big pharma learns how to love information technology
The pharmaceutical industry has tended to be sceptical about the value of large-scale spending on computing. But in November this year, 250 or more senior executives from many of the sector's largest companies will be travelling to the World Pharma IT Congress in London to find out what their peers - heads of IT, knowledge management, informatics, e-business (or what you will) - are up to. Louisa Carson, conference organiser for Oxford International, said interest in the meeting is strong. 'Pharma needs to catch up with developments in information technology. It has been lagging behind.'
Why the change? The influx of data from genome research means major changes in the computational input into drug R&D. But one of the conference speakers, Dr Richard Scott, head of discovery informatics at drug design company DeNovo Pharmaceuticals, fingers the main reason for the burst of interest - money. The big management consultancies and others all agree that: '[Large Pharma] organisations aren't producing NCEs (new chemical entities) at a fast enough rate to sustain their business,' he said.
Pharmaceutical companies are facing a sharp squeeze on profits, despite the seemingly inexorable rise in healthcare spending. Patents are close to expiring on many top-selling drugs, and fewer new products are making it to the market. With hungry generics companies quick to launch a copycat version of a blockbuster drug as soon as its patent expires, profits from a single drug can collapse with alarming speed.
Although healthcare spending continues to rise, governments (and health insurance firms) around the world are keen to keep the lid on drug costs. Even in the US, the industry's most receptive and lucrative market, individual State governments are putting pressure on manufacturers to offer substantial discounts on drugs for publicly-funded medical provision for the poor.
According to the US manufacturers' organisation PhRMA, the average cost of developing a new drug has soared from $138m in 1975 to $802m today. The cost, which was revealed in a study carried out by Tufts University, is so high because of the number of failures early in the process: just one out of 5,000 screened drug candidates makes it to the market as a new medicine. Even that doesn't guarantee a profit, since only three out of every 10 marketed drugs produce revenues that match or exceed development costs. The Tufts study estimated that it takes 10 to 15 years to take a new drug from the laboratory to regulatory approval by the US Food and Drug Administration.
The success of the human genome project has added to the problem. In part, this is because companies are now tackling more complex diseases using more sophisticated scientific tools. Regulatory authorities are enforcing tighter safety controls too. Drug company managers have reasoned that, if they can reduce time to market by even a small amount, they can cut costs and at the same time extend the profitable period when the drug is both on sale and under patent protection.
To help them, companies have turned increasingly to information management solutions. But many of the issues under discussion in November will be about people rather than technology. Ivan Walrath, global head of records management for Pfizer, said that the conference will stress the need for good communications between information technology staff and their scientific colleagues. '[They] need to be able to speak the same language on a project, so that IT can understand the business need and so that the business line can understand the ramifications and opportunities of the technology,' said Walrath.
Dr James Fickett, global director of bioinformatics for AstraZeneca - who will talk at the conference about 'fulfilling the potential of bioinformatics' - agreed. Although bioinformatics has been around for a while, it's new to much of the industry. 'Managers haven't really figured out how to deal with it yet,' he said. 'Unlike with chemistry, they don't really know what questions to ask.' In the past, conversations between bio-informaticians and managers have tended to be about technology. 'They should be about what the key information is that needs to be managed, who uses it, and how it gets delivered, not about Oracle databases or C++ or Java programming,' said Fickett.
The experience of chemists and biologists points the way to a resolution. 'Pharma companies have had to mix chemistry and biology for a long time. There's a lot about chemistry that biologists don't understand, but a former biologist that heads a discovery unit learns to ask the right questions, such as tell me about the results,' he said. For bioinformatics, the right questions to ask are about the flow of information.
Fickett mixes informal and formal approaches to improve communication. At AstraZeneca, there is a steering group for bioinformatics, with group leaders from all around the company. At their last meeting, 'I had the steering group meet at the same time as the molecular biologists' so that they could meet and chat with scientists from another discipline. And each time the molecular biologists hold their quarterly meeting, Fickett picks one bioinformatics tool and gives them a half-hour talk on what it does and who uses it. 'It's a gradual education process,' he said.
The tools Fickett talks about are both 'off-the-shelf' and internally developed. 'I would prefer to buy, any time I can. To be honest I don't really understand why you can't buy more,' he complained. 'As a sector, bioinformatics is still maturing; as a community it's pretty much there; but as a business, it's not there yet.' The problem is that there were unrealistic expectations among some of the small bioinformatics companies founded in the 1990s. 'People haven't really figured out what model, for what software, to produce what services, to sell.'
Sometimes, Fickett finds that he needs to tame the unrealistic expectations users have of the technology. 'People are keen to get into new technology - systems' biology or pathway modelling. It's important to know what's realistic and what's not'. For example, he said: 'You can't at the moment build a good quantitative model of any interesting physiological process.'
Another great unachievable goal is seamless data integration - 'It's not going to happen anytime soon,' said Fickett. 'There are database integration tools out there, but they don't do what you want them to do. And at the moment it's much more realistic to take two or three data types that you need to put together to answer a question today, and build a system to integrate them on a very limited scale.'
Fickett focuses on how people work, what tools they are using, and to how to improve them. 'We built a central repository for gene expression data. It was worth doing but it doesn't get a whole lot of use, because most people are only really interested in their own expression data. On the other hand, we've got half a dozen groups around the company sticking data from clinical trials together with expression data in Excel spreadsheets. Because of the emphasis on biomarkers that's common throughout the industry right now, people need to relate the clinical to the genomic. They're already doing it, and if we make it easier for them they're going to use it.'
DeNovo's Scott believes that: 'In order to get to market faster and get what limited time you have on patent, you've got to do something smarter with the information you generate.' But not everyone seems to be taking the right road, in his view. 'A lot of people [in big pharma companies] have industrialised target-selection with gene arrays and gene chips. All that this has done is generate more data. That's not the right way to do business.'
For Scott, being smarter means better integration of existing data and, at the conference, he plans to present a case study in informatics integration, using data flow and web services based on the company's newly completed cheminformatics platform SkelGen II. The technology, which was developed in collaboration with the Swiss pharma giant Roche, builds upon DeNovo's original SkelGen, which generated novel lead molecules for development as potential drugs. SkelGen II extends the ability to generate new molecules by incorporating synthetic chemistry know-how. It provides pharmaceutical chemists with a means of rapidly generating multiple novel chemical entities within protein target sites, while at the same time giving synthetic reaction schemes for their preparation. DeNovo needed to think hard about how they were going to make the tool - which was programmed in Fortran - available to researchers. 'When you present users with tools that help them find novel drugs faster, or novel compounds that help them in their basic research, you don't want to be teaching them how to use yet another interface or piece of application software,' said Scott. 'But what everyone will use, with almost zero resistance, is a web browser.'
With a preference for open-source tools, DeNovo programmers used JavaServer Pages technology, which allows most of the work to be done by the server, rather than the browser. They wrote a compact platform-independent SOAP client using the standard Java Development Kit 1.4.1 and used XML (Extensible Markup Language) for all of the data transformation. According to Scott, the beauty of XML is that it allows validation of the data that is sent to the server for calculation: 'I've done a straw poll of my colleagues in the industry. Most of them - a mixture of chemists, biochemists, and molecular modellers - are now starting to use XML.'
The answer that comes back from a researcher's query is presented as a two-dimensional chemical structure by a small applet 'because chemists like to see output as 2-D structures', said Scott. 'They can then view and sort them by various properties that come back from the calculation - by a score or predictor of how good a compound is.'
The project was undertaken principally to allow DeNovo's pharma partners access to its technology, but the company's own staff liked the application so much that they now all use it too.
But cutting the costs of research and speeding drugs to market together represent just one route to increasing efficiency. Companies must also aim to maximise profits, once they are on the market. Better integration of information technology into the everyday running of the company helps to achieve that goal, according to Mark DiPaolo, another conference speaker and global head of e-business at Novartis.
This can involve some fundamental reshaping of the business. 'Technology implemented to reproduce a paper process wastes an opportunity to redesign business processes. Using fully electronic processes enables new business processes to evolve. For example, work formerly performed in one location can now be shared across locations and time zones for 24 hour work-days,' said Pfizer's Ivan Walrath. 'Cost, time saving, space saving and working globally are all important, but working globally is most important.'
DiPaolo agrees - but varies the degree of control according to need. The management of clinical trials, for example, is one of several IT functions that is organised globally, because of the need to compile a single dossier for submission to the US Food and Drug Administration. 'There's a whole IT group that manages the systems and processes behind clinical trials,' said DiPaolo. 'For marketing and sales it's a very different picture.' He described how 'e-initiatives have popped up all over the place', with the global office overseeing broad policy. The US is a huge part of Novartis's business - more than 40 per cent: 'They developed a lot of e-marketing and sales capability.' Best practices for these and other locally inspired programmes are shared via an e-business intranet, one of several Novartis internal networks.
A two- or three-day twice-yearly meeting of e-business managers around the world discusses important strategy issues such as security. In addition, an annual e-business report introduces new technology and new approaches to staff around the world, sharing best practice and knowledge.
DiPaolo has recently been formulating a global strategy for e-retailing - the use of interactive web sites, video clips and videoconferencing - to promote drugs to doctors. The practice was pioneered in the US, but he decided to go beyond guidelines to develop 'a functional capability' and to promote its use by other parts of Novartis. Now with a strategic partner 'we've developed a platform and we have over a dozen programmes we are executing on the platform' which will be available to marketing departments in other regions. Other hot topics for DiPaolo are patient compliance through messaging, RFID (radio-frequency identification) tags for various purposes, and interactive digital television.
But whether considering the latest technology or simple rules for company web sites, information technology must serve the wider business. 'We really focus on our top 10 countries and our top five brands. This is where Novartis puts nearly all of its resources,' said DiPaolo.