OR WAIT 15 SECS
The authors describe the operational qualification of test accuracy with regard to temperature drift using a thermal-compensation algorithm on several freeze dryers.
The sterilizing grade filter on a freeze dryer is a fundamental element for maintaining sterility throughout the process of lyophilization. According to Annex 1 in the European Guidelines, the integrity of critical gas filters should be confirmed after use (1). Risk analysis could also confirm the need for testing before use because of the value of a particular product.
The water intrusion test (WIT) for hydrophobic PTFE membrane cartridges enables in situ integrity testing without the interference of contaminative wetting media (e.g., alcohol) and obviates the need for downstream manipulation. WIT measures the decay rate of a pressure level imposed upon a hydrophobic membrane surrounded by water on the upstream side. Its dependability derives from the demonstrated correlation of the test values to organism retention levels in accordance with the HIMA ASTM F-838-05 guideline (2).
WIT has been demonstrated to apply to mono membranes of expanded PTFE since 1991. It may or may not extend to other constructions. Importantly, WIT reflects the hydrophobicity of the filter and thus assesses the suitability of air filters for reuse in equipment, such as freeze dryers, autoclaves, and all other venting applications.
Temperature drift may have a significant effect on the integrity test value and may result in false conformity or false non-conformity test results. To avoid this thermal impact, a thermal compensation algorithm can be used for applications (e.g., freeze dryers and autoclaves) on which an integrated WIT-system has been implemented.
This article will describe the operational qualification of the test accuracy using the thermal-compensation algorithm. Sartorius Stedim and the Global Technical Service for Secondary Technologies of GSK Biologicals qualified the test jointly on several freeze dryers (IMA Life).
Materials and methods
The following equipment and materials were used: freeze dryers (LMX33, IMA Life), hydrophobic sterilizing grade filter (Sartofluor 0.2 µm, 10 in., Sartorius), cartridge dummy (hand-made from filter cartridge components), stainless steel filter housing (Sartorius), Nitra pneumatic 4-mm diameter tubing (Nitra, AutomationDirect), needle valve for water (Swagelok), high precision scales (Sartorius Stedim), electro heating ribbon (Tibtech), and a sports timer (EA Combs).
For all accuracy trials, a standard WIT was launched from the supervisory control and data acquisition (SCADA) system on the freeze dryer using the filter dummy. The WIT involved the following steps:
All test parameters were visualized on SCADA and printed.
For the drying qualification, a real membrane cartridge was used. The detailed information in this article will be restricted to accuracy qualification and temperature-drift correction.
Integrating WIT into a freeze dryer from the design phase
The integration of WIT into a freeze dryer makes the integrity test a fully integrated process step, thus eliminating the need for any additional effort for manual or semiautomatic filling of the filter housing with test water. The test can be performed before and after the freeze drying cycle.
The integration consists of the following steps:
Programming the software of the freeze dryer with the necessary algorithms and valve sequences as required.
Figure 1: Simplified piping and instrumentation diagram of the freeze dryer including set-up for water intrusion test.
WIT may be integrated into existing freeze dryers, but the effort of requalification is substantial because of the significant software and hardware modifications required.The water used for the integrity test can be directly supplied from a water for injection (WFI) line or generated by condensing steam. A cooling system, including a control loop, ensures that the test water temperature is within specification, and a system to homogenize the water avoids thermal stratification. Once the water has been generated and stabilized in terms of temperature, the filter housing can be cooled down and refilled for test.
Figure 2: Water filling phase of filter housing for predefined volume and overflow valve to generate known gas net volume.
A headspace volume, validated by weighing, on top of the filter housing allows for the highest accuracy. This technique is important because the final compacted headspace gas volume after filling is calculated based on the ratio between the actual atmospheric pressure and the actual test pressure (see Figures 2 and 3). The value of the compacted headspace gas volume is also corrected for any temperature drift during the stabilization phase.
Figure 3: Stabilization phase and compaction of predefined volume after closing of overflow valve.
When the system has been stabilized for 10 min and the pressure, temperature, and intrusion rate are constant, the filling valve at the bottom of the housing is closed. The water intrusion value is then calculated from the measured pressure drop and the compacted headspace gas volume derived from the ideal gas equation as follows (see also Figure 4):
where WIT is water intrusion value in ml/min, Δp is pressure drop, Vcomp is compacted headspace gas volume, pref is reference pressure of 1000 mbar (regardless of atmospheric pressure), and t is test time (10 min).
Figure 4: Expansion of compacted gas volume due to water intrusion.
Temperature influence on the test value during the measurement phase
Temperature variations of the headspace can lead to significant errors in the determination of WIT. A temperature drift of as little as 1 °C can lead to an error of 24%. The following calculation is derived from the perfect gas law (pV = nRT):
where pT1 is absolute pressure before the temperature change (typical test pressure is 3500 mbar), VT1 is volume before the temperature change, T1 is temperature in Kelvin before the temperature change, pT2 is absolute pressure after the temperature change, VT2 is equal to VT1, and T2 is temperature in Kelvin after the temperature change.
Because VT1 = VT2 when the intrusion rate is zero, then the following equation can be used:
Thus pressure variation (ΔpT ) can be calculated according to the following equation:
where ΔpT is pressure variation due to temperature drift during the measurement phase. Therefore, a 1 °C temperature drift inside the net volume at an absolute test pressure of 3500 mbar, generates a pressure variation of 12 mbar.
The typical pressure drop for a WIT is 50 mbar in 10 min. The influence of one degree of temperature drift inside the headspace volume is then estimated by the following equation:
Temperature influence during the stabilization phase
During the stabilization phase preceding the pressure drop measurement, the compacted gas net volume is also influenced by temperature variation, although to a much smaller extent. As the pressure remains constant during the stabilization phase (pstab) the equation 2 can be written as:
During the stabilization phase there is no variation of the pressure. Therefore, one can say that pT1 = pT2 = pstab, so:
Thus volume variation (ΔVT ) due to the eventual temperature variation during the stabilization phase can be calculated according to the following equation:
Using the above equation, the gas net volume is influenced by less than 0.4% per degree of temperature drift.
Effective temperature influence from the environment
A variation of 1 °C in the environment of the filter housing that is being tested is not likely to influence the test result in the above mentioned manner (i.e., by 24%). The housing is filled with water, which has a great heat capacity, and it takes time to transfer the energy from the stainless steel into the gas net volume. Nevertheless, during risk analysis, the temperature influence has to be taken into account, especially because the environment of a freeze dryer or an autoclave can be subject to temperature variations caused by the previous steaming cycle or other equipment.
Insulating the top part of the headspace volume is often sufficient to protect against temperature influence. However, from a risk-analysis point of view, it is safer to monitor and record any temperature variation using a temperature sensor inside the headspace volume. The integration of an algorithm to compensate for influence caused by an eventual temperature drift gives higher process security and avoids false conformity and false nonconformity test results.
Qualification of test accuracy
The important considerations for the qualification of this system include the following:
All individual steps were qualified after delivery of the complete system to the end user.
The main concerns of the end user were the test accuracy, its capacity to detect a nonconforming filter, and its capacity to compensate for temperature drift. The accuracy was expected to be ± 5% or better.
For accuracy qualification, a test-cartridge dummy is used, which allows a small quantity of water to be removed during the measurement phase (typically 3–4 g/10 min/10 in. cartridge) and thereby simulate the water intrusion phenomenon. A high precision balance with a minimum accuracy of ± 0.005 g is used for weighing the water, and a temperature sensor is used to define water density.
The correlation between water weight and the water-intrusion reference value is shown in the following equation:
where WITref is the water intrusion reference value in ml/test time (10 min), Wscale is weight on the balance in grams. Pabs is the actual absolute test pressure at the beginning of the test in mbar (approximately 3500 mbar), δwater is water density in g/mL, 1000 is the reference pressure of 1000 mbar, which is different from the atmospheric pressure.
During the qualification, three different values were generated under stable temperature conditions, and WIT values of the freeze dryer were compared with the calculated reference values.
A forth value corresponding to a nonconformity test (3.9 g/10 min/10 in.) was generated and gave a nonconformity test result with the same accuracy (± 5% or better). Note that 3.9 g/10 min corresponds to a WIT value of 13.8 mL/10 min. The maximum WIT value for a Sartofluor 0.2 µm 10 in. cartridge is 13 mL/10 min.
Qualification of temperature-drift compensation
As the stainless steel part containing the compacted gas net volume was insulated, it was decided that the most appropriate way to simulate a temperature drift was to use a electro heating ribbon in direct contact with the insulation, thus representing a clear worst case because of a higher heat transfer than ambient air. The accuracy measurements were performed using the same method as for the stable temperature qualification but with the addition of heating the insulation (see Figure 5).
Figure 5: Set-up for qualification of temperature drift compensation using electro-heating ribbon (heating resistance).
Results of accuracy and temperature qualification
The recorded test parameters and the resulting test values are shown in Table 1, where Atm is atmospheric pressure (mbar absolute), W is weight (g) of water removed on the scale during the measurement phase, T0 is the gas temperature in the net volume before compaction, T1 is the gas temperature in the compacted net volume in the beginning of the measurement phase, T2 is the gas temperature in the compacted net volume at the end of the measurement phase before draining, "Deviation with Tcorr" is the deviation of the integrated measurement using the temperature correction algorithm versus the reference measurement, and "Calculation without Tcorr" is manual calculation of the deviations that would have been existed without temperature correction.
Table I: Recorded test parameters and resulting test values.
Draining the housing and drying the filter
After WIT, the filter housing is drained. Drying the hydrophobic filter after WIT is particularly important with regard to freeze dryers because the exposure of a wetted filter to high vacuum levels would lead to ice crystal formation in the membrane and damage the filter cartridge.
The standard method of drying, which involves dry air or nitrogen flow over the filter (without going through the membrane), together with heating the upstream gas flow, has been shown to be inefficient within an acceptable period of time. The drying procedure used for the freeze dryer in this article uses a water ring vacuum pump. The residual moisture in the filter was then dried under a low vacuum and exhausted by the induced gas flow through the filter.
The drying-procedure parameters for one filter manufacturer are not necessarily transferable to another manufacturer as filter construction has a strong influence; a single layer construction is easier to dry than a double layer construction when humidity is trapped between the two membranes.
At high temperatures, it is preferable to use nitrogen gas to avoid oxidization of the polypropylene fleeces in the cartridge.
Qualification of drying
The qualification of the drying is based on the weight of the cartridge. The cartridge is weighed out of its box, and its value is recorded. After WIT, the cartridge is weighed again, and its value is recorded. The drying cycle is then launched and interrupted on a regular basis (e.g., every 10 min) and the weight of the cartridge is recorded again. The cartridge must return to its original weight within ± 0.5 g/10 inch.
Integrating WIT within SCADA and the freeze dryer control system (or any other type of equipment) gives straight forward test results without imposing additional downtime and without imposing operator influence. It therefore answers the request for increased process security in accordance with regulatory requirements.
The operational qualification described in this paper demonstrates an accuracy of ± 5% or better. The qualification of the temperature drift compensation had to be conducted under extreme conditions because of the insulation. Despite the electro heating ribbon being in direct contact with the insulation, the low temperature drift (max 0.7 °C) shows that this kind of insulation is sufficient for preventing temperature drift caused by environmental conditions. Nevertheless, it is clear that from a process safety point of view, temperature compensation gives undeniable evidence for test result accuracy.
The large difference between the reference WIT value and the calculated WIT value when no temperature compensation was used shows the great influence of temperature variation on the test value and the risk of getting false conformity or false nonconformity test results. Environmental conditions must therefore always be taken into account in the risk analysis for filter integrity testing.
The deviation of –4.7% when heating 0.7 °C (read from the temperature probe), compared with a deviation of around +2% under stable conditions, most probably comes from the fact that the heating was unevenly done and therefore generated temperature gradients within the gas net volume. Also, the temperature probe did not cover the full length of the gas volume. This point could be improved upon from an engineering perspective. Nevertheless, the test accuracy was within the defined limit. This comprehensive qualification of the test process responded to the expectation of quality assurance in a pharmaceutical environment and gave a traceable guarantee for the integrity of the cartridge when performed within the limits of the defined parameters.
Magnus Stering* is head of application specialists for filtration technology southern Europe at Sartorius Stedim Biotech, firstname.lastname@example.org. Nicolas Debruyne is senior engineer, global technical services at GSK Biologicals, and Gianfranco Castiglioni is senior field service and qualification expert at IMA Life.
*To whom all correspondence should be addressed.
Submitted: Aug. 22, 2011. Accepted: Sep. 26, 2011.
1. European Commission, EudraLex Vol. 4: Good Manufacturing Practice (GMP) Guidelines, Annex 1: Manufacture of Sterile Medicinal Products (2009).
2. T. H. Meltzer, M. Jornitz, and P. J. Waibel, Pharm. Technol. 18 (9) 76–84 (1994).