Tuesday, May 08, 2007

Why should I get my prognostic markers done?

I think that this is a genuine dilemma for patients with CLL. In one sense whether you know what your prognostic markers say or not is not going to affect the rate of progression of your disease. Your disease will either progress or not, and when and if it does progress there is time enough to consider treatment, but until it does you might as well forget about it. Undoubtedly, some patients are able to put it out of their minds. On the other hand some patients are happier if they can know all there is to know about their disease, be it good news or bad.

Let's go back to the beginning. Before the prognostic markers came along patients were classified according to stage. The Rai system was used in America and the Binet system in Europe. Although they are largely interchangeable there are important differences. Rai stage 0 consists of patients who just have a lymphocytosis. Their lymph nodes are not enlarged and their liver and spleen cannot be felt on clinical examination. They also have normal or near normal hemoglobin and platelet levels. Although such patients are mostly benign, hidden among them are patients who have been picked up early in their disease and within a few weeks will have progressed to stage III or IV, and some who have enlarged lymph nodes, liver or spleen, but they cannot be felt on clinical examination although they would be revealed by a CT. The most benign Binet group, stage A, does allow enlargement of liver, spleen or lymph nodes, but only in 2 of the five possible areas (neck, armpit, groin, spleen or liver). More than this they are stage B. Rai Stage I or II overlaps with Binet stage B, but also contain some patients who are Binet stage A (with enlargement in fewer than 3 areas). So Rai stage 0 is mainly benign, but includes some early stage aggressive diseases. Binet stage A is similar though it contains rather more early stage aggressive diseases. Rai I or II are regarded as an intermediate stage, as is Binet B, but some of these are late stage non-aggressive disease. Someone with 1cm lymph nodes in groin, armpit and neck would be Binet stage B yet may have a very benign disease.

You would think that having anemia or thrombocytopenia (Binet C or Rai III and IV) would automatically be bad news, but this is only true if the cytopenias are due to heavy marrow infiltration. They could be due to an enlarged spleen or to autoimmunity.

In fact most new patients (around 80%) have Binet stage A disease, so these staging systems that have been around since the 1970s are increasingly less helpful for treatment decisions.

Oncologists have known about this for a long time and have been looking for something better. Two of the most relied upon are bone marrow biopsy and lymphocyte doubling time.

Bone marrow biopsy is unnecessary for diagnosis. It is useful to distinguish the cause of a cytopenia - diffuse infiltration implies that this is the cause, and compared to any other pattern it is a poor prognostic feature. Bone marrow biopsy can help with prognosis, and in the context of a clinical trial can be used before and after treatment to assess response, though these days tests for minimal residual disease give more information. But it can be a painful investigation and it is not cheap. Its use as a prognostic indicator could easily be replaced by a better test.

Lymphocyte doubling time is a simple and cheap test that just needs regular CBCs. But it has its drawbacks. You need several tests showing an upward trend. The tests have to be done at the same lab, because of inter-lab variation. Allowance has to be made for infection or vaccination and other things that can transiently raise the lymphocyte count. The increase has to be measured on a meaningful part of the curve. An increase from 4 to 8 is not significant, while an increase from 40 to 80 probably is.

Other tests like the measurement of serum LDH or beta-2 microglobulin have their advocates, but they are often difficult to interpret. Both these and other substances (like CD23, thrombopoietin, IL-8 etc) are made by CLL cells, so measuring how much there is, is dependent on how many CLL cells there are and how quickly they are growing. You can get the same result from a large slow growing tumor or a small rapidly-growing tumor. They are also affected by how well the kidneys work in excreting the surplus.

At this time CLL was regarded as a continuum. We didn't know when it started - it was diagnosed at an arbritary time during its progression and its rate of progression was a closed book. Any patient could suddenly change from being stable to progressive, and all patients needed to be watched carefully for when that happened.

In this context the two papers that appeared in Blood in 1999 (Hamblin et al and Damle et al) were remarkable. They appeared to say that CLL was not really a continuum, but there were two different types, one relatively benign and one relatively aggressive, determined by something set in stone, the kind of cell that the leukemia had developed from. If you were benign from the beginning, you remained benign. If you were aggressive now, you had always been aggressive. Far from being a marker of prognosis, to know your VH gene mutational status was part of the diagnostic procedure; did you have this type of CLL or that type? It was as important as knowing whether you had follicular lymphoma or diffuse large cell lymphoma.

As time has passed, we have learned that things are more complicated than that. In 2000 Dohner et al published in New England Journal of Medicine how the FISH test for chromosomal abnormalities could be used as a prognostic factor. In a sense this was 10 year old news. In 1990 Juliusson et al had published in New England Journal of Medicine about the importance of chromosomal abnormalities. The problem at that time was the technique of karyotyping was unsuccessful in most laboratories. (In fact most of the patients in this paper came from our series; although we are able to detect an abnormal karyotype in about 90% of patients where one is present, only one other laboratory can match this, and most have results of around 40%). FISH testing allowed laboratories to detect thye 4 commonest chromosomal abnormalities (trisomy 12 and deletions on chromosomes 11, 13 and 17) easily and reproducibly.

The Dohner lab have demonstrated that del 17p is particularly evil. Patients who have this had (in 2000) an average survival of 2 and a half years. This is because they lack one copy of the p53 gene, and lacking this the cells are not protected from further genetic damage. Fortunately, only about 5% of patients have this deletion at presentation. Not quite as bad, but still worse than just having unmutated VH genes, is having del 11q. This implies a lack of one copy of the ATM gene. (ATM stands for 'ataxia telangectasis mutated'. It is the gene that is mutated in the rare congenital disease ataxia telangectasia). The ATM gene is part of the same biochemical pathway as p53. The reason we think it isn't as bad as a p53 deletion is that you need both ATM genes to be knocked out to lose its function in the cell, whereas knocking out just one p53 gene is sufficient. Preliminary work by Tanya Stankovic in Birmingham suggests that this is so. About 15-20% of CLL patients have del 11q at presentation. Del 11q is almost always associated with unmutated VH genes, del 17p usually so.

FISH testing is very useful, but it misses all the less common abnormalities and especially missed translocations. Ideally I would have both FISH and conventional karyotyping, but at the moment conventional karyotyping is only reliable in about two laboratories in the world.

CD38 and ZAP-70 started out as quick and easy ways to get the same information as the VH gene mutations, but both have proved to be unsatisfactory surrogates. CD38 proved to be an independent prognostic factor but if VH genes and FISH were both done CD38 added nothing extra. ZAP-70 is unsatisfactory because there isn't a standardised assay. With some laboratories it doesn't differ very much from VH genes, but the assay done at Tom Kipps lab actually seems to be a better predictor than VH genes. This is the one that commercial labs have tried to copy, but in commercial hands it is much less successful and gives unreliable results.

Until recently the evaluation of prognostic factors has been retrospective. They were tested on samples that were sometimes more than 20 years old from patients who were treated in many different ways, often suboptimally. However since 1999 the tests have been applied in many prospective trials to see how they perform when patients are given standard treatments. So how do they do?

The most important finding has been for FISH for del 17p. These patients do uniformly badly in clinical trials. In particular patients who fail the fludarabine + cyclophosphamide combination have a very poor survival. One of the commonest causes for this is p53 deletion. It is not the only reason. P53 function can be abnormal in the presence of a normal chromosome 17, and what is needed is a functional test of p53. Andy Pettitt at Liverpool has developed one of these and it is a good candidate for being evaluated in the next round of clinical trials. In the meantime it seems foolhardy to begin conventional treatment without doing the FISH test. The 5% with del 17p should be treated in a different way. Treatments that may be effective in p53 deficient CLL include high dose steroids, Campath, flavopiridol, revlimid and allograft. This is important for patients receiving their first treatment, but even more so for patients receiving subsequent treatments. As many as 30% of patients in clinical trials eventually develop p53 abnormalities.

For del 11q the picture isn't so clear. In some clinical trials it is an independent adverse predictive factor while in others it doesn't show, disappearing into the adverse effect of unmutated VH genes. In general the response rate is the same as with other treated cased, but respopnses appear to be shorter. At the moment there is no justification for offering a different treatment from that given to other unmutated cases.

In most trials unmutated VH genes seems to be an independent adverse factor. Overall survival is better for those with mutated VH genes who receive the same treatment. It seems to me that these is justification for stratifying treatment according to VH gene mutations. I envision trials comparing standard treatment with something less aggressive for those with mutated VH genes and standard treatment versus something more aggrtessive for those with unmutated VH genes.

What about doing these tests at presentation? I still think that this is a matter of patient choice. It is important to remember that the median is not the message. Median survival is simply the time at which half the patients are alive and half are dead. Obviously if 0% are dead at two years, 50% dead at three years and 100% dead at four years, the median is very important, but if the first death occurs at 1 year, the last death at 30 years and the 50% cut off is at 15 years, then the median survival figure is not much use to the individual patient who might die this year or in 30 years time.

VH genes are a bit better than that. In a clinical trial the difference in survival after treatment at 5 years between patients with mutated and unmutated VH genes was 16%. But added to that is the fact that around half of the patients with mutated VH genes never need any treatment whereas nearly all those with unmutated VH genes do.

I recently studied 150 patients with Binet stage A disease whose lymphocyte count was less than 30. 50 had poor prognostic markers and all of them needed treatment. All had eventually died of their CLL even though one of them lived beyond 20 years. On the other hand 100 had good prognostic markers. Only 10% ever needed treatment and none have died of CLL so far. In this case the best single marker at predicting the future was CD38 with a 30% cut off.

As I said, some patients relish this sort of information; some disdain it. I am not going to judge between them. However, I think that everybody should be able to access it if they want it.

10 comments:

Anonymous said...

Do you mean 16 percentage points? I assume that is what you mean (i.e. the difference between 50% and 66% is 16 percentage points, not 16%). Percentages of percentages are meaningless.

Since a post in Wikipedia has explained it, I don't have to:

"Percentage points (PP) are the proper unit for the arithmetic difference of two percentages.

Consider the following hypothetical example: in 1980, 40 percent of the population smoked, and in 1990 only 30 percent smoked. We can thus say that from 1980 to 1990, the incidence of smoking decreased by 10 percentage points even though smoking has not decreased by 10 percent; since percentages indicate ratios—not differences."

Terry Hamblin said...

Yes, of course, thanks for the correction. Clumsy of me. It's the difference between 69% and 53% and the gap is widening every year.

Anonymous said...

Terry.

Do I understand you correctly?

If you have CLL and your WBC never gets over 30k and you have good prognostic indicators you have a 100% chance of of survival, compared to an age and sex matched cohort at 20 years?

If this is true, and I am sure it is from your sample of patients, it means that 66% of CLL people should be told to forget about it.

Is the quality of life of these people affected? What constitutes "death by CLL"?

Terry Hamblin said...

Steve,

Yes, that is pretty well what the figures say, though admittedly the data have not been tested by being published in a peer-reviewed publication. We are expanding the numbers and doing all the prognostic markers to be sure, but the data so far say that if you have a Hb greater than 12, a platelet count greater than 100, a CD38 less than 30%, no lymph nodes, spleen or liver to feel then the actuarial survival in our series (censoring for non-CLL deaths) is 100%, and the treatment-free survival is 89%.

CLL-related deaths are deaths from infection, treatment, transformation, marrow failure. Cancer deaths were not regarded as CLL-related unless they were from melanoma, lymphoma or leukemia. In this population the common cancers are common in both normals and patients with CLL and the sample is too small to sort out noise from signal.

However, this is not 60% of CLLs. In our series it comprises about a quarter of patients.

One caveat is that although the longest survivor is out at more than 25 years, some patients have only been observed for 3 years. this is an actuarial curve and as time passes it might change.

Anonymous said...

I am one of those who relish knowing all the numbers.

About your study of 150 Binet A patients with absolute lymphocyte counts less than 30....is this at initial diagnosis only? I think one of your earlier posts said so. If this ALC less than 30 criterion applied throughout the course of the disease, then it could be a self-fulfilling prophecy.

Thanks.

justme said...

"It's the difference between 69% and 53% and the gap is widening every year."

Why is the gap widening, exactly?

Terry Hamblin said...

Because the unmutated subset is dying more rapidly than the mutated subset.

Anonymous said...

Sorry I must have misunderstood your post.

50 out of 150 had poor prognostic indicators and died.

100 with good indicators have not died.

I assumed that 100 was two thirds of the sample ie 66%.

Anonymous said...

Dr. Hamblin,
What can you tell me about having an extra 12 chromosome? With all the discussion of deletions of other chromosomes I haven't come across much about this extra 12. If I am unmutated, what does this mean for me exactly? Thanks!

Terry Hamblin said...

Trisomy 12 adds nothing to being unmutated. It is a rather different disease from the point of view of the pathologist, but it makes no difference to the patient