ZAP-70 again.

One of the most pressing topics to come up in Canada was the usefulness or otherwise of ZAP-70 as a prognostic factor. In fact John Byrd said that patients coming to see him with their ZAP-70 assay was the bane of his life.

Now I have already blogged about ZAP-70 once and my views on it have not changed.

ZAP-70 is the signaling molecule for the T cell receptor. The T cells receptor is made up of several chains that protruse from the cell membrane ready to be tweaked by the target molecule which the T cell is programmed to react with. It is the zeta chain that actually reacts with ZAP-70 (which stands for Zeta Associated Protein with a molecular weight of 70 kiloDaltons). When it is twitched, ZAP-70 relays the message to the nucleus of the cell where instructions are issued for the cell to divide or secrete or whatever.

In B cells ZAP-70 is not normally involved. The B cell receptor is also made up of several chains, and the one that does the signaling is called CD79b. This reacts, not with ZAP-70, but with a molecule from the same family, called syk. Syk relays the message to the nucleus in a similar way and instructions are issued for division or secretion or whatever.

In CLL cells ZAP-70 is sometimes present when it shouldn't be. Even in these cells, it's still syk and not ZAP-70 that does the signaling. So what does ZAP-70 do?

All signaling has an on switch and an off switch. When you think about it, there would have to be, wouldn't there? Once a signal is sent, you don't want to keep repeating the same signal do you? In fact there is an automatic damping down of the signal once it has been sent. As syk reacts with downstream molecules, compounds are formed that inhibit syk from passing any more messages for a while. What ZAP-70 does is bind to these inhibitory compounds so that syk signaling can continue unimpaired.

So why is ZAP-70 there in this anomolous way? We don't know why, but we know how. Heat shock protein 90 (HSP90) acts as a molecular chaparone. It sticks close by certain moleculews and protects them from harm. It is a common feature of cancer cells that HSP90 binds to molecules that are supposed to be destroyed by natural processes and allows them to persist. ZAP-70 in some CLLs is one of these molecules with 'protection'. Why it gets this protection no-one knows. It's certainly not because it knows Tony Soprano.

A trial of treatment to remove the protection is under way in San Diego. They do this with a derivative of the antibiotic geldanamycin, known as 17-AAG.

The presence of ZAP-70 in some CLLs was discovered by the NIH group of Lou Staudt. After we had discovered the difference between mutated and unmutated CLLs other workers began looking for an easier assay. CD38 had proved to be only 71% concordant with VH gene mutations. The NIH group developed something they called a lymphochip (this was about the same time that FISH was getting popular and teh headline writers had fun with FISH and chips). The lymphochip allows thousands of genes to be examined at the same time for whether they are switched on or off. The glow red if on and green if off. They compared patients with mutated and unmutated VH genes - eventually using 70 of our patients for confirmation. The result was that of all the genes they examined the closest correlation was with ZAP-70. At the same time they began to look for antibodies that would detect the protein ZAP-70 in the cell rather than just the active gene. Both our lab and Montserrat's lab developed assays for ZAP-70 using flow cytometry. We presented our results at the same ASH meeting, but they were rather quicker in getting them into print - we were waiting for rather larger numbers. Both our methods were similar.

There are problems with the assay. First, unlike most flow methods we were trying to detect a molecule inside the cell rather than on its surface. This means that the cell has to be made permeable to the antibody with something like paraformaldehyde. Then ZAP-70 is only one of several proteins that belong to the syk family. They are all rather alike, and an antibody made against one tends to have some reactivity against the others. Then antibodies are selected because they give a clear reaction in a particular test. When used in a different test they may not show the same characteristics. We were looking at antibodies that showed reactions by immunohistochemistry or gel diffusion or immunoprecipitation. We looked at 8 different antibodies before we found one that was both sensitive and specific by flow.

The Spanish group chose a different way of expressing the results to us. The problem is that T cells and NK cells contain a lot of ZAP-70 and CLL blood has a lot of T cells and NK cells. We needed to separate the two populations, so you have to stain the T cells with anti-CD2 stained with a different color so that they appear in a different quadrant. The Spanish group used the most positive T cells as the zero, whereas we put in a negative control. Nevertheless, we got very similar results with 94% and 95% concordance with VH genes.

It would be more satisfactory to use a directly labelled antibody to do the test. The Spanish group and we used a fluorescently labelled second anti-antibody to detect the presence of the primary antibody. Using a single antibody would make it easier to commercialize the test. However, when you directly labelled our antibody, you got false positive results. The San Diego group eventually produced a test using a new antibody that seems to give reproducible results, and this paper later came out in the New England Journal of Medicine. Unfortunately, the concordance with VH genes was only 77%, hardly better than CD38. What they did suggest was that patients who were mutated/ZAP-70 positive did better than those that were unmutated ZAP-70 negative, and this seems to hold now they have accumulated over 1000 cases.

There is a difference, though. Our results were based on overall survival, the Rassenti results are based on length of first remission.

Although, I said that a direct test would be easier to commercialize, when Quest and other labs took up the challenge, there were problems. Reports started coming back that patients were getting different results from different labs. Because of this a workshop was organized by teh journal, Cytometry B. A lot of labs partcipated, and it was clear that there was a lot of disagreement on what was important. Results were different when heparin was used as an anticoagulant compared with EDTA. How much time passed between drawing the blood and testing it was important. and there were lots of technical differences. As a result ZAP-70 testing by flow is back in the melting pot. Individual research labs are fairly sure what there own test means, but there is no comercial test that is reliable, and my advice today is don't pay good money for a ZAP-70 test.

Finally, there has been a paper which tries to explain the animalous results between VH genes and ZAP-70. The Germans say that the VH mutated/ZAP-70 positive cases are the V3-21s, and the VH unmutated/ZAP-70 negative cases are thaose with p53 or ATM problems. All I can say to this is I have our examined this hypothesis in teh light of our own cases, and it does not hold water.