Cleaning Validation Practices: Using a One-Pot Processor

During recent years, concern regarding the cleanability of pharmaceutical processing equipment and operator exposure to active pharmaceutical ingredients (APIs) has been steadily growing, driven mainly by the increasingly strict regulations regarding operator safety and the occurrence of more potent active compounds. As a result, most process equipment is now equipped with integrated clean-in-place (CIP) systems. This article examines the CIP system of a one-pot processor and describes how effective it is at removing two different types of drug compound - water-soluble theophylline and water-insoluble mebendazole.

Methods and materials: theophylline

Materials. The theophylline batch formulation comprised 65% lactose 200 mesh (DMV International, The Netherlands), 5% PVP K25 (BASF, Germany) and 30% theophylline (Medeva Pharma, Belgium). Two per cent P3-cosa CIP-95 detergent (Ecolab, Belgium) was used to wash the theophylline. P3-cosa CIP-95 is a detergent based on alkali and complexing agents, and a 1% solution has a pH of 12.3–12.7 at 20 °C in deionized water. It does not contain tension-active ingredients. The one-pot processor used was the UltimaPro 75 (Collette NV, Belgium).

Batch production. All raw materials were loaded into the system's bowl using its vacuum loading system and then dry mixed. Water (1.8 kg) was added using a pressure vessel to granulate the material, and the batch was dried using vacuum, microwaves and the swinging bowl. A dust-free method was used to discharge the batch into a drum. Total production time was 50 min, during which the processor was kept closed to simulate actual conditions when working with potent compounds.

Cleaning cycle. The cleaning cycle began by prewashing the product feed tube, the gas-assisted vacuum system (Transflo), the liquid addition system and the bowl with water (35 L, 40 ºC).

After discharging the prewash water, the product filter was cleaned using hot water and detergent. The swinging bowl was filled with detergent, and both the bowl and lid were cleaned whilst the mixer and chopper operated at high speed. The solution was discharged and the discharge valve was cleaned with hot water and detergent. The total volume of hot water (60 °C) used was 80 L.

Table I: Theophylline sample analysis.

Cold water (60 L, 20 °C) was used to remove any residual detergent followed by a final rinse with demineralized water (50 L, 20 °C). The supply lines for the cleaning media were blown dry with purified process air. Finally, the machine was dried using vacuum, gas-assisted vacuum and the heated jacket. The total cleaning cycle took approximately 90 min, 30 min of which were spent drying and preparing the machine for the next batch.

Four theophylline batches were produced to evaluate the cleaning phase's reproducibility. The system was cleaned immediately after each production run with the exception of one batch for which cleaning was delayed by 24 h.

Sampling. Following each batch production, samples were taken from six areas of the processor - the product filter, impeller, chopper, bowl and lid, gas-assisted vacuum system and discharge valve. These regions were selected because it is critical that they are cleaned between runs. Both the stainless steel surfaces of the areas and their surrounding seals were evaluated (Table I). Samples were taken using the swab method, wetting a swab tissue with 2 mL of ultra-pure water. Swab surfaces were predetermined and their exact size was known.

Product filter. Although the total surface in contact with the product is 1524.5 cm2, the sample from the product filter was taken from a surface area of 100 cm2 from the lowest part of the filter, that is, the area in closest contact with the product (particularly during swinging).

Impeller. The wingnut was located centrally at the bottom of the impeller and, because of its location, was included in the study. The impeller blade was one of the largest parts in direct contact with the product. To ascertain exactly how much product remained on the impeller blades (front and back surfaces), it was completely swabbed.

The lipseal of the impeller shaft was deemed to be a critical region of the processor because it is not flush with the lid and no spray balls are available in the lid for this size of machine. This area was included in the study to see whether agitating the water with the mixer was sufficient to clean this part.

Chopper and bowl. The chopper knife and a 100 cm2 surface area of the bowl were deemed critical because these parts were in direct contact with the product. Any remaining residue would, therefore, contaminate subsequent batches.

Discharge valve. Although the contact time between the product and the discharge valve was short, there is considerable risk that residues remaining in this area would contaminate the next batch.

Despite the design of the processor preventing direct contact between the bowl seal and the product, the seal was included as a critical area because water could infiltrate the small gap between the bowl and lid rims. Finally, the Transflo opening was included to assess the possibility of water seeping through the filter and being trapped there. After the cleaning cycle, samples of the final rinse water were collected to ensure elimination of the detergent.

Swab sample analysis. After transfer of the swab in a borosilicate tube (Corning Glass Works, USA), 5 mL of distilled water was added. The tube was vortex-mixed for 10 s, homogenized in an ultrasonic bath for 1 min and 250 mL of the solution was injected into a column.

The chromatographic system consisted of an isocratic pump (L-7110), a variable wavelength UV detector (L-7400) and a PC-interface (D-7000) (complete system manufactured by Merck-Hitachi, Germany), while the integration of the peak area was performed using the HPLC system manager V3.0 (Merck-Hitachi, Germany). The analytical column (25034 mm i.d.) was packed with 5 µm RP-C18 particles (Licrosphere; Merck, Germany). Samples were manually injected using a syringe-loaded injector (Valco six-channel injector; Valco Instruments, USA) and a loop of 250 µL. The mobile phase, consisting of 50% (v/v) MeOH and 50% (v/v) distilled water, was degassed before use and had a flow rate of 1.0 mL/min.

The UV detector monitored the theophylline concentration at 280 nm. All experiments were run at ambient temperature. Stock solutions of theophylline at a concentration of 100 mg/mL were prepared in distilled water, stored at 8 °C in stoppered flasks and used to prepare a set of six calibration standards; concentrations varied (0.01–100 µg/mL). The method was linear across the concentration range (R250.9991). The within-day variability (n55) was 0.28% whereas the inter-day variability compared with the above-mentioned concentration range was determined at 0.33%. The detection limit was determined at 2.1 ng/mL, thus the quantification threshold was 14.7 ng/swab.

Analysis of the rinse samples. To determine detergent residues in the final rinse water, a conductivity measurement was made using the LF 325 conductivity meter (WTW GmbH & Co. KG, Germany). The samples of the final rinse water were compared with blank samples of the demineralized water used for rinsing and samples of a 1:1000 dilution of detergent solution.

Acceptance criteria. Two criteria were calculated to ensure the cleaning cycle was validated - the 10 ppm criterion and the absolute mass criterion. The acceptable quantity of a product per swab can be calculated as follows:

Based on formulation density and the bowl's requirement to be at least one-third full to be impacted by the chopper during granulation, the smallest theophylline batch was 12 kg. Initial trials on the swab determined that the theophylline recovery was between 95–100%. Nevertheless, the Factor 2 for swab yield was maintained in the formula for calculating the acceptance criteria, to account for any operator influence.

For absolute mass, the acceptance limit for an active substance is defined as 1 µg/cm2, which was used as the acceptance criterion in this study. The acceptance criterion for residual detergent traces is that the conductivity of the final rinsing water should be lower than the conductivity of a 1:1000 dilution of the detergent solution.

Methods and materials: mebendazole

Materials. The raw materials for the mebendazole batches were provided in a premixed blend by Janssen Pharmaceutica (Beerse, Belgium). Three detergents (all Ecolab) were used for the cleaning cycle:

• P3-cosa CIP-92, an alkaline detergent containing surfactants. A 1% solution in 20 °C in deionized water has a pH of 11.5–12.5. A 2% concentration solution was used in this study

• P3-cosa PUR-84, a liquid additive for the alkaline detergent based on complexing agents, surfactants and sequestrants; used together with the alkaline detergent at a concentration of 0.75%

• P3-cosa CIP-72, an acid detergent containing different organic acids, used in a concentration of 2% for the neutralization of the alkaline detergent and the removal of the additive.

Batch production. All raw materials were loaded into the bowl using the vacuum loading system and dry mixed. Water (12 kg) was added using a pressure vessel for granulation of the material. The batch was dried using vacuum, microwaves and the swinging bowl before being discharged into a drum using a dust-free method. Total production time was 125 min. During production, the processor was closed to simulate actual conditions when working with potent compounds.

Cleaning cycle. The mebendazole cleaning cycle differed from the theophylline cleaning cycle in the following ways:

• hot water (175 L, 60 °C) was used in the prewash

• for the first cleaning cycle, an alkaline detergent combined with an additive was used

• the cold water rinse was replaced by cleaning with acid detergent to neutralize the alkaline detergent

• after rinsing with acid detergent, a "once-through" cycle of the Transflo, the product filter, bowl, lid and discharge valve with hot water was performed to remove traces of acid detergent.

The total volume of hot water (60 °C) used for all steps in the cleaning cycle was 175 L. The final steps of the cleaning cycle (including 50 L demineralized water) were identical to those of the cleaning cycle for theophylline. The total cleaning cycle took approximately 100 min, 30 min of which were spent drying and preparing the machine for the next batch.

Table II: Mebendazole sample analysis.

To evaluate the reproducibility of the cleaning cycle, three production runs were performed, followed by cleaning. The equipment was cleaned 24 h after the run.

Sampling. Samples were taken from the same areas as those taken for the theophylline runs, that is, product filter, impeller, chopper, bowl and lid, gas-assisted vacuum system and discharge valve. Again, both stainless steel surfaces and seals were evaluated (Table II). Sampling was done using the swab method, although the swab tissue was wetted with analytical grade concentrated methanol. The swab surfaces were predetermined and the exact size known.

The sample area of the discharge valve was the piston rather than the cover as used for the theophylline runs to get a better overview of the cleanability of the valve's parts. Also, for the mebendazole study, samples were taken from the final rinse water to determine detergent residuals.

Swab sample analysis. The swab was dried in the sample tube, after which 20 mL of a formic acid/dimethylformamide mixture (1:9, v/v) was added. The tube was shaken for 60 min and filtered through an Acrodisc 0.45 mm filter (Pall Corporation, USA) and 10 mL of the filtrate was injected into the column.

The chromatographic system comprised an isocratic pump and a variable wavelength UV detector. The analytical column (10034.6 mm i.d.) was packed with 3.5 µm RP-C18 particles (Xterra; Waters, USA). The mobile phase consisted of a mixture of 65% (v/v) demineralized water, 0.5% (w/v) ammonium acetate and 35% (v/v) acetonitrile; its flow rate was 1.0 mL/min. The UV detector monitored the mebendazole concentration at 312 nm and the column temperature was a constant 35 °C. The system suitability was verified with standard solutions of 0.1 and 5 mg/mL. The detection limit was determined at 1.0 ng/mL, thus the quantification threshold was 20 ng/swab.

Rinse sample analysis. The samples of the final rinse water were analysed using TOC (Sievers 820 TOC Analyzer; Ionics Instruments, USA).

Acceptance criteria. As with the theophylline runs, the 10 ppm criterion was considered. The acceptable quantity of a product per swab could be calculated as previously described.

The smallest mebendazole batch size that could be used in the processor was 8 kg, based on the density of the formulation and the requirement for the bowl to be filled to at least one-third of its total volume for adequate impact of the chopper during granulation. The second criterion, and the one used as the acceptance criterion in this study, was the absolute mass. The acceptance limit for an active substance is defined as 1 µg/cm2. The acceptance criterion for detergent residuals is 10 ppm.

Results and discussion

Theophylline. Results of theophylline swab analysis samples are presented in Table I. The product filter, impeller blade, chopper knife, bowl bottom and outlet valve cover show a low and reproducible level of theophylline, even when the processor was cleaned 24 h after the production run.

As these are large areas in direct contact with the product, it is important that no residues remain after cleaning to avoid contaminating the next batch. The results demonstrate that the cleaning cycle is capable of achieving the required level of cleanliness for these regions.

The results for the impeller wingnut, the bowl seal and the opening showed a wider variation among the batches, but the observed residues remained well below maximum acceptable levels.

The larger variation was because of the unpredictability of water movement in the bowl during cleaning. Depending on this motion, the water was able to come into contact with the wingnut, seep through the filter or penetrate the small gap between the metal surface of bowl and lid to touch the bowl seal.

For all batches, the results of the lipseal showed that the recovered amount of theophylline was greater than maximum acceptable levels. As explained earlier, this area is considered critical because of the lipseal design and lack of spray nozzles in the one-pot processor's lid. Initially, the results were thought to be because of a design issue; however, results from the mebendazole runs showed no problems with the lipseal, indicating that changing the cleaning cycle might reduce the recovered theophylline levels or that the API water-solubility was the cause.

The conductivity of the final rinse water samples was consistently lower then the conductivity of the 1:1000 detergent solution dilution and equalled the conductivity of the blank (demineralized water) samples. These results indicate that the CIP cycle rinsing steps are capable of removing all detergents from the equipment.

Mebendazole. The results of the mebendazole swab analysis are presented in Table II. Of the nine regions sampled, four were below the quantification limit for mebendazole (520 ng/swab) - the product filter surface, impeller wingnut, lipseal of the mixer shaft and the piston of the outlet valve.

These results differed with the results obtained from the theophylline samples, particularly the lipseal. The consistently low results for the impeller blade, chopper knife and the Transflo opening indicate excellent cleaning cycle reproducibility for these areas.

The results observed for the bowl bottom and the bowl seal were slightly higher than the results obtained for other areas. All results, however, remained well below the maximum acceptable levels and showed good reproducibility. Results obtained during the mebendazole cleaning validation provide sufficient proof that the CIP system of the one-pot processor is also capable of cleaning water-insoluble drugs.

The total organic carbon (TOC) measurements of the final rinse water samples showed a TOC content of (on average) 6.26 ppm. The TOC content of the blank samples was always less than 1 ppm. These results indicate that there was some residual detergent in the bowl, but at a level lower than the acceptable maximum. Removal of detergent by the final rinsing step was found to be acceptable.

Conclusions

This study has shown that the CIP system of this one-pot processor is capable of removing both water-insoluble mebendazole and water-soluble theophylline from the system to a level significantly less than acceptable maxima. Although certain areas show a larger variation in results than others, the reproducibility of the cleaning cycle can be considered good as the results for all areas were always consistent.

Acknowledgments

We would like to thank Professor Remon, Professor Chris Vervaet and Dr Geert Vergote of the Laboratory of Pharmaceutical Technology at the State University of Ghent (Belgium) for the use of their analytical equipment and their kind assistance during theophylline sample analyses.

We would also like to thank

Dr Lieven Baert and Dr Filip Kiekens of Johnson & Johnson Pharmaceutical R&D, department of Janssen Pharmaceutica for the use of their analytical equipment and their kind assistance.

Bibliography

1. Cleaning Validation Handbook (CD-ROM), Ecolab NV, Zellik-Asse, Belgium (2002).

2. www.ecolab.com

3. D. Rohsner and W. Serve, "The Composition of Cleaning Agents for the Pharmaceutical Industry," Pharm. Eng. 15(2), 20-29 (1995).

4. PDA Technical Report No. 29, "Points to Consider for Cleaning Validation," PDA J. Pharm. Sci. Technol. 52(6) sup. (1998).