How is the experiment performed? As I understood, you have a sample of patients, all having a viral infection and all do need an anti-viral therapy (how is this judged?). Are these patients randomized into two groups: with or without T-cell infusion? Is this done before they get anti-viral therapy? And if so, how long do you wait until you judge (how?) if a patient then still needs anti-viral therapy? Is the "non-infused" (control) group treated similarily (i.e. getting a placebo infusion, withold the anti-viral therapy for the same time), and does the "T-cell infusion" group (treated) recieve a placebo instead of the anti-viral drug?

What is the outcome and how is it measured? A viral titer? Severity of a symptom? Presence of a symptom? Something else? How do you measure the effect of the anti-viral therapy? How would you judge whether or not a T-cell treated patient needs to recieve an anti-viral drug?

Jochen Wilhelm : Thanks for the response. It's a single arm phase II trial (no control group) - using T cells for viral infections in transplant recipients. Current practice is all these patients with a confirmed viral infection, by PCR copy numbers like u said, will get anti-viral therapy. We are looking if giving pre-emptive virus specific donor T cells will bring down the viral load without giving anti-viral therapy. Hypothesis is that it will in atleast 50% of the patients.

Outcome will be the best response seen at 6 weeks post first infusion. If the patient doesn't achieve a partial or better response after first infusion, then he will receive the anti-viral drug which is the current first line management.

Sorry, I still don't get it.

Do all patients in your study get the infusion and the anti-viral drug? Or do all patients get the infusion but not the drug?

And what is "the best response"?

Jochen Wilhelm : All the patients will get the infusion alone initially.... If they do not show a partial or better response, they will be switched over to anti-viral...

We will measure weekly PCR copy numbers. Response will be graded from complete ( no detectable viral load), partial (>50%), inadequate (20-5-%) and no response. The best response achieved in the first 6 weeks will be used for outcome.

If they do not achieve a partial or complete response in the first post infusion PCR assay, they will be switched over to anti-viral drug.

[hypothesis - atleast 50% patients will have a partial or complete response (and hence not require anti-viral therapy).

Ok, so the actual experiment is to keep the patients (all got an infusion) for one week without antiviral-drug treatment, and then to decide whether or not they do get antiviral-drug based on a reduction of the viral load by at least 50%.

You consider the infusion as clinically relevant if, after one week, only 50% of the patients will have a viral load reduction of less than 50% (and therefore getting the drug).

Am I correct? I am still not sure how to consider the subsequent time-points (up to 6 weeks). Selecting the "best response" clearly introduces a bias - that's not a good way to go. You lack a control group, so you can not control for time-effects, and it's impossible to interpret the "survival curve" ("survival" here means "no reduction in virus titer by at least 50%"; the "event" would be the fact that the reduction is >50%, which happens some time after infusion [time-to-event]).

Yes. We plan to follow this with an RCT comparing two arms. But for this single arm study, I was stuck as to how to get the sample size :-). Historical data is that all the patients would require anti-viral but comparing 100% vs 50% made no sense statistically i think...

(There was also a prophylactic arm where we said historically 80% haplo transplant recipients would acquire viral infections or reactivations post-transplant and by giving PROPHYLACTIC T cell infusions after transplant we would bring down this number to 40%. This was a straight forward calculation). Somehow I couldn't solve the preemptive arm calculation with my limited statistic knowledge

And thanks for the suggestion. Let me discuss whether it will be better to stick to the immediate post infusion response rather that best response as you said.....