Phase 3 clinical trial set up

Phase 3 studies are typically double-blind, randomized trials. If an effective treatment already exists for the targeted intervention, the study should be comparator-controlled.

Indication

The indication (target pathogen(s), site of infection, health condition) targeted in a phase 3 trial will almost always be the same as the one targeted at phase 2. For each indication, regulatory authorities publish guidance on key aspects of phase 3 study design, including the appropriate definition of the indication, the expected patient population, study endpoints and timepoints, and non-inferiority margin.

For a given indication, guidance from regulatory authorities may be unavailable or not recently updated. For this reason, regulatory precedent is also important. Information on recently approved drugs is published by the European Medicines Agency (EMA) (European Public Assessment Reports, EPARs) and by the US Food and Drug Administration (FDA) (Summary Basis of Approval, SBOA) documents.

Similarly, the treatment regimen to be evaluated (i.e. whether an investigational drug is used in combination with existing therapies) will already have been defined at phase 2.

Comparator/control group

Although a placebo-controlled trial provides the most information on efficacy, a placebo control is not acceptable if an effective treatment already exists.

Patients must be randomized to receive either an antibiotic known to be effective for treatment of their infection (the standard treatment or standard of care) or the investigational drug. Dose, dosing regimen, and infusion rate (for intravenous drugs) for the comparator should be consistent with its labelling.

Use of an “active comparator” is important for optimizing the benefit/risk trade-off for trial participants. In addition, this design provides regulators, physicians and payors with valuable information about the comparative efficacy and safety of a new drug, and how its characteristics differ from existing treatments.

For some less serious infections, including sinusitis and bronchitis, the FDA requires placebo-controlled trials.

Minimizing risk of bias

Ideally, phase 3 trials should be double-blind – neither clinician nor patient knows which treatment is being given. This will ensure that clinical assessments are not biased by an investigator knowing the treatment to which a subject was randomized.

If the treatment regimens for the investigational drug and comparator are too different to permit blinding, mechanisms should be put in place such that the physician making clinical assessments remains blinded. This is typically done by identifying a designated investigator for study evaluations, who does not administer the initial treatment. Because of the added complexity of this approach, the treatment regimen of a comparator may also be considered in defining the comparator for a trial.

Demonstrating efficacy

In general, trials must be statistically powered to meet regulatory guidelines to demonstrate non-inferiority to the comparator, an appropriate approved antibiotic for the given indication – the trial must show that the efficacy of a new treatment does not fall a certain percentage (the non-inferiority margin) below the efficacy of the existing antibiotic.

Trial size

The size of a phase 3 trial is dictated by the need to achieve statistical power to demonstrate efficacy, while minimizing sample size to reduce the cost and length of a trial.

The optimal size of a trial varies across indications. In a non-inferiority trial, it will also depend on the size of the non-inferiority margin. The nature of a development programme is also significant:

For a standard development programme for initial therapy of a given indication, non-inferiority margins are likely to be strict, calling for a larger sample size.
For streamlined development to support an indication for patients with no or few available treatment options, larger non-inferiority margins may be acceptable (but must be agreed with the FDA); smaller sample sizes may then be feasible.

Trial size is also dependent on the need to obtain sufficient safety data. If broad use of a drug is anticipated, safety data on around 1500 patients is likely to be required; if the drug is for more restricted use, data from a minimum of 300 patients may be sufficient.

Statistical analysis

A statistical analysis plan (SAP) must be developed before data collection begins. Statistical analyses must be agreed in advance, so that analyses of efficacy are not influenced by results at the end of the trial. Developing a statistical analysis plan requires high levels of statistical expertise, particularly for complex trial designs such as adaptive trials.

A statistical analysis plan must be consistent with the clinical trial protocol and is often included as part of the protocol. It forms part of the clinical trial application and the final licensing submission.

Interim analyses

An interim analysis is a planned analysis of data at a specified point part-way through a trial. Often this is when 50% of participants have been followed for an agreed period.

Interim analyses are used to assess whether the data support the continuation of the trial. A trial may be halted early if:

There is already strong, statistically robust evidence of the safety and efficacy of a new treatment in a placebo-controlled trial: in this case, it would be unethical for trial participants to be randomized to treatment or placebo.
There is evidence of harm in a treatment arm in excess of that seen in a comparator of the trial; a trial may be paused temporarily while adverse events are investigated to determine whether they are associated with new drug use, or permanently if adverse events are found to be linked to new drug use and are seen as sufficiently serious.
There is no chance that efficacy will be demonstrated even if further patients are recruited (termination due to futility).

Interim analyses may also lead to modification in the design or conduct of a trial. However, any changes must have been pre-specified in a trial protocol or must be agreed by trial regulators through a protocol amendment. Interim analyses are particularly important in adaptive trials, representing points at which decisions are made on how a trial will be adapted in light of data collection to date.

It is highly desirable for interim analyses to be carried out by a group independent of the team running a trial, to avoid any potential for bias in interpretation of data. Usually, interim analyses are the responsibility of an independent data and safety monitoring board (DSMB), which makes recommendations on continuation or modification of the trial.

Drug-resistant pathogens

Due to the need to ensure that participants receive effective therapy, phase 3 trials typically enroll patients with drug-susceptible or readily treatable drug-resistant infections. If an infection is found to be caused by bacteria with resistance mechanisms for which there are few (or no) effective antibiotics available, a patient is generally excluded from the trial.

Patients whose infections are due to pathogens with reduced susceptibility (i.e. a high minimum inhibitory concentration (MIC) value), which might be considered as non-susceptible or resistant to the comparator, may be retained in the trial if they are clinically improving. However, the approach to be taken must be specified in advance in the study protocol.

Separate trials are usually organized to collect efficacy data on pathogens with specific, defined resistance mechanisms. These are known as resistant pathogen or target pathogen studies. Guidance on the design of these studies is provided by regulatory authorities such as the FDA.