The reason animals cannot be used to predict human responses to drugs and disease is that both are examples of evolved complex systems. Because of the characteristics of evolved systems and complex systems, very small differences in the genomes result in profound differences to responses like drugs and disease. Not every drug acts differently in every species, but there are enough differences that animal models per se are of no predictive value. Nevertheless, there is still money to be made pretending that animal models do have predictive value.
Rudolf Jaenisch, creator of the first transgenic mouse in 1974, has now invented a process whereby several genes can be altered in both copies of a gene. In a press release from MIT, Jaenisch stated: “This new method is a game changer. We can now make a mouse with five mutations in just three to four weeks, whereas the conventional way would take three to four years. And it's rather straightforward, probably even easier than the conventional way.” Yes, they can make such mice but that does not mean the mice will be of predictive value for human response to drugs and disease.
A press release from UCSF states: “Epilepsy cured in mice using brain cells. UC San Francisco cell therapy raises hope for severe human forms.” The scientists implanted medial ganglionic eminence (MGE) cells in the hippocampus. Scott C. Baraban, PhD and William K. Bowes Jr. Endowed Chair in Neuroscience Research at UCSF stated: “Our results are an encouraging step toward using inhibitory neurons for cell transplantation in adults with severe forms of epilepsy. This procedure offers the possibility of controlling seizures and rescuing cognitive deficits in these patients.” I am sure that this technique worked in mice, as claimed, but again, despite the claim that this raises hopes for humans, in reality it should not.
A press release from Albert Einstein College of Medicine announced: “Discovery may help prevent chemotherapy-induced anemia.” The drug discovery was in mice. And finally a press release from Lund University states: “New mouse model confirms how type 2 diabetes develops.” The release continues: “Researchers at Lund University in Sweden have developed a new mouse model that answers the question of what actually happens in the body when type 2 diabetes develops and how the body responds to drug treatment. Long-term studies of the middle-aged mouse model will be better than previous studies at confirming how drugs for type 2 diabetes function in humans.” The article referred to is .
Human-based research is locating genes that are relevant to humans.
In a press release, The Universitat Autonoma de Barcelona announced that by studying humans, scientists discovered “a gene that controls 3 different diseases.” Scientists have “discovered that a good number of our antibody genes, how well they operate and, potentially, what they fight off, actually vary from person to person. That means even though drugs, treatments and vaccinations are designed to treat whole populations our response to them could be individualistic. After completely sequencing the immensely repetitive DNA in the human genome's one million nucleotide-long immunoglobulin heavy (IGH)-chain locus, these scientists have also found ethnicity influences immunity.”  A similar study  found that how well people respond to the smallpox vaccine is determined by their genome.
A recent editorial in Nature Medicine  stated: “A recent study showing that mice do not reproduce the patterns of gene expression induced by human inflammatory disease has provoked renewed discussion of the validity of animal models in translational research.” This is referring to the Seok study  in Proc. Natl. Acad. Sci. USA. The editorial continues: “Highlighting just how poorly the mice reflected the human disease, the correlations of the gene changes in the mouse models with their human disease counterparts came close to those expected by random chance alone.” (Emphasis added.) I have frequently stated that studying animal models yields results similar to what would have been expected from random chance. The editorial continues by saying that mouse models must be good for something and that animal models are valuable but offers no hard evidence that such is the case. The editorial ends with the following: “Rather than overrelying on animal models to understand what happens in humans, isn't it time to embrace the human 'model' to move forward?” 
Not if that means disrupting the status quo that so many universities and researchers profit from.
1. Wang, H., et al., One-Step Generation of Mice Carrying Mutations in Multiple Genes by CRISPR/Cas-Mediated Genome Engineering. Cell, 2013.
2. Hunt, R.F., et al., GABA progenitors grafted into the adult epileptic brain control seizures and abnormal behavior. Nat Neurosci, 2013. advance online publication.
3. Lucas, D., et al., Chemotherapy-induced bone marrow nerve injury impairs hematopoietic regeneration. Nat Med, 2013. advance online publication.
4. Omar, B.A., et al., Enhanced beta cell function and anti-inflammatory effect after chronic treatment with the dipeptidyl peptidase-4 inhibitor vildagliptin in an advanced-aged diet-induced obesity mouse model. Diabetologia, 2013: p. 1-9.
5. Bogliolo, M., et al., Mutations in ERCC4, Encoding the DNA-Repair Endonuclease XPF, Cause Fanconi Anemia. American journal of human genetics, 2013. 92(5): p. 800-806.
6. Watson, C.T., et al., Complete haplotype sequence of the human immunoglobulin heavy-chain variable, diversity, and joining genes and characterization of allelic and copy-number variation. Am J Hum Genet, 2013. 92(4): p. 530-46.
7. Kennedy, R.B., et al., Transcriptomic profiles of high and low antibody responders to smallpox vaccine. Genes Immun, 2013.
8. Editorial, Of men, not mice. Nat Med, 2013. 19(4): p. 379.
9. Seok, J., et al., Genomic responses in mouse models poorly mimic human inflammatory diseases. Proceedings of the National Academy of Sciences of the United States of America, 2013.