Clinical Pharmacology and Therapeutics

CLINICAL DRUG DEVELOPMENT 89
Too many statistical
comparisons performed
differences should be reported as confidence intervals (usually
95% confidence intervals). Such intervals will diminish as
the sample size is increased. Confidence intervals reflect the
effects of sampling variability on the precision of a procedure,
and it is important to quote them when a ‘non-significant’
result is obtained, and when comparing different estimates of
effectiveness (e.g. drug A in one trial may have performed
twice as well as placebo, whereas drug B in another trial may
have performed only 1.5 times as well as placebo; whether
drug A is probably superior to drug B will be apparent from
inspection of the two sets of confidence intervals).
If many parameters are analysed, some apparently ‘significant’
differences will be identified by chance. For example, if
100 parameters are analysed in a comparison of two treatments,
one would expect to see a ‘significant’ difference in
approximately five of those parameters. It is therefore very
important to prespecify the primary trial end-point and secondary
end-points that will be analysed. Statistical corrections
can be applied to allow for the number of comparisons made.
One must also consider the clinical importance of any statistically
significant result. For example, a drug may cause a statistically
significant decrease in blood pressure in a study, but if
it is only 0.2 mmHg it is not of any clinical relevance.
CLINICAL DRUG DEVELOPMENT
For most new drugs, the development process – following a
satisfactory preclinical safety evaluation – proceeds through
four distinct phases. These are summarized below. Figure 15.3
illustrates the overall decision-making process for determining
whether or not a new therapy will be clinically useful.
PHASE I
TYPE I ERROR
False-positive result
(significant difference found
when no difference present)
Too small sample size
(insufficient power)
Figure 15.2: Different types of statistical error.
TYPE II ERROR
False-negative result
(no significant difference found
when difference present)
The initial studies of drugs in humans usually involve healthy
male volunteers unless toxicity is predictable (e.g. cytotoxic
agents, murine monoclonal antibodies). The first dose to be
administered to humans is usually a fraction of the dose that
produced any effect in the most sensitive animal species
tested. Subjective adverse events, clinical signs, haematology,
biochemistry, urinalysis and electrocardiography are used to
assess tolerability. Depending on the preclinical data, further,
more specific evaluations may be appropriate. The studies are
placebo controlled to reduce the influence of environment and
normal variability. If the dose is well tolerated, a higher dose
will be administered either to a different subject in a parallel
design, or to the same group in an incremented crossover
design.
This process is repeated until some predefined end-point
such as a particular plasma concentration, a pharmacodynamic
effect or maximum tolerated dose is reached. Data from
the single-dose study will determine appropriate doses and
dose intervals for subsequent multiple-dose studies. If the
drug is administered by mouth, a food interaction study
should be conducted before multiple-dose studies.
The multiple-dose study provides further opportunity for
pharmacodynamic assessments, which may demonstrate a
desired pharmacological effect and are often crucial for the
selection of doses for phase II. Having established the dose
range that is well tolerated by healthy subjects, and in some
cases identified doses that produce the desired pharmacological
effect, the phase II studies are initiated.
Key points
Phase I studies:
• initial exposure of humans to investigational drug;
• assessment of tolerance, pharmacokinetics and
pharmacodynamics in healthy subjects or patients;
• usually healthy male volunteers;
• usually single site;
• 40–100 subjects in total.
PHASE II
Phase II studies are usually conducted in a small number of
patients by specialists in the appropriate area to explore efficacy,
tolerance and the dose–response relationship. If it is ethical and
practicable, a double-blind design is used, employing either a
placebo control or a standard reference drug therapy as control.
These are the first studies in the target population, and it
is possible that drug effects, including adverse drug reactions
and pharmacokinetics, may be different to those observed in
the healthy subjects. If the exploratory phase II studies are
promising, larger phase III studies are instigated, using a
dosage regimen defined on the basis of the phase II studies.
Key points
Phase II studies:
• initial assessment of tolerance in ‘target’ population;
• initial assessment of efficacy;
• identification of doses for phase III studies;
• well controlled with a narrowly defined patient
population;
• 100–300 patients in total;
• usually double-blind, randomized and controlled.
PHASE III
Phase III is the phase of large-scale formal clinical trials in which
the efficacy and tolerability of the new drug is established.