The failure of hitherto promising drug candidates when they go into animal tests and clinical trials occurs far too often for the pharmaceutical industry's comfort. It represents costly failures for the companies involved and pushes up the overall cost of drug development.

Improving pipeline productivity is “clearly the biggest challenge”, according to Kevin Hrusovsky, president and chief executive of microfluidics specialists Caliper Life Sciences in Hopkinton, Massachusetts.

Others agree. “The attrition rate of candidate drugs in later clinical development is still way too high,” says Gary Franklin, industrial sector specialist in marketing communications at molecular-interactions company Biacore, based in Uppsala, Sweden, and recently acquired by GE Healthcare.

Kevin Hrusovsky: getting more drugs through the late stages of development is the big challenge. Credit: CALIPER LIFE SCIENCES

As well as interacting with its molecular target, a successful drug must also behave properly in the human body. It must be easily absorbed, not be broken down too quickly, and have no side effects that make it too toxic to use. Known as ADMET or ADMETox (for absorption, distribution, metabolism, excretion and toxicity), these tests are key to efficient drug discovery. ADMETox testing is initially carried out in cell-based and in vitro assays to determine properties such as a compound's solubility, its ability to cross cell membranes, and its cellular toxicity. But these tests are still far from perfect predictors, and improving them is a priority area in drug research. Undesirable ADMETox properties account for some 50–60% of drugs that fail at the preclinical stage.

Franklin believes that the biggest challenge is to obtain more comprehensive data much earlier in the drug-development process. He points to the Critical Path Initiative, published in 2005 by the US Food and Drug Administration, which aims to stimulate a national effort to develop improved evaluation methods that provide better information and maximize the chances of clinical success.

To address this pressing need, researchers will need faster and more precise instruments that make measurements with greater information content than ever before. This is a big demand, but there are some new tools of the trade that may make a difference.

Automating ADMETox

Equipment companies have their sights set on improving ADMETox testing, including the tests that all new drugs must undergo for their possible adverse effects on the cardiac potassium channel (see 'Ion channels get automated').

Another routine ADMETox test predicts cellular uptake and efflux of a compound. The absorption of a compound across the epithelial lining of the human gut is conventionally predicted by assaying its transport into and out of Caco-2 cells. Caco-2 is a human colonic adenocarcinoma cell line that will differentiate in culture to express characteristic features of mature intestinal cells. The cells must be grown for 21 days in strictly controlled culture conditions to form a monolayer of differentiated cells before the test can be done. This tedious preparation process can be automated using a liquid-handling workstation such as the Biomek 3000, from Beckman Coulter of Fullerton, California, to perform the culture manipulations. “We have demonstrated intact and functional Caco-2 monolayers after 21 days of culture,” says Keith Roby, product manager for strategic marketing automation at Beckman Coulter. “The workstation can also automate the compound permeability and efflux testing and sample collection.”

The new Biomek FXP automated platform from Beckman Coulter can be used as a complete ADMETox workstation, and offers a faster in vitro assay for drug absorption than the Caco-2 assay. The new assay is based on parallel artificial membrane permeability analysis (PAMPA) and µSOL technologies for measurement of in vitro drug permeability and solubility, respectively, developed by pION of Woburn, Massachusetts.

In PAMPA, passive absorption through an artificial phospholipid membrane is intended to mimic the movement of a compound across the gut lining. “The PAMPA method provides significant time savings compared with the Caco-2 cell-based method, which takes more than three weeks to complete,” says Roby. “With PAMPA, data can be collected in less than two hours,” he says. All aspects of the permeability assay, including all liquid-handling steps and analysis, are controlled by PAMPA Evolution 96 software.

Model virus: a potential influenza virus inhibitor (purple) modelled with Tripos's SYBYL software and redocked to the viral neuraminidase (yellow). Credit: TRIPOS

The Biomek FXP also supports metabolism assays, such as cytochrome P450 inhibition studies and microsomal stability assays. The toxicology assays it supports include ones based on cell proliferation, cytotoxicity and apoptosis.

The small-footprint Biomek NX Assay Workstation can also be used to automate cell-based screens for apoptosis. The system will schedule a six-hour method to process ten assay plates from the initial induction of the apoptosis pathway to detection of activity of the effector enzyme caspase 3/7, without the intervention of an operator.

Mark Collins, associate director of strategic marketing at Thermo Fisher Scientific, based in Waltham, Massachusetts, also considers late-stage failure of drug candidates a big problem for the pharmaceutical industry. “The attrition rate is one of the biggest issues; too many compounds fail at preclinical or phase I clinical tests,” he says.

Collins thinks that high-content screening (HCS) could be one of the answers. “ADMETox testing is starting to use HCS for automated genotoxicity assays as well as for predictive assays to rank compounds for animal testing. This research may pave the way for HCS assays to be a better predictor of certain types of toxicity than animals,” he says.

Look, no cells: Beckman Coulter's Biomek FX ADMETox Workstation can be used with pION's PAMPA technology for permeability assays using synthetic phospholipid membranes.

On the software front, both Inpharmatica of London, UK, (soon to be part of Belgian drug-discovery company Galapagos) and ChemSilico of Tewksbury, Massachusetts, have software tools to predict ADME properties.

Intelligent cell screening

“The continued growth of cell-based assays is driving a need for automated systems to prepare and run these assays,” says Marc Feiglin, chief technology officer in life-science research at Tecan in Männedorf, Switzerland. “Automating this in a cost-effective manner requires a single system that can pipette both large volumes of media and small volumes of compounds.”

The Freedom EVO platform from Tecan offers versatile robotic liquid-handling. It is available in four sizes, with a variety of robotic arms and other liquid-handling accessories. According to Feiglin, it offers the widest dynamic range of pipetting volumes currently available on one platform — from less than 1 µl to hundreds of millilitres.

The Cellomics family of automated imaging instruments from Thermo Fisher Scientific are designed for HCS and high-content analysis. They enable researchers to analyse the responses of cells in vitro to stimulation by small molecules or the knockdown of gene expression by RNA interference (RNAi). “HCS provides temporal and spatial information about such cellular responses as well as phenotypic profiles,” says Collins. The core of a typical Cellomics HCS system is an imager such as their ArrayScan VTI. This is packaged with quantitative image-analysis software modules that address the experimental area of interest and data-management software, necessary because of the huge quantities of imaging data that HCS generates.

A feature of the Cellomics HCS system is intelligentAcQuisition (iQ), which allows images to be analysed on-the-fly in real time, and measurements from the analysis fed back to allow the image-acquisition process to adapt to the biology being measured. “IQ provides significant productivity gains compared with HCS systems that acquire images and then do the analysis offline, as it performs in parallel all the steps of the HCS process during the time the plate is being scanned,” says Collins. He likens Cellomics iQ to a digital camera, which allows you to see immediate results. Without it you're more like someone taking pictures with a film camera: “You don't know how good the picture is until you have the film developed, but by then it might be too late, or you have too many pictures of the same thing,” he says.

Picking up the signal

One application of Cellomics HCS systems is to study intracellular signalling pathways in oncology, as they enable a pathway to be 'walked' using multiplexing.

Another company pursuing ways to streamline signalling research is DiscoveRx of Fremont, California, which received a 2006 Frost & Sullivan award for technology innovation for its PathHunter assay platform, which is used to study signalling pathways in intact cells. A recent addition to the range is a β-arrestin assay to study the activation of G-protein-coupled receptors (GPCRs), a common drug target.

An attractive feature of PathHunter assays is that they are relatively simple: they are adapted to microtitre plates, involve only one or two additions of reagents, and do not require cell washing or fixation. The assays are compatible with many different cell lines, especially those most commonly used in high-throughput screening, such as CHO and HEK cells.

PathHunter assays detect receptor activation by a technique called enzyme fragment complementation. Two modified β-galacto-side fragments, fused to the receptor and the β-arrestin molecule, respectively, are brought together when β-arrestin binds to the activated receptor and form an active enzyme that generates a chemiluminescent signal.

DiscoveRx claims that PathHunter is the first chemiluminescence assay technology that can measure protein trafficking directly inside the cell — as such, it is especially suited to evaluating how cellular physiology is perturbed by drug candidates. It can be used to study translocation, degradation, secretion, protein–protein interactions and membrane trafficking. The chemiluminescent signal generated does not need dedicated imaging technology and can be read in 96-, 384- or 1,536-well microplates with a standard luminometer.

Tecan's new Infinite series of microplate readers are compatible with its Freedom EVO liquid-handling workstation and are available as either filter- or monochrometer-based systems. According to Feiglin, the sensitivity of the luminescence readers makes them a popular choice for assays of GPCR activity. Using the 1,536-well optics options on the Infinite F500 filter-based microplate reader, a 1,536-well plate can be read in less than 30 seconds. Also new from Tecan is the HydroFlex automated microplate washing and vacuum filtration system, which can be integrated into the Freedom EVO platform.

Tecan's HydroFlex platewasher can be integrated into the Freedom EVO system. Credit: TECAN

CyBio of Jena, Germany, made its reputation as one of the first companies to offer multipipettors. It recently unveiled a new robotic liquid-handling platform, CyBi-RoboSpense, which will work with almost any type of sample tube or plate, and can even transport plates to other instruments such as other pipettors, microplate readers or liquid chromatography or mass spectroscopy equipment. The CyBi-RoboSpense can also be used for complicated sample preparation and purification routines such as tryptic digests and magnetic bead separation.

Microfluidics applications

More companies are incorporating the principles of microfluidics into their instruments, making the movement of microlitre — or even nanolitre — volumes of solutions do what used to take hundreds of millilitres, not to mention a saving in time and labour. An innovator in this area is Gyros of Uppsala, Sweden, whose Gyrolab Bioaffy microlabs for protein purification and quantification are based on affinity-capture techniques. The Gyrolab Bioaffy 200 is ideal for high-sensitivity assays to detect low concentrations of proteins. It can be used, for example, to quantify protein markers of disease present in the nanograms per millilitre range. The Gyrolab Bioaffy 20HC is designed for bio-pharmaceutical applications such as process optimization and the quantification of monoclonal antibody drugs intended for therapeutic use, especially for measurements on the milligrams per millilitre scale, which means that samples can be run without dilution.

Micromanagement: a computer rendition of the surface of a Caliper LabChip shows the circular wells through which the channels are loaded.

The CD-sized microlabs run on Gyrolab workstations; the Gyrolab Workstation LIF can run up to five CDs per batch, and includes a laser-induced fluorescence (LIF) detector. “What made this system particularly interesting to the customer,” says Christina Burtsoff Asp, vice-president of marketing at Gyros, “was that it provided a measurement range from micrograms to grams per litre, which allowed dilution-free analysis in concentration ranges relevant for the entire purification process, from cell supernatant to purified product — all on a single system.”

Another new product using microfluidic technology is Caliper's Desktop Profiler, a bench-top kinase-profiling system that is pre-loaded with microplates with protein kinases, substrates and reagents. “Kinase profiling enables the identification of potential side effects in therapeutic candidates, thus helping researchers to better understand how a compound will react in the human body,” explains Hrusovsky. The Desktop Profiler uses Caliper's LabChip microfluidics-based screening technology, and is designed for researchers seeking to qualify drug candidates — that is, to select the more promising drug candidates from those that might be too toxic or ineffectual.

Structure from sequence: a homology model (white) can match a crystal structure (orange) remarkably well. Credit: TRIPOS

Fluidigm in San Francisco, California, recently introduced a new digital array nanofluidic chip for a type of quantitative PCR that the company calls digital PCR, which runs on its BioMark system. Digital PCR is based on limiting-dilution analysis, and estimates the absolute number of copies of, for example, a target sequence in a sample by successively diluting the sample and partitioning it into a large number of chambers, such that each chamber contains either one or no target molecules.

Nanofluidics makes this application of limiting-dilution analysis practicable, in that after the sample is loaded, it is automatically distributed into about 1,000 nanolitre-volume reaction chambers. Amplification of the target sequence by PCR in chambers that contain a molecule can be detected and the number of original molecules in the sample calculated.

The new technology is already proving useful: one group of researchers has used it to correlate specific copies of genes with the regulatory states of single stem cells; another study used digital arrays to identify individual bacterial cells carrying marker genes. This technology promises to compete with the microwell plates currently used for high-throughput quantitative PCR. Digital arrays offer higher throughput, use smaller amounts of reaction solutions, and avoid the thousands of pipetting steps that would be needed for a similar experiment in microwell plates.

Going in vivo

However sophisticated cell-based and in vitro tests become, promising candidate drugs have eventually to be tested in living organisms. Hrusovsky points to the problem of the poor correlation of in vitro experiments with what happens in vivo. “Pipeline productivity can be improved by preventing the drug attrition that happens when a candidate compound moves from in vitro to in vivo studies,” he says.

Caliper's IVIS Spectrum in vivo optical-imaging system uses tomography to give three-dimensional images of fluorescent and bioluminescent reporters in whole organisms. It is designed to help researchers “better visualize, track and understand how a therapeutic compound reacts, at a molecular level, in vivo, thus bridging in vitro and in vivo experimentation”, says Hrusovsky. Spectral unmixing capabilities allow multiple fluorescent reporters to be used. Illumination can be switched from above the specimen to below it, enabling the tomographic reconstruction of, for example, both shallow and deep tumours.

Getting more selective

Several general trends may speed up the drug-discovery process and reduce the rate of attrition of drug candidates. Franklin notes a shift towards structure-based techniques for drug discovery and the identification of new druggable targets (see 'In silico screening with chemical informatics'), using dramatically smaller compound libraries compared with high-throughput screening. “This has the potential to significantly shorten drug-discovery timelines and, hopefully, to increase overall productivity in the industry,” he says. “The trend towards fragment-based screening in particular shows great promise.”

“More rational screening — that is, for higher information content — of smaller libraries would help,” agrees Collins. “Too many targets from molecular biology and genomics have little or no clinical relevance; high-content screening and other technologies such as RNAi can help here.” Collins also sees pressures on the drug-discovery industry to produce 'personalized medicine', which would provide a drug that best fits the individual patient. Traditionally, the drug industry “has sought the one-size-fits-all blockbuster, so to respond to the idea of personalized medicine, technological, economic, and ethical changes will be needed”, he says.

“Some believe that the industry is poised to see the benefits of the investments made in new technologies over the past few years, and that state-of-the-art informatics is finally capable of supporting the complexity of life-science discovery,” says Kathleen Mensler, vice-president of marketing and corporate development at modelling software company Tripos in St Louis, Missouri. “With advances in understanding biological pathways and the unprecedented amount of valuable data available, breakthroughs are inevitable. The question is when.”