Alcoholism: Developing Drugs for Treatment

Administrative draft guidance, not enacted legislation.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

Alcoholism:
Developing Drugs 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DRAFT GUIDANCE
This guidance document is being distributed for comment purposes only.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publication in the Federal Register of the notice announcing the availability of the draft
guidance.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 http://www.regulations.gov. Submit written
comments to the Division of Dockets Management (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 Rachel Skeete at 301-796-2280.
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Drug Evaluation and Research (CDER)
February 2015
Clinical/Medical
15003dft.doc
02/09/15

Alcoholism:
Developing Drugs for
Treatment
Guidance for Industry
Additional copies are available from:
Office of Communications, Division of Drug Information
Center for Drug Evaluation and Research
Food and Drug Administration
10001 New Hampshire Ave., Hillandale Bldg., 4th Floor
Silver Spring, MD 20993
Tel: 855-543-3784 or 301-796-3400; Fax: 301-431-6353; Email: druginfo@fda.hhs.gov
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Drug Evaluation and Research (CDER)
February 2015
Clinical/Medical

TABLE OF CONTENTS
I. 	INTRODUCTION............................................................................................................. 1
II. 	DEVELOPMENT PROGRAM ....................................................................................... 2
A. 	General Considerations .................................................................................................................3
1. Early Phase Clinical Development Considerations .........................................................................3
2. Drug Development Population .........................................................................................................3
3. Efficacy Considerations ...................................................................................................................4
4. Safety Considerations.......................................................................................................................4
B. 	Specific Efficacy Trial Considerations .........................................................................................4
1. Trial Design......................................................................................................................................4
2. Trial Population ...............................................................................................................................5
3. Entry Criteria ...................................................................................................................................6
4. Special Populations ..........................................................................................................................6
5. Choice of Comparators ....................................................................................................................6
6. Efficacy Endpoints............................................................................................................................7
7. Endpoint Adjudication ......................................................................................................................7
REFERENCES.............................................................................................................................. 9
APPENDIX 1: TERMINOLOGY RELATED TO PROBLEM ALCOHOL USE.............. 10
APPENDIX 2: SOURCES PROVIDING SUPPORT FOR NO HEAVY DRINKING
AS A VALID SURROGATE FOR CLINICAL BENEFIT .................................................... 11

Contains Nonbinding Recommendations
Draft — Not for Implementation
1 	Alcoholism: Developing Drugs for Treatment
2 	Guidance for Industry1
3
4
5
6
7 	This draft guidance, when finalized, will represent the Food and Drug Administration’s (FDA’s) current
8 	thinking on this topic. It does not create or confer any rights for or on any person and does not operate to
9 	bind FDA or the public. You can use an alternative approach if the approach satisfies the requirements of
10 	the applicable statutes and regulations. If you want to discuss an alternative approach, contact the FDA
11 	staff responsible for implementing this guidance. If you cannot identify the appropriate FDA staff, call
12 	the appropriate number listed on the title page of this guidance.
13
14
15
16
17 	I. 	INTRODUCTION
18
19 	The purpose of this guidance is to assist sponsors in the clinical development of drugs for the
20 	treatment of alcoholism.2 There are many different terms, definitions, and diagnostic criteria that
21 	have been used to describe this condition. However, in this guidance, we use the term
22 	alcoholism to describe patients with alcohol use problems that would make them candidates for
23 	treatment with medication. As the World Health Organization (WHO) notes,3 alcoholism is a
24 	“term of long-standing use” and is “generally taken to refer to chronic continual drinking or
25 	periodic consumption of alcohol which is characterized by impaired control over drinking,
26 	frequent episodes of intoxication, and preoccupation with alcohol and the use of alcohol despite
27 	adverse consequences.” Further discussion of terminology can be found in Appendix 1.
28
29 	We are issuing this guidance to better communicate our current thinking on the appropriate
30 	endpoints for clinical trials of drugs to treat alcoholism, and to apprise sponsors of possible
31 	alternatives to abstinence-based endpoints, which have often been considered an unattainable
32 	threshold in the clinical trial setting, and which may be considered a hindrance to clinical
33 	development for drugs to treat alcoholism. This guidance provides supporting information for
34 	the endpoints as appropriate measures of clinical benefit. This draft guidance is intended to
35 	serve as a focus for continued discussions among the Division of Anesthesia, Analgesia, and
1
1 This guidance has been prepared by the Division of Anesthesia, Analgesia, and Addiction Products in the Center
for Drug Evaluation and Research (CDER) at the Food and Drug Administration.
2 For the purposes of this guidance, all references to drugs include both human drugs and therapeutic biological
products unless otherwise specified.
3 http://www.who.int/substance_abuse/terminology/who_lexicon/en/ (accessed 2/1/14)

Contains Nonbinding Recommendations
Draft — Not for Implementation
36 	Addiction Products (DAAAP), pharmaceutical sponsors, the academic community, and the
37 	public.4
38
39 	This guidance does not contain discussion of the general issues of statistical analysis or clinical
40 	trial design. Those topics are addressed in the ICH guidances for industry E9 Statistical
41 	Principles for Clinical Trials and E10 Choice of Control Group and Related Issues in Clinical
42 	Trials, respectively.5
43
44 	FDA’s guidance documents, including this guidance, do not establish legally enforceable
45 	responsibilities. Instead, guidances describe the Agency’s current thinking on a topic and should
46 	be viewed only as recommendations, unless specific regulatory or statutory requirements are
47 	cited. The use of the word should in Agency guidances means that something is suggested or
48 	recommended, but not required.
49
50
51 	II. 	DEVELOPMENT PROGRAM
52
53 	In all diagnostic schemes, alcoholism is identified by behavior — continued self-administration
54 	of alcohol despite physical and psychosocial consequences. Alcoholism is understood to be a
55 	chronic, relapsing disorder that may require long-term and even lifelong treatment. The aim of
56 	treatment is often expressed as an effort to modify drinking behavior, but the actual desired
57 	effect is improvement in physical and psychosocial consequences. Therefore, drinking behavior
58 	(particularly that snapshot of behavior that can be observed during the brief window of a clinical
59 	trial) is considered a surrogate endpoint, not a direct measure of how the patient feels or
60 	functions. Trials intended to show direct effects on physical or psychosocial consequences of
61 	drug use would need to be long and large, and may be impractical. As such, sponsors do not
62 	need to demonstrate a direct effect on the physical and psychosocial consequences of alcoholism
63 	in alcoholism clinical trials, but they should show modifications in drinking behavior ascribed to
64 	a particular treatment that are likely to translate to improvement in the physical and psychosocial
65 	consequences.
66
67 	Because drinking behavior is considered a surrogate endpoint, sponsors should document a
68 	pattern of behavior that can be reasonably predictive of clinical benefit (e.g., improvement in the
69 	way the patient feels or functions). Patients who attain and sustain complete abstinence from
70 	alcohol may be assumed to accrue clinical benefit. Thus, trials showing a difference in the
71 	proportion of patients who attain and sustain complete abstinence may support an indication of
72 	treatment of alcoholism. We believe analyses of existing data also support the use of another
73 	valid surrogate endpoint defined by a pattern of reduced drinking, described as no heavy drinking
74 	days. Heavy drinking days are defined by the National Institute on Alcohol Abuse and
75 	Alcoholism (NIAAA) as days when the patient consumes more than four standard drinks (men)
4 In addition to consulting guidances, sponsors are encouraged to contact the division to discuss specific issues that
arise during the development of drugs to treat alcoholism.
5 We update guidances periodically. To make sure you have the most recent version of a guidance, check the FDA
Drugs guidance Web page at
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm.
2

PREA, originally enacted on December 3, 2003 (Public Law 108-155), codified many of the elements of the
pediatric rule, and established requirements for studies of certain drugs and biological products used in pediatric
patients. PREA (section 505B of the Federal Food, Drug, and Cosmetic Act, 21 U.S.C. 355c), reauthorized by the
Food and Drug Administration Amendments Act of 2007, as Title IV, on September 27, 2007 (21 U.S.C. 355c), and
made permanent in 2012 with the passage of the Food and Drug Administration Safety and Innovation Act (Public
Law 112-144), requires pediatric studies for certain drugs and biological products.
Contains Nonbinding Recommendations
Draft — Not for Implementation
76 	or more than three standard drinks (women). Standard drinks are defined in the United States as
77 	containing 14 grams of alcohol, such as would be found in a standard shot of hard liquor, a 12-
78 	ounce bottle of beer, or a 5-ounce glass of wine. An analysis of the proportion of patients who
79 	attain and sustain a pattern of drinking that never exceeds the heavy drinking definitions may be
80 	appropriate.
81
82 	We anticipate this pattern to be attained within a reasonable, behaviorally, and
83 	pharmacologically justified grace period, and that it be sustained for at least 6 months of
84 	treatment. We do not necessarily expect that efficacy will be sustained after the drug is
85 	withdrawn if the drug is intended to be administered chronically.
86
87 	DAAAP’s current recommendation is for trials of 6 months’ duration, with a primary endpoint of
88 	the proportion of patients who do not have any heavy drinking days during the observation
89 	period (percent no heavy drinking days). Background explanations to support the validity of this
90 	endpoint as a surrogate for clinical benefit may be found in Appendix 2.
91
92 	A. 	General Considerations
93
94 	1. 	Early Phase Clinical Development Considerations
95
96 	As part of their early phase clinical development program, sponsors should consider drug-
97 	alcohol interactions and may need to conduct formal drug-alcohol interaction trials. When
98 	designing the early phase clinical development program, sponsors should also be aware of, and
99 	give consideration to, the possibility that patients with chronic alcoholism or with hepatic
100 	impairment might have different pharmacokinetic/pharmacodynamic profiles from the general
101 	population.
102
103 	2. 	Drug Development Population
104
105 	In general, the target population for this indication should be adults who are seeking treatment
106 	for alcoholism. Early phase trials in which alcohol is administered generally should be
107 	conducted in nontreatment-seeking individuals, although there may be some circumstances under
108 	which administration of alcohol to treatment-seeking individuals may be justified.
109
110 	To fulfill the requirements of the Pediatric Research Equity Act (PREA),6 studies in adolescents
111 	(12 through 16 years old) may be required. Sponsors should assess the size of the treatment-
112 	seeking population to determine whether studies in this population may be practicably
113 	conducted. A waiver will be considered based on this assessment.
114
3
6

Contains Nonbinding Recommendations
Draft — Not for Implementation
115 	3. 	Efficacy Considerations
116
117 	Generally, two adequate and well-controlled trials will be needed to support an efficacy claim for
118 	this indication.
119
120 	4. 	Safety Considerations
121
122 	The safety database should be sufficiently sized, from both the standpoint of sample size and
123 	length of observation, to assess the safety of a drug intended for the treatment of a chronic
124 	disease.
125
126 	The size of the safety database depends on a number of factors, including whether the drug is a
127 	new molecular entity (NME) or a reformulation of a known drug substance, the nature of the
128 	safety findings from the clinical trials, and the nonclinical data for the drug under development.
129 	For the safety evaluation of an NME intended for treatment of alcoholism, we recommend
130 	sponsors refer to the ICH guidance for industry E1A The Extent of Population Exposure to
131 	Assess Clinical Safety: For Drugs Intended for Long-Term Treatment of Non-Life-Threatening
132 	Conditions for drugs intended for long-term treatment of non-life-threatening conditions and to
133 	the guidance for industry Premarketing Risk Assessment. These guidances make
134 	recommendations on the minimum size of the database. A safety database larger than
135 	recommended in these guidances may be warranted for a number of reasons (many of which are
136 	discussed in these guidances), including safety signals emerging as more clinical data become
137 	available.
138
139 	For reformulations of drugs with existing alcoholism indications, the size of the safety database
140 	should reflect the differences from existing formulations of the drug and the gap in safety data
141 	expected from these differences. In general, in the case of reformulated drugs, the amount of
142 	safety data that should be collected to support safe use depends on differences in
143 	pharmacokinetics and route of administration. To determine an appropriate number of patients
144 	for the safety database for a drug previously approved for an alcoholism indication, or other
145 	indication, sponsors should consider the extent of differences between the previous patient
146 	population studied and the alcoholism population under evaluation, and whether the differences
147 	alter the risk for adverse reactions.
148
149 	B. 	Specific Efficacy Trial Considerations
150
151 	1. 	Trial Design
152
153 	Alcoholism clinical trials should be designed as randomized, placebo-controlled, superiority
154 	trials with a minimum duration of 6 months and a primary endpoint based on the response rate.
155 	Responders can be defined either as patients who do not drink at all during the observation
156 	period, or as patients who do not have any heavy drinking days during the observation period. A
157 	responder analysis is recommended because calculations of group mean values such as percent
158 	days abstinent or percent of heavy drinking days across a treatment group are difficult to
159 	interpret with respect to the clinical benefit for individual patients. Responder analyses illustrate
160 	the clinically important effect of a treatment in an individualized fashion, and facilitate risk-
4

Contains Nonbinding Recommendations
Draft — Not for Implementation
161 	benefit comparisons. Responder analyses also may reveal the effect of a drug that has a
162 	clinically important effect in a small subset of patients.
163
164 	The primary endpoint and responder definition were chosen based on unpublished analyses
165 	commissioned by NIAAA of longitudinal data from both clinical trial settings and observational
166 	settings. Patients who never exceeded the heavy drinking limits had minimal alcohol-related
167 	consequences and were much less likely to have relapsed at follow-up.
168
169 	Abstinence also can be used as a responder definition for the primary endpoint.
170
171 	The recommended trial duration is based on data indicating that drinking patterns over shorter
172 	durations of time, such as 12 weeks, may not be stable or representative of future experience
173 	(Zweben and Cisler 2003) and may be too brief to predict ongoing treatment response. It is
174 	acknowledged that many other chronic diseases are studied in trials of only 3 months’ (12
175 	weeks’) duration using direct measures of clinical benefit. However, the problems associated
176 	with alcoholism are not readily reversible with cessation of drinking or with the avoidance of
177 	heavy drinking. Sustained adherence to the target change in drinking behavior, an indirect
178 	measure, is needed to accrue clinical benefit.
179
180 	It is also understood that periods of abstinence are quite common among alcohol-dependent
181 	individuals: in one survey (Schuckit, Tipp, et al. 1997), periods of abstinence lasting at least 3
182 	months were reported by 62.3 percent. This could make it hard to show a treatment effect in a
183 	brief alcoholism treatment trial.
184
185 	Some might suggest that trials of 1 year’s duration would be more appropriate. In the alcoholism
186 	field, the duration of abstinence considered to represent a stable condition, or sustained
187 	remission, is often set at 12 months. 7 Once well-established, abstinence from alcohol appears,
188 	for many patients, to be a stable pattern, sustained over several years of follow-up (Dawson,
189 	Goldstein, et al. 2007). However, based on data indicating that abstinence at 6 months has been
190 	shown to be a predictor of abstinence at 5-year follow-up (Weisner, Ray, et al. 2003), and in the
191 	interest of practicality, we recommend trials with a minimum of 6 months on-treatment
192 	observation.
193
194 	Neither abstinence nor no heavy drinking responder definitions need any additional data to
195 	support the pattern of drinking behavior as a valid surrogate for clinical benefit.
196
197 	Sponsors can choose other definitions of treatment responders, but would need to submit data
198 	that demonstrate the target drinking pattern they select is a valid surrogate for clinical benefit.
199
200 	2. 	Trial Population
201
202 	The trial population should be patients with alcoholism who require pharmacologic treatment.
203 	The Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria are commonly used to
204 	define addiction populations. However, the latest version of the DSM (DSM-V) subsumes all
7 Diagnostic and Statistical Manual of Mental Disorders, 2000, Fourth Edition, Text Revision (DSM-IV-TR),
American Psychiatric Publishing, Inc.
5

Contains Nonbinding Recommendations
Draft — Not for Implementation
205 	problematic use of alcohol under the term alcohol use disorders (AUD). The blanket designation
206 	of AUD introduces ambiguity with respect to defining a trial population for alcoholism clinical
207 	trials, because patients with mild or early alcohol use problems would be included if all patients
208 	with DSM-V AUD were enrolled. Therefore, sponsors should create eligibility criteria adequate
209 	to define a population of patients with a degree of AUD severity that may benefit from
210 	pharmacologic treatment. The DSM diagnostic criteria for AUD can be used as a foundation,
211 	augmented by other key factors that would identify that set of patients with AUD for whom
212 	pharmacologic treatment would be appropriate. These might include a requirement that
213 	particular DSM-V diagnostic criteria are met, or that other features are present, such as a
214 	subjective loss of control over drinking.
215
216 	It is important to highlight that patients with mild degrees of AUD are likely to benefit from
217 	nonpharmacologic interventions and are therefore not the target population for drugs to treat
218 	alcoholism. This means both that the risk-benefit calculation is different for these patients than
219 	for those who are in need of pharmacologic treatment, and that they may have a high rate of
220 	placebo response that can complicate the demonstration of efficacy in clinical trials. Thus, we
221 	recommend that patients whose problems fall into the category of mild alcohol use disorder,
222 	some patients who would meet criteria for moderate alcohol use disorder, or those who are
223 	perceived to have a problem of abuse but not addiction, are not generally the ideal population for
224 	study.
225
226 	3. 	Entry Criteria
227
228 	Patients should have a history of episodes of heavy drinking in the period before screening that
229 	would permit detection of a change in drinking behavior in this regard as a result of
230 	pharmacotherapy. Patients with a history suggestive of clinically significant withdrawal
231 	symptoms can be enrolled, but the protocol should include procedures to monitor for withdrawal
232 	and to provide necessary treatment.
233
234 	The decision to enroll patients who are drinking at baseline or patients who have ceased drinking
235 	at the time of enrollment should be based on the presumed mechanism of action of the drug.
236 	Some drugs may be hypothesized to be effective in helping patients to stop drinking by blocking
237 	the effects of alcohol (i.e., via extinction of the behavior). Other drugs might be useful in
238 	reducing the risk of relapse once drinking has ceased.
239
240 	4. 	Special Populations
241
242 	Patients with a range of comorbid conditions typically seen in patients with alcoholism,
243 	including hepatic impairment, should be enrolled in clinical trials.
244
245 	5. 	Choice of Comparators
246
247 	Sponsors should use placebo comparators. An active comparator also can be included in the
248 	trial. Claims of comparative superiority, however, involve specific planning in trial design to
249 	demonstrate that there is a clinically meaningful benefit of the drug in question over the
6

Contains Nonbinding Recommendations
Draft — Not for Implementation
250 	comparator. The comparator drug should be used in an effective dose and in an appropriate
251 	population to support a comparative claim.
252
253 	6. 	Efficacy Endpoints
254
255 	Trials should measure the proportion of patients in each treatment group who attain, and sustain
256 	over the observation period, a target drinking pattern that is considered a valid surrogate for
257 	clinical benefit. The following two options can be used as target drinking patterns and do not
258 	need any additional data to support the pattern as a valid surrogate for clinical benefit. Sponsors
259 	should discuss other options with the division.
260
261 	(1) Abstinence. As noted above, trials that use complete abstinence as the target drinking
262 	pattern can be used.
263
264 	(2) No Heavy Drinking. Trials that use no heavy drinking as the target drinking pattern can
265 	be used. This is based on several lines of evidence that provide support for this pattern as
266 	a valid surrogate for clinical benefit. Several of these lines of support are from
267 	unpublished analyses, but there are also published studies that confirm these analyses.
268 	Support for this endpoint is summarized in Appendix 2.
269
270 	7. 	Endpoint Adjudication
271
272 	Information about patients’ drinking can be collected using the Time-Line Follow-Back Method
273 	(TLFB) (Sobell, Maisto, et al. 1979). Briefly, the TLFB is a calendar-assisted retrospective
274 	reconstruction of how many drinks were consumed per day. Initially, the TLFB was developed
275 	to be administered by a research assistant, but other techniques including computer-based
276 	administration have also been developed. The retrospective window is as long as 3 months. It is
277 	generally understood that the TLFB data are not a precise reflection of a patient’s drinking, but
278 	the TLFB has been widely accepted as providing a reasonable estimate that is sensitive to
279 	change.
280
281 	Note that if sponsors are interested in documenting only abstinence versus any drinking, or
282 	adherence to nonheavy limits versus any violation of heavy drinking limits, it may not be
283 	necessary to use the TLFB method of reconstruction of drinking day by day. Other methods may
284 	be sufficient for obtaining the information necessary to adjudicate the patient as a responder or
285 	nonresponder. For example, the Alcohol Research Group/National Alcohol Research Center’s
286 	2009 – 2010 National Alcohol Survey 8 used the following question: “Think of all kinds of
287 	alcoholic beverages combined, that is, any combination of bottles or cans of beer, glasses of
288 	wine, drinks containing liquor of any kind, or coolers, flavored malt beverages or pre-made
289 	cocktails. In this question, 1 drink is equal to a 12 ounce bottle or can of beer or cooler, a five
290 	ounce glass of wine, or a 1 shot of liquor (1.5 ounces). During the last 12 months, what is the
291 	largest number of drinks you had on any single day?”.
292
293 	Either self-report method should be supplemented by some type of biological verification.
294 	Currently, there are no ideal biomarkers of drinking that can be used to reliably capture any
8 http://www.arg.org/downloads/arg/N12%20FINAL%20Landline%20Questionnaire.pdf (accessed 4/25/14)
7

Contains Nonbinding Recommendations
Draft — Not for Implementation
295 	instance of drinking or of heavy drinking, but some attempt to collect biological data may have
296 	the effect of increasing the veracity of self-report. Additionally, patients who are acutely
297 	intoxicated cannot give reliable retrospective accounts. Therefore, at a minimum, a breath
298 	alcohol measurement at each visit should be incorporated.
299
300 	It is recommended that the data on alcohol use be collected by staff who are not providing
301 	counseling. This is intended to reduce the likelihood that patients will conceal drinking to avoid
302 	disappointing the therapist.
303
304 	If sponsors are interested in documenting only whether patients are responders or nonresponders,
305 	it is not necessary to accurately reconstruct the amount consumed on each day of the trial, and
306 	therefore there are methods to ensure that missing data should be relatively rare. A patient who
307 	has already had a heavy drinking day during the efficacy ascertainment period is already
308 	adjudicated even if lost to follow-up. If a patient who met the responder definition up to the
309 	point of dropping out can be located by telephone, he or she can be asked “What is the largest
310 	number of drinks you had on any one occasion since the last time we saw you?” If the patient
311 	indicates that he or she has had at least one heavy drinking day, the outcome for that patient is
312 	adjudicated and is not considered missing. Patients who self-report ongoing adherence to the no
313 	heavy drinking limits may present a challenge because of the lack of biological verification or
314 	other sources of confirmation of self-report; these patients might be included as either responders
315 	or nonresponders in different sensitivity analyses. Only patients who met the responder
316 	definition up to the point of loss to follow-up and cannot be located should be considered truly
317 	unadjudicated. Careful attention to obtaining contact information at the time of trial enrollment
318 	can limit the number of patients for whom outcome data are truly missing.
319
8

Contains Nonbinding Recommendations
Draft — Not for Implementation
320 	REFERENCES
321
322 	Dawson, DA, RB Goldstein, and BF Grant, 2007, Rates and Correlates of Relapse Among
323 	Individuals in Remission From DSM-IV Alcohol Dependence: A 3-Year Follow-Up, Alcohol
324 	Clin Exp. Res., 31, 2036-2045.
325
326 	Delucchi, KL and C Weisner, 2010, Transitioning Into and Out of Problem Drinking Across
327 	Seven Years, J Stud. Alcohol Drugs, 71, 210-218.
328
329 	Greenfield, TK, 2000, Ways of Measuring Drinking Patterns and the Difference They Make:
330 	Experience with Graduated Frequencies, J Subst. Abuse, 12, 33-49.
331
332 	Sanchez-Craig, M, DA Wilkinson, and R Davila, 1995, Empirically Based Guidelines for
333 	Moderate Drinking: 1-Year Results From Three Studies With Problem Drinkers, Am. J Public
334 	Health, 85, 823-828.
335
336 	Schuckit, MA, JE Tipp, TL Smith, and KK Bucholz, 1997, Periods of Abstinence Following the
337 	Onset of Alcohol Dependence in 1,853 Men and Women, J Stud. Alcohol, 58, 581-589.
338
339 	Sobell, LC, SA Maisto, MB Sobell, and AM Cooper, 1979, Reliability of Alcohol Abusers’ Self-
340 	Reports of Drinking Behavior, Behav. Res. Ther., 17, 157-160.
341
342 	Weisner, C, GT Ray, JR Mertens, DD Satre, and C Moore, 2003, Short-Term Alcohol and Drug
343 	Treatment Outcomes Predict Long-Term Outcome, Drug Alcohol Depend., 71, 281-294.
344
345 	Zweben, A and RA Cisler, 2003, Clinical and Methodological Utility of a Composite Outcome
346 	Measure for Alcohol Treatment Research, Alcohol Clin Exp. Res., 27, 1680-1685.
347
9

Contains Nonbinding Recommendations
Draft — Not for Implementation
348 	APPENDIX 1: TERMINOLOGY RELATED TO PROBLEM ALCOHOL USE
349
350 	The WHO notes that alcoholism may be considered to be synonymous with alcohol addiction,
351 	but does not endorse the use of either term. Addiction, per the WHO, is “Repeated use of a
352 	psychoactive substance or substances, to the extent that the user (referred to as an addict) is
353 	periodically or chronically intoxicated, shows a compulsion to take the preferred substance (or
354 	substances), has great difficulty in voluntarily ceasing or modifying substance use, and exhibits
355 	determination to obtain psychoactive substances by almost any means. Typically, tolerance is
356 	prominent and a withdrawal syndrome frequently occurs when substance use is interrupted.”9
357
358 	More recently, the term alcohol dependence was substituted for both alcoholism and alcohol
359 	addiction in diagnostic criteria by both the WHO and the American Psychiatric Association
360 	(APA), because of concerns that the word addiction carried an unwanted stigma, and in turn,
361 	could be a barrier to seeking treatment. However, this created ambiguity, because the term
362 	dependence came to have dual meanings connoting both a physical neuroadaptation (sometimes
363 	called physical dependence) and the notion of addiction.
364
365 	Another term, alcohol abuse has been applied when individuals use alcohol to the point of
366 	experiencing problems caused by drinking, but do not manifest features of alcoholism. Notably,
367 	there have been concerns voiced that the term abuse also carries a stigma that would prevent
368 	individuals from self-identifying or seeking treatment and suggestions have been made to
369 	abandon this term as well, replacing it with misuse. For FDA purposes, the terms addiction,
370 	dependence, abuse, and misuse are distinct from one another, but we acknowledge that they may
371 	be used inconsistently, and sometimes interchangeably. We have retained the historical term,
372 	alcoholism, in this guidance because of ambiguity and ongoing evolution in the use of other
373 	terms.
374
375 	In the most recent version of the APA’s DSM, a new diagnostic approach subsuming all
376 	problematic use of alcohol under the term AUD has been put forth. This construct eliminates the
377 	distinction between alcohol abuse and alcohol dependence (the latter term being essentially
378 	synonymous with alcohol addiction or alcoholism), and creates a diagnosis for individuals with
379 	problems mild enough that they are merely markers for future problems related to drinking. This
380 	facilitates early identification and intervention, but it also creates a problematic level of
381 	heterogeneity in the group of people who meet criteria for the diagnosis of AUD.
382
9 http://www.who.int/substance_abuse/terminology/who_lexicon/en
10

Contains Nonbinding Recommendations
Draft — Not for Implementation
383 	APPENDIX 2: SOURCES PROVIDING SUPPORT FOR NO HEAVY DRINKING
384 	AS A VALID SURROGATE FOR CLINICAL BENEFIT
385
386 	This Appendix summarizes select sources of information that provide support for the distinctive
387 	pattern of drinking reduction, referred to as no heavy drinking days in this guidance, as a valid
388 	surrogate for clinical benefit.
389
390 	Project MATCH
391
392 	The first unpublished analysis, commissioned by the Treatment Research Branch at NIAAA,
393 	explored the dataset from Project MATCH, a trial comparing 3 different behavioral
394 	(nonpharmacologic) treatments delivered over 12 weeks in 1,726 patients with diagnoses of
395 	alcohol abuse or dependence who had been actively drinking in the 3 months before trial entry.
396 	Assessments were conducted every 3 months, capturing both alcohol consumption and various
397 	measures of drinking-related consequences, and patient function or dysfunction. The analysis
398 	examined the relationship of problems and functioning to various measures of drinking. 10 The
399 	investigator found a high degree of variability using continuous measures of drinking such as
400 	percent days abstinent or drinks per day, but that a consumption quantity cut-off was related
401 	strongly to an array of consequences and functioning variables. The conclusion was that the best
402 	single predictor of nonconsequential drinking was never exceeding the daily heavy drinking
403 	limits. The recommendations based on this analysis were that the target pattern of drinking
404 	should be defined as being abstinent or never exceeding three drinks on a single occasion
405 	(women) or four drinks on a single occasion (men). In the sample analyzed, 22 percent of
406 	patients met this definition over the full 12-month post-treatment follow-up.
407
408 	If a target drinking pattern based on percent days abstinent was of interest, the analysis suggested
409 	that similar functional outcomes would require a pattern of 92 percent days abstinent. Because
410 	of high degrees of fluctuations in consumption and status across time, the analysis also suggested
411 	that longer follow-up periods (6 to 12 months) were needed to provide insight into more
412 	sustained status.
413
414 	National Alcohol Surveys
415
416 	The second NIAAA-commissioned analysis used data from the 1995 and 2000 National Alcohol
417 	Surveys, which collected information on alcohol consumption using a graduated frequencies
418 	measure (Greenfield 2000), alcohol dependence criteria, and information about alcohol-related
419 	social consequences. Participants included 7,447 current drinkers, but the analysis focused on
420 	the subset of 820 respondents who either reported having had prior treatment for alcohol
421 	problems or endorsed a concern about their drinking, to better approximate the target population
422 	for alcoholism treatment drugs.
423
424 	The investigator concluded that treated or concerned drinkers who restrict intake to low volume
425 	(averaging fewer than 2 drinks per week) and whose quantities in a day never exceeded
10 The analysis of the Project MATCH data was conducted by Dr. Ron A. Cisler.
11

Contains Nonbinding Recommendations
Draft — Not for Implementation
426 	3 (women)/4(men) carry low risk of 12-month dependence or abuse (less than 5 percent).11
427 	Those drinking 4 plus/5 plus even on occasion have significantly higher risks (10 to 20 percent)
428 	of meeting criteria for AUD. The report noted that “If [treated or concerned] individuals drink at
429 	all, the only somewhat ‘safe’ level appears to be drinking less than 2 drinks/week on average and
430 	never exceeding 4 drinks for a man or 3 drinks in a day for a woman.”
431
432 	National Epidemiologic Survey on Alcohol and Related Conditions
433
434 	Findings from two waves of data from the National Epidemiologic Survey on Alcohol and
435 	Related Conditions were published by Deborah Dawson and colleagues (Dawson, Goldstein, et
436 	al. 2007). Wave 1, collected in 2001 to 2002, identified 4,422 individuals who had met criteria
437 	for alcohol dependence before the past year. Of these:
438
439 	 	25.0 percent were still classified as dependent in the past year
440
441 	 	27.3 percent were classified as being in partial remission
442
443 	 	11.8 percent were asymptomatic risk drinkers who demonstrated a pattern of drinking
444 	that put them at risk of relapse
445
446 	 	17.7 percent were low-risk drinkers (no heavy drinking days)
447
448 	 	18.2 percent were abstainers
449
450 	The last 3 categories comprise 2,109 individuals in full remission from alcohol dependence.
451
452 	At Wave 2, collected 2004 to 2005, 1,772 of those 2,109 individuals were re-interviewed.
453 	Recurrence of AUD symptoms occurred in 51 percent of asymptomatic risk drinkers (any heavy
454 	drinking days); 27.2 percent of low-risk drinkers (no heavy drinking days); and 7.3 percent of
455 	abstainers. The adjusted odds ratios of recurrence of AUD symptoms compared to abstainers
456 	was 14.6 for asymptomatic risk drinkers and 5.8 for low-risk drinkers.
457
458 	The proportion of individuals who had been in remission at Wave 1 who met criteria for alcohol
459 	dependence at Wave 2 was 10.2 percent for asymptomatic risk drinkers (any heavy drinking
460 	days); 4 percent for low-risk drinkers (no heavy drinking days); and 2.9 percent for abstainers.
461 	The adjusted odds ratios of recurrence of dependence, relative to abstainers, was 7.0 for risk
462 	drinkers and 3.0 for low-risk drinkers. Thus, compared to abstinence, no heavy drinking days
463 	does still represent three times the risk of relapse to dependence compared to abstinence and
464 	nearly six times the risk of relapse to AUD symptoms. However, drinking patterns including any
465 	heavy drinking days seem to carry 7 times the risk of relapse to dependence compared to
466 	abstinence and nearly 15 times the risk of relapse to AUD symptoms compared to abstinence.
467 	Those who engaged in risk drinking (equals heavy drinking days) even fewer than 1 time per
468 	month at Wave 1 were significantly more likely to meet criteria for dependence at Wave 2 than
11 The analysis of National Alcohol Survey data was conducted by Dr. Thomas Greenfield.
12

Contains Nonbinding Recommendations
Draft — Not for Implementation
469 	those who did not, and there was no association between frequency of risk drinking and the
470 	adjusted prospective risks of chronic medical conditions other than liver disease.
471
472 	Weisner
473
474 	Delucchi and Weisner (Delucchi and Weisner 2010) examined transitions into and out of
475 	problem drinking across 7 years in a longitudinal study of 1,350 problem drinkers sampled from
476 	one county’s general population (general population sample) and individuals entering the
477 	county’s public and private chemical dependency programs (treatment sample). Problem
478 	drinking was defined as 2 or more of the following in the previous 12 months: (1) 5 plus drinks
479 	per day at least once a month for men or 3 plus drinks in a day weekly for women; (2) 1 or more
480 	alcohol-related social consequences (from a list of 8); and (3) 1 or more alcohol dependence
481 	symptoms (from a list of 9). Follow-up interviews were conducted 1, 3, 5, and 7 years after
482 	baseline. The extent to which problem drinkers transition into and out of problem drinking was
483 	examined using Markov modeling.
484
485 	The authors reported that a latent Markov model with heterogeneous transitions and five patterns
486 	fit the data, and the estimated transition probabilities are displayed in the following figure found
487 	in the article.
488
489
490 	Source: Delucchi and Weisner, 2010, p. 215.
491
492 	The authors demonstrated that individuals transitioning into a status of nonproblem drinker are
493 	likely to remain in that status over time. Conversely, if individuals maintain a problem drinker
494 	status over time, it becomes increasingly difficult to transition out of that status.
495
496 	Sanchez-Craig
497
498 	Data from three independent trials involving two distinct populations of problem drinkers were
499 	pooled with the intent of refining guidelines on moderate drinking for heavy drinkers (Sanchez-
500 	Craig, Wilkinson, et al. 1995). The trial patients were 235 individuals who participated in 3
501 	trials of secondary prevention of alcohol problems and were interviewed at the 12-month follow-
502 	up period. Patients were classified as problem-free (reporting no alcohol-related problems in the
503 	past 6 or 9 months) or problem (reporting 1 or more problems). In the problem-free group, the
504 	average number of drinks per day and the upper limit of the confidence interval were less than
13

Contains Nonbinding Recommendations
Draft — Not for Implementation
505 	four. For the group reporting a problem, the mean quantity per day drinking was 5.5. The
506 	authors also conducted an analysis grouping patients at 12-month follow-up into 4 categories
507 	based on both the amount of drinks consumed per day and the frequency of drinking in a week.
508 	The authors concluded that the two groups above the cutoff on quantity of drinks per day (i.e.,
509 	five or more for men and four or more for women) had similar prevalence of all problem types
510 	and higher prevalence, while the two groups below this cutoff experienced a low likelihood of
511 	problems. This is an additional source of support that was already available at the time that
512 	NIAAA had commissioned the formerly described analyses.
513
14