The cost of drug development has surpassed one billion dollars per drug, and candidate attrition rates continue to rise, driven in part by toxicities that reduce therapeutic windows. Current methods used to identify activities at the biochemical and/or cellular level, which would be predictive of downstream human toxicity, are largely insufficient.
This conclusion is supported by the large number of clinical failures due to hepatotoxicity that are later confirmed by retrospective studies looking at the predictability of existing in vitro toxicology assays.
Failure to adequately predict compound toxicity leaves the discovery of these undesirable activities in the hands of the clinic, which is the final (and most expensive) filter in the process. In the worst cases, adverse pharmacology is discovered in the patient population, and a drug is pulled from the market—a costly event that erodes the perception that the pharmaceutical industry can deliver safe, efficacious drugs.
What are the root causes for inadequate profiling of early drug candidates? Certainly capacity to complete the right assay in the right timeframe plays a key role. As more pharmacologically relevant assays are brought into the process, shortages of critical reagents like primary cells become acute, limiting the number of assays that can be done.
Likewise, more complex assay formats like phenotypic endpoints are often carried out at low throughput by highly skilled experts. Combining these factors with increased competitive pressure has resulted in reliance on low-biological content approaches to compound risk assessment during the early drug discovery process.
Here, iterative rounds of chemical modification/biological evaluation are performed, looking for increased potency, specificity, and bioavailability with reduced toxicity. During this lengthy stage, termed the structure activity relationship (SAR), the lack of timely, highly predictive toxicity assessment can lead to a silent risk buildup, manifesting itself far downstream in human studies.
Therefore, in order to improve the clinical success rate of the drug discovery process, there is a critical need for higher capacity, higher biological content assays that can be implemented during the SAR.
High-Content Screening Approach
Over the past decade, the development of automated imaging as a means of measuring multiple phenotypic events inside intact cells has led to a new class of cell- based assays that have several key advantages over current methods. High-content screening (HCS) generates multiple endpoints of size, shape, texture, and intensity. These assays measure individual cells in each well, thereby reducing losses in data integrity due to population averaging.
Because each cell is measured independently, subtle changes in physiology can be detected and robustly quantified. By combining multiple, spectrally distinct targets fluorescently tagged in a variety of ways (dyes, GFP) in each cell, a highly correlated picture of cell response emerges.
To date, the Thermo Scientific Cellomics HCS platform has been used in several key studies of predictive toxicology—measuring subtle changes in organelle integrity (size, shape, membrane potential) that has been empirically linked to downstream effects of the liver in vivo.
In addition, mitochondrial potential, lysosome mass, and multiparameter assessment including nuclear morphology, membrane permeability, and mitochondrial function are examples of successful application of HCS to cytotoxicity endpoints.
Furthermore, HCS has been shown to increase the ability to predict human hepatoxicity in several retrospective studies on hundreds of compounds with known pharmacologies. One study reported increases in sensitivity (93% vs. <25%) and specificity (98% vs. ~90%) when using a single, multiplexed HCS assay compared to the best combination of seven biochemical assays that are routinely used in the industry.
A different study reported applying a phenotypic panel of HCS assays to over 300 drugs and chemicals (including rare and idiosyncratic ones) known to cause liver damage, returning an impressive true positive rate of 50–60% with <5% false positive rate.
HCS Tools for Predictive Toxicology
To answer the need for better predictive toxicology, the Cellomics HCS Platform offers a platform of instrumentation, validated reagent kits, image analysis algorithms, and flexible data analysis tools. The platform is designed to address key toxicological endpoints in an efficient and robust manner.
For example, Thermo Scientific offers over 18 biologically validated HCS reagent kits for cell health and general cytotoxicity, several of which simultaneously investigate multiple indicators of cytotoxicity to get an accurate and complete picture of what is occurring to each cell in each well during treatment (Figure 1).
These kits examine key organelle health markers such as nuclear shape and size, cell membrane permeability, mitochondrial transmembrane potential, p53 and p21 activation, caspase activation, cytochrome C release, and oxidative stress. After using a Cellomics HCS reagent kit, the Cellomics BioApplication image-analysis algorithms provide hundreds of cytotoxicity-specific cell-level and well-level measurements to automatically assess samples for those indicators labeled by the kits.
For instance, cell health profiling is a bioapplication that is used in conjunction with a number of the reagent kits to evaluate 2-to 6-channel assays looking at multiple markers of cell health. Assay development and optimization with the BioApplication is straightforward, as the cell images and data produced by the algorithm can be examined with only two mouse clicks in the software’s convenient Protocol Interactive window (Figure 2).
Interacting with the images and seeing the resulting field’s data before a scan is completed allows for any necessary tuning of the algorithm before one commits to an entire plate scan, without the need to be an image analysis expert. For optimal decision-making power the BioApplication Event Wizard allows the user to combine multiple outputs to further subcategorize cells in a cytotoxicity assay (Figure 3).
User-defined criteria within the wizard classify populations at both the cell and well levels. By combining different algorithmic outputs in a cytotoxicity assay, the wizard can label cells in early/reversible toxicity if the cell has a decreased mitochondrial transmembrane potential but no cell membrane permeability, versus late/irreversible toxicity when the nuclei have fully condensed and the cell membrane has been impaired.
This rule-based approach allows rapid ranking of compounds based on multiple outputs, providing rich information to guide SAR and toxicology groups.
For those performing genotoxicity studies, Thermo Scientific offers 15 biologically validated reagents kits, again with an accompaniment of genotox-associated BioApplication image analyses. Whether investigating micronucleus formation, the MDM2 and p53 pathway, ATM, or Ku70/80, the kits cover a broad range of genotoxicity and DNA repair applications.
The Cellomics Micronucleus BioApplication was made specifically to identify genotoxicity in the form of micronuclei formation and was the first of its kind in the image-analysis industry. The algorithm has replaced many traditional in vitro toxicity assays as it is automated, more sensitive, scalable, and more objective than former methods.
The Thermo Scientific offering of Cellomics products for toxicology is primarily aimed at predictive toxicology, so that the compounds that make it through the preclinical testing phases have a greater chance of safely helping patients. This goal should benefit both the general public and pharmaceutical companies.
A recent industry report on early toxicology states, “With better predictions come increased efficiency, lower development costs, and drugs that reach the market faster. This would have a major effect on how drug discovery is perceived, both inside and outside the industry. This improvement would be a major milestone, adding credibility to the changes that have been taking place in drug discovery approaches.” Thermo Scientific considers HCS a tool for improving clinical outcomes and is continually working to develop and improve such tools.