Skip to main content
Premium Trial:

Request an Annual Quote

For Drug Toxicity Testing, microRNAs Hold Promise But More Work Remains


While much attention has been focused on the use of microRNAs for diagnostic applications, a number of pharmaceutical firms have been exploring the non-coding RNAs' potential as biomarkers of drug-related toxicity.

However, these efforts are early stage and require additional work before miRNAs could reliably replace traditional biomarkers, according to data presented this week at the Cambridge Healthtech Institute microRNA as Biomarkers and Diagnostics meeting in Boston.

Although a number of biomarkers exist for predicting a drug's toxicity both preclinically and clinically, these typically are only detectable after injury in a particular organ, Greg Falls, manager of discovery and molecular toxicity at GlaxoSmithKline, noted during a talk at the conference.

For instance, an increase in the protein complex troponin, which is involved in heart muscle contraction, is commonly used to detect cardiovascular toxicity. But this is only "reporting damage" that has occurred, rather than predicting it, he said.

A similar situation exists with the widely used liver toxicity biomarker creatinine, Rounak Nassirpour, principal scientist of drug safety research and development at Pfizer, said. "An ideal biomarker needs to come up way before the levels of serum creatinine increase."

Another important limitation of many existing toxicity biomarkers cited by researchers is their lack of translatability. For example, in preclinical testing a drug is tested in healthy animals that are treated with a toxicant to induce a disease state, Falls said. But in a clinical setting the compound is being administered to patients who almost always have some sort of pre-existing condition that could impact their responses to treatment.

"That's a big gap we have to overcome," he said.

Amid a growing body of literature implicating them in a variety of biological conditions and disease states, miRNAs have become attractive biomarker candidates, in part because many of them are tissue specific and, therefore, can help pinpoint toxicity in specific organs.

Additionally, miRNAs have been found to be stable in biofluids such as serum and urine, offering the potential for monitoring their expression without the need for invasive procedures, Nassirpour noted. Further, they are conserved across species, which would help address translational concerns, and they can be easily detected with existing technologies.

Already there is evidence suggesting that miRNAs might be able to outperform traditional toxicity biomarkers, including work published in 2009 by a team at Merck that was outlined at the CHI meeting by Merck molecular biomarker and diagnostic scientist Xuemei Zhao.

In that paper, the liver-specific miRNA miR-122 was found overexpressed in the plasma of rats that had been treated with liver toxicants but not muscle toxicants. Similarly, plasma levels of miR-133a, which is muscle specific, were only increased in the animals only in response to muscle toxicants.

In contrast, the liver toxicity biomarkers alanine aminotransferase (ALT) and aspartate aminotransferase were elevated in animals with toxicity in either organ. Additionally, miR-122 proved to have a higher diagnostic sensitivity than ALT when correlated to liver histopathological results.

In light of such data, Nassirpour and her colleagues at Pfizer began testing whether miRNAs obtained in urine could be used to specifically detect kidney toxicity.

"Nephrotoxicity from drug-induced exposure is estimated to contribute a quarter of all cases of acute kidney injury," she said during a presentation at this week's meeting. "There is a really high unmet need for sensitive biomarkers … for detecting nephrotoxicity early on."

The Pfizer group administered the antibiotic gentamicin — a known nephrotoxicant — to rats at either 25 mg/kg or 50 mg/kg once a day for seven days. Urine was collected on the first and last days, and the expression patterns in the biofluid were evaluated using either qRT-PCR or next-generation sequencing (NGS).

Analysis with qRT-PCR revealed 178 miRNAs in the urine, 32 of which were significantly altered after gentamicin treatment. NGS revealed 146 miRNAs in the samples, 14 of which were differentially regulated in response to the antibiotic.

A total of 60 miRNAs were identified by both platforms, three of which — let-7d, miR-203, and miR-320 — were dysregulated in response to gentamicin.

Nassirpour noted that the low concordance between the platforms was a concern and probably due to a number of factors including NGS's apparent sensitivity to contaminants, which could affect the detectability of low-expressing miRNAs; limitations of the qRT-PCR primers used, which are annotated only to miRBase and don't include the isomiRs; and the possible differences in sensitivities inherent in NGS and qRT-PCR.

Overall, however, she said that the results of the experiments "build on the growing data indication that microRNAs may be useful biomarkers of toxicity," noting that all of the miRNAs detected by the two platforms were involved in pathways relevant to kidney toxicity and necrosis.

While the data presented by Zhao and Nassirpour centered on toxicity in the liver and kidneys, respectively, Falls' work focused on drug-induced injury in the heart, which he said has become a pre-eminent issue for the pharmaceutical industry in recent decades as highlighted by the high-profile withdrawals of approved drugs such as the pain relievers Vioxx and Darvocet.

When it comes to evaluating cardiotoxicity, the industry has largely relied on measuring troponin levels, as well as evaluating electrical disturbances such as QT prolongation, he said. But these do not indicate a problem before morphological or functional adverse events.

As such, he has been working to uncover changes in gene expression that can bridge the gap between the mechanisms and manifestations of cardiac toxicity, with miRNAs as a promising avenue of exploration.

Previously, investigators from Japan's National Cardiovascular Center correlated overexpression of the heart-specific miRNA miR-208 in plasma to troponin increases following drug-induced injury. And scientists at GlaxoSmithKline have linked increases in miR-21 and miR-24 to cardiac hypertrophy, he said.

In a more recent study, described at the CHI conference, Falls and his collaborators treated rats with the environmental toxin bis(2-chloroethoxy)methane, or CEM, to induce cardiac injury.

He described CEM as a "sledgehammer hitting the heart," boosting levels of troponin and inducing necrosis in the rats. The agent also increased levels of miR-208, which is encoded by an intron of the alpha-myosin heavy chain gene and is required for the upregulation of beta-myosin heavy chain that occurs in response to cardiac stress and injury.

Meanwhile, plasma levels of miR-133, which is implicated in cardiac hypertrophy, was increased in all CEM-treated rats and correlated with increased troponin in three of six animals.

"We believe this shows a proof of principle regarding heart-selective microRNA detection in circulation during cardiac necrosis," Falls said. However, it is not clear whether these miRNAs offer any advantages over existing biomarkers.

Falls said that GlaxoSmithKline is currently testing this in more clinically relevant animal models.