jbb-s1-002

8/13/2019 JBB-S1-002

1/8

Special Issue 1 2011J Bioequiv Availab

ISSN:0975-0851 JBB, an open access journal

Research Article Open Access

Chow,J Bioequiv Availab 2011, S1

http://dx.doi.org/10.4172/jbb.S1-002

Research Article Open Access

Bioequivalence & Bioavailability

Keywrds: Follow-on biologics; Biosimilars; Interchangeability;Switching; Alternating

Background

In the United States (US), or traditional chemical (small molecule)drug products, when an innovative (brand-name) drug product isgoing off patent, pharmaceutical and/or generic companies mayfile an abbreviated new drug application (ANDA) or approval ogeneric copies o the brand-name drug product. In 1984, the US Foodand Drug Administration (FDA) was authorized to approve genericdrug products under the Drug Price Competition and Patent Term

Restoration Act, which is known as the Hatch and Waxman Act. Forapproval o small molecule generic drug products, the FDA requiresthat evidence in averageo bioavailability in terms o the rate and extento drug absorption be provided. Te assessment o bioequivalence asa surrogate endpoint or quantitative evaluation o drug saety andefficacy is based on the Fundamental Bioequivalence Assumptionthat i two drug products are shown to be bioequivalent in averagebioavailability, it is assumed that they will reach the same therapeuticeffect or they are therapeutically equivalent and hence can be usedinterchangeably. Under the Fundamental Bioequivalence Assumption,regulatory requirements, study design, criteria, and statistical methodsor assessment o bioequivalence have been well established (see, e.g.,[1-7]).

Unlike small molecule drug products, the generic versions obiologic products are viewed similar biological drug products (SBDP).Te SBDP are notgeneric drug products, which are drug products withidenticalactive ingredient(s) as the innovative drug product. Tus, theconcept or development o SBDP, which are made o living cells, isvery different rom that o the generic drug products or small moleculedrug products. Te SBDP are usually reerred to as biosimilars byEuropean Medicines Agency (EMA) o European Union (EU), ollow-on biologics (FOB) by the US FDA, and subsequent entered biologics(SEB) by the Public Health Agency (PHA) o Canada. As a number obiologic products are due to expire in the next ew years, the subsequentproduction o ollow-on products has aroused interest within thepharmaceutical/biotechnology industry as biosimilar manuacturersstrive to obtain part o an already large and rapidly-growing market.

Te potential opportunity or price reductions versus the originatorbiologic products remains to be determined, as the advantage o aslightly cheaper price may be outweighed by the hypothetical increased

risk o side-effects rom biosimilar molecules that are not exact copieso their originators.

In this article, the ocus will not only be placed on the undamentaldifferences between small molecule drug products and biologic products,but also issues surrounding quantitative evaluation o bioequivalence(or small molecule drug products) and biosimilarity (or biosimilarsor ollow-on biologics). In the next section, undamental differencesbetween small molecule drug products and biologic drug products arebriefly described. Sections 3 and 4 provide brie descriptions o currentprocess or quantitative evaluation o bioequivalence and biosimilarity,

respectively. A general approach using biosimilarity index orassessment o bioequivalence and biosimilarity, which was derivedbased on the concept o reproducibility probability was proposed anddiscussed in Section 5. Section 6 summarizes some current scientificactors and practical issues regarding the assessment o biosimilarity.Brie concluding remarks are given in the last section o this article.

Fundamental Differences

Biosimilars or ollow-on biologics are undamentally differentrom those o traditional chemical generic drugs. Unlike traditionalchemical generic drug products which contain identical activeingredient(s), the generic versions o biologic products are madeo living cells. Unlike classical generics, biosimilars are not identicalto their originator products and thereore should not be brought to

market using the same procedure applied to generics. Tis is partly areflection o the complexities o manuacturing and saety and efficacycontrols o biosimilars when compared to their small molecule genericcounterparts (see, e.g., [8-11]).

*Corresponding author: Shein-Chung Chow, PhD, Professor, Department of

Biostatistics and Bioinformatics, Duke University School of Medicine, 2424 Erwin

Road, Hock Suite 1102, Room 11068, Durham, NC 27705, E-mail:sheinchung.

[email protected]

ReceivedAugust 12, 2011; AcceptedSeptember 06, 2011; PublishedSeptember

08, 2011

Citation:Chow SC (2011) Quantitative Evaluation of Bioequivalence/Biosimilarity.

J Bioequiv Availab S1. doi:10.4172/jbb.S1-002

Copyright: 2011 Chow SC. This is an open-access article distributed under theterms of the Creative Commons Attribution License, which permits unrestricted

use, distribution, and reproduction in any medium, provided the original author and

source are credited.

Abstract

As more biologic products are going off patent protection, the development of follow-on biologics (biosimilars)

products has received much attention from both biotechnology industry and the regulatory agencies. Unlike traditional

small-molecule (chemical) drug products, the development of biologic products is very different and variable with

respect to the manufacturing process and environment. The complexity and heterogeneity of the molecular structure,

complicated manufacturing process, different analytical methods, and possibility of severe immunogenicity reactions

make quantitative evaluation of follow-on biologics a great challenge to both scientic community and regulatory

agencies. In this article, an overview of current criteria, study design, and statistical methods for quantitative

evaluation of bioequivalence for the traditional small molecule generic drug productsand biosimilarity for biosimilars

products is provided. In addition, a general approach for development of a biosimilarity index based on the concept

of reproducibility probability for quantitative evaluation of bioequivalence/biosimilarity is proposed. Some scientic

factors and practical issues are also discussed.

Quantitative Evaluation of Bioequivalence/BiosimilarityShein-Chung Chow*

Biostatistics and Bioinformatics, Duke University School of Medicine Durham, North Carolina, USA
http://dx.doi.org/10.4172/jbb.S1-002http://dx.doi.org/10.4172/jbb.S1-002

8/13/2019 JBB-S1-002

2/8



Citation: Chow SC (2011) Quantitative Evaluation of Bioequivalence/Biosimilarity. J Bioequiv Availab S1. doi:10.4172/jbb.S1-002

Page 2 of 8

Some o the undamental differences between biosimilars andgeneric chemical drugs are summarized in able 1. For example,biosimilars are known to be variable and very sensitive to the

environmental conditions such as light and temperature. A smallvariation may translate to a drastic change in clinical outcomes(e.g., saety and efficacy). In addition to differences in the size andcomplexity o the active substance, important differences also includethe nature o the manuacturing process. Since biologic products areofen recombinant protein molecules manuactured in living cells,manuacturing processes or biologic products are highly complexand require hundreds o specific isolation and purification steps. Tus,in practice, it is impossible to produce an identicalcopy o a biologicproduct, as changes to the structure o the molecule can occur withchanges in the manuacturing process. Since a protein can be modifiedduring the process (e.g., a side chain may be added, the structuremay have changed due to protein misolding, and so on), differentmanuacturing processes may lead to structural differences in the finalproduct, which result in differences in efficacy and saety, and may havea negative impact on the immune responses o patients. It should benoted that these issues occur also during the post-approval changes othe innovators biological products.

Tus, SBDP are not generic products. Hence, the standard genericapproach is not applicable and acceptable due to the complexity obiological/biotechnology derived products. Instead, similar biologicalapproach depending upon the state-o-art o analytical proceduresshould be applied.

Quantitative Evaluation of Bioequivalence

For approval o small molecule generic drug products, the FDArequires that evidence o average bioequivalence in drug absorptionin terms o some pharmacokinetic (PK) parameters such as the areaunder the blood and/or plasma concentration-time curve (AUC)and peak concentration (C

max) be provided through the conduct o

bioequivalence studies. In practice, we may claim that a test drugproduct is bioequivalent to an innovative (reerence) drug product ithe 90% confidence interval or the ratio o geometric means o theprimary PK parameter is completely within the bioequivalence limits o(80%, 125%). Te confidence interval or the ratio o geometric meanso the primary PK parameter is obtained based on log-transormeddata. In what ollows, study designs and statistical methods that arecommonly considered in bioequivalence studies are briefly described.

Study design

As indicated in the Federal Register[Vol. 42, No. 5, Sec. 320.26(b)and Sec. 320.27(b), 1977], a bioavailability study (single-dose or multi-

dose) should be crossover in design, unless a parallel or other designis more appropriate or valid scientific reasons. Tus, in practice, astandard two-sequence, two-period (or 22) crossover design is ofen

considered or a bioavailability or bioequivalence study. Denote by and R the test product and the reerence product, respectively. Tus,a 22 crossover design can be expressed as (R, R), where R is thefirst sequence o treatments and R denotes the second sequence otreatments. Under the (R, R) design, qualified subjects who arerandomly assigned to sequence 1 (R) will receive the test product first and then cross-overed to receive the reerence product Rafer a sufficient length o wash-out period. Similarly, subjects whoare randomly assigned to sequence 2 (R) will receive the reerenceproduct (R) first and then receive the test product () afer a sufficientlength o wash-out period.

One o the limitations o the standard 22 crossover design is thatit does not provide independent estimates o intra-subject variabilities

since each subject will receive the same treatment only once. In theinterest o assessing intra-subject variabilities, the ollowing alternativehigher-order crossover designs or comparing two drug products areofen considered: (i) Balaams design, i.e., (, RR, R, R), (ii) two-sequence, three-period dual design, e.g., (RR,R), and (iii) our-sequence, our-period design, e.g., (RR, RR, RR, RR).

For comparing more than two drug products, a Williams designis ofen considered. For example, or comparing three drug products,a six-sequence, three-period (63) Williams design is usuallyconsidered, while a 44 Williams design is employed or comparing 4drug products. Williams design is a variance stabilizing design. Moreinormation regarding the construction and good design characteristicso Williams designs can be ound in Chow and Liu [7].

In addition to the assessment o average bioequivalence(ABE), there are other types o bioequivalence assessment such aspopulation bioequivalence (PBE) which is intended or addressingdrug prescibability and individual bioequivalence (IBE) which isintended or addressing drug switchability. For assessment IBE/PBE, the FDA recommends that a replicateddesign be considered orobtaining independent estimates o intra-subject and inter-subjectvariabilities and variability due to subject-by-drug product interaction.A commonly considered replicate crossover design is the replicate oa 22 crossover design is given by (RR, RR). In some cases, anincomplete block design or an extra-reerence design such as (RR,RR) may be considered depending upon the study objectives o thebioavailability/bioequivalence studies [12].

Statistical methods

As indicated earlier, ABE is claimed i the ratio o averagebioavailabilities between test and reerence products is within thebioequivalence limit o (80%, 125%) with 90% assurance based onlog-transormed data. Along this line, commonly employed statisticalmethods are the confidence interval approach and the method ointerval hypotheses testing. For the confidence interval approach,a 90% confidence interval or the ratio o means o the primarypharmacokinetic response such as AUC or C

max is obtained under an

analysis o variance model. We claim bioequivalence i the obtained90% confidence interval is totally within the bioequivalence limito (80%, 125%). For the method o interval hypotheses testing, theinterval hypotheses that

0 : Bioinequivalence vs. : BioequivalenceaH H (1)Note that the above hypotheses are usually decomposed into two

Chemical drugs Biologic drugs

Made by chemical synthesis Made by living cells

Dened structureHeterogeneous structure

Mixtures of related molecules

Easy to characterize Difcult to characterize

Relatively stable

No issue of immunogenicity

Variable

Sensitive to environmental conditions

such as light and temperature

Issue of immunogenicity

Usually taken orally Usually injected

Often prescribed by a generalpractitioner

Usually prescribed by specialists

Table 1: Fundamental Differences.

8/13/2019 JBB-S1-002

3/8

8/13/2019 JBB-S1-002

4/8




Page 4 of 8

aggregated criterion would provide a stronger evidence o biosimilaritydue to potential offset (or masked) effect between the average andvariability in the aggregated criterion. Further researchor establishing

the appropriate statistical testing procedures based on the aggregatecriterion and comparing its perormance with the disaggregatecriterion may be needed.

Chow et al. [14] compared the moment-based criterion withthe probability-based criterion or assessment o bioequivalence orbiosimilarity under a parallel group design. Te results indicate thatthe probability-based criterion is not only a much more stringentcriterion, but also has sensitivity to any small change in variability. Tisjustifies the use o the probability-based criterion or assessment obiosimilarity between ollow-on biologics i a certain level o precisionand reliability o biosimilarity is desired.

Study design: As indicated earlier, a crossover design is ofen

employed or bioequivalence assessment. In a crossover study,each drug product is administered to each subject. Tus, estimate(approximate) within-subject variance can be sued to address switchability and interchangeability. For a parallel-group study, each drugproduct is administered to a different group o subjects. Tus, we canonly estimate total variance (between and within subject variances) notindividual variance components. For ollow-on biologics with longhal-lives, crossover study would be ineffective and unethical. In thiscase, we need to under take study with parallel groups. However, aparallel-group study does not provide an estimate or within-subjectvariation (since there is no R vs. R).

Statistical methods: Similar to the assessment o averagebioequivalence, Shuirmanns two one-sided tests procedure or the

confidence interval are recommended or assessment o biosimilarityi similar criteria are adopted. On the other hand, i similar criteria orassessment o population/individual bioequivalence are considered,the 95% confidence upper bound can be used or assessing biosimilaritybased on linearized criteria o population/individual bioequivalence.

Interchangeability

As indicated in the Subsection (b)(3) amended to the Public HealthAct Subsection 351(k)(3), the terminterchangeableor interchangeabilityin reerence to a biological product that is shown to meet the standardsdescribed in subsection (k)(4), means that the biological product maybe substituted or the reerence product without the intervention o thehealth care provider who prescribed the reerence product. Along thisline, in what ollows, definition and basic concepts o interchangeability(in terms o switching and alternating) are given.

Definition and basic concepts: As indicated in the Subsection (a)(2) amends the Public Health Act Subsection 351(k)(3), a biologicalproduct is considered to be interchangeable with the reerence producti (i) the biological product is biosimilar to the reerence product; and(ii) it can be expected to produce the same clinical result in any given

patient. In addition, or a biological product that is administered morethan once to an individual, the risk in terms o saety or diminishedefficacy o alternating or switching between use o the biologicalproduct and the reerence product is not greater than the risk o usingthe reerence product without such alternation or switch.

Tus, there is a clear distinction between biosimilarity and

interchangeability. In other words, biosimilarity does not implyinterchangeability which is much more stringent. Intuitively, i a testproduct is judged to be interchangeable with the reerence product then

it may be substituted, even alternated, without a possible intervention,or even notification, o the health care provider. However, theInterchangeability is expected to produce the sameclinical result in any

given patient, which can be interpreted as that the same clinical resultcan be expected in every single patient. In reality, conceivably, lawsuitsmay be filed i adverse effects are recorded in a patient afer switchingrom one product to another.

It should be noted that when FDA declares the biosimilarity o twodrug products, it may not be assumed that they are interchangeable.Tereore, labels ought to state whether or a ollow-on biologic whichis biosimilar to a reerence product, interchangeability has or has notbeen established. However, payers and physicians may, in some cases,switch products even i interchangeability has not been established.

Switching and alternating: Unlike drug interchangeability (interms o prescribability and switchability [7], the US FDA has slightly

perception o drug interchangeability or biosimilars. From the FDAsperspectives, interchangeability includes the concept o switching andalternating between an innovative biologic product (R) and its ollow-on biologics (). Te concept o switching is reerred to as not only theswitch rom R to or to R (narrow sense o switchability), butalso to and R to R (broader sense o switchability). As a result,in order to assess switching, biosimilarity or R to , to R, to , and R to R need to be assessed based on some biosimilaritycriteria under a valid study design.

On the other hand, the concept o alternating is reerred to as eitherthe switch rom to R and then switch back to (i.e., to R to )or the switch rom R to and then switch back to R (i.e., R to to R.Tus, the difference between the switch rom to R or the switch

rom R to and the switch rom R to or the switch rom to Rneeds to be assessed or addressing the concept o alternating.

Study design: For assessment o bioequivalence or chemical drugproducts, a standard two-sequence, two-period (22) crossover designis ofen considered, except or drug products with relatively long hal-lives. Since most biosimilar products have relatively long hal-lives, it issuggested that a parallel group design should be considered. However,parallel group design does not provide independent estimates ovariance components such as inter- and intra-subject variabilities andvariability due to subject-by-product interaction. Tus, it is a majorchallenge or assessing biosimilars under parallel group designs.

In order to assess biosimilarity or R to , to R, to , and

R to R, the Balaams 42 crossover design, i.e., (, RR, R, R) maybe useul. For addressing the concept o alternating, a two-sequence,three-period dual design, i.e., (R, RR) may be useul. For addressingboth concepts o switching and alternating or drug interchangeabilityo biosimilars, a modified Balaams crossover design, i.e., (, RR,R, RR) is then recommended.

Remarks: With small molecule drug products, bioequivalencegenerally reflects therapeutic equivalence. Drug prescribability,switching, and alternating are generally considered reasonable.With biologic products, however, variations are ofen higher (otherthan pharmacokinetic actors may be sensitive to small changesin conditions). Tus, ofen only parallel-group design rather thancrossover kinetic studies can be perormed. It should be noted that

very ofen, with ollow-on biologics, biosimilarity does not reflecttherapeutic comparability. Tereore, switching and alternating shouldbe pursued only with substantial caution.

8/13/2019 JBB-S1-002

5/8




Page 5 of 8

A General Approach for Assessment of Bioequivalence/Biosimilarity

As indicated earlier, the concept o biosimilarity andinterchangeability or ollow-on biologics is very different rom thato bioequivalence and drug interchangeability or small molecule drugproducts. It is debatable whether standard methods or assessmento bioequivalence and drug interchangeability can be applied toassessing biosimilarity and interchangeability o ollow-on biologicsdue to the undamental differences as described in Section 2. Whileappropriate criteria or standards or assessment o biosimilarity andinterchangeability are still under discussion within the regulatoryagencies and among the pharmaceutical industry and academia, wewould like to propose the a general approach or assessing biosimilarityand interchangeability by comparing the relative difference between atest product vs. a reference productand the reference vs. the referencebased on the concept o reproducibility probability o claimingbiosimilairty between a test product and a reerence product in a uturebiosimilarity study provided that the biosimilarity between the testproduct and the reerence product has been established in the currentstudy.

Development of biosimilarity index

Shao and Chow [15] proposed a reproducibility probability asan index or determining whether it is necessary to require a secondtrial when the result o the first clinical trial is strongly significant.Suppose that the null hypothesis H

0 is rejected i and only i ||>c,

where c is a positive known constant and is a test statistic. Tus, thereproducibility probability o observing a significant clinical resultwhen H

ais indeed true is given by

( ) | (| | | ),ap P T c H P T c= > = > (3)

where

is an estimate o, which is an unknown parameter or vector oparameters. Following the similar idea, a reproducibility probability canalso be used to evaluate biosimilarity and interchangeability between atest product and a reerence product based on any pre-specified criteriaor biosimilarity and interchangeability. As an example, biosimilarityindex proposed by Chow et al. [16] is illustrated based on the well-established bioequivalence criterion by the ollowing steps:

Step 1. Assess the average biosimilarity between the test productand the reerence product based on a given biosimilaritycriterion. For illustration purpose, consider bioequivalencecriterion as biosimilarity criterion. Tat is, biosimilarity is

claimed i the 90% confidence interval o the ratio o meanso a given study endpoint alls within the biosimilarity limito (80%, 125%) based on log-transormed data.

Step 2. Once the product passes the test or biosimilarity inStep 1, calculate the reproducibility probability basedon the observed ratio (or observed mean difference) andvariability. We will reer to the calculated reproducibilityprobability as the biosimilarity index.

Step 3. We then claim biosimilarity i the ollowing null hypothesisis rejected:

H0: P p

0vs. H

a: P>p

0. (4)

A confidence interval approach can be similarly applied. In otherwords, we claim biosimilarity i the lower 95% confidence bound o thereproducibility probability is larger than a pre-specified numberp

0. In

practice, p0

can be obtained based on an estimated o reproducibilityprobability or a study comparing a reerence product to itsel (thereerence product). We will reer to such a study as an R-R study.

In an R-R study, define

concluding average biosimiliarity between the test and the

reference products in a future trial given that the average

biosimiliarity based on ABE criterion has been established

in first trial

TRP P

=

(5)

Alternatively, a reproducibility probability or evaluating thebiosimilarity o the two same reerence products based on ABEcriterion is defined as:

concluding average biosimiliarity of the two same reference

products in a future trial given that the average biosimilarity

based on ABE criterion have been established in first trialRRP P

=

(6)

Since the idea o the biosimilarity index is to show that thereproducibility probability in a study or comparing ollow-on biologicwith the innovative (reerence) product is higher than a reerenceproduct with the reerence product. Te criterion o an acceptablereproducibility probability (i.e., p

0) or assessment o biosimilarity

can be obtained based on the R-R study. For example, i the R-R studysuggests the reproducibility probability o 90%, i.e., P

RR = 90%, the

criterion o the reproducibility probability or bioequivalence studycould be chosen as 80% o the 90% which isp

0=80% P

RR= 72%.

Te above described biosimilarity index has the advantages that (i)it is robust with respect to the selected study endpoint, biosimilaritycriteria, and study design, (ii) it takes variability into consideration (oneo the major criticisms in the assessment o average bioequivalence),

(iii) it allows the definition and assessment the degree o similarity(in other words, it provides partial answer to the question that howsimilar is considered similar? and (iv) the use o biosimilarity indexwill reflect the sensitivity o heterogeneity in variance.

Most importantly, the biosimilarity index proposed by Chowet al. [16] can be applied to different unctional areas (domains) obiological products such as good drug characteristics such as saety(e.g., immunogenicity), purity, and potency (as described in BPCIAct), pharmacokinetics (PK) and pharmacodynamics (PD), biologicalactivities, biomarkers (e.g., genomic markers), and manuacturingprocess, etc. or an assessment o global biosimilarity. An overallbiosimilar index across domains can be obtained by the ollowing steps:

Step 1. Obtain Pi

, the probability o reproducibility or the i-thdomain,i=1,.., K.

Step 2. Define the global biosimilarityindex1

K

i iiP w P

=

= , wherei

w is the weight or the i-th domain.

Step 3. Claim global biosimilarity i the lower 95% confidencebound o the reproducibility probability (P) is larger thana pre-specified number P

0, where P

0 is a pre-specified

acceptable reproducibility probability.

Remarks

Hsieh et al. [17] studied the perormance o the biosimilarityindex under a R-R study or establishing a baseline or assessment obiosimilarity based on current criterion or average bioequivalence.

Te results indicate that biosimilarity index is sensitive to thevariability associated with the reerence product. Te biosimilarityindex decreases as the variability increases. As an example, Figure 1

8/13/2019 JBB-S1-002

6/8




Page 6 of 8

gives reproducibility probability curves under a 22 crossover design

with sample sizes n1= n

2= 10, 20, 30, 40, 50, and 60 at the 0.05 level o

significance and (L

,U

) = (80%, 125%) when d

= 0.2 and 0.3, where d

is

the standard deviation o period difference within each subject.

In practice, alternative approaches or assessment o the proposed

biosimilarity index are available (see, e.g., [17,18]). Te methods

include maximum likelihood approach and Bayesian approach. For

the Bayesian approach, letp() be the power unction, where is an

unknown parameter or vector o parameters. Under this Bayesian

approach,is random with a prior distribution assumed to be known.

Te reproducibility probability can be viewed as the posterior mean o

the power unction or the uture trial

( ) ( | )p x d , (7)where (|x)

is the posterior density o , given the data set xobserved

or the previous trial (s). However, there may exist no explicit ormor the estimation o the biosimilarity index. As a result, statisticalproperties o the derived biosimilarity index may not be known. In thiscase, the finite sample size perormance o the derived biosimilarity

index may only be evaluated by clinical trial simulations.

As an alternative measure or assessment o global biosimilarity

across domains, we may consider1 ,

K

i i ird w rd

== where TRii

RRi

Prd

P=

which is the relative measure o biosimilarity between and R ascompared to that o between R and R. Based on rd

i, i=1, , K, we may

conduct a profile analysisas described in the 2003 FDA guidance onBioavailability and Bioequivalence Studies for Nasal Aerosols and NasalSprays for Local Action[5]. However, statistical properties o the profile

analysis based on rdi, i=1, , K are not ully studied. Tus, urtherresearch is required.

Scientific Factors and Practical Issues

Following the passage o the BPCI Act, in order to obtain input

on specific issues and challenges associated with the implementationo the BPCI Act, the US FDA conducted a two-day public hearing

onApproval Pathway for Biosimilar and Interchangeability Biological

Products held on November 2-3, 2010 at the FDA in Silver Spring,

Maryland, USA. In what ollows, some o the scientific actors and

practical issues are briefly described.

Fundamental biosimilarity assumption

Similar to Fundamental Bioequivalence Assumption or

assessment o bioequivalence, Chow et al. [14] proposed the ollowing

Fundamental Biosimilarity Assumption or ollow-on biologics:

When a biosimilar product is claimed to be biosimilar to an

innovators product based on some well-defined product characteristics

and is therapeutically equivalent provided that the well-defined product

characteristics are validated and reliable predictors of safety and efficacy

of the products.

For the chemical generic products, the well-defined product

characteristics are the exposure measures or early, peak, and

total portions o the concentration-time curve. Te Fundamental

Bioequivalence Assumption is assumed that the equivalence in the

exposure measures implies therapeutically equivalent. However, due to

the complexity o the biosimilar drug products, one has to veriy that

some validated product characteristics are indeed reliable predictors

o the saety and efficacy. It ollows that the design and analysis or

evaluation o equivalence between the biosimilar drug product and

innovator products are substantially different rom those o thechemical generic products.

Note: The reproducibility probability decreases when 1/

2(original scale) moves away from 1 and

d(log scale) is larger.

Figure 1: Impact of Variability on Reproducibility Probability.

8/13/2019 JBB-S1-002

7/8




Page 7 of 8

Endpoint selection

For assessment o biosimilarity o ollow-on biologics, the ollowing

questions are commonly asked. First, what endpoints should be usedor assessment o biosimilarity? Second, should a clinical trial alwaysbe conducted?

o address these two questions, we may revisit the definition obiosimilarity as described in the BPCI Act. A biological product thatis demonstrated to be highly similar to an FDA-licensed biologicalproduct may rely on certain existing scientific knowledge about saety,purity (quality), and potency (efficacy) o the reerence product. Tus,i one would like to show that the saety and efficacy o a biosimilarproduct are highly similar to that o the reerence product, thena clinical trial may be required. In some cases, clinical trials orassessment o biosimilarity may be waived i there exists substantialevidence that surrogate endpoints or biomarkers are predictive o the

clinical outcomes. On the other hand, clinical trials are required orassessment o drug interchangeability in order to show that the saetyand efficacy between a biosimilar product and a reerence product aresimilar in any given patient o the patient population under study.

How similar is similar?

Current criteria or assessment o bioequivalence/biosimilarityis useul or determining whether a biosimilar product is similar to areerence product. However, it does not provide additional inormationregarding the degree o similarity. As indicated in the BPCI Act,abiosimilar product is defined as a product that is highly similar tothe reerence product. However, little or no discussion regarding thedegree o similarity or highly similar was provided. Besides, it is also o

concern to the sponsor that what i a biosimilar product turns out tobe superior to the reerence product?. A simple answer to the concernis that superiority is not biosimilarity.

Practical issues

Since there are many critical (quality) attributes o a potentialpatients response in ollow-on biologics, or a given critical attribute,valid statistical methods are necessarily developed under a validstudy design and a given set o criteria or similarity, as described inthe previous section. Several areas can be identified or developingappropriate statistical methodologies or the assessment o biosimilarityo ollow-on biologics. Tese areas include, but are not limited to:

Criteria for biosimilarity (in terms of average, variability, or

distribution):o address the question that how similar is similar?,we suggest establishing criteria or biosimilarity in terms o average,variability, and/or distribution.

Criteria for interchangeability: In practice, it is recognized thatdrug interchangeability is related to the variability due to subject-by-drug interaction. However, it is not clear whether criterion orinterchangeability should be based on the variability due to subject-by-drug interaction or the variability due to subject-by-drug interactionadjusted or intra-subject variability o the reerence drug.

Bridging studies for assessing biosimilarity:As most biosimilarsstudies are conducted using a parallel design rather than a replicatedcrossover design, independent estimates o variance components

such as the intra-subject and the variability due to subject-by-druginteraction are not possible. In this case, bridging studies may beconsidered.

Other practical issues include (i) the use o a percentile method orthe assessment o variability, (ii) comparability in biologic activities,(iii) assessment o immunogenicity, (iv) consistency in manuacturing

processes (see, e.g., [19-21]), (v) stability testing or multiple lots and/or multiple labs (see, e.g., [19]), (vi) the potential use o sequentialtesting procedures and multiple testing procedures, (vii) assessingbiosimilarity using a surrogate endpoint or biomarker such as genomicdata (see, e.g., [22]).

Further research is needed in order to address the above mentionedscientific actors and practical issues recognized at the FDA PublicHearing.

Concluding Remarks

As indicated earlier, we claim that a test drug product isbioequivalent to a reerence (innovative) drug product i the 90%

confidence interval or the ratio o means o the primary PK parameteris totally within the bioequivalence limits o (80%, 125%). Tis onesize-fits-all criterion only ocuses on average bioavailability and ignoresheterogeneity o variability. Tus, it is not scientifically/statisticallyjustifiable or assessment o biosimilarity o ollow-on biologics. Inpractice, it is then suggested that appropriate criteria, which can takethe heterogeneity o variability into consideration be developed sincebiosimilars are known to be variable and sensitive to small variations inenvironmental conditions [14,23,24].

At the FDA public hearing, questions that are commonly asked areHow similar is considered similar? andHow the degree of similarityshould be measured and translated to clinical outcomes (e.g., safety andefficacy)? Tese questions closely related to drug interchangeability

o biosimilars or ollow-on biologics which have been shown to bebiosimilar to the innovative product [11,25].

For assessment o bioequivalence or chemical drug products,a crossover design is ofen considered, except or drug productswith relatively long hal-lives. Since most biosimilar products haverelatively long hal-lives, it is suggested that a parallel group designshould be considered. However, parallel group design does not provideindependent estimates o variance components such as inter- andintra-subject variabilities and variability due to subject-by-productinteraction. Tus, it is a major challenge or assessing biosimilars underparallel group designs.

Although EMA o EU has published several product-specific

guidances based on the concept papers (e.g., [26-34]), it has beencriticized that there are no objective standards or assessment obiosimilars because it depends upon the nature o the products.Product-specific standards seem to suggest that aflexible biosimilaritycriterion should be considered and the flexible criterion should beadjusted or variability and/or the therapeutic index o the innovative(or reerence) product.

As described above, there are many uncertainties or assessmento biosimilarity and interchangeability o biosimilars. As a result, it is amajor challenge to both clinical scientists and biostatisticians to developvalid and robust clinical/statistical methodologies or assessmento biosimilarity and interchangeability under the uncertainties. Inaddition, how to address the issues o quality and comparability in

manuacturing process is another challenge to both the pharmaceuticalscientists and biostatisticians. Te proposed general approach using thebiosimilarity index (derived based on the concept o reproducibility

8/13/2019 JBB-S1-002

8/8




Page 8 of 8

probability) may be useul. However, urther research on the statisticalproperties o the proposed biosimilarity index is required.

References

1. Schuirmann DJ (1987) A comparison of the two one-sided tests procedure and

the power approach for assessing the equivalence of average bioavailability. J

Pharmacokinet Biopharm 15: 657-680.

2. EMEA (2001) Note for Guidance on the Investigation of Bioavailability and

Bioequivalence. The European Medicines Agency Evaluation of Medicines for

Human Use. EMEA/EWP/QWP/1401/98, London, United Kingdom.

3. FDA (2001) Guidance on Statistical Approaches to Establishing

Bioequivalence. Center for Drug Evaluation and Research, the US Food and

Drug Administration, Rockville, Maryland, USA.

4. FDA (2003) Guidance on Bioavailability and Bioequivalence Studies for Orally

Administrated Drug Products General Considerations, Center for Drug

Evaluation and Research, the US Food and Drug Administration, Rockville,

Maryland, USA.

5. FDA (2003).Guidance on Bioavailability and Bioequivalence Studies for Nasal

Aerosols and Nasal Sprays for Local Action, Center for Drug Evaluation andResearch, the US Food and Drug Administration, Rockville, Maryland, USA.

6. WHO (2005) World Health Organization Draft Revision on Multisource

(Generic) Pharmaceutical Products: Guidelines on Registration Requirements

to Establish Interchangeability, Geneva, Switzerland.

7. Chow SC, Liu JP (2008) Design and Analysis of Bioavailability and

Bioequivalence Studies. 3rd edition, Chapman Hall/CRC Press, Taylor &

Francis, New York, New York, USA.

8. Schellekens H (2004) How similar do biosimilar need to be? Nat Biotechnol

22: 1357-1359.

9. Chirino AJ, Mire-Sluis A (2004) Characterizing biological products and

assessing comparability following manufacturing changes. Nat Biotechnol 22:

1383-1391.

10.Crommelin D, Bermejo T, Bissig M, Damianns J, Kramer I, et al. (2005)

Biosimilars, generic versions of the rst generation of therapeutic proteins: dothey exist? Contrib Nephrol 149: 287-294.

11. Roger SD, Mikhail A (2007) Biosimilars: opportunity or cause for concern? J

Pharm Sci 10: 405-410.

12. Chow SC, Shao J, Wang H (2002) Individual bioequivalence testing under 2x3

crossover designs. Stat Med 21: 629-648.

13.Haidar SH, Davit B, Chen ML, Conner D, Lee L, et al. (2008) Bioequivalence

approaches for highly variable drugs and drug products. Pharm Res 25: 237-

241.

14.Chow SC, Hsieh TC, Chi E, Yang J (2010) A comparison of moment-based

and probability-based criteria for assessment of follow-on biologics. J Biopharm

Stat 20: 31-45.

15.Shao J, Chow SC (2002) Reproducibility probability in clinical trials. Stat Med

21: 1727-1742.

16. Chow SC, Endrenyi L, Lachenbruch PA, Yang LY, Chi E (2011) Scienticfactors for assessing biosimilarity and drug Interchangeability of follow-on

Biologics. Biosimilars 1: 1-14.

17. Hsieh TC, Chow SC, Yang LY, Chi E (2011) Statistical test for evaluation of

biosimilarity based on reproducibility probability. Submitted for publication

consideration.

18. Yang LY, Chow SC, Hsieh TC, Chi E (2011) Bayesian approach for assessment

of biosimilarity based on reproducibility probability. Submitted.

19. ICH (1996) Q5C Guideline on Quality of Biotechnological Products: Stability

Testing of Biotechnological/Biological Products. Center for Drug Evaluation

and Research, Center for Biologics Evaluation and Research, the US Food and


20. ICH (1999) Q6B Guideline on Test Procedures and Acceptance Criteria

for Biotechnological/Biological Products. Center for Drug Evaluation and

Research, Center for Biologics Evaluation and Research, the US Food and


21. ICH (2005) Q5E Guideline on Comparability of Biotechnological/Biological

Products Subject to Changes in Their Manufacturing Process. Center for Drug

Evaluation and Research, Center for Biologics Evaluation and Research, the

US Food and Drug Administration, Rockville, Maryland, USA.

22. Chow SC, Shao J, Li L (2004) Assessing bioequivalence using genomic data.

J Biopharm Stat 14: 869-880.

23. Chow SC, Liu JP (2010) Statistical assessment of biosimilar products. J

Biopharm Stat 20: 10-30.

24. Hsieh TC, Chow SC, Liu JP, Hsiao CF, Chi E (2010) Statistical test for

evaluation of biosimilarity of follow-on biologics. J Biopharm Stat 20: 75-89.

25. Roger SD (2006) Biosimilars: How similar or dissimilar are they? Nephrology

11: 341-346.

26. EMEA (2003a) Note for Guidance on Comparability of Medicinal Products

Containing Biotechnology-derived Proteins as Drug Substance Non Clinical

and Clinical Issues. The European Medicines Agency Evaluation of Medicines

for Human Use. EMEA/CHMP/3097/02, London, United Kingdom.

27. EMEA (2003b) Rev. 1 Guideline on Comparability of Medicinal Products

Containing Biotechnology-derived Proteins as Drug Substance Quality

Issues. The European Medicines Agency Evaluation of Medicines for Human

Use. EMEA/CHMP/ BWP/3207/00/Rev 1, London, United Kingdom.28. EMEA (2005a) Guideline on Similar Biological Medicinal Products. The

European Medicines Agency Evaluation of Medicines for Human Use. EMEA/

CHMP/437/04, London, United Kingdom.

29. EMEA (2005b) Draft Guideline on Similar Biological Medicinal Products

Containing Biotechnology-derived Proteins as Drug Substance: Quality Issues.

The European Medicines Agency Evaluation of Medicines for Human Use.

EMEA/CHMP/49348/05, London, United Kingdom.

30. EMEA (2005c) Draft Annex Guideline on Similar Biological Medicinal Products


and Clinical Issues Guidance on Biosimilar Medicinal Products containing

Recombinant Erythropoietins. The European Medicines Agency Evaluation of

Medicines for Human Use. EMEA/CHMP/94526/05, London, United Kingdom.

31. EMEA (2005d) Draft Annex Guideline on Similar Biological Medicinal Products


and Clinical Issues Guidance on Biosimilar Medicinal Products containingRecombinant Granulocyte-Colony Stimulating Factor. The European Medicines

Agency Evaluation of Medicines for Human Use. EMEA/CHMP /31329 /05,

London, United Kingdom.

32. EMEA (2005e) Draft Annex Guideline on Similar Biological Medicinal Products

Containing Biotechnology-derived Proteins as Drug Substance Non-Clinical


Somatropin. The European Medicines Agency Evaluation of Medicines for

Human Use. EMEA/CHMP/94528/05, London, United Kingdom.

33. EMEA (2005f) Draft Annex Guideline on Similar Biological Medicinal Products



Recombinant Human Insulin. The European Medicines Agency Evaluation of


34. EMEA (2005g) Guideline on the Clinical Investigating of the Pharmacokinetics

of Therapeutic proteins. The European Medicines Agency Evaluation of


Submit your next manuscript and get advantages of OMICS

Group submissions

Unique features:

User friendly/feasible website-translation of your paper to 50 worlds leading languages

Audio Version of published paper

Digital articles to share and explore

Special features:

100 Open Access Journals

10,000 editorial team

21 days rapid review process

Quality and quick editorial, review and publication processing

Indexing at PubMed (partial), Scopus, DOAJ, EBSCO, Index Copernicus and Google Scholar etc

Sharing Option: Social Networking Enabled

Authors, Reviewers and Editors rewarded with online Scientic Credits

Better discount for your subsequent articles

Submit your manuscript at: http://www.editorialmanager.com/jbiobio
http://dx.doi.org/10.4172/jbb.S1-002http://www.ncbi.nlm.nih.gov/pubmed/3450848http://www.ncbi.nlm.nih.gov/pubmed/3450848http://www.ncbi.nlm.nih.gov/pubmed/3450848http://pubs.amstat.org/doi/pdf/10.1198/jasa.2003.s254http://pubs.amstat.org/doi/pdf/10.1198/jasa.2003.s254http://pubs.amstat.org/doi/pdf/10.1198/jasa.2003.s254http://pubs.amstat.org/doi/pdf/10.1198/jasa.2003.s254http://pubs.amstat.org/doi/pdf/10.1198/jasa.2003.s254http://www.nature.com/nbt/journal/v22/n11/box/nbt1104-1357_BX2.htmlhttp://www.nature.com/nbt/journal/v22/n11/box/nbt1104-1357_BX2.htmlhttp://www.nature.com/nbt/journal/v22/n11/abs/nbt1030.html?lang=enhttp://www.nature.com/nbt/journal/v22/n11/abs/nbt1030.html?lang=enhttp://www.nature.com/nbt/journal/v22/n11/abs/nbt1030.html?lang=enhttp://books.google.co.in/books?hl=en&lr=&id=h0qCtQWLCpcC&oi=fnd&pg=PA287&dq=Biosimilars,+generic+versions+of+the+first+generation+of+therapeutic+proteins:+do+they+exist%3F&ots=YNYdk3Qklm&sig=JYFRJAAAaKM2Qkx6EqtgpX25qo8http://books.google.co.in/books?hl=en&lr=&id=h0qCtQWLCpcC&oi=fnd&pg=PA287&dq=Biosimilars,+generic+versions+of+the+first+generation+of+therapeutic+proteins:+do+they+exist%3F&ots=YNYdk3Qklm&sig=JYFRJAAAaKM2Qkx6EqtgpX25qo8http://books.google.co.in/books?hl=en&lr=&id=h0qCtQWLCpcC&oi=fnd&pg=PA287&dq=Biosimilars,+generic+versions+of+the+first+generation+of+therapeutic+proteins:+do+they+exist%3F&ots=YNYdk3Qklm&sig=JYFRJAAAaKM2Qkx6EqtgpX25qo8http://www.ualberta.ca/~csps/JPPS10_3/ReviewArticle_1308/R_1380.htmlhttp://www.ualberta.ca/~csps/JPPS10_3/ReviewArticle_1308/R_1380.htmlhttp://www.ncbi.nlm.nih.gov/pubmed/17891552http://www.ncbi.nlm.nih.gov/pubmed/17891552http://www.ncbi.nlm.nih.gov/pubmed/17891552http://www.ncbi.nlm.nih.gov/pubmed/20077247http://www.ncbi.nlm.nih.gov/pubmed/20077247http://www.ncbi.nlm.nih.gov/pubmed/20077247http://onlinelibrary.wiley.com/doi/10.1002/sim.1177/abstracthttp://onlinelibrary.wiley.com/doi/10.1002/sim.1177/abstracthttp://www.ncbi.nlm.nih.gov/pubmed/15587969http://www.ncbi.nlm.nih.gov/pubmed/15587969http://localhost/var/www/apps/conversion/tmp/scratch_10/Statistical%20assessment%20of%20biosimilar%20productshttp://localhost/var/www/apps/conversion/tmp/scratch_10/Statistical%20assessment%20of%20biosimilar%20productshttp://www.ncbi.nlm.nih.gov/pubmed/20077250http://www.ncbi.nlm.nih.gov/pubmed/20077250http://www.ncbi.nlm.nih.gov/pubmed/16889575http://www.ncbi.nlm.nih.gov/pubmed/16889575http://www.ncbi.nlm.nih.gov/pubmed/16889575http://www.ncbi.nlm.nih.gov/pubmed/16889575http://www.ncbi.nlm.nih.gov/pubmed/20077250http://www.ncbi.nlm.nih.gov/pubmed/20077250http://localhost/var/www/apps/conversion/tmp/scratch_10/Statistical%20assessment%20of%20biosimilar%20productshttp://localhost/var/www/apps/conversion/tmp/scratch_10/Statistical%20assessment%20of%20biosimilar%20productshttp://www.ncbi.nlm.nih.gov/pubmed/15587969http://www.ncbi.nlm.nih.gov/pubmed/15587969http://onlinelibrary.wiley.com/doi/10.1002/sim.1177/abstracthttp://onlinelibrary.wiley.com/doi/10.1002/sim.1177/abstracthttp://www.ncbi.nlm.nih.gov/pubmed/20077247http://www.ncbi.nlm.nih.gov/pubmed/20077247http://www.ncbi.nlm.nih.gov/pubmed/20077247http://www.ncbi.nlm.nih.gov/pubmed/17891552http://www.ncbi.nlm.nih.gov/pubmed/17891552http://www.ncbi.nlm.nih.gov/pubmed/17891552http://www.ualberta.ca/~csps/JPPS10_3/ReviewArticle_1308/R_1380.htmlhttp://www.ualberta.ca/~csps/JPPS10_3/ReviewArticle_1308/R_1380.htmlhttp://books.google.co.in/books?hl=en&lr=&id=h0qCtQWLCpcC&oi=fnd&pg=PA287&dq=Biosimilars,+generic+versions+of+the+first+generation+of+therapeutic+proteins:+do+they+exist%3F&ots=YNYdk3Qklm&sig=JYFRJAAAaKM2Qkx6EqtgpX25qo8http://books.google.co.in/books?hl=en&lr=&id=h0qCtQWLCpcC&oi=fnd&pg=PA287&dq=Biosimilars,+generic+versions+of+the+first+generation+of+therapeutic+proteins:+do+they+exist%3F&ots=YNYdk3Qklm&sig=JYFRJAAAaKM2Qkx6EqtgpX25qo8http://books.google.co.in/books?hl=en&lr=&id=h0qCtQWLCpcC&oi=fnd&pg=PA287&dq=Biosimilars,+generic+versions+of+the+first+generation+of+therapeutic+proteins:+do+they+exist%3F&ots=YNYdk3Qklm&sig=JYFRJAAAaKM2Qkx6EqtgpX25qo8http://www.nature.com/nbt/journal/v22/n11/abs/nbt1030.html?lang=enhttp://www.nature.com/nbt/journal/v22/n11/abs/nbt1030.html?lang=enhttp://www.nature.com/nbt/journal/v22/n11/abs/nbt1030.html?lang=enhttp://www.nature.com/nbt/journal/v22/n11/box/nbt1104-1357_BX2.htmlhttp://www.nature.com/nbt/journal/v22/n11/box/nbt1104-1357_BX2.htmlhttp://pubs.amstat.org/doi/pdf/10.1198/jasa.2003.s254http://pubs.amstat.org/doi/pdf/10.1198/jasa.2003.s254http://pubs.amstat.org/doi/pdf/10.1198/jasa.2003.s254http://www.ncbi.nlm.nih.gov/pubmed/3450848http://www.ncbi.nlm.nih.gov/pubmed/3450848http://www.ncbi.nlm.nih.gov/pubmed/3450848http://dx.doi.org/10.4172/jbb.S1-002

jbb-s1-002

Documents