The "extra chromosomal" part of ecDNA means that the gene is located on a circular plasmid, that copies on its own. In normal human genetics, we think of genes having two copies, one per chromosome. In cancer (and pre-cancer) that all goes out the window. There are large chromosomal rearrangements resulting in the number of the copies of genes going up and down.

Having a higher or lower copy number of a gene can change its dosage: how much of the protein gets expressed. For chromosomal changes, changing the copy number and gene dosage takes big mutational events. But with ecDNAs, it's extremely flexible to dose the amount of the gene. Because the plasmid gets copied some number of times inside the cell, and the copy number could be very high or very low.

And now, when a cell with an ecDNA in it divides, the copies get split into the two daughter cells with a stocastic amount in one versus the other. Which is another time that the gene dosage can be optimized for greater cell division. Those cells that divide the most are those that have the perfectly tuned about of the ecDNA for the right gene dosage.

This is a pattern in cancer that used to be in text books in the 1980s, then fell out of textbooks as its significance wasn't fully understood and it didn't seem to provide many therapeutic options. The first time I saw it again was in glioblastoma samples from The Cancer Genome Atlas that had whole-genome-sequencing and could find stuff like this easily. The people who found it had not been around in the 1980s to read those textbooks!

Now, we have more diagnostic tools that can detect these ecDNAs in, say, blood. And because they tend to be at very high copy number, it can be easier to detect these in blood than other DNA from cancer cells. Also, there's potential for these ecDNAs to hop from one cell to another, without direct descent. Which has concerning implications for controlling cancer, perhaps.

So, that's all background. What's new to me in this paper (and I haven't looked at ecDNA in a decade, so maybe not all of this is new to the field): ecDNA is showing up in pre-tumors. For the gene EGFR, these ecDNAs show up without one of the common mutations to EGFR that usually only happen in cancer (called EGFRv3). And the ability for the ecDNA to accumulate with a non-mutated (i.e. wild-type) sequence means that there's many more chances for the EGFRv3 mutation to happen.

This also provides a basic survey of what's out there in a variety of pre-tumor and tumor samples. Which helps to understand what these things are.

I'm not sure why this was posted to HN. I love it, but I can't imagine that many people get much out of this!!

Extrachromosomal DNA–Driven Oncogene Evolution in Glioblastoma