Abstract
A monoclonal antibody drug product manufacturing process was transferred to a different production site, where aseptic filling took place within an isolator that was decontaminated (sanitized) using vapor phase hydrogen peroxide (VPHP). A quality-by-design approach was applied for study design to understand the impact of VPHP uptake on drug product quality. Both small-scale and manufacturing-scale studies were performed to evaluate the sensitivity of the monoclonal antibody to hydrogen peroxide (H2O2) and characterize VPHP uptake mechanisms in the filling process. The acceptable H2O2 uptake level was determined to be 100 ng/mL for the antibody in the H2O2 spiking study; protein oxidation was observed above this threshold. The most prominent sources of VPHP uptake were identified to be the silicone tubing assembly (associated with the peristaltic pumps) and open, filled vials. Silicone tubing, an effective depot to H2O2, absorbs VPHP during different stages of the filling process and transmits H2O2 into the drug product solution during filling interruptions. A small-scale isolator model, established to simulate manufacturing-scale conditions, was a useful tool in understanding H2O2 uptake in relation to tubing dimensions and VPHP concentration in the isolator air (or atmosphere). Although the tubing assembly had absorbed a substantial amount of VPHP during the decontamination phase, the majority of H2O2 could be removed during tubing cleaning and sterilization in the subsequent isolator aeration phase, demonstrating that H2O2 in the final drug product solution is primarily taken up from residual VPHP in the isolator during filling. Picarro sensor monitoring demonstrated that the validated VPHP aeration process generates reproducible residual VPHP profiles in isolator air, allowing small-scale studies to provide relevant recommendations on tubing size and interruption time limits for commercial manufacturing. The recommended process parameters were demonstrated to be acceptable and rendered no product quality impact in six consecutive manufacturing batches in the process validation campaign. Overall, this case study provides process development scientists and engineers an in-depth understanding of the VPHP process and a science-based approach to mitigating drug product quality impact.
LAY ABSTRACT: While the use of vapor phase hydrogen peroxide as a sanitizing agent for isolator and cleanroom decontamination has gained popularity in recent years, its impact on product quality during aseptic manufacturing of biopharmaceutical drug products is yet to be fully understood. With this scope in mind, this case study offers a detailed account of defining process parameters and developing their operating ranges to ensure that the impact to product quality is minimized. Both small-scale and manufacturing-scale studies were performed to assess the sensitivity of a monoclonal antibody to hydrogen peroxide, to characterize hydrogen peroxide uptake sources and mechanisms, and to eventually define process parameters and their ranges critical for minimizing product quality impact. The approach and outcome of this study is expected to benefit scientists and engineers who develop biologic product manufacturing processes by providing a better understanding of drug product process challenges.
- Vapor phase hydrogen peroxide
- Monoclonal antibody
- VPHP uptake
- Drug product quality
- Quality-by-Design
- Drug product filling
- Vaporized hydrogen peroxide
- VHP
1. Introduction
A monoclonal antibody (mAb) drug product (DP) was transferred to a different manufacturing site for production. The manufacturing processes between the two sites were identified with a major difference in decontamination (or sanitization) of the aseptic filling core. At the new site, the vial-filling operation takes place within an isolator, which is decontaminated using vapor phase hydrogen peroxide (VPHP). In contrast, the original site performs aseptic filling in a restricted access barrier system (RABS) that is decontaminated by a non-VPHP decontaminated agent (an isopropyl alcohol and water solution). This study intended to raise the awareness of process development engineers and scientists about the potential impact of VPHP on mAb product quality as well as to gain insights into developing risk-mitigation measures.
The Quality-by-Design (QbD) approach was adopted for this transfer project. QbD employs principles of scientific knowledge and appropriate quality risk management to enhance product and process understanding (1⇓⇓–4). The aim of process understanding is to minimize the process impact on product quality attributes (QAs). A risk tool, called risk ranking and filtering (RRF), was applied for each process unit operation to identify potential critical process parameters (pCPPs) based on filtering and ranking each pCPP's impact on QAs or key performance indicators. The outcome of the RRF exercises helped guide the study design (5).
Hydrogen peroxide (H2O2) is one of the most common agents for gaseous decontamination (6, 7). VPHP is effective against spores, bacteria, and viruses, and it is environmentally safer than other sanitizing agents in releasing nontoxic end products after catalytic breakdown (i.e., water and oxygen) (8, 9). Thus, the use of isolators with VPHP decontamination is becoming a preferred enclosure system for manufacturing parenteral products in the pharmaceutical industry to ensure meeting the highest aseptic requirements. However, as an effective oxidizing agent, H2O2 is known to oxidize proteins (10⇓⇓⇓–14), which can be a general concern for manufacturing biopharmaceutical products.
H2O2 can be introduced into protein formulations during production from different sources (15⇓–17). VPHP decontamination of the isolator certainly provides an obvious source of H2O2 in the manufacturing process. Several recent studies have raised awareness to the impact of residual H2O2 from the VPHP source on protein degradation (18⇓–20). H2O2 spiking studies (20) can be performed on the protein formulation to assess the sensitivity to peroxide-induced oxidation and help set the limit for H2O2 uptake during DP manufacturing. However, how H2O2 residues eventually enter the DP solution during production is scarcely reported in the literature. It is important technical knowledge because only proper controls to the manufacturing processes can ensure H2O2 uptake in final DP containers (e.g., vials, prefilled syringes, etc.) to be within acceptable limits. Hubbard and Eppler reported the relationship of H2O2 residues and product exposure on isolator and RABS filling lines, and they provided an overview of potential H2O2 uptake sources (21). Residual VPHP levels in the isolator atmosphere after the aeration phase of the decontamination cycle are critical to H2O2 uptake in four likely sources which may eventually lead to H2O2 absorption by the DP solution: empty glass surfaces (vials or syringes), the exposed product at the tip of the filling needle, open (unstoppered) filled vials or syringes, and the tubing of the filling line (21). The latter two are the primary uptake sources and the focus of this study.
In the isolator, various polymeric components (e.g., filtration equipment, tubing, etc.) are known to absorb and desorb H2O2 and often cause unwanted effects (e.g., second spike in H2O2 level) in the final aeration phase of a decontamination cycle (22, 23). Of these polymeric materials, the most critical are product-contacting, including silicone tubing. Almost all DP filling lines use platinum-cured silicone tubing in the product flowpath, especially in the peristaltic filling system (21). Silicone tubing may absorb VPHP during the decontamination cycle and the entire filling process from the isolator atmosphere. It may desorb H2O2 and readily release it to the DP solution during process interruptions, rendering a serious threat to protein product quality. Therefore, this study investigated H2O2 uptake and release in relation to tubing properties (i.e., sizes, types, etc.) and processing conditions (e.g., tubing cleaning conditions and procedures, interruption times, pre-flush, etc.).
Protein DP solution in open, filled vials or syringes that are idle on the manufacturing line prior to stoppering due to process interruptions is directly exposed to the isolator atmosphere and can readily absorb VPHP. Hubbard and Eppler suggested that the rate of absorption depends on the vial size and fill volume and that an interruption longer than 10–15 min may result in the H2O2 concentration in the vial to rise (21). This study assessed H2O2 uptake by the filled vial in relation to atmospheric VPHP concentration.
All uptake tests in this study were performed using a combination of laboratory-based small-scale isolator studies and manufacturing-scale studies. The vial configurations and tubing assembly in the small-scale isolator mimicked manufacturing-scale setup and could serve as a predictor for the performance of manufacturing-scale conditions. Worst-case exposure conditions were applied if small-scale processes or procedures were unable to mimic corresponding manufacturing-scale procedures. H2O2 concentration was determined using a qualified Amplex® UltraRed Hydrogen Peroxide assay.
2. Materials and Methods
The mAb, based on a human immunoglobulin G1 (IgG1) framework, was prepared into a drug substance (DS) formulation bulk. The DS formulation bulk was shipped frozen in stainless steel vessels to the DP manufacturing facility. At the DP site, the frozen DS bulk was thawed, pooled, mixed, and filtered prior to being filled into two different vial configurations: 21.4 mL in 20 cm3 glass vials and 15 mL in 15 cm3 vials. The DP formulation was 60 mg/mL mAb containing sucrose and acetate. The mAb DP solution was sterile-filtered into a surge tank. From the surge tank until vial filling and stoppering, all operations occurred inside an isolator.
2.1. Drug Product (DP) Process Overview
The process flow in the isolator is shown in Figure 1. The filling equipment and the tubing assembly were set up inside the isolator before a VPHP decontamination cycle was activated. A validated VPHP-based cycle was applied for isolator decontamination and followed by an aeration phase to reduce the level of VPHP in the isolator air. During aeration, the tubing assembly was sterilized via clean-in-place (CIP), steam-in-place (SIP), and drying. Filling production could officially begin immediately after the aeration phase ended, or it could be started at a later time based on the production schedule. Fill parameters and fill weight were adjusted and optimized before vial filling was officially started.
The process flow for vial-filling operation inside the isolator.
2.2. Risk Ranking and Filtering (RRF) Tool
Potential process parameters associated with VPHP decontamination were assessed based on scientific understanding, platform knowledge, and previous experiences. These process parameters may have an impact on critical product QAs and key process indicators. Per the process in Reference 5, each process parameter was assessed for its main effect and potential interactions with other process parameters against pre-defined responses (product QAs and process attributes and performance). Detailed assessment approaches have been previously described (5).
2.3. H2O2 Spiking Study
The sensitivity of the mAb to peroxide-induced oxidation can be determined through an H2O2 spiking study. H2O2 solution was spiked into the mAb formulation (60 mg/mL mAb containing sucrose and acetate) to obtain final H2O2 concentrations ranging from 50 to 3000 ng/mL. The mAb formulation without H2O2 spiking served as the control for the study. Each of the H2O2-spiked samples and unspiked control was aseptically filled into 15 cm3 glass vials with a fill volume of 14 mL for real-time (5 °C), accelerated (25 °C), and stressed (40 °C) stability testing for up to 12, 6, and 1 months, respectively. At each of these time points, the samples were analyzed for residual peroxide using the fluorometric Amplex® UltraRed Assay (Life Technologies, Carlsbad, CA) and for protein oxidation using reverse phase ultra-high performance liquid chromatography (RP-UHPLC).
2.4. Filling Assembly
For routine filling, 12 sets of the tubing assembly, including all the components from the surge tank to filling needles (Figures 2a and 2b), were set up inside an isolator. Each filling assembly included three pieces of silicone tubing (Versilic® SPT-60L, Flexicon Corp., Bethlehem, PA) and a stainless-steel Y-connector (with PEEK threaded fittings) for use with a peristaltic pump (Bausch+Ströebel, Ilshofen, Germany). The peristaltic pump was located between the surge tank and the Y-connector (Figure 2a). Table I lists the materials and dimensions of all components of the tubing assembly (Figure 2b). The holdup volume of the solution within each segment of the tubing flowpath was calculated and is listed in Table I. The holdup volume is important for tubing assembly flush, that is, the required number of filling cycles to clear the tubing assembly.
(a) Graphical representation of the filling line setup including the surge tank, the filling needles, and the filling assembly, which includes the tubing assembly (A–F), and the peristaltic pump (G). (b) Photograph of the tubing assembly (with component information listed in Table I).
Component and Dimensional Information of the Tubing Assembly Leading from the Surge Tank to the Filling Needles
2.5. VPHP Decontamination Cycle
In normal isolator-based filling operations, the interior surfaces of the entire isolator and the filling line (piping, tubing, vessels, surge tank, process equipment, filters, and associated fittings) are decontaminated before vials are introduced for filling. During decontamination, a predetermined amount of liquid peroxide is vaporized and distributed throughout the working chamber of the isolator and held for a predetermined period of time to sanitize all surfaces. After this cycle, the isolator switches into an aeration mode, where humidity-controlled air is circulated throughout the chamber and exhausted from the isolator until the VPHP concentration has decreased enough to begin filling with minimal risk to the product.
2.6. Clean-In-Place (CIP), Steam-In-Place (SIP), and Drying
During the VPHP aeration phase, the entire assembled filling line was sterilized via a validated program of CIP, SIP, drying, and cooling. In general, CIP cleans the filling line by flushing cold water (purified water at 20 °C) for ∼60 min and hot water (water at 80 °C for injection) for ∼40 min. SIP sterilizes the filling line with steam (121 °C) for 70 min. Air (20–30 °C) is flushed for 130 min to dry and cool the filling line. The overall aeration phase, including the CIP, SIP, and drying processes, lasts ∼10 h.
2.7. Monitoring VPHP Concentration in Isolator
VPHP concentrations were monitored via a Picarro H2O2 sensor (Model G1114 or G2114, Picarro Inc., Santa Clara, CA), which uses cavity ring-down spectroscopy to detect H2O2 in the vapor phase. The Picarro analyzer has a precision of 1.1 ppb and a lower detection limit of 4 ppb VPHP. As VPHP tends to be absorbed by many plastics, the Picarro sensor must use polytetrafluoroethylene (or similar material) sampling lines that do not readily absorb H2O2. Monitoring a large isolator with a complicated internal geometry and airflow such as a filling line requires the sampling point of the Picarro to be carefully chosen to deliver a representative reading of product exposure to VPHP. In the manufacturing-scale studies described here, the Picarro sampling point was placed close to the filling needles. The small-scale isolator features a small and unobstructed work cell capable of maintaining a homogenous VPHP concentration (data not presented). Thus, the location of the Picarro sensor for VPHP concentration monitoring was not critical.
2.8. Scale-Down Isolator Experiments
A Pharmaceutical Safety Isolator (PSI-M, Skan AG, Switzerland) was used to test the filling components in a controlled VPHP environment. In addition to normal isolator functionality, this isolator is able to maintain a constant peroxide concentration in the atmosphere by continuously vaporizing H2O2 while all other functions of the isolator are in normal production mode. By changing the dosing rate and the concentration of the H2O2 stock bottle, any VPHP level can be targeted and maintained. This level is monitored by a Picarro H2O2 sensor (model G1114 or G2114).
To test H2O2 uptake by the tubing, a Flexicon PD12I (Flexicon Corp., Bethlehem, PA) pump head was placed inside the isolator while a Flexicon MC12 pump controller was placed outside the isolator and connected to the pump head through a sanitary fitting port. The tubing assembly was loaded into the pump head and kept exposed to the isolator atmosphere. The two ends of the tubing ran outside the isolator through separate filling ports; one end was connected to a water bottle (upstream) and the other to a filling needle (downstream). In this way, the tubing assembly inside the isolator could be exposed to various concentrations of VPHP. Alternatively, a measured section of bulk tubing could be exposed to VPHP in the isolator (as opposed to the complete tubing assembly). In this manner, individual components could be studied.
H2O2 uptake by filled vials was also tested in the isolator. These vials were placed on the deck of the isolator, and then their stoppers were removed to be exposed to the controlled VPHP atmosphere. The vials were re-stoppered at various time points and removed from the isolator for testing.
2.9. Analytical Methods and Tests
2.9.1 Determination of H2O2 Concentration in mAb Solutions:
H2O2 concentrations in the mAb formulation were measured using the fluorometric Amplex® UltraRed Assay. The method of testing has been previously described (19). With this mAb formulation, the assay was qualified with a limit of detection of 7 ng/mL and limit of quantification of 25 ng/mL.
2.9.2 Detection of Protein Oxidation by RP-UHPLC:
Heavy chain Fc and Fab oxidation of the mAb in the H2O2-spiked samples was determined by a RP-UHPLC method, which has been previously described (24). For this mAb, the assay was qualified with 0.6% and 0.8% variability for different oxidation sites.
3. Results and Discussion
Assessing the impact of isolator decontamination by VPHP on DP quality and filling process performance requires in-depth understanding of the H2O2 uptake mechanism and its relationship with all relevant process parameters. This understanding can facilitate the design of studies. The outcome of these studies would define the operating ranges of key process parameters. Operating these parameters within the ranges would ensure that the risks that affect product quality are mitigated. Thus, the development approach in this study began with the establishment of technical and risk assessment. A RRF tool was used to identify and assess all relevant process parameters.
3.1. Technical and Risk Assessment Outcome
The technical and risk assessment, as summarized in Table II, was divided into three process stages: VPHP decontamination phase, VPHP aeration phase, and the filling process. The impact of process parameters in these three phases on H2O2 uptake by open, filled vials and the tubing assembly was determined. The rationales for determining the impact was provided. Some parameters may have interactions with other parameters; for example, residual VPHP level in the isolator air may depend on filling start time (after the aeration phase). The later the filling operation begins, the lower the level of residual VPHP in isolator air, and thus, the lower the tendency of H2O2 uptake by either the open, filled vials or the tubing assembly. Tubing size may further affect H2O2 uptake in relation to residual VPHP level in isolator air; smaller tubing presents a large surface area–to–volume ratio, which promotes a higher concentration of H2O2 absorbed by the DP solution in the tubing. Overall, product sensitivity to H2O2, tubing size, CIP, SIP, and drying procedures, residual VPHP level in isolator air, and interruption time are the pCPPs that should be evaluated based on a combination of small-scale and manufacturing-scale studies, which are described in the following sections.
Summary of Technical and Risk Assessment for VPHP Decontamination and H2O2 Uptake in the Isolator for Filling Operation
3.2. H2O2 Spiking Study
The purpose of the H2O2 spiking study was to determine the sensitivity of the mAb to peroxide-induced oxidation and establish an acceptable H2O2 level that does not cause oxidation to the protein DP. This level may serve as guidance for VPHP uptake studies. In this study, residual H2O2 and oxidation of the mAb (Fc and Fab regions of the heavy chain) were monitored and the results are presented in Tables III and IV, respectively. The peroxide in all the spiked mAb samples was consumed after 1 week at 40 °C, 1 month at 25 °C, and 3 months at 2–8 °C (except for positive control spiked with 3000 ng/mL H2O2). Residual H2O2 in the positive control was not consumed until 6 months at 2–8 °C. Peroxide-induced oxidation of both Fc and Fab in the heavy chain of the mAb was not observed for samples spiked with up to 100 ng/mL H2O2 after 12 months at 2–8 °C, 6 months at 25 °C, and 1 month at 40 °C when compared to the unspiked control stored under the same conditions. For samples spiked with 200 ng/mL and 300 ng/mL H2O2, an increase in Fc oxidation of 1.1–1.3% was observed after 6 months at 25 °C and 0.6–0.8% after 1 month at 40 °C when compared to their unspiked control under the same storage conditions. At 2–8 °C, Fc oxidation was not observed after 12 months for the sample spiked with 200 ng/mL H2O2, but an increase of 0.9% in Fc oxidation was observed for the sample spiked with 300 ng/mL H2O2. For Fab oxidation, no increase was observed for all the spiked samples at 2–8 °C and 25 °C storage. However, an increase of 2.9% in heavy chain Fab oxidation was only observed for the sample spiked with 300 ng/mL H2O2 after 1 month at 40 °C. Based on these results, the acceptable peroxide limit for the sample mAb at 60 mg/mL is 100 ng/mL. This threshold was applied for subsequent H2O2 uptake investigations. Note that other protein DPs may have different acceptable limits because the sensitivity of protein to H2O2-induced oxidation is product-specific.
Residual Hydrogen Peroxide in mAb Formulation (60 mg/mL) Spiked with Various Concentrations of H2O2 as Measured by Amplex® UltraRed Assay
Oxidation of mAb in Formulation (60 mg/mL) Spiked with Various Concentrations of H2O2 as Measured by RP-UHPLC
It was previously shown that the methionine residues of the Fc and H2O2 react in a one-to-one stoichiometric ratio (19), meaning the amount of H2O2 required for oxidation increases with increasing protein concentration. Therefore, a more concentrated protein solution (i.e., >100 mg/mL) may allow for a higher acceptable peroxide limit. In addition to protein concentration, other formulation factors such as formulation excipients, protein framework, and DP physical state (i.e., liquid versus lyophilized) should also be considered when designing the H2O2 spiking study to establish an acceptable H2O2 level.
3.3. Worst-Case H2O2 Uptake Assessed in a Manufacturing-Scale Study
A manufacturing-scale water fill was performed to understand VPHP uptake during filling following a routine manufacturing process with worst-case H2O2 uptake conditions. The worst-case parameters included (a) filling was initiated immediately after the VPHP aeration phase, (b) a 30 min interruption was executed at the beginning of filling, and (c) using a smaller tubing size [i.e., 4.8 mm internal diameter (ID) tubing for 15 mL fill in 15 cm3 vials]. In (a) and (b), a high level of residual VPHP in the isolator air was expected (although actual atmospheric VPHP concentration was not monitored in this study). In (c), a higher tendency of H2O2 uptake was expected than the larger tubing (6.0 mm ID tubing for 21 mL fill in 20 cm3 vials) due to the larger surface area–to–volume ratio. During the 30 min interruption, 24 filled vials were left open for direct VPHP exposure. After the interruption, the line was restarted and those 24 vials were stoppered and sampled. Based on the holdup volume in the tubing assembly (i.e., 42.97 mL listed in Table I), it would take three 15 mL filling cycles to clear the tubing line. Thus, the first four vials filled from each needle after re-start were also sampled. In addition, vials were sampled from the beginning, middle, and end of the run in normal production conditions (not sampled after any interruptions). H2O2 levels from these vials are presented in Figure 3.
Summary of H2O2 uptake in vials throughout the first manufacturing-scale water fill batch.
The open vials exposed to 30 min interruption showed a H2O2 level in the range of 50–100 ng/mL (∼60 ng/mL on average). Although the H2O2 level in these vials was still within the 100 ng/mL limit determined by the spiking study, the lack of the safety margin in case of additional uptake from the tubing assembly poses a concern.
All vials from the first three filling cycles (12 vials in each cycle) showed a high level of H2O2, ranging from 100 to 150 ng/mL. The H2O2 level in the fourth filling cycle, however, decreased substantially to below 50 ng/mL (∼30 ng/mL on average), which was only slightly higher than the level detected in vials filled in the middle (after ∼10,000 filled vials) and in the end (∼20,000 filled vials) of the filling process. This result confirmed that the silicone tubing assembly absorbs VPHP and later releases it as H2O2 into the DP solution during the 30 min interruption. This absorption and desorption action spanned the entire silicone tubing flowpath. Overall, this study suggested that H2O2 uptake may pose a risk to product quality, so the filling process needs to be thoroughly understood and carefully controlled.
3.4. Process Understanding via Small-Scale Studies
A laboratory-based small-scale isolator was used to perform uptake studies to better understand the manufacturing-scale process. This small-scale system is capable of providing well controlled, constant VPHP levels for H2O2 uptake by vials and tubing assemblies identical to those used in commercial manufacturing. The CIP, SIP, and drying processes are not available in the small-scale isolator. Despite the lack of certain capabilities, this small-scale isolator should be able to serve as a feasible worst-case model.
Of all process parameters listed in Table II, only the residual VPHP level in isolator air was not available (not monitored) from the first manufacturing-scale study. Our platform knowledge and data (not published) could estimate the residual VPHP level to be around 500 parts per billion (ppb) given the results from the initial manufacturing-scale study. Thus, bulk tubing (without the Y-connector, but measured to the same total length as the tubing assembly) was exposed to 500 ppb VPHP in the small-scale isolator, and then filled with water for a 30 min incubation (to simulate filling interruption). After incubation, water in the tubing was filled into vials (14.8 mL per filling cycle). H2O2 concentration in each vial was analyzed using the Amplex® UltraRed Assay. The same experiments were performed for other incubation times: 15 and 60 min. The profile of H2O2 uptake versus the number of filling cycles is presented in Figure 4a. There was a significant H2O2 uptake in the first three filling cycles, identical to what was observed in the manufacturing-scale study. The uptake level increased with increasing incubation time. The 30 min incubation resulted in ∼150 ng/mL H2O2 uptake, also consistent with the manufacturing-scale study. It suggested that during commercial manufacturing, the residual VPHP level in isolator air remains high enough to cause too much peroxide uptake in DP immediately after the aeration phase.
H2O2 uptake after exposure to 500 ppb VPHP in a small-scale isolator by the 4.8 mm tubing set (a) without a Y-connector and (b) with a Y-connector; (c) comparison of H2O2 uptake rate by water versus mAb DP solution.
To assess H2O2 uptake by the Y-connector of the tubing assembly, the experiments above were repeated on the complete tubing assembly using a smaller filling volume, that is, 3 mL per filling cycle. The H2O2 uptake profile (Figure 4b) displayed a trough at around Filling Cycle #4 (i.e., ∼12 mL). The holdup volume of Tubing C (4.8 mm ID in Table II) was 12.4 mL, suggesting that there was no H2O2 uptake by the stainless-steel Y-connector and H2O2 uptake occurred primarily via the silicone tubing. (Note that holdup volume flush followed the sequence of Tubing C, Y-Connector, and Tubing A or B in the tubing assembly.)
Water has thus far been used as a surrogate for uptake studies. To demonstrate water's comparability to the mAb solution, 15 cm3 vials filled with 15 mL of water for injection (WFI) and mAb solution, respectively, were exposed to 100 ppb VPHP for various periods of time (triplicate vials for each time point). H2O2 uptake in relation to exposure time is summarized in Figure 4c. The slopes of the two uptake curves matched well, with water picking up H2O2 slightly faster than the mAb solution. This suggested that water is a suitable surrogate for the mAb solution.
3.5. Relative Significance of Two H2O2 Uptake Sources by the Tubing Assembly
The tubing assembly underwent two stages of H2O2 uptake. First of all, the tubing was assembled and placed on the pump prior to VPHP decontamination; therefore, it would pick up a substantial amount of H2O2 upon exposure to high VPHP concentrations during the decontamination phase. The tubing may actually become H2O2-saturated in this process. The second stage is the tubing's continuous H2O2 uptake during the filling process from the residual VPHP in the isolator air. Understanding the relative importance of these two uptake stages would help develop risk mitigation strategies.
To differentiate these two uptake sources, empty 4.8 mm ID tubing with a total length of 212.5 mm (the lengths of Components A + B + C in Table I) was decontaminated in the small-scale isolator and then moved outside the isolator (to discontinue the uptake from residual VPHP in the isolator air). The H2O2-saturated tubing assembly was primed with water, held for 20 min, flushed, and sampled. This experiment was repeated three more times, that is, a total of four extraction cycles. The relationship of H2O2 uptake of water versus the number of post-extraction filling cycles (4 mL fill volume) was plotted and is shown in Figure 5a. The first 20 min extraction showed a very high level of H2O2, ∼700–800 ng/mL. The H2O2 level in the second and the third extraction decreased substantially to 200–300 and 100 ng/mL, respectively, while the level in the fourth extraction primarily overlapped with that of the third extraction. These levels of H2O2 are not acceptable and will increase the risk of mAb oxidation.
H2O2 extraction profiles by water from the 4.8 mm tubing set (a) after VPHP decontamination in a small-scale isolator, (b) after VPHP decontamination + 1 min water flush, and (c) after VPHP decontamination + 5 min water flush; (d) H2O2 extraction profiles by water from the 4.8 mm tubing set after VPHP decontamination in a manufacturing-scale isolator.
After the VPHP decontamination cycle during commercial manufacturing, the tubing assembly was further cleaned and processed by CIP, SIP, and drying. This operation was expected to further remove H2O2 from the tubing. Without the full CIP, SIP, and drying capability in the laboratory facility, two additional experiments were performed in the small-scale isolator by repeating the study above; each was followed by either a 1 min or a 5 min flush of hot (∼70 °C) deionized water before four 15 min extractions. This was to partially mimic the commercial CIP cycle. The 1 min hot water flush greatly removed H2O2 from the tubing to result in 100–250 ng/mL of H2O2 in the first 15 min extraction, 50–150 ng/mL in the second extraction, and <50 ng/mL in the third and fourth extractions (Figure 5b). This decreasing trend continued with the 5 min hot water flush; the H2O2 level diminished to 50–100 ng/mL in the first 15 min extraction and <50 ng/mL for the three subsequent extractions. This outcome supported the hypothesis that CIP and SIP operations can eliminate most of H2O2 taken up during VPHP decontamination. A manufacturing-scale study was performed to confirm this.
A commercial decontamination procedure was performed on 12 tubing assemblies (Assembly 1– 12) in the manufacturing-scale isolator. After the VPHP aeration phase, including CIP, SIP, and drying, all tubing assemblies were removed from the isolator. The Y-connector and the filling needle were immediately dismantled from the tubing (Components A, B, and C in Figure 2b). All three tubing pieces were filled with WFI and sealed with tubing clips. The 12 tubing assemblies were divided into four extraction groups with extraction time at 10 min for Assemblies 1–3), 20 min (Assemblies 4–6), 30 min (Assemblies 7–9), and 60 min (Assemblies 10–12). The extraction was repeated three more times for all extraction groups except for the 60 min group, which was extracted once more. The WFI from all tubing pieces of each extraction group was pooled for H2O2 analysis, and the results are summarized in Figure 5d. Regardless of the extraction time, all WFI samples contained a low level of H2O2, even for the worst-case sample (i.e., the first fraction in the 60 min extraction group), being ∼50 ng/mL. All fractions after the first showed 15–20 ng/mL of H2O2, suggesting that no additional H2O2 can be extracted from the silicone tubing and confirming that the CIP, SIP, and drying operations may have completely removed H2O2 absorbed during VPHP decontamination. Therefore, it is expected that H2O2 uptake during filling is primarily contributed from the residual VPHP in the isolator air.
3.6. H2O2 Uptake by Open, Filled Vials
In the same manufacturing-scale study (Section 2.7), H2O2 uptake by the open, water-filled, 15 cm3 vials was also tested. Three groups of five vials were exposed to the isolator air for 15, 30, and 60 min, respectively. The residual VPHP level in the isolator during vial exposure was monitored by the Picarro VPHP sensor and determined to be in the range of 360 to 410 ppb. The uptake results are presented in Figure 6. There was a slight increasing trend with increasing exposure time. The 60 min exposure resulted in H2O2 uptake of 40–70 ng/mL, while vials after 3 min exposure showed ∼50 ng/mL. The result of the 30 min exposure was slightly lower than what was observed in the first manufacturing-scale water fill study (Figure 3), suggesting the residual VPHP level after the first water fill might be higher than 400 ppb.
H2O2 uptake by open, water-filled 15 cm3 vials after exposed to the isolator air for 15, 30, and 60 min.
3.7. Monitoring of Residual VPHP Level during Manufacturing-Scale Technical Fills
All studies so far suggested that residual VPHP level in isolator air during filling is the most important parameter and should be monitored. In two manufacturing-scale technical fills (one each for the 15 mL and 21.4 mL dose configuration), a Picarro VPHP sensor was used to measure residual VPHP in isolator air after the VPHP decontamination phase. Picarro VPHP traces of the two fills are similar, as shown in Figure 7a (15 mL) and Figure 7b (21.4 mL), respectively. The decontamination phase typically ended when the VPHP level decreased to 300 ppb, immediately followed by the aeration phase. During aeration, the tubing assembly flowpath was sterilized by CIP, SIP, and drying, and, interestingly, the VPHP level did not continue decreasing but instead it increased (to ∼500 ppb) first. This trend could be attributed to the release of H2O2 originally absorbed by equipment and components in the isolator during the decontamination phase. The aeration operation lasted ∼6–7 h. At the end of aeration (or the beginning of filling production), the VPHP concentration in isolator air went down to ∼300 ppb again. Fill weight adjustment was performed before the official vial filling. Fill weight adjustment involved glove movement inside the isolator, which again caused a bump in the VPHP profile. From that point, the VPHP concentration continued declining to the end of the filling operation.
Profile of atmospheric VPHP concentration monitored via a Picarro sensor during two technical filling batches.
3.8. Confirmation Studies
Given the residual VPHP dissipation profile (Figure 7), the worst-case VPHP exposure (i.e., interruption) for both the tubing and open, filled vials occurs at the beginning of filling (immediately after fill weight adjustment). The VPHP concentration at this worst-case point was estimated to be ∼250 ppb.
Laboratory-based H2O2 uptake experiments were performed on the 4.8 mm and 6.0 mm tubing assemblies with 250 ppb VPHP exposure. The H2O2 uptake rate of these two tubing sets in response to interruption times of 30, 60, and 120 min is depicted in Figures 8a and 8b, respectively. The 4.8 mm tubing set took up H2O2 at a rate of 50–100% faster than that of the 6.0 mm tubing set, although the 4.8 mm tubing's surface area-to-volume ratio is roughly 25% higher. The 4.8 mm tubing set failed even in the 30 min interruption period, picking up ∼100 ng/mL H2O2. As expected, the 6.0 mm tubing set performed better. It could allow for 30 min interruption after which H2O2 uptake was ∼50 ng/mL. With the understanding that the 6.0 mm tubing assembly provided filling performance in fill weight accuracy and precision comparable to the 4.8 mm tubing set (unpublished data), these results prompted two recommendations to the manufacturing sites: (1) using 6.0 mm tubing assembly for filling both 21.4 mL and 15 mL configurations and (2) applying a filling line flush after any interruption duration exceeding 30 min.
H2O2 uptake after exposure to 250 ppb VPHP in a small-scale isolator by (a) the 4.8 mm tubing set and (b) the 6.0 mm tubing set.
These recommendations were tested in the process validation campaign, which consisted of three consecutive production batches for each of the 21.4 mL and 15 mL configurations. In each batch an interruption was executed immediately after fill weight adjustment as the worst-case condition. The duration of the interruption was either 15, 30, or 60 min. Table V summarizes the time events of these batches and the corresponding VPHP levels as measured by Picarro. The VPHP decaying profile of each of the process validation batches appears to be consistent with that of the two technical fills. In the beginning of filling production (and during fill weight adjustment), the VPHP level was tight in the range of 310 and 324 ppb (except 266 ppb for Batch 0003 of the 15 mL configuration and 62 ppb for Batch 0002 of the 21.4 mL configuration), suggesting a good reproducibility of the operation and a potential worst-case VPHP exposure with a filling interruption taking place around this time. The VPHP level ranged between 260 and 290 ppb during interruption in all batches except for Batch 0002 of the 21.4 mL vial configuration. This batch immediately followed Batch 0001, thus receiving no separate VPHP decontamination cycle in the isolator. As expected, its VPHP level is very low, ∼60 ppb during interruption.
VPHP Monitoring and H2O2 Uptake Data from Process Validation Batches
The vial sampling plan and H2O2 uptake testing results are also summarized in Table V. Samples of H2O2 uptake taken after each interruption period included six open, filled vials (directly exposed to VPHP during the interruption) and vials filled from the DP solution in the tubing flowpath (i.e., five filling cycles for both the 21.4 mL and 15 mL vial configurations). The level of H2O2 uptake in vials was tested by the Amplex® UltraRed Assay. All H2O2 values are within 60 ng/mL for vials from direct exposure and from tubing migration, even in the worst-case 60 min interruption (for Batch 0002 of the 15 mL configuration). In the batch with the lowest VPHP level (Batch 0002 of the 21.4 mL configuration), vials contained much lower levels of H2O2 (12 ± 3 ng/mL after direct exposure and 22 ± 15 ng/mL from tubing). To understand the general effect, vials filled in the beginning, the middle, and the end of the entire filling operation were also taken and measured for H2O2 content. Their values (the average of six vials) are generally <30 ng/mL, only slightly greater than the reference values (the DP solution sampled prior to entering the isolator), which ranged between 10 and 22 ng/mL (average of three replicates).
Process parameters (interruption limit and tubing size) recommended by the small-scale isolator study proved to be effective in minimizing the impact of VPHP decontamination on protein oxidation via the worst-case H2O2 uptake approach.
4. Conclusions
This case study presents a comprehensive view of the VPHP decontamination process and its potential impact on protein quality. A risk-based tool was successfully applied to identify potential process parameters that might critically influence H2O2 uptake and helped design small-scale and manufacturing-scale studies for process understanding. Although various steps of the filling process could contribute to H2O2 uptake into the vial product, direct exposure of open, filled vials and migration through silicone tubing proved to be the most important sources of H2O2 uptake. Filling interruption time and tubing size were demonstrated to be the most critical process parameters. Careful control of these two parameters could effectively reduce the risk of H2O2 uptake and has been confirmed and proven during the process validation campaign. Overall, this case study will benefit process scientists and engineers in better understanding this unique process as well as designing essential studies for mitigating the risk of H2O2 uptake during process performance qualification and commercial manufacturing.
Conflict of Interest Declaration
The authors declare that they have no competing interests.
Acknowledgments
The authors would like to thank Dr. Alexander Streubel and Mr. Fabian Sager for their assistance in Amplex® UltraRed Assay testing support as well as Mr. Michael Jeske for his transfer leadership.
- © PDA, Inc. 2018
Reference
Jump to section
Related Articles
Cited By...
- Development and Validation of a Customized Amplex UltraRed Assay for Sensitive Hydrogen Peroxide Detection in Pharmaceutical Water
- Vapor-Phase Hydrogen Peroxide Uptake by Silicone Tubing and Primary Packaging Components during Protein Drug Product Aseptic Filling: Impact of Pretreatment and Sterilization Process
- Stability Evaluation of Hydrogen Peroxide Uptake Samples from Monoclonal Antibody Drug Product Aseptically Filled in Vapor Phase Hydrogen Peroxide-Sanitized Barrier Systems: A Case Study