Identification and Comparison of

Extractables in Drug Container

Closure Systems
State-of-the-art Data Mining Process Taking Complete
Advantage of Electron Ionization, Chemical Ionization,
Collision-induced Dissociation, and Accurate Mass
Information using an Agilent 7200 GC/Q-TOF System

Application Note
Pharmaceutical

Author Abstract
Syed Salman Lateef Regulatory authorities and working groups provide guidelines on the evaluation
Agilent Technologies, Inc. of drug containers to ensure the safety and efficacy of drug products. The
guidelines for container qualification suggest a risk-based approach when applying
an extraction study. This approach identifies and classifies the toxicological
relevance of nonvolatiles, semivolatiles, and elemental contaminants found in
extracts from components or from the entire drug container closure system. In
this work, four drug containers originating from different manufacturers have been
investigated, generating profiles for semivolatile extractable compounds. Data
were acquired on an Agilent 7200 GC/Q-TOF in both electron ionization (EI) and
chemical ionization (CI) modes to ensure a comprehensive identification of the
extracted compounds. El data have been acquired to identify compounds using the
NIST library. Compound distribution across the different samples was visualized
based on a chemometric approach using Agilent MassHunter Mass Profiler
Professional software. Compounds that could be only tentatively identified after EI
were confirmed using parent ion mass data generated in CI mode. Approximately
170 compounds were identified in each container. The results indicate that the
combination of EI and CI data increased the number of identified extractable
compounds, enabling a thorough evaluation of container systems.
Introduction Known reactive functional groups that This Application Note shows that a
fall under Class III are aliphatic secondary high resolution Agilent 7200 GC/Q-TOF
Extractable studies on drug containers
amino-, cyano-, N-nitroso-, diazo-, system was used for extractable
provide valuable information that ensures
triazeno-, quarternary N, strain-ringed studies on four different drug container
the safety and efficacy of drugs. An
lactones, epoxides, quinones, and systems. The sample preparation
extractable study of the drug container
α,β-unsaturated ketones4. Recently, aimed to analyze only the semivolatile
closure system during the early phase
Jenke3 developed and justified a compounds. The chromatograms were
of the manufacturing process efficiently
semiquantitative risk evaluation matrix acquired in both EI and CI modes. Using
supports early material assessment
used to determine the amount and testing the Agilent MassHunter Unknown
and selection processes1. Changes in
necessary to establish whether the Analysis software, the EI spectra were
the manufacturing process when using
container is suitable for its intended use. automatically deconvoluted and matched
containers from different vendors change
to NIST 14 library. The data acquired from
the extractable profiles. Evaluation of An extractable study, like any untargeted the extracts of four different containers
the toxicity of extractable compounds study, attempts to identify compounds were visualized using a Venn diagram,
from the drug container closure without prior knowledge of the sample which is a feature of Agilent Mass
systems is performed using a risk-based contents. Compounds showing strong Profiler Professional (MPP) software.
approach outlined by the Food and Drug fragmentation in electron ionization Compounds that were tentatively
Administration (FDA)2,3. This approach (EI) mode will increase the number of identified due to low EI library matching
allows for patient population, route of unknowns detected. However, by applying scores were converted to a custom
drug administration, and the potential an orthogonal soft ionization technique, database. These low custom databases
for interaction between formulation such as chemical ionization (CI), highly were used to mine CI accurate mass
and container systems. For ophthalmic labile compounds amiable to softer data to confirm some of the compounds.
drug products (ODP), the degree of ionization can be more easily identified. Figure 1 shows the workflow used for this
concern associated with the route of Experiments performed on accurate study.
administration, and the likelihood of mass instrumentation while using
interaction between the liquid formulation custom-made accurate mass databases
and the packaging material has been would invariably increase the number of
defined as high. compounds identified. Accurate mass
instrumentation would also facilitate the
The approach to evaluate the toxicity
identification of unknowns by formula
is to rely on available literature on the
generation of molecular peaks and their
absorption, distribution, metabolism, and
fragments.
elimination (ADME) of the compound.
If such information is not available, the
compounds must be classified based Workflow
on their structure using the Cramer
classification approach. The Cramer Container closure system
classification classifies compounds into Sample: Empty pharmaceutical bottles treated with solvent
different classes: Control: Extraction solvent

• Class I – low toxicity GC/Q-TOF aquisition EI and CI mode

• Class II – intermediate toxicity

EI mode CI mode
• Class III – significant toxicity Library
Agilent MassHunter Unknowns editor Database/Library
Analysis software Agilent MassHunter Qualitative Analysis
Targeted or untargeted analysis

Structure confirmation by CI/MS/MS

Agilent Mass Profiler
Professional Software analysis

E&L list

Figure 1. Extractable and leachables workflow for analysis of semivolatiles using accurate mass
GC/Q-TOF system.

2
Experimental Table 1. GC/Q-TOF instrument parameters used in this experiment.
HPLC grade n-hexane, 99%, was GC conditions
purchased from RCI Labscan (Thailand).
GC Agilent 7890A
Injection port Multimode inlet (MMI)
Sample preparation
Mode Splitless
Four empty formulation bottles Septum purge flow 3 mL/min
were purchased locally. Two were
Inlet program 70 °C (0.2 minutes) to 325 °C (7 minutes) at 600 °C/min
made of low-density polyethylene
(containers 1 and 2), and two were made Liner Ultra Inert Splitless, single taper, glass wool (p/n 5192-3163)
of polyethylene (containers 3 and 4). Carrier gas Helium
Five milliliters of n-hexane was added to Flow 1.3 mL/min (constant)
each, and sonicated for 1.5 hours. After Purge flow to split vent 60 mL/min at 2.73 minutes
sonication, the solvent was analyzed. Gas saver 20 mL/min at 3 minutes
n-hexane extraction solvent was used as Oven program 50 °C (3 minutes) to 320 °C (7 minutes) at 6 °C/min
a solvent blank.
Equilibration time 1 minute
Run time 55 minutes
Data acquisition and processing
Columns Agilent DB-5ms, 30 m × 250 µm, 0.25 µm (p/n 122-5532)
The following Agilent software was used
Injection volume 2 µL
for data acquisition and processing:
MS conditions
• Agilent MassHunter Acquisition MS Agilent 7200
Software (B.07.02) Tune Autotune
Transfer line 280 °C
• Agilent MassHunter Qualitative MS source (EI and CI) 300 °C
Analysis Software including PCDL MS quad 175 °C
Manager Standalone tool (B.07.00). Mass range 55 to 700 amu
Acquisition rate 5.00 spectra/sec
• Agilent MassHunter Quantitative
Election ionization
Analysis Software including Library
EI emission current 35 µA
Editor and Unknown Analysis
standalone tools (B.07.01). EI electron energy 70 eV
Chemical ionization
• Agilent Mass Profiler Professional CI emission current 240 µA
Software (Ver. 13.1) CI gas flow 20 % EPC
CI electron energy 115 eV
Instrument parameters Mode Positive
Table 1 shows the instrument parameters CI reagent gas Methane
used in this analysis. Collision cell EPC Nitrogen, 1.5 mL/min

Data analysis
EI source data analysis
The data files were processed using
MassHunter Unknowns Analysis software
for deconvolution, and matched against
the NIST 14 library match. A match score
of > 80 was used to select identified
compounds.

3
Creating an accurate mass EI library CI data analysis Results and Discussion
The EI results with scores from 50 to 79 The CI data were processed in
were exported to Library Editor Software MassHunter Qualitative Analysis software
EI mode analysis
to form a low score EI library. Scores ≥ 80 using the Find by Formula algorithm with The data acquired in EI mode were
were exported to form a high score possible adducts [M+H]+, [M+C2H5]+, and processed using the Unknowns Analysis
EI library. The library (in .xml format) [M+C3H5]+. The low score EI library was tool for chromatographic deconvolution
contained compound information such as used as a formula database. The CI data and library matching. Although
name, formula, retention time (RT), and were also searched for other extractables height-based filtering of compounds can
spectra. using the custom user-created be performed by Unknowns Analysis
extractable library. Software, no filtering was applied.
MPP analysis Benzene, (1-ethylundecyl)-, also called
Structure elucidation using CI/MS/MS 2-phenyl tridecane, is an extractable
The EI data were reprocessed using the compound identified at 27.3 minutes
Unknowns Analysis tool to deconvolute CI/MS/MS data files were processed (Figure 2) in container 2. The Extracted
and match spectra and RT using the using the Find by Targeted MS/MS Ions Chromatograms (EIC) of this
accurate mass, high score EI library. This feature within MassHunter Qualitative deconvoluted component coeluted, and
step helped to filter the results to be Analysis software. The fragment had the same peak shape (Figure 2C)
exported into MPP software. structures were confirmed and drawn while its EI spectrum had a unit mass
using ACD software (ACD Labs, Toronto). (NIST) library match with a score 85.3.
Agilent Personal Compound Database
(PCD) Toxicological evaluation
A custom database of literature-reported In silico prediction for Cramer
extractables and leachables was created. classification was performed using
The database entries consisted of Toxtree v 2.6.135
chemical formula, accurate mass, and
CAS ID.

A Components results B EIC

×10 7
6 27.3317
27.1722

27.7912
5
26.6548

26.9702
26.8204

27.0796

4
Counts

27.6710

28.0806

28.3559
3
2
1
0
26.4 26.6 26.8 27.0 27.2 27.4 27.6 27.8 28.0 28.2
Acquisition time (min)
C Ion peaks
Component RT: 27.3317 D Mirror plot
×10 7
×10 2 Component RT: 27.3335
1.5 91.0548
1.4 Component
1.3 91.0546 0.8
1.2 119.0855 Experimental spectra
0.6
1.1 105.0702 119.0856
1.0 231.2116 0.4
Counts

0.9
Counts

260.2506 0.2 231.2120

0.8
0.7 0
0.6 71.0
-0.2 41.0 231.0
0.5
0.4 -0.4
0.3 Library spectra
0.2 -0.6
119.0
0.1 -0.8 91.0
0
27.3 27.35 27.4 25 50 75 100 125 150 175 200 225 250 275
Acquisition time (min) Mass-to-charge (m/z)

Figure 2. Agilent MassHunter Unknown Analysis Software identified Benzene (1-ethylundecyl), by deconvolution and NIST library search. Component results (A),
deconvoluted component chromatograms (B), overlay of EICs of individual component (C), and mirror plot of deconvoluted component spectrum and library hit (D).

4
Compounds with acceptable spectral Entity list 1: Container 1 Entity list 4: Container 4
matches with a score > 80, were exported 150 entities 207 entities
to MPP software for data interpretation.
The compound distribution across
samples was visualized using a Venn
diagram in MPP software. Figure 3 shows
the Venn diagram of the distribution
of identified extractables from the four
different container systems. The Venn
diagram shows 22 compounds (identified)
commonly found in all containers
independent of the manufacturer.
Comparison of extractables distribution
Entity list 2: Container 2 Entity list 3: Container 3
shows more than 200 compounds present 141 entities 204 entities
in containers 3 and 4. Approximately
150 compounds have been extracted Figure 3. An Agilent Mass Profiler Professional Venn diagram showing the distribution of EI identified
from containers 1 and 2, wherein compounds among the four containers.
container 2 shows the lowest number
of compounds. The study shows that Table 2. Selected list of compounds identified in extract of container 2 (14 compounds from the Venn
containers 3 and 4 are less suitable as a diagram in Figure 3) and compounds common between all samples.
container system, compared to containers
Extractable compound list found only in container 2
1 and 2. Table 2 shows a selected list
of extractables identified in container 2, Octadecane
and the compounds found common to all Benzene, (1-methylundecyl)-
containers. Pentadecane, 2,6,10,14-tetramethyl-
Heptadecane, 3-methyl-
Benzene, (1-ethylundecyl)-
Eicosane, 2-methyl-
1,2-Benzenedicarboxylic acid, bis(2-methylpropyl) ester
Heneicosane
Nonadecane
Tetradecane
Pentacosane
Commonly found extractables in all containers
Pentadecane
trisiloxane, 1,1,1,5,5,5-hexamethyl-3-[(trimethylsilyl)oxy]-
Heptadecane, 3-methyl
Pentadecane, 2,6,10-trimethyl-
Nonane, 4,5-dimethyl
Tetracosane, 11-decyl
Pentadecane, 8-hexyl
(E)-Hex-3-enyl (E)-2-methylbut-2-enoate
Dodecane
3-Ethyl-3-methylnonadecane
Cyclohexasiloxane, dodecamethyl-
Dodecane, 2,6,10-trimethyl
Octane, 3,5-dimethyl-
Tridecane
Nonane, 5-(1-methylpropyl)-
Tridecane, 6-methyl-
Sulfurous acid, 2-ethylhexyl isohexyl ester
Cycloheptasiloxane, tetradecamethyl-
6,6-Diethylhoctadecane

5
CI mode analysis was confirmed using the CI mode with using the literature-derived custom
The CI data acquisition mode increased a 0.58 ppm mass error for the most database. Accurate mass CI data can be
the number of identified compounds that abundant molelcular ion (Table 3A). used to distinguish between compounds
were not considered as identified after Additionally, a database of EI high score that cannot be determined using CI
EI data acquisition due to low library results could be made to mine CI data mode unit resolution single quad data.
matching scores. To confirm the presence to confirm EI results, as previously For example, m/z 194.094 and 194.0577
of low score EI results, a library of low shown6. To identify more compounds, an have been identified as two different
score hits from the EI data analysis were in-house built database was compiled compounds: benzoic acid, 4-ethoxy-
made using the Library Editor Software. containing extractables reported in the ethyl ester and 1,3-benzenedicarboxylic
For example, benzoic acid, 4-ethoxy-, literature. Accurate mass information acid, 1,3,-dimethylester, respectively.
ethyl ester had an EI matching score of helps to identify compounds by formula. This identification of compounds having
67 of 100 points because it was buried Table 3B shows the list of contaminants the same nominal m/z requires high
in the chemical noise. The compound identified by data mining the CI files, resolution accurate mass data.

Table 3. CI-MS data were searched against low score EI results (A) and extra compounds detected using in-house databases (B) from container 2. The mass
error refers to the most abundant molecular ion.

A) Extractable by low score EI databases B) Extractable by literature-derived custom databases

Mass error Mass error
ID Mass (ppm) ID Mass (ppm)
(-)-Aristolene 204.1880 1.4 1,2-(1,8-napthalenediyl)benzene 202.0784 -0.6
(2S,6R,7S,8E)-(+)-2,7-Epoxy-4,8-megastigmadiene 192.1510 -3.3 1,3-Benzenedicarboxylic acid 166.0266 0.1
(3S,4aR,8aR)-1,1,3,6-Tetramethyl-3-vinyl-3,4,4a,7,8, 220.1830 3.6 1,3-Benzenedicarboxylic acid, 194.0577 0.9
8a-hexahydro-1H-isochromene 1,3-dimethyl ester
(4aS,8aS)-8-Isopentyl-4,4,7,8a-tetramethyl-1,2,3,4,4a,5,6,8a- 262.2660 4.9 1-Heptadecanol, 1-acetate 298.2865 2.4
octahydronaphthalene
1,1’-Bicyclooctyl 222.2350 1.6 1-Heptene 98.1094 1.1
1,2-Dimethoxy-4-(adamantyl-1)benzene 272.1780 -2.5 1-Octene 112.1252 -0.1
1,3-di-iso-propylnaphthalene 212.1570 3.8 1-Phenanthrenecarboxylic acid, 314.2233 4.1
1,2,3,4,4a,9,10,10a-octahydro-1,4a-
dimethyl-7-(1-methylethyl)-, methy
1,3-Dimethyl-5-n-decylcyclohexane 252.2820 2.6 2,5-Cyclohexadiene-1,4-dione 108.0207 3.7
1,4,5,8-Tetramethylnaphthalene 184.1250 0.5 2-Cyclohexen-1-one, 3,5,5-trimethyl- 138.1043 1.1
10,18-Bisnorabieta-5,7,9(10),11,13-pentaene 238.1720 1.9 2-Cyclopenten-1-one, 2-methyl- 96.0572 3.1
13,15-Octacosadiyne 386.3910 3.2 2-Hexanone 100.0884 4.6
1-Naphthalenol, 1,2,3,4,4a,7,8,8a-octahydro-1,6-dimethyl-4- 222.1980 1.5 2-Naphthol 144.0569 4.4
(1-methylethyl)-
1-Nonadecene 266.2970 1.6 2-Nonenal, 2-pentyl- 210.1978 2.6
1R,2c,3t,4t-Tetramethyl-cyclohexane 140.1570 3.4 2-Propanol, 1-ethoxy- 104.0840 -2.6
1-Undecene, 9-methyl- 168.1880 0.8 3(2H)-Furanone, 5-(1,2-dihydroxyethyl)- 144.0424 -0.9
2,2,3-Triethyloxirane 128.1200 1.6 3-Octanone 128.1198 2.5
2,2’-Dimethylbiphenyl 182.1100 3.9 4,7-Methano-1H-indene, 3a,4,7,7a- 132.0939 -0.2
tetrahydro-
2H-1-Benzopyran-5-carboxylic acid, 3,4-dihydro-2-methyl-4-oxo- 206.0580 0.3 4-Methylbenzophenone 196.0884 1.9
2-Methyl-6-methyleneoct-7-en-4-one 152.1200 0.7 4-Octylphenol; p-Octylphenol 206.1675 -1.9
Benzoic acid, 4-ethoxy-, ethyl ester 194.0940 -0.6 Benzoic acid, 4-ethoxy-, ethyl ester 194.0941 0.9

6
Confirmation by CI/MS/MS Table 4. Cramer classification of compounds from four different containers exceeding 20 µg/mL
The accurate mass CI/MS/MS was concentration.
used to confirm the structures for diethyl Compound Container Cramer classification
phthalate detected in EI. Phthalates, 5,5-Diethylpentadecane 4 Class I (Low)
in general, have a missing molecular
Hexadecane 1,2,3,4 Class I (Low)
peak and a dominant fragment peak at
m/z 149. Therefore, it is a challenge to 3-Methyloctacosane 4 Class I (Low)
assign correct identification to phthalates. 1-Hexadecanol 4 Class I (Low)
The GC/Q-TOF provides not only accurate Eicosane 4 Class I (Low)
mass for the molecular and fragment cis-11-Eicosenamide 4 Class III (High)
ions, but also for MS/MS fragments, 1-Decanol, 2-hexyl- 4 Class I (Low)
which enables formula generation for 3,3,13,13-Tetraethylpentadecane 4 Class I (Low)
product ions. Figures 4A and 4B show
1-Hexacosanol 4 Class I (Low)
the CI and CI/MS/MS analyses of diethyl
2-Methylpentacosane 4 Class I (Low)
phthalate. The CI/MS/MS spectra
were interpreted (Figure 4C) using ACD Cyclotetradecane 4 Class I (Low)
software (ACD Labs, Toronto), and the 3,3-Diethylpentadecane 4 Class I (Low)
product ions of precursor at m/z 223.0937 2-Methylheptacosane 4 Class I (Low)
were identified. The characteristic Cyclopentane, undecyl- 4 Class I (Low)
fragment peak of m/z 149.0239 for 1-Octadecanol 4 Class I (Low)
phthalates was also observed (Figure 4B). 3-Methyltriacontane 4 Class I (Low)
The fragments and their accurate mass
5,5-Diethylheptadecane 4 Class I (Low)
helped to confirm the compound identity.
Octacosanol 4 Class I (Low)
3-Methylhexacosane 4 Class I (Low)
3,3-Diethylheptadecane 4 Class I (Low)
Heptadecane 1,2,3 Class I (Low)
Dodecane 1 Class I (Low)
Tetradecane 1 Class I (Low)
2,2,4,4, tetramethyloctane 1 Class I (Low)
Octadecane 3 Class I (Low)

223.0963 ×102
×103 A ([C12H14O4]+H)+ 4.5 B C Diethyl phthalate
CI MS CI MS/MS O
3.5 4.0 149.0239
O CH3 (C12H14O4)+H+
3.5
3.0 O CH3 m/z 223.0937
3.0
2.5 O
Counts
Counts

2.5
2.0
2.0 O
223.0937
1.5
1.5 OH (C8H6O4)+H+
1.0 OH m/z 167.0278
1.0
251.1288 167.0278 O
0.5 ([C12H14O4]+H2H5)+ 0.5 57.0743
0 0
O
140 160 180 200 220 240 260 280 300 50 100 150 200 250 300
Mass-to-charge (m/z) Mass-to-charge (m/z) + (C8H4O3)+H+
OH
m/z 149.0239
O

Figure 4. Structure elucidation of diethyl phthalate.

7
Toxicity evaluation match scores were stored in a custom 4. Li, K.; et al. Creating a Holistic
The concentration of compounds from library to mine CI data, providing an Extractables and Leachables (E&L)
all containers detected by both EI and CI increase of 14 % in compounds identified Program for Biotechnology Products.
mode were determined semiquantitatively in one of the samples. The CI data files PDA Journal of Pharmaceutical
using triphenyl phosphate as an internal were also used to search against an Science and Technology 2015,
standard7. In the evaluation of ophthalmic in-house database by formula containing 69590–619.
drug containers, extractables present common extractables known from
literature. Additionally, CI/MS/MS was 5. IdeaConsult Ltd. Toxtree (Estimation
at > 20 µg/mL would be considered for of Toxic Hazard—A Decision Tree
risk evaluation. In this study, compounds performed to confirm the identity of these
compounds. The Cramer classification Approach) v.2.6.13. http://toxtree.
exceeding 20 µg/mL were categorized sourceforge.net/.
based on the Cramer classification and MPP distribution plot of the detected
to determine their potential toxicity, extractables show that container 2 had
6. Lateef, S. S.; et al. Differential
assuming that the containers are being the lowest number of extractables and no
Analysis in Screening Assays for an
used for ophthalmic drug products. significant amounts of Cramer Class III
Extractables and Leachables Study
Table 4 shows the results. One high risk compounds.
Using an Agilent 7200 GC/Q-TOF
compound was found in container 4, System Combined with Data Mining
which also included many more References Software, Agilent Technologies
extractables than the other containers. 1. Mire-Sluis, A.; et al. Extractables and Application Note, publication number
Therefore, container 4 is a less suitable Leachables. Challenges and Strategies 5991-6688EN, 2016.
choice for formulation. in Biopharmaceutical Development.
BioProcess International Feb 2011. 7. Jenke, D.; et al. Utilization of
Conclusion Internal Standard Response Factors
2. Guidance for Industry, Container to Estimate the Concentration of
An Agilent 7200 GC/Q-TOF system was Closure Systems for Packaging Human Organic Compounds Leached from
used to perform qualitative screening and Drugs and Biologics. U.S. Department Pharmaceutical Packaging Systems
identify extractables from four different of Health and Human Services, Food and Application of Such Estimated
container closure systems. An extended and Drug Administration: Rockville, Concentrations to Safety Assessment.
number of identified compounds was MD, May 1999. Journal of Chromatographic Science
obtained by acquiring data in both EI 2012, 50206–212.
and CI mode. Agilent MassHunter Mass 3. Jenke, D. Development and
Profiler Professional software provided Justification of a Risk Evaluation
a versatile tool in automated data Matrix to Guide Chemical
mining workflows. Unique and common Testing Necessary To Select
compounds within the different group and Qualify Plastic Components
of samples were determined. EI spectral Used in Production Systems for
data were identified using the NIST Pharmaceutical Products. PDA Journal
14.0 library. Compounds with low library of Pharmaceutical Science and
Technology 2015, 69, 677–712.

www.agilent.com/chem
For Research Use Only. Not for use in diagnostic
procedures.

This information is subject to change without notice.

© Agilent Technologies, Inc., 2016

Published in the USA, May 1, 2016
5991-6901EN