Summary:
Bioequivalence studies is the most popular method for demonstrating therapeutic equivalence between an original and a generic formulation. However, in recent years there has been an ongoing effort to find new approaches that can ensure comparable efficacy and safety between the two compared products in a more economical, fast and ethical way. One of these approaches is the ability to use the direct in vitro dissolution curve comparison indices (DCCIs), which to date are only used as tools in routine quality control, between reference and test formulations. Thus, the goal of the present postgraduate thesis is to examine the theoretical sensitivity of the in vitro dissolution curve comparison indices, namely f1 (difference factor), f2 (similarity factor), and ξ1, ξ2 (Rescigno indices) in reflecting differences in early exposure, between the test and the reference formulations, based on the pharmacokinetic parameters Cmax (maximum drug concentration in blood) and AUCTmax, R (partial area under the blood concentration - time curve up to the time of maximum concentration, Tmax of the reference formulation, R).
In order to investigate the eventual agreement between similarity of the in vitro dissolution curves and bioavailability, simulated cumulative in vitro dissolution and in vivo blood concentration-time data were regenerated. Direct curve comparison indices up to two specific cut-off points (ref85 and fast85) were calculated from in vitro data. The ratios test (T)/reference (R) of the pharmacokinetic parameters Cmax, Tmax and AUCTmax,R were estimated from in vivo data. Finally, in vitro - in vivo correlation (IVIVC) graphs between the indices and the ratios of the pharmacokinetic parameters were constructed. In all cases dissolution/input kinetics are described by the Weibull function. In vivo data were generated assuming that dissolution of the drug fully reflects the drug input into the general circulation and that the drugs follow one-compartment distribution model kinetics. Simulation of in vivo data was performed by numerical convolution. Two scenarios of simulations were examined: in the first scenario (simulations A) both the dissolution/absorption rate constant (ka) and the shape parameter (u) of the Weibull function values of the test formulation are different from the corresponding parameter values of the reference formulation. In the second scenario (simulations B) only the dissolution rate constant value between the two formulations is different. For each of these two simulation scenarios, cases of drugs having the same dissolution rate but differing in their elimination characteristics (kel) were further studied. Finally, a special case scenario of simulations (Γ) was investigated, in which different test formulations have the same shape parameter value, that is different from that of the reference formulation, the same kaR/kel ratio and the same kaT/kaR ratios per examined pair of absolute kaR and kel values. For all simulation scenarios focus was on whether or not there was agreement between in vitro dissolution curves similarity and bioequivalence, and also whether or not the results were consistent with the conclusions drawn in other recent studies where first order dissolution/input kinetics was assumed.
The results obtained when both the shape parameter and the dissolution rate constant values of the test formulation are different from those of the reference formulation (simulations A) show that when uT < 1 there is agreement with the conclusions drawn from the first-order input model in more cases than for uT > 1. Specifically, in cases where uT = 0.5 and uT = 0.7, AUCTmax,R appears to be more sensitive than Cmax when kaT > kaR, whereas when uT > 1 this applies only when uT = 1.5 and kaT > kaR. In these cases, AUCTmax,R, (especially when kaR/kel < 1), reflects the differences in early exposure better than Cmax, whereas in general indices, especially f1 and f2, also indicate dissimilarity in dissolution curves, resulting in in vitro-in vivo agreement, mostly based on AUCTmax,R and especially when kaR/kel < 1. Nevertheless, there is one case (uT = 1.5) where in vitro indices do not detect differences in dissolution profiles, whereas AUCTmax,R shows differences in early exposure (when 0.5 < kaT/kaR < 0.76 and kaR/kel < 1).
For the above shape parameter values results obtained regarding the other formulations, that is formulations when kaT < kaR in cases where uT = 0.5 and uT = 0.7 as well as formulations when kaT > kaR in cases where uT = 1.5, are not in agreement with the first-order model. In those cases, AUCTmax,R seems to be more sensitive when kaR/kel > 1 and Cmax better reflects differences in early exposure. These differences appear to be detectable by indices as well, resulting in in vitro-in vivo agreement, predominantly based on Cmax, especially when kaR/kel < 1. Even greater deviations from the first-order model are observed when the shape parameter values of the test formulation differs substantially from the value of the reference formulation (uT = 0.2, 2, 5). In these cases Cmax appears to be overall more effective in detecting differences in early exposure, as AUCTmax,R, especially for cases where kaR/kel < 1 shows low sensitivity and therefore, mainly in the case of drugs with pronounced sigmoidal concentration-time curves fails to detect differences in early exposure. According to the above, it becomes evident that the shape parameter value of the test formulation is important for the in vitro-in vivo agreement.
The results obtained in case when only the dissolution rate constant value of the test formulation differs from that of the reference formulation (simulations B) show that there is general agreement with the conclusions of the first-order model, especially when uR = uT = 0.5, but in this case the pharmacokinetic parameters appear to be slightly more sensitive. Thus, Cmax, fails to detect differences in early exposure only in some cases when kaR / kel > 1, whereas indices ξ1 and ξ2 do not detect differences when kaR / kel < 1. When uT = uR = 2 agreement is only observed when differences are detected based on AUCTmax,R, as the parameters, and particularly Cmax, appear less sensitive than in the case of the first-order model. Therefore mainly Cmax fails to detect differences in early drug exposure for some formulations with kaT/kaR < 0.67 and kaT/kaR > 1.3. In the specific case of drugs with the same shape parameter values of the test formulation which are different from that that of reference formulation, same kaR/kel ratio and same kaT/kaR ratios per pair of absolute kaR and kel values (simulations Γ), the results drawn from the same ratio values but different absolute rate constant values kaR, kaT and kel are not in agreement. Therefore, it is concluded that when the kinetics of the test formulation differs from that of the reference formulation, the sensitivity of both the pharmacokinetic parameters and the direct curve comparison indices depends not only on the parameter ratios, but also on the absolute values of these rate constants. Finally, based on the simulations studied in this work, it was observed that lower parameter values led to a greater in vitro-in vivo agreement.
Keywords:
difference factor f1, similarity factor f2, Rescigno indices ξi, in vitro dissolution comparison, Weibull function, bioequivalence, pharmacokinetic parameters, in vitro - in vitro correlations, IVIVC, Cmax, AUCTmax,R, drug absorption
