Again I think we're getting finicky here & absence of proof is far from proof of absence IMO. If horses are frequently hurt by being ridden, if we can't prove whether it is ONLY the saddle(wonder if a Perfectly Fitting Saddle even exists...) or ONLY the weight or ONLY the bad riding, that to me is not a reason to discount weight ratios. And discounting stoicness because 'you could say it about anything' is not reasonable either to me...
It is impossible to have a rational discussion if proof and evidence are to be discarded before the discussion begins. If "evidence" consists of 'I feel this is true' rather than 'we can see XYZ', then only competing feelings are left. Knowledge does not advance on competing feelings.
As for stoic - you admit that leg problems cannot be ignored, no matter how 'stoic' the horse. That certainly implies leg problems create a level of pain that overcomes any stoic beliefs the horse may have, while back problems do not. It follows that leg problems are more likely to result in an unrideable horse than back problems.
The study folks like to point out which used very out of shape horses also measured stress to back muscles. Not atrophy of back muscles, which can happen. Not damage to the spine. Stress to muscles. The chemical work indicated this happened at 30% and the massagers thought there were signs of it at 25%, but there was no sign of muscle atrophy.
Here is an example of back problems causing significant problems and what was done about it. Tom Roberts writes about his experience in World War One. During a long retreat, the horses were left saddled 24 hours a day and were breaking down. He also noted that the British Army preferred strong horses, since a 10.5 stone cavalryman resulted in a total weight (including tack and field gear) of 21 stone - roughly 150 & 300 pounds, respectively.
To solve the problem, they were ordered to stop every 2 hours, remove the saddle, and massage the horse's back for 5 minutes on each side, 10 minutes total. When that was done, the horses stopped breaking down from back problems.
When muscle tissue becomes too sore, it can be solved with rest. At 57, I like rest and ibuprofen. This is very different than damage to bone structures.
I think the evidence, along with some thinking about reasons, indicates the area of danger for a horse carrying weight is the legs. With rare exceptions, the back will suffer muscle damage. That damage, IIRC, occurs at a certain PSI - a sustained weight of 4.5 pounds/sq inch IIRC, but don't quote me because it has been a while since I've seen the figures. Thus saddle fit and weight distribution become important.
There is a good article here: How to Ease the Burden | The Trail Rider
An example of why I think it is harmful to think in terms of percentage of body weight:
If we use any percentage basis, that horse could carry far more weight than the horse in my avatar...but I'd bet on Bandit, Trooper or Mia being able to handle more rider weight, not less.