`
`www.elsevier.com/locate/ejpb
`
`Research paper
`
`Drug adsorption to plastic containers and retention of drugs
`in cultured cells under in vitro conditions
`
`Joni J. Palmgre´n a,*, Jukka Mo¨ nkko¨ nen b, Timo Korjamo b, Anssi Hassinen a,
`Seppo Auriola a
`
`a Department of Pharmaceutical Chemistry, University of Kuopio, Kuopio, Finland
`b Department of Pharmaceutics, University of Kuopio, Kuopio, Finland
`
`Received 19 April 2006; accepted in revised form 23 June 2006
`Available online 29 June 2006
`
`Abstract
`
`Loss of drug content during cell culture transport experiment can lead to misinterpretations in permeability analysis. This study anal-
`yses drug adsorption to various plastic containers and drug retention in cultured cells under in vitro conditions. The loss of various drugs
`to polystyrene tubes and well plates was compared to polypropylene and glass tubes both in deionised water and buffer solution. In cel-
`lular uptake experiments, administered drugs were obtained from cultured cells by liquid extraction. Samples were collected at various
`time points and drug concentrations were measured by a new HPLC–MS/MS method. Acidic drugs (hydrochlorothiazide, naproxen,
`probenicid, and indomethacin) showed little if any sorption to all tested materials in either water or buffer. In the case of basic drugs,
`substantial loss to polystyrene tubes and well plates was observed. After 4.5 h, the relative amount remaining in aqueous test solution
`stored in polystyrene tubes was 64.7 ± 6.8%, 38.4 ± 9.1%, 31.9 ± 6.7%, and 23.5 ± 6.1% for metoprolol, medetomidine, propranolol,
`and midazolam, respectively. Interestingly, there was no significant loss of drugs dissolved in buffer to any of the tested materials indi-
`cating that buffer reduced surficial interaction. The effect of drug concentration to sorption was also tested. Results indicated that the
`higher the concentration in the test solution the lower the proportional drug loss, suggesting that the polystyrene contained a limited
`amount of binding sites. Cellular uptake studies showed considerable retention of drugs in cultured cells. The amounts of absorbed drugs
`in cellular structures were 0.45%, 4.88%, 13.15%, 43.80%, 23.57% and 11.22% for atenolol, metoprolol, medetomidine, propranolol,
`midazolam, and diazepam, respectively. Overall, these findings will benefit development and validation of further in vitro drug perme-
`ation experiments.
`Ó 2006 Elsevier B.V. All rights reserved.
`
`Keywords: Drug loss; Plastic instruments; Cultured cells; HPLC–MS/MS; Drug analysis
`
`1. Introduction
`
`Cell culture models like Caco-2 cells are commonly
`used to predict intestinal absorption properties of various
`drugs [1–3]. For transport experiments, cells are typically
`cultured in flasks and seeded on plastic membrane filters,
`where they form a monolayer. Each insert is placed in a
`
`* Corresponding author. Present address: Department of Pharmaceuti-
`cal Chemistry, University of Kuopio, P.O. Box 1627, FIN-70211 Kuopio,
`Finland. Tel.: +358 17 163465; fax: +358 17 162456.
`E-mail address: joni.palmgren@uku.fi (J.J. Palmgre´n).
`
`0939-6411/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved.
`doi:10.1016/j.ejpb.2006.06.005
`
`well of a polystyrene plate in the presence of buffer solu-
`tion. Test compounds are generally added to the apical
`side of the cell monolayer and after some incubation
`time samples from the basolateral side are collected for
`permeability analysis. The loss of drug content during
`experiments, however, can lead to a false assessment of
`permeability. Drug loss may arise from interactions with
`plastic surfaces or from absorption and retention within
`cultured cells [4,5]. Drug loss due to metabolism in
`Caco-2 cells is limited or insignificant, due to low expres-
`sion levels of metabolizing enzymes of the cytochrome
`P450 class [6].
`
`Petition for Inter Partes Review of US 8,648,106
`Amneal Pharmaceuticals LLC – Exhibit 1017 – Page 369
`
`
`
`370
`
`J.J. Palmgre´n et al. / European Journal of Pharmaceutics and Biopharmaceutics 64 (2006) 369–378
`
`In transport experiments, both initial sample and end-
`stage sample from the donor compartment are measured.
`Mass balance is evaluated based on the sum of the amount
`transported and amount remaining in the donor compart-
`ment against initial amount of donor. Reduced mass bal-
`ance is generally observed in transport experiments. For
`example, 80–90% recoveries were reported for nine hetero-
`geneous drugs that encompassed both acids and bases and
`71% recovery was measured for procaine [4]. Recently, it
`has been reported that pH and concentration had an effect
`on the recovery of the acidic drug indomethacin [5]. Results
`showed that mass balance values decreased with decreasing
`pH and concentration. At the lowest pH and concentration
`recovery for indomethacin was only 50%. There was no sig-
`nificant adsorption of indomethacin to the plastic wells.
`Therefore, it was suggested that indomethacin had accu-
`mulated in the cell monolayer, but it was not directly
`shown.
`Most cell culture instruments are made from polystyrene
`plastic, a long carbon chain polymer in which every other
`carbon is covalently bound to a phenyl group. Polystyrene
`is an amorphous, clear, and breakable all-plastic, which is
`used for many applications. The surface of untreated poly-
`styrene is very hydrophobic and disallows the attachment
`of most cells. Thus, a variety of chemical (sulfuric acid)
`and physical (gas plasma, corona discharge, or irradiation)
`administration methods have been utilised to modify poly-
`styrene plastic surfaces [7–11]. These methods modify the
`surface through addition of different chemical groups
`(hydroxyl, ketone, aldehyde, carboxyl, or amine) onto the
`polymer so that the surface becomes hydrophilic and/or
`charged [7–9]. Modified polystyrene (TC) allows for more
`efficient cell attachment and thus growth.
`Sorption of drugs to plastic infusion bags composed of
`polyvinylchloride (PVC) and to plastic intravenous tubing
`is well documented, since drug loss in this manner might
`cause treatment failure. Generally, the sorption of samples
`has been analysed by UV spectrophotometry [12,13] or by
`UV-HPLC methods [14–18]. However, UV-based methods
`can have limitations with sensitivity and selectivity. During
`the last decade, liquid chromatography–tandem mass spec-
`trometry (LC–MS/MS) has shown its usefulness in diverse
`analytical fields. LC–MS/MS analysis is suitable for detect-
`ing small amounts of compounds in a heterogeneous mix-
`ture, and is fast, accurate, and reliable [2,3,19]. Due to
`the high selectivity and sensitivity of MS/MS detection, it
`is a very promising analytical method also for the study
`of drug sorption.
`The aim of this study was to evaluate drug loss during
`through
`in vitro cell permeability experiments either
`adsorption to plastic cell culture material or retention in
`cultured cells. The LC–MS/MS-based assay system was
`developed for this purpose and a comprehensive set of test
`drugs with diverse physicochemical properties were select-
`ed. Many of the studied drugs are listed in the FDA Guid-
`ance for Industry as model drugs for permeability studies
`[20] and some drugs (e.g., diazepam, midazolam, and
`
`medetomidine) are known to interact with PVC and poly-
`styrene plastic [15,18,21]. In the experiments, glass and
`polypropylene (PP) tubes were compared to TC well plates
`and TC tubes. To our knowledge, this is the first report
`which details both speculated elements of drug loss, that
`is (1) drug adsorption to the plastic instruments and (2)
`retention of drugs in cultured cells. Interaction between
`the heterogeneous drugs and negatively charged polysty-
`rene is also illustrated. The results described here will be
`important in development and validation of in vitro drug
`permeation experiments.
`
`2. Experimental
`
`2.1. Chemicals and materials
`
`The compounds atenolol, propranolol, metoprolol,
`antipyrine, diazepam, midazolam, naproxen, probenicid,
`ibuprofen, hydrochlorothiazide and indomethacin were
`obtained from Sigma (St. Louis, MO). Medetomidine
`was from DomitorÒ (Orion, Finland). Buffer solution
`components were purchased from Bio Whittaker (Belgium)
`and water was purified and deionised by a Milli-Q system
`(Millipore). Acetonitrile and methanol (HPLC S grade)
`were obtained from Rathburn (Walkerburn, UK). Ammo-
`nium acetate and formic acid were from Riedel-de Haen
`(Seelze, Germany). All compounds and reagents were of
`the highest quality. Borosilicate glass tubes (16 · 100 mm,
`PyrexÒ), modified
`polystyrene
`culture
`tubes
`(16 ·
`125 mm), and well plates (12 well, Costar and TranswellÒ)
`were purchased from Corning Incorporated (NY) and
`polypropylene test
`tubes (10 mL) were from Sarstedt
`(Australia).
`
`2.2. Drug recovery assay
`
`2.2.1. Surficial binding of drugs to plastic and glass
`All drugs were solubilised in both Hanks’ balanced salt
`solution (HBSS) containing 25 mM of N-[2-hydroxyeth-
`yl]piperazine-N 0-[2-ethanesulfonic acid] (HEPES, pH 7.4)
`and in Milli-Q water. Final concentrations of each com-
`pound are presented in Tables 3 and 4. Test solutions con-
`tained the mixture of all six basic (pH 7.05) or all four
`acidic (pH 6.65) drugs in water or correspondingly in buffer
`(pH 7.4). Recovery experiments were performed using
`methods and conditions from traditional in vitro perme-
`ability studies. Test solutions (1.5 mL) were added to TC
`well plates, TC culture tubes, glass, and polypropylene
`tubes. Tubes and well plates were placed in an orbital hor-
`izontal shaker (Heidolph Inkubator 1000, Titramax 1000,
`Germany) with constant stirring (300 rpm) at either 37 or
`3 °C. Initial samples (200 lL) were collected from each test
`solution. Sample aliquots (200 lL) from well plates were
`collected at 15, 30, 60, 120, 180, and 270 min, and sample
`aliquots
`from test
`tubes were
`collected
`at
`120
`and 270 min. Equal amounts of internal standard (I.S.)
`(antipyrine and ibuprofen to basic and acidic mixtures,
`
`Petition for Inter Partes Review of US 8,648,106
`Amneal Pharmaceuticals LLC – Exhibit 1017 – Page 370
`
`
`
`J.J. Palmgre´n et al. / European Journal of Pharmaceutics and Biopharmaceutics 64 (2006) 369–378
`
`371
`
`respectively) were added to each sample to a final concen-
`tration of 90 nM. All recovery experiments were conducted
`in triplicate and samples were analysed during the experi-
`ment and at least within 2 h. The extent of binding to poly-
`carbonate membranes was also tested. Membranes from
`well plates (insert membrane: 0.4 lm pore size, 12 mm
`diameter) were cut out and placed to glass tubes. Test solu-
`tions (1.5 mL) in buffer or in water were added to glass
`tubes, and samples were collected and prepared as stated
`above in the case of test tubes.
`
`2.2.2. Extraction of drugs from TC culture tubes
`After initial surface binding experiments, the remaining
`solution was discarded and TC-tubes were flushed twice
`with Milli-Q water with 2 min of shaking. Afterwards,
`methanol (1.5 mL) was added to TC-tubes, which were vig-
`orously mixed for 5 min by vortex. Samples (200 lL) were
`then collected for the recovery determination. Extraction
`was also performed through addition of crystalline NaCl
`to the physiological concentration of 0.9%. After NaCl
`addition both TC-tubes and glass tubes were mixed for
`15 min by vortex. Samples were collected for quantification
`both before and after addition of salt.
`
`2.2.3. Cell culture and cellular uptake
`Caco-2 cells, a human colon adenocarcinoma cell line,
`were purchased from ATCC (Manassas, VA). Cells were
`grown on filters as described previously [22]. Buffered solu-
`tions containing basic drugs at 23.75 lM, except for mede-
`tomidine at 3.125 lM, were administered to cells for 2 h at
`37 °C in a temperature-controlled orbital shaker. Cells were
`washed twice with PBS buffer and lysed by addition of 0.1%
`Triton X-100 solution. Finally, cells were carefully scraped
`off the membranes, suspended by pipetting, and removed
`to microcentrifuge tubes. Donor samples (apical side) were
`collected before and after the experiments. All samples were
`stored at 20 °C until prepared, extracted, and analysed.
`Sample preparation and extraction were performed as
`described previously [22] with slight modification. Internal
`standard (antipyrine) was added to each sample at a concen-
`tration similar to that reported previously. Matrix effect for
`all the basic drugs (internal standard included) in cultured
`cells were determined as described earlier [22].
`
`2.2.4. Liquid chromatography
`The HPLC system included a Finnigan Surveyor MS
`pump and a Finnigan Surveyor autosampler (serial 1.4,
`San Jose, CA) with a 30 lL injection volume. The tray tem-
`perature and column oven control were set to +15 °C. The
`chromatographic separation was performed using a Xterra
`MS C18 reversed-phase column (2.1 · 20 mm, 2.5 lm,
`Waters, Milford, MA) with a flow rate of 200 lL/min.
`Solution A was water containing either 0.2 mM ammoni-
`um acetate or 0.1% formic acid for acidic or basic drugs,
`respectively. Solution B was composed of 80% acetonitrile
`and 20% of the corresponding solution A. The gradient
`profile for all the drugs was 0–80% acetonitrile in 6 min,
`
`and the column was re-equilibrated with solution A for
`4 min before the next injection.
`
`2.2.5. Mass spectrometry
`Measurements were performed with a LTQ quadrupole
`ion trap mass spectrometer equipped with an electrospray
`ionisation (ESI) source (Finnigan Surveyor LTQ, serial
`1.0 SPI, San Jose, CA). The mass spectrometer was operat-
`ed in the positive and negative ion modes for basic and
`acidic compounds, respectively. The quantification was
`based on multiple reaction monitoring (MRM) of the most
`intense fragment ions (m/z). In the MS/MS experiments,
`precursor molecular ions ([M + H]+ or [M H] ) were
`selected and fragmented in the ion trap. Mass spectromet-
`ric parameters were optimized by constant addition of a
`single analyte in water to the HPLC flow via a T-connec-
`tor. The conditions and parameters employed for acidic
`drugs were: capillary temperature 250 °C, spray voltage
`4.2 kV, sheath gas flow rate 35 (arbitrary units), capillary
`voltage 18 V, tube lens 65 V, and for basic drugs were:
`capillary temperature 275 °C, spray voltage 4.5 kV, sheath
`gas flow rate 35 (arbitrary units), capillary voltage 26 V,
`tube lens 75 V. In the ion trap, the relative collision energy
`ranged from 40% to 60% for all the monitored drugs. The
`flow from the HPLC was diverted to waste for the first
`1.5 min and after 6 min to decrease ion source contamina-
`tion. Data were processed using the Xcalibur software
`package version 1.4 SRI.
`
`2.2.6. Standard solutions, calibration, and accuracy
`Individual stock solutions (10 mM) of compounds were
`prepared separately in methanol, except medetomidine,
`which was commercially available in aqueous solution.
`Stock solutions were further diluted to 1 mM in Milli-Q
`water. Working solutions (40–400 lM) were prepared by
`diluting the stock solutions (1 mM) in water or in buffer
`solution. Furthermore, the working solutions were com-
`bined and further diluted with water or buffer solution.
`This mixture of basic or acidic compounds was used both
`for test solutions used in Section 2.2.1, and for calibration
`and quality control (QC) standards after serial diluting.
`The test solutions, calibration solutions, and QC standards
`contained either six basic or four acidic compounds. The
`calibration range and QC standard values of each com-
`pound are shown in Tables 1 and 2. Similarly, I.S. working
`solutions (1 lM, antipyrine or ibuprofen) were prepared by
`diluting stock solutions with water or buffer solution.
`Equal amounts of I.S. were added to each standard and
`sample solution to 90 nM. All stock and working solutions
`were stored in the dark at 20 °C until used. Test solu-
`tions, calibration solutions, and QC standards were pre-
`pared daily and analysed immediately after preparation.
`Calibration curves were constructed by plotting chroma-
`tographic peak ratios of standard area/I.S. area versus con-
`centration of the standard using linear regression. From
`these curves the coefficients of correlation (r2) were calcu-
`lated. The lowest limit of quantification (LLOQ) for each
`
`Petition for Inter Partes Review of US 8,648,106
`Amneal Pharmaceuticals LLC – Exhibit 1017 – Page 371
`
`
`
`372
`
`J.J. Palmgre´n et al. / European Journal of Pharmaceutics and Biopharmaceutics 64 (2006) 369–378
`
`Table 1
`Calibration range, linearity (r2), and LLOQ of the LC–ESI-MS/MS method for various compounds in buffer solution
`Linearity (r2)a
`
`Compounds in test solution
`
`Calibration range (nM)
`
`LLOQ (n = 3)
`
`Bases
`Atenolol
`Metoprolol
`Medetomidine
`Propranolol
`Midazolam
`Diazepam
`
`Acids
`Hydrochlorothiazide
`Naproxen
`Probenicid
`Indomethacin
`
`2.5–50.0
`2.5–50.0
`1.0–20.0
`5.0–100.0
`1.0–20.0
`2.5–50.0
`
`20.0–320.0
`20.0–320.0
`2.5–80.0
`10.0–80.0
`
`a Correlation coefficient from six calibration points (n = 3).
`
`0.9939
`0.9962
`0.9966
`0.9919
`0.9899
`0.9970
`
`0.9946
`0.9960
`0.9957
`0.9944
`
`nM
`
`RSD%
`
`Accuracy (%)
`
`2.5
`2.5
`1.0
`5.0
`1.0
`2.5
`
`20.0
`20.0
`2.5
`10.0
`
`9.03
`8.72
`5.28
`8.78
`10.50
`6.84
`
`4.22
`8.25
`5.80
`2.19
`
`100.33
`96.11
`102.62
`97.07
`109.62
`102.08
`
`111.08
`106.07
`99.44
`106.68
`
`compound was calculated based on the FDA Guidance for
`Industry, Bioanalytical Method Validation [23]. Briefly, the
`analyte response at LLOQ should be five times level of the
`
`baseline noise, and the analyte response at LLOQ should
`be determined with precision of <20% and accuracy of
`80–120%.
`
`Table 2
`Within-day and between-day precision and accuracy of the LC–ESI-MS method for the various compounds used in this study
`
`In Milli-Q water
`
`In buffer solution
`
`Within-day
`variation
`(n = 6)
`QCb
`(nM)
`
`Between-day
`variation
`(3 days, n = 9)
`
`RSD
`(%)
`
`Accuracy
`(%)
`
`RSD
`(%)
`
`Accuracy
`(%)
`
`Within-day
`variation
`(n = 6)
`QCb
`(nM)
`
`Between-day
`variation
`(5 days, n = 15)
`
`RSD
`(%)
`
`Accuracy
`(%)
`
`RSD
`(%)
`
`Accuracy
`(%)
`
`Basesa
`Atenolol
`
`Metoprolol
`
`Medetomidine
`
`Propranolol
`
`Midazolam
`
`Diazepam
`
`12.5
`20.0
`12.5
`20.0
`5.0
`8.0
`25.0
`40.0
`5.0
`8.0
`12.5
`20.0
`
`3.64
`4.55
`3.19
`4.03
`4.89
`6.88
`5.66
`9.05
`7.09
`9.94
`5.38
`7.45
`
`96.44
`104.39
`107.42
`97.60
`102.91
`92.85
`101.82
`92.55
`101.43
`103.13
`96.90
`103.27
`
`5.90
`3.39
`7.17
`2.20
`6.45
`4.29
`4.90
`4.63
`5.07
`4.91
`6.17
`2.94
`
`95.66
`97.24
`97.62
`98.73
`95.27
`97.04
`92.62
`98.36
`92.98
`105.69
`93.85
`104.49
`
`12.5
`20.0
`12.5
`20.0
`5.0
`8.0
`25.0
`40.0
`5.0
`8.0
`12.5
`20.0
`
`4.27
`3.35
`8.35
`3.49
`7.67
`5.23
`5.03
`5.89
`5.43
`4.67
`6.58
`3.96
`
`97.17
`96.60
`99.54
`97.05
`92.40
`94.20
`99.31
`99.92
`105.22
`106.60
`92.94
`101.53
`
`In Milli-Q water
`
`In buffer solution
`
`3.11
`5.21
`2.46
`2.48
`4.33
`3.24
`4.05
`4.62
`7.98
`6.83
`4.70
`4.11
`
`102.30
`98.76
`100.57
`100.78
`99.56
`99.49
`99.22
`101.73
`97.05
`106.27
`95.25
`103.81
`
`Between-day
`variation
`(3 days, n = 9)
`
`Within-day
`variation
`(n = 6)
`QCb
`(nM)
`
`Between-day
`variation
`(3 days, n = 9)
`
`RSD
`(%)
`
`Accuracy
`(%)
`
`RSD
`(%)
`
`Accuracy
`(%)
`
`Within-day
`variation
`(n = 6)
`QCb
`(nM)
`
`RSD
`(%)
`
`Accuracy
`(%)
`
`RSD
`(%)
`
`Accuracy
`(%)
`
`Acidsa
`Hydrochlorothiazide
`
`Naproxen
`
`Probenicid
`
`Indomethacin
`
`64.0
`100.0
`64.0
`100.0
`16.0
`25.0
`16.0
`25.0
`
`8.91
`8.71
`4.88
`6.79
`8.21
`8.28
`4.43
`6.37
`
`109.26
`103.51
`101.73
`107.19
`99.77
`102.01
`102.02
`97.86
`
`6.43
`4.51
`2.61
`3.16
`3.69
`3.58
`5.36
`3.74
`
`96.11
`102.83
`100.07
`100.95
`98.38
`102.55
`105.58
`104.99
`
`64.0
`100.0
`64.0
`100.0
`16.0
`25.0
`16.0
`25.0
`
`3.20
`4.28
`2.49
`3.02
`2.55
`2.13
`2.12
`2.52
`
`94.60
`90.65
`99.13
`94.97
`91.36
`92.34
`106.12
`109.49
`
`2.67
`3.17
`1.65
`2.36
`3.32
`2.87
`2.43
`4.21
`
`98.21
`93.24
`101.24
`95.38
`92.10
`91.77
`103.48
`107.28
`
`a Bases and acids are in chromatographical order.
`b QC, Quality control sample (nominal concentration).
`
`Petition for Inter Partes Review of US 8,648,106
`Amneal Pharmaceuticals LLC – Exhibit 1017 – Page 372
`
`
`
`J.J. Palmgre´n et al. / European Journal of Pharmaceutics and Biopharmaceutics 64 (2006) 369–378
`
`373
`
`Within-day accuracy and precision of the assay were
`determined by repetitive measurements (n = 6) of QC stan-
`dards at two different concentrations. Precision was calcu-
`lated as the relative standard deviation (RSD%) and
`accuracy was determined as the mean% [(mean measured
`concentration)/(expected concentration) · 100]. Between-
`day accuracy and precision were evaluated by performing
`repeated measurements of the same QC standards on three
`or five different days and calculated in the same manner as
`the within-day values. Both accuracy and precision were
`also tested according to FDA guidance through the follow-
`ing criteria: the accuracy and precision deviation values
`should be within 15% of the actual values.
`
`3. Results and discussion
`
`3.1. Liquid chromatography
`
`All compounds were separated within 6 min (Fig. 1).
`The Xterra column properties allowed for fast analysis as
`the total chromatographic runtime was only 10 min. In sur-
`face binding experiments, the short analysis time was desir-
`able because sample aliquots were collected within 15 min
`intervals (at the beginning of study) and the total experi-
`ment time was only 4.5 h. The runtime potentially could
`be shorter, however, the wide ranging lipophilicity of the
`set of drugs used here required a total run time to 10 min
`for the desired degree of separation. Because hydrophilic
`compounds such as atenolol and hydrochlorothiazide ini-
`tially eluted quite early, the column oven was set to
`+15 °C, which resulted in uniform peak shapes,
`longer
`retention times, and better separation from salts and impu-
`rities. Due to the high specificity and selectivity of MS/MS
`detection, no interfering peaks from other compounds were
`found in ion channels specific for a given m/z value.
`Furthermore, the elution profile was sufficient to elute all
`
`of the drugs in a mixture, since pure water samples did
`not present any traces of carry-over.
`
`3.2. Mass spectrometry
`
`ESI source coupled with MS was chosen for this study
`because ESI-MS-based methods are commonly used and
`suitable for low molecular weight pharmaceutical com-
`pounds. Detection of acidic drugs using ion trap instru-
`ments has
`typically been performed using full MS
`mode [24,25]. In this study, both acidic and basic drugs
`were monitored by MS/MS detection. The operating
`parameters
`for ESI-MS were manually optimized to
`maximize the detection sensitivity, and general settings
`were used for each compound. The ionisation in the
`positive ion mode for all the basic drugs was sufficient,
`since abundant [M + H]+ ions were observed for each
`compound. However, tuning of negative ion source for
`detection of acidic drugs was laborious and it was neces-
`sary to obtain a high level of ionisation with intense
`[M H] ions. The protonated and deprotonated mole-
`cules were both selected as precursor ions and product
`ions were detected by the MS/MS mode. The most
`intense product ion for each compound was used for
`quantification. Representative precursor and product ions
`are listed in Fig. 1. Furthermore, three different buffer
`compositions were used in the mobile phase to optimize
`peak intensity and retention times of acidic compounds.
`We observed that 10 mM ammonium acetate suppressed
`the signal of acidic compounds as reported previously
`[24]. Therefore, ammonia solution at pH 8.2 was tested,
`which resulted in better sensitivity, but peak shapes were
`uneven. Weak ammonium acetate (0.2 mM) buffer was
`ultimately selected for the analysis because it gave both
`uniform peak shapes and similar sensitivity as the ammo-
`nia solution.
`
`100
`
`100
`
`100
`
`100
`
`100
`
`0
`
`ecnadnubAevitaleR
`
`85.2
`
`692z/medizaihtorolhcordyH
`
`962
`
`922z/mnexorpaN
`
`581
`
`482z/mdicineborP
`
`042
`
`502z/mneforpubI
`
`161
`
`653z/mnicahtemodnI
`
`213
`
`92.4
`
`26.4
`
`31.5
`
`22.5
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`)nim(emiT
`
`100
`
`100
`
`100
`
`100
`
`100
`
`100
`
`100
`
`0
`
`ecnadnubAevitaleR
`
`03.2
`
`29.2
`
`99.2
`
`71.3
`
`23.3
`
`63.3
`
`762z/mlolonetA
`
`522
`
`862z/mlolorpoteM
`
`191
`
`pitnA
`
`y
`
`981z/menir
`
`641
`
`102z/menidimotedeM
`
`59
`
`062z/mlolonarporP
`
`381
`
`623z/mmalozadiM
`
`192
`
`582z/mmapezaiD
`
`752
`
`44.4
`
`1
`
`2
`
`3
`)nim(emiT
`
`4
`
`5
`
`6
`
`Fig. 1. Chromatography and tandem mass spectrometry of test compounds. Quality control mixtures of either acidic or basic drugs prepared in buffer
`solution were separated by reverse-phase LC over a 6-min gradient. Product ions of acidic compounds (left) or of basic compounds (right) were generated
`and measured by ESI/MS/MS.
`
`Petition for Inter Partes Review of US 8,648,106
`Amneal Pharmaceuticals LLC – Exhibit 1017 – Page 373
`
`
`
`374
`
`J.J. Palmgre´n et al. / European Journal of Pharmaceutics and Biopharmaceutics 64 (2006) 369–378
`
`3.3. Calibration and accuracy
`
`Validation was performed for all four different condi-
`tions (acidic/basic drugs in water and buffer). The calibra-
`tion curves of
`test drugs were linear over the used
`concentration ranges and correlation coefficients (r2) were
`greater than 0.9899 (Table 1). Linearity was similar both
`in water and buffer solution, so only coefficients obtained
`from buffered samples are shown in Table 1. The equations
`for the curves were calculated using six calibration points
`with three replicate standards for each point (n = 3) per
`curve. To compensate for analyte losses during sample
`preparation and analysis, the sample assay was based on
`the internal standard method, which was calculated from
`the peak area ratios of unknown/I.S. versus the calibration
`curve. The LLOQ data in buffered solution are presented in
`Table 1. The accuracy range was 96–111% and the RSD
`precision was lower than 10.5% (n = 3) for all LLOQ val-
`ues. Further, the signal-to-noise ratios obtained at the
`LLOQ were at least 20:1. The sensitivity of this assay is
`reflected by the LLOQ value of diazepam, which was
`5000–50,000 times lower than concentrations measured
`by UV spectrophotometry [12,13].
`Within-day precision was evaluated by performing six
`repetitive analyses of QC standards, which gave RSD val-
`ues between 2.12% and 9.94%. The accuracy range was 92–
`110% (Table 2). The between-day accuracy of the methods
`ranged from 91% to 108% and the RSD precisions were
`lower than 8%. As seen in Table 2, RSD and accuracy val-
`ues were similar both for samples in buffer and water.
`Overall, the tested parameters surpass the FDA criteria
`[23].
`
`3.4. Adsorption experiment
`
`In a typical in vitro permeability experiment, the donor
`drug concentration can range from 20 to 1000 lM [26,27].
`Correspondingly, the final receiver sample can range from
`1 to 50 lM. The first 60 min of the experiment is most often
`used for the calculation of coefficients and during that time
`the drug concentration can be observed at nM levels. In this
`study, both lM and nM concentration levels were studied.
`Each drug concentration was selected individually according
`to the detection intensity that was about 10–20 times the
`LLOQ value of each drug. Even though the concentrations
`in the test solutions were low, they were similar to those
`observed in traditional permeability studies. The accuracy
`and precision of the assay method was remarkable, which
`provided the ability for quantification of small losses of
`drugs. Therefore, the LC–MS/MS method development
`was validated extensively.
`
`3.4.1. Loss of acidic drugs
`For the most part, acidic drugs showed little if any sorp-
`tion to all tested materials in either water or buffer. The
`lack of binding was seen both in the presence of glass
`and plastic materials. The only appreciable binding
`
`Table 3
`Sorption of acidic drugs in water or buffer to different containers and
`materials after 270 min
`
`Drug and material
`
`% remaining ± SD (n = 3)
`
`Acids in test solution (nM)
`
`Water +37 °C
`
`Buffer +37 °C
`
`Hydrochlorothiazide (100.0)
`Glass tube
`Polypropylene tube
`TC tube
`TC well plate
`Polycarbonate membrane
`
`Naproxen (100.0)
`Glass tube
`Polypropylene tube
`TC tube
`TC well plate
`Polycarbonate membrane
`
`Probenicid (25.0)
`Glass tube
`Polypropylene tube
`TC tube
`TC well plate
`Polycarbonate membrane
`
`Indomethacin (25.0)
`Glass tube
`Polypropylene tube
`TC tube
`TC well plate
`Polycarbonate membrane
`
`103.7 ± 5.4
`103.7 ± 10.2
`93.7 ± 10.8
`87.6 ± 9.4
`–
`
`95.3 ± 7.1
`90.9 ± 10.8
`94.6 ± 1.7
`92.2 ± 4.1
`–
`
`90.8 ± 5.0
`87.2 ± 8.0
`88.6 ± 0.5
`91.0 ± 6.5
`–
`
`94.9 ± 8.3
`69.5 ± 4.7
`83.3 ± 2.6
`86.2 ± 5.1
`–
`
`85.6 ± 3.7
`87.4 ± 8.3
`92.2 ± 3.2
`91.6 ± 5.0
`91.9 ± 5.8
`
`97.3 ± 2.7
`100.7 ± 3.8
`99.3 ± 2.0
`101.6 ± 3.7
`101.1 ± 7.0
`
`88.8 ± 1.5
`92.9 ± 5.4
`91.0 ± 2.2
`96.7 ± 3.6
`95.6 ± 5.7
`
`92.1 ± 4.4
`98.8 ± 1.3
`102.4 ± 0.8
`101.4 ± 1.8
`101.6 ± 2.9
`
`observed was with the lipophilic compound indomethacin
`in water to PP-tubes, where the concentration decreased
`to 69.5 ± 4.7% (n = 3) from that initially added (Table 3).
`
`3.4.2. Loss of basic drugs in buffer (+37 °C)
`Cell culture permeability screening is typically per-
`formed in buffered solutions (pH 7.4), so loss of drugs
`was studied in the presence of buffer. As summarized in
`Table 4 there was no significant loss of drugs dissolved in
`buffer to any of the tested materials indicating that buffer
`reduced surficial
`interaction. All drugs remained above
`79.9% of their initial value, except propranolol, which
`had 72.7 ± 5.5% remaining in solution in TC-tubes.
`
`3.4.3. Loss of basic drugs in water (+37 °C)
`The loss of basic drugs to polystyrene well plates and
`TC-tubes in water was a rapid process. All the drug losses
`were achieved within the first 15 min (Fig. 2). After 4.5 h,
`the relative amount remaining in TC-tubes in aqueous
`solution was 64.7 ± 6.8%, 38.4 ± 9.1%, 31.9 ± 6.7%, and
`23.5 ± 6.1% for metoprolol, medetomidine, propranolol,
`and midazolam, respectively (Table 4). Hydrophilic ateno-
`lol did not show loss in any of the different containers. As
`seen in Table 4, the loss of basic drugs to TC-plastic was
`much higher than to glass and PP-tubes. Midazolam had
`a strong affinity to TC-plastic in aqueous solution, but this
`affinity was greatly diminished in the presence of buffer.
`Fig. 3 summarizes the loss of midazolam in test solutions
`in different containers.
`
`Petition for Inter Partes Review of US 8,648,106
`Amneal Pharmaceuticals LLC – Exhibit 1017 – Page 374
`
`
`
`J.J. Palmgre´n et al. / European Journal of Pharmaceutics and Biopharmaceutics 64 (2006) 369–378
`
`375
`
`Glass tube
`
`PP tube
`
`TC tube
`
`TC well plate
`
`120
`
`100
`
`80
`
`60
`
`40
`
`20
`
`0
`
`Midazolam in test solution (%)
`
`In water
`
`In buffer
`
`Fig. 3. Sorption of midazolam to various surfaces. Midazolam was
`prepared in a test solution of either water or buffer and was added to
`different containers for 270 min at +37 °C. The remaining concentration
`was determined by HPLC–ESI/MS/MS. Each value indicates mean ± SD
`(n = 3).
`
`3.4.4. Loss of basic drugs in water (+3 °C)
`Sorption experiments were repeated at +3 °C since this
`temperature is commonly used to study sorption kinetics
`to plastic [17] as well as transport mechanisms of drugs
`in cultured cells [5]. Generally, at +3 °C loss of basic drugs
`to TC-tubes was less than at +37 °C. On the contrary, loss
`of basic drugs increased in well plates at +3 °C compared
`to that at +37 °C (Table 4). This result may be due to a
`lower extent of evaporation at the colder temperature,
`since the concentration in wells increased over time at
`+37 °C. The effect of evaporation can be seen in Fig. 3,
`where the concentrations in well plates are higher than in
`TC-tubes. Sorption to polycarbonate membranes was also
`investigated. Membranes were cut out from inserts and
`placed to glass tubes for analysis. Results indicated that
`the loss of medetomidine, propranolol, and midazolam to
`polycarbonate membrane was similar to that observed in
`TC-tubes in water at +3 °C (Table 4).
`
`3.4.5. The effect of drug concentration to sorption
`The sorption of medetomidine, propranolol, and midaz-
`olam to TC-tubes was tested at a high concentration
`(1000 nM of each drug in water). At this concentration,
`these drugs remained above at 58% of the initial value
`and the losses were clearly lower than observed at
`8–40 nM. When the drug concentration in the test solution
`was 2500 nM, no drug loss was detected at all. This sug-
`gests that the higher the concentration the lower the
`amount of drug loss, furthermore, the surface of tubes
`can interact with only a limited amount of drugs.
`
`3.4.6. Extraction of drugs from culture tubes (TC-tubes)
`To account for the drug loss observed in TC materials
`after initial sorption experiments, TC-tubes were flushed
`with water and treated with methanol to extract any bound
`drugs. Methanol addition recovered 24.0 ± 4.5%, 46.5 ±
`9.2% and 72.5 ± 13.9% of initial amounts of medetomi-
`dine, propranolol, and midazolam, respectively. As seen
`in Table 5, methanol wash recovered almost all of the frac-
`tion lost for midazolam. Since buffer solution decreased the
`
`Table 4
`Sorption of basic drugs in water or buffer to different containers and
`materials after 270 min
`
`Drug and material
`
`% remaining ± SD (n = 3)
`
`Bases in test solution
`(nM)
`
`Atenolol (20.0)
`Glass tube
`Polypropylene tube
`TC tube
`TC well plate
`Polycarbonate membrane
`
`Metoprolol (20.0)
`Glass tube
`Polypropylene tube
`TC tube
`TC well plate
`Polycarbonate membrane
`
`Medetomidine (8.0)
`Glass tube
`Polypropylene tube
`TC tube
`TC