This can happen that some particular difference between species makes inferring risk of a molecule inaccurate. The real LD50 of nicotine is probably a good bit higher than thought now, but we can't just go poisoning people to really find out so we rely on animal models.

I suspect that MDMA and related are somewhat physically harmful but is it more neurotoxic and has a greater addictive potential than legal drugs like alcohol? I don't think so.