Log in
Register|
jag009 ★★★ NJ, 2012-12-05 17:04 (4153 d 05:20 ago) (edited by jag009 on 2012-12-05 17:24) Posting: # 9674 Views: 5,364 |
|
|
Hi all, Would appreciate all your comments with the following scenario... I recently ran a FDA BE study involving an IR product with a long and variable t1/2. The literature value is around 25 hrs with an upper range of 60 hr. I ran a small pilot study (n=6) and the results showed that 96 hours sampling was more than adequate to capture 90% of the PK profile (Helmut will scream at me for using n=6  ).My pivotal study finished with 44 subjects. The study population showed a mean t1/2 of 25 hrs (Range 12-67 hrs) for each treatment. I went ahead with PK blood sampling for 96 hrs (timept before 96 hr was 72 hr). Based on 96 hour sampling, the mean(AUCt/AUCinf) ratio was about 92% for each treatment. Individual data showed 3 subjects to have AUCt/AUCinf ratio of 60% with a t1/2 of greater than 60 hours. Maybe this is a silly question... Do you think that I might get question from the agency about the 3 subjects? Since the mean t1/2 from each treatment arm was 25 hrs with an upper range of 67 hrs, I can re-evaluate the data as a truncated design (if being asked) and compute AUC0-72. The result as is for both AUC and AUCi are dead since the 90% CIs were well within 80-125% (Ratio~100%, intra CV~10%). Washout was adequate(3 wks) with no carryovers. I don't have an issue with computing AUC0-72(pass anyway) but I was just curious if I would even get a question from FDA regarding the 3 subjects. Thanks John |
|
Helmut ★★★   Vienna, Austria, 2012-12-05 20:41 (4153 d 01:42 ago) @ jag009 Posting: # 9677 Views: 4,680 |
|
|
Hi John, ❝ […] IR product with a long and variable t1/2. The literature value is around 25 hrs with an upper range of 60 hr. I ran a small pilot study (n=6) and the results showed that 96 hours sampling was more than adequate to capture 90% of the PK profile What was the longest half-life in the pilot? Did you try to calculate a CL predicting the worst case? Example (slides 63–65). ❝ (Helmut will scream at me for using n=6 Not necessarily. n=6 is OK if used for checking the sampling schedule, bioanalytics (hey, samples from the ‘real world’), etc. Only for sample size estimation n=6 is useless. ❝ My pivotal study finished with 44 subjects. The study population showed a mean t1/2 of 25 hrs (Range 12-67 hrs) for each treatment. I went ahead with PK blood sampling for 96 hrs (timept before 96 hr was 72 hr). Based on 96 hour sampling, the mean(AUCt/AUCinf) ratio was about 92% for each treatment. Individual data showed 3 subjects to have AUCt/AUCinf ratio of 60% with a t1/2 of greater than 60 hours. ❝ ❝ Do you think that I might get question from the agency about the 3 subjects? Maybe.  Interestingly FDA’s Guidance doesn’t state anything about the AUCt/AUC∞-ratio (like EMA’s GL). EMA is more strict, because extent of absorption has to be demonstrated only for AUCt. They state that a “reliable estimate of the extent of exposure is achieved if AUC(0-t) covers at least 80% of AUC(0-∞)”. My gut feeling is that FDA likes ≥80% as well. I would say that a reliable estimate is achieved already at 2–4 × tmax – but that’s another story. On the other hand for the FDA you have to show BE of both metrics. If you are able to estimate elimination in these three subjects reliably (≥3 data points, sufficient quality of the fit) it shouldn’t be a big deal.❝ Since the mean t1/2 from each treatment arm was 25 hrs with an upper range of 67 hrs, I can re-evaluate the data as a truncated design (if being asked) and compute AUC0-72. ❝ The result as is for both AUC and AUCi are dead since the 90% CIs were well within 80-125% (Ratio~100%, intra CV~10%). Washout was adequate(3 wks) with no carryovers. I don't have an issue with computing AUC0-72(pass anyway) but I was just curious if I would even get a question from FDA regarding the 3 subjects. Since AUCt, AUC∞, and AUC72 pass I would submit the latter also as a sensitivity analysis. Regulators love them. BTW, I (still) don’t understand why FDA states “For drugs demonstrating high intrasubject variability in distribution and clearance, AUC truncation warrants caution. In such cases, we also recommend that sponsors and/or applicants consult the appropriate review staff.” At least distribution and clearance are properties of the drug, not the formulation. That’s why truncation is acceptable for EMA for all IR formulations without any justification (not even t1/2 > 24h or alike).— Dif-tor heh smusma 🖖🏼 Довге життя Україна! ![[image]](https://static.bebac.at/pics/Blue_and_yellow_ribbon_UA.png) Helmut Schütz ![[image]](https://static.bebac.at/img/CC by.png) The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
jag009 ★★★ NJ, 2012-12-05 23:04 (4152 d 23:19 ago) @ Helmut Posting: # 9679 Views: 4,478 |
|
|
Thanks Helmut! ❝ What was the longest half-life in the pilot? Did you try to calculate a CL predicting the worst case? From the pilot data n=6, mean t1/2 was ~ 21 hrs, longest t1/2 was 27 hrs. ❝ If you are able to estimate elimination in these three subjects reliably (≥3 data points, sufficient quality of the fit) it shouldn’t be a big deal. Well FDA doesn't have the AUCt/AUCi ratio but they do ask for adequate sampling. For those 3 subjects with AUCt/AUCi ratio ~ 60%, 4 datapts which showed very good linear decline were used to determine t1/2. Only one subject had Cmax at 24 hrs(avg Tmax 4-6 hrs). By the way, these subjects showed AUCt/AUCi ratio of ~ 60% for both T and R. I am leaning toward submitting the study as is since the 3 subjects in question demonstrated bioequivalence between T and R anyway. Some of my colleague suggested analyze the data with and without these 3 subjects and I considered that nonsense. John |
Ing. Helmut Schütz