Monday, June 4, 2007

Applying systems theory to produce better medicines - SDM Pulse, Summer 2007

Editor’s note: This is the first in a series of articles that will follow Ragu Bharadwaj’s progress through the System Design and Management Program. In this piece, Bharadwaj introduces the problems inherent to the drug development activities of today’s pharmaceutical industry. He hopes to find ways to improve these processes through the strategies and techniques taught in SDM.

Current and alumni SDM fellows are invited to contribute their thoughts on how best to address these issues by writing SDM Industry Codirector John M. Grace, Suggestions may be featured in a future issue of the SDM Pulse@MIT.

By Ragu Bharadwaj, SDM ’07

As a computational chemist who works in the pharmaceutical industry, I joined the 2007 SDM cohort to find ways to improve the industry through systems thinking.

Drug discovery and development are long processes—it typically takes 10 to 15 years and well over a billion dollars to bring a new drug to market. Much of the industry’s knowledge and expertise is tacit, so knowledge capture is difficult and not well implemented. And, the stakes are high—only about one in 10,000-15,000 compounds synthesized makes it to clinical trials—and that makes pharmaceutical companies secretive.

I am interested in introducing efficiencies to drug discovery and development—which today is a poorly understood, continually evolving system of processes, with poorly integrated supply chains and very high failure rates.

There are three main steps to bringing a drug to market: drug discovery, drug development and commercialization.

Drug discovery begins with evaluating the benefits of developing a drug for a particular disease or condition. Issues to be considered include cost, intellectual property rights, biological target validation and assay development. Drugs that get past this stage proceed to lab testing on animals and, with luck, to clinical trials.

During drug development, candidate molecules are tested in human trials. Drug materials and placebos must be available in the right doses at the right time, which makes it important to understand the supply chain. The supply chain takes on further significance when elements in the design and development process are globally distributed. Clinical trials usually cost $100 million to $200 million per year and involve simulation and statistics experts as well as doctors.

During commercialization, the FDA-approved drug is marketed to doctors and sometimes to patients who can influence their doctors.

How can systems theory, analysis and design improve these processes?

Drug discovery and development involve iterative cycles with feedback loops and decisions, currently addressed mainly by aggregated domain expertise.

The drug discovery process often starts with chemists evaluating literature, patents and assay results from compound libraries (assortments of diverse compounds) to identify promising "hits." New compounds are synthesized using input from medicinal chemists, computational chemists, pharmacokinetic experts and toxicologists. Variable cycle times for chemistry and assays introduce time delays in the information feedback cycles. System effects work in devious ways to slow down and reduce useful information obtained from each cycle.

I’m hoping that we can improve these processes using ideas from systems product development, systems dynamics, lean thinking and decision analysis.

After three to five years and about 10,000 compounds, multiple candidates are proposed to the development team, which tests them in animals for toxicity and other properties. There is a high chance of failure. Animal data takes a long time to obtain and is highly variable. Hard decisions are made with poor data during development.

Next, FDA permission is sought for clinical trials. Reliable data capture and statistical analysis are critical at this stage, yet trials are often carried out in multiple, remotely located hospitals. The documentation submitted to gain final FDA approval for a drug can easily exceed a million pages. Managing all this information requires precise coordination and control.

When a new drug is finally approved, there is still the hurdle of selling it to recoup costs and make a profit. What efficiencies can be introduced to this part of the system? Convincing a risk-averse doctor to adopt a new treatment is a costly exercise requiring a knowledgeable salesforce.

I’m hoping we can apply ideas from Systems Theory and Systems Dynamics to identify and change the slowest and least efficient parts of the system. Perhaps we can leverage ideas developed in other industries such as manufacturing.

Certainly, the challenges posed by this complex system are well worth tackling. After all, solving the problems of the pharmaceutical industry holds out the promise of better medicines for everyone.

No comments:

Post a Comment