Measurable Residual Disease: What Fellows Need to Know
Note: Please review ASH’s disclaimer regarding the use of the information contained in these articles.
Malignant hematology is famous for its three-letter acronyms: MDS, CML, ALL, CLL, and the list goes on. The newest and most talked-about kid on the block is MRD, or measurable residual disease (previously known as minimal residual disease). Indeed, the term MRD has become so ubiquitous that it has elbowed out its older MRD cousin from the transplant world (“matched related donor,” now often referred to as a sibling donor).
If you’re like us though, you have likely heard about MRD only from conferences or in clinic. As part of a new series for ASH TraineE-News, we will review these sorts of learned-on-the-fly concepts in more detail. Our first topic? You guessed it, MRD!
What does MRD refer to?
In short, an MRD assay can detect small quantities of malignant cells that previous methods might miss. While the sensitivities of conventional assays can vary widely — for example, how many cells a hematopathologist manually reviews or whether a multiple myeloma fluorescence in situ hybridization assay is CD138-enriched — a good rule of thumb is that immunohistochemical or immunofluorescent assays can detect malignant cells down to a level of approximately 1 of 100 total cells.1
In contrast, MRD assays make it possible to detect malignant cells down to a sensitivity of 10-5 (0.001% or 1 malignant cell in 100,000) or even 10-6 (0.0001% or 1 in 1,000,000) using next-generation sequencing (NGS) or next-generation flow cytometry (NGF) assays. These MRD assays can sometimes be run off of peripheral blood (e.g., in CLL where circulating disease is common2) but are generally checked from bone marrow aspirates for multiple myeloma and acute leukemia.
For patients who are ostensibly disease-free, NGS or NGF testing may be able to demarcate MRD positivity (i.e., some malignant cells still detected above a predefined threshold) from MRD negativity (i.e., no such cells detected at this predefined threshold). This predefined threshold may vary from disease to disease (or trial to trial) but is often set at 10-4 in acute myeloid leukemia (AML)3 and 10-5 in multiple myeloma.4
How do NGS and NGF detect MRD?
NGS uses polymerase chain reaction probes to detect malignancy-specific genetic sequences. In chronic myeloid leukemia (CML), we have been using the equivalent of NGS for detection of the classic bcr-abl mutation for decades without referring to it as MRD. An example of NGS MRD testing is the ClonoSEQ® assay, which uses an initial “ID” test to detect specific genetic sequences for an individual patient (called clones) to track changes in the quantity of clones over time. In contrast, NGF uses color-coded antibodies to detect malignancy-specific cell surface marker profiles at a high level of sensitivity. Examples of NGF techniques for MRD detection include EuroFlow® for multiple myeloma and institutional assays run by the University of Washington and the Mayo Clinic.
How does one decide between NGS and NGF? The answer is surprisingly complex, but there are two practical factors that we trainees should know:
- NGS can be run off stored samples, while NGF requires a fresh sample (since dead cells generally slough off their cell surface markers).
- NGF can be assessed on any sample, while NGS will only work if a tumor-specific clonal sequence is available and known (e.g., through the ClonoSEQ® ID sample described earlier).
Under optimal conditions though, both NGS and NGF assays can reach a sensitivity of between 10-5 and 10-6 for MRD detection.5
Does MRD negativity mean that a patient is in remission?
Not always. One reason is that a single MRD assay doesn’t always tell the whole story. For example, a patient with myeloma can have MRD-negative bone marrow but radiographic evidence of active plasmacytomas. Conversely, because our working definition of a complete response in myeloma requires M-spike clearance from the blood (which can sometimes take months to clear after intensive therapies6), a patient may truly be MRD-negative without a synchronous complete response based on serum testing.
Further, sampling error or errors with technique can interfere with MRD assays. For diseases like chronic lymphoblastic leukemia (CLL) or acute lymphoblastic leukemia (ALL), peripheral blood MRD assessments can bypass sampling error (compared to a single bone marrow aspirate) but may not necessarily be as sensitive. For bone marrow MRD assessments, hemodilution can profoundly affect the denominator of any MRD calculation, which is why MRD testing should always be sent from the first pull of bone marrow aspirate.
What is the point of MRD testing in clinical practice?
Outside of a research setting, there are currently few (if any) situations in which costly MRD assessments will unambiguously change management. Perhaps the closest such scenario is in patients with B-cell ALL who have achieved a complete response after induction and are planning to proceed to allogeneic transplantation. In this setting, pretransplantation MRD positivity is associated with poorer post-transplant outcomes7 than MRD negativity, and the addition of blinatumomab — a bispecific antibody that allows native T-cells to recognize CD19-positive tumor cells — has a growing evidence base for MRD eradication8 in this setting.
Additionally, there are several ongoing studies using MRD assays to determine if therapy can be de-escalated or even stopped, including the CAPTIVATE study9 with ibrutinib/venetoclax in CLL or the MASTER study10 with Dara-KRd (daratumumab, carfilzomib, lenalidomide, and dexamethasone) in myeloma. However, every MRD assay has limitations based on sampling error, testing characteristics, and disease heterogeneity. As such, MRD testing requires dedicated and prospective study in each specific disease.
What is the point of MRD testing in research? Is MRD negativity an acceptable surrogate endpoint for clinical trials?
MRD is indeed prognostic, and several meta-analyses (in myeloma,11 B-ALL,12 and AML13) have shown that patients who achieve MRD negativity have better outcomes compared to patients who do not. However, studies that compare responders with non-responders are flawed in terms of assessing a test’s prognostic value. Why? Because this type of study is inherently a comparison of patients with tumor biology that is intrinsically responsive to treatment versus patients whose tumors are not. As such, achievement of MRD negativity reflects not only the efficacy of treatment but also the underlying tumor biology. Or, to borrow a parenthood analogy that one of us is currently struggling with: If a baby can sleep through the night by 12 weeks of age, congratulations to the parents! However, rather than attributing it to intensive sleep training or a brand of pacifier, it was probably a skill the baby had all along.
In summary, the use of MRD testing in malignant hematology will likely continue to grow in coming years, but only as the science behind its methodology and trials investigating its clinical applications expand. The details of what we’ve discussed are outlined in the Table. While MRD testing is an exciting concept, only carefully designed trials of MRD-guided decision-making will help determine whether MRD negativity is an appropriate surrogate for hard endpoints such as overall survival.
Table. MRD: The Good, the Bad, and the Ugly.
| Technical Details | Clinical Implications | |
| Good | Can possibly detect one malignant cell out of (up to) a million total cells | Can inform need for pre-transplantation blinatumomab in patients with B-ALL who achieve an MRD-positive complete remission |
| Bad | Can be influenced by a hemodiluted sample (hence, always pull MRD first from the bone marrow aspirate) | Currently unclear if MRD is a marker of therapy effectiveness versus indolent disease biology |
| Ugly | Does not account for sampling error or other sites of disease: for example, plasmacytomas on a PET scan | Can be quite costly and has not yet been shown to change clinical management in most settings |
Disclaimers: Many details are intentionally omitted in this article for simplicity’s sake. Brand names are included for convenience and do not represent endorsements.
- Ravandi F, Walter RB, Freeman SD. . Blood Adv. 2018;2:1356-1366.
- Lew TE, Anderson MA, Lin VS, et al. . Blood Adv. 2020;4:165-173.
- Schuurhuis GJ, Heuser M, Freeman S, et al. . Blood. 2018;131:1275-1291.
- Kumar S, Paiva B, Anderson KC, et al. . Lancet Oncol. 2016;17:e328-e346.
- Medina A, Puig N, Flores-Montero J, et al. . Blood Cancer J. 2020;10:108.
- Singhal S, Powles R, Milan S, et al. . Bone Marrow Transplant. 1995;16:537-540.
- Shen Z, Gu X, Mao W, et al. . BMC Cancer. 2018;18:755.
- Gökbuget N, Dombret H, Bonifacio M, et al. . Blood. 2018;131:1522-1531.
- Wierda WG, Tam CS, Allan JN, et al. . Blood. 2020;136 (Supplement_1):16-17.
- Costa LJ, Chhabra S, Godby KN, et al. . Blood. 2019;134 (Supplement_1):860.
- Munshi NC, Avet-Loiseau H, Anderson KC, et al. . Blood Adv. 2020;4:5988-5999.
- Bassan R, Brüggemann M, Radcliffe HS, et al. . Haematologica. 2019;104:2028-2039.
- Short NJ, Zhou S, Fu C, et al.. JAMA Oncol. 2020;6:1890-1899.
