I think you are missing the fact that it looks like they only tested CTx in patients with cancers where bone mets are reasonably common. Thus in the non-prostate cohort where they tested Ctx, over 60% (46/73) in fact had bone mets. So it's not too surprising that the effect was similar.
I would suggest that it is non-credible that such occult bone mets would be of same effect on CTx as bone scan observable bone mets.
I would suggest that an interesting metric would be the average beginning CTx of the no-mets group vs of the observable mets group. I'd bet it is much closer to normal. Further I'd wager that Cabo is working to shut down the normal bone metabolic pathway even in patients with normal bones and the open question is how much anti-cancer effect is created by interfering with a pathway that bone mets normally exploit. Will it kill the cancer? Mitigate pain? ... I can make reasonable arguments both ways.