As I was saying before I got sidetracked onto the Brambell receptor, Geoff Hale was telling us about the half lives of antibodies. Zenapax, Herceptin and Avastin last for the expected three weeks, and rituximab for almost three weeks, but other antibodies, like Remicade and Simulect last for only a week. This seems mainly to be caused by the antibody being removed from the circulation by binding to its target. Typical therapeutic levels of antibody vary from as little as 0.1 to 10 micro gm/ml, while total IgG levels are 100,000 times this. This is obviously a technical challenge for measurement, but ELISA assays can be designed.
He then went on to talk about Campath. The CD52 antigen is an unusual molecule. It is a glycoprotein (meaning it is made from amino acids and sugars). The peptide element is only 12 amino acids long: from the C terminus it goes ser-pro-ser-ser-thr-gln-ser-thr-asp-asn-gln-gly. The C terminus is linked to a glycosylphosphatidylinositol (GPI) anchor (many CD antigens are linked to the cell surface by GPI anchors). At the asn residue is a large carbohydrate, rich in sialic acid and therefore strongly negatively charged. Campath recognized an epitope containing the 5 amino acids nearest the cell surface. Geoff has developed a flow assay using either a T-cell or B-cell line as substrate to satisfactorily measure blood levels.
In CLL, as you might expect, highest blood levels were achieved in those patients with lowest blood counts and response was related to levels achieved. Although iv Campath gave the highest blood levels quickest, the same levels could be achieved by subQ Campath, eventually. In patients with negative tests for minimal residual disease, the half life of Campath was about 3 weeks.
Geoff presented similar results for rituximab blood levels. What he is saying is that we ought to adopt a more scientific approach than giving everybody the same dose. With antibodies being so expensive, it would make sense to tailor the dose to individuals according to the blood levels achieved.
Peter Sonderman from Glycart gave a talk on the importance of glycosolation in monoclonal antibodies. As you remember, the sugar molecules on Campath were mentioned by Geoff Hale in his talk. It is important to remember that the monoclonal antibodies that we use to treat patients are artificial creations. In order to produce them in the quantities that are needed for treatment the pharmaceutical companies have genetically engineered them so that they will grow in cell lines like Chinese Hamster Ovary (CHO) cells. Unfortunately, CHO cells lack the enzymes that put the right sugars on the antibody. Some years ago, Martin Dyer made the observation that the current Campath wasn’t as good as the Campath1H that he got from Geoff Hale and Herman Waldmann. The difference was that the original Campath was made (as antibodies should be) in plasma cells, but Alemtuzumab is made in CHO cells. He began to suspect that it was all down to the glycosolation.
The Glycart technology involves transfecting the gene for an enzyme, GNTIII into the CHO cells. This catalyzes the addition of N-acetylglucosamine (GLcNAc) to the string of sugars (in the diagram it bisects the two mannoses), and this reaction prevents the addition of fucose to a different GLcNAc. I don’t pretend to understand this, but the effect is to make the antibody bind 50 times more efficiently to the human FC gamma IIIa receptor (this is the important receptor on NK cells that works in ADCC). Sure enough, Campath that has been treated in this way is far more effective in an ADCC assay, but does this matter clinically?
The evidence that it does comes from some clinical experiments with Rituximab. There is known to be a polymorphism in the FC gamma III receptor. At position 158 there may be either a valine or a phenylalanine. Only 10-15% of the population are homozygous for valine at this site, but these are specially favored if they get follicular lymphoma, because their response to rituximab is 100% as opposed to 67% for those with phenylalanine. Unfortunately this does not seem to be the case for CLL, which is why we believe that ADCC may not be an important mechanism for CLL and rituximab.
Nevertheless, applying the Glycart technology to both anti-CD20 and anti CD52 makes them much more effective in the test tube, and a trial has just begun with a glycarted anti-CD52 in CLL. Preliminary results are encouraging.