GTC Biotherapeutics (fka GTCB)

[MTB]

Dew's right on target as far as parsing the intentional vs deliberate bias that might underlying the KS data 'showing' decreased mortality in the ATIII subgroup.

But the bedrock of any assessment of the 'validity' (read reproducibility, or if a Jon Stewart fan, the 'truthiness') of these results has to be in the recognition that this was post-hoc SUBGROUP analysis -- and so HUGELY suspect.

By definition, subgroup analysis does not offer statistically valid results -- and you are on damn tenuous ground to begin to think otherwise. Subgroup analysis may offer some insight overlooked in the initial study design and it might make us happy, but it is nothing more than that.

An intern-level journal club would appropiately dismiss post-hoc subgroup analysis as 'data mining' (this a perjorative), and note that the end points of a study must be prospectively choosen and sufficiently powered for that end point for the results to be valid. Subgroup analysis obviously does not meet these criteria.

If you go looking BACK through a study for 'results,' by simple random chance, you are going to find positive and negative findings -- that are in no way reproducible.

Given that ATIII has 1) viable mechanism for activity in DIC, and 2) on the basis of the prospective animal data to date, I think it reasonable that ATIII will turn out to be helpful in DIC treatment. The KS subset analysis on ATIII is intriguing, but in no way qualifies as a #3 in this list.