ArticlePDF Available

A Comparative Analysis of Biopharmaceutics Classification System and Biopharmaceutics Drug Disposition Classification System: A Cross-Sectional Survey with 500 Bioequivalence Studies


Abstract and Figures

Although policies of waiving bioequivalence studies are part of the legal framework of various regulatory agencies, there is no harmonization with regard to extension of the biowaiver to drugs other than those with high solubility and high permeability, nor is there any consensus or official endorsement of the biopharmaceutics drug disposition classification system (BDDCS). To better understand the applicability of the biowaiver, we carried out a cross-sectional survey to estimate the relative risk of obtaining nonbioequivalent (non-BE) or bioinequivalent (BIE) results for drug products containing drugs belonging to each of the biopharmaceutics classification system (BCS) and BDDCS classes. Five hundred bioequivalence studies were randomly sampled from a database of the Brazilian Health Surveillance Agency (ANVISA). The drugs were classified according to the BCS and BDDCS, to evaluate how characteristics related to drug and dosage form influence the outcome of bioequivalence studies. The relative risk of obtaining a non-BE result was approximately four times lower for drugs in classes 1 and 3 of BCS or BDDCS when compared with class 2 drugs. Thus, it seems that the final outcome of a bioequivalence study is strongly influenced by the solubility of the drug, but not by its intestinal permeability or extent of metabolism. © 2013 Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci.
Content may be subject to copyright.
A Comparative Analysis of Biopharmaceutics Classification
System and Biopharmaceutics Drug Disposition Classification
System: A Cross-Sectional Survey with 500 Bioequivalence
1Division of Bioequivalence, Brazilian Health Surveillance Agency (ANVISA), Bras´
ılia, Brazil
2Institute of Mathematics and Statistics, University of S˜
ao Paulo, S˜
ao Paulo, Brazil
3Institute of Pharmaceutical Technology, Goethe University, Frankfurt am Main, Germany
4Faculty of Pharmaceutical Sciences, University of S˜
ao Paulo, S˜
ao Paulo, Brazil
Received 13 December 2012; accepted 4 March 2013
Published online 11 April 2013 in Wiley Online Library ( DOI 10.1002/jps.23515
ABSTRACT: Although policies of waiving bioequivalence studies are part of the legal frame-
work of various regulatory agencies, there is no harmonization with regard to extension of the
biowaiver to drugs other than those with high solubility and high permeability, nor is there
any consensus or official endorsement of the biopharmaceutics drug disposition classification
system (BDDCS). To better understand the applicability of the biowaiver, we carried out a
cross-sectional survey to estimate the relative risk of obtaining nonbioequivalent (non-BE) or
bioinequivalent (BIE) results for drug products containing drugs belonging to each of the bio-
pharmaceutics classification system (BCS) and BDDCS classes. Five hundred bioequivalence
studies were randomly sampled from a database of the Brazilian Health Surveillance Agency
(ANVISA). The drugs were classified according to the BCS and BDDCS, to evaluate how char-
acteristics related to drug and dosage form influence the outcome of bioequivalence studies.
The relative risk of obtaining a non-BE result was approximately four times lower for drugs in
classes 1 and 3 of BCS or BDDCS when compared with class 2 drugs. Thus, it seems that the
final outcome of a bioequivalence study is strongly influenced by the solubility of the drug, but
not by its intestinal permeability or extent of metabolism. © 2013 Wiley Periodicals, Inc. and
the American Pharmacists Association J Pharm Sci 102:3136–3144, 2013
Keywords: biowaiver; BCS; BDDCS; bioequivalence; dissolution; permeability; solubility
In 1995, Amidon and coworkers1–4 published the fun-
damentals of the biopharmaceutics classification sys-
tem (BCS). These were based on the results from a
set of absorption models, which demonstrated that
oral drug absorption is chiefly controlled by two char-
acteristics of the drug substances, that is, solubility
under physiological conditions and intestinal perme-
This article reflects the scientific opinion of the authors and
not necessarily the policies of the Brazilian Health Surveillance
Agency (ANVISA).
Correspondence to: Rodrigo Cristofoletti (Telephone: +55-
61-81114955; Fax: +55-61-81598587; E-mail: rodrigocristofol@
Journal of Pharmaceutical Sciences, Vol. 102, 3136–3144 (2013)
© 2013 Wiley Periodicals, Inc. and the American Pharmacists Association
ability, and by the dissolution rate of the drug from
the dosage form.1–4 According to the BCS, a drug could
be categorized under one of the four classes of BCS
depending on its solubility and intestinal permeabil-
ity, at which point it would be possible to define the
rate-limiting step in drug absorption: gastric empty-
ing time or intestinal permeability (for a rapid dis-
solving drug product containing a BCS class 1 or BCS
class 3 drug, respectively), in vivo dissolution and/or
solubility (BCS class 2), or a mix of all variables (BCS
class 4).4In the same year, the US Food and Drug Ad-
ministration (FDA) started using BCS in its guidance
related to scale-up and postapproval changes to de-
fine the tests needed to support each level of change.5
The application of BCS principles in setting regula-
tory boundaries for waiving in vivo bioequivalence
studies in the approval of new generic drug products
was first implemented 5 years later in 2000.6In gen-
eral, the FDA has accepted biowaiver requests only
for immediate-release (IR) solid oral dosage form con-
taining a highly soluble and highly permeable drug.6
This policy has been challenged as too conservative
by some authors, who suggest extending biowaiver-
based decisions to drugs belonging to other BCS
categories.7–9 Currently, there is no harmonization
on this point among the various regulatory agencies
because some authorities accept biowaiver requests
for more than one BCS class and/or apply different
classification boundaries with regard to the defini-
tions of solubility and permeability.6,10–13 These dis-
crepancies have been summarized and discussed by
Potthast14 and Zheng et al.15
The European Medicines Agency (EMA) extended
the BCS-based biowaiver to highly soluble but poorly
permeable drugs (i.e., class 3), provided they are
formulated as very rapid dissolving dosage forms
whose compositions are qualitatively the same and
quantitatively very similar to the respective refer-
ence drug products. These restrictions were applied
to circumvent unpredictable effects on the membrane
transporters.12 This guideline is consistent with other
reports that had already suggested the extension of
biowaiver for BCS class 3 drugs. However, the concept
that a more stringent in vitro dissolution requirement
is needed to ensure that the oral drug absorption can
only be limited by its intestinal permeability lacks
In turn, the WHO recommends an even more flex-
ible approach because it not only embraces the con-
cepts of biowaiver based on BCS proposed by FDA
and EMA but also suggests its extension to certain
BCS class 2 drugs. According to WHO, weak acidic
drugs with high permeability and high solubility un-
der conditions of pH 6.8 could also be candidates for
biowaiver because such drugs would behave similarly
to BCS class 1 drugs in the small intestine, which is
the primary site of absorption.10 However, this regula-
tory approach has not found consensus among the sci-
entific community; despite being supported by some
authors,22–24 it has been questioned by others.25,26
In 2005, Wu and Benet27 reviewed the concepts of
BCS to resolve a possible ambiguity in the definition
of permeability. They recognized that highly perme-
able compounds are usually eliminated primarily by
metabolism, an observation that subsequently formed
the basis of the biopharmaceutics drug disposition
classification system (BDDCS), which classifies drugs
according to their solubility and extent of metabolism.
This concern about the ambiguity of the permeability
definition was further addressed in other reports, but
it does not seem to have been resolved as yet.28–30
Nonetheless, supporters of both the BCS and BDDCS
agree that the extent of drug metabolism, as charac-
terized by the fraction of metabolites formed by oxida-
tive (phase 1) or conjugative (phase 2) enzymes, would
add value to the present BCS requirements because
it would better reflect the fraction of drug absorbed.31
The EMA guidance was the first to mention metabo-
lites specifically as contributing to the overall fraction
of drug absorbed.12 As a caution, some authors have
noted that the polymorphism of genes related to drug
metabolism could generate divergent classifications
according to BDDCS because of variations in allelic
distribution among different populations.32
Given the lack of harmonization in the applica-
tion of BCS/BDDCS to allowing biowaiver-based drug
product approvals, a cross-sectional survey to eval-
uate the relative risk of obtaining nonbioequivalent
(non-BE) or bioinequivalent (BIE) results for each
class of BCS and BDDCS was conducted. The survey
also evaluated how well a partial set of in vitro dis-
solution profiles could predict the bioequivalence out-
come for both classification systems. The aim was to
use retrospective data to determine the feasibility of
biowaiver extensions to classes 2–4 drug substances.
A cross-sectional survey was performed on a random
sample of bioequivalence studies collected from an in-
ternal database of the Brazilian Health Surveillance
Agency (ANVISA). This database, referred to as the
Brazilian System of Bioequivalence and Pharmaceu-
tical Equivalence (SINEB), is a computerized record
of all bioequivalence studies conducted in Brazil since
2008, including studies showing bioequivalent (BE),
non-BE, and BIE results (with BIE being defined
as when the point estimate is outside the range of
0.80–1.25 in a study with statistical power greater
than 80%). The sample size necessary to achieve a
priori statistical power of 80% at the 0.05 level of sig-
nificance (two sided) was calculated using OpenEpi R
version 2.3.133 by considering the prevalence of non-
BE results that has been reported for each class of
the BCS.34,35 Because of the very low difference in
the prevalence of non-BE results for drugs belonging
to BCS classes 1 and 3, the sample size required to cor-
rectly reject a false null hypothesis (H0: relative risk =
1; total sample size of 13,300 bioequivalence studies)
was much higher than the total content of the SINEB.
Thus, the sample size was calculated considering only
comparisons between classes 1, 3, and 4 versus class
2 of the BCS. This calculation indicated the need to
consider at least 96 bioequivalence studies for each
of the BCS classes. Taking into account the distribu-
tion of drugs belonging to each class of BCS in a set
of registered generic drug products,36 we randomly
sampled 500 bioequivalence studies from SINEB, en-
abling us to statistically evaluate how the charac-
teristics related to the drug (solubility, intestinal
permeability, and extension of drug metabolism) and
dosage form (in vitro dissolution) can influence the
outcome of a bioequivalence study. All studies sam-
pled were approved by the ethics committees and
conducted by the contract research organizations cer-
tified for good laboratory practices, good clinical prac-
tices, and in accordance to the Helsinki declaration
and the Brazilian law. In vitro dissolution data, per-
formed under pharmacopeial or validated conditions,
were also collected, where available.
Drug Classification
Drugs were classified according to the BDDCS by re-
ferring to a previous list.37 This list also provided
the classification of drug solubility over pH range
of 1.0–7.5 and the calculated log P(clog P) value,
which was chosen as a parameter of lipophilicity to
provisionally classify the sampled drugs into BCS.
Metoprolol was chosen as the reference compound for
intestinal permeability because its absorbed fraction
was around 90%.38 On this basis, drugs showing clog
Pvalues higher than or equal to the corresponding
value for metoprolol were assumed to be highly per-
Statistical Analysis
The proportions of non-BE results were compared
among all BCS or BDDCS classes using a chi-square
test (or Fisher’s exact as appropriate). Multiple 2 ×
2 contingency tables were developed to compare all
combinations of the independent classes of BCS or
BDDCS. The relative risk of obtaining a non-BE or
a BIE result and its 95% confidence interval (95%
CI) were also calculated. Data were analyzed using
for Windows, version 17 (SPSS Inc., Chicago,
Illinois). A pvalue of less than 0.05 was considered
The predictability of the in vitro dissolution test
with respect to the bioequivalence outcome was
evaluated by the following diagnostic parameters:
sensitivity, specificity, likelihood ratio positive (LR+),
likelihood ratio negative (LR), and posttest prob-
ability. Sensitivity is the conditional probability of
detecting a similar in vitro dissolution profile given
that a generic drug product is indeed BE. In turn,
specificity is the conditional probability of detecting
a nonsimilar in vitro dissolution profile given that
a generic drug product is non-BE. The complemen-
tary probabilities of sensitivity and specificity are the
false-negative and false-positive probabilities, respec-
tively. The LR+is the ratio between sensitivity and
the false-positive probability (1 specificity). In other
words, LR+for dissolution tests is a statistic for sum-
marizing how many times more likely BE generic
drug products are to have similar in vitro dissolution
profiles than non-BE drug products. On the contrary,
LRis the ratio between the false-negative proba-
bility (1 sensitivity) and specificity, showing how
many times less likely BE generic drug products are
to have nonsimilar in vitro dissolution profiles than
non-BE drugs products. A LR+higher than 1 indi-
cates that similar in vitro dissolution profiles are as-
sociated with BE results, whereas a LRlower than 1
indicates that nonsimilar in vitro dissolution profiles
are associated with non-BE results. Thus, a LR whose
95% CI contains 1 lacks diagnostic value. Although
these four indicators summarize diagnostic accuracy,
they can only address how well a BE (or a non-BE)
result “predicts” an in vitro dissolution result because
they represent conditional probabilities whose a pri-
ori condition is the result of the in vivo study (i.e.,
given a BE result what is the probability of finding a
similar in vitro dissolution?). However, in the regula-
tory field, it is of greater interest to know how well the
in vitro test can predict the bioequivalence outcome.
This was assessed by calculating the posttest prob-
ability, which combines the prevalence (also known
as the pre-test probability) of BE and non-BE re-
sults with LR+and LR, respectively. According to
Bayes’ theorem, such a combination describes, for in-
stance, how a similar or a nonsimilar in vitro disso-
lution profile changes the prior knowledge about the
probability of obtaining a BE or a non-BE result. To
apply the posttest probability, some additional calcu-
lations (described in the following references) were
required to convert odds to probabilities.39–43 Alge-
braically, these diagnostic parameters can be defined
as follows (please see Table 1):
Sensitivity =a/(a+c)
Specificity =d/(b+d)
LR+=sensitivity/(1 specificity)
LR−=(1 sensitivity)/specificity
Posttest probabilityBE =posttest oddsBE/(1 +
posttest oddsBE), where
Posttest oddsBE =pretest oddsBE ×LR+
Pretest oddsBE =prevalenceBE/(1
Table 1. General 2 ×2 Contingency Table
In vivo Result
In vitro Result Bioequivalent (BE) Nonbioequivalent (non-BE)
Similar dissolution profile a (true positive) b (false positive)
Nonsimilar dissolution profile c (false negative) d (true negative)
PrevalenceBE =pretest probabilityBE =(a+c)/(a
Posttest probabilitynon-BE is defined analogously,
where posttest oddsnon-BE =pretest oddsnon-BE ×
The 95% CIs for sensitivity, specificity, LR+,and
LRwere calculated according to previously reported
All sampled bioequivalence studies were conducted
with drug products formulated as IR dosage forms
that were not intended for absorption in the oral
cavity. Moreover, only non-BE studies whose statis-
tical power a posteriori was higher than 80% were in-
cluded in this survey. In this context, it was assumed
that a non-BE study with a point estimate outside the
bioequivalence boundaries was BIE; in other words,
the non-BE result was because of the differences be-
tween the test and the reference formulations.
A total of 107 out of 114 drugs were classified ac-
cording to BCS and BDDCS, representing 95% (475)
of the studies previously sampled. Figure 1 summa-
rizes the number of drug substances grouped at each
class of BCS and BDDCS.
Drugs with at least 70% metabolism were clas-
sified as being extensively metabolized; however, if
Figure 1. Distribution of drugs sampled between BCS
and BDDCS classes.
its fraction metabolized was lower than this cutoff
value, the drug was classified as poorly metabolized.37
The lipophilicity of 66 extensively metabolized drugs
and 23 poorly metabolized drugs were, respectively,
higher and lower than the metoprolol value. On the
contrary, nine drugs with fraction metabolized higher
than 70% of the administered dose had a clog Pvalue
lower than metoprolol, whereas nine drugs showing
high lipophilicity were classified as being poorly me-
tabolized. This lack of a one to one correspondence
has already been addressed by several authors and
has mainly been attributed to the differences in ac-
cess to the metabolizing enzymes within the hep-
atocytes or the predominance of a drug absorption
mechanism based on paracellular or carrier-mediated
transport.27,32, 46,47
In all classes of both classification systems, there
were non-BE results for Cmax (peak plasma con-
centration) and AUC0–t(area under the plasma
concentration-time curve). The general prevalence of
non-BE results ranged from 10% (class 3) to 40%
(class 2). Moreover, BIE results were also observed
in all BCS and BDDCS categories (see Table 2).
Table 3 shows a comparison between the relative
risks of obtaining a non-BE or a BIE result for each
class of BCS or BDDCS taken two at a time. Summa-
rizing, the risk of non-BE for drug products containing
class 2 drug substances according to either the BCS
or BDDCS was between 2.5-fold and fourfold higher
than those containing highly soluble drugs. In all of
these comparisons, the statistical power was higher
than 99% (calculated on OpenEpi R
version 2.3.1).33
On the contrary, we were not able to ensure an ad-
equate statistical power for comparing the classes 1
and 3, as expected, or for any comparison involving
class 4 drugs (mainly because of the very low preva-
lence of BCS class 4 drugs in our sample—only 2.8%).
As none of the comparisons involving BCS or BDDCS
class 4 drugs showed statistical validity, studies of
drug products containing these drugs were excluded
from subsequent analysis.
Table 2. Distribution of Non-BE Results Between the Four Classes of BCS and BDDCS
System Classes
Non-BE Results
Only for Cmax
Non-BE Results
Only for AUC0–t
Non-BE Results for
Cmax and AUC0–t Non-BE Results
Number of
Studies Sampled
BCS 1 18 (7) 0 4 (1) 22 (8) 140
2 52 (21) 5 (0) 26 (11) 83 (32) 206
3 7 (4) 1 (0) 4 (1) 12 (5) 115
4 1 (0) 1 (0) 1(1) 3 (1) 14
BDDCS 1 18 (8) 1 (0) 4 (2) 23 (10) 150
2 48 (19) 5 (0) 25 (10) 78 (29) 191
3 7 (4) 0 4 (1) 11 (5) 105
4 5 (2) 1 (0) 2 (1) 8 (3) 29
The numbers between parentheses represent the BIE results (point estimate outside 80–125 in a study with statistical power is greater than 80%) among
those already declared as non-BE.
Table 3. Relative Risk of Obtaining a Non-BE or a BIE Result
95% Confidence Interval 95% Confidence Interval
System Classes Compared Risk of Non-BEaLower Upper Risk of BIEaLower Upper
BCS 1 ×20.3900.257 0.592 0.3680.175 0.774
1×31.506 0.779 2.910 1.314 0.442 3.908
2×33.8612.204 6.764 3.5731.432 8.916
BDDCS 1 ×20.3750.248 0.567 0.3950.193 0.809
1×31.464 0.746 2.871 1.575 0.498 4.980
2×33.8982.173 6.994 3.9861.440 11.030
aThese columns show the relative risks of the first class compared with the second class, in each comparison 2 ×2 (i.e., for the first line: BCS 1 compared
with BCS 2).
p<0.01 showing a significant difference in the proportion of non-BE or BIE results between the classes compared.
The predictability of in vitro dissolution profile per-
formed under a single experimental condition, phar-
macopeial or validated, over the bioequivalence out-
come was also evaluated for all classes of BCS and BD-
DCS. A total of 29 out of 475 drug products sampled
did not present a comparison of in vitro dissolution
profile; thus, the diagnostic parameters sensitivity,
specificity, LR+,LR, and posttest probability were
calculated with data provided by 432 and 417 drug
products containing drug substances from classes 1–3
of BCS and BDDCS, respectively (Tables 4 and 5).
A similar in vitro dissolution profile was approx-
imately 1.7 times more likely for a BE than for
a non-BE generic drug product containing a BCS or
a BDDCS class 1 drug. The probability of detecting
a BE drug product was increased to 90% when the
in vitro dissolution profile of the test drug product had
been similar to the reference drug product (posttest
probability). Also, the predictability of a non-BE out-
come was significantly increased after performing an
in vitro dissolution test for drug products containing
highly soluble drugs. The posttest probability of de-
tecting a non-BE result for drug products containing
BCS classes 1 or 3 drugs was 4.5- or 2.5-fold higher
than the respective pretest probability, after obtain-
ing a nonsimilar in vitro dissolution profile.
Table 4. Similarities Between in vitro and in vivo Results Inside Each Class of BCS and BDDCS
System Classes BE and DS Non-BE and DS BE and DNS Non-BE and DNS Total
BCS 1 112 12 4 10 138
2 98 60 17 10 185
2a13 7 1 1 22
3 91 7 8 3 109
3b74 3 25 7 109
BDDCS 1 123 13 4 10 150
2 90 57 15 9 171
379 6 8 3 96
aResults for a set of five highly permeable acidic drugs.
bThe in vitro test was considered similar only when both test and reference drug products showed a very rapid dissolution (Q >85% in 15 min).
BE, bioequivalent result; non-BE, nonbioequivalent result; DS, similar in vitro dissolution profile; DNS, nonsimilar in vitro dissolution profile.
Table 5. Predictability of BCS and BDDCS over the Bioequivalence Outcome
Systems Classes Sensitivity Specificity LR+
of BE
of BE LR
of Non-BE
of Non-BE
BCS 1 0.96 (0.91–0.99) 0.45 (0.27–0.65) 1.77 (1.21–2.60) 0.84 0.90 0.07 (0.03–0.22) 0.16 0.71
2 0.85 (0.78–0.90) 0.14 (0.08–0.24) 1.00 (0.88–1.12) 0.62 0.62 1.03 (0.50–2.13) 0.38 0.38
2a0.93 (0.70–0.99) 0.12 (0.02–0.47) 1.07 (0.79–1.43) 0.64 0.64 0.53 (0.04–7.44) 0.36 0.36
3 0.92 (0.84–0.96) 0.30 (0.11–0.60) 1.31 (0.87–1.97) 0.91 0.91 0.27 (0.09–0.87) 0.09 0.27
3b0.75 (0.65–0.82) 0.70 (0.40–0.89) 2.49 (0.96–6.12) 0.91 0.91 0.36 (0.21–0.61) 0.09 0.22
BDDCS 1 0.97 (0.92–0.99) 0.43 (0.26–0.63) 1.71 (1.20–2.46) 0.85 0.90 0.07 (0.02–0.21) 0.15 0.71
2 0.86 (0.78–0.91) 0.14 (0.07–0.24) 0.99 (0.88–1.12) 0.61 0.61 1.05 (0.49–2.26) 0.39 0.39
3 0.91 (0.83–0.95) 0.33 (0.12–0.64) 1.36 (0.85–2.17) 0.91 0.91 0.28 (0.09–0.86) 0.09 0.27
aResults for a set of five highly permeable acidic drugs.
bThe in vitro test was considered similar only when both test and reference drug products showed a very rapid dissolution (Q>85% in 15 min).
On the contrary, performing an in vitro dissolution
test for a drug product containing a class 2 drug did
not add any value to diagnosis of a BE or a non-BE
result because both LR+and LRlack statistical sig-
nificance (both 95% CI contain 1).
Distribution of Non-BE Results Among BCS and BDDCS
The prevalence of non-BE results among the BCS
classes (Fig. 1 and Table 2) was similar to those previ-
ously reported.34,35 The highest prevalence of non-BE
results occurred for the poorly soluble, highly perme-
able drugs. This result was expected because accord-
ing to the BCS theory, the absorption of a class 2 drug
is controlled by in vivo drug dissolution or even by its
solubility (as in the case of drugs with very high dose
number like that of griseofulvin), which means that
pharmaceutics properties rather than emptying gas-
tric time or intestinal permeability control the drug
Also, the proportion of non-BE results were sim-
ilar in each BCS/BDDCS class, even though these
two systems employ different cutoff values to clas-
sify the extent of drug absorption: more than 90% for
BCS and more than 70% for BDDCS. Indeed, the frac-
tion absorbed of extensively metabolized drugs might
be even lower than 70% of the administered dose in
some cases because the metabolism criterion used for
BDDCS assignment did not limit the metabolic pro-
cesses to oxidative (phase 1) and conjugative (phase 2)
enzymes, which occur only after drug absorption,31,37
and many drugs can be metabolized presystemically
by bacterial enzymes.49,50
Relative Risk of Obtaining a Non-BE Result for Each
Class of BCS and BDDCS
The relative risk of obtaining a non-BE or a BIE result
was not different from unity when comparing drug
products containing drugs belonging to the classes 1
and 3 of BCS or BDDCS. However, it should be men-
tioned that the type 2 error was too high for these
analyses (around 70%, with H0:μclass 1 =μclass 3) be-
cause we were not able to collect an adequate sample
size, as already explained. On the contrary, the rel-
ative risk of getting a non-BE or a BIE result was
almost fourfold higher for poorly soluble and highly
absorbed drugs when compared with highly soluble
drugs (see Table 3). Because the estimated risks were
similar between both comparisons, class 1 or 3 ver-
sus class 2 of both systems, it seems that solubility
outweighs any effect of the extent of absorption with
regard to the bioequivalence outcome. These results
can aid in assessing the risk of extending biowaiver
beyond BCS class 1 drugs, as suggested by Polli et al.8
However, a potential weakness in the above anal-
ysis is that the BCS classification relied on clog P
as a measure of permeability. Clog Pdata are not
recognized in any regulatory guidance of biowaiver
as being sufficient for or even supportive of a per-
meability classification. So the provisional BCS clas-
sification based on solubility and clog Pcould lead
to false negative results (in the case of drugs that
are substrates for uptake with regard to permeability
assessment). On the contrary, there is a reasonably
good correlation between lipophilicity and intestinal
permeability, at least for drugs absorbed mainly by
passive mechanisms.46,47
Predictability of
In Vitro
Dissolution Test of
Bioequivalence Outcome—Diagnostic Parameters for
Highly Soluble Drugs
Considering that a comparative in vitro dissolution
profile and a bioequivalence study can have dichoto-
mous outcomes, it was possible to compare both in 2 ×
2 contingency tables. Indicators of test performance
derived from such tables were used to evaluate the
predictability of in vitro test over in vivo outcomes
(Tables 1 and 5).
The sensitivity of the single in vitro dissolution
test was significantly high for all BCS or BDDCS
classes, indicating that 75%–96% of the BE drug prod-
ucts showed similar in vitro dissolution profiles to
their respective reference drug products, whereas the
specificity was not statistically significant for drugs
belonging to classes 1 and 3 of BCS and BDDCS; in
other words, 50% of the non-BE drug products showed
nonsimilar in vitro dissolution profiles, whereas the
other half showed similar in vitro results. However,
the biowaiver based on the BCS requires not one but
a set of three in vitro dissolution profiles performed
under conditions relevant to pH in the upper gastroin-
testinal tract. This set can be interpreted as a serial
diagnostic test because the result of the first deter-
mines whether the second test is performed.51 For
example, if the in vitro dissolution profiles at pH 1.2
are similar, the second in vitro dissolution test at pH
4.5 is performed, and if it is also similar, the third test
at pH 6.8 is conducted. Only if all three in vitro dis-
solution profiles are similar, the drug product will be
approved according to the biowaiver procedure, oth-
erwise the diagnostic is negative. This combination
of multiple tests leads to a higher overall specificity
than is possible with a single in vitro test.52 Consider-
ing that the diagnostic indicators in this survey were
calculated using only a single in vitro result from a
dissolution test performed under pharmacopeial or
manufacturer-validated conditions to assess quality
control, it seems safe to expect a higher accuracy of
the diagnostic test when performing the complete set
of in vitro dissolution profiles required to support a
biowaiver request.
The absorption of a class 3 drug is more complex
than that of a class 1 drug because its absorption is
segmentally dependent along the small intestine,53–57
and it is also mediated by uptake transporters.27,58
Therefore, the absorption of a class 3 drug could
be more affected by certain excipients, which inter-
act with gastrointestinal motility, reducing the in-
testinal transit time,59 and/or interact with uptake
transporters.58 As Brazilian law does not require a
quantitative or qualitative similarity between generic
and reference formulations,60 the composition of such
drug products generally differ. As the formulations of
the 500 drug products sampled were not compared
with the respective reference formulations, we can-
not exclude the possibility of different excipients pre-
sented in a generic formulation having affected a
membrane carrier or the gastrointestinal motility, re-
sulting in a non-BE that could not have been predicted
by in vitro dissolution testing. This might explain the
lower posttest probability presented by drug products
containing class 3 drug substances when compared
with class 1 drugs (27% vs. 71%). Further studies
would be necessary to confirm this point. Neverthe-
less, because of the generally low risk of obtaining
a non-BE result for class 3 drugs, it seems safe to
extend biowaiver decisions for generic drug products
containing highly soluble, poorly absorbed drugs as
long as they are formulated with the same excipients
present in the reference formulation, along the lines
of the EMA guideline.12
Rapid Dissolution Versus Very Rapid Dissolution for
Biowaiving BCS Class 3 Drugs
The diagnostic indicators for drug products contain-
ing BCS class 3 drugs were also calculated using two
dissolution criteria, rapid dissolution (Q>85% in
30 min, with similarity being demonstrated by f2)
versus very rapid dissolution (Q>85% in 15 min,
without f2 calculation). As the 95% CI for the di-
agnostic parameters calculated for both dissolution
criteria almost completely overlapped (see Table 5),
it seems that the predictability of the in vitro dis-
solution profile of the BE outcome is not improved
by employing a more restrictive dissolution criterion.
Further, this procedure increased the probability of
false negative results (1 sensitivity, from 8% to
25%) and decreased the posttest probability of non-
BE results (from 27% to 22%). Previous reports have
already demonstrated that a very rapid dissolution
criterion is too conservative for biowaiving oral for-
mulations containing BCS class 3 drugs.16,17 ,21,61,62
Predictability of
In Vitro
Dissolution Test of BE
Outcome—Diagnostic Parameters for Class 2 Drugs
The probability of false positive results obtained in
in vitro test for class 2 drugs was almost 90%, show-
ing a statistically significant 95% CI for the speci-
ficity, suggesting that the pharmacopeial dissolution
methods are not biorelevant. Furthermore, LR+and
LRlack diagnostic value because its 95% CI con-
tains unity. So, the posttest probability is equal to the
pretest probability; in other words, the predictabil-
ity of in vitro dissolution test, performed according to
the pharmacopeial methods, with respect to the bioe-
quivalence outcome was not different from the inher-
ent prevalence of BE or non-BE results (see Table 5).
However, as we did not have access to the set of in
vitro dissolution profiles, which would be required by
guidelines on the BCS-based biowaiver for these drug
products, no conclusion can be reached about their
Diagnostic Parameters for Highly Permeable Weak
Acidic Drugs Showing High Solubility at pH 6.8
As the WHO recommends an extension of biowaiver
based on BCS for drug products containing class 2
drugs that show high solubility and rapid dissolution
at pH 6.8, we also calculated the diagnostic indica-
tors utilizing 22 results of in vitro dissolution profiles
performed at neutral conditions (phosphate buffer
under pH 6.8–7.5) for a set of eight non-BE (all of
them because of Cmax) and 14 BE drug products con-
taining five weak acidic drugs with pKaoflessthan
5.5 and which have dose numbers lower than one at
pH values ranging from 6.8 to 7.4.22 However, these
diagnostic parameters proved not to be different from
those calculated for all drugs belonging to BCS class
2, in accordance with the previous reports for drug
products containing ibuprofen that showed that dis-
solution tests under neutral conditions were unable
to detect differences in absorption rate on one hand25
or overdiscriminated in cases of in vivo equivalence
on the other hand.26 In fact, it seems that the disso-
lution of weak acidic drugs under gastric conditions
might be able to detect in vivo differences related to
Cmax,18,23 ,25 but the low amount of drug dissolved pre-
cludes a meaningful calculation of the f2 statistics.
Recently, some authors suggested scaling the data
to 100% release from the reference formulation be-
fore applying the f2 statistics (as the f2 represents an
absolute rather than a relative difference). However,
the discussion in this arena has just started, and the
best way to handle the dissolution data for products
containing highly permeable weak acidic drugs still
needs to be agreed upon.24
With this cross-sectional survey, we demonstrated
that solubility outweighs any effect of the extent of
drug absorption, measured either as intestinal per-
meability or extension of drug metabolism, with re-
gard to the bioequivalence outcome. Also, as the esti-
mated risks were similar between the comparisons of
classes 1 and 3 versus class 2 of both systems, BDDCS
could be considered as adequate as BCS in the classifi-
cation of drugs with a view to applying the biowaiver.
As in some cases, drug metabolism data are more
readily available than permeability data; it seems
useful to accept both ways of classifying drugs for reg-
ulatory purposes. Moreover, because of the generally
low risk of obtaining non-BE or BIE results for class
3 drugs, it seems safe to extend biowaiver decisions
for generic drug products containing highly soluble,
poorly absorbed drugs as long as they are formulated
with the same excipients present in the reference for-
mulation and in similar amounts. On the contrary,
because of the high risk of obtaining non-BE or BIE
results and the lack of significance of LR+and LR
seen with the quality control in vitro dissolution test,
extending biowaiver for class 2 drugs does not seem
to be as straightforward.
1. Amidon GL, Sinko PJ, Fleisher D. 1988. Estimating human
oral fraction dose absorbed: A correlation using rat intesti-
nal membrane permeability for passive and carrier-mediated
compounds. Pharm Res 5(10):651–654.
2. Sinko PJ, Leesman GD, Amidon GL. 1991. Predicting fraction
dose absorbed in humans using a macroscopic mass balance
approach. Pharm Res 8(8):979–988.
3. Oh DM, Curl RL, Amidon GL. 1993. Estimating the fraction of
dose absorbed from suspensions of poorly soluble compounds
in humans: A mathematical model. Pharm Res 10(2):264–270.
4. Amidon GL, Lennern ¨
as H, Shah VP, Crison JR. 1995. A the-
oretical basis for a biopharmaceutic drug classification: The
correlation of in vitro drug product dissolution and in vivo
bioavailability. Pharm Res 12(3):413–420.
5. US Food and Drug Administration (FDA). 1995. Guidance
for industry: Scale-up and post approval changes: Chem-
istry, manufacturing, and controls, in vitro dissolution testing,
and in vivo bioequivalence documentation. Accessed July 10,
2012, at:
6. US Food and Drug Administration (FDA). 2000. Guidance for
industry: Waiver of in vivo bioavailability and bioequivalence
studies for immediate-release solid oral dosage forms based on
a biopharmaceutical classification system. Accessed July 10,
2012, at:
Shah VP, Lesko LJ, Chen ML, Lee VHL. 2002. Biopharma-
ceutics classification system: The scientific basis for biowaiver
extensions. Pharm Res 19(7):921–925.
8. Polli JE, Yu LX, Cook JA, Amidon GL, Borchardt RT, Burn-
side BA, Burton PS, Chen ML, Conner DP, Faustino PJ. 2004.
Summary workshop report: Biopharmaceutics classification
system—Implementation challenges and extension opportu-
nities. J Pharm Sci 93(6):1375–1381.
9. Polli JE, Abrahamsson BSI, Yu LX, Amidon GL, Baldoni JM,
Cook JA, Fackler P, Hartauer K, Johnston G, Krill SL. 2008.
Summary workshop report: Bioequivalence, biopharmaceutics
classification system, and beyond. AAPS J 10(2):373–379.
10. World Health Organization (WHO). 2006. Proposal to waive
in vivo bioequivalence requirements for the WHO Model List
of Essentials Medicines immediate release solid oral dosage
forms. Accessed July 20, 2012, at:
info general/documents/TRS937/WHO TRS 937 annex8 eng.
11. Gupta E, Barends D, Yamashita E, Lentz K, Harmsze A, Shah
V, Dressman J, Lipper R. 2006. Review of global regulations
concerning biowaivers for immediate release solid oral dosage
forms. Eur J Pharm Sci 29(3):315–324.
12. European Medicines Agency (EMA). 2010. Guideline on
the investigation of bioequivalence. Accessed July 20, 2012,
at: GB/document library/
Scientific guideline/2010/01/WC500070039.pdf.
13. Brazilian Health Surveillance Agency (ANVISA). 2011.
Resoluc¸ ˜
ao RDC n. 37/11: Disp˜
oe sobre o Guia para isenc¸˜
e substituic¸˜
ao de estudos de biodisponibilidade relativa/
encia e d´
a outras providˆ
encias. Accessed July 20,
2012, at:
2011/res0037 03 08 2011.html.
14. Potthast H. 2012. Update on regulations on in vitro equiva-
lence testing. In Biowaiver monographs 2004–2012; Dressman
JB, Ed. The Hague, The Netherlands: FIP, pp 34–50.
15. Zheng N, Lionberger RA, Mehta MU, Yu LX. 2012. The
biowaiver monographs: BCS based biowaivers—A regulatory
overview. In Biowaiver monographs 2004–2012; Dressman JB,
Ed., The Hague, The Netherlands: FIP, pp 52–71.
16. Blume HH, Schug BS. 1999. The biopharmaceutics classifica-
tion system (BCS): Class III drugs—Better candidates for BA/
BE waiver? Eur J Pharm Sci 9(2):117–121.
17. Cheng CL, Yu LX, Lee HL, Yang CY, Lue CS, Chou CH. 2004.
Biowaiver extension potential to BCS class III high solubility–
low permeability drugs: Bridging evidence for metformin
immediate-release tablet. Eur J Pharm Sci 22(4):297–304.
18. Kortej¨
arvi H, Urtti A, Yliperttula M. 2007. Pharmacoki-
netic simulation of biowaiver criteria: The effects of gastric
emptying, dissolution, absorption and elimination rates. Eur
J Pharm Sci 30(2):155–166.
19. Stavchansky S. 2008. Scientific perspectives on extending the
provision for waivers of in vivo bioavailability and bioequiva-
lence studies for drug products containing high solubility-low
permeability drugs (BCS-class 3). AAPS J 10(2):300–305.
20. Homˇ
sek I, Parojˇ
c L, Jovanovi´
2010. Justification of metformin hydrochloride biowaiver cri-
teria based on bioequivalence study. Arzneimittelforschung
21. Tsume Y, Amidon GL. 2010. The biowaiver extension for BCS
class III drugs: The effect of dissolution rate on the bioequiv-
alence of BCS class III immediate-release drugs predicted by
computer simulation. Mol Pharm 7(4):1235–1243.
22. Yazdanian M, Briggs K, Jankovsky C, Hawi A. 2004. The “high
solubility” definition of the current FDA guidance on biophar-
maceutical classification system may be too strict for acidic
drugs. Pharm Res 21(2):293–299.
23. Shohin IE, Kulinich JI, Ramenskaya GV, Abrahamsson
B, Kopp S, Langguth P, Polli JE, Shah VP, Groot DW,
Barends DM, Dressman JB. 2012. Biowaiver monographs
for immediate-release solid oral dosage forms: Ketoprofen.
J Pharm Sci 101(10):3593–3603.
24. Barends DM, Shah VP, Dressman J. 2012. Biowaiver mono-
graphs—What have we learned? In Biowaiver monographs
2004–2012; Dressman JB, Ed. The Hague, The Netherlands:
FIP, pp 15–32.
25. ´
Alvarez C, N´
nez I, Torrado JJ, Gordon J, Potthast H, Garc´
Arieta A. 2011. Investigation on the possibility of biowaivers
for ibuprofen. J Pharm Sci 100(6):2343–2349.
26. Shohin I, Kulinich J, Vasilenko G, Ramenskaya G. 2011.
Interchangeability evaluation of multisource ibuprofen drug
products using biowaiver procedure. Indian J Pharm Sci
27. Wu CY, Benet LZ. 2005. Predicting drug disposition via ap-
plication of BCS: Transport/absorption/elimination interplay
and development of a biopharmaceutics drug disposition clas-
sification system. Pharm Res 22(1):11–23.
28. Benet LZ. 2010. Predicting drug disposition via application of
a Biopharmaceutics Drug Disposition Classification System.
Basic Clin Pharmacol Toxicol 106(3):162–167.
29. Benet L, Larregieu C. 2010. The FDA should eliminate the
ambiguities in the current BCS biowaiver guidance and make
public the drugs for which BCS biowaivers have been granted.
Clin Pharmacol Ther 88(3):405–407.
30. Amidon K, Langguth P, Lennern ¨
as H, Yu L, Amidon G. 2011.
Bioequivalence of oral products and the biopharmaceutics clas-
sification system: Science, regulation, and public policy. Clin
Pharmacol Ther 90(3):467–470.
31. Benet LZ, Amidon GL, Barends DM, Lennern ¨
as H, Polli JE,
Shah VP, Stavchansky SA, Yu LX. 2008. The use of BDDCS
in classifying the permeability of marketed drugs. Pharm Res
32. Chen ML, Yu L. 2009. The use of drug metabolism for predic-
tion of intestinal permeability. Mol Pharm 6(1):74–81.
33. Dean AG, Sullivan KM, Soe MM. 2012. OpenEpi: Open source
epidemiologic statistics for public health, Version 2.3.1. Ac-
cessed January 15, 2012, at:
34. Lamouche S, Leonard H, Shink ´
E, Tanguay M. 2008. The
biopharmaceutical classification system: Can it help predict
bioequivalence outcome? A CRO retrospective analysis. Ac-
cessed January 10, 2012, at:
AM 2008/AAPS2008–002992.PDF.
35. Ramirez E, Laosa O, Guerra P, Duque B, Mosquera B, Borobia
AM, Lei SH, Carcas AJ, Frias J. 2010. Acceptability and
characteristics of 124 human bioequivalence studies with ac-
tive substances classified according to the Biopharmaceu-
tic Classification System. Br J Clin Pharmacol 70(5):694–
36. Anir AK, Anand O, Chun N, Conner DP, Mehta MU, Nhu DT,
Polli JE, Yu LX, Davit BM. 2012. Statistics on BCS classifica-
tion of generic drug products approved between 2000 and 2011
in the USA. AAPS J 14(4):664–666.
37. Benet LZ, Broccatelli F, Oprea TI. 2011. BDDCS applied to
over 900 drugs. AAPS J 13(4)519–547.
38. Regårdh CG, Borg KO, Johansson R, Johnsson G, Palmer L.
1974. Pharmacokinetic studies on the selective$1-receptor an-
tagonist metoprolol in man. J Pharmacokinet Pharmacodyn
39. Altman DG, Bland JM. 1994. Diagnostic tests. 1: Sensitivity
and specificity. Br Med J 308:1552.
40. McGee S. 2002. Simplifying likelihood ratios. J Gen Intern
Med 17(8):647–650.
41. Loong TW. 2003. Understanding sensitivity and specificity
with the right side of the brain. Br Med J 327:716–
42. Deeks JJ, Altman DG. 2004. Diagnostic tests 4: Likelihood
ratios. Br Med J 329:168–169.
43. Akobeng AK. 2007. Understanding diagnostic tests 2: Likeli-
hood ratios, pre- and post-test probabilities and their use in
clinical practice. Acta Paediatr 96(4):487–491.
44. Simel DL, Samsa GP, Matchar DB. 1991. Likelihood ratios
with confidence: Sample size estimation for diagnostic test
studies. J Clin Epidemiol 44(8):763–770.
45. Newcombe RG. 1998. Interval estimation for the difference be-
tween independent proportions: Comparison of eleven meth-
ods. Stat Med 17(8):873–890.
46. Kasim NA, Whitehouse M, Ramachandran C, Bermejo M,
Lennernas H, Hussain A, Junginger HE, Stavchansky S,
Midha K, Shah V, Amidon GL. 2004. Molecular properties of
WHO essential drugs and provisional biopharmaceutical clas-
sification. Mol Pharm 1(1):85–96.
47. Takagi T, Ramachandran C, Bermejo M, Yamashita S,
Lawrence XY, Amidon GL. 2006. A provisional biopharma-
ceutical classification of the top 200 oral drug products in the
United States, Great Britain, Spain, and Japan. Mol Pharm
48. Lobenberg R, Amidon GL. 2000. Modern bioavailability, bioe-
quivalence and biopharmaceutics classification system. New
scientific approaches to international regulatory standards.
Eur J Pharm Sci 50(1):3–12.
49. Sousa T, Paterson R, Moore V, Carlsson A, Abrahamsson B,
Basit AW. 2008. The gastrointestinal microbiota as a site for
the biotransformation of drugs. Int J Pharm 363(1–2):1–25.
50. Grundmann O. 2010. The gut microbiome and pre-systemic
metabolism: Current state and evolving research. J Drug
Metabol Toxicol 1(2):104–111.
51. Zhou XH, Obuchowski NA, Mcclish DK. 2002. Statistical meth-
ods in diagnostic medicine. New York: John Wiley & Sons.
52. Weinstein S, Obuchowski NA, Lieber ML. 2005. Clinical eval-
uation of diagnostic tests. Am J Roentgenol 184(1):14–19.
53. Wacher VJ, Salphati L, Benet LZ. 1996. Active secretion and
enterocytic drug metabolism barriers to drug absorption. Adv
Drug Deliv Rev 20(1):99–112.
54. Lee VHL. 2000. Membrane transporters. Eur J Pharm Sci
55. Cao X, Lawrence XY, Barbaciru C, Landowski CP, Shin
HC, Gibbs S, Miller HA, Amidon GL, Sun D. 2005. Perme-
ability dominates in vivo intestinal absorption of P-gp sub-
strate with high solubility and high permeability. Mol Pharm
56. Dahan A, Amidon GL. 2008. Segmental dependent transport
of low permeability compounds along the small intestine due
to P-glycoprotein: The role of efflux transport in the oral ab-
sorption of BCS class III drugs. Mol Pharm 6(1):19–28.
57. Dahan A, Miller JM, Hilfinger JM, Yamashita S, Yu L, Lenner-
nas H, Amidon GL. 2010. High-permeability criterion for BCS
classification: Segmental/pH dependent permeability consid-
erations. Mol Pharm 7(5):1827–1834.
58. Shugarts S, Benet LZ. 2009. The role of transporters in the
pharmacokinetics of orally administered drugs. Pharm Res
59. Chen ML, Straughn A, Sadrieh N, Meyer M, Faustino P,
Ciavarella A, Meibohm B, Yates C, Hussain A. 2007 A mod-
ern view of excipient effects on bioequivalence: Case study of
sorbitol. Pharm Res 24(1):73–80.
60. BRASIL. 1999. Lei n. 9.787/99: Disp ˜
oe sobre a vigilˆ
aria estabelece o medicamento gen´
erico, disp˜
oe sobre a
ao de nomes gen´
ericos em produtos farmacˆ
euticos e
a outras providˆ
encias. Accessed July 20, 2012, at: http:// 99.htm.
61. Polli JE. 1997. In vitro-in vivo relationships of several “imme-
diate” release tablets containing a low permeability drug. Adv
Exp Med Biol 423:191–198.
62. Jantratid E, Prakongpan S, Dressman J, Amidon G, Junginger
H, Midha K, Barends D. 2006. Biowaiver monographs for im-
mediate release solid oral dosage forms: Cimetidine. J Pharm
Sci 95(5):974–984.
... Two conference abstracts [5,6] report work on large databases of BE studies (918 and 1200, respectively), but information available from these reports is very limited. On the other hand, two full-length research articles [7,8] report the analyses of 124 and 500 BE studies, respectively, and focus mainly on the impact of BCS or Biopharmaceutical Drug Disposition Classification System (BDDCS), on acceptability of the BCS-based biowaiver approach and discriminative power of the in vitro methods. Comprehensive research for finding additional discriminatory features within each BCS Class that could help us additionally improve the risk assessment is limited. ...
... Significantly different percentages of non-BE studies were found across different BCS classes in our database (Table 4 and Fig. 1), indicating high association between BCS and BE study outcome. This is in line with a number of publications that supported in vivo predictive nature of BCS [6][7][8]. ...
... In theory, BCS class I and III drugs were presented as less risky for non-BE outcome compared with classes with poorly soluble APIs, while the publications often reported similarly low failure rate for classes I, III, and IV, ranging from 10% to 16%. Similarity of BCS class IV APIs to BCS class I and III was usually attributed to smaller sample size of BCS IV group (i.e., insufficient power to detect differences) [6][7][8]. Our estimated failure rate of 12.5% within BCS class IV seemed comparable to that reported in the literature; however, in our case the difference between highly soluble group of BCS and BCS class IV was obvious due to negligible failure rate in the highly soluble BCS classes. ...
Full-text available
Background and objectives: Understanding predictive potential of parameters to perform early bioequivalence (BE) risk assessment is crucial for good planning and risk mitigation during product development. The objective of the present study was to evaluate predictive potential of various biopharmaceutical and pharmacokinetic parameters on the outcome of BE study. Methods: Retrospective analysis was performed on 198 Sandoz (Lek Pharmaceuticals d.d., A Sandoz Company, Verovskova 57, 1526 Ljubljana, Slovenia) sponsored BE studies [52 active pharmaceutical ingredients (API)] where characteristics of BE study and APIs were collected for immediate-release products and their predictive potential on the study outcome was assessed using univariate statistical analysis. Results: Biopharmaceutics Classification System (BCS) was confirmed to be highly predictive of BE success. BE studies with poorly soluble APIs were riskier (23% non-BE) than with highly soluble APIs (0.1% non-BE). APIs with either lower bioavailability (BA), presence of first-pass metabolism, and/or being substrate for P-glycoprotein substrate (P-gP) were associated with higher non-BE occurrence. In silico permeability and time at peak plasma concentrations (Tmax) were shown as potentially relevant features for predicting BE outcome. In addition, our analysis showed significantly higher occurrence of non-BE results for poorly soluble APIs with disposition described by multicompartment model. The conclusions for poorly soluble APIs were the same on a subset of fasting BE studies; for a subset of fed studies there were no significant differences between factors in BE and non-BE groups. Conclusion: Understanding the association of parameters and BE outcome is important for further development of early BE risk assessment tools where focus should be first in finding additional parameters to differentiate BE risk within a group of poorly soluble APIs.
... The key method of classification impacting assessment of risk related to bioavailability is the Biopharmaceutics Classification System (BCS) [1], which is widely discussed within the literature. There are analyses available on real sets of data which assess BCS impact on bioequivalence outcome [2][3][4][5]. BCS is also implemented in assessments related to waiving of in vivo studies [6]. One of the extensions to BCS, the Biopharmaceutics Drug Disposition Classification System (BDDCS) [7], is a powerful tool for further insights into the mechanism of issues related to bioavailability. ...
... One of the extensions to BCS, the Biopharmaceutics Drug Disposition Classification System (BDDCS) [7], is a powerful tool for further insights into the mechanism of issues related to bioavailability. Nevertheless, it was not found to be advantageous for predicting bioequivalence study outcome, when compared with BCS [3]. ...
Full-text available
Background and objectives: Risk assessment related to bioequivalence study outcome is critical for effective planning from the early stage of drug product development. The objective of this research was to evaluate the associations between solubility and acido-basic parameters of an active pharmaceutical ingredient (API), study conditions and bioequivalence outcome. Methods: We retrospectively analyzed 128 bioequivalence studies of immediate-release products with 26 different APIs. Bioequivalence study conditions and acido-basic/solubility characteristics of APIs were collected and their predictive potential on the study outcome was assessed using a set of univariate statistical analyses. Results: There was no difference in bioequivalence rate between fasting and fed conditions. The highest proportion of non-bioequivalent studies was for weak acids (10/19 cases, 53%) and neutral APIs (23/95 cases, 24%). Lower non-bioequivalence occurrence was observed for weak bases (1/15 cases, 7%) and amphoteric APIs (0/16 cases, 0%). The median dose numbers at pH 1.2 and pH 3 were higher and the most basic acid dissociation constant (pKa) was lower in the non-bioequivalent group of studies. Additionally, APIs with low calculated effective permeability (cPeff) or low calculated lipophilicity (clogP) had lower non-bioequivalence occurrence. Results of the subgroup analysis of studies under fasting conditions were similar as for the whole dataset. Conclusion: Our results indicate that acido-basic properties of API should be considered in bioequivalence risk assessment and reveal which physico-chemical parameters are most relevant for the development of bioequivalence risk assessment tools for immediate-release products.
... The Biopharmaceutics Classification System (BCS) categorizes drug molecules into four groups based on their solubility and permeability profiles: -Class I: high permeability, high solubility compounds -Class II: high permeability, low solubility compounds -Class III: low permeability, high solubility compounds -Class IV: low permeability, low solubility compounds Up to 50% of all the authorized drug are categorized in classes II and IV [2]. For molecules belonging to classes II and IV, the main goal in formulation development is increasing their solubility [3]. ...
Full-text available
The issue of poor aqueous solubility is often a great hitch in the development of liquid dosage forms for those drugs that the Biopharmaceutics Classification System (BCS) includes in classes II and IV. Among the possible technological solutions, inclusion of the drug molecule within polymeric micelles, and particularly nanomicelles, has been proposed in the last years as a valid strategy. Our attention has been recently attracted by Soluplus ® , an amphiphilic polyvinyl caprolactam-polyvinyl acetate-polyethylene glycol graft copolymer able to form small and stable nanomicelles. The aim of this study was to characterize Soluplus ® nanomicelles to enhance the apparent solubility of three model APIs, categorized in BCS class II: ibuprofen (IBU), idebenone (IDE), and miconazole (MIC). Drug-loaded Soluplus ® micelles with a mean size around 60–70 nm were prepared by two methods (direct dissolution or film hydration method). The prepared nanosystems were characterized in terms of mean particle size and Zeta potential, physical stability, drug solubility, and in vitro drug release. The solubility of the tested APIs was shown to increase linearly with the concentration of graft copolymer. Soluplus ® can be easily submitted to membrane filtration (0.2 µm PES or PTFE membranes), showing the potential to be sterilized by this method. Freeze-drying enabled to obtain powder materials that, upon reconstitution with water, maintained the initial micelle size. Finally, viscosity studies indicated that these nanomicelles have potential applications where a bioadhesive material is advantageous, such as in topical ocular administration. Graphical abstract
... Limitations of C max as a BE metric are well described (26)(27)(28)(29). In a retrospective study performed in Brazil, 12 of 115 studies of Class III drug products provided nonbioequivalent (i.e., non-BE, where confidence interval exceeds 80-125% range) result, with 5 of those being bioinequivalent (i.e., point estimate is outside the range of 80-125%) (30). Specifically, among the 12 non-BE studies, 7 were due to only C max , 4 were due to both C max and AUC 0-t , and 1 was due to only AUC 0-t . ...
Full-text available
The objective of this review article is to summarize literature data pertinent to potential excipient effects on intestinal drug permeability and transit. Despite the use of excipients in drug products for decades, considerable research efforts have been directed towards evaluating their potential effects on drug bioavailability. Potential excipient concerns stem from drug formulation changes (e.g., scale-up and post-approval changes, development of a new generic product). Regulatory agencies have established in vivo bioequivalence standards and, as a result, may waive the in vivo requirement, known as a biowaiver, for some oral products. Biowaiver acceptance criteria are based on the in vitro characterization of the drug substance and drug product using the Biopharmaceutics Classification System (BCS). Various regulatory guidance documents have been issued regarding BCS-based biowaivers, such that the current FDA guidance is more restrictive than prior guidance, specifically about excipient risk. In particular, sugar alcohols have been identified as potential absorption-modifying excipients. These biowaivers and excipient risks are discussed here. Graphical Abstract
... BE is achieved when the bioavailabilities of two drugs "lie within acceptable predefined limits" to ensure "similarity in terms of safety and efficacy" (European Medicines Agency, 2010), thus demonstrating "the absence of significant difference in the rate and extent of absorption under similar experimental conditions" (U.S. Food and Drug Administration, 2013;Cristofoletti et al., 2018). BE can be demonstrated in vivo and in vitro (Chow, 2014), although in vitro assessment has limited acceptance, i.e., only for drugs with high solubility and permeability (Cristofoletti et al., 2013). A standard approach for demonstrating BE is a two-way crossover (2 × 2) clinical trial conducted in healthy subjects (Chow, 2014). ...
Full-text available
Demonstration of bioequivalence (BE) is mandatory while developing generic drugs. The scientific concept of BE applies equally to different regulatory agencies. However, the application of the concept may differ for each agency, which can affect the design of BE studies. To evaluate the study practices in terms of the BE concept in South Korea, we retrospectively analyzed BE study reports available from Ministry of Food and Drug Safety between 2013 and 2019. Statistical estimation of the pharmacokinetic parameters, including peak concentration and area under the concentration–time curve to the last measurable concentration, as well as study design, number of subjects in a study, study duration, fasting status, and formulation of specific drugs were obtained. The drugs were classified per World Health Organization Anatomical Therapeutic Chemical Classification and Biopharmaceutics Classification System. Post-hoc intrasubject coefficient of variation and corresponding sample sizes were calculated from the 90% confidence intervals of pharmacokinetic parameters. A total of 143 generic drugs in 588 BE studies were analyzed. The largest number of studies were performed in the area of Cardiovascular system (172 studies), followed by Nervous system (143 studies) and Alimentary tract and metabolism (92 studies). Overall, BE studies in South Korea were conducted in accordance with the global guideline despite the differences in details. BE studies were focused on the several therapeutic areas and conducted in a similar manner. The number of subjects was generally larger than that estimated with 90% power.
... BCS 2 drugs were the most commonly reported with DRBA, perhaps unsurprisingly as poor solubility is associated with greater variability in plasma exposure (108,109). The proportion of BCS 2 drugs (55.5%) listed here was lower to the proportion of BCS 2 drugs mentioned in a previous study (based on AUC criteria in healthy, young volunteers) where 63% of drugs were BCS class 2 DRBA products (110). ...
Full-text available
Generally, bioequivalence (BE) studies of drug products for pediatric patients are conducted in adults due to ethical reasons. Given the lack of direct BE assessment in pediatric populations, the aim of this work is to develop a database of BE and relative bioavailability (relative BA) studies conducted in pediatric populations and to enable the identification of risk factors associated with certain drug substances or products that may lead to failed BE or different pharmacokinetic (PK) parameters in relative BA studies in pediatrics. A literature search from 1965 to 2020 was conducted in PubMed, Cochrane Library, and Google Scholar to identify BE studies conducted in pediatric populations and relative BA studies conducted in pediatric populations. Overall, 79 studies covering 37 active pharmaceutical ingredients (APIs) were included in the database: 4 bioequivalence studies with data that passed BE evaluations; 2 studies showed bioinequivalence results; 34 relative BA studies showing comparable PK parameters, and 39 relative BA studies showing differences in PK parameters between test and reference products. Based on the above studies, common putative risk factors associated with differences in relative bioavailability (DRBA) in pediatric populations include age-related absorption effects, high inter-individual variability, and poor study design. A database containing 79 clinical studies on BE or relative BA in pediatrics has been developed. Putative risk factors associated with DRBA in pediatric populations are summarized.
... Pharmacokinetics of silybin in rats was altered (AUC and Cmax were increased) through co-administration of tangeretin, and moreover, the protective effect of silybin against liver damage, in the presence of tangeretin, was significantly enhanced [15,20]. Taking into account the low solubility (lower than 20 μg/ml in the water at 35 • C) [21]), high gut-wall efflux and presystemic metabolism, silymarin should be considered as class IV drug in accordance with biopharmaceutics drug disposition classification system [22]. ...
Silymarin is a mixture of flavonolignans obtained from the seeds of milk thistle (Silybum marianum L. Gaertner). Silymarin behaves as a weak acid and is categorised as a class IV drug substance in accordance with biopharmaceutics drug disposition classification system, possessing low solubility, as well as low bioavailability. The scope of this study was to identify possible formulation strategies of silymarin. Then, the main aim was to manufacture silymarin solid dispersions using a solvent evaporation approach and to characterise the physicochemical and drug release properties of the two formulations containing two different porous carriers, namely Avicel® PH-102 and Syloid® XDP 3150, and different concentrations of Tween® 80. Silymarin Log P was determined to be 1.6 (±0.14) negating the possibility of bypassing first pass metabolism via lymphatic transport. Utilising alkaline titration, the apparent pKa of silymarin was found to be similar to that of the silybin pKa (5.68). The crystallinity of raw silymarin was confirmed using powder X-ray diffraction and differential scanning calorimetry, and its thermal degradation was observed at a temperature higher than 220°C (thermogravimetric analysis). Avicel® PH-102 and Syloid® XDP 3150 were characterised in terms of morphology using scanning electron microscopy, particle size distribution (laser diffraction spectroscopy), pore size distribution and intra-particle porosity using mercury intrusion porosimetry. Solid dispersions were manufactured using an organic solvent method incorporating silymarin, the carrier and optionally Tween® 80. The amorphous state of silymarin in all prepared formulations was confirmed using differential scanning calorimetry and powder X-ray diffraction. Silymarin dissolution kinetics were faster for Syloid® XDP 3150 versus Avicel® PH-102 and explained through carrier properties. The addition of Tween® 80 and increasing the concentration from 0.3 to 1.6% (w/w) significantly increased the drug release kinetics of Avicel® PH-102 formulations but had no effect on Syloid® XDP 3150 formulations. Drug release from prepared formulations was compared with Legalon® 70 using the similarity factor (F2). Syloid® XDP 3150-based formulations showed F2>50%. Tween® 80 had a negligible effect on the silymarin release from Syloid® XDP 3150-based formulations. Interestingly, the ability of Tween® 80 to inhibit gut wall efflux is well known. Thus, the inclusion of this excipient offers an opportunity to modulate the silymarin bioavailability without changing the drug release profile. A six-month stability study (at room temperature and 40% RH) confirmed that solid dispersions were still powder X-ray diffraction and differential scanning calorimetry amorphous. Acetone was used for both silymarin extraction and preparation of solid dispersions. Thus, there is an opportunity to use a single step to both load silymarin and form solid dispersions within a single-step.
As a drug advances through the late stages of clinical development, formulation changes are common to meet clinical, manufacturing, and/or business needs. Since some formulation changes may alter in vivo drug absorption, it is critical to understand the impact of these changes on in vivo PK performances to support the transition between pre- and post-change formulations and ensure the drug’s efficacy and safety. While clinical RBA/BE studies are time-consuming and expensive, other formulation bridging approaches that bring opportunities to expedite drug development by waiving clinical formulation bridging studies are summarized. This review discussed the current formulation bridging options based on in vitro dissolution, physiologically-based biopharmaceutics modeling (PBBM), in vitro – in vivo correlation (IVIVC), and risk-based assessment during the early and late stages of clinical development. By increasing the understanding of the opportunities and challenges associated with different formulation bridging approaches, this review helps with the selection/design of formulation bridging studies in a phase appropriate manner for formulation change during product development.
The objective of the present study was to develop a physiologically based biopharmaceutics (PBBM) approach to predict the bioequivalence of dosage forms containing poorly soluble drugs. Aripiprazole and enzalutamide were used as model drugs. Variations in the gastrointestinal (GI) physiological parameters of fasted humans were taken into consideration in in vitro biorelevant dissolution testing and in an in silico PBBM simulations. To estimate bioequivalence between dosage forms, the inter-individual variabilities in their performance in virtual human subjects were predicted from the in vitro studies and variability in e.g. gastric emptying and fluid volume in the stomach was also taken into account. Formulations with different in vitro dissolution performance, a solution and a tablet formulation, were used in order to evaluate the accuracy of bioequivalence prediction using the PBBM approach. The bioequivalence parameters, i.e. geometric mean ratio and 90% confidence interval, for both drugs were predicted well in the virtual studies. In order to achieve even more precise predictions, it will be important to continue characterizing GI physiological parameters, along with their variabilities, on both an inter-subject and inter-occasion basis.
Despite having adequate solubility properties, bioequivalence (BE) studies performed on immediate release formulations containing BCS1/3 drugs occasionally fail. By systematically evaluating a set of 17 soluble drugs where unexpected BE failures have been reported and comparing to a set of 29 drugs where no such reports have been documented, a broad assessment of the risk factors leading to BE failure was performed. BE failures for BCS1/3 drugs were predominantly related to changes in Cmax rather than AUC. Cmax changes were typically modest, with minimal clinical significance for most drugs. Overall, drugs with a sharp plasma peak were identified as a key factor in BE failure risk. A new pharmacokinetic term (t½Cmax) is proposed to identify drugs at higher risk due to their peak plasma profile shape. In addition, the analysis revealed that weak acids, and drugs with particularly high gastric solubility are potentially more vulnerable to BE failure, particularly when these features are combined with a sharp Cmax peak. BCS3 drugs, which are often characterised as being more vulnerable to BE failure due to their potential for permeation and transit to be altered, particularly by excipient change, were not in general at greater risk of BE failures. These findings will help to inform how biowaivers may be optimally applied in the future.
Full-text available
A theoretical approach for estimating fraction dose absorbed in humans has been developed based on a macroscopic mass balance that incorporates membrane permeability and solubility considerations. The macroscopic mass balance approach (MMBA) is a flow model approach that utilizes fundamental mass transfer theory for estimating the extent of absorption for passively as well as nonpassively absorbed drugs. The mass balance on a tube with steady input and a wall flux of J_w = P_wC_b results in the following expression for fraction dose absorbed, F: F = 2 An ∫_0^1C*_b dz* where the absorption number, An = L/R · P_w/〈v_z〉, L and R are the intestinal length and radius, P_w is the unbiased drug wall permeability, 〈v_z〉 is the axial fluid velocity, C*_b = C_b/C_o and is the dimension-less bulk or lumen drug concentration, C_b and C_o are the bulk and initial drug concentrations, respectively, and z* is the fractional intestinal length and is equal to z/L. Three theoretical cases are considered: (I) C_o ≤ S, C_m ≤ S, (II) C_o > S, C_m ≤ S, and (III) C_o > S, C_m > S, where S is the drug solubility and C_m is the outlet drug concentration. Solving the general steady-state mass balance result for fraction dose absorbed using the mixing tank (MT) and complete radial mixing (CRM) models results in the expressions for the fraction dose absorbed in humans. Two previously published empirical correlations for estimating fraction dose absorbed in humans are discussed and shown to follow as special cases of this theoretical approach. The MMBA is also applied to amoxicillin, a commonly prescribed orally absorbed -lactam antibiotic for several doses. The parameters used in the correlation were determined from in situ or in vitro experiments along with a calculated system scaling parameter. The fraction dose absorbed calculated using the MMBA is compared to human amoxicillin pharmacokinetic results from the literature with initial doses approximated to be both above and below its solubility. The results of the MMBA correlation are discussed with respect to the nonpassive absorption mechanism and solubility limitation of amoxicillin. The MMBA is shown to be a fundamental, theoretically based model for estimating fraction dose absorbed in humans from in situ and in vitro parameters from which previously published empirical correlations follow as special cases.
Several existing unconditional methods for setting confidence intervals for the difference between binomial proportions are evaluated. Computationally simpler methods are prone to a variety of aberrations and poor coverage properties. The closely interrelated methods of Mee and Miettinen and Nurminen perform well but require a computer program. Two new approaches which also avoid aberrations are developed and evaluated. A tail area profile likelihood based method produces the best coverage properties, but is difficult to calculate for large denominators. A method combining Wilson score intervals for the two proportions to be compared also performs well, and is readily implemented irrespective of sample size. © 1998 John Wiley & Sons, Ltd.
The evolution of the complex metabolic interaction between intestinal microbiota in the human gut with its host is multidimensional. Our understanding of this complex interaction has evolved in the past years either with the use of more sophisticated analytical techniques or by reported adverse drug effects that have been associated with intestinal drug metabolism such as with sorivudine. The composition of the intestinal microbiome is initially determined by environmental and genetic factors although external influences as well as host immune reactions provide for adjustment of the delicate balance in both health and disease conditions. The metabolism of drugs by both intestinal bacteria and further by enterocytes leading to their systemic absorption deserves further attention and may provide valuable insights into pre-systemic drug metabolism, delivery, and toxicity. A better understanding of the metabolic pathways may aid in the drug development and toxicity evaluation process.
Literature and experimental data relevant to the decision to allow a waiver of in vivo bioequivalence (BE) testing for the approval of immediate-release (IR) solid oral dosage forms containing ketoprofen are reviewed. Ketoprofen's solubility and permeability, its therapeutic use and therapeutic index, pharmacokinetic properties, data related to the possibility of excipient interactions, and reported BE/bioavailability (BA)/dissolution data were taken into consideration. The available data suggest that according to the current Biopharmaceutics Classification System (BCS) and all current guidances, ketoprofen is a weak acid that would be assigned to BCS Class II. The extent of ketoprofen absorption seems not to depend on formulation or excipients, so the risk of bioinequivalence in terms of area under the curve is very low, but the rate of absorption (i.e., BE in terms of peak plasma concentration, Cmax) can be altered by formulation. Current in vitro dissolution methods may not always reflect differences in terms of Cmax for BCS Class II weak acids; however, such differences in absorption rate are acceptable for ketoprofen with respect to patient risks. As ketoprofen products may be taken before or after meals, the rate of absorption cannot be considered crucial to drug action. Therefore, a biowaiver for IR ketoprofen solid oral dosage form is considered feasible, provided that (a) the test product contains only excipients present also in IR solid oral drug products containing ketoprofen, which are approved in International Conference on Harmonisation or associated countries, for instance, as presented in this paper; (b) both the test drug product and the comparator dissolve 85% in 30min or less in pH 6.8 buffer; and (c) test product and comparator show dissolution profile similarity in pH 1.2, 4.5, and 6.8. When one or more of these conditions are not fulfilled, BE should be established in vivo.
The aim was to assess dynamic and static parameters on routine computed tomography pulmonary angiography (CTPA) that may detect pulmonary hypertension (PH). Fifty patients underwent CTPA and echocardiograms. Twenty-six patients had PH, and 24 patients did not have PH. The following parameters were measured on CTPA: density of the pulmonary artery (PA), ratio between the density in the PA and the thoracic aorta (TA), the time between the start of contrast injection to the time the scan trigger density was reached, and PA diameter. All measured parameters showed significant correlation with PH detected by echocardiogram. The best combination of parameters for detection of PH was contrast density ratio between PA and thoracic aorta of greater than or equal to 1.5 and/or a time to scan trigger of greater than or equal to 8 seconds. The parameters measured correlate well with PH by echocardiography. This suggests that CTPA can potentially be used to detect PH.
Binary-Scale Data Ordinal- and Continuous-Scale Data Tests of Equivalence
Sensitivity and Specificity Combined Measures of Sensitivity and Specificity Receiver Operating Characteristic (ROC) Curve Area Under the ROC Curve Sensitivity at Fixed FPR Partial Area Under the ROC Curve Likelihood Ratios ROC Analysis When the True Diagnosis Is Not Binary C-Statistics and Other Measures to Compare Prediction Models Detection and Localization of Multiple Lesions Positive and Negative Predictive Values, Bayes Theorem, and Case Study 2 Optimal Decision Threshold on the ROC Curve Interpreting the Results of Multiple Tests
The simplest diagnostic test is one where the results of an investigation, such as an x ray examination or biopsy, are used to classify patients into two groups according to the presence or absence of a symptom or sign. For example, the table shows the relation between the results of a test, a liver scan, and the correct diagnosis based on either necropsy, biopsy, or surgical inspection.1 How good is the liver scan at diagnosis of abnormal pathology?View this table:View PopupView InlineRelation between results of liver scan and correct diagnosis1One approach is to calculate the proportions of patients with normal and abnormal liver scans who are correctly “diagnosed” by the scan. The terms positive and negative are used to refer to the presence or absence of the condition of interest, here abnormal pathology. Thus there are 258 true positives and 86 true negatives. The proportions of these two groups that were correctly diagnosed by the scan were 231/258=0.90 and 54/86=0.63 respectively. These two proportions have confusingly similar names.Sensitivity is the proportion of true positives that are correctly identified by the test.Specificity is the proportion of true negatives that are correctly identified by the test.We can thus say that, based on the sample studied, we would expect 90% of patients with abnormal pathology to have abnormal (positive) liver scans, while 63% of those with normal pathology would have normal (negative) liver scans.The sensitivity and specificity are proportions, so confidence intervals can be calculated for them using standard methods for proportions.2Sensitivity and specificity are one approach to quantifying the diagnostic ability of the test. In clinical practice, however, the test result is all that is known, so we want to know how good the test is at predicting abnormality. In other words, what proportion of patients with abnormal test results are truly abnormal? This question is addressed in a subsequent note.References↵Drum DE, Christacapoulos JS.Hepatic scintigraphy in clinical decision making.J Nucl Med1972;13: 908–15.OpenUrlFREE Full Text↵Gardner MJ, Altman DGGardner MJ, Altman DG.Calculating confidence intervals for proportions and their differences. In: Gardner MJ, Altman DG eds.Statistics with confidence.London: BMJ Publishing Group,1989: 28–33.