Wipe-sampling procedure optimisation for the determination of 23 antineoplastic drugs used in the hospital pharmacy

Abstract

Purpose

Optimise a wipe sampling procedure to evaluate the surface contamination for 23 antineoplastic drugs used in the hospital pharmacy.

Methods

The influence of various parameters (ie, sampling device, sampling solution, desorption modes) was evaluated using a validated liquid chromatography–mass spectrometry (LC-MS/MS) method able to quantify 23 antineoplastic drugs widely used in the hospital pharmacy: 5-fluorouracil, busulfan, cyclophosphamide, cytarabine, dacarbazine, daunorubicin, docetaxel, doxorubicin, epirubicin, etoposide, etoposide phosphate, fludarabine phosphate, ganciclovir, gemcitabine, idarubicin, ifosfamide, irinotecan, methotrexate, paclitaxel, pemetrexed, raltitrexed, topotecan and vincristine. Best conditions were tested with real samples from a hospital pharmacy chemotherapy compounding unit.

Results

Polyester swabs (TX714 and TX716) gave satisfactory results for the desorption step for all compounds with mean recoveries of 90% and 95%, respectively. For the wiping step, higher recoveries were obtained using TX716 and isopropanol 75% as wiping solution. As anticipated, most intense contaminations were found close to the chemotherapy production site, on surfaces the most frequently in contact with operators’ hands.

Conclusion

Wipe sampling method was successfully developed and applied to real samples to determine surface contamination with 23 antineoplastic agents in trace amounts.

Introduction

The toxic effects of most antineoplastic agents used for cancer treatment have been well known since their introduction on the drug market.¹ However, occupational risks for healthcare professional daily handling these compounds still need to be fully assessed. Indeed, today no exposure limit exists for anticancer drugs in any work environment except the recommendation of United States Pharmacopeia which indicates only a maximum threshold for cyclophosphamide. To fill the gap, the scientific community made a particular focus on the determination of antineoplastic drug traces in the environment (hospital pharmacies, care units or effluents) and in biological fluids from staff handling these compounds, to control exposure. In the past decade, surface monitoring by wipe sampling has been a useful tool to evaluate contamination trends, implement corrective measures and increase workers’ awareness about the risk related to the manipulation of toxic compounds.^2–5 This non-invasive and relatively easy method allows the identification and quantification of contaminants on various surfaces from workbenches inside production facilities to patients' bedside in care units. The chapter <800> of the United States Pharmacopeia recently recommended to routinely use environmental wipe sampling to evaluate the residual contamination at least every 6 months.⁶ However, no indication was given on which antineoplastic drug to assay or on the contamination threshold criteria. Some authors reported the use of a single antineoplastic agent as a marker of contamination, but considering the difference of toxicity, the wide range of physico-chemical properties and variation in quantities used, the determination of a high number of compounds appeared more relevant. Multi-targeted liquid chromatography (LC) with tandem mass spectrometry (MS) detection is a technique of choice to determine trace amounts of antineoplastic drugs due to its high selectivity and sensitivity. Generally, the development of wipe sampling methods involves two different steps: the definition of operating conditions for the sample wiping (ie, wipe support, moistening solution, solvent and modality of desorption) and those of the separation and detection of target analytes. The purpose of this work was the optimisation of a fast and straightforward wipe sampling procedure coupled to a previously validated LC-MS multi-targeted method, allowing the simultaneous determination of 23 antineoplastic drugs, described elsewhere.^{7 8}

Methods

Sample analysis

Chemicals and reagents

All solvents were MS grade and all chemicals were obtained in the highest analytical quality available. Acetonitrile (ACN), isopropanol (IPA), acetic acid and ammonium hydroxide were purchased from Merck (Darmstadt, Germany). Dimethyl sulfoxide (DMSO) was purchased from Sigma-Aldrich (Buchs, Switzerland). Ultrapure Type 1 water was obtained from a Milli-Q purification system from Millipore (Bedford, MA, USA). Water for injection (WFI) was obtained from Laboratorium Dr. G Bichsel AG, Unterseen, Switzerland,

5-fluorouracil, methotrexate and dacarbazine were purchased from Tokyo Chemical Industry (Zwijndrecht, Belgium). Gemcitabine was obtained from Acros Organic (Geel, Belgium). Busulfan was obtained from Sigma-Aldrich (Buchs, Switzerland). Ganciclovir was obtained from Roche Pharma (Reinach, Switzerland) as cymevene lyophilisate for injection, cytarabine, epirubicin and topotecan were obtained from Toronto Research Chemicals (North York, ON, Canada). Raltitrexed, pemetrexed, docetaxel, paclitaxel, vincristine, doxorubicin, daunorubicin, idarubicin, etoposide, etoposide phosphate, irinotecan and fludarabine phosphate were obtained from Pharmaserv (Stansstad, Switzerland). Cyclophosphamide and ifosfamide were obtained from Baxter AG (Opfikon, Switzerland) as endoxan and holoxan lyophilisate for injection, respectively. [¹³C¹⁵N₂]-fluorouracil, [¹³C²H₃]-methotrexate, [¹³C₆]-irinotecan, [²H₈]-cyclophosphamide and [²H₅]-paclitaxel used as internal standards (IS) were purchased from Alsachim (Strasbourg, France).

Stock solutions of standards were prepared by dilution of standard compounds in DMSO (or WFI grade water for pemetrexed) at 1 mg/mL and were kept at −80°C. Stock solutions of internal standards were prepared by dilution of individual radiolabelled compounds in DMSO at 0.1 mg/mL and were kept at −80°C. Stock solutions were thawed at room temperature for 30 min and vortexed a few seconds before use.

Wiping and desorption material

The evaluated materials for wipe sampling were Western Blotting Paper Disc (1030–025 Grade 3 MM) from Whatman (GE Healthcare Life Sciences, Buckinghamshire, UK), polyurethane foam swabs TX712A from Texwipe (Kernersville NC, USA) and polyester swab TX714K & TX716 from Texwipe (Kernersville NC, USA).

Equipment and LC-MS/MS conditions

All samples were determined with a freshly prepared calibration curve for each compound. An adequate volume of internal standards was added during the desorption step to obtain a final concentration of 50 ng/mL for each IS. The samples were stored at 5°C in the autosampler immediately after preparation and analysis sequence time was limited to 10 hours to avoid degradation of samples.

Safety consideration on antineoplastic agents' handling

Because antineoplastic agents are highly toxic compounds, their processing required strict safety precautions to limit analyst and environment exposure. All powders were weighted and solubilised under horizontal laminar airflow safety cabinet equipped with HEPA H14 filters. To limit the volume of toxic waste generated and intensive exposure of analyst, all dilutions to prepare calibration and validation samples were performed with an automated liquid handling workstation (Freedom EVO, Tecan, Männedorf, Switzerland). All instruments and materials in contact with toxic compounds were treated as hazardous waste. Personal protective equipment (gloves, gown, mask …) were chosen following literature recommendations.^9–12 Moreover, all experiments were performed in a laboratory dedicated to the manipulation of toxic compounds. Specific ventilation was used to ensure a lower pressure inside the laboratory to contain any potential contamination.

Wipe sampling optimisation

Desorption step development

Standard solution of the 23 cytotoxic drugs at 2000 ng/mL in isopropanol was spotted on wiping devices to obtain a final amount of 200 ng per sample for each compound. Then, desorption was performed in 10 mM acetic acid pH 5.1 with 2% acetonitrile solution. The five internal standards were added at a final concentration of 50 ng/mL each. Different desorption procedures (sonication for 5 min and vortexing for 5 min) were tested. Each desorption procedure was repeated in triplicate (N=3). The solutions were transferred into glass vials for UHPLC-MS/MS analysis. Results were expressed as recoveries (in per cent) as mean±RSD (relative SD)

Surface sampling step development

Standard solution of the 23 cytotoxic drugs at 2000 ng/mL in isopropanol was spread over a stainless-steel plate and a Type I borosilicate glass plate (10×10 cm) with an adjustable volume micropipette to obtain a final amount of 200 ng for each drug. After solvent evaporation (15–30 min), the surface was wiped with TX716 swabs moistened with aqueous isopropanol solutions at different concentrations (50%, 75% and 100%, v/v). Then, desorption was performed in 10 mM acetic acid pH 5.1 with 2% acetonitrile solution. The five internal standards were added at a final concentration of 50 ng/mL each. The samples were vortexed for 5 min, and the solutions were transferred into glass vials for UHPLC-MS/MS analysis. Each wiping procedure was repeated in triplicate (n=3). Results were expressed as recoveries (in per cent) as mean±RSD.

PCA analysis method

Principal component analysis (PCA) was carried out with SIMCA software (version 15, Umetrics, Umeå, Sweden) after Pareto scaling. PCA is a projection method that allows an easy visualisation of the samples' distribution and the detection of the major trends in the dataset without any prior information. It summarises multivariate data using a limited number of dimensions that account for the largest sources of variance, that is, the principal components (PCs).

Application to real samples

Contamination of workplace surfaces was evaluated in a hospital pharmacy chemotherapy compounding unit. Sample location included not only critical locations (workbench surfaces and objects inside class III biosafety cabinets, BSC), but also other areas that could come into contact with technicians’ hands, such as other places in clean rooms (surrounding the biosafety cabinets) or logistics rooms (room where all the logistical and administrative tasks are conducted, that is, from prescription validation to preparation of materials and products, and then final reconciliation outside the clean rooms). For each location, a surface representing approximately 100 cm² was wiped with a moistened swab. For non-planar location, the whole surface was wiped. After complete drying, the swab was desorbed, and the solution was transferred into a glass vial for analysis by LC-MS/MS. Sampling was performed for all locations on the same working day, just before the weekly decontamination process.

Results

Desorption step

On the four sampling devices tested, paper disc and polyurethane foam (TX712) exhibited the lowest recoveries for most of the compounds with either sonication or vortexing desorption modes (figure 1). Mean recoveries for paper disc were 69% in both modes, whereas polyurethane foam mean recoveries were 74% and 84% for sonication and vortexing, respectively. However, with TX712, very low recoveries were observed for docetaxel and paclitaxel (<10% with sonication). The polyurethane foam swab was only advantageous for anthracycline compounds (doxorubicin, epirubicin, idarubicin, daunorubicin) and etoposide phosphate in both modes. Polyester swabs (TX714 and TX716) gave globally higher recoveries for all the compounds with mean values of 90% and 95%, respectively. Sonication or vortexing did not have any significant effect on global recovery.

Figure 1.

Superimposed histograms of desorption recoveries obtained for all compounds after sonication for 5 min (A) or vortexing for 5 min (B) with the 4 sampling devices tested.

Wiping step

PCA results

As a starting point for the wiping results, a table containing the percentage recovery obtained with various concentrations of IPA for the 23 compounds on two surfaces (stainless steel or Type I borosilicate glass) was considered. In order to take into account the efficiency of the extraction process, a PCA model was computed after Pareto scaling. The two first principal components summarised 64.6% and 16.2% of the total variance, respectively. Samples obtained with 100% IPA were located on the left of the score plot, 25% IPA in the middle and 75% IPA on the right (figure 2). The loading plot indicated a common pattern for all measured compounds, with positive loading values. The lowest recoveries were therefore associated with 100% IPA wiping solutions, an intermediate situation was observed with 25% IPA, while the best recoveries were obtained with 75% IPA, independently from the desorption material (stainless steel or glass).

Figure 2.

PCA analysis of the impact of isopropanol concentration on the wiping recoveries on stainless steel and glass plates.

Application to real samples

On the 20 tested surfaces, 18 were contaminated with at least one antineoplastic agent (table 1). Total contamination for the 23 tested drugs ranged from 8 ng to 8989 ng with a median value of 109 ng for the 18 contaminated samples. 5-fluorouracil, ganciclovir, topotecan and cyclophosphamide were the most detected drugs (in total amount) with 9137, 909, 554 and 304 ng, respectively. Pemetrexed, raltitrexed and vincristine were not detected in any collected sample.

Table 1.

Results (in ng/swab) of wipe sampling in chemotherapy compounding unit (BSC=Type III biosafety cabinet)

	5-fluorouracil	Cytarabine	Fludarabine	Ganciclovir	Gemcitabine	Dacarbazine	Methotrexate	Pemetrexed	Busulfan	Raltitrexed	Etoposide phosphate	Topotecan	Ifosfamide	Cyclophosphamide	Etoposide	Daunorubicin	Irinotecan	Doxorubicin	Epirubicin	Idarubicin	Vincristine	Docetaxel	Paclitaxel
BSC glove	40	19		3	2		1							13			1		6				8
BSC glove	1	5		333	208							428		11	11			5
BSC PC mouse		6		521	10							66	1	10	1
BSC PC mouse	65	148	178	4	5		5		1			23	1	18			4	7	15				55
BSC touch screen	13	8	8	1			1							3					3	1
BSC weighing scale	1	8		2			2							20						1
BSC workbench	60	12	14	1			5		1			11		30		6		6	5	3
Chlorhexidine bottle (inside isolator)	46	50	27	1			2		2			11		12			1	10	7	2			3
BSC door handle	3			2			8						3	145									2
Production room – weighing scale				21	23									17									64
Production room – door handle
Production room – transfer box handle	2	3				2								1								1
Production room –transfer box	1													4
Production room floor	1	3		2	5							7		5		8		6	5	2
Production room floor	1	3		2	4							8		1									2
Logistic room floor			6		3									1									9
Logistic room – storage box (inside fridge)	6	9.8		3	5							295		30	7		1			2		10	15
Logistic room – 5-fluorouracil storage box	8902	9	26	8	5								1	5						2			31
Logistic room – fridge handle
Logistic room – PC mouse	2	13		8	3						4			5						1

Discussion

The development of a generic wipe sampling method to assay more than 20 antineoplastic drugs may be considered as a tedious task owing to the wide range of physico-chemical properties and the number of parameters that might influence the overall recovery of the method. Three main steps can be distinguished during the wipe sampling procedure: wiping of cytotoxic drugs from the investigated surface to the sampling device; desorption of the compounds from the sampling device to the solution; and the analysis by LC-MS/MS which was already optimised for the determination of trace amounts of antineoplastic drugs. Therefore, this study focused on the development of the two first steps.

Desorption step

Three main parameters might influence the recovery of the desorption step: the desorption solvent, the sampling device and the desorption modality. The nature of the solvent used to desorb compounds from the sampling device might also have a strong influence on the recovery. However, desorption is the last step of sample preparation and the solvent must be compatible with the beginning of the gradient used in the LC-MS/MS method. Therefore, only 10 mM acetic acid at pH 5.1 with 2% acetonitrile was tested as a desorption solvent. In this study four sampling devices were tested with two desorption modalities (vortex and sonication). On the four tested devices, paper discs were detrimental for most of the compounds in term of recovery (figure 1). Moreover, their handling was not ideal as they required the use of tweezers to hold the discs and they disintegrated when the sampled surfaces were not totally smooth. Polyurethane swab demonstrated better resistance to abrasion but recoveries were not satisfactory. These two sampling devices were therefore excluded for the next optimisation step. Polyester swabs TX714 & TX716 gave comparable results in both desorption modalities with a higher average recovery with TX716. Moreover, as TX716 was indicated as clean room-compatible by the manufacturer, this sampling device was selected for the next optimisation step.

Wiping step

The nature and the composition of wiping solvent may affect the global recovery of sampling procedure. However, this solution is directly in contact with the sampled surfaces made from different materials (ie, plastics, stainless steel, glass …) and must not alter them. Therefore the choice of organic solvent was restrained to those already used routinely for microbial decontamination, for instance ethanol and isopropanol. Because IPA was more volatile, only the influence of the concentration of this solvent was studied.

PCA was carried out to provide an unsupervised overview of the dataset. A clear trend related to IPA concentration was highlighted on the scores of the first PC (64.6% of total variance). Samples' distribution clearly highlighted the 75% IPA condition as the wiping method offering the best recoveries for all compounds. Some chemically relevant groupings were observed, for example, anthracyclines compounds (idarubicin, daunorubicin, doxorubicin and epirubicin). However, these results demonstrated the relatively aspecific extraction of the analytes.

Application to real samples

As already reported by other authors2 4 5 13 most intense contaminations were found close to the chemotherapy production site (inside biosafety cabinets), on surfaces the most frequently touched by operators’ hands (ie, chlorhexidine bottle, touch screen, PC mouse …). The results revealed three types of sampling areas: surfaces with high exposure to antineoplastic drugs but with frequent and easy cleaning procedures (ie, biosafety cabinet workbench, weighing scales or clean room door handle) on which only a residual contamination was found; regularly cleaned but hardly washable surfaces (ie, computer mouse inside biosafety cabinet) on which an intermediate contamination was found; and surfaces with lower cleaning frequency on which a high contamination was found (biosafety cabinet gloves, chlorhexidine bottle).

Outside the production area, the same observation could be made regarding the contamination of areas. Most contaminated places were the less frequently cleaned and the more frequently handled (storage boxes, PC mouse).

Even if cleaning procedures exist and are applied routinely, some additional actions are still needed to reduce the contamination of the working environment in the chemotherapy compounding unit.

Conclusion

A wipe-sampling procedure allowing the determination of 23 antineoplastic drugs used in the hospital pharmacy in trace amounts was optimised using a previously validated LC-MS/MS method. A sampling device, a sampling solution and a desorption modality were clearly defined. The method was successfully applied for the determination of surface contamination at the cytotoxic production unit of our hospital. Therefore, this approach is particularly appropriate for environmental monitoring and can be used to identify the exposure of hospital staff who handles cytotoxic drugs and to validate decontamination procedures. A service available to all European healthcare professionals has now being launched based on the developed procedure (www.cytoxlab.ch).

Key messages.

What is already known on this subject

• Handling cytotoxic agents present a major risk for healthcare professionals, and surface contamination is the major route of exposure

• Several studies were reported on the surface contamination in hospital pharmacies and usually a few antineoplastic compounds were determined by the authors.

What this study adds

• To our knowledge it is the first validated wipe-sampling method reported to determine more than 20 antineoplastic drugs

• The developed wipe-sampling method is simple, fast, reliable and safe for operators.

• The applicability of the method was demonstrated with real samples from the chemotherapy compounding unit.