Corky,
I think that this is an interesting article--The fact that computational methods can be used to improve antibody affinity. However, IMO all of the different methods for geneating antibodies (phage-display, transgenetics animals, CDR-grafting, computational methods, etc.) are multiple roads to the same end. I do not believe that any one method is best suited to produce the "perfect" antibody candiate. Regardless of how an AB is generated it must be validated functionally, in vitro and in vivo.
Another point to consider is that higher affinity may not translate into improved therapeutic efficacy (I wonder if this was a problem with the AMGN/ABGX antibody, in addition to it being IgG2). In some of IMCL's papers they have addressed this issue and stated that increased affinity may reduce tumor penetrating ability. In other words the affinity needs to be high enough but further increasees in affinity may reduce efficacy.
IMCL enhanced the affinity of its anti-VEGFR through testing various light-chains with the heavy chain of 2C6. In this case it seemed to work as tumor penetration may not be as critical in anti-angiogenic theray. Based on my reading of the literature, it does not appear that IMCL has pursued such a strategy (testing various light-chains) with its human anti-EGFR AB. It is unclear to me what the exact affinity of IMCL's human anti EGFR AB is? They have stated that it is similar to Erbitux; however, I seem to recall IR stating that the affinity was higher. We will see what the clinical trials show. Do you have any expectations that the IMCL's human anti-EGFR could have enhanced efficacy in addition to reduced side effects?
Regards,
biophud