Exploiting Pharma 4.0 Technologies in the Non-Biological Complex Drugs Manufacturing: Innovations and Implications

Abstract

The pharmaceutical industry has entered an era of transformation with the emergence of Pharma 4.0, which leverages cutting-edge technologies in manufacturing processes. These hold tremendous potential for enhancing the overall efficiency, safety, and quality of non-biological complex drugs (NBCDs), a category of pharmaceutical products that pose unique challenges due to their intricate composition and complex manufacturing requirements. This review attempts to provide insight into the application of select Pharma 4.0 technologies, namely machine learning, in silico modeling, and 3D printing, in the manufacturing process of NBCDs. Specifically, it reviews the impact of these tools on NBCDs such as liposomes, polymeric micelles, glatiramer acetate, iron carbohydrate complexes, and nanocrystals. It also addresses regulatory challenges associated with the implementation of these technologies and presents potential future perspectives, highlighting the incorporation of digital twins in this field of research as it seems to be a very promising approach, namely for the optimization of NBCDs manufacturing processes.

Keywords: Pharma 4.0; digital twins; in silico modeling; machine learning; non-biological complex drugs; three-dimensional (3D) printing.

Figure 1

Emerging tools of Pharma 4.0 and their main advantages in drug manufacturing.

Figure 2

Illustrative depiction of non-biological complex drugs (NBCDs) and descriptions of their main technological advantages.

Figure 3

(A) Images obtained via scanning electron microscopy of (a) top layer of Eudragit in the printfills, (b) CPT crystals within the printfills, (c) micelles loaded with CPT inside of the printfills, and (d) magnified polymeric micelles inside the printfills. (B) Cumulative release of CPT from the printfills in simulated gastrointestinal fluids. The release of CPT-loaded micelles is depicted with squares, while the release of free CPT is indicated with circles. (C) The cell viability of the dissolution medium sourced from the printfills containing CPT-loaded micelles (depicted as blue columns) and free CPT (illustrated with pink columns) was assessed following 4, 6, and 24 h of incubation with Caco-2 cells. (D) Intestinal permeability and respective TEER values of CPT across a standard Caco-2 cell model (a) and across a 3D intestinal model (b) and apparent permeability coefficients of CPT-loaded micelles (blue) and free CPT (pink) in both Caco-2 monoculture model and the 3D model (c), where * p < 0.05, *** p < 0.001 and **** p < 0.0001 (E) H&AND staining to assess cellular integrity following exposure to the dissolution medium containing CPT-loaded micelles and free CPT during permeability assay. The cytoplasm was stained pink, while the nucleus was stained purple. The transwell membrane, positioned just beneath the cellular monolayer, remains transparent [77]. CPT: camptothecin; H&AND: hematoxylin, and eosin; TEER: transepithelial electrical resistance.

Figure 4

(A) Schematic depiction of the ISGS manufacturing process via MeltDrops technology. (B) Molecular interactions forming between the components of the formulation at two different temperatures: 300 K (a) and 308 K (b): Green CPK model = TIM; Red = DRZ; Blue = Sodium chloride; Yellow = Benzalkonium chloride; Pink = HPMC; Orange = Poloxamer 407; Green wire model = Poloxamer 188; Blue = water. (C) Cumulative release of timolol maleate (TIM) and dorzolamide hydrochloride (DRZ) over time from both MeltDrops and commercially available solutions. (D) Percentage decrease in IOP following the administration of MeltDrops and commercially available eyedrops. (E) HET-CAM test results after the application of (a) MeltDrops with no signs of irritation, (b) 0.9% w/v saline solutions, also showing no signs of irritation, and (c) 0.1 N Sodium hydroxide solution, revealing features such as vascular lysis, coagulation, and hemorrhage [86]. DRZ: dorzolamide hydrochloride; IOP: intraocular pressure; ISGS: in situ gelling system; TIM: timolol maleate.

Figure 5

(A) Comparative illustration of MIMO model (a) and MISO model (b). (B) Graphical user interface of the ANN for liposome particle size and PDI prediction in continuous liposome manufacturing. (C) Comparison of predicted vs. target values for liposome particle size in the training and testing sets, without molecular descriptors. (D) Comparison of predicted vs. target values for liposome PDI in the training and testing sets, without molecular descriptors. (E) Evaluation of MRE with and without incorporating molecular descriptors in the ANN input: (a) particle size and (b) PDI [103]. ANN: artificial neural network; MIMO: multiple-input–multiple-output; MISO: multiple-input–single-output; PDI: polydispersity index; MRE: mean relative error.

Figure 6

(A) Snapshots captured during 1 microsecond of molecular dynamics simulations, alongside the experimental size distribution of NAP liposomes (a–c,d1) and PAL liposomes (a1–c1,d2). (B) Heat map assessment illustrating the connection between logS, molecular complexity, and XLogP3 of the drug with encapsulation within liposomes. The numerical values within each matrix indicate the frequency of encapsulation occurrences corresponding to each specific property. (C) A scatter plot displaying the comparison between experimental and predicted values for size, PDI, zeta potential, and encapsulation [98]. NAP: naproxen; PAL: palmatine HCL; PDI: polydispersity index.

Figure 7

(A) Scatter plots comparing the predicted values obtained from the machine learning algorithm with the experimental values for nanocrystal size on the training, validation, and test subsets within the BWM, HPH, and ASP size datasets. (B) Contrasts in the predicted values generated by the algorithm with the actual sizes of nanocrystals prepared with BWM (a–d), HPH (e–g), and ASP (h–k) [105]. ASP: antisolvent precipitation; BWM: ball wet milling; HPH: high-pressure homogenization.

Figure 8

(A) Scatter plots comparing the predicted values obtained from the machine learning algorithm with the experimental values for nanocrystal PDI on the training, validation, and test subsets within the BWM, HPH, and ASP size datasets. (B) Contrast the predicted values generated by the algorithm with the actual PDI values of nanocrystals prepared by BWM (a–d), HPH (e–g), and ASP (h–k) [105]. ASP: antisolvent precipitation; BWM: ball wet milling; HPH: high-pressure homogenization; PDI: polydispersity index.