MicroRNAs in toxic acute kidney injury: Systematic scoping review of the current status

Abstract Acute kidney injury induced by nephrotoxic agents is common, increasing in incidence and associated with considerable morbidity and mortality in developing countries. MicroRNAs are stable biomarkers that can be detected in extracellular fluids. This systematic scoping review aims to describe published research on urinary and circulating microRNAs in toxic acute kidney injury in both animal and human studies. We conducted a literature search, using EMBASE and Medline, for articles on urinary and circulating microRNA in nephrotoxic injuries to February 2020. A total of 21 publications studied acute kidney injury from 12 different toxic agents. Cisplatin was the most common nephrotoxic agent (n = 10), followed by antibiotics (n = 4). There were no randomized controlled trials. An increase in urinary miR‐218 predicted acute kidney injury in six different studies, suggesting it is a promising biomarker for nephrotoxin‐induced acute kidney injury. There were many factors that prevented a more comprehensive synthesis of microRNA performance including highly variable models, no consistent protocols for RNA isolation, cDNA synthesis and PCR amplification, and variability in normalization methods using reference controls. In conclusion, while microRNAs are promising biomarkers to study nephrotoxic acute kidney injury, the replication of most positive findings is not assessable due to deficient reporting of negative outcomes. A very narrow range of poisons have been studied, and more human data are required. In particular, further studies are needed on the most important causes of nephrotoxic injury, such as pesticides, chemicals, snake envenoming, and medicines other than aminoglycosides and cisplatin.


| INTRODUC TI ON
Acute kidney injury (AKI) caused by drugs, chemicals, and toxins is a common clinical problem around the world. Deliberate self-poisoning with pesticide and snake envenomation are typical causes of toxic AKI in developing countries, 1 while toxic AKI from drugs such as chemotherapeutic agents is most common in western countries. 2 A major challenge in the management of AKI is the only modest sensitivity of current injury biomarkers for early diagnosis. 3,4 These biomarkers have had limited translation into clinical medicine 5 and there is still an ongoing search for more sensitive and specific biomarkers for diagnosis of AKI. Better biomarkers of toxic AKI may help in earlier diagnosis, allowing early and appropriate treatment, and a means to monitor treatment effects.
MicroRNAs (miRNAs) are a class of RNAs that are potential diagnostic and prognostic biomarkers. MicroRNAs are endogenous, noncoding RNAs, 21-25 nucleotides in size that regulate gene expression by binding to target messenger RNAs (mRNAs). [6][7][8] They have potential advantages over protein biomarkers due to higher stability and tissue specificity, and the ease of quantifying large numbers of them simultaneously. 9 Even though miRNAs were discovered in 1993, publications on miRNAs and kidneys only started from 2007, with rapidly growing interest in the last decade ( Figure S1). Toxic agents lead to many changes in tissue miRNA expression. This may be detected in miRNA released into the systemic circulation, including when there is cell/tissue lysis due to injury. MicroRNAs can be detected in a range of biological samples, such as blood, urine, cerebrospinal fluid, feces, and breast milk. 10 The majority of the reviews published so far focus on noncoding RNAs in various types of AKI and the information on miRNAs in nephrotoxic AKI is not organized. There are recent reviews published on noncoding RNAs in other types of AKI, but the limited information on miRNAs due to nephrotoxic AKI is not separable. 11 Many studies report tissue miRNA regulation in nephrotoxic AKI. However, no systematic reviews have focused on miRNAs that might be easily and repeatedly measured in urine or blood as biomarkers of renal toxicity.
This systematic scoping review aims to describe published research on urinary and circulating miRNAs in nephrotoxic AKI in both animal and human studies. This will provide a useful basis for future studies that address cell-free miRNA as biomarkers for toxin-induced AKI.

| Search strategy
We conducted literature searches using EMBASE (1993 to February 2020) and Medline (1993 to February 2020) for articles on circulating miRNA in toxic AKI. We used a combination of sets of search terms: [(acute toxicity OR nephrotoxicity OR kidney injury OR kidney damage OR organ toxicity OR acute kidney injury OR toxin OR snakebite OR snake bite envenomation) AND (serum microRNA OR urinary mi-croRNA OR plasma microRNA OR circulating microRNA OR cell-free microRNA OR microRNA OR serum miRNA OR plasma miRNA OR Urinary miRNA OR circulating miRNA OR miRNA)]. We used Covidence systematic review management system for the title/abstract and fulltext screening. We restricted the search to the English language.

| Screening for study relevance and eligibility
Randomized/nonrandomized trials, controlled/noncontrolled trials, cohort studies, and case-control studies that investigated the use of miRNAs as biomarkers for the diagnosis or risk stratification of AKI following nephrotoxic agents (drugs and chemicals) were considered eligible. Furthermore, we included studies reporting on the detection (iii) studies with no available full text (eg, conference abstracts).

| Data extraction
We extracted data into a prespecified data collection form to capture studied agents, experimental method, animal model or human study results, time and dose responses, and the number of miRNAs profiled (Table S1). We extracted data on the miRNAs identified within the site of injury and their expression in urine and serum.

| Literature selection and study characteristics
Our search identified 1378 unique peer-reviewed journal articles published since 2007. We then performed a detailed evaluation of each title and abstract and 1343 were excluded due to being other forms of AKI, miRNA studied in cells and tissues, review articles, conference abstracts or editorial/commentary, and book chapters.
Finally, 35 articles were eligible for full-text screening and we selected 21 for final analysis.  The application of the screening process and selection criteria is summarized in Figure 1

| Nephrotoxic agents studied
There have been no studies on circulating or urinary miRNA in AKI induced by pesticides, chemicals, or snakebite envenoming. Studies to date have focused on chemotherapeutic agents such as cisplatin, 13

| miRNA profiling models and type of specimen
miRNA studies have been predominantly performed in animal models (16/21 and three studies in both animal and humans), usually rat models, (Table 1). All studies had limited power. Animal studies consistently had small numbers per group (ie n = 5 to 10); human studies ranged from 24 to 92 patients per group (Supplementary Table 2). The two largest studies were both on contrast medium-induced AKI in humans. 21,27 Only two studies simultaneously examined miRNA expression in the circulation, urine and tissue in animal models. 16,32 One reported increases in two miRNA in kidney tissue with decreased levels in plasma and urine. 32 The other only reported statistically significant changes in miRNAs, and reported minimal overlap in results from kidney, plasma, and urine. 16 Rat studies generally used outbred stocks (Wistar and Sprague-Dawley, 14/15) while mice studies more often used inbred strains (Balb/c, 2/4) (Table S2).

| miRNA detection techniques
The majority of studies used TaqMan low-density array (TLDA) card RT-qPCR detection technique (9/21) or SYBR Green RT-qPCR (5/21) (Table S2). Few studies profiled candidate miRNAs and then validated the best candidates in the same model. 16,29 F I G U R E 1 Search strategy flow diagram. The PRISMA diagram shows the process of literature identification, screening, eligibility, and study included in the review

| Comparison of miRNA profiling in animal model and humans
Two studies on contrast medium-induced AKI showed higher expression of serum miR-30a-5p, miR-30e-5p, and miR-30b in both animal models and humans (Table 2). 21,27 There were no urine miRNAs that had comparable changes in both animals and humans, noting there was only one human study examining urinary miRNA. 23

| Confirmation of kidney injury and categorization of different stages of AKI
Most animal studies confirmed AKI by histopathological examination of kidney tissue alone (11/16) or histology with serum creatinine (SCr) (5/16). In contrast, human studies used SCr, 21,24,27 12 or histopathology 23 to confirm AKI (Table S2). Grading of severity of AKI was generally done in animal studies but not in human studies.

| Quality measures in the selected studies
We systematically analyzed the studies further to assess the quality of the study in terms of study design, the methods, and the outcome ( Figure 3A and B).

Study design
We found only two studies performed global profiling and validated the selected miRNAs in an independent study sample using the same model and same sample type. 22,31 Only three studies examined miRNA expression in both circulation and urine. 16,31,32 Method The majority of studies used healthy controls (ie, sham group in animal models) while only three studies also used disease controls (poisoned but no kidney injury). 15,26,29 TaqMan probe-based RT-qPCR technique was widely used with exogenous controls for normalization.

Study outcome
In general, the studies focused reporting on positive results. Only

| DISCUSS ION
This review identified only 21 studies on circulating or urinary miR-NAs in nephrotoxic AKI in animal models and humans, mainly concerning drug-induced AKI. We found no studies that investigated important nondrug agents such as pesticides, chemicals, and envenoming induced AKI. for toxicological studies is often suggested as they are more stable, better defined, more uniform, and provide more repeatable data than outbred stocks. 35,36 Complete inbreeding is an unusual genetic state for a mammal that does not accurately reflect the human genetic makeup. Therefore, the use of Diversity Outbred mice, a newly developed mouse population derived from progenitor lines of the Collaborative Cross is highly recommended as a

F I G U R E 3 Quality measures included in 21 studies (A) Summary as heat map and (B)
percentage. Quality of study determined by the authors. Green, light brown, and red indicate reported, unclear/not relevant, and not reported, respectively powerful model system that more closely reflects the genetic variability seen in humans. 37 Advantages of Diversity Outbred mice in genetics are higher mapping resolution, increased heterozygosity, and uniformly distributed genetic variation across the genome. 38 Knowledge of the specificity of the miRNA is also required to define the clinical utility of potential biomarkers. The change in miRNA expression might be specific to one or a wide range of toxic agents, they may or may not be species specific. We found only a few miR-NAs that were shown multiple times to detect AKI after nephrotoxic agent exposure. Notably, miR-218 was associated with cisplatin, gentamicin, and puromycin-induced AKI. 14,15,20,25,29,30 MiR-218 regulates cell migration via the Slit-Robo pathway and promotes the motility of renal tubular epithelial cells that undergo epithelial mesenchymal transition. 39 Thus upregulation of miR-218 might indicate tubular AKI and cell apoptosis from any tubular nephrotoxic process. 40 In contrast, some miRNA may be specific to a particular toxic injury. 41 Rat models were generally the models used for in vivo animal studies. However, different miRNA profiles were often reported for the same nephrotoxic agent. For example, three cisplatin rat models did not identify any common urinary miRNAs after AKI. 16,28,30 This concerning discrepancy in reported miRNA expression was not explored by the authors, but might feasibly be due to variable rat strains, doses, sample collection times, or different techniques used in each study. Even greater discrepancies would be expected between animal and human studies. 42 Useful new clinical diagnostic biomarkers need to detect AKI earlier than the traditional biomarker SCr, ideally with no additional assays, adjustments, or calculations. Interestingly, adjusting for age and gender did not change the diagnostic performance of miR-21 and miR-423 in paracetamol-induced AKI. 23 UCr is often used to normalize for urine dilution in assays of urinary protein biomarkers. Results adjusted for urine dilution were presented by roughly half the studies. However, normalizing miRNA to UCr or urine volume may be superfluous as typically the miRNA concentrations were elevated around a million-fold. Ideally studies would report both absolute and normalized values for urinary protein biomarkers, 43 and demonstrate if this was unnecessary; however, this was not done by any studies.

| Limitations of the review
One limitation of this systematic review is that we only included articles in English; however, we did not identify any relevant non-English articles with our search terms. Additionally, we did not perform any meta-analysis on the specificity and sensitivity of miRNAs due to ex- Profile and validate using the same model and sample type (serum or urine).
Use both disease controls and healthy controls as well as the kidney injury group.

| Method
Normalize using combinations of control miRNAs (exogenous and endogenous).
Validate the results using the same or secondary technique such as real time or digital PCR.

| Outcome
Compare the results with healthy controls and disease controls.
Report both absolute quantification and the fold change.
Report all analyzed miRNAs (significant and nonsignificant).

| CON CLUS I ON AND FUTURE DIREC TIONS
While miRNA are promising biomarkers to study nephrotoxic AKI, we identified (i) lack of attempts to replicate most findings, (ii) no consistent protocols for RNA isolation, cDNA synthesis, and PCR amplification, (iii) variability in methods used to normalize to reference controls (iv) some major discrepancies between study findings (perhaps explained by i-iii, above), (v) a very narrow range of poisons studied, and (vi) a dearth of human data. Thus, we require more studies in both humans and animals on other important causes of nephrotoxic injury, such as pesticides, chemicals, envenoming, and medicines other than aminoglycosides and cisplatin.
Standardizing techniques and reference controls and/or reporting absolute quantification would allow for better comparisons between studies.

| E THIC S APPROVAL S TATEMENT
No ethical approval required for the systematic review.

D I SCLOS U R E S
The authors declare that they have no conflict of interest.