what is the chance of these LoF or PVT variants to be causal, instead of only associated with the phenotype ?

A first thing you can look at is their frequency. If your disease is very rare, but your LOF variants are common (which is usually the case in GWAS) then it is unlikely to be causal. Second, it's worth checking if you can find unaffected individuals without the disease, which would argue for an association rather than causal, depending on the penetrance. The third question could be to see if the proteins can be linked to ocular diseases. A protein only active in sperm, or kidney, or muscle, is unlikely to be your variant of interest.

But eventually, you'll have to do some functional validation in the lab to show the role of this variant in the disease.