Derek Lowe’s, “In the Pipeline“, had an interesting discussion on animal models and their ability to predict the efficacy of drugs in people…
Interestingly from what I have observed, being in the area, is that the success rate in the clinic of antiviral drugs against HIV, at least in the context of efficacy, is very high despite the fact that there are no animal models.
That’s right…what most people outside of the field might not realize is that the only pre-clinical “efficacy” model used for HIV drugs is cell culture antiviral activity. In almost every case bona fide activity in cell culture translates into clinical efficacy. The only thing standing in the way of a cell culture active compound and a clinical candidate is the usual ADME-Tox studies, but cell culture activity is usually (but not always) good enough to provide drug-like compounds. The caveat is that the anti-viral activity must be validated in order to weed out the “promiscuous” false-positive compounds.
This might be the reason that a lot of start-up companies are pursuing anti-HIV agents even thought the market is becoming very crowded (>20 drugs and counting)