Kinetic and mechanistic studies on the hydrolysis and photodegradation of diazepam and alprazolam

Article abstract of DOI:10.1002/poc.861

The corresponding o-aminobenzophenones are usually reported as the main degradation products in the hydrolysis of 1,4-benzodiazepinones. Nevertheless, in previous studies of diazepam (DZ) in acidic aqueous medium we isolated and characterized seven unexpected degradation products. Kinetic measurements under several reaction conditions shed some light on the mechanisms of the complex reactions that are taking place, and photochemical studies give hints on the mechanisms of chlorination and annelation processes. Alprazolam (AL) seems to exhibit an unusually high stability against hydrolysis under several conditions; nevertheless, the structure is sensitive to photolytic cleavage. The photodegradation of AL was studied in aqueous and methanolic solutions. Characterization of the isolated products by ¹H and ¹³C NMR and mass spectrometry revealed that electron transfer, oxidation and rearrangement reactions take place. The influence of several variables such as pH, solvent composition and light irradiation were examined and mechanisms for the formation of three photoproducts are proposed. A specific method for the determination of AL in the presence of photoproducts was developed, which allowed kinetic determinations of the photostability of AL. The photosensitivity observed in some patients treated with AL seems to be due to one of the characterized photoproducts. Copyright

Full text of DOI:10.1002/poc.861

JOURNAL OF PHYSICAL ORGANIC CHEMISTRY
J. Phys. Org. Chem. 2005; 18: 156–161
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/poc.861
Kinetic and mechanistic studies on the hydrolysis
and photodegradation of diazepam and alprazolamy
C. Gallardo Cabrera, R. Goldberg de Waisbaum and N. Sbarbati Nudelman*
Departamento de Quı´mica Orga´ nica, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Pab. II, P. 3,
Ciudad Universitaria, 1428 Buenos Aires, Argentina
Received 12 August 2003; revised 2 November 2003; accepted 31 May 2004
ABSTRACT: The corresponding o-aminobenzophenones are usually reported as the main degradation products in the
hydrolysis of 1,4-benzodiazepinones. Nevertheless, in previous studies of diazepam (DZ) in acidic aqueous medium
we isolated and characterized seven unexpected degradation products. Kinetic measurements under several reaction
conditions shed some light on the mechanisms of the complex reactions that are taking place, and photochemical
studies give hints on the mechanisms of chlorination and annelation processes. Alprazolam (AL) seems to exhibit an
unusually high stability against hydrolysis under several conditions; nevertheless, the structure is sensitive to
photolytic cleavage. The photodegradation of AL was studied in aqueous and methanolic solutions. Characterization
1
of the isolated products by H and 13C NMR and mass spectrometry revealed that electron transfer, oxidation and
rearrangement reactions take place. The influence of several variables such as pH, solvent composition and light
irradiation were examined and mechanisms for the formation of three photoproducts are proposed. A specific method
for the determination of AL in the presence of photoproducts was developed, which allowed kinetic determinations of
the photostability of AL. The photosensitivity observed in some patients treated with AL seems to be due to one of the
characterized photoproducts. Copyright # 2004 John Wiley & Sons, Ltd.
KEYWORDS: diazepam; alprazolam; drug stability; hydrolysis; photostability; anxiolytics; benzodiazepinone pharma-
ceuticals; photosensitivity
INTRODUCTION
heroin in laboratory animals.2b Recently it was reported
that AL was involved in the inhibition of monocyte
chemoattractant protein.3
Diazepam (DZ) and alprazolam (AL) are among the
most frequently prescribed benzodiazepinones (BDPs),
which belong to an important class of drugs that exhibit
different therapeutic applications such as anxiolytics,
anticonvulsants, antithrombotics,HIVTatantagonists, anti-
tumor antibiotics, antipsychotics, anticonvulsants, hypno-
tics and muscle relaxants with different durations of
action.1 Recent studies carried out with DZ show that
it also inhibits shock-induced ultrasonic vocalization in
adult rats,2a while studies with AL demonstrate that a non-
rewarding dose of AL potentiates the affective response to
Since they are so widely used for the treatment of a
wide range of clinical disorders, several sensitive methods
are currently being developed for their quantitative deter-
mination in many matrices. Gas chromatography–ion trap
tandem mass spectrometry was recently reported for the
simultaneous detection of 22 BDPs.4 In the particular case
of DZ and AL, reversed-phase liquid chromatography–
electrospray ionization mass spectrometry5 and liquid
chromatography–tandem mass spectrometry (LC–MS–
MS)6 have been described as especially suitable for the
accurate determination of these drugs and their metabo-
lites in biological matrices such as plasma and hair.
Photosensitivity associated with BDP has recently
been reported,7,8 particularly in patients treated with
AL.9 The ICH Harmonized Tripartite Guideline10 and
the recent FDA draft guidance11 state that light testing
should be an integral part of stress test. Therefore, it was
of interest to test the stability of DZ and AL under both
hydrolytic and photochemical stress conditions.
´
*Correspondence to: N. S. Nudelman, Departamento de Quımica
Organica, Facultad de Ciencias Exactas y Naturales, Universidad de
´
Buenos Aires, Pab. II, P. 3, Ciudad Universitaria, 1428 Buenos Aires,
Argentina.
E-mail:Nudleman@qo.feen.uba.ar
ySelected article presented at the Seventh Latin American Conference
on Physical Organic Chemistry (CLAFQO-7), 21–26 September 2003,
EXPERIMENTAL
Diazepam was obtained from Roche, and used as rece-
ived. 2-(N-Methyl)amino-5-chlorobenzophenone (1),
´
Florianopolis, Brazil.
Contract/grant sponsors: University of Buenos Aires; CONICET.
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 156–161

HYDROLYSIS AND PHOTODEGRADATION OF DIAZEPAM AND ALPRAZOLAM
157
identified by CG–MS and NMR of the compounds iso-
lated by column chromatography; then the GC retention
times were checked against authentic samples indepen-
dently synthesized, as described previously.12 Acridi-
nones 6 and 7 were determined by analytical TLC,
using silica gel G60 plates with 90% toluene–cyclohexane
as eluent (RF ¼ 0.11 and 0.28 for 6 and 7, respectively) and
compared with authentic samples independently pre-
pared. Semi-quantitative determinations were carried
out by UV-visible spectrophotometry.
For kinetic measurements, the corresponding substi-
tuted aminobenzophenones (3 ꢀ 10ꢂ3 M) were dissolved
in methanol and appropriate volumes of dilute HCl were
added to obtain concentrations of 0.5, 1.0 and 1.5 M in
HCl in methanol–water (84:16 and 1:1). The ampoules
were immediately placed in a thermostat at 80 ꢃ 0.1 ꢁC.
Ampoules were taken at appropriate time intervals and
aliquots were worked up with CH2Cl2 by standard pro-
cedures and analysed by GC.
The reactions were run under pseudo-first-order con-
ditions using a 100-fold excess (at least) of HCl over the
substrate. The reactions were followed for at least three
half-lives and the final values were determined after more
than 10 half-lives.
For light-stress exposure testing, a photoreactor pro-
vided with an HPA-400 W medium–pressure metal halide
lamp (Philips) and mirrors was used. This metal halide
lamp has a close resemblance to sunlight and the output
spectrum is fairly uniform across the 350–650 nm
region.14 The photoreactor geometry and the methods
for the calibration of the lamp intensity were reported
previously.15
2-amino-5-chlorobenzophenone (2) 2-N-dimethylamino-
5-chlorobenzophenone (3), 2-amino-3,5-dichlorobenzo-
phenone (4), 2-(N-methyl)amino-3,5-dichlorobenzophe-
none (5), 2,4-dichloro-10-methylacridin-9-(10H)-one
(6) and 2,4-dichloroacridin-9-(10-H)-one (7) were syn-
thesized as previously described.12 2-N-Dimethylamino-
3,5-dichlorobenzophenone (8) was identified and
characterized in the reaction mixture by GC–MS. AL
and the synthetic intermediates were obtained from
Gador and used as received. HPLC-grade acetonitrile,
citric acid, disodium acid phosphate and triethylamine
from Aldrich were used as purchased. LC-grade metha-
nol was distilled immediately prior to use. Phosphate–
citrate buffer solutions (pH 2.0, 3.6 and 5.0) were
prepared according to standard methods. Solvents were
purified according to described procedures.13
Mass spectra were recorded on a VG (Manchester, UK)
Trio-2 mass spectrometer. High-resolution mass spectra
were measured on a BG ZAB-SEQ4F mass spectrometer.
NMR spectra were recorded on a Bruker (Karlsruhe,
Germany) AM-500 spectrometer. NMR samples were
dissolved in chloroform-d (Aldrich) and NMR spectra
were referenced using tetramethylsilane as internal refer-
ence. HPLC experiments were carried out on a Hewlett-
Packard (now Agilent Technologies) (Avondale, PA, USA)
1100 HPLC system consisting of an HP-G1311A Quat
pump, HP 61315A UV detector and LiChrosorb RP-8
(5 mm) reversed-phase HPLC column (200 ꢀ 4.6 mm
i.d.). Data acquisition and treatment of GC and HPLC
experiments were carried out using a Hewlett-Packard HP-
G2170AA Chem Station.
RESULTS AND DISCUSSION
Mechanism of hydrolysis
The diazepinone seven-membered ring in BDP is known
to be highly sensitivity to hydrolysis in both acidic and
basic media. The ring opening can occur by initial
—
breakage of the amide functionality or the C N bond,
—
in both cases the final product being the substituted
2-aminobenzophenone (BP, Scheme 1). Most of the
Quantitative GC determinations were carried out with
Model 5890 Series II Plus system (Hewlett-Packard) with
flame ionization detection (FID), using dried nitrogen as
carrier gas and an HP5 (5% phenyl, 95% dimethyl-
polysiloxane) capillary column. A typical temperature
programme that allowed the separation of the o-amino-
benzophenones was a range from 170 to 180 ꢁC at 2 ꢁC
minꢂ1, then from 180 to 200 ꢁC (held for 10 min) at
0.5 ꢁC minꢂ1. Retention times, tr (min) ꢃ 0.03, were as
follows: 5, 12.00; 1, 13.00; 2, 15.48; 3, 16.05; 4, 18.00;
and 8, 16.80. The products in the reaction mixtures were
Scheme 1
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 156–161

158
C. G. CABRERA, R. G. DE WAISBAUM AND N. S. NUDELMAN
reported methods for stability studies of BDP are based
on the determination of these very well characterized
compounds.
Previous studies from our group on the reaction of DZ
with HCl in aqueous MeOH media showed that, in
addition to the expected 2-methylamino-5-chlorobenzo-
phenone (1), other products, 2–8, were formed. Forma-
tion of products 1–3 in MeOH–HCl media is easily
explained, but the products 4–8 were completely unex-
pected and a careful isolation and full characterization
was therefore undertaken and reported.12
Products 4–8 were thought to arise from further reac-
tions of products 1 and 2, and in order to provide some
clues about the routes of formation of these rare products,
kinetic studies of the reactions of 1 and 2 with HCl at
different concentrations in MeOH–H2O (1:1) were car-
ried out and reported.16 The disappearance of the starting
2-aminobenzophenone, 1 or 2, and the appearance of the
reaction products was followed by GC of the reaction
mixtures and the rate constants for the reactions giving
rise to the different products were calculated. Similar
studies were then carried out at higher MeOH content.
Figure 1 shows a representative plot for the reaction of
2-(N-methyl)amino-5-chlorobenzophenone with 1.0 M
HCl in methanol–water (84:16) at 80 ꢁC.
Figure 1. Reaction of 2-methylaminobenzophenone (1)
with 1.0 M HCl in methanol:water (84:16) at 80 ꢁC. Data as
a function of time for (&) the disappearance of 1 and the
formation of ( þ ) 2; ( ꢀ ) 4; (*) 5; and (~) 3
(ꢄ2) or at high pH (9–11), AL was recovered unchanged.
Only by NMR analysis of the reaction mixture signals
was it possible to detect the presence of the substituted
aminobenzophenone (see Fig. 2).
In contrast, AL was insensitive to acid or base hydro-
lysis. The aromatic triazole ring seems to inhibit the
hydrolytic ring opening. Even after reflux at very low pH
Figure 2. 1H NMR spectrum of the reaction mixture of A at pH ꢅ 2. The signal for the protonated amino group of the
corresponding BP (B), the singlet for the methylene group H-4B (between the double doublet for H-4A) and the singlet for the
protons of the MeB group are clearly seen
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 156–161

HYDROLYSIS AND PHOTODEGRADATION OF DIAZEPAM AND ALPRAZOLAM
159
These results suggest that the ring opening is a rever-
sible reaction and the starting structure of AL was
recovered intact in the attempted isolation of the hydro-
lysis products. Similar results were observed when ac-
celerated thermal degradation was attempted. The
compound was then exposed to photochemical stress.
Photochemical studies
The stress studies carried out on AL by UV irradiation
revealed that the photolability of the drug is the most
adverse stability factor. The main photodegradation
products were isolated and characterized as 1-methyl-
6-phenyl-4H-s-triazo-[4,3-ꢀ][1,4]benzodiazepinone (8-
hydroalprazolam) (9), 5-chloro[500-methyl-4H-1,2,4-tria-
zol-4-yl]benzophenone (10) and triazolaminoquinoleine.
Accelerated pH-dependent studies show that the photol-
ability increases as the pH decreases; at pH 9.0 photo-
degradation does not occur; therefore, photochemical
degradation studies of AL were carried out in acidic
media. The rate of reaction was followed by a spectro-
flurometric method specially developed for studying the
stability of AL tablets under photochemical stress.17
Table 1 lists the rate coefficients of the photodegradation
of AL in buffered solutions at pH 2.0, 3.6 and 5.0.
The reaction rate constants were calculated using a
first-order rate equation. The reactions were followed at
least for four half-lives and, as shown in Table 1, the
largest rate coefficient was observed at pH 2.0.
The rate of disappearance of AL and the rate of
appearance of the main photodegradation products were
followed by HPLC. Figure 3 shows a typical plot for the
reaction in pH 3.6 buffer solution. The photodegradation
of AL gives triazoloquinoleine as the main degradation
product. This compound exhibits high fluorescence and
might be the responsible for the photosensitivity observed
in some patients treated with AL. 8-Hydroalprazolam (9)
is the second in importance, followed by the triazoloben-
zophenone. Usually, the substituted aminobenzophe-
nones are the main hydrolytic degradation products of
1,4-benzodiazepinones.
Figure 3. Photodegradation of AL in pH 3.6 buffer. Con-
centrations measured by HPLC as a function of time: (^)
alprazolam; ( þ ) 8-hydroalprazolam; (*) triazolquinoleine;
(~) triazolbenzophenone
in aqueous methanol media, using similar techniques to
those described above for AL. In these cases, the reac-
tions were carried out in sealed ampoules. The experi-
mental results allowed us to conclude that degradation is
accelerated by photochemical stress. Gas evolution was
observed which was identified as molecular hydrogen, by
¨
¨
the following test using Drager tubes (Dragenwerk,
Germany): (a) with a catalyst of the reaction H2 þ O2
! H2O and a moisture indicator, a blue colour was
observed (hydrogen detection); (b) no colour develop-
ment was observed with o-tolidine, which led to a
yellow–orange product with chlorine.
Mechanism of formation of degradation
products
For AL, the simplest photodegradation observed is de-
chlorination to give the 8-hydroalprazolam. Dehalogena-
tion of aromatic compounds is a well-known reaction in
the photochemistry of aryl halides, and occurs via by an
electron transfer mechanism between a donor and an
acceptor (the aryl halide).18
The mechanism of the reaction can be intermolecular
or intramolecular if the donor and the acceptor are in the
same molecule.19,20 Only a few studies have been carried
out with drugs with clinical activity, but diclofenac
undergoes dechlorination21 and midazolam defluorina-
tion22 under photolytic stress. Nitrogen atoms are known
to be very good electron donors and since AL has four
nitrogen atoms, an intramolecular electron transfer can be
postulated to occur with any of them. Scheme 2 shows a
plausible mechanism for electron transfer from the nitro-
gen bond to the chlorinated ring showed as: NR0R00 in
Scheme 2, where [AL]* is used to indicate AL in the
Some photodegradation studies were also carried out
on DZ. In relation to the unexpected hydrolytic behaviour
described above, reactions of 2-methylaminobenzophe-
none were carried out under photochemical stress in HCl
Table 1. Photodegradation of AL in aqueous media:
pseudo-first order rate coefficientsa,b
pH
kobs (10ꢂ5 sꢂ1
)
t (h)
2.0
3.6
5.0
3.61
1.67
0.55
5.3
11.5
34.7
a [Alprazolam] ¼ 2.75 ꢀ 10ꢂ3 M in phosphate–citrate buffer.
b Spectrofluorometric method (ꢁexc 260 nm, ꢁem 435 nm).
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 156–161

160
C. G. CABRERA, R. G. DE WAISBAUM AND N. S. NUDELMAN
Scheme 4
unstable intermediate could be stabilized to a 3,5-di-
chloroaminobenzophenone (4 or 5) or react with the o-
hydrogen of the second phenyl ring affording 3,5-dichloro-
acridinones (6 or 7). Consistent with this mechanism is
the fact that no monochlorinated acridinones were ob-
served and also that the formation of nitrenium ion from
amines is known to be increased by UV irradiation.18
(Since the nitrenium ions are extremely unstable and
difficult to generate,23 one of the referees proposed an
alternative concerted TS for both the molecular hydrogen
elimination and chloride ion attack. This mechanism may
equally account for the obtained results; we prefer the
mechanism proposed in Scheme 4 on the basis of the
effects observed by UV irradiation.)
Scheme 2
CONCLUSIONS
Scheme 3
The isolation of the degradation products and the kinetic
determinations under photochemical stress shed light on
the photostability of AL and DZ and the mechanisms of
formation of the degradation products. The results in-
dicate that light exposure and acidic media should be
avoided during the process formulation and handling of
AL. When studying acid hydrolysis of diazepam, not
only formation of the expected 2-(N-methylamino)ben-
zophenone should be monitored but also the other pro-
ducts described in this paper.
excited state. Detachment of a chloride ion followed by
proton abstraction from the solvent gives product 9.
For the formation of the triazolbenzophenone under
photochemical stress, photooxidative opening of the
diazepinone ring is postulated followed by detachment
of a carbamic HO2CNH2 molecule. Although this type of
reaction has not been observed with simpler BDPs, it is
likely that the additional stabilization given by the aro-
matic triazole ring could be a driving force for the
operation of this route. The isolation as traces of product
11 supports the proposed mechanism (Scheme 3).
A plausible mechanism for the formation of the main
degradation product, the fluorescent triazolaminoquino-
line, can be postulated: a similar opening of the diazepi-
none ring (as in Scheme 3) followed by reaction between
the activated methylene and the carbonyl group, giving
rise to a highly stabilized six-membered ring.
Finally, all the observations on the photochemical
reactions of 2-methylaminobenzophenone in HCl in
aqueous MeOH media offer an explanation for the for-
mation of the unexpected products observed in the acid
hydrolysis of DZ. Although somewhat speculative, a
nitrenium intermediate could account for the formation
of the double chlorinated products and the acridinones as
shown in Scheme 4.
Acknowledgements
C. G. C. is a grateful recipient of a fellowship from
COLCIENCIAS (the National Research Council of
Colombia). N. S. N. gratefully acknowledges the hospital-
ity of the Institute for Materials Chemistry and Engineer-
ing, Kyushu University, Fukuoka, Japan, where this paper
was written. Financial support from the University of
Buenos Aires and from CONICET (the National Research
Council of Argentina) is gratefully acknowledged.
REFERENCES
1. Prut L, Belzung C. Eur. J. Pharmacol. 2003; 463: 3–33.
´
2. (a) Sanchez C. Eur. J. Pharmacol. 2003; 463: 133–143; (b)
Walker BM, Ettenberg A. Pharmacol. Biochem. Behav. 2003;
75: 75–80.
The elimination of molecular hydrogen from the cor-
responding aminobenzophenone would give an unstable
nitrenium intermediate that could react with chloride ions
in excess giving the 3,5-dichloro intermediate. This
3. Oda T, Ueda A, Shimizu N, Handa H, Kasahara T. J. Immunl.
2002; 9: 3329–3335.
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 156–161

HYDROLYSIS AND PHOTODEGRADATION OF DIAZEPAM AND ALPRAZOLAM
161
4. Pirnay S, Ricordel I, Libong D, Bouchonnet S. J. Chromatogr. A
2002; 54: 235–245.
5. Toyo’oka T, Kumakia Y, Kanboria M, Katoa M, Nakaharab Y.
J. Pharm. Biomed. Anal. 2003; 30: 1773–1787.
6. Kronstrand R, Nystram I, Josefsson M, Hodgins J. Anal. Toxicol.
2002; 7: 479–484.
14. Moore DE. In Photostability of Drugs and Drug Formulations,
ch. 4, Tonnesen HH (ed). Taylor and Francis: London, 1996;
65–66.
15. Nudelman NS, Cabrera CG. J. Pharm. Sci. 2002; 91: 1274–1286.
16. Nudelman NS, Goldberg RN. J. Phys. Org. Chem. 1997; 10:
97–106.
7. Harth Y, Rapoport M. Drugs Saf. 1996; 14: 252–259.
8. Reid CG. Can. Pharm. J. 1995; 128: 31–56.
17. Nudelman NS, Cabrera CG. J. Pharm. Biomed. Anal. 2002; 30:
887–893.
9. (a) Kanwar AJ, Gupta R, Das Metha S. Dermatologica 1990; 181:
7–12; (b) Beijersbergen van Henegouwen GMJ. Adv. Drug Res.
1997; 29: 79–170.
18. Albini A, Sulfizio A. In Photoinduced Electron Transfer, vol. 1.
Fox MA, Chanon M (eds). Elsevier: Amsterdam, 1988; Part C
88–93.
10. International Conference on Harmonization. Photostability Testing
of New Drug Substances and Products, ICH, November 1996.
11. Food and Drug Administration. Guidance for Industry: Stability
Testing of Drug Substances and Drug Product, Draft Guidance.
FDA: Washington, DC, 1998.
12. Nudelman NS, Waisbaum RG. J. Pharm. Sci. 1995; 84: 208–
211.
13. (a) Nudelman NS, Doctorovich F. Tetrahedron 1994; 50: 4651–
4666; (b) Nudelman NS, Marder M, Gurevich A. J. Chem, Soc.,
Perkin Trans. 2 1993; 229–236; (c) Nudelman NS, Lewkowicz E,
Furlong JJP. J. Org. Chem. 1993; 58: 1847–1852.
19. Pac C, Ishitani O. Photochem. Photobiol. 1988; 48: 767–772.
20. Fox MA. Photochem. Photobiol. 1990; 52: 617–621.
21. Moore DE, Roberts-Thomson S, Zhen D, Duke CC. Photochem.
Photobiol. 1990; 52: 685–690.
22. Andersin R, Mesilaakso M. J. Pharm. Biomed. Anal. 1995; 13:
667–670.
23. (a) McClelland RA, Kahley MJ, Davidse PA, Hadzialic G. J. Am.
Chem. Soc. 1996; 118: 4794–4796; (b) McClelland RA, Ren D,
Ghobrial D, Gadosy TA. J. Chem. Soc., Perkin Trans 2 1997; 451–
456; (c) Sukhai P, McClelland RA. J. Chem. Soc., Perkin Trans 2
1996; 1529–1530, and references cited therein.
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 156–161