Journal of Analytical Toxicology, Vol. 20, March/April 1996
I Case Report
Analytical Findings in a Suicide Involving Sodium Azide
Pierre Marquet 1,*, Sophie Clement 2, Hayat Lotfi 1-3, Marie-France Dreyfuss 1, Jean Debord 1, Daniel Dumont 2, and
G~rard Lach~tre 1-3
15ervicede Pharmacologieet Toxicologieet 2Servicedes Urgences,CHRU Dupuytren, 87042 LimogesCedexand 3Laboratoirede
Toxicologie, Facult~de Pharmacie,87025 LimogesCedex, France
I Abstract I
A 47-year-old laboratory assistantingested approximately 9 g of
sodium azide powder and died 4 h later at a hospital. A highperformance liquid chromatographic method using diode-array
detection has been developed for the determination of an azide
benzoyl derivative in blood (after a simple deproteinization) and in
several tissues (after homogenization in a neutral buffer and
deproteinization of the supernatant). The blood concentration in
this case was lower than those previously published. The highest
azide concentration was found in lung tissue. A complete
toxicological screening revealed the presence of cyanide in blood,
which has been previously reported twice, but for the first time, it
was confirmed by massspectrometry. Whether the production of
cyanide in the presence of azide took place in vivo or postmortem
remains unknown; the nature of the metabolic pathway involved
also remains unknown.
Introduction
Sodium azide, a highly toxic chemical, is used in industry as
an explosive, in cars as a safety bag (AIRBAG| inflation
reagent, and in chemical and biomedical laboratories as an
antifungal and conservative for biological samples and diagnostic reagents. It has also been used as a potent hypotensive
agent (1), but clinical trials in the 1950s were abandoned because of side effects.
Sodium azide normally occurs as a white powder, is highly
soluble in water, and has a PKa of 4.6. In acidic solutions, it is
rapidly converted to hydrazoic acid, a toxic and pungent gas. In
humans, sodium azide is rapidly absorbed after ingestion (1),
through the skin, and, together with hydrazoic acid, by inhalation. It has been shown in vitro, on rat liver mitochondria,
that it acts by uncoupling mitochondrial oxidative phosphorylation by cytochrome oxidase inhibition (2). It also
inhibits horse liver catalase (3). The LDs0by the oral route in
white mice is 27 mg/kg (4).
Probably because of a relatively limited availability, intoxi* Author to whom correspondenceshouldbe addressed.
134
cation with sodium azide is rare and generally involves people
working in laboratories or factories using this chemical. To the
best of our knowledge, approximately 30 cases have been reported in the literature, 14 of which were lethal (5-16). Very
few fatalities have been documented with azide concentration
determinations in blood or in tissues.
A direct spectrophotometricassay, first proposed for titration
of azide in diagnostic solutions (17), has been applied to postmortem samples (15). But such a method is not reliable for
precise determinations in complex matrices or in mixtures.
Colorimetric methods, using ferrochloride addition to the
samples or volumetric measurement of nitrogen liberated by
iodine reaction with hydrazoic acid, have also been employed
(8). Both lack specificity and sensitivity.
Recently, Lambert et al. (16) developed a high-performance
liquid chromatographic (HPLC) method very similar to ours,
which we did not know of at the time of its development. This
method was also based on previous work by Swarin and Waldo
(18), designed for various industrial media. Their method used
diode-array detection (DAD) and was applied to postmortem
samples; derivatization of hydrazoic acid was performed with
3,5-dinitrobenzoyl chloride, whereas we used benzoyl chloride. Both methods are simple, rapid, and specific and allowed
reliable determinations of azide in blood and various organs of
individuals who committed suicide.
Case History
A 47-year-old male laboratory technician intentionally ingested approximately 9 g of sodium azide powder at 8 p.m. He
was found several minutes later, vomiting and confused. He
had written a suicide note. When he arrived at the hospital
emergency department at 9:30 p.m., he showed no other clinical signs. A gastric lavage was immediately performed, and
cardiac and blood pressure monitoring were established. The
patient quickly fell into a coma, and he showed signs of
laryngeal dyspnea. At 10:15 p.m., bradycardia and a drop in
blood pressure necessitated treatment with atropine and
Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.
Journal of Analytical Toxicology, Vol. 20, March/April 1996
from Aldrich (St. Quentin Fallavier);and RP Normapur KH2PO4
was purchased from Prolabo (Fontenay-sous-Bois, France). A
1-g/L stock solution of sodium azide was prepared in deionized
water and kept frozen. Working solutions were obtained by
dilution of the stock solution in deionized water.
Procedure.Solid tissues were pretreated prior to liquid-liquid
extraction as follows: 20 g of tissue was homogenized in
40 mL 1M Tris buffer, pH 7.4, then separated into two equal
parts. To one part was added 10 mg of subtilisin, and hydrolysis
was performed at 37~ for 2 h. Then, after centrifugation, 1 mL
of the supernatant of each hydrolyzedand nonhydrolyzedtissue
extract was further processed like liquid biological samples. In
a 10-mL glass tube, 1 mL of liquid sample was deproteinized
with 1 mL acetonitrile, shaken for 15 rain on an oscillating
agitator, and centrifuged at 900 xg for 5 rain. The supernatant
was transferred in another 10-mL glass tube containing 50 pL
1M Tris buffer, pH 4.6, and 50 t~Lofa 100-g/L benzoyl chloride
methanolic solution. After shaking for 15 min, extracts were incubated at ambient temperature for 15 rain. Fifty microliters
was then injected into the chromatograph.
Chromatography. HPLC was performed using a reversedphase isocratic method with a Waters 501 pumping system
and a 7125 Rheodyne| injector equipped with a 20-1JLloop. A
column (150 x 4.6-ram i.d.; 5-pm particle size) (Waters SA., St.
Quentin en Yvelines, France) packed with Novapack Cls was
used. The mobile phase was a 0.05M KH2PO4 Tris buffer
(pH 4.6)-acetonitrile mixture (55:45, v/v). The analysis was
performed at ambient temperature and at a flow rate of
1.1 mL/min.
ephedrine. The patient was then intubated and mechanically
ventilated. A few minutes later, electrical shocks and vasopressive drugs (adrenaline and isoprenaline) had to be administered because of cardiac arrest. A continuous supraventricular
arrhythmia followed for some minutes, and there were electrocardiographic signs of myocardial ischemia. A second heart
arrest then occurred; resuscitation was ineffective. Death was
declared at 11:45 p.m.
The autopsy was performed on the following day. A deep
cyanosis of the face and the fingers and purple and bluish
lividities on the body were noted. The lungs were congested,
and pleura, like epicardium, presented petechia. The liver bled
when sectioned. Meninges were congested, and the brain was
very pale. Gastric contents and bile, kidney, liver, and brain
samples were collected for toxicological analysis. Blood sampled during the sectioning of the thoracic wall, which probably
came from congested intercostal veins (called thoracic blood),
and from the suprahepatic veins (suprahepatic blood) was also
collected for analysis.
Experimental
Analytical reagents. Sodium azide (99% pure) was purchased
from Merck (Paris, France); HPLC-grade acetonitrile was
obtained from Carlo-Erba (Nanterre, France); benzoyl chloride
was purchased from Sigma (St. Quentin Fallavier,France); tris(hydroxymethyl)-methylamine(Tris) (99% pure) was obtained
0.01
-
A
B
0.008 -
_E
0.02
0.006 9
I
I
0.004 '
0.01
0.002
I
~
E E
9
~
988m,o
Benzoylazide
0
,,,i
I
"
10
Time (rain)
I ' I
200
....
i ....
220
i ....
i ....
240
i i . . . i
....
260
i ....
i ....
280
i ....
i ....
i ....
300
i ....
320
i..].i
....
340
i...
i
--i
....
360
i ....
i ....
380
Wavelength (nm)
FigureI. (A) Chromatogram and (B) UV spectrum of sodium azide as benzoyl derivative, obtained from one-fifth diluted postmortem blood.
135
Journal of Analytical Toxicology, Vol. 20, March/April 1996
Detection. A Waters 991 diode-array detector (Waters S.A.)
continuously recorded the ultraviolet (UV)spectrum from 200
to 400 nm. The spectrum of an eventual peak in the chromatogram of an unknown sample was statistically compared
with the recorded spectrum of a derivatized azide standard.
Chromatograms were represented at 252 nm, and benzoylazide (azide derivative) concentrations were computed, comparatively to a calibration graph, at the same wavelength.
Validations. To study linearity and reproducibility, test samples were prepared at 0, 500, 1000, 2000, and 5000 IJg/L by
adding 100 laL of an appropriate working solution to 900 IJL of
whole blood. Each of these standards was prepared, extracted,
and analyzed daily for 5 days. Repeatability was assessed on
500-, 1000-, and 5000-1~g/Lstandard concentrations by performing six different extractions and quantitations of a same
sample test on the same day. All the calculations were performed from chromatographic peak areas. Percent recovery
was directly obtained from these same experiments, as peakarea ratio between each extract and a respective nonextracted
derivatized solution of sodium azide, corresponding to 100%
extraction. Accuracy was evaluated by comparison of mean
computed concentrations with target ones.
Drug screening. A screening on more than 400 drugs and
toxicants in blood samples and gastric contents has been performed by means of a double detection method using
HPLC-DAD and gas chromatography-mass spectrometry
(GC-MS), after two separate nonselective liquid-solid extrac-
tions, for acidic and basic compounds, respectively. More sensitive chromatographic screenings have also been performed,
using GC-MS after acetylation for neuroleptics; GC for ethanol
and barbiturates; HPLC-DADfor alpidem, zolpidem, and zopiclone; and spectrophotometry for carboxyhemoglobin and
methemoglobin. Cyanideshave been assayed in blood by means
of a colorimetric technique measuring cyanocobalamin formed
from hydroxocobalamin (19). Confirmation was performed by
a headspace GC-MS method, based on a previously published
GC method with thermoionic detection (20). This previous
method utilized blood cyanide conversion in an acidic medium
to volatile hydrogen cyanide in a tightly closed vial. Then,
hydrogen cyanide was immediately derivatized to cyanogen
chloride in the headspace of the vial by reaction with chloramine T.
Results and Discussion
By using our present method, the chromatographic retention time of benzoylazide was approximately 10 min, and its UV
spectrum showed a maximal absorbance at 252 nm (Figure 1).
The limit of detection was slightly lower than 200 IJg/L,as the
signal-to-noise ratio measured for this concentration was
better than 3. Within-day coefficient of variation (CV) values
were better than 10 and 5% for the 500- and 5000-1Jg/Lsodium
Table I. Repeatability and Reproducibility of Sodium Azide Determination in Whole Blood
Added
concentration
(pg/L)
Repeatability(n = 6)
CV*
area*
(%)
Reproducibility (n = 5)
CV
area*
(%)
Recovery
(%)
Calculated
concentration*
(pg/L)
Accuracy
(%)
19.21
500
0.0180
6.95
109.4
0.0176
14.45
596
1000
0.0345
4.56
104.3
0.032
10.08
989
1.15
2000
-
-
-
0.0655
9.29
1871
6.45
5000
0.1968
3.23
103.3
0.186
6.05
5032
0.64
* CV = Coefficient of variation.
"~The values given are the mean.
Table II. Present and Previously Published Sodium Azide Concentrations in Postmortem Samples
Study
Kozlicka-Gajdzinska et al. (1966)
Klug et al. (1987)
Gastric
contents
(pg/g)
Intestinal
contents
(pg/g)
Blood
(mg/L)
Bile
(mg/L)
Liver
(pg/g)
Kidney
(pg/g)
-
867
429
-
-
NF*
58,900
7360*
85
-
6.1
Brain
(pg/g)
Lung
(pg/g)
Urine
-
-
-
-
Peclet et al. (1991)
8.8*
-
40
-
NF
-
-
NF
Lambert et al. (1995)
754
-
262
1283
14
205
-
-
Present study
1036
-
7.4w 8.3 n
21.4
NF (<0.4)
NF (<0.4)
17.6
-
* NF = Not found.
* This value was found in the duodenum; no azide was found in the jejunum.
* Total content (value given in grams).
w Value found in thoracic blood.
u Value found in suprahepatic blood.
136
2.7
Journal of Analytical Toxicology, Vol. 20, March/April 1996
azide concentrations, respectively. Interday CV values were
less than 15 and 10% for the same respective concentrations
(Table I). Accuracy,based on the interday results, showed a 19%
overestimation of the 500-1ag/Ltheoretical concentrations. On
the other hand, accuracy was very good for the 1000-Iag/L
target concentration. This could be due in part to the moderately increased extraction recovery at 500 lag/L(Table I), which
directly affects the calculated concentration and is probably
caused by the lack of a chemical analogous as internal standard. For this reason, the limit of quantitation retained in this
study was 500 IJg/L, although repeatability and reproducibility
were still satisfactory below this concentration.
Regression on 20 points (five concentrations from 500 to
5000 IJg/L,four determinations per concentration) was excellent (r = 0.9937, p < .001), and nonlinearity, assessed by
ANOVA, was insignificant. Linearity has even been verified
from 500 up to 50,000 lJg/L on two different days (seven concentrations, one determination per concentration); correlation coefficients were 1.0000 and 0.9998, respectively.
The lack of an internal standard was due to the difficulty of
rapidly finding a chemical analogue of azide, as these forensic
results could not be delayed. Nevertheless, on the condition of
the above range limitation and accurate filling of the injector
loop, the present method is precise, reproducible, and accurate.
Sodium azide concentrations were measured in gastric contents, thoracic blood, suprahepatic blood, bile, and various
tissues (Table II). Most of these concentrations were greater
than the calibration range of the method, so biological liquids
or supernatants of ground tissues had to be diluted before a
new extraction. Enzymatic hydrolysis of ground tissues did
not improve azide recovery.
We could find only four papers presenting postmortem azide
concentrations in the digestive tract, blood, or tissues. These
values are also represented in Table II. The large concentration
variability between studies is not directly representative of the
actual ingested doses, as some subjects had vomited or been
lavaged or both after ingestion. Also, the time between ingestion and death was not always known.
Blood concentrations varied from 8 to 262 mg/L, the lowest
value being ours and the highest having been obtained by
Lambert et al. (16) with a very similar analytical method.
100% = 30544
Base peak = 419
1:57
2:51
I
RIC = 61
Time
3:45
4:38
5:33
|
I
I
6:27
Unfortunately, clinical signs and time between ingestion and
death were unknown in the latter case.
We failedto detect any azide in the liver and kidneys, notwithstanding extractions at different pH values, with or without
enzymatic hydrolysis; this was opposed to a high concentration
found in the bile. The absence of azide in the liver was confirmed by two of the other four studies (5,15), whereas low
concentrations were found in the two studies that reported the
highest blood concentrations (8,16). Only Lambert et al. (16) determined sodium azide in the kidneys, at a concentration almost
as high as blood concentrations (205 pg/g and 262 lJg/mL,
respectively), but as in our case, no urine could be sampled.
Peclet and Ponton (15), who reported a fairly high blood concentration (40 rag/L), found no azide in urine, but they used
direct spectrophotometry, which has a low sensitivity.
We reported here for the first time azide concentrations in
the brain and lung. The former was low, and it was noticeable
that the latter was higher than the concentration in blood or
in other tissues. It suggested that azide is either preferentially
bound to lung tissue or preferentially excreted in expired gas
as hydrazoic acid.
Among the other drugs or toxicants analyzed, the only findings were ethanol in gastric contents (0.79 g/L), probably by
postmortem production as it was absent in blood, and cyanide
in blood (0.38 mg/L). The presence of cyanide was confirmed by
GC-MS; the single-ion chromatogram (m/z 61) of CNCI is
shown in Figure 2. This finding has been reported twice before
in the literature: the cyanide concentration was 1.6 mg/L (unknown method) in the plasma of a male who had ingested
1.2-2 g of sodium azide (11) and 9 mg/L (Koenig's colorimetric
method) in the blood of a female who had ingested an unknown but high dose, as suggested by the obvious rapidity of
death (16). Whether the production of cyanide in the presence
of azide took place in vivo or postmortem remains unknown, as
well as the nature of the metabolic pathway involved.
Reported deaths with sodium azide were usually related to
voluntary ingestion of the equivalent of 1.5-20 g of the powder
form, when it was known or could be evaluated. Delay before
death ranged from 40 rain (5) to 31/2 days (12), during which
neurological signs (agitation alternating with coma), dyspnea,
lactic acidosis, and cardiac rhythm disorders (resistive to most
treatments) were always found. Clinically, this case was thus
very similar to these reports. Moreover, the 4-h delay before
death, the vomiting, and the gastric lavage were in agreement
with the low azide concentrations in blood and tissues as compared with the few other documented studies.
References
m
,
200
I
300
I
Figure 2. Single-ion chromatogram of CNCI
mortemblood.
I
400
500
Scan number
I
600
700
(m/z 61), obtained from post-
1. M.M. Black, B.W. Zweifach, and F.D. Speer. Comparison of
hypotensive action of sodium azide in normotensive and hypertensive patients. Proc. Soc. Exp. Biol. Med. 85' 11-16 (1954).
2. F. Palmieri and M. Klingerberg. Inhibition of respiration under the
control of azide uptake by mitochondria. Eur. J. Biochem. 1:
439-46 (1967).
3. E.C. Foulkes and R. Lemberg. The azide inhibition of catalase.
Enzymologia 13:302-12 (1949).
137
Journal of Analytical Toxicology,Vol. 20, March/April 1996
4. J.D.R Graham. Actions of sodium azide. Br. J. Pharmacol. 4(1):
1-6 (1949).
5. H. Kozlicka-Gajdzinska and J. Brzyski. A case of fatal intoxication
with sodium azide. Arch. Toxikol. 22:160-63 (1966).
6. E.A. Emmet and J.A. Ricking. Fatal self-administration of sodium
azide. Ann. Intern. Med. 83(2): 224-26 (1975).
7. T.E. Albertson, S. Reed, and A. Siefkin. A case of fatal sodium
azide ingestion. Clin. Toxicol. 24(4): 339-51 (1986).
8. E. Klug and V. Schneider. Suizid mit natriumazid. Z. Rechtsmed.
98:129-32 (1987).
9. J. Abrams, R.S. EI-Malakh, and R. Meyer. Suicidal sodium azide
ingestion. Ann. Emerg. Med. 16:1378-80 (1987).
10. K.W. Judge and N.E. Ward. Fatal azide-induced cardiomyopathy
presenting as acute myocardial infarction. Am. J. Cardiol. 64:
830-31 (1989).
tl. W. Klein-Schwartz, R.L. Gorman, G.M. Oderda, B.P. Massaro,
T.L. Kurt, and J.C. Garriott. Three fatal sodium azide poisonings.
Med. Toxicol. Adverse Drug Exp. 4:219-27 (1989).
12. J.D. Howard, K.J. Skogerboe, G.A. Case, W.A. Raisys, and E.Q.
Lacsina. Death following accidental sodium azide ingestion.
]. Forensic 5ci. 35(1): 193-96 (1990).
13. L. Binder and L. Fredrickson. Poisonings in laboratory personnel
and health care professionals. Am. J. Emerg. Med. 9:11-15
(1991).
138
14. RE. Senecal, J.E. Dyer, J.D. Osterloh, and K.R. Olson. Toxic
volatile hydrazoic acid (HN 3) from contact of sodium azide
(NaN 3) with acids. Vet. Hum. Toxicol. 33(4): 364 (1991).
15. C. Peclet and G. Ponton. Fatal sodium azide poisoning. Int.
Assoc. Forensic Toxicol. 21(3): 28-29 (1991 ).
16. W.E. Lambert, M. Piette, C. Van Peteghem, and A.P. De Leenheer.
Application of high-performance liquid chromatography to a
fatality involving azide. J. Anal. Toxicol. 19:261-64 (1995).
17. E.A. Terpinski. Spectrophotometric determination of sodium
azide. Analyst 110:1403-1405 (1985).
18. S.J.Swarin and R.A. Waldo. Liquid chromatographic determination of azide as the 3,5-dinitrobenzoyl derivative. J. Liq. Chromatogr. 5(4): 597-604 (1982).
19. M. Laforge, F. Buneaux, P. Houeto, F. Bourgeois, R. Bourdon,
and P. Levillain. A rapid spectrophotometric blood cyanide
determination applicable to emergency toxicology. J. Anal. Toxicol. 18:173-75 (1994).
20. M. Odoul, B. Fouillet, B. Nouri, R. Chambon, and P. Chambon.
Specific determination of cyanide in blood by headspace gas
chromatography. J. Anal. Toxicol. 18:205-207 (1994).
Manuscript received March 6, 1995;
revision received July 19, 1995.