1) This is not a therapy. This is basic research that could potentially lead to a therapy, or it could merely lead to a marginal improvement. Most of science are like this, massive Deep Learning style leaps are relatively uncommon.

2) The research is about a group of chemical structures called Porphyrins, commonly found as part of a human proteins. They are apparently a good fit as delivery vehicles to send anti cancer drugs into cells. Apparently these particular chemical structures have inherent anti-cancer properties too. This is a very simplified explanation and I have not read the paper itself. This is a chemistry/biochemistry paper as much as it is a biology paper.

3) It is very difficult to explain to a layman whether a research is truly useful, or whether it will end up as a dependency (to use a software engineering term) of a bigger research. Imagine somebody made a marginal improvement to a neural network model. By itself it is not revolutionary, but the incremental gains add up over time. This is an incremental gain, a lot more work and research will be needed to turn it into a possible therapy. And even then it may hit some snag down the road that will render it useless. Science is tedious and expensive.