Potential Sampling Errors in Stability Studies Due to Dead Volume in Closed System Transfer Devices

Abstract

Background:

Closed system transfer devices (CSTD) help to reduce the exposure of healthcare professionals to hazardous drugs. They may be used in stability studies conducted on anticancer drugs. During a stability study about polyolefin bags of gemcitabine, Tevadaptor® device was suspected of causing a bias in the evaluation of the concentrations of the first aliquots extracted from the bags.

Objective:

The objectives are to determine whether the use of a CSTD to prepare a drug solution and to withdraw it from a bag can interfere on the measured concentration compared to the expected one and to suggest hypothesis to explain the phenomenon.

Method:

In the first experiment, three polyolefin bags of gemcitabine (5.4 mg/mL) were prepared under aseptic conditions using the Tevadaptor Luer Lock Adaptor®. The day of preparation, five aliquots of 3.8 ml each were sequentially withdrawn from each polyolefin bag using the same device. After one day, a new aliquot was withdrawn from each bag. In the second experiment, three polyolefin bags of gemcitabine (5.4 mg/ml) were prepared under aseptic conditions using a needle. One aliquot was extracted using a needle after the preparation from each bag, and another aliquot was extracted after one day. The concentrations of all aliquots were measured by liquid chromatography coupled to a photodiode array detector during the same run.

Results and discussion:

The concentrations of the first aliquots extracted on day zero from the polyolefin bags using the Tevadaptor Luer Lock Adaptor® exhibit an overestimation of 26% ([95%CI: 23%-29%] P<0.001) compared to the others. Overestimation is not found for subsequent aliquots, or while using a needle to bypass the Tevadaptor® device.

Conclusion:

This case highlights the bias that may arise when using CSTDs in stability studies. They should be used with comprehensive understanding of their technical specifications.

Introduction

The use of closed system transfer devices (CSTDs) to prepare hazardous drugs helps to reduce drug contaminants. CSTDs are defined as “devices designed to prevent the transfer of contaminants into the environment during drug transfer between the vial and syringe by maintaining a closed connection.” ¹ They also prevent the escape of hazardous vapors outside the system during drug compounding and administration. ² Many brands of CSTDs are available on the market, all serving the same purpose.

The Tevadaptor devices from Braun were used in a previous long-term stability study about gemcitabine in dose-banding conditions to prepare the bags and to periodically withdraw aliquots of 3.8 mL for stability testing.^3,4 Preliminary results of the study indicated that the concentrations obtained from aliquots withdrawn on day 0 (the day of preparation) were approximately 25% higher than the expected concentrations. However, from day 1 onward, the measured concentrations aligned with the expected concentrations for the entire duration of the study period. A recent study suggested that the transfer of products via CSTDs may over- or underdose the final concentrations of preparations, especially for small-volume preparations. ⁵ A potential contamination of the solution of the first withdrawn aliquot by the concentrated solution kept in the dead volume of the transfer device must be assessed.

The objectives of this study were to determine whether the use of a CSTD to prepare a drug solution and to withdraw it from a bag can interfere with the measured concentration compared with the expected one and to suggest a hypothesis to explain this phenomenon.

Methods

Solution Preparation

For the first experiment, three polyolefin bags of gemcitabine were prepared aseptically under a vertical laminar airflow hood (CA/RE/S Biological Safety Cabinet, CleanAir, Burladingen, Germany). The preparation was made manually by an injection of 42 mL of the brand solution of gemcitabine (38 mg/mL; lot 87210442BC, Fresenius Kabi, Bad Homberg vor der Höhe, Germany) using a 50-mL syringe with a Tevadaptor Syringe Adaptor (B. Braun, Sheffield, United Kingdom) into a polyolefin bag of 250 mL 0.9% sodium chloride solution (lot 22020902, Carelide, Mouvaux, France) using the Tevadaptor Luer Lock Adaptor (B. Braun) to produce banding doses of 1600 mg (5.4 mg/mL). The company indicates a dead-space volume of 0.16 mL for the Tevadaptor Luer Lock Adaptor. ⁶

On the day of the preparation, five aliquots (day 0: 1–5) of 3.8 mL each were sequentially removed from each polyolefin bag using the Tevadaptor Luer Lock Adaptor coupled with a 10-mL syringe and were immediately frozen. ⁷ After 1 day, a new aliquot (day 1) was withdrawn from each bag and was immediately frozen.

For the second experiment, three polyolefin bags of gemcitabine were prepared under the same process but without using the Tevadaptor device coupled with a 10-mL syringe. On the day of preparation, one aliquot (day 0) was removed from each polyolefin bag using a needle and immediately frozen. After 1 day, a new aliquot (day 1) was withdrawn from each bag and immediately frozen.

Chromatographic Assay

Gemcitabine was measured with an ultra-high performance liquid chromatography system (UHPLC; Acquity UPLC-H Class, Waters, Milford, CT) coupled with a photodiode array detector (PDA; Acquity, Waters) following the method previously described. ⁴ To avoid variability due to experimental conditions, all aliquots were defrosted together and were manually diluted 100-fold in purified water just before chromatographic analysis. The analyses were performed in triplicate.

Statistical Method

The statistical analysis is based on a linear mixed model where the response is the logarithm of the concentration of gemcitabine. This model includes the time (day 0 vs day 1), the device (Tevadaptor vs needle), and a binary variable indicating whether the aliquot is the first from the Tevadaptor (1 = yes; 0 = no) as a fixed effect, and a random intercept per bag to account for correlation between measurements made on the same bag. With this model, we can estimate the expected overestimation occurring with the first aliquot withdrawn through the Tevadaptor device.

This analysis was performed using R 4.3.0 (R Foundation for Statistical Computing, Vienna, Austria, Vienna) and the following packages: ggplot2 (for the graphical representation of data) and lme4 (to fit linear mixed models).

Results

Figure 1 shows the measured concentration of gemcitabine of each aliquot from the three bags withdrawn through a needle on days 0 and 1. It also shows the measured concentration of gemcitabine of each aliquot from the three bags withdrawn through the Tevadaptor device on days 0 (five successive aliquots) and 1. In both cases, the expected concentration was 5.4 mg/mL.

Figure 1.

Evolution Gemcitabine Concentration in Aliquots Withdrawn from the Three Polyolefin Bags on Days 0 and 1 Using a Needle (left) and in the Five Successive Aliquots Withdrawn on Days 0 and 1 from the Three Polyolefin Bags Using the Tevadaptor Device (right).

The statistical method used to estimate the expected overestimation occurring with the first aliquot of the Tevadaptor device gives a mean bias between 23% and 29% (estimation, 26%; 95% CI, 23%–29%; P < .001).

The potential contamination of the first aliquot (day 0: 1) by the high-concentrate brand solution included in the dead volume of the Tevadaptor device can be calculated as follows:

$$\frac{\begin{array}{l}\left(38\mathrm{mg}/\mathrm{mL}\times 0.16\mathrm{mL}\right)\\ +\left(5.4\mathrm{mg}/\mathrm{mL}\times 3.64\mathrm{mL}\right)\end{array}}{3.8\mathrm{mL}}=6.77\mathrm{mg}/\mathrm{mL}$$

where 38 mg/mL is the concentration of the brand solution, 0.16 mL is the dead volume of the brand solution withdrawn from the Tevadaptor, 5.4 mg/mL is the concentration of the polyolefin bag, 3.64 mL is the volume withdraw from the polyolefin bag, and 3.8 mL is the total volume of the first aliquot withdrawn

Discussion and Conclusion

The first experiment revealed a difference in concentrations between the initial and subsequent aliquots that exceed the variability of the method (1.4%). ⁴ Although subsequent aliquots exhibited a measured concentration close to the expected concentration (5.4 mg/mL), the concentration of the first aliquot was 26% higher (6.71 ± 0.31 mg/mL). These results align with those obtained in the preliminary stages of the stability study and suggest that the differences between the concentrations of the first and subsequent aliquots are not related to a time effect. ⁴ This experiment supports the hypothesis of the presence of residual highly concentrated fluid in the CSTD dead space. Calculation of the theoretical concentration of the first aliquot withdrawn via the Tevadaptor device considering a contamination with 0.16 mL of a highly concentrated solution (38 mg/mL) gives a result of 6.77 mg/mL. This is in concordance with the measured concentration (6.71 ± 0.31 mg/mL) and reinforces the hypothesis.

To confirm this hypothesis, the second experiment aimed to bypass the Tevadaptor device by withdrawing aliquots using a needle. In this case, the concentrations of day 0 (5.38 ± 0.11 mg/mL) and day 1 (5.32 ± 0.04 mg/mL) remained constant given the variability of the method. ⁴ This highlights the fact that bypassing the Tevadaptor allowed us to find the expected concentration.

In this study, the impact of the dead volume of the Tevadaptor Luer Lock Adaptor was proportional to the small volume of the first aliquot withdrawn for the stability test. However, this dead volume would have no impact when the entire infusion bag (292 mL) is administered to a patient. Moreover, the homogenization of the solution through the Tevadaptor device during preparation could eliminate the highly concentrated solution in the dead volume of the device. In practical terms, after injecting the 42 mL of drug solution into the bag, the same volume should be withdrawn back into the syringe and injected again into the bag. The two- or threefold repetition of the operation would ensure homogeneity between the solution in the bag and the solution in the dead space of the Tevadaptor, as recommended by Braun. In some cases, devices may have a dead volume as high as 0.58 mL. ⁸ When drugs are packaged in small volumes, the use of a CSTD vial adaptor with a large dead volume may result in under- or incorrect dosing, ⁹ which reinforces the importance of carefully following the instructions for use.

Other challenges faced by companies that commercialize CSTDs include particle formation and incompatibility during dose preparations.^{9 –11} Nevertheless, the risks associated with the use of CSTDs should not overshadow their advantages. First, the use of CSTDs reduces drug leakage. ¹² Second, they help to minimize surface contamination, especially from cytotoxic infusion preparations. ¹³ Third, they contribute to reduce occupational exposure to anticancer drugs, ¹⁴ although a 2018 review challenges this. ¹⁵

Limitations

Several limitations affect this study. First, only one anticancer drug has been studied, and the results obtained cannot be extrapolated to a whole range of pharmaceutical agents. Second, only one model of CSTD is described, and thus it does not predict the potential pitfalls of other systems.

This study highlights one bias that may arise when using a CSTD in a stability study. Despite the protective benefits they provide, such devices should always be used cautiously in stability studies, with a comprehensive understanding of the technical specifications of the system and their potential impact on the study’s design.