SIEMENS

Detectives are at work at Siemens. They are sifting through lines of genetic code for evidence, developing smarter, faster tools for discovering the fugitives they are pursuing, and working closely with authorities in universities and government agencies to capture and liquidate their targets. “It’s a predator-prey relationship,” comments Gayle Wittenberg, Program Manager at Siemens Corporate Research (SCR) in Princeton, New Jersey. ”We are getting better and better at what we do. But the bad guys - the viruses and bacteria we are tracking - are hard to identify because their genetic characteristics are constantly evolving.”

Although the battle against invisible invaders such as the wily methicillin-resistant Staphylococcus aureus (MRSA) bacterium, which accounts for about 20,000 hospital deaths per year in the U.S. is just getting started, Wittenberg and other molecular detectives at Siemens have scored significant successes, most notably with regard to identifying the unique characteristics of what is now known as “the pandemic 2009 H1N1 flu virus.”

The story of how Siemens researchers identified the 2009 H1N1 Flu “suspect” unfolded like a rapid-fire whodunit. On April 14, 2009, after numerous deaths in Mexico, the World Health Organization (WHO) and the U.S. Centers for Disease Control (CDC) announced seven cases of the unique hemagglutinin-neuraminidase (H1N1) flu strain in the U.S. On April 27, the WHO declared a pandemic alert level 4 on a six-point scale (with 6 being the highest level). And on April 28, the U.S. National Institutes of Health (NIH) published all available versions of the virus’s genetic code on the Internet.

“As soon as that information became available, we began running it through our biomarker discovery process for the development of rapid diagnostic tests, which we call the RAPID2 pipeline,” says Dorin Comaniciu, PhD, who heads Siemens’ Medical Informatics Global Technology Field at Siemens Corporate Research in Princeton, New Jersey.

Comaniciu explains that the key to the identification of any pathogen is the ability to zero in on those parts of its genome that all members of its group have in common. “No two genetic sequences in the same group are entirely identical,” he says. “But some parts of their sequences are. And that is where our technology comes in - specifically our experience with machine learning and face recognition, which is a similar kind of pattern recognition problem.” Using a cluster of high-performance interconnected computers, Wittenberg, Comaniciu and other SCR researchers compared the NIH’s H1N1 sequences to a publicly-available database of sequences from common flu strains. Their effort paid off quickly. Within less than two days following the NIH’s posting of the H1N1 genotype, RAPID2 had identified the unique sequences - also known as “signatures” or “primers” - that distinguished the H1N1 virus from all others.

The resulting information - a digital representation of part of the virus’s genetic code that is equivalent to a suspect’s fingerprints - was transferred from Siemens Corporate Research to Siemens Molecular Diagnostics’ world-class research and development facility in Berkeley, California. There, according to Vice President for Global R&D Management Dr. Norbert Piel, “One of our most experienced scientists was able to translate RAPID2’s computergenerated signatures into their nucleic-acid counterparts, otherwise known as detection reagents.” Once these reagents had been produced, they were tested on known copies of the virus using Siemens’ VERSANT automated platform for kinetic polymerase chain reaction (kPCR)-based genetic detection. “The result,” says Piel, “was that on the very first try we obtained a perfect match between the signatures we had generated and those of the actual virus.”

As soon as the test was ready, it was added to Siemens Molecular Diagnostics’ panel of flu assays, transferred to a Mexican state lab at the heart of the pandemic, and tested in a double blind evaluation on the automated VERSANT kPCR platform. “There, results indicated that the assay was as sensitive as and possibly more specific to the 2009 H1N1 virus than a test developed by the Centers for Disease Control,” says Siemens Healthcare Diagnostics Director James Uzgiris, PhD. “In this comparison to the CDC assay, our test was more specific for the 2009 H1N1 virus. We attribute this to the excellent capabilities of the RAPID2 algorithm and to the performance of the VERSANT kPCR platform.” Siemens has filed a patent application on the methods used and is analyzing whether there is a business case for developing a commercial H1N1 assay to add to its panel of flu tests.