Though it accounts for just a tiny percentage of overall chip sales, a thumbnail-size glass plate on which intricate patterns are printed is a tool with the power to transform drug research and improve the health of millions of people. It's called a biochip, and the patterns hold tens of thousands of "probes" -- segments of DNA that represent genes.
In a typical experiment, a drug researcher places a sample of diseased tissue that has been tagged with a fluorescent dye onto a gene-laden chip. A scanner then reads the chip, and if the DNA in the sample matches any of the genes on the chip, that part of the chip lights up. Matches that occur over and over provide invaluable clues about the role of that specific gene in cancer, AIDs, and other diseases.
With the introduction of the first biochip in the late 1980s by Affymetrix (AFFX) in Santa Clara, Calif., drug researchers have been able to dramatically speed their pace of investigation. And as researchers' interests have become more varied and nuanced, chipmakers have come up with new iterations of biochips.
GENOME ON A CHIP. Today, some chips help scientists see which genes contribute to a particular disease and to what extent. Others help track the tiniest mutations between the same gene sequences -- or strands of DNA -- in people with the same disease. Still other chips aim to illuminate the activity of proteins -- the hormones, enzymes, and antibodies that are the foot soldiers of disease and a body's defenses against it.
Over the past decade, the power of these chips has multiplied. Nowadays, a single chip can hold the contents of an entire human genome -- 30,000 to 100,000 genes, depending on who's counting. Just two years ago, it took two chips to hold that amount of data, and a decade ago it took five. Labs at drug companies and universities can't get enough of the latest versions. Though they get cheaper every year, just like their cousins in computers and electronics, dollar-sales figures continue to rise -- as much as 50% in 2003, to perhaps $500 million, says Aaron Geist, an analyst at Robert W. Baird in Milwaukee.
Attracted by this trend, more competitors are entering the market -- including General Electric (GE), which acquired British biochip maker Amersham in 2003. GE, plus No. 2 Agilent Technologies (A) and more recent entrants, such as Illumina (ILMN) and Applied Biosystems (ABI), are stepping up their efforts to grab chunks of Affymetrix' estimated 70% market share by attacking it high and low.
FALLING SHARE? Illumina has beat the market leader to the punch with biochips that hold up to six human genomes. At the same time, plenty of small competitors are vying to sell chips that are focused on limited groups of specific genes. Affymetrix, meanwhile, is looking beyond customers that do pure research to those that do diagnostic tests -- a market it expects will be robust.
Opening new markets is becoming an imperative for Affymetrix. Though it remains the biggest biochip maker by a wide margin, its sales have been flat or declining over the last three quarters, says Paul Knight, an analyst at Thomas Weisel Partners in New York. "I think Affymetrix' share could fall to 60% as others come into this market over the next 24 months," he says.
Much of the new demand for chips, Knight adds, is for those more customized than Affymetrix makes. Once the mapping of the human genome was completed in 2001, researchers quickly turned up a vast number of possible "targets" for potential drugs -- often a protein that counteracts disease-causing genes. Now, the need is to narrow the field by validating which targets deserve further study. This new mission requires chips whose focus is small groups of specific genes or the genetic makeup of obscure organisms. "Now that you have a better idea of what gene you're interested in, that creates demand for more customizable chips," Knight says.
"MORE ADAPTABLE." Agilent, No. 2 in the industry, is benefiting from this trend. It estimates that its share of the biochip market rose to 15% in 2003, up from what analysts say was single digits previously. That growth, says Mel Kronick, Agilent's chief scientist for bioresearch solutions, came from sales of "catalog chips" -- which contain the genetic maps of organisms widely used in research, such as human, rat, and mouse -- and from sales of customized chips.
Customization may be Agilent's primary edge. It uses an ink-jet manufacturing process for its chips that is "much more adaptable to customization" than the photolithography process that Affymetrix uses, Kronick claims.
Agilent has plenty of company. Emile Nuwaysir, vice-president for business development at NimbleGen in Madison, Wis., says his company makes chips to the specification of "hundreds" of customers -- and then does the experiment for them and sends back results. This service is popular with labs that want to explore the characteristics of very specific sets of genes or of organisms whose genome isn't available off-the-shelf.
"TRYING TO CATCH UP." NimbleGen, which has yet to turn a profit, will introduce several new niche services in 2004. It'll build so-called promoter chips, which will track the interaction of proteins with genes. And it'll create chips to help in the study of genetic variations between microbes -- knowledge that could come in handy for finding new antibiotics.
Other players are trying to make a name for themselves as the limits to how much can be put on a single chip keep expanding. Illumina's six-genome chip, announced in mid-January and scheduled to be available at midyear, will let customers run six of the same experiments on a single chip, saving time and money, says Jay Flatley, Illumina's president and CEO. He hopes that this product will take some market share from Affymetrix. "We'd be naïve to think we can unseat Affymetrix, but we're trying to catch up," he says.
Of course, Affymetrix, which last fall was the first large player to put the entire human genome on a single chip, has plans to avoid being outdone. And it continues to add to its large library of chips containing whole genomes of heavily researched organisms.
"GREAT DIAGNOSTIC." Trevor Nicholls, chief commercial officer of global operations at Affymetrix, says researchers are increasingly finding its mix of products to be the right one. He doesn't see customization becoming a big part of Affymetrix' overall business, because "we can put so many whole genomes on a chip, there is less need for custom chips." He says its latest chips and chip-reading hardware allow for increasingly automated and more sensitive testing.
Affymetrix is also ahead in exploring the next frontier: developing real-world uses for chips. In 2003, Swiss drugmaker Roche said it would use Affymetrix biochips to develop diagnostic tests. One of the first that's likely to emerge from the Roche partnership will detect variations in a gene called p450, which is known to affect the body's ability to process about 25% of drugs on the market. Knowing more about a patient's p450 gene should make it easier to prescribe the right medicine for that person. "It will be a really great diagnostic," says Robert Baird analyst Geist.
In February, moreover, Affymetrix and French partner Biomerieux will introduce a chip that can test food DNA. Such information could be of interest to farmers who want to make sure that they're using untainted animal feed.
NO MORE GUESSWORK? Other players, such as Amersham, see the emerging field of proteomics -- the study of the structure and function of proteins -- as a differentiating strength of their biochips. Trevor Hawkins, senior vice-president for development at Amersham's discovery-systems division, says it has been actively working out how to put a large number of proteins on chips, at the request of customers. "[Protein chips] are one of the next major growth areas," says Hawkins, who adds that Amersham hopes to be selling such products a year from now.
Perhaps the ultimate achievement would be to use biochips to provide "personalized" medicine, a goal that scientists are gradually closing in on. In late 2002, researchers in the Netherlands used Agilent chips to read the gene expression pattern of women with breast cancer and determine what treatments would work best for their particular tumors. Many drug companies are already using these chips to learn more about the genetic makeup of the recruits in their clinical trials. This information should help make clearer which patients will and won't benefit from the drugs being tested.
True, efforts to understand the links between genes and disease are still in their early stages. Much more work will have to be done before custom treatments are the norm. But they're almost certain to become that at some point, as innovative companies come closer to unleashing the potential of biochips. By Amy Tsao
in New York