Miron Ludzinski is a senior research scientist at Biopharmaceutics R&D, Drug Product Processing and Packaging Technologies, Bristol-Myers Squibb Company.

Miron Ludzinski

Lyophilization or freeze drying of injectable pharmaceuticals is a well-established and extensively practiced process (1–3). Although used for several conventional and biotechnology-based drug products, the process remains complex, time consuming, and cost intensive. For successful lyophilization, each of the three stages (i.e., freezing, primary drying, and secondary drying) must be fully executed before the cycle proceeds to the next stage. Furthermore, various mechanical and control systems of the lyophilization unit must maintain process conditions within narrow limits to ensure uniform quality and dryness throughout the entire batch. For production-scale batch sizes on the order of 10,000–100,000 vials, the importance of tight process control is evident, even more so when expensive products such as biotechnology-based drugs are processed. 

The lyophilization cycle recipes in current use can be broadly classified into two groups: those based on product-temperature cycles and those based on shelf-temperature time cycles. In the first, product temperatures are monitored continuously throughout the cycle by means of temperature probes that are inserted into representative vials within the load. The cycle conditions required to advance to the next portion of the cycle are product temperature–based or progress of the cycle is controlled by product temperatures. For example,   

Cool shelves to –45 °C or below and hold until all product temperatures (or average of all product temperatures) are below –40 °C. The cycle will not move to the next step until step requirement is met. 

Freeze all product at –40 °C or below for at least 3 h.

Alternatively, in a cycle based on shelf-temperature time, the conditions required to advance to the next step are independent of the product temperatures. For example, 

Maintain shelf-fluid temperature at ≤ –40 °C for 4 h 

In cycles based on shelf-temperature time, product-temperature probes are often used to help evaluate the effect of cycle deviations on product quality. By definition, the probes are not used to control the progress of a cycle. In practice, the distinction between these two methods of process control is important because measurement of product temperature(s) during routine production operation tends to vary because of the number of factors such as inconsistency in manual placement of the probes, differences in geometry of the probes, location of the probe tips, and so forth. Moreover, with modern lyophilizer units where loading and unloading of the vials is performed using an automated system through a partial door ("pizza door"), placement of temperature probes has become almost impossible. In cycles based on product temperature, variability in the length of various segments of the cycle may occur because of the dependence on product-temperature probes. Cycles based on shelf-temperature time, by comparison, are highly reproducible and vary relatively little with respect to the length of the individual cycle segments, or the length of the overall cycle. Other methods for monitoring the progress of lyophilization cycles have been reported such as barometric, manometric, and pressure-rise measurement (4–6). These methods, however, need retrofitting to the existing equipment with new hardware and/or more extensive process qualification and validations.  

Lyophilization cycles are frequently transferred from one lyophilizer unit to another because of a number of factors such as change of site of manufacture, a change of scale, or to allow flexibility within a plant to accommodate scheduling, maintenance, or repair. The approach required for a successful transfer will depend to some extent on the type of cycle being transferred. Cycles based on product temperature are less affected by differences in the lyophilization equipment used because the effect of these differences on product temperatures is automatically compensated for by the method of cycle step control. For shelf-temperature-time–based cycles, functional equivalence of the equipment used is required for consistent product-cycle transfer. The transfer of product cycles between units that are not functionally equivalent may require modifications to the cycle recipe and considerable revalidation effort. Currently, there exist reports on lyophilization process qualifications and transfer of cycles (7–9), but there are no specific industry procedures or regulatory guidelines available that can be used for the transfer of lyophilization cycles across lyophilizers. A recent general report describes some of the factors that must be considered during such a transfer process (10). 

In this article, the authors develop a comprehensive methodology to establish functional equivalence between different lyophilizers. Once functional equivalence between units is established, test requirement and validation protocols may be optimized, resulting in considerable savings of time, resources, and capital. Such a procedure is extremely desirable in large manufacturing operations, where multiple lyophilizer units are used within or across the plants. The following approach was used:  

Evaluate comparability of relevant technical characteristics of the lyophilizer units

Compare sublimation rates using a model compound  

Compare lyophilization process parameters during the cycles (i.e., the ability of the units to properly maintain all independent parameters within the narrow limits of specifications throughout the trials). 

Comparability of relevant technical characteristics of the lyophilizer units. 

As long as the critical process parameters such as shelf temperature, chamber pressure, condenser temperature, and duration of various cycle steps are held within a specified range, the drying profile as well as the quality of the final product will remain consistent from batch-to-batch, irrespective of the lyophilizer unit used. The majority of commercial lyophilization cycles in use today are very conservative with respect to equipment capacity. That is, the maximum operating capacities of various systems such as refrigeration, vacuum system, shelf-temperature control, load of water to be sublimed 

 condenser capacity, and so forth are not challenged during the course of routine production cycles. This fact enhances the reliability of production and allows for a successful cycle transfer between apparently dissimilar units. In such case, as long as the selected cycle parameters are within the performance envelope of a given lyophilizer, the cycle can be reliably executed within narrow process limits. However, in some instances, not only good control of the lyophilization parameters (including the chamber pressure, shelf temperature, duration, and so forth) are necessary for scale up, but also there must be comparability of the lyophilizer unit dependent factors such as shelf area, finish of the shelf material (e.g., thickness of the shelf wall), and the heat-transfer rate (e.g., ice sublimation rate in primary drying). 

There are, however, several equipment characteristics that may have a pronounced effect upon the course of the typical lyophilization cycle. These characteristics include the type of vacuum instrumentation (e.g., capacitance manometer or thermocouple gauge), method of pressure control (e.g., bleed gas or vacuum valve stifling), shelf-temperature uniformity, and range of shelf-temperature thermo regulation. These technical characteristics are easily evaluated by surveying the equipment and equipment specifications and reviewing data generated during the initial and subsequent routine qualifications (11). The authors reviewed selected technical characteristics of three production-scale lyophilizers (labeled A420FT, B420FT, and C220FT). 

Comparison of sublimation rate studies using mannitol solution. 

 Each of the three consecutive stages in lyophilization cycle (i.e., freezing, primary drying, and secondary drying) must be fully executed to yield a dry product with consistently low residual moisture. Although the final moisture level throughout the lot reaches a constant low value at the end of terminal drying, the drying rates of individual vials during earlier stages of the cycle may be different in the load such that some vials may dry faster than others. As the drying progresses, the remaining vials eventually catch up and the entire lot is fully dry and attains a low uniform residual moisture level after a sufficiently long period of drying. Such intermediate non-uniformity of drying rate generally occurs because of vial-related factors (e.g., differences in vials, freezing-time differences, location of vials, variability in stopper placement) or equipment-related factors (e.g., shelf temperature gradients during temperature ramps, radiant heat from the chamber walls and door).  

In experiments performed in the laboratory using a pilot-scale lyophilizer (25 ft

, 40-L condenser capacity), the drying rate profiles during the early part of sublimation (primary drying) were within ±10% of each other in three separate but identical runs. The runs were performed using trays of vials filled with 62-mL mannitol solutions in 100-cm

Because a given lyophilizer unit is, by definition, functionally equivalent to itself, application of this same limit to a comparison of drying rates between different lyophilizer units provides a rational approach for functional equivalence. Thus, the limit of ±10% variability within different units was selected as the acceptance criteria for functional equivalence.  

For the direct comparison of lyophilizer units A420FT (420-ft

 shelf surface area), B420FT (same model as A420FT but has longer vapor path between the condenser and vacuum pumps), and C220FT (220-ft

 shelf surface area), vials containing mannitol solutions were used and the initial rate of sublimation drying was studied in Type I borosilicate glass vials of 100-cm

  capacity. The concentration of mannitol was kept at 5.82% w/v in water for injection (which corresponded to the percent solid composition of a solution for lyophilization of a commercial product). After completion of the freezing process, sublimation drying was conducted for as long as 16 h at the first shelf-temperature regimen of primary drying. The cycle was interrupted and the amount of water lost because of sublimation was calculated by the difference in weights. The drying rates of trays of corresponding locations among three lyophilizer units were compared. 

Cycle parameters for these trials were selected on the basis of existing production cycles executed in the units under study. The rationale for conducting only a partial cycle was to magnify the profiles of drying-rate kinetics, which is highly variable among individual vials during the early part of primary drying of a lyophilization cycle. If the cycle was allowed to proceed to completion, then all the vials would be equally dry to a very low level of residual moisture and comparison of the kinetics of drying would not reveal any meaningful differences.  

Comparison of process parameters during cycles. 

 The drying-rate studies were conducted in full and partially loaded conditions. Data from these cycles were used for comparison. Under full-load conditions (210 trays), a large amount of solution (~800 kg) is subjected to the lyophilization process. Such a large load of water to be sublimed represents the worst-case scenario in which the aspect ratio of the load 

 the condenser capacity was close to 90%, when normally, only 60–70% range is recommended by the equipment manufacturers. Under such conditions, the system capabilities are fully challenged. Hence, the flawless execution of the lyophilization cycles (i.e., maintaining all the independent and dependent parameters within narrow limits) confirms that these units are indeed capable of the performance according to the set parameters and that the profiles generated are comparable to each other. 

 Review of selected technical parameters for the lyophilizer units was performed using installation and operational manuals, as well as installation, operational, and performance qualification test results (see Table I). Those parameters that are generally expected to have the greatest influence on the lyophilization process are indicated as critical; a number of additional characteristics are shown for information purposes only. 

Table I: Overview of selected technical characteristics of lyophilizers. 

Sublimation rate studies using mannitol solution. 

 Solutions of mannitol in water for injection (5.82% w/v) were filtered through 0.22-μm membrane filters, and about 62 mL of this solution was aseptically filled into 100-cm

  molded glass vials. Each vial was semi-stoppered with lyophilization stoppers (20-mm, double vented), and the vials were placed in close packing (60 vials each) in aluminum transfer trays with the surrounding rings or brackets. Some trays were identified with a suitable code. These trays were weighed using a top-loading scale situated inside the aseptic area. The vial-filled trays were then transported to the lyophilizer chamber, the bottoms of the trays were pulled out (leaving the ring in place), and vials were directly transferred onto the surface of the shelves. 

Figure 1: Placement of weighed trays of vials containing mannitol solutions. Five weighed trays were located in the top, middle, and bottom shelves. One weighed tray was located in all the remaining shelves. In the full-load studies, the remaining space of the shelves was filled with unweighed trays containing vials with mannitol solutions. Total number of trays in partial load studies: A420FT = 23; B420FT and C220FT = 25. Total number of trays in full load studies: C220FT = 110; A420FT and B420FT = 210. (FIGURE IS COURTESY OF THE AUTHOR) 

Figure 1 shows the placement of the weighed trays containing vials in the lyophilizer chambers. Weighed trays were distributed throughout the chamber to obtain water loss data from all the locations. Once loaded, the lyophilization cycle was initiated as described in Table II. Upon completion of the lyophilization cycle sequence, the trays were unloaded and the marked trays were weighed using the same top-loading scale. The loss of water caused by sublimation drying in each tray was calculated on the basis of the difference in weights (final weight of the tray minus the initial weight). Using this methodology, drying rates were compared in lyophilizers with partially loaded trays (23 trays in C220FT and 25 trays in A420FT and B420FT) in the lyophilizer chambers as well as in fully loaded (110 trays in C220FT and 210 trays in A420FT/B420FT) conditions. 

Table II: Lyophilization cycle used for sublimation rate studies. 

 Shelf-temperature, condenser-temperature, and chamber-pressure profiles during the lyophilization runs were compared in lyophilizers under partial-load and full-load conditions, and the process parameter data were analyzed following the runs to determine the extent of agreement from unit to unit. 

An overview of the relevant technical characteristics of the lyophilizers (see Table I) includes only equipment parameters that are generally regarded as having the greatest influence on the course of a lyophilization cycle. The parameters that are most critical to maintain a desired lyophilization profile are the shelf temperature, the condenser temperature (higher temperatures in response to greater condensation rates of ice), the chamber pressure, the duration of various stages, and the control–monitoring systems. Table I shows that all of the critical hardware and the process control–monitor mechanisms between the lyophilization units are comparable. Similarity suggests that the independent programmable-cycle parameters (shelf temperature, chamber pressure, and duration of various steps) will be executed identically in these units. Therefore, the resultant lyophilization cycles will be equivalent as long as the process conditions do not over-burden the system's capabilities.  

Table III: Weight-loss data during sublimation from each tray containing mannitol solutions (kg).

The temperatures of the shelf inlet–outlet, the condenser inlet–outlet, and of products are monitored by identical resistance temperature devices and are routinely calibrated. Similarly, the chamber and condenser pressures are monitored and controlled by manometer diaphragm-based gauges. These absolute gauges ensure that the pressure registered is a true pressure without any artifacts that may arise because of the composition of gases, as in the case of Pirani-type gauges. The chamber pressure is maintained within a narrow range (±20 mTorr) with dry sterile nitrogen or air by modulation of a proportional solenoid valve. This control mechanism ensures very tight control of desired chamber pressure. Similarity of these lyophilizer units in this regard (i.e., monitoring and controlling of chamber pressure) is an important factor in ensuring their equivalency.  

Table IV:  Comparison of sublimation rate studies: weight loss irrespective of the locations of trays. 

 Sublimation rate studies using mannitol solutions as a model system were conducted to confirm the similarity suggested by the general capabilities of the units. Table III shows the weight-loss data during the sublimation phase. The differences in drying rates between trays or vials at various positions in the lyophilizer unit are expected to be greatest during the early parts of the drying process and tend to diminish as the cycle proceeds to completion. This difference exists because as the drying proceeds and a considerable cake width starts accumulating at the top of the cake structure, resistance from the dried cake, peripheral drying because of glass walls on the sides, and secondary drying processes start contributing to the overall rates and these effects may not be consistent in all the vials during the intermediate stage. At the end of the cycle, following an extended period of primary and secondary drying, the residual moisture levels reduce to very low levels and a constant value is obtained in all the locations of the chamber. 

Table V:  Comparison of sublimation rate studies:  weight loss according to the locations of trays. 

To detect differences in drying rates, it was necessary to terminate the lyophilization cycle after initial primary drying stage of the process (i.e., approximately first 40–50% of the sublimation stage), when the differences were expected. Based on the average loss of 1.75 kg of water per tray, the amount of total water lost during this process would be 45% of the total amount that could be removed during the entire process. This result allowed comparison of trays or vials having different amounts of moisture remaining during initial primary drying. 

Figure 2: Lyophilizer performance (full load): chamber pressure, shelf temperature, mean product temperature, and condenser temperature for FD-103, FD-104, and FD-105 (A420FT, B420FT, and C220FT units are FD-103, FD-104, and FD-105, respectively). (FIGURE IS COURTESY OF THE AUTHOR) 

Tables IV and V show that even during this early period of drying, the rates of drying among various units were quite close to each other (i.e., 1.79% and 3.21%) in fully loaded and partially loaded conditions, respectively. These results were irrespective of the locations of the trays within the units. The differences in the drying rates among vials located at corresponding locations were slightly more variable but within 10% of each other. 

Figure 3: Lyophilizer performance (partial load): chamber pressure, shelf temperature, mean product temperature, and condenser temperature for FD-104 (B420FT) and FD-105 (C220FT). (FIGURE IS COURTESY OF THE AUTHOR) 

 The equipment parameter data from the drying rate studies conducted under partial- and full-load conditions were analyzed for this comparison. Under full-load conditions (i.e., 210 trays), a large amount of solution (~800 kg) is subjected to the lyophilization process. Under such a large load, the system capabilities are fully challenged. In spite of such a full-load condition, the flawless execution of the lyophilization cycles while maintaining all the independent and dependent parameters within the narrow limits clearly confirms that these units are capable of performing according to the set parameters and that the profiles generated are comparable with each other. 

Table VI: Comparison of lyophilization process parameters (partial load): shelf temperature, condenser temperature, and chamber pressure in B420FT and C220FT. 

The temperature-pressure rate profiles reflect close agreement of the lyophilization parameters during the entire run (see Figures 2, 3). (Note: lyophilizer units A420FT, B420FT, and C220FT are referred to as FD-103, FD-104, FD-105, respectively) The independent parameters (i.e., the shelf temperature, chamber pressure, and the condenser temperature) were within the overlapping range throughout the course of the cycle in all three lyophilizers (see Tables VI and VII). Moreover, the mean product temperatures were within the overlapping range throughout the course of the cycle in all three lyophilizers (see Figures 2 and 3). These parameters are the actual indicators of equipment performance under stress conditions and their close agreement with the programmed recipe confirms excellent capability of these units.  

Table VII:  Comparison of lyophilization process parameters (full load): shelf temperature, condenser temperature, and chamber pressure in A420FT, B420FT, and C220FT. 

All of the critical hardware and the process control and monitoring mechanisms among the lyophilization units studied were comparable to each other. This similarity supports the hypothesis that the programmable cycle parameters such as shelf temperature, chamber pressure, and time duration of various stages will be executed identically in these units. Even during the early period of drying (when the greatest variability in the sublimation dry rate is expected), the observed rates of sublimation drying were quite close to each other, i.e., 1.79% and 3.21% in fully loaded and partially loaded conditions, respectively. Even under fully loaded conditions and relatively aggressive cycle conditions of the shelf temperature regimen, the lyophilization cycles were executed smoothly as seen by the maintenance of all the independent and dependent parameters (as reflected by the mean product temperatures) within the narrow limits of specifications confirming functional equivalency of these units. In a follow-up study, a set of four commercial large-scale lyophilizers (420-ft

 shelf area) were evaluated for functional equivalency. The validation matrix to use these multiple lyophilizer units for a commercial product was minimized, thereby saving resources and time. 

These results provide the basis to establish functional equivalence among different lyophilizers. Once established, functional equivalence can facilitate successful implementation of scale-up and transfer of lyophilization cycles. Such equivalency procedures are desirable in large manufacturing operations, where multiple lyophilizers are used for processing a common lyophilized product, and reduced validation campaigns based upon a matrixing approach can save considerable resources and time.  

is a senior principal scientist at Biopharmaceutics R&D, Drug Product Processing and Packaging Technologies, Bristol-Myers Squibb Company, One Squibb Drive, New Brunswick, NJ 08993, 

Articles

The authors describe a comprehensive methodology for establishing functional equivalence among various lyophilizers.