For folks who really, actually, truly think that there's a difference between murder and not preventing a death when the means and knowledge to do so are available, I don't think there's anything you can say, since a logical consequence of this belief is that there's no moral imperative to do anything when an asteroid is about to kill everyone.
But for those who merely profess this belief for in-group reasons, you can pull the conversation sideways, talking about how you'd probably want to prevent a small rock from falling on your head, and how if there was a really big rock falling on a bunch of people's heads, you'd want to work together to prevent it from falling on your head as well as theirs, and if there was a really big rock, you'd do the same, and if there were an asteroid you want to work together with the whole world, only the asteroid is metaphorical.
Show them that their own belief system already prioritizes existential risk.
For folks who really, actually, truly think that there's a difference between murder and not preventing a death when the means and knowledge to do so are available, I don't think there's anything you can say, since a logical consequence of this belief is that there's no moral imperative to do anything when an asteroid is about to kill everyone.
Doesn't follow. Even if the two are different, this doesn't necessarily mean that there's no moral imperative in regards to the latter, it may just be a different or lesser moral imperative.
Many people on Less Wrong believe reducing existential risk is one of the most important causes. Most arguments to this effect point out the horrible consequences: everyone now living would die (or face something even worse). The situation becomes even worse if we also consider future generations. Such an argument, as spelt out in Nick Bostrom's latest paper on the topic, for instance, should strike many consequentialists as persuading. But of course, not everyone's a consequentialist, and on other approaches it's far from obvious that existential risk should come out on top. Might it be worth to spend some more time investigating arguments for existential risk reduction that don't presuppose consequentialism? Of course, "non-consequentialism" is a very diverse category, and I'd be surprised if there were a single argument that covered all its members.