Biotechnology stocks are years away from profits – but don’t tell that to Wall Street. In the weeks following September 11, biotech stocks held their ground as the fear of bioterrorism grew. Now that the U.S. actually faces a bioterrorism attack – so far three people have died of anthrax and on October 23 federal officials said anthrax was found at a mail center serving the White House – these stocks have gone through the roof. Shares of Cepheid, a California company that makes testing products to detect contamination, have risen by more than 400% in recent weeks. (It closed at $5.89 a share on October 22.) Bruker Daltonics, a Massachusetts-based developer of life sciences products, has seen its share price more than double – the stock closed at $19.89 on October 22. These companies show a great deal of promise, but the truth is that most of them will not have a commercially available product for at least five to seven years. In fact, the enormous gains in their stock prices during the last few weeks show how difficult it is for investors to value biotech companies. Typically equity analysts value such companies using proxy business drivers such as the dollar size of partnerships, number of patents, or number of discovered drug targets. But these drivers do not fully capture the ability of these companies to turn their research into marketable drugs.
Two Approaches to Drug Discovery
Traditionally, drug discovery in a biotechnology company has been driven by a science-based approach. Scientists at the company develop a hypothesis, perform experiments and ultimately deliver results. Some of these results are commercially viable – or so the company hopes.
A recent research paper by Wharton management professor Bruce Kogut and Michelle Gittelman, a professor at the Stern School of Business in New York City, describes the process as an organizational mechanism that combines the “capabilities of scientists within and outside the boundaries of the firm and ultimately intervenes in the normative selection process of science to produce valuable technical innovations.” (The paper is titled, “Does Good Science Lead to Valuable Knowledge? Biotechnology Firms and the Evolutionary Logic of Citation Patterns.”)
Recently, however, advances in technology with names such as ultra-high throughput screening, microfluidics, and computational biology have led to so-called industrialized methods of research where tens of thousands of experiments can be done simultaneously. Entrepreneur Wei-Wu He, a partner at venture-capital firm Emerging Technology Partners, says, “Certain biological processes, especially the ones that scientists do repetitively, can be automated and lead to an industrialized approach to discovering new genes.”
The sequencing of the human genome earlier this year provides a dramatic example of both approaches. Craig Venter, founder and CEO of Celera Genomics, who spoke at Wharton earlier this year, did the seemingly impossible when the company fully sequenced the human genome in just three years at a fraction of the cost of the publicly funded Human Genome Project (HGP). Using a cutting-edge product – a so-called automated gene-sequencer manufactured by Perkin-Elmer Biosystems – Venter devised an assembly-line method of sequencing the human genome. This technique was appropriately dubbed “whole-genome shotgun sequencing.”
This industrialized method of sequencing contrasted sharply with the methods that HGP initially used. Scientists at HGP painstakingly sequenced human chromosomes in a precise, bottom-up fashion. Base-pair by base-pair, scientists built up gene fragments, which led to complete gene sequences and ultimately whole chromosomes and then the entire human genome.
The success of Celera Genomics’ high-volume, computationally intensive, industrialized shotguns against HGP’s precise methods has led to optimism that automated methods of experimentation and discovery such as ultra-high throughput screening will ultimately lead to better drugs faster than the lab-bench type experiments of classical molecular biology. Another cause for optimism is the emerging technology known as microfluidics, which makes it possible to perform tasks such as analyzing DNA sequences by taking “advantage of the chemical properties of liquids and gases with the electrical properties of semi conductors by combining them on a single microchip,” according to OE Reports, a publication that tracks trends in the optical engineering industry.
So which approach to drug discovery is more effective – the plodding, classic one or the high-speed shotgun one? That is a question that Ulrich’s research helps answer – and it has crucial implications for the way biotech firms are valued.
Product Architectures
In his paper, Ulrich describes product architecture as the scheme by which the function of a product is allocated to its constituent components. He describes two types of product architectures: integral and modular. A product with a modular product architecture allows a division of its components by having standardized interfaces. The components are interchangeable and individually upgradeable. A personal computer is an example of a modular product architecture, where different components can be mixed and matched to create a complete product.
In contrast, a product with an integral product architecture is composed of components that are tightly bound to one another. In general, the individual components perform many functions, are spaced close to each other and are tightly synchronized. Typically, a high level of performance calls for integral processes while market-driven issues such as speed to market or flexibility make modular processes necessary. Biopharmaceutical drugs have an integral product architecture. “A drug can’t be made modular,” says Christian Terwiesch, a Wharton professor of operations and information management.
Pharmaceutical companies do have a certain level of modularity in their development chains, however. According to Terweisch, “In most pharmaceutical companies, though everybody is under one umbrella, work is done by highly specialized professionals and thus there are a lot of full handovers.There is relatively low interdependence among tasks. The interfaces between tasks are clean.”
These clean interfaces are between the various stages of drug discovery. These stages include target identification and target validation (a ‘target’ refers to the spot where a drug attacks). Then, there is lead identification and lead optimization (a ‘lead’ refers to a potential drug). Finally, the lead undergoes clinical development and ultimately, the drug is manufactured and sold as a commercial product.
On the surface, these clean interfaces allow the value chain in drug discovery to be disintegrated, or operated in separate, relatively independent modules. A deeper look reveals, however, that the processes are highly integrated. The processes have much greater similarity to small, focused labs than factories.
Here’s how Vivian Liu, senior scientist at Signature Bioscience, a California-based firm that has developed a new technique for analyzing biological data, describes the environment: “The principal investigator fosters an environment that promotes individual creativity and teamwork at the same time. Everyone approaches the same scientific question from different perspectives using a variety of technologies and works together with a common goal in mind. The end result is usually a prolific publication including a long list of authors on each paper, which is indicative of good management of individual thinkers in a team environment.”
Biotech companies, in most cases, provide inputs into the laboratories of pharmaceutical drug manufacturers. In some cases, these inputs consist of information from databases. In others, they are drug targets or lead drug compounds. These are not integrated solutions but merely fragments of a solution. Pharmaceutical companies take these inputs and internalize them through their proprietary processes, which results in outputs that are consistent and at a performance level with their internal research. Some representative performance levels between stages are that only .02% of drugs screened ever come to market while 23% of drugs entering Phase I clinical trials reach the market.
Biotech companies, however, are no longer content to be mere suppliers to pharmaceutical companies. Many biotech companies are reshaping themselves into drug discovery companies themselves, trying to capture the huge margins of the pharmaceutical industry. Their main competitive advantage is an intimate knowledge of their core technology platform. Will they succeed in doing this? That depends on how integral their drug discovery processes are.
Theory suggests that at the present level of understanding of molecular biology, drug discovery processes would be difficult to modularize. Based on academic research, several conditions need to be satisfied to modularize these development steps. There must be clearly demarcated functional specifications between the components and the rest of the system. There has to be technology that can accurately measure those attributes to ensure that they are properly achieved. The whole process needs to be so well understood that there is explicit codification of design rules.
Ulrich says that when the production processes are understood and stable, “it is often possible to establish design rules that express the constraints of the production process.” At the current level of biological understanding, these conditions cannot be satisfied. In addition, as can be seen from the small percentage of compounds that actually make it to the stage of becoming marketable drugs, it is clear that we are nowhere near the technology performance demanded by industry.
Interviews by Knowledge at Wharton indicate that industry experts, in general, agree with the theoretical analysis that developing commercial innovations out of biological science requires integral processes. Ron Garren, editor of the influential Biotech Insight newsletter, says: “My bias is that this is still a molecular biology problem. Ultimately someone has a smart idea and figures out the underlying molecular biology or comes up with a novel way of solving a problem.” As for the automated, shotgun approach, Garren has his doubts about their efficacy. “All these automated approaches get you a little closer but they don’t give you the answer,” he says. “Finding a drug is still like finding an oil well. Sometimes you can use rational approaches like satellites to locate wells, and at other times it’s just luck.” In the same way, he adds, “you can use these tools to try to understand the biology or you can just bet on a star scientist.”
Despite the consensus opinion, biotechnology companies are deploying a wide variety of product development processes in their attempt to forward integrate. The modular approaches typically involve platform technologies with high-throughput, automated technology aimed at a wide variety of disease conditions. Affymetrix, Aurora Biosciences and Celera Genomics are companies that fall within this category. Affymetrix sells DNA microarrays, products that are more commonly called DNA chips. They consist of DNA implanted on wafers and allow researchers to analyze thousands of genes at a time to see which ones are active in particular conditions. Aurora Bioscience sells the Ultra-high Throughput Screening System (UHTSS) Platform, which can screen more than 100,000 compounds a day with more than 2,400 re-tests, accessed from a store of over one million compounds. Finally, as mentioned earlier, Celera Genomics industrialized gene sequencing with advanced instrumentation.
Companies with integral development processes typically start with a molecular biology specialization, then bring in appropriate technologies as needed. This discovery capability typically is focused on specific biological disease areas. Representative companies in this category include Exilixis, Geron and Onyx. Exilixis has a technology platform that uses different species to discover gene function, elucidate disease and biochemical pathways and validate novel drug targets. Geron uses sophisticated technology, called telomere studies, to understand and ultimately treat age-related diseases including cancer. Finally, Onyx is studying anti-cancer therapy based on exploring differences between cancer cells and normal cells.
Merck’s Integral Premium
These different approaches, however, do enable observers to differentiate the valuations of a wide variety of biotechnology companies, at least on a relative basis. One way is to calculate the “integral” premium that the markets bestow on highly integrated discovery organizations.
Merck is considered to have the most integrated drug discovery engine. Some 50% of Merck’s sales come from direct sales of pharmaceuticals outside its pharmaceutical benefits subsidiary, Merck-Medco. As a result, it is possible to assume that 50% of its market capitalization of $141 billion can be attributed to its drug sales and pipeline. It is also possible to estimate how many drugs Merck has in all stages of its pipeline from the laboratory to the marketplace. The company has 24 drugs in the market, and using the company’s annual report and other public sources of information, one can draw some conclusions about Merck’s drug pipeline. The report describes one New Drug Application (NDA), 18 in Phase IIb and beyond, and three in Phase I tests. However, this is likely to be an underestimate.
For drugs in various stages of development, the odds for reaching the market are well known: 20% for drugs in Phase I, 30% for Phase II, 60% for Phase III, and 80% for NDAs. Utilizing these probabilities, Merck’s pipeline consists of 33.5 drugs, which results in $2.1 billion in market capitalization per drug.
Using various scenarios, the same type of analysis can be done on the biotechnology “drugs” which works out to approximately $1 billion in market cap per drug. That gives a integral premium of 2.1. One way to look at this premium is the amount of shareholder value left on the table. A biotechnology company could unleash shareholder value by organizing its discovery capabilities into an “integral” development structure and properly conveying that to investors.
This framework can also be used for valuing biotech companies. Certainly, these concepts can provide insights into the distributions and values for a Monte Carlo Simulation, a tool often used in valuing early technology companies. However, consider a simple approach. Equity analysts often use valuation by multiples to arrive at a company’s market value. This involves comparing some performance measure of the firm such as revenue to an average of other comparable firms. For example, investors can compare the market/revenue ratio of a firm to representative multiples of similar firms. The trick is to make sure that the comparable firms are truly similar to the firms that are being analyzed.
To employ the method described in this article, it is important to ensure that all the firms being compared are similar in their developmental architecture. Typically, investors would use a set of comparable firms in valuing a firm but for illustration purposes, consider only one example. Exilixis – a company pioneering the use of genetically manipulatable model systems for biomedical research – should be compared to a company like Millenium, which has a highly integral development architecture, using a variety of technologies to develop biopharmaceuticals. The analysis shows that Exilixis is undervalued relative to Millennium.
A quick glance at equity reports shows Celera Genomics is compared to a variety of companies including Millennium. A company such as Celera, at this point in its development, has more in common with a “modular” company such as Incyte – and analysis shows it is overvalued in relation to Incyte. Of course, to be fair to Celera, it is attempting to forward integrate into a structure similar to Millennium. The market suggests that it is midway in the continuum between modular and integral development structures.
Economist Hernando de Soto wrote in The Mystery of Capital, “The great practitioners of capitalism, from the creators of integrated title systems and corporate stock to Michael Milken were able to reveal and extract capital where others saw only junk by devising new ways to represent the invisible potential that is locked up in the assets we accumulate.” Ulrich’s concepts help understand the invisible potential locked up in the organizational structure of biotechnology companies.
