Dr. Kasturi Mahadik
Senior Research Associate
Centre for Predictive Human Model Systems (CPHMS), AIC-CCMB

Have you wondered about the journey of a routine tablet like aspirin before it reaches your cabinet? Did you know scientists spend over a decade (if not more!) testing these pint-sized pills?

Newbies on the block

We have been saying this for years and yet, very little has changed, 9 out of 10 drug candidates fail to make it from the laboratory to market. Most often, candidates that are deemed safe in animals do not perform similarly in humans. It is becoming increasingly evident that animals fail to emulate humans. What,then, is the alternative?

One of today’s most promising preclinical research models are palm-sized chips that resemble human organs through their 3D human cell culture and fluidics (broadly categorized as microphysiological systems or MPS). However, the question remains, are these Organ-Chips scalable for use in the drug discovery pipeline? Have they undergone rigorous assessments? Are they reproducible?

Emulate, a leader in the world of Organ-Chips decided to take this challenge and carry out the largest-ever testing of Organ-Chips (870 chips in total) in their recent study steered by Dr. Lorna Ewart. While many drugs have desired effects on target organs, they undergo attrition due to their toxicity i.e. their ability to cause liver damage (popularly referred to as Drug Induced Liver Injury or DILI). Hence, Emulate decided to test its Liver-Chips with drugs that have documented clinical outcomes and measure their predictive power as preclinical tools of tomorrow. For a critical, yet unbiased and blinded study of drug behaviour, they partnered with an independent third party, the Innovation and Quality (IQ) consortium.

The IQ consortium

The IQ consortium comprises a collaboration of pharmaceutical and biotechnology companies that aim to advance science and technology to enhance drug discovery programs.Recognising the promise of MPS for drug development, an IQ MPS affiliate was formed in 2018 and in due course, it recommended a series of criteria that a new MPS model must meet. Their ultimate aim is to aid MPS regulatory acceptance and promote their broad industrial adoption. Ewart et al. is a first-of-its-kind study that carried out Organ-Chip performance validation in accordance with the IQ MPS-listed structural and performance criteria. Further, IQ MPS suggested Emulate 27 hepatoxic and nontoxic drugs for blinded testing (neither clinical impact, name nor concentration of the drug was known to those working with the Liver-Chips).

Design of the Liver-Chip

Each Liver-Chip consisted of two parallel channels separated by a porous membrane. The upper channel was lined by hepatocytes (cells used across the 870 chips spanned three different human donors) while the lower was lined by endothelial cells, stellate cells and Kupffer cells. Media was passed through each of the channels to mimic blood perfusion.

Assessing drug toxicity

Cell viability in the Liver-Chip was assessed by measuring debris, cellular morphology injury score and staining for markers (apoptotic cell death and mitochondrial injury) on days 1, 3 and 7 after drug dosing. The top channel effluents were also collected for albumin and alanine transaminase (ALT) analyses. Inhibition of albumin production and increase in ALT were used as clinical measures of liver damage. Typically, when albumin production was inhibited, morphological injury scores and ALT levels also increased. The authors found that a decrease in albumin production was the most sensitive marker of hepatocyte toxicity in the Liver-Chip.

Using the above analyses, 7 out of 14 drugs were correctly identified as toxic. In addition to the seven matched pairs, the IQ MPS guidelines require that an effective human MPS DILI model predict liver responses to 6 additional drugs associated with clinical DILI (Ewart et al., only had access to 5 of these), which Liver-Chip identified correctly.

Lastly, the study expanded the Liver-Chip assessment to include 8 other drugs whose toxicities were poorly predicted by hepatic spheroids (a widely used preclinical model at present). Overall, Liver-Chip correctly predicted toxicity in 12 out of 15 toxic drugs, that is, it demonstrated 80% sensitivity. This was double the sensitivity of hepatic spheroids for the same set of drugs. Further, the chips did not wrongly classify any drug as toxic (corresponding to 100% specificity, while spheroids displayed a mere 67% specificity). Specificity is an important parameter as false positives can greatly limit the usefulness of a model by failing safe and effective compounds.

Importantly, each of the toxic drugs tested in this study was evaluated using animal models in the past. And in each case, these drugs were deemed safe and efficacious to progress into clinical trials. Therefore, the ability of Liver-Chips to flag 80% of these drugs for their toxicity risk at the preclinical stage represents a giant leap for mankind.

Economic value model

The team also built a model to assess the economic impact of incorporating LiverChip into pre-clinical research. The model revealed that supplementing existing preclinical models with Liver-Chips for the prediction of DILI had the potential to generate an estimated $3 billion annually due to improved research and development productivity. In addition to liver toxicity, the other top causes of drug safety failures are cardiovascular, neurological, immunological and gastrointestinal toxicities. If four additional Organ-Chips could be set up to model these common toxicities and their performance could match that of the Liver-Chip, the productivity gain could potentially extend to an estimated $24 billion annually!

Taken together, today there is an urgent need to evaluate Organ-Chips across laboratories and sectors for reproducibility. Should these studies be consistent, Organ-Chip technology could herald a new era in drug development, where more drugs could enter the market at remarkably lower costs at never before-seen record times.