Molecular monitoring for minimal residual disease

Jan. 18, 2015
Editor’s Note:

With this installment, MLO is proud to begin Year Three of our exclusive monthly feature, “The Primer: A guide to molecular diagnostics.” The editors thank Dr. John Brunstein for agreeing to provide these columns for a third consecutive year, sharing with our readers his deep knowledge of this vital topic—presented in a reader-friendly style. We are pleased that we continue to receive nothing but positive feedback for “The Primer.”

In this month’s edition of The Primer, we’re going to examine a topic in oncology. More specifically, it is one of particular relevance in some leukemias and lymphomas such as acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), and chronic myelogenous leukemia (CML). This is Minimal Residual Disease monitoring, or MRD.

Some background first, with the usual caveats; the information as presented here is a simplified summary of a very complex topic, so some statements are generalizations which an expert in the topic will recognize are not universally correct. As presented, however, this information should help readers unfamiliar with the methodology and be a starting point for asking more focused questions as needed of their molecular laboratory. It should help readers in the interpretation of lab results related to this methodology as well.

Genetic disruptions and leukemias

As most readers will know, leukemias are cancers of white blood cells. While these can arise in association with a wide range of genetic disruptions, some specific types of genetic damage are common causes of particular classes of leukemias. For example, translocations between chromosomes 12 and 21 causing fusions of the TEL and AML1 genes are observed in approximately one quarter of all ALL cases. Similarly, translocations between chromosomes 9 and 22 (known sometimes as the Philadelphia chromosome, after its place of discovery in 1960) resulting in fusions of the BCR and ABL genes are highly associated with CML. In either case, the expressed fusion gene products (TEL-AML1 or BCR-ABL) are critical to the progression of the cancer. While they may not be acting alone in the complete underlying pathology of the disease, these characteristic genetic rearrangements can serve as distinctive markers for cancerous cells in the bone marrow or peripheral blood in appropriate cases. For our purposes, we’ll focus mostly on CML as an example of how MRD measurement is applied. 

After a diagnosis of CML including confirmation of the relevant translocation as described above, generally by fluorescence in-situ hybridization (FISH), Sanger sequencing, and/or next-generation sequencing (NGS), appropriate therapies are started. These can be highly effective, with clearance of the majority of cancer cells in a relatively short time leading to disease remission. Historically, however, it has been found that remissions are often of short duration, and relapses often show less susceptibility to the same therapeutic agents used in the first course of treatment. The net result has been a rather poor long-term prognosis for most leukemias.

The cause of the relapses is the survival of a very small pool of cancer cells during the treatment. Known as the Minimal Residual Disease (MRD), these can be a very small fraction of the original cancer cell population as present at time of diagnosis and are generally below the level of detection by FISH. This state is referred to as a Complete Cytogenetic Response or CCyR. Given the capacity of these few cells to replicate very rapidly, it is not hard to understand how they give rise to relapses; nor is it hard to imagine how they may well have been selected for some level of resistance to the therapy agents they have been subjected to. By analogy, the situation would be similar to the bacterial population left in an infection where there was an incomplete course of antibiotic therapy. CCyR is thus not a good enough marker for disease clearance, and some more sensitive method is needed to quantitatively detect the very rare cells making up the MRD.

Molecular approaches to MRD

Readers who are already familiar with this series may sense that I am about to suggest that this is a prime sort of application for molecular methods, and indeed I am. Applications such as quantitative PCR (qPCR) for the fusion gene, quantitative reverse transcription PCR (qRT-PCR) for the fusion gene transcript, or possibly NGS (applied at significant depth to detect very rare fusions in a mostly wild-type background) are feasible candidates. 

Of these, qPCR and qRT-PCR are currently the most common approaches in our example of CML. The very high (arguably, complete) association of the BCR-ABL fusion with this disease led to early efforts at standardizing a molecular method of MRD quantification. In 2005, the National Institutes of Health (NIH) issued recommendations that traditional qRT-PCR be employed. (For a review of how qRT-PCR would be performed, I direct the reader to the August 2013 installment of this series, “Quantitative PCR methods,” found on page 32 of that issue. Or, visit

Briefly, patient samples from bone marrow or peripheral blood are obtained, RNA is extracted, and qRT-PCR (generally by a probe-based method for maximal specificity, and in triplicate for statistical validity) is performed against a reference standard template serial dilution. Ideally, a standard, presumed invariantly expressed “housekeeping gene” would similarly be assessed by qRT-PCR on the same samples, allowing for not only a measure of BCR-ABL fusion per sample but a ratio of it to an internal standard level. This standardization compensates for sample-to-sample variations in efficiencies of extraction, reverse transcription, amplification, and detection. It should be noted that there are several well characterized housekeeping genes employed as references in this context, and that not every lab uses the same reference. 

Measurements by this approach prior to treatment and then at regular intervals during therapy allow for a highly sensitive and numerically accurate assessment of MRD levels. An International Scale (IS) for evaluating MRD level changes in a patient is based on the patient’s initial pre-treatment level (100% BRC-ABL load) and 0.1% of this value. The 0.1% mark, or 3 log decrease in BCR-ABL transcript, is defined as the Major Molecular Response or MMR. (By comparison, CCyR occurs at approximately 10%–1%) A Complete Molecular Response (CMR) is defined as when BCR-ABL fusion transcripts are no longer detectable; with methods commonly employed, this limit of detection corresponds to around a 5 log decrease in load from pre-treatment levels or 0.001%.

Implications for therapy decisions

The importance of these terms relates to their utility in determining when to adjust or change therapy. While, as discussed above, achieving CCyR alone may be effective only at causing short-term remission, patients who have achieved MMR or CMR have vastly better long-term remission prospects.(Readers interested in specific values on this are directed to the primary literature, as the exact values and interpretations in context of exact molecular assay, sample type, and sampling frequency exceed the scope of our current overview). 

Via ongoing periodic monitoring of a patient during and subsequent to a course of therapy, it is therefore possible to not only directly observe the effectiveness of the treatment (time to CCyR and MMR are good prognostic markers), but also to detect the emergence of therapy resistance as evidenced by a significant increase in BCR-ABL fusion transcript between time points, allowing for potential changes to therapy before full relapse occurs. In short, much better disease management is provided through molecular MRD monitoring.

All of the above is fairly straightforward. Challenges for the clinician can arise, however, in knowing what constitutes a significant change between time points, as opposed to “normal” fluctuations. Communication with the testing laboratory may help to clarify this in the context of the assay as performed at that site, but in any case a 0.5 to 1 log change is suggested as significant in literature.1 Larger challenges can arise when trying to compare results on a single patient as tested in different laboratories, such as when a patient moves or a test provider is switched. Although efforts to standardize this testing have been underway for some time and actively continue, differences in methodology between laboratories, such as choice of reference gene, PCR instrument, or reference material, can lead to large apparent changes in fusion load in the absence of any real change. One possible approach to handling this is to have a sample split between old and new laboratories (or have both participate in a sample exchange or common EQA program), in each case allowing for the establishment of a correlation factor between tests. 

How does this approach apply to the other types of leukemias mentioned in the introduction? The same general principles apply to them as discussed for CML, although for some, there is less certainty of what particular molecular marker is appropriate in an individual case. This may increasingly lead to personalized approaches, which would originate through identification of a significant genetic marker for an individual during diagnosis, most likely through NGS methods. A directed qPCR or qRT-PCR can be developed and used specific to the patient marker; however, the challenges of this approach are in effective validation of the method. There will also be less a priori significance of any given percentage decrease in marker than in the well studied case of CML, although one would expect values to still be informative of treatment effectiveness and likely be able to signal early signs of therapy resistance and remission in time to most effectively alter therapy regimens. For the time being, personalized, patient-specific approaches remain primarily in translational research labs and on a small scale. 

Alternate molecular methodologies

What about methods other than the well-established probe-based q(RT)-PCR for this application? Digital PCR is still a qPCR approach, but its inherent ability to provide absolute quantitative values would lend itself well to this application; absolute values would provide a more direct method of comparing results between laboratories. Not surprisingly, there are already several extant publications on this application. In addition to benefits in quantification accuracy, these reports indicate digital PCR to have a two-to-three-log higher sensitivity than traditional qRT-PCR in the case of CML. This raises the potential for still better therapeutic monitoring and earlier detection of therapy resistance development. Together, these aspects make dPCR attractive for laboratories looking to establish MRD testing.  

As mentioned above, NGS is also a possible approach. It would provide data likely to be readily comparable between laboratories when a common marker such as CML is measured, and would have innate ability to be personalized to individual specific markers for other cancers; however, in order to work well in this setting, a very large sequencing depth would be required in order to detect the marker in a minute fraction of cells. This required depth has implications for throughput and cost which can be addressed somewhat through the use of focused gene panels, targeting just those gene regions (particularly fusion breakpoints) of interest. 

For now however, most clinical exposure to molecular MRD monitoring—most probably encountered in the context of  CML, ALL, or AML—is in the format of probe based qRT-PCR. An appreciation for the method, possible challenges around comparing results of different labs, and the meanings of terms such as CCyR, MMR, and CMR are of help in allowing non-specialists to understand and interpret these highly useful results. 


  1. Radich JP. How I monitor residual disease in chronic myeloid leukemia. Blood.2009;114(16):3376-3381.  Accessed December 11, 2014.