Molecular monitoring in CML—a modern example of an old proverb

“You cannot see the forest for the trees” (John Haywood, London 1546)

Chronic myeloid leukemia (CML) is an uncommon leukemia but one that has been on the cutting edge of therapeutic and diagnostic discovery over the last three decades. With the identification of a cytogenetic abnormality half-century ago, then a gene rearrangement, and ultimately an abnormal tyrosine kinase [1], this disease has become the poster child for drug development in what is now called targeted therapy [2],—not just for CML, but also for other malignant disorders [3]. What was once a disease that was only curable with a allogeneic stem cell transplant, is now a “chronic” disease treatable and potentially curable with a targeted therapy pill and with survival no different from age-matched controls in almost all patients [4], if the disease is diagnosed in chronic phase. Only with her2 therapy in breast cancer, did any other disease have this option at that time. TKI toxicity and compliance issues have become more of a reality than disease progression to acute leukemia [5]. We continue to develop newer drugs that are of value in the resistant population, although their value in early care is controversial.

Understanding the biology of CML has also led to the development of options for diagnosing and monitoring of the disease and the impact of treatment [6]. What used to require labor-intensive and expensive cytogenetic testing to identify the Philadelphia chromosome, can now be done with a standardized relatively simple blood molecular test with roughly a thousand times greater sensitivity. Polymerase chain reaction (PCR) testing for the gene rearrangement RNA has become the standard for monitoring CML therapy outcomes in the vast majority of patients using commercially BCR::ABL1 available kits and analyzers, with the exception of those with uncommon gene splicing, where patient-specific primers are needed or less sensitive FISH analysis used. This kind of monitoring is now common in other hematological malignancies [7]. What has really been spectacular is some of the work that has made for sample transfer (sometimes just on paper blots) from places with no testing, to central labs, making this technology available to more patients. This has also helped reduce the cost of testing.

Molecular testing has now gone even further with CML. Gene sequencing, and in some cases specific probes, now allow the identification of binding site mutations for either bcr::abl1 or the myristoyl pocket, and can help redirect therapy to alternate drugs in the case of resistance. This test is not as readily available in most labs but can be extremely valuable. With standard Sanger sequencing, mutations can be detected at around the 15% range. More sensitive techniques can improve sensitivity to less than 1% [8]. This is probably of value in the rare cases of compound mutation resistance. Many of my colleagues around the world are looking at these and their efforts to improve this have been heroic. The big question however, is; what is too sensitive? At what level does a mutation actually confer resistance? Are the mutations that are now being detected at alternate sites in advanced or resistant disease actually responsible for resistance [9, 10]? Or are they just manifestations of aging, demographics, “normal” clonal hematopoiesis, etc.? The problem is that with an uncommon disease and an even less common frequency of disease progression, and that we generally look at these only in patientswho manifest CML resistance. We really do not have a good idea of how common these findings are in a normal population or a non-progressing population, and we definitely do not have much data on progression of these mutations in time. I do not mean that we should not continue to look, and as we develop less expensive technology to do this more easily on a broader scale, we will be able to chart these as they develop and better predict outcomes [11]. At this time, however, we may have outraced our ability to interpret accurately, but this should not stop the work in this area.

What are some of the issues in using molecular or PCR testing for monitoring? If one is a regular CML treater and uses the test routinely, an expectation of understanding the limitations and problems should be straightforward. If however a physician uses this rarely, they may not understand the concepts of control genes, breakpoints, amount of RNA, etc. that go into generating a number. Why a result is “undetectable” at one time and then has a number the next, is not always clear to a non-specialist and those of us in referral centers very often have to deal with questions of “loss of response” or “relapse” when undetectable has become 0.01% IS with a very sensitive assay.

Where this latter issue is amplified is with the patient. It is now common and “expected” that patients have access to their results. Some of them sit on their phones or computers and look for these to show up or are set up to get them the moment they are available, usually before the ordering physician has seen them. The most unfortunate situation occurs when there really is a problem with the results and the patient sees them and goes into panic mode before the treating physician has seen them and can discuss with the patient. This is most horrible when the result is a biopsy or imaging study, which truly shows a bad situation. When molecular testing and the report of undetectable going to detectable at a truly sensitive level, which is really no change, is seen by most patients, this will often generate calls or visits that are not needed, but are generated by misinterpreting such a result. In patients on treatment-free remission monitoring, this can often lead to a lot of stress even though low levels are now tolerated with no intervention. Although we do our best to educate our patients and discuss issues with them, the concept of the presence and absence of disease is often disturbing [12].

Finally, we are dealing with what is called treatment-free remission (TFR) or treatment discontinuation [13, 14]. This is not done for medical reasons, which is a separate issue. TFR is done when a patient gets a sustained deep molecular remission over at least two years, and an effort is made to stop treatment. Attempting this in a case where there is an insufficient response, will fail. This is the “holy grail” of CML therapy. Most people want it – patients and their physicians. It is no reality for the majority of patients – response is not deep enough, sustained, patients do now want to try, or about a third to a half of those who do try, fail. Survival does not depend on this. Results show that the 10-year survival of patients achieving a BCR::ABL1 transcripts (IS) < 1% is virtually the same as those achieving a 0.001% response. Not achieving a response deep enough may be distressing to patients. Of greater concern is the drive to achieve this sometimes pushes physicians to switch TKIs even in otherwise well patients, with a resulting new set of adverse events. In general, as new CML TKIs have come into use, more chronic nagging issues related to first-generation drugs; can sometimes be replaced by more severe problems with second or third-generation drugs. Chasing molecular numbers and not patient quality can be a problem. Changing drugs for severe side effects or resistance is necessary. Changing to what may be a riskier or unknown risk new drug to achieve lower numbers needs to be carefully thought through.

Molecular monitoring in CML has been a boon to patients and physicians in terms of accuracy, convenience, and cost, and by no means should be downplayed. It has allowed us to improve patient care and outcome but is not without potential problems when it becomes the primary goal. We must use and interpret it wisely. Patient survival, safety and QOL are a priorityfor many [15] and the “forest” in this analogy, although not always in agreement with physicians. TFR is the ideal bonus, the icing on the cake, and molecular typing is the “trees”. We can hope for it and work for it and encourage our patients to be compliant and honest in reporting side effects, but realize that not achieving TFR is not a failure.

Molecular monitoring is right up there with survival, monitoring of side effects and adverse events, optimizing QOL and cost control. As CML treaters, we must look at the road to response, but also look to all the lanes on the highway. This is very important to the other people on the journey with us. I wonder if John Haywood treated CML.

Competing interests

The author declares no competing interests.