Show Summary Details
More options …

# Pharmaceutical Technology in Hospital Pharmacy

Editor-in-Chief: Lagarce, Frédéric

Ed. by Crauste-Manciet, Sylvie

Free Access
Online
ISSN
2365-242X
See all formats and pricing
More options …
Volume 1, Issue 3

# Implementation of Analytical Control of Low Volume Pediatric Cytotoxic Drugs Preparations using a UV/Raman Spectrophotometer

Thibaut Chouquet
• Corresponding author
• Department of Pharmacie, Hôpital Armand trousseau, 26 Avenue du Dr Arnold Netter, Paris 75012, France
• Email
• Other articles by this author:
/ Guy Benoit
• Department of Pharmacie, Hôpital Armand trousseau, 26 Avenue du Dr Arnold Netter, Paris 75012, France
• Other articles by this author:
/ Karine Morand
• Department of Pharmacie, Hôpital Armand trousseau, 26 Avenue du Dr Arnold Netter, Paris 75012, France
• Other articles by this author:
Published Online: 2016-09-27 | DOI: https://doi.org/10.1515/pthp-2016-0013

## Abstract

Background: Pediatric chemotherapy preparations are usually not analytically controlled, for several reasons. First, they are generally made in syringe, which does not allow to take a sample without changing the final volume. Secondly, the percentage of the dose consecrated to control is important and finally low concentrations can cause sensitivity problems. This lack of quality control, greatly reduces the security of the chemotherapy circuit.

Objective: The main objective is to develop an analytical control to low volume pediatric preparations, made in syringes or in infusion bags with a final volume from 20 to 50 mL.

Material and Method: The development of analytical control automatons, like QCPrep+®, allows a fast and reliable qualitative and quantitative analysis, using UV and Raman spectroscopy. Two protocols were tested. The protocol 1 was intended to test a small sample volume further diluted automatically by the QCPrep+® in order to obtain a sufficient volume for the analysis. The protocol 2 was aimed to test a one milliliter sample volume and several analysis volume without dilution.

Results: Our protocol implies the overfilling with one milliliter of solvent followed by a sampling of one milliliter. The analysis is performed with 900 µL. Ten cytotoxic drugs commonly used in pediatric oncology have been validated according to ICH recommendations. The development of analytical control for low volume preparations allows young patients to benefit the same level of safety and quality than adult patients.

## Introduction

Antitumor cytotoxic drugs are hazardous drugs with high toxicity and narrow therapeutic range. For patient, an overdose may result in serious or fatal side effects at the expense of treatment efficiency [15]. In children with oncological or hematological malignancies, this overdose is even more potent. As such, the French National Agency of Drug Safety (ANSM) classifies it among the twelve events that should never happen (“never events”) [6]. On the other hand, a sub-dosage can compromise treatment efficiency and potential recovery. Pharmacists responsible for centralized units’ reconstitution play an important role in order to secure delivery of injectable cytotoxic drugs. Indeed several types of quality controls are implemented [714] to ensure conformity of the preparations. The first level of quality control is a visual control. During preparation, pharmacy technicians or pharmacist controls the manipulator and verify all critical steps related to drug identity or volumes withdrawn. This method is flawed as error detection is solely based on the human factor. To improve the effectiveness of visual verification, photographic or video-control were developed [15]. However, despite the interesting potential of this surveillance, we lack long term experience and feedback. A secondary type of control can be achieved with gravimetric control. It can be done on the final preparation or during production. It’s a good tool used for quality control of chemotherapy preparations [12] and it brings an upper level of security than visual inspection [16]. The analytical control, allows to identify and quantify the drug and identify the solvent. By regulation, it is not mandatory to carry out such controls. However, the French Society of Oncology Pharmacy (SFPO) strongly recommends to implement analytical controls to ensure the quality of preparations and patients’ safety [17]. In recent years, analytical control automatons were developed to meet the demand of analytical controls [8, 1825]. The first one combined UV and Infrared spectrometry (Multispec®, Microdom, France). Its successor, named QCPrep+® (IconeService, France) was improved by replacing the infrared by Raman spectroscopy. This is a destructive analysis using a fraction of the preparation. The sample volume required is around one milliliter. UV spectrometry coupled with Raman spectroscopy offers very good performance to identify and quantify drugs. Analyzers which combine both these functions are more efficient than UV-HPLC (identification with UV spectrum and retention time) or UV-FTIR (Multispec®) [20, 21]. In our centralized preparation unit, visual and gravimetric controls were first implemented. To complete and improve our practices, we decided to acquire a QCPrep+®. Only preparations in infusion bag with a volume greater than 50 mL were controlled analytically and represented 1/3 of our production. Indeed, in pediatrics, doses and volumes are often lower than in adults. Therefore, the majority of preparations are made into syringes with a final volume of less than 50 ml. Usually these preparations are not analytically controlled, for several reasons: [1] a syringe is a measuring device that allows a visual check and when sampling, there is a discrepancy between the final volume of the preparation and the labeling; [2] the percentage of the dose dedicated to control is important, [3] low concentrations can cause analytical sensitivity problems. Thus, our objective was to develop an analytical control to low volume pediatric preparations, made in syringes or in infusion bags with a final volume from 20 to 50 mL.

## Chemicals

Commercially available cytotoxic drugs were used to validate the analytical methods. The 10 different drugs used and the corresponding chemotherapy protocols, standard doses and the minimum and maximum concentrations analyzed are detailed in Table 1.

Table 1:

Details of 10 drugs analyzed: INN (International Nonproprietary name), brand names or generic names, chemotherapy protocols, dose, dilution volume, maximum and minimum concentrations analyzed.

## Equipment

The QCPrep+® analyzer is an “all in one” automaton composed of a sampling and injection system, a spectrometric analysis system and two management softwares. Sample vials are placed on a motorized carousel with 110 slots coupled to a rotatable turret performing sampling and injections. This turret is composed of a 2.5 mL syringe with a micrometer screw (accuracy of 0.1 % v/v), and a needle able to penetrate the septum. Unlike the Multispec® analyzer, the QCPrep+® is able to dilute the samples and the analysis volumes can be modified. The analysis system consists in a spectrophotometer and a RAMAN spectrometer (BWTek® (USA)). Each spectrometer is equipped with a sample cell: the UV cell is first filled with 500 μL of sample before the Raman cell. If the sample volume is less than one milliliter, the UV cell is completely filled while Raman cell is partially filled and the quality of the analysis is poor. In most cases, UV analysis is sufficient to identify and quantify the drugs. For drugs with close molecular conformations, such as anthracyclins or oxazophosphorins, Raman analysis is necessary. Identification is performed by comparing the acquired spectra with prerecorded ones, with a correlation algorithm. Some points of the acquired spectra are compared with those of all the bank’s spectra. After identification, quantification is carried out from the drug calibration curve. The spectra are untreated, they are not normalized. One analysis is performed in less than 2 min. The calibration curves are plotted automatically by the QCPrep+® through its ability to perform dilutions. Thus, for each drug, only the highest point of the range was prepared and further automatically diluted.

## Optimization of analysis parameters

To implement the control of low-volumes pediatric chemotherapies, the main prerequisites were:

• The sample volume must not represent more than 5 % of the final volume. Although there is no recommended threshold, we consider this value as the maximum fraction of the total dose that can be used for the control.

• This volume must be sufficient for a good quality analysis (quantitation and identification of the drug)

• The final volume of the preparation must match the labeled volume

Tests were carried out with solutions of cytarabine 6.00 mg/mL and 30.00 mg/mL and preparations of doxorubicin 1.00 mg/mL. The concentrations were chosen to assess the impact of dilution on a low level concentration (cytarabine 6.00 mg/mL) and concentration at which the diluted sample remains in the linear range (cytarabine 30.00 mg/mL). The concentration of doxorubicin is an average level of quality control for which we had good results. Controls are prepared from commercial drugs. Cytarabine and doxorubicin are ready to use. The dilutions are performed in appropriate conditions for handling hazardous compounds as cytotoxic agents. Dilutions were made with micropipette and volumetric flasks.

Two protocols were tested. The protocol 1 was intended to test a small sample volume (volume withdrawn from the preparation) further diluted automatically by the QCPrep+® in order to obtain a sufficient volume for the analysis (analysis volume=volume withdrawn from the vial by the analyzer). The protocol 2 was aimed to test a one milliliter sample volume (an overfilling of the preparation with one ml of diluent is necessary) without further dilution. All analyses were performed in triplicate. Data were collected as followed: identification of the drug and solvent, sample quantification (percentage error over the theoretical concentration), smooth progress of the analysis (absence of air bubble and good acquisition of spectra).

## Analytical method validation

After the identification of new analytical parameters, the method was validated for 10 drugs (asparaginase, cyclophosphamide, cytarabine, doxorubicin, daunorubicin, ifosfamide, methotrexate, vincristine etoposide and etoposide phosphate) as per ICH guidelines [26]. Validation parameters were:

• Linearity: a calibration curve was obtained for each drug. The concentration range depend on the minimal and maximal concentrations used in patients. For each drug, a stock solution, representing the maximal concentration was prepared and further automatically diluted. Correlation coefficient (R²) was used as criterion of linearity. A single calibration range was performed on Day 1 of the validation. Controls at day 2 and day 3 were made with this calibration range.

• Accuracy: is the closeness between the measured value and the nominal true value. It was obtained by calculating the percentage of the nominal value (relative error %) of three concentrations. The mean concentration should be within ±5 % of the nominal concentration.

• Precision: is assessed by the repeatability and reproducibility. Three levels of concentrations were considered for each range: low, medium, high. On day 1, six analyzes were performed for each level to determine repeatability. On days 2 and 3, six analyzes were performed, each day, to determine the reproducibility. For repeatability (RPA), the coefficient of variation (RSDrpa) was calculated from the standard deviation (σrpa) and average (mrpa) in the first series:

$\mathrm{R}\mathrm{S}{\mathrm{D}}_{\mathrm{r}\mathrm{p}\mathrm{a}}=100×{\mathrm{\sigma }}_{\mathrm{r}\mathrm{p}\mathrm{a}}/{\mathrm{m}}_{\mathrm{r}\mathrm{p}\mathrm{a}}$

For reproducibility (RPO), the variation coefficient (RSDrpo) was calculated from the standard deviation (σrpo) and average (mrpo) of the three series as: $\mathrm{R}\mathrm{S}{\mathrm{D}}_{\mathrm{r}\mathrm{p}\mathrm{o}}=100×{\mathrm{\sigma }}_{\mathrm{r}\mathrm{p}\mathrm{o}/}{\mathrm{m}}_{\mathrm{r}\mathrm{p}\mathrm{o}}$

In each case, the RSD must be less than 5 %.

• The limits of detection (LOD) and quantification (LOQ): Were calculated according to the ICH recommendations:

$\mathrm{L}\mathrm{O}\mathrm{D}=3.3.\mathrm{\sigma }/\mathrm{S}\phantom{\rule{1em}{0ex}}\mathrm{L}\mathrm{O}\mathrm{Q}=10.\mathrm{\sigma }/\mathrm{S}$

Where σ is the standard deviation of the response and S is the slope of the calibration curve.

## Identification

Suitable identification tests should be able to discriminate compounds with closely related structures. Among these compounds: anthracyclins, oxazophosphorins and vinca-alkaloids are often used in pediatric oncology and cannot be discriminated easily. Thus, the ability of the QCPrep+® analyzer to discriminate: doxorubicin from daunorubicin, cyclophosphamide from ifosfamide and vinblastine from vindesine was tested. Chemical structures of doxorubicin, epirubicin and cyclophosphamide, ifosfamide are shown in Figures 1 and 2. For each couple of compounds, specific spectral areas within the Raman or the UV spectra allowing to differentiate them were determined and control samples were analyzed using these settings. For each drug, several calibration ranges were tested, 18 controls of three concentration levels were carried out on 3 different days. For each control, was assessed the drug identification: resemblance percentage of the acquired spectrum with the bank’s spectrum and ability to discriminate structurally related drugs. Although QCPrep® is delivered with ready to use spectra, the bank was rebuilt with the drugs we used. Each new drug (new manufacturer or generic) spectra and calibration were updated and validated.

Figure 1:

Chemical structures of doxorubicin and epirubicin.

Figure 2:

Chemical structures of ifosfamide and cyclophosphamide.

## Method development

Protocol 1 was carried out in order to determine if a smaller sample volume allowed a good-quality analysis. The results are shown in Tables 2 (cytarabine 6.00 mg/mL) and 3 (cytarabine 30.00 mg/mL). For both concentrations of cytarabine, drug and solvent were properly recognized and the Raman spectra were well acquired. The dilution tests showed that the percentage of error increases with dilution. Thus, for a 10-fold dilution of the 6.00 mg/mL solution, the error is +37.4 %. Finally, no test with cytarabine 6.00 mg/mL was consistent. The analysis carried out with cytarabine 30.00 mg/ml showed better results but the percentages of error were greater than 5.0 % in all cases. Only the 2-fold dilution was satisfying with a percentage of error of 2.0 %. No air bubbles were observed for any volumes and concentrations.

Protocol 2 was secondly conducted, following the results of protocol 1. The results are shown in Tables 4 (cytarabine 6.00 mg/mL) and 5 (doxorubicin 1.00 mg/mL). Cytarabine was properly recognized and quantified for all tested parameters.

Table 2:

Results of protocol 1 – cytarabine 6.00 mg/mL – linear from 0.0 to 37.5 mg/mL.

Table 3:

Results of protocol 1 – cytarabine 30.00 mg/mL – linear from 0.0 to 37.5 mg/mL.

Table 4:

Results of protocol 2 – cytarabine 6.00 mg/mL – linear from 0.0 to 37.5 mg/mL.

Table 5:

Results of protocol 2 – doxorubicin 1.00 mg/mL – linear from 0.25 to 5 mg/mL.

Nevertheless, with analysis volumes of 600 µL, 700 µL and 1 mL, solvents were badly identified. The removal of the entire sample of one milliliter led to the formation of air bubbles (approximately 200 µL). This is due to the shape of the vial which does not allow to recover all the sample. This air was injected into the Raman cell and interfered with the analysis. For doxorubicin, which is identified by Raman spectrometry, only analysis volumes of 800 µL and 900 µL are efficient to correctly quantify and identify the samples (Figure 3). Figure 4 shows the typical profile of a Raman spectrum obtained in the presence of air bubble when the analysis was conducted with an insufficient volume. In the first part of the spectrum between 12,500 and 11,111 cm–1 (800–900 nm) the diffusion is negative, and rapidly increases to zero but no characteristic peak of the drug appears. In Raman spectrometry, the solvent is identified in the spectral area 9,901–9,709 cm–1 (1,010–1,030 nm). In this area, glucose diffuses unlike NaCl 0.9 %. This range is the specific area of solvents and not cytotoxic drugs emits. Yet, on Figure 4, the identification of the solvent was also impossible. This poor spectrum acquisition, generated by air bubbles, corresponds to the most informative part of the spectrum, resulting in a significant loss of information. It is therefore impossible to make an identification of the drugs and the solvent. Finally, regarding the results obtained with protocol 1 and 2, it was decided to overfill low volume preparations with one milliliter of solvent and to dedicate this milliliter to the analytical control. The optimal analysis volume was set at 900 µL.

Figure 3:

Normal Raman spectrum of doxorubicin 2.5 mg/mL in G5 %. X-axis represent wavelength in Raman from 800 nm (12,500 cm–1) to 1,040 nm (9,615 cm–1). Y-axis represent intensity of the signal.

Figure 4:

Raman spectrum of doxorubicin 2.5 mg/mL in G5 % acquiredwith a 700 µl sample. X-axis represent wavelength in Raman from 800 nm (12,500 cm–1) to 1,040 nm (9,615 cm–1). Y-axis represent intensity of the signal.

## Identification

The identification tests showed that discrimination of molecules with closed conformations is possible. Both anthracyclins and oxazophosphorins are discriminated using Raman spectra whereas the discrimination of vinca-alkaloids is based on the second part of the UV spectra. The chosen specific spectral areas (Figures 5, 6 and 7) were validated and are now routinely used. Identification of vinca-alkaloids is performed in UV, in a wide spectral region between 255–332 nm (Figure 5). To discriminate structurally related drugs in Raman, it’s necessary to select spectral areas with differences, even minor. The identification algorithm does not allow discrimination between two close spectra. Therefore, the pharmacist must select specific areas for discrimination. Different areas were evaluated to discriminate oxazaphosphorins and anthracyclins. Identification of anthracyclins is carried out in Raman, in two spectral regions between 11,905–11,792 cm1 (840–848 nm) and 11,534–11,261 cm1 (867–888 nm) (Figure 6). Identification of oxazophosphorins is carried out in Raman, in a single spectral region between 12,092–11,962 cm1 (827–836 nm) (Figure 7).

Figure 5:

Specific spectral area for discriminating vinca-alkaloïds (vincristine 0.5 mg/mL in NaCl 0.9 % and vinblastine 0.1 mg/mL in NaCl 0.9 %). X-axis represent wavelength in UV from 190 nm to 400 nm. Y-axis represent intensity of the signal.

Figure 6:

Specific spectral area for discriminating anthracyclins (doxorubicine 2.5 mg/mL in G5 % and daunorubicine 1 mg/mL in G5 %). X-axis represent wavelength in Raman from 800 nm (12,500 cm–1) to 1,040 nm (9,615 cm–1). Y-axis represent intensity of the signal.

Figure 7:

Specific spectral area for discriminating oxazophosphorins (cyclophosphamide 5 mg/mL in G5 % and ifosfamide 10 mg/mL in G5 %). X-axis represent wavelength in Raman from 800 nm (12,500 cm–1) to 1,040 nm (9,615 cm–1). Y-axis represent intensity of the signal.

## Method validation

After determining the new analytical parameters, an analytical validation was realized for 10 drugs. The results are presented in Tables 6 and 7. For each drug, all validation data were within the expected ranges. R² was greater than 0.95, the LOD and LOQ were compatible with our routine use.

Table 6:

Linearity data and limits of detection (LOD) and quantification (LOQ).

Table 7:

Accuracy data of each drug performed with new analysis parameters.

## Discussion

In our centralized unit, the pediatric preparations of chemotherapy are mainly made in syringes (60 % of production) with a final volume less than 50 mL. Some preparations are made with a final volume less than 20 mL. The study’s objectives do not include these preparations because they represent a small part of the production and the final volume is too low to allow a sample without using a large amount of the total dose. For reasons given above (discrepancy between the final volume and labelling, sampling represents a substantial proportion of the total dose) most of our production was not controlled analytically. Therefore, it was important to develop a specific analytical method for the quality control of such preparations. To our knowledge, no unit controls low-volumes and syringe preparations. Thus, Dziopa et al. [24] and Bazin et al. [19] did not control preparations with a final volume of less than 100 mL. Nardella et al. did not control preparations in infusion bag, with a final volume below or equal to 20 mL [25]. Havard et al. advised to collected samples that should not exceed 0.5 % of the total volume of the preparation. Also they did not control low-volume preparations [8]. The method retained for the analytical control of cytotoxic drugs preparations with a final volume of 20 to 50 mL, consists in overfilling the preparation with one milliliter of solvent and in the withdrawing of this milliliter for the analysis. The analysis is then carried out with a volume of 900 µL. The method does not involve a significant increase in processing time. The time necessary to add one milliliter of solvent in the preparation is negligible. Only the sampling stage, before unrealized, increases the preparation time without impacting the production. The overfilling can not be manage in our software which is not highly adaptable, so a human intervention is necessary. To our knowledge, no chemotherapy prescription software does currently support removal or addition of solvent for analytical control.

## Method development

To develop this method of sampling and analysis, two protocols were tested with doxorubicin and cytarabine. These two drugs were chosen for several reasons: they are commonly used in pediatric onco-hematology; they are ready-to-use, which avoids the bias due to the reconstitution step, and are inexpensive. The identification of cytarabine is performed by UV spectrometry whereas the doxorubicin is identified by Raman spectrometry. Indeed, anthracyclins (doxorubicin, epirubicin, daunorubicin) are structurally very similar and UV spectrum is not sufficient to discriminate them. The identification of the solvent also requires Raman spectroscopy: glucose diffuses in the 9,901–9,709 cm1 (1,010–1,030 nm) range while sodium chloride 0.9 % does not. Thus, tests must be carried out with cytarabine, doxorubicin, glucose and NaCl to evaluate the results obtained with both UV and raman spectroscopy.The results of the tests made with protocol 1 have shown limits due to the dilutions. Indeed, no analysis was quantitatively satisfying while all the concentrations were higher than the LOQ. However, the percentage of error increased with the dilution. Therefore, the smaller dilution factor must be preferred in case of an over-concentrated sample. Level 5 of Protocol 1 with cytarabine 6.00 mg/mL (Table 2) and level 1 of Protocol 1 with cytarabine 30.00 mg/mL (Table 3) have the same concentration: 3.00 mg/mL. However, the results obtained are significantly different, +11.0 % for cytarabine 6.00 mg/mL and –11.5 % for cytarabine 30.00 mg/mL. Theoretically, the automatic dilution system is accurate enough and is used regularly to perform the calibration ranges, without showing linearity worries. In any case, another important point is the discrepancy between the final volume of the preparation and labeling. This problem can be corrected by incorporating the sampling in manufacturing sheet. However, our software (Phedra®) does not allow it. Finally, this method has been forsaken. Protocol 2 was designed to evaluate the minimum volume enabling a Raman spectrometric analysis. In summation, the UV cell is first filled with 500 µL of the sample and the Raman cell is then filled with 500 µL or with the remaining volume in the case of an injection volume less than 1,000 µL. The results showed that volumes of 600 and 700 µL are not sufficient to achieve a Raman analysis. Indeed, the spectra acquired with these volumes do not allow to properly identify the solvent and doxorubicin. The analysis carried out with a sample volume of 1,000 µL and an analysis volume of 1,000 µL is quantitatively satisfying but is not consistent from a qualitative point of view. Indeed, in the same manner as for 600 and 700 µL samples, air bubbles disrupted the Raman analysis. Indeed, 200 µL of air were observed due to the shape of the vial which does not allow to withdraw the totality of the liquid. We have concluded that a good analysis requires a volume of 800 or 900 µl. Although the results obtained with 800 µL and 900 µL were similar, we have chosen to perform the analysis with 900 µL to fill the Raman cell to the maximum and avoid the presence of air. The accuracy of syringes used for preparations with a volume of 20 to 50 mL is one milliliter. In practice, for collecting a volume of less than one milliliter, it is necessary to use a dedicated 1 mL syringe. This adds a sampling device, to the detriment of the protocol 1. Conversely, taking a sample of one milliliter (protocol 2) can be made directly with the final syringe, saving time and material.

## Identification

The identification of drugs may be carried out in UV or in Raman, on the whole spectrum or on a specific spectral area. Several drugs with very similar structures must be discriminated: vincristine and other vinca-alkaloids, doxorubicin and daunorubicin, cyclophosphamide and ifosfamide. UV spectrometry is generally not enough informative to differentiate these molecules, except for vinca-alkaloids for which the very small difference between 255–332 nm is sufficient (Figure 5). In practice, anthracyclins and oxazophosphorins are difficult to identify by UV-HPLC or UV-FTIR [19, 23, 24]. In Raman, some differences are found in very narrow spectral regions (Figures 6 and 7). To ensure the proper identification of these drugs, it is indispensable to build specific libraries, which may differ from those proposed by the manufacturer. This problem is also found for the control of monoclonal antibody preparations, mainly in adult oncology [27].

## Method validation

To validate this method of sampling and analysis, ten drugs assays were validated. All tested parameters were within range defined by the ICH guidelines. These results showed that the new settings have no impact on the linearity, repeatability and reproducibility of assay methods and that it can be used routinely. Our validation results are similar to data from other studies [19, 24, 25]. The LOD and LOQ calculations are theoretical. The formula may tend to overstate values. However, despite the pediatric doses, these concentrations are close to the theoretical LOQ calculated, expect for three drugs: doxorubicin, daunorubicin and cytarabine. However these extreme concentrations represented a very few preparations, mainly in the newborn babies. For these preparations others control are made as a complement to analytical control: a weighing of the final preparation and a photographic control. In order to improve the robustness of our method, it is possible to achieve validation with more conditions: different commercial forms, various solvent (0.9 % NaCl, G5 %, Ringer). However, the results validated our method for routine use.

## Limitation of the method

The newly implemented method meets all our requirements. However, the fraction of the prescribed dose used for the control is still important. In adults, the final volume of the majority of cytotoxic drugs preparations is larger than in children. One-milliliter sample represents a fraction of the total dose of less than 1 % and is therefore negligible. For low volume pediatric preparations, this fraction is 4.76 % for preparations with a final volume of 20 ml (sampling 1 mL in a preparation of 21 mL) 1.96 % for 50 ml (sampling 1 mL in a preparation of 51 mL). There are no recommendations on the maximum fraction of the prescribed dose, which can be used for the control. To answer this question, we compared with the dose banding [28]. This principle was defined by Plumridge [29]: “It is a system in which, after an agreement between prescribers and pharmacists, doses of injectable cancer drugs calculated on an individual basis (skin surface) are rounded to determine standard intervals. The maximum variation between the calculated doses and intervals standards should not exceed plus or minus 5 % of the initially calculated dose”. Dose banding was developed after a survey of 1,104 British oncologists who were willing to reduce a prescribed dose of 5 % and 10 % for 92 % and 40 % of them respectively, in order to facilitate the preparation and to accelerate the providing to the patients [30]. Therefore, according to this study, it is possible to dedicate 5 % or less of the dose to secure the pediatric preparations. The developed method can also be used to control some adults’ preparations made with a volume less than 100 mL. In case of non-conformity, it is not possible to make a second sample. Indeed, taking a second milliliter causes a mismatch between the final volume and the labeling. It could be possible to add a new milliliter of solvent, and to withdraw a second sample of one milliliter, but for the preparations with a final volume of 20 mL, the fraction of the total dose used for the control would be of 9.3 %. For preparation of 50 mL, this percentage would be 3.88 %. In order to harmonize our practices, no new sample is taken in case of non-conformity and the preparation is destroyed. This is a drawback, which is counter balanced by the low rate of non-conformity of our preparations and by the low cost of the majority of the anticancer drugs used in children. The next step should be the application of a narrower acceptance range (±10 %), as recommended by the SFPO. Indeed, for preparations close to the lower limit of the current acceptance range (–15 %), the fraction of the dose not administer to the patient can reach 20 %. Finally, the developed method meets our validation criteria: the volume taken from the preparation must be as low as possible and the volume of analysis must be sufficient to achieve good identification and drugs quantification. It provides some advantages: the final volume of the preparation corresponds to the labeling, the handling is simple (one milliliter of solvent added and taken) and corresponds to the accuracy of the 20 and 50 mL syringes used to make the preparations. To our knowledge and according to the drug stability data, the dilution produced by the addition of one milliliter of solvent is negligible, for cytotoxic drugs used in our unit. The developed method is not specific to syringe preparations. Thus, in order to harmonize our practices, all chemotherapy preparations (low or high volume, in syringe or infusion bag) are controlled by the same protocol.

## Conclusion

Securing the circuit of the cytotoxic drugs is a major issue to improve patients’ safety [31]. The use of an analytical control is a reliable and effective method to ensure the quality of chemotherapy preparations. The development of an analytical method suitable for low volume preparations, allows young patients to benefit of the same level of safety and quality than adult patients. The drawbacks of proposed method are largely outweigh by the safety improvements.

## References

• 1. Al Omar S, Amayiri N, Madanat F. Safety of inadvertent administration of overdose of intrathecal Cytarabine in a pediatric patient. J Oncol Pharm Pract Off Publ Int Soc Oncol Pharm Pract 2015 Oct;21(5):384–7. Google Scholar

• 2. Curran CF. Acute doxorubicin overdoses. Ann Intern Med 1991 Déc 1;115(11):913–14. Google Scholar

• 3. Hachimi-Idrissi S, Schots R, DeWolf D, Van Belle SJ, Otten J. Reversible cardiopathy after accidental overdose of mitoxantrone. Pediatr Hematol Oncol 1993 Mars;10(1):35–40.

• 4. Jurek T, Rorat M, Dys P, Swiatek B. Fatal cisplatin overdose in the treatment of mediastinal lymphoma with the ESHAP regimen – analysis of the causes of the adverse drug event. Onkologie 2013;36(1–2):49–52.

• 5. Bäck H, Gustavsson A, Eksborg S, Rödjer S. Accidental doxorubicin overdosage. Acta Oncol Stockh Swed 1995;34(4):533–6.

• 6. ANSM. Les événements qui ne devraient jamais arriver – Never Events [Internet]. Disponible sur: http://ansm.sante.fr/Dossiers/Securite-du-medicament-a-l-hopital/Les-evenements-qui-ne-devraient-jamais-arriver-Never-Events/%28offset%29/0

• 7. Limat S, Drouhin JP, Demesmay K, Tissot E, Jacquet M, Woronofl-lemsi MC. Incidence and risk factors of preparation errors in a centralized cytotoxic preparation unit. Pharm World Sci 2001;23(3):102–6.

• 8. Havard L, Bonan B, Huen JM, Roche Y, Prognon P. Quality control in centralized cytotoxic reconstitution unit: dream or reality? STP Pharma Prat 2005;15(2):148–60. Google Scholar

• 9. Bourget P, Paci A, Rey JB, Mercier L, Demirdjian S. Contribution of high-performance thin-layer chromatography to a pharmaceutical quality assurance programme in a hospital chemotherapy manufacturing unit. Eur J Pharm Biopharm. nov 2003;56(3):445–51.

• 10. Bouma M, Nuijen B, Jansen MT, Sava G, Picotti F, Flaibani A, et al. Development of a LC method for pharmaceutical quality control of the antimetastatic ruthenium complex NAMI-A. J Pharm Biomed Anal 2003 févr 26;31(2):215–28.

• 11. Bouligand J, Paci A, Mercier L, Vassal G, Bourget P. High-performance thin-layer chromatography with a derivatization procedure, a suitable method for the identification and the quantitation of busulfan in various pharmaceutical products. J Pharm Biomed Anal 2004 févr 18;34(3):525–30.

• 12. Lecordier J, Heluin Y, Plivard C, Bureau A, Mouawad C, Chaillot B, et al. Safety in the preparation of cytotoxic drugs: How to integrate gravimetric control in the quality assurance policy? Biomed Pharmacother 2011 févr;65(1):17–21.

• 13. Bazin C, Cassard B, Caudron E, Prognon P, Havard L. Comparative analysis of methods for real-time analytical control of chemotherapies preparations. Int J Pharm 2015 Oct 15;494(1):329–36.Google Scholar

• 14. Basuyau F, Brunelle P. Comparison of two fast physical measurements for the dose control of cytotoxic injectable preparations before administration. Application to 5-fluorouracile. J Pharm Clin 2000;19(1):5–8. Google Scholar

• 15. Benizri F, Dalifard B, Zemmour C, Henriquet M, Fougereau E, Le Franc B. DrugCam(®)-An intelligent video camera system to make safe cytotoxic drug preparations. Int J Pharm 2016 avr 11;502(1-2):198–207. Google Scholar

• 16. CARREZ L. Fiabilité de la préparation des chimiothérapies : double contrôle ou système gravimétrique ? [Internet]. GERPAC; 2013 [cité 15 déc 2014]. Disponible sur: http://www.gerpac.eu/spip.php?article942

• 17. Blanc-Legier, Carvalho, Daouphars, Desmaris, Escalup L, Helvig, et al. Recommandations SFPO 2012. In 2012. p. 12.

• 18. Bazin C, Vieillard V, Astier A, Paul M. Contrôle rapide et fiable des préparations de chimiothérapies à base d’anticorps monoclonaux à l’aide de l’automate Multispec®. Ann Pharm Fr 2010 mai;68(3):163–77.

• 19. Bazin C, Vieillard V, Astier A, Paul M. Implementation of real-time identification analysis and quantification of chemotherapies preparations with a Multispec® analyser. Ann Pharm Fr 2014 janv;72(1):33–40.

• 20. Bourget P, Amin A, Moriceau A, Cassard B, Vidal F, Clement R. The Raman Spectroscopy (RS): A new tool for the analytical quality control of injectable in health settings. Comparison of RS technique versus HPLC and UV/Vis-FTIR, applied to anthracyclines as anticancer drugs. Pathol Biol 2012 déc;60(6):369‑79.

• 21. Bourget P, Amin A, Vidal F, Merlette C, Lagarce F. Comparison of Raman spectroscopy vs. high performance liquid chromatography for quality control of complex therapeutic objects: Model of elastomeric portable pumps filled with a fluorouracil solution. J Pharm Biomed Anal 2014 mars 25;91:176–84.

• 22. Bourget P, Amin A, Vidal F, Merlette C, Troude P, Corriol O. Raman spectroscopy applied to analytical quality control of injectable drugs: analytical evaluation and comparative economic versus HPLC and UV/visible-FTIR). J Pharm Belg 2013 Sept;(3):32–45.

• 23. Delmas A, Gordien JB, Bernadou JM, Roudaut M, Gresser A, Malki L, et al. Quantitative and qualitative control of cytotoxic preparations by HPLC-UV in a centralized parenteral preparations unit. J Pharm Biomed Anal 2009 juillet;49(5):1213–20.

• 24. Dziopa F, Galy G, Bauler S, Vincent B, Crochon S, Tall ML, et al. Aquantitative and qualitative method to control chemotherapeutic preparations by Fourier transform infrared–ultraviolet spectrophotometry. J Oncol Pharm Pract 2013 juin 1;19(2):121–9.

• 25. Nardella F, Beck M, Collart-Dutilleul P, Becker G, Boulanger C, Perello L, et al. A UV-Raman spectrometry method for quality control of anticancer preparations: Results after 18 months of implementation in hospital pharmacy. Int J Pharm 2016 févr 29;499(1‑2):343–50.

• 26. International Conference on Harmonisation. VALIDATION OF ANALYTICAL PROCEDURES: TEXT AND METHODOLOGY Q2(R1) [Internet]. ICH; 1994 oct [cité 29 déc 2014]. Report No.: Version 4. Disponible sur: http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q2_R1/Step4/Q2_R1__Guideline.pdf

• 27. Bazin C, Vieillard V, Astier A, Paul M. Reliable real-time analytical control of monoclonal antibodies chemotherapies preparations on Multispec automaton. Ann Pharm Fr. mai 2010;68(3):163–77.

• 28. Chatelut E, White-Koning ML, Mathijssen RH, Puisset F, Baker SD, Sparreboom A. Dose banding as an alternative to body surface area-based dosing of chemotherapeutic agents. Br J Cancer 2012 Sept 25;107(7):1100–6. Google Scholar

• 29. Plumridge RJ, Sewell GJ. Dose-banding of cytotoxic drugs: a new concept in cancer chemotherapy. Am J Health-Syst Pharm AJHP Off J Am Soc Health-Syst Pharm 2001 Sept 15;58(18):1760–4. Google Scholar

• 30. Kaestner SA, Sewell GJA. National Survey Investigating UK Prescribers’ Opinions on Chemotherapy Dosing and ‘Dose-Banding’. Clin Oncol 2009 mai;21(4):320–8.

• 31. Pelus E, Dellanegra M, Chaterly D. Cytotoxics centralized preparation unit: quality assessment. J Pharm Clin 1998;(17):97–102.

#### Thibaut Chouquet

Thibaut Chouquet is a hospital pharmacist. During his studies and his residency, he specialized in pharmaceutical technology and oncology in several French hospitals. It performs various works on the control of pediatric chemotherapy, transportation chemotherapy and environmental contamination by cytotoxic drugs (assessment by LC-MSMS and mutagenesis study) and obtained his pharma.D in 2015, a university degree on hospital pharmaceutical technology and a Master 2 in human toxicology and risk assessment. He works at the Armand Trousseau Hospital as a pharmacist assistant in the preparatory, laboratory of control and parenteral nutrition.

#### Guy Benoit

Guy Benoit is a hospital pharmacist, head of pharmaceutical technology sector in the University Hospital Group “Est-Parisien” (Saint-Antoine, Rothschild, Armand-Trousseau, Tenon, La Roche-Guyon). His area of expertise is the production (sterile and non-sterile preparations, chemotherapy, parenteral nutrition, formulation, galenic), control (analytical method development, stability studies) and quality assurance. He supervises many pharmacists and residents projects, thesis (pharma D, PhD, publications).

#### Karine Morand

Karine Morand is a hospital pharmacist, currently working in the child hospital Armand Trousseau, in Paris. She provides pharmacy services to the inpatient hematology/oncology unit and the outpatient infusion center. Her area of expertise is oncology pharmacy, in adults and children. She had worked in several centralized cytotoxic preparation units in hospitals or in specialized anticancer centers. She was involved in the improvement of preparation and administration of anticancer drug, in stability studies and in the development of analytical control of cytotoxic preparations.

Revised: 2016-08-26

Accepted: 2016-08-26

Published Online: 2016-09-27

Published in Print: 2016-09-01

Conflicts of interest statement: The authors state no conflict of interest. They have read the journal’s Publication ethics and publication malpractice statement available at the journal’s website and hereby confirm that they comply with all its parts applicable to the present scientific work.

Citation Information: Pharmaceutical Technology in Hospital Pharmacy, Volume 1, Issue 3, Pages 151–162, ISSN (Online) 2365-242X, ISSN (Print) 2365-2411,

Export Citation