The polio virus receptor (aka CD155) is on many many different cells, not just tumor cells. The virus they're using here should be able to enter most of them. There's little to no cancer specificity there.
Having just scanned the article, I think it works similar to how the HPV based oncolytic viruses work.
So - a normal cell has tumor suppressor genes, the two main ones being p53 and RB. They're both multifunctional, but generally act to stop cell growth or cause apoptosis (cell death) if things seem to be going wrong. There's a very long list of cancers in which either or both of these genes are mutated, including glioblastoma.
link. (TP53 at the top)
When HPV infects a normal cell, it makes two genes, (among others) E6 and E7, which bind to and inactivate RB and p53. This allows (or causes) the cell to grow and makes it more difficult to kill. Rarely it can even make the cell cancerous.
Now, what happens when you make an HPV without E6 and E7? It can't grow in a normal cell (even though it can enter it) because it RB and p53 are there stop it. It's an abortive infection. But if it gets in a tumor cell it can grow due to no functional p53/RB, and eventually lyse the cell. Thus, the cancer cell is permissive to the virus, normal cells aren't. That's where the specificity is.