This is a very interesting observation which is probably (because of the symmetry) equivalent to

k = sqrt(q*r)

I have my doubts about existence of a synthetic proof, but, even with trigonometry, the above needs some perseverance.