IARC 50th Anniversary Conference
Genomic interrogation of tumour samples reveals essential targets
Dr Roland Eils - German Cancer Research Center (DKFZ), Heidelberg, Germany
When it comes to cancer genome sequencing, so where we try to understand the genetic basis of a particular tumour in one individual, we are producing an awful lot of data. So if you want to have a comparison, so the kind of data which are produced, is about the size which would require ten iPhones or smart phones to store your data. Now if you look at thousands of individuals, as we are doing in Heidelberg, the type of data throughput is at about the scale of Twitter which is like 10TB of data per day.
Like finding a needle in a haystack?
Absolutely, you have a massive amount of data but there is only a tiny little bit in this data which might be relevant for your particular cancer and for the way how we will treat this tumour.
How are you putting these data to use?
First of all it’s just kind of data crunching, so without any background knowledge you try to find repetitive patterns, recurrent patterns, in those big data sets but this is only the first step. The second step, which is much more demanding and much more difficult is that you have to put those findings into a context of the scientific literature, the clinical findings about the kind of tumour you are affected with. That requires a lot of interactive work between bioinformaticians on one side and clinical oncologists or physicians on the other side.
Have you discovered anything that is of practical use?
At the moment we are rounding out cancer genome sequencing into our comprehensive cancer centre and we have sequenced around about 1,000 cancer patients. Our biggest surprise, positive surprise, in about two-thirds of all the patients where we sequenced their tumour we could find a genetic alteration, so a genetic lesion, where a drug option is available for treatment. That had direct consequences for the treatment or the way how we treat those cancer patients.
How will this be practically exploited?
It still requires a lot of interaction between computational scientists on the one side, so bioinformaticians, and clinical oncologists because you cannot just look at the data and derive data-driven results out of this cancer genome sequence data and use this as a recommendation for treatment. You really have to interpret this data in a context.
What is your message to cancer doctors?
If you have a cancer patient where you’re not very sure how to diagnose and treat this particular tumour, find a comprehensive cancer centre where cancer genome sequencing is offered as an opportunity and try to leverage upon the existing technology and expertise in those centres.
Is it possible to scan a cancer and figure out how to target it with existing and accessible drugs?
Absolutely. As I said, in two-thirds of our cancer patients which we have sequenced we could find a genetic alteration where a drug option which is a licensed drug option is available. In about a quarter of all cases where we found such a drugable lesion our recommendations were taken up by the physicians and were directly translated into a therapy.
What is your take-home message?
We should invest more into understanding the molecular mechanisms of cancer in the clinical practice environment and take leverage of the existing technology and also developments, both on the competition but sequencing and clinical oncology side.