Performance evaluation of the HPLC diode array and evaporative light scattering detection for the implementation of dose-banded gemcitabine infusion bags

Abstract

Background

Dose banding (DB) was used to optimise the individualisation of patient treatments with gemcitabine (Gem) in order to improve workload planning at the pharmacy of the University Hospital Centre of Besançon (UHCB). A new simple and fast high-performance liquid chromatographic (HPLC) method was also developed for the quantification of Gem without dilution of the infusion bags.

Methods

Individual doses of Gem preparations were retrospectively analysed over a 1-year period to determine the frequency of prepared doses. Using a maximum gap of 7.5% around the doses chosen, the selected Gem standard doses were 1400 mg, 1600 mg, 1800 mg and 2000 mg. Following the DB scheme, the frequency of prescription of standard and individualised Gem doses was analysed over a period of 10 months. The four selected Gem standard doses were aseptically prepared in polyolefin infusion bags. Each series of 20 bags was stocked under refrigerated storage conditions (4°C) for up to 84 days. The quantification of Gem without dilution of the infusion bags was obtained by the development of a HPLC method coupled to a diode array detector (DAD) or an evaporative light scattering detector (ELSD).

Results

During the 10-month period following implementation of the DB, 75.6% of the 1266 prescribed doses were covered by the four standardised preparations. The number of different Gem doses was reduced from 183 to 55. Concerning the Gem quantification, both heteroscedasticity and non-linearity were observed with DAD. Using an ELSD, the trueness values were between 98.59% and 101.52% with excellent repeatability values between 0.66% and 1.42%.

Conclusion

A new HPLC method has been developed for the quantification of Gem without dilution of the infusion bags prepared in advance as a result of a target DB scheme successfully implemented in our pharmacy department.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• The preparation of gemcitabine (Gem) infusion bags is a resource-consuming process. Dose banding allows a reduction in patient waiting time and improvement in pharmacy capacity planning.

• Reconstituted Gem is chemically/physically stable at 4°C for at least 84 days in the concentration range 7–23 mg/mL in a 0.9% sodium chloride injection packaged in prefilled polyolefin infusion bags.

• An automated compounding system enables the hospital pharmacy to ensure sterile preparation of Gem infusion bags.

• Gem shows maximal ultraviolet absorbance at 270 nm.

WHAT THIS STUDY ADDS

• A dose banding scheme was used for individually calculated doses of 1300–2150 mg with 200 mg increments between each standard dose with a maximum deviation of 7.5% from the prescribed dose.

• An aseptic preparation of Gem infusion bags in series was obtained using an automated compounding system.

• This is a new, simple and fast Gem quantification reverse phase high-performance liquid chromatographic (RP-HPLC) method in infusion bags without dilution using an evaporative light scattering detector (ELSD).

• Identification of Gem in the infusion bags was obtained by its retention time on the ELSD-RP-HPLC chromatogram and confirmed by acquisition of the ultraviolet spectrum using a diode area detector.

• The ELSD could be applicable to other Gem derivative drugs for parenteral administration.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• There is much interest in developing a prospective analytical control during the implementation of a dose banding scheme in the hospital pharmacy. Based on the data obtained in this work, the ELSD can be used for the quantification of drugs in infusion bags prepared in advance, which overcomes the difficulties associated with high concentration levels. Currently, this ELSD is not commonly used in hospital pharmacies but, in the future, its use should be intensified.

Introduction

The role of the pharmacy department in hospitals is to dilute or dissolve commercially available pharmaceutical formulations (including cytotoxic drugs) in appropriate conditions to ensure the protection of nurses and the sterility of the injectable solution. For these reasons, cytotoxic drug infusions are prepared in pharmacy-based centralised cytotoxic drug preparation units.^1–4 Modern cancer therapy developed the concept of personalised medicine involving adaptation of the treatment to the individual patient. Many cytotoxic molecules in oncology are administered in body size-based dosing schedules, but some are now administered using fixed dosing for adult patients.⁵

Dose banding (DB) is a concept which allows the advance preparation of standardised doses of cytotoxic drugs. DB is a system whereby individually calculated doses of intravenous cytotoxic drugs are grouped within defined bands (dose ranges) and predetermined standard doses (band midpoint) are prepared and dispensed.⁶ Standardised doses are the most frequently prescribed doses rounded to 7.5%, although it is recommended that the maximum variation between the exact dose prescribed and the adjusted dose is not more than ±10%.^{7 8} DB thus allows the preparation of standardised products in series. Standard doses can be prepared in advance by batch to increase production capacity and, at the same time, to regulate pharmacy workflow as well as reducing patient waiting time.⁹ Indeed, according to Hendrikx et al, fixed dosing at a suitable selected dose can improve medication safety and decrease healthcare costs without reducing efficacy or safety margins.¹⁰ When the chemotherapy bags are aseptically prepared, microbiological stability is no longer a problem and overall stability depends mostly on the physicochemical stability.

Analytical control is necessary for every bag of a batch prepared by DB. Different strategies are applied in the hospital pharmacy. One approach consists of Fourier Transform Infrared-UV spectrophotometry¹¹ and Raman spectroscopy,^{12 13} and other techniques have been developed such as flow injection analysis and reverse phase high-performance liquid chromatography (RP-HPLC) with a diode array detector (DAD).^{14 15}

Gemcitabine (Gem) is an antitumour agent used in the treatment of several human malignancies including ovarian, cervical, endometrial, lung and breast cancer.¹⁶ Over a 1-year period, 1753 infusion bags of Gem have been prepared in the pharmacy department of the University Hospital Centre of Besançon (UHCB). Among these preparations, it appears that 183 different dosages were made; 1502 of these preparations could have been made in advance. Gemcitabine is a molecule that has frequent digestive, hematologic and pulmonary adverse effects. Due to its harmful effects, dose reductions and treatment postponements are frequent. After dilution of the commercial 38 mg/mL Gem concentrate for infusion solution, the pharmaceutical company (Hospira UK Ltd) indicates in the Summary of Product Characteristics a chemical and physical in-use stability of 84 days for the diluent 0.9% sodium chloride solution for infusion in the concentration range 7–23 mg/mL, stored at 4°C in the absence of light in polyolefin infusion bags.¹⁷ This long-term stability and the quantity of infusion bags prepared paved the way for DB.

Only a few methods are available for the determination of Gem in pharmaceutical formulations, such as ultraviolet spectroscopy¹⁸ and stability indicating validated liquid chromatographic methods.^{19 20} Unfortunately, these analytical methods mostly require a dilution step in the linearity zone of the calibration curve which results in time wasting and sometimes miscalculation of dilution factors.

The aim of this study was to set up DB for the prescribing of Gem infusion bags and to develop a simple and fast HPLC analytical control method for the determination of the Gem quantity in these solutions prepared in advance in the pharmacy department of UHCB. The problem was that the concentration of Gem in the infusion bags was in a concentration range of 7–23 mg/mL. At this high concentration range, the use of a DAD can result in heteroscedasticity of response and the ordinary least square regressions could thus result in an inaccurate estimation of the Gem concentration. As we do not want to do dilution to save time and reduce handling errors, the use of an evaporative light scattering detector (ELSD) was compared with that of DAD.

Methods

Individual doses of Gem preparations were retrospectively analysed over a 1-year period (August 2020 to August 2021) regarding the frequency of prepared doses. Using a maximum gap of 7.5% around the doses chosen, the ranges of the most frequently prescribed Gem doses were identified. The proposed Gem standard doses were thus 1400 mg, 1600 mg, 1800 mg and 2000 mg of Gem per intravenous preparation bag. Following the DB scheme, the frequency of prescription of standard and individualised Gem doses was analysed over a period of 10 months (August 2021 to June 2022). Data were retrieved from the preparation software (BPC Chimio, France), which is installed in the cytotoxic drug preparation unit and allows the entry of prescriptions and the transfer in the pharmacy department of UHCB.

Aseptic preparation of standardised GEM infusion bags in series

Gem standard doses (1400 mg, 1600 mg, 1800 mg and 2000 mg) were prepared in polyolefin infusion bags containing a nominal mean volume of 108 mL 0.9% sodium chloride (NaCl) vehicle solutions (Freeflex, Fresenius Kabi, Germany). The corresponding Gem standard concentrations were thus equal to 12.96 mg/mL, 14.81 mg/mL, 16.67 mg/mL and 18.52 mg/mL, respectively. Each series included 20 infusion bags per standard dose and were aseptically prepared and analysed. Infusion bags were labelled, single packed in plastic bags and heat-sealed. Each series was identified by a batch number and stocked under refrigerated storage conditions (4°C) in the pharmacy department for up to 84 days.

Gem quantification in the infusion bags

HPLC material

The Waters Alliance e2695 HPLC system consists of a DAD (PDA detector 2998), an evaporative light scattering detector (ELSD 2424) with a nitrogen generator Genius SQ (Peak, UK), a column heater, an autosampler and a computer/Empower software. The Polaris C18-A microparticulate (5 µm particle size) column (150 mm length, 4,6 mm internal diameter) was supplied by Agilent Technologies. Water for UHPLC-MS was obtained from Carlo Erba (France) and all the organic solvents for the preparation of the mobile phase were of HPLC analytical grade. Acetonitrile (ACN) and trifluoro acetic acid (TFA) were obtained from VWR (France). NaCl (0.9%) used for pharmaceutical formulations was obtained from Fresenius Kabi (Germany). The pharmaceutical specialities of Gemcitabine Hospira (38 mg/mL) were obtained from Pfizer (France).

Sample preparation

A validation procedure was performed to estimate quantitative parameters of the HPLC method for the analysis of Gem in pharmaceutical formulations. Quantitative performance of the HPLC method was estimated in three separate series at five concentration levels with three repetitions for the calibration standard (CS) and four concentration levels with three repetitions for the validation standard (VS). The concentrations of Gem corresponding to the standard dose of the infusion bags were considered for the determination of the concentration range used in the validation and were prepared by appropriate dilution in NaCl 0.9%. The CS (7.6 mg/mL, 11.4 mg/mL, 15.2 mg/mL, 19.0 mg/mL, 22.8 mg/mL) and VS (14.07 mg/mL, 15.97 mg/mL, 17.87 mg/mL, 20.15 mg/mL) were independently prepared. In addition, an analysis of five bags with three repetitions of each dose of standardised preparations (1400 mg, 1600 mg, 1800 mg and 2000 mg Gem in a bag of 108 mL NaCl 0.9%) was carried out at three different times. The autosampler temperature was 4°C with a 1 µL sample injection volume. Each series involved freshly prepared calibration, validation samples and solvents, washing of the column and liquid chromatography system.

Analytical method validation

The best regression calibration model should have a determination coefficient r² of ≥0.999, a narrow horizontal band of randomly distributed percentage relative error (%RE) values around the concentration axis with the least Σ of the absolute %RE across the whole concentration range. The homogeneity of the curve was determined using a Cochran test. ANOVA tests were applied to determine applicability. Each day for 3 days, three separate series of the CS solutions of Gem were prepared, analysed and quantified using a calibration curve prepared the same day. The repeatability of the method was estimated by the calculation of the relative standard deviation (RSD) of intra-day analysis and intermediate precision was calculated using an RSD of inter-day analysis. Both RSDs were considered acceptable if they were lower than 5%. Method trueness (accuracy profile) was estimated by the evaluation of the recovery of the concentrations of three separate series of the CS solutions of Gem to experimental values found using the mean square equation, and the results should be within the range 95–105%. The overall accuracy profile was constructed following the instructions from the reference.²¹

Results

Evaluation of the DB scheme

Before implementing the DB scheme, a total of 1753 Gem infusion bags corresponding to 183 different doses were prepared during a 1-year period. Using a maximum variance of 7.5% around the standard doses, a total of 1502 bags (85.68%) were covered in the different ranges of the doses (table 1). The number of bags prepared that were not covered by the ranges was 251 (14.32%), with 150 (8.56%) for doses below 1300 mg and 101 (5.76%) for doses above 2150 mg. Figure 1 shows the subdivision of the different bags of Gem prepared before DB. Each bar shows the different doses with a gap of 20 mg between them.

Table 1.

Subdivision of the different doses of gemcitabine prepared (A) before dose banding as a function of the proposed standard doses and (B) after dose banding

(A)
Proposed standard doses (mg)	Gap of doses around the standard dose (mg)	Maximum variance around the standard dose	No of bags prepared	% of total
1400	1300–1500	7.14%	265	15.12%
1600	1500–1700	6.25%	345	19.68%
1800	1700–1900	5.56%	426	24.30%
2000	1900–2150	7.5%	466	26.58%
Total			1502	85.68%

(B)
Doses prepared (mg)	No of bags prepared	% of total
1400	190	15.00%
1600	348	27.49%
1800	226	17.86%
2000	193	15.24%
Total	957	75.59%
Other doses	309	24.41%

Figure 1.

Frequency of the different doses of gemcitabine bags prepared (A) before dose banding (DB) and (B) after DB.

During the 10-month period following the implementation of DB, 957 (75.59%) of the 1266 prescribed doses were covered by the four standardised preparations (table 1). The number of different doses was reduced from 183 to 55. Figure 1 shows the subdivision of the different bags of Gem prepared after DB. Each bar represents each dose prepared.

Analytical method validation

Chromatograms and chromatographic data

A total of 1 µL of the Gem solution (15.97 mg/mL) freshly prepared was eluted isocratically with an isocratic elution of (A) H₂O/0.1% TFA and (B) ACN/0.1% TFA: 80% A + 20% B at a column temperature equal to 25°C. With the DAD, the Gem peak was observed at a retention time (RT) equal to 2.02 min at 270 nm which allowed maximum sensitivity (figure 2A). With the ELSD, two peaks were observed on the chromatogram (figure 2B) with RT equal to 1.85 min and 2.08 min. The peak at 2.08 min corresponded to the Gem peak and the peak at 1.85 min corresponded to the NaCl peak. The peak at 1.85 min was not obviously observed with the DAD detector. So, in order to obtain an analysis time of <3 min and a resolution (Rs) at least equal to 1.5 (Rs=1.70), the flow rate was fixed at 0.8 mL/min. Decreasing the temperature of the nebuliser gas from 70°C to 40°C produced an improvement in the signal/noise (S/N) ratio. Due to the low elution point of the elution solvents used, a temperature of 40°C of the nebuliser gas appeared to be the best condition. Equally, using a nitrogen pressure of 3.45 bars also reduced the baseline noise.

Figure 2.

Reference chromatogram of a 15.97 mg/mL gemcitabine solution (A) at 270 nm and with diode array detector screening with ultraviolet spectrum and (B) with an evaporative light scattering detector. Mobile phase: (H₂O+0.1% TFA) (80%)/(ACN+0.1% TFA) (20%). Flow rate: 0.8 mL/min. Column temperature: 25°C. ACN, acetonitrile; TFA, trifluoro acetic acid

Precision of retention time (RT) and area

The average RT and peak area and the corresponding RSDs for the Gem from three replicates at two different concentrations of Gem solutions were analysed.

For DAD and ELSD, the RT of the Gem peak was 2.0212 min and 2.0846 min, respectively, with an RSD value of 0.074% and 0.053%. The RT RSDs were thus excellent and were found to be less than ±0.10%. These values confirmed that the peak retention was an excellent chromatographic parameter for evaluating the presence of Gem in the infusion bags. This was confirmed by the simultaneous acquisition of the Gem infusion solution ultraviolet spectrum using DAD (figure 2A).

With DAD, for a Gem concentration of 7.6 mg/mL and 22.8 mg/mL, the peak area was equal to 12 281 593 and 22 068 935, respectively, with an RSD value of 1.34% and 0.60%, respectively.

With ELSD, for a Gem concentration of 7.6 mg/mL and 22.8 mg/mL, the peak area was equal to 828 719 and 3 540 610, respectively, with an RSD value of 2.47% and 1.60%, respectively.

For the lowest and highest Gem concentration in the infusion bag, the peak area RSDs were thus excellent and were found to be less than 3% for the two detectors.

Linear or non-linear detector responses for DAD and ELSD

DAD response

Figure 3A shows the variation of the Gem peak area versus the Gem concentration with the use of the DAD. For concentrations ranging from 7 mg/mL to 23 mg/mL and for three replicates of the CS, the mean determination coefficient r² was equal to 0.9632.

Figure 3.

Calibration curves and corresponding residual plots of the regression models for the quantification of gemcitabine (Gem) infusion solutions with (A) DAD and linear regression model; (B) DAD and quadratic regression model; and (C) ELSD and log–log regression model. DAD, diode array detector; ELSD, evaporative light scattering detector.

In this concentration range, a lack of linearity was thus observed for this calibration curve. This was confirmed by the residual plots (figure 3A) and the accuracy profile (online supplemental data) for this linear regression model. Trueness, repeatability and Σ%RE are shown in table 2.

Table 2.

Analytical and regression parameters of the reversed HPLC method for GEM quantification in infusion bags prepared in advance with the use of the DAD and ELSD

Detector	RT (min)	m2	m1	m0	r2	Trueness (%)	Repeatability (%)	Σ%RE
DAD/Linear*	2.021	−	5.90×105	9.95×106	0.9632	113.25–116.18	1.03–1.05	65.904
DAD/Quadratic†	2.021	−2.75×104	1.43×106	3.11×106	0.9960	93.46–100.63	2.17–6.63	61.310
ELSD/Log–log‡	2.084	−	1.39	5.03	0.9999	98.6–101.50	0.66–1.42	1.649

*Peak area=m₁x+m₀.

†Peak area=m₂x²+m₁x+m₀.

‡Log(peak area)=m₁log(x)+m₀ (x: concentration of Gem in mg/mL).

DAD, diode array detector; ELSD, evaporative light scattering detector; Gem, gemcitabine; RT, gemcitabine peak retention time.

Looking at the experimental data it is evident that, with DAD, the measured peak area (A) can be related to the Gem concentration (x) by a quadratic function: peak area=m₂x²+m₁x+m₀ (figure 3B). In this equation, m₂, m₁ and m₀ were constants. By using non-linear regression, for concentrations ranging from 7 mg/mL to 23 mg/mL and for three replicates of the CS, the non-linear regression coefficients of this quadratic equation were determined: m₂=−2.75×10⁴, m₁=1.43×10⁶ and m₀=3.11×10⁶ with a mean determination coefficient r²=0.9960. From this full regression model, a Student's t-test was used to provide the basis for the decision as to whether or not the model coefficients were significant. This test showed that the difference between the groups was statistically significant (t=14.15 with 16 degrees of freedom) with p≤0.001 and a power of performed test α=0.05. Therefore, the results of the Student’s t-test show that no variables can be excluded from the model. The corresponding residual plots and accuracy profile of this quadratic regression model are shown in figure 3B and in the online supplemental data, respectively. The corresponding trueness, repeatability, and Σ%RE are shown in table 2.

ELSD response

With the ELSD, it was usually observed that the measured peak area (A) can be related to the Gem concentration (x) by a log–log linear regression (figure 3C). For concentrations ranging from 7 mg/mL to 23 mg/mL and for three replicates of the CS, the log–log mean linear regression was equal to log (peak area) = m₁*Lnx +m₀ where m₁=1.39 and m₀=5.03 with a mean determination coefficient r² equal to 0.9999. The corresponding residual plots and accuracy profile of this log–log model are shown in figure 3C and in the online supplemental data, respectively. The corresponding trueness, repeatability, and Σ%RE are shown in table 2.

Discussion

Chemotherapy is given to patients as an individually calculated dose based on their body surface area. The DB system means that doses are grouped into discrete bands that lie within 5–10% of a patient’s calculated dose. In our case, the dose variance varied from 5.5% to 7.5%. A DB scheme for individually calculated doses of 1300–2150 mg with 200 mg increments between each standard dose with a maximum deviation of 7.5% from the prescribed dose was accepted by the medical staff of the pharmacy department of the UHCB. DB enables preparation of the standard infusion bags in series and as stock. Qualitative analysis of the Gem infusion bags prepared in advance must be done before administration. The analytical protocol must be simple and fast and directly applicable to the Gem infusion bags without dilution to avoid time wasting and sometimes miscalculation of dilution factors.

Calibration curves are usually used to determine the concentration of a drug with an unknown concentration. Linear regression is the usual way to fit experimental data, but it might be expected that variance of each data point may be quite different (heteroscedasticity).²² Linear regression by the least square method assumes that each data point in the range has a constant absolute variation; however, many analytical methods produce data that are heteroscedastic.²² The use of linear least squares regression in the construction of calibration curves leads to inaccurate estimation of the concentration. Residual plots can be used to evaluate the performance of the linear model.²³ If the data adequately fit the linear model, then the residual should be randomly distributed in a horizontal band centred on the concentration axis. In the present study, the residual plot for linear regression clearly showed that the residual was not randomly distributed around the concentration axis and formed a curved pattern (a ∩-shaped pattern; figure 3A). Both heteroscedasticity and non-linearity were thus observed in the calibration curve of Gem ranging from 7 to 23 mg/mL. Least square linear regression could thus result in large errors in the calculation of Gem concentrations in the infusion bags.

This was confirmed by close examination of the shape of the calibration curve which revealed a smoothly curving nature. There was also a detectable pattern in the residual plots for quadratic regression (figure 3B). The mean value of the data points was above the residual=0 line in the entire concentration range x demonstrated that the predicted value was too low and was thus not randomly distributed around the concentration axis. Quadratic regression could thus result in large errors in the calculation of Gem concentrations in the infusion bags like the linear regression.

This was confirmed by the value of r²=0.9960 obtained for the quadratic regression. Even though this value was higher than that obtained with the linear model (0.9632), it was lower than the acceptable limit of acceptance of 0.999. For quantitative determinations, we conclude that DAD is not the most effective method for quantification of Gem in infusion bags in the 7–23 mg/mL concentration range.

We thus propose alternative solutions. Dilution of the Gem bags before analysing by HPLC-DAD is one possibility. However, this option was rejected because concern for the pharmacy workflow and the necessity to obtain a fast method meant it was not possible to dilute Gem bag samples before analysis. The method must be simple and rapid and should not require any preliminary treatment (including no dilution) of the 108 mL of the solution inside the chemotherapeutic bags. Also, to keep the same use and for the comfort in the care services, it is important to continue to use bags of 108 mL of NaCl 0.9% without dilution before analysis. Therefore, it is necessary to use another detector to select an efficient regression model for Gem calibration data in the 7–23 mg/mL concentration range.

For this, we turned to the ELSD. This detector responds to the amount of material (mass) rather than the absorptivity or ionisation efficiency.²⁴ Therefore, it is a non-selective detector for relatively non-volatile compounds such as Gem. The ELS detection mode was thus an excellent candidate for determination of the peak area with excellent reproducibility for quantitative analysis. The response for ELSD can be related to the following relationship: log (peak area) = m₁*Lnx +m₀,²⁴ where m₁ represents the response intensity and m₀ the shape of the response curve. Using this detector, the residual plots for log–log regression clearly showed that the residuals are scattered approximately randomly around zero, and there is no trend in the spread of residuals with concentration (figure 3C). Theoretically, values of m₁ between 0.66 and 2 have been reported for ELSD.²⁵ This is because ELSD presents different sensitivities regarding the scattering domain of work (reflexion-refraction, Mie and Rayleigh).²⁶

These domains are determined by the ratio of the dried aerosol solute particle diameter (d_s) to the wavelength of the incident light beam (λ).²⁶ In our case of Gem quantification, the slope of the log–log regression gave a m₁ value of 1.39. This value was strikingly close to the theoretical exponent 4/3 corresponding to the Mie domain where 0.1<d_s/λ<1.0. This means that, using the ELSD, a substantial proportion of particles are in the Mie domain—that is, there were a lot of homogenous and small droplets that easily evaporate, creating more surface area for light scattering to occur with a scattered light intensity proportional to d_s ⁴.²⁶ The excellent value of the mean determination coefficient r² (0.9999) confirmed this result and indicated an excellent dose-dependent correlation between peak area and Gem concentration. The relative mean bias coefficients were less than 3.0% for the calibration standard. The mean repeatability RSD coefficient and mean intermediate precision RSD coefficient were less than 2%. The accuracy profile constructed with the data showed that the limits of 95% interval coefficients were all within 2% of the accepted value with trueness values between 98.6% and 101.5%.

Our results therefore clearly showed that, with DAD, the linear and quadratic calibration model misjudged the concentrations of Gem. However, ELSD with the log–log model gave the least value for the Σ%RE in the Gem calibration curve, providing the most adequate approximation of variance. Furthermore, from the accuracy profiles (see online supplemental data) it can be concluded that, for the quantification of Gem, the method is valid only for the log–log model because all tolerance intervals are narrower, better centred around the 100% recovery line and included within the acceptance limits of ±5%. This log–log model reduces the total error of measurement and improves the performance of the analytical procedure.

Using this log–log equation, the accuracy and precision of the back-calculated concentrations were excellent with a high r² value of 0.9999. Consequently, the ELSD could be considered accurate for Gem over the tested range and with a short analysis time (Gem retention time <3 min).

To confirm the applicability of this new reversed phase HPLC-ELS method, four Gem infusion solutions with three repetitions of each dose of standardised preparations (1400 mg, 1600 mg, 1800 mg and 2000 mg Gem in bags of 108 mL of NaCl 0.9%) in a total of 60 intravenous Gem infusion bags were analysed (table 3). The concentration of Gem was calculated with reference to a calibration curve constructed on the same day as described above. As shown in table 3, concentrations of the tested pharmaceutical preparations were excellent and were found to be less than ±2% of the prescribed concentration, which corresponds to a value 80% lower than the acceptable limit value usually set at ±10% for routine analysis of chemotherapy infusion bags in the pharmacy department at the UHCB. The RSD was less than 3.8%. In addition to this quantification, Gem identity can be confirmed via its retention time and the ultraviolet spectrum acquired during injection thanks to the DAD. The method established is rapid, simple, sensitive, accurate and reliable, and it can be used for routine analysis and quality control of Gem in DB in infusion bags.

Table 3.

Analysis of Gem by the reversed HPLC-ELS method in infusion bags prepared in advance at the four standard doses (1400 mg, 1600 mg, 1800 mg, 2000 mg). Peak retention times are given for GEM recognition

Gem infusion bags	Quantity	RT (min)
1400 mg Gem in 108 mL NaCl 0.9%	101.1±1.4%	2.08
1600 mg Gem in 108 mL NaCl 0.9%	100.4±0.7%	2.08
1800 mg Gem in 108 mL NaCl 0.9%	98.6±1.1%	2.08
2000 mg Gem in 108 mL NaCl 0.9%	101.5±0.9%	2.08

ELS, evaporative light scattering; Gem, gemcitabine; HPLC, high-performance liquid chromatography; RT, retention time.

Conclusion

A target DB scheme for prescribing and centralised preparation of standard Gem infusion bags has been successfully implemented in the pharmacy department of the UHCB. In addition, a new reversed phase HPLC method was developed for the quantification of Gem without dilution of the infusion bags prepared in advance. For the first time it has been shown that, unlike DAD, ELSD allowed an excellent quantitative performance in terms of accuracy and precision with an analysis time of <3 min for the four standard doses of Gem. In addition, 100% recognition concerning the identification of Gem in the infusion bags was obtained thanks to the Gem peak RT and confirmed with the acquisition of the ultraviolet spectrum thanks to the DAD. Therefore, this HPLC-ELSD method can be used as a simple technique in quality control and was successfully applied in the pharmacy department of the UHCB.

Data availability statement

No data are available.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

Not applicable.