Read more

February 07, 2023
4 min read
Save

Researchers explore connections between cancer and genetic ancestry

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

Researchers at Cold Spring Harbor Laboratory have developed software that correctly deduces continental ancestry from tumor DNA and RNA.

“One can point to specific genes that predispose people to certain kinds of cancer, but in this case we’re talking about something else,” Alexander Krasnitz, PhD, research professor at Cold Spring Harbor Laboratory, told Healio. “We’re talking about how the appearance and biology of tumors differ in different people depending on their ancestry.”

Quote from Alexander Krasmitz, PhD

Krasnitz and his research team trained their software tools using hybrid DNA profiles created from cancerous and unrelated cancer-free genomes of a known background. They then tested the software’s performance against pancreatic, ovarian, breast and blood cancer specimens from patients with known ancestry. The software showed 95% accuracy in matching these hybrid profiles to continental populations.

Krasnitz spoke with Healio about the potential applicability of the software and its implications for cancer treatment.

Healio: What prompted you to conduct this study?

Krasnitz: We would like to know a lot more about ancestry and cancer. There have been a number of important publications on this from important groups. The obstacle is that such studies require a lot of data. Ideally, these will need to be systematic data, meaning you have a genomic, molecular portrait of cancer. At the same time, however, you also need a cancer-free genomic portrait of the individual.

You need to know the genome in the individual, and from this genome, you can determine the genetics of the cancer. You need to put them together. The problem is that for most of the existing data on cancer, we have the portrait of the tumor but not the matching genetic portrait of the individual. Enormous databases, such as Sequence Read Archives, are maintained by the government, as well as other very large databases. We set out to do something that would allow us to learn something about the genomic portrait of the individual from the cancer-derived data.

Another aspect of this involves biobanks or tissue archives. Typically, if you go to a large medical center and talk to the pathologist or specialist who runs the biobanks or tissue archives, you can find a cancer specimen from a tumor. However, it would be exceedingly difficult to find a specimen from the same patient that is cancer-free. Comparing the two side-by-side, you can learn about the relationship between ancestry and cancer. There are epidemiologic examples of how, for example, certain subtypes of breast cancer, such as triple-negative breast cancer, are much more prevalent in people of African ancestries.

Healio: How does your software address these concerns?

Krasnitz: Using the tools we’ve been trying to build, we can now go to that tissue archive, pull out the tissue, and analyze it genomically, through molecular analysis. From there, you can also learn something about the individual and, importantly, their ancestral background. We developed computational methods to do this.

We learned how to restore the ancestral portrait of the individual from so-called exome sequences. This is basically a DNA sequence, but it’s restricted to the coding region of the genome. We can do it even better if we have whole genome sequences, but these are a bit less common and a lot more expensive. We have also learned how to do this from RNA sequences, which are distinct from DNA sequences. We can look at RNA sequences of the individual and learn something about the ancestry of the individual in some detail from that.

We can do this from a gene panel, as well. With gene panels, you do not sequence the whole genome — you don’t even sequence the coding part of the genome; but from the coding part of the genome, you only sequence a few hundred genes. You get less information, but even based on that, you can learn something about the individual's ancestry.

In addition to the genetic aspect of the cell’s molecular structure, there are also specialized epigenetic assays one can perform, and even though the sequence that you get with that is quite limited, you can still learn about the individual's ancestry. We’ve learned all this and are now looking forward to two applications.

Healio: What are these applications?

Krasnitz: We are contributing to a study being run by our colleagues here at Cold Spring Harbor Laboratory and our colleagues in our medical partner institutions, Northwell Health and SUNY Downstate Medical Center. We have surgical specimens of colorectal cancer, but we do not have matching cancer-free tissue from the patient. This is a direct application of the software.

There is another way to apply this more broadly, not just to cancer but to any human disease. When you try to figure out a person’s ancestry from molecular data, you face two difficulties. One is that there may be cancer, and often cancer rearranges the genome in such a way that it is hard to recognize. Another issue is not all data types are optimal for learning ancestry. The examples I gave, such as RNA sequences, panels and epigenetic assays, are not ideal for this purpose, but nonetheless, you can do it. You can have, for example, a postmortem specimen of a brain from a patient with Alzheimer’s disease, but maybe you only have an RNA specimen. If you are in this constrained situation, what do you do? You can still learn from this, using other methods. That is, in a nutshell, what we did.

Healio: Could this be used to develop targeted cancer treatments?

Krasnitz: In the future, it’s quite possible. If you learn something about the patient’s ancestral background and then you have an unrelated retrospective observation, that background may indicate a greater likelihood of a certain cancer subtype. You can make clinical decisions with better information.

Healio: Is there anything else you’d like to say about this?

Krasnitz: This result would not have been possible without our first author, my postdoctoral fellow Pascal Belleau. His effort in this project has been nothing short of heroic. He came up with the key ideas, wrote thousands of lines of code and analyzed enormous amounts of data.

For more information:

Alexander Krasnitz, PhD, can be reached at: Cold Spring Harbor Laboratory, 1 Bungtown Road, Cold Spring Harbor, NY 11724; email: krasnitz@cshl.edu.