YouTube Facebook LinkedIn Google+ Twitter Xinginstagram rss  

By Michael Greeley

 May 9, 2003 | Venture capitalists are obsessed with funding the "next big thing." And occasionally we actually do that, but unfortunately too often we go to great excess. The promise of introducing emerging information technologies to the process of drug discovery is intoxicating. That said, one development with potential I've been watching cautiously is predictive modeling, which begins to move the drug discovery process from the wet lab to the computer lab.

The cost and time to bring new compounds to market are increasing. Ironically, the explosion of genomic data in recent years may have only exacerbated the challenge of discovering new therapies. In response to those developments, new venture-backed companies have emerged to provide new data-capture and analytical tools. These tools allow pharma to model particular organs and compound interactions in the body at sub-cellular levels. "We have all the parts and pieces now; predictive modeling will allow us to reassemble the systems," says Keith Elliston, CEO of Genstruct, a developer of disease-specific mechanistic models.

The focus to understand disease states at the molecular level will lead to targeted treatment regimens. This approach promises to create more effective therapies and allow pharma to treat smaller patient populations, reducing development costs and thereby breaking the blockbuster drug paradigm.

The pressures on existing drug discovery processes are numerous and complex. None of the bio-IT solutions at hand are adequate or comprehensive enough to completely respond to this tremendous increase in potential targets. According to Barbara Handelin, CEO of Kenna Technologies, a developer of in silico biological simulation models, "Predictive modeling is the discipline that one must turn to when complexity becomes too large." The challenge now is to create knowledge out of data.

Venture capitalists rushed into the bio-IT market in the 1990s, funding companies that successfully created this explosion of genomic data. Few companies anticipated that the evolution to greater value-added solutions would be required to sustain their businesses. At the end of the day, the drug discovery process for a single drug still requires eight to 10 years to reach preclinical trials, with unacceptable attrition rates.

Here Comes the Flood 
Bio-IT predictive modeling solutions in development by several vendors should dramatically improve the drug discovery model of the future. Many extraordinary entrepreneurs out of pharma, biotech, and IT sectors have started exciting new companies focused on these solutions.

The predictive power of these solutions promises to reduce to about two years the compound validation phase prior to preclinical testing. The resultant cost savings will be dramatic, as will the expected improvement in drug attrition rates through the clinical phases. Some analysts estimate that the oft-cited $800 million of pre-launch costs may drop to less than $200 million with the use of these new tools.

So what is there not to like about this sector? It has attracted a high degree of entrepreneurial talent. Unfortunately, compelling market opportunities often attract too much capital too quickly and too many companies get created. This has the devastating effect of fractioning a nascent market across too many vendors. A recent Frost & Sullivan study reported more than a dozen private venture-backed companies already exist, in addition to a similar number of public competitors. And this is for a market that Frost & Sullivan today measures at less than $200 million.

Nearly half of the top 40 big pharmas are piloting predictive modeling tools.
Here's the rub. Frost & Sullivan also projects that in the next five years this market will grow to around $6 billion. What is misleading about this projection is that it is predicated on the success of a robust discovery partnership revenue model. Partnering revenues are notoriously difficult to negotiate, much less to generate sustainable revenue streams. Elliston of Genstruct concurs. "It is difficult to establish a price if you are selling to a customer who has never purchased this before and they cannot adequately value the product." Venture investors are being asked to accept a software-based licensing revenue business model that will instead generate "drug discovery-like" revenues. Furthermore, the capital required to carry a company through lean periods is hard to come by. Venture investors are focusing obsessively on burn rates. Regardless, some observers estimate that nearly half of the top 40 big pharmas are piloting some form of predictive modeling and simulation tools.

However, many investors will experience the ever-present uncertainty of whether the product they're backing actually can predict molecular pathways. Chaotic and complex, biological systems may ultimately not lend themselves to mathematical modeling. As many pharma executives have told me, you cannot impose structure on what is an inherently unstructured environment.

Entrepreneurs need to keep these concerns in mind as they contemplate raising venture capital. Many bio-IT investors have experienced the euphoria of the genomic and proteomic revolutions, only to see their portfolio companies struggle to build sustainable revenue models. The excitement of this next wave of companies should address the lack of deep value-added services missing in the first wave of bio-IT companies. The promise may yet become reality.

Michael A. Greeley is the managing general partner of IDG Ventures, a global family of venture capital funds operating in North America, Europe, and Asia, with approximately $600 million under management. He can be reached at 

For reprints and/or copyright permission, please contact Angela Parsons, 781.972.5467.