100% visual inspection of tablets produced with continuous direct compression and coating.

Visual appearance of tablets is an important property for patients. Since the visual appearance is most strongly influenced by the applied coating, this necessitates a high level of process control and homogeneity in the coating process. In recent years, a number of tablet coaters have been developed that can be used in combination with continuous tablet production lines. In this study, 180 kg of tablets were produced using a continuous direct compaction line with a throughput of 25 kg/h. Tablets were consequently subdivided into 12 lots and coated in a semi-batch drum coater directly after compression. For a detailed understanding of intra-lot and lot-to-lot variability, a 100% visual inspection of the tablets was performed using an automatic tablet inspection and sorting machine. All tablets were analyzed from all 6 sides and the unsuitable tablets were sorted out. In the worst lot, only 1 out of around 300 tablets was sorted out due to color mismatch. For some tablets, edge chipping was also observed, which would presumably not be detected during routine sampling. Root causes for the defects could be found in the intentionally chosen set of old punches and in the operation parameters of the coater. Nonetheless, the lot-to-lot variability according to all criteria was very low.

Optimization of a semi-batch tablet coating process for a continuous manufacturing line by design of experiments.

The aim of the study was to optimize a tablet coating process for a continuous manufacturing line. High throughputs should be achieved while inter-tablet coating variability should be as small as possible. Drug-free cores were coated with a colored suspension. All processes were monitored in-line with Raman spectroscopy. A statistical design of experiment was performed to find optimum process parameters. Tablet loading, spray rate and drum rotation speed were studied. Image analysis was performed using a computer scanner. Tablet hue and saturation were evaluated to obtain information about the inter-tablet color variabilities and the numbers of outliers. Low variabilities could be achieved using low spray rates and high rotation speeds and they were independent from the tablet batch sizes in the studied factor space. For the prediction of the coating thickness, univariate analysis was compared to PLS-regression. Calibration models were built based on the three center points of the statistical design of experiment resulting in RMSEC of 1.07% of sprayed suspension with R2 of 0.9989 and Q2 of 0.9987. Model prediction was possible independent from loading, spray rate and drum rotation speed. The experiment with lowest color variability was conducted with a desired throughput rate of 25 kg/h and with a RMSEP of 2.5%.

Monitoring of tablet coating processes with colored coatings.

Endpoints of coating processes for colored tablets were determined using in-line Raman spectroscopy. Coatings were performed with six commercially available formulations of pink, yellow, red, beige, green and blue color. The coatings were comprising pigments and/or dyes, some causing fluorescence and interfering the Raman signal. Using non-contact optics, a Raman probe was used as process analytical technology (PAT) tool, and acquired spectra were correlated to the sprayed mass of aqueous coating suspension. Process endpoints were determined using univariate (UV) data analysis and three multivariate analysis methods, namely Projection to Latent Structures (PLS)-regression, Science-Based Calibration (SBC) and Multivariate Curve Resolution (MCR). The methods were compared regarding model performance parameters. The endpoints of all coating experiments could be predicted until a total coating time of 50min corresponding to coating thicknesses between 21 and 38µm, depending on the density of the coat formulation. With the exception of SBC, all calibration methods resulted in R2 values higher than 0.9. Additionally, the methods were evaluated regarding their capability for in-line process monitoring. For each color, at least two methods were feasible to do this. Overall, PLS-regression led to best model performance parameters.

Evaluation of in-line Raman data for end-point determination of a coating process: Comparison of Science-Based Calibration, PLS-regression and univariate data analysis.

A multivariate analysis method, Science-Based Calibration (SBC), was used for the first time for endpoint determination of a tablet coating process using Raman data. Two types of tablet cores, placebo and caffeine cores, received a coating suspension comprising a polyvinyl alcohol-polyethylene glycol graft-copolymer and titanium dioxide to a maximum coating thickness of 80µm. Raman spectroscopy was used as in-line PAT tool. The spectra were acquired every minute and correlated to the amount of applied aqueous coating suspension. SBC was compared to another well-known multivariate analysis method, Partial Least Squares-regression (PLS) and a simpler approach, Univariate Data Analysis (UVDA). All developed calibration models had coefficient of determination values (R2) higher than 0.99. The coating endpoints could be predicted with root mean square errors (RMSEP) less than 3.1% of the applied coating suspensions. Compared to PLS and UVDA, SBC proved to be an alternative multivariate calibration method with high predictive power.