Log in
Register|
sciguy ☆ Canada, 2013-08-02 17:35 (4356 d 03:25 ago) Posting: # 11170 Views: 6,433 |
|
|
Dear all, I was wondering if I could get an opinion from the forum regarding covariate selection in PK studies. I realize covariate analysis in classic BE crossover studies is unusual but here I am assessing relative bioavailability (based on the usual AUC and Cmax metrics) of two formulations in a parallel design, phase III study. This is a patient population so I am trying to decide what covariates should be looked at in the PK analysis. I am thinking age, gender, race, weight (vs. bmi?), renal function, smoking. I think these are 'popular' covariates to look at but I wonder for the purpose of submissions if there is anything else regulators would be looking for? Does anyone have any suggestions or more experience with this? Thank you for any input. -sciguy Edit: Category changed. [Helmut] |
|
Helmut ★★★   Vienna, Austria, 2013-08-02 18:24 (4356 d 02:37 ago) @ sciguy Posting: # 11171 Views: 5,474 |
|
|
Hi sciguy, ❝ […] I am assessing relative bioavailability (based on the usual AUC and Cmax metrics) of two formulations in a parallel design, phase III study. This is a patient population so I am trying to decide what covariates should be looked at in the PK analysis. I am thinking age, gender, race, weight (vs. bmi?), renal function, smoking. I think these are 'popular' covariates to look at but I wonder for the purpose of submissions if there is anything else regulators would be looking for? Does anyone have any suggestions or more experience with this? Two-stage analysis (either classical modeling or NCA; both followed by regression) is somewhat outdated. Nowadays most people use it only in screening for covariates which might be worthwhile including in a PopPK model. In my experience BMI rarely gives much information. Additionally to body weight you can try body surface area. If your drug is renally excreted to a substantial degree, creatinine (from lab. exams) is quite often a very good covariate. IMHO classical two-stage makes only sense if you run regression on the covariates in different models. Multivariate regression (likely with different weighting schemes) is a plain nightmare. PopPK has the advantage that you can model the error of the covariate in many different ways (additive, proportional, mixed,…). — 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 |
|
ElMaestro ★★★ Denmark, 2013-08-02 18:51 (4356 d 02:10 ago) @ sciguy Posting: # 11172 Views: 5,349 |
|
|
Hi sciguy, ❝ I was wondering if I could get an opinion from the forum regarding covariate selection in PK studies. I realize covariate analysis in classic BE crossover studies is unusual but here I am assessing relative bioavailability (based on the usual AUC and Cmax metrics) of two formulations in a parallel design, phase III study. Erm... bioavailability as phase III??? I haven't understood what is going on, could you please elaborate? ❝ This is a patient population so I am trying to decide what covariates should be looked at in the PK analysis. I am thinking age, gender, race, weight (vs. bmi?), renal function, smoking. I think these are 'popular' covariates to look at but I wonder for the purpose of submissions if there is anything else regulators would be looking for? Does anyone have any suggestions or more experience with this? Plain BE usually implies anova (or similar) but not ancova. There must be some detail I have overlooked. A guidance specific to some special API perhaps? Would like to know more. Have a great weekend. — Pass or fail! ElMaestro |
|
sciguy ☆ Canada, 2013-08-02 21:04 (4355 d 23:57 ago) @ ElMaestro Posting: # 11177 Views: 5,394 |
|
|
Hi ElMaestro, ❝ Erm... bioavailability as phase III??? ❝ I haven't understood what is going on, could you please elaborate? BA/BE is not the primary endpoint, more of a secondary objective to support early phase data to compare a combo drug vs each of the mono ingredients. They want to show no drug-drug interactions in the combo (ie: no difference vs. mono) ❝ Plain BE usually implies anova (or similar) but not ancova. There must be some detail I have overlooked. A guidance specific to some special API perhaps? I guess because this is a parallel design, ancova is not far-fetched? Strict BE is not being evaluated although the 90% CI will be reported to give some measure of similarity (ie: 80-125%). ❝ Would like to know more. ❝ Have a great weekend. You too! -sciguy |
|
ElMaestro ★★★ Denmark, 2013-08-02 22:51 (4355 d 22:10 ago) @ sciguy Posting: # 11179 Views: 5,379 |
|
|
Hi sciguy, ❝ BA/BE is not the primary endpoint, more of a secondary objective to support early phase data to compare a combo drug vs each of the mono ingredients. They want to show no drug-drug interactions in the combo (ie: no difference vs. mono) Well...this could come down to what you consider "pivotal" proof of whatever you wish to show. If the drug-drug interaction thing is only ancillary -which may actually be fair- then I guess co-variates could be explored. Perhaps it could be an idea to define a tertiary objective, this being the investigation of relative BA using no covariates, or at least to also explore and present without those covariates. — Pass or fail! ElMaestro |
Ing. Helmut Schütz