Hematologic Malignancies: Regulatory Considerations for Use of Minimal Residual Disease in Development of Drug and Biological Products for Treatment: Guidance for Industry

Administrative guidance, not enacted legislation. This captured publisher edition preserves the listed document and attachments; later developments are not independently consolidated.

Hematologic Malignancies:
Regulatory Considerations for
Use of Minimal Residual
Disease in Development of Drug
and Biological Products for
Treatment
Guidance for Industry
U.S. Department of Health and Human Services
Food and Drug Administration
Oncology Center of Excellence (OCE)
Center for Drug Evaluation and Research (CDER)
Center for Biologics Evaluation and Research (CBER)
January 2020
Clinical/Medical

Hematologic Malignancies:
Regulatory Considerations for
Use of Minimal Residual
Disease in Development of Drug
and Biological Products for
Treatment
Guidance for Industry
Additional copies are available from:
Office of Communications, Division of Drug Information
Center for Drug Evaluation and Research
Food and Drug Administration
10001 New Hampshire Ave., Hillandale Bldg., 4th Floor
Silver Spring, MD 20993-0002
Phone: 855-543-3784 or 301-796-3400; Fax: 301-431-6353; Email: druginfo@fda.hhs.gov
https://www.fda.gov/drugs/guidance-compliance-regulatory-information/guidances-drugs
and/or
Office of Communication, Outreach, and Development
Center for Biologics Evaluation and Research
Food and Drug Administration
10903 New Hampshire Ave., Bldg. 71, Room 3128
Silver Spring, MD 20993-0002
Phone: 800-835-4709 or 240-402-8010; Email: ocod@fda.hhs.gov
https://www.fda.gov/vaccines-blood-biologics/guidance-compliance-regulatory-information-biologics/biologics-
guidances
U.S. Department of Health and Human Services
Food and Drug Administration
Oncology Center of Excellence (OCE)
Center for Drug Evaluation and Research (CDER)
Center for Biologics Evaluation and Research (CBER)
January 2020
Clinical/Medical

TABLE OF CONTENTS
I. 	INTRODUCTION............................................................................................................. 1
II. 	BACKGROUND ............................................................................................................... 2
III. 	DEVELOPMENT OF MRD AS A BIOMARKER FOR REGULATORY USE ........ 3
A. 	Regulatory Uses of Biomarkers .................................................................................................... 3
B. 	Mechanisms for Novel Surrogate Endpoint Acceptance or Qualification ................................ 4
C. 	Meta-Analyses for Validation of MRD as a Surrogate Endpoint .............................................. 5
D. 	MRD as an Endpoint in Clinical Trials ....................................................................................... 7
E. 	MRD for Patient Selection or Enrichment .................................................................................. 8
IV. 	TECHNOLOGY CONSIDERATIONS .......................................................................... 8
A. 	Assay Considerations ..................................................................................................................... 8
1. Cellular Technology Platforms ....................................................................................................................9
2. Molecular Technology Platforms .................................................................................................................9
3. All Technology Platforms ...........................................................................................................................10
B. 	Sampling Considerations............................................................................................................. 10
V. 	DISEASE-SPECIFIC CONSIDERATIONS ................................................................ 11
A. 	Acute Lymphoblastic Leukemia ................................................................................................. 11
B. 	Acute Myeloid Leukemia ............................................................................................................ 11
C. 	Acute Promyelocytic Leukemia .................................................................................................. 12
D. 	Chronic Lymphocytic Leukemia ................................................................................................ 13
E. 	Chronic Myeloid Leukemia ........................................................................................................ 13
F. 	Multiple Myeloma ........................................................................................................................ 14
VI. 	REGULATORY SUBMISSIONS THAT USE MRD .................................................. 15
GLOSSARY................................................................................................................................. 18

Contains Nonbinding Recommendations
1
Hematologic Malignancies: Regulatory Considerations for Use of	1
Minimal Residual Disease in Development of Drug and Biological	2
Products for Treatment	3
Guidance for Industry1
4
5
6
7
This guidance represents the current thinking of the Food and Drug Administration (FDA or Agency) on	8
this topic. It does not establish any rights for any person and is not binding on FDA or the public. You	9
can use an alternative approach if it satisfies the requirements of the applicable statutes and regulations.	10
To discuss an alternative approach, contact the FDA office responsible for this guidance as listed on the	11
title page.	12
13
14
15
I. 	INTRODUCTION	16
17
This guidance is intended to help sponsors planning to use minimal residual disease (MRD) as a	18
biomarker in clinical trials conducted under an investigational new drug application (IND) or to	19
support marketing approval of drugs and biological products2 for treating specific hematologic	20
malignancies.	21
22
The use of MRD as a biomarker in drug development is distinct from FDA’s requirement for	23
investigating, clearing, or approving an in vitro diagnostic device for clinical use in measuring	24
MRD. Manufacturers interested in developing a specific MRD assay for clinical use should	25
consult the Office of In Vitro Diagnostics and Radiological Health in the Center for Devices and	26
Radiological Health (CDRH).	27
28
In general, FDA’s guidance documents do not establish legally enforceable responsibilities.	29
Instead, guidances describe the Agency’s current thinking on a topic and should be viewed only	30
as recommendations, unless specific regulatory or statutory requirements are cited. The use of	31
the word should in Agency guidances means that something is suggested or recommended, but	32
not required.	33
34
1 This guidance has been prepared by the Oncology Center of Excellence in cooperation with the Center for Drug
Evaluation and Research and the Center for Biologics Evaluation and Research at the Food and Drug
Administration.
2 For the purposes of this guidance, all references to drug products include both human drugs and biological
products.

Contains Nonbinding Recommendations
2
35
II. 	BACKGROUND	36
37
Despite development of treatments that eliminate morphologically detectable malignant cells,	38
some patients with hematologic malignancies who have achieved complete remission or	39
complete response (CR), even of considerable duration, will experience relapse. Conventional	40
morphologic detection for hematologic malignancies has a threshold limit of one tumor cell in	41
100 cells. Technology exists that can detect the persistence of malignancy at orders of	42
magnitude below the limit of conventional morphologic detection, a level of disease burden	43
known as MRD. These technologies measure cell characteristics such as genetic mutations, cell	44
surface markers, or specific DNA gene rearrangements.	45
46
MRD as a general measure of tumor burden has multiple potential regulatory and clinical uses as	47
a biomarker. Depending upon the clinical setting, MRD may be used to reflect a patient’s	48
response to treatment or as a prognostic tool to assess a patient’s risk of future relapse. As such,	49
MRD can be used to enrich clinical trial populations or guide allocation into specific treatment	50
arms in clinical trials. There are challenges within each context of use that need to be addressed,	51
such as the underlying disease, patient heterogeneity, therapeutic context, target of therapy, or a	52
combination of disease parameters, to allow effective use of MRD in regulatory decision-	53
making.	54
55
MRD assessments can vary among laboratories and technologies, which can result in discrepant	56
results. Many clinical laboratories develop their own protocols that can affect MRD	57
measurements. Technologies can have different performance characteristics. Sample collection	58
procedures can also differ. However, standardized methodologies can ensure that results	59
obtained between technologies and laboratories are consistent. This includes standardized	60
posttreatment timing for when a bone marrow (BM) or blood sample is collected, standardized	61
sample processing, predetermined MRD thresholds, and accurate reporting of the performance	62
characteristics of the test (e.g., accuracy, precision, specificity, sensitivity). For example,	63
reporting MRD negative results without information regarding limit of detection is not	64
meaningful.	65
66
The evidence to support the clinical validity of MRD as a biomarker varies across hematologic	67
cancer types and patient populations. To gain a better understanding of the state of the science of	68
MRD, FDA cosponsored public workshops on MRD in acute lymphoblastic leukemia (ALL),	69
chronic lymphocytic leukemia (CLL), and acute myeloid leukemia (AML) as well as a	70
symposium on MRD in multiple myeloma (MM) from 2012 through 2014. In addition, a public	71
workshop, Minimal Residual Disease as a Surrogate Endpoint in Hematologic Cancer Trials,3	
72
was held on September 7, 2016, under a cooperative agreement with FDA to discuss the clinical,	73
statistical, and technical barriers to implementing use of MRD in clinical trials. As a result of	74
these workshops and an analysis4 of marketing applications showing inconsistent quality of	75
3 See https://healthpolicy.duke.edu/events/minimal-residual-disease-surrogate-endpoint-hematologic-cancer-trials.
4 Gormley N, V Bhatnagar, LA Ehrlich, B Kanapuru, H-Z Lee, AE McKee, A Farrell, and R Pazdur., 2017, FDA
Analysis of MRD Data in Hematologic Malignancy Applications, J Clin Oncol, 35:2541.

Contains Nonbinding Recommendations
3
MRD data, FDA identified a need to provide sponsors with guidance on use of MRD as a	76
biomarker in regulatory submissions.	77
78
79
III. 	DEVELOPMENT OF MRD AS A BIOMARKER FOR REGULATORY USE	80
81
A. 	Regulatory Uses of Biomarkers	82
83
The term biomarker is commonly understood as referring to a characteristic that is measured as	84
an indicator of normal biologic processes, pathogenic processes, or responses to an exposure or	85
intervention, including therapeutic interventions.5 MRD can be used as a biomarker. The	86
terminology listed below is derived from the BEST Resource6 definitions and the draft guidance	87
for industry and FDA staff Qualification Process for Drug Development Tools (December	88
2019)7 but has been slightly modified to reflect applicability to MRD. Sponsors can potentially	89
use MRD status as any of the following types of biomarkers:	90
91
• 	Diagnostic biomarker: A biomarker used to detect or confirm presence of a disease or	92
condition of interest or to identify individuals with a subtype of the disease.	93
94
• 	Prognostic biomarker: A biomarker used to identify likelihood of a clinical event,	95
disease recurrence or progression in patients who have the disease or medical condition	96
of interest. A prognostic biomarker informs about the natural history of the disease in	97
that particular patient in the absence of a therapeutic intervention.	98
99
• 	Predictive biomarker: A biomarker used to identify individuals who are more likely	100
than similar individuals without the biomarker to experience a favorable or unfavorable	101
effect from exposure to a drug product.	102
103
• 	Efficacy-response biomarker: A biomarker that is used to show that a response has	104
occurred in an individual who has been exposed to a drug product.	105
106
• 	Monitoring biomarker: A biomarker measured serially and used to detect a change in	107
degree or extent of the disease.	108
109
5 See FDA-NIH Biomarker Working Group, 2018, BEST (Biomarkers, EndpointS, and other Tools) Resource,
accessed September 9, 2019, https://www.ncbi.nlm.nih.gov/books/NBK338448/. See also Section 507 of the
Federal Food, Drug, and Cosmetic Act, which defines biomarker for purposes of that section, in relevant part, as “a
characteristic (such as a physiologic, pathologic, or anatomic characteristic or measurement) that is objectively
measured and evaluated as an indicator of normal biologic processes, pathologic processes, or biological responses
to a therapeutic intervention.”
6 FDA-NIH Biomarker Working Group, 2018, BEST (Biomarkers, Endpoints, and other Tools) Resource.
7 When final, this guidance will represent the FDA’s current thinking on this topic. For the most recent version of a
guidance, check the FDA guidance web page at https://www.fda.gov/regulatory-information/search-fda-guidance-
documents.

Contains Nonbinding Recommendations
4
An efficacy-response biomarker could be a surrogate endpoint. A surrogate endpoint does not	110
measure the clinical benefit of primary interest; instead, it predicts the clinical benefit based on	111
epidemiologic, therapeutic, pathophysiologic, or other scientific evidence. Specifically, a	112
surrogate endpoint predicts a specific clinical outcome of the patient at some later time and can	113
be used as the basis of marketing application approval decisions.	114
115
Understanding which of these biomarker attributes applies to the proposed use of MRD is	116
important to consider when validating MRD for that proposed use and for the trial design. There	117
are challenges within each MRD context of use that should be adequately justified, such as	118
underlying disease, patient heterogeneity, therapeutic context, target of therapy, or a combination	119
of disease parameters.	120
121
B. 	Mechanisms for Novel Surrogate Endpoint Acceptance or Qualification	122
123
Two mechanisms exist to obtain the Agency’s feedback on the use of a novel surrogate endpoint	124
to support approval. One mechanism is through the formal drug development tool (DDT)	125
qualification process, specifically the biomarker qualification process. The DDT qualification	126
process is an initiative undertaken in response to FDA’s Critical Path Initiative and updated	127
under the 21st Century Cures Act, adding section 507 to the Federal Food, Drug, and Cosmetic	128
Act. The purpose of the DDT qualification process is to qualify a DDT for a specific context of	129
use, such that a sponsor and FDA can rely on the DDT to have a specific interpretation and	130
application in drug development and regulatory review. FDA will make information about a	131
DDT that has been formally qualified for a specific context of use publicly available to expedite	132
drug development and review of regulatory applications. A qualified DDT can be included in	133
submissions of INDs, new drug applications (NDAs), or biologics license applications (BLAs)	134
without the need for FDA to reconsider and reconfirm the suitability of the DDT. Qualifying a	135
biomarker requires robust scientific evidence, and there is a higher evidentiary standard if the	136
biomarker is to be used as a surrogate endpoint.8	
137
138
A second mechanism to obtain the Agency’s feedback on the use of a novel surrogate endpoint	139
to support approval is through discussions with the specific Center for Drug Evaluation and	140
Research (CDER) or Center for Biologics Evaluation and Research (CBER) review division. In	141
this setting, the pharmaceutical sponsor or interested group meets with the FDA review division	142
to present scientific data in support of the proposed surrogate endpoint. These data may be from	143
previous clinical trials conducted by the sponsor, a meta-analysis of several trials conducted in	144
the disease area, or other data, including product-nonspecific data, that support the use of the	145
proposed surrogate endpoint. An example of this mechanism for a surrogate endpoint reasonably	146
likely to predict clinical benefit is pathologic complete response in neoadjuvant treatment of	147
breast cancer.9 An example of a validated surrogate endpoint that used this mechanism is the	148
8 For additional information on the DDT qualification process, see the DDT Qualification Programs web page at
www.fda.gov/Drugs/DevelopmentApprovalProcess/DrugDevelopmentToolsQualificationProgram/default.htm and
the draft guidance for industry and FDA staff Qualification Process for Drug Development Tools. When final, this
guidance will represent the FDA’s current thinking on this topic.
9 See the guidance for industry Pathological Complete Response in Neoadjuvant Treatment of High-Risk Early-
Stage Breast Cancer: Use as an Endpoint to Support Accelerated Approval (October 2014). We update guidances

Contains Nonbinding Recommendations
5
surrogate of complete response at 30 months in follicular lymphoma. A surrogate endpoint that	149
is reasonably likely to predict clinical benefit can be used to support accelerated approval, and a	150
validated surrogate endpoint can support traditional approval.10 To explore this approach	151
further, sponsors should request a meeting with the relevant review division.	152
153
With either approach, the strength of evidence to support surrogacy depends on (1) biological	154
plausibility of the relationship, (2) demonstration in epidemiological studies of the prognostic	155
value of the surrogate endpoint for the clinical outcome, and (3) evidence from clinical trials that	156
treatment effects on the surrogate endpoint correspond to effects on the clinical outcome.11	
157
158
C. 	Meta-Analyses for Validation of MRD as a Surrogate Endpoint	159
160
Various statistical criteria have been proposed for validating a surrogate endpoint; often, meta-	161
analytical approaches have been used. The terminology and definitions below provide further	162
detail about statistical principles relevant to the validation of a surrogate endpoint.	163
164
• 	Individual-level association is the strength of the association between the surrogate and	165
the true clinical endpoint.	166
167
• 	Trial-level association is the strength of the association between the effects of treatment	168
on the surrogate and the true endpoint.	169
170
Although single-arm trial data may be used to demonstrate individual-level association and	171
assess efficacy outcome of interest in subgroups by MRD level for the purposes of hypothesis	172
generation, the meta-analysis to validate MRD at the trial level should include only randomized	173
trials. The issues pertinent to meta-analyses in general have been discussed in another	174
guidance.12	
175
176
Sponsors should discuss with the Agency and provide details of the meta-analysis plan, which	177
should include but not be limited to considering the following aspects:	178
179
• 	Details of trial designs, inclusion and exclusion criteria, MRD assessment, and disease	180
setting. Sponsors should justify poolability of the data.	181
182
periodically. To make sure you have the most recent version of a guidance, check the FDA guidance web page at
https://www.fda.gov/regulatory-information/search-fda-guidance-documents.
10 For additional information on expedited programs and surrogate endpoints used to support accelerated or
traditional approval, see the guidance for industry Expedited Programs for Serious Conditions—Drugs and
Biologics (May 2014).
11 See the ICH guidance for industry E9 Statistical Principles for Clinical Trials (September 1998).
12 For additional information on meta-analyses, see the draft guidance for industry Meta-Analyses of Randomized
Controlled Clinical Trials to Evaluate the Safety of Human Drugs or Biological Products (November 2018). When
final, this guidance will represent FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
6
• 	Inclusion of trials that include a patient population representative of the population in	183
which the surrogate endpoint ultimately will be used.	184
185
• 	Inclusion of an adequate number of randomized trials with sufficient follow-up time.	186
Sponsors should justify the number of trials to be included in the meta-analysis.	187
188
• 	Inclusion of trials that demonstrate both positive and negative results. For example,	189
sponsors should present randomized trials that failed to meet their primary endpoint, and	190
trials that had divergent MRD and event-free survival/progression-free survival/overall	191
survival (EFS/PFS/OS) results, if available, should also be presented. Sponsors should	192
explain the divergent results if possible.	193
194
• 	Analysis based on individual patient-level data to allow an assessment of individual-level	195
surrogacy.	196
197
• 	Prespecified criteria for concluding surrogacy based on both trial-level and patient-level	198
association measures, including prespecified timing and window of the MRD assessment.	199
If a fixed time point is not feasible, the MRD assessments in a window of the trial should	200
be prespecified. Sponsors should explore sensitivity analyses based on different time	201
windows. Sponsors should discuss with the Agency the time window chosen. For	202
example, sponsors can prespecify for patients with newly diagnosed ALL the MRD	203
assessment at the time of the first complete response (CR1), 28 days plus or minus a	204
window of a specific number of days.	205
206
• 	Inclusion of long-term clinical endpoints, such as EFS/PFS and OS that have been clearly	207
and consistently defined across studies. Sponsors should explore alternative event	208
definitions for EFS/PFS or alternative censoring schemes for EFS/PFS/OS as sensitivity	209
analyses.	210
211
• 	Discussion of missing MRD assessments and reasons for missing data (e.g., caused by	212
sample collection issues, loss to follow-up). Sponsors should explore the effects of	213
missing data on the results.	214
215
• 	Consideration of the statistical handling of unevaluable samples.	216
217
• 	Potential confounding factors, which sponsors should incorporate into the planned	218
validation analyses.	219
220
• 	Sensitivity analyses to demonstrate robustness of the surrogate (e.g., alternative statistical	221
methods for evaluation of association,13 cross validation) and subgroup analyses.	222
13 Shi Q, CR Flowers, W Hiddemann, R Marcus, M Herold, A Hagenbeek, E Kimby, H Hochster, U Vitolo, BA
Peterson, E Gyan, M Ghielmini, T Nielsen, S De Bedout, T Fu, N Valente, NH Fowler, E Hoster, M Ladetto, F
Morschhauser, E Zucca, G Salles, and DJ Sargent., 2017, Thirty-Month Complete Response as a Surrogate End
Point in First-Line Follicular Lymphoma Therapy: An Individual Patient-Level Analysis of Multiple Randomized
Trials, J Clin Oncol, 35(5):552–560.

Contains Nonbinding Recommendations
7
223
• 	Discussion and analyses using different MRD cutoffs (e.g., 10 -4 , 10-5 ). For assisting in	224
the interpretation of the results, sponsors can present analyses such as surrogate threshold	225
effect.14	
226
227
Even if a meta-analysis supports validation of MRD as a surrogate endpoint, applying these	228
results to a new trial requires a certain amount of extrapolation. Some caveats regarding the use	229
of MRD as a surrogate endpoint include the following:	230
231
• 	Even if MRD can be validated as a surrogate endpoint, the use of MRD may not be	232
applicable to subgroups of the patient population or future trial populations if there are	233
important differences (e.g., prior therapy, disease status, line of treatment) between the	234
population evaluated in the meta-analysis and the to-be-enrolled population. This may	235
represent a different context of use, and as such, any differences should be justified.	236
Sponsors should perform sensitivity and subgroup analyses to evaluate the strength of the	237
surrogate endpoint in different disease settings or patient characteristics.	238
239
• 	When a new drug product is under investigation, it may not be reasonable to assume that	240
the quantitative relationship between the drug product’s effects on the surrogate endpoint	241
and the clinical benefit endpoint will be the same as previously studied drug products’	242
effects. This is especially true for drug products that have a markedly different	243
mechanism of action (e.g., cytotoxic therapy versus immunotherapy). Although the	244
credibility of this extrapolation will be primarily based on biological considerations, the	245
meta-analyses mentioned above can provide supportive evidence. To obtain best	246
estimates of the relationship between the surrogate and clinical benefit endpoints, the	247
meta-analysis should include drug products with varying mechanisms of action and	248
evaluate the relationship in mechanistic subtypes.	249
250
D. 	MRD as an Endpoint in Clinical Trials	251
252
MRD analyses should be based on the intent-to-treat (ITT) population. A patient may not have	253
an MRD assessment because of a missed assessment, test failure, or not meeting clinical criteria	254
for assessment (i.e., lack of CR). For ITT-based analyses, sponsors should consider any patient	255
without an MRD assessment as not responsive to treatment. Analyses based on the MRD	256
evaluable population are appropriate for sensitivity analyses.	257
258
Missing and unevaluable assessments of MRD should be kept to a minimum. Sponsors should	259
collect and summarize reasons for missing MRD assessments and consult with the Agency	260
before finalizing the statistical analysis plan. Sponsors also should perform further exploratory	261
or sensitivity analyses to evaluate comparability of the results using different evaluation	262
populations.	263
264
14 Burzykowski T and Buyse M, 2006, Surrogate Threshold Effect: An Alternative Measure for Meta-Analytic
Surrogate Endpoint Validation, Pharm Stat, 5(3):173–186.

Contains Nonbinding Recommendations
8
Sponsors can also include MRD in the clinical trial as a secondary or an exploratory endpoint. If	265
MRD-negative response (e.g., MRD-negative CR) is used as a secondary endpoint and is	266
planned for inclusion in the prescribing information, it should be included as a key secondary	267
endpoint with appropriate control for multiplicity.15	
268
269
E. 	MRD for Patient Selection or Enrichment	270
271
Many clinical risk classifications may not be able to accurately predict relapse in patients with	272
hematologic malignancies, which may result in inappropriate use or timing of treatments. MRD	273
has been regarded as an important prognostic factor for predicting disease recurrence, which may	274
improve risk classification. Sponsors can use MRD level to serve as a stratification factor, select	275
patients at high risk, or enrich the trial population.16	
276
277
FDA recommends that sponsors consult the Agency about incorporating any MRD assay into a	278
trial before submitting the protocol for trials that include MRD for patient selection or as an	279
endpoint (primary or key secondary).	280
281
282
IV. 	TECHNOLOGY CONSIDERATIONS	283
284
A. 	Assay Considerations	285
286
Currently, four general technologies are used for MRD assessment in hematologic malignancies:	287
multiparametric flow cytometry (MPFC), next-generation sequencing (NGS), quantitative	288
reverse transcription polymerase chain reaction (RT-qPCR) of specific gene fusions, and allele-	289
specific oligonucleotide polymerase chain reaction (ASO-PCR). These cellular (MPFC) and	290
molecular (NGS, RT-qPCR, and ASO-PCR) technology platforms have different advantages and	291
limitations in terms of sample input, cost, robustness, and reproducibility.	292
293
FDA is agnostic as to which technology platform is used in clinical trials assessing MRD.	294
However, sponsors should fully prespecify the selected platform (in terms of assay procedure,	295
reagents, and analysis) and analytically validate the platform for its context of use. Also, in the	296
context of a clinical trial, ideally sponsors should use a single technology to assess MRD to	297
compare results directly. Although use of multiple technologies is discouraged, if the sponsor	298
includes multiple technologies in the trial and plans for the primary analysis to be based on data	299
from multiple technologies, the sponsor should prespecify the methodology for combining these	300
technologies into a single MRD determination and discuss this with the Agency.	301
302
Analytical validation ensures that the assay measures the analyte or analytes that it is intended to	303
measure in the intended tissue type. The process of analytical validation is defined as	304
establishing that the performance characteristics of the assay are acceptable in terms of the	305
15 See the draft guidance for industry Multiple Endpoints in Clinical Trials (January 2017). When final, this
guidance will represent the FDA’s current thinking on this topic.
16 See the guidance for industry Enrichment Strategies for Clinical Trials to Support Determination of Effectiveness
of Human Drugs and Biological Products (March 2019).

Contains Nonbinding Recommendations
9
assay’s sensitivity, specificity, accuracy, precision, and other relevant performance	306
characteristics using a specified technical protocol, which may include specimen collection,	307
handling, and storage procedures. Analytical validation is concerned with the assay’s technical	308
performance and does not address clinical utility.	309
310
MRD assay validation should encompass the entire assay system from sample collection (e.g.,	311
BM aspirate versus blood) to system output (e.g., decision-making threshold for MRD positive	312
versus negative) and should use relevant clinical samples. Where technically feasible, the	313
detection threshold of the MRD assay should be at least 10-fold below the clinical decision-	314
making threshold (the definition of MRD). For example, if MRD positive or negative is defined	315
as detection of greater or less than 1 x 10 -5 cells, respectively, then the assay should be optimized	316
and validated to have an analytical sensitivity of at least 1 x 10-6 . If this level of detection is not	317
feasible with the proposed assay, sponsors should provide appropriate justification that the assay	318
is adequate to fulfill its intent in the trial. Additionally, to ensure that the assay performance	319
achieved in validation testing is replicated in the clinical trial, sponsors should strictly adhere to	320
the assay protocol in all clinical trial laboratory sites. The following sections detail specific	321
considerations for the different technology platforms.	322
323
1. 	Cellular Technology Platforms	324
325
When using cellular technology platforms for MRD assessments in clinical trials, sponsors	326
should do the following:	327
328
• 	Prespecify the total number of events to be collected to support the quantitative	329
assessment of MRD	330
331
• 	Use a consistent panel of antibodies and fluorochromes, as no single antigen is specific	332
for any neoplasm	333
334
• 	Consider sample stability, which may limit the utility of flow cytometry	335
336
• 	Use a consistent analysis template (e.g., gating strategy)	337
338
• 	Determine whether the therapy affects the expression and therefore detectability of the	339
specific antigens targeted by the antibody panels of the flow cytometry assay	340
341
• 	Evaluate the potential for the flow assay to detect normal BM cells that are regenerating	342
after chemotherapy to reduce the likelihood that those cells are misinterpreted as	343
abnormal cells	344
345
2. 	Molecular Technology Platforms	346
347
When using molecular technology platforms for MRD assessments in clinical trials, sponsors	348
should do the following:	349
350
• 	Prespecify nucleic acid quantity (e.g., micrograms) and quality metrics	351

Contains Nonbinding Recommendations
10
352
• 	Consider the need for an internal control when a cell number is derived from DNA	353
content calculations because poor DNA quality may cause artificially low MRD levels	354
355
• 	Store diagnostic samples to be used for clone identification in case of assay changes	356
357
• 	Consider how to account for shifts in clonality as assessed by molecular markers (i.e., the	358
specific molecular marker may be lost as a result of treatment while the disease remains	359
present)	360
361
• 	Track assay failures (i.e., failures of the assay to identify the relevant clone for a patient)	362
and consider this failure rate for clinical endpoint calculations	363
364
3. 	All Technology Platforms	365
366
When using any technology platform for MRD assessments in clinical trials, sponsors should do	367
the following:	368
369
• 	Prespecify preanalytical procedures and ensure that the sample collection and storage	370
procedures used are appropriate to obtain the desired cell population	371
372
• 	Take hemodilution into account (specifically, the amount of blood needed for the	373
procedure to obtain the required number of events or amount of nucleic acid) and request	374
that investigators use the first BM pull for MRD assessments	375
376
• 	For all testing, especially if centralized testing is not used, assay protocols and result	377
interpretation should be standardized to ensure MRD measurements are comparable	378
between laboratories	379
380
B. 	Sampling Considerations	381
382
Target levels of MRD for use in a regulatory setting are disease-specific and dependent upon the	383
proposed use of the biomarker. In a clinical trial, the protocol should prespecify the	384
measurement of MRD, which sponsors should conduct at prespecified times using a consistent	385
and validated assay. The MRD assessment at a prespecified postinduction therapy time point is	386
anticipated to be a sensitive measure of CR to induction therapy in either a frontline or	387
relapsed/refractory setting. Consistent time-point specification would provide an opportunity to	388
assess the kinetics of an MRD response and its duration, which may provide supportive evidence	389
of drug activity. The timing of MRD assessment also is important when considering using MRD	390
before allogeneic hematopoietic cell transplantation to predict transplant outcomes.	391
392
393

Contains Nonbinding Recommendations
11
V. 	DISEASE-SPECIFIC CONSIDERATIONS	394
395
A. 	Acute Lymphoblastic Leukemia	396
397
MRD has emerged as one of the most significant prognostic factors in ALL, independent of	398
patient age, B- or T-cell origin, or genetic subtype. Additional considerations for using MRD in	399
ALL treatment trials include the following:	400
401
• 	BM is the preferred substrate for measuring MRD. If blood samples are used for	402
assessing MRD in the clinical trial, sponsors should include justification for using blood	403
rather than BM.	404
405
• 	CR with recovery of blood counts is the preferred time point to assess MRD. For	406
regimens for which the efficacy-response evaluation is based on a calendar-driven time	407
point rather than waiting for blood count recovery, at least an M1 marrow (marrow with	408
leukemic blasts less than 5%) should be documented for patients being assessed for	409
MRD.	410
411
• 	When MRD is used as an efficacy endpoint for ALL, the absence of extramedullary	412
disease should be documented concurrently with assessment of BM and blood counts.	413
However, FDA does not expect the conduct of invasive procedures to test for	414
extramedullary disease if the procedures are not within the clinical standard of care at the	415
time of the efficacy evaluation.	416
417
• 	FDA has accepted an MRD level of 0.1% or more to define patients with ALL in first or	418
second CR with high risk of relapse. For trials that use MRD levels of less than 0.1%	419
with CR for patient selection, the submission should include information to justify use of	420
the lower MRD level.	421
422
• 	For new drugs that have a demonstrated durable CR in patients with relapsed or	423
refractory ALL, FDA has accepted MRD of less than 0.01% as supporting evidence of	424
efficacy. As technologies improve and new clinical findings emerge, the level of MRD	425
needed to support an efficacy claim may change.	426
427
B. 	Acute Myeloid Leukemia	428
429
The molecular heterogeneity of AML poses substantial challenges to the use of MRD as a	430
biomarker. Additional considerations for use of MRD in AML treatment trials include the	431
following:	432
433
• 	BM is the preferred substrate for measuring MRD. If blood samples are used for	434
response assessment of MRD in the clinical trial, sponsors should include justification for	435
using blood rather than BM.	436
437
• 	CR with recovery of blood counts is the preferred time point to assess MRD. If	438
assessments are made at CR without count recovery or at lesser responses (e.g., complete	439

Contains Nonbinding Recommendations
12
remission with incomplete hematologic recovery), sponsors should include data to justify	440
the plan.	441
442
• 	For the marker (e.g., cell surface or genetic mutation) selected to assess MRD, sponsors	443
should provide data showing that the marker reflects the leukemia and not underlying	444
clonal hematopoiesis (false-positive result). Sponsors should also describe the false-	445
negative rate that might result from relapse from a marker-negative clone. If multiple	446
markers and/or multiple platforms are used, sponsors should provide an analysis of the	447
risk of false-positive and false-negative results for each marker individually and for the	448
panel as a whole.	449
450
• 	For studies of targeted therapies (e.g., IDH1, IDH2, or FLT3 inhibitors) for which the	451
MRD marker is the target of the therapy, sponsors can use nonclinical data to identify the	452
mutations in the marker that are known to be sensitive to the therapy and those that are	453
known to be resistant to the therapy. If using only the target of therapy as the MRD	454
marker, sponsors should provide justification for not using other MRD markers to avoid	455
false-negative results when clonality changes.	456
457
C. 	Acute Promyelocytic Leukemia	458
459
The standard-of-care use of MRD testing and monitoring is established for the initial treatment	460
of patients with acute promyelocytic leukemia (APL) using tretinoin with arsenic and/or	461
anthracycline. Whether the same guidelines for use of MRD apply to other drug classes needs to	462
be confirmed as new drugs are evaluated for initial or salvage therapy. Additional specific	463
considerations include the following:	464
465
• 	BM is the preferred substrate for measuring MRD. If blood samples are used for	466
response assessment of MRD in the clinical trial, sponsors should include justification for	467
using blood rather than BM.	468
469
• 	CR following recovery of blood counts is the preferred time point to assess MRD. If	470
assessments are made at CR without count recovery or at lesser responses, sponsors	471
should include data to justify the plan.	472
473
• 	MRD should be assessed at the end of consolidation rather than at the end of induction,	474
when differentiating agents are used, to avoid false-positive results. For new drug	475
products for treatment of APL, sponsors should use data from early-phase trials to	476
establish the optimal timing for MRD assessment in the pivotal trials.	477
478
• 	Patients with low-risk APL who achieve confirmed MRD negativity after	479
arsenic/tretinoin-based therapy are generally considered cured and require no further	480
monitoring. For new drug products for treatment of APL, long-term monitoring may be	481
required in the pivotal trial if data from early-phase trials are not sufficient to confirm	482
that MRD negativity is also durable with the new drug product.	483
484

Contains Nonbinding Recommendations
13
• 	An MRD level less than 0.01% is generally considered negative after first-line	485
arsenic/tretinoin- or idarubicin/tretinoin-based induction. For new drug products for	486
treatment of APL, sponsors should use data from early-phase trials to confirm this	487
threshold for defining MRD negativity for the new drug product.	488
489
• 	Although an MRD level less than 0.01% is generally considered negative after first-line	490
treatment, marketing applications for treatment of molecular relapse may need clinical	491
outcomes (i.e., EFS) if data are not available to support a proposed MRD threshold as the	492
sole criterion for response to salvage therapy.	493
494
D. 	Chronic Lymphocytic Leukemia	495
496
Current literature suggests that there is an association between MRD negativity and OS in	497
patients with CLL treated with chemoimmunotherapy. The therapeutic paradigm with small	498
molecule inhibitors of the B-cell receptor signaling pathway and other novel products continue to	499
rapidly evolve in this area. Additional specific considerations include the following:	500
501
• 	MRD status should be measured by a standardized method with a quantitative lower limit	502
of detection sufficient to evaluate the prospective cutoff in the trial and at least less than	503
10-4 (0.01%). Currently, MRD is most commonly assessed using RT-qPCR and flow	504
cytometric methods, but NGS can also reliably assess MRD in CLL.	505
506
• 	A challenge in MRD testing is that CLL is a multicompartmental disease involving the	507
BM, blood, lymph nodes, liver, and spleen; after treatment, one or more of these sites	508
may serve as a reservoir for residual disease. Sponsors should carefully consider for	509
assessment the sample source, which should be the same throughout the trial. This is	510
especially important as therapeutic intervention differentially affects MRD measurement	511
in peripheral blood and BM, as has been demonstrated with certain therapeutics (e.g.,	512
anti-CD20 monoclonal antibodies, alemtuzumab).	513
514
• 	The timing of when to test for MRD has yet to be standardized and the time to response	515
and response durations may vary by type of therapeutic regimen. Sponsors should	516
prespecify the timing and method of MRD testing and provide adequate justification in	517
the protocol. MRD should also be measured at the end-of-treatment response assessment	518
to fully capture the treatment effect.	519
520
• 	MRD should be assessed in patients who are in CR. If MRD assessments are to be made	521
in patients in other response categories (e.g., partial response), sponsors should include	522
data to justify the plan.	523
524
E. 	Chronic Myeloid Leukemia	525
526
There have been dramatic improvements in clinical outcomes in patients with chronic myeloid	527
leukemia (CML) from targeting the BCR-ABL1 oncoprotein. The detection and monitoring of	528
MRD has become standard of care in patients with CML. Specific considerations include the	529
following:	530

Contains Nonbinding Recommendations
14
531
• 	Monitoring MRD in CML should use assays with results based on the International Scale	532
(IS) with the standardized baseline set to 100%. Molecular response is expressed as log	533
reduction from 100%.	534
535
• 	Currently, RT-qPCR(IS) is the preferred assay to monitor response to therapy. In	536
general, RT-qPCR assays with a sensitivity of more than 4.5-log reduction from the	537
standardized baseline are recommended for measuring BCR-ABL1 transcripts.	538
539
• 	Major molecular response (MMR) is defined as BCR-ABL(IS) of less than 0.1% or more	540
than 3-log reduction in BCR-ABL1 mRNA from the standardized baseline if RT-	541
qPCR(IS) is not available.	542
543
• 	There is evidence that achieving an MMR predicts superior long-term clinical outcomes	544
(PFS/EFS).	545
546
• 	Achieving MMR has become a consensus goal of CML therapy, and durable MMR can	547
be a measure of clinical benefit.	548
549
• 	In addition, MRD can be used to select and monitor patients who are eligible for	550
treatment discontinuation of tyrosine kinase inhibitor therapy.	551
552
F. 	Multiple Myeloma	553
554
Significant improvements in clinical outcomes of MM have spurred interest in the use of MRD	555
as a potential surrogate endpoint to expedite drug development. Multiple trials have evaluated	556
the relationship between MRD status and PFS/OS.	557
558
Additional specific considerations for use of MRD in trials of new drug products for treatment of	559
MM include the following:	560
561
• 	Most published literature to date has evaluated MRD in the newly diagnosed	562
posttransplant setting. Fewer studies have evaluated MRD in the setting of	563
relapsed/refractory disease or newly diagnosed patients with myeloma who are not	564
eligible for transplant. The relationship between MRD and clinical benefit and the test	565
performance characteristics should be demonstrated in each disease setting (e.g., relapsed	566
refractory, newly diagnosed, nontransplant eligible, smoldering MM) to validate MRD as	567
a surrogate endpoint in MM. This is especially important in disease settings, such as	568
smoldering myeloma, in which there is a lower disease burden and the potential for	569
toxicity or other nondisease-related factors influence long-term outcomes.	570
571
• 	MRD should be assessed only in patients who are in CR. If MRD assessments are to be	572
made in patients in other response categories (e.g., partial response, very good partial	573
response), sponsors should include data to justify the plan.	574
575

Contains Nonbinding Recommendations
15
• 	MRD is currently assessed using MPFC and NGS methods in the BM. These	576
methodologies are not able to detect extramedullary disease. There has been interest in	577
using imaging techniques (e.g., positron emission tomography/computed tomography,	578
magnetic resonance imaging) in combination with MRD to assess response. When	579
considering using MRD in MM clinical trials, sponsors should discuss with FDA how	580
extramedullary disease will be assessed and whether imaging should be incorporated into	581
the assessment of response.	582
583
• 	At this time, the relationship between MRD and clinical benefit in patients with different	584
cytogenetic abnormalities and their associated risks is unclear. When considering using	585
MRD in clinical trials, sponsors should consider the patient’s cytogenetic risk. For	586
example, given the prognostic effect of cytogenetics, the trial may benefit from	587
stratification to ensure that there is no imbalance between the arms that would affect the	588
MRD assessment. Alternatively, trials may be designed to intervene in patients who are	589
MRD positive and have poor risk cytogenetics because this may represent a group at risk	590
for particularly poor outcomes.	591
592
593
VI. 	REGULATORY SUBMISSIONS THAT USE MRD	594
595
As indicated above, FDA views MRD as a biomarker that is a reliable quantitation of tumor	596
burden, independent of assay. As such, FDA does not foresee the need to codevelop an MRD	597
assay with a drug product.17 However, for FDA to adequately assess the safety of a proposed	598
clinical trial that uses MRD (e.g., for patient selection) or to determine the credibility of a	599
clinical trial outcome based in part on MRD, submissions that use MRD for regulatory purposes	600
or for critical treatment purposes should include sufficient information to address the following	601
two main issues:	602
603
• 	Is MRD, as assessed (sample, timing, threshold, etc.), a clinically valid biomarker for the	604
context of use (disease, disease status, type of therapy, etc.)?	605
606
• 	Is the MRD assay used (or to be used) in the clinical trial analytically valid for the range	607
of results that are important to the trial?	608
609
When the MRD assay used is FDA-cleared or -approved for the specific malignancy and	610
specimen type, identifying the assay with the required number of cells to be evaluated or the	611
DNA input requirements will be sufficient to address the analytical validity in most cases. When	612
the MRD assay is not FDA-cleared or -approved, FDA would expect additional information,	613
such as those listed in Table 1, to be submitted for review.	614
615
17 A potential exception might be when the MRD marker is the direct target of the drug product under study, such as
for selecting patients for treatment in a clinical trial of an Fms-related tyrosine kinase 3 (FLT3) inhibitor when the
MRD assay is for an FLT3 mutation. In such a circumstance, sponsors should consult with FDA about the need for
a companion diagnostic early in clinical development.

Contains Nonbinding Recommendations
16
Table 1. Information to Help Review of Regulatory Submissions That Use MRD	616
IND Clinical Trial Submission 	NDA or BLA Submission
1. Justification that MRD as used is clinically valid
for the proposed context and
2. Letter of authorization to cross-reference the
investigational device exemption or other device-
related regulatory submission for information
about the assay or
• A statement of intended use;
• The specific test method (including instruments,
reagents, and specimen handling);
• Confirmation that the lab has a process in place
for reagent control;
• A brief discussion of how the test method was
validated analytically for each specimen type;
and
• A summary of the performance obtained for
accuracy, precision, specificity, and sensitivity;
and
3. Indicate in the clinical trial informed consent
document that the MRD assay is investigational.
1. Justification that MRD as used is clinically valid
for the context of the proposed claim and
2. Letter of authorization to cross-reference the
investigational device exemption or other device-
related regulatory submission for information
about the assay or
• A statement of intended use;
• The specific test method (including instruments,
reagents, and specimen handling);
• Confirmation that the lab has a process in place
for reagent control;
• A brief discussion of how the test method was
validated analytically for each specimen type;
and
• A summary of the performance obtained for
accuracy, precision, specificity, and sensitivity;
and
3. A SAS XPORT file (xpt file extension) with results
of MRD testing. For each result, specify the
sample type, date of sample, assay used, input
quantity, assay sensitivity, and assay result.
617
618
For an IND clinical trial submission, when using an MRD assay that is not FDA-cleared or -	619
approved for the intended use and the trial is considered a significant risk device trial (e.g.,	620
eligibility criterion, allocation to a specific treatment, departures from standard of care, etc.),	621
FDA may require an investigational device exemption to use the assay in the clinical trial.18	
622
Sponsors can submit a letter of authorization to cross-reference the investigational device	623
exemption, which will then provide the necessary information regarding the assay. When the	624
trial is considered a nonsignificant risk device study, the sponsor should submit abbreviated	625
information about the assay (see Table 1) to the IND for review to allow FDA to confirm that	626
results from the device will be interpretable. An NDA or BLA submission should include	627
similar information about the assay (see Table 1) in addition to a data file with the results of	628
MRD testing.	629
630
Although general principles outlined in this guidance should help sponsors with crucial questions	631
about potential MRD use for marketing applications, FDA recommends that sponsors meet with	632
18 As an alternative, sponsors can consider the streamlined approach to codeveloping the MRD assay with the drug
product under the IND as described in the draft guidance Investigational In Vitro Diagnostics in Oncology Trials:
Streamlined Submission Process for Study Risk Determination (April 2018). When final, this guidance will
represent the FDA’s current thinking on this topic. See 21 CFR 812; for information on risk determination for
investigational use of devices, see the guidance for industry and FDA staff Requests for Feedback and Meetings for
Medical Device Submissions: The Q-Submission Program (May 2019).

Contains Nonbinding Recommendations
17
FDA before starting a drug development pathway incorporating MRD assessment intended to	633
support an NDA or a BLA. FDA will ensure that these meetings include a multidisciplinary	634
team of review staff from CBER, CDER, and CDRH, as needed. Sponsors can submit protocols	635
using MRD after these meetings and request a special protocol assessment for eligible protocols,	636
if they choose, that provides confirmation of the acceptability of assessments, endpoints, and	637
protocol design to support drug marketing applications. Ultimately, marketing approval depends	638
not only on the design of clinical trials but also on FDA review of the results and data from all	639
studies in the drug marketing application.	640
641
642
643

Contains Nonbinding Recommendations
18
GLOSSARY	644
645
ALL 	acute lymphoblastic leukemia	646
AML 	acute myeloid leukemia	647
APL 	acute promyelocytic leukemia	648
ASO-PCR 	allele-specific oligonucleotide polymerase chain reaction	649
BLA 	biologics license application	650
BM 	bone marrow	651
CBER 	Center for Biologics Evaluation and Research	652
CDER 	Center for Drug Evaluation and Research	653
CLL 	chronic lymphocytic leukemia	654
CML 	chronic myeloid leukemia	655
CR 	complete response or complete remission	656
DDT 	drug development tool	657
EFS 	event-free survival	658
IDE 	investigational device exemption	659
IND 	investigational new drug application	660
IS 	International Scale	661
ITT 	intent-to-treat	662
MM 	multiple myeloma	663
MMR 	major molecular response	664
MPFC 	multiparametric flow cytometry	665
MRD 	minimal residual disease	666
NDA 	new drug application	667
NGS 	next-generation sequencing	668
OS 	overall survival	669
PFS 	progression-free survival	670
RT-qPCR 	quantitative reverse transcription polymerase chain reaction	671