^*p<0.05 defined as statistically significant.

SD, standard deviation; Delta, Defined as admission-baseline laboratory value.

WBC, White Blood Cell Count; ARC, Absolute Reticulocyte Count; SCD, Sickle Cell Disease; Severe SCD, HbSS or HbSB0 thalassemia; Mild SCD, HbSC or HbSB+ thalassemia.

We evaluated the impact of having received ketorolac on participants who presented to the ER without AKI. Sixty four percent of these participants received at least one dose of ketorolac in the ER or hospital. Among patients admitted on ketorolac, they received the medication for a mean of 2.2 days (s.d. 1.3). No difference in the development of AKI was identified among participants who received at least one dose of ketorolac as compared to participants who did not receive any ketorolac during the hospitalization (p=0.14). However, among the participants that received at least one dose of ketorolac, the odds of developing AKI increased by 63 % (OR=1.63, 95% CI: 1.08, 2.47) for every additional day a participant received ketorolac (p=0.02). When adjusting for age, gender and SCD genotype, the total days of ketorolac remained associated with increased odds of developing AKI (OR=1.81, 95% CI: 1.16, 2.82). An increase in the dose of ketorolac (mg/kg) was also associated with increased odds of developing AKI in unadjusted (p=0.04) and adjusted models (p=0.04). Finally, patients who developed AKI received more doses of ketorolac than patients who did not develop AKI (5.9 vs. 4.3 doses, p=0.02) but the total doses did not reach statistical significance in survival modelling for time to develop AKI (p=0.28).

Patients admitted for VOC with AKI required a significantly longer length of stay than patients without AKI. Patients with AKI required a mean length of stay of 6.1 days as compared to 4.5 days for patients without AKI (p=0.002) (Figure 1). In a generalized linear mixed Poisson model, the number of hospital days required for those with AKI was 32 % longer than those without AKI (OR 1.32, 95% CI 1.10–1.58).

Figure 1.

The association of AKI events on length of stay for vaso-occlusive pain hospitalizations. LOS, Length of Stay; AKI, Acute Kidney Injury

Discussion

While the impact of single or repeated episodes of AKI on progression to CKD is lacking in SCD patients, this link has been clearly defined in other diseases [14]. This current study suggests that SCD patients can present with AKI or develop AKI during hospitalization for SCD pain episodes. Perhaps a very important risk factor which may help us understand the underlying pathophysiology of AKI in SCD is that those with AKI had a significantly larger drop in hemoglobin. This association between an acute drop in hemoglobin and development of AKI during ACS was also noted in our prior study [8]. A drop in hemoglobin during pain or ACS events could lead to organ pathophysiology due to complications from either hypoxic-ischemic events, hemolysis, or inflammation. In murine SCD models, brief episodes of hypoxic-ischemic events produce profound acute renal injury [25, 26]. Second, the murine model of SCD has shown that an increase in hemolysis or exposure to excess cell free hemoglobin can also lead to renal injury [27]. While not demonstrated yet in SCD, hemolysis or increases in plasma free heme/hemoglobin in patients undergoing cardiac surgery have been associated with renal injury and AKI [28, 29]. In the setting of a pain crisis, patients with a larger drop in hemoglobin may be subject to a higher risk for kidney injury due to an acute, direct toxic effect of plasma free heme or hemoglobin [30, 31]. While our data suggests that a drop in hemoglobin may directly impact the development of AKI during pain crisis, other potential etiologies for the development of AKI include complications associated with increased inflammatory state, volume depletion, or frequent NSAID use that occurs during a SCD pain crisis [30, 32, 33]. While all SCD patients should be monitored for AKI during their admission (as defined by a rise in SCr or decrease in urine output), special attention should be placed on patients with an acute drop in hemoglobin.

The NIH consensus moderately recommends, with low quality of evidence, the use of oral NSAIDs as an adjuvant therapy for pain in the absence of contraindications [19]. Recent data from animal and human studies highlight the potential risk of AKI on CKD [14]. Our study provides new data that could suggest caution to this NIH moderate recommendation, as 9 % of patients with pain crisis presented to the ER with findings of AKI and an additional 8 % developed AKI during their hospitalization. A second concern is that some patients may treat their prolonged pain crises at home with NSAIDs and oral opioids rather than seek ER care but do not have access to continuous intravenous fluids. These patients are also at risk for developing AKI but are not captured in this study. A third concern for frequent NSAID use is that some pediatric SCD patients have CYP2C9 allele variants that alter NSAID metabolism and may place these patients at increased risk for NSAID toxicity [34]. A final issue relevant to inducing AKI is that ER and inpatient teams may initially manage patients with ketorolac as part of standardized clinical pain pathways or based on individual physician preference [35, 36]. Two randomized clinical trials failed to demonstrate that ketorolac can reduce hospital days, or total morphine dose [37, 38]. While our study did not show a difference in AKI prevalence between patients who received at least one dose vs no dose, one prior published case report identified a SCD patient who developed irreversible renal failure while receiving ketorolac during a pain crisis despite aggressive hydration starting in the ER [39]. Our data did identify that multiple days and higher doses of ketorolac were risk factors for AKI among those patients who received at least one dose of ketorolac. Without a clear benefit to IV NSAIDs and potential risk for AKI among patients who receive multiple doses of ketorolac, physicians should consider not using or early discontinuation of ketorolac and closely monitoring patients who receive ketorolac for a decrease in urine output or increase in SCr from baseline clinic levels.

We acknowledge several limitations inherent to retrospective cohort studies. First, according to KDIGO definitions, urine output of <0.5 mL/kg/hr for six hours also fulfills the definition for AKI. As most patients do not have strict quantification of urine output in our hospital, we did not utilize oliguria as part of the definition for AKI in this retrospective study and instead relied solely on SCr-based definitions. Second, not all patients had SCr on every hospital day, so it is very possible that the true incidence of AKI in our cohort may be significantly higher. Third, we do not have data on the total days/doses of oral NSAIDs at home, nor did we have tools to adequately assess for hydration status at the time of presentation to the ER. Third, while we evaluated the impact of an IV fluid bolus in the ER as a potential surrogate for hydration status at the time of admission, we could not ensure accuracy of our records for oral intake prior to or during hospitalization in this retrospective review. Therefore we could not analyze the impact of total fluid intake prior to or during the hospitalization on the development of AKI. Finally, the definition of AKI includes an increase in SCr by ≥1.5 times baseline, which is known or presumed to have occurred within the prior seven days. As we often did not have a “well-clinic” visit baseline serum creatine obtained within seven days, we “presumed” that the SCr obtained at their last office visit also represented their baseline SCr within seven days prior to admission. Despite these limitations, this data improves the current scant literature on AKI in pediatric SCD patients and attempts to highlight the importance of monitoring for AKI during SCD crisis.

In conclusion, we have shown that at least 17 % of pediatric SCD patients admitted for pain crisis will develop AKI, including 9 % that present with AKI to the ER. An acute drop in hemoglobin was an identified risk factor for developing AKI and the pathophysiology of AKI and acute anemia should be explored. The impact of AKI is significant as patients who develop AKI require longer hospitalizations for their pain crisis, independent of potential confounders. In light of these findings, we believe that clinicians should closely monitor for AKI as defined by either: 1) 50 % increase in SCr from baseline, 2) ≥0.3 mg/dL increase in serum creatinine from baseline, or 3) <0.5 mL/kg/hr of urine output for 6–12 hours during hospital admission for SCD. Clinicians should consider the risk of NSAIDs (IV or oral) on the kidneys when developing institutional or individualized pain treatment plans. Future research should focus on the pathophysiology of AKI in patients with SCD and the impact of AKI on progression to SCD nephropathy. In addition, the risk and benefits of NSAID administration during pain crises should be prospectively evaluated. Finally, novel strategies for pain management that do not rely on known nephrotoxic medications should be evaluated in clinical trials for this vulnerable population whose long-term life-expectancy is defined by kidney failure [3, 4].