Acceptable print sizes involves much more than just megapixel count. It's the same idea that a higher megapixel camera isn't necessarily better than a lower megapixel camera - it depends upon the quality of the pixels too.
A better way to think of it is in terms of "signal to noise", like with capturing images. Signal is the picture you want to show - colours, shapes, edges, fine detail, etc. Noise is anything that detracts from that - pixellation, thermal noise, grain, abnormal transitions, etc. Longer viewing distances effectively minimise the "noise".
If you have properly sampled stars (e.g. 1''/px in 2'' seeing) with accurate tracking, then even if you enlarge the image by a *HUGE* factor it won't look pixellated. On the other hand, an image scale of 4''/px in 2'' seeing will give you square stars - which looks horrible very quickly upon enlargement.
Another factor to consider is that a CCD images' pixels are usually higher quality than for a DSLR - 16 bits for most CCDs vs 12 bits for most DSLRs. If you shoot with a mono CCD, the resolution will be much higher compared to a Bayer matrix too.
On an aesthetic level, your subject matter influences your maximum reasonable print sizes too. That's why night scape images often look spectacular, even though at the pixel level the stars are distorted, the image is noisy, etc. On the other hand, a close-up image of a single nebula requires a lot of stacking and photon capture to get the same "wow" effect.
Short answer: if an image looks spectacular on screen, it'll look spectacular in print if you use quality materials and a good print process
