Telaprevir - 18 yes 0 no

[frijole]

Whew - that was dicey.  One of the FDA advisory committee panelists was agressive in his pursuit of issues with the rash, but in the long run, he and everyone else on the panel voted to approve.

Now it goes to the FDA.  Approval date (or not) is May 23. Some people are betting that the FDA will announce both BOC and TEL on May 7 - the date for Boceprevir approval (or not)

[Andiamo1]

Wouldn't you need to use multiple trials and calculate the difference between each of them for the same values?  Since this is a sample population of a larger group, isn't there a slightly different formula where you need to subtract 1 from the denominator?.

In any case, one standard deviation is likely to make some comparisons irrelevant, don't you think?

[willing]

coeric: anyone doubtful about whether triple tx will work given their past ifn response should look at the data directly - it's well summarized and the difference is stark. The 90-page Merck submission is at

http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/AntiviralDrugsAdvisoryCommittee/UCM252343.pdf

The SPRINT-2 results (naive, 1097 pts) are in Table 8, pg. 42 and the RESPOND-2 results (previous tx failure,  413 pts) are in table 15 pg 53.

Having a 1 log drop by w4 makes approximately a  40% difference in SVR outlook - nothing vague about that.

Any early cutoff decision will exclude some potential SVRs - even continuing with detectable VL at w24 leaves a 2-3% chance of SVR.  And for many continuing with a 30% chance will  still be the right choice. However it's good to know what you're signing up for.

Testing earlier can give an early peek at how it plays out ( I tested at 24h and 48h post) but there's no equivalent data similar to what's given in those two tables. Hopefully Vertex will at some point let us know their side of the story.

Andiamo: though you didn't clear until w20 it looks as if you'd still be classified as a relapser by the Vertex criterion so your odds with them would have been about 80%.

Re standard deviations, you're right that they don't appear to be explicitly included but since what's being measured is a difference in response rates they are easily calculated from  counts. The standard error of the difference of two response rates is
sqrt( (p1*(1-p1)/n1+ p2(1-p2)/n2 )
where p1 = svr1/n1 and p2=svr2/n2 are svr %s in the two groups.

Agreed  that confidence intervals (which incorporate the standard error) are more informative than point estimates. For example, it makes no sense that a lead-in would reduce your SVR odds. The improvement may be small but shouldn't make things worse.

In fact looking at the confidence intervals coeric quoted above shows that the CI for the partial responders is much wider (over twice) than the CI for relapsers, making that point estimate more suspect. Of the 662 pts in the REALIZE, 354 were relapsers but only 124 partial responders so you'd expect a noisier estimate in that subgroup.

[Andiamo1]

I think a lead in is imprecise at best as a screen.  I was a slow responder and became undetectable at week 20.  I relapsed after EOT.  I was a very rapid responder when I treated with triple therapy and am SVR 3 years post treatment.

The data presented is too vague to use as a way of seeing the effectiveness of a lead in as a screen. I have not seen the standard deviations published for any trial, so I assume it's large enough to make the differences between a lead in and no lead in meaningless.  I don't have any proof of that, but a trial of a few thousand people used to predict the response of a few million people must have a large standard deviation.  If it's +- 5% then a difference of up to 10% between 2 trials is meaningless.  I hope it's not that large, but who knows?  And that's saying the trial selection is totally unbiased and I don't think many people believe that.

Eric

[coeric]

thank you. you have a great point about screening people out of therapy with a lead-in. Do you suppose there are patients in the boce trials that would have achieved SVR if they had been allowed to continue?

your points are relevant.  i am thinking that a shorter than 4 week lead-in might be better.  a while back i recall a study that suggested interferon response and SVR could be predicted with-in 24 hours after the first interferon shot.

it might be useful to know my response to interferon/ribivirin to facilitate response-guided therapy.  say instance i have a weak interferon response and after 2 weeks of innterferon/ribavirin the viral load is reduced only 75%.  and at this point i want to continue on.   the way forward might be to try to maintain a significant dose of ribivirin with the aid of epogen when the anemia kicks in.   on the other hand suppose by some miracle i am undetected by 2 weeks.  i might be more inclined to reduce the ribavirin when the anemia starts.  

i am also  experiencing loss of attention.  don't know whether its due to advancing age or this silly disease.   i want to thank all the folks like yourself who have been cured and continue helping us newbies out.  thank you one and all.

eric

[Andiamo1]

I apologize for these scattered posts: I have severe insomnia and ADD that has plagued me since I treated with SOC in 2002.

I think we have to analyze how many people that would SVR were screened out due to the lead in test; in other words what errors took place in the screening process.  I think enough data is there to do it, but it's beyond my ability these days.    The question is what percentage were screened out in the Boceprevir trial and how does that compare with the failure rate when there is no lead in screen?

Given the standard deviation of these trials, say 5%, which we can only guess at, the results of the two trials are identical.

Lots more analyses here before we conclude that a lead in is beneficial.  I wish I could do it :(.

[Andiamo1]

Personally, I would never take interferon 4 weeks just to see the effectiveness of the treatment.  I would sooner hang by my thumbs.  There is a strong possibility that potential SVR candidates would be screened out.  I would wait for a better test than a lead in.