Tuesday, February 13, 2007

Minimal residual disease

This is a concept that is confusing for most people, but it is vital if you are to understand the treatment of leukemia.

There are 1,000,000,000,000,000 cells in the body. 90% of them aren't even human, they're bacterial, but although you may be only 10% human by cell number, bacterial cells are so small that they make up only a trivial amount of your volume. It is impracticable to talk about a number with so many zeros, so in real life we talk about such large numbers as 10 to the power of 15; that is, a one with 15 zeros after it. It is usually written as a 10 with a tiny superscript 15 after it. Since I haven't worked our how to do superscripts on the blog, I shall refer to such numbers as 10 to ten power of whatever.

So in the human body there are 10 to the power of 14 human cells. In a patient with so much leukemia that they are just about to die there may be as many as 10 to the power of 13 leukemic cells; that is as many as 10% of the cells in the body are leukemic. When the disease presents there are typically about 10 to the power of eleven cells in the body. (Put the figure in your mind this way. 10 to the power 6 is a million. In America, 10 to the power 9 is a billion [in England it is a thousand million or milliard], 10 to the power 12 is a trillion [in England it's a billion]; so, 10 to the power of eleven is 100 billion in American money.

A complete remission in acute leukemia means that you can no longer detect leukemic cells in the bone marrow. This equates to less than a billion cells in the body (ten to the power of nine). It is normal to go on treating at this level (called consolidation treatment); it is hoped that by this time there are fewer than 10 to the power of 5 cells (one hundred thousand) left in the body. This is a stupendous result meaning that 99.9999% of the leukemic cells have been eliminated. We don't really know whether this is good enough. It is assumed that the body's defenses can handle small numbers of cells, and this seems to be true for some people some of the time.

Assays have been developed to test whether there is any disease left in the body - these are known as tests for minimal residual disease (MRD). Mostly these tests rely on one of two techniques: flow cytometry or the polymerase chain reaction PCR). Flow techniques rely on identifying a combination of antigens on the leukemic cells that is not on normal cells. For example CD2, CD7, CD19, and CD45. Flow techniques can be very sensitive, able to detect one cell in a thousand. PCR is generally more sensitive, detecting one cell in one hundred thousand. In acute leukemia, the PCR is used to detect a unique piece of DNA in the leukemia such as the DNA coding for the fusion protein generated by a chromosomal translocation.

In CLL a complete remission is less stringent than in acute leukemia. The NCI have defined a CR as all of the following being present for at least 2 months: absence of lymphadenopathy, hepatomegaly or splenomegaly by physical examination and appropriate imaging, absence of constitutional symptoms, polymorphs > 1500 per microlitre, platelets greater than 100,000 per microlitre, hemoglobin > 11.0 g/dL, bone marrow biopsy showing normal cellularity, fewer than 30% lymphocytes and no lymphoid nodules. Although patients with a CR have longer remissions than those who only get a partial response, most CRs can be shown to have residual disease by more sensitive tests.

In CLL there are usually no chromosomal translocations to use for a PCR test for MRD. The CLL cells are clonal and the DNA sequence of the immunoglobulin genes is the same in every cell. A PCR technique has been developed making use of this. Since in most cases nobody knows what this sequence is the PCR has to use what are called 'consensus primers' ie it will pick up any monoclonal population. It will detect MRD, but it is not very sensitive. It will detect one in a thousand cells. This means that if 10 thousand cells are loaded onto the PCR machine and among them are 10 the same, then it will pick it up, but if only 8 are the same then it won't. Some patients have had their immunoglobulin DNA sequenced to see if they were mutated or unmutated. If this has ben done then it is possible to design a PCR using sequence specific primers, and this makes the PCR much more sensitive picking up one cell in 100,000.

The alternative method is to use flow. Many people have hit upon the fact that CLL cells are CD5 and CD19 positive and have used this as a flow assay. Unfortunately this is neither sensitive nor specific - you get false positives and false negatives. This assay has been used in many famous units as a method of detecting MRD. Honestly, it is quite useless. Andy Rawstron developed a four color flow assay that was almost as sensitive as the best PCR technique. Using antibodies against CD19/CD5/CD20/CD79b he was able to detect somewhere between 2 and 5 cells in 100,000. This assay won't do for patients who have been treated with rituximab, and for these he has devised another assay employing antibodies against CD81/CD22/CD19/CD5.

These assays work in practice. In a study employing Campath to remove MRD, MRD negative patients had longer treatment-free and overall survivals than those whose CRs were MRD positive. The actuarial figures were startling; at 7 years only 20% of the MRD negative patients had require retreating, whereas patients with MRD positive CRs needed retreatment after a median 20 months and those with PRs after a median 13 months. Achieving an MRD negative CR led to an overall survival of 80% at 7 years. Not achieving an MRD negative CR led to an overall median survival of 19 months. However of those who became MRD negative, half had become MRD positive again by 28 months. This does not mean that Campath is a wonder treatment. Only 20% of those treated became MRD negative, and this may have more to do with the nature of the CLL than the fact that they had Campath. Campath may just be a good, if expensive way, of picking winners.


Anonymous said...

I'm not exactly sure what you mean by the following:

Only 20% of those treated became MRD negative, and this may have more to do with the nature of the CLL than the fact that they had Campath. Campath may just be a good, if expensive way, of picking winners.

Do you mean that CLL is capricious? That there is some genetic variation that's at play? Or some other difference? It is hard for me to believe that there is a simple random variation at play here.

Terry Hamblin said...

Exactly so. The biggest variation is whether the VH genes are mutated or unmutated. Responses are better in the mutated subset. Other determining factors are del 11q and del 17p which tend to lead to more resistant disease. There may well be other unidentified factors that affect response rates. What I am saying is that rather than wade through all the prognostic factors to pick out the good responders, it might be more sensible to give a course of Campath. Those who get a molecular remission are the responsive ones and those who do not are the more resistant ones. That doesn't mean that Campath is necessarily the best treatment for the responsive ones, indeed it may be counter-productive because of the potential side effects of Campath like CMV reactivation.