This article requires a subscription to view the full text. If you have a subscription you may use the login form below to view the article. Access to this article can also be purchased.
Abstract
Background and objectives MicroRNAs (miRNAs) are small ribonucleotides regulating gene expression. MicroRNAs are present in the blood in a remarkably stable form. We tested whether circulating miRNAs in the plasma of critically ill patients with acute kidney injury (AKI) at the inception of renal replacement therapy are deregulated and may predict survival.
Design, setting, participants, & measurements We profiled miRNAs using RNA isolated from the plasma of patients with AKI and healthy controls. The results were validated in 77 patients with acute kidney injury, 30 age-matched healthy controls, and 18 critically ill patients with acute myocardial infarction by quantitative real-time PCR.
Results Circulating levels of miR-16 and miR-320 were downregulated in the plasma of kidney injury AKI patients, whereas miR-210 was upregulated compared with healthy controls (all P < 0.0001) and disease controls (miR-210 and miR-16: P < 0.0001; miR-320: P = 0.03). Cox regression (P < 0.05) and Kaplan–Meier curve analysis (P = 0.03) revealed miR-210 as an independent and powerful predictor of 28-day survival.
Conclusions Circulating miRNAs are altered in patients with kidney injury AKI. MiR-210 predicts mortality in this patient cohort and may serve as a novel biomarker AKI reflecting pathophysiological changes on a cellular level.
Introduction
Acute kidney injury (AKI) in critically ill patients has been identified as an independent risk factor for adverse outcome (1). Mortality of patients with AKI in the intensive care unit (ICU) is still unacceptably high despite significant advances in supportive care (2). A recent multinational, multicenter study including 29,000 critically ill patients revealed that the in-hospital mortality of patients with AKI exceeds 60% (3). Early markers identifying patients at risk, a prerequisite for therapeutic interventions, are not currently available. Thus, the search for new biomarkers in this setting remains an area of utmost interest.
MicroRNAs (miRNAs) are endogenous, single-stranded molecules consisting of approximately 22 noncoding nucleotides. Genetic gain- and loss-of-function studies show a disease-specific role for selected miRNAs (4). Thus, deregulation of miRNAs likely results in a severe disturbance of downstream gene networks and signaling cascades within the cell, culminating in disease initiation and/or progression (5). However, recent studies also demonstrated miRNAs to be detectable in the circulation and useful as biomarkers for diseases (6–8). MiRNAs are present in the blood in a remarkably stable form that even withstands repetitive freezing/thawing cycles and is protected against RNases (6,7,9).
In study presented here, we tested the hypothesis that the circulating miRNAs detected in the circulation of critically ill patients with AKI requiring renal replacement therapy (RRT) are altered and might serve as biomarkers predicting survival. Circulating miRNAs were measured in the plasma of 77 patients, 30 age-matched healthy controls, and 18 disease with AKI controls with acute myocardial infarction.
Study Population and Methods
This study is a post hoc measurement of prospectively collected blood samples from the Hannover Dialysis Outcome (HANDOUT) trial (10). The study protocol was approved by the Hannover Medical School Ethics Committee (project/approval no. 2905) and was conducted in accordance with the Declaration of Helsinki and German federal guidelines. Patients in seven ICUs of the tertiary care center at Hannover Medical School suffering from AKI were evaluated for inclusion. The inclusion criteria were non post-renal AKI with RRT dependence indicated by a loss of kidney function of >30% of the estimated GFR (eGFR) calculated with the Modification of Diet in Renal Disease or Cockroft–Gault equation and/or cystatin C-GFR within 48 hours before inclusion and oliguria/anuria (<30 ml/h >6 hours before inclusion) or hyperkalemia (>6.5 mmol/L) or severe metabolic acidosis (pH < 7.15, bicarbonate < 12). Exclusion criteria were pre-existing chronic kidney disease as defined by eGFR <60 ml/min or a serum creatinine concentration >1.7 mg/dl >10 days before initiation of the first RRT. Further exclusion criteria were participation in another study, consent denial or withdrawal, and need for extracorporeal membrane oxygenation therapy. Attending nephrologists performed the enrollment after obtaining written informed consent from a patient or his/her legal representatives. If the patient was recovering and able to communicate, he/she was informed of the study purpose and consent was required to further maintain his/her status as a study participant.
After inclusion, the specific medical condition leading to RRT initiation was documented out of a list of five possible causes requiring immediate RRT (10). All patients received nutritional intake of at least 25 to 30 kcal/kg per day, preferentially delivered as enteral nutrition. The prescribed protein intake was >1.2 g/kg per day. RRT in all patients was performed in a slow, extended dialysis mode using the GENIUS dialysis system (Fresenius Medical Care, Bad Homburg, Germany) as described in detail elsewhere (11). We used high-flux polysulfone dialyzers (F60S, 1.3 m2, Fresenius Medical Care, Bad Homburg, Germany) for all treatments. The dose of the RRT was tailored according to the patient's individual need, starting with at least one treatment daily. RRT was discontinued in patients meeting the following criteria for renal recovery: urine output >1000 ml/d and/or increased solute clearance (i.e., decline in pretreatment serum creatinine concentration with eGFR >15 ml/min [by Modification of Diet in Renal Disease equation, Cockroft–Gault equation, and/or cystatin C-GFR]). Serum cystatin C, serum creatinine, and serum C-reactive protein (CRP) levels were determined by routine laboratory methods.
Sequential Organ Failure Assessment (SOFA) score (12) and Acute Physiology and Chronic Health Evaluation II (APACHE II) score (13) were obtained for each patient immediately before initiation of RRT. The presence of sepsis was defined according to the Society of Critical Care Medicine/European Society of Intensive Care Medicine/American College of Chest Physicians/American Thoracic Society/Surgical Infection Society International Sepsis Definitions (14). AKI was classified post hoc by means of the RIFLE criteria at initiation of RRT (15).
Sampling and Quantification of circulating miRNAs
Plasma samples of all patients were obtained before the start of RRT. RNA was isolated using the MasterPure RNA purification kit (Epicenter Biotechnologies) according to the manufacturer's instructions. We supplemented the samples with 5 fmol/μl Caenorhabditis elegans miR-54 (cel-miR-54) and normalized circulating miRNAs to this control as described previously (9).
MiRNA Transcriptome Profile Analysis
Pooled total RNA from five patients with AKI and five healthy age-matched controls was used for genome-wide miRNA array analysis. Total RNA was labeled and processed as described earlier (5). We used Affymetrix miRNA GeneChip array analysis using GeneChip miRNA arrays. For normalization and further data analysis, the XRAY Excel software tool (Biotique) was used. In brief, for data normalization we used Affymetrix CEL files. The input files were normalized by scaling to a common median value. A scaling factor was added to each probe expression value such that for each chip, the median probe value was equal to the probe median expression across all arrays. Arrays are corrected for background by the method described in the Robust Multiarray Average (16). After normalization, the array probe values are adjusted for background by assuming that perfect match probe scores are distributed as the sum of independent exponential (S = signal) and normal (B = background) distributions. After normalization, background correction, and transformation, the perfect match probe scores were combined to estimate a gene level for each chip by means of median polish applied to microarray analysis by Irizarry et al. (16). In conclusion, we provide fold changes of deregulated miRNAs between healthy controls and patients with AKI as well as normalized raw expression data of the initial Affymetrix screen (see Table 2).
Detection and Quantification of miRNAs by Quantitative Real-Time PCR
MiR-16, miR-210, and miR-320 were validated by quantitative miRNA real-time PCR (RT-PCR) technology (TaqMan MicroRNA Assays, Applied Biosystems, Foster City, CA) in 77 AKI patients, 30 age-matched healthy controls, and 18 patients with acute myocardial infarction as disease controls. Values were normalized to spiked-in cel-miR-54. Mir-1249 was amplified as a control miRNA as shown previously (17). To assess the prognostic value of circulating miRNAs in patients with AKI, levels of circulating angiopoietin-2 (Ang-2) were measured as an established biomarker of AKI (18) and compared with levels of circulating miR-210. Ang-2 was quantified in a blinded fashion by in-house ELISA as described previously (19).
Study Outcomes and Statistical Analyses
The main objective of the study was to analyze the predictive value of circulating miRNAs concerning mortality and renal recovery of critically ill patients with AKI receiving RRT. The study endpoint was defined as survival 4 weeks after initiation of RRT.
Variables were analyzed for normal distribution using the Kolmogorov–Smirnov test. Correlation analysis for normally distributed variables was performed using the Pearson correlation coefficient. Non-normally distributed variables were assessed by Spearman correlation analysis. Continuous variables are expressed as medians with corresponding 25th and 75th percentiles (interquartile range) and were compared by using the Mann–Whitney rank sum test or the Kruskal–Wallis one-way ANOVA. Categorical variables were compared using the χ2 test. Parameters independently associated with survival were identified by univariate and multivariate Cox proportional hazards models. To fulfill the assumptions of normality, values of miR-210 were subjected to logarithmic transformation. Variables found to be statistically significant at a 10% level in the univariate analysis were included in the multivariate model using backward elimination. The distribution of the time-to-event variables was estimated using the Kaplan–Meier method with log-rank testing. For comparison of the prognostic values of parameters, receiver operating characteristic (ROC) curves were generated. Two-sided P values < 0.05 were considered statistically significant for all statistical procedures used. All statistical analyses were performed with the SPSS package (SPSS, Inc., Chicago, IL).
Results
To assess an effect of AKI on circulating miRNAs, we conducted a global miRNA expression analysis using RNA isolated from the plasma of patients with AKI at inception of RRT (n = 5) and healthy age-matched controls (n = 5). The clinical characteristics of the whole cohort of AKI patients (n = 77) are shown in Table 1. The levels of 13 circulating miRNAs differed between patients and healthy controls (Table 2). We screened the levels of circulating miR-16, miR-320, and miR-210 in a validation set of 77 patients, 30 age-matched controls, and 18 patients with acute myocardial infarction by using TaqMan quantitative PCR. As shown in Figures 1a through 1c, the results of the miRNA profile could be confirmed in the entire cohort, which identified miR-16 and miR-320 to be downregulated (both P < 0.0001) and miR-210 to be upregulated (P < 0.0001) in patients with AKI compared with healthy controls. None of the circulating miRNAs were significantly elevated in healthy controls compared with the disease control group (miR-16 and miR-320: P = 0.2; miR-210: P = 0.09). However, levels of all miRNAs differed significantly between disease controls and AKI patients (miR-210 and miR-16: P < 0.0001; miR-320: P = 0.03).
Demographic, clinical, and laboratory characteristics of patients
Raw data of deregulated miRNAs in a global miRNA expression analysis (Affymetrix gene arrays)
Levels of circulating microRNAs (miRNAs) are strongly deregulated in patients with acute kidney injury (AKI) versus healthy controls (CTL) and patients with acute myocardial infarction (non-AKI). (A) MiR-16 and (C) miR-320 are downregulated compared with AKI patients (both P < 0.0001), whereas (B) miR-210 is upregulated (P < 0.0001). Compared with non-AKI critically ill controls, all miRNAs are significantly deregulated ([B] miR-210 and [A] miR-16: P < 0.0001; [C] miR-320: P = 0.03). (D) Differences in baseline levels of circulating miR-16 did not reach statistical significance in survivors versus nonsurvivors (P = 0.9). Baseline levels of circulating (E) miR-210 and (F) miR-320 were significantly lower in survivors versus nonsurvivors at 4 weeks after initiation of renal replacement therapy (RRT) (P = 0.02 and P = 0.03, respectively).
The only significant associations were as follows: baseline miR-16 (r = 0.3, P = 0.02), miR-210, and miR-320 levels (both r = 0.4, P < 0.0001) correlated with lactate levels, and baseline miR-210 additionally correlated with heart rate (r = 0.4, P < 0.0001). MiR-320 also correlated with heart rate (r = 0.4, P < 0.0001) and noradrenaline dose (r = 0.3, P = 0.04).
We did not find an association between circulating miRNAs and APACHE II or SOFA score. There were no significant differences in levels of circulating miRNAs in patients with or without sepsis (miR-16: P = 0.3; miR-210: P = 0.4; miR-320: P = 0.6), surgery (miR-16: P = 0.4; miR-210: P = 0.7; miR-320: P = 0.8), or shock (miR-16: P = 0.5; miR-210: P = 0.5; miR-320: P = 0.4). Similarly, levels of circulating miRNAs did not differ between patients on a medical, general surgical, or cardiothoracic ICU or RIFLE class (miR-16: P = 0.4; miR-210: P = 0.8; miR-320: P = 0.5) as assessed by ANOVA (miR-16: P = 0.4; miR-210: P = 0.3; miR-320: P = 0.08). Levels of circulating miRNAs did not correlate with serum creatinine levels at the start of RRT (miR-16: r = −0.14, P = 0.2; miR-210: r = −0.3, P = 0.8; miR-320: r = −0.4, P = 0.8). Twenty-three patients (30%) did not receive heparin before the start of RRT because of bleeding complications and/or liver insufficiency. Heparin treatment did not have an effect on circulating levels of miRNAs (miR-16: P = 0.9; miR-210: P = 0.9; miR-320: P = 0.7). MiR-1249, which did not show differential regulation in AKI patients versus healthy controls in the initial Affymetrix screen, was analyzed as a control miRNA. Levels of miR-1249 did not differ between AKI patients and healthy controls (P = 0.5).
Patients were grouped as survivors and nonsurvivors at 4 weeks after initiation of RRT. Patient groups were comparable with respect to baseline demographics and the proportion of RIFLE categories (Table 1). Survivors and nonsurvivors differed with regard to the proportion of patients with metabolic acidosis (P = 0.005) and SOFA score (P = 0.05). Patients with sepsis had higher SOFA and APACHE II scores that resulted from more severe cardiovascular and respiratory impairment compared with nonseptic patients (data not shown). A total of 28 patients died in our cohort. Levels of circulating miR-210 (P = 0.02) and miR-320 (P = 0.03) were significantly elevated in nonsurvivors, whereas miR-16 did not show differential regulation in nonsurvivors compared with survivors (Figures 1d through 1f). To determine the relationship between circulating miRNA levels at the initiation of RRT and mortality, we initially performed univariate Cox proportional hazards analyses. In our cohort of 77 critically ill patients with AKI, CRP levels, age, gender, and body mass index were not significantly associated with survival (Table 3). Among the variables tested, miR-210, Ang-2 levels, major surgery, sepsis, and APACHE II and SOFA scores displayed prognostic significance at a 10% level and were subsequently subjected to multivariate Cox regression analysis (Table 3). Only major surgery (P = 0.02) and miR-210 (P = 0.049) remained independent predictors of survival in multivariate analysis. The positive predictive value of miR-210 is 0.412 (95% confidence interval: 0.38 to 0.41), whereas the negative predictive value is 1.0 (95% confidence interval: 0.72 to 1.0). In ROC curve analysis, miR-210 yielded an area under the ROC curve (AUC) value of 0.7 (SEM: 0.06; 95% confidence interval: 0.53 to 0.78; P = 0.03), whereas miR-320 showed a similar level of significance (AUC: 0.7; SEM: 0.07; 95% confidence interval: 0.53 to 0.78; P = 0.03). MiR-16 levels did not reach statistical significance (AUC: 0.55; SEM: 0.07; 95% confidence interval: 0.41 to 0.69; P = 0.4). For comparison, the SOFA score yielded an AUC value of 0.7 (SEM: 0.06; 95% confidence interval: 0.57 to 0.82; P = 0.005). Figure 2 illustrates the Kaplan–Meier curve of survival 4 weeks after initiation of RRT stratified to miR-210 above and below the median. Log-rank test confirmed statistical significance for miR-210 (P = 0.03), whereas miR-320 (P = 0.08) and miR-16 (P = 0.4) were NS.
Univariate and multivariate Cox regression analyses for survival
Kaplan–Meier curve of survival in critically ill patients with AKI stratified to circulating miR-210 above and below median. Log-rank test confirmed statistical significance for miR-210 (P = 0.03).
Discussion
Our study is the first clinical evaluation of circulating levels of miRNAs in critically ill patients with AKI requiring RRT. The results are as follows: (1) the detection of circulating miRNAs in the plasma of critically ill patients with AKI is feasible, (2) miR-16 and miR-320 are downregulated whereas miR-210 is upregulated in AKI patients, (3) baseline levels of miRNAs correlated with lactate levels and heart rate, (4) baseline levels of circulating miR-210 and miR-320 are elevated in nonsurvivors compared with survivors, and (5) miR-210 was identified as a strong independent prognostic factor for 28-day survival in the multivariate Cox proportional hazards regression analysis and Kaplan–Meier curve analysis.
We previously analyzed the prognostic value of osteopontin in patients with AKI (20). Although samples in the study presented here as well as our previous osteopontin study were obtained from patients included in the HANDOUT trial, the sample size in this study was smaller than in the osteopontin trial (77 versus 109). Secondly, a different set of patients of the HANDOUT trial was included in the trial presented here. Thus, there was only minimal overlap between patients of the two studies.
Here we show the successful isolation and measurement of circulating miRNAs in patients with AKI. Due to the intricate regulation of multiple miRNAs and the unknown effect of factors that modulate circulating miRNA levels, we did not normalize levels of circulating miRNAs to a circulating “housekeeping” miRNA. Instead, we supplemented the plasma samples with recombinant cel-miR-54, which can be specifically detected by TaqMan quantitative RT-PCR, to normalize potential differences in the efficiency of RNA isolation as shown previously (9). In our study we focused on miR-210 because of its established role in the molecular response to hypoxia (21). Hypoxia-inducible factor is a crucial inductor of miR-210 in endothelial cells (21). Additionally, miR-210 hypoxic induction was shown to be dependent on the von Hippel–Lindau tumor suppressor gene in RCC4 renal carcinoma cells (22). It is thus conceivable that miR-210 is released from renal endothelial cells, among others, in response to hypoxia in AKI. This hypothesis is buttressed further by findings in other ischemic diseases such as myocardial infarction, in which miR-210 was shown to be upregulated (23). In our study, miR-210 showed borderline upregulation in the disease control group (patients with acute myocardial infarction). However, the level of upregulation in patients with AKI was much greater, reaching high statistical significance compared with our disease control group and healthy controls. The level of deregulation in our cohort may thus be viewed as AKI specific. Heparin was previously shown to influence the results of quantitative RT-PCR analysis (24). We thus analyzed the exposure of patients in our cohort to heparin before the start of RRT. Twenty-three patients did not receive heparin. The exposure to heparin did not have an effect on circulating levels of miRNAs.
We provide here evidence of miR-210 as an independent prognostic survival factor in AKI. We also assessed levels of circulating Ang-2 as an established biomarker of AKI and levels of CRP as an inflammatory marker in our cohort of patients to compare levels of circulating miRNAs to other markers of AKI. Although Ang-2 levels displayed prognostic significance in univariate analysis, miR-210 and major surgery were the only independent prognostic factors in multivariate Cox regression analysis. However, it was not the purpose of our study to suggest that circulating miRNAs in the setting of AKI might outperform other well established markers. Detection of circulating miRNAs merely represents an interesting, novel approach of stratifying the prognosis of critically ill patients with AKI as compared with other prognostic scores such as SOFA or other biomarkers. Because tissues and circulating cells release circulating miRNAs in response to toxic insults, their level in blood might reflect alterations in the setting of AKI on a cellular level. In addition, circulating miRNAs might also be a reflection of crosstalk between different cell types because of their release in exosomes and uptake by neighboring cells.
We have shown that certain miRNAs are downregulated in the setting of AKI. This might be because circulating miRNAs are secreted to a lower extent by cells, are taken up by other circulating or resident cells, or are degraded. The metabolic alterations of AKI (i.e., hyperkalemia, metabolic acidosis) might lead to the degradation of certain circulating miRNAs, although this hypothesis has not yet been addressed so far.
There are certain limitations to our study. First and foremost, we cannot provide molecular insights into the underlying mechanisms that cause a dysregulation of circulating miRNAs in AKI patients. Multiple parameters may have affected the levels of circulating miRNAs, including the change of tissue expression or the release of the miRNAs by circulating cells. In addition, the dysregulation may also reflect a modulation of the miRNA processing apparatus because it has been shown that cellular stressors may affect the expression of the enzyme Dicer, which is essential for the biogenesis of mature miRNAs (25). Forty-eight of 77 AKI patients (62%) were anuric/oliguric. Because of the limited value of patients with preserved urinary output, we did not analyze urinary miRNA levels in this cohort. Urinary miRNA levels are interesting biomarkers as a possible reflection of kidney miRNA production, which should be analyzed in future studies. Experimental studies are warranted to further elucidate the mechanisms by which the development of AKI induces the release of circulating miRNAs by kidney tissue and circulating cells.
In conclusion, the study presented here provides first insights into levels of circulating miRNAs in critically ill patients with AKI. An array of circulating miRNAs can be reliably detected in the plasma of AKI patients. We identified miR-210 as a strong and independent predictor of survival in critically ill patients with AKI. Larger, prospective clinical studies are needed to confirm the results of our single-center study.
Disclosures
J.L., J.T.K., and T.T. filed a patent application about the use of circulating miRNAs in kidney disease.
Acknowledgments
We acknowledge the support of the Deutsche Forschungsgemeinschaft (DFG LO 1736/1-1) to J.L. and T.T. and a grant from the Integrated Research Center Transplantation to T.T. J.M.L. and J.T.K. contributed equally to this manuscript.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
See related editorial, “Circulating Micro-RNAs in Acute Kidney Injury: Early Observations,” on pages 1517–1519.
- Received January 15, 2011.
- Accepted March 7, 2011.
- Copyright © 2011 by the American Society of Nephrology
References
Jump to section
More in this TOC Section
Original Articles
Cited By...
- MicroRNAs in AKI and Kidney Transplantation
- Biomarkers for the Early Detection and Prognosis of Acute Kidney Injury
- Neurotensin Promotes the Development of Colitis and Intestinal Angiogenesis via Hif-1{alpha}-miR-210 Signaling
- Impairment of Wound Healing in Patients With Type 2 Diabetes Mellitus Influences Circulating MicroRNA Patterns via Inflammatory Cytokines
- Noninvasive Micromarkers
- Human miRNome Profiling Identifies MicroRNAs Differentially Present in the Urine after Kidney Injury
- Increased Serum and Urinary MicroRNAs in Children with Idiopathic Nephrotic Syndrome
- Circulating and Urinary microRNAs in Kidney Disease
- Micro-RNA-34a Contributes to the Impaired Function of Bone Marrow-Derived Mononuclear Cells From Patients With Cardiovascular Disease
- The TGF-{beta} Route to Renal Fibrosis Is Not Linear: The miR-21 Viaduct
- Circulating Micro-RNAs in Acute Kidney Injury: Early Observations