That's one of those kinds of questions that I think can only be answered experimentally. The cascade of miRNA effects could very well affect telomorase, and beyond that, most of the contents of the exosomes remain unknown. In fact, I've seen a few reports that suggest that miRNAs only constitute about 1% of the total RNAs in the exosomes (This has led me to my current investigations into the abundance of other short non-coding RNAs found within). Also, who knows what other mechanisms that we are completely blind to that might be involved, like miRNAs such a short time ago.
Personally, I do my best to take definitions like the 'hallmarks of cancer' with a grain of salt, as definitions can obscure the important assumptions that underpin them, or they might lead you down the garden path. Of course, such definitions are useful for discussion but if you focus upon them too much, you find yourself having philosophical or taxonomic discussions, or start saying silly things like 'junk DNA'. My approach is to just wonder within the realm of what is physically possible, and if something seems physically possible, then test whether or not it is so.
One thought that I have been having regarding these findings, is whether or not all metastasis is actually cellular. Is it possible that metastatic tumors could originate via exosomal influence? After a certain exosome load, might the body start spontaneously sprouting tumor? Probably much more likely the exosome influence creates fertile ground for metastatic cells, but at this point we really can't exclude the possibility. That said, I am not sure the extent to which met tumors have been exhaustively linked to the parent tumors. I work in glioma, and they are rarely metastatic.
My guess is that after a certain point, these converted cells could produce progeny that fit the bill, especially once they start acquiring chromosomal aberrations. I actually had an astrocytic cell line turn very tumorigenic on me after a transfection with one protein. I've since performed a miRNA array on the parent and the tumor, in addition to three other rat gliomas. Interestingly, the astrocytes that turned malignant have a miRNA fingerprint that is much closer to the three other tumor lines than it is to the parent cells. If you consider the hundreds, if not thousands, of targets for those miRNAs, that's pretty striking. But then again, Dolly the sheep was a mammary gland cell. If epigenetic influence can make a mammary nucleus build a sheep, then I would imagine that non-tumorigenic cells can be epigenetically coerced into the real deal.