DNA can form structures other than just the classical 'helix' shape. For example a https://en.wikipedia.org/wiki/G-quadruplex which form at the end of the helix.
This paper shows that deleting one 'TET' gene (https://en.wikipedia.org/wiki/TET_enzymes) leads to an increase in certain types of cancer. The protein encoded by that gene removes methyl (-CH3) groups from DNA, in a process called demethylation.
The 'opposite' process of methylation, in this case, is performed by the enzyme DNMT1 - and if you delete both TET and DNMT1 genes, there are less of these non-helical DNA structures and less cancer.
So the conclusion is that an imbalance between metylation and demethylation of the DNA can lead to cancer.
We know that if a group of specific enzymes (TET) are disabled for some reason, you have increased cancer rates. We also know that in some cancers you observe unusual structures in the DNA called G-quadruplexes and R-loops, which are different from the usual double-helix. We also know that these unusual DNA structures destabilize the genome, leading to more mutation which in turn leads to more cancer.
What this paper showed that disabling these specific TET enzymes lead to an increase in these unusual DNA structures and to cancer. So it establishes the causal relationship between the lack of these enzymes and the unusual DNA structures.
What I find interesting is that the function of the TET enzymes we know, demethylation of DNA doesn't have anything direct to do with the formation of G-quadruplexes or R-loops.