Identification of chlorothiazide and hydrochlorothiazide UV-A photolytic decomposition products

Article abstract of DOI:10.1021/js9601501

Methanol solutions of hydrochlorothiazide and chlorothiazide were irradiated with fluorescent UV-A lamps in order to simulate degradation under normal conditions. The degradation products were identified by comparison to synthetic standards featuring electrospray ionization mass spectroscopy, ultraviolet spectroscopy, and high performance liquid chromatography: The standards were characterized by high resolution fast atom bombardment MS and ¹H NMR. The photolysis of chlorothiazide resulted in photodehalogenation products exclusively, while the irradiation of hydrochlorothiazide primarily yielded photodehalogenation products with significant yields of photodehydrogenation products and minor amounts of thermal hydrolysis products.

Full text of DOI:10.1021/js9601501

Identification of Chlorothiazide and Hydrochlorothiazide UV-A Photolytic
Decomposition Products
†
X
†
†
L
ARRY K. REVELLE , STEVEN M. MUSSER‡, BRADLEY J. ROWE
,
AND
IRINA C. FELDMAN
Received March 28, 1996, from the †Food and Drug Administration, Center for Drug Evaluation and Research, Division of Drug Analysis,
1114 Market Street, Room 1002, St. Louis, MO 63101 and ‡Food and Drug Administration, Center for Food Safety and Applied Nutrition,
Instrumentation and Biophysics Branch, 200 C Street, S. W., Washington, DC 20204.
received January 15, 1997 .
Accepted for publication January 27, 1997X.
Final revised manuscript
mediately pipetted into a quartz photolysis cuvette (1 cm × 1 cm × 3
Abstract
0 Methanol solutions of hydrochlorothiazide and chlorothiazide
cm), Starna Cells, Atascadero, CA. The cuvette was immediately
capped with
a rubber septum. The cuvette was centrally and
were irradiated with fluorescent UV-A lamps in order to simulate
degradation under normal conditions. The degradation products were
identified by comparison to synthetic standards featuring electrospray
ionization mass spectroscopy, ultraviolet spectroscopy, and high perfor-
mance liquid chromatography. The standards were characterized by high
resolution fast atom bombardment MS and 1H NMR. The photolysis of
chlorothiazide resulted in photodehalogenation products exclusively, while
the irradiation of hydrochlorothiazide primarily yielded photodehalogenation
products with significant yields of photodehydrogenation products and
minor amounts of thermal hydrolysis products.
horizontally located, 3 cm underneath two UV-A 24 inch, 20 watt
Sylvania fluorescent lamps, Model F20T12/BLB, Starbeam Supply,
St. Louis, MO. The lamps were mounted horizontally, parallel, and
6 cm apart. The cuvette solution was irradiated. Aliquots were
analyzed immediately by HPLC-MS and HPLC-UV. Several hours
later after photolysis termination they were analyzed again by
HPLC-UV.
HPLC-MSsChromatography was performed on
a YMC, Inc.
(Wilmington, NC) J sphere ODS-M80 HPLC column (2 × 250 mm).
Chromatography conditions for electrospray analysis started with 0%
acetonitrile then changed by a linear gradient to 30% acetonitrile in
20 min. A buffer concentration of 0.1% formic acid was maintained
throughout the run. A Hewlett-Packard (Palo Alto, CA) Model 1050
HPLC pump was employed to provide a linear gradient and constant
flow rate of 200 mL/min. The entire column effluent passed into the
standard ion source of a Finnigan Model TSQ-7000 triple-quadrupole
mass spectrometer (San J ose, CA). Nitrogen was used as a nebulizing
gas and the capillary temperature was 260 °C. The instrument was
scanned over the range of 180-500 amu at 1 s/scan.
Introduction
Hydrochlorothiazide (1) and chlorothiazide (1A) have proven
to be clinically effective diuretics for more than 35 years since
their original syntheses.1,2 Degradation of these thiazides by
hydrolysis has been well-studied.3 Thiazides 1 and 1A have
been reported to be potent photosensitizers which increase
the sensitivity of skin to solar radiation.4 The irradiation of
1 at wavelengths greater than 310 nm (medium-pressure
mercury lamp with glass filter)5 and the irradiation of 1 and
1A at 313 nm (high pressure mercury lamp with filters)6 have
been reported.
The International Conference on Harmonization (ICH) has
developed guidelines for photostability studies. The ICH
Photostability Expert Working Group has proposed guidelines
for optional photolysis sources.7 One of the optional ICH lamp
sources was defined as a UV-A fluorescent lamp with wave-
lengths ranging from 300 to 400 nm. This lamp source
simulates the UV portion of sunlight filtered through window
glass.8 To study the photodegradation of 1 and 1A due to UV
radiation that might be encountered during storage, manu-
facture, and skin exposure, we have photolyzed methanol
solutions of these thiazides under the ICH-defined UV-A lamp
and identified the degradation products. In the photolysis of
hydrochlorothiazide we did not find after exhaustive effort the
principal hydrolysis product reported by Tamat and Moore.5
In addition they5 failed to find the photodehydrogenation
products we identified. In the photolysis of chlorothiazide Ulvi
and Tammilehto6 failed to find the photodehalogenation
products we identified. We have synthesized independently
a standard for each degradation product reported (Figure 1).
Toxicity studies will be reported separately.
HPLC-UVsA Waters 600 HPLC pump was employed to provide
a linear gradient and constant flow rate of 1.0 mL/min. The mobile
phase was pumped through a Hewlett-Packard 1090M (Palo Alto, CA)
ChemStation equipped with a diode array detector, a Rheodyne
manual injector (Model 7125), and a Beckman Ultrasphere, C-18, 5
µm HPLC column (150 × 4.6 mm). The wavelength was monitored
at 270 nm. The chromatographic conditions for analysis started with
0% acetonitrile then changed by a linear gradient to 30% acetonitrile
over a period of 12 min. A phosphoric acid solution (0.1%) was
maintained throughout the run.
High Resolution MSsSpectra were measured on a MS-50 three-
sector (KRATOS, Ramsey, NJ ) mass spectrometer. Samples were
dissolved in methanol and mixed with 4-nitrobenzyl alcohol (Aldrich,
St. Louis, MO) prior to introduction to the mass spectrometer. The
ionization mode was FAB with Argon as a primary beam. The spectra
were recorded twice for each sample by the peak matching method;
a mixture of glycerol and CsI was used as a cocalibrant.
High Resolution NMRsProton NMR spectra were recorded on a
Unity-Plus 500 (Varian, Palo Alto, CA) spectrometer. All NMR
samples were dissolved in methanol-d4 (Cambridge Isotope Labora-
tories, Woburn, MA) and referenced to residual protonated methanol
at 3.30 ppm.
Photolysis of Hydrochlorothiazide (1)sA 9.6 mg/mL methanol
solution of 1 (Mylan Pharmaceuticals, Morgantown, WV) was added
to fill a photolysis cuvette. The solution was irradiated according to
the photolysis method previously described. At 3 h intervals over a
21 h period, a 25 mL aliquot was removed via syringe to be diluted
with 1.00 mL of an internal standard solution for HPLC-UV analysis.
(The internal standard solution was prepared by dissolving 380 mg
of ethylparaben in 1.00 L of 0.1% phosphoric acid aqueous). The above
photolysis experiment was repeated using the same concentration of
1 in methanol without deoxygenation.
Experimental Section
Photolysis of Chlorothiazide (1A)sA 1.2 mg/mL methanol solution
of 1A (Merck, Sharp & Dohme, West Point, PA) was added to fill a
photolysis cuvette. The solution was photolyzed according to the
photolysis method previously described. At 24 h intervals over a 5
day photolysis period a 75 mL aliquot was removed via syringe and
diluted with 75 mL of internal standard solution for HPLC-UV
analysis. (The internal standard solution was prepared as described
P a r t 1sPhotolysis MethodsMethanol solutions of 1 or 1A were
presaturated with nitrogen by bubbling for 60 min to remove
molecular oxygen. The deoxygenated thiazide solution was im-
X Abstract published in Advance ACS Abstracts, March 1, 1997.
Published 1997, American Chemical Society
and American Pharmaceutical Association
S0022-3549(96)00150-5
This article not subject to U.S. copyright.
Journal of Pharmaceutical Sciences / 631
Vol. 86, No. 5, May 1997

above.) The above photolysis experiment was repeated using the
same concentration of 1A in methanol without deoxygenation.
P a r t 2sSynthesis of Thiazide DerivativessSynthetic thiazide
derivatives described below (Scheme 1) were dissolved in methanol
and applied to tapered, preparative TLC plates (Analtech, Newark,
DE). The plates were developed in ethyl acetate:hexane 80:20 eluant.
The proper band was scraped and eluted with ethyl acetate for
analysis. Synthetic standards showed greater than 98% purity by
HPLC-UV analysis.
4-Amino-6-iodo-1,3-benzenedisulfonamide (6)sChlorosulfonic acid
(40 mL) was reacted with m-iodoaniline (8 g) by analogy to reported
procedures.1,2 The final product was recrystallized from boiling water
to yield intermediate 6 (4 g, 29% yield) as a colorless solid: ESI-MS
(M - H) 376 (100).
4-Amino-1,3-benzenedisulfonamide (5)sInto 100 mL of anhydrous
ethanol was dissolved 6 (100 mg). The solution was photolyzed
according to the photolysis method previously described except that
a 100 mL, 35 cm × 2.5 cm quartz tube (Ace Glass, Louisville, KY)
was employed with a microspin bar for stirring during photolysis.
The dark solution was purified by preparative TLC to afford 5 (12
mg, 18% yield): ESI-MS (M - H) 250 (100); high-resolution FAB (M
+ H) C6H10N3O4S2 calcd 252.0112, found 252.0122; 1H NMR (CD3-
OD) δ 6.88 (1H, d, J ) 8.79 Hz, 5), 7.70 (1H, dd, J ) 2.44, 8.79 Hz,
6), 8.20 (1H, d, J ) 2.44 Hz, 2).
Figure 1
sHPLC−MS chromatogram of hydrochlorothiazide photolysis products
after 24 h. See Scheme 1 for peak component identification.
2H-1,2,4-benzothiadiazine-7-sulfonamide 1,1-Dioxide (3A)sStandard
5 (17 mg) was refluxed for 2 h with 5 mL of formic acid, by analogy
to reported procedures.2 The final purification by preparative TLC
afforded 3A (3.5 mg, 20% yield) as a colorless solid: ESI-MS (M - H)
260 (100); high-resolution FAB (M + H) C7H8N3O4S2 calcd 261.9956,
found 261.9964; 1H NMR (CD3OD) δ 7.39 (1H, d, J ) 8.79 Hz, 5),
7.83 (1H, s, 3) 8.07 (1H, dd, J ) 8.79, 1.96 Hz, 6) 8.33 (1H, d, J )
1.96, 8).
3,4-Dihydro-2H-1,2,4-benzothiadiazine-7-sulfonamide 1,1-Dioxide
(3)sStandard 5 (34 mg) was reacted with 6 mg of paraformaldehyde
by analogy to reported procedures.1 Final purification by preparative
TLC resulted in 3 (5.9 mg, 17%) as a colorless solid: ESI-MS (M -
H) 262 (100); high-resolution FAB (M + H) C7H10N3O4S2 calcd
264.0113, found 264.0116; 1H NMR (CD3OD) δ 4.76 (2H, s, 3), 6.84
(1H, d, J ) 8.79 Hz), 7.71 (1H, dd, J ) 2.44, 8.79 Hz, 6), 8.05 (1H, d,
J ) 2.44 Hz, 8).
4-Amino-6-methoxy-1,3-benzenedisulfonamide (4)sChlorosulfonic
acid (70 g) and m-anisidine (8.0 g) were reacted as described by
reported procedures5 to give 4 (4.95 g, 27% yield). The final purifica-
tion by a series of hot methanol extractions produced a slightly brown
solid: ESI-MS (M - H) 280 (100); high-resolution FAB (M + H)
C7H12N3O5S2 calcd 282.0218, found 282.0210; 1H NMR (CD3OD) δ
3.93 (3H, s, OCH3), 6.48 (1H, s, 5), 8.15 (1H, s, 2).
Scheme 1sHydrochlorothiazide (1), chlorothiazide (1A), their derivatives and
reactions.
6-Methoxy-3,4-dihydro-2H-1,2,4-benzothiadiazine-7-sulfonamide 1,1-
Dioxide (2)sStandard 4 (100 mg) was reacted with paraformaldehyde
(27.5 mg) as described by reported procedures.5 Final purification
by preparative TLC afforded 2 (31.5 mg, 30% yield): ESI-MS (M -
H) 292 (100); high-resolution FAB (M + H) C8H12N3O5S2 calcd
294.0218, found 294.0213; 1H NMR (CD3OD) δ 3.92 (3H, s, OCH3),
4.74 (2H, s, 3), 6.40 (1H, s, 5), 7.97 (1H, s, 8).
6-Methoxy-2H-1,2,4-benzothiadiazine-7-sulfonamide 1,1-Dioxide
(2A)sStandard 4 (25 mg) was heated to reflux for 1.5 h with 5 mL of
formic acid by analogy to reported procedures.2 The final product
2A (6.8 mg, 26.3% yield) was purified by preparative TLC as a
colorless solid: ESI-MS (M - H) 290 (100); high-resolution FAB (M
+ H) C8H10N3O5S2 calcd 292.0062, found 292.0060; 1H NMR (CD3-
OD) δ 4.05 (3H, s, OCH3), 6.89 (1H, s, 5), 7.85 (1H, s, 3), 8.25 (1H, s,
8).
afford 3 or by OCH3, from the methanol solvent, to produce
2. After a 24 h photolysis, the concentrations of 3 and 2 were
nearly equal, as determined by integration of the HPLC-MS
chromatogram. From Figure 2, the photolysis of 1 as followed
by HPLC-UV shows that 3 and 2 are major photolysis
products. A discussion of the photoreaction mechanisms of 1
has been reported.5 Parallel photolysis reactions in which the
methanol solution of 1 was not deoxygenated resulted in lower
yields of 2 and 3. A discussion of the mechanisms of how
oxygen scavenges free radicals in photoreactions has been
reported.9 The photodehalogenation reactions of 1 we ob-
served were consistent with those reported by other groups.5,6
PhotodehydrogenationsThe photodehydrogenation reac-
tions, in which hydrothiazides 1, 2, and 3 were dehydroge-
nated to thiazides 1A, 2A, and 3A, are shown in Scheme 1.
After 21 h of hydrochlorothiazide photolysis, the yields of
photodehalogenation products 2 and 3 were many times
greater than the yields of photodehydrogenation products 1A,
2A, and 3A, as viewed by HPLC-UV integration (Figure 2).
When the above photolysis of 1 was repeated without deoxy-
genating the methanol solution, the yields of the photodehy-
drogenation products were lowered. Previous photolysis
studies5,6 of 1 do not report photodehydrogenation reactions.
Thermal HydrolysissThe hydrolysis of thiazides 2 and 3
afforded the ring-opened thiazides 4 and 5 (Scheme 1).
Photolysis aliquots of 1 were analyzed immediately by HPLC-
Results
Hyd r och lor oth ia zid e (1)sAfter 1 was photolyzed for 24
h, the resulting solution was analyzed directly by HPLC-MS.
The chromatogram generated (Figure 1) shows the relative
intensities of each of the seven degradants and the starting
material. In Scheme 1 the decomposition reactions can be
divided into three types: photodehalogenation, photodehy-
drogenation, and thermal hydrolysis.
PhotodehalogenationsPhotodehalogenation is the primary
degradation process in which the Cl of 1 is replaced by H to
632 / Journal of Pharmaceutical Sciences
Vol. 86, No. 5, May 1997

way of confirming the structure, by use of the diode array
detector, the UV spectrum of the standard 3A was superim-
posed on the the UV spectrum of the photoproduct 3A (Figure
4). Previous photolysis studies report no photodehalogenation
products of 1A.6
Discussion
A discussion of a previously reported photolysis of hydro-
chlorothiazide (1)5 is presented. The source of UV light
employed was a medium-pressure mercury vapor lamp with
glass filter. The primary photoprocesses were photodehalo-
genation and photohydrolysis reactions. The rates of the two
reactions were about equal. No photodehydrogenation reac-
tions were reported. Some of the degradates were character-
ized by GC-MS. Standards were synthesized to help identify
some of the photoproducts reported. The photomechanisms
involving radical cation intermediates were discussed.
Our hydrochlorothiazide (1) photolysis results differ sig-
nificantly from those of previously reported studies.5,6 The
most significant difference is that Tamat and Moore5 report
hydrolyzed 1 (compound 7, Scheme 1) as a major photolysis
product. Ulvi and Tammilehto6 reported detecting hydrolysis
product 7 in small amounts. Compound 7 was not detected
in any of our hydrochlorothiazide photolysis studies.
Differences in the photolysis sources may explain this
apparent discrepancy. Tamat and Moore5 and Ulvi and
Tammilehto6 utilized medium- and high-pressure mercury
vapor photolysis sources, respectively. In our hydrochlorothi-
azide photolysis studies, a fluorescent UV-A photolysis source
was employed. The fluorescent source intensity wavelength
distribution differs remarkably from that of mercury vapor
sources.8,10
Figure 2sPhotolysis of hydrochlorothiazide (1) monitored by HPLC−UV.
Specifically, the UV-A radiation emitted by a medium-
pressure mercury vapor source is dominated by radiation at
313 and 365 nm.10 Although a high-pressure mercury vapor
source emits continuous radiation throughout the UV-A
spectral region, Ulvi and Tammilehto6 using filters only
utilized radiation near 313 nm. In contrast, a fluorescent
UV-A source emits a broad continuous radiation band cen-
tered at about 350 nm with less than 1% of the radiation at
or below 315 nm.8 Thus, it is possible that photohydrolysis
of hydrochlorothiazide (1) to compound 7 is produced pre-
dominately by 313 nm radiation from the mercury vapor
photolysis sources. This high-energy radiation is only a minor
component of the radiation from a fluorescent UV-A photolysis
source.
Our studies indicate a hydrochlorothiazide photodehydro-
genation process, yielding small amounts of compounds 1A,
2A, and 3A (Scheme 1). Tamat and Moore5 did not observe
this photolysis process. It is likely that small amounts of these
compounds were simply not detected by Tamat and Moore’s5
isocratic HPLC method monitored at 254 nm. Our HPLC
monitoring system required gradient elution and small par-
ticle (5 µm) C-18 stationary phase conditions to resolve the
dehydrogenation products from the corresponding saturated
species (Scheme 1, compounds 1-3). Further, in our studies
the HPLC monitoring wavelength of 270 nm, rather than 254
nm used by Tamat and Moore, offers greater sensitivity for
detecting photodehydrogenation products (see Figure 4, UV
spectra of compounds 3 and 3A).
Figure 3sPhotolysis of chlorothiazide (1A) monitored by HPLC−UV.
UV (Figure 2) and showed yields of 4 and 5 to be many times
lower than the yields of photodehalogenation products 2 and
3. After the photolysis aliquots were allowed to stand in the
dark for a few hours, the aliquots were analyzed again by
HPLC-UV. The results showed that the rate of hydrolysis
to 4 and 5 in the dark was approximately equal to the rate of
hydrolysis to 4 and 5 under UV light exposure. It appears
that the hydrolyses of 3 and 2 to 5 and 4 go well without UV
light. Previous studies5 report that the rate of photohydrolysis
of 1 was approximately equal to the rate of photodehaloge-
nation reactions. Although considerable effort was made to
identify the hydrolysis product 7 in the photolysate of 1, the
reported major photolysis product5 was not observed by
HPLC-UV, HPLC-MS, or HPLC coinjection experiments.
Ch lor oth ia zid e (1A)sPhotodehalogenationsAfter 1A was
irradiated for 2 days, the resulting solution was analyzed by
HPLC-MS. From the (M - H) molecular ions of each
degradant, m/z 260 and 290, structures 3A and 2A, respec-
tively, were proposed on the basis of photodehalogenation
reactions (Scheme 1). Standards 2A and 3A were synthesized
and shown to be identical to the corresponding degradants
by HPLC coinjection experiments and by HPLC-MS and
HPLC-UV. From Figure 3, after 5 days, the yield of 3A (Cl
replacement by H) is many times the yield of 2A (Cl replace-
ment by OCH3) as viewed by HPLC-UV integration. As a
The unique photodegradation product distributions for 1
and 1A we have reported can be attributed to the UV-A lamps
employed and the method of product detection and identifica-
tion. The photolysates generated in these experiments were
analyzed directly by HPLC-MS and HPLC-UV with no
pretreatment or derivatization. The sensitivity observed was
sufficient to identify and quantify decomposition products at
Journal of Pharmaceutical Sciences / 633
Vol. 86, No. 5, May 1997

Scheme 2sSynthesis of thiazide derivatives.
and Moore5 in that we did not find hydrolysis product 7 after
diligent effort by several methods. In addition they5 did not
find the dehydrogenation products which we identified. The
results of our photolysis of chlorothiazide (1A) differ from
those reported by Ulvi and Tammilehto6 in that we identified
photodehalogenation products. The differences in product
distribution are attributed mainly to differences in the emis-
sions of the lamps employed.
References and Notes
Figure 4sRepresentative UV spectral match of standards and photoproducts.
1. Werner, L. H.; Halamandaris, A.; Ricca, S.; Dorfman, L.;
Stevens, G. J . Am. Chem. Soc. 1960, 82, 1161-1166.
2. Novello, F. C.; Sprague, J . M. J . Am. Chem. Soc. 1957, 79, 2028-
2029.
3. Mollica, J . A.; Rehm, C. R.; Smith, J . B.; Govan, H. K. J . Pharm.
Sci. 1971, 60, 1380-1384.
4. Moore, D. E. J . Pharm. Sci. 1977, 66, 1282-1284.
5. Tamat, S. R.; Moore, D. E. J . Pharm. Sci. 1983, 72, 180-183.
6. Ulvi, V.; Tammilehto, S. Acta Pharm. Nord. 1989, 1, 195-200.
7. International Conference on Harmonization (ICH) Draft Step 2
Tripartite Guidelines for the Photostability Testing of Drug
Substances and Products, J uly, 1995.
8. Riehl, J . P.; Maupin, C. L.; Layloff, T. P. Pharmacopeial Forum
1995, 21, 1654-1663.
9. Moore, D. E. J . Pharm. Biomed. Anal. 1987, 5, 441-453.
10. Calvert, J . G.; Pitts, J . N. Photochemistry; Wiley and Sons: New
York, 1966; p 696.
very low concentration, for example 4 and 5 in Figure 1. The
(M - H) molecular ion suggested the structure of the photo-
degradant. If the tentatively identified structure was logical
from the standpoint of known photochemical processes, a
standard was synthesized and shown to be identical to the
corresponding photodegradant by HPLC coinjection experi-
ments and by HPLC-MS and HPLC-UV. By use of a diode
array detector, the UV spectrum of the standard was super-
imposed on the UV spectrum of the corresponding photoprod-
uct to confirm its identity (Figure 4). The structure of the
standard was further defined by high-resolution FAB-MS and
1
by H NMR.
Previous investigators have expressed difficulty in synthe-
sizing standard 3.5 We attempted unsuccessfully to make
standard 3 by the synthetic route previously reported.5
Eventually we solved the problem by following our own route
(Scheme 2). The key step in our synthesis of derivative 3 was
the preparative photolysis (photodehalogenation) of the iodo
intermediate 6 to give standard 5.
Acknowledgments
We thank D. Andre d'Avignon of Washington University, St. Louis,
MO, for NMR support. We thank Henry D. Drew and Walter L.
Zielinski of the Division of Drug Analysis, St. Louis, MO, for many
helpful discussions. We thank Dennis D. Senderovich of the Division
of Drug Analysis for technical support. High-resolution mass spec-
trometry was provided by the Washington University Mass Spec-
trometry Resource, St. Louis, MO, an NIH Research Resource (Grant
No. P41RR0954). Certain equipment, instruments, or materials are
identified in this report to specify adequately the experimental
procedure. Such identification does not imply recommendation or
endorsement by the Food and Drug Administration nor does it imply
that the materials or equipment identified are necessarily the best
available for the purpose.
Conclusions
The primary photochemical process in the photolysis of
thiazides 1 and 1A employing a UV-A lamp defined by ICH
is photodehalogenation. Irradiation of 1 also produces sig-
nificant amounts of photodehydrogenation products and ther-
mal hydrolysis products. The results of our photolysis of
hydrochlorothiazide (1) differ from those reported by Tamat
J S9601501
634 / Journal of Pharmaceutical Sciences
Vol. 86, No. 5, May 1997