Veterinarians have known for a long time that cats do not handle certain drugs as well a dogs and other animals do. Drugs such as aspirin, acetaminophen, and other NSAIDs can be much more toxic to cats than to other animals. If we give them to cats at all, they are given in very low, infrequent doses (aspirin) or not at all (acetaminophen). We have known for a long time that cats have much lower levels of the liver enzymes which humans and other animals use to metabolize these drugs. Cats are considered "hypercarnivores", meaning they are adapted to a diet that is 70% or more animal matter. Since most drugs are quite recent inventions, it is obvious that these enzymes did not evolve in response to the drugs which they metabolize but to something else.
A recent paper on PLoS One has worked out the genetics behind this enzyme difference in cats (and a few other animals). These enzymes (Glucuronyltransferases) have developed over many millions of years to deal with toxins found in plants used for food. In species such as cats that do not eat many plants, there is no selective pressure to maintain this type of enzyme function, and mutations can cause the genes coding for them to become inactive. This is what has happened in cats with these particular enzymes, with the result that certain drugs that are chemically similar to some types of plant toxins are much more toxic to cats than they are to other species. This type of issue with drug metabolism is known as a "species defect". While the specific difference in metabolism being discussed here is one of the best known and documented of these species defects, other less well documented differences occur as well.
This has important implications for not only the use of pharmaceuticals of known and controlled dosages, but even more importantly for the use of herbal medications in different species. The problems with herbal medicines are best described in David Colquhoun's Patients Guide to Magic Medicine as "Herbal medicine: giving patients an unknown dose of an ill-defined drug, of unknown effectiveness and unknown safety." The bit about unknown safety is especially applicable in the case of the types of species defects in metabolism that we are discussing here. While some of them are well known, giving an herbal medication that has many different chemical compounds in it may increase the risk of adverse reactions. In addition, the actual dose of active ingredients in herbal products is often very variable, so a product that seems safe when taken once may be toxic the next time. Just because herbs are natural does not mean they are safe, and animals have evolved a lot of natural variability in sensitivity to plant compounds, as this paper documents.