The changing shape of pharmaceutical R&D

“It’s a shape-shifter. . . . It will assume the shape of whatever it thinks will frighten us the most.”

Just as pharmaceutical researchers and developers are becoming comfortable with such simple but crucial ideas as the “innovation funnel” and the “technology pyramid”, they are watching the shapes that embody these ideas change almost beyond recognition. Designing R&D strategies, processes, and technologies that mold to changing situations is not easy, and increasingly these organizations are turning to consulting partners for help.

Good consulting partners help organizations manage change. They develop an understanding of why R&D takes on the shape it does, what is driving the changes in shape, and where to focus resources to take advantage of the new shape when it emerges.

Accelrys Consulting is no stranger to change—the company was recently formed from a combination of five well-established companies in the simulation and informatics marketplace—but the company’s greatest asset is its dedication to long-lasting partnerships and complete customer satisfaction. Dedicated teams design, develop, and deliver customized systems that combine world-class informatics and modeling technologies to meet the exacting needs of pharmaceutical research.

Consulting partnerships are valuable when the R&D shape seems to be changing especially quickly. Three such situations are transitions from

Innovation funnels to bathtubs
The “innovation funnel” is the familiar image of the R&D process. A wide range of drug candidates and research choices are narrowed down over time until the one or two compounds that are commercially marketable emerge from the end of the funnel. Normally, the funnel has a smooth, possibly linear shape. Until recently, the typical pharmaceutical IT vision was “funnel compression”, in which it was hoped that increased productivity would shorten the overall funnel length by a few years.

However, to paraphrase Sangtae Kim, vice president and information officer of Lilly Research Laboratories, the funnel is effectively becoming a bathtub (2). The emergence of new technologies such as genomics and robotics means that more and more data are flooding in from discovery, only to try to pass through the small hole defined by the limited growth in the number of clinical trials. Kim explains,

This “tidal wave” of data will propagate along the preclinical development segment and onto the doorsteps of the clinic. From there, advances in data mining are required to help filter the greater volumes of data and translate the new wealth of choices in the preclinical segments into a set of more robust candidates for the clinical trials, there by achieving less attrition in the higher cost portions of the pipeline. The message . . . is alignment with a changing profile as well as the conventional goal of cycle compression.

In effect, the strategy is to know more, sooner. The key to effective innovation funnel management is to design and place your strongest, quickest, least expensive screening processes as far upstream as possible. This seems like common sense, but it requires an acceptance of the importance of information, which pharmaceutical companies appear slow to demonstrate.

On average, the pharmaceutical industry spends only ~5% of sales on IT, far less than banks and other information-intensive industries (3). A recent PricewaterhouseCoopers report estimated that as much as 80% of the information held by many large pharmaceutical companies is unstructured and therefore not easily searchable (4). When pharmaceutical companies begin devoting more resources to their IT systems, will these systems effectively convert a “tidal wave of data” into a large number of new products?

As a first step, solutions must be put into place to manage key data and decisions along various points of the funnel, such as target identification, lead optimization, or formulation. Accelrys implemented such a solution for Procter & Gamble (see box, “An answer to the data deluge”), with excellent results.

Given the shifting shape of the complete funnel, point solutions—individual tools that support only a part of the R&D cycle—will ultimately provide a partially effective answer. Managing the flow of information and decision points up- and downstream in the R&D cycle requires that the entire process be integrated.

Such a broad integration may appear daunting; however, the technology to solve many of the informatics challenges in life sciences exists today. New distributed-object computing technologies have been applied successfully in a variety of other industries. The knowledge-led R&D infrastructure is based on using the existing available databases, processes, and applications already familiar to scientists, and simply making them work together better in the scientists’ hands.

The existing vertical applications that the scientists have been using for analysis in single domains (e.g., biological activity, chemical inventory, analytical results, and computed results) simply cannot represent a sufficiently wide set of data or provide flexible enough analysis tools across all of these domains. Demands for custom applications arise frequently and are often urgent. This is an area in which the interface between IT and R&D groups has traditionally been most strained.

Speed and functionality can be achieved, either by in-house information systems (IS)–IT teams or by outside consultants, using component-based development and links to standard desktop tools. Of all the ways to bind an enterprise together, the middleware approach is perhaps “the best path to a solution”, according to David Linthicum, CTO of SAGA Software Inc. (Reston, VA) (5). One of the middleware methods acknowledged as having one of the most mature development environments is JavaBeans (6). Novel interfaces can be built quickly from components in this development environment, combining the components to assemble the functionality and user interfaces that researchers require. The distributed architecture, which manages data in a range of federated and distributed resources, provides robustness.

When scientists and IS–IT staff are freed from the constraints of existing domain-specific systems, they can begin to think more naturally about information from all sources and use it in creative and innovative ways to accelerate decision making. Using an integrated informatics environment, scientists working in pharmaceutical R&D can find answers to questions such as:

This is the key to managing the flood of incoming data to obtain the most useful information output.

Technology pyramids to office blocks

A “technology pyramid” is a simple shape that embodies the complex interplay between information technology and the value it brings to a company over its lifetime. At the lowest, earliest level is the underlying software framework, which must be incorporated into an organization’s IT structure. The middle levels are typically automation and business intelligence, which become integrated with and used by the workflows (business processes) of the organization, and in doing so, optimize them. Finally, the longer term benefits of the best IT solutions are at the top: allowing an organization to identify new opportunities, and so be more responsive to change, while decreasing reliance on and maintenance requirements of outdated systems.

However, this simple shape is changing. Many pharmaceutical companies have recognized that the value grows as you go upward, but the volume (the number of people involved in value-adding activities) decreases. Now as more organizations emphasize the need to give software users as much power as possible to add value, they are effectively enlarging the ranks on the higher levels of the organization. The pyramid is turning into an office block.

Unfortunately, pharmaceutical companies lag behind other industries in two out of four basic areas related to knowledge and information management competencies, according to Elisabeth Goodman, assistant director of information management at GlaxoSmithKline. They are on a par with other industries by holding initial discussions on building their “office blocks”, and by having some groupware and intranet interfaces in place, but they are well behind in defining basic competencies and including them in development plans, as well as in having any training programs.

Training the end users of the new office block model to use the software is essential to the acceptance of the technology and is therefore essential to success. However, the availability and quality of training and the software quality assurance (QA) processes used can vary greatly. At Accelrys, we have almost 70 people in the consulting business unit, and we devote significant resources to training, documentation, and QA.

Training is also essential to swell the middle layers of the pyramid. Routine menu-based training in the application must be replaced by training that is domain-specific and based on business cases. Although it is easy to do routine training even with few resources, users must discover the value and intricacies of the software on their own, and more often than not, they become discouraged. If the training is grounded in domain knowledge and aimed toward business knowledge, the value of the software is more immediately apparent, and user acceptance is more readily gained. A consulting organization must have sufficient technical support staff with the domain and software knowledge to offer value-added training. Perhaps the most critical consideration, however, in changing the shape of technology uptake is to understand the processes that under lie user acceptance of a complex IT solution.

Raymond C. Rowe, professor of industrial pharmaceutics (School of Pharmacy, University of Bradford, West Yorkshire, U.K.) and company research associate at AstraZeneca, summarized an interesting point of view in an editorial (7):

I have always assumed, like most researchers, that if a technology had certain advantageous features, benefits would flow automatically on implementation, this of course being the driving force behind the urge for continuous technological improvement. However, I now know that this is a naive assumption and that the link between the provision of advantageous technological features and successful implementation is, at best, a gross simplification. It is not completely incorrect because there are many examples where the provision of advantageous features has led to real benefits downstream. However, it cannot be taken for granted because, in many cases, no benefits ensue. There is a real need for a model that will enable researchers to understand not only the usage and benefits of a new technology, and hence predict its success or failure on implementation, but also to leverage its successful implementation by fostering those desirable benefits through design at the outset, that is, using foresight.

Rowe then describes the importance of the tasks people perform and the roles they fill in an organization in assessing the features and benefits of any new technology. Tasks and roles should be assessed at the design stage of a project, when the benefits of any changes in a task or role supported by the new technology are much easier to identify.

Some of this is simply good practice. The first and most important stage in any consulting project is to understand the customer’s business and technical needs. This requires listening, reflection, and observation with a number of groups at the client’s sites. However, the key to success is in knowing the tasks and roles of the scientists for whom the software is being designed.

Good consultants have extensive pharmaceutical research experience and provide a cultural fit, which they use to facilitate communication and collaboration across the organization, from discovery to development. They speak the language of the geneticist, the medicinal chemist, the synthetic chemist, the assay biologist, and the formulator. Through daily contact with these people, they continue to learn the tasks and roles as they evolve. They identify where software can support a better way of working for clients and enhance the role they play in their organization. They identify where clever, but inappropriate, software would decrease researchers’ effectiveness. Throughout the design, implementation, and uptake process, user success comes first.

This philosophy has an important consequence: Putting customer success before everything means that consultants may decide to implement a system with components provided by other companies. Accelrys consultants have an unparalleled breadth of knowledge of IT solutions in the pharmaceuticals industry and continually invest in relationships with those who are “best-of-breed”. Most organizations do not want to be tied to a single vendor, and because we value our independence, we support them in this.

Technology life-cycle curves to bubble charts

Figure 1. The technology life-cycle curve. In the past, organizations chose an appropriate time to adopt a new technology based on IT resources and corporate culture.

Figure 2. Organizations now manage portfolios of technologies, investing their resources to optimize the balance of risk and reward. The radius of each circle is proportional to the resources that an organization invests in a given technology. The circle in the lower-left quadrant represents “I don’t know what we’re doing here or whether it’s important, but everyone else is doing this”—an approach that may be followed without realizing it, but that can become apparent when this type of chart is used.

The technology life-cycle curve embodies the familiar idea that there are different approaches to embracing new technology. The groups of people coming to a new technology over time can be described as innovators, early adopters, early and late majorities, and laggards; the numbers in each group give rise to the normal distribution depicted in Figure 1.

Using this model, an organization could examine its IT resources and culture with relatively little difficulty, and with a high degree of confidence decide where it was on the life-cycle curve. However, this simple one-dimensional picture is changing, and IT departments must now cope with being at several points on the curve at once.

Eric Degn, vice president of Business Information Technology at Novo Nordisk (Denmark), described the motivations for managing a portfolio of projects (8). He said that the goal is to manage the spread of projects because this best matches resources to opportunities, ensures control of overall IT investment, maintains compliance with the IT strategy, and aids top management’s awareness and ownership of projects.

Degn then described the use of a bubble chart in Novo Nordisk’s project prioritization process. A bubble chart is one way of representing a portfolio of technology projects (Figure 2). The probability of success is plotted along the y axis and rewards along the x axis; each project is represented by a circle (bubble) whose radius is proportional to the resources invested.

A key skill of the technology portfolio manager is knowing how and where to nucleate new bubbles and control their motion in the context of their moving neighbors. This places three constraints on a consulting organization:

Accelrys meets these constraints by having a large group of domain-experienced consultants, who work on a portfolio of projects that covers the range of business areas from combinatorial chemistry to e-commerce (see box, “Tackling combinatorial enumeration”). Recent and ongoing projects include

Bubble charts are rarely developed without an accompanying discussion of risk, which in R&D “comes with the turf, virtually by definition,” according to Michael J. Smith (9). Risk is especially important for pharmaceutical IT departments in an environment moving toward effective portfolio management. These departments must recognize the benefits of taking risks (the word itself comes from the Italian risicare, meaning to dare, notes Smith), but this depends critically on the accuracy of assessing the investment needed, rewards to be gained, and roadblocks to success. In performing these assessments, the value of a partner who has successfully managed pharmaceutical informatics consulting projects over the course of many years can prove to be valuable.

Bathtubs, office buildings, and bubble charts
The changes facing pharmaceutical R&D will place high hurdles in front of the organizations devoted to supporting them. Accelrys’s goal of providing an integrated technology platform to support pharmaceutical discovery and development depends in a key way on the unique skills of its consulting business. These skills include

These skills, combined with a commitment to value for the client, place our company in a unique position to surmount those hurdles.

David Nicolaides is a senior consultant at Accelrys Inc. (230/250 The Quorum, Barnwell Rd., Cambridge CB5 8RE, U.K.; +44-0-1223-402-846; dave@accelrys.com). His Ph.D. from Cornell University (Ithaca, NY) is in computational chemistry, working on optimizing computer programs. He did postdoctoral research at Rutgers University (Piscataway, NJ) studying the classical theory of liquids; at the physics department of the University of Edinburgh studying critical phenomena; and at the University of Bristol physics department working on computer simulations of wetting and adsorption, high-T_c superconductors, and Gibbs ensemble simulations. Before joining Accelrys in 1996, he worked in the chemical engineering department of the University of Bradford (West Yorkshire, U.K.) studying computer simulations of mesoscale systems, such as powders, pastes, and suspensions.