Lyophilization is widely used for pharmaceuticals / biopharmaceuticals to improve stability and provide adequate shelf life. In spite of well established records of process development [1-8] scale-up/tech-transfer, [9-16] and validation  of optimized lyophilization process, deviations to the validated process still occur during commercial manufacturing. Each process deviation requires an assessment of the impact to product quality and stability.
Deviations may occur even for the best lyophilization process due to uncontrollable circumstances such as equipment malfunction, power outages and human error. While the possible deviations to the lyophilization process are endless, they will often fall into one of the following three failure modes: (1) Shelf temperature excursions from set point, (2) Chamber pressure leak into the lyophilizer system, and (3) Variables extrinsic to the lyophilization process itself, such as the filling process or container closure components. The implications of such process deviations can be severe- ranging from delay in release to even rejection of expensive drug product lots. In this paper, the authors present case studies of process deviations they have encountered and the approaches that can be adopted to minimize such occurrences, where applicable. This paper also suggests a general product impact assessment strategy for each of these three failure modes. Through this review article, we hope to raise greater awareness to these real life manufacturing issues, so that it may be of practical benefit for the pharmaceutical/ biotechnology industry so as to be prepared when handling such occurrences.
Results and Discussion
Temperature Excursions during Lyophilization
As part of pre-validation or characterization activities performed prior to commercial production, the robustness of the lyophilization cycle is tested by evaluating the product quality impact when subject to worst case “envelope” conditions. The operating parameters critical to the lyophilization cycle (such as shelf temperature, chamber pressure, annealing time, freezing rate, primary drying time) are varied within defined operating ranges. A representative example of such a scheme is shown in Table 1. Typically, several lyophilization robustness runs are performed at various combinations (high, target, and low) of these parameters, and the product quality is evaluated based on long-term stability data. Other approaches have also been recently proposed to characterize the robustness of the process. 
Table 1. Representative example of lyophilization parameters tested as part of characterization runs
Once characterized, the “envelope” conditions help define the “acceptable” operating ranges within which variation from target conditions can occur during manufacturing without impacting the quality of the final product. However, any deviations (even small or instantaneous) outside of these conditions need to be investigated through a non-conformance process and a product quality evaluation completed prior to the release of the lot. While deviations can occur for any of the operating parameters shown in Table 1, the most common occurrences are probably temperature excursions. Hence, in the subsequent section, some case studies of such excursion (based on the author’s experience) are presented and the approach to analyze potential impact on product quality outlined.
Temperature Excursion During Initial Freezing
A power outage occurs within 1 hour of the initial freezing step. As a result the shelf temperature deviates from the set point temperature of -13.2 ± 3°C to a maximum temperature of approximately 2.5°C (in ~3 hours), before immediately returning to -13.2 ± 3°C, and continuing through the rest of the cycle. A disruption of the initial freezing rate can potentially lead to disruption in the formation of uniform ice-crystal structure resulting in incomplete crystallization of crystalline excipients and/or heterogeneous moisture distribution in the lyophilized product. In such occurrences, the first step is to determine if there is an annealing step as part of the lyophilization cycle and if the cycle progressed as planned. An annealing step at the end of preliminary freezing ensures formation of uniform ice-crystal formation and/ or excipient crystallization,  through a mechanism termed “Ostwald ripening”. Confirmation of a successful completion of annealing step should negate any product quality concern. However, in most cases, a theoretical assessment is not sufficient and in the event of such excursions, additional samples are usually collected across the lyophilizer and tested for product quality (in addition to final lot release testing). A suggested sampling scheme (which provides entire shelf coverage and can be repeated for all impacted shelves) is shown in Figure 1. Test results from these additional sample locations should be compared against historical performance established during characterization/ validation runs. The combination of these test results along with lot release results provides crucial evidence of minimal product impact and enables release of the lot.
Figure 1. Sampling scheme for additional testing of samples
Temperature Excursion Prior to Start of Primary Drying
Before the start of the temperature ramp from freezing to the primary drying set point (the chamber is already under vacuum); the shelf temperature accidentally increases from the set point temperature of – 45 ± 3°C to a maximum temperature of approximately - 34°C and then returns immediately to – 45 ± 3°C, before proceeding with the rest of the cycle. This deviation is outside the validated shelf temperature range and a non-conformance/ investigation is required to assess potential product impact.
Any temperature excursion during the freezing step of the cycle needs to be evaluated against critical thermal properties of the formulation such as the glass transition (Tg’) and collapse temperature (Tc). Temperature excursions that result in exceeding these thermal properties can result in product collapse or meltback leading to a negative impact on product quality . In the case of this excursion an evaluation was performed to determine if this temperature excursion exceeded the critical temperatures of the product. Since product temperature is not typically measured for commercial production batches, it may be necessary to assume a certain delta between shelf and product temperature based on prior characterization data. Alternatively, predictive modeling can be used to estimate the product temperature based on prior heat transfer characterization. In this case it was determined that the product temperature did not exceed the Tg’ or Tc, leading to a preliminary conclusion of no product impact. Additional samples were collected as per the sampling scheme shown in Figure 1 for analysis, which, coupled with lot release testing, enabled release of the particular lot.
Temperature Excursion During Secondary Drying
The dryer loses shelf control for 30 minutes during secondary drying as a result of a brief power outage. This results in the shelf temperature being maintained at 5°C cooler than the set point of 25°C. Secondary drying during lyophilization is intended to desorb bound water until the target residual moisture content specific for the product is achieved (typically < 1% w/w).  It is also understood that during this phase of drying, the drying temperature is the more critical determinant of final moisture content over drying time. . Hence, the main focus was on understanding the impact of the slightly lower cooling temperature on the final moisture content of the product.
In case of a deviation that occurs due to a power outage, there is not much that can be done in terms of mitigation through manufacturing controls. However a systematic approach can be adopted to evaluate product impact and disposition of the lot. The starting point could be to compare the period for which the shelf temperature deviated relative to the overall drying time allowed. In the case of such deviations, the first step would be to evaluate if this temperature excursion falls within the “envelope” conditions characterized (Table 1). If it is outside of what is deemed acceptable, then additional product impact evaluation needs to be performed. In this case, the impact of 30 minutes drying at 5°C colder was considered negligible compared to overall secondary drying time of 10 hours, hence no impact on the final residual moisture numbers was expected. However, as mentioned in the previous sections, this technical assessment needs to be supported by additional testing of samples for final moisture content. In this particular instance, since final moisture content is the more critical determinant, it is recommended to collect additional samples from the top, middle and bottom shelves of the lyophilizer and also from various sampling location within a given shelf. The residual moisture content from these samples should not only be compared to the final specification of the product, but also to historical moisture levels obtained. Demonstrating that the residual moisture results meet the additional testing criteria will provide adequate assurance on the product quality and stability of the product.
One of the most common equipment-related deviations in freeze-drying is a pressure leak. Although it is standard practice to perform a leak test prior to every GMP batch, some leaks may go undetected because they only manifest at low temperature or are intermittent and undetectable until triggered by a nearby vibration or other disturbance. The most obvious consequence of a pressure leak is the ability of the vacuum system to maintain and control the appropriate pressure levels. If the vacuum system is not able to achieve the chamber pressure set point then the lyophilization process will not proceed to drying; if a leak develops during drying and the vacuum control is suddenly interrupted, the product temperature may be impacted and a product impact assessment similar to those described in the previous section will be required.
However, the vacuum system will often be able to control pressure within the manufacturing tolerance despite the existence of a pinhole leak. In this case, the leak may be detected by an increased variability in the pressure signal compared to that of a batch without a leak; this corresponds to a higher frequency of valve openings (either of the isolation valve or nitrogen inlet valve depending on the method of pressure control) required by the vacuum system to keep the chamber pressure near the set point. Since the chamber pressure stays within the allowable limits in this scenario, no impact to product temperature would be expected- however, the sterility of the system would need to be evaluated.
Any leak, large enough to have a measureable impact on the vacuum control system would likely be larger than 0.2 μm and therefore constitute a sterility risk depending on the location of the leak. The simplest case to assess would be if the leak was contained to a controlled aseptic area (i.e. a leak at the door). In this case, the air leaking into the lyophilizer would be HEPA-filtered and have a known bioburden level; based on this level, the leak rate, and lyophilizer volume, the theoretical maximum bioburden concentration of the lyophilizer can be calculated. This value can then be compared to an acceptance criterion, such as the 1 CFU per 10 ft3 limit suggested by Jennings, et al. , and a passing result would enable release of the lot.
A leak located at the condenser will allow ingress of air from an uncontrolled environment. However, the risk of product contamination is low because of the constant negative pressure difference between the lyophilization chamber and condenser as well as the long distance between the condenser and chamber in most production size lyophilizers. Additionally, the very low temperature and pressure in the condenser will hinder free particle gas flow or Brownian movement.
A vacuum leak allowing air from an uncontrolled environment, such as a mechanical area, is more difficult to assess from a sterility point of view. The pressure difference between the sublimation surface and the condenser and the resulting flow of water vapor out of the vial may prevent the ingress of contaminants into the vial, but this is difficult to show conclusively in a production environment. Of course, any theoretical assessment should be accompanied by a review of the available product quality data to confirm agreement. In such instances however, it is difficult to salvage the lot due to overall sterility concerns.
Failures Due to Variables Extrinsic to the Lyophilization Process
In this section we further discuss examples of manufacturing issues encountered during lyophilization but not necessarily related to a single root cause but rather due to a combination of factors. These failure modes may be identified during commercial technology transfer prior to process qualification.
Two separate cases of raised stopper observations (0–2 mm of stopper raised above the vial opening after complete stoppering) that have a direct impact on product sterility are discussed. These types of issues are typically observed during visual inspection of vials after completion of lyophilization (including complete stoppering) and prior to vial unloading from the shelves. In the first instance, it was determined that the stoppers were sticking to the top of the lyophilizer shelf when the shelves retracted after stoppering leading to raised stoppers. It was determined that the design of the stopper along with the shelf temperature during stoppering could contribute to the raised stoppers. Lowering of the shelf temperature along with changes in the stopper design to reduce the adhesive nature (between stopper and lyophilizer shelf) eliminated the occurrence of raised stoppers.
Figure 2. Diagram of a lyophilizer showing three types of leak location. Leak 1 is allowing ingress from a controlled aseptic area into the chamber, Leak 2 is allowing ingress from an uncontrolled area into the condenser, and Leak 3 is allowing ingress from an uncontrolled area into the lyophilization chamber
In the second instance of raised stoppers, it was determined that the critical dimensions of stoppers and vials used during the run were at the edge of the allowable range, thereby leading to an improper container closure fit. This led to the stoppers being raised slightly after stoppering in the lyophilizer. To avoid such failure modes during tech transfer, it is important to perform a “stacked tolerance” analysis of the critical vial and stopper dimensions, and specifications appropriately created and controlled to ensure an acceptable fit.
Spots and Streaks
Figure 3. Representative example of spots and streaks typically seen in lyophilized products. A) Vials with different degrees of spots/ streaks, B) Product “halo” on the shoulder of a vial
Spots and streaks refer to lyophilized product that is deposited above the dry cake on the side or and/ or neck of the glass vials. The manifestation of spots and streaks can be in the form of product around the vial neck (often termed as “halos”) to random structures/ spots on the neck and/ or side of the vials. Presence of extensive spots and streaks may raise concerns regarding manufacturing process controls, container closure integrity, and deliverable dose. In addition the presence of spots and streaks may render a product pharmaceutically in-elegant for the end user. It is important to understand the various factors and their mechanisms that may cause appearance of such spots and streaks in a lyophilized product, so as to implement appropriate corrective actions. In the subsequent sections, we provide a high level overview of a range of factors that can lead to spots and streaks.
One of the more obvious and common cause is product splashing that can occur during the filling process, whereby in liquid droplets are deposited on the inner wall of vials and subsequently manifest as spots/ streaks after lyophilization (Figure 3A). In such cases, optimization of the vial line filling speed, fill nozzle diameter and design are typical corrective actions. Also during the vial washing, drying and depyrogenation steps of manufacturing, there might be an accumulation of static charges on the vial surface. During filling, these static charges may play a role in preferentially attracting the liquid near the vial opening depending on the formulation properties.
In these cases dried product is typically seen around the shoulder of the vials, which are also termed as “halos” (Figure 3B)
Figure 4. Liquid migration on the surface of a vial
However in other cases, the mechanism of formation of spots and streaks may not be easily discernible or obvious. In one case study, following observation of spots and streaks in a lyophilized product, a thorough investigation was carried where product splashing during filling or static charges were ruled out as causative factors. In this case, liquid migration on the vial due to surface tension effects was established to be the root cause. This was evident when the product in question was carefully dispensed in a vial with a handheld pipette with care being taken not to cause product splashing. After a few minutes, product was found to rise on its own along the side of the vials (Figure 4). In such cases, use of a coated vial may be most effective in alleviating this issue, since it is usually burdensome to re-formulate a commercial product.
Due to the inherent complexity of the lyophilization process, process deviations are not un-common during commercial manufacturing. These deviations can occur due to single or multiple unexpected failure modes. In this paper, through the use of specific case studies we discuss examples of some of these failure modes as well outline general approaches that that can be adopted to evaluate product quality impact. In addition we also discuss specific actions (through raw material and process controls) that can be implemented to minimize such occurrences. While it may be difficult to be prepared for all failure modes (especially in the case of gross equipment failure or human errors), there are best practices that be adopted during commercialization of lyophilized products to be better prepared for such unexpected events. Lessons learned from process deviations and troubleshooting can be incorporated into risk assessments and execution of process characterization studies around the appropriate design space prior to process validation can help mitigate the occurrence of process deviations . For example, the causative factors for spots and streaks can be investigated and incorporated as part of selection of vial and glass type suitable for a particular molecule and formulation. In the case of the stopper-related issues discussed in this article, process issues can be prevented by identifying such issues at the design stage and through pilot scale characterization.
The authors would like to thank Disha Ahuja, Ph.D. for the spots and streaks section outlined in the manuscript. Dr. Ahuja is a Senior Scientist in the Drug Product Engineering group at Amgen, Inc.
Shouvik Roy, Ph.D., is a Sr. Scientist in the Drug Product Engineering Group at Amgen. In his current role, Dr. Roy manages a group of Scientists and Engineers providing process development and manufacturing support to Amgen’s commercial protein therapeutic products. Dr. Roy’s expertise lies in the area of formulation, fill, finish operations for vials and syringe products and with product/device interfaces. Dr. Roy received a Ph.D. in Pharmaceutical Sciences from the University of Colorado Health Sciences Center and an M.S. in Industrial Pharmacy from the University of Toledo. Dr. Roy is also currently a reviewer for several peer-reviewed journals such as J. Pharm Sci, AAPS Sci Tech, and Biotechnology Progress.
Ananth Sethuraman, Ph.D., is a Principal Scientist in the Drug Product Engineering group at Amgen. Dr. Sethuraman manages a group of scientists and supports commercialization of late-stage protein therapeutic products. Dr. Sethuraman received an M.S. and Ph.D. in Chemical and Biological Engineering from Rensselaer Polytechnic Institute, Troy, NY.
Christian Ruitberg is a Principal Engineer in the Drug Product Engineering Group at Amgen. He has worked on the process development of lyophilized products for 10 years. Christian currently leads a team at Amgen focused on the development, technical transfer, and manufacturing support of a commercial lyophilized biopharmaceutical product. He received his undergraduate degree in Chemical Engineering from Rennsselaer Polytechnic Institute and M.Eng. in Chemical Engineering from Lehigh University
- Blue,J. and Yoder,H. (2009) “Successful lyophilization development of protein therapeutics.” Am. Pharm. Rev. Jan/Feb: 90 – 96.
- Geiseler,H.; Kramer,T.; and Pikal,M.J. (2007) “Use of manometric temperature measurement (MTM) and SMART™ freeze dryer technology for development of an optimized freeze-drying cycle.” J. Pharm. Sci. 96(12): 3402 – 3418.
- Kuu,W.Y. and Nail,S.L. (2009) “Rapid freeze-drying cycle optimization using computer programs developed based on heat and mass transfer models and facilitated by tunable diode laser absorption spectroscopy (TDLAS).” J. Pharm. Sci. 98(9): 3469 – 3482.
- Pikal,M.J. (1985) “Use of laboratory data in freeze drying process design: Heat and mass transfer coefficients and the computer simulation of freeze drying.” J. Parenter. Sci. Technol. 39(3): 115 – 139.
- Searles,J.A.; Carpenter,J.F.; and Randolph,T.W. (2001) “The ice nucleation temperature determines the primary drying rate of lyophilization for samples frozen on a temperaturecontrolled shelf..” J. Pharm. Sci. 90(7): 860 – 871.
- Searles,J.A.; Carpenter,J.F.; and Randolph,T.W. (2001) “Annealing to optimize the primary drying rate, reduce freezing-induced drying rate heterogeneity, and determine T(g)’ in pharmaceutical lyophilization.” J. Pharm. Sci. 90(7): 872 – 887.
- Tang,X. and Pikal,M.J. (2004) “Design of freeze-drying processes for pharmaceuticals: practical advice.” Pharm Res. 21(2): 191 – 200.
- Tang,X.C.; Nail,S.L.; and Pikal,M.J. (2005) “Freeze-drying process design by manometric temperature measurement: design of a smart freeze-dryer.” Pharm Res. 22(4): 685 – 700.
- Kuu,W.Y.; Hardwick,L.M.; and Akers,M.J. (2005) “Correlation of laboratory and production freeze drying cycles.” Int. J. Pharm. 302: 56 – 67.
- Rambhatla,S. and Pikal,M.J. (2003) “Heat and mass transfer scale-up issues during freezedrying, I: Atypical radiation and the edge vial effect.” AAPS PharmSciTech. 4(2): E14.
- Rambhatla,S. and Pikal,M.J. (2004) “Heat and mass transfer scale-up issues during freezedrying, II: Control and characterization of the degree of supercooling.” AAPS PharmSciTech. 5(4): e58.
- Rambhatla,S.; Tchessalov,S.; and Pikal,M.J. (2006) “Heat and mass transfer scale-up issues during freeze-drying, III: Control and characterization of dryer differences via operational qualification tests.” AAPS PharmSciTech. 7(2): E39.
- Sane,S.U. and Hsu,C.C. (2007) “Strategies for successful lyophilization process scale-up.” Am. Pharm. Rev. Jan/Feb: 132 – 136.
- Speaker,S.M. and Teagarden,D.L. (2008) “Practical considerations of scale-up and technology transfer of lyophilized drug products.” Am. Pharm. Rev. Sep/Oct: 54 – 61.
- Tchessalov,S. and Warne,N. (2008) “Lyophilization: Cycle robustness and process tolerances, transfer and scale-up.” Eur. Pharm. Rev. 8(3): 76 – 83.
- Tchessalov,S.; Dixon,D.; and Warne,N. (2007) “Principles of lyophilization cycle scale-up.” Am. Pharm. Rev. Mar/Apr: 88 – 92.
- Jameel,F. and Paranandi,M. (2006) “Critical considerations in the transfer and validation of a lyophilization process.” Am. Pharm. Rev. Mar/Apr: 53 – 55.
- Knight,C.A.; Hallett,J.; and DeVries,A.L. (1988) “Solute effects on ice re-crystallization. An assessment technique.” Cryobiology 25: 55 – 60.
- Carpenter,J.F. and Chang,B.S. Lyophilization of Protein Pharmaceuticals. 1996. Biotechnology and Biopharmaceutical Manufacturing, Processing and Preservation. Editors: Kenneth E. Avis and Vincent L. Wu.
- Pikal,M.J.; Shah,S.; Roy,M.L.; and Putman,R. (1990) “The secondary drying stage of freeze drying: Drying kinetics as a function of temperature and chamber pressure.” Int. J. Pharm. 60: 203 – 217.
- Jennings,T.A. (1990) “A model for the effect of real leaks on the transport of microorganisms into a vacuum freeze-dryer.” J. Parenter. Sci. Technol. 44(1): 22 – 25.
- Sundaram,J.; Hsu, C.C.; Shay, Y-H,M; Sane, S.U. (2010) “Design Space Development for Lyophilization Using DOE and Process Modeling” BioPharm International Volume 23, Issue 9.