Dynamic range remains proportional to FWC and read noise. This with resolution are the two essential parameters. At least at the start, a bad camera design can spoil good incoming data.
QE is basically how efficient you are at capturing photons. A sensor with a very high QE but a small FWC or high read noise would be much worse than a sensor with lower QE but higher FWC or lower read noise. All other things being equal, a lower QE means that you have expose a bit more or use a bigger collector (lens) to record the same number of photons.
Also note that R. Clarck numbers are often more specifically defined than QE in the generic sense of the term. He'll say, for example, that the green sensels have a QE of 30% for a specific wavelength. But since the green pixels occupy 50% of the sensing area, it would actually be 15% compared to an unfiltered KAF sensor.
Assuming a camera given with 60% QE as a whole, it would mean 30% in green, 15% in red, 15% in blue compared to a bare sensor. Also QE varies a lot with wavelength. See
http://www.starsensesci.com/product/product_astro/QHY9/images/KAF8300QE.jpg for example.
In that case, the 60% (30% effective at 550nm) is just a peak in green. At 700nm, in red, you are at 40%, 10% effective.
But it's true that better micro-lenses have enabled us to do more with the same flux of incoming photons, or do the same with smaller optics.
