Exploring molecular phenotypes of incident breast cancer through population-based proteogenomics

Proteins play a key role in malignant transformation and represent an important hub for oncogenic signaling. Elevated levels of circulating proteins could potentially be identified years before the onset of cancer and its detection by current diagnostics.

In this on-demand SelectScience^® webinar, join Dr. Anders Mälarstig, Director of Target Sciences Pfizer Worldwide Research and Development, as he explores the power of proteomic discoveries, and the integration of proteogenomics to generate key biological insights into risk prediction that may lead to translational applications for breast cancer prevention.

Read on for highlights from the live Q&A session or watch the webinar on demand, at a time that suits you.

What are your next steps for the analysis of the 3,000 proteins in the Kalmar study, and how do you plan to validate any of these findings?

AM: With a very data-rich analysis such as this, we have to think very carefully about how well disciplined we are with the analysis. What I find the most exciting about expanding the part of the protein that we're able to look at in circulating blood, is that it opens up opportunities to not just study single proteins, but also see where classes of proteins converge on the same pathways. That is one of our key steps, to classify the different proteins that we now have, and to reconstruct pathways to see if any of those give a clearer picture on how we can separate the breast cancer cases and controls.

In addition to that, we are going to investigate the genetic variance linked to all of these 3,000 proteins. Essentially, we’ll integrate genomics information with the plasma protein levels, and map the protein quantitative trait loci and test using the Mendelian randomization method to see if any of these proteins are also likely to have a causal role in breast cancer risk.

The third area that I want to mention, is that we are very interested in the question of whether certain breast cancer subtypes would elicit a greater response that we can measure at the proteomic level. Part of the analysis also involves classifying the breast cancer patients into five different subtypes and studying the protein biomarkers in relation to those subtypes.

Of course, aggressiveness is another measure that we have in Kalmar. If breast cancer has been diagnosed in between the mammography visits, or a tumor has been detected between two mammographic examinations, this would indicate that it is a fast-growing tumor that might be more aggressive.

Is this sort of framework commonly being used in drug discovery and development today? If not, how close are we to implementing this type of approach for identifying drug targets across the pharma industry?

AM: The genetics field has been very collaborative ever since the first Genome-Wide Association Studies were conceived and executed by the Wellcome Trust. SCALLOP (Systematic and Combined AnaLysis of Olink Proteins) is one example of a genetics collaboration that has extended to proteomics and allowed us to ask questions that we couldn't before.
Thanks to SCALLOP and other research in this area, both pharmaceutical companies and the biotech industry are becoming more aware of the Mendelian randomization method. It has reached quite a lot of press in the past few years, and it's becoming increasingly recognized as a key tool to define targets for therapeutic modulation.

To answer the question directly, we are very close to implementing it, but there is still a lot of development that will certainly pave the way for more drug targets being discovered in this way. It is also expected that the community will keep collaborating around this question.

In order to detect the presence of tumor proteins in the blood, what is the minimum size of the tumor that's needed in order to perform early screening?

AM: This is exactly the type of question that we are trying to answer now with the Kalmar cohort. We had some samples that were taken up to three years before the diagnosis, and some samples that were taken only three months before the diagnosis, and I think it logically follows that it would be easier to detect the profile in the samples with a shorter period of time between blood draw and diagnosis. This is something that we are looking at in the Kalmar study and we are using different cutoffs to be able to understand precisely that.

How important is the half-life of a protein?

AM: If you're measuring a dynamic change that is triggered by either a medical insert or drug, the dynamics of the protein will matter. With the selection of proteins that we have looked at now, especially with the Olink Explore 3072 panel, there will be some proteins that are only expressed under certain conditions, certain disease states, or other states, and in this case, the protein’s half-life can be important for the detectability.

Could liquid biopsies eventually eliminate the need for tissue biopsies, and is it feasible that we could get all of the information we need from plasma?

AM: This depends on what you are trying to capture. If the aim is to get sufficient information to inform a medical course of action, then perhaps looking at the blood proteome will be sufficient, at least for certain cancer types. At the same time, we are in the era of precision medicine and matching the drug to the specific tumor subtypes is critical.

I'm hoping that we will see more studies where there is a combination of RNA or DNA sequencing of somatic cells in the tumor, and then in parallel, to look at the blood proteome and see how those two match up. It may well be possible that the blood proteome has the information that we need, but we have to carry out some very well thought out studies to address that.

SelectScience runs 10+ webinars each month, discover more of our upcoming webinars>>