Bioequivalence and Bioavailability Forum • IVIVC with literature IV/IR data to serve as impulse func.

jag009
★★★

NJ,
2012-07-12 19:37
(5096 d 18:53 ago)

Posting: # 8941
Views: 5,038

IVIVC with literature IV/IR data to serve as impulse func. [Regulatives / Guidelines]

Hi,

Is it valid to generate an IVIVC without a 4-way study that involves 3 test ER formulations and IV/IR treatment? Can I generate one with the IR/IV data obtained from literature to serve as the impulse function for deconvoution and convolution steps? Data from 3 test ER formulations will be from one study, i.e., 4-way crossover study compared 3 test ER formulations vs reference ER product.

Thanks

John

JMCardot
☆

France,
2012-07-13 14:13
(5096 d 00:18 ago)

@ jag009
Posting: # 8942
Views: 4,317

IVIVC with literature IV/IR data to serve as impulse func.

Post reply

Dear John,

NO you cannot use literature data, lets modulate now the answer. It depends:

If the drug behaves as a one cp model you can use Wagner Nelson and in this case a deconvolution is not necessary (IV or IR not necessary) but the simulation/prediction is harder afterward and you have to check the absence of flip-flop.
Use of literature or data from different studies is sometime made at posteriori reanalyzing old studies to have a first idea about IVIVC feasibility without any regulatory purpose. In your case if you want to play with the data without any regulatory purpose: not to be included in a dossier, no use afterwards for biowaiver or surrogate of in vivo, only to have a first approach of the relevance of the dissolution method for example after a first small pilot study (to have a first rough idea) you can try but it is not recommended as often if results are encouraging you will have a tendency to say that it is a pity not to use the data! So please try to take this opinion in consideration before taking a final decision: if you use literature data you might have the impression to gain for the current study but if the results are encouraging you might have to redo a study with IR or IV to confirm the results … so you do not gain at the end. The fact not to use literature is that often you do not have any idea about quality, subjects (are they similar or different from your subjects), analytical method, formulation, calculation made (mean values), etc… and you cannot extrapolate the published data to your own study. It is like pooling data from different studies together it is not a recommended process.
if you want to use the data afterward for any part of the dossier, it is better to have either an IV or and IR fast formulation arm in your study to perform deconvolution and after convolution to predict (i) internal or external predictability (ii) new formulations having the same route of administration, the same release mechanisms and within the dose strength and linearity domain of the drug based on the dissolution.

So to conclude it is better to include an arm with fast IR or IV when you want to make IVIVC. If you have a good knowhow of the pharmacokinetic behavior of your drug and if the drug behaves as a one compartment model (and that confirm with recent studies and not studies described in the 70ties with high LOQ methods) you can use WN for example, but even in this case the prediction are not simple and not straightforward to be explained in a dossier, so include an IR/IV arm.

Best regards,

Jean-Michel

jag009
★★★

NJ,
2012-07-19 01:18
(5090 d 13:13 ago)

(edited on 2012-07-19 15:08)
@ JMCardot
Posting: # 8958
Views: 4,208

IVIVC with literature IV/IR data to serve as impulse func.

Post reply

Thank you Jean-Michel,

Questions:

1) I agree with most of what you said. Correct me on this.. --> Since the UIR (Unit impulse function) only serves as a "Denominator" for the IVIVC process for the convolution and deconvolution steps, if one keeps it constant throughout the IVIVC modelling process (and later stages) then shouldn't it be okay if literature IR or IV data is used?

2) There is a chapter in the IVIVC book edited by David Young et al. Bill Gullesbie wrote a chapter (2.3) on "Method Not Requiring an IV or IR Reference Dose". He suggested that the UIR can be estimated by fitting the overall convolution model to the plasma concentrations resulting from the ER dosage forms.

John

Edit: Dear John, please avoid typewiter-style multiple blanks in future posts. They are not displayed anyway. [Helmut]

JMCardot
☆

France,
2012-07-20 08:46
(5089 d 05:45 ago)

@ jag009
Posting: # 8961
Views: 4,242

IVIVC with literature IV/IR data to serve as impulse func.

Post reply

Dear John,

1) Yes in theory and in an ideal world, NO in practice as in this case you have to assume that (i) the two sets of subjects respond exactly in the same way: exactly the same behavior/reaction and subjects composition of the datasets (for example not in one case slow acetylators and in the second fast, sex influence and different composition of dataset, for IR no influence of the formulation, no meal influence, same clearance between the datasets, etc…) => in other words the model and your data are reflecting the same overall subject population and the variability is low between the two (ii) that the mean value published reflect the individual data (iii) that the published data are accurate (analytical method, enough sampling points, model used, etc…) (iv) that variability is of no importance (invariance of the function in particular same clearance) (v) all other concerns ;-)

. Even if you keep it constant throughout the IVIVC modeling process, it is not possible. To give you an extreme example: imagine you have set up an animal model IV for your drug will you use this IV model to deconvolate, establish correlation and then convulate in Human if you keep it constant?
In addition I remind you that you are not allowed to use the results for any regulatory purpose and if you have made a mistake selecting the publication and the data, the results are going to be non-sense.
If you want to apply such a method I can only tell you the risks associated.

2) I do not have the book in my hands but I think that you are referring to the (ii) below. Without IR or IV you can use (i) Wagner Nelson if you have a one compartment model (ii) a pure convolution approach (one step approach) which use often methods linked with population pharmacokinetic and is based on individual subjects convolution, in this case you do not have the absorption derived from the deconvolution process but you use only the dissolution observed. Based on dissolution you see which function (IVIVC and time scaling to simplify) must be applied to modify and make it a suitable input function for your convolution (using the IV model) which must fit your observed data adjusting the IVIVC function. In this case the quality of the IV model is of importance and must exist for each subject. You have 4 models to be implemented to do it: the dissolution data, the IV or IR data, the SR data and the IVIVC.

See Gillespie WR. Convolution-based approaches for in vitro-in vivo correlation modeling. in: Young D., Devane J., Butler J. editors. Advances in Experimental Medicine and Biology, In vitro-in vivo Correlations, Plenum Press, New York. 1997. p. 53-65.
Gaynor C, Dunne A and Davis J. A Comparison of the Prediction Accuracy of Two IVIVC Modelling Techniques. J Pharm Sci. 2008;97:3422-32.
O’Hara T, Hayes S, Davis J, Devane J, Smart T and Dunne A. In vivo-in vitro (IVIVC) modeling incorporating a convolution step. J Pharmacokinet Phar. 2001;28:277-298.

Best regards,

JM