Phases of clinical trials – 0, 1, 2, 3, 4
14.3.1 Phase 0
Phase 0 trials are preclinical trials and are intended to expedite the clinical evaluation of new molecular entities. They are often performed in animals. The aim is to look for a dose–response pattern and gather information to guide transition into phase I.
14.3.2 Phase I
After demonstrating safety in animals, the next step is to document safety in humans. Stage I trials are usually small, and the study population may be comprised of volunteers who are in good health or those for whom there is no effective treatment (e.g., metastatic disease). Phase I trials aim to establish safety and perform pharmacokinetic studies that help to determine maximum tolerated dose.
14.3.3 Phase II
The purpose of phase II trials is to determine efficacy, i.e., whether the treatment at a particular dose/regimen is effective in ideal situations. They determine therapeutic activity and further evaluate toxicity and/or side effects.
14.3.4 Phase III
After demonstrating that a given drug is safe and has a reasonable chance of improving or affecting a disease, phase III trials are undertaken to demonstrate a new drug or treatment is superior to current standard of care. These are usually randomized controlled clinical trials, with large numbers of patients. If only one drug is being evaluated, the design is often placebo controlled and done in parallel. If two or more drugs are being evaluated, a factorial or crossover design is also possible . See Fig. 14.2.
Clinical trial design – parallel, crossover, and factorial
Parallel–group design is a simple design in which each participant is randomly assigned to receive a particular treatment or not. Multiple drugs or treatments may be tested versus each other, and there is often a control or placebo group. Different doses of the same drug can also be simultaneously evaluated. This design is simple and eliminates the potential for drug interaction.
In a crossover design, in general, two treatments are evaluated. Each group receives two treatments but the order is randomly assigned. Prior to the crossover, the design appears to be similar to a parallel-group design. However, the crossover allows one to determine the benefit of adding one drug/treatment to another in a sequential fashion and which order is most beneficial. This design is appropriate for chronic conditions that are stable over time and for interventions that last a short time. It is important that the treatment drugs do not react with each other to avoid making inaccurate conclusions about the results. It should be noted that crossover designs have also been used to evaluate timing of surgical treatments; for example, in the NSABP B-18 trial, patients were randomized to receive either (neoadjuvant) chemotherapy followed by surgery or surgery followed by chemotherapy in order whether one sequence was superior to the other.
In a factorial design, two drugs or interventions can be simultaneously evaluated. With two drugs, four combinations of treatments and placebo are possible. Patients are randomly assigned to each group. For example, one group might receive drug A and drug B. Another group would receive drug A and placebo. Another would receive drug B and placebo, and another would receive two placebos. The factorial design can be very efficient, as data is gathered on two drugs at the same time. A drawback of the factorial design is the concern over potential drug interactions; however, this design also allows for the determination of synergies between two treatments.
14.3.5 Phase IV
Phase IV studies are post-marketing studies that are conducted after the drug has been approved for use. The purpose of phase IV studies is to gather information regarding additional potential side effects in diverse populations and gather long-term follow-up data.
For surgical trials of new equipment or implants, a parallel series of trials has been proposed .
14.3.6 Phase I: Laboratory Study
This phase, similar to phase 0 trials for drug, is preclinical and often involves animal models.
14.3.7 Phase II: Cohort Study
As devices and implants do not have a “dose” that needs to be titrated, this phase is akin to phase II trials where a tightly controlled study is done on a cohort of patients to determine efficacy of a particular device or treatment.
14.3.8 Phase III: Randomized Controlled Trial
As with other phase III randomized controlled trials, the purpose of these studies is to compare outcomes with a particular device, implant, or treatment vs. standard of care.
14.3.9 Phase IV: Surveillance Study
Again, careful evaluation must continue for long-term sequelae after a phase III randomized controlled trial; for devices or surgical implants, this is similar to phase IV trials for drugs.
The gold standard to determine if one intervention is better than another is the randomized controlled trial, which provides the strongest evidence for a cause-effect association between an intervention and an outcome. The process of randomization effectively limits bias, such that differences in outcome may be attributable to differences in treatment groups, rather than to inherent differences or confounding factors between the two groups [5, 6].
Several randomization strategies exist. Simple randomization is analogous to a coin toss; heads means the patient will be randomized to the treatment arm, and tails means they will be in the control group. Randomization tables can simply allocate patients to one arm or another in whatever ratio the trial dictates. Another (slightly more complicated) randomization strategy is block randomization, where patients are divided into blocks (which are often based on some characteristic) and randomized within each block. This is also called stratified randomization, as patients are initially stratified according to a key variable or set of variables (e.g., age, race, sex, stage) and then randomized within each stratum. In surgical trials, stratifying by surgeon has been used to adjust for the experience of individual surgeons . In very large trials, stratification is usually unnecessary, as large imbalances tend to diminish with greater numbers of subjects.