Quite apart from these general trends blurring
the distinctions between phases I, II and III, there
are (and always have been) sound medical or phar-
macological reasons for doing so. Good examples
might be the following:
It would be unreasonable to study the pharma-
cokinetics of relatively toxic agents, at poten-
tially therapeutic doses, in normal volunteers
due to the near-certainty of the adverse events.
Typically, this information can be gained in
patients with diseases potentially responsive to
these agents. Thus, the first-in-man studies in
this case are ‘phase II’, using the classic nomen-
clature. Cytotoxic and antiviral drugs are two
important classes of agent where this is com-
monly the case.
There is little point in testing the tolerability of
drugs in normal volunteers, when only patients
with the disease of interest are able to demon-
strate a relevant pharmacodynamic effect. The
doses at which tolerability must be confirmed are
unknown until the exposure of patients can indi-
cate the doses that may be effective. The devel-
opment of potent opioids such as alfentanil,
sufentanil and remifentanil as anesthetic agents
are a good example.
There are some diseases which have neither ani-
mal model nor relevant pharmacodynamic or sur-
rogate end point in normal volunteers. Such
diseases may also alter the pharmacokinetics of
the drug, thus invalidating anything that might be
learned from normal volunteers. A good example
is the migraine syndrome. No animal species has
migraine, and normal volunteers cannot report an
anti-migraine effect. Nausea, vomiting and gas-
tric stasis are common during migraine attacks
and may be expected to alter the pharmacoki-
netics and effectiveness of oral therapies.
There is nonetheless little hope that the phase I–III
aphorism will die. Nevertheless, it is quitewrong to
assume that these ‘classical’ terms and definitions
still apply to how drugs are developed according to
modern practice. The classical four-phase strategy
of drug development is far too stereotyped,
simplistic and pedestrian to have survived into
the modern era of drug development. None of
today’s successful companies actually use such a
strategy. We are simply shackled with an outmoded
terminology.
9.2 Concepts of bias and
statistical necessities
Bias is a general consideration in clinical trial
design, regardless of the type of trial being con-
ducted. It is considered here as an overarching
issue, to be applied to the systematic description
of the types of study design considered below.
The word bias has many definitions, but in this
context, it is best described as a distortion of, or
prejudice toward, observed effects that may or may
not truly be due to the action of the test drug(s).
Many things can distort the true measurement of
drug action, and bias is the trialist’s most unremit-
ting enemy. This enemy comes from many quarters
(Table 9.1). The clinical trialist must be sufficiently
humble to realize that he or she, himself or herself,
may be a source of bias.
The pharmaceutical physician may not be
expected to be a specialist statistician, and statistics
are not the subject of this chapter. However, the
ability to talk to and understand statisticians is
absolutely essential. Sine qua non: Involve a
good statistician from the moment a clinical trial
is contemplated. Furthermore, the pharmaceutical
physician should be confident of a sound under-
standing of the concepts of type I and type II error,
and the probabilitiesaandb(e.g. Freimanet al.,
1978). This is one of your best defences against
bias.
9.3 Prospective definitions: the
only way to interpret what
you measure
It does not require a training in advanced statistics
to hold a common sense and accurate approach to
creating clinical hypotheses, translate theminto the
precise quantities of a measured end point and then
102 CH9 PHASE II AND PHASE III CLINICAL STUDIES