You are here: MEDICA Portal. MEDICA Magazine. Topic of the Month. Volume archives. Our Topics in 2012. July 2012: Personalized Medicine.
Topic of the Month July: Personalized Medicine
"We can arrive considerably faster at the patient-specific mutations"
Up to 20 terabyte of scientific
data is accumulated per patient
in our projects;© panther-
MEDICA.de spoke with tumor researcher Doctor Christian Regenbrecht, Team Leader of the “Tumor Stem Cells” Task Force at the Institute of Pathology of the Charité – University Hospital, Berlin, Germany, about this new project.
MEDICA.de: The topic of personalized medicine is of tremendous importance especially in cancer treatment. Nevertheless, physicians and scientists are still working in many different areas. Which obstacles do you currently battle against in tumor research?
Christian Regenbrecht: Aside from the ethical and legal challenges, there are technical hurdles to overcome. The challenge lies in processing enormous amounts of patient data in everyday clinical life. By now, we are able to collect more data on patients than ever before. We need suitable solutions for this, that enable analysis and evaluation in real-time as much as possible. Aside from clinical data, up to 20 terabyte of scientific data is accumulated per patient in our projects.
This is a scale of about 20 larger hard drives per patient. Generally, there are still many unanswered questions in this area. That is why we especially look into how these large data volumes can be handled and processed. Physicians who work with important patient data, need clever solutions for example, that help in bringing relevant data to the fore. Ultimately, special information criteria about a patient help the physician to reach real-time, evidence-based treatment decisions.
MEDICA.de: Recently, you started a project with the Hasso-Plattner-Institute. Among other things, it is about launching a “supercomputer“ that in the future perhaps is able to create even more targeted personalized therapy.
Regenbrecht: We cooperate with the Hasso-Plattner-Institute in Potsdam. And yes, for our purposes, this is a super computer – but it is not a “supercomputer” in terms of information technology. In their local Future SOC Lab, since May 2012 an additional computer network with 1,000 cores, consisting of 25 servers with 80 calculating engines each is available for scientific projects. This project with the pathology institute is also among them. The benefit of such a network for genome science for instance is that this way one processor does not have to successively compare the more than three billion individual base pairs of the human genome, but for this to take place up to 1000-fold in parallel.
This way we can arrive considerably faster at the individual genome characteristics, meaning the patient-specific mutations. This does not mean that the computer decides the therapy or treatment recommendation. Instead the enormous computing power can be utilized to interpret and analyze medical data in real-time for the first time ever. Normally, commercial sequencer companies in the medical field guarantee a turnover time of up to three months. This is an amount of time that is far too long to provide meaningful additional information for making a decision on a therapy. However, if in the future this can be created faster by a factor of 1000, this information could become relevant for patient treatment.
MEDICA.de: How does the practical process work, when a tumor is meant to be examined?
Regenbrecht: After we receive some tumor and normal tissue from a surgery or biopsy, for example during melanoma surgery, we extract the nucleic acids from the tissue, so DNA and RNA. DNA is sequenced and normally is available as fragments of approximately 200 to 250 base pairs. Through mapping, the fragments are aligned with the so-called reference genome, which consists of over three billion base pairs. In doing so, potential mutations are meant to be identified as the cause for the tumor. We require bioinformatics for this, which helps to calculate the best solution to assign as many sequences as possible correctly to the exact complementary piece on the reference genome.
Picture this like a puzzle with 3 billion pieces, where every piece also does not exactly fit. If you can parallelize the corresponding algorithms and calculations, the process will then accelerate by approximately the factor of the number of processors and the degree of parallelism. We are currently researching with the Hasso-Plattner-Institute, how the in-memory technology that is being researched there can -among other things-accelerate mapping algorithms.
The enormous computing power can be utilized to interpret and analyze medical data in real-time for the first time ever; © panthermedia / Manfred Grafweg
MEDICA.de: What exactly is improved by this in personalized cancer therapy?
Regenbrecht: Until now, DNA was obtained from sequencing service providers. You waited about 90 days for the results. This waiting period would now be cut down to two weeks at the most. In the future, these results could be available for the physician significantly faster to make a decision on therapy. Today, this process is still not clinical practice yet, but merely the subject of research projects. One very large project is the OncoTrack project.
However, you must not deduce the promise of a cure from personalized medicine. A human being has approximately 30,000 genes. The function and the exact interplay between many genes are not even known yet. That is why for the time being, this is a scientific sub area and fundamental research. We are currently in the process of learning how we can establish certain connections. For many years, we have tried with our international research partners to biologically understand these connections and reconstruct them in silico.
MEDICA.de: What differentiates this computer network from other systems?
Regenbrecht: For the time being, there are no other possibilities. However, you have to point out that personalized medicine actually is not a problem of bioinformatics, but that of expertise. The more we learn about the fundamentals of biology, for instance about translational medicine or systems biology, the simpler and much more predictable treatments become. The computer assists us in making personalized, individual decisions for the respective disease pattern of the patient. The more frequently general mechanisms are being identified, the better this can be illustrated on the computer.
MEDICA.de: What is going on in other countries? Is there an international trend?
Regenbrecht: Personalized medicine is up and coming, however, there are still obstacles with regard to contents and social policy that we have to overcome. Starting with this year, there are more and more professional meetings, particularly in the U.S., that concentrate on the ”big data“ topic. These will answer questions on how we can utilize the enormous amounts of data to better help patients. Internationally, this is an important subject in cancer research.
MEDICA.de: Is one goal to introduce these computer clusters all over the world?
Regenbrecht: Absolutely. For one entire day and for the first time ever, this year’s World Health Summit in Berlin, dedicates itself to the topic of health and IT. Representatives of pharmaceutical companies, patient advocates of health insurance companies and representatives of physicians and information technology will sit together and discuss, how they can jointly create more structures to implement this knowledge for the patients as quickly as possible.
The interview was conducted by Diana Posth and translated by Elena O'Meara.