There is no doubt that rituximab was the drug of the decade as far as sufferers from lymphoid malignancies are concerned, but the big question is why does it work? After all, I spent nearly thirty years testing monoclonal antibodies as treatments for lymphoma with very disappointing results. Studies using anti-idiotype, anti-CD5, anti-CD19, anti-CD22, anti-CD23 and anti-CD37 were all very disappointing, but by a lucky chance the anti-CD20s were effective. Why?
We know that the CD20 molecule is an unusual one, but there is one loop that stands out from the cell surface, though it never gets far from the surface. It is against this loop that most of the antibodies (including rituximab) react. There is a much smaller loop (on the left in the diagram) that hardly protrudes from the surface at all and it is against this that the antibody ofatumumab reacts.
My ex-colleagues in Southampton, Mark Cragg and Martin Glennie have been beavering away at this topic for years and in three recent papers that have described what they think is going on (Lim et al Haematologica 2010; 95:135-143; Beers et al Blood 2010; 115:5191-5201; and Beers et al Semin Hematol 2010; 47:107-114.). In this article I will try to distill what they have discovered into simple language.
Simply binding an antibody to a leukemia cell isn't enough to kill that cell. Usually some sort of effector mechanism is necessary to kill the cell. When we experiment with B cells and antibody in a test tube, it becomes apparent that there are a number of means of killing those cells. The easiest to demonstrate is Complement Dependent Cytotoxicity (CDC). Complement is a series of proteins labelled C1 to C9 which form a cascade of reactions eventually ending up by punching holes in the target cell. When you have a blood transfusion and get the wrong blood group it is CDC that destroys your blood cells (and may kill you), but it is not clear that this is an important mechanism inside the body even if it does work in the test tube.
You can destroy Complement by heating serum to 56 degrees Celsius or in an animal by injecting Cobra Venom Factor. In this way you can determine whether CDC is an important mechanism. In general terms, most investigators do not think it is an important mechanism in human antibody therapy.
Not all anti-CD20 antibodies permit CDC and Cragg and Glennie have divided monoclonal anti-CD20s into those that do (Type I) and those that don't (Type II). Rituximab is a Type I antibody. Type I antibodies have other characteristics, chief of which is to redistribute CD20 into lipid rafts which are then internalized, thus clearing the antigen from the surface. This process is known as antigenic modulation, which I have written about earlier.
There was much confusion about antigenic modulation, and it has always been said that CD20 is one of the few antigens that doesn't modulate, which is why antibodies against it work so well. That turns out to be true for some types of lymphocytes and not for others. Cells from diffuse large B-cell lymphoma and from follicular lymphoma and from some lymphoblastoid cell lines are poor at antigenic modulation, whereas other cell lines and cells from mantle cell lymphoma are much better at it. Best of all are CLL cells. Normal B cells are also good modulators.
Antigenic modulation is distinct from antigen shaving in which the antigen/antibody complex is skimmed from the surface of the malignant B cell by macrophages, usually in the spleen. With antigenic modulation the antibody is internalized by the malignant B cells. Note that both these mechanisms consume the monoclonal antibody and may account for the more rapid than expected clearance of antibody from the circulation.
Most people believe that the most important method of killing that rituximab uses is ADCC which stands for Antibody Dependent Cellular Cytoxicity. This process uses other cells (either NK cells or macrophages) to kill the tumor cell. When the antibody latches on to the CD20, the other end of the molecule sticks out away from the cell surface. This is known as the Fc portion of the molecule (the name comes from the 1950s when immunoglobulin molecules were broken up by various chemicals; the Fc portion was the fraction that could be crystalized). NK cells and macrophages have receptors that recognize Fc (called Fc receptors and labelled CD16, CD32 and CD64). Certain individuals have minor molecular variations (polymorphisms) of CD16 and CD32 which make their Fc receptors less effective. If these people get follicular lymphoma then rituximab works less well than in people without the polymorphisms. However these polymorphisms do not affect the outcome of CLL patients treated with FCR, suggesting that rituximab may act differently in CLL.
However, Complement activation may still be important. It seems likely that activation of Complement is the cause of the acute 'allergic' reaction caused by the first dose of rituximab, which is especially bad where there is a high white count. It is my opinion that the anaphylactoid fragments of Complement, C3a and C5a are responsible. Furthermore, there is evidence that both C3b and C5b inhibit ADCC.
What about Type II antibodies? There aren't very many. The best known is tositumomab, the antibody that is labelled with radioactive iodine in Bexxar. These antibodies do not activate complement and do not push CD20 into lipid rafts. Neither do they cause antigenic modulation. But do they kill tumor cells? They can certainly activate ADCC, but they also have another mechanism of cytotoxicity. This form of killing involves homotypic adhesion, which simply means that two tumor cells stick together. It is known that the process requires cholesterol and is energy dependent. It involves mitochondria moving within the cells to the area where they are in contact. Cell death is caspase independent, which means that it doesn't use apoptosis and is therefore likely to be effective in p53 deficient tumors. The killing requires lysosomes which swell and release their contents into the cytoplasm and extracellular space. Lysosomes are spherical organelles that contain enzymes (acid hydrolases) that break up endocytized materials and cellular debris. If they release their enzymes into the cytoplasm, it's like letting a tiger out of its cage. Something's going to die - in this case, the cell.
Of the new antibodies available in the clinic or in trials, ofatumumab is a Type I antibody whose ability to stay on the surface (because it goes for the little loop, not the big one) for a long time negates the negative effect of antigenic modulation. On the other hand, GA101 is a type II antibody. It has good direct cell killing and its Fc has been optimized by adding the correct sugars to the amino acids so that ADCC has been improved.