Acute Myeloid Leukemia: Developing Drugs and Biological Products for Treatment: Draft Guidance for Industry

Administrative draft guidance, not enacted legislation.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

Acute Myeloid
Leukemia: Developing
Drugs and Biological
Products for Treatment
Guidance for Industry
DRAFT GUIDANCE
This guidance document is being distributed for comment purposes only.
Comments and suggestions regarding this draft document should be submitted within 60 days of
publication in the Federal Register of the notice announcing the availability of the draft
guidance. Submit electronic comments to https://www.regulations.gov. Submit written
comments to the Dockets Management Staff (HFA-305), Food and Drug Administration, 5630
Fishers Lane, Rm. 1061, Rockville, MD 20852. All comments should be identified with the
docket number listed in the notice of availability that publishes in the Federal Register.
For questions regarding this draft document, contact (CDER) Donna Przepiorka at 301-796-5358
or Donna.Przepiorka@fda.hhs.gov, or (CBER) Office of Communication, Outreach and
Development at 800-835-4709 or 240-402-8010.
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)
August 2020
Clinical/Medical

Acute Myeloid
Leukemia: Developing
Drugs 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., 4 th 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)
August 2020
Clinical/Medical

Contains Nonbinding Recommendations
Draft — Not for Implementation
TABLE OF CONTENTS
I. 	INTRODUCTION............................................................................................................. 1
II. 	BACKGROUND ............................................................................................................... 1
III. 	DEVELOPMENT PROGRAM ....................................................................................... 2
A. 	General Drug Development Considerations ................................................................................ 2
1. Nonclinical ....................................................................................................................................... 2
2. Devices ............................................................................................................................................. 3
3. Clinical Pharmacology .................................................................................................................... 3
4. Special Populations ......................................................................................................................... 5
5. Safety Reporting Considerations ..................................................................................................... 7
B. 	Efficacy Endpoints ......................................................................................................................... 8
1. Time-to-Event Endpoints Used Commonly for AML ....................................................................... 8
2. Binary Endpoints Used Commonly for AML ................................................................................. 10
3. Minimal Residual Disease (MRD)-Based Endpoints ..................................................................... 13
4. Other Potential Measures of Efficacy for AML ............................................................................. 13
C. 	Exploratory Trial Considerations .............................................................................................. 14
1. First-in-Human (FIH) Trials ......................................................................................................... 14
2. Exploratory Trial Population ........................................................................................................ 15
3. Dose-Escalation Trials .................................................................................................................. 16
4. Exploratory Expansion Cohorts .................................................................................................... 18
D. 	Confirmatory Trial Considerations ........................................................................................... 19
1. Confirmatory Trial Population ...................................................................................................... 19
2. Dose Selection and Treatment Plan ............................................................................................... 20
3. Confirmatory Trial Design ............................................................................................................ 20
4. Confirmatory Trial Procedures ..................................................................................................... 23
IV. 	REGULATORY SUBMISSIONS.................................................................................. 24
A. 	Investigational New Drug Applications ..................................................................................... 24
B. 	Marketing Applications ............................................................................................................... 25
1. Assessment of Efficacy ................................................................................................................... 25
2. Assessment of Safety ...................................................................................................................... 25
3. Clinical Pharmacology .................................................................................................................. 26
GLOSSARY................................................................................................................................. 27
APPENDICES ............................................................................................................................. 29
Appendix 1: Example DLT Criteria for Drugs for AML ................................................................ 29
Appendix 2: Additional Statistical Discussion .................................................................................. 30
Appendix 3: Additional Data Files for Marketing Applications for AML Drugs ......................... 32

Contains Nonbinding Recommendations
Draft — Not for Implementation
1
Acute Myeloid Leukemia: Developing Drugs and Biological	1
Products for Treatment	2
Guidance for Industry1
3
4
5
This draft guidance, when finalized, will represent the current thinking of the Food and Drug	6
Administration (FDA or Agency) on this topic. It does not establish any rights for any person and is not	7
binding on FDA or the public. You can use an alternative approach if it satisfies the requirements of the	8
applicable statutes and regulations. To discuss an alternative approach, contact the FDA staff responsible	9
for this guidance as listed on the title page.	10
11
12
13
14
I. 	INTRODUCTION	15
16
The purpose of this guidance is to assist sponsors in the clinical development of drugs and	17
biological products2 for the treatment of acute myeloid leukemia (AML). Specifically, this	18
guidance addresses FDA’s current thinking regarding the overall development program and	19
clinical trial designs for the development of drugs to support an indication of treatment of	20
AML, including indications limited to an individual phase of treatment (e.g., maintenance,	21
transplantation preparative regimen, etc.).3	
22
23
In general, FDA’s guidance documents do not establish legally enforceable responsibilities.	24
Instead, guidances describe the Agency’s current thinking on a topic and should be viewed only	25
as recommendations, unless specific regulatory or statutory requirements are cited. The use of	26
the word should in Agency guidances means that something is suggested or recommended, but	27
not required.	28
29
30
II. 	BACKGROUND	31
32
AML is a malignant neoplasm arising from a myeloid-lineage progenitor. Although the disease	33
is clonal, the molecular pathogenesis is highly heterogeneous. The International Agency for	34
Research on Cancer classifies AML and related neoplasms on the basis of morphological,	35
1 This guidance has been prepared by the Oncology Center of Excellence, the Center for Drug Evaluation and
Research, and the Center for Biologics Evaluation and Research in consultation with the Center for Devices and
Radiological Health at the Food and Drug Administration.
2 For the purposes of this guidance, references to drugs include drugs approved under section 505 of the Federal
Food, Drug, and Cosmetic Act (the FD&C Act) (21 U.S.C. 355) and biological products licensed under section 351
of the Public Health Service Act (42 U.S.C. 262).
3 In addition to consulting guidances, sponsors are encouraged to contact the appropriate review division to discuss
specific issues that arise during the development of drugs for the treatment of AML.

Contains Nonbinding Recommendations
Draft — Not for Implementation
2
clinical, and genomic parameters, including specific genetic abnormalities.4 The median age at	36
diagnosis is 68 years, but the disorder occurs in patients of all ages from neonates to the elderly.5	
37
For decades, the standard treatment6 for patients with AML was intensive cytotoxic	38
chemotherapy for induction and consolidation with or without postremission allogeneic	39
hematopoietic stem cell transplantation (HSCT), and the only intent of treatment was cure.	40
Investigations of cytotoxic drugs with or without targeted drugs continue in an effort to increase	41
the fraction of patients with AML who are cured. However, many patients with AML who had	42
just mild pathological or age-related organ impairment at diagnosis were considered to have too	43
high a risk of life-threatening or fatal organ toxicity from such intensive therapy and therefore	44
were offered only palliative treatments or no treatment at all.	45
46
New classes of drugs, including drugs that target the specific pathogenetic mutations or a	47
disordered epigenome, are being developed as alternatives to cytotoxic drugs for the treatment of	48
AML. In some cases, these newer approaches may extend survival without the prospect for cure,	49
but extending survival may be a meaningful benefit for patients who would live for only weeks if	50
left untreated. Inducing temporary control of disease with minimal treatment burden and	51
palliation of symptoms are two additional outcomes that might also be considered meaningful in	52
certain circumstances (see discussion in III.B below).	53
54
The expansion of treatment intent, broadening of the intended population, and development of a	55
wide range of new drug classes as alternatives to cytotoxic drugs contribute substantially to the	56
complexity of clinical development programs for new drugs for AML. This guidance addresses	57
these considerations and provides recommendations regarding the design and conduct of clinical	58
trials and the types of supporting data that would facilitate efficient development of drugs for the	59
treatment of AML.7	
60
61
62
III. 	DEVELOPMENT PROGRAM	63
64
A. 	General Drug Development Considerations	65
66
1. 	Nonclinical	67
68
• 	The Agency’s expectations for the nonclinical programs for treatments of	69
malignancies are summarized in the ICH guidances for industry S9 Nonclinical	70
4 For examples, see Swerdlow SH, Campo E, Harris NL, et al (eds), WHO Classification of Tumours of
Haematopoietic and Lymphoid Tissues, 2017. Consult www.iarc.fr for resources with the latest diagnostic criteria
for AML classification (accessed July 16, 2020).
5 National Cancer Institute SEER Stat Fact Sheets: Acute myeloid leukemia. Available from:
http://seer.cancer.gov/statfacts/html/amyl.html (accessed July 16, 2020).
6 See the Glossary for definitions of the AML treatment and disease-related terms used in this guidance.
7 This guidance does not contain discussion of the general issues of statistical analysis or clinical trial design. Those
topics are addressed in the draft ICH guidance for industry E8(R1) General Considerations for Clinical Studies
(May 2019; when final, this guidance will represent the FDA’s current thinking on this topic) and the ICH guidances
for industry E9 Statistical Principles for Clinical Trials (September 1998) and E10 Choice of Control Group and
Related Issues in Clinical Trials (May 2001). 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
Draft — Not for Implementation
3
Evaluation for Anticancer Pharmaceuticals (March 2010) and S9 Nonclinical	71
Evaluation for Anticancer Pharmaceuticals Questions and Answers (June 2018).	72
These guidances apply to drugs for AML.	73
74
• 	For cellular or gene therapy products being developed for the treatment of AML,	75
sponsors should also consult the guidances for industry Preclinical Assessment of	76
Investigational Cellular and Gene Therapy Products (November 2013) and Long	77
Term Follow-Up after Administration of Human Gene Therapy Products (January	78
2020).	79
80
2. 	Devices	81
82
• 	For drugs with a specific therapeutic target, an in vitro companion diagnostic device	83
(referred to as a “companion diagnostic” herein) may be essential for the safe and	84
effective use of the drug. Sponsors developing a targeted drug for AML should take	85
into consideration the need for a companion diagnostic early in the drug development	86
timeline.8	
87
88
• 	Minimal residual disease (MRD) is a biomarker of subclinical tumor burden in	89
patients with AML. In clinical development programs for new AML drugs, MRD	90
assays might be used for selection of patients for participation in protocols,	91
assignment of treatments by prognostic subcategories, or as a measure of efficacy.	92
The guidance for industry Hematologic Malignancies: Regulatory Considerations for	93
Use of Minimal Residual Disease in Development of Drug and Biological Products	94
for Treatment (January 2020) provides recommendations about use of MRD and	95
MRD assays in regulatory submissions for drugs or biologics, including those	96
applicable to AML drugs.	97
98
3. 	Clinical Pharmacology	99
100
• 	Patients with AML are commonly prescribed concomitant medication that are	101
substrates, inducers, or inhibitors of cytochrome P450 (CYP) enzymes. In particular,	102
triazole antifungals are moderate to strong CYP3A inhibitors commonly prescribed to	103
reduce the risk of invasive fungal infections in patients with AML. Such drugs may	104
increase the systemic exposure of new AML drugs that are metabolized by CYP3A	105
and may decrease the tolerability of new AML drugs that are CYP3A substrates.	106
Additional studies should be used to address this potential for harm:	107
108
− 	Sponsors should conduct in vitro metabolism studies to determine if a new AML	109
drug is a substrate, inhibitor, or inducer of CYP3A prior to conduct of the first-in-	110
human (FIH) trial.9	
111
112
8 For guidance pertaining to companion diagnostics, see the CDRH internet page on companion diagnostics
(https://www.fda.gov/MedicalDevices/ProductsandMedicalProcedures/InVitroDiagnostics/ucm407297.htm).
9 See the draft guidance for industry In Vitro Metabolism- and Transporter- Mediated Drug-Drug Interaction
Studies (October 2017). When final, this guidance will represent the FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
Draft — Not for Implementation
4
− 	If an AML drug is a CYP3A substrate, sponsors should proactively incorporate	113
strategies for dose modification with concomitant use of moderate and strong	114
CYP3A inhibitors early in their clinical development programs. If available,	115
sponsors may leverage pharmacokinetic data (e.g., exposure-response	116
relationships for safety and effectiveness, clinical drug interaction studies) from	117
patients with other malignancies who have received the new drug to estimate the	118
potential effect of the co-administration of the new drug with CYP3A inhibitors	119
and determine an appropriate dose of the new drug with moderate or strong	120
CYP3A inhibitors in patients with AML. The development of physiologically	121
based pharmacokinetic models may aid in assessing the effect of some CYP3A	122
modulators on the AML drug and should be considered.	123
124
− 	If the new AML drug is a substrate of, inhibits, or induces any major CYP	125
enzyme or other metabolic enzymes in vitro, sponsors should conduct clinical	126
drug interaction studies to determine appropriate mitigation strategies. FDA’s	127
draft recommendations regarding such studies are described in the draft guidance	128
for industry Clinical Drug Interaction Studies – Study Design, Data Analysis and	129
Clinical Implications.10	
130
131
• 	Common supportive care medications for patients with AML, including antimicrobial	132
prophylaxis (e.g., fluoroquinolones) and antiemetics (e.g., 5-HT3 receptor	133
antagonists), are known to prolong the QT interval. Sponsors should conduct an	134
adequate assessment early in clinical development to assess the QT prolongation	135
potential of the AML drug as described in FDA's guidance.11 If the AML drug has	136
the potential to prolong the QT interval, the protocols should include appropriate	137
strategies for mitigation of QT prolongation, including a list of prohibited	138
concomitant medications associated with QT prolongation and/or more frequent	139
monitoring of ECG and electrolytes, particularly in patients with nausea, vomiting, or	140
diarrhea.	141
142
• 	Patients with AML, especially the elderly, may have impaired hepatic or renal	143
function. Prior to enrolling patients with organ impairment on trials of treatments for	144
AML, the sponsor should identify elimination pathways of the parent drug and its	145
active metabolites. If renal or hepatic elimination pathways are identified, the	146
sponsor should characterize the impact of organ impairment on the pharmacokinetics	147
of the parent drug or active metabolites early in clinical development as described in	148
the FDA’s guidances.12 This provides the basis of dose modifications for patients	149
with organ impairment in late phase clinical studies.	150
151
10 October 2017. When final, this guidance will represent the FDA’s current thinking on this topic.
11 See the ICH guidance for industry E14 Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic
Potential for Non-Antiarrhythmic Drugs (October 2005).
12 See the draft guidance for industry Pharmacokinetics in Patients with Impaired Renal Function — Study Design,
Data Analysis, and Impact on Dosing and Labeling (March 2010) (when final, this guidance will represent FDA’s
current thinking on this topic) and the guidance for industry Pharmacokinetics in Patients with Impaired Hepatic
Function: Study Design, Data Analysis, and Impact on Dosing and Labeling (May 2003).

Contains Nonbinding Recommendations
Draft — Not for Implementation
5
4. 	Special Populations	152
153
a. 	Pediatric Patients	154
155
• 	FDA encourages sponsors to address the pediatric population early in their clinical	156
development program for drugs for the treatment of AML. For example, adolescent	157
patients should be considered for enrollment along with adults in trials for the	158
treatment of AML.13	
159
160
• 	When it is not clear that dosing for pediatric patients can be derived with certainty	161
from adult data, or for FIH studies in younger age groups, studies in children should	162
begin with a phase 1 trial of the new drug as monotherapy. The phase 1 monotherapy	163
trial population need not be limited to patients with AML, but the acceptability of the	164
recommended phase 2 dose (RP2D) should be confirmed in a small cohort of	165
pediatric patients with AML before conduct of larger trials for AML in children.	166
167
• 	Section 505B of the Federal Food, Drug, and Cosmetic Act (FD&C Act) requires that	168
certain marketing applications, those for certain drugs that are directed at a molecular	169
target that FDA determines to be substantially relevant to the growth or progression	170
of a pediatric cancer, contain reports of molecularly targeted pediatric cancer	171
investigations, unless a deferral or waiver is granted. The requirement for pediatric	172
investigations applies even if the drug is for an indication for which orphan	173
designation has been granted.14 Sponsors of molecularly-targeted AML drugs should	174
discuss the applicability of these requirements to their drug as early as end-of-phase 1	175
to allow sufficient time to develop a pediatric study plan, if needed.15	
176
177
b. 	Older Adult Patients	178
179
• 	For clinical trials of AML drugs, sponsors should enroll a population that is	180
representative of the age range of patients with the disease. It is acknowledged,	181
however, that older adults with AML may have age-related comorbidities that place	182
them at higher risk for adverse outcomes when treated with intensive chemotherapy.	183
FDA has accepted, but does not require, use of age 75 years as an upper limit for	184
inclusion in trials of intensive chemotherapy. FDA, however, encourages use of no	185
age limit for trials of nonintensive treatments for AML.	186
187
• 	Dose reductions may be required for older patients (e.g., age 65 years and older).	188
Safety, pharmacokinetic, and exposure data from older adults in early phase trials of a	189
13 See the guidance for industry Considerations for the Inclusion of Adolescent Patients in Adult Oncology Clinical
Trials (March 2019) and the guidance for industry and IRBs Cancer Clinical Trial Eligibility Criteria: Minimum
Age Considerations for Inclusion of Pediatric Patients (July 2020).
14 For additional information, see the draft guidance for industry FDARA Implementation Guidance for Pediatric
Studies of Molecularly Targeted Oncology Drugs: Amendments to Sec. 505B of the FD&C Act (December 2019).
When final, this guidance will represent the FDA’s current thinking on this topic.
15 For additional information see the guidance for industry Pediatric Study Plans: Content of and Process for
Submitting Initial Pediatric Study Plans and Amended Initial Pediatric Study Plans (July 2020).

Contains Nonbinding Recommendations
Draft — Not for Implementation
6
new AML drug should be used to justify the dose or dose modifications of the drug	190
for older adults to be tested in later phase trials.	191
192
• 	FDA recommends an assessment of older adults (e.g., age 65 years and older) for	193
physiologic function at study baseline to assist in identifying subgroups that may be	194
at risk for an adverse outcome when treated for AML. Sponsors may consider using	195
an available geriatric assessment tool or propose a new tool for use in the clinical	196
trials. A simple assessment tool evaluating single or multiple aspects of function with	197
limited burden to the patient is preferred. Sponsors are encouraged to request a	198
meeting as early as possible with FDA to discuss the incorporation of an existing or a	199
new assessment tool for older adult patients in AML clinical trials.	200
201
c. 	Patients with Organ Impairment	202
203
• 	For late phase clinical trials of AML drugs, sponsors should enroll a population that is	204
representative of patients diagnosed with AML, including those with impaired organ	205
function.16 Appropriate organ impairment studies should have been conducted or the	206
impact of organ impairment on the exposure of the parent drug and its active	207
metabolites assessed adequately to provide appropriate dose modifications as stated in	208
section III.A.3.	209
210
• 	For drugs being developed specifically for the treatment of AML in patients with pre-	211
existing comorbidities that preclude use of intensive chemotherapy, FDA has	212
accepted the following criteria to define the population to be included in the trials that	213
will support marketing approval:	214
215
‒ 	ECOG performance status > 2,	216
217
‒ 	Severe cardiac disorder (e.g., congestive heart failure requiring treatment, ejection	218
fraction ≤ 50%, or chronic stable angina),	219
220
‒ 	Severe pulmonary disorder (e.g., DLCO ≤ 65% or FEV1 ≤ 65%),	221
222
‒ 	Creatinine clearance < 45 mL/min, and	223
224
‒ 	Hepatic disorder with total bilirubin > 1.5 time the upper limit of normal.	225
226
FDA will consider additional criteria if sponsors can provide data to justify their	227
proposal.	228
229
16 For additional information, see the guidance for industry Cancer Clinical Trial Eligibility Criteria: Patients with
Organ Dysfunction or Prior or Concurrent Malignancies (July 2020).

Contains Nonbinding Recommendations
Draft — Not for Implementation
7
d. 	Pregnant Patients	230
231
• 	The AML population includes a substantial proportion of young adult females.	232
Pregnant women may be diagnosed with AML during the course of their pregnancy.	233
The standard of care in this circumstance is to administer intensive chemotherapy	234
after the first trimester. As such, pregnant women with AML in certain	235
circumstances may be considered for inclusion in AML clinical trials based on a	236
thorough benefit-risk evaluation and when the trial offers the possibility of direct	237
benefit to the woman and/or fetus that is unavailable outside the research setting.	238
239
• 	Data from relevant nonclinical studies to support safety in pregnant patients should be	240
available prior to enrolling pregnant women in AML clinical trials. In addition,	241
safety data for the drug from previous human exposure, even for indications other	242
than AML, should be included in the assessment of risks.	243
244
• 	When a pregnancy has been identified during an AML clinical trial, the risks and	245
benefits of continuing versus stopping investigational treatment should be reviewed	246
with the pregnant woman. A second informed consent process reflecting additional	247
benefit-risk considerations is advisable for women who choose to continue treatment	248
with the investigational drug during pregnancy.	249
250
• 	Sponsors should consider meeting with FDA early in drug development to discuss	251
when and how to include pregnant women in clinical trials. For a draft of additional	252
general points to consider when pregnant women are included in clinical trials, see	253
the draft guidance for industry Pregnant Women: Scientific and Ethical	254
Considerations for Inclusion in Clinical Trials.17	
255
256
5. 	Safety Reporting Considerations	257
258
• 	Patients with AML may have adverse events due to the underlying leukemia.	259
Additionally, many AML drugs are designed to be myelosuppressive and are	260
expected to result in complications from the cytopenias. Nonclinical studies and the	261
analysis of class effects may also establish expected toxicities for the investigational	262
drug. Sponsors should submit a list of the anticipated serious adverse events that the	263
sponsor does not plan to report individually in an expedited manner to FDA. An IND	264
safety report must be submitted to FDA if an aggregate analysis indicates that the	265
adverse events are occurring more frequently in the drug treatment group per 21 CFR	266
312.32(c)(1)(i)(C). Additional information can be found in the guidance for industry	267
and investigators Safety Reporting Requirements for INDs and BA/BE Studies	268
(December 2012).	269
270
• 	Although investigators are required to report all serious adverse events to the sponsor	271
immediately (312.64(b)), this requirement may be burdensome and not useful when a	272
large proportion of the serious adverse events are expected at a high rate, such as	273
17 April 2018. When final, this guidance will represent the FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
Draft — Not for Implementation
8
might occur with the cytopenic complications of treatment of AML. Sponsors may	274
propose an alternative reporting arrangement for investigators in the protocol or in a	275
specific waiver request to FDA, and FDA will provide comment on whether the	276
alternative reporting arrangement is acceptable. For early phase trials, the alternative	277
reporting arrangement is likely to be limited to an alternative timeframe for the	278
investigator to report a serious adverse event to the sponsor; not reporting a serious	279
adverse event at all would be unacceptable.	280
281
• 	Patients with AML may experience relapse while on treatment or during study	282
follow-up. AML-related events, such as relapse or death from relapse, should not be	283
submitted by the sponsor as an IND safety report unless there is evidence suggesting	284
a causal relationship between the investigational drug and the adverse event, such as	285
an aggregate analysis showing that relapse occurred more frequently in the	286
investigational treatment group.	287
288
B. 	Efficacy Endpoints	289
290
1. 	Time-to-Event Endpoints Used Commonly for AML	291
292
a. 	Overall Survival (OS)	293
294
• 	OS is defined as the time from randomization to the date of death from any cause.	295
296
• 	For patients who are alive at the data cut-off, the observations for time-to-event are	297
censored at the last date of documented survival.	298
299
b. 	Event-Free Survival (EFS)	300
301
• 	For studies of drugs for the treatment of AML, EFS is defined as the time from	302
randomization to the date of:	303
304
− 	Induction treatment failure (ITF),	305
306
− 	Relapse for those who have induction treatment success (e.g., complete remission	307
(CR)), or	308
309
− 	Death from any cause,	310
311
whichever comes first. For patients who achieve induction treatment success and are	312
alive and in remission at the data cut-off, EFS should be censored at the last	313
assessment date. See the discussion of duration of remission in section III.B.2.a.	314
315
• 	ITF is defined as failure to achieve the initial interim efficacy endpoint within a	316
prespecified period of time. For example, for studies of intensive induction regimens	317
for first-line treatment of AML, the recommended definition of ITF is failure to	318
achieve morphological CR within 42 days of start of the last cycle of induction	319

Contains Nonbinding Recommendations
Draft — Not for Implementation
9
chemotherapy. Because the induction period can be variable and prolonged, it raises	320
ambiguities about how to define time to treatment failure. Therefore, day 1 of	321
treatment should be assigned as the event date for patients with ITF.	322
323
• 	The credibility of the results of EFS analyses are highly dependent on the quality of	324
the data. Many of the data quality issues for EFS are similar to those encountered	325
when using progression-free survival for studies of treatments for solid tumors. For	326
additional general points to consider when using such an endpoint, see the guidance	327
for industry Clinical Trial Endpoints for the Approval of Cancer Drugs and Biologics	328
(December 2018).	329
330
c. 	Relapse-Free Survival (RFS)	331
332
• 	RFS may be acceptable as an endpoint specifically in studies of treatments for	333
patients with AML in remission, such as the consolidation or maintenance phases.	334
335
• 	For studies of drugs for the treatment of AML, RFS is defined as the time from	336
randomization to the date of relapse or the date of death from any cause, whichever	337
comes first. For patients alive and in remission at the data cut-off, RFS should be	338
censored at the last assessment date.	339
340
d. 	Statistical Considerations for Time-to-Event Endpoints	341
342
• 	The general principles for the design and analysis of clinical trials as outlined in ICH	343
E9 apply to trials of treatments for AML. The bullets below are additional	344
considerations specific to AML trials and can also be thought of as discussing	345
specific attributes of the estimand concept, which is further discussed in the draft ICH	346
guidance for industry E9(R1) Statistical Principles for Clinical Trials: Addendum:	347
Estimands and Sensitivity Analysis in Clinical Trials.18	
348
349
• 	For time-to-event endpoints in a randomized trial for AML, the primary analysis set	350
consists of all randomized subjects. With respect to the primary analysis method,	351
FDA has accepted the log-rank test. Although FDA is open to discussion about other	352
methods, it is incumbent on the sponsor to provide the required justification.	353
Additional summary metrics that should be reported include the estimated medians	354
(where meaningful), hazard ratios, and 95% confidence intervals.	355
356
• 	It is common for some but not all patients with AML to undergo allogeneic HSCT	357
after or in conjunction with an investigational drug, which may impact EFS or RFS.	358
Additionally, as more effective drugs for AML are approved, post-study treatment	359
may impact OS. As these treatments are integral to the practice of medicine, the	360
primary analysis of these endpoints should be conducted without censoring for such	361
18 June 2017. When final, this guidance will represent the FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
Draft — Not for Implementation
10
treatment.19 For additional discussion about survival analyses when HSCT is a post-	362
study treatment, see Appendix 2.	363
364
• 	Trials designed to cure AML often result in survival contours characterized by an	365
initial drop followed by a plateau. For additional discussion about analysis when	366
there is a survival plateau, see Appendix 2.	367
368
• 	Secondary and sensitivity analyses of time-to-event endpoints should follow a	369
prespecified statistical analysis plan. These analyses may include the use of	370
alternatively-defined endpoints (e.g., alternative definition of time to ITF other than	371
day 1 when using EFS), alternatively-defined populations, or using alternative	372
analysis methods.	373
374
2. 	Binary Endpoints Used Commonly for AML	375
376
a. 	Complete Remission (CR)	377
378
• 	For documentation of CR, FDA has used the following definition:	379
380
− 	Marrow blasts < 5% by morphological examination,	381
− 	Absolute neutrophil count (ANC) > 1 Gi/L,	382
− 	Platelet count > 100 Gi/L,	383
− 	Absence of leukemic blasts in the peripheral blood by morphological	384
examination, and	385
− 	No evidence of extramedullary disease.	386
387
• 	The protocol should provide for maximum windows of time between marrow	388
sampling and peripheral blood tests used to establish CR. For the response	389
assessment of extramedullary disease, invasive testing should be limited only to sites	390
involved with AML at baseline that cannot be evaluated directly by general physical	391
examination, unless invasive testing is considered standard of care. The date of	392
marrow sampling is assigned as the CR date. Missing data is considered failure to	393
achieve CR. Additional considerations may be needed depending on the extent of	394
missingness, how differential it is between the arms when the AML study is	395
randomized, and whether the study is open-label. See section III.D.4 for a	396
discussion of trial procedures critical to the assessment of CR and section IV.B.1 for	397
the discussion of the adjudication of CR for the purpose of labeling.	398
399
• 	For CR, the duration of remission (DOR) is defined as the time from CR to	400
hematological relapse or death from any cause, whichever comes first. Adequate	401
follow-up is required in order to establish that the durability of CR is meaningful.	402
403
19 See the draft treatment policy discussion in the draft ICH E9(R1). When final, this guidance will represent the
FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
Draft — Not for Implementation
11
• 	Hematological relapse is defined as marrow blasts > 5% by morphology, persistent	404
reappearance of blasts in the peripheral blood by morphology, or the occurrence of	405
extramedullary disease. In general, once CR is confirmed by marrow examination,	406
further follow-up for relapse may be limited initially to physical examination and	407
peripheral blood tests. The known time to relapse for the regimen in the control arm	408
or from other historical data should be used when planning the frequency and	409
duration of testing for relapse, but in order to determine DOR, EFS, and RFS as	410
accurately as possible, the assessments would likely be performed more frequently	411
than in standard practice. When relapse is suspected on the basis of the follow-up	412
physical examination or peripheral blood counts, additional testing may be performed	413
to confirm the finding, but the date of relapse is set to the date of the first test that	414
suggests relapse.	415
416
b. 	CR with Partial Hematological Recovery (CRh)	417
418
• 	Use of CRh as an endpoint is applicable to drugs that are relatively nontoxic and	419
nonmyelosuppressive, as might be used for palliative purposes.	420
421
• 	For documentation of CRh, FDA has used the following definition:	422
423
− 	Marrow blasts < 5% by morphological examination,	424
425
− 	ANC > 0.5 Gi/L and platelet count > 50 Gi/L, but the count recovery criteria for	426
CR are not met,	427
428
− 	Absence of leukemic blasts in the peripheral blood by morphological	429
examination, and	430
431
− 	No evidence of extramedullary disease.	432
433
• 	Since the potential utility of CR as an endpoint is similar to that of CRh in this	434
setting, the actual endpoint used is CR+CRh. Adequate follow-up is needed in order	435
to establish that the durability of CR+CRh is meaningful.	436
437
c. 	Transfusion-Independence (TI)	438
439
• 	Durable TI as an endpoint is applicable to drugs that are relatively nontoxic and	440
nonmyelosuppressive, as might be used for palliative purposes.	441
442
• 	When durable TI is used, this endpoint should be supported by evidence showing an	443
effect of the treatment on an endpoint reflecting antileukemia activity. TI as an	444
endpoint for the treatment of AML should also be distinguished from TI as used in	445
the evaluation of hematopoietic growth factors (e.g., for the treatment of anemia)	446
where the effect of the drug is directed at normal hematopoietic cells rather than at	447
the leukemia.	448
449

Contains Nonbinding Recommendations
Draft — Not for Implementation
12
• 	TI is defined as the absence of red blood cell and platelet transfusions for a	450
prespecified period of time during continued treatment. The credibility of the data is	451
dependent on the protocol specifying the minimal parameters for use of transfusions	452
and documentation that the instructions were followed. Hence, an important	453
supporting analysis would include an assessment of serial measurements of	454
hemoglobin and platelet counts to ensure that the observed TI was an actual treatment	455
effect and not a bias in the administration of transfusions by the investigator.	456
457
• 	TI should be assessed as a response achieved in the subgroup of patients who were	458
transfusion dependent (TD) at baseline (conversion from TD to TI with treatment)	459
separately from the subgroup of patients who were TI at baseline (maintenance of TI	460
with treatment). For patients with active AML, transfusion dependence at baseline is	461
based on the receipt of any red blood cell or platelet transfusions within at least 28	462
days prior to the start of study treatment. Analyses of red blood cell TI and platelet	463
TI separately should be used to establish consistency of the components of the TI	464
endpoint.	465
466
d. 	Statistical Considerations for Binary Endpoints	467
468
• 	For single-arm AML trials, the analysis set consists of all patients treated with	469
investigational drug.20 If the labeling claim is limited by the target of the drug (e.g.,	470
AML with a FLT3 mutation for a drug that is a FLT3 inhibitor), the analysis set	471
should include only those patients confirmed positive for the target using the	472
proposed companion diagnostic or bridged clinical trial assay. For binary endpoints,	473
proportions and their 95% confidence interval should be reported.	474
475
• 	For randomized AML trials, the analysis set consists of all randomized patients. For	476
binary endpoints, the primary analysis may be based on Fisher’s Exact test; the	477
Cochran-Mantel-Haenszel test may apply when stratification factors were used at	478
randomization. Proportions and their 95% confidence intervals should be reported.	479
Any additional metrics to quantify the treatment effect, such as the difference in	480
proportions, ratio of proportions or odds ratio, should be prespecified. For targeted	481
drugs, a secondary analysis should be performed where the analysis set is restricted to	482
patients confirmed positive for the target.	483
484
• 	DOR may be calculated using the Kaplan-Meier method using relapse or any-cause	485
death as events. Estimated median and range should be reported. When the number	486
of study subjects is small, or when follow-up is short, the Kaplan-Meier estimate may	487
not be stable. In this circumstance, the observed median and range of observed DOR	488
may be reported. Sensitivity analyses may include calculation of DOR including	489
nonprotocol antileukemia treatment in the absence of documented relapse as an	490
additional event, or calculation of DOR with censoring at HSCT.	491
492
20 In cases of personalized products with the potential for a high rate of manufacturing failure, additional efficacy
analyses based on enrolled patients may be needed even in a single-arm trial in order to assess the impact of
manufacturing failure on the efficacy endpoint.

Contains Nonbinding Recommendations
Draft — Not for Implementation
13
3. 	Minimal Residual Disease (MRD)-Based Endpoints	493
494
• 	For new drugs that have a demonstrated durable CR in patients with relapsed or	495
refractory acute leukemia, FDA has accepted marrow MRD of less than 0.01% as	496
supporting evidence of efficacy. As technologies improve and new clinical findings	497
emerge, the level of MRD needed to support an efficacy claim for AML may change.	498
499
• 	CR as defined in section III.B.2 is the preferred timing to assess MRD as a response	500
endpoint. If assessments are made at CR without count recovery (CRi), or at lesser	501
responses, to support a claim of efficacy, the sponsor should include data to justify	502
the validity of the plan. The recommended analyses of MRD-based response	503
endpoints are similar to those for CR discussed in section III.B.2. When used as a	504
binary endpoint, the denominator for the analysis of MRD response should be all	505
treated patients (single-arm trial) or the ITT population (randomized trial), and the	506
numerator should be all patients who achieved CR and the required level of MRD.	507
Missing data should be imputed as a failure.	508
509
• 	Using MRD-based definitions to identify relapse for the purposes of determining	510
DOR, EFS, or RFS can be challenging in studies of new treatments for AML, since	511
that would require frequent marrow sampling. It may be more practical to monitor	512
for hematological relapse as described in section III.B.2 for the DOR, EFS, and RFS	513
definitions unless there is a validated MRD assay using peripheral blood samples.	514
515
• 	For additional information on the use of MRD as an efficacy endpoint, see the	516
guidance for industry Hematologic Malignancies: Regulatory Considerations for	517
Use of Minimal Residual Disease in Development of Drug and Biological Products	518
for Treatment.	519
520
4. 	Other Potential Measures of Efficacy for AML	521
522
• 	FDA acknowledges that as technology progresses and clinical trial data accumulate,	523
alternative biomarkers or measures of efficacy may be proposed for use as endpoints	524
in AML clinical trials. When considering the use of efficacy endpoints other than	525
those listed above, especially in a trial to be used to support a marketing application,	526
sponsors should obtain advice from FDA about the acceptability of the proposed	527
novel endpoint prior to initiating the trial.	528
529
• 	Key efficacy endpoints may also include well-defined and reliable patient-focused	530
outcome measures. When used as the basis of a claim of treatment of AML, such	531
endpoints should be supported by data showing that the treatment also has a direct	532
effect on the leukemia. For additional information, refer to the guidance for industry	533
Patient-Reported Outcome Measures: Use in Medical Product Development to	534
Support Claims (December 2009).	535
536

Contains Nonbinding Recommendations
Draft — Not for Implementation
14
• 	Sponsors planning to use real world data21 to support an AML drug marketing	537
application should obtain advice from FDA at the time of protocol development to	538
ensure that the data sources will provide the data needed to assess the treatment	539
effect. Important considerations include whether the sources capture the data	540
elements (e.g., marrow results, peripheral blood differentials, etc.) to derive clinically	541
accepted endpoints for demonstrating efficacy, and if so, the extent of	542
misclassification, the timing of assessment, and the frequency of assessment.	543
Sponsors should plan for additional discussions regarding alternative measures if the	544
data sources do not capture the key elements of the clinically accepted endpoints.	545
546
C. 	Exploratory Trial Considerations	547
548
1. 	First-in-Human (FIH) Trials	549
550
• 	Conducting an FIH trial in patients with rapidly progressing acute leukemias has	551
several challenges; the doses used in the first cohorts may be subtherapeutic, and the	552
assessment of toxicity may be confounded by adverse events due to the underlying	553
leukemia. Where feasible, sponsors should consider alternative designs for the FIH	554
trial that would identify a pharmacologically-active dose before commencing the	555
dose-escalation trial in patients with AML. For example, the sponsor may consider a	556
limited window study over a short interval (days to weeks) prior to the administration	557
of a standard treatment or conducting the initial dose escalation in patients with more	558
slowly growing tumors (solid tumors or lymphoma). Where applicable, sponsors	559
may also consider the FDA’s Model Informed Drug Development (MIDD)22 pathway	560
to help select the appropriate doses for efficacy and safety evaluation.	561
562
• 	Historically, the most effective regimens for the treatment of AML have been	563
combination regimens. Nonetheless, the FIH trial should be limited to assessment of	564
one drug at a time, and study of the combination should not commence until there is	565
adequate information about safety and tolerability of the individual drugs. Rare	566
exceptions to this principle are described in the guidance for industry Codevelopment	567
of Two or More New Investigational Drugs for Use in Combination (June 2013).	568
569
• 	An FIH trial of a myeloablative drug to be used as a single-agent preparative regimen	570
for HSCT for the treatment of patients with AML may be feasible, but prior to	571
submission of the investigational new drug application (IND), sponsors should obtain	572
advice from FDA about the optimal approach for development of such drugs. An	573
FIH trial of a new drug in combination with a preparative regimen is rarely	574
acceptable.23	
575
21 For additional information, see “Framework for FDA’s Real-World Evidence Program” at
https://www.fda.gov/media/120060/download and the draft guidance for industry Submitting Documents Using
Real-World Data and Real-World Evidence to FDA for Drugs and Biologics (May 2019). When final, this guidance
will represent the FDA’s current thinking on this topic.
22 See the Federal Register (83 FR 16868, April 17, 2018).
23 An example of an exception would be for a cell therapy where there is no scientific justification for study of the
cell therapy in the absence of a prespecified standard preparative regimen.

Contains Nonbinding Recommendations
Draft — Not for Implementation
15
576
• 	Although many drugs developed for the treatment of AML are highly	577
myelosuppressive and/or genotoxic, in select cases it may be possible to conduct the	578
FIH trial in healthy volunteers. The advantage to this approach is that the safety	579
profile may be simpler to determine in the absence of confounding adverse events due	580
to the underlying leukemia. FDA recommends that sponsors request feedback on the	581
design of FIH trials of new AML drugs in healthy volunteers, including the	582
limitations in exposure and other restrictions needed to protect healthy volunteers	583
participating in such studies.24	
584
585
• 	For AML drugs that are CYP3A substrates, sponsors should consider enrolling	586
patients on azole antifungals or other CYP3A inhibitors in FIH trials to generate data	587
needed to select a safe dose with these concomitant drugs (see section III.A.3).	588
589
• 	Sponsors developing cellular or gene therapy products for the treatment of AML	590
should also consult the guidance for industry Considerations for the Design of Early-	591
Phase Clinical Trials of Cellular and Gene Therapy Products (June 2015) and the	592
guidance for industry Long Term Follow-Up after Administration of Human Gene	593
Therapy Products.	594
595
2. 	Exploratory Trial Population	596
597
• 	For dose-escalation trials being conducted to determine the maximum tolerated dose	598
(MTD), the eligible population is usually limited to patients who have failed all	599
conventional drugs. Patients with subtypes of AML that respond very poorly to	600
conventional drugs, such as those with high-risk genetic abnormalities, might also be	601
considered for such trials even without prior treatment, but if doing so, the consent	602
form should clearly state the implications of foregoing conventional drugs in order to	603
participate in the clinical trial.	604
605
• 	The benefit-risk ratio for participation in a dose-escalation trial may also be	606
acceptable for patients with MRD after treatment with conventional drugs for AML,	607
but such protocols should include a description of the evidence that justifies the risks	608
of such a study compared to the prognosis based on the level of MRD proposed for	609
eligibility.	610
611
• 	For dose-escalation trials being conducted to determine the RP2D, the eligibility	612
criteria that address organ function and comorbidities should be commensurate with	613
the target patient population. For example, if developing a drug for the treatment of	614
AML in patients with renal or hepatic impairment, including patients with only	615
normal renal or hepatic function might conclude with a dose that is safe in patients	616
with normal organ function but that is too toxic for the target population with organ	617
impairment.	618
24 See also the guidance for industry Estimating the Maximum Safe Starting Dose in Initial Clinical Trials for
Therapeutics in Adult Healthy Volunteers (July 2005).

Contains Nonbinding Recommendations
Draft — Not for Implementation
16
619
• 	Multiple genetic mutations and molecular pathways have been identified as	620
contributing to the pathogenesis and persistence of AML. For new drugs proposed to	621
target these mutations or pathways, the clinical development program should have an	622
early phase trial that includes patients with and without the putative target in order to	623
assess the need in later phase trials to select patients based on the presence of the	624
target. Including marker-negative patients might not be necessary for drugs that	625
target a cell surface receptor, especially when preclinical data suggest no potential for	626
a therapeutic effect in the absence of the cell surface receptor.	627
628
3. 	Dose-Escalation Trials	629
630
• 	For dose-escalation trials, the general principles for selection of the safe starting dose	631
and the frequency of administration as described elsewhere25 also apply to drugs	632
being developed for the treatment of AML. As discussed in section III.C.1, the safe	633
starting dose for a study in patients with active AML may differ from the starting	634
dose for a study in healthy volunteers. The nonclinical data should also be used to	635
determine the slope of the dose-toxicity curve, the anticipated therapeutic dose range,	636
and the maximal exposure in order to plan the increments in dose between cohorts in	637
the escalation. For drugs that are CYP3A substrates, the selection of a safe starting	638
dose should also consider the concomitant use of drugs that are CYP3A4 inhibitors	639
such as azole antifungals (see section III.A.3).	640
641
• 	The protocol should describe the specific rule-based or model-based criteria used to	642
guide the decision on whether to proceed with escalating the dose in subsequent	643
cohorts. For dose-escalation trials of conventional outpatient chemotherapy for	644
patients with cancer, escalation to higher doses is generally limited by the rate of	645
severe, life-threatening, or fatal events (grades 3-5) termed dose-limiting toxicities	646
(DLTs), and the MTD as identified by the 3+3 rule has no more that 17% DLTs. This	647
paradigm, however, is not applicable to all types of treatments for AML. For	648
example, such a rule would allow far greater toxicity than acceptable for continued	649
treatment or maintenance that extends for years. On the other hand, the rule would	650
likely result in premature closure of a trial of a preparative regimen for HSCT, where	651
grade 3 toxicities are common. Hence, the criteria proposed to guide dose-escalation	652
decisions should take into account the types, severities, and rates of toxicities	653
accepted with standard regimens of similar intensity in the intended population (see	654
Appendix 1 for examples). The protocol should describe the data that support the	655
assumptions used to develop the criteria for guiding dose-escalation.	656
657
• 	For many cytotoxic drugs used for the treatment of AML, there is a strong dose-	658
response effect, and in order to achieve the highest response rate, the cited goal of the	659
dose-escalation trial is to identify the MTD. This is not necessarily true for targeted	660
drugs, for which the pharmacodynamic effect may plateau at doses lower than	661
maximally-tolerated. Hence, the goal of the dose-escalation trial should be to	662
25 See ICH S9 and ICH S9 Questions and Answers.

Contains Nonbinding Recommendations
Draft — Not for Implementation
17
determine the RP2D instead. The protocol should include a definition of the RP2D,	663
and the determination of the RP2D should take into consideration the safety,	664
tolerability, pharmacokinetic, pharmacodynamic, and efficacy data (see also section	665
III.D.2).	666
667
• 	Based on the design of the dose-escalation trial, participants in the initial cohorts of	668
the trial may not receive optimal treatment, which may be a disadvantage for patients	669
with active AML who are in need of cytoreductive treatment. Despite the desire to	670
ensure that patients with AML are treated with pharmacologically-active doses of	671
drug, intra-patient dose-escalation based on lack of very early response may not be	672
scientifically valid; a complete characterization of safety, tolerability, and efficacy at	673
any dose level usually requires treatment for multiple cycles. Intra-patient dose-	674
escalation may be considered in select circumstances where risks can be minimized	675
objectively. For example, if there is an established pharmacodynamic biomarker for	676
safety, intra-patient dose escalation may be feasible with frequent monitoring of the	677
biomarker. Additionally, for patients who have received multiple cycles of treatment	678
without evidence of cumulative toxicity or therapeutic activity, it may be beneficial to	679
escalate the individual patient's dose to a higher level if that higher dose has been	680
established as safe in subsequent cohorts. The protocol should specify the criteria for	681
when intra-patient dose escalation is allowed, how the new dose is assigned, any	682
changes in the monitoring plan needed to accommodate the change in dose, and how	683
the safety and efficacy data will be evaluated for such patients.	684
685
• 	The planned duration of treatment should be described clearly in the protocol. Long-	686
term treatment may be considered in the dose-escalation trial, typically for patients	687
with relapsed or refractory AML, but there should be objective criteria for when to	688
discontinue treatment permanently, including high-grade toxicities. When treatment	689
in the dose-escalation trial is planned to extend beyond achievement of CR, a	690
rationale should be provided for the proposed duration of treatment after remission.	691
For patients who are taken off the investigational drug after achieving a CR, the	692
protocol may also address retreatment in case of relapse.	693
694
• 	Early phase trials are also the place to determine the expected time to response,	695
allowing study treatment to continue in the absence of toxicity unless prespecified	696
levels of disease response have not occurred within a maximum number of cycles.	697
Such information will provide support for the treatment plan proposed for	698
confirmatory trials designed to test for efficacy.	699
700
• 	For early phase trials of intensive AML drugs given with curative intent, a	701
maintenance phase is generally not acceptable in settings where there is no	702
established benefit of maintenance; in such cases, a randomized control arm is	703
recommended.	704
705
• 	Certain toxicities of treatment, such as anemia or tumor lysis syndrome, are expected	706
with almost any treatment of patients with AML. Treatment of such usual toxicities	707
is considered standard practice, and detailed instructions on the practice of medicine	708

Contains Nonbinding Recommendations
Draft — Not for Implementation
18
need not be included in the protocol unless a specific treatment is critical for safe use	709
of the investigational drug. Based on established class toxicities, mechanism of	710
action and/or nonclinical studies, there may also be unusual drug-specific toxicities,	711
such as differentiation syndrome or cytokine release syndrome. Until treatment is	712
standardized in practice, instructions for management of patients with such unusual	713
drug toxicities should be included in the protocol.	714
715
4. 	Exploratory Expansion Cohorts	716
717
• 	A small cohort of 6-12 subjects treated at the presumptive RP2D can be useful to	718
confirm safety prior to start of additional trials. In the absence of data from a safety	719
expansion cohort, the confirmatory trial should include a very early interim safety	720
analysis to corroborate safety of the RP2D.	721
722
• 	When the new drug is being studied as an add-on and the background regimen has	723
substantial toxicity (i.e., a standard intensive AML induction regimen), a randomized	724
comparison may be necessary to detect even large differences in toxicity that might	725
not be noticed in the single-arm setting.	726
727
• 	Responses as defined in section III.B.2 are generally acceptable measures of activity	728
that should be included in exploratory early phase clinical trials in AML. Lesser	729
responses (e.g., partial remission, shorter term transfusion-independence, etc.) may	730
reflect activity of the drug, but such lesser responses should guide development to	731
alternative strategies to leverage that activity (i.e., different schedules or use in	732
combinations) rather than being viewed as a success.	733
734
• 	A small cohort of patients treated at the presumptive RP2D can also be used to	735
provide an estimate of efficacy to support design of additional trials. Such a cohort	736
generally includes approximately 20 subjects. Large single-arm expansion cohorts	737
solely for exploratory purposes are discouraged. Any large single-arm trial should	738
have a design based on clear hypothesis testing, and the protocol should include	739
justification of the sample size proposed.	740
741
• 	Time-to-event endpoints are difficult to interpret in single-arm trials and, therefore,	742
are generally not useful in assessing efficacy in exploratory early phase trials. Data	743
for such endpoints, however, should still be collected, since such data could be useful	744
in designing the confirmatory trials if other objective measures of efficacy support	745
further development of the drug.	746
747
• 	To ensure the safety of study participants, the expansion cohort plan should include	748
stopping rules for excessive toxicity that would require pausing enrollment to	749
evaluate whether the treatment plan should be modified. The acceptable rate and type	750
of toxicities will depend on the treatment setting as discussed for development of	751
DLT criteria in section III.C.3. The protocol should describe the exact bounds for the	752
stopping criteria, the statistical method used to calculate the bounds, and the basis for	753
the clinical assumptions used in the calculation. FDA recommends that the bounds be	754

Contains Nonbinding Recommendations
Draft — Not for Implementation
19
calculated using nonstringent parameters (i.e., 70% power or 80% posterior	755
probability), so that the trial can be paused at the earliest sign of excessive toxicity.	756
For patients with active AML, toxic events for stopping rules might include	757
treatment-related deaths, prolonged neutropenia lasting past cycle day 42 in the	758
absence of disease, and high-grade nonhematological adverse reactions.	759
760
D. 	Confirmatory Trial Considerations	761
762
1. 	Confirmatory Trial Population	763
764
• 	The protocol should use the most updated diagnostic criteria for AML or for a	765
specific AML type to describe the eligible population. Sponsors should seek advice	766
from FDA rather than using outdated criteria solely to match a population used in	767
support of a past approval.	768
769
• 	Patients with newly-diagnosed AML, patients with AML in late first relapse (e.g.,	770
first remission > 6 months), and patients with other relapsed or refractory AML (e.g.,	771
primary refractory, early first relapse, and any second or later relapse) represent three	772
distinct indications. A separate trial for each indication is recommended, but separate	773
cohorts in a single trial may be used for analyses to support each indication	774
independently. In the latter circumstance, the protocol should describe clearly the	775
eligibility criteria for each cohort. HSCT is considered standard practice in the	776
treatment of AML, and relapse post HSCT would fall under either treatment of first	777
relapse or treatment of later relapse rather than being a separate indication.	778
779
• 	For clinical trials of a biomarker-selected AML population, the eligibility criteria	780
should state clearly what assay is to be used to select patients with the cognate target,	781
the tissue (blood, marrow, etc.) used for the assay, and the level of the target needed	782
to meet eligibility.	783
784
• 	For clinical trials planned to support a marketing application for the intended	785
population of patients with comorbidities that preclude use of intensive induction	786
chemotherapy, the eligibility criteria should include detailed parameters that describe	787
the population. See section III.A.4.c for examples of criteria for organ impairment	788
that FDA has accepted to describe this subgroup of patients for AML trials.	789
790
• 	For clinical trials being designed to support a marketing application, the eligibility	791
criteria should reflect the characteristics of the general population with AML.	792
Exclusion criteria should be limited to disease- or patient-related factors associated	793
with a lack of benefit or an unacceptable risk of toxicity from the investigational drug	794
based on data in early phase trials.26	
795
796
26 For additional discussion, see the draft guidance for industry Enhancing the Diversity of Clinical Trial
Populations — Eligibility Criteria, Enrollment Practices, and Trial Designs (June 2019). When final, this guidance
will represent the FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
Draft — Not for Implementation
20
• 	For clinical trials of noncurative drugs for AML, the eligibility criteria should	797
specifically exclude patients willing and able to receive intensive curative treatment.	798
799
2. 	Dose Selection and Treatment Plan	800
801
• 	The dose and schedule of the investigational drug in the treatment regimen should be	802
optimized before initiating the confirmatory trials. Clinical pharmacokinetic data,	803
pharmacodynamic data, clinical activity measures, clinical safety data, and	804
nonclinical pharmacology data, should be pooled for conduct of integrated dose-	805
response and exposure-response analyses for activity and safety for dose	806
optimization. The results of such analysis should be included in the protocol to	807
justify the dose.	808
809
• 	For drugs planned to be administered for multiple cycles, and especially for drugs	810
given long-term on an outpatient basis, tolerability should be taken into consideration	811
when choosing the dose to be used in the confirmatory trial. In general, for drugs	812
planned to be given long-term or over multiple cycles, it is expected that dose	813
modifications or discontinuations for adverse reactions are limited to less than 20% of	814
the patients, and that at least 80% dose intensity is achieved over multiple cycles for	815
at least 80% of the patients.	816
817
• 	The protocol should include dose adjustment strategies for specific populations (e.g.,	818
with organ impairment or with concomitant use of moderate and strong CYP3A	819
modulators) and in response to emerging adverse events. The experience with these	820
instructions during study conduct provides the basis for dose modification	821
instructions in labeling.	822
823
3. 	Confirmatory Trial Design	824
825
a. 	General Considerations for Confirmatory Trial Designs	826
827
• 	The principles of designing trials to demonstrate efficacy for the purposes of	828
supporting a marketing application are described in general guidance,27 and these	829
general principles are applicable to trials for AML drugs. The bullets below provide	830
additional advice specific for the trials of treatments for AML.	831
832
• 	To prevent bias in study conduct or in selection of poststudy treatments, the use of	833
blinded treatments where feasible is recommended for randomized trials.	834
835
• 	The use of specific genetic targets and other prognostic factors used for eligibility or	836
risk stratification should be described in detail. For patients with relapsed or	837
27 See the draft ICH E8(R1) and E9(R1) (when final these guidances will represent the FDA’s current thinking on
these topics); the ICH guidances for industry E9 and E10; and the draft guidance for industry Demonstrating
Substantial Evidence of Effectiveness for Human Drug and Biological Products (December 2019) (when final, this
guidance will represent the FDA’s current thinking on this topic).

Contains Nonbinding Recommendations
Draft — Not for Implementation
21
refractory AML, the protocol should state clearly whether these prognostic factors are	838
measured at the time of diagnosis or at the time of relapse.	839
840
• 	Because treatment for AML involves discrete stages of treatment with different goals,	841
the purpose of treatment with the investigational drug should be stated clearly in the	842
protocol. Potential objectives may include remission induction alone, remission	843
induction followed by consolidation, consolidation of remission alone, remission	844
maintenance (after chemotherapy or transplantation), or control of complications of	845
the disease in the relapsed/refractory setting.	846
847
• 	If the clinical trial has goals in multiple stages of treatment, sequential	848
randomizations may be needed. For example, if a maintenance indication is planned	849
in addition to initial treatment, patients should be rerandomized prior to maintenance	850
to allow for isolation of the treatment effect of study drug(s) during maintenance.	851
852
• 	A detailed statistical analysis plan stating the trial hypotheses, sample size, analysis	853
timing, and analysis methods should be submitted before trial initiation. The sample	854
size calculation should be based on the expected efficacy in the control arm and the	855
anticipated treatment effect of the investigational drug with respect to the primary	856
endpoint in the planned patient population. Estimating the outcome for the control	857
arm in a molecular subgroup may be challenging for treatments of AML with new	858
molecular targets that were not studied previously with standard care regimens.	859
When there is little extant data to support the assumptions for the anticipated	860
treatment effect, sponsors may consider an adaptive design or other novel approach.28	
861
In such a case, the sponsor should request feedback from FDA on the proposed	862
design prior to initiating the trial.	863
864
• 	When the design requires an active comparator, the treatment should be standard of	865
care for the study population (e.g., study drug vs. 7+3). Placebo comparators may be	866
considered in add-on trials (e.g., study drug+7+3 vs. placebo+7+3) if appropriate	867
treatment for the control arm. Comparative efficacy studies of combinations that do	868
not isolate the effect of the study drug (study drug+azacitidine vs. 7+3) may also be	869
acceptable if the control is standard of care for the population, the activity of the	870
study drug was demonstrated in other trials, and the contribution of each drug in the	871
new regimen is supported by other data in the context of use.	872
873
• 	It is common for multiple efficacy endpoints (i.e., OS, EFS, CR) to be assessed in a	874
clinical trial for AML. The statistical analysis plan should prespecify a multiple	875
testing strategy for important secondary endpoints that adjusts for multiplicity	876
conditioned on demonstrating a positive outcome for the primary endpoint. Note that	877
effects on secondary endpoints are generally not sufficient to support a marketing	878
application in the absence of demonstration of an effect on the prespecified primary	879
endpoint. Additionally, even if an effect on a secondary endpoint is demonstrated, it	880
may not be acceptable for labeling if it is not an established efficacy endpoint; for	881
28 For example, see the guidance for industry Adaptive Designs for Clinical Trials of Drugs and Biologics
(November 2019).

Contains Nonbinding Recommendations
Draft — Not for Implementation
22
example, the composite of CR+CRi may not be suitable for labeling due to the	882
inclusion of CRi.	883
884
• 	In large randomized trials, an interim analysis for futility is strongly recommended to	885
ensure that the benefit-risk ratio for enrolled patients continues to be favorable. FDA	886
has accepted group sequential/early stopping designs for interim analyses. However,	887
for certain endpoints, such as EFS or RFS, FDA discourages early stopping for	888
efficacy based on a positive assessment at the interim analysis. More follow-up may	889
be needed to assess other important endpoints, such as duration of response and	890
safety, that would be needed to determine the overall benefit-risk. FDA is willing to	891
discuss the potential pitfalls in a timely fashion when the sponsor is considering early	892
study termination based on interim efficacy analysis results.	893
894
• 	The timing of analysis of continued response (e.g., DOR, RFS, etc.) should be	895
prespecified to mitigate bias in study result interpretation.	896
897
b. 	Treatment of AML with Curative Intent	898
899
• 	Treatments with an expectation of a survival plateau are considered treatment with	900
curative intent. Examples include standard intensive chemotherapy as first-line	901
treatment for AML.	902
903
• 	FDA has accepted OS, EFS, and RFS as clinical endpoints that represent clinical	904
benefit for traditional approval for treatments with curative intent.	905
906
• 	AML is a heterogenous disease, and historical controls are severely limited in their	907
ability to accurately parallel the intended population for the indication. Therefore, the	908
use of historical controls in AML is not appropriate for studies of treatment with	909
curative intent. Trials intended to support a marketing application for this indication	910
should have a randomized control arm.	911
912
c. 	Treatment of AML without Curative Intent	913
914
• 	Treatments with no expectation of a survival plateau, but where the goal is to extend	915
survival or greatly improve durable CR relative to a control, are considered treatment	916
without curative intent for AML.	917
918
• 	FDA has accepted OS and EFS as clinical endpoints that represent clinical benefit for	919
traditional approval for treatments without curative intent. For studies in populations	920
with a very high rate of induction treatment failure or when OS is expected to be	921
short, OS may be the more practical endpoint to establish clinical benefit. Durable	922
CR may also support traditional approval depending on the disease setting and	923
benefit-risk ratio.	924
925
• 	Trials intended to support a marketing application for this indication may be	926
randomized or single-arm in design, depending on the endpoint, patient population,	927

Contains Nonbinding Recommendations
Draft — Not for Implementation
23
and available therapy. FDA recommends that sponsors request advice from FDA on	928
proposed study designs for this indication.	929
930
d. 	Treatment of AML with Palliative Intent	931
932
• 	Nonintensive treatments without substantial associated toxicities administered with	933
the goal of temporary disease control and minimal treatment burden are considered	934
treatment with palliative intent in AML.	935
936
• 	Durable TI may represent a direct clinical benefit resulting from the relief from the	937
burdens of insufficient hematopoiesis due to active AML. FDA has accepted durable	938
CR and durable CR/CRh with TI as clinical endpoints that represent clinical benefit	939
for traditional approval for treatments with palliative intent.	940
941
• 	Trials intended to support a marketing application for this indication may be	942
randomized or single-arm in design depending on the endpoint, patient population,	943
and available therapy. Best supportive care may be acceptable as a comparator in a	944
randomized trial only for a patient population without available therapies. In certain	945
clinical settings, a single-arm trial may be appropriate for traditional approval if there	946
are adequate historical data to support the null hypothesis.	947
948
4. 	Confirmatory Trial Procedures	949
950
• 	Baseline demographic and disease characteristics are used to ensure consistency of	951
the benefit-risk by subgroup analyses. The following key AML-specific information	952
should be documented and collected on the case report forms:	953
954
− 	Disease (WHO-based diagnosis29),	955
956
− 	Disease status at enrollment (e.g., newly-diagnosed, 2nd relapse, etc.),	957
958
− 	Response status at enrollment (primary refractory vs. untreated vs. refractory	959
relapse),	960
961
− 	Duration of first remission,	962
963
− 	Genetic profile and/or risk group at diagnosis and at enrollment (use of the most	964
contemporary accepted risk stratification is recommended),	965
966
− 	All prior treatments for AML,	967
968
− 	Baseline functional assessments (where applicable, geriatric assessment is	969
recommended), and	970
971
29 See footnote 4.

Contains Nonbinding Recommendations
Draft — Not for Implementation
24
− 	Relevant comorbidities (see section III.A.4.c).	972
973
• 	Patients with AML receiving intensive chemotherapy or high-dose chemotherapy for	974
transplantation are expected to have a high rate of low-grade adverse reactions. For	975
studies of drugs with well-established safety profiles, consideration should be given	976
to collection of a limited amount of safety data.30 For new drugs with unclear safety	977
profiles, all adverse events should be collected regardless of grade or attribution.	978
979
• 	To ensure that data will be available for the assessment of potential interactions	980
between new drugs and other drugs used commonly for patients with AML, the dates	981
and doses of concomitant medications, especially antifungal medications, should be	982
accurate.	983
984
• 	To assess confounding in efficacy analyses due to subsequent post-study treatments,	985
the following post-study information should be documented and collected on the case	986
report forms:	987
988
− 	At least the first post-study salvage treatment and the reasons for the treatment	989
choice and	990
991
− 	HSCT and CAR T cell dates for patients proceeding to transplantation with an on-	992
study response or as a post-study salvage treatment.	993
994
995
IV. 	REGULATORY SUBMISSIONS	996
997
A. 	Investigational New Drug Applications	998
999
• 	General requirements for INDs apply to AML. See sections III.A and III.C for	1000
recommendations on submission of FIH trials in AML as the IND-initiating study.	1001
Sponsors may request advice from FDA through the pre-IND program.	1002
1003
• 	FDA supports the use of innovative trial designs, such as master protocols, for	1004
efficient drug development in AML. For IND submissions that contain innovative	1005
trial designs, FDA recommends consultation through the pre-IND program. For	1006
additional draft recommendations, see the draft guidance for industry Master	1007
Protocols: Efficient Clinical Trial Design Strategies to Expedite Development of	1008
Oncology Drugs and Biologics.31	
1009
1010
• 	A companion diagnostic may be essential for patient selection in IND protocols for	1011
targeted AML drugs. Sponsors may request a study risk determination directly from	1012
CDRH or in concert with the IND (see the guidance for industry Investigational In	1013
30 See the guidance for industry Determining the Extent of Safety Data Collection Needed in Late-Stage Premarket
and Postapproval Clinical Investigations (February 2016).
31 September 2018. When final, this guidance will represent the FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
Draft — Not for Implementation
25
Vitro Diagnostics in Oncology Trials: Streamlined Submission Process for Study	1014
Risk Determination (October 2019) to determine whether an IDE is needed). See also	1015
section III.A.2.	1016
1017
B. 	Marketing Applications	1018
1019
1. 	Assessment of Efficacy	1020
1021
• 	Assessments of efficacy in AML clinical trials are generally based on objective	1022
criteria, such as neutrophil counts and marrow blast percentage. To allow FDA to	1023
confirm the analyses of the treatment effect, the raw data supporting the study	1024
endpoints should be submitted in the marketing application.	1025
1026
− 	If bone marrow pathology results exceed the character limit for a variable in an	1027
xpt file, a pdf of the report may be acceptable.	1028
1029
− 	To assist with the adjudication of responses, the submission should include a	1030
summary response file (see Appendix 3) for the confirmatory study and for the	1031
integrated efficacy population.	1032
1033
− 	For studies with an endpoint of TI (see section III.B.2.c), the submission should	1034
include a summary transfusion analysis data file (see Appendix 3) for at least the	1035
confirmatory study.	1036
1037
− 	To assist with the assessment of response and TI, the submission should include a	1038
file with the dates of RBC and platelet transfusions and the number of units	1039
transfused.	1040
1041
2. 	Assessment of Safety	1042
1043
• 	Patients with AML have a high background of adverse events due to the leukemia.	1044
Assessment of toxicities of the new AML drug in different disease settings (e.g., solid	1045
tumor patients) and in healthy volunteers is helpful in ascertaining causality of	1046
adverse events.	1047
1048
• 	To assist with the adjudication of causality of fatal adverse events, the submission	1049
should include a data file with the date of death, study day of death, proximate cause	1050
of death (usually as reported by the investigator), and the root cause of death as	1051
determined by the sponsor. The root cause is generally categorized as a direct effect	1052
of active AML, an adverse reaction, or an unrelated intercurrent event (such as car	1053
accident). When the sponsor is considering additional categories for root cause, such	1054
as “early death,” feedback on the proposed categories should be sought at the	1055
presubmission meeting.	1056
1057
• 	For drugs with unusual adverse reactions, such as differentiation syndrome, FDA	1058
encourages sponsors to meet with FDA review staff prior to submission of a new drug	1059

Contains Nonbinding Recommendations
Draft — Not for Implementation
26
application (NDA) or biologics license application (BLA) to develop a detailed	1060
methodology for identifying cases, determine when additional narratives should be	1061
included in the submission, and to discuss the structure of the data files to be used for	1062
the analysis of such cases.	1063
1064
• 	When the study drug is used in multiple stages of AML treatment (e.g., in	1065
combination with induction, in combination with consolidation, and as maintenance),	1066
safety and laboratory data should be assessed by treatment stage.	1067
1068
• 	When a randomized trial has a comparator arm with a different duration of treatment	1069
(e.g., continuous oral therapy vs. a fixed duration of intensive salvage chemotherapy),	1070
it is important to compare toxicities between study arms for a similar duration of	1071
treatment. For long-term continuous treatment with investigational drug, safety	1072
beyond the period of comparison should be analyzed separately and compared to	1073
early-period toxicities to identify unique late-onset adverse reactions.	1074
1075
• 	For myelosuppressive AML drugs, an analysis should be performed to determine the	1076
incidence of prolonged thrombocytopenia (platelets < 50 Gi/L) or neutropenia (ANC	1077
< 0.5 Gi/L) past cycle day 42 in the absence of active leukemia.	1078
1079
3. 	Clinical Pharmacology	1080
1081
• 	If the AML drug is a CYP3A substrate, the submission should include analyses of the	1082
effect of concomitant drugs, including moderate and strong CYP3A inhibitors and	1083
inducers on the systemic exposure of parent drug and its active metabolites, on safety	1084
and efficacy, and whether the available safety and efficacy data support the proposed	1085
dose modifications for concomitant treatment with moderate and strong CYP3A	1086
inhibitors and inducers (see section III.A.3). If the AML drug or its major	1087
metabolite(s) is an inhibitor or inducer of metabolism enzymes or transporters, the	1088
submission should include analyses of the effect of the parent drug and major	1089
metabolites on the systemic exposure of concomitant drugs that are substrates of	1090
metabolism pathway or transporter and have a likelihood of coadministration (e.g.,	1091
commonly-used antibiotics, other AML drugs in the combination regimen).	1092
1093
For submissions specifically for indications that target the population of patients with	1094
comorbidities that preclude use of intensive chemotherapy for AML, the submission should	1095
include the results of studies on the effects of renal and hepatic impairment on the systemic	1096
exposure of the parent drug and its active metabolites (see section III.A.3).	1097

Contains Nonbinding Recommendations
Draft — Not for Implementation
27
GLOSSARY	1098
1099
A. Terms referring to the types of AML treatment are defined as follows when used in this	1100
guidance	1101
1102
Episodic treatment: A treatment plan of multiple cycles of short-term administrations of	1103
intensive treatment. A typical course of episodic first-line treatment for AML consists of 1-2	1104
cycles of induction and 2-4 cycles of consolidation with or without HSCT.	1105
1106
Continuous treatment: Repeated cycles of treatment, usually without a drug-free period. A	1107
typical course of continuous treatment of AML consists of daily dosing.	1108
1109
B. Terms referring to phases of AML treatment are defined as follows when used in this	1110
guidance	1111
1112
Induction: A limited course of treatment, usually intensive, with the objective of achieving	1113
CR.	1114
1115
Consolidation: A limited course of treatment, usually intensive, given after achievement of	1116
remission with the objective of reducing the risk of early relapse.	1117
1118
Maintenance: An extended but time-limited course of treatment, usually relatively	1119
nontoxic, given after achievement of CR with the objective of reducing the risk of relapse	1120
beyond the period of treatment. When the treatment plan allows for extended therapy for	1121
patients without achieving CR, the course is considered continued treatment rather than	1122
maintenance.	1123
1124
Continued treatment: An extended course of treatment after induction phase with the	1125
objective of controlling the AML disease burden while on therapy. Continued treatment may	1126
be time-limited, but it is generally administered until unacceptable toxicity or recurrence	1127
after a response.	1128
1129
C. Terms referring to intensities of AML treatment are defined as follows when used in this	1130
guidance	1131
1132
Intensive therapies: regimens expected to cause high-grade organ toxicity (including	1133
neurologic, pulmonary, cardiac, gastrointestinal, genitourinary, renal, hepatic, or cutaneous	1134
toxicities) or where the expected duration of neutropenia may approach 42 days from the	1135
start of the treatment cycle. Intensive regimens include 1-2 cycles of induction followed by	1136
consolidation with chemotherapy or HSCT.	1137
1138
Nonintensive therapies: lower doses of cytotoxic chemotherapy or targeted drugs with	1139
limited or no expected organ toxicities.	1140
1141

Contains Nonbinding Recommendations
Draft — Not for Implementation
28
D. Disease status is assessed at the time of study enrollment. Terms relevant to AML	1142
disease status are defined as follows when used in this guidance	1143
1144
Primary refractory disease: The patient did not experience CR in response to first-line or	1145
any subsequent induction therapy.	1146
1147
Untreated relapse: The patient experienced CR in response to the last prior therapy, then	1148
demonstrated relapse and has not yet received definitive re-induction therapy for that relapse.	1149
1150
Refractory relapse: The patient experienced disease remission in response to past therapy,	1151
then demonstrated relapse and was treated with definitive re-induction therapy but did not	1152
experience CR with this re-induction.	1153
1154
Line of therapy: A line of therapy is defined as the planned therapy consisting of one or	1155
more cycles of episodic treatment or a defined period of continuous treatment. This may	1156
consist of single-agent or combination therapy as well as a planned sequence of treatment	1157
phases. For example, first-line treatment of AML with induction, consolidation, and	1158
allogeneic HSCT is considered one line of therapy. A line of therapy ends when the patient	1159
fails to achieve a response within a prespecified period (refractory) or relapses after	1160
achieving CR.	1161

Contains Nonbinding Recommendations
Draft — Not for Implementation
29
APPENDICES	1162
1163
Appendix 1: Example DLT Criteria for Drugs for AML	1164
Setting 	Hematological SAR Criteriaa 	Nonhematological SAR Criteria
Healthy Volunteer Study 	Any grade > 2 	Any grade > 2
Continuous Long-Term
Treatment (e.g.,
maintenance, extended
treatments like imatinib)
Any grade > 3 ANC or PLTS
lasting more than 7 days
Any grade 3 lasting > 72 hours
Any grade > 4
Hy's law cases
Any AR that leads to dose reduction or
withdrawal
Short-Term Episodic
Outpatient Therapy (e.g.,
CHOP-like)
Any grade > 4 ANC or PLTS
lasting more than 7 days
Any grade 3 (with exceptions) b
Any grade > 4
Hy's law cases
Any AR that leads to dose reduction or
withdrawal
Episodic Reduced Intensity
(e.g., azacitidine)
Any grade > 4 ANC or PLTS
lasting past cycle day 28
Any grade 3 (with exceptions) b
Any grade > 4
Hy's law cases
Any AR that leads to dose reduction or
withdrawal
Episodic Intensive
Chemotherapy with
Curative Intent (e.g., 7+3 -
based)
Any grade > 4 ANC or PLTS
lasting past cycle day 42
Grade > 4 organ toxicityc
Hy's law cases
CAR T Cells 	Any grade > 4 ANC or PLTS
lasting past day 42, or marrow
cellularity < 5% at day 42
Grade > 3d CRS (with exceptions)b
Grade 3 neurotoxicity (with exceptions)b
Grade 4 neurotoxicity
Other grade > 3 toxicity to vital organs
(with exceptions)b,c
Myeloablative Preparative
Regimen (e.g., high-dose
busulfan)
No ANC recovery to > 0.5
Gi/L by day 21 (PBSC), 28
(marrow), or 42 (UCBT)
Grade > 4 organ toxicityc
Abbreviations: ANC - absolute neutrophil count, AR – adverse reaction, CAR - chimeric antigen receptor, CRS - cytokine
release syndrome, PBSC - peripheral blood stem cells, PLTS - platelet count, SAR - suspected adverse reaction, and UCBT -
umbilical cord blood transplantation.
a Not applicable in the presence of active leukemia. Patients with active leukemia are not evaluable for a hematological DLT.
b May exclude grade 3 toxicities that resolve within a prespecified time frame (e.g., 72 hours).
c Adverse reactions involving neurologic, pulmonary, cardiac, gastrointestinal, genitourinary, renal, hepatic, or cutaneous
systems.
d Refers to Lee Criteria for CRS. In the remainder of the table, grade number refers to NCI-CTCAE criteria.
1165

Contains Nonbinding Recommendations
Draft — Not for Implementation
30
Appendix 2: Additional Statistical Discussion	1166
1167
Postrandomization HSCT Subsequent Poststudy Treatments	1168
1169
For AML trials, it is common for patients to receive subsequent anti-AML treatments post-	1170
randomization, which include HSCT and/or anti-AML drugs (not to be confused with	1171
concomitant therapies). As use of subsequent treatments is consistent with the practice of	1172
medicine, FDA recommends that the primary analysis of a time-to-event endpoint (e.g., OS,	1173
EFS) not censor for subsequent treatments.1 This approach implies that HSCT and/or	1174
subsequent anti-AML treatments are viewed as part of the as-needed AML treatment regimen	1175
taken after the initial study drug and that the treatment effect is the result of both the study drug	1176
and the subsequent anti-AML treatment.	1177
1178
To help ensure that the treatment effect is interpretable, AML trials should be designed such that	1179
investigators are blinded to patients’ assigned treatment. Regardless of the feasibility of blinding	1180
and as HSCT extends survival, rules or criteria should be clearly prespecified in the protocol	1181
prior to study initiation to determine how patients are to be selected for HSCT. In addition,	1182
where patients are still on study, follow-up of patients should continue even after initiation of	1183
subsequent treatments.	1184
1185
It has been suggested that the true treatment effect should be free from the influence of HSCT or	1186
subsequent treatments.2 Under this approach, the treatment effect may be interpreted as the	1187
difference in the endpoint between patients who initiated the investigational drug and patients	1188
who initiated the control treatment if HSCT and/or subsequent anti-AML treatments had not	1189
been available, or if available, were withheld from patients. In settings where HSCT and/or	1190
subsequent treatments are integral to the practice of medicine, this approach to thinking about the	1191
treatment effect is currently not recommended for the primary analysis for the following reasons:	1192
1193
• 	First, it may not be possible to design a clinical trial to estimate this treatment effect if	1194
patients are provided HSCT or subsequent treatments as needed. This implies that this	1195
treatment effect can only be estimated by modeling, using causal inference methods	1196
developed for observational studies where the assumptions therein are difficult, if not	1197
impossible, to justify.	1198
1199
• 	Second, the clinical relevance of such an estimand is still an open question if it can never	1200
be realized in practice.	1201
1202
Plateauing Effect	1203
1204
Trials designed to cure AML often result in survival contours characterized by an initial drop	1205
followed by a plateauing effect after some time point post randomization. This is an example of	1206
nonproportional hazards. While the log-rank test is somewhat robust to nonproportionality, it	1207
1 See the draft treatment policy discussion in the draft ICH E9(R1). When final, this guidance will represent the
FDA’s current thinking on this topic.
2 For additional reference, see discussion on hypothetical strategies in draft ICH E9(R1). When final, this guidance
will represent the FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
Draft — Not for Implementation
31
generally results in loss of power. Furthermore, nonproportionality can cause difficulty in	1208
describing the treatment effect. FDA is open to discussion about analyses based on other	1209
approaches, such as weighted Cox regression or other weighted methods, or summarizing the	1210
treatment effect using restricted mean survival time (RMST) or landmark survival analysis.	1211
Plans that use these alternative approaches should include:	1212
1213
• 	justification for what constitutes clinically meaningful difference,	1214
1215
• 	justification of design parameters, such as sample size and follow-up duration, based on	1216
this endpoint, and	1217
1218
• 	justification for the value of the threshold that will be used to calculate the RMST.	1219

Contains Nonbinding Recommendations
Draft — Not for Implementation
32
Appendix 3: Additional Data Files for Marketing Applications for AML Drugs	1220
1221
The following variables are recommended for custom data files to assist with endpoint	1222
adjudication	1223
1224
Variables That Assist Morphological Response Assessment	1225
1226
• 	Study identification number	1227
• 	Site identification number	1228
• 	Unique subject number	1229
• 	Treatment arm	1230
• 	Date of start of study drug	1231
• 	Date of last study drug	1232
• 	Study day of last study drug	1233
• 	Date of last platelet transfusion	1234
• 	Study day of last platelet transfusion	1235
• 	Date of last RBC transfusion	1236
• 	Study day of last RBC transfusion	1237
• 	Date of CR*	1238
• 	Study day of CR*	1239
• 	Date of ANC used for CR response*	1240
• 	Study day of ANC used for CR response*	1241
• 	ANC used for CR response*	1242
• 	Date of platelet count used for CR response*	1243
• 	Study day of platelet count used for CR response*	1244
• 	Platelet count used for CR response*	1245
• 	Date of marrow used for CR response*	1246
• 	Study day of marrow used for CR response*	1247
• 	Marrow blasts percentage used for CR response*	1248
• 	Date of assessment of Auer rods (yes/no) at CR response*	1249
• 	Date of assessment of extramedullary disease for CR response*	1250
• 	Study day of assessment of extramedullary disease for CR response*	1251
• 	Absence of extramedullary disease (yes/no) at CR response*	1252
• 	Date of relapse	1253
• 	Study day of relapse	1254
• 	Date of transplantation	1255
• 	Study day of transplantation	1256
1257
* If CRh is an endpoint in the study, these measures should also be provided for CRh.	1258

Contains Nonbinding Recommendations
Draft — Not for Implementation
33
Variables That Assist the Transfusion Independence Assessment	1259
1260
• 	Study identification number	1261
• 	Site identification number	1262
• 	Unique subject number	1263
• 	Treatment arm	1264
• 	Date of start of study drug	1265
• 	Date of last study drug	1266
• 	Study day of last study drug	1267
• 	RBC transfusion dependence at baseline (yes/no)	1268
• 	Platelet transfusion dependence at baseline (yes/no)	1269
• 	Transfusion dependence for either RBC or platelets at baseline (yes/no)	1270
• 	RBC transfusion independence (TI) criteria met post baseline (yes/no)	1271
• 	Platelet TI criteria met post baseline (yes/no)	1272
• 	TI criteria met for both RBC and platelet transfusions post baseline (yes/no)	1273
• 	Date of start of RBC TI	1274
• 	Study day of start of RBC TI	1275
• 	Date of end of RBC TI	1276
• 	Duration of RBC TI post baseline	1277
• 	Date of start of platelet TI	1278
• 	Study day of start of platelet TI	1279
• 	Date of end of platelet TI	1280
• 	Duration of platelet TI post baseline	1281
• 	Date of start of RBC and platelet TI	1282
• 	Study day of start of RBC and platelet TI	1283
• 	Date of end of RBC and platelet TI	1284
• 	Duration of RBC and platelet TI post baseline	1285
• 	Date of last contact	1286
• 	Study day of last contact	1287
• 	Status at last contact (alive and TI, alive and transfusion-dependent, dead, or lost)	1288