| TERRY HAMBlIN |
The problem with these new prognostic tests
is that they are new. They have been developed in research laboratories
and are still only being used in a relatively few labs. Most lab tests
are subject to external quality assurance. This means that wherever you
go in the world you get the same answer. The new prognostic tests are
not controlled in this way.
Flow cytometry
This is a method of determining whether an antibody reacts with a cell.
You can do this down the microscope, but if you do you can only count a
few hundred cells. The flow cytometer is a machine that does the same
thing, but counts tens of thousands of cells, so it is much more
accurate. Most routine labs now have a flow cytometer - there are quite
cheap bench top machines, and the methods have become standardised. The
main use of flow cytometry in CLL is to ensure the diagnosis. CLL cells
should be CD5+, CD19+, CD23+, and have weak surface immunoglobulin and
weak or negative CD79b. They are mostly FMC7 negative too, and CD20 is
usually quite weak. There are a number of other B-cell lymphomas that
sometimes resemble CLL, but flow cytometry for these is different and
therefore you can be sure that you are dealing with CLL and not an
impostor.
Recently labs have been testing for CD38. This tells something about
prognosis. The problem is that CD38 is also on T cells, so you have to
be sure that you are looking at the right population (B-cells not
T-cells), and when it was introduced even quite expert research labs
made a hash of it.
The reason that CD38 was thought to be useful was that it seemed to give
the same results as whether or not there were somatic mutations in the
immunoglobulin genes. Our lab had discovered that those patients without
mutations had a much shorter survival (about 8 years) than those with
mutations (about 25 years). The New York group came to the same
conclusion as us and showed that unmutated cases were CD38 positive.
Since somatic mutations is a hard to do test and very time consuming, it
was thought that CD38 could be a substitute. Unfortunately, this turned
out not to be true. We showed that the tests give different results in
30% of cases and in 25% of cases the level of CD38 changes during the
course of the disease.
It was because we were searching for another surrogate that ZAP-70
turned up. After we had suggested there might be 2 types of CLL,
scientists at the NIH in Maryland examined the 2 types to see which
genes were switched on and which were switched off. They did this using
a chip for microarrays. They found that the two types differed by about
240 genes, and the gene that most completely separated them was ZAP-70.
ZAP-70 is a molecule used to signal between the cell surface and the
nucleus. It is part of the normal signalling apparatus of T-cells but
not of B-cells. However it does seem to be involved in signalling in the
more severe type of CLL. Measuring ZAP-70 in the way that the NIH did it
is just as complicated as looking for somatic mutations.
At this point we collaborated with the NIH and we devised some other
quicker tests. One was RT-PCR, where you extract the RNA from the CLL
cells to see whether it will bind to complementary nucleic acid probes.
This turns out to be complicated because you have to remove the T-cells
(which have even more ZAP-70 then positive CLL cells do. Another used an
antibody to stain histology slides of lymph node or bone marrow. This
was OK, but only semi-quantitative. Another antibody test was Western
blotting, but this also meant removing the T-cells.
We then set about designing a flow cytometry test for ZAP-70. This was
done by Jenny Orchard in our lab. It was quite a difficult task because
ZAP-70 in CLL cells is quite weak and inside the cells rather than on
the surface. The antibodies available were not designed for this sort of
assay. Many labs around the world gave up because they couldn't get it
to work, but Jenny persevered and finally developed an assay. We had to
check that it gave the same result as all the previous tests including
RT-PCR, and Western blotting. At the same time as us Dr Montserrat's lab
in Barcelona, Spain also produced a Flow assay. Both assays were
reported at the ASH meeting in December 2002. Both assays gave closely
similar results to somatic mutations - <10% discrepancies.
Late last year the CLL Consortium reported an American test developed at
San Diego. Unfortunately this gave 25% discrepancies with somatic
mutations. They suggested that if somatic mutations and ZAP-70 gave
different results then the ZAP-70 was more likely to be right.
Unfortunately this is not what the British, the Spanish and the Germans,
who now have a test of their own, find. So the question remains open and
I still think that we should do somatic mutations as well. Fortunately a
more streamlined test for somatic mutations is becoming available, and
we can now get answers more quickly and on older material. We can do the
test on samples arriving by post rather than having the patient appear
at the lab for testing.
FISH testing is another way of looking at chromosomes. There is no
single chromosomal abnormality in CLL. The earliest finding was trisomy
12 - three number 12 chromosomes. In the mid 1980s, David Oscier in our
lab discovered the commonest abnormality, a number 13 chromosome missing
its long arm (del 13q). The problem with conventional analysis of
chromosomes is that CLl cells don't divide very easily, and you can only
analyse chromosome down the microscope when they are dividing. With FISH
you can recognise chromosomes in non-dividing cells. It is like an
antibody test where a fluorescently labelled antibody is reacted with a
protein on the cell. Here it is a fluorescently labelled piece of DNA
which is reacted with apiece of DNA in the cell with a matching
sequence. Just at fluorescent antibodies are targeted at a single
protein, so the fluorescent DNA is targeted at a particular piece of
DNA. This means that with FISH you don't see the whole set of
chromosomes only the ones that you specifically target.
Generally, the targets are the long arm of 13, the middle of 12, the
long arm of 11 and the short arm of 17. A missing bit of 13 is a good
prognosis, a missing bit of 17 is a very bad prognosis, a missing bit of
11 is often bad news. but some patients are not affected by it and we
are still trying to work out why. Three number 12s is Ok if there are
somatic mutations but bad news if there are not. I would prefer to have
the full set of chromosomes tested, but the truth is that most labs are
unable to do this, and even we have abandoned it for most routine cases.
So what are the most important test to have done. We do somatic
mutations (VH mutations), ZAP-70, CD38 and FISH. And I only trust the
labs that have developed the tests with appropriate controls. Until
there are proper quality control services I don't trust the commercial
labs on these very difficult tests.
A simple question. I'm sorry the answer is so complicated. |
