Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis

journal homepage: www.elsevier.com/locate/jpba

Identification and quantification of ethylene oxide in sterilized

medical devices using multiple headspace GC/MS measurement
Pascal Gimeno a,∗,1 , Marie-Laure Auguste a , Vagn Handlos b,1 , Anne Mette Nielsen c ,
Stephan Schmidt d,1 , Nelly Lassu a , Martin Vogel d , Antonius Fischer d , Charlotte Brenier a ,
Françoise Duperray a
a
Agence nationale de sécurité du médicament et des produits de santé (ANSM), Direction des Contrôles (CTROL), 143/147 boulevard Anatole France, 93285
Saint-Denis Cedex, France
b
Capital Region Pharmacy, Marielundvej 25, DK-2730 Herlev, Denmark
c
Danish Medecines Agency, Denmark
d
Sanofi-Aventis Deutschland GmbH, R&D PDP Analytical Sciences, Germany

a r t i c l e i n f o a b s t r a c t

Article history: This manuscript, based on the ISO 10993-7 approach, describes a multiple HS-GC measurement of resid-
Received 10 January 2018 ual EO present in sterilized plastic samples. The quantification of EO is done, according to the ISO standard,
Received in revised form 17 May 2018 by addition of EO amounts extracted for each repeated extraction. During the method development, the
Accepted 19 May 2018
specificity of the detection of EO regarding acetaldehyde (structural isomer of EO) which may be formed
Available online 22 May 2018
from EO has been ensured and different tests were performed to check a possible influence of the sam-
ple preparation. Assays to maximize EO extraction were performed for different materials (Cyclo-olefine
Keywords:
Copolymer (COC), Cyclo-olefine Polymer (COP), Silicon, Polyurethane (PUR)) changing extraction tem-
Ethylene oxide
Sterilized medical devices
peratures and times for the headspace and the pre-thermal treatment. Results highlight that depending
ISO 10993-7 on the material, EO can be more or less retained and thus thermal extraction conditions to maximize the
Headspace/gas chromatography–mass amount of extractible EO from plastics may change accordingly. For COC syringes a validation according
spectrometry to ICH guidelines and an inter-laboratories study were performed. The method has been used for a market
Cyclo-olefine copolymer survey of EO sterilized medical devices, results obtained are reported in this manuscript.
Thermal extraction © 2018 Elsevier B.V. All rights reserved.

1. Introduction vents. When EO is heated to about 400 ◦ C, or to 150–300 ◦ C in

Ethylene oxide (EO) is mainly used as a chemical intermedi-
ate in the manufacture of ethylene glycol (used to make antifreeze
and polyester) or used as sterilant [1]. Because of its molecu-
lar structure, EO reacts with various compounds with opening of
the ring. Its typical reactions are with nucleophiles which pro-
ceed via the SN2 mechanism. This property leads to bactericidal,
fungicidal, and sporicidal disinfection. It is effective against most
micro-organisms, including viruses. The effect of EO on microor-
ganisms is obtained by the alkylation of proteins essential for cell
reproduction. It is used as a fumigant for foodstuffs and textiles
and as an agent for the gaseous sterilization of heat-labile pharma-
ceutical and surgical materials. EO is isomeric with acetaldehyde
and with vinyl alcohol and readily soluble in many organic sol-
∗ Corresponding author.
E-mail address: pascal.gimeno@ansm.sante.fr (P. Gimeno).
1
Member of Ph. Eur. group 16 “Plastic Containers for Pharmaceutical Use”.
the presence of a catalyst (Al2 O3 , H3 PO4 , etc.), it isomerizes into
acetaldehyde [2]. In absence of a catalyst, the thermal isomeriza-
tion of EO yields significant amount of by-products [3]. EO is used
as a sterilizing agent in the healthcare industry because of its non-
damaging effects for delicate instruments and devices that require
sterilization, and for its wide range of material compatibility includ-
ing silicone [4], poly-methylmethacrylate (PMMA) [5], rubbers [4]
and biodegradable polyurethanes [6] unlike other sterilization type
such as ␥-irradiation. The sterilization process involves exposing
products to ethylene oxide gas in presence of carbon dioxide, added
due to risk of explosion of pure EO and air, under vacuum in a
sealed chamber pressure vessel. EO penetrates porous packaging
and sterilizes all accessible surfaces of the product. EO can pen-
etrate multiple layers of porous packaging, making it suitable for
the sterilization of a wide range of materials not compatible with
other methods of sterilization. EO can cause cancer [7], particularly
leukemia cancer of the pancreas [8]. The effects of exposure to EO
become more severe as the exposure level increases. Exposure to
high levels of EO can cause seizures, paralysis, and coma, and dam-

https://doi.org/10.1016/j.jpba.2018.05.035
0731-7085/© 2018 Elsevier B.V. All rights reserved.
120 P. Gimeno et al. / Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127
age the liver and kidneys. Long-term exposure can cause brain and
nervous system problems and cataracts. Skin contact with EO can
cause dermatitis, blisters, and burns. The International Agency for
Research on Cancer classifies ethylene oxide into group 1, meaning
it is a proven carcinogen [9].
Requirements regarding EO residues are reported in the ISO
10993-7 standard [10]. This harmonized standard gives useful gen-
eral guidelines for the assay of residual amounts of EO, ethylene
chlorohydrin (ECH) or ethylene glycol (EG) after the sterilization
process of medical devices, it specifies EO residues limits and
highlights the necessity for manufacturer to reduce EO residues
in the sterilized products. General informations regarding ana-
lytical methods to asses EO residues are also described in this
standard. Two types of extraction are proposed depending on the
assay of total amount or leachable EO. This standard proposes
for limited or prolonged exposure devices to assay EO by sim-
ulation using water as extracting simulant and for permanent
contact medical devices (>30 days) to use an exhaustive extrac-
tion process (consecutive sample extraction). Total exhaustion is
for practical reasons defined when the EO amount extracted is
less than 10% of the first extraction or until there is no signifi-
cant increase in the cumulative residue levels detected. The EO
amount in the sample is calculated by summing the EO values
obtained for the mean peak area measurements made in each
of the sample extraction. Different publications dealing with the
assay of residual EO in medical devices have been published among
which some manuscripts focused on the assay of EO in specific
medical device or type of plastic: syringes [11], hemodialysis [12],
Intraocular lenses [13], high-density PVC-PE [14], Polyurethane
[15], Polymer [16], Polypropylene [11] or more generally polymeric
materials [17]. Most published methods used an extraction process
consistent with the ISO standard [10]: multiple HS-GC measure-
ment [13–15,17–20] or simulation extraction [12–14,16,19–23].
Two papers used non-ISO methods: colorimetric [24] or involving a
dissolution with DMF [13]. Regarding simulation methods, extrac-
tion simulants used are numerous compared with the ISO standard
(only water): HCl [16], acetone [14], biological serum [19], water
[20]. All assays are performed using gas chromatographic methods:
HS-SPME-GC/FID [18], SPME-GC/FID [19], HS-GC/MS [13–15,20],
GC/ECD [23], GC/FID [14,16], HS/GC/FID [17] except one using
photoionization PID [12]. EO residues tested on samples are very
different, some papers reports focused on traces detection allow-
ing an EO detection at ppb levels [18,19]. Depending on material
tested, a manuscript report using thermal EO extraction, points out
that re-sterilization process clearly increases the quantity of EO
residues in the devices compared to the 1st sterilization and thus
the necessity to increase the aeration period before the release
of products [25] whereas another manuscript reports that mul-
tiple re-sterilizations showed no real difference on the amount
of residual EO [15]. Some manuscripts also highlight the influ-
ence of the polymer structure (crystalline/amorphous), the size of
molecular chain or the material porosity in the caption/release of
EO during/after sterilization [15,17]. Regarding simulation meth-
ods, most papers point out that water could be considered as the
best simulant for EO extraction [14,19] including though inter-
laboratory trials [20]. Whereas, another publication tests different
simulant and highlights that depending on the matrix, water could
or not be the best simulant to be use [21]. A comprehensive
study examines a series of polymers and explore different exhaus-
tive extraction conditions (water, 3 different organic solvents or
air using thermal desorption) to determine residual EO. Results
obtained highlight that EO exhaustively extracted vary signifi-
cantly depending on the material and the extraction solvent used
and that in some cases exhaustive extraction methods using a
solvent could extract more EO than thermal desorption methods
[17].
Through the revision/elaboration of its monographs, the Euro-
pean Pharmacopoeia (Ph. Eur.) wants to improve the assay of EO
proposed for sterilized medical devices, especially for EO sterilized
Cyclo-olefine copolymer (COC) syringes. In this way, a multiple HS-
GC measurement of residual EO present in sterilized COC samples
has been developed by the French National Agency for Medicines
and Health Products Safety (ANSM), based on a method used for
market survey. During tests performed, results obtained clearly
highlight the need for different EO extraction conditions depend-
ing on the type of plastic tested. As this important aspect is few
evoked in the literature for thermal exhaustive extraction and
mentioned on an evasive way that could be misleading in the
Annex K of the ISO 10994-7:2008 standard “... heat in an oven at
an appropriate temperature for an appropriate amount of time. The
time/temperature regimen is relatively arbitrary ...”, results obtained
in the context of this study have been proposed for publication.
The quantification of EO was done, according to the ISO 10993-7
standard, by addition of EO amounts extracted for each repeated
thermal extraction until exhaustion. Different preliminary tests
were performed during the method development either to check
a possible influence of the sample preparation or to improve
EO extraction regarding headspace temperatures and pre-thermal
treatments. Thus, comparative assays were performed on differ-
ent EO sterilized materials available in the laboratory (COC, COP,
silicone, PUR . . .) using crushed and cut samples (small and large
pieces particles). For COC syringes, formerly prepared crushed sam-
ples stored in a freeze were compare to freshly prepared ones
in order to check a possible leak of EO once crushed. Compar-
ative assays to maximize the EO extraction were performed for
different materials (silicone, PUR and COC) changing extraction
temperatures and times for the headspace temperature and the
pre-thermal treatment. For COC syringes, multiple HS-GC measure-
ment were performed using extraction conditions less stringent
than conditions allowing to maximize the EO extraction (regard-
ing pre-thermal treatment or HS temperatures) in order to check
if performing more repeated extraction could lead to similar EO
amounts. At least, to ensure the relevance of the pre-thermal
extraction step for slow diffusing materials, residual EO was deter-
mined on same COP samples with and without EO pre-thermal
extraction, within the same analytical run. Once conditions maxi-
mizing the extraction of EO from COC syringes has been established,
a method validation and an inter-laboratories study were per-
formed and the method tested, without any modification, to COP
samples.
2. Experimental: materials and methods
2.1. Medical devices
Different materials were used to develop/validate and test the
method: Cyclic-olefin copolymer (COC) syringes either sterilized in
March 2015 or in November 2015 or not sterilized were used for the
method development and validation; Silicone and Polyurethane
(PUR) tubing’s sterilized in 2014 were used to test the influence of
the plastic nature on the EO extraction; Cyclic-olefin polymer (COP)
vials either formerly sterilized, sterilized mid-September 2016 or
not sterilized were assayed to ensure the relevance of the pre-
thermal extraction step used for slow diffusing materials and to
check the possibility to extend the method validated for COC to
COP samples (similar to COC material). All sterilized samples were
frozen at −20 ◦ C when received and kept at this temperature until
analysed. Samples were frozen in March 2016 for COC samples,
2015 for Silicone and PUR samples and September 2016 for COP
samples.
 P. Gimeno et al. / Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127
2.2. Drying oven/crushing machine
For a better extraction of EO from slow diffusing materials, such
as COC or COP, a crushing machine used to crush materials and a
pre-thermal extraction treatment consisting in warming the sam-
ple preparation at 120 ◦ C at least for 12 h in a drying oven before
the first injection into the HS/GC system were used. The crushing
®
machine used was a Retsch SM300 Verder Scientific and the drying
®
oven apparatus a Firlabo incubator Air Concept AC60.
2.3. Gas chromatography and mass spectrometry
All analyses were carried out on an Agilent 6890 gas chro-
matograph interfaced with an Agilent 5973 quadrupole mass
spectrometer (Santa Clara, California – USA) using a CP-SilicaPLOT
capillary column 30 m × 0.32 mm (i.d.) × 4.0 ␮m film thickness. To
prevent dislodged particles from the PLOT column to reach and
may injure the MS detector (scarring the rotors) a trap column
was installed after the column, as recommended by the supplier
of the column. The GC oven temperature was settled as follows:
100 ◦ C held for 2.0 min, ramped to 225 ◦ C at 20 ◦ C min−1 and held
for 5.0 min. The injection was performed in split mode one fifti-
eth using an headspace HP 7694 settled as follows: equilibration
temperature (120 ◦ C); equilibration time (60 min); transfer-line
temperature (130 ◦ C); carrier gas (He); vial pressurisation time
(0.5 min); loop fill (0.15 min); injection time (3.00 min) and shak-
ing mod (High agitation). The ion source temperature was settled
at 230 ◦ C, and the helium carrier gas flow rate at 1 mL min−1 . The
mass spectrometer was tuned on electron impact ionization (EI) at
70 eV. The acquisition was performed on full-scan (m/z = 10–350)
and on SIM mode. For EO quantification, m/z = 44 was used as quan-
tifier ion whereas m/z = 29 and 15 were used as qualifier ions. On
SIM mode three different ions were monitored according to litera-
ture [26] and after preliminary injections using the full-scan mode.
HS/GC–MS analytical parameters (split ratio, injection time. . .)
have been tested/optimized using standard solutions (not reported
in this manuscript).
2.4. Reference standards
Reference standard solutions are prepared extemporaneously
introducing 1.0 mL of a not previously opened commercial EO stan-
dard (50 000 ␮g/mL) into a 50.0 mL volumetric flask and diluting
with ethanol (Reference solution (a) – [C] = 1000 ␮g/mL). From the
reference solution (a) different standard solutions (b)–(d) respec-
tively at 200 ␮g/mL, 50 ␮g/mL and 10 ␮g/mL EO are prepared by
dilution with ethanol GC quality. Using reference solutions (b–d),
standard calibration vials ranging from 0.5 ␮g to 6.0 ␮g per vial are
prepared using solvent volumes ranging from 20 to 50 ␮L. Appro-
priate volumes of each standard solution are placed into 20 mL
headspace crimp-capped vials using PTFE septum and immediately
closed prior to the headspace analysis. For safety reason, ethylene
oxide (EO) standards of known and certified concentrations were
purchased from a commercial source (Supelco – ref: 48838–50
000 ␮g/mL prepared in ethanol). An alternative could be to pre-
pared EO stock solution from the pure gas compound as described
in the Ph. Eur. for “Ethylene oxide stock solution. 1036401” (not
used in this manuscript).
2.5. Sample preparations
For low released materials (COC and COP), liquid nitrogen
(Messer France, referred as “food contact” quality) has been used
to freeze samples before been crushed using a crushing machine.
Depending on the amount of EO residue present in the sample,
0.10–3.00 g of plastic are weighted into a 20 mL headspace vial after
121
having crushed (COC, COP) or cut (silicone, PUR) into pieces less
than 0.5 cm2 . The weight of sample has to be adjusted in order to
have an EO residue amount within the validated calibration range
and to limit the number of consecutive extraction to the exhaus-
tion. The vial is closed and for EO slow diffusing materials allowed
to stand at 120 ◦ C at least 12 h in a drying oven before the first
injection into the HS/GC system. The vial is then injected into the
HS/GC system, the cap removed from the vial under a hood and
the vial is purged for 30 s with dry nitrogen (Messer France – gas
purity 4.5). The cap is replaced using a new PTFE septum and the
heating and injection repeated to exhaustion. Most often, if max-
imizing extraction conditions are used (see 3.2), the exhaustion is
obtained after one or two consecutive injections. As some materials
resorb EO partially or completely as the temperature equilibrates
to room temperature, during the analysis of these materials, sam-
ples may need to be injected on to the column while they are still
hot or warm and then purged (as described above) without further
cooling.
During the method validation (see paragraph 3.6.2.5), COC
samples used were sent to SANOFI to be tested with their own
HS-GC method (inter-laboratory reproducibility). About 200 mg of
COC samples are weighed exactly and transferred into a 10 mL
headspace vial. The vial is closed with PTFE laminated septum and
3 ␮L of EO-d4-standard is added with a microliter syringe. Quan-
tification is calculated from the peak areas of the extracted ion
chromatograms of fragment m/z = 29 for the native EO and fragment
m/z = 30 for the deuterated internal standard.
3. Results and discussion
3.1. Standard calibration curves
Due to the high volatility of ethylene oxide, not previously open
commercial standards were preferably used for the preparation of
reference solutions. Attention should be paid to solvent volumes
used to perform standard solutions, volumes used should not be
higher than 50 ␮L to allow a complete evaporation of EO into the
20 mL HS vial. Comparative tests performed twice (N = 2) regarding
solvent volume seems to highlight no differences on the calibra-
tion curve as long as solvent volumes are less or equal to 50 ␮L.
For routine tests, volume ranging from 20 to 50 ␮L depending on
the calibration level were used for convenient reasons. An alter-
native could be to use the same volume for the preparation of all
standard preparation (i.e. 20 ␮L). Furthermore, the high volatility
of EO (boiling point: 10.4 ◦ C) may rise problems in the process of
preparing dilutions of the standard. The problem can be solved by
carefully minimising the free surface of the dissolved EO in contact
with the atmosphere and by handling solutions as quick as possible.
An alternative could be the use of cold vials or ice [17] (not tested
in this manuscript).
3.2. Influence of sample preparations
Comparative results performed on COC samples (N = 6) clearly
highlight for low EO released materials the necessity to crush the
sample in order to clearly improve the EO extraction (Table 1).
Whereas for soft materials like silicone (N = 2) or PUR (N = 2), no dif-
ference on the EO extraction is observed using either small or large
cut samples. These experimental results are consistent with results
already published in literature regarding the influence of the poly-
mer structure (crystalline/amorphous), the size of molecular chain
or the material porosity in the caption/release of EO during/after
sterilization [15,17]. Indeed, for EO slow diffusing materials, the
size of pieces is very important to maximize the EO released dur-
ing the extraction process. The smaller is the size the better is the
122 P. Gimeno et al. / Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127

Table 1 significant higher EO amounts compared to re-used crushed prepa-

Influence of sample preparations.
rations stored in the freezer, old crushed samples several times
/ Silicone PUR COC tested during the method validation (see 3.6.2.5) give compara-
(sterilized in March 2015) ble results pointing the need for controlled storage conditions of
Crushed Not performed Not performed 5,0 ppm crushed samples in the freezer (–20 ◦ C).
(N = 6, RSD = 2.9%)
Small cuts 0,86 ppm 0,40 ppm 2,0 ppm
3.3. Maximization of extraction conditions (Headspace and pre
(N = 2) (N = 2) (N = 6, RSD = 7.5%)
Large cuts 0,86 ppm 0,43 ppm Not performed thermal extraction)
(N = 2) (N = 2)
Comparative qualitative studies, regarding the variation of the
/ COC amount of the EO extracted (and acetaldehyde present) from COC
(sterilized in November 2015)
(Fig. 1 – N = 1–6), silicone (Fig. 2 – N = 1) and PUR (Fig. 3 – N = 1) when
Freshly prepared crushed sample 31.2 ppm increasing either headspace injection conditions or pre-thermal
(N = 6, RSD = 7.3%) treatment temperatures, have been performed in order to esti-
Stock Frozen crushed sample 23.2 ppm
(N = 6, RSD = 7.2%)
mate analytical conditions needed to maximize the amount of EO
extracted from each material. For these comparative tests, com-
mercial EO standard solutions were re-used once open (economic
reasons). Indeed, even if due to the high volatility of EO atten-
tion should be paid to the re-use of commercial EO standards once
opened, the purpose of these tests was to compare the influence of
different parameters on the EO extraction during consecutive tests
(performed most often within the same analytical run or within the
week) assuming that the loss of EO from the open commercial solu-
tions within the test could be considered as negligible. Thus may
explain slightly higher EO amounts assayed during these studies
compared with the method validation (3.6.2).
Qualitative tests performed clearly highlight the need for differ-
ent extraction conditions depending on the type of plastic tested in
accordance with literature [17]. For COC samples (Fig. 1), a higher
efficiency of the pre-thermal treatment (Fig. 1B) for the extraction
of EO residues was clearly stated compared to HS extraction con-
ditions (Fig. 1A). Indeed, for low released materials as COC, the EO
extracted increases to more than 10 ppm when we increase the

Fig. 1. Variation of the amount of the EO extracted (and acetaldehyde formed) from
COC syringes: When increasing headspace injection temperature and extraction
time (1 h and 2 h) using a fixed EO pre-thermal extraction of one night (∼12 h) in
a drying oven at 70 ◦ C (Fig. 2A); When increasing the temperature of the drying
oven (EO pre-thermal extraction – 1 night) using fixed headspace conditions (80 ◦ C
during 2 h) (Fig. 2B).

extraction – low particle size give high extraction yield for equal
time. By crushing hard plastic materials the EO initially trapped
in the material structure is released and the effect of the thermal
desorption improved.
Results obtained on COC samples sterilized in November 2015
(N = 6) highlight a loss of EO from samples once crushed even if
frozen compared to crushed samples freshly prepared (samples Fig. 2. Variation of the amount of the EO extracted (and acetaldehyde formed) from
silicone tubings: When increasing headspace injection temperature without prelim-
crushed and tested within the same day). According to these tests
inary EO pre-thermal extraction (Fig. 3A); When increasing the temperature of the
(see Table 1), some EO seems to be quickly lost after the crushing drying oven (EO pre-thermal extraction – 1 night) using fixed headspace conditions
process. Nevertheless, even if fresh crushed samples assayed give (100 ◦ C during 1 h) (Fig. 3B).
 P. Gimeno et al. / Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127 123

pre-thermal treatment temperature and up to only 5.6 ppm using
stringent headspace conditions (120 ◦ C – 2 h) pointing out the need
of a pre-thermal extraction step. Nevertheless, if the pre-thermal
treatment temperature is too high (T◦ C > 120–140 ◦ C) the amount
of EO drop down probably due to a degradation of EO in the gas
phase (Fig. 1B). This degradation of EO leads to a slight formation
of acetaldehyde. It is known that in the absence of a catalyst, the
thermal isomerization of EO is not selective, acetaldehyde or EO
may react to form significant amount of by-products [3]. Note that
ethanol may contain low residues of ethanal (acetaldehyde) up to
10 ppm. For silicone samples (Fig. 2), results highlight that the EO
amount extracted from silicone is directly linked to the headspace
temperature. Up to 160 ◦ C during 1 h the EO amount extracted
increased to 2.1 ppm, whereas, Fig. 2B highlights a degradation of
the EO extracted for a temperature higher than 100 ◦ C if this tem-
perature is applied for 1 night (>12 h) pointing out the inability to
use a pre-thermal extraction step for this type of material. These
results may be explained by a quite quick extraction of EO from sil-
icone materials leading to a degradation of EO if high temperatures
(>100 ◦ C) are applied during a long time (>1 h). The degradation of
EO was experimentally observed using standard EO solutions under
the same conditions. This degradation justify for low released mate-
rials the impossibility to use same analytical conditions (regarding
the pre-thermal treatment) for standards and samples. Unlike sil-
icone or COC, the thermal resistance of PUR material is quite low
(≤120 ◦ C). When the temperature increased above 120 ◦ C, the PUR
become yellow and melt. This physical criteria limits EO extrac-
tion conditions to T◦ C ≤ 120 ◦ C. Results obtained for PUR (Fig. 3)
shows that the EO extraction is directly linked to the headspace
temperature up to 120 ◦ C and that acetaldehyde (AA) is also formed
in parallel to the increase of EO (Fig. 3A). EO extracted and AA
formed follow both the same profile. Fig. 3B points out an increase
of acetaldehyde formed if a pre-thermal extraction of EO is per-
formed using a pre-thermal treatment one night. After 1 night even
at 100 ◦ C, the amount of acetaldehyde formed is higher than the EO
extracted. As for silicone, EO seems to be quickly extracted from this
material or at least the additional amount extracted within time
do not compensate the degradation of EO observed using standard
solutions.
3.4. Multiple HS-GC measurement for COC samples
Additional tests were performed to check if the amount of EO
extracted after multiple HS-GC measurement using extraction con-
ditions less stringent than those needed to maximize the amount
of EO extracted, could lead to similar results. As shown in Fig. 4,
it is necessary to proceed up to 4 steps to extract EO amounts
lower than 10% of the EO amount extracted during the 1st injec-
tion. These experimental conditions are not suitable for routine
tests. During tests performed simultaneously on the same sample
using extraction conditions maximizing the EO extraction from COC
samples (i.e. 120 ◦ C one night + HS-GC conditions of 120 ◦ C 1 h) the
EO amount obtained was estimated around 11 ppm after only one
extraction step (the EO amount determined during the 2nd extrac-

Fig. 3. Variation of the amount of the EO extracted (and acetaldehyde formed) from PUR tubings: When increasing headspace injection temperature (1 h) without preliminary
EO pre-thermal extraction (Fig. 4A); When increasing the temperature of the drying oven (EO pre-thermal extraction – 1 night) using fixed headspace conditions (120 ◦ C
during 1 h) (Fig. 4B).
124 P. Gimeno et al. / Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127
Fig. 4. Amount of EO extracted from COC syringes after multiple HS-GC measure-
ment using a pre-thermal extraction of 1 night at 100 ◦ C and a headspace injection
at 80 ◦ C during 1 h.
tion was <10% of the amount obtained during the 1st extraction).
Results obtained (11 ppm vs 9 ppm) point out for COC that the use
of stringent extraction conditions allows to faster accelerate the
EO extraction process and to extract slightly higher EO than after
multiple HS-GC measurement using softer less stringent extraction
conditions. According to results obtained, the method for the assay
of EO should therefore be adapted to the plastic tested. Depend-
ing on the type of plastic, multiple HS-GC measurement using less
demanding, e.g. lower temperature, extraction conditions may lead
to an underestimation of the EO residues.
3.5. Comparative EO residues measurements with and without
EO pre-thermal extraction step
To ensure the relevance of the pre-thermal extraction step used
for low released EO materials, the residual EO was determined using
a multiple HS-GC measurement on same COP samples with and
without EO pre-extraction step. To avoid any biases, these com-
parative tests (N = 3) were performed within the same analytical
run (same calibration curve/same day). Results obtained for each
COP samples tested are given in Table 2. If we compare EO amount
extracted with and without EO pre-thermal extraction step, the dif-
ference observed is for both samples approximately 20%. For COP
and COC, results obtained clearly point out the relevance of the EO
pre-extraction step (pre-thermal treatment 120 ◦ C – 12 h) to speed
up the EO extraction (1 extraction vs 4 extractions – see 3.4), to
improve the sensitivity of the method and increase up to 20% the
final EO amount extracted. Results obtained for COC samples are
quite similar to COP with an EO amount extracted of about 11 ppm
using a pre-thermal extraction step (see Fig. 1B) and only 9 ppm
with a not relevant pre-thermal extraction step (see Fig. 4).
3.6. Method validation for COC samples
3.6.1. Validation protocol
6 preparations per day of each sterilized COC syringes (March
and November 2015) were prepared according to part 2.5, during
Table 2
EO amount extracted with/without pre-extraction step.
EO amount extracted
with EO pre-thermal extraction step
Not sterilized COP <0.05 ppma
COP formerly sterilized 9.7 ppm
(N = 3, RSD = 3.3%)
COP freshly sterilized(mid-September 2016) 67.5 ppm
(N = 3, RSD = 10.2%)
a
Detection limit ∼0.05 ppm for 3.0 g sample weight.
3 non consecutive days. The not sterilized COC sample was used
to assess the method selectivity. The intra-laboratory repeatabil-
ity was tested preparing independent calibration curves, using not
previously opened commercial EO standard, and performing tests
by 2 different analysts. The method reproducibility and accuracy
was tested through an inter-laboratories study.
3.6.2. Validation results
The method validation was performed according to ICH Q2 (R1)
validation guideline (2005) using extemporaneous preparation of
all standard solutions. Intra-laboratory validation data are gather
in Table 3.
3.6.2.1. Method specificity. The specificity of the EO detection is
ensured by: the use of mass spectrometry as detection mod
(m/z = 44, 29, 15 – using ion ratios determined on standard solu-
tions); the chromatographic separation of EO (RT = 6.92 min) from
its structural isomer (same m/z ions) acetaldehyde (RT = 7.06 min);
the systematic assay of non-sterilized samples and the injection
of potential EO degradation products according to the ISO 10993-
7 (ethylene chlorohydrin which might be formed in presence of
chlorine and ethylene glycol which might be formed in presence of
water). None of both EO degradation product interfere with the EO
peak.
3.6.2.2. Limits of detection (LoD) and quantification (LoQ). The LoD
and LoQ were determined on the quantification ion in SIM mode by
injecting a 0.2 ␮g/vial and a 0.5 ␮g/vial standard solutions. LoD and
LoQ were estimated from the signal to noise ratio (s/n) based on
3.0 g of plastic sample. The LOD was set at ∼0.05 ppm (S/N ∼ 3) and
the LOQ at ∼0.1 ppm (S/N ∼ 10). These values were considered as
consistent with the sensitivity needed for market survey (i.e. health
safety issues).
3.6.2.3. Linearity/System conformity. Calibration linearity was
checked on standard solutions. The EO peak area relative standard
deviation (RSD) obtained (2.1%) for 6 consecutive injections of the
same standard solution (2.0 ␮g per vial) was <5.0% as required by
the Official Medicine Control Laboratories (OMCL) guideline for
this type of assay [27]. For each calibration curve, determination
coefficients R2 > 0.995 as well as mean relative bias (absolute
bias/calibration level × 100) in agreement with “in-house” spec-
ifications (≤10%) point out an acceptable linear correlation on
the calibration range considered (0.5–6.0 ␮g per vial). Assays
performed during linearity tests highlight for EO concentration
higher than 8.0–10.0 ␮g per vial a possible detector signal plateau
(non-linearity) possibly due to the formation of 2-methoxyethanol
a reaction product of EO and ethanol. The 0 value was not forced
through the origin. A statistical process of linearity using analytical
software (i.e. ANOVA) was performed and led to satisfactory
results.
3.6.2.4. Repeatability and intra-laboratory reproducibility. RSDs cal-
culated on different sample preparations (N = 6 on 3 days) after have
Difference
without EO pre-thermal extraction step
<0.05 ppma
7.8 ppm 19.6%
(N = 3, RSD = 6.3%)
52.7 ppm 21.9%
(N = 3, RSD = 9.0%)
 P. Gimeno et al. / Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127
Table 3
Intra-laboratory validation data.
Validation data
System suitability
(2.0 ␮g standard solution)
Limit of Detection/(LoD)
Limit of Quantification (LoQ)
Linearity
(Methacrylic acid/ISTD areas)
Mean Repeatability (n = 6 on 3 days using ANOVA)
a
Value was accepted even if considered as quiet high regarding result obtain for the inter-runs repeatability.
been processed using ANOVA are given in Table 3. Results obtained
for intra-run and inter-runs repeatability were considered as quiet
satisfactory regarding the analytical technique used. Indeed, assays
were performed using headspace injection mode, no internal stan-
dard is used and a multiple HS-GC measurement is performed (i.e.
cumulative extraction). Furthermore, due to high EO residual levels,
for COC syringes sterilized in November 2015 a low sample amount
was used (∼0.1 g).
3.6.2.5. Inter-laboratory reproducibility. A review of the analytical
method proposed for the assay of EO was kindly performed by the
Danish Medicines Agency (DMA). The work was focused on the
assay of EO trapped in COC using prefillable EO sterilized syringes.
In parallel to this review, COC samples used for the method vali-
dation and the DMA review were also sent to SANOFI to be tested
with their own HS-GC method (see 2.5). Inter-laboratory results
are given in Table 4. Results obtained during this inter-laboratories
study highlight on one hand a difference on the EO amount assayed
if we compared tests performed during the review of the method
performed by the DMA and the ANSM/SANOFI results and on the
other hand really similar results using 2 different methods (ANSM
and SANOFI). Higher EO residues determined by the DMA were
not clearly explained and were supposed to be due to different
time and temperature of the goods under transport which was not
been recorded. Regarding tests performed by SANOFI using their
own HS-GC method, results clearly highlight satisfactory values
regardless samples assayed. EO amount assayed were considered
as similar and repeatable using both HS/GC methods.
3.6.2.6. Accuracy. Intra-laboratory accuracy tests were not per-
formed. Indeed, Common Testing Samples (CTS) are not available
and spiked preparations were considered not representative of
the EO trapped in the sterilized sample. Nevertheless, similar
results obtained during the Inter-laboratory reproducibility (see
3.6.2.5) using 2 different methods (ANSM and SANOFI) actu-
ally highlight a quiet good accuracy (based on cross-validation
approach) of the method proposed (7.6 ± 0.3 ppm vs 8.0 ± 0.1 ppm
and 22.2 ± 0.7 ppm vs 24.6 ± 0.7 ppm).
3.7. Application of the method validated for COC to the assay of
COP samples
Results obtained for each COP samples tested, using the method
validated for COC, are given in Table 2. Results (EO amounts and
125
injection repeatability:
Retention time precision 0.1%
Peak area repeatability 2.1%
Resolution
Ethylene Oxide/Acetaldehyde >2.0
Estimated from the 0,2 ␮g per vial calibration curve (S/N ∼ 3) 0,05 ppm (for 3.0 g weighted)
Estimated from the 0,5 ␮g per vial calibration curve (S/N ∼ 10) 0,1 ppm (for 3.0 g weighted)
Linearity range 0.5–6.0 ␮g per vial
Determination coefficient R2 >0.995
Biases ≤10%
Not sterilized COC syringe <0.05 ppm
RSD (intra-run)
COC syringe sterilized in March 2015 3.3%
COC syringe sterilized in November 2015 8.1%a
RSD (inter-runs)
COC syringe sterilized in March 2015 6.2%
COC syringe sterilized in November 2015 7.5%
RSD) determined for COP samples using a pre-thermal extraction
step are consistent with results obtained during the method vali-
dation for similar sterilized COC samples. The method developed
and validated for the assay of residual EO in COC samples may
directly be applied without any modification to the assay of EO
in COP samples.
3.8. Results on 70 medical devices controlled during a market
survey
Information provided by several neonatal wards in France
highlight the use of EO as sterilizing agent for almost 85% of single-
use sterile devices (especially tubes and catheters for medicinal
or nutrition product administration). This sterilization technique
requires from manufacturer the need to manage ethylene oxide
(EO) residuals in devices after the sterilization process to avoid
patient exposure. The ISO 10993-7 standard mentioned require-
ments relating to EO sterilization residuals in medical devices and
specifies the allowable EO residual limits that manufacturers must
ensure. The level of exposure to these substances should be as
low as possible. During a market survey performed between 2014
and 2015, the ANSM checked the good implementation of the
ISO 10993-7 standard regarding EO residues focused on medical
devices used in service of neonatology and paediatric clinics [28],
a total of 70 medical devices were controlled. Results obtained
point out the use of 3 plastics (PUR, Polyvinylchloride (PVC) and
silicone) in tubings and 2 additional plastics (Acrylonitrile Butadi-
ene Styrene (ABS) or Methylmethacrylate Acrylonitrile Butadiene
Styrene (MABS)) for connection parts. Among sample tested, some
of them do not clearly mention the nature of the plastic used for
connections. The repartition of plastics used in tubing and con-
nections are given in Fig. 5. The chart given in Fig. 6 shows the
samples distribution regarding EO residues. When several parts of
a sample were tested, the sample result reported in the graph cor-
responds to the highest part content. From an analytical point of
view results clearly highlight a higher EO residue level in connec-
tions parts compared to tubbings. Indeed, Whereas EO residues in
tubbings are not higher than 11.0 ppm (most of them <3.0 ppm), EO
residues in connections parts could reach more than 800 ppm. This
difference could be explained by the nature of the plastic used. For
tubing’s, soft plastics are used whereas for connections hard plas-
tics parts are used which release EO slowly compared to more soft
materials used for tubing’s [25]. More generally speaking, results
obtained during this market survey also showed discrepancies in
126 P. Gimeno et al. / Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127
Table 4
Inter-laboratory reproducibility.
Not sterilized COC syringe
COC syringe sterilized in March 2015
COC syringe sterilized in November 2015
Fig. 5. Plastic repartition in tubbings and connections.
the implementation of standard NF EN ISO 10993-7 on EO resid-
uals. Most manufacturers did not take into account a neonate low
birth weight or the concomitant use of other EO-sterilized devices
in their calculation of the allowable residue limits. The ANSM there-
fore issued a reminder of current legislation to manufacturers and
sterilization companies in July 2014 and provided a framework for
the market authorization of medical devices used in neonates and
infants. Furthermore, during this survey, the ISO 10993-7 stan-
dard also appeared unclear from an analytical point of view on
EO amounts
DMA ANSM SANOFI
Not detected Not detected Not detected
12.0 ppm 7.6 ± 0.3 ppm 8.0 ± 0.1 ppm
(N = 3, RSD = 5.9%)
35.9 ppm 22.2 ± 0.7 ppm 24.6 ± 0.7 ppm
(N = 1)
Fig. 6. EO residual amount in tubbings and connection parts.
the way how to proceed to extract EO residues (extraction by sim-
ulation/exhaustive extraction, extraction conditions. . .) and from
a health safety aspect on the way to established residuals lim-
its. Therefore, different technical and general comments regarding
either analytical methods or human health safety were proposed
by the ANSM to the ISO group to modify the ISO standard during a
revision process. As few remarks were implemented, in September
2015, the ANSM took a human health care decision regarding the
ISO 10993 standard to des-harmonized this standard. Since March
2016 a compliance to the ISO requirements is not considered in
France as sufficient to ensure human health safety. Manufactur-
ers have also to comply with additional requirements given in a
French human health care decision [29]. This survey participated to
a more comprehensive approach which aim is to reduce the patient
exposure to CMR substances, without challenging the use of EO as
sterilizing agent.
4. Conclusion
Through the revision/elaboration of its monographs, the Euro-
pean Pharmacopoeia (Ph. Eur.) wants to improve the method
proposed to assay EO in sterilized medical devices, especially for
Cyclo-olefine copolymer (COC) syringes. In this way, a multiple HS-
GC measurement of residual EO had been developed by the ANSM.
 P. Gimeno et al. / Journal of Pharmaceutical and Biomedical Analysis 158 (2018) 119–127
The quantification of EO is done according to the ISO 10993-7 stan-
dard by addition of EO amounts extracted for each repeated thermal
extraction. Depending on the type of material and on the amount of
EO residues present in the sample, 0.1–3.0 g of plastic are weighted
after having crushed or cut into an HS vial. The vial is closed and for
EO slow diffusing materials allowed to stand at 120 ◦ C at least 12 h
in a drying oven before the first injection into the HS/GC system.
Comparative results on the sample preparation clearly highlight the
necessity to crush low released materials as COP or COC to improve
the EO extraction. Differences observed between samples crushed
and cut reach +250% of EO extracted. For fast released materials
as silicone, no differences on the EO extraction was observed using
either crushed, or cut samples. The method should be adapted to the
plastic tested. Indeed, depending on the type of plastic, multiple HS-
GC measurement using less demanding extraction conditions may
lead to an underestimation of the EO residues. For low released EO
materials such as COP and COC samples, results obtained point out
the relevance of an EO pre-extraction step (120 ◦ C – 12 h) in order to
speed up the EO extraction, to improve the sensitivity of the method
and to increase up to 20% the final EO amount extracted. Whereas,
for other materials such as silicone and PUR, a direct extraction
of EO (without using pre-thermal treatment) should preferably
be used. Regarding the HS/GC–MS chromatographic method, the
specificity of the detection of EO regarding acetaldehyde which may
be formed from EO has been ensured. For each calibration curve,
determination coefficients R2 > 0.995 as well as mean relative bias
in agreement with “in-house” specifications (≤10%) point out an
acceptable linear correlation on the calibration range considered
(0.5–6.0 ␮g per vial). Results obtained during the method validation
highlight an acceptable sensitive and repeatability of the method
proposed. LoD and LoQ were respectively determined at ∼0.05 ppm
and ∼0.1 ppm with an intra-run and inter-runs repeatability quiet
satisfactory regarding the analytical technique used (headspace
injection mode, no internal standard and a multiple HS-GC mea-
surement). Between laboratories reproducibility tested using 2
different methods highlight a quiet good accuracy of the method
proposed based on a cross-validation approach (7.6 ± 0.3 ppm vs
8.0 ± 0.1 ppm and 22.2 ± 0.7 ppm vs 24.6 ± 0.7 ppm).
Acknowledgements
Authors wish to gratefully acknowledge Jürgen Thürk and Horst
Koller from Schott who kindly supplied us on COC and COP samples.
Authors also wanted to thanks all experts and the secretariat of the
Ph. Eur. group 16 who give their opinion on the method proposed
and advised us.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.jpba.2018.05.035.
References
[1] Agency for Toxic Substances and Disease Registry (ATSDR), Toxicological
Profile for Ethylene Oxide, U.S. Public Health Service, U.S. Department of
Health and Human Services, Atlanta, GA, 1990, December. https://www.atsdr.
cdc.gov/toxprofiles/TP.asp?id=734&tid=133. (Accessed 10 April 2018).
[2] Gunther Prauser, Guillermo Fischer, Kurt Dialer, Acetaldehyde as main
product of the oxidation of ethylene oxide on a silver catalyst, Angew. Chem.
92 (1980) 389.
[3] P.V. Zimakov, O.H. Dyment, Ethylene Oxide, Khimiya, 1967, pp. 57–85
(Chapter II).
[4] E. Gautriaud, K.T. Stafford, J. Adamchuk, et al., Effect of sterilization on the
mechanical properties of silicone rubbers, Bio. Process. Int. (April) (2010),
http://www.bioprocessintl.com/manufacturing/monoclonal-antibodies/
effect-of-sterilization-on-the-mechanical-properties-of-silicone-rubbers-
239379/. (Accessed 10 April 2018).
127
[5] T.J.A.G. Münker, S.E.C.M. van de Vijfeijken, C.S. Mulder, et al., Effects of
sterilization on the mechanical properties of poly(methylmethacrylate) based
personalized medical devices, J. Mech. Behav. Biomed. Mater. 81 (2018)
168–172.
[6] K. Gorna, S. Gogolewski, The effect of gamma radiation on molecular stability
and mechanical properties of biodegradable polyurethanes for medical
applications, Polym. Degrad. Stab. 79 (2003) 465–474.
[7] IARC Monographs on the Evaluation of Carcinogenic Risks to Humans,
1,3-Butadiene, Ethylene Oxide and Vinyl Halides (Vinyl Fluoride, Vinyl
Chloride and Vinyl Bromide), vol. 97, International Agency for Research on
Cancer, Lyon, 2008, pp. 185–287, ISBN 978-92-832-1297-3.
[8] Agency for Toxic Substances and Disease Registry, U.S. Public Health Services,
Toxicological Profile for Ethylene Oxide, 1990, https://www.atsdr.cdc.gov/
ToxProfiles/tp137.pdf. (Accessed 10 April 2018).
[9] IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Some
Industrial Chemicals, vol. 60, International Agency for Research on Cancer,
Lyon, 1999, ISBN 978-92-832-1297-3.
[10] ISO 10993-7:2008 – Biological Evaluation of Medical Devices – Part 7:
Ethylene Oxide Sterilization Residuals.
[11] C. Chioua, J. Calop, P. Leglise, et al., Analysis of residual ethylene oxide in
disposable syringes, Ann. Pharm. Fr. 46 (3) (1988) 185–195.
[12] K. Hoy, V. Handlos, Release of gas residues from an ethylene oxide sterilized
haemodialyser, Biomaterials 4 (4) (1983) 321–322.
[13] N.T. Lao, R.R. Rajan, H.F. Bushar, Residual ethylene oxide determination in
polymethyl methacrylate disks and intraocular lenses by headspace gas
analysis and N, N-dimethylformamide extraction, J. Pharm. Sci. 78 (1) (1989)
40–43.
[14] N.T. Lao, H.T. Lu, A. Rego, et al., Interlaboratory comparison of analytical
methods for residual ethylene oxide at low concentration levels in medical
device materials, J. Pharm. Sci. 84 (5) (1995) 647–655.
[15] A.D. Lucas, K. Merritt, V.M. Hitchins, et al., Residual ethylene oxide in medical
devices and device material, J. Biomed. Mater. Res. B Appl. Biomater. 66 (2)
(2003) 548–552.
[16] K.M. Wojcik-O’Neill, M. Ello, Equivalency of hydrochloric acid and distilled
water extraction media for determining residual ethylene oxide in medical
devices, J. Pharm. Sci. 80 (8) (1991) 783–784.
[17] A.D. Lucas, C. Forrey, D.M. Saylor, et al., Solvent or thermal extraction of
ethylene oxide from polymeric materials: medical device considerations, J.
Biomed. Mater. Res. B Appl. Biomater. (2017), http://dx.doi.org/10.1002/jbm.
b.34052.
[18] M.P. DiCicco, B. Lang, T.I. Harper, Optimization of a novel
headspace-solid-phase microextraction-gas chromatographic method by
means of a Doehlert uniform shell design for the analysis of trace level
ethylene oxide residuals in sterilized medical devices, Biomed. Chromatogr.
23 (6) (2009) 647–657.
[19] T. Harper, L. Cushinotto, N. Blaszko, et al., Round-robin evaluation of a
solid-phase microextraction-gas chromatographic method for reliable
determination of trace level ethylene oxide in sterilized medical devices,
Biomed. Chromatogr. 22 (2) (2008) 136–148.
[20] D.E. Marlowe, N.T. Lao, A.R. Eaton, et al., Interlaboratory comparison of
analytical methods for residual ethylene oxide in medical device materials, J.
Pharm. Sci. 76 (4) (1987) 333–337.
[21] A.D. Lucas, M.E. Stratmeyer, Extraction and stability of ethylene oxide residue
in medical devices, Biomed. Instrum. Technol. 42 (1) (2008) 76–79.
[22] D.T. Centola, K.I. Ayoub, N.T. Lao, et al., Variables affecting simulated use
determination of residual ethylene oxide in medical devices, J. AOAC Int. 84
(2) (2001) 512–518.
[23] H. Kikuchi, A. Nakamura, K. Tsuji, Gas chromatographic determination with
electron capture detection of residual ethylene oxide in intraocular lenses, J.
Assoc. Off. Anal. Chem. 71 (5) (1988) 1057–1062.
[24] X. Liu, D. Zheng, Y. Geng, et al., Discussion about the sampling positions of the
bag-type infusion sets for single use in the ethylene oxide residues detection,
Chin. J. Med. Instrum. 38 (1) (2014) 68–70.
[25] I. Buben, V. Melichercíková, N. Novotná, et al., Problems associated with
sterilization using ethylene oxide. Residues in treated materials, Cent. Eur. J.
Public Health 7 (4) (1999) 197–202.
[26] U.S. Department of Commerce, National Institute of Standards and Technology
(NIST), WebBook of Chemistry. https://webbook.nist.gov/chemistry/.
[27] OMCL Network of the Council of Europe, Quality Assurance Document,
PA/PH/OMCL (16) 17 R, Qualification of Equipment Annex 2: Qualification of
GC Equipment. https://www.edqm.eu/fr/guidelines-management-qualite-
deqm-86.html. (Accessed 10 April 2018).
[28] ANSM, Rapport: Dispositifs médicaux utilisés en néonatalogie et pédiatrie et
stérilisés à l’oxyde d’éthylène. Mise en œuvre de la norme NF EN ISO 10993-7,
2015, Octobre. http://ansm.sante.fr/S-informer/Points-d-information-Points-
d-information/Dispositifs-medicaux-utilises-en-neonatalogie-et-en-
pediatrie-sterilises-a-l-oxyde-d-ethylene-Mise-en-aeuvre-de-la-norme-NF-
EN-ISO-10993-7-Point-d-Information. (Accessed 10 April 2018).
[29] Décision de police sanitaire fixant les conditions particulières de mise sur le
marché et de distribution de certains dispositifs médicaux stérilisés à l’oxyde
d’éthylène, ANSM, 2015, 10 septembre. http://ansm.sante.fr/S-informer/
Points-d-information-Points-d-information/Dispositifs-medicaux-utilises-
en-neonatalogie-et-en-pediatrie-sterilises-a-l-oxyde-d-ethylene-Mise-en-
aeuvre-de-la-norme-NF-EN-ISO-10993-7-Point-d-Information. (Accessed 10
April 2018).