Saturday, June 9, 2007

Not all drugs benefit all man!

While Bertalan Mesko at ScienceRoll is introducing us to the best medicine 2.0 tools, i am reviewing some results from the American Society of Clinical Oncology meeting last week.

In another example of how we need to examine patients pharmacogenomics prior to instituting therapy the SWOG (South Western Oncology Group) in the U.S. releases results of a collaborative effort with two clinical groups in Japan (Japan Multinational Trial Organization).

The researchers were interested in how these different groups metabolized certain chemotherapeutic agents Paclitaxel and Carboplatin. Now what is interesting about this study presented at the ASCO conference is the fact that they were able to isolate two gene polymorphisms responsible for these effects.

In patients with certain variations in the CYP3A4 gene, it took 2.75 times longer for their lung cancer to progress than in patients without the variations. A variation in another gene, ERCC2, appeared to interfere with how well patients responded to treatment.

The problems with this type of analysis are threefold.

  1. The study was too small a size to not need replication

  2. The study did not involve the genome of the tumors (Perhaps the DNA repair mechanisms in the Caucasian tumors were better. This would like to decreased cell death i.e. response to chemo. As with the tumors and ERCC1)

  3. The dose of chemo was not controlled (although the lower dose worked better in the Japanese)

The Sherpa Says: In order to make useful sense of Personalized Oncology we must look at genomes of both the cancer and the person. I feel that we are introducing erroneous data to confuse us. I hope The Cancer Genome Atlas will show us some better data!

No comments: