Measurable Residual Disease Moves Deeper Into Hematologic Cancer Development

Key Takeaways

• Measurable residual disease (MRD) is increasingly used in hematologic cancer development to identify risk, stratify and enrich clinical trial populations, assess depth of response, and support treatment decisions.

• MRD negativity can be strongly prognostic without necessarily serving as a validated surrogate for progression-free survival or overall survival, making context of use critical.

• Acute lymphoblastic leukemia provides an established regulatory precedent for MRD-defined treatment populations, while multiple myeloma is at the forefront of efforts to use MRD as an earlier regulatory endpoint.

• The FDA’s January 2026 draft guidance proposes MRD-negativity rate and complete response as potential primary endpoints supporting accelerated approval in multiple myeloma under defined conditions.

• As MRD becomes more consequential in development, assay sensitivity, specimen selection, assessment timing, analytical validation, and reporting become integral to clinical and regulatory strategy.

Response assessment in hematologic malignancies has historically depended in part on methods that cannot detect very small populations of malignant cells. Conventional morphologic assessment has a detection threshold of approximately one tumor cell among 100 cells, whereas measurable residual disease (MRD) technologies can identify disease several orders of magnitude below that level.¹ That additional sensitivity creates distinctions among patients who might otherwise fall within the same conventional response category.

Those distinctions can affect much more than prognosis. MRD can help identify higher-risk populations, stratify patients within clinical trials, contribute to treatment decisions, characterize the depth of response, and serve as an efficacy endpoint. As a result, it is becoming relevant to decisions throughout development rather than only to the assessment of whether a therapy produced a deeper remission.

Regulatory policy is evolving with this expanded role. In January 2026, the U.S. Food and Drug Administration (FDA) issued draft guidance proposing MRD-negativity rate, along with complete response, as a potential primary endpoint to support accelerated approval of therapies for multiple myeloma (MM).² The document remains draft guidance and does not establish final agency policy, but it illustrates how directly MRD is now entering discussions about the structure and timing of drug development.

The significance of an MRD result, however, depends on the decision it is intended to support. A measurement can be strongly associated with an individual patient’s prognosis without necessarily being suitable for determining treatment or substituting for a longer-term clinical endpoint. Increasing analytical sensitivity therefore expands what developers can observe, while making the intended context of use increasingly important to how an MRD result is interpreted and validated.

The clearest foundation for MRD is its ability to separate patients with different risks of subsequent disease progression or relapse. In acute myeloid leukemia (AML), a systematic review and meta-analysis encompassing 81 publications and 11,151 patients found estimated five-year disease-free survival of 64% among patients without detectable MRD compared with 25% among patients with MRD.³ Estimated overall survival (OS) was 68% and 34%, respectively. Consensus recommendations in AML likewise recognize MRD for prognostic, predictive, monitoring, and treatment-response assessment.⁴

Development programs can use that information in several ways. FDA guidance identifies MRD as a possible stratification factor, a means of selecting higher-risk patients or enriching trial populations, a measure of treatment response, and a potential efficacy endpoint.¹ Each application asks a different question of the same underlying measurement.

Using MRD to identify patients who are more likely to relapse is not equivalent to using it to decide whether therapy should stop. Neither establishes that a treatment’s effect on MRD will reliably predict its effect on progression-free survival (PFS), OS, or another clinical outcome. The FDA specifically distinguishes associations between MRD status and outcomes at the individual-patient level from the evidence needed to demonstrate surrogacy at the trial level .

The shift from prognostic marker to development variable therefore depends less on the existence of an MRD result than on whether the evidence supports the intended use of that result in a particular disease, treatment setting, assay, specimen, and time point.

One of the most established development applications is the use of MRD to distinguish patient populations that conventional response categories do not fully separate. The FDA guidance explicitly allows MRD to be used as a stratification factor, to identify patients considered at high risk, or to enrich a clinical trial population.

Acute lymphoblastic leukemia (ALL) provides an important example. MRD has been identified as a major prognostic factor in ALL, and the FDA has accepted an MRD threshold of at least 0.1% to identify patients in first or second complete remission who nevertheless remain at high risk of relapse.¹ In 2018, the FDA granted accelerated approval for blinatumomab to treat adults and children with B cell precursor ALL in first or second complete remission with MRD of at least 0.1%. The indication therefore defined a treatment population by residual disease detectable despite morphologic remission.⁵ The FDA’s current surrogate endpoint table continues to list MRD response rate for patients with B cell precursor ALL in first or second complete remission as an endpoint appropriate for accelerated approval.⁶

This approach allows developers to identify risk before overt relapse and to construct populations around biological information that conventional remission categories do not capture. However, thresholds cannot be transferred indiscriminately across indications. In adult ALL, the interpretation of MRD depends on factors including assay sensitivity, specimen source, assessment timing, and disease biology.⁷ The value of MRD for trial eligibility, enrichment, or stratification therefore rests on a disease- and context-specific definition of what the measurement represents.

The consequences become more substantial when MRD moves beyond describing risk and begins influencing what therapy a patient receives or how long treatment continues.

In adult ALL, MRD information is increasingly incorporated into treatment algorithms. Expert recommendations describe its use in selecting patients for MRD-directed therapies and in decisions involving allogeneic hematopoietic stem cell transplantation, while emphasizing that interpretation must account for disease characteristics, assay performance, specimen source, and timing.⁷

Chronic lymphocytic leukemia (CLL) provides a clear example of a trial in which MRD was incorporated directly into treatment duration. In the phase III FLAIR study, the duration of ibrutinib–venetoclax treatment was individualized according to MRD assessments in peripheral blood and bone marrow.⁸ Treatment duration was defined as twice the amount of time required for a patient to reach undetectable MRD, subject to the study’s maximum treatment period.

MRD-guided de-escalation is also being investigated in MM. In the MRD2STOP trial, patients receiving maintenance therapy discontinued treatment after multimodal confirmation of MRD negativity. Among 47 patients who stopped maintenance, three-year PFS was estimated at 92% in patients whose disease was undetectable below 10^–7, compared with 49% among those with detectable disease at that threshold. The proposed threshold requires further validation, making these findings evidence of an emerging approach rather than an established standard for treatment cessation.⁹

The attraction of these strategies is clear: MRD creates the possibility of tailoring treatment to the depth and persistence of response rather than using duration alone. Yet a negative MRD result does not establish that clinically relevant disease has been eradicated or that additional effective therapy has no value.

The E1910 trial makes that limitation particularly clear. Adults with B cell precursor ALL who had already achieved MRD-negative remission after induction and intensification were randomized to consolidation chemotherapy with or without blinatumomab. Three-year OS was 85% with blinatumomab and 68% with chemotherapy alone, while three-year relapse-free survival was 80% and 64%, respectively.¹⁰ Patients who had crossed the assay-defined threshold for MRD negativity could still derive substantial benefit from additional therapy.

MRD negativity is therefore an operational measurement of disease burden at a defined level of sensitivity and at a particular point in treatment. Crossing that threshold does not establish eradication of clinically relevant disease or eliminate the possibility of benefit from subsequent therapy. Treatment adaptation requires evidence linking the result not simply to prognosis, but to the clinical consequences of acting on that result.

The potential development impact becomes greatest if an MRD endpoint can provide reliable evidence of treatment efficacy before PFS, OS, or another longer-term outcome becomes available. This is also where the evidentiary standard becomes more demanding.

A prognostic biomarker does not automatically become a surrogate endpoint. Patients who achieve MRD negativity may have better outcomes than those who do not, but that patient-level association does not prove that a treatment that produces more MRD-negative responses will necessarily generate a corresponding improvement in later clinical outcomes. FDA guidance therefore calls for evidence at both the patient and trial levels when evaluating MRD for surrogacy and cautions against extrapolating a surrogate relationship across populations or therapeutic mechanisms without adequate justification.¹

AML demonstrates the strength of the prognostic relationship while also illustrating the distinction. The large meta-analysis of MRD in AML found substantial differences in disease-free and overall survival between patients with and without detectable MRD and supported consideration of MRD as a clinical trial endpoint.³ That association provides an important foundation, but it does not independently establish that treatment effects on MRD can universally substitute for treatment effects on long-term clinical outcomes.

MM now has trial-level evidence directly evaluating surrogacy. A 2025 meta-analysis of randomized controlled trials comprising 15,304 patients examined associations between treatment effects on MRD negativity and survival endpoints. It found significant associations with both PFS and OS and concluded that the results support MRD as a surrogate for survival in MM.¹¹

Evidence from CLL shows why similar assumptions cannot simply be transferred across hematologic malignancies. A 2026 analysis of 43 CLL trials involving 9,628 participants found that detectable MRD remained strongly associated with progression or death at the individual-patient level. At the trial level, however, the correlation between treatment effects on MRD and PFS was weak overall and varied according to treatment approach, therapy type, and the source of the sample used for MRD assessment.¹²

An MRD result can therefore provide excellent information about an individual patient’s prognosis while performing much less reliably as a measure of whether the relative effects of two treatments will translate into different long-term outcomes across a clinical trial.

For sponsors, the critical question is not simply whether MRD correlates with survival. It is whether the difference a treatment produces in MRD reliably predicts the difference that treatment will produce in an outcome that matters clinically. That claim must be demonstrated in the relevant development context.

Regulatory use of MRD has evolved incrementally rather than through broad acceptance of the biomarker across hematologic malignancies. The blinatumomab precedent established that MRD could help define a high-risk treatment population and support accelerated approval in ALL, while the FDA’s 2020 guidance provided a broader framework for incorporating MRD into hematologic cancer development.^1,5

Multiple myeloma has since become a particularly active area for endpoint development. European regulatory guidance adopted in 2022 addresses undetectable MRD as an intermediate efficacy endpoint in controlled, randomized MM studies and describes its potential to permit earlier approval while confirmatory evidence from harder clinical endpoints continues to mature.¹³

The January 2026 FDA draft guidance advances that discussion in the United States. It proposes MRD-negativity rate and complete response as potential primary endpoints capable of supporting accelerated approval in MM.² The draft states that an MRD endpoint could be used in either a single-arm or randomized trial, although randomized trials are preferred. It also describes a development pathway in which a randomized study supporting accelerated approval based on an earlier endpoint can continue to evaluate PFS or OS for traditional approval.

The draft defines important boundaries around that proposal. The FDA recommends assessing MRD negativity in bone marrow at a threshold of at least one residual tumor cell in 10⁵ cells. It states that available data are insufficient to support MRD as an accelerated-approval endpoint in the maintenance setting, smoldering MM, monoclonal gammopathy of undetermined significance, or extramedullary disease. The agency also considers the available data limited for imaging-based MRD and sustained MRD negativity as primary MRD endpoints.²

The scope is equally significant. The 2026 draft explicitly does not address MRD for patient selection, enrichment, trial stratification, or treatment decisions. Those applications remain distinct from the question the guidance is addressing: when an MRD endpoint can provide sufficiently persuasive evidence of efficacy to support accelerated approval.

An assay can therefore be appropriate for stratifying a trial without being validated as a surrogate endpoint, and an MRD threshold useful for treatment adaptation may not be the threshold required for a regulatory efficacy analysis. Regulatory acceptance attaches to a defined context of use rather than to MRD as a generic concept.

As MRD carries greater clinical and regulatory weight, the analytical method used to generate the result becomes inseparable from its interpretation.

The FDA does not prescribe a single MRD technology for clinical development. Its 2020 guidance is technology-agnostic and discusses approaches including multiparametric flow cytometry, next-generation sequencing, reverse transcription quantitative polymerase chain reaction, and allele-specific oligonucleotide polymerase chain reaction. The selected platform, however, should be prespecified and analytically validated for its intended context of use.¹

A negative MRD result only establishes that disease was not detected above the sensitivity achieved by a particular assay, specimen, and testing process. The FDA therefore emphasizes factors including assay performance, thresholds, timing of assessment, sample processing, quality controls, and reporting. Its guidance states that an MRD-negative result reported without the assay’s limit of detection is not meaningful.

This makes assay sensitivity part of the clinical definition rather than a technical detail. Two patients labeled MRD-negative may not have been assessed to the same depth if different methods, specimens, or analytical thresholds were used. Consensus recommendations in CLL similarly emphasize reporting the detection limit together with the MRD result.¹⁴

MM adds another complication because marrow assessment may not capture all manifestations of disease. Standardized myeloma response criteria incorporate sensitive bone-marrow methods and include imaging-based absence of extramedullary disease within deeper MRD response assessment.¹⁵ MRD2STOP likewise used multimodal confirmation incorporating positron emission tomography, flow cytometry, and sequencing when evaluating maintenance discontinuation.⁹

These considerations become development requirements as soon as MRD affects a consequential decision. If the result determines eligibility, stratification, treatment duration, or a primary efficacy endpoint, variability in sensitivity, specimen collection, sample quality, timing, or interpretation can alter the meaning of the clinical result. Assay selection and validation therefore have to follow from the role MRD is expected to play in the program.

Greater sensitivity can also uncover distinctions whose clinical significance is not yet established. In MRD2STOP, deeper sequencing separated patients who would otherwise have shared an MRD-negative classification at a less sensitive threshold. Detecting additional residual disease may improve risk resolution, but analytical depth does not itself establish what intervention should follow.

That distinction is increasingly important for trial design. If MRD will support a regulatory endpoint, the assay has to generate evidence that is reproducible, interpretable, and aligned with the threshold and time point specified in the clinical strategy. If MRD will guide treatment, developers need evidence that acting on the result improves the relevant outcome. The same measurement technology can therefore sit within very different evidentiary frameworks depending on what the protocol asks it to do.

The growing influence of MRD reflects more than the ability to detect progressively smaller amounts of residual disease. It reflects the increasing number of decisions developers want those measurements to support.

The contrasting evidence across hematologic malignancies shows why that distinction matters. MRD can define risk without being a validated surrogate, and an MRD-negative patient can still benefit from further therapy. A threshold suited to clinical stratification may not be appropriate for treatment cessation or regulatory efficacy assessment. More sensitive assays can reveal additional residual disease without establishing the clinical action that should follow.

The central development question is therefore becoming increasingly specific: what decision is an MRD measurement intended to support? The answer determines the required assay sensitivity, specimen, timing, threshold, analytical validation, clinical evidence, and regulatory strategy. The ability to measure residual disease more deeply creates the opportunity. The rigor with which developers define and validate its context of use will determine how much of that opportunity can be translated into faster and more informative hematologic cancer development.

1. “Hematologic Malignancies: Regulatory Considerations for Use of Minimal Residual Disease in Development of Drug and Biological Products for Treatment: Guidance for Industry.” U.S. Food and Drug Administration. 24 Jan. 2020.

2. “Minimal Residual Disease and Complete Response in Multiple Myeloma: Use as Endpoints to Support Accelerated Approval: Draft Guidance for Industry.” U.S. Food and Drug Administration. Jan. 2026.

3. Short, Nicholas J, et al. “Association of Measurable Residual Disease With Survival Outcomes in Patients With Acute Myeloid Leukemia: A Systematic Review and Meta-analysis.” JAMA Oncology. 6: 1890–1899 (2020).

4. Heuser, Michael, et al. “2021 Update on MRD in Acute Myeloid Leukemia: A Consensus Document from the European LeukemiaNet MRD Working Party.” Blood. 138: 2753–2767 (2021).

5. Jen, Emily Y, et al. “FDA Approval: Blinatumomab for Patients with B-cell Precursor Acute Lymphoblastic Leukemia in Morphologic Remission with Minimal Residual Disease.” Clinical Cancer Research. 25: 473–477 (2019).

6. “Table of Surrogate Endpoints That Were the Basis of Drug Approval or Licensure.” U.S. Food and Drug Administration. Accessed 02 Sep. 2026.

7. Short, Nicholas J, et al. “Clinical Use of Measurable Residual Disease in Adult ALL: Recommendations from a Panel of US Experts.” Blood Advances. 9: 1442–1451 (2025).

8. Munir, Talha, et al. “Chronic Lymphocytic Leukemia Therapy Guided by Measurable Residual Disease.” New England Journal of Medicine. 390: 326–337 (2024).

9. Derman, Benjamin A, et al. “Discontinuation of Maintenance Therapy in Multiple Myeloma Guided by Multimodal Measurable Residual Disease Negativity (MRD2STOP).” Blood Cancer Journal. 14: 170 (2024).

10. Litzow, Mark R, et al. “Blinatumomab for MRD-Negative Acute Lymphoblastic Leukemia in Adults.” New England Journal of Medicine. 391: 320–333 (2024).

11. Ntanasis-Stathopoulos, Ioannis, et al. “Evaluating Minimal Residual Disease Negativity as a Surrogate Endpoint for Treatment Efficacy in Multiple Myeloma: A Meta-Analysis of Randomized Controlled Trials.” American Journal of Hematology. 100: 427–438 (2025).

12. Wang, Yun, et al. “Measurable Residual Disease Is Not a Universally Reliable Surrogate for Progression-Free Survival in Clinical Trials of New Chronic Lymphocytic Leukemia Therapies.” Leukemia. 40: 1439–1447 (2026).

13. “Reflection Paper on the Use of Measurable Residual Disease as a Clinical Endpoint in Multiple Myeloma Studies.” European Medicines Agency. 12 Jul. 2022.

14. Wierda, William G, et al. “Measurable Residual Disease in Chronic Lymphocytic Leukemia: Expert Review and Consensus Recommendations.” Leukemia. 35: 3059–3072 (2021).

15. Kumar, Shaji, et al. “International Myeloma Working Group Consensus Criteria for Response and Minimal Residual Disease Assessment in Multiple Myeloma.” Lancet Oncology. 17: e328–e346 (2016).