CARDIOVASCULAR JOURNAL OF AFRICA • Vol 21, No 2, March/April 2010
AFRICA
69
Editorial
Bioequivalence: tried and tested
Two drug products are considered bioequivalent ‘if their bio-
availabilities ... are similar to such a degree that their effects,
with respect to both efficacy and safety, will essentially be the
same’.
1
The bioequivalence of two drug products is generally
demonstrated through a clinical study in healthy volunteers, the
so-called bioequivalence study. If bioequivalence is shown for
two drug products, therapeutic equivalence of the drug products
is implied. Chow and Liu
2
call this assumption, namely that
bioequivalence implies therapeutic equivalence, the ‘fundamen-
tal bioequivalence assumption’.
Most drug products on the market today have been subjected
to bioequivalence assessment at various stages in their devel-
opment. As is well known, generic drug products require the
demonstration of bioequivalence to the relevant innovator prod-
uct for regulatory approval. What is perhaps less well known
is that most innovator products, too, require some form of
bioequivalence testing. New drugs typically undergo pharma-
cokinetic dose-proportionality studies, and drug–drug and drug–
food interaction studies, all of which use the bioequivalence
concept. The site of development and production of the drug
product could be changed. Most importantly, when the innova-
tor formulation to be marketed is different from the formulation
used previously in pivotal efficacy trials, as is often the case,
bioequivalence of the marketed formulation to the clinical trial
formulation must be shown. In this sense, many innovator drug
products on the market are in fact ‘generic copies’ of the clinical
trial formulation for which therapeutic efficacy and safety had
been shown in patients.
Consumers of drug products, therefore, of both generic and
innovator products, need assurance on the question whether
bioequivalence implies therapeutic equivalence. All drug manu-
facturers of either generic or innovator products need to know
whether the bioequivalence concept and bioequivalence meth-
odology serve their products well during development. On both
these questions relatively recent developments have shed some
light.
History of the bioequivalence concept
Public concern and ongoing discussion about bioequivalence
started in the early 1970s with reports about digoxin intoxica-
tions. At the time, generic digoxin formulations were increasingly
prescribed in the United States, and a change in the manufactur-
ing process of a company in Great Britain led to an unintentional
increase in the bioavailability of one brand of digoxin tables.
3,4
It became clear that drug products that are pharmaceutically
equivalent, that is, products that contain the same drug in the
same dose, are not necessarily bioequivalent.
5
Over the years, various regulatory guidelines on the design,
conduct and statistical analysis of bioequivalence studies have
been published. Many years of research, discussion and contro-
versy culminated in the seminal Food and Drug Administration’s
(FDA) 1992 guidance
6
on the statistical analysis of bioequivalence
studies. That guidance established such well-known concepts as
the pharmacokinetic characteristics for rate and extent of drug
absorption and the statistical decision rule for the demonstration
of bioequivalence (90% confidence interval for the test/refer-
ence ratio of mean bioavailability must fall completely in the
bioequivalence acceptance range of 80–125%.)
Switchability of drug products: the individual
bioequivalence intermezzo
With the publication of the 1992 FDA guidance,
6
one might
have thought that agreement had been reached, finally, among
researchers and regulators on the central concepts of bioequiva-
lence. Ironically, almost exactly around that time, the new
concept of individual bioequivalence
7
was formulated and
sparked a new era of research and discussion, and probably more
controversy than ever before.
Two US biostatisticians, Anderson and Hauck,
7
pointed out
that the traditional way of bioequivalence assessment, as circum-
scribed, for example, in the contemporary 1992 FDA guidance,
ensured merely that the bioavailability of two drug products was
similar (‘equivalent’) on average. They raised the following clini-
cally very relevant question: does equivalence of average bio-
availability, which they termed average bioequivalence, ensure
that the bioavailability of two drug products is equivalent in
individual patients? In other words, does average bioequivalence
imply switchability of drug products in individual patients?
Following the groundbreaking article of Anderson and Hauck,
numerous statistical approaches to individual bioequivalence
were published (Schall
8
provides a unified view of most of the
approaches; see also the reviews of Hauschke, Steinijans and
Pigeot
9
and Chow and Liu
2
). Eventually, in 2001, the individual
bioequivalence concept was adopted in an FDA guidance.
10
However, ‘responses [to the guidance] were doubt-filled as to
whether the new bioequivalence criteria really provided added
value compared to average bioequivalence’.
9
Crucially and rather illuminating on the question of the
general validity of the fundamental bioequivalence assumption
was the observation that ‘there has been no published evidence
of clinical failure with a formulation demonstrated to be equiva-
lent to the reference product under average bioequivalence’.
11,12
Individual bioequivalence was called a ‘theoretical’ solution to
a ‘theoretical’ problem.
13
In response to the widespread criticism
and doubts, in 2003 the FDA omitted the individual bioequiva-
lence concept from a subsequent guidance,
14
and returned, almost
full circle, to the conventional (average) bioequivalence concept
of the 1992 guidance.
6
Bioequivalence and therapeutic equivalence
The concept of individual bioequivalence proved to be an inter-
