当前位置：文档库 › Good statistics practice in the drug development and regulatory approval process

Good statistics practice in the drug development and regulatory approval process

GOOD STATISTICS THE DRUG DEVELOPMENT AND REGULATORY APPROV AL PROCESS

S HEIN-C HUNG C HOW,P H D

Executive Director,Biostatistics and Data Management,Covance,Inc.,Princeton,New Jersey

During the development and approval process of a new drug,the concept of good statistics

practice(GSP)is necessarily implemented.GSP is a set of principles which assures the

validity of the design and analysis of the intended studies conducted at various stages

of the process of drug development and regulatory approval.GSP provides a fair assess-

ment of the drug product with the desired accuracy,precision,and reliability.In essence,

GSP not only concerns the validity of statistical inference regarding drug efficacy and

safety,but also provides assurance of the proper identity,strength,quality,purity,and

stability of the drug product.This paper describes regulatory requirements for statistics

and the role of statistics in the drug development and regulatory process.The concept

and importance of GSP are illustrated through some practical and/or regulatory statistical

issues that commonly occur during the drug development and regulatory approval process.

Key Words:Code of Federal Regulations;United States Pharmacopeia and National

Formulary;Good laboratory practice;Good clinical practice;Current good manufacturing

practice;Good research practice;Good statistics practice

INTRODUCTION clinical evaluation due to the toxicity or intol-

erability in animals or humans.After a prom-THE DEVELOPMENT OF a pharmaceutical

ising compound is obtained,the clinical de-entity is a lengthy process involving drug

velopment and regulatory approval process discovery,formulation,laboratory develop-

is also time consuming,which is necessary ment,toxicological studies,clinical develop-

to assure the effectiveness and safety of the ment,and regulatory registration.The whole

promising compound.

process is not only time consuming but also

For the development of pharmaceutical very costly.For example,for drug discovery,

products,different countries,such as the it may require the screening of a large num-

United States,the European Community ber of compounds in order to obtain a few

(EC),and Japan have similar but slightly dif-potential promising compounds.In practice,

ferent sets of regulatory requirements.This it is very likely that these potential promising

paper focuses on the United States regulatory compounds may never reach other stages of

requirements for drug development.The drug development such as animal studies or

United States regulations for drug develop-

ment are developed based on the Federal

Food,Drug and Cosmetic Act(FD&C Act) Presented at the DIA“First International Taipei Sympo-passed in1938.The FD&C Act requires sium,”August29–30,1996,Taipei,Taiwan.pharmaceutical companies to submit full re-Reprint address:Shein-Chung Chow,PhD,Biosta-

ports of investigations on the safety of new tistics and Data Management,Covance,Inc.,210Car-

negie Center,Princeton,NJ08540.drugs.In1962,the significant Kefauver-Har-

1157

1158Shein-Chung Chow ris Drug Amendments to the FD&C Act were 1.Representative samples be drawn for test-passed.The Kefauver-Harris Amendments

ing,and

not only strengthen the safety requirements 2.A valid design and appropriate statistical for new drugs but also established for the

analysis be justified to ensure the accuracy first time an efficacy requirement for new and reliability of the test results. drugs.In1984,the United States Congress

passed the Price Competition and Patent As a result,the concept of good statistics Term Restoration Act to provide increased

practice(GSP)is the key to the success of patent protection to compensate for patent GLP,GCP,GRP,and cGMP,and conse-life lost during the approval process.Based

quently,the success of the drug product.The on this act,the Food and Drug Administra-objective of this paper is not only to introduce tion(FDA)was authorized to approve ge-

the concept of GSP but also to illustrate the neric drugs based on bioavailability and bioe-importance and the implementation of GSP quivalence trials on healthy male subjects.

through some practical and/or regulatory is-The United States regulations for drug devel-sues that commonly occur during the devel-opment are codified in the United States

opment of a drug product.

Code of Federal Regulations(CFR)which In the next section,the United States regu-are to be carried out by the Food and Drug

latory requirements for drug development are Administration.briefly outlined.The role of statistics in the Before the approval of a drug,the FDA

drug development and regulatory approval requires that substantial evidence of the ef-process is discussed in the section on“Good fectiveness and safety of the drug product be

Statistics Practice.”Also included in this sec-provided through the conduct of two well-tion are the concept of GSP,some difficulties/ controlled clinical studies.Before the drug

concerns that commonly are encountered be-can be tested in humans,however,it is also tween statisticians and scientists,and sugges-

tions for the implementation of GSP.In the required that an appropriate analytical meth-

od be developed and a number of animal section on“Practical Issues,”the concept of studies be conducted to assess the pharmaco-

GSP is illustrated by means of some practical logical characteristics,efficacy,and safety of and/or regulatory issues from the areas of the drug product in animals.To assist the

nonclinical,preclinical,and clinical drug de-sponsors in fulfilling the requirements,the velopment.A brief conclusion is given in the

last section.

FDA has issued a number of guidelines and

guidances at various stages of drug develop-

ment.The United States regulations codified

REGULATORY REQUIREMENTS

in the CFR and FDA guidelines and/or guid-

ances are then considered standards for good For approval of drug products,the FDA has laboratory practice(GLP),good clinical

published regulations at various stages of (GCP),and good regulatory practice(GRP).drug development.These regulations include

the investigational new drug application After the drug is approved,the FDA also

requires that the drug product be tested for its(IND)and new drug application(NDA)for identity,strength,quality,purity,and stability

new drugs,orphan drugs and over-the-before it can be released for use.For this counter human drugs and the abbreviated purpose,the FDA also issued current good

new drug application(ANDA)for generic manufacturing practice(cGMP)guidelines to drugs.

Table1summarizes CFR requirements at assure that the drug product possesses good

drug characteristics such as proper identity,various stages of new drug and generic drug strength,quality,purity,and stability.

development.The cGMP as codified in21 In drug development,almost all of the CFR210and211provides minimum require-GLP,GCP,GRP,and cGMP guidelines re-

ments for the manufacture,processing,pack-quire that:ing,and holding of a drug product.Its pur-

GSP in Drug Development and Regulatory Approval1159

TABLE1

The United States Regulations for Drug Development

CFR Number Stage of Pharmaceutical Development

21CFR312Investigational New Drug Application(IND)

21CFR314New Drug Application(NDA)

21CFR314Abbreviated New Drug Application(ANDA)

21CFR314Supplements

21CFR10Stability

21CFR50,56,312,314Good Clinical Practice(GCP)

21CFR210&211Current Good Manufacturing Practice(cGMP)

21CFR320Bioavailability/Bioequivalence

pose is to assure that the drug product meets strate substantial evidence of the effective-the standard for the identity,strength,quality,

ness and safety of the drug.As indicated in and purity of the drug product.The standards Section314of CFR Part21,an NDA should are usually referred to as those specified in

include information on CMC,nonclinical the United States Pharmacopeia and National pharmacology and toxicology,human phar-Formulary(USP/NF)(1).The USP/NF con-

macology and bioavailability,clinical data, sists of compendia of legally public standards and statistics for review by various related

divisions at the FDA before an approval deci-for drug identity,strength,quality,purity,

packaging,and labeling of drug products in sion can be reached.

the United States which are recognized as

Note that as indicated earlier,the develop-official compendia in the FD&C Act.The ment and approval of a new drug is a lengthy USP/NF describes not only the legally recog-

process.This lengthy process does not allow nized standard testing methods and assay the access of promising drugs or therapies

to patients with serious or life-threatening procedures for drug identity,strength,qual-

ity,and purity but also the number of samples illness.Under the so-called treatment IND to be tested and the acceptance of

published in1987,the FDA permits promis-

various tests for release standards.ing drugs or therapies currently under inves-For new drug development,an

tigation to be available to patients with seri-mits the sponsors to gather the data on clini-ous or life-threatening diseases.In1992,the safety and effectiveness that are needed

Parallel Track Regulations began allowing

an NDA.An IND should contain informa-promising drugs or therapies for serious or regarding chemistry,manufacturing,and

life-threatening diseases to become available controls(CMC)of the drug substance and with considerably fewer data than required

for approval.The FDA also published the drug product to ensure the identity,strength,

quality,and purity of the investigational Accelerated Approval based on surrogate drug.In addition,the sponsors are required

endpoints to accelerate the approval process to provide adequate information about phar-for promising drugs or therapies indicated macological studies for absorption,distribu-

for life-threatening diseases.

tion,metabolism,and excretion(ADME)and For generic drug development,the spon-

sors are required to include the information acute,subacute,and chronic toxicological

studies and reproductive tests in various ani-regarding product information(eg,potency mal species to support that the investiga-

and stability),pharmacokinetic data and tional drug is reasonably safe to be evaluated analysis,statistical analysis,analytical meth-in clinical trials in humans.In addition,for

odology and validation,and clinical data in approval of new drugs,the FDA requires at an ANDA submission for review.A complete

ANDA is then subject to a request for plant least two adequate well-controlled clinical

studies be conducted in humans to demon-inspection,chemistry/micro review,labeling

1160Shein-Chung Chow review,and bioequivalence review.The tic effect.The FDA requires that analytical ANDA would be approved if positive results

methods for the active ingredients of the drug are obtained from the plant inspection and be developed and validated so that the assay

results are in compliance with some estab-all reviews.Note that a new policy recently

issued by the Office of Generic Drugs of the lished specifications according to the USP/ FDA indicates that for drug product manu-

NF requirements.Statistics are necessarily factured at a foreign facility and shipped to applied in order to meet certain standards of the United States for packaging,release and

accuracy and reliability.At the same time, stability testing can now be conducted at any stability studies are usually conducted to es-

tablish the expiration dating period of the facility,United States or foreign.

To assist sponsors in fulfilling regulatory drug product according to the design and requirements,the FDA has published a num-

analysis as specified in the FDA stability ber of guidelines and guidances at various guideline.The FDA requires that statistical stages of drug development and regulatory

inference of the estimated expiration dating approval.For example,to establish an appro-period be included in the NDA submission

of the drug.

priate expiration dating period and product

storage requirements,the FDA published a Before the drug can be approved,the FDA stability guideline under21CFR10.90(2).

requires that substantial evidence of the ef-This guideline provides a means of meeting fectiveness and safety of the drug be provided regulatory requirements for an IND(21CFR

in the Technical Section of Statistics of an 312.23),an NDA(21CFR314.50),and an NDA submission.Since the validity of statis-

tical inference regarding the effectiveness ANDA(21CFR314.55).For clinical devel-

opment of new drugs,the FDA issued a and safety of the drug is always a concern, guideline which described general considera-

it is then suggested that a careful review be tions for clinical evaluation of drug products performed to ensure an accurate and reliable (3).In addition,for the approval of new

assessment of the drug product.In addition, drugs,the FDA also developed a guideline to have a fair assessment,the FDA also estab-

lished advisory committees,each consisting to assist sponsors in the preparation of sub-

missions(4).This guideline describes an ac-of clinical,pharmacological,and statistical ceptable format for submission of clinical

experts and one advocate(not employed by and statistical information on the drug prod-the FDA)in designated drug classes and sub-uct.For the approval of generic drugs,a guid-

specialities,to provide a second but indepen-ance on statistical procedures for bioequiva-dent review of the submission.The responsi-

bility of the statistical expert is not only to lence studies was issued by the Division of

Bioequivalence,Office of Generic Drugs,ensure that a valid design is used but also to FDA(5).This guidance suggests that a cross-

evaluate whether statistical methods used are over model with log-transformed data be appropriate for addressing the scientific and used to assess average bioequivalence for ap-

medical questions regarding the effective-proval of generic drugs.ness and safety of the drug.

After the drug is approved,the FDA also

requires that the drug product be tested for its GOOD STATISTICS PRACTICE identity,strength,quality,purity,and stability

before it can be released for use.For this The Role of Statistics in

purpose,the cGMP is necessarily imple-Drug Development

mented to:

When a new drug is discovered,it is impor-

tant to design an appropriate dosage form for 1.Validate the manufacturing process,

the drug so that it can be delivered efficiently

2.Monitor the performance of the manufac-to the site of action for the optimal therapeu-turing process,and

GSP in Drug Development and Regulatory Approval1161 3.Provide quality assurance of the final agents.The implementation of GSP involves product.

some key factors which have an impact on

the success of GSP.These factors include: At each stage of the manufacturing process, 1.Regulatory requirements for statistics, the USP/NF requires that sampling plans,ac- 2.The dissemination of the concept of statis-ceptance criteria,and valid statistical analy-tics,

ses be performed for the intended tests such 3.An appropriate use of statistics,

as potency,content uniformity,and dissolu- https://www.wendangku.net/doc/42887717.html,munication and flexibility,and tion.For each test,sampling plan,acceptance 5.Statistical training.

criteria,and valid statistical analysis are cru-

cial for determining whether the drug product These factors are briefly described below. can pass the test based on the results from a

In the drug development and approval representative sample.process,regulatory requirements for statis-

tics are the key to the implementation of GSP.

They not only enforce the use of statistics

but also establish standards for statistical Good Statistics Practice

evaluation of the drug product under investi-As discussed above,statistics plays an impor-gation.Statistical evaluation provides useful tant role during the development and regula-information for pharmaceutical scientists and tory approval process of new drug and ge-regulatory agents for determining whether neric drug products.A valid design and an the drug product under investigation has the appropriate use of statistical methods provide claimed effectiveness and safety for the in-an accurate and reliable assessment of the tended diseases,and whether the drug prod-drug under investigation.Therefore,good uct possesses good drug characteristics such statistics practice is necessarily applied in the as the proper identity,strength,quality,pu-drug development and regulatory approval rity,and stability.

process.GSP is defined as a set of principles In addition to regulatory requirements,it which assures the validity of the design and is always helpful to disseminate the concept analysis of the intended studies conducted at of statistics whenever possible.GSP not only various stages of the development and regu-fulfills regulatory requirements but it also latory approval process of a drug product.A improves the accuracy and precision of the valid design is one which can reflect the evaluation of the drug under investigation study objectives and address scientific or and the quality of the drug product manufac-medical questions regarding the drug product tured after approval.It is then important for under investigation.Under the valid design,pharmaceutical scientists to understand the a representative sample is then selected from concept of statistics during the process of the target population at random for study.drug development and regulatory approval. Statistical inference drawn based on appro-It is also critical for pharmaceutical scientists priate statistical methods under a valid design and regulatory agents to recognize that: can then provide a fair assessment with the

desired accuracy,precision,and reliability. 1.A valid statistical inference is necessary As a result,GSP is the key to the success

to provide a fair assessment with certain of the development and regulatory approval assurance regarding the uncertainty of the process of a drug product.

drug under investigation, The implementation of GSP,however,is 2.A larger sample size is often required to a team project which requires mutual com-

increase such assurance,and munication,confidence,respect,and cooper- 3.An invalid design and analysis may result ation among statisticians,pharmaceutical

in a misleading conclusion regarding the scientists in related areas,and regulatory drug under investigation.

1162Shein-Chung Chow

A commonly encountered problem in the The purpose of such training program is drug development and regulatory approval

threefold:

process is the misuse of design and statistical

methods in some studies.The misuse of de- 1.It enhances communications within the sign and statistical methods may result in

statistical community.Statisticians can either having the right question with the certainly benefit from such training pro-

grams by acquiring more practical experi-wrong answer,or having the right answer for

the wrong question.As a result,it is not clear ence from each other,

what the question is and what the answer is

2.It provides the opportunity to exchange because the design may not be able to address ideas and/or concepts regarding regulatory the question or statistical analysis cannot re-

requirements between professional socie-flect the design and hence cannot address the ties,and

3.Most importantly,it identifies critical question.As a result,GSP recommends that

statistical methods should be chosen to re-practical and/or regulatory issues that are flect the design which should be able to ad-

commonly encountered in the drug devel-dress the scientific or medical questions re-opment and regulatory approval process. garding the intended study objectives.

A panel discussion from different disci-

Communication and flexibility are impor-plines may result in some consensus to

resolve the issues.

tant factors in the success of GSP.Inefficient

communication between statisticians and

pharmaceutical scientists or regulatory agents

PRACTICAL ISSUES

may result in a misunderstanding of the in-

tended study objectives and consequently an This section illustrates the concept of GSP invalid design and/or inappropriate statistical

by means of some practical and/or regulatory methods.Thus,efficient communications issues commonly encountered during new

drug and generic drug development.For il-among statisticians,pharmaceutical scien-

tists,and regulatory agents is essential for lustration purposes,drug development is GSP.In addition,in many studies,the as-

classified into the areas of nonclinical,pre-sumption of a statistical design or model may clinical,and clinical.

not be met due to the nature of the drug under

study,experimental environment,and/or

Nonclinical Applications

other causes related/unrelated to the studies.

In this case,the traditional approach of doing In the pharmaceutical industry,a scale-up everything by the book does not help.A con-

experiment is usually employed to ensure cern from a pharmaceutical scientist or the that the results from a laboratory batch or regulatory agent may translate into a con-

small-scale production batch can be predict-straint for statistical design and analysis.In ive of a regular scale production batch.The this case,GSP suggests that a successful stat-

purpose of the scale-up experiment is not istician be flexible in the sense that a valid only to identify,evaluate,and optimize criti-statistical design and analysis should be de-

cal formulation and/or(manufacturing)pro-veloped under the constraints.cess factors of the drug product,but also to Since regulatory requirements for the drug

maximize or minimize excipient ranges.If development and approval process may vary there are a large number of formulation and/ from drug to drug,various designs and/or

or process factors,then the scale-up program statistical methods may be required to pro-could be very substantial.For example,sup-vide a valid assessment of a drug product.

pose a pharmaceutical company is interested Therefore,GSP suggests that statistical train-in studying the effects of eight process and ing programs be routinely held for both stat-

ingredient factors on various properties of isticians and nonstatisticians,including phar-tablets of a drug product.If each factor has maceutical scientists and regulatory agents.

two levels,a full28factorial design would

GSP in Drug Development and Regulatory Approval1163 require a total of256runs which it is almost with partial sampling time points has become impossible for the pharmaceutical company

very popular because it reduces the number to conduct.As an alternative,a fractional of assays tremendously.For example,if

one only performs stability testing on two factorial design is suggested.Suppose that it

is believed that only three or four factors will strengths per batch,the number of assays will have a major impact on the properties of the

be reduced by33.3%(from243to162). tablets of the drug product and that there are The above two examples demonstrate that no interactions involving four or more factors

an appropriate use of statistics is not only based on prior information.Then,a28?4frac-very cost effective but can only assist in

achieving the goal of the desired accuracy tional factorial design is useful.This design

reduces the total of256runs to only16runs.and reliability within the timeframe.It should For another example,consider the issue

be noted,however,that the choice of an ap-of stability analysis for establishment of an propriate fractional factorial design for a expiration dating period of a drug.In1993,

scale-up program or a reduced design such the International Conference on Harmoniza-as a matrixing design or a bracketing design tion(ICH)issued guidelines on stability

for stability studies should be statistically based on a strong industrial interest in inter-justified so that the results will have certain national harmonization of requirements for

accuracy,precision,and reliability without marketing in the EC,Japan,and the United losing much statistical power.

States(6).The ICH guideline is similar to

the current FDA stability guideline.For ex-

Preclinical Applications

ample,the ICH guideline suggests that test-

ing under the defined long-term conditions In recent years,as more generic drug prod-be normally done every three months over

ucts became available,whether the quality the first year,every six months over the sec-and the therapeutic effect of the generic drugs ond year,and then annually.It also requires

were comparable to the innovator drug be-that stability information from accelerated came a concern.Although the FDA indicates and long-term testing be provided on at least

that an approved generic drug can be used three batches.As a result,if there are three as a substitute for the innovator drug,it is

recognized that the current FDA regulation different strengths and three package types

of a drug product,a total of243assays are on average bioequivalence does not guaran-required based on a full factorial design over

tee drug interchangeability between generic four years.In practice,it is almost impossible drugs and the innovator drug(10).

to conduct a stability study of this size due

Drug interchangeability is usually classi-to the limited resources available and budget fied as drug switchability and drug prescriba-

bility.Drug prescribability is usually referred constraints.

Alternatively,as suggested by the ICH to as the physician’s choice for prescribing guideline,a matrixing design or bracketing

an appropriate drug product for his/her new design may be used.A matrixing design is patients between an innovator drug and a any subset of a full factorial design(7,8).A

number of generic drugs which have been bracketing design is the design of a stability shown to be bioequivalent to the innovator schedule such that at any time point only the

drug.The underlying assumption of prescri-samples on the extremes of container size bability is that the innovator drug and its and/or dosage strengths are tested(6).Nord-

generic drugs can be used interchangeably brock(9)examined a number of commonly in terms of the efficacy and safety of these used stability designs based on fractional fac-

drug products.

torial designs with partial sampling time Under current practice,the FDA only re-points.Some of these designs are,in fact,

quires that evidence of bioequivalence in av-special cases of matrixing designs.The use erage bioequivalence of the generic drugs to of a matrixing design or a bracketing design

the same innovator drug be provided.Bioe-

1164Shein-Chung Chow quivalence among generic drugs is not re-statistical difference and a clinically mean-quired.In practice,it is very likely that two

ingful difference in the assessment of the generic drugs may not be bioequivalent when effectiveness and safety of a drug product.

This inconsistent result often creates confu-both drugs are bioequivalent to the same in-

novator drug.To overcome this disadvan-sion and/or an argument among clinicians tage,it is suggested that population bioequi-

and statisticians.A significant difference valence be assessed in addition to average may be referred to as a statistically signifi-bioequivalence.To ensure drug prescribabil-

cant difference or a clinically significant dif-ity,Chow and Liu(11)proposed performing ference.A difference which is unlikely to

occur by chance alone is considered a statisti-a meta analysis for bioequivalence review

based on data from all submissions.The pro-cally significant difference.A statistically posed method provides a useful tool for mon-

significant difference may be relatively small itoring the performance of the generic drugs as compared to the treatment mean.A large approved by the FDA based on current regu-

difference may not be of statistical signifi-lation on average bioequivalence.cance if the sample size is too small.

A clinically significant difference is a dif-

Drug switchability is related to the switch

from a drug(eg,an innovator)to an alterna-ference that is considered clinically meaning-tive drug(eg,a generic drug)within the same

ful and important to the investigators.Basi-subject whose concentration of the drug has cally,there are four different outcomes for been titrated to a steady,efficacious,and safe

an observed difference in a clinical trial.The level.For drug switchability,the FDA recom-observed difference may show that: mends that the assessment of individual bioe-

quivalence be considered(12).Based on the 1.The difference is both statistically and concept of individual bioequivalence,several

clinically significant,

methods have been proposed.None of these 2.There is a statistically significant differ-

ence yet the difference is not clinically methods,however,seems to serve the pur-

pose of ensuring drug switchability due to the significant,

discrepancy from statistical and regulatory

3.The difference is of clinical significance point of views in terms of the concept,defini-yet it is not statistically significant,and tion,and criteria of individual bioequivalence

4.The difference is neither statistically sig-(13,14,15).nificant nor clinically significant.

Note that bioequivalence testing as a sur-

rogate for clinical evaluation is based on the If the difference is both clinically and sta-fundamental assumption that if the two drug

tistically significant or it is neither clinically products are bioequivalent(in terms of drug nor statistically significant,then there is no

confusion and a consistent conclusion can be absorption),then they are therapeutically

equivalent.Therefore,it is very important at drawn.In many cases,however,a statisti-least to assure that the two drugs have similar

cally significant difference does not agree drug absorption profiles and consequently with the clinically significant difference.For have a similar therapeutic effect.Thus,alter-

example,a statistical test may reveal that native regulatory criteria for bioequivalence there is a statistically significant difference.

The difference is too small(which may be should be developed in such a way that they

can address both drug prescribability and due to a unusual small variability or a rela-drug switchability.This task,however,pro-

tively large sample size),however,to be of vides a challenge not only to regulatory any clinical importance and hence it is not agents but also to biostatisticians.

clinically significant.In this case,a small p-

value may lead one to conclude the effective-Clinical Applications ness of the treatment.On the other hand,the In clinical trials,it is common to observe

result may indicate that there is a clinically an inconsistent result between a significant significant difference but the sample size

GSP in Drug Development and Regulatory Approval1165 may be too small(or variability is too large)decreasing improvement among T,A,and P; to claim a statistically significant difference.

however,T is superior to P and hence is In this case,the evidence of effectiveness is effective),or P>TA(ie,both T and A are

ineffective).

not substantial due to a large p-value.

For another example,consider the use of As a result,showing equivalence or supe-active control trials in clinical development.

riority may imply that T and A are both For approval of a new drug,conducting a equally effective or equally ineffective.This clinical trial comparing the new drug with a

statement is especially true for some drug control is required.Section314.126in Part products such as antianxiety agents,antide-

pressants,antianginal agents,or appetite 21of the CFR indicates that a control could

be a placebo control,no treatment control,suppressants which may not necessarily beat positive control,or historical control.In prac-

the placebo.For these drug products,the tice,since it may not be ethical to conduct a FDA prefers placebo control trials,while the placebo control study with very ill patients

EC resists placebo control trials.It is then with severe or life-threatening diseases to suggested that trials including the test drug,

an active control,and a placebo be used with establish efficacy of a new drug,a positive

control trials is often considered as an alter-a different ratio of patients assigned when-native to evaluate the effectiveness and safety

ever possible.

of the new drug by comparing it with a posi-The above two examples indicate that the tive control which has been shown to be ef-

communication of statistical concept be-fective and safe for the intended disease.A tween clinicians and statisticians is essential

for the implementation of GCP in clinical positive control trial is also known as an ac-

tive control trial.development for establishment of drug effi-As indicated by Pleager and Hall(16),the

cacy and safety.

primary objective of an active control trial

could be to:

CONCLUSIONS

1.Establish the efficacy of the test drug,

During the development and regulatory ap-2.Show that the test drug is equivalent to an proval process,GLP,GCP,GRP,and cGMP active control,or

are necessarily applied to:

3.Demonstrate that the test drug is superior

to the active control. 1.Assure the effectiveness and safety of the

drug under investigation before approval, The equivalence and superiority to the active

and

control,however,do not guarantee that the 2.Ensure that the drug product possesses

good drug characteristics such as proper test drug is effective.As an example,con-

sider the case where a test drug(denoted by identity,strength,quality,purity,and sta-T)is superior to an active control(denoted

bility in compliance with the standards as by A),denoted by T>A(ie,a statistically specified in the USP/NF after regulatory or clinically significant difference between

approval.

T and A is observed).In this case,it is possi-

ble that the actual outcome is T>A>P(ie,In essence,GSP is the foundation of GLP,

GCP,GRP,and cGMP.The implementation both T and A are effective),T>P>A(ie,

only T is effective),or P>T>A(ie,both A of GSP is a team project which involves stat-and B are ineffective),where P denotes the

isticians,pharmaceutical scientists,and regu-placebo.For another example,when T and latory agents as well.The success of GSP A are equivalent(denoted by TA),it may fall

depends upon mutual communication,confi-in one of the following possible outcomes:dence,respect,and cooperation among statis-

ticians,pharmaceutical scientists,and regu-TA>P(ie,both T and A are equally effec-

tive),TAP,T>AP(ie,there is no significant latory agents.

1166Shein-Chung Chow

Harmonization Guideline.International Conference REFERENCES

on Harmonization;1993.

7.Chow SC.Statistical design and analysis of stability https://www.wendangku.net/doc/42887717.html,P/NF.The United States Pharmacopeia XXIII

studies.Presented at the48th Conference on Applied and the National Formulary XVIII.Rockville,MD:Statistics,Atlantic City,New Jersey:1992.

United States Pharmacopeidal Convention,Inc.;

8.Chow SC,Liu JP.Statistical Design and Analysis

1995.in Pharmaceutical Science.New York,New York:

Marcel Dekker,Inc.;1995.

2.FDA.Guideline for Submitting Documentation for

the Stability of Human Drugs and Biologics.Rock-9.Nordbrock E.Statistical comparison of stability ville,Maryland:Center for Drugs and Biologics,

study designs.J Biopharm Stat.1992;2:91–113.

Food and Drug Administration;1987.10.Chow SC,Liu JP.Current issues in bioequivalence 3.FDA.Guideline for General Considerations for Clin-

trials.Drug Inf J.1995;29:795–804.

ical Evaluation of Drugs.Rockville,Maryland:Cen-11.Chow SC,Liu JP.Meta-analysis for bioequivalence ter for Drug Evaluation and Research,Food and

review.J Biopharm Stat.1997;7(1):97–111.

Drug Administration;1977.12.Chen ML.Individual bioequivalence—a regulatory 4.FDA.Guideline for the Format and Content of the

update.J Biopharm Stat.1997;7(1):5–11.

Clinical and Statistical Sections of an Application.13.Wellek S.A comment on so-called individual criteria Rockville,Maryland:Center for Drug Evaluation

of bioequivalence.J Biopharm Stat.1997;7:17–21.

and Research,Food and Drug Administration;1988.14.Endrenyi L.Some issues for the consideration of 5.FDA.Guidance on Statistical Procedures for Bioe-

individual bioequivalence.J Biopharm Stat.1997;

quivalence Studies Using a Standard Two-Treatment7:35–39.

15.Liu JP,Chow SC.Some thoughts on individual bioe-

Crossover Design.Rockville,Maryland:Division of

Bioequivalence,Office of Generic Drugs,Center for quivalence.J Biopharm Stat.1997;7:41–48.

Drug Evaluation and Research,Food and Drug Ad-

16.Pledger GW,Hall D.Active control trials:do they

ministration;1992.address the efficacy issue?Proceedings of the Bio-6.ICH.Stability Testing of New Drug Substances and

pharmaceutical Section of the American Statistical Products.Tripartite International Conference on Association;1986:1–7.