(M1230-11-63) Prediction Model for the Dry Film Thickness in the Continuous Manufacturing of Oral Films

Purpose: Oral film strips are thin, flexible dosage forms designed to adhere to the oral mucosa, where they rapidly disintegrate and release the active pharmaceutical ingredient(s). Their advantages over conventional dosage forms include enhanced patient compliance, ease of administration without the need for water, rapid onset of action due to fast dissolution and potential for oromucosal absorption and stability of the active ingredients. The thickness of the wet film during the manufacturing process is a surrogate parameter for the content of the API measured in mass per area. While the wet film thickness (WFT) is a critical process parameter (CPP), the thickness of the dried film is a critical quality attribute (CQA) of the finished drug product.¹ Oral films require mechanical stability for subsequent cutting, packaging, and patient application. To the best of the authors knowledge, literature does not describe prediction models for the dry film thickness (DFT) of oral films.By employing a predictive model for the dry film thickness within the continuous manufacturing process, adherence to the demands on quality can be improved. Additionally, the model may facilitate early intervention in the process reducing waste. Through the utilization of process analytical technologies (PAT), real-time monitoring and model predictive control of the oral film production process can enable precise regulation of CPP, thereby enhancing efficiency, reproducibility and ultimately ensuring quality control.^2,3

Methods: Continuous manufacturing of oral films: Films were manufactured utilizing the continuous coating machine TGM-K-1.4 (Optimags, Dr. Zimmermann GmbH, Karlsruhe, Germany) (Figure 1). The polymer solutions were continuously conveyed via a gear pump through the coating knife onto a release liner. The coated film was subsequently transported through the drying channel which is equipped with two independently adjustable heating plates. After the drying step the dried films were coiled up onto a roll. Wet film thickness determination: For the WFT determination a chromatic confocal probe (CHRocodile S, Precitec, Gaggenau, Germany) was equipped. The measuring position was adjacent to the coating knife. A measuring range of 3 mm and a scan rate of 2000 Hz were applied. Mass and dry film thickness determination: To characterize the resulting mass and dry film thickness, samples of 10 cm² were taken just before the upcoiling of the oral film. The thickness was analyzed using a micrometer screw (Series 331, Mitutoyo, Kawasaki, Japan), while the films were weighed using a balance (CP 224 S Sartorius, Göttingen, Germany). Data processing: The data was generated in a Design of Experiments (DOE) approach, wherein influential factors were identified and utilized to develop a predictive model via an inhouse Python script. For the coefficient evaluation, a response surface modeling approach was employed. The model underwent tuning via Monte Carlo cross-validation to optimize the R-squared metric. Polymers: Various concentrations of different polymers, such as hydroxypropyl mehtylcellulose (HPMC), polyvinyl alcohol (PVA), and polyvinylpyrrolidone (PVP), were investigated. Regarding the solid content of the films, 3 % glycerol (Glycerol 85 %) was employed as a plasticizer. Demineralized water or a mixture of water and ethanol (1:1, w/w) was used as solvent. The films were applied onto a single-sided silicone-coated polyethylene terephthalate liner (Silphan S 75 M, Siliconature, Godega di Sant'Urbano, Italy).

Results: Figure 2 shows the coefficient plot for HPMC solutions. As expected, the experiments reveal that the concentration of polymer solution and WFT exhibit positive coefficients. The liner speed, affecting drying time in the drying tunnel, possesses a negligible negative coefficient. Given the pronounced coefficients observed for the WFT, continuous and accurate measurement thereof allows the prediction of the DFT. These findings could be extrapolated to investigations involving other polymers. It was observed that experiments conducted with varying polymer concentrations provided sufficient data output, enabling the incorporation of rheological properties relevant to the casting process. However, the rheological properties of the polymers appear to be less relevant for predicting DFT Through the experiments, we successfully developed a predictive model for the DFT across various polymers, as exemplified shown for HPMC in Figure 3. Looking at the summary of fit for our model we observe a R² of 0.890 and a R² _LOOCV of 0.848.

Conclusion: The predictive model facilitates the identification of optimal process parameters by elucidating the relationship between polymer concentration, wet film thickness, and dry film thickness. This insight enables manufacturers to tailor their processes for consistent and precise film thickness control. Real-time monitoring and prediction of dry film thickness enable proactive quality control measures, ensuring the uniformity and consistency of film products. Also, the early detection of deviations in predicted dry film thickness allows a previous intervention and for removal of non-conforming products. In conclusion, the development of a predictive model for dry film thickness across various polymers, based on polymer concentration and wet film thickness, offers a holistic approach to process optimization, quality assurance, resource management, and product performance enhancement.

References: [1] Preis, M.; Knop, K.; Breitkreutz, J., Mechanical strengh test for orodispersible and buccal films. Int J Pharm, 461, 22-29 (2014).
[2] Rantanen, J.; Khinast, J., The future of pharmaceutical manufacturing sciences. J Pharm Sci, 104, 3612-3638 (2015).
[3] Niese, S; Quodbach, J. Application of a chromatic confocal measurement system as new approach for in-line wet film thickness determination in continuous oral film manufacturing processes. Int J Pharm, 551, 203-211 (2018).


Figure 1:Schematic drawing oft the TGM-K-1.4 (Optimags, Dr. Zimmermann, Germany) with the mounted CHRocodile S probe (blue. Precitec, Germany).


Figure 2: Coefficient plot after Monte Carlo cross-validation for HPMC.


Figure 3: Resulting predictive model for HPMC.