Prediction of in vivo performance of ibuprofen immediate-release products using different dissolution models

Characterization of IBU IR commercial products

Full tablet characterization was carried out using a 3-in-1 hardness, diameter, and thickness tablet tester (Sotax-MT 50 MultiTest 50, Switzerland). The uniformity of weight for each product was calculated using electric balance. The mean of 20 tablets for each product was calculated.

Content uniformity

The assay of IBU in different IR commercial products was assessed by the UV spectrophotometric method (DU–650 UV-Vis Spectrophotometer, Beckman, USA) at λ_max of 221 nm (Emara et al., 2014a). Briefly, 20 tablets of each product were weighed and ground, and the weight equivalent to one tablet was dissolved in methanol, then vortexed, and filtered (Millex, 0.45 um). The filtrate was diluted with phosphate buffer of pH 7.2 and analyzed for drug content.

Comparative in vitro dissolution studies of IBU products

In vitro dissolution (n = 6) of IBU IR commercial products (200 mg IBU/tablet) was carried out in filtered, degassed 900 mL phosphate buffer of pH 7.2 (as recommended by USP, USP-32, 2009) at 37°C ± 0.2°C employing the three compendial dissolution apparatuses as described below:

USP I (basket) and USP II (paddle, pharmacopeial) methods

IBU dissolution profiles were determined in an automated dissolution USP apparatus I (basket method) and USP apparatus II (paddle method) dissolution tester (AT8-Xtend, Sotax, Switzerland), with a rotation speed of 50 rpm and an autocontrolled multichannel pump employed.

USP IV (flow-through cell) method

IBU dissolution profiles were carried out in the FTC, USP apparatus IV (Dissotest CE-6 equipped with a CY 7-50 piston pump, Sotax, Switzerland). In all experiments, a tablet was loaded in the 22.6 mm dissolution cell, with a ruby bead located at the cone entry and no glass beads present to ensure a turbulent flow pattern. Built-in filtration of 0.7 μm Whatman glass microfiber, followed by GF/F and GF/D and then glass wool, was employed, with pump speed set at 8 ± 0.2 ml/minute.

For all dissolution testing, samples were withdrawn from dissolution media at predetermined time intervals and replaced with fresh media. Collected samples were analyzed spectrophotometrically at λ_max of 221 nm against a blank buffer (Emara et al., 2014a).

Comparison and analysis of in vitro dissolution data

An independent model was applied for comparison among different dissolution profiles of IBU IR commercial products.

The model-independent approaches provide a direct comparison of dissolution profiles. Fit factors (f₁ and f₂) (Moore and Flanner, 1996), mean dissolution time (MDT) (Costa and Lobo, 2001), and dissolution efficiency (DE) (Emara et al., 2014b) were employed.

Fit factors

The difference factor f₁ evaluates the percentage difference between the two curves: f₁ values ranging from 0 to 15 indicate similar dissolution profiles, while values >15 indicate dissimilar profiles, according to the following equation: ${\mathit{f}}_1={\sum }_{\mathit{t}=1}^{\mathit{n}}[{\mathit{R}}_{\mathit{t}}-{\mathit{T}}_{\mathit{t}}]/\sum _{\mathit{t}-1}^{\mathit{n}}{\mathit{R}}_{\mathit{t}}]\times 100,$

where R_t and T_t represent the percent of drug dissolved for reference and test, respectively, at each sample point t and n is the number of time intervals.

On the other hand, the similarity factor f₂ is a measurement of the similarity of any two dissolution curves according to the following equation: ${\mathit{f}}_2=50\times \log \{[1+(1/\mathit{n}{\sum }_{\mathit{t}=1}^{\mathit{n}}({\mathit{R}}_{\mathit{t}}–{\mathit{T}}_{\mathit{t}}{)}^2{]}^{-0.5}\times 100\}.$

f₂ values ranging from 50 to 100 indicate similar dissolution profiles, while values ≤50 indicate dissimilar dissolution profiles (Costa and Lobo, 2001; Moore and Flanner, 1996; Shah et al., 1997). As stated by the FDA guidelines, f₁ and f₂ values of ≤15 and ≥50, respectively, ensure equivalence in the dissolution profiles (Shah et al., 1997).

Mean dissolution time (MDT)

MDT was calculated, as described by Costa and Lobo (2001), as follows: $\mathrm{MDT}={\sum }_{\mathit{j}=1}^{\mathit{n}}\mathit{[}\mathit{t}{^}_{\mathit{j}.}?{\mathit{M}}_{\mathit{j}}\mathit{]}/{\sum }_{\mathit{j}=1}^{\mathit{n}}?{\mathit{M}}_{\mathit{j}},$

where j represents sample number, n is the number of dissolution sample times, t^_j represents the time at the midpoint between t_j and t_j−1, and ΔM_j is the additional amount of drug dissolved between t_j and t_j−1.

Dissolution efficiency (DE)

Percent dissolution efficiency (%DE) was also calculated to evaluate the relative performance of IBU IR commercial products (Emara et al., 2014b). The value of %DE at 60 minutes (%DE₆₀) for each tablet was quantified as % ratio of area under the dissolution curve till 60 minutes relative to the area of the rectangle defined by 100% dissolution at the same time, as follows: $\%\mathrm{DE}=[{\mathrm{AUC}}_{0-60}/{\mathit{Q}}_{100.\mathrm{t}}]\times 100.$

Statistical significance was determined for MDT and %DE_{60 minutes} using Student’s t-test in the comparison between means of two groups. Data analysis was carried out using SPSS software (version 17.0). Differences were considered significant when p < 0.05.

Prediction of in vivo plasma response

The deconvolution approach was employed for the prediction of IBU in vivo plasma response. The in vitro dissolution data of different IBU IR commercial products, in different USP apparatuses, were converted into predicted plasma concentration–time curves, based on the back-calculation of the WN equation (Ostrowski et al., 2010), according to the following equation: ${\mathit{C}}_{(\mathit{t}+1)}=[(2\times ?{\mathit{F}}_{\alpha }\times \mathit{f}\times \mathit{D}/{\mathit{V}}_{\mathrm{d}})+{\mathit{C}}_{(\mathit{t}+1)}\times (2-{\mathit{K}}_{\mathit{e}}\times ?\mathit{t})]/2+{\mathit{K}}_{\mathit{e}}\times ?\mathit{t}$

where D is dose, ΔF_a = F_a(t+1) – F_a(t), and Δt = t_(t+1) – t_(t).

Validity of prediction

FDA guidance specifies the importance of assessment of internal and/or external predictability/validity as a requirement for the employment of IVIVC models for official regulatory submissions (Ostrowski et al., 2010).

The %prediction error (PE%) for AUC_0-t and C_max was calculated according to the following equation:

%PE = [Observed Parameter – Predicted parameter) / Observed Parameter] × 100.

The IVIVC model is assumed to be valid if %PE is equal to or less than 10, while %PE between 10% and 20% suggests inconclusive predictability, which requires additional data. %PE greater than 20% indicates inadequate or lack of predictability/validity of the IVIVC model (Ostrowski et al., 2010).

Tablet characterization

Table 1 shows the tablet characterization and drug content for IBU products. The three IBU products (200 mg IBU/tablet) under investigation showed acceptable weight variations, thickness, diameter, and hardness characteristics. The content uniformity of all IBU products met the assay specifications in the USP, where the percentages of IBU were between 90 and 110% (USP-32, 2009).

Table 1. Tablet characterization and content uniformity of IBU commercial products (200 mg/tablet). Data represent mean ± SD (n = 20). [Click here to view]

Figure 1. Dissolution profiles of IBU IR commercial products (200 mg) employing different USP models in phosphate buffer of pH 7.2 at 37°C. Data represent mean ± SD (n = 6). [Click here to view]

In vitro dissolution results

Figure 1 shows the dissolution profiles of IBU products using different dissolution models (USP I, USP II, and USP IV). Table 2 presents IBU multipoints percent dissolved along the 60- minutes dissolution.

As evident from Figure 1 and Table 2, the dissolution of IBU was highly influenced by different dissolution models employed and the manufacturing sites and/or excipient compositions exhibited by various IBU products.

As shown in Figure 1, upon employing the USP I model (basket method), all products showed the lowest percentage of IBU dissolved. Table 2 shows that Q_{60 minutes} ranged from 45% to 76%.

However, upon employing the USP II model (paddle method, the official dissolution model), all IBU products complied with the pharmacopeial dissolution criterion [not less than 80% (Q) of IBU dissolved within 60 minutes] (USP-32, 2009). As observed in Table 2, Advil showed the fastest dissolution profile [80% (Q) in 10 minutes], followed by Nurofen [80% (Q) in 30 minutes], which might result in a rapid onset of action (Emam et al., 2020).

These vast differences in the dissolution patterns among IBU IR products in USP I and USP II could be attributed to the fact that the hydrodynamic environment below the basket in USP I is not as well mixed as that of the paddle in USP II (Todaro et al., 2017), which might cause the product to stick to the basket mesh, hence affecting the dissolution uniformity. In addition, the high agitation force exhibited by the paddle shaft in USP II could cause the higher dissolution of IBU in this model when compared to USP I (Hashem et al., 2019).

On the other hand, Figure 1 and Table 2 show that upon using the USP IV model (FTC method), all IBU products gave fast dissolution results, where more than 80% of IBU dissolved at 30 minutes with the lowest SD values indicating higher reproducibility and uniformity. Previously (Yoshida et al., 2016), the FTC method proved to mimic ideal hydrodynamic conditions for homogenous and mild agitation in contrast to other dissolution models (basket and paddle). FTC uses a piston pump that produces a sinusoidal or semisinusoidal flow, which in turns allows a continuous, uniform flow of the dissolution medium around the dosage form that mimics the natural environment in the GIT (Medina et al., 2017), in addition to maintaining perfect sink conditions which is important for the in vitro dissolution study of poorly soluble drugs such as IBU (Emara et al., 2014b; Forrest et al., 2018).

A previous study compared the in vitro dissolution performance of two IBU generic suspensions available in Mexico against the reference (Advil^® suspension, 2 g/100 ml), employing both the USP II and IV models (Medina et al., 2017). They concluded that the FTC method under the proposed setup (i.e., open system, 22.6 mm cells, laminar flow pattern at a flow rate of 16 ml/min) was suitable for studying the in vitro dissolution of IBU suspensions. They recommended carrying out future in vivo studies to estimate the proposed dissolution methodology (Medina et al., 2017).

Table 2 Percent IBU dissolved after 5 (Q5 minutes), 10 (Q10 minutes), 15 (Q15 minutes), 30 (Q30 minutes), and 60 (Q60 minutes) minutes from IR products in different dissolution models. Data represent mean ± SD (n = 6). [Click here to view]

Table 3. Fit factors’ values comparing the dissolution profiles of IBU commercial products in each dissolution model. [Click here to view]

Lu and Fassihi studied the dissolution of commercial IBU IR 200 mg tablets available in the United States using both apparatuses (USP I and II) in phosphate buffer pH 7.2 at rotation speeds of 50 and 100. Their results confirmed comparable dissolution profiles among the studied IBU products in both USP models, proposing the capability of apparatus interchangeability (Lu and Fassihi, 2017).

Comparative evaluation of in vitro dissolution data

Table 3 shows the fit factor values (f₁ and f₂) upon comparing the dissolution profiles of IBU products with each other in each dissolution model. Fit factors data showed dissimilar dissolution profiles of IBU products employing USP I and II models. For USP IV, only Advil and Nurofen exhibited similar dissolution profiles (with f₁ and f₂ values of 11 and 50, respectively).

From another perspective, Table 4 shows the values of the fit factors comparing the dissolution profiles of each IBU product upon employing different dissolution models with each other. Fit factors’ data revealed dissimilar dissolution profiles for each IBU product among the three USP methods.

Overall results concluded from Tables 3 and 4 emphasize that varying the in vitro dissolution testing model, and variations related to changes in the manufacturing sites and/or formulation differences within the tested products, led to the changes in dissolution profiles of IBU products.

In addition to fit factors, Table 5 presents the MDT and DE at 60 minutes (%DE_{60 minutes}) mean values for comparing dissolution profiles of IBU products in different dissolution models. Significant differences were found in MDT and %DE_{60 minutes} values (p? 0.05) for all IBU products using the USP I, II, and IV models.

Prediction of IBU in vivo plasma concentration-time profile from in vitro data

Inspection of the above dissolution results showed the possibilities of generating IVIVC for the three commercial products due to the effect seen upon changing the dissolution model employed as well as product manufacturing site and/or formulation variables.

Table 4. Fit factors’ values comparing dissolution profiles of each IBU commercial product using different dissolution models. [Click here to view]

Table 5. Dissolution parameters calculated with mean dissolution time (MDT) and dissolution efficiency at 60 minutes (%DE60 minutes). Data represent mean ± SD (n = 6). [Click here to view]

For the establishment of the IVIVC approach, simple spreadsheet software was employed for data calculation and conversion, with the in vitro dissolution data used as the input function. The actual PK data of the IBU IR commercial product (200 mg) were obtained from a previously published study (Shin et al., 2017), where the bioavailability of the IBU IR tablet (Brufen, 200 mg) was investigated in 36 healthy human volunteers under fasting conditions. The maximum observed plasma concentration (C_max) was 24.1±4.10 mg/L. The area under the plasma concentration–time curve AUC_0-12 and AUC_0-∞ were 79.60 ± 13.90 and 80.70 ± 15 mg.hour/L, respectively. These data were used for IVIVC purposes.

Figure 2 shows the corresponding predicted IBU plasma concentrations for the three commercial products in the USP I, II, and IV apparatuses. The observed results indicated that drug in vitro input data variations were well reflected on the predicted in vivo profiles. The observed high predicted plasma profiles for the IBU commercial products in USP II and IV were well reflected with their corresponding in vitro results, where more than 80% of the drug was dissolved within 60 minutes (Figures 1 and 2). On the other hand, USP I represented a scenario in which less than 80% of the drug dissolved (Figure 1), which was well reflected in the in vivo results (Figure 2).

Figure 3 shows IVIVC models plotted between the predicted plasma concentrations in different dissolution apparatuses against actual plasma concentrations obtained from (Shin et al., 2017) to determine which USP model provided the best correlation with the in vivo results. The suggested in vitro/in vivo relationships are described by linear regression and presented by R², slope, and intercept, with the results tabulated in Table 6.

Good correlation, with R² values ≥ 0.96, could be observed for all IBU products in the USP II and IV models (Figure 3 and Table 6). The best IVIVC models with R² values ≥ 0.99 and intercept values close to zero were observed in the following cases: Advil product in the USP II and IV models as well as Brufen and Nurofen in the USP IV model.

Validation of the IVIVC

Table 7 presents predicted pharmacokinetic parameters, C_max, AUC_0-12, and AUC_0-∞ and the percentages prediction error (%PE), of the IBU IR products in the studied USP models. In the USP I dissolution model, the %PE between the actual and predicted C_max, AUC_0-12, and AUC_0-∞ values were >20% for the three commercial products (Table 7), which is indicative of the lack of validity of the suggested IVIVC model (Ostrowski et al., 2010).

Figure 2. Plasma profiles of IBU IR commercial products in different dissolution models predicted by back-calculation of the Wagner-Nelson method. Results are presented as mean ± SD (n = 6). [Click here to view]

Figure 3. Predicted concentration against observed concentration for IBU IR commercial products using deconvolution approach, dotted lines represent the best linear regression for each IBU product. [Click here to view]

On the other hand, in the USP II dissolution model, %PE regarding C_max values was >20%. However, %PE of AUC_0-12 and AUC_0-∞ values for Brufen and Nurofen ranged from 10.72 to 13.03%, indicating inconclusive validation, and hence, additional data is preferably needed. Only the Advil product exhibited %PE values of 6.21% and 5.22% for AUC_0-12 and AUC_0-∞, respectively (Table 7).

Yet, upon employing the USP IV model, %PE regarding C_max values were 10%–20%; also, %PE for AUC_0-12 and AUC_0-∞ were ≤10%, which met the US-FDA acceptance limit (i.e., the ideal IVIVC model is assumed if %PE equals or is less than 10) (Ostrowski et al., 2010). Low prediction error values observed for the USP IV dissolution model strongly confirm the validity of the IVIVC model and its predictive ability to estimate IBU in vivo performance from IR commercial products.

It is worth mentioning that the FDA requirements for IR products focus on a single-point sampling approach, where a single sample at one specified time point (60 minutes) is required to meet an acceptance criterion (≥80% Q). This approach is usually employed for QC purposes. Yet, such an approach is not appropriate during product and/or method development stages or minor changes of excipients after product approval, which should be better reflected by the multipoint sampling approach (Zhang et al., 2007). Hence, for a more discriminating dissolution method and reducing the incidence of unexpected results, multiple sampling time points and sample withdrawals are preferred (Zhang et al., 2007).

The present study outlined the differences observed among commercially available IBU products during the early, mid, and late stages of dissolution employing different USP dissolution models. Both USP II and IV succeeded as single-point dissolution assays in yielding around 80% of IBU dissolved in 60 minutes. However, actual differences related to formulation variables or changing manufacturing sites were better explained by sampling at several discrete time points.

Table 6. The obtained R2, slope, and intercept values for the IVIVC models applied for IBU commercial products. [Click here to view]

Table 7. Predicted pharmacokinetics parameters and percentage prediction error (%PE) of IBU IR commercial product (200 mg) in different USP dissolution models. Actual PK parameters were obtained from Shin et al. (2017). [Click here to view]

To date, the ability of a certain dissolution model to predict the bioavailability of commercial products is still under investigation. The present study pointed out the ability of the FTC model (USP IV) to simulate more adequately the in vivo performance of IBU and hence could reflect the bioavailability of the drug from IR tablets. The effective use of such a simple predictive approach could support in saving valuable resources in terms of budgets and increased costs associated with drug development within pharmaceutical industries.