The drug release rate of pharmaceutical drug products is critical for ensuring product effectiveness and patient safety. This is controlled by the thickness of the coating layer applied to drug product particles. This study demonstrated that in-line particle size analysis is a viable method to predict dissolution performance and thus can be used to control a coating process.
Process Analytical Technology (PAT) is a mechanism for measuring and controlling process performance parameters and Critical Quality Attributes (CQA’s) with the goal of ensuring final product quality. Comprehensive implementation of PAT can compensate for process variability within normal operating conditions to ensure process and product quality is managed. Off-line measurement to monitor and control processes introduces a delay between process sampling and the availability of results. In some traditional sampling approaches, the physical act of sampling introduces additional variations that may impact resulting data. In-line PAT facilitates immediate access to process data enabling rapid decision making to optimise process conditions during development and manufacture. Additionally, the application of in-line PAT supports the automated collection of process data for continuous process verification. Using real time analysis, feedback control tools can be implemented to modify conditions for process optimisation.
Many PAT tools are currently available that support the real time measurement of critical quality attributes based on physical or chemical characteristics. Dissolution, a CQA that directly impacts the final drug product quality, cannot be directly measured in-line due to sampling time requirements of standardised dissolution testing. Many solid dosage forms, including pellets and multiparticulates, are coated for modified release. Enteric or delayed coatings can increase bioavailability and improve patient acceptance while sustained release coatings can increase patient compliance and enhance convenience. Fluid Bed or Wurster coating processes are routinely applied to a range of core materials to produce the required drug release profile. The correlation of dissolution performance to data currently measured with the implementation of PAT provides an opportunity to predict dissolution test results for a modified release product.
Presented here are several elements of a study to predict the dissolution performance of polymer-coated multiparticulates using in-line PAT. In-line multiparticulate data is first assessed, and an in-line and off-line particle size measurement technique comparison is presented for a selection of substrates and functional coating processes. The correlation between multiparticulate film thickness and predicted weight gain is investigated and finally dissolution performance predictions and results are compared.
The experimental processes and associated formulations are presented in Table 1. Chlorpheniramine Maleate (CPM) and Propranolol HCI (PRP) coated Colorcon Suglets® were coated with Surelease® and/or Opadry® EC functional coatings. CPM-SR-1 and CPM-SR-2 are identical with the intention of demonstrating repeatability of results and dissolution performance prediction.
Process settings for functional coating materials were based on recommendations by Colorcon and applied for a predicted weight gain of 18-20 per cent. Colorcon also recommend a 1-hour cure time for the aqueous-based Surelease functional coating.
Based on the quantity of coating solution sprayed, samples were extracted from the process at points corresponding to a predicted weight gain of 2.5 per cent, 5 per cent, 7.5 per cent, 10 per cent, 12.5 per cent, 15 per cent, 17.5 per cent and 20 per cent. Surelease based samples were also taken at 30 mins and 1 hour of curing. In-line particle size data was collected at these intervals also for comparative studies. Collected samples were sent for analysis by Colorcon’s analytical lab to determine dissolution and separately measure off-line Particle Size Distribution (PSD) data using a Camsizer system.
For process experiments CPM-SR-1 and CPM-SR-2, sugar spheres (Suglets®, Colorcon) mesh size 18/20 (850-1000μm) were layered with Chlorpheniramine maleate (CPM). The CPM layered pellets were coated with Surelease aqueous ethylcellulose dispersion (E-7-19040, Colorcon) as a barrier membrane coating and Opadry Hypromellose based coating system (YS-1-19025-A, Colorcon) as a pore former at 80:20 ratios. Preparation of the coating dispersion solution comprised the dissolution of Opadry in deionized water. This was then added to Surelease to obtain total solid content of 15 per cent w/w.
In process experiments CPM-EC and PRP-EC,Opadry EC ethylcellulose organic coating system (505O190028, Colorcon) was used as an alternative fully formulated barrier membrane organic coating to evaluate the performance on CPM and PRP layered pellets respectively. Opadry EC coating solution was prepared in Ethanol and water at 90:10 ratios.
Sugar spheres (Suglets®, Colorcon) mesh size 20/25 (710-850 μm) were layered with propranolol HCl (PRP) for process experiment PRP-SR. The PRP layered pellets were coated with Surelease aqueous ethylcellulose dispersion (E-7-19040, Colorcon) and Opadry Hypromellose based coating system (YS-1-19025-A, Colorcon) as a pore form.
er at 80:20 ratios as described previously in CPM-SR-1.
The targeted coating weight gain was 18-20 per cent for all experiments and samples were taken at every 2.5 per cent predicted weight gain (WG).
The Wurster coating process is commonly used to apply coating films of high quality and uniformity. It can be applied to a wide range of base materials of varying particle size and shape. The use of differential air flow streams to create a cyclic flow of particulates inside a chamber with a spray nozzle located at the bottom of the fluidised bed ensures a uniform application of desired coating thickness. A diagram of the Wurster bowl and component parts is presented in Figure 1. Critical process parameters include Spray Rate, Atomizing Air Pressure, Air Volume, Product Temperature, Orifice Plate Configuration and Partition Height. These require monitoring and control to ensure the process produces the desired quality.
A Glatt GPCG-2 lab system was deployed for this experiment. It is a lab scale unit commonly used for formulation and process development, offering flexibility for a range of fluid bed processes. Figure 2 shows the product container used for this study, including the SD-55windows added to accept non-product contact PAT. The Eyecon2 device was installed on the lowest positioned window of the product container to deliver optimal measurement of particulates during Wurster processing. In general, the application of in-line PAT to capture data such as PSD or moisture content during fluid bed coating processes allows for greater process understanding and control from product design through to commercialisation.
The Eyecon2 from Innopharma Technology was used to measure in-line particle size distribution data. The Eyecon2 employs the principles of direct-imaging to capture images of flowing or static material and uses advanced digital image processing to provide data on the particle size and shape distribution. The Eyecon2 can be used at-line or in-line and has application in typical oral solid dose processes including granulation, milling and fluid bed coating. Implementation of the Eyecon2 can significantly reduce analytical time and increase process knowledge from development to commercial manufacturing.
Figure 3 shows the Eyecon2 mounted on the lower window of the Glatt GPCG-2 lab system to provide an in-line measurement device without the necessity of product contact. Such positioning offered the opportunity to capture dense images of the multiparticulates, maximising the number of particles captured per image, and therefore minimising the time required to obtain representative measurements.
An image captured by Eyecon2 of the CPM-coated multiparticulates partway through the Surelease / Opadry functional coating process is presented in Figure 4.
An off-line instrument, Camsizer, using the principle of backlight imaging and digital processing was used as an alternative method to measure particle size distribution for comparison with the In-line data.
Drug release was measured using a UV spectrometer of 1 gram of CPM and PRP barrier membrane coated pellets in a dissolution bath using USP apparatus I (baskets) at 100 rpm. USP purified water was used as a dissolution media (1000 ml) at 37.0 ± 0.5 °C.
Graph 1 presents Eyecon2 in-line particle size data collected during experimental process CPM-SR-1 at the sample intervals of predicted weight gain discussed earlier i.e. at every 2.5 per cent predicted weight gain up to a value of 20 per cent, plus two additional samples during the curing phase. The particle size distribution is defined at each sample interval using the Dv10, Dv50 and Dv90 values. Dv50 represents the median particle diameter for the volume distribution, while 10 per cent and 90 per cent of the population distribution lies below the Dv10 and Dv90 values respectively. These values confirm a steady rate of particle growth over time, with a negligible rate of growth during the curing step as expected.
Real-time on-screen images captured with the Eyecon2 during the start and end of the film coating process are presented in Figures 5 and 6 respectively. In-line images can provide subjective supporting information for process understanding and performance investigation.
To demonstrate in-line process measurement repeatability, the Eyecon2 in-line particle Dv50 data collected during process experiments CPM-SR-1 and CPM-SR-2 is presented for comparison in Graph 2. The formulation for both process experiments is identical and the resulting plots demonstrate little variation. Use of the Eyecon2 over time will facilitate the identification of control limits and the accepted variability of the process.
A comparison of the particle analyser distribution data from the Eyecon2 and off-line Camsizer is presented in Graph 3. Reported are the Dv10, Dv50 and Dv50 values produced by each particle analyser instrument for process CPM-SR-1. Some variation is expected as different instruments can use different measuring techniques and hence yield marginally different results. However, a strong correlation between the particle size distribution data for each method for the duration of the process is apparent.
The correlation of the on-line and off-line data is further explored in Graph 4 where a linear plot of the Eyecon2 in-line values against the equivalent Camsizer off-line values are presented. The resulting trendline R-squared values for each data set i.e. Dv10, Dv50 and Dv90, indicate a high correlation particularly for Dv50 with a value over 0.98.
To further determine the correlation of the in-line and off-line measuring techniques, all Dv50 data collected during the sample intervals of each of the five process experiments is presented in Graph 5. The resulting trendline r-squared value of 0.9895 offers a high degree of confidence that the Eyecon2 particle analyser is consistently producing results comparable to the offline Camsizer system.
The data clustered around 850um on the X-axis represent the smaller, PRP-coated starting substrates while the data points clustered around 975um represent the larger CPM-coated substrates.
Dissolution performance of multiparticulates is influenced by particle size. The thickness of applied functional coatings for multiparticulates will directly impact the dissolution process. For each of the five experimental processes, the functional coating was applied to achieve a predicted weight gain of 20 per cent. As coating densities vary due to differences in functionality, variations in total film thickness are expected across the experimental processes to achieve the relevant predicted weight gain.
In-line PSD data from the Eyecon2 can be used to evaluate coating thickness and demonstrate the correlation between film thickness and predicted weight gain. The film thickness (f ) at a given sample interval can be defined as the increase in reported particle size diameter divided by two, as seen in Figure 7. As a range of PSD descriptive parameters are available from Eyecon2, three methods to calculate coating thickness are explored to determine the most suitable for dissolution predictions. These are the difference in Dv50 percentiles, the difference in the average of the most common percentiles of Dv10, Dv50 and Dv90, and the difference in the average of all the percentile data measured by the Eyecon2.
A comparative representation of the data collected in the experimental process CPM-SR-1 presented in Graph 6 indicates a close match between the three methods.
A good correlation for in-line and off-line Dv50 parameter measurements was previously demonstrated. Due to the similarities of the three coating thickness calculation methods, the difference in Dv50 values has been selected for coating thickness representation in the dissolution prediction element of this study. The relationship between calculated coating thickness and predicted weight gain for experiment CPM-EC is presented in Graph 7. Results confirm a similar trend to the CPM-SR-1 process however, the coating thickness is lower as expected due to the difference in the functional coating compositions.
Several factors affect drug dissolution rate. Prediction of the dissolution rate using in-line measurement techniques necessitates a formulation model specific to the coated multiparticulates under test. A dissolution prediction model was designed using the dissolution test results of experimental process CPM-SR-1. The formulation and process parameters were repeated in CPM-SR-2 and the dissolution prediction model was used to predict CPM-SR-2 dissolution performance and a comparison is made against actual results.
The dissolution rate data for the multiparticulate samples collected during experimental process CPM-SR-1 at the predicted weight gain intervals is presented in Graph 8. As previously discussed, each predicted weight gain sample correlates to an equivalent coating thickness.
The relationship between dissolution performance and equivalent film thickness calculated is presented in Graph 9. Each best fit polynomial represents a specific time in the dissolution medium. For example, a film thickness of 20um yields a much lower dissolution performance at 15 minutes compared with 240 mins as expected. These polynomials can be used to predict dissolution performance based on calculated film thickness as determined from in-line particle size measurements.
Using the polynomial equations displayed in Graph 10 and in-line particle distribution data collected in experimental process CPM-SR-2, dissolution performance is predicted. The results are presented in Table 2. A graphical representation of the data is presented in Graph 10.
To corroborate the validity of the prediction model, the predicted (P) and actual (A) dissolution data is presented in Graph 11. The close overlay generally observed between the predicted and measured data shows the viability of the prediction model. To improve the accuracy of the dissolution prediction model for deployment, an expansion of the data set is recommended. Inclusion of CPM-SR-2 actual data in building the prediction model will further enhance the viability and value of the prediction model for subsequent repetition and dissolution prediction of the process.
• A strong correlation was demonstrated between in-line and off-line digital image analysis to measure particle size distribution
• Data from the Eyecon2 demonstrated a direct relationship between predicted weight gain and functional coating thickness
• A clear correlation was established between functional coating thickness and dissolution profile
• It was established that in-line particle size data analysis is a viable method to predict dissolution performance using formulation specific models.
It is also noted that the application of PAT in this study facilitated real time process understanding with the support of in-line data and images. This has the potential to enhance process development, control, optimisation, investigation, scale up and transfer.
Using Colorcon products and process knowledge, multiple datasets are potentially available to develop robust predictive models for deployment.
The Glatt GPCG2 lab system unit offers flexibility to facilitate experimental requirements with different Wurster processing parameters.