In the world of proteomics, Europroteome chief medical officer Marc Reymond is a bit of a maverick. Rather than take a brute force approach to analyzing proteins relevant to disease, the founder of the Hennigsdorf, Germany-based company prefers to look at the emerging discipline this way:
“If you go hunting, you can come with 20 machine guns and fire into the forest, and then you can go and you look what’s lying on the ground. That’s not our philosophy. We observe, we look at the animals, [and if] we think this animal is interesting, then we fire. Of course to be complete, because of the patent situation now, you have to be in the forest before someone comes with the machine guns.”
From Patient to Protein and Back to Patient
Reymond’s analogy does much to explain Europroteome’s scheme for deriving value from studies of the human proteome, but it doesn’t provide the whole story. In fact, as Reymond demonstrated at a recent IBC proteomics conference in Geneva, the company is focusing on a “patient first” approach to understanding disease: instead of first assembling molecular biology data to apply to studies of disease, Europroteome, with its network of 40-odd clinical researchers and their access to clinical samples, is first assembling clinical and pathology data, and then integrating gene expression and proteomics information to fill out the whole picture.
“We decided to start with conventional medicine, and then plug in molecular biology,” he said. “We truly go from patient to proteomics, and then back to the patient.”
Reymond’s presentation at the recent IBC meeting of the company’s unpublished research provides a good example. In it, he described an algorithm developed in partnership with Phase-It, a Heidelberg, Germany-based bioinformatics company, that the companies designed using clinical and pathological data collected prospectively from 2759 patients with colorectal cancer.
Europroteome and Phase-It used the data to train the rule-based algorithm, called OncoAssist, to identify which patients in the subset of those with stage I and II colorectal cancer would go on to experience an often fatal recurrence after their initial tumors were removed. Reymond and his colleagues were surprised to find that the algorithm could successfully predict survival and the tendency of the cancer to metastasize — even without additional molecular biology information.
Currently, Europroteome is incorporating the proteomics and transcriptomics data it has collected from patient samples using 2D gels into its models of colorectal and other cancers, to gain a deeper insight into the pathways responsible for disease. Eventually, Reymond hopes the approach will translate not just into better predictions of patient prognosis, but more importantly, biomarkers and drug targets to help treat patients with cancer.
To make this happen, Europroteome is relying on proteomic data collected over the last five years using a proprietary sample preparation method. The method uses a ubiquitous epithelial antigen in conjunction with mechanical and immunological separation steps, providing Europroteome’s clinical network with a standard procedure for handling clinical samples, as well as a means to sort out epithelial cells, from which adenocarcinomas arise. Reymond said the company has been awarded a patent on the technique in Australia, and has patents pending in the US and Europe.
Proteomics Technology to Match Sample Prep Expertise
Europroteome has linked its sample pool with a suite of proteomics tools. At the company’s 13,500 square-foot research facility in Hennigsdorf, the company has installed systems for separating proteins using 2D gels and LC-MS, identifying them with MALDI-TOF mass spectrometry, and studying their interactions using Biacore’s surface plasmon resonance technology, Reymond said.
In addition, Europroteome performs experiments using Ciphergen’s affinity arrays, and has access to higher-end mass spectrometry systems through collaborators at the Max Planck Institute for protein folding and enzymology in nearby Halle.
And the company is not just relying on its proteomics platform to collect information on the molecular biology behind its patient samples. Europroteome has also designed experiments for analyzing mRNA expression from the same samples using technology from Agilent and Rosetta Inpharmatics, now owned by Merck.
Europroteome has bankrolled its efforts through equity investments from its non-exclusive, first drug development partner BioMérieux, a Marcy l’Etoile, France-based company that also bought two of Europroteome’s early diagnostic targets for colorectal cancer in the fall of 2000.
Almost a year later, Europroteome raised Euro 9.5 million ($8.2 million) in an oversubscribed second round of private financing. The funds allowed the company to build a management backbone and “develop a more conventional company,” Reymond said. Last year, Europroteome attracted Dr. Ulrich Traugott, a former top-executive at Glaxo-Wellcome Europe, to lead the company.
In the short term, Reymond said the company is looking to license its OncoAssist algorithm to pharmaceutical companies, who might use the technology to help define patient cohorts that would respond to a given therapeutic. Europroteome is also looking to find pharmaceutical partners to further develop its findings around caveolin-1, a protein associated with tumor suppression. With revenue from successful partnerships, the company intends to “go as far as possible” with its own target development, Reymond said.
But just as an observant hunter does not shoot everything that moves, Europroteome does not intend to overburden itself with hundreds of drug targets that it cannot validate as having functions related to disease. “At Europroteome you will never hear anybody telling you that we have 150 potential targets,” Reymond said. “We have a max of three or five, but the right ones.”