Minimal Residual Disease in Leukemia:
The Case for Drug Development Tools
J. Milburn Jessup
Scientific Director
Precision Cancer Care Program

ecent history has shown that pathologic complete response (pCR) in breast cancer is a critical drug development tool (DDT). Particularly, the I-SPY 2 trial has supported development of agents for FDA accelerated approval based on pCR data obtained in the neoadjuvant setting in breast cancer. Leukemias have similar characteristics in that tumor burden can be measured directly in the bone marrow as well as in the blood after treatment. As discussed below, the key concept of Minimal Residual Disease (MRD) is that there is a threshold of disease below which the likelihood of relapse is low. It is somewhat easier to identify leukemia cells than to measure pCR, because usually (not always) the leukemic cells express antigens that are distinct or in a pathological profile that distinguishes leukemic cells from normal cells.

Additionally, leukemic cells may be quantitated by flow cytometry or molecular methods so that numerical cut-offs can be analyzed for association with clinical outcome. For this reason, we felt that leukemia might be a good use case for MRD as a DDT, and that acute lymphoblastic leukemia (ALL) might be the best model since it has fewer molecular variations than other leukemias and because MRD is already used extensively to monitor childhood ALL. Also, there was concern that adult ALL was different from the more prevalent childhood ALL because of its much lower survival rate and the low tolerance that adult patients have for the robust chemotherapy regimens used in childhood leukemia. Thus, there was a need for better methods to support development of milder treatments for adults. All these factors prompted investigation into MRD as a DDT for ALL.

Drug Development and the Importance of Minimal Residual Disease

ALL is an orphan disease with less than 6,000 new cases annually, and is nearly ten times more frequent in children than in adults. Although childhood ALL has a five-year survival of 85-90%, it is only 30-40% in adults depending on such factors as age at diagnosis, comorbidities, and the ability to undergo rigorous treatments such as bone marrow transplant. By 2011, it was clear that hematologic malignancies had a major advantage over solid tumors for drug development in that each leukemia was identified by distinct sets of either Cluster Determinant (CD) antigens or clonal immunoglobulin or T-cell receptor rearrangements that differentiated ALL from normal cells or other leukemias. It was apparent as early as 2001 that survival was improved if the number of blasts in the marrow was decreased below some detection limit, as measured with standardized protocols using real-time quantitative polymerase chain reaction (RQ-PCR). This limit became known as MRD, and the MRD threshold above which treatment should be considered is now 1 blast in 10,000 (0.01%) nucleated cells. Since RQ-PCR has a sensitivity that exceeds 0.001%, quantitation to the MRD threshold rapidly became integrated into clinical trials, with therapy decreased or intensified based on the MRD threshold.

RQ-PCR, adopted primarily in Europe, is quite sensitive, but it is time-consuming, labor intensive, expensive, and unable to measure blasts in some patients. So multiparametric flow cytometry (mpFCM), first with four-color and then with six-color assays, was developed by the US Children’s Oncology Group (COG) for its trials in ALL, employing a central laboratory for all studies.

Rise of Targeted Therapies for ALL

By 2011, targeted therapies were entering the clinic for adoptive cellular as well as biologic therapy of ALL in both children and adults. Use of targeted therapies in many instances is based on the MRD rate; thus, it became imperative to standardize MRD assessment in multicenter clinical trials. Furthermore, these therapies targeted the membrane antigens that mark ALL, first CD19 and later CD22. Adoptive cellular therapy uses chimeric antigen receptors (CAR) to CD19 that are long-acting and engineered to transduce a patient’s normal T-cells to kill leukemic blasts. The first biologic is blinatumomab, a bispecific monoclonal antibody that targets CD19 with one arm while the other binds to CD3, bringing T-cells to CD19 ALL blasts and stimulating cytotoxic events. Early trials of both approaches were highly successful in relapsed/refractory and newly diagnosed patients with ALL. Since these antigens, especially CD19, were the cornerstone for identifying ALL blasts by mpFCM, better standardization of MRD measurement was needed because it was shown that resistance to therapy would likely derive from outgrowth of blasts that did not express the target antigen, spurring demands for better standardization of MRD and strategies to detect relapsing clones without CD19.

FDA-ASCO-NCI Workshops on MRD

In 2013, FDA and the American Society for Clinical Oncology (ASCO) held workshops to review the role and quantitation of MRD for Chronic Lymphocytic Leukemia (CLL) and Acute Myelogenous Leukemia (AML); NCI supported a similar workshop in ALL, leading to consensus statements available elsewhere. At this time, MRD was measured in the US by independent Laboratory Developed Tests (LDTs) instead of by FDA-cleared tests; therefore, there was concern that it would be difficult to get consistent high quality results for clinical trials, let alone for patient treatment, without harmonization.

As a result, the NCI workshop reported on a “bake-off” which compared the COG reference laboratories with two adult oncology group reference laboratories. The satellite COG laboratory was highly concordant with the base COG reference laboratory; however, the other laboratories were much less concordant with the base laboratory. The low concordance further demonstrated the need for standardizing MRD assays.

Questions in Standardizing MRD

The “bake-off” raised other questions as well:

  • Was adult ALL like childhood ALL in the pattern of CD antigen presentation?
  • What was the best MRD level for a positive test?
  • Given the future use of targeted therapies that would alter clonal evolution, would mpFCM still be able to identify blasts which deleted the CDs that were targeted?
  • Similar questions were raised in the FDA-ASCO workshops.

FDA supported initial focus on ALL since the disease is molecularly a little simpler than either CLL or AML. CLL has the problem of somatic hypermutation that mature B cells use to generate the B-cell antigen receptor repertoire, while AML is extremely heterogeneous with multiple subtypes that confound a simple strategy for identification of blasts. Since CAR T cells and biologics were beginning to come to the clinic, the FDA suggested that MRD may be useful as a DDT if it was shown to be a predictor of response. This led to the formulation of a project that was supported by a public-private partnership managed by the Foundation for the National Institutes of Health (FNIH).

MRD Detection in Adult ALL

The FNIH project had two major components: (1) Meta-analysis led by Drs. Donald Berry of the UT MD Anderson Cancer Center and Jerry Radich of the Fred Hutchinson Cancer Research Center; and (2) Standardization tasks led by Michael Keeney of the London Ontario Health Sciences Centre, Dr. Brent Wood of the University of Washington Medical Center, and Dr. Michael Borowitz of Johns Hopkins University.

The meta-analysis evaluated existing international literature to determine the utility of MRD for predicting event-free (EFS) and overall (OS) survival in pediatric and adult ALL. The major findings were: MRD behaves similarly in pediatric and adult ALL for EFS and overall survival (OS), e.g., for EFS the HR was 0.23 (0.18-.28 95% CT, N= 11249 patients) for MRD-negative patients compared to MRD-positive patients, while the EFS HR in adults was similar 0.28 (0.24-0.33, N = 2065 patients).

Subset analysis did not demonstrate significant interactions with the timing of MRD assessment, method of analysis, cytogenetics, or even the MRD cut-off. OS results were also alike in that both had HR’s of 0.28. Thus, this analysis established that MRD in adult ALL has the same importance as in childhood ALL as a prognostic factor. The analysis also demonstrated that MRD after therapy is likely to be a predictive biomarker after therapy and could be used as a DDT.

The standardization work was carried out in a cadre of eight laboratories including the cooperative oncology group reference laboratories, an academic laboratory, a government laboratory, the laboratory of a nonprofit healthcare organization, and an international laboratory. The goal was to develop a method for standardizing the demonstration of MRD by mpFCM. This goal had two parts: 1) harmonize the COG standard six-color panel; and 2) use the cadre of harmonized laboratories to develop new eight-color kits that could be commercialized as another method of MRD testing. Becton-Dickinson and Beckman Coulter, two of the main manufacturers of flow cytometers for clinical use, generously supported this effort by providing reagents for the project. Each manufacturer understood that the project would enable them to develop platform-specific reagents that could then be cleared by FDA. Commercialization of kits for these platforms would be a tremendous standardization driver because the kits would need to be used as cleared by the FDA.

A standard protocol was developed based on the COG panel, including a process for in silico training using electronic list mode data files from cases of ALL. This was a critical step because the files could be loaded into each laboratory’s analytic system independent of the platform on which it was collected, and each group could go through MRD assessment exercises in an anonymous manner. After several rounds and a wet lab challenge followed by more in silico training, the group was well harmonized and some of the laboratories were cleared to perform in COG trials. (This process is documented in detail in a publication by Keeney et al., along with addressing the challenging problem of identifying so-called “hematagones” [regenerating normal B cells] in adult bone marrow after recovery from treatment.) While a manuscript on standardizing MRD assessment by mpFCM in ALL is good, it does not necessarily promote adoption of the practice. However, the system has been accepted into the proficiency testing program of College of American Pathologists (CAP) Clinical Laboratory Improvement Amendments (CLIA) certification process based on these results. Another success of the standardization effort is increasing the number of laboratories certified to support NCI’s Cancer Therapy Evaluation Program from two to 18. Decentralization helps because a major problem for MRD by mpFCM is that samples must be analyzed fresh, within at most 96 hours of collection.

Manufacturers are also in the late stages of validating eight-color panels for mpFCM on their newest cytometers. We hope that the manufacturers will be able to bring their data forward to FDA for evaluation and possible clearance as in vitro medical devices.

Rise of Next Generation Sequencing

Next generation sequencing (NGS) holds the promise that clones of leukemic cells can be followed individually by their molecular signatures and therapy altered quickly based on the rise and fall of individual clones. The advantage that NGS has over mpFCM is that it can use banked specimens and does not require fresh material for clinical assay. Thus, it is possible to assess NGS-based technology in samples banked from clinical trials where mpFCM was performed even years earlier.

Adaptive Biotechnologies performed an NGS analysis of ~600 COG samples. These and other data enabled the company to gain de novo approval for its ClonoSeq assay in September 2018. ClonoSeq is now cleared for use to measure MRD in multiple myeloma and ALL and will enable laboratories that use it to perform standardized assessments of MRD.


The research described has demonstrated that MRD is a potentially useful DDT for ALL, enabling the quantitation of disease burden in a standardized way. The standardization process continues with the CAP-CLIA proficiency testing, eight-color panel validations, the expansion of certified mpFCM-based laboratories, the understanding that adult ALL behaves similarly to childhood ALL, and now the clearance of a NGS-based assay. FDA is now continuing the effort with an individual patient-based meta-analysis of MRD in ALL to confirm and expand the findings of the literature-based meta-analysis.

And what about the original impetus for the development of the MRD project in ALL, blinatumomab? The eight-color panels will compensate for loss of CD19 or CD22. In addition, on March 29, 2018, FDA expanded its approval for the use of blinatumomab in patients with pre-B ALL who have an increased risk of relapse because the MRD level meets or exceeds 0.1%. MRD is now part of a drug approval!

Besides ALL, efforts in multiple myeloma and AML are in planning. All these hematologic malignancies support Dr. Radich’s statement: “MRD is the disease.”