The Pharmaceutical Journal
Vol 268 No 7181 p68-69
19 January 2002

Joint Pharmaceutical Analysis Group/Royal Society of Chemistry

Speeding drug discovery and development

In preclinical lead optimisation technologies (PLOT) in pharmaceutical development, in vitro testing is increasingly being used for efficacy screening despite a preponderance of animal tests to meet current official safety requirements, said Professor CHRIS ATTERWILL (director of biosciences, Huntingdon Life Sciences).

Given a current attrition rate of candidate drug failures of at least 5,000:1, drug developers need to reduce costs, bring candidate substances faster to the market, employ automation and in vitro cell culture technology, recognise biomarkers of toxicity and make more use of theoretical computation (so-called in silico testing).

In vitro PLOT screening was much cheaper and quicker than animal models and used smaller quantities of compound. It could replace extensive animal testing for the basic ADMET regimes (absorption, distribution, metabolism, excretion and toxicity) before regulatory preclinical animal testing.

In new drug development, 70 per cent of early failure was due to pharmacokinetic problems and lack of drug efficacy. Particular problems were poor bioavailability, unexpected toxicity, inappropriate animal safety tests and drug/drug interactions. Each of these failure categories could be minimised using early lead optimisation strategies.

Overall, Professor Atterwill noted: significant progress with integrated ADMET in vitro screens at late discovery stage; greater use of high throughput screens requiring a minimal amount of compound; the major impact of cell culture; growing application of toxicogenomics and proteomics; and the predictive value of computational tools in pre-clinical testing. He hoped that regulatory agencies would begin to accept in vitro procedures to reduce the need for animal testing.

Candidate selection

Reviewing candidate selection in making the step from discovery to development, JEFF MOORE (Vertex Pharmaceuticals Europe) described an integrated approach to drug discovery involving high throughput screening, combinatorial and computational chemistry, molecular biology, enzymology, pharmacology and chemogenomics (a parallel drug design across gene families in which knowledge of the active site of one target was used to design compounds to inhibit related targets). Mr Moore gave the example of kinases, where patents had been filed covering over 100 distinct active drug scaffolds.

One analytical tool that had a unique impact was LC-MS/MS (liquid chromatography combined with tandem mass spectrometry), which was rapid, sensitive to picogram levels, selective for single analytes and required minimal sample pre-treatment. Such analysis enabled rapid method development, with robust and highly productive assays with wide applicability.

The candidate selection process included evidence of biochemical and pharmacological activity, assessment of in vivo specificity and duration of action, pharmacokinetic and tissue distribution studies, and relevant toxicity data. This was complemented by manufacturing and formulation issues, assessment of solubility in water across a physiologically relevant pH range, and accelerated stability studies in the solid state. Key questions in candidate selection were: is the molecule potent enough, safe enough, sufficiently bioavailable, easy enough to make economically and formulate reliably, economic in terms of market potential with an advantage over competition and protected by patent.

Separation sciences

Dr ROMAN SZUCS (Pfizer, Sandwich) said that in exploratory development separation sciences had a role as an underpinning strategy that optimised analyte-specific robust methodology, with minimal analytical validation. For discriminating the products of synthesis, he favoured variants of orthogonal chromatography, that is, the concomitant use of two distinct separatory phases.

He used double column gas chromatography (GC) with mass spectrometric (MS) detection for relatively volatile products; and for the less volatile he advocated four parallel high performance liquid chromatography (HPLC) columns, using a variety of detection options, such as diode array detectors (DAD), evaporative light scattering detectors and time-of-flight MS. The four columns could be identical, to maximise comparison and throughput, or different, to optimise discrimination of individual products. Correlation was improved by using tandem MS detectors. He had validated this analytical armoury by extensive trials, monitoring peak correlation and resolution.

In impurity analysis, he relied on chiral phase GC for volatile optical isomers and for involatile isomers he used sequentially either reverse or normal phase chiral HPLC systems that were fully automated and could be left running overnight. For ionic impurities, he suggested orthogonal capillary electrophoresis (CE) with ultraviolet (UV) and MS detectors, with the alternative of ion chromatography or capillary ion analysis (CIA), for which he prescribed a range of buffer systems and presented a list of counter ions that could be characterised. For heavy metals he advocated an inductively coupled plasma (ICP)-MS system. For residual solvent impurities, depending on the required application, he used either specific headspace "fast" GC analysis or generalised non-specific screening.

He concluded that the pharmaceutical analyst had to be prepared to respond to a vast array of questions from synthetic process and development colleagues.

Single-chip technology

Introducing a single-chip technology joint venture by Imperial College London and the University of Glasgow, Dr DARRIN DISLEY (Adaptive Screening Ltd, Cambridge) outlined his company's miniaturised molecular and cellular profiling systems and bio-informatics tools. These comprised specific and non-specific protein microarrays immobilised onto flat surfaces, plus cellular microarrays within three-dimensional bio-analytical chip devices. By enabling high-information parallel compound analysis regimes to be implemented on a single chip, they promised to be among the most important pharmaceutical research and development tools of the post-genomic era,

Dr Disley attributed the high failure rates in current screening to taking a single "selective" protein target out of context of its non-linear binding relationships with other proteins, and then identifying compounds that bound specifically to that protein target, without considering the pharmacological and toxicological effects of non-specific binding. His "adaptive arrays" mapped "chemical space", using recombinant protein array technology, low-light array imaging and Bayesian neural net technology to recognise the multidimensional binding relationships of compounds to a selected set of "specific" and "non-specific" protein models designed to be a surrogate of the human proteome. He demonstrated clear distinction in a comparison of five non-steroidal anti-inflammatory drugs with five terpene natural products and five nitroaromatic mutagens.

A secondary profiling "cellular array" platform, which integrated parallel microfluidics, optics and microelectronics on a chip, was the basis of a high information, high density, cellular profiling system that allowed direct monitoring of physiological processes in the cellular environment.

Finally, a new software environment combining a sophisticated relational database with scheduling and control systems would allow the company's technology platform to interface with existing experimental design processes as well as combinatorial synthesis, compound and cell culture automation.

Solid state characterisation

Solid state characterisation techniques were the key to determining a compound's physical form, said Dr RICHARD STOREY. He emphasised the need for early collaboration with drug discovery colleagues, prioritising support when necessary, in obtaining more information on the solid form and the morphology. He distinguished between polymorphs and solvates and discussed their transformations, itemising the parameters that influenced the solid form. A range of solid state analytical techniques was available for screening crystallinity, hygroscopicity, solvation and he contrasted the laborious manual, and conveniently robotic, forms of "sitting drop" method for screening salt formation.

Dr Storey stepped through the logic in screening an optimal solid form as a lead drug candidate, eliminating metastable forms. He provided three diverse examples of extending the role of conventional solid state analytical techniques, offering a bespoke combination service outside the normal remit. Previous results from one technique might be ambiguous (and often were), but complex issues could be resolved by a combination of techniques, including X-ray and dynamic vapour sorption; thermal analysis and MS, and isothermal microcalorimetry.

Chemical and biological characteristics

Professor HUBERTUS IRTH (Vrije University, Amsterdam) described analytical screening technologies for simultaneously determining chemical and biological characteristics of active compounds.

In traditional drug screening, biochemical and chemical information was typically not correlated. Screening and chemical analysis were performed on different instrumental set-ups, by different scientists, often in different laboratories. Bottlenecks often occurred. Co-ordinating these activities became particularly laborious and time-consuming when active compounds were present in complex mixtures such as biological samples, solution-phase combinatorial chemistry products or natural products.

Professor Irth's research had focused on screening technologies in which real-time measurement of biospecific interactions was an integral part of fully automated chemical analysis. A complex mixture was subjected to gradient HPLC separation and then a biological interface divided the flow. The bioactivity chromatogram was compared in real time with a library of active substances and correlated with the parallel chemical stream identified by diagnostic MS/MS data and DAD profiles. The bioactivity was measured by molecular recognition at binding sites, where the targets were macromolecular receptors, using low protein concentrations labelled with a fluorogenic ligand. It was possible to use a continuous flow system if the fluorogen label was replaced by MS detection.

Professor Irth recognised challenges in the need for a fast assay (1–2 minutes) and conditions that were compatible for HPLC and bioassay. The main application areas he addressed involved on-line screening of focused combinatorial libraries, active metabolite and natural product screening, and evaluation of orphan receptors. His examples included an optimised model assay for a natural enzyme and miniaturised on-line screening of a protein mixture. He emphasised that novel biological systems needed integration of bioscreens with the chemical analysis and that high throughput technologies must harness bioinformatics and advanced analytical methodology.

Accelerating drug development

Speaking on accelerating drug development using chip-based technologies, Dr COULTON LEGGE (GlaxoSmithKline, Harlow) said that microsystems were moving away from microwell plates to microfluidic channels". The new regime was a "lab-on-a-chip" system, combining high throughput chemistry with capillaries for sample introduction, preparation, processing and analysis.

Contemporary microsystems reduced capacity, time, consumption, operator skill and cost, and increased throughput and flexibility. Pitfalls were that poor mixing resulted from diffusion-limited laminar flow, and small volumes reduced detection levels and caused susceptibility to debris and bubbles.

Dr Legge described the development of an analyser for DNA sizing, running 12 samples in series in a microchip capable of analysing complete DNA from a 1µl sample in 30 minutes. He commented that conventional miniaturisation of microplates brought new problems in liquid handling, evaporation and detection, so that new enclosed formats were needed. He described fabrication of LC microcolumns with electrokinetic pumps for high throughput analysis, facilitated by diode array, fluorescence, electrochemical and refractive index detection.

For searching the genetic code, DNA chips were prepared by photolithography, with integral fluorescence detectors: these could reveal a single mismatch of two nucleotides. In an interesting technique of imprinted microchannels in compact discs, the flow rate depended on rotation speed, and rapid analysis using time-of-flight MS.

Physicochemical properties

Describing the measurement and use of physicochemical properties for drug discovery, Dr CHRIS BEVAN (GlaxoSmithKline, Stevenage) said that in high throughput screening, the bottleneck had become in vivo testing and selection of drug candidates, while industry was under pressure to become more efficient — to minimise costs, reduce headcount, and compress R&D timelines from 14 to 10 years or less. Quality information was needed earlier, but traditional methods might no longer be adequate.

His company had funded a world-wide project to deliver high throughput screens for physicochemical, pharmacokinetic, metabolic, and toxicological factors. The screening benefited discovery projects through a large international repository of consistent data that could reveal more about mechanisms regulating kinetics and toxicology. The data could also be used to build predictive models, which aided further drug design.

Dr Bevan discussed assessment of lipophilicity using an arbitrary but standardised chromatographic hydrophobicity index; and measurement of solubility by equilibration of a solid sample with a buffer solution which, after filtration, was measured by gradient HPLC. This work complemented ADMET bio-studies involving in vitro metabolism, permeation cell culture, in vivo pharmacokinetics and a general mutagenicity assay. For these various purposes, his company worldwide employed a comprehensive toolkit of high throughput methods, employing standardised general procedures suitable for large compound sets. Solubility could be measured by laser nephelometry, using 96-well microtitre plates with stepwise serial dilutions until the compound redissolved.

This process was reproducible for test compounds, with results similar to those produced by HPLC. Rapid measurement of pKa could be attained by gradient titration, using a prototype instrument developed in collaboration with Sirius Instruments, and for which a patent had been filed and published. The novelty accrued from continuous flow titration and replacement of pH measurement by measurement of time.

Data mountain

Dr HUW LOARING (Inna Phase Ltd) examined means of collating and integrating the "data mountain" of pharmacokinetic information and analytical measurement; and the need for a data repository to demonstrate credibility in good laboratory practice. Computerised systems were essential to consolidate, process and archive scientific data, and data handling software was crucial. A convenient laboratory information management system (LIMS) captured, processed, displayed and calculated raw data and concomitantly undertook sample tracking, data storage and calibration, and provided appropriate structures. Although a "generic" LIMS might be satisfactory for general analytical work, it was not suitable for bioanalysis.

Data handling systems had four aims: to recognise assay standards, to operate a manipulation algorithm at the data interface while maintaining data integrity unaltered in transfer, to provide for regression evaluation with quality control outliers, and to operate a reliable decision tree indicating whether to reassay. Ideal software made provision for method validation, stability analysis and pharmacokinetic studies; it even provided a uniform research report format. Dr Loaring claimed that his company produced a LIMS ready for a bioanalytical laboratory without needing customisation. His vision was a total business intelligence framework based on a data "warehouse" that embraced all information arising from the LIMS, ADMET studies, pharmacokinetic and toxicity investigations, and even third party work.