IN
DEGREE PROJECT CHEMICAL SCIENCE AND ENGINEERING, SECOND CYCLE, 30 CREDITS
STOCKHOLM SWEDEN 2020 ,
Development and validation of an ultrafiltration-UHPLC-MS/MS
method for quantification of
unbound Beta-Lactam antibiotics cefotaxime, piperacillin, cloxacillin and flucloxacillin in plasma
LEONA CLARIN
KTH ROYAL INSTITUTE OF TECHNOLOGY
Development and validation of an ultrafiltration-UHPLC- MS/MS method for the quantification of unbound Beta- Lactam antibiotics cefotaxime, piperacillin, cloxacillin and
flucloxacillin in plasma
Leona Clarin October 2020
Degree Project in Chemistry KD200X
Examiner: Åsa Emmer, professor, Analytical Chemistry, KTH
Supervisors: Victoria Barclay, PhD and Jennie Östervall, PhD, Clinical Pharmacology
Department, Karolinska University Hospital
Abstract
Infections in critically ill patients are a problem for the healthcare system and at any one time, 70 % of all intensive care unit (ICU) patients are treated with antibiotics. Antibiotics bind to proteins in the blood, but only unbound drug can diffuse over capillary membranes and bind to the targeted receptor. Standard protein binding percentages for antibiotics have been developed from studies on healthy volunteers and dosing regimens for patients are adapted accordingly. The determination of the total concentration of antibiotics in patients’ blood samples is, based on the standard percentages, ordinarily representative for the
pharmacological effect of the antibiotic. However, certain conditions that are common in critically ill patients can alter protein binding percentages, resulting in a larger or smaller unbound fraction. This in turn can result in toxicity or therapeutic failure.
The aim of this project was to develop an analytical method for the determination of the unbound concentration of the Beta-Lactam antibiotics cefotaxime, flucloxacillin, cloxacillin and piperacillin in plasma. A method was successfully developed using ultrafiltration for the extraction of unbound analytes and ultra high performance liquid chromatography tandem mass spectrometry, UHPLC-MS/MS, for their quantification. The method was partly
validated according to the European Medicines Agency’s guidelines on bioanalytical method validation.
Keywords: Unbound concentration, free concentration, ultrafiltration, LC-MS/MS,
cefotaxime, flucloxacillin, cloxacillin, piperacillin
Sammanfattning
Kritiskt sjuka patienter med infektioner är en börda för sjukvården och 70 % av alla patienter på intensivvårdsavdelningar är ordinerade antibiotika. Antibiotika binder till proteiner i blodet, men enbart den icke-proteinbundna (fria) fraktionen kan diffundera över kapillära membran och binda till receptorer. Standardproteinbindningsgrad för olika antibiotika har utvecklats från studier på friska frivilliga och doseringen av läkemedlen är anpassade därefter.
Den totala koncentrationen av antibiotika i patienters blod är vanligen representativ för den farmakologiska effekten. Dock kan vissa sjukdomar påverka proteinbindningsgraden vilket resulterar i en större eller mindre mängd fria antibiotika i blodcirkulationen. Det här kan i sin tur resultera i toxicitet eller otillräcklig effekt av läkemedlet.
Syftet med det här projektet var att utveckla en analytisk metod för att bestämma den fria koncentrationen av Beta-Lactam antibiotikan cefotaxim, flukloxacillin, kloxacillin och piperacillin i plasma. En metod utvecklades med ultrafiltrering för extraktion av den fria fraktionen och högupplösande vätskekromatografi och tandem masspektrometri, UHPLC- MS/MS, för kvantifiering av analyterna. Metoden validerades delvis enligt den Europeiska Läkemedelsmyndighetens riktlinjer för bioanalytisk metodvalidering.
Nyckelord: Fri fraktion, ultrafiltrering, LC-MS/MS, cefotaxim, flukloxacillin, kloxacillin,
piperacillin
Acknowledgements
I would like to express my gratitude to my supervisors Jennie Östervall and Victoria Barclay for their continuous support and guidance throughout this project. I would also like to say thank you to the entire staff at the Clinical Pharmacology department for their encouraging words.
Lastly, I would like to thank my friends and family, especially Omar, for always believing in
me.
Table of contents
Abstract ... 2
Sammanfattning ... 3
Acknowledgements... 4
Abbreviations ... 7
1. Background ... 8
1.1 Beta-Lactam antibiotics... 8
1.2 Protein binding and unbound concentration ... 9
1.3 In-vitro and in-vivo methods to determine the unbound concentration ... 10
1.4 Previous research ... 10
1.5 Validation according to EMA guidelines ... 11
1.5.1 Calibration curve and quality controls ... 11
1.5.2 Accuracy and precision ... 11
1.5.3 Carry-over ... 11
1.5.4 The matrix effect ... 12
1.5.5 Recovery ... 12
1.6 Sustainability ... 13
1.7 Aim ... 13
2. Experimental ... 14
2.1 Chemicals and materials ... 14
2.2 Instruments ... 14
3. Method ... 15
3.1 Determination of total concentration ... 15
3.2 Stock solutions ... 15
3.3 Calibration range ... 16
3.4 Internal standard ... 18
3.5 Sample preparation ... 19
3.5.1 Sample preparation of calibration curve, QC levels and blank matrix ... 19
3.5.2 Sample preparation of plasma samples ... 19
3.6 Chromatography ... 19
3.5 Mass spectrometry ... 20
3.6.1 Tuning ... 20
3.6.2 Infusion ... 20
3.6.3 Instruments ... 20
3.7 Validation ... 21
3.7.1 Ex-vivo QCs ... 21
3.7.2 Stability during incubation ... 21
3.7.3 Recovery of ultrafiltration ... 21
3.7.4 Volume for ultrafiltration ... 22
3.7.5 Matrix effect ... 22
3.7.6 Calibration curve, accuracy and precision ... 22
3.7.7 Carry-over ... 22
3.7.8 Ultrafiltrate QCs ... 22
4. Results and discussion ... 23
4.1 Calibration range ... 23
4.2 Internal standard ... 23
4.3 The chromatographic system ... 24
4.4 Mass spectrometry ... 24
4.5 Method validation ... 25
4.5.1 Stability during incubation ... 25
4.5.2 Recovery of ultrafiltration ... 26
4.5.3 Volume of ultrafiltration ... 27
4.5.4 Matrix effect ... 28
4.5.5 Calibration curve... 29
4.5.6 Accuracy and precision ... 30
4.5.7 Carry-over ... 34
4.5.8 Ultrafiltrate QCs ... 34
4.5.9 Ex-vivo QCs ... 37
5. Conclusions and future research ... 38
References ... 39
Appendix A – post column infusion experiments ... 41
Abbreviations
CV Coefficient of variance
EMA European Medicines Agency
FCTot A method for determining the total
concentration of flucloxacillin and cloxacillin in plasma.
ICU Intensive care unit
IS Internal standard
LC-MS/MS Liquid chromatography tandem mass
spectrometry
LLOQ Lower limit of quantification
MeOH Methanol
MQ MilliQ water
PCTot A method for determining the total
concentration of piperacillin and cefotaxime in plasma.
QC Quality control
QCH High quality control
QCL Low quality control
QCM Medium quality control
UF Ultrafiltrate
UHPLC Ultra high performance liquid
chromatography
ULOQ Upper limit of quantification
1. Background
1.1 Beta-Lactam antibiotics
Penicillin was discovered more than 90 years ago by Alexander Fleming and was the first antibiotic to be used in clinical medicine [1] [2]. Penicillin was, through X-ray
crystallography, characterized as a Beta-Lactam antibiotic i.e. an antibiotic which contains a four membered Beta-Lactam ring in its molecular structure. Another member of the Beta- Lactam family is cephalosporins, whose structure resembles penicillin, see Figure 1 [3].
Figure 1 - Core structure of penicillins (1) and cephalosporins (2). The Beta-Lactam ring is marked in red.
Beta-Lactams are the most commonly prescribed category of antibiotics and commonly used for treatment of critically ill patients in intensive care [4]. Studies have established that this family of antibiotics is time-dependent, meaning that the bacterial killing is determined by the time during which the unbound antibiotic concentration is maintained above the minimum inhibitory concentration (MIC) [5]. The consequences of insufficient antibiotic use i.e. an unbound concentration below MIC can result in antibiotic resistance and in worst case even death due to therapeutic failure [4]. Flucloxacillin, cloxacillin and piperacillin are categorized as penicillins, see Figure 3, Figure 4 and Figure 5. Cefotaxime is categorized as a
cephalosporin, see Figure 2 [2].
Figure 2 - Molecular structure of cefotaxime Figure 3 - Molecular structure of flucloxacillin
Figure 4 - Molecular structure of cloxacillin Figure 5 - Molecular structure of piperacillin
1.2 Protein binding and unbound concentration
Many drugs exist mainly bound to plasma proteins at therapeutic concentrations. Albumin, the most abundant protein in human plasma, is an important binding protein for many antibiotics. The free drug hypothesis states that it is the unbound drug that is
pharmacologically active. Only nonionized unbound drugs can permeate the capillary membrane and bind to its targeted receptor. The ability for an antibiotic to permeate membranes is crucial since the majority of bacterial infections occur in interstitial fluid of tissues, i.e. the fluid surrounding the cell. This means that small changes in protein binding, see Equation 2, can have a large effect on the unbound concentration, especially for
antibiotics with protein binding percentages > 80 %, such as flucloxacillin and cloxacillin [1]
[6] [7] [8]. These pharmacokinetic and pharmacodynamic changes that affect protein binding can be a result of certain illnesses, such as hypoproteinemia, liver disorders and uremia; all common in critically ill patients. Many studies have shown that usage of standard doses of Beta-Lactam antibiotics can result in drug toxicity as well as therapeutic failure in patients with changed protein binding [7] [9]. Antibiotic dosing regimens and standard protein binding data are developed from studies on healthy volunteers and do not regard these potential pharmacokinetic differences. The determination of the total concentration of antibiotics i.e.
both protein-bound and unbound concentration in plasma, is normally representative of the pharmacological effect by using the standard protein binding percentages. However, only the unbound concentration is representative in patients with altered protein binding. Therefore, a better understanding of protein binding of Beta-Lactams in critically ill patients is necessary in order to achieve an optimal dose adjustment [9] [5]. The unbound fraction of a drug is calculated according to Equation 1. The protein binding percentage of a drug is calculated according to Equation 2.
𝑈𝑛𝑏𝑜𝑢𝑛𝑑 𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛 = 𝑈𝑛𝑏𝑜𝑢𝑛𝑑 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑇𝑜𝑡𝑎𝑙 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛
Eq. 1
𝑃𝑟𝑜𝑡𝑒𝑖𝑛 𝑏𝑖𝑛𝑑𝑖𝑛𝑔 [%] = (1 − 𝑈𝑛𝑏𝑜𝑢𝑛𝑑 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛
𝑇𝑜𝑡𝑎𝑙 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 ) ∗ 100 Eq. 2
1.3 In-vitro and in-vivo methods to determine the unbound concentration
There are different methods used to extract an unbound drug. Equilibrium dialysis and ultrafiltration are the most widely used in-vitro methods for determining the unbound
concentration of drugs [7]. In equilibrium dialysis, a plasma sample is placed on one side of a semipermeable membrane and a buffer is placed on the other. The membrane only allows smaller unbound molecules to permeate. An equilibrium is normally reached after 4-24 hours [10]. Equilibrium dialysis has often been considered the golden standard method, as precise measurements are obtained. However, the method is labor intensive and time-consuming [8].
Ultrafiltration uses centrifugal forces to separate an ultrafiltrate with unbound drug from plasma with the protein-bound drug by the use of a filter. It is a fast and easy method. The main drawback is potential adsorption of drug to the filter. This can be compensated for by measuring a recovery ratio of the filter [8].
The in-vivo method, micro dialysis, consists of a probe that is inserted in the interstitial fluid of different tissues. A dialysate solution flows through the probe’s membrane which is in contact with the intestinal fluid. Unbound drug in the fluid diffuses across the membrane to the dialysate solution, while larger molecules such as proteins do not permeate the membrane.
This in-vivo method can also be modified to an in-vitro method by placing the micro dialysis probe in a plasma sample [8].
1.4 Previous research
Previous methods by S.E. Briscoe et al. [6] and B.C. McWhinney et al. [4] have been validated for the determination of unbound concentration for many different Beta Lactam antibiotics; among them piperacillin and flucloxacillin. One validated method covered the calibration range 0.1-50 µg/mL for both analytes, and the other in the calibration range 10- 1000 µg/mL for piperacillin and 5-500 µg/mL for flucloxacillin. In both methods, HPLC was used with UV detection and a reverse phase C18 column. The analytes were divided in different isocratic methods with different mobile phases [6] [4]. Both methods extracted unbound analytes by ultra-filtration. In one of the methods, the quality controls (QCs) were prepared in plasma and stored at -70 °C up to 12 months. The calibration curve was prepared in deionized water and stored at -70 °C up to 12 months [6].
Another article by K.H.M. Larmené-Beld et al. [11] presented a validated UPLC-MS/MS method for determination of flucloxacillin and cloxacillin in micro dialysis samples prepared in Ringer’s solution i.e. a saline solution. The method used a C18 column. The calibration range was 0.05-5 µg/mL for flucloxacillin and 0.1-5 µg/mL for cloxacillin in micro dialysate.
The micro dialysate samples were stable at -20 °C for 7 days, and stable after three freeze- thaw cycles.
Many previous studies have investigated the unbound concentration of piperacillin,
cloxacillin and flucloxacillin as well as other analytes by using ultrafiltration. In these studies,
a plasma volume of 200 µl was used [6] [12] [13] [4] for ultrafiltration. This also concurs
with the current and accredited sample preparation for unbound anticonvulsant used in the
Clinical Pharmacology department at Karolinska University Hospital. Different ultrafiltration
filters were used in the methods, among them Amicon Ultra-0.5 mL 30,000 molecular weight cut off centrifugal filter device [6].
A previous study performed at the Clinical Pharmacology department concluded that the optimal method to obtain a physiological pH of 7.4±0.1 in plasma is incubation at 11 % CO
2at 37 °C for 30 minutes [14]. The same study also concluded that there was no significant difference in the obtained unbound analyte concentration when ultrafiltration was performed at 22 °C and 37 °C [14].
1.5 Validation according to EMA guidelines
The Clinical Pharmacology department validates all LC-MS/MS methods according to the European Medicines Agency’s guideline on bioanalytical method validation [15].
1.5.1 Calibration curve and quality controls
A minimum of six calibration levels should be used within the calibration range and at least four quality controls (QCs) should be prepared [15]. The calibration curve and the QCs should be prepared from different stock solutions. The concentration levels should be prepared according to Table 1 [16] [15].
Table 1 – A description of QCs
Control level Description
LLOQ Lower limit of quantification is the same
concentration as the lowest calibration level.
QCL A low QC that is three times the
concentration of LLOQ.
QCM A medium QC that is 30-50 % of the
calibration range.
QCH A high QC that has the concentration of 75
% of the highest concentration in the calibration range.
1.5.2 Accuracy and precision
At least 5 replicates of each QC level should be analyzed at three occasions with a calibration curve. A precision, given as CV, of 15 % and an accuracy of ±15 % is allowed for the QC levels and calibration levels except for LLOQ/S1 (lowest calibration level) for which an accuracy of ±20 % and a precision of 15 % is allowed. This is an acceptance criterion for both inter-day and intra-day measurements. The validation of the calibration curve with
corresponding QC levels is accepted if these criteria are achieved. The three analyses need to be conducted using at least two instruments and at the hands of two laboratory staff.
1.5.3 Carry-over
A carry-over experiment is conducted by placing a blank matrix without internal standard
after the highest calibration level in the autoinjector. If the obtained concentration is less than
20 % of the concentration of LLOQ, the carry-over is regarded insignificant.
1.5.4 The matrix effect
In biological samples, there are many components, such as phospholipids, salts and proteins, that can interfere with the ionization in mass spectrometry analysis. This is called the matrix effect and can be considered by comparing the detection area of post-ultra filtrated spiked matrix samples to the detection area of spiked neat samples, according to the post-extraction spike method [13] [17]. Each matrix, such as serum, EDTA plasma and citrate plasma, contains different amounts of interfering components, which can also differ between patients.
Therefore, the matrix effect must be established for each matrix used and also regard
differences between plasma samples from different patients. An internal standard normalized matrix factor should be calculated in order to correct potential differences resulting from ionization and variations in the method, for instance variations in injection volume. The internal standard was added to the sample post extraction. Matrix samples from at least six different patients need to be investigated for each matrix. The matrix effect is calculated according to Equation 3. A precision, given as CV, of 15 % is allowed.
% 𝑀𝑎𝑡𝑟𝑖𝑥 𝑒𝑓𝑓𝑒𝑐𝑡
= (
𝐴𝑛𝑎𝑙𝑦𝑡𝑒 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒
𝑝𝑜𝑠𝑡−𝑢𝑙𝑡𝑟𝑎 𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑒𝑑 𝑠𝑝𝑖𝑘𝑒𝑑 𝑚𝑎𝑡𝑟𝑖𝑥 𝑠𝑎𝑚𝑝𝑙𝑒𝐼𝑆 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒
𝑝𝑜𝑠𝑡−𝑢𝑙𝑡𝑟𝑎 𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑒𝑑 𝑠𝑝𝑖𝑘𝑒𝑑 𝑚𝑎𝑡𝑟𝑖𝑥 𝑠𝑎𝑚𝑝𝑙𝑒𝐴𝑛𝑎𝑙𝑦𝑡𝑒 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒
𝑠𝑝𝑖𝑘𝑒𝑑 𝑛𝑒𝑎𝑡 𝑠𝑎𝑚𝑝𝑙𝑒𝐼𝑆 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒
𝑠𝑝𝑖𝑘𝑒𝑑 𝑛𝑒𝑎𝑡 𝑠𝑎𝑚𝑝𝑙𝑒)
∗ 100
Eq. 3
1.5.5 Recovery
The recovery ratio of the analyte is investigated in order to ensure that the filter used in ultrafiltration is suitable for the sample preparation. The analyte recovery ratio is calculated by comparing the peak area of the ultra-filtrated samples to the peak area of the same solution that has not undergone ultrafiltration. This establishes any possible interaction between the analyte and the filter [13] [17]. Internal standard is added post extraction. The recovery ratio is calculated according to Equation 4. A precision, given as CV, of 15 % is allowed.
% 𝑅𝐸 = (
𝐴𝑛𝑎𝑙𝑦𝑡𝑒 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒
𝑒𝑥𝑡𝑟𝑎𝑐𝑡𝑒𝑑 𝑠𝑎𝑚𝑝𝑙𝑒𝐼𝑆 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒
𝑒𝑥𝑡𝑟𝑎𝑐𝑡𝑒𝑑 𝑠𝑎𝑚𝑝𝑙𝑒𝐴𝑛𝑎𝑙𝑦𝑡𝑒 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒
𝑢𝑛𝑒𝑥𝑡𝑟𝑎𝑐𝑡𝑒𝑑 𝑠𝑎𝑚𝑝𝑙𝑒𝐼𝑆 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒
𝑢𝑛𝑒𝑥𝑡𝑟𝑎𝑐𝑡𝑒𝑑 𝑠𝑎𝑚𝑝𝑙𝑒)
∗ 100
Eq. 4
1.6 Sustainability
To minimize contagion and pharmaceutical contamination of water supply, all biological waste, drug waste and containers in contact with said waste were sorted and disposed of by burning. Glass vials used in analysis were disposed of likewise. Larger batches of samples were prepared in 96 deep-well plates in order to minimize the number of disposable products.
All solvents used, such as methanol and acetonitrile were disposed of by evaporation in a fume hood in order to minimize environmental impact. The only hazardous chemical used was formic acid, which was very diluted in the mobile phase solution (0.1 %). All standard and QC samples were prepared in Milli-Q water, which is considered a Class 3 solvent, i.e. a preferred solvent according to the 12 principles of green chemistry. Acetonitrile, which was used as mobile phase, is considered a Class 2 solvent; less environmental-friendly but still a usable option [18]. Due to acetonitrile’s flammable, toxic and volatile properties, it was stored and handled according to necessary safety precautions. UHPLC-MS/MS requires less solvent use than other similar analyses, such as LC-MS/MS, and therefore generates less solvent waste.
The consequences of insufficient antibiotic use i.e. an unbound concentration below MIC can result in antibiotic resistance, which is an increasing threat to global health [4] [19]. Dose regimens for critically ill patients may be altered as a result of the determination of unbound concentration of antibiotics. Therefore, the method developed in this project may lead to a more efficient antibiotic use which is an important step to prevent and control the spread of antibiotic resistance [19].
1.7 Aim
The aim of this project was to develop and validate an analytical method for the determination
of unbound concentration of cefotaxime, flucloxacillin, cloxacillin and piperacillin in plasma
using ultrafiltration for the extraction of unbound analytes, and UHPLC-MS/MS for their
quantification.
2. Experimental
2.1 Chemicals and materials
Cefotaxime sodium (purity 0.97), piperacillin-d5 (purity 0.95), flucloxacillin sodium (purity 0.97) and cloxacillin-13C4 sodium salt (purity 0.95) were purchased from Toronto Research Chemicals (North York, Canada). Piperacillin sodium (purity 0.964) and cloxacillin sodium salt monohydrate (purity 0.987) were purchased from Sigma Aldrich (St Louise, MA, USA).
Cefotaxime-13C-d3 (purity 0.984) was purchased from Alsachim (Illkirch, France). Milli-Q water was produced by a Q-pod water system of model Millipak Express 40 (MerckMillipore, Darmstadt, Germany). Acetonitrile and formic acid (purity >0.99) used for the preparation of mobile phases were purchased from VWR Chemicals (Radnor, PA, USA).
Filters that were used were Sartorius Vivaspin 500 with a 10k MWCO PES membrane
(Göttingen, Germany), Sartorius Vivascon 500 with a 10k MWCO HY membrane (Göttingen, Germany) and Amicon Ultra 0.5 mL with an Ultracel-30k NMWL filter from MerckMillipore Ltd. (Cork, Ireland). Two Eppendorf vial types were used; Micro Tube 1.5 mL and Micro Tube 1.5 mL Easy Cap, both from Sarstedt AG & Co (Nümbrecht, Germany). Ellerman vials were purchased from Kimble Chase (Meiningen, Germany). 1.2 mL champagne vials were purchased from Scantec Nordic (Jonsered, Sweden) and 1 mL Nunc 96 deep-well plates were purchased from Thermo Fisher Scientific (Waltham, MA, USA)
Citrate plasma was gathered from healthy volunteers. Heparin and EDTA plasma samples were gathered from leftovers from anonymized patient samples.
2.2 Instruments
A VX-2400 multi-tube vortexer from Troemner (New Jersey, NY, USA) was used to vortex the prepared samples. A Hettich Mikro 200R (Vlotho, Germany) was used for centrifugation of the samples. An INCOmed246 Memmert (Buechenbach, Germany) CO
2incubator was used for regulation of pH and heat in samples.
A liquid chromatography tandem mass spectrometry (LC-MS/MS) system was used for the analysis of the samples. The instruments consisted of a Dionex Ultimate 3000 RS LC system coupled to a TSQ Quantis triple quadrupole mass spectrometer with an electrospray ionization ion source. One of the instruments had a TriPlus RSI autosampler coupled to the LC system.
All parts of the LC-MS/MS system were purchased from Thermo Fisher Scientific (Waltham,
MA, USA). The software used was TraceFinder, also from Thermo Fisher Scientific. The LC
system used a Kinetex 2.6 µm RP Biphenyl column (100 Å pore size, 50x2.1 mm) from
Phenomenex (Værløse, Denmark).
3. Method
3.1 Determination of total concentration
For the determination of the total concentration, i.e. both protein-bound and unbound concentration of flucloxacillin and cloxacillin in plasma samples, a method developed at the Clinical Pharmacology Laboratory was used. The concentration was determined using a UHPLC-MS/MS and a Kinetex 2.6 µm Biphenyl column (100 Å pore size, 50x2.1 mm). 0.1
% formic acid in Milli-Q was used as mobile phase A and acetonitrile as mobile phase B.
Samples were prepared through protein precipitation using mobile phase B. This method will, in this report, be referred to as FCTot.
For the determination of the total concentration, i.e. both protein-bound and unbound concentration of cefotaxime and piperacillin in plasma samples, a method developed at the Clinical Pharmacology Laboratory was used. The concentration was determined using a UHPLC-MS/MS and a Hypersil Gold RP 1.9 µm C18 column (50x2.1 mm). 0.1 % formic acid in Milli-Q was used as mobile phase A and MeOH as mobile phase B. Samples were prepared through protein precipitation using mobile phase B. This method will, in this report, be referred to as PCTot.
3.2 Stock solutions
Two stock solutions for each analyte were prepared; one for the preparation of calibration curve and one for the preparation of QCs, see Table 2 and Table 3. The stock solutions were kept separately in polypropylene containers at -80 °C.
For preparation of the calibration curve, working solutions with concentration 100 µg/mL, 10 µg/mL and 0.5 µg/mL of each analyte were prepared in Milli-Q water and stored separately in polypropylene containers at -80 °C.
For preparation of QCs, working solutions with concentration 50 µg/mL and 1 µg/mL of each analyte were prepared in Milli-Q water and stored separately in polypropylene containers at - 80 °C.
Table 2 - Preparation of stock solutions used for calibration curve
Molar mass [g/mol]
Molar mass with counterion [g/mol]
Solvent Concentration stock solution [µg/mL]
Cefotaxime sodium salt
455.5 477.5 MQ 1793.9
Piperacillin sodium salt
517.6 539.5 MQ 4802.3
Flucloxacillin sodium salt
453.9 475.9 MQ 1779.0
Cloxacillin sodium salt monohydrate
435.9 475.9 MQ 1795.2
Table 3 - Preparation of stock solutions used for QCs.
Molar mass [g/mol]
Molar mass with counterion [g/mol]
Solvent Concentration stock solution [µg/mL]
Cefotaxime sodium salt
455.5 477.5 MQ 1661.7
Piperacillin sodium salt
517.6 539.5 MQ 4395.5
Flucloxacillin sodium salt
453.9 475.9 MQ 1686.5
Cloxacillin sodium salt monohydrate
435.9 475.9 MQ 1525.2
3.3 Calibration range
Before carrying out the validation of the analytical method, The European Medicines Agency recommends that an expected range of concentration should be known [15]. The total
concentration ranges for piperacillin, cefotaxime, cloxacillin and flucloxacillin are known from the current accredited methods. The expected calibration range for the unbound analytes should be based on the calibration range of the total concentration and adapted to the standard protein binding percentage of the analytes in healthy patients. However, the LLOQ might be limited by the purity of the internal standard [20].
The range of the calibration curve was developed from the calibration range from the FCTot
and PCTot methods and altered according to standard protein binding percentages for each
analyte, see Table 4. The calibration levels and QCs were prepared in Milli-Q according to
Table 5 and Table 6. After preparation, they were kept in Eppendorf vials at -80 °C. Newly
thawed calibration samples and QCs were used for each analysis, after which they were
disposed of. Other calibration ranges were also investigated during the method development.
Table 4 - The calibration range for the LC-MS/MS determination of total concentration of analytes and their respective standard protein binding percentage.
Analyte Calibration range total concentration in FCTot and PCTot [µg/mL]
Standard protein binding percentage [21] [22] [23]
[24]
Expected calibration range unbound [µg/mL]
Piperacillin 0.1-150 30 % 0.07-105
Cefotaxime 0.1-50 25-40 % 0.06-37.5
Cloxacillin 0.1-100 92 % 0.008-8
Flucloxacillin 0.1-100 94-95 % 0.005-6
Table 5 – The concentration of the prepared calibration levels (S-Standard). All concentrations are given in µg/mL.
Cefotaxime Piperacillin Flucloxacillin Cloxacillin
S1 0.01 0.01 0.01 0.01
S2 0.25 0.25 0.25 0.25
S3 0.5 1.5 0.5 0.5
S4 1 3 1 1
S5 2.5 7.5 2.5 2.5
S6 5 15 5 5
S7 10 30 10 10
S8 15 45 15 15
S9 20 60
Table 6 - The concentration of the prepared QCs. All concentrations are given in µg/mL.
Cefotaxime Piperacillin Flucloxacillin Cloxacillin
LLOQ 0.01 0.01 0.01 0.01
QCL 0.03 0.03 0.03 0.03
QCM 11 25 7.5 7.5
QCH 16 45 11 11
3.4 Internal standard
The same stable isotope labeled (SIL) internal standards used in this method are used in the FCTot and PCTot methods. An SIL internal standard was available for flucloxacillin.
However, its mass spectrum overlaps with the mass spectrum of flucloxacillin, resulting in incorrect detection. Therefore, the internal standard for cloxacillin is also used for the
quantification of flucloxacillin. Stock solutions of each internal standard were prepared with a concentration of 1 mg/mL. The internal standard solution was prepared in Milli-Q water with concentrations stated in Table 7. 50 µL of this internal standard was added to each sample.
The solution of internal standard was kept in a polypropylene container at -80 °C and thawed before each analysis. Other concentrations of internal standard were also investigated during the method development.
Table 7 - The selected internal standards and their concentration in solution.
Analyte Internal standard Concentration in IS
solution [µg/mL]
Cefotaxime Cefotaxim-13Cd3 0.67
Cloxacillin Cloxacillin-13C4 0.27
Flucloxacillin Cloxacillin-13C4 0.27
Piperacillin Piperacillin-d5 1.33
3.5 Sample preparation
3.5.1 Sample preparation of calibration curve, QC levels and blank matrix
50 µL of each calibration level, QC and blank matrix were transferred to champagne vials upon which 50 µL of internal standard was added to each sample. A lid was placed on the vials and the vials were thereafter vortexed for 10 seconds. The samples were then placed in the autoinjector. Larger sets of samples were prepared in 96 deep-well plates.
3.5.2 Sample preparation of plasma samples
200 µL of plasma sample was added to an Amicon Ultra filter. Protein binding is dependent on temperature and pH. Therefore, the plasma samples were stored in an incubator for 30 minutes at 37 °C and at 11 % CO
2before ultrafiltration to obtain the physiological pH of 7.4 and temperature of 37 °C [8] [14]. The samples were ultra-filtrated by centrifugation for 10 minutes at 10900 RPM. 50 µL of the ultrafiltrate from plasma samples was added to
champagne vials. 50 µL of internal standard was added to each sample. The vials were capped, and thereafter vortexed for 10 seconds. The samples were then placed in the
autoinjector with the calibration curve, QCs and blank matrix samples. Larger sets of samples were prepared in 96 deep-well plates.
3.6 Chromatography
The chromatographic separation of cefotaxime, flucloxacillin, cloxacillin and piperacillin was obtained on a reverse phase Kinetex 2.6 µm Biphenyl column (100 Å pore size, 50 x 2.1 mm) heated to 30 °C. The mobile phases consisted of 0.1 % formic acid in Milli-Q water as mobile phase A, and acetonitrile as mobile phase B. The gradient was programmed according to Table 8 with a flowrate of 0.4 ml/min, and a total analysis time of 3.5 minutes per sample.
Table 8 - The mobile phase gradient used in the method.
Time [min] % A % B
0 15 85
0.70 37 63
3.00 37 63
3.01 99 1
3.48 99 1
3.49 15 85
3.5 Mass spectrometry 3.6.1 Tuning
The eluate was transferred to a Thermo Scientific TSQ Quantis triple quadrupole mass spectrometer with an electrospray ionization ion source. The optimal tuning parameters for each analyte and their internal standard are displayed in Table 9.
Table 9 - Transitions for the analytes and their internal standards.
Analyte Precursor ion [m/z] Product ion [m/z] Collision energy [V]
Cefotaxime 456.09 396.07 9.38
Cefotaxime-13Cd3 460.09 400.05 9.25
Flucloxacillin 454.18 160.13 12.37
Cloxacillin 436.09 277.00 12.62
Cloxacillin-13C4 440.14 281.07 12.67
Piperacillin 518.18 143.13 19.79
Piperacillin-d5 523.21 148.13 20.039
3.6.2 Infusion
In order to estimate the qualitative matrix effect, an infusion trial was conducted. Analyte was injected post-column with eluate carrying blank matrix i.e. ultrafiltrate from citrate, heparin and EDTA plasma. The mixture was injected into the mass spectrometer. The same
experiment was thereafter completed but without analyte solution. The obtained matrix effect profiles were then compared in order to establish any potential ion suppression or
enhancement as well as estimate the qualitative matrix effect.
3.6.3 Instruments
Standard practice requires validation of methods on two equal LC-MS/MS instruments to ensure a backup procedure [16]. The two instruments used for validation of this method were both Thermo Scientific LC systems coupled to a TSQ Quantis triple quadrupole mass
spectrometer with electrospray ionization ion source. One of the instruments used a two-
channel injection system with two identical columns, and the other instrument operated with a
single-channel system. Due to the difference in injector type, different cleaning procedures of
the injection needle were operated.
3.7 Validation 3.7.1 Ex-vivo QCs
2 mL ex-vivo QCs for each analyte was prepared in citrate plasma with concentrations according to Table 10. The concentration was chosen based on standard protein binding for each analyte. The ex-vivo QCs were prepared twice in six replicates per QC level according to 3.3.2 to determine the unbound concentration of each analyte.
Table 10 - Concentration of analyte in the ex-vivo QCs.
Analyte Concentration [µg/mL]
Cefotaxime 25
Cloxacillin 76
Flucloxacillin 71
Piperacillin 44
3.7.2 Stability during incubation
Protein binding is dependent on temperature and pH. Therefore, the plasma samples were stored in an incubator for 30 minutes before ultrafiltration to obtain the physiological pH of 7.4 and temperature of 37 °C [8]. In order to ensure the stability of the antibiotic analytes at this temperature, a stability test was performed. A determination of the total concentration of piperacillin, cloxacillin, flucloxacillin and cefotaxime was completed, analyzing spiked plasma samples that were thawed and thereafter kept at 37 °C for 30 minutes, 1 hour, 2 hours, 8 hours and 24 hours and compared to plasma samples that were thawed and immediately analyzed. The current and accredited sample preparations and LC-MS/MS methods, FCTot and PCTot, for the investigated analytes were used in this experiment.
3.7.3 Recovery of ultrafiltration
Three filters were compared in terms of recovery ratio in order to establish any interactions between analyte and filter. The filters were chosen based on availability in the laboratory (Sartorius Vivacon and Sartorius Vivaspin) and on previous research (Amicon Ultra) [6] [25]
[9].
Four Milli-Q water solutions with concentration 5 µg/mL of each analyte were prepared. The
analytes were kept in separate samples in order to avoid any competing interactions between
different analytes and filter. 500 µL of each solution was placed in triplicate in each filter and
centrifuged for 10 minutes at 10900 RPM. The analyte recovery ratio was then calculated by
comparing the analyte area of the ultra-filtrated samples to the analyte area of the same
solution that had not undergone ultrafiltration. Internal standard was added to each sample
post extraction and an IS compensated recovery ratio was calculated for each filter.
3.7.4 Volume for ultrafiltration
The ultrafiltrate from different volumes of sample added to the Amicon Ultra filter before centrifugation were investigated; 500 µL, 200 µL, 150 µL and 100 µL. The samples were ex- vivo QCs prepared in citrate plasma. The ultrafiltrate from a 500 µL sample was regarded as reference, as this volume is recommended by the manufacturer. Each analyte was analyzed separately in order to disregard any competing interactions between different analytes and filter. The samples were prepared in three replicates. The filters were placed in incubation at 11 % CO
2at 37 °C for 30 minutes, and thereafter centrifuged 10 minutes at 10900 RPM. An IS compensated ratio between analyte area for each volume and the analyte area for the reference was calculated.
3.7.5 Matrix effect
Ultrafiltrate from citrate, heparin and EDTA plasma was obtained by ultra-filtrating 500 µL of plasma acquired from six different patients per plasma matrix. Each ultrafiltrate as well as a Milli-Q water solution was spiked to the same concentration as QCL and QCH. An IS compensated matrix effect was calculated. Milli-Q water solution was used as neat solution as it was used for preparation of QCs and calibration curve.
3.7.6 Calibration curve, accuracy and precision
After the calibration curve and IS concentration were confirmed, three measurements of accuracy and precision were performed. One calibration curve and six replicates of each QC were included per analysis batch. The three analyses were divided over two days, two instruments and with two preparations of mobile phases and at the hands of two laboratory staff.
3.7.7 Carry-over
In each analysis, one blank Milli-Q sample and one blank ultrafiltrate from citrate plasma with and without internal standard were included after the highest calibration concentration level in order to determine any carry-over.
3.7.8 Ultrafiltrate QCs
3.7.8.1 LLOQ and QCM in ultrafiltrate
5 mL of LLOQ and QCM were prepared in ultrafiltrate from citrate plasma. Six replicates of each QC were analyzed three times.
3.7.8.2 Stability of ultrafiltrate QCL
An ultrafiltrate solution from citrate plasma and a Milli-Q water solution were spiked to the
same concentration level as QCL for all analytes. The solutions were prepared and thereafter
analyzed in six replicates per matrix. Thereafter, the solutions were frozen in -80 °C in three
different containers; one in an Eppendorf vial, one in an Eppendorf vial Easy Cap (EC) and
one in an Ellerman vial. After five days, six replicates of each matrix and container were
analyzed once again. This experiment was conducted in order to investigate the stability of
the analytes after freezing and thawing as well as to investigate the effect of the container in
which the analytes are kept.
4. Results and discussion
4.1 Calibration range
The calibration range was chosen based on a combination of factors. A lower concentration of 0.005 µg/mL for each analyte was tested as LLOQ. However, small variations in internal standard area resulted in large deviations in analyte concentration. In addition to this, the carry-over from the highest calibration level was significant. The low purity of cloxacillin’s internal standard resulted in large deviations in concentration for LLOQ for cloxacillin and flucloxacillin.
A wider calibration range of 0.005-30 µg/mL for cefotaxime and 0.005-100 µg/mL for
piperacillin resulted in a large curvature. This caused both QCs at low concentrations (LLOQ) and high concentration (QCH) to deviate. Therefore, the final range of 0.01-15 µg/mL for cloxacillin and flucloxacillin, 0.01-60 µg/mL for piperacillin and 0.01-20 µg/mL for
cefotaxime was selected for further validation. The reduction of the calibration in the higher range means that a dilution method for samples with concentrations higher than ULOQ needs to be developed.
4.2 Internal standard
During the method development, piperacillin’s internal standard was used for quantification of cloxacillin and flucloxacillin, as the low purity of cloxacillin’s internal standard caused a contamination, which was problematic at lower concentrations. However, at higher
concentrations of piperacillin, a saturation of the ion source was reached. This was not an issue for determining the concentration of piperacillin in the samples, as the saturation affected piperacillin and piperacillin’s internal standard equally. The same saturation of the ion source was not reached for flucloxacillin and cloxacillin, which resulted in incorrect measurements at higher concentrations when using piperacillin’s internal standard. Therefore, cloxacillin’s internal standard was used for the calibration of cloxacillin and flucloxacillin, despite its low purity.
Both higher and lower concentrations of internal standard were investigated. At higher
concentrations, the low purity of cloxacillin’s internal standard resulted in interfering peaks,
which was evident in LLOQ and QCL. At lower concentrations, small variations in signal
from the internal standard resulted in large deviations in the quantification.
4.3 The chromatographic system
Figure 6 - A graphic illustration of the gradient program used for the chromatographic method. Included in the illustration are the peaks for each analyte in the order of cefotaxime, piperacillin, cloxacillin and flucloxacillin.
All chromatographic peaks were fully separated using the chosen mobile phase gradient. The gradient and the peaks with corresponding time of elution are illustrated in Figure 6.
4.4 Mass spectrometry
A post column infusion experiment was run to evaluate qualitative matrix effects, see Table 32 – Table 38 in Appendix A.
Cefotaxime elutes soon after the elution of endogenous substances which could result in ion suppression. However, the internal standard of cefotaxime elutes at a similar time, which compensates for the ion suppression. The difference in qualitative matrix effect between citrate and the other matrices is evident, but since this matrix effect is similar between
analytes and their respective internal standard, this difference should be compensated as well.
Further results from the quantitative matrix effect are presented in 4.5.4.
4.5 Method validation
4.5.1 Stability during incubation
Figure 7 - graph illustrating the stability of analytes in QCs in citrate plasma after incubation.
None of the four analytes were stable after more than 8 hours in incubation, see Figure 7. At 30 minutes, 1 hour and 2 hours, the deviations from the nominal concentrations vary, even with concentration values that exceed the reference concentration values. This can be due to water evaporation since some samples (0 minutes, 30 minutes, 1 hour and 2 hours) were kept in incubation without a lid in order to obtain the pH of 7.4. The samples (8 hours and 24 hours) were capped in order to avoid extensive evaporation.
Due to prioritization of clinical patient samples and a resulting lack of autosampler positions in the instrument, the analysis of the flucloxacillin and cloxacillin samples were delayed four days, during which time the prepared samples were kept at +8 °C. This may have affected the obtained results.
Only one QC per level, analyte and time in incubation was investigated. Therefore, without a complete statistical analysis, it is difficult to conclude how long samples can endure a
temperature of 37 °C without analyte breakdown. Further experiments with more QCs per level and analyte should be completed in order to determine the stability during incubation.
-20%
0%
20%
40%
60%
80%
100%
120%
0 min 30 min 1 hour 2 hours 8 hours 24 hours
Stability during incubation
QCL Cefotaxime QCL Piperacillin QCM Cefotaxime QCM Piperacillin QCH Cefotaxime QCH Piperacillin QCL Cloxacillin QCL Flucloxacillin QCH Cloxacillin QCH Flucloxacillin
4.5.2 Recovery of ultrafiltration
Table 11 - The recovery of cefotaxime, flucloxacillin, cloxacillin and piperacillin in Sartorius Vivacon filter
Analyte IS compensated recovery [%] CV [%]
Cefotaxime 86.5 2.1
Flucloxacillin 75.0 0.36
Cloxacillin 85.3 4.6
Piperacillin 86.8 2.4
Table 12 - The recovery of cefotaxime, flucloxacillin, cloxacillin and piperacillin in Sartorius Vivaspin filter
Analyte IS compensated recovery [%] CV [%]
Cefotaxime 94.0 5.1
Flucloxacillin 60.1 12.0
Cloxacillin 63.6 8.6
Piperacillin 76.9 3.7
Table 13 - The recovery of cefotaxime, flucloxacillin, cloxacillin and piperacillin in Amicon Ultra filter
Analyte IS compensated recovery [%] CV [%]
Cefotaxime 97.9 1.8
Flucloxacillin 82.1 3.8
Cloxacillin 89.6 2.5
Piperacillin 73.4 2.8
Table 14 - The recovery of cefotaxime, flucloxacillin, cloxacillin and piperacillin in Sartorius Vivacon, Sartorius Vivaspin and Amicon Ultra filters
Filter IS compensated
recovery [%]
CV [%]
Sartorius Vivacon 83.4 0.066
Sartorius Vivaspin 73.7 0.12
Amicon Ultra 85.8 0.11
The IS compensated recovery is highest for the Amicon Ultra filter for all analytes except
piperacillin, see Table 11, Table 12, Table 13 and Table 14. This filter was chosen for further
analysis in the validation due to these results and results from previous studies [6]. Further
recovery rate experiments should be conducted for this filter and all analytes at different
concentrations. The adsorption of analyte to the filter, which is ultrafiltration’s main
drawback, might be compensated by correcting the obtained results using the calculated
recovery ratio.
4.5.3 Volume of ultrafiltration
Table 15 - Measured mean response ratio and CV for all four analytes using different volumes of ex vivo-QCs. Included in the table are the deviations from the reference sample volume of 500 µL.
Cefotaxime response ratio
Cloxacillin response ratio
Piperacillin response ratio
Flucloxacillin response ratio 500 µL
Mean 55.7 27.6 46.2 16.9
CV % 2.1% 1.2% 5.2% 6.1%
200 µL
Mean 56.0 28.1 48.1 16.8
CV % 1.9% 5.3% 1.6% 8.2%
Deviation from 500 µL. %
-0.6 % -1.7% -4.3% 0.7%
150 µL
Mean 57.8 23.8 46.1 14.9
CV % 2.8% 1.6% 2.1% 5.2%
Deviation from 500 µL, %
- 3.7% 13.7% 0.2% 12.2%
100 µL
Mean 63.8 32.0 41.9 19.6
CV % 25.8% 27.2% 25.0% 24.7%
Deviation from 500 µL, %
- 14.5% -15.8% 9.3% -15.7%
The results were presented in response ratio instead of concentration since no accepted
calibration curve was prepared with this analysis. The filters that used 200 µL of sample
displayed the lowest overall deviation from the reference sample for all analytes, see Table
15. 100 µL indicated differences between the analytes, as well as large CV values. A sample
preparation was selected where 200 µL of plasma sample was centrifuged for the extraction
of ultrafiltrate. The sample preparation using this volume was used for all plasma samples
during the validation. However, a sample volume of 150 µL is acceptable if there is an
insufficient sample volume.
4.5.4 Matrix effect
Table 16 - Calculated IS compensated matrix effect with citrate, heparin and EDTA plasma compared to Milli-Q water.
Cefotaxi me QCL
Cefotaxi me QCH
Flucloxa cillin QCL
Flucloxac illin QCH
Cloxacil lin QCL
Cloxacil lin QCH
Piperaci llin QCL
Piperaci llin QCH Citrate
plasma UF Mean matrix effect
94.44 99.16 91.76 99.18 96.26 100.86 104.17 101.23
CV % 2.4 3.5 4.4 3.3 4.0 4.2 2.9 4.3
Heparin plasma UF Mean matrix effect
99.09 101.28 98.21 100.23 101.08 103.08 107.25 109.92
CV % 3.9 3.3 5.5 4.3 3.7 2.9 4.3 2.9
EDTA plasma UF Mean matrix effect
99.4 100.71 98.26 100.86 100.89 101.76 107.43 104.53
CV % 2.5 1.7 5.2 3.0 4.7 1.9 5.0 1.8
The obtained results from the matrix effect experiment are presented in Table 16. The matrix effect should be close to 100 % in order to best correspond to the neat matrix, Milli-Q water.
The obtained intra- and inter-matrix CV should not exceed 15 %. This acceptance criterion was achieved for ultrafiltrate from citrate, heparin and EDTA plasma. Therefore, all three plasma matrices can be used in this method. In the current accredited methods for analysis of total concentration of cefotaxime, flucloxacillin, cloxacillin and piperacillin, patient samples are accepted in heparin plasma. The same plasma samples can therefore be used both for analysis of total concentration and unbound concentration. Citrate plasma, which there is an abundance of in the laboratory, can be used for preparation of ex-vivo QC samples. Since EDTA plasma is also an accepted matrix, patient samples of this matrix can be analyzed as well.
A matrix effect experiment was also conducted with 9 mg/mL NaCl in Milli-Q used as neat
solution since it is isotonic with body fluids [26]. The results obtained from the matrix effect
experiment were acceptable for all matrices and all analytes except for piperacillin. The NaCl
solution was not an optimal matrix for piperacillin. Therefore, 9 mg/mL NaCl in Milli-Q
water was rejected as a neat solution for the preparation of the calibration curve and QCs.
4.5.5 Calibration curve
Table 17 - Summary of the inter-day run accuracy and precision for the calibration curve for cefotaxime.
Cefotaxime S1 S2 S3 S4 S5 S6 S7 S8 S9
Nominal concentration 0.01 0.1 0.5 1 2.5 5 10 15 20
Mean concentration [µg/mL] 0.011 0.094 0.494 1.001 2.58 4.88 10.2 14.8 20.1
CV % 4.5 3.3 2.6 1.8 1.1 0.8 1.3 1.5 0.8
Deviation from nominal concentration % 5.5 −5.7 −1.2 0.1 3.0 −2.3 1.6 −1.4 0.4
Table 18 - Summary of the inter-day run accuracy and precision for the calibration curve for flucloxacillin.
Flucloxacillin S1 S2 S3 S4 S5 S6 S7 S8
Nominal concentration 0.01 0.1 0.5 1 2.5 5 10 15
Mean concentration [µg/mL] 0.012 0.098 0.455 0.919 2.48 5.17 10.2 14.8
CV % 0.8 2.3 0.9 1.5 1.8 1.9 1.8 0.8
Deviation from nominal concentration % 16.2 −2.0 −9.0 −8.1 −0.9 3.3 1.7 −1.4
Table 19 - Summary of the inter-day run accuracy and precision for the calibration curve for cloxacillin.
Cloxacillin S1 S2 S3 S4 S5 S6 S7 S8
Nominal concentration 0.01 0.1 0.5 1 2.5 5 10 15
Mean concentration [µg/mL] 0.011 0.091 0.475 0.967 2.55 5.10 9.84 15.1
CV % 4.5 3.6 2.0 0.6 1.4 2.5 2.1 0.8
Deviation from nominal concentration % 13.7 −8.5 −5.0 −3.3 2.0 2.0 −1.6 0.5
Table 20- Summary of the inter-day run accuracy and precision for the calibration curve for piperacillin.
Piperacillin S1 S2 S3 S4 S5 S6 S7 S8 S9
Nominal concentration 0.01 0.1 1.5 3 7.5 15 30 45 60
Mean concentration [µg/mL] 0.010 0.100 1.47 2.91 7.66 14.8 30.5 44.5 60.1
CV % 5.8 3.4 1.3 1.8 2.4 1.7 1.2 0.8 0.2
Deviation from nominal concentration % 3.5 −0.3 −1.9 −2.9 2.1 −1.5 1.8 −1.1 0.2 All calibration levels passed the acceptance criteria, i.e. the deviation of <15 % from the
nominal value for S2-S9 and a maximum deviation of 20 % from the nominal value for S1, as well as a CV <15 %, see Table 17, Table 18, Table 19 and Table 20 All calibration curves were fitted quadratically and weighted 1/X. This resulted in a coefficient of determination, r
2> 0.99 for all runs and all analytes. EMA states that calibration concentration levels can be excluded if > 75 % of all calibration levels and a minimum of six points fulfills the
acceptance criteria. However, all calibration levels fulfilled these criteria, and no
concentration levels were therefore excluded.
4.5.6 Accuracy and precision
Table 21 - Summary of the intra- and inter-day precision for the QCs for cefotaxime.
Cefotaxime LLOQ QCL QCM QCH
Nominal concentration [µg/mL] 0.01 0.03 11 16 Measurement Day 1
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00916 0.0266 10.7 15.3
Intra-day CV % 4.6 3.0 1.5 1.4
Intra-day deviation % -8.5 -11.4 -2.6 -4.4 Measurement 1 Day 2
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.0107 0.0295 11.0 16.0
Intra-day CV % 2.0 1.5 0.6 1.1
Intra-day deviation % 7.0 -1.78 -0.2 -0.12
Measurement 2 Day 2
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00905 0.0272 10.8 15.7
Intra-day CV % 4.7 2.3 3.1 2.2
Intra-day deviation % -9.5 -9.5 -1.8 -2.0
Total measurements
Number of samples 18 18 18 18
Mean concentration [µg/mL] 0.00963 0.0277 10.8 15.7
Inter-day CV % 8.8 5.1 2.2 2.4
Inter-day deviation % -3.7 -7.6 -1.5 -2.2
Table 22 - Summary of the intra- and inter-day precision for the QCs for flucloxacillin.
Flucloxacillin LLOQ QCL QCM QCH
Nominal concentration [µg/mL] 0.01 0.03 7.5 11 Measurement Day 1
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00919 0.0288 8.23 11.8
Intra-day CV % 6.5 5.0 3.7 1.8
Intra-day deviation % -8.1 -4.0 9.8 7.2
Measurement 1 Day 2
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.010 0.0291 7.97 10.8
Intra-day CV % 2.9 2.3 1.8 1.1
Intra-day deviation % 1.5 -3.2 6.2 -1.7
Measurement 2 Day 2
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00913 0.0283 8.14 12.2
Intra-day CV % 7.2 1.8 3.0 1.7
Intra-day deviation % -8.7 -5.8 8.6 11.0
Total measurements
Number of samples 18 18 18 18
Mean concentration [µg/mL] 0.00949 0.0287 8.11 11.6
Inter-day CV % 7.4 3.3 3.1 5.4
Inter-day deviation % -5.1 -4.3 8.2 5.5
Table 23 - Summary of the intra- and inter-day precision for the QCs for cloxacillin.
Cloxacillin LLOQ QCL QCM QCH
Nominal concentration [µg/mL] 0.01 0.03 7.5 11 Measurement Day 1
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00837 0.0263 7.53 11.2
Intra-day CV % 6.1 4.4 3.5 5.5
Intra-day deviation % -16.3 -12.2 0.4 2
Measurement 1 Day 2
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00912 0.0273 7.2 10.9
Intra-day CV % 3.4 0.8 1.6 0.9
Intra-day deviation % -8.8 -9.0 -3.9 -1.3
Measurement 2 Day 2
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00855 0.0270 7.29 11.2
Intra-day CV % 7.1 4.8 3.3 2.0
Intra-day deviation % -14.5 -10.1 -2.8 2.0 Total measurements
Number of samples 18 18 18 18
Mean concentration [µg/mL] 0.00868 0.0269 7.34 11.1
Inter-day CV % 6.6 3.9 3.4 3.6
Inter-day deviation % -13.2 -10.4 -2.1 0.9
Table 24 - Summary of the intra- and inter-day precision for the QCs for piperacillin.
Piperacillin LLOQ QCL QCM QCH
Nominal concentration [µg/mL] 0.01 0.03 25 45 Measurement Day 1
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00888 0.0264 23.6 42.6
Intra-day CV % 4.5 5.9 2.0 1.4
Intra-day deviation % -11.2 -12.1 -5.6 -5.3 Measurement 1 Day 2
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00912 0.0264 24.3 42.2
Intra-day CV % 3.2 1.5 0.9 0.8
Intra-day deviation % -8.8 -11.9 -2.7 -6.3 Measurement 2 Day 2
Number of samples 6 6 6 6
Mean concentration [µg/mL] 0.00910 0.0265 23.6 42.4
Intra-day CV % 3.2 3.7 2.3 3.4
Intra-day deviation % -9.0 -11.8 -5.5 -5.7 Total measurements
Number of samples 18 18 18 18
Mean concentration [µg/mL] 0.00903 0.0264 23.9 42.4
Inter-day CV % 3.7 3.9 2.2 2.1
Inter-day deviation % -9.7 -11.9 -4.6 -5.8
The results from the intra- and inter-day precision and accuracy for LLOQ, QCL, QCM and
QCH for all analytes are presented in Table 21, Table 22, Table 23 and Table 24. All QCs
passed the acceptance criteria of a maximum of intra- and inter-day 15 % deviation from the
nominal value for QCL, QCM and QCH and a maximum of intra- and inter-day 20 %
deviation from the nominal value for LLOQ as well as a CV < 15 %.
4.5.7 Carry-over
The instrument using a two-channel injector displayed no significant carry-over. This is due to its thorough cleaning procedure. Carry-over was, however, observed in the analysis using the instrument with a single-channel injector. For cefotaxime and flucloxacillin no significant analyte carry-over could be detected. For cloxacillin and piperacillin an analyte carry-over of
> 20 % the analyte area of LLOQ was observed after calibration level S7. This carry-over was only observed at the first injection of blank matrix, not the second. This means that at higher concentrations (> 5 µg/mL for cloxacillin and >15 µg/mL for piperacillin) i.e. at
concentrations higher than calibration level S6, at least two injections of blank matrix must be carried out before reinjecting any following sample. Before the method is fully validated and added to the order assortment of the clinical laboratory, the single-channel injector will be replaced by a two-channel injector, eliminating carry-over on both instruments.
4.5.8 Ultrafiltrate QCs
4.5.8.1 Ultrafiltrate LLOQ and QCM
Table 25 - Measured mean concentration and inter-day deviation and CV of the analytes in LLOQ and QCM prepared in ultrafiltrate from citrate plasma.
Cefota xime LLOQ UF
Cefota xime QCM UF
Fluclox acillin LLOQ UF
Fluclox acillin QCM UF
Cloxaci llin LLOQ UF
Cloxac illin QCM UF
Pipera cillin LLOQ UF
Pipera cillin QCM UF Nominal
concentrati on [µg/mL]
0.01 11 0.01 7.5 0.01 7.5 0.01 25
Number of samples
18 18 18 18 18 18 18 18
Mean concentrati on [µg/mL]
0.0093 6
9.60 0.0106 6.86 0.0095 0
6.61 0.0078 7
20.2
Inter-day CV %
7.5 2.1 4.9 6.0 6.4 2.9 8.9 7.8
Deviation
%
-6.4 -12.8 6.4 -8.6 -5.0 -11.8 -21.3 -19.0
The ultrafiltrate from citrate plasma that had been spiked according to LLOQ and QCM was
analyzed three times in six replicates for each analysis. The inter- and intra-day deviations
were < 15 % for all analytes, except piperacillin, that displayed large negative deviations, see
Table 25. Although the deviations were large, the overall precision was high. This could
result from an incorrectly prepared spiking solution or low solubility of piperacillin. To
exclude the former reason, ultrafiltrate was spiked again, see 4.5.8.2.
4.5.8.2 Stability of ultrafiltrate QCL after freezing
Table 26 - Analysis of QCL prepared in ultrafiltrate (UF) and Milli-Q water (MQ) immediately after preparation.
Cefot axime QCL UF
Cefotax ime QCL MQ
Flucloxa cillin QCL UF
Flucloxa cillin QCL MQ
Cloxaci llin QCL UF
Cloxacil lin QCL MQ
Piperac illin QCL UF
Piperaci llin QCL MQ Number of
samples
6 6 6 6 6 6 6 6
Mean concentratio n [µg/mL]
0.0275 0.0273 0.0308 0.0296 0.0259 0.0273 0.0380 0.04 22
CV %
3.8 2.8 13.8 7.3 13.3 4.0 55.4 70.4Deviation from nominal concentratio n %
-8.4 -9.0 2.6 -1.4 -13.7 -9.1 26.5 40.7
In a second experiment, a Milli-Q solution and an ultrafiltrate solution from citrate plasma were prepared according to the same concentration as QCL and immediately analyzed post- preparation without freezing and thawing. The experiment was conducted in order to investigate the stability of the analytes in -80°C and discard any potential errors in
concentration from the preparation. However, the results from analysis of piperacillin showed a low accuracy and precision, likely due to low solubility of piperacillin. Therefore, only the obtained results for cefotaxime, flucloxacillin and cloxacillin from the stability test were compared to the concentrations in Table 26, see Table 27, Table 28 and Table 29. The obtained results for piperacillin in the stability test were compared to the nominal concentration, see Table 30.
Table 27 - Cefotaxime: stability after freezing and thawing after five days at -80°C in Eppendorf, Eppendorf EC and Ellerman containers.