Name: Hanno Langen
Position: Head of biology research, Roche Center for Medical Genomics, Basel.
Background: PhD, University of Zurich, 1979-1988. Studied DDT-binding peptides. Postdoc, Rockefeller University, working with Bruce Merrifield, 1989-1990
Hanno Langen, head of biology research at the Roche Center for Medical Genomics in Basel, Switzerland, was an early adopter of proteomics methods in drug discovery. His lab at Roche acquired its first mass spectrometer for protein characterization in 1992 — before the field was even called proteomics.
Langen is also a senior editor of the journal Proteomics.
In 2002, ProteoMonitor spoke to Langen [PM 12-23-02] about his background and his outlook for proteomics in drug discovery. This week, we caught up with him again to discuss new developments in the field and the growing emphasis on biomarker discovery and validation.
When we last spoke, you saw a role for proteomics at various stages of pharmaceutical research, including target discovery, target validation, lead validation, toxicology, and biomarker discovery. Which of these areas has proven most fertile for proteomics? How has the role of proteomics in pharmaceutical research developed over the last few years, in particular at Roche?
The main development we have, and maybe in the whole pharmaceutical industry, is that the focus is not so much on target discovery anymore. One role of proteomics is in is target validation, and the biggest role at the moment is in biomarker discovery and biomarker validation.
It seems straightforward to come up with biomarker candidates but it seems to be a great challenge to move from candidates to validated biomarkers. What has your own experience been at Roche for validating candidates?
First of all, if you look at this from the diagnostic perspective, [diagnostic developers] are not used to attrition rates. We are used to high attrition rates from the pharma side, but not from the diagnostic side, and now the diagnostics division is facing attrition rates in the field of biomarker validation. For example, for a screening marker, you [need to] have an extremely high level of sensitivity and specificity. And to reach [that level] is extremely difficult. It think it’s OK to validate a biomarker to a certain extent to improve the diagnostic setting, but really to have it to a stage to put it on the market is where the major difficulties are.
When you ask me what is coming out in proteomics discovery that has made it to market, there is very little. This is a challenge we are facing more and more.
We see that the validation phase is the more expensive phase than the discovery phase. This also means that the discovery phase has to be done in a very diligent way so you don’t have too many false positives.
We are not using the usual 2D gel approach. We are not using image comparison. We cut out every spot on the 2D gels and compare on the basis of the mass spec level. This approach is amazingly robust and most of the proteins we discovered to be changed in our tissue approach can be verified on the western blot level. Bruker with their unique sample preparation method was instrumental for our success in this approach.
Looking back four years ago, I think it was my strategy to focus on 2D gels because the false discovery rate is much lower than the false discovery rate with 2D LC/MS-MS methodology. It’s changing now a little bit [as] the 2D LC/MS-MS approach is now getting better because the accuracy of these machines is helping to reduce the false discovery rate. Therefore this type of technology has also, in our hands, [played] a bigger role in the discovery phase.
A few years ago, people were talking about discovering biomarkers in plasma or serum. Some are now saying this is too difficult and are switching to tissues for discovery work. Do you agree?
That was fairly early on our approach. Our discovery phase was not in plasma because we have clearly seen early on that to go to a sensitivity level of [nanograms per milliliter] in plasma was not possible a few years ago. Therefore we started this with tissues, and we have several examples where we can validate that these markers are occurring in cancer patients.
The biomarkers we validated up to now are at least as good or better as known biomarkers, for example, in colorectal cancer. However, screening markers have to be even better. Maybe the combination of these markers will overcome the hurdles to enter in the screening market, which is one of the most attractive for diagnostics.
I think that the recent advances in fractionation methodologies are making it possible to go to plasma samples. The inherent problem is still that you can use only very few samples to do the discovery and that you have to go then to an intermediate validation phase. The initial validation is in the so called black-and-white situation, where you have late-stage cancer and healthy controls. To find their difference in this population group is relatively easy. Many academic groups stop at this point, but when you include other diseases in the validation phase the situation is very different, because the markers might be not specific for the disease you have done your discovery in.
You also have to include in the sample selection non-related diseases. Maybe you have found an inflammation marker that can be found in many other situations coming from the cancer. Therefore you also have to look into other diseases, not only in the healthy controls, in your validation phase. You have several sets of validation. You have to go through and show the clinical utility. And finally, I think the biggest hurdle is when you go on the market, [you have to] show that you can, for example, improve the mortality in cancer with your diagnostic test.
Now that the FDA is beginning to show more of an interest in proteomic data, what role do you see for regulatory agencies in driving the development of proteomic biomarkers?
I think it is still an early dialogue we have with the FDA. I think more dialogue with the FDA is needed. I think there is also a big interest when you say, for example, you want to have personalized medicine to show that you increase your efficacy with a certain biomarker, or you have a faster way to perform your clinical studies. But still, it is a difficult process because you have to show that your biomarker is correlated with your clinical endpoint.
I think there is an incentive now from the FDA to do this. FDA is driving this process, but also it’s in the interest of the companies to drive this process because there is quite a lot of hope in this approach.
Where do you see the greatest potential for biomarkers as diagnostic products? In what areas do you see the first products coming out?
I think you can split this in two potentials: when you go for the diagnostic industry, it’s screening markers. There is no potential for pharma with this type of markers, there is purely a diagnostic potential. And when you look at the pharma potential, then [the potential] is that you can show early on efficacy with your biomarker, security in your development path, and finally that you can reach a higher efficacy using the combination of the biomarker and the drug.
For example, Herceptin. When women have an overexpression of the biomarker on the cell surface, the efficacy is significantly higher. This is a huge potential.
Regarding the proteomic techniques you are using these days. When I last talked to you, you said you were using 2D gels and LC-MS/MS. Has that changed? What are you using now?
We are still using the 2D technology but we are also incorporating new investments in the LC/MS-MS technology, because it has additional benefits. On the other hand, we are also using now more and more antibody technologies like reverse protein arrays or multiplexed ELISA technologies because we have to validate enough samples for our biomarker candidates.