Log in
Register|
bjkim97 ☆ Korea / Seoul, 2010-12-22 03:42 (5239 d 15:32 ago) (edited on 2010-12-22 05:22) Posting: # 6321 Views: 14,401 |
|
|
Dear all. I have been working for a Pharmaceutical Company... And my English is very poor. Please be understanding even if I make a English conversation. I have a question. I determined seleted the sample size as follows Based upon the mean ISCV from the literature, a sample size is determined to have a power of over 0.8 (80%) with T/R ratio of 5%. However, if no reviewed literatures suggest ISCV of over 30%, the sample size is determined to have a power of ≥ 0.8 (80%) with T/R ratio of 5% based upon the maximum ISCV. For example As no reviewed literatures indicated that the ISCV is over 30%, the sample size of Pantoprazole bioequivalence trial was determined based upon ISCV of 22.5% and T/R ratio of 5%. The sample size was determined to include 23 subjects. However, according to the local bioequivalence trial standards, a total number of subjects should be over 24 (12 per treatment group). Given the possible dropouts, the number of subjects was finally determined to be 30 However other review of Pantoprazole is regarded as a "HVD" Therefore I think Pantoprzole has to be treated as a "HVD" I want to hear your opinon Do you think my calculation is right? Please do not hestitate to give comments to me after review ASAP because my boss is "Angel"  Thank you for your help in advance Literature #1 Comparative BA study with two pantoprazole delayed-released Tablet formulations administered with and without food in healthy subjects Arzneimittel-Forschung(Drug Research),2008;58(3):141-148 AUClast, Cmax Mean ratio : 95.66, 104.65 90% CI : 0.8526~1.0732, 0.9086~1.2054 CV Intra-subject : 33.82%, 30.12% Literature #2 Bioequivalence of Two enteric coated formulations of pantoprzole in healthy volunteers under fasting and fed conditions Arzneimittel-Forschung(Drug Research),2007;57(6):309-314 AUClast, AUCinf, Cmax Mean ratio : 94.31, 94.36, 98.76 90% CI : 0.89~0.99, 0.89~0.99, 0.94~1.03 CV Intra-subject : 11.23%, 11.16%, 10.16% Byung-Ju Kim. Bioequivalence Scientist Tel: +82 2 317 2081 / +82 10 3955 1601 Edit: E-mail address removed (spam harvesters are eager in collecting…) I activated the e-mail link in your personal profile – other registered users may contact you through a web-form (your e-mail address is not displayed). [Helmut] |
|
boonchai_l ☆ Thailand, 2010-12-22 06:55 (5239 d 12:19 ago) @ bjkim97 Posting: # 6322 Views: 13,027 |
|
|
Dear bjkim97 First, I understand your English and English is not my mother language, don't worry. I used to be a rookie in this business and your problem was my problem and I'm sure this was everyone problem. I think your problem is not sample size calculation because your calculated sample size from initial parameters is correct but the problem might be CV selection. You give 2 papers and I also have a few pantoprazole BE papers but I cannot access full text your first one and its abstract did not give number of subjects in the study, so the intra-subject CV of the first one could not be accessed by only the information from its abstract. However, the rest literature (that I have) indicated this drug is not high CV drug, it's around 11-18% but it seem to be higher in fed condition and also in MR. If I were you I will set 20% CV into the sample size calculation and find more information for fed condition or MR. The intra-subject CV is a random variable, so it can be any value around its mean with its variance in each condition. It's not surprising if CV from any study are difference because the condition is not same at least time condition. IMHO, CV selection depend on you and your sponsor, you should explain them and then cooperate with them to decide with full any support and select the optimum CV that make you and your sponsor feel safe under the budget limitation. Boonchai L. |
|
VRP ☆ India, 2010-12-22 07:54 (5239 d 11:20 ago) @ boonchai_l Posting: # 6323 Views: 13,005 |
|
|
Dear Both, though I'm not a pro in this business, I'll try to answer based on whatever knowledge I've garnered. when you have multiple papers mentioning different intra-subject CVs you have to calculate a combined CV considering all values you have... you can refer Helmut's lecture series on calculation of sample size which explains this thing in detail and also helps you calculate giving examples. Regards, Dr. Vivek Phirke |
|
bjkim97 ☆ Korea / Seoul, 2010-12-22 08:40 (5239 d 10:34 ago) @ boonchai_l Posting: # 6324 Views: 13,091 |
|
|
Dear boonchai ❝ I used to be a rookie in this business and your problem was my problem and I'm sure this was everyone problem. Me too, I don't know it when I worked the Bioequivalence study at ten years ❝ I think your problem is not sample size calculation because your calculated sample size from initial parameters is correct but the problem might be CV selection. Yes, CV selection is a very prominent question in my mind. ❝ You give 2 papers and I also have a few pantoprazole BE papers but I cannot access full text your first one and its abstract did not give number of subjects in the study, so the intra-subject CV of the first one could not be accessed by only the information from its abstract However, the rest literature (that I have) indicated this drug is not high CV drug, ❝ it's around 11-18% but it seem to be higher in fed condition and also in MR. If I were you I will set 20% CV into the sample size calculation and find more information for fed condition or MR. Hmm. Pantoprazole is not "HVD" in your opinion? If your opinion Pantoprazole is not "HVD" and do I have more review the literature ❝ The intra-subject CV is a random variable, so it can be any value around its mean with its variance in each condition. It's not surprising if CV from any study are difference because the condition is not same at least time condition. And I want your opinion, What is ISCV study difference of main factors? (eg., Formulation or on Washout period Subject Condition, anything factor? ❝ IMHO, CV selection depend on you and your sponsor, you should explain them and then cooperate with them to decide with full any support and select the optimum CV that make you and your sponsor feel safe under the budget limitation. I refer to your opinion. Thank you very much for your time in my question. |
|
boonchai_l ☆ Thailand, 2010-12-22 10:06 (5239 d 09:08 ago) @ bjkim97 Posting: # 6325 Views: 12,991 |
|
|
Dear Kim Who define HVD? Does it subject to the regulator in each country? Is it useless? if you known you are going to run study with 40% CV in Cmax but your regulator don't allow to extent the BE acceptance criteria. In this case, HVD is only the information telling you that you have to run the risky project if the number of subjects are not enough. The first paper you referred is MR and The evidence that I have with standard design, IR, fasting condition there is no one which the CV exceed 20% that's why I don't think this drug (for above condition) is a high CV drug. BTW: Combined CV is an option. Suppose that if you have 7 papers support the CV is around 10-15% CV but there is one paper that have 27% CV. How do you set the CV? There are many way to set CV into the formula e.g. select from median or select from mode or subjective setting up or even assuming and etc. Finally, beside the sample size calculation, the number of subjects also depend on your many restrictions e.g. your team discussion, budget, trust of your evidence, sponsor opinion and etc... that you and your team should be the best one who know about. It is the basically statistical idea. That's why there are a number of statistics textbooks stated about sample size determination under budget limitation. Regards, Boonchai L. |
|
Dr_Dan ★★ Germany, 2010-12-22 12:06 (5239 d 07:08 ago) @ boonchai_l Posting: # 6326 Views: 13,133 |
|
|
Dear Kim, dear Boonchai L. Reliable information on the intrasubject coefficient of variation is needed for sample size planning, you are right. Unfortunately, this information is usually not presented in publications on bioequivalence studies, and only the pooled inter- and intrasubject coefficient of variation for either test or reference formulation is reported. Thus, the essential information for sample size planning of future studies is not made available to other researchers. The standard approach to the analysis of a two-treatment, two-sequence, two-period crossover trial is an analysis of variance (ANOVA) for the log-transformed pharmacokinetic parameters, where the factors formulation, period, sequence and subject nested within sequence are used to explain overall variability in the observations. The residual coefficient of variation (CV) is a measure of the variability that is unexplained by the aforementioned factors. Among others, within-subject variability, formulation variability, analytical errors, and subject by formulation interaction can contribute to this residual variance. A drug product is called highly variable if its intra-individual (i.e. within-subject) variability is greater than 30%. A high CV as estimated from the ANOVA model is thus an indicator for high within-subject variability. Only by using a replicate design it is possible to estimate the within-subject variability of the test and of the reference product as well as the subject × formulation interaction. Please have a look at the revised EMA guideline CPMP/QWP/EWP/1401/98 Rev. 1 "If an applicant suspects that a drug product can be considered as highly variable in ist rate and/or extent of absorption, a replicate cross-over design study can be carried out. Those HVDP for which a wider difference in Cmax is considered clinically irrelevant based on a sound clinical justification can be assessed with a widened acceptance range. If this is the case the acceptance criteria for Cmax can be widened to a maximum of 69.84 – 143.19%. For the acceptance interval to be widened the bioequivalence study must be of a replicate design where it has been demonstrated that the within-subject variability for Cmax of the reference compound in the study is >30%. The applicant should justify that the calculated intra-subject variability is a reliable estimate and that it is not the result of outliers. The request for widened interval must be prospectively specified in the protocol." I hope this helps Kind regards Dan — Kind regards and have a nice day Dr_Dan |
|
Helmut ★★★   Vienna, Austria, 2010-12-22 17:59 (5239 d 01:15 ago) @ boonchai_l Posting: # 6330 Views: 13,283 |
|
|
Dear Boonchai! ❝ Who define HVD? Convention (aka old hat). Was first stated at the Bio-International ’89 in Toronto, Canada as 25% CV being “problematic” and 30% considered highly variable. ≥30% Was agreed upon at the Bio-International ’92 in Bad Homburg, Germany.1,2 Nitpicking terminology: We distinguish between HVDs (Highly Variable Drugs: CV of a solution ≥30%) and HVDPs (Highly Variable Drug Products: CV of a formulation ≥30%). EMA adopted this wording in the current guideline. ❝ Does it subject to the regulator in each country? Is it useless? No and no. In some countries widening of the BE-limits is acceptable:
![[image]](img/uploaded/image66.png) Note that the bioequivalence limits are displayed in ln-scale in order to show the symmetry around 1: ln(1) = 0 and ln(0.80, 1.25) = ∓0.2231.
— 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 |
|
d_labes ★★★ Berlin, Germany, 2010-12-22 13:05 (5239 d 06:09 ago) @ bjkim97 Posting: # 6328 Views: 13,760 |
|
|
Dear Byung-Ju, have a look at the EMA Public Assessment Reports to be found here for another source of CV's for your sample size estimation. Unfortunately there are only 90% confidence intervals given. But with a little math as explained in Helmut's lectures (f.i. this one) you will get the ISCV's. A hint: If you can use R have a look at the package PowerTOST which will free you from the burden to do the math by paper and pencil (function CVfromCI()). If you look at the resulting CV's from the studies mentioned in the PAR's you will find that at least for Cmax for studies in fed state there is the danger of having a CV >30% which is by convention seen as highly variable. See for instance the PAR for the application of Wockhard Ltd. which gives for Cmax CV=34.8% in fasting state and CV=59.3% in fed state. Note the replicate design of the fed study. Since there are a number of PARs its time to read Helmut's lectures with respect to pooling of CVs. No simple mean, median! Hope this helps. — Regards, Detlew |
|
boonchai_l ☆ Thailand, 2010-12-22 22:26 (5238 d 20:48 ago) @ d_labes Posting: # 6331 Views: 13,001 |
|
|
Dear all, Please read carefully because I afraid my poor English make someone misunderstand. Thanks for all comments that try to explain me or even speak to me sarcastically. I would like to reply in all issues but I am exhausted to explain them in English, so I will reply only issue of combined CV. I am sorry and I afraid my opinion will bother someone again. I have read the pooling of CV in Helmut's lectures around three week ago, the concept was developed using sampling distribution of sample variance from classical statistics or parametric approach. I appreciated Helmut because he understands and be able to apply statistical concept even if he does not grow up in statistical field, while I am as a statistician but I never try to apply that concept and excuse from them by "I am being busy" always. Although I fully agree that concept, but not at all and I still stand that the pooling of CV concept is an option under some situations. Anyone can choose their own CV as long as they want to. It is quite subjective and there are many different reasons to support their own CV. There are pro and con in each way, for example, Mary select the minimum CV because she is optimistic and she trust that paper, Ronaldo exclude some CV before he pool the rest because those were studied in Thailand and he doesn’t trust them , Lady K. select the average CV because it's very simply, Kim select 25% CV because of his sponsor, Filipe M. select 14% CV because it is his team solution, Tom select the maximum one because he is conservative and he has a lot of budget…Though the pooling of CV concept would be developed from statistical knowledge, it has some disadvantages. First, it does not simplify. Second, in case that one value of many CV is an outlier CV. Last and very important, as everyone knows this concept comes from assumption of parametric statistics, so each variance as samples or random variable (RV) should be iid from chi-square distribution. Are there anyone check them? How much its robustness if it breaks the assumptions? Are there any methods to combine them if it's not come from the same population? And bla bla bla… No need to answer these questions. Because I only want to elucidate my idea and I don't intent to insult the pooling of CV. Finally, I would like to say sorry again to d_labes or anyone if my opinion bothers you and seem to be irrespective to the pooling of CV. Sincerely, Boonchai L. |
|
ElMaestro ★★★ Denmark, 2010-12-22 23:22 (5238 d 19:52 ago) @ boonchai_l Posting: # 6332 Views: 13,024 |
|
|
Dear Boonchai_L, that's an excellent post, and certainly worthy of some thoughts. From a practical perspective, I think real life is often rather cruel. When a sponsor calculates the sample size for a BE study, literature is a good starting point. Yet a lot of sponsors are seeing their studies failing because they are not realistic about the CV's. I don't think the solution is just pooling or any other statistical approach. In stead, I think a (but not necessarily the [only]) solution is get real about CV that can be produced. It is often virtually impossible to tell if a CRO is capable of getting the CV as low as you hope, especially if the CRO does not indicate they have worked with the API in question before. The analytical method may differ, the practices at the clinical facility differ etc. Unless the sponser has a good reason to believe the actual CV will be in the low end of the published CV's I think the expectation should be in the high end of the published CV's. Here's a well kept secret: Audits (not just the ordinary motherhood-and-applepie-system-audits) are a sponsor's best friend in this regard. S. — Pass or fail! ElMaestro |
|
Ohlbe ★★★ France, 2010-12-23 02:23 (5238 d 16:51 ago) @ boonchai_l Posting: # 6333 Views: 12,827 |
|
|
Dear Boonchai, ❝ Finally, I would like to say sorry again to d_labes or anyone if my opinion bothers you and seem to be irrespective to the pooling of CV. No need to apologise. That's what forums, including this one, are there for: to allow people to express their opinion. All people do not share the same opinion: that's good, that's how science sometimes moves forward. It's good to "think out of the box" too. Regards Ohlbe — Regards Ohlbe |
|
Helmut ★★★ Vienna, Austria, 2010-12-23 16:29 (5238 d 02:45 ago) @ boonchai_l Posting: # 6337 Views: 13,211 |
|
|
Dear Boonchai! ❝ […] afraid my poor English make someone misunderstand. Don’t worry, native English speakers are in the minority ’round here. ❝ […] the pooling of CV […] was developed using sampling distribution of sample variance from classical statistics or parametric approach. […] from assumption of parametric statistics, so each variance as samples or random variable (RV) should be iid from chi-square distribution. Are there anyone check them? How much its robustness if it breaks the assumptions? Are there any methods to combine them if it's not come from the same population? And bla bla bla… Wonderful – you are right in every aspect. IID is also an assumption in the classical 2×2 cross-over model, which may not hold as well. If the reference has a higher variance than the test, you are penalized. In a nonreplicate design we have no means to check the validity of the assumption, so bad luck. It would be good statistical practice the check the assumptions and have a contingency plan a priori if they are violated – rendering the model invalid. In the past many protocols in Europe checked the ANOVA’s residuals and evaluated the study by a nonparametric method if the assumptions were violated. Regulators didn’t like that (why the hell?)… BTW, in Japan such a procedure is still acceptable according to the current guidelines. When it comes to pooling, we have no means to check the assumptions.  I don’t know whether anybody challenged the method with real data (at least it’s not published). ❝ I appreciated Helmut because he understands and be able to apply statistical concept even if he does not grow up in statistical field, THX from an interested amateur! ❝ Anyone can choose their own CV as long as they want to. It is quite subjective and there are many different reasons to support their own CV. […] for example, Mary, Ronaldo, Lady K., Kim, Filipe M., Tom… Great summary! Not least Dave Dubbin’s FARTSSIE is popular with its – “We know the sample size beforehand and tweak the input (CV, GMR, power) to get a justification for the protocol.” That’s clearly not the idea behind ICH E9. Pooling is just a member of the statistical toolbox. It’s up to the responsible persons to use their brains to give individual studies more or less credit (as ElMaestro suggested). The best estimate one gets always from a – reasonably large – pilot study. Only then you can control all the side-conditions (clinical performance, bioanalytics,  ) which may influence variability. We also must not forget the GMR. Sometimes sponsors are too optimistic about their product. If the GMR in a pilot study comes as 81%, what would you do? Retreat to statistics (random variability and crossing fingers) or go straight ahead and reformulate?  — Dif-tor heh smusma 🖖🏼 Довге життя Україна! Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
d_labes ★★★ Berlin, Germany, 2011-01-03 09:43 (5227 d 09:31 ago) @ boonchai_l Posting: # 6387 Views: 12,517 |
|
|
Dear Boonchai, back from my season's holidays I only can say: Well roared, lion!  .Just to cite Stephen Senn: "The sample size calculation is an excuse for a sample size and not a reason." (Chapter 13 of Statistical Issues in Drug Development, Second Edition, Wiley 2007). — Regards, Detlew |
|
bjkim97 ☆ Korea / Seoul, 2010-12-23 15:43 (5238 d 03:31 ago) @ d_labes Posting: # 6336 Views: 12,858 |
|
|
Dear d_labes Thank you very much. Your good advice... I looked through the "MU2010-CD2.pdf" and I almost understand it. But I don't understand one chapter that "pooling of CV" Please tell me about this in detail. I know "Calculate the variance from CV" I don't know "Calculate the total variance weighted by df" Naturally I don't know "Calculation the pooled CV from total variance" How is the calculate the total variance? I want explain this by Excel or Electronic File  Regards bjkim97 |
|
Helmut ★★★ Vienna, Austria, 2010-12-23 17:11 (5238 d 02:03 ago) @ bjkim97 Posting: # 6338 Views: 12,863 |
|
|
Dear Kim! ❝ I know "Calculate the variance from CV" OK. ❝ I don't know "Calculate the total variance weighted by df" ❝ Naturally I don't know "Calculation the pooled CV from total variance" ❝ How is the calculate the total variance? ❝ I want explain this by Excel or Electronic File See the examples (slides 18–20). In a 2×2 cross-over study you have 2 sequences (TR|RT). Therefore, with a sample size of \(\small{n}\) we have \(\small{df=n-2}\). In my first example (slide 18), first study: \(\small{n=12}\), \(\small{df=12-2=10}\), \(\small{CV=0.2\;(20\%)}\), \(\small{\sigma^2=\log_{e}(CV^2+1)=\log_{e}(0.04+1)\approx0.0392}\), weighted variance \(\small{\sigma_\textrm{w}^2=\sigma^2\times df\approx 0.3922}\). Repeat these steps for all studies and calculate the “total variance weighted by df” \(\small{\sigma_\textrm{tot}^2}\). In the example that’s \(\small{(0.3922+0.8618)/(10+10)=0.0627}\). Then \(\small{CV_\textrm{pooled}=\sqrt{\exp(\sigma_\textrm{tot}^2)-1}\approx0.2544}\). No need for M$-Excel. — Dif-tor heh smusma 🖖🏼 Довге життя Україна! Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
Ing. Helmut Schütz