CLIN. CHEM.

20/7, 820-824 (1974)

Gas-Chromatographic Determination

of Ethylene Glycol in Serum

R. L. Peterson1 and D. 0. Rodgerson

We have developed a method for detection and Several chemical methods have been reported for
quantitation of ethylene glycol in serum, based on estimating ethylene glycol (6-10). These methods
the gas chromatography of its dibenzoate ester. No depend on the quantitative oxidation of ethylene
special reagents or equipment are required. The glycol to formaldehyde by periodic acid. Either peri-
method is virtually free of interference and sensitive
odic acid remaining is titrated (8, 10) or the formal-
to less than 0.02 mg of ethylene glycol per milliliter.
The total analysis time is about 90 mm. dehyde produced is condensed with a color-devel-
oping reagent (6, 7, 9). Neither type of assay is spe-
cific for ethylene glycol. It has been reported (7) that
.The tOxicity of ethylene glycol for humans is well
known (1-4). This compound is relatively rarely in- one method, in which color is developed by reaction
of a dye with formaldehyde, gave false positives in
gested, but the consequences of its ingestion are
23% of all sera examined.
drastic (1, 2). However, if the toxic agent can be iden-
Gas-chromatographic methods have also been re-
tified soon enough, therapeutic measures can be
taken to prevent the severe organ damage or death ported (11-13), in two of which (11, 12) special col-
that otherwise can occur (2). umn-packing materials are used; these methods have
not been adapted to the analysis of biological mate-
Ethylene glycol intoxication proceeds through
three stages (1-4). Up to 12 h after ingestion, the pa-
rials. A third procedure (13) employs “Porapak Q,”
a material often used in gas chromatography of alco-
tient’s symptoms resemble those of ethanol intoxica-
hols. We found that the flame ionization detector re-
tion or of diabetic ketosis. In the second stage, car-
sponse to underivatized ethylene glycol is too low to
diopulmonary complications arise. This period lasts
permit its direct estimation by gas chromatography.
from 12 to 24 h. Evidence for kidney damage is seen
The discovery that ethylene dibenzoate can be
during the third stage, which extends from 24 to 36
chromatographed on “OV- 17,” a column-packing
h. Death may occur during any of these three stages
material often used in the gas chromatography of a
if sufficient ethylene glycol has been drunk.
wide variety of drug materials (14), has prompted us
The metabolism of ethylene glycol begins with an
to develop a method exploiting this property. Only
oxidation step catalyzed by alcohol dehydrogenase
40 t1 of serum is required in this procedure. Analysis
(5) and terminates in the production of oxalic acid
time is about 90 mm.
(1,2). Because it is a relatively poor substrate of al-
cohol dehydrogenase, the production of oxalic acid Materials and Methods
and other metabolites of ethylene glycol can be
Apparatus
avoided by infusing the patient with ethanol (2).
Forced diuresis and hemodialysis may also be used A Model 900 gas chromatograph (Perkin-Elmer
to renrove the glycol (2, 3). Detection of ethylene Corp., Norwalk, Conn. 06852) equipped with a dual-
glycol in the first stage of intoxication is essential, flame ionization detector was used in these determi-
because organ damage may be irreversible in the see- nations. It was attached to a “Chromocorder,”
Ond and third stages (1-3). Model 8700 (Barber-Colman Chromatography Divi-
sion, Nuclear Chicago Corp., Des Plaines, Ill. 60018).
Mixing was done with a “Van-Whirl Vortex
Department of Pediatrics, University of Colorado Medical Cen- Mixer” (Van Waters and Rogers Scientific, Bronx,
ter, Denver, Cob. 80220.
1 Present address: Bristol Hospital, Brewster Rd., Bristol, Conn. N.Y. 10452, Cat. No. 58810-006).
06010. A GLC-1 centrifuge (Ivan Sorvall, Inc., Norwalk,
Received Oct. 17, 1973; accepted May 9, 1974. Conn. 06852) was used.

820 CLINICAL CHEMISTRY, Vol. 20, No. 7, 1974

 All spectrophotometnic measurements were per- fuge to separate the phases. Decant all the liquid
formed with a Model 300-N spectrophotometer (Gil- phase and re-extract the solid phase with another 0.5
ford Instrument Laboratories, Inc., Oberlin, Ohio ml of ethanol. Combine the ethanol extracts in a
44074). 15-mi conical stoppered centrifuge tube and evapo-
rate just to the point of dryness under a stream of ni-
Reagents trogen at 30-35 #{176}C.
Unless otherwise noted, all chemicals were AR Dissolve the residue in 100 il of pynidine and add
grade (J. T. Baker Chemical Co., Phillipsburg, N. J. 15 il of a 5 mg/mI solution of ct-naphthol in pyni-
08865). dine. Finally, add 40 il of benzoyl chloride, mix, and
Ethylene glycol, stock solutions. The hygroscopicity heat 5 mm at 60 #{176}C, cool, add 50 l ethanol, and re-
of this compound makes it mandatory to prepare heat 5 mm at 60#{176}C.
stock solutions immediately upon opening the con- Add 400 zl of chloroform and 2 ml of dilute (0.1
tainer. All aqueous or serum solutions were prepared mol/liter) hydrochloric acid to the pyridine solutions
from an aqueous stock of 111 mg of ethylene glycol and shake for 1 mm to extract. Separate the layers
per milliliter, stored at -20 #{176}C.
A solution in pyri- by centrifugation and decant the aqueous (upper)
dine of 4.44 mg/mi ethylene glycol was stored tightly layer.
sealed in the presence of a desiccant at room temper- Inject 4 to 7 izl of the chloroform layer onto an
ature. OV-17 column, at an oven temperature of 210 #{176}C.
a-Naphthol, 5 mg/mi. The solid was dissolved in Set the injector and manifold temperatures at 310 #{176}C
the appropriate amount of pyridine and stored tight- and the detector temperature at 320 #{176}C.
The carrier
ly sealed in the presence of a desiccant at room tem- gas is nitrogen, regulated to a flow rate of 25 ml/
perature. Solutions of this compound are stable for mm.
no more than a week, at which time they develop an The procedure used in the reference method has
amber color. been described (6).
The monohydrate of magnesium sulfate was pre-
pared by heating the heptahydrate 2 or more hours Results
at 180 #{176}C.
The anhydrous salt may also be used. Chromatograms of a processed serum sample, with
The column packing material used for gas chroma- and without added ethylene glycol, are illusttated in
tography was a pre-coated packing consisting of 3% Figure 1. The relative retention time of ethylene di-
by weight of “OV-17” on 100 and 120 mesh “Gas benzoate compared to that of a-naphthyl benzoate
Chrom Q” (both from Applied Science Laboratories, was determined in 70 separate trials by using ex-
Inc., State College, Pa. 16801). Silanized glass wool tracts of sera containing from 0.11 mg to 2.78 mg of
was obtained from the same supplier. The OV-17
column packing material was packed in a 1.83 X 20
mm (i.d.) coiled glass column (Perkin-Elmer Corp.).
Reagents used in the reference method were pre- B
A
pared as described by Kirela (6). Basic fuchsin dye
was a product of Allied Chemical Corp., Morristown,
N. J. 07960.
Serum specimens were obtained from the discards
of the routine chemical laboratory. All had been
stored refrigerated and were used two days on less
after collection. A portion of each specimen was used
to dilute ethylene glycol to a known concentration.
From this point on, specimens with or without added
ethylene glycol were stored frozen. The 50 specimens
were from patients of ages ranging from six months
to 68 years, in various states of health.
Each serum specimen, with and without added
ethylene glycol, was processed by the proposed and
reference assay methods. Added ethylene glycol
ranged in final concentration from 0.111 mg/mI to
2.78 mg/mi.
0123456789 0123456789

Procedure TIME (mm)

To a 40-zl serum specimen contained in a 10 X 75 Fig. 1. Gas chromatography of benzoylated ethanol ex-
tracts of serum (A) and serum containing ethylene glycol
mm glass tube, add 0.5 ml of ethanol. Disperse well (B)
on the mixer. Add 75 mg of magnesium sulfate mo- 3% OV-17 on 100-120 mesh Gas Chrom 0; 1.83 m X 20 mm (id.)
nohydrate and mix again. Allow this mixture to coiled glass column; oven temperature, 210 ‘C; attenuation, x 32. Ethyl-
ene dibenzoate emerges at approximately 4.4 mm; cr-naphthyl benzoate
stand 5 mm at room temperature, remix, and centni- emerges at 5.9 mm

CLINICAL CHEMISTRY, Vol. 20, No. 7, 1974 821

ethylene glycol per milliliter and a pyridine solution
containing 1.11 mg of ethylene glycol per milliliter. C

The mean and standard deviation were 0.790 ± C

0.012 (range, 0.78 to 0.83). Lower concentrations give

longer retention times.
A calibration curve was constructed by plotting C5
the ratio of the ethylene dibenzoate peak height to
the a-naphthyl benzoate peak height obtained from CC
N

the chromatogram of each processed serum specimen 0

against that serum’s ethylene glycol concentration a
C

(Figure 2). Each point is the mean of separate deter-

minations on five different sera containing the same l.0
concentration of glycol. The curve of absorbance at Serum Ethylene Glycol, mg/mi
555 nm vs. ethylene glycol concentration, obtained by
Fig. 2. Standard curve for ethylene glycol in serum by
applying the reference method (6) to the same serum
gas chromatography of its dibenzoate ester
specimens, is shown in Figure 3. Precise timing of the Each point represents the mean of determinations on five different serum
addition of the oxidizing and color development re- specimens containingthe same concentrationof ethylene glycol. The
bars indicate the range of the results
agents and of reading of absorbances was necessary,
because the final absorbance readings did not stabil-
ize after the incubationperiod recommended in the
reference method (6).
All sera to which no ethylene glycolhad been -.-- Aqueous standards
added were carriedthrough both procedures.In the
#{149} Serum specimens
gas-chromatographic method, a peak that co-chro-
matographed with authentic ethylene dibenzoate E
C
U)
was considered a positive response. The apparent U,
U,
concentration of serum ethylene glycol was comput- a
ed by comparing the ratio of the height of that peak a
U
0
C
to the height of the a-naphthyl benzoate peak to the a
.0
0
same ratio for a pyridine solution containing 1.11 mg a
.0
of ethylene glycol per milliliter. Any absorbance at 4
555 nm in the reference method procedure was simi-
larly considered a positive response. This absorbance
was converted to an apparent ethylene glycol con-
0.0
centration by referring to the aqueous standard 0.0 0.4 0.8

curve. Results of these determinations are shown in Ethylene Glycol, mg/mI

Table 1. Fig. 3. Determination of ethylene glycol in serum by the
The recoveryofethyleneglycoladded to the serum reference method
specimens was evaluated in both procedures. In the Each point on the serum curve represents the mean of determinations on
five or more different serum specimens, containingthe same finalethyl-
gas-chromatographic procedure, the recovery was ene glycol concentration. The points on the aqueous curve are single de-
evaluated by comparing the ethylene dibenzoate terminations
peak heightto the a-naphthyl benzoate peak height
for extracted serum samples to the same ratio for a
pyridine solution containing 1.11 mg of ethylene gly- the aqueous standard curve. No corrections were
col per milliliter. Recovery of ethylene glycol from made for positive responses in serum containing no
serum by the reference method was calculated from glycol in either method, because this would be im-
possible in practice. The mean and standard devia-
tion for any one ethylene glycol concentration repre-
sent single determinations on five different sera. The
results are given in Tables 2 and 3.
Table 1. Response of Assay System to Sera A single serum specimen, containing 0.833 mg of
Containing No Ethylene Glycol ethylene glycol per milliliter, was processed by the
Ethylene glycol Reference method Proposed method gas-chromatographic method in 20 replicate deter-
found, mg/mI No. , %oftotal No. %of total minations. The recovery, evaluated as described
0 33 66 5 10 above, was 72 ± 7%, with a range of 63 to 86%.
<0.02 5 10 45 90 When the final chloroform extract from one of the 20
0.02 to 0.10 10 20 0 0 replicates was injected into the gas chromatograph2o
0.12 1 2 0 0 times, the recovery was 70.9 ± 0.4%, with a range of
0.26 1 2 0 0 70.2 to 71.6%. The ratio of peak heights for a 1.11
mg/ml pyridine solution of ethylene glycol, per-

122 CLINICAL CHEMISTRY, Vol. 20, No. 7, 1974

 Table 2. Recovery Studies, Gas-Liquid Table 3. Recovery Studies,
Chromatographic Method Reference Colorimetric Method
Ethylene glycol Ethylene glycol
concn, mg/mi Mean ± SD, % Range, % concn, mg/mi Mean ± SD, % Range, %
0.111 55±9 40.8to61.6 0.111 108±36 80to163
0.333 68 ± 6 57.5 to 73.5 0.333 120 ± 9 111 to 132
0.555 67 ± 4 62.2 to 73.2 0.555 115 ± 11 95.5 to 123
0.833 64 ± 5 54.8 to 66.9 0.833 110 ± 8 100 to 119
1.11 70 ± 2 66.5 to 71.5 1.11 101 ± 4 97.3 to 106
1.39 63±2 62.3 to67.6 1.39 117±5 liOto 134
1.67 71 ± 5 65.2 to 77.5 1.67 104 ± 4 100 to 106
1.94 74±5 67.4to80.6 1.94 108± 3 106 to 112
2.22 70 ± 6 64.4 to 78.5 2.22 109 ± 5 105 to 117
2.78 63±5 62.1to72.4

formed in 10 separate determinations on five differ- drug-screening method of Barrett (14), in which
ent days was 1.207 ± 0.021 with a range of 1.167 to OV-17 packing material is used, column tempera-
1.240. tures are several degrees higher or lower than 210 #{176}C.
Several extraction procedures were tried in addi- Furthermore, chloroform extracts of at least 1 ml of
tion to the one described. Substitution of methanol serum must be concentrated 200-fold to give a detect-
or 1-butanol for ethanol decreased both recoveries able response (14). Most of the drugs detectable in
and precision. Precipitation of serum proteins by serum are unlikely to react with benzoyl chloride. A
tungstic acid or trichloroacetic acid followed by ex- lower limit of 0.02 mg/mi for the detection of ethylene
traction with methanol, ethanol or 1-butanol also glycol by this method must be accepted until a means
gave worse recoveries and poorer precision. is devised to eliminate the peak on the chromato-
We unsuccessfully searched for an internal stan- grams of serum extracts that mimics ethylene diben-
dard that could be added to serum prior to extrac- zoate.
tion. The benzoate esters of methanol, ethanol, 1- Reagent stability is poor for the reference method
propanol, 1-octanol, glycerol and benzyl alcohol are (6), as has been pointed out (7). In contrast, benzoyl
not detectable on OV-17 at several temperatures. chloride, pyridine, and anhydrous magnesium sulfate
The benzoate ester of propylene glycol has a reten- or its monohydrate may be stored indefinitely if pre-
tion time of 0.73, relative to that of a-naphthyl ben- cautions are taken to protect them from moisture. As
zoate, at several temperatures. It can easily be dis- was pointed out, pyridine solutions of a-naphthol
tinguished from ethylene dibenzoate, but the overlap must not be stored for more than one week.
of the two chromatographic peaks is too extensive to Conversion of ethylene glycol to its dibenzoate
allow accurate quantitation of either in the presence ester results in a very considerable amplification of
of the other. the flame ionization detector signal. We could not
The recoveries of ethylene glycol extracted from detect ethylene glycol in serum by gas chromatogra-
serum can be normalized by using a standard con- phy with the use of a flame ionization detector, even
sisting of ethylene glycol contained in a serum ma- in concentrations exceeding 1.0 mg/mi, on Porapak Q
trix. When sera containing known concentrations of at the temperature recommended by Brown et al.
ethylene glycol are compared to one another,, the ap- (13). Less than 45 ng of ethylene glycol is detectable
parent recoveries average 98% with the expected as its dibenzoate ester by the proposed method.
high coefficient of variation (11%). The greatest disadvantage of the proposed method
is its relatively poor precision, undoubtedly originat-
Discussion ing in the extraction step. Evidence for this claim is
The linearity of response of the proposed method found in the comparison of results from 20 replicate
to ethylene glycol, its relative freedom from interfer- determinations on a single serum containing ethyl-
ences, the availability and stability of the reagents ene glycol with those from 10 repetitions of the ben-
used, and the simplicity of the procedure are its zoylation of ethylene glycol in a pyridine solution.
major advantages. Its most severe limitation is the The coefficient of variation was nearly 10% in the
rather poor precision. first case, but less than 2% in the second case, which
Besides the several alcohols tested for their possi- avoided the extraction step.
ble interference in the proposed method, one can Coupled with this precision is the apparently in-
reasonably eliminate virtually all the common drugs complete recovery of ethylene glycol from serum. It
as potential interferences. In the gas-chromatographic is unlikely that this result is caused by incomplete

CLINICAL CHEMISTRY, Vol. 20. No. 7, 1974 823

benzoylation of the glycol, because in that case a lin- prove to be a useful addition to the clinical toxicolo-
ear response to ethylene glycol would not be expect- gist’s battery of procedures.
ed over a 25-fold concentration range. Furthermore,
a strongly exothermic reaction was always observed References
when excess ethanol was added to the reaction solu- 1. Friedman, E., Greenburg, J., Merrill, J., and Dammin, G.,
tion just before the chloroform extraction. Residual The consequences of ethylene glycol poisoning. Amer. J. Med. 32,
benzoyl chloride was apparently still present after 891(1962).
2. Picconi, L., A new approach to the management of ethylene
reaction with the serum extract. glycol (antifreeze) poisoning. Amer. J. Hosp. Pharm. 23, 226
Removal of ethylene glycol from the aqueous and (1966).
proteinaceous medium of serum is necessary, be- 3. Joly, J. B., Hault, G., Frossard, C., Thieffry, S., and Fabiani,
cause the water and protein concentrations in serum P., Intoxication grave par l’ethylene glycol. Presse Med. 75, 2041
(1967).
relative to even the most lethal serum concentrations 4. Curry, A. S., Oxalates. In Advances in Forensic and Clinical
of ethylene glycol would render its benzoylation im- Toxicology. The Chemical Rubber Co., Cleveland, Ohio, 1972, p
possible. Nonaqueous solvents of ethylene glycol 161.
other than alcohols are almost non-existent. By trial 5. von Wartburg, J. P., Bethune, J. L., and Vallee, B. L., Human
liver alcohol dehydrogenase: Kinetic and physicochemical proper-
and error, we found ethanol to be the optimum sol- ties. Biochemistry 3, 1775(1964).
vent and’that the addition of magnesium sulfate mo- 6. Kirela, E. W., Ethylene glycol. In Manual of Analytical Toxi-
nohydrate to the ethanolic extraction solution im- cology. Irving Sunshine, Ed. The Chemical Rubber Co., Cleve-
proved the recovery of glycol. land, Ohio, 1971, p 157.
7. Harger, R. N., and Forney, R. B., A simple method for detect-
An ideal procedure for the extraction of ethylene ing and estimating ethylene glycol in body materials. J. Forensic
glycol from serum would incorporate an internal Sci. 4, 136(1959).
standard with solubility properties in serum and in 8. Adler, N., Residual ethylene oxide and ethylene glycol in eth-
the extracting solvent identical to those of ethylene ylene oxide sterilized pharmaceuticals. J. Pharm. Sci. 54, 735
(1965).
glycol. Propylene glycol, the most likely candidate, is
9. Russel, J. C., McChesney, E. W., and Goldberg, L., Reap-
benzoylated to a product for which the retention time praisal of the toxicology of ethylene glycol. Food Cosmet. Toxicol.
at several temperatures is too close to that of ethylene 7,107(1969).
dibenzoate to allow quantitation of either in the 10. Hamilton, G. E., and Metzner, A. B., Vapor phase oxidation
of ethylene oxide. md. Eng. Chem. 49,838(1956).
presence of the other.
11. Davis, A., Roald, A., and Tufts, L. E., Determination of trac-
Apparent recoveries of ethylene glycol from serum es of glycols by gas chromatography. J. Gas Chromatogr. 2, 306
of approximately 100% are obtained by this proce- (1964).
dure when a serum to which a known concentration 12. Spitz, H. D., and Weinberger, J., Determination of traces of
of ethylene glycol has been added is used as a stan- ethylene oxide, ethylene chlorohydrin and ethylene glycol by gas
chromatography. J. Pharm. Sci. 60,271(1971).
dard. This technique tends to normalize the varia- 13. Brown, D. J., Jam, N. C., Forney, R. B., Hughes, F. W., and
tions because of the incomplete extraction of the gly- Richards, A. B., Gas chromatographic assay of glycol-ethanol
col from serum. The coefficient of variation remains combinations in biological materials. J. Forensic Sci. 13, 537
(1968).
at about 11%, as expected.
14. Barrett, M. J., An integrated gas chromatographic program
The convenience and specificity of this method off- for drug screening in serum and urine. Clin. Chem. Newslett. 3,
set its deficiency in precision. We think that it will No. 1, p 1 (1971). (A publication of Perkin-Elmer Corp.)

824 CLINICAL CHEMISTRY, Vol. 20, No. 7, 1974