For Christina Bligaard Pedersen, cancer is not only a terrible disease that affects one in three of us, but it is also a huge matrix of figures on the computer screen in which she is trying to find patterns. She is a bioinformatician and close to finishing her PhD studies at DTU Health Tech, and she is also attached to the Department of Genomic Medicine at Rigshospitalet in Copenhagen. And the many figures come from tissue samples taken from patients diagnosed with breast cancer.
She receives cells from the hospital which have been extracted from the patients’ cancerous nodules and translated into data sets. She perhaps looks at 1,000 cells and 20,000 genes in a huge table, and then uses a computer tool to search for cells that resemble each other. Which are most similar and which are most different? By answering these questions, she can arrive at a subdivision, and knowledge about which cell type is dominant in the individual patient’s tumour.
Growing data volume
Breast cancer takes many forms, and the patients’ tissue is not the same either. Therefore, the research aims to subdivide the patients into subtypes and link each type with the treatment that has been shown to be most effective.
“We’re looking at the transcriptomic profile of the tumour, that is at which genes are expressed in the individual cells. If you find cells that overexpress a specific gene, which results, for example, in uninhibited cell growth, you can then consider whether it is possible in the long term to target a drug that specifically counteracts this,” says Christina Bligaard Pedersen, and elaborates:
“In addition, we can gain deeper insight into the composition of the tumour: A tumour can contain 90 per cent of one type of cells and 10 per cent of another. And if the treatment only catches the 90 per cent, the remaining 10 per cent have a free reign to continue to spread.”
Some of the tools Christina uses are based on mathematical solutions that were presented decades ago. But the area is developing by leaps and bounds in these years, not least because more data are constantly being generated:
“Previously, there was perhaps one data point for a whole tumour. Now we can have points for 1,000 single cells. Therefore, we can also perform much more complicated analyses. But this also entails enormous challenges. Multiple mathematical tools combined with statistics must be used to interpret the large data volumes. Bioinformaticians like me are specialists in this. However, we will never be able to give definitive answers and we will not replace doctors. All my findings are to be used as support for the doctor’s decision on treatment,” says Christina Bligaard Pedersen.
New tools
As an engineer, Christina is not only interested in finding a way to support the doctor’s decisions. She has also worked with developing new IT analysis tools. She has thus found a method for comparing different data sets, which should resemble each other, but do not really do so.
If, for example. gene expression is measured on the same data in two different ways, the result should, in fact, be the same. However, due to technical differences in the methods, the results may not be immediately comparable. Using existing mathematical models and combining them in new ways, Christina has demonstrated that it is possible to compare the measured data on a one-to-one basis. And this detective work has led to a very concrete IT tool, which is now available for free use.
The tool is intended for researchers and clinicians who want to have directly comparable data sets that have been collected over an extended period. This could, for example, be in connection with clinical trials or when comparing data from a new patient group with a larger, publicly available data set.
Immunotherapy for breast cancer patients
In February 2020, Christina received one of the Danish Ministry of Higher Education and Science’s Elite Research (EliteForsk) travel grants. However, immediately afterwards, the outbreak of the coronavirus pandemic prevented all travel activity. So it will not be until in early 2022—when she has completed her PhD—that she can leave for Harvard Medical School in Boston, a world leader in single-cell RNA sequencing. She will be spending two months there to continue her work with cancer and single cell analyses.
She then has a postdoc position at DTU until the summer, where she will work with some of her PhD plans, which were interrupted due to the coronavirus lockdown. This work consists of examining whether single cells can predict which breast cancer patients could benefit from immunotherapy.
Cancer cells can acquire the ability to down-regulate the immune system, and if it is possible to see that a cell has acquired this property, it may then be possible to find ways of awakening the immune system. In some cases, however, the immune system may not recognize the cancerous tissue as being dangerous at all, and then it may not make sense, because the immunotherapy will not produce an immune reaction in any case. Christina’s research should help find ways of identifying the cases where it might help to push the immune system with immunotherapy.
“I will never solve the riddle of cancer or find the ultimate cure for the disease, but it’s very satisfying that—through my nerdy approach to numbers—I can help bring us closer to a more individualized cancer treatment with better effect and fewer side effects. Going from numbers in a matrix to seeing a clear pattern and thinking that this might help someone is almost magical,” says Christina Bligaard Pedersen.