4 questions in one ........

answer 1  basically no different than other clinical trials 

for some lenghty but good docs on trial design look at the Cochrane reviews and all literature behind all of that from these groups  you wil learn a lot about the pitfalls.

answer 2 depends on how you formulate your hypothesis and how strong your effect is and if any research done by others can be used as a frame of reference in  a convincing manner. One such approach is -using the patient as its own control

In radiotherapy and in oncology this is only seldom done but in other branches of medicine it is done.

There is no law against it but you should be able to defend your trial design amongst peers. A one armed study will be difficult te get approval for.

answer 3  see answer 2

answer 4 re: sample size calculations: there exists no universal recipe.

It depends on your hypothesis , your design, the knowledge you already have on any effect, the [subjectively choosen] level of significance     the [subjectively choosen] required power.

I can understand you might think my answer is rather vague, but say we look back at the time penicillin was invented. Severe pneumonia would kill at least 60%  and penicillin reduce that numer to below 10%.

If we then had to apply our present knowledge on statistics and methodology and we had already a few case reports of survivors, a one armed trial would have been quite ethical.

In oncology similar examples may be found i.e. by refering to cases where drugs were used in a palliative setting but turned out surprisingly effective.

4 questions in one ........

answer 1  basically no different than other clinical trials 

for some lenghty but good docs on trial design look at the Cochrane reviews and all literature behind all of that from these groups  you wil learn a lot about the pitfalls.

answer 2 depends on how you formulate your hypothesis and how strong your effect is and if any research done by others can be used as a frame of reference in  a convincing manner. One such approach is -using the patient as its own control

In radiotherapy and in oncology this is only seldom done but in other branches of medicine it is done.

There is no law against it but you should be able to defend your trial design amongst peers. A one armed study will be difficult te get approval for.

answer 3  see answer 2

answer 4 re: sample size calculations: there exists no universal recipe.

It depends on your hypothesis , your design, the knowledge you already have on any effect, the [subjectively choosen] level of significance     the [subjectively choosen] required power.

I can understand you might think my answer is rather vague, but say we look back at the time penicillin was invented. Severe pneumonia would kill at least 60%  and penicillin reduce that numer to below 10%.

If we then had to apply our present knowledge on statistics and methodology and we had already a few case reports of survivors, a one armed trial would have been quite ethical.

In oncology similar examples may be found i.e. by refering to cases where drugs were used in a palliative setting but turned out surprisingly effective.

There is not a need of a comparison in phase 2 trials; the main aim of these trials is to observe the safety and the efficacy of a new treatment (or drug) with a recommended dose calculated from phase 1 trial in Oncology. Of course, you can do a phase 2 trial with a single arm. But, randomized phase 2 trials exist also. 

safety and efficacy are to be compared, these are not absolute safety and absolute efficacy

Great discussion.

I was thinking that every phase 2 clinical trial must have a comparasion group. However For safety and feasibilty I might can come with single group however comparasion group is always better I suppose.

Comparison is essentially the subject of phase 3 trials. If an investigational drug or treatment is not safe, a phase 3 trial is not necessary. In phase 3 trials, sample size of all treatment arms is extremely important to demonstrate a significant difference between standard old drug (or drugs) and an investigational treatment. The aim of comparison is to show the superiority, inferiority or equivalence. And phase 3 trials have an end-point of efficacy such as overal survival, progression-free survival, objective response rate etc. 

If your results are far superior to wel documented normal practice AND you can ethically justify you choices, there are NO objections to a study with only a single treatment arm.

But you should in advance know very wel how to convince the peers that are going to judge you, both ethically and scientific.

I gave a good example: If you gave drug X [already known for other uses, so with an established safety record] in a clearly palliative setting and much to your suprise, survival increased dramatically in a wel documented cohort, then such a study would stand a fair chance.

But you must do youre homework very well.

Say [hypothetically] you achieve at least 36 months survival in all patients of a cohort of 50 patients with advanced stage cancer of the pancreatic head [inoperable], you would have such a strong case that it would be unethical not to do so.

I have some references:

* 'Guide to clinical studies and developing protocols', Bert Spiker, Raven Press, 1984, ISBN 0-88167-018-9

* 'Fundamentals of clinical trials', Lawrence M. Friedman et al, PSG Publishing Co, 1985. ISBN 0-88416-499-3

* 'Clinical trials in Oncology', S Green et al, CRC, 2012, ISBN 9781439814482

The first two were recomendations by people inside the Upjohn Co around 1988, and the last, I found it by searching for the latest edition of one of the 90 books about the subject: 'Clinical Trials Oncology' I found in abebooks.com

Hope it's useful. Regards. Salut †

If you are performing multiple studies you may add another work of Bert Spilker:

Guide to Planning and Managing Multiple Studies Bert Spilker Raven Press 1987 ISBN 0-88167-264-5 

Another excellent allround reference:

Clinical Trials  Design, Conduct, and Analysis   Curtis L.Meinert  1986 Oxford University Press  ISBN 0-19-503568-2

As we always want to learn as much as possible from as few patients as possible, there is an alternative approach, without fixed sample sizes [there is a fixed uper limit], but with built in stopping rules as early as possible.

The book is tough reading, but if and when you do multiple studies it may be worth the effort, as -on average- [this is no guarantee for individual studies] you can complete trials with about half the patients you would have needed with fixed sample sizes, but it promises more trials and faster results for your research budget. However occasionally for an individual study it may result in doubling the sample size. So use this approach only

The design and analysis of sequential clinical trials.  J.Whitehead 1983

Published by Ellis Horwood Ltd.  and Halsted Press

ISBN 0-85312-404-3

ISBN 0-85312-532-5

ISBN 0-470-27355-0

Phase II trials are planned with prior assumptions (hypothesis) for sample size calculation; they can test efficacy or safety parameters. You can also introduce, for instance, separate boundaries to test two research questions simultaneously, or adopt 2-stage designs to discard early a suboptimal experimental research question. Phase II designs are versatile and fit to support (or not) a future randomised trial. A fellow used to say 'a phase II trial is to convince myself, a phase III to convince other people'.

@ Sandro J Martins    Phase 2 does not preclude studies with a single arm only. It all depends on all other data available. With phase II you obtain data [i.e. efficacy estimates] that are of sufficient quality to help design phase 3 or to quickly eliminate most -would-be- phase 3 studies.

Totally agree with Dr Borm. Phase II trials are mainly single-arm experiments, both for fast recruitment as to reduce their costs. Ideally, every phase III trial should be planned based on previous strong data from phase II trial- when missing this point, probability of a negative phase 3 trial soars as commonly seing in oncology. Hurry comes with its price.   

Pharmaceutical companies are money driven enterprises. For these organizations and their share holders trials serve several purposes, including marketing. Read about the Cochrane projects and you know why trials 'driven by' pharmaceutical companies should unconditionally be distrusted. Actively inducing publication bias is only  the least of their crimes against science and humanity. This is a very, very major problem in medicine. And the publication bias is virtually everywhere. Cooperation with a pharmaceutical company will at times be unavoidable, but one must strongly guard against meddling with trials and reporting. In recent years, the counterpressure against this phenomenum seems to result in more and more trials being conducted outside western europe / USA, often in countries with less controls against such interference.