• Login

• Design
• Clinical trials: an overview
• Trial definitions
• Phase I trials
• Phase II trials
• Phase III trials
• Sample size
• Reference texts

Phase II Trials

Phase II trials are performed to assess the effectiveness and safety of a treatment. They are not designed to quantify treatment effect but to indicate if a treatment affect is present. As a result they are usually fairly small trials (20-40 patients).

Phase II trials for cytostatic drugs

The main aims of phase II trials are to determine the drugs therapeutic effects by observing any anti-tumour activity in specific tumour types. Further information is also gathered on toxicity, methods to manage toxicity and administration of the drug. Patients recruited to such trials will have a specified tumour tyupe. If several drugs or dosing schedules are of interest, then phase II trials may be used as a screening method to identify the moste ffective treatment. For such situations the patients should be randomised into treatment groups to ensure unbiased allocation. These randomised trials are not comparitive trials. The therapeutic effect of a drug is often measured byt the proportion of observed response. A drug is deemed worthy of further investigation (phase III) if the proportion of response is greater than some pre-defined threshold. The conclusion of a trial is subject to two types of error:

• False positive (concluding a drug is effective when it is not)
• False negative (concluding a drug is ineffective when it is)

The conclusion reached is then a trade-off of the two types of error. Allowing weak evidence to be used to demonstrate drug efficacy decreases the probability of a false negative conclusion, but increases the chance of a false positive; accepting onyl strong evidence implies the converse. A decision as to which error is more important needs to be made so that the chance of occurrence can be minimised. Increasing the sample size reduces the probability of an overall erroneous conclusion.

There are three main designs for phase II trials:

• Single stage design:
  This is the simplest design. The sample size is determined before any patients have been recruited using the levels decided for the false positive and negative errors. Once the required nukber of patinets has been recruited, interest is in the total number of responses observed.
  
  There is a need to terminate phase II trials early if there is strong evidence that the treatment is ineffective. This motivation led to the sequential patient accrual designs.

• Single sequential design:
  In this design the sample size is not pre-determines. Instead the decision to continue the trial is made after each new patient is entered, the calculation controls for false positive and negative errors. A consequence of this design is that patient accrual time will generally be slow.
• Multi-stage design:
  This is a mixture of single stage and sequential design. Patients are entered in batches of size n and the decision to continue the trial is taken after each of the accrued batches.

Further details of these designs can be found in J Herson and further statistical detail in L Mariani.

Historically, phase II trials often accrued 14 patients and if there was one or more response the drug was accepted as having potential. This was derived from usinf the first stage of Gehan's two-stage design and shoud not on its own be considered conclusive (see Machin et al.).