Disrupting Pharma: Personal medicine, maybe too personal?

One of the chapters in "The Innovators Prescription" by Clayton Christensen et al. is about how pharmaceutical research is going to work in the future.
One good thing about the book is that the authors clearly distinguish between what they call "precision medicine" and the current buzzword du jour "personalized medicine". For them precision medicine is about using technology to go from intuition-based medicine to a clearly analytical and data driven approach, where diagnostics, biomarkers and such are going to play a major role, and not the individual experience and, well, intuition of the MD you are talking to.

Instead of developing a single "block buster" drug that is going to do a little for a whole lot of people, they believe that the drugs of the future are going to be targeted on sub-populations of people for which the impact is going to be much larger, since the drug is going to affect pathways with clear clinical significance for that sub-population, i.e. a big bang for a smaller group of people than the ho-hum effect of the block buster.
If you have a drug that works well for a certain pathway, all you need is a diagnostic test to check if that particular pathway is the right one to target and you can be pretty sure the drug is going to work (the precision medicine part).

Question is: How do you identify the targets and the sub-population initially?
This is where the term "personalized medicine" and genotyping are typically used. The thinking is that all you need to do is to look at the genes of patients, and the variations therein, and presto, the differences tell you for which pathway to develop a drug for.

There is one little problem with this: The emerging research indicates that the sub-populations might be smaller than you think, and the big bang might also not be what everybody hoped for.

The story goes like this:
When researchers currently look at variations in the genome, they look at single changes in the sequence, a so-called SNP. You take a whole bunch of people, some healthy, others suffering from some ailment you want to investigate, and you look at known SNPs in the two populations, trying to find disease relevant SNPs.
In the ideal case, one of the changes occurs only in the sick population, but not in the healthy.

In reality, the ideal case never happens.
The more experience researchers get in running these studies, the more it looks like that the sub-populations are very small indeed, or alternatively, if you find something, a lot of people that have that SNP are perfectly fine.
Even for diseases like Schizophrenia, where it is known that genetic predisposition is increasing the likelihood of suffering from it dramatically, the most frequently associated SNPs identified so far can only explain a few percent of the Schizophrenia cases.

There might be a good explanation for this: If a large percentage of a population suffers from a disease that affects their chances of survival, over time evolutionary pressure would eliminate that sub-population, or would lead to other changes in the genome that would make an individual more robust for the effects of that single change.
So after a while of evolution running its course, what is left are disease-related genetic changes in the gpopulation that either occur only rarely (affect only a very small portion of people), or are not very significant (people have that SNP, but some other pathway can compensate for that change and the SNP in itself can not really help distinguish between the sick and the healthy).

This might explain why for some relatively new diseases, like HIV, the analysis of SNPs works well, but for "older" diseases, like hypertension, the analysis of SNPs has not really worked that well, since evolution had a chance to take care of this.

Since pretty much every pharma company on the planet seems to be jumping on the "personal medicine" bandwagon, it might be a challenge to come up with a business model that would work for a truly, very personal medicine.

References:

Need, A., Ge, D., Weale, M., Maia, J., Feng, S., Heinzen, E., Shianna, K., Yoon, W., Kasperavičiūtė, D., Gennarelli, M., Strittmatter, W., Bonvicini, C., Rossi, G., Jayathilake, K., Cola, P., McEvoy, J., Keefe, R., Fisher, E., St. Jean, P., Giegling, I., Hartmann, A., Möller, H., Ruppert, A., Fraser, G., Crombie, C., Middleton, L., St. Clair, D., Roses, A., Muglia, P., Francks, C., Rujescu, D., Meltzer, H., & Goldstein, D. (2009). A Genome-Wide Investigation of SNPs and CNVs in Schizophrenia PLoS Genetics, 5 (2) DOI: 10.1371/journal.pgen.1000373

McCarthy, M., Abecasis, G., Cardon, L., Goldstein, D., Little, J., Ioannidis, J., & Hirschhorn, J. (2008). Genome-wide association studies for complex traits: consensus, uncertainty and challenges Nature Reviews Genetics, 9 (5), 356-369 DOI: 10.1038/nrg2344

GIBSON, G., & GOLDSTEIN, D. (2007). Human Genetics: The Hidden Text of Genome-wide Associations Current Biology, 17 (21) DOI: 10.1016/j.cub.2007.08.044