Isolation and structural elucidation of degradation products of alprazolam: Photostability studies of alprazolam tablets

Article abstract of DOI:10.1002/jps.10141

Accelerated thermal, hydrolytic, and photochemical degradations of alprazolam were performed under several reaction conditions. The stress studies revealed the photolability of the drug as the most adverse stability factor; the main photodegradation products were isolated and properly characterized as: triazolaminoquinoleine; 5-chloro-[5-methyl-4H-1,2,4-triazol-4-yl]benzophenone, and 1-methyl-6-phenyl-4h-s-triazo- [4,3-α][1,4]benzodiazepinone. Accelerated pH-dependent studies show that the photoinstability increases as the pH decreases; at pH 9.0, photodegradation does not occur, therefore, the photochemical degradation of alprazolam was performed in buffered solutions at pH 2.0 and 3.6. The higher rate of reaction was observed at pH = 2.0; consequently, acidic conditions should be avoided and appropriate light protection is recommended during the drug-development process, storage, and handling. The main degradation route for alprazolam tablets is also photochemical.

Full text of DOI:10.1002/jps.10141

Isolation and Structural Elucidation of Degradation Products
of Alprazolam: Photostability Studies of Alprazolam Tablets
NORMA S. NUDELMAN, C. GALLARDO CABRERA
Depto. Qu ´ı mica Org a´ nica, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires,
Pab. II, P.3 Ciudad Universitaria, 1428 Buenos Aires, Argentina
Received 1 August 2001; revised 22 January 2002; accepted 31 January 2002
ABSTRACT: Accelerated thermal, hydrolytic, and photochemical degradations of alpra-
zolam were performed under several reaction conditions. The stress studies revealed the
photolability of the drug as the most adverse stability factor; the main photodegradation
products were isolated and properly characterized as: triazolaminoquinoleine; 5-chloro-
``
[
5-methyl-4H-1,2,4-triazol-4-yl]benzophenone, and 1-methyl-6-phenyl-4H-s-triazo-
[
4,3-a][1,4]benzodiazepinone. Accelerated pH-dependent studies show that the photo-
instability increases as the pH decreases; at pH 9.0, photodegradation does not occur,
therefore, the photochemical degradation of alprazolam was performed in buffered
solutions at pH 2.0 and 3.6. The higher rate of reaction was observed at pH ¼ 2.0;
consequently, acidic conditions should be avoided and appropriate light protection is
recommended during the drug-development process, storage, and handling. The main
degradation route for alprazolam tablets is also photochemical. ß 2002 Wiley-Liss, Inc.
and the American Pharmaceutical Association J Pharm Sci 91:1274–1286, 2002
Keywords: alprazolam; photochemical degradation; alprazolam stability; triazolo-
benzodiazepines; alprazolam photodegradation products
INTRODUCTION
methods have been recently developed for forensic
and clinical applications.
2,5
Alprazolam (AL) is a popular triazolobenzodiaze-
pine; compared with the older benzodiazepines,
AL has a shorter half-life, a smaller dose is requir-
ed for therapeutic efficacy (clinical doses range
Stability studies of other triazolobenzodia-
6
7
zepines, such as triazolam and midazolam, have
been reported, but despite the extensive use of AL,
to the best of our knowledge, a thorough study of
its stability has not been published. The kinetics
and equilibrium of the reversible AL ring-opening
reaction at 258C were carefully studied over a pH
from 0.25 mg three times daily to a total dose of
1
4
mg per day), and it is extensively used for the
control of panic attacks and in the management of
2
8
anxiety disorders. Many of the routine human
range of 0.5–8.0. Hydrolysis of the benzodiaze-
studies of AL have utilized conventional gas chro-
matography with electron capture detection,
pinone ring is one the most frequently observed
degradation routes for benzodiazepinones;
3
9,10
or high-pressure liquid chromatography (HPLC)
with ultraviolet (UV) detection. Nevertheless,
because of the low concentrations found in bio-
logical specimens, highly sensitive analytical
nevertheless, in the case of AL, we have deter-
mined that hydrolysis under several different
4
1
1
conditions, is not a major degradation source.
Similar observations have been reported for
1
2,13
this
and other related triazole benzodiaze-
1
4
pinones. The reversible 1,4-benzodiazepinone
ring-opening under aqueous acidic conditions was
previously described for AL and for triazolam.
Correspondence to: Norma S. Nudelman (Telephone: 5411-
4
5763355; Fax: 5411-45763346;
8
6
E-mail: nudelman@qo.fcen.uba.ar)
In preliminary forced stress testing, we have
observed that the drug is photolabile, and a
Journal of Pharmaceutical Sciences, Vol. 91, 1274–1286 (2002)
ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association
1
274
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

PHOTOSTABILITY STUDIES OF ALPRAZOLAM
1275
photosensitivity reaction has been recently repor-
ted in some patients treated with AL.
There is a strong demand for harmonized
guidelines for photochemical stability studies of
drugs and drug products from regulatory bodies
mirrors was used. This metal halide lamp has a
close resemblance to sunlight; the output spec-
trum is rather uniform across the 350–650 nm
1
5–17
23
region. Distance between the lamp and the
samples was 5 cm; likely heating of the samples
was monitored and the temperature was always
below 608C. The UV dose from the lamp was
1
8,19
dealing with drug registration worldwide.
The
2
0
ICH Harmonized Tripartite Guideline as well as
the recent FDA draft guidance state that light
testing should be an integral part of the stress
2
1
24
measured by a reported method using quinine
monohydrochloride dihydrate (2% solution in
water) as a chemical actinometer, for calibrating
the UV radiation in sunlight-simulating condi-
tions. The ICH establishes that the minimum
2
2
testing. The present work describes stability
testing of AL under several stress conditions, isola-
tion and characterization of the main photodegra-
dation products of AL buffered solutions, and
kinetic studies of the photochemical degradation
of pharmaceuticals containing AL in solid forms.
2
desired exposure is 200 W per m which corre-
sponds to a quinine actinometer change of 0.5 AU
at 400 nm; this change was observed after a 3-h
exposure time with a new halide lamp. The
calibration of the photolysis system was per-
formed following the recent recommendations
EXPERIMENTAL SECTION
Materials
25
for the photostability testing. Linear regression
analysis of the absorbance change for the quinine
actinometry versus the photolysis exposure time
for independent quinine solutions showed almost
the same slope within experimental error (ꢁ 2–
AL, and the synthetic intermediates, were obtain-
ed from Gador Labs (Buenos Aires, Argentina) and
used as received. HPLC-grade acetonitrile, citric
acid, disodium acid phosphate, and triethylamine
from Aldrich Chemical Company (Milwaukee, WI)
were used as purchased. Liquid chromatography
grade methanol was distilled immediately before
use. Phosphate-citrate buffer solutions (pH 2.0,
3
%). The quinine solutions and the drug samples
were exposed after the UV lamp was operating for
h to ensure proper equilibrium of the system.
2
Calibration of the lamp was performed at the
beginning of each study; a new lamp replaced it
when a change in the slope was observed. Thermal
controls were performed with the quinine solu-
tions as well as with the drug samples by exposing
tightly wrapped cells of each of them during the
entire photolysis exposure time: no changes in
the absorbances were observed in any of the
wrapped cells. In separate runs, the drug samples
were exposed for longer times (6–7 h) to enlarge
the photolysis extent. The integrated intensity of
radiation DAbs, at 400 nm after 6-h exposure was
of DAbs ¼ 0.9 AU of the quinine actinometer. No
change in the nature or the relative yields of the
photodegraded products was observed when com-
pared with the 3-h exposure time.
3
.6, and 5.0) were prepared according to standard
methods.
Mass spectra were recorded on a VG Trio-2
mass spectrometer (Manchester, UK). High-reso-
lution mass spectra were measured on a ZAB-
SEQ4F mass spectrometer (Manchester, UK).
Nuclear magnetic resonance (NMR) spectra were
recorded on a Bruker AM-500 spectrometer
(
Karlsruhe, Germany). NMR samples were dis-
solved in chloroform-d (Aldrich), and NMR spec-
tra were referenced using tetramethylsilane
as internal reference. HPLC experiments were
performed on a Hewlett-Packard (now Agilent
Technologies) 1100 HPLC system consisting of an
HP-G1311A Quat pump, HP 61315A UV detector,
and reversed-phase HPLC column Lichrosorb RP-
Preparation of Photodegradation Products
8
, 5 mm (200 ꢀ 4.6 mm). Data acquisition and treat-
AL 1.0 g, in a 1000-mL volumetric flask, was
dissolved in 130 mL of methanol and made up to
volume with citrate buffer (pH 3.6). The solution
was irradiated with an HPA lamp for 8 days. The
temperature inside the reactor was nearly 608C,
and it was widely demonstrated that AL does not
undergo thermal degradation under these condi-
tions (see Results and Discussion). After irradia-
tion, the solution was treated with NaOH up
ment were performed by using a Hewlett-Packard
HP-G2170AA Chem Station (Avondale, PA).
Photostability Studies
For the light-stress exposure testing, a photo-
reactor provided with a medium-pressure metal
halide lamp HPA-400 W (Philips, Belgium) and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

1
276
NUDELMAN AND GALLARDO CABRERA
to pH 8.0; upon cooling, a yellowish precipitate
appeared which was removed by filtration. The
filtrate was extracted four times with 100 mL of
dichloromethane each time. The extracts were
washed with water several times and the organic
A was 0.025 M acetate buffer (pH 4.5)/acetonitrile
(65:35 v/v) and mobile phase B was 0.025 M
acetate buffer (pH 4.5)/acetonitrile (20:80). The
following gradient program was used: 0–10 min
(linear gradient mobile phase A: 100% to 50%),
10–14 min (isocratic, mobile phase A: 50%) and
14–20 min (linear gradient mobile phase A: 50%
to 0%). The flow rate was 1.0 mL/min, with UV
detection at 254 nm.
[5-Chloro-2-[5-methyl-4H-1 ,2 ,4 -triazol-4 -
yl]-benzophenone],1 (triazolbenzophenone).
This photoproduct was firstly purified by pre-
parative TLC silica gel 60 F254 (Merck), using
isopropyl alcohol (100%) as eluent (Rf alprazolam
0.20; R f triazolbenzophenone 0.27). Spots were visua-
lized by 254-nm UV light, and extracted from the
silica with methanol. Solvent was removed by
distillation at reduced pressure. The extract was
dissolved in chloroform, filtered by suction, and
the solvent distilled at reduced pressure. The
extract was then purified by preparative TLC
cellulose (Aldrich). Plates were developed in
aqueous 0.2 N NaOH (Rf alprazolam 0.35; Rf triazol-
layer (dried with MgSO ) was evaporated to dry-
4
ness under reduced pressure. This mixture of
photodegradation products is called the ‘‘extract’’
below.
`
0
0
0
0
Tablets (0.5 mg of AL) were irradiated with the
lamp, described previously, for 6 months. After
irradiation, the tablets were dissolved in metha-
nol by sonication for 30 min, filtered, and the
filtrate was evaporated to dryness under reduced
pressure: three main degradation products were
isolated and identified. In separate runs, four
brands of commercial tablets (containing 0.5
to 2.0 mg of AL as the only clinical drug) were
exposed to sunlight for 400–700 days. Two main
products were isolated and identified.
Isolation and Purification of
Photodecomposition Products
0.49). Spots were removed as descri-
benzophenone
bed above. The residue was further purified by
HPLC using method 1 before the spectroscopic
determinations. Sample solutions were prepared
by diluting the residue with water/acetonitrile
(40:60 v/v); 20-mL aliquots of the sample solution
were injected in the HPLC. The fractions corre-
sponding to the main part of each peak were
The ‘‘Extract’’
To isolate the several photodegration products,
different chromatographic systems were tested.
Repeated attempts of separation by thin-layer
chromatography (TLC), using several eluent
systems, showed that the ‘‘extract’’ was a complex
mixture very difficult to resolve; similar unsuc-
cessful results were obtained using different kinds
of column chromatography. This suggests that the
products might have very similar structures.
By several successive TLCs, three main pro-
ducts could be isolated. The first sample resolu-
tion was performed using preparative silica gel
plates (Merck KGaA, TLC silica 60 F254 Darm-
stadt, Germany) with 70:30:0.4 EtOAc/MeOH/
triethylamine as eluent. UV light (254 and
manually collected (tr
14.3 min;
alprazolam
tr triazolbenzophenone 12.0 min). Then, each fraction
was evaporated to dryness under reduced pres-
sure. The residue was dissolved in chloroform and
sonicated for 15 min, after which the solution was
filtered and evaporated to dryness under reduced
pressure. It is worthwhile to note that this com-
pound, together with compound 3, were the main
degradation products isolated from the irradiated
tablets, following the described procedures.
3
66 nm) showed four very close spots that were
1-[methyl]-6-phenyl-4H-s-triazol-[4,3a][1-
4]benzodiazepine, 2 (8H-alprazolam). This
product was purified by preparative TLC silica
gel (Merck TLC silica 60 F254). Plates were
developed in chloroform/methanol (100:5 v/v)
(R f alprazolam 0.56; Rf 8H alprazolam 0.52). Then the
extract was purified by preparative TLC cellulose
(Aldrich) using aqueous 0.2 N NaOH as solvent
(Rf 8H alprazolam 0.65). The residue was redissolved
in 0.025 M acetate buffer (pH 4.5)/acetonitrile
(40:60 v/v) and then submitted to further purifica-
removed, eluted with MeOH, and worked up with
standard procedures. The residue was redissolved
in chloroform, filtered, and the filtrate evapora-
ted to dryness. The observed fractions were
purified by successive TLCs and HPLC as des-
cribed below for each photodegradation product.
For HPLC method 1, a mobile phase of water/
acetonitrile (60:40) was pumped at 1 mL/min
through an HPLC column with UV detection at
2
20 nm.
HPLC method 2 was a binary gradient
tion by HPLC using method 2. (tr
alprazolam
reversed-phase HPLC procedure. Mobile phase
10.7 min; tr 8H alprazolam 8.1 min)
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

PHOTOSTABILITY STUDIES OF ALPRAZOLAM
1277
Triazolaminoquinoleine, 3. This photopro-
duct was firstly purified by preparative TLC on
silica gel plates (Merck TLC silica 60 F254) using
a mixture of ethyl acetate/dichloromethane/tri-
ethylamine (70:30:0.4 v/v) as eluent. Spots were
visualized by 366-nm UV light. The extract
corresponding to 3 was further purified by pre-
parative HPLC using method 1. Because the MW
of this compound is the same as that of AL, its
identity was confirmed by independent synthesis.
Synthesis of triazolaminoquinoleine, 3.
AL 186 mg was dissolved in 2 mL of methanol
and 4 mL of 0.33 N HCl was added to generate
that were further purified by column chromato-
graphy. MeOH pH 2.8 formic buffer (50:50
followed by 70:30) and then MeOH/formic acid
50%, 10:0.8, gave one compound which we call
compound 4. However, TLC analysis of fraction 2
using alumina plates gave three close spots that
could be separated by column chromatography
using successive solvent mixtures. Thus, ethyl
acetate/methanol/triethylamine 50:5:0.5 followed
by 50:10:0.5 gave compound 5 and 30:70:0.5 gave
compound 6 and compound 4. The amounts of
isolated compounds 4–6 were not enough to per-
form a reliable characterization, and efforts were
concentrated on the main products 1–3.
0
0
0
0
‘
‘in situ’’ the 5-cloro-2-[5 -methyl-4H-1 ,2 ,4 -tria-
0
zol-4 -yl]-benzophenone. The pH of this solution
was immediately adjusted to 9.0 with 2.0 N KOH,
and 2.0 mL of acetic anhydride in 4 mL of THF
RESULTS AND DISCUSSION
(
3
tetrahydrofurane) was added with stirring. After
h at room temperature, the reaction mixture
Preliminary studies of AL stability were per-
formed under thermal, hydrolytic, and oxidative
stress conditions. Solid samples of AL were stored
at 808C: periodic quantitative HPLC determina-
tion did not show significant degradation even
after 5 months at 808C. Buffered AL solutions at
pH ¼ 3.5 were heated in sealed ampoules at 808C
for 88 days: HPLC determinations did not show
any significant change in AL concentration or
appearance of any degradation product. Previous
studies reported the observation of changes in the
UV-visible spectra of pH < 0.5 AL solutions at
was evaporated to dryness under vacuum. The
residue was dissolved with 0.5 mL of methanol,
1
was heated for 6 h at 918C. After cooling, the
solution was extracted with 5 mL of benzene. The
extract was purified by TLC silica gel using chlo-
roform/isopropyl alcohol/methanol (80:15:5 v/v).
0 mL of 2.0 N KOH was added, and the solution
(
Rf triazolquinoleine 0.77; Rf alprazolam 0.62; R
f subst.
2
6
0
.49). 3 was properly characterized ;
benzophenone
the overall yield was 83%.
2
58C, but the degradation products were not
4,5
identified. Attempts to isolate the degradation
The Precipitate
27
products by TLC also failed. AL solutions at pHs
ranging from 2.0 to 9.0 were studied in our
laboratory at temperatures of 50–808C to detect
hydrolysis but AL seems to be stable under those
conditions. Attempts to isolate any degradation
product were unsuccessful likely because of the
small amounts and/or to the equilibrium between
The yellow crystals isolated from the irradiated
buffered solution (pH 3.8, 8 days) were sparingly
soluble only in methanol; the addition of acid
or basic cosolvents did not improve dissolution.
A new run was performed by irradiating AL in
citrate buffer (pH 2.0) during 10 days. By neu-
tralizing the solution, a yellowish precipitate was
obtained. This solid was soluble in ethyl acetate,
thus allowing its separation from the buffer salts
after several washings. The organic layer was
1
the structures. Only by H-NMR determination of
the reaction mixture at pH ¼ 0.5 was it possible to
`
characterize the presence of the [5-chloro-2-[3-
0
0
0
0
0
methylamino-5 methyl-1 ,2 ,4 -triazol-4 -yl]-ben-
1
dried (MgSO ) and the solvent distilled at reduced
zophenone] in small amounts. The H-NMR of AL
4
pressure. The TLC analysis revealed the presence
of several products. The solid was firstly treated
by column chromatography separating fraction 1
is shown in Figure 1 for further comparisons; the
spectrum exhibits the AX system of both geminal
protons in C4 as two duplets at 5.5 and 4.3 ppm,
J ¼ 12 Hz, and the sharp singlet due to the methyl
(
EtAcO/MeOH, 7:3) and fraction 2 (MeOH/
1
triethylamine, 50:0.1). The TLC analysis showed
that fraction 1 is a mixture of the three products
previously described.
Attempts to resolve fraction 2 by several TLC
systems using silica gel were unsuccessful; then
by using reverse phase with methanol and formic
acid as eluent, several close spots were obtained
group at 2.65 ppm. In the H-NMR of the mixture
obtained by hydrolysis (not shown), the distinct
signals due to the open substituted aminobenzo-
phenone are clearly a broad signal at 8.8 ppm
attributed to the protonated amino group and the
relatively broad singlet for the two equivalents
protons at C4 which appears at 4.5 ppm (2H).
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

1278
NUDELMAN AND GALLARDO CABRERA
1
Figure 1. H-NMR spectrum of alprazolam.
Several oxidants were tested on AL solutions in
degradation products. This gives confidence to the
artificial irradiation stress to reproduce natural
photochemical degradation. Longer exposure to
sunlight (400–700 days) increases the concentra-
tion of the photodegradation products but it did
not change the structures and no new compounds
appeared.
methylene chloride and in chloroform, but even
after 5 days at the solvent boiling temperature, no
significant degradation was observed. The describ-
ed results showed the relatively high stability of
AL under the usual accelerated aging conditions.
The main degradation route for AL was found
to be photochemical and most of the following
studies were conducted to determine the effect
of light irradiation under different conditions.
Preliminary studies show that AL solutions at
pH > 9.0 are insensitive to photochemical reac-
tions, and the rate of degradation increases as the
pH decreases. Figure 2 shows the changes in the
UV visible spectra of pH 3.5 buffered AL solutions
upon irradiation: an increase in the absorbance at
240 nm and a decrease at the AL lmax ¼ 260 nm is
observed. No new peaks appeared; identical AL
solutions protected from light showed no changes
in the UV visible spectra after 48 h, which
indicates that the changes observed in Figure 2
are not due to conventional hydrolysis as
previously reported. These changes are not
significant to follow the photostability by spectro-
photometric determinations and a fluorometric
assay was developed. Solid AL samples as well as
tablets containing AL as the only clinical drug
were exposed to natural sunlight; after 88 days of
irradiation, they showed by TLC the same photo-
Figure 2. UV spectrum of alprazolam in pH 3.6
buffered solution after irradiation with a metal halide
lamp. Irradiation times (h) for curves (a) to (i) equal to:
0, 1, 2, 3, 4, 5, 6, 7, and 9 h, respectively.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

PHOTOSTABILITY STUDIES OF ALPRAZOLAM
1279
Structure Elucidation of the
Photodegradation Products
gen atoms in the AL molecule contributes to
the reaction occurrence. Photo-dehalogenation for
other drugs has been previously reported: e.g.,
2
9
30
As stated in the Experimental section, the
separation and isolation of the degradation
products of AL were very difficult likely because
of their similar chemical properties. The most
important photochemical degradation was obser-
ved at pH ¼ 2.0 (citrate buffer). After 8–10 days of
irradiation of buffered AL solutions, three main
products, 1–3, were isolated and fully identified.
dichlofenac
and chlorpromazine
give the
reduction product and also hydroxylated pro-
30,31
ducts
by photochemical degradation. The
typical dd signal for the geminal coupling
(J ¼ 12.9 Hz) of the two H4s is clearly observed
1
in the H-NMR spectrum (Figure 4), and the
singlet for the methyl group which also appears at
the same chemical shift as that of AL. The
aromatic protons zone is quite different, showing
one additional proton; the corresponding shifts
and multiplicities for the other aromatic protons
are consistent with the assigned structure. The
``
5
-Chloro-[5-methyl-4H-1,2,4-triazol-4-yl]-
benzophenone, 1 (HRMS MW 294.0667). The
electron ionization mass spectra (EI-MS) of this
compound showed the molecular ion and some
key ion fragments with a pattern typical of
compounds containing one chlorine atom.
Furthermore, this is the only one that shows a
key ion fragment of mass 105 which revealed an
aminobenzophenone with one unsubstituted phe-
nyl ring, as previously observed in the EI-MS of
degradation products of other benzodiazepi-
13
C-NMR (not shown) is also very similar to that
of AL.
Triazolaminoquinoleine, 3 (HRMS MW
308.0827). This compound shows a strong fluor-
escence at lexc 260 nm; the emission spectrum
shows a lmax ¼ 410 nm. The first approach to the
structure of compound 3 was performed through
the analysis of the EI-MS. Unexpectedly for a
degradation product, the molecular ion shows the
same mass as AL (308 amu), but the rest of the
spectrum is completely different. The [M ꢂ 1] ion
base peak, typical for amines, and the presence of
[M ꢂ 1 þ 2] [M] [M þ 2] ions, indicate an amine
group and one chloro atom in the molecule. Other
main ion fragments are 103, 89, and 77 which are
also observed in the EI-MS of AL. The main clues
for the elucidation of this unexpected structure
9
,10
nones.
Another important ion fragment corre-
sponds to the loss of 29 amu (HNN) which is
typical of 1,2,4-triazoles. The ion fragment (m/z
1
92) complementary of that of mass 105, and the
other ion fragments of minor relative abundance
confirm the structure. Analysis of the H-NMR
1
spectrum showed in Figure 3 indicates a methyl
group (singlet at high field), the typical pattern for
the protons in both phenyl groups with multi-
plicities consistent with the given structure and
the singlet at low field which indicates the unique
1
13
arose from a careful analysis of the H- and C-
NMR spectra. The singlet for the methyl group
appeared downfield, suggesting a major depro-
tective effect that could be due to an increased
resonance in the rings. The broad signal at 4.69
`
`
proton (H3) in the triazole ring. A signal at 192.59
1
3
ppm in the C-NMR of 1 (not shown) confirms the
carbonyl C, as well as the other C assignments for
the remainder of the signals.
1
1
-[Methyl]-6-phenyl-4H-s-triazol-[4,3a][1-
ppm observed in the H-NMR spectrum shown in
4
]benzodiazepine (8H-alprazolam), 2 (HRMS
Figure 5, integrates for 2 H which are exchange-
able upon treatment with MeOD. That these 2 H
are bonded to the same N (like in a primary
MW 274.1216). The EI-MS shows an intense
molecular ion of 274 amu, in which the [M þ 2] ion
is clearly absent. Both results suggest that this
compound could be the product of photodechlor-
ination of AL. Consistent with this assumption,
the rest of the EI-MS shows fragment ions that
are very similar to those observed in the EI-MS of
AL, after the separation of one Cl atom (main
fragment ions: 245, 204, 103, 89, and 77).
Halogenated aryl compounds easily undergo
intra or inter-electron transfer photo-dehalogena-
tion when there is a good electron donor present,
given the reduction and/or the hydroxylated (or
methoxylated if the reaction is performed in
1
3
amine), was indicated by the C spectrum whose
peculiar downfield zone is expanded in Figure 6.
Only one signal for aliphatic carbon appears
which is due to the methyl group. (The typical
signal for the methylene C, which is part of the
diazepinone group in AL, is not present.) The
resonance of two carbons typical for alkenes is
observed at 112.44 and 116.59 ppm; there are only
two (instead of three) imine C, which appear at
147.35 and 145.46 ppm, respectively. Also, notice-
1
able in the H-NMR is the absence of the dd signal
typical of the geminal coupling of the protons
bonded to C4 in the diazepine ring. In the
2
8
methanol) products. The accumulation of nitro-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

1
280
NUDELMAN AND GALLARDO CABRERA
1
``
Figure 3. H-NMR spectrum of 5-chloro-[5-methyl-4H-1,2,4-triazol-4-yl]benzophe-
none.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

PHOTOSTABILITY STUDIES OF ALPRAZOLAM
1281
1
Figure 4. H-NMR spectrum of 1-methyl-6-phenyl-4H-s-triazo-[4,3-a][1,4]benzodia-
zepinone.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

1282
NUDELMAN AND GALLARDO CABRERA
1
Figure 5. H-NMR spectrum of triazolaminoquinoleine.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

PHOTOSTABILITY STUDIES OF ALPRAZOLAM
1283
1
3
Figure 6.
C-NMR spectrum of triazolaminoquinoleine.
1
aromatic zone of the H-NMR spectrum, the
signals due to the unsubstituted phenyl ring
were assigned, according to the integration and
the expected multiplicities for the ortho- and
para-coupling, as shown in Figure 5. The other
three signals, which integrate for one H each one,
exhibit the classical AMX multiplicity of a tri-
substituted ring: the doublet at 8.07 ppm (H-9),
the dd at 7.32 ppm (H-8), and the d at 7.22 (H-6).
The signal that appears at the most downfield
position is assigned at H-9 because of the
anisotropy that affects the ipso H. Because
various rearrangement structures could be con-
sistent with the above results, the structure was
fully confirmed by independent synthesis, as
described in the Experimental section.
Photostability of Pharmaceuticals Containing AL
To assure that the photodegradation products
identified by irradiation of buffered solutions of
AL could be related to the degradation undergone
by exposure of pharmaceuticals to natural irra-
diation, commercial tablets containing 0.5, 1.0,
and 2.0 mg of AL were irradiated by sunlight for
long periods (400–700 days). The contents of AL
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

1284
NUDELMAN AND GALLARDO CABRERA
and of the degradation products were periodically
determined. The knowledge of the different struc-
tures of the photodegradation products allowed
the development of a reliable HPLC assay for AL
and its degradation product. Four commercial
pharmaceuticals were studied. A (2.0 mg) and B
of AL. Furthermore, compounds 1 and 3 could be
easily isolated from the irradiated tablets, pur-
ified, and fully characterized by NMR and MS
which were identical with the spectra determined
with the photodegraded products of AL. Under
the present experimental conditions, the degra-
dation of AL and the formation of compounds
1 and 3, follow a zero order kinetic law; the
regression coefficients for the four samples were
within acceptable ranges for solid forms. The
8-hydroalprazolam, 2, appeared only in small
amounts. It was observed that the relative rates
of degradation of the tablets packed in blister
packs are higher than those in large doses;
although it was shown that AL is not very
sensitive to hydrolysis, it could be possible that
the blister better retains the humidity and other
likely deleterious factors that could accelerate the
degradation. Because there are also small differ-
ences in the excipients, their specific effects as
well as the possible surface catalysis, cannot be
fully neglected.
(
1.0 mg) are tablets expended as large doses; C
and D (0.5 mg) are packed in 10 tablet blisters.
Tablets A and B were separately put in sealed
transparent flasks, whereas samples C and D
were irradiated in the original package. The
tablets become yellowish upon irradiation but no
changes in the blister packs were observed even
after 2 years of sunlight irradiation. The contents
of AL and of the degradation products were
periodically determined. Figure 7 shows a typical
HPLC run corresponding to tablets D after 408
days of sunlight exposure. The peaks appear in
the chromatogram in the order: 8H-alprazolam,
2
, as traces; triazolbenzophenone 1 (5.3%); AL
(
93.2%), and triazolquinoleine, 3 (1.3%).
The four commercial samples were exposed to
sunlight between 495 (tablets D) and 781 (tablets
C) days; the concentrations of AL and of the main
degradation products were periodically deter-
mined. The two main degradation products,
namely: triazolbenzophenone, 1, and triazolami-
noquinoleine, 3, showed identical TLC and HPLC
chromatographic behavior with the standards of
the previously characterized compounds from the
photochemical degradation of buffered solutions
CONCLUSIONS
AL shows an unusually high stability against the
most common degrading factors: humidity, oxy-
gen, and heat. Hydrolytic degradation products
resulting from the opening of the diazepinone ring
were not amenable of isolation due to the rapid
reverse cyclization at the various pHs assayed.
However, we have found that the drug is sensi-
tive to both artificial and sunlight, and the
photolability increases at low pH. Irradiation of
buffered solutions of AL allowed isolation and
characterization of three main products: tria-
``
zolaminoquinoleine (fluorescent); 5-chloro-[5-
methyl-4H-1,2,4-triazol-4-yl]benzophenone, and
1
-methyl-6-phenyl-4H-s-triazo-[4,3-a][1,4benzo-
diazepinone (also called 8H-alprazolam). Irra-
diation of AL tablets by artificial and sunlight
showed the same main degradation products,
being 8H-alprazolam formed only as traces. The
higher rate of the photochemical reaction was
observed at pH ¼ 2.0, consequently, acidic con-
ditions should be avoided, and appropriate light
protection is recommended during the drug-
development process, storage, and handling.
Figure 7. Typical HPLC determination of alprazo-
lam tablets D after 408 days of exposure to sunlight in
blister packs. Peaks appear in the order: 8H-alprazolam
ACKNOWLEDGMENTS
``
(
traces); 5-chloro-[5-methyl-4H-1,2,4-triazol-4-yl]ben-
zophenone (5.3%); alprazolam (93.2%); triazolquino-
leine (1.3%).
CGC is a grateful recipient of a COLCIENCIAS
fellowship. The authors are indebted to Gador
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

PHOTOSTABILITY STUDIES OF ALPRAZOLAM
1285
Lab. for the provision of AL and other drugs.
Financial support from the University of Buenos
Aires, the CONICET (National Research Coun-
cil), and the Agency for the Promotion of Science
and Technology from Argentine is acknowledged.
a-hydroxyalprazolam in human plasma by gas
chromatography/negative-ion chemical ionization
mass spectrometry. J Mass Spectrom 31:1033–
1
038.
6
. Konishi M, Hirai K, Mori Y. 1982. Kinetics and
mechanism of the equilibrium reaction of triazolam
in aqueous solution. J Pharm Sci 71:1328–1334.
REFERENCES
7. Andersin R, Mesilaakso M. 1995. Photochemical
decomposition of midazolam. IV. Study of pH-
dependent stability by high-performance liquid
chromatography. J Pharm Biomed Anal 13:667–
670.
1
. Laihanen N, Muttonen E, Laaksonen M. 1996.
Solubility and intrinsic dissolution rate of alpra-
zolam crystal modifications. Pharm Dev Technol
1
4:373–380.
8. Cho MJ, Scahill TA, Hester JB. 1983. Kinetics and
equilibrium of the reversible alprazolam ring-open-
ing reaction. J Pharm Sci 72:356–362.
9. Nudelman NS, Waisbaum RG. 1997. Kinetic study
of the reaction of 2-amino-5-chlorobenzophenone
2
. (a) Dresser GK, Spence JD, Bailey DG. 2000.
Pharmacokinetic–pharmacodynamic consequen-
ces and clinical relevance of cytochrome P450 3A4
inhibition. Clin Pharmacokinet 38(1):41–57; (b)
Gommans J, Hijzen TH, Maes RA, Olivier B.
with HCl in MeOH-H
106.
2
O. J Phys Org Chem 10:97–
2000. Discriminative stimulus properties of alpra-
zolam. Psychopharmacology 148(2):146–152; (c)
Yuan R, Flockhart DA, Balian JD. 1999. Pharma-
cokinetic and pharmacodynamic consequences of
metabolism-based drug interactions with alpra-
zolam, midazolam, and triazolam. J Clin Pharma-
col 39(11):1109–1125; (d) Santagostino G, Cucchi
ML, Frattini P, Zerbi F, Di Paolo E, Preda S,
Corona GL. 1998. The influence of alprazolam on
the monoaminergic neurotransmitter systems in
dysthymic patients: Relationship to clinical
response. Pharmacopsychiatry 31(4):131–136.
. (a) Greenblatt DJ, Javaid JI, Locniskar A, Harmatz
JS, Shader RI. 1990. Gas chromatographic analysis
of alprazolam in plasma: Replicability, stability and
specificity. J Chromatogr 534:202–207; (b) Green-
blatt DJ, Divoll M, Moschitto LJ, Shader RI. 1981.
Electron-capture gas chromatographic analysis of
the triazolobenzodiazepines alprazolam and tria-
zolam. J Chromatogr 225:202–207.
. (a) Hall MA, Robinson A, Brissie RM. 1995. High-
performance liquid chromatography of alprazolam
in postmortem blood using solid phase extraction.
J Anal Toxicol 19:511–513; (b) Schmith VD, Cox
SR, Zemaitis MA, Kroboth PD. 1991. New high-
performance liquid chromatographic method for
the determination of alprazolam and its metabo-
lites in serum: Instability of 4-hydroxyalprazolam.
J Chromatogr 568:253–260.
10. (a) Nudelman NS, Waisbaum RG. 1995. Isolation
and structure elucidation of novel products of
the acidic degradation of diazepam. J Pharm Sci
84:208–211; (b) Nudelman NS, Waisbaum RG.
1995. Acid hydrolysis of diazepam: Kinetic study
of the reactions of 2-(N-methylamino)-5-chloroben-
zophenone, with HCl in MeOH-H
84:998–1004, and refs. therein.
O. J Pharm Sci
2
11. (a) Nudelman NS, Gallardo C. 1999. Factores que
afectan estabilidad de alprazolam: Elucidaci o´ n
estructural de los principales productos de degra-
daci o´ n. Vitae 6:29–40; (b) Nudelman NS, Wais-
baum RG, Gallardo C. 1998. Stability studies of
benzodiazepines: Solvolysis and photodegradation.
Atualidades Fisico Quim Org 478–497.
12. Allen LE, Erickson MA. 1998. Stability of alpra-
zolam, chloroquine phosphate, cisapride, enalapril
maleate and hydralazine hydrochloride in extem-
poraneously compounded oral liquids. Am J Health
Syst Pharm 55:1915–1920.
13. Caira MR, Easter B, Honiball S, Horne A, Nassim-
beni LR. 1995. Structure and thermal stability of
alprazolam and selected solvates. J Pharm Sci
84:1379–1384.
14. Stemm NL, Skoug JW, Robins RH. 1995. Gradient
high performance liquid chromatography assay for
degradation products of adinazolam mesylate in a
sustained release tablet formulation. Pharm Res
12:738–745.
15. Kanwar AJ, Gupta R, Das Metha S. 1990. Photo-
sensitivity due to alprazolam. Dermatologica 181:
75.
16. Harth Y, Rapoport M. 1996. Photosensitivity
associated with antipsychotics, antidepressant and
anxiolytics. Drug Saf 14:252–259.
17. Reid CG. 1995. Photosensitivity: Not a picture of
health. Can Pharm J 128:31–56.
18. Tonnesen HH. 1996. Introduction: Photostability
testing of drug formulations—Why and how? In:
3
4
5
. (a) Hold KM, Crouch DJ, Rollins DE, Wilkins DG,
Maes RA. 1997. Detection of alprazolam in hair by
negative ion chemical ionization mass spectro-
metry. Forensic Sci Int 84:201–209; (b) Crouch
DJ, Rollins DE, Canfield DE, Andrenyak DM,
Schulties JE. 1999. Quantitation of alprazolam
and a-hydroxyalprazolam in human plasma using
liquid chromatography electrospray ionization
MS-MS. J Anal Toxicol 23:479–485; (c) Hold KM,
Rollins DE, Crouch DJ, Wilkins DG, Canfield DV,
Maes RA. 1996. Determination of alprazolam and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002

1286
NUDELMAN AND GALLARDO CABRERA
Tonnesen HH, editors. Photostability of drugs
and drug formulations. London: Taylor & Francis,
pp 1–7.
Photostability testing: Design and interpretation of
tests on drug substances and dosage forms. In:
Tonnesen HH, editors. Photostability of drugs and
drug formulations. London: Taylor & Francis, pp
305–321.
1
2
9. Piechocki JT. 1998. Selecting the right source for
pharmaceutical photostability testing. In: Albini A,
Fasani E, editors. Drugs: Photochemistry and
photostability. Cambridge: Royal Society of Chem-
istry, pp 247–271.
25. United States Pharmacopeial Convention Inc.
Photostability testing. 2000. Pharmacopeial Forum
26:384–387.
0. (a) International Conference on Harmonization,
Q1B. 1996. Photostability testing of new drug
sustances and products, November; (b) Fed. Regist.
26. Cho MJ, Sethy VH, Haynes LC. 1986. Sequentially
labile water-soluble prodrugs of alprazolam. J Med
Chem 29:1346–1350.
1
997, 62, 27115.
27. Jimenez RM, Dominguez E, Bahia D, Alonso RM,
Vicent F. 1987. On the mechanism of hydrolysis of
the triazolobenzodiazepine, triazolam, spectro-
scopic study. J Heterocycl Chem 24:421.
2
2
1. Food and Drug Administration. 1998. Guidance for
industry: Stability testing of drug substances and
drug product, draft guidance, June.
2. Valvini SC. 2000. Industrial stability testing in
the United States and computerization of stability
data. In: Carstensen JT, Rhodes CT, editors. Drug
stability, principles and practices, 3rd ed. New
York: Marcel Dekker, pp 515–552.
3. Moore DE. 1996. Standardization of photodegrada-
tion studies and kinetic treatment of photochemical
reaction. In: Tonnesen HH, editor. Photostability of
drugs and drug formulations. London: Taylor &
Francis, pp 65–66.
4. (a) Hong DD, Shah M. 2000. Development and
validation of HPLC stability-indicating assays. In:
Carstensen JT, Rhodes CT, editors. Drug stability,
principles and practices, 3rd ed. New York: Marcel
Dekker, pp 329–384; (b) Anderson NH. 1996.
28. Greenhill JV. 1996. Is the photodecomposition of
drugs predictable? In: Tonnesen HH, editor. Photo-
stability of drugs and drug formulations. London:
Taylor & Francis, pp 83–110.
29. Moore DE, Roberts-Thomson S, Zhen D, Duke CC.
1990. Photochemical studies on the anti-inflamma-
tory drug diclofenac. Photochem Photobiol 52:685–
690.
30. Schoonderwoerd SA, Beijerssbergen van Henegou-
wen GM, van Belkum S. 1989. In vivo photodegrada-
tion of chlorpromazine. Photochem Photobiol 50:659.
31. Greenhill JV, McLelland MA. 1990. Photodecom-
position of drugs. In: Ellis GP, West GB, editors.
Progress in medicinal chemistry, Vol. 27. Amster-
dam: Elsevier, pp 51–121.
2
2
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 5, MAY 2002