I think it's a case of comparing enough things until something shows up. It looks like being Met/Met homozygous for rs4680 resulted in significantly higher cortisol when compared to the heterozygous type, but not when compared alone to the other Val/Val homozygous type. And the mean values were the same for the IgE types, but they're making a big deal out of high out-liers in a combined Met/Met and Met/Val group while sort of ignoring that there were a ton of very low values as well.
On the plus side, the patients had both CCC and Fukuda ME/CFS.
But overall, I don't think they've found anything of particular interest, if anything at all.