Mostly this: targetting the metastatic process (that is the dissemination of cells originating from the primary tumor) is in most cases irrelevant because a) in the great majority of cases of cancers that will eventually lead to death, the diagnosis occurs when a tumor has already metastasized. This approach won't cure pre-existing mets, which are generally responsible for the lethality of cancer. And b) if you don't have mets yet, in the vast majority of cases, surgical removal of the primary tumor is the obvious solution. The only application that I can see is if you have a non-metastatic tumor which is not operable. It happens but is certainly by far not the majority of cases. So implying that you might stop 90% of cancer deaths is just plain wrong.

There are additional mechanistic reasons but I think that alone is sufficient to relegate this process quite low in terms of likelihood of sucessful prevention of cancer-related death.

>However, it is what the article implies.

No, it doesn't: while it is true that cancer death is due to metastasis in 90% of cases, this approach wouldn't affect pre-existing metastasis at all (it only affects the motility of cancer cells, so the propensity of a cancer to metastasize). So if you have metastasis at the time of diagnosis, which is the case in most deadly cancer, this approach is most likely completely useless. And if you don't have metastasis at the time of diagnosis, this approach is also in most cases irrelevant, because the way to go is simply surgical removal of the primary tumor. Diagnosis without mets is generally refered to as stage 1 cancer and often has very low mortality, with some exceptions.