Profiling indomethacin impurities using high-performance liquid chromatography and nuclear magnetic resonance

Article abstract of DOI:10.1016/S0928-0987(01)00198-1

The anti-inflammatory drug indomethacin was investigated regarding new related impurities. Therefore, related substances 2-9 were prepared by independent synthesis and physicochemically characterized. To determine indomethacin and its related substances, a new HPLC-UV method was developed and validated. Indomethacin and its impurities were eluted on a C₁₈ column with a mobile phase consisting of methanol and an aqueous solution of 0.2% phosphoric acid at a flow rate of 1.5 ml/min and were quantified by UV detection at 320 nm. Overall, the HPLC-UV method was simple and reliable for the detection of eight impurities in indomethacin. In addition to the HPLC-UV method, ¹H nuclear magnetic resonance (NMR) was used to investigate indomethacin regarding impurities. For that purpose, related substances 2-9 were systematically added to indomethacin and investigated. The NMR method was found to be very useful for the identification of impurities in bulk substance without prior separation. Both HPLC-UV and NMR were used to analyze 38 batches of indomethacin available on the European market. The outcome was that 42% of the batches did not meet the compendial requirements although they met the specifications of current compendial methods. Some batches contained the previously undescribed impurity 8, while other batches contained by-products from two distinct synthetic routes. The methods presented herein are important contributions to the ongoing efforts to reduce impurities and therefore the risk of adverse side-effects in drugs that are no longer under patent protection.

Full text of DOI:10.1016/S0928-0987(01)00198-1

European Journal of Pharmaceutical Sciences 14 (2001) 301–311
www.elsevier.nl/locate/ejps
Profiling indomethacin impurities using high-performance liquid
chromatography and nuclear magnetic resonance
*
Sonja Hess , Uwe Teubert, Jutta Ortwein, Kurt Eger
Institute of Pharmacy, University of Leipzig, Liebigstrasse 18, D-04103 Leipzig, Germany
Received 7 December 2000; received in revised form 29 May 2001; accepted 8 August 2001
¨
Dedicated to Professor Dr H. Oelschlager, Institute of Pharmacy, University of Jena, on the occasion of his 80th birthday.
Abstract
The anti-inflammatory drug indomethacin was investigated regarding new related impurities. Therefore, related substances 2–9 were
prepared by independent synthesis and physicochemically characterized. To determine indomethacin and its related substances, a new
HPLC–UV method was developed and validated. Indomethacin and its impurities were eluted on a C18 column with a mobile phase
consisting of methanol and an aqueous solution of 0.2% phosphoric acid at a flow rate of 1.5 ml/min and were quantified by UV detection
at 320 nm. Overall, the HPLC–UV method was simple and reliable for the detection of eight impurities in indomethacin. In addition to the
HPLC–UV method, 1H nuclear magnetic resonance (NMR) was used to investigate indomethacin regarding impurities. For that purpose,
related substances 2–9 were systematically added to indomethacin and investigated. The NMR method was found to be very useful for the
identification of impurities in bulk substance without prior separation. Both HPLC–UV and NMR were used to analyze 38 batches of
indomethacin available on the European market. The outcome was that 42% of the batches did not meet the compendial requirements
although they met the specifications of current compendial methods. Some batches contained the previously undescribed impurity 8, while
other batches contained by-products from two distinct synthetic routes. The methods presented herein are important contributions to the
ongoing efforts to reduce impurities and therefore the risk of adverse side-effects in drugs that are no longer under patent protection.
2001 Published by Elsevier Science B.V.
Keywords: Reversed-phase chromatography; Indomethacin; Impurities; Validation; Nuclear magnetic resonance
1. Introduction
protection (Denk et al., 2000a,b; Eger et al., 1999; Hess et
al., 2001a,b; Lehr et al., 1999; Toussaint et al., 2000). Due
to small modifications in the synthesis of these drugs, new
and unknown impurities can occur and may not be
observed when investigated with compendial methods.
Recently, the observation of new impurities in trimetho-
prim batches led to the revision of this monograph in the
European Pharmacopoeia (Eur. Pharm. Comm., 2000a;
Eger et al., 1999; Hess et al., 2001a,b).
With the aim of reducing the risk of adverse side-effects
arising from widely administered drugs such as indo-
methacin, 38 industrial batches were investigated on
purity. All batches met the specifications of current
compendial methods. However, the current European
Pharmacopoeia test (Eur. Pharm. Comm., 2000b) on
related substances in indomethacin is limited to the
detection of 4-chlorobenzoic acid, investigated by thin-
layer chromatography. Therefore, it was desirable to
develop other methods that allowed for the investigation of
Indomethacin
([1-(4-chlorobenzoyl)-5-methoxy-2-
methylindol-3-yl]acetic acid, 1; Fig. 1) was introduced to
the market in 1963, and since then it has been widely used
in musculoskeletal and joint disorders such as rheumatoid
arthritis. It may also be used in inflammation, pain and
oedema (Hardman et al., 1990). For the determination of
indomethacin (1) in blood or plasma several chromato-
graphic procedures have been developed (Tsai and Naito,
1981; Sato et al., 1997). So far little has been published on
impurities in the drug substance.
As has been reported earlier, there is a need for the
reevaluation of drugs that are no longer under patent
*
Corresponding author. Present address: LBC, NIDDK, Bldg. 8, Rm.
B2A27, National Institutes of Health, Bethesda, MD 20892-0805, USA.
Tel.: 11-301-402-2197; fax: 11-301-402-1967.
E-mail address: sonja]hess@nih.gov (S. Hess).
0928-0987/01/$ – see front matter
PII: S0928-0987(01)00198-1
2001 Published by Elsevier Science B.V.

302
S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
Fig. 1. Compounds investigated in this study.
a variety of related substances. This is especially true
because other compendial tests (Brit. Pharm. Comm.,
2000; USP24, 1999a) are also limited to the investigation
of 4-chlorobenzoic acid.
benzoic acid, indomethacin batches were analyzed using
HPLC–UV and NMR. Among the batches tested, 42%
showed impurities that could not be identified with com-
pendial methods. Therefore, the HPLC–UV and NMR
methods described herein are important contributions for
the quality control of indomethacin.
The search for impurities in indomethacin is difficult
because at least four different synthetic routes have been
described (Hartke et al., 1997; Shen et al., 1963; Yama-
moto, 1968). Syntheses are based on a [3.3] sigmatrope
cyclization (Fischer indole synthesis) of an appropriate
methoxyphenylhydrazone derivative to give the corre-
sponding indole derivative. Acylation with an activated
chlorobenzoyl derivative is accomplished either before,
simultaneously, or after the [3.3] sigmatrope cyclization.
Depending on the chosen procedure and the used protect-
ing group (methyl, ethyl or tert. butyl), different impurities
can arise.
Impurities 2–9 (Figs. 1 and 2) were synthesized based
on theoretical considerations of possible process im-
purities. Standard procedures have been applied to syn-
thesize and characterize known (2–4, 7) and previously
undescribed (5, 8–9) impurities (Shen et al., 1963; Yama-
moto, 1968).
In the course of this study, an HPLC–UV method was
developed which allowed for the separation and quantifica-
tion of the possible process impurities 2–9 from indo-
methacin (1). The HPLC–UV method described herein is
validated with the following parameters: precision, accura-
cy, recovery, linearity and range, limit of quantification,
and ruggedness of the method. Additionally, the identity of
the impurities was verified using 1H nuclear magnetic
resonance. As previously described, it was possible to
identify impurities directly in bulk drug substance without
prior separation of the impurities (Hess et al., 2001a;
Holzgrabe et al., 1999; USP24, 1999b). To test whether
industrial batches contain impurities other than 4-chloro-
2. Experimental
2.1. Chemicals and reagents
All chemicals and reagents were of the highest purity
commercially available. As the standard, indomethacin
chemical reference standard (CRS) was used. 5-Methoxy-
2-methyl-3-indoleacetic acid (6, 98.3% purity), 2,3-di-
methyl-5-methoxyindole (11) and levulinic acid (13) were
purchased from Aldrich (Steinheim, Germany). All other
impurities were synthesized de novo. The purity was
determined using HPLC and was higher than 98.8%.
Methanol was HPLC gradient grade quality and supplied
by Merck (Darmstadt, Germany); 85% phosphoric acid
¨
was analytical grade and supplied by Grussing (Filsum,
¨
Germany). Ion exchanged water (UO 100, Letzner, Huck-
eswagen, Germany) was further purified by bidistillation.
The aqueous phosphoric acid solution was filtered off
through a cellulose acetate filter (0.45 mm) before use.
2.2. Syntheses
NMR spectra were measured as described under Section
2.5. Infrared spectra were recorded on a Perkin-Elmer
FT-IR PC 16 spectrophotometer. All compounds were

S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
303
Fig. 2. Syntheses of impurities 2–5 and 7–9. Conditions: (a) thionyl chloride, absolute methanol or ethanol; (b) dicyclohexylcarbodiimide, 4-
dimethylaminopyridine, tert. butanol, alcohol, chloroform; (c) 1,19-carbonyldiimidazole, 4-dimethylaminopyridine, 10, THF; (d) NaH, DMF, 4-chloro-
benzoyl chloride; (e) acetic acid, DT; (f) 4-chlorobenzoyl chloride, pyridine, 08C.
checked for purity by TLC on 0.2-mm aluminum sheets
with silica gel 60 F254 (Merck). Melting points were taken
try, Leipzig, Germany, and were within 60.4% of the
theoretical values.
¨
on a Buchi 535 melting point apparatus and are un-
corrected. Thermospray and electron ionization mass spec-
tra were recorded on an MS Engine HP 5989A mass
spectrometer (Hewlett-Packard). Elemental analyses for C,
H, N were determined at the Institute of Organic Chemis-
2.2.1. Indomethacin methyl ester, [1-(4-chlorobenzoyl)-
5-methoxy-2-methylindol-3-yl]acetic acid methylate (2)
A suspension of 0.5 g (1.4 mmol) 1 in 10 ml of absolute
methanol was cooled down and treated with 0.1 ml of

304
S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
thionyl chloride. The mixture was allowed to reach room
temperature and was stirred for 12 h. The precipitate was
collected by filtration, and the residue was recrystallized
from methanol.
1H, indole-CH, J53.5, 9.0 Hz), 6.93 (d, 1H, indole-CH,
J59.0 Hz), 7.01 (d, 1H, indole-CH, J53.5 Hz), 7.66 (m,
4H, phenyl-CH); 13C NMR d [ppm] 13.9 (CH3), 28.5
(C(CH3)3), 31.5 (CH2), 56.1 (OCH3), 81.0 (C(CH3)3),
102.3, 112.2, 115.3 (indole-CH), 114.0 (C-3), 129.8, 131.9
(phenyl-CH), 131.0, 131.3 (C-3a, C-7a), 134.9 (C-19),
135.9 (C-2), 138.4 (C-49), 156.3 (C-5), 168.6, 170.5
(C=O); MS (EI) 413 [M]1.
Yield: 0.4 g (77%); M.p. 91.38C; IR n [cm21] 3446,
2931, 1732, 1672, 1477, 1467, 1434, 1365, 1326, 1235,
1
1193, 806; H NMR d [ppm] 2.22 (s, 3H, CH3), 3.63 (s,
3H, COOCH3), 3.76 (s, 3H, OCH3), 3.78 (s, 2H, CH2),
6.71 (dd, 1H, indole-CH, J52.5, 9.1 Hz), 6.92 (d, 1H,
indole-CH, J59.1), 7.03 (d, 1H, indole-CH, J52.5 Hz),
7.66 (dt, 4H, phenyl-CH, J52.2, 8.7 Hz); 13C NMR d
[ppm] 13.1 (CH3), 29.1 (CH2), 51.8 (COOCH3), 55.4
(OCH3), 101.6, 111.4, 114.6 (indole-CH), 112.7 (C-3),
130.2, 130.5 (C-3a, C-7a), 134.1 (C-19), 135.4 (C-2),
137.6 (C-49), 155.6 (C-5), 129.0, 131.2 (phenyl-CH),
167.8, 171.0 (C=O); MS (EI) 371 [M]1.
2.2.4. 4-Chlorobenzoic acid N9-h2-[1-(4-chlorobenzoyl)-
5-methoxy-2-methyl-1H-indole-3-yl]-acetylj-N-(4-meth-
oxyphenyl)-hydrazide (5)
To a suspension of 0.25 g (0.80 mmol) N-(4-methoxy-
phenyl)-4-chlorobenzoic acid hydrazide HCl (10) in 5 ml
of tetrahydrofuran 0.064 ml (0.8 mmol) of pyridine was
added and stirred at room temperature for 1 h. This
suspension was filtered off. The filtrate was added to a
solution of 0.30 g (0.84 mmol) 1, 0.14 g (0.85 mmol)
1,19-carbonyldiimidazole, and a catalytic amount of 4-
dimethylaminopyridine in 10 ml of tetrahydrofuran under
argon atmosphere. The mixture was refluxed for 8 h. The
solvent was removed in vacuo, and the residue was
dissolved in ethyl acetate and chromatographed on silica
(4532 cm) using ethyl acetate. The fractions were com-
bined and the solvent was removed in vacuo. The residue
was recrystallized from ethyl acetate.
2.2.2. Indomethacin ethyl ester, [1-(4-chlorobenzoyl)-5-
methoxy-2-methylindol-3-yl]acetic acid ethylate (3)
Prepared analogously to 2; ethanol was used instead of
methanol. To separate 3 from indomethacin traces, the raw
material was chromatographed on silica (4032 cm) using
dichloromethane/acetone (9/1).
Yield: 0.42 g (78%); M.p. 99.58C; IR n [cm21] 3440,
2993, 1728, 1676, 1478, 1467, 1374, 1360, 1324, 1236,
1186, 1090, 1036, 868; 1H NMR d [ppm] 1.27 (t, 3H,
CH2CH3, J57.1 Hz), 2.39 (s, 3H, CH3), 3.66 (s, 2H,
CH2), 3.84 (s, 3H, OCH3), 4.16 (q, 2H, CH2CH3, J57.1
Hz), 6.69 (m, 1H, indole-CH), 6.88 (d, 1H, indole-CH,
J59.1 Hz), 6.98 (d, 1H, indole-CH, J52.5 Hz), 7.57 (dt,
4H, phenyl-CH, J51.9, 6.6 Hz); 13C NMR d [ppm] 13.4
(CH3), 14.3 (CH2CH3), 30.5 (CH2), 55.7 (OCH3), 61.0
(CH2CH3), 101.4, 111.7, 115.0 (indole-CH), 112.8 (C-3),
130.7, 130.9 (C-3a, C-7a), 134.0 (C-19), 135.9 (C-2),
139.3 (C-49), 156.1 (C-5), 129.1, 131.2 (phenyl-CH),
168.3, 170.9 (C=O); MS (EI) 385 [M]1.
Yield: 0.22 g (44%); M.p. 175.68C; IR n [cm21] 3446,
2958, 1689, 1510, 1478, 1439, 1400, 1360, 1360, 1313,
1251, 1224, 1091, 834; 1H NMR d [ppm] 2.34 (s, 3H,
CH3), 3.72 (s, 2H, CH2), 3.76 (s, 3H, OCH3), 3.81 (s, 3H,
OCH3), 6.71 (m, 3H, indole-CH, phenyl-CH), 6.93 (m, 2H,
indole-CH, phenyl-CH), 7.09 (d, 2H, indole-CH, phenyl-
CH, J511.8 Hz), 7.17 (d, 2H, phenyl-CH, J511.8 Hz),
7.34 (d, 2H, phenyl-CH, J511.8 Hz), 7.45 (d, 2H, phenyl-
CH, J511.8 Hz), 7.59 (d, 2H, phenyl-CH, J511.8 Hz),
8.12 (s, br, 1H, NH); 13C NMR d [ppm] 13.4 (CH3), 30.7
(CH2), 55.5 (OCH3), 55.8 (OCH3), 100.9, 112.6, 114.3
(indole-CH, arom. CH), 111.5 (C-3), 115.1, 128.19,
128.24, 129.2, 131.2 (arom. CH), 139.6 (C-49), (C-5),
127.7, 127.8, 130.08, 130.13, 130.9, 133.5, 136.6, 136.7 (C
arom. quart.), 156.4, 158.7, 168.4 (C=O); MS (EI) 616
[M1H]1.
2.2.3. Indomethacin tert. butyl ester, [1-(4-chloro-
benzoyl)-5-methoxy-2-methylindol-3-yl]acetic acid tert.
butylate (4)
A solution of 0.5 g (1.4 mmol) 1, 0.5 g (6.9 mmol) tert.
butanol, 0.32 g (1.54 mmol) dicyclohexylcarbodiimide,
and 0.017 g (0.14 mmol) 4-dimethylaminopyridine in 15
ml of chloroform was stirred at room temperature for 24 h.
Precipitated dicyclohexylurea was filtered off and washed
with 30 ml of chloroform. The organic layer was washed
once with 50 ml of 5% acetic acid, and once with 50 ml of
water. The organic layer was dried over Na2SO4, and the
solvent was removed in vacuo and chromatographed on
silica (4532 cm) using ethyl acetate/petrol ether/acetic
acid (6/10/0.1).
2.2.5. 1-(4-Chlorobenzoyl)-5-methoxy-2,3-dimethylindole
(7) (Merck and Co., 1968)
To a suspension of 0.28 g (6.8 mmol) sodium hydride
(60%) in 10 ml of DMF in an ice-bath, a solution of 1.0 g
(5.7 mmol) 2,3-dimethyl-5-methoxyindole (11) in DMF
was added dropwise. The mixture was allowed to reach
room temperature and stirred for 30 min. Then 0.81 ml
(6.8 mmol) 4-chlorobenzoyl chloride in 3 ml of DMF were
added and stirred for 12 h at room temperature. The
solution was added slowly with stirring to 250 ml of
ice-cold 5% HCl. The aq. phase was extracted 53 with 50
ml of dichloromethane. The combined organic layer was
Yield: 0.4 g (69%); M.p. 109.5–1118C; IR n [cm21
]
2970, 1731, 1729, 1690, 1686, 1682, 1480, 1457, 1400,
1368, 1330, 1318, 1205, 1162, 1149, 1087, 1076, 838, 795,
1
753; H NMR d [ppm] 1.40 (t, 9H, C(CH3)3), 2.21 (s, 3H,
CH3), 3.64 (s, 2H, CH2), 3.76 (s, 3H, OCH3), 6.71 (dd,

S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
305
washed twice with 100 ml of water and 100 ml of sat.
NaCl solution. The organic phase was dried over Na2SO4
and the solvent was evaporated in vacuo. The residue was
chromatographed on silica (column 532 cm) using ethyl
acetate/petrol ether (1/9). The fractions were combined
and the solvent was removed in vacuo. The residual oil
was suspended in 20 ml of methanol; the precipitate was
collected by filtration and dried in vacuo.
solvent was removed in vacuo. The residue was recrystal-
lized from toluene.
Yield: 0.4 g (62%); M.p. 147–1488C; IR n [cm21
]
3245, 1696, 1636, 1593, 1509, 1484, 1360, 1302, 1251,
1170, 1094, 1015, 843; 1H NMR d [ppm] 3.76 (s, 3H,
OCH3), 6.75 (d, 2H, phenyl-CH, J58.9 Hz), 7.19 (dd, 4H,
phenyl-CH, J57.8, 8.3 Hz), 7.28 (t, 2H, phenyl-CH, J5
10.1 Hz), 7.39 (d, 2H, phenyl-CH, J510.1 Hz), 7.72 (t,
2H, phenyl-CH, J58.3 Hz), 9.66 (s, 1H, NH); 13C NMR d
[ppm] 55.5 (OCH3), 114.4, 128.1, 128.4, 128.9, 128.9,
130.4 (phenyl-CH), 129.8, 132.3, 137.1, 138.8 (C quart.),
159.0, 165.5 (C=O); MS (TS-POS) 415 [M1H]1.
Yield: 0.5 g (28%); M.p. 81–82.58C; IR n [cm21] 3440,
3058, 2924, 1687, 1610, 1482, 1365, 1313, 1262, 1224,
1156, 1047, 835, 797; 1H NMR d [ppm] 2.20 (s, 3H,
CH3), 2.32 (s, 3H, CH3), 3.86 (s, 3H, OCH3), 6.69 (dd,
1H, indole-CH, J52.7 Hz), 6.92 (d, 2H, indole-CH, J58.8
Hz), 7.56 (dt, 4H, phenyl-CH, J56.6 Hz, 1.9 Hz); 13C
NMR d [ppm] 8.8, 13.4 (23CH3), 55.8 (OCH3), 101.4,
111.2, 115.0 (indole-CH), 115.5 (C-3), 130.9, 132.0 (C-3a,
C-7a), 133.8 (C-2), 134.4 (C-19), 138.9 (C-49), 156.0
(C-5), 129.1, 131.1 (phenyl-CH), 168.3 (C=O); MS (TS-
POS) 313 [M1H]1.
2.3. Liquid chromatography system
The LC system included a Bischoff HPLC pump, a
Gynkotek GINA 50 Autosampler, a Bischoff UV detector
LAMBDA 1000, and a Shimadzu registrator and integrator
C-R6A. Chromatographic separation was achieved iso-
cratically at ambient temperature in reversed phase col-
umns of the following type: Nucleosil 120-5, C18, 250
mm34.6 mm I. D. with a pore diameter of 10 nm
2.2.6. [1-(3,4-Dichlorobenzoyl)-5-methoxy-2-methylindol-
3-yl]acetic acid (8)
A solution of 1.0 g (2.9 mmol) N-(4-methoxyphenyl)-
3,4-dichlorobenzoic acid hydrazide HCl (10) and 0.7 g
(5.8 mmol) levulinic acid (13) in 20 ml of glacial acetic
acid was refluxed for 1 h at 858C. After cooling, the
solution was filtered off, and the filtrate was reduced in
vacuo. The residue was dissolved in 30 ml of ethanol and
precipitated by dropwise addition of water. The precipitate
was collected by filtration and recrystallized from ethanol/
water (1/1).
¨
(Macherey-Nagel, Duren, Germany).
The mobile phase consisted of 75 vol. of methanol and
25 vol. of 0.2% (v/v) phosphoric acid. The mobile phase
was degassed with ultrasound (15 min) before use. Flow
rate was 1.5 ml/min. The system was equilibrated for 30
min. Indomethacin (1) and its by-products were monitored
at 320 nm. The injection volume was 20 ml. The attenua-
tion of the registrator was adjusted to 6.0 (more sensitive)
and 8.0 (less sensitive).
Yield: 0.22 g (44%); M.p. 130–132.58C; IR n [cm21
]
3432, 2933, 1684, 1609, 1478, 1437, 1385, 1357, 1323,
1
1289, 1224, 1179, 1154, 1034, 833, 751; H NMR d [ppm]
2.22 (s, 3H, CH3), 3.66 (s, 2H, CH2), 3.76 (s, 3H, OCH3),
6.75 (dd, 1H, indole-CH, J53.2, 12.3 Hz), 7.02 (dd, 2H,
indole-CH, J53.2, 12.3 Hz), 7.61 (dd, 1H, phenyl-CH,
J52.5, 11.5 Hz), 7.83 (d, 1H, phenyl-CH, J511.5 Hz),
7.93 (d, 1H, phenyl-CH, J52.5 Hz); 13C NMR d [ppm]
13.3 (CH3), 29.5 (CH2), 55.4 (OCH3), 101.8, 111.4, 114.8
(indole-CH), 113.8 (C-3), 130.1, 131.8, 131.2 (C-3a, C-7a,
C-39), 135.1, 135.4, 135.9 (C-19, C-2, C-49), 155.7 (C-5)
129.2, 131.0, 131.2 (phenyl-CH), 166.5, 172.0 (C=O); MS
(TS-NEG) 392 [M–H]2.
2.4. Preparation of the standard and sample solutions
Indomethacin standard (CRS) stock solutions were
prepared (1 mg/ml) in methanol. The solutions of the
impurities were prepared analogously. Calibration refer-
ence solutions were prepared by spiking the CRS solution
with each impurity stock solution separately in concen-
trations ranging from 0.1 up to 1.9% of the nominal
concentration of 1. Calibration curves for standard and
impurities were recorded. Quantification was calculated
from calibration curves.
2.2.7. 4-Chlorobenzoic acid N9-(4-chlorobenzoyl)-N-(4-
methoxyphenyl)-hydrazide (9)
Sample solutions containing indomethacin at a nominal
concentration of 1.0 mg/ml in the mobile phase were
prepared separately for each batch and investigated. To
check the system suitability, a resolution test was run
before and after each sequence. A mixture of 5 mg/ml
indomethacin CRS and 5 mg/ml [1-(4-chlorobenzoyl)-5-
methoxy-2-methylindol-3-yl]acetic acid ethylate (3) was
used for the test. The resolution Rs of the two principal
peaks had to be at least 11, and was determined according
to the method of the European Pharmacopoeia (Eur.
Pharm. Comm., 1997) with the following equation:
To a solution of 0.5 g (1.6 mmol) N-(4-methox-
yphenyl)-4-chlorobenzoic acid hydrazide HCl (14) in 5 ml
of pyridine in an ice-bath, 0.25 g (1.9 mmol) 4-chloro-
benzoyl chloride was added and the suspension was stirred
under cooling for 3 h. The mixture was then slowly added
to 100 ml of ice water. The precipitate was collected by
filtration. The residue was dissolved in 50 ml of chloro-
form and washed with 50 ml of 5% HCl and 50 ml of
water. The organic layer was dried over Na2SO4 and the

306
S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
3.2. Chromatography
(trb 2 tra)
0.5a 1 b0.5b
]
Rs 5 1.18*
(1)
b
An HPLC–UV method was developed to separate the
following impurities from indomethacin (1):
where tra and trb are the retention times of the first and
second peak, respectively. The terms b0.5a and b0.5b
represent the width of the peaks at the peak height divided
by 2.
• [1-(4-chlorobenzoyl)-5-methoxy-2-methylindol-3-yl]-
acetic acid methylate (2),
• [1-(4-chlorobenzoyl)-5-methoxy-2-methylindol-3-yl]-
acetic acid ethylate (3),
• [1-(4-chlorobenzoyl)-5-methoxy-2-methylindol-3-yl]-
acetic acid tert. butylate (4),
• 4-chlorobenzoic acid N9-h2-[1-(4-chlorobenzoyl)-5-
methoxy-2-methylindol-3-yl]acetylj-N-[4-methoxy-
phenyl)-hydrazide (5),
• 5-methoxy-2-methyl-3-indoleacetic acid (6),
• [1-(4-chlorobenzoyl)-5-methoxy-2,3-dimethylindole
(7),
• [1-(3,4-dichlorobenzoyl)-5-methoxy-2-methylindol-3-
yl]acetic acid (8),
2.5. Nuclear magnetic resonance
1H NMR and 13C NMR spectra were determined on a
Varian Gemini 300 spectrometer (1H: 300 MHz, 13C: 75
MHz) using d6-DMSO. The chemical shifts of the remain-
ing protons of the deuterated solvents served as internal
standard.
The number of pulses in the 1H experiments for the
determination of the purity of the batches was 500. As a
reference, 10 mg indomethacin CRS was spiked with 0.1%
of each impurity separately, investigated, and compared to
the samples. Samples consisted of a solution of 10 mg of
each indomethacin batch, dissolved in 1 ml of DMSO.
• and 4-chlorobenzoic acid N9-(4-chlorobenzoyl)-N-(4-
methoxyphenyl)-hydrazide (9) (Fig. 1).
The retention times are given in Table 1. A sample
spiked with all investigated impurities (2–9) is shown in
Fig. 3. Compounds 6 and 9 were eluted before indo-
methacin with relative retention times of 0.47 and 0.85,
respectively. The remaining impurities were observed after
the indomethacin peak at 5.8 min. The esters 2, 3, and 4
were eluted at 8.7, 9.9 and 16.1 min, respectively. The
dichloro derivative (8) was eluted at 8.8 min, the hydrazide
(5) at 10.2 min, and the dimethylindole (7) at 16.4 min.
The easily available ethyl ester 3 was selected as the
standard for the system suitability test.
Impurities 2 and 8 coeluted and could therefore not be
unambiguously assigned without further data collection
such as DAD, MS, or NMR. Although impurities 3 and 5,
as well as compounds 4 and 7, did not show complete
baseline assignment, the peaks could be assigned to one
impurity by means of their retention times. The main goal,
the separation of impurities 2–9 from indomethacin, was
achieved with the presented isocratic method. Therefore,
the LC method provided the necessary specifity and
selectivity for the investigation of impurities in indo-
methacin.
3. Results and discussion
3.1. Syntheses
In conversion to the synthesis of the drug indomethacin,
compounds 2–5 were prepared using indomethacin CRS as
starting material (Fig. 2). Ester derivatives 2–5 were
synthesized with standard procedures using either thionyl
chloride or dicyclohexylcarbodiimide and 4-dimethyl-
aminopyridine or 1,1-carbonylimidazole and 4-dimethyl-
aminopyridine in order to facilitate the ester formation. In
batches of indomethacin, impurities 2–4 resulted from
insufficient cleavage of the methyl, ethyl, or tert. butyl
ester in the last step of each synthetic route. Impurity 5
could be obtained from reaction with hydrazide 10 fol-
lowed by insufficient cleavage. Derivative 6 (Fig. 1)
represented an impurity arising from incomplete reaction
with chlorobenzoyl chloride, whereas impurity 7 could be
formed after thermal or photochemical decarboxylation of
indomethacin (1). Impurity 7 (Fig. 2) was synthesized
from dimethylindole derivative 11, which was reacted with
4-chlorobenzoyl chloride (Merck and Co., 1968). Dichloro
compound 8 was prepared in analogy to the synthesis of
indomethacin developed by Yamamoto (Yamamoto, 1968).
Dichlorobenzoic acid was first reacted with 4-methoxy-
phenylhydrazone followed by [3.3] sigmatrope cyclization
with levulinic acid. Because one chloro substituent facili-
tates the introduction of a second chloro substituent in
aromatic rings, this by-product formation seemed to be
very likely. Finally, the reaction of N-(4-methoxyphenyl)-
4-chlorobenzoic acid 14 with an excess of 4-chlorobenzoyl
chloride yielded the potential impurity 9.
3.3. Linearity
Quantification of indomethacin and its impurities was
based on the calibration of the UV response of indo-
methacin CRS and the impurities determined at 320 nm.
This wavelength was found to be the most appropriate
because indole derivatives like indomethacin show a
maximum at this wavelength. Linearity of indomethacin
was checked ranging from 0.005 to 1.00 mg/ml, and
linearity of the investigated impurities was checked rang-
ing from 0.0005 to 0.1 mg/ml. After having checked the

S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
307
Table 1
Retention times and characteristic validation data of the investigated compoundsa
Compound
Retention
time (min)
Relative
retention time
Linearity
Correlation
coefficient, r
Precision
(%)
Recovery
(%)
(1–19 mg/ml)b
1
5.8
1.0
13 129 5203220 542
(5–1000 mg/ml)
12 785.0324693.6
8887.6312109.2
10 258327383.3
8456.43122 068
8986.53114 240
15 5993213 771
12 182317080.1
3813.6327581.7
0.999
0.86
100.23
2
3
4
5
6
7
8
9
8.8
9.9
16.1
10.2
2.7
16.4
8.7
4.9
1.51
1.71
2.77
1.75
0.47
2.83
1.50
0.85
0.999
0.996
0.992
0.991
0.998
0.999
0.999
0.997
0.99
0.63
0.93
1.22
0.96
1.02
0.88
0.84
99.78
100.82
102.31
98.52
98.36
100.60
96.26
101.05
a Conditions: isocratic HPLC; stationary phase: Nucleosil 120-5, C18, 250 mm34.6 mm ID; mobile phase: 75 (v/v) methanol and 25 (v/v) 0.2%
phosphoric acid; flow rate 1.5 ml/min; UV detection at 320 nm.
b Unless otherwise noted.
linearity of indomethacin and its impurities separately, the
linearity of the by-products systematically added to a
standard solution of 1.0 mg/ml indomethacin was addi-
tionally tested. Spiked impurities were checked ranging
from 1.0 to 1.9 mg/ml (corresponding to 0.1–0.19%
referred to the nominal concentration of 1.0 mg/ml
indomethacin). The linearity of the spiked solution is given
in Table 1. For all derivatives a high correlation coefficient
was determined. Thus, a very accurate method has been
developed.
an indomethacin solution eight times. The concentration
was compared to the theoretical concentration. The relative
standard deviation (R.S.D.) was 0.86%. The precision of
the impurities was determined from spiked solutions at 10
mg/ml and the R.S.D. was between 0.63 and 1.22% for all
impurities (Table 1).
3.5. Accuracy and recovery
To determine the accuracy, impurities were added to an
indomethacin standard solution ranging from 0.19 to 0.5%
of the nominal concentration of indomethacin. For ethyl
ester 3, an average recovery of 100.80% was obtained for
that range. Recoveries found for the other derivatives were
between 96.26 and 102.31% and are listed in Table 1.
3.4. Precision
The precision of the method was evaluated by injecting
3.6. Limit of quantification (LOQ)
To test the limit of quantification, spiked solutions were
investigated. The limit of quantification of the impurities
2–9 was determined from the signal to noise ratio greater
than 10 and was found to be below 1 mg/ml for each
compound. Relationships between concentration and UV
responses are given in Table 1 for the range 1–19 mg/ml.
3.7. Ruggedness
In order to determine the ruggedness of the method, the
stability of the indomethacin solution was tested. There-
fore, test solutions of six batches of indomethacin were
prepared according to the control test and stored in the
autosampler at room temperature. They were chromato-
graphed at the beginning, and after 24 and 48 h. The
relative standard deviation was determined from 1.73 to
2.24%. Indomethacin solutions are stable within the in-
vestigated 48 h. The ruggedness of the mobile phase was
additionally checked. Variations in the concentration of the
Fig. 3. HPLC separation of all impurities from indomethacin. Conditions:
isocratic HPLC, stationary phase: Nucleosil 120-5, C18, 250 mm34.6
mm ID; mobile phase: 75 (v/v) methanol and 25 (v/v) 0.2% phosphoric
acid, flow rate 1.5 ml/min, UV detection at 320 nm.

308
S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
mobile phase led to variations in the retention times, so
care was required with the composition of the mobile
phase.
group (Fig. 4A). The remaining protons of 3 were chemi-
cally and magnetically identical to those from indo-
methacin and, therefore, could not be distinguished from
indomethacin under the conditions measured.
The spectra of indomethacin spiked with 0.1% of the
dichloro compound 8 appeared to be more complex due to
the chemical shifts induced by the second chloro atom in
the phenyl group. Therefore, the downfield range (6.6–8.0
ppm) was more influenced than the upfield range of each
spectrum. As shown in Fig. 5A, the signals for the indole
protons and the phenyl protons of the dichloro derivative 8
were shifted, both up- and downfield, when compared to
indomethacin. Thus, a typical pattern was observed for
indomethacin and impurity 8 (Fig. 5A).
The investigations were extended to all other impurities
described in this study (data not shown). NMR data
supported the analysis of indomethacin batches, and ascer-
tained the assignment of the by-products, especially when
ambiguities occurred in case of coelution in the HPLC.
Prior independent synthesis of all investigated impurities
and preparation of spiked solutions increased the level of
confidence for assignment of the peaks. The results
3.8. Nuclear magnetic resonance
1H NMR spectroscopy is a useful tool for investigation
of impurities of the samples, even without prior separation
from the drug substance (Hess et al., 2001a; Holzgrabe et
al., 1999; USP24, 1999b). This method was successfully
applied to the identification and determination of all
impurities in indomethacin batches in concentrations as
low as 0.1%. In a series of experiments, impurities have
been added to indomethacin CRS substance and NMR
spectra were recorded and compared with 38 authentic
batches. As presented in Figs. 4 and 5, it was possible to
unambiguously assign impurities in the NMR spectra of
indomethacin at a 0.1% level. Fig. 4 shows the NMR
spectrum of indomethacin spiked with 0.1% ethyl ester 3.
Ethyl ester 3 was basically identified on the triplet at 1.27
ppm and the quadruplet at 4.16 ppm representing the ethyl
Fig. 4. 1H NMR spectrum of indomethacin spiked with ethyl ester 3. The full spectrum of 1 spiked with ethyl ester 3 is shown at the bottom. The
identification of ethyl ester 3 was possible due to its additional ethyl group. As shown in expansion A, ethyl ester 3 was identified by a triplet at 1.27 ppm
(a) and a quadruplet at 4.16 ppm (b).

S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
309
Fig. 5. 1H NMR spectrum of indomethacin spiked with dichloro derivative 8. At the bottom, the full spectrum of 1 spiked with dichloro derivative 8 is
shown. The additional chloro atom in 8 induced shifts of the phenyl and indole hydrogens which facilitated the identification of 8 in drug substance.
Expansion A shows the assignment of the indole- (f–h) and phenyl-hydrogens (i–k) of 8.
indicated the high potential of NMR spectroscopy in the
search for impurities in drug substances.
observed. There seems to be no other explanation than the
attempt to reduce the amount of impurity 2 and/or 3 by
mixing the bulk product with substances derived from two
distinct synthetic routes. In view of good manufacturing
practices, this seems to be unacceptable. The new dichloro
compound 8 was identified in concentrations higher than
0.2%. According to compendial guidelines, no impurity
should be greater than 0.2%, and the sum of all impurities
should not be greater than 0.4%. Although all investigated
batches met current compendial specifications, 16 out of
the 38 investigated indomethacin batches did not meet
these specifications when investigated with the newly
developed HPLC–UV and NMR methods. This means that
42% of the batches available on the European market did
not meet the regulations. The development of these
3.9. Investigation of industrial batches
To investigate the quality of currently available in-
dustrial indomethacin batches, 38 batches were investi-
gated by compendial methods and by the new HPLC–UV
and NMR method. Although all batches met the compen-
dial tests, new impurities could be identified when investi-
gated by HPLC–UV and NMR. The main impurities found
in the investigated industrial batches were the methyl ester
2, the ethyl ester 3 and the dichloro compound 8 (Table 2).
Interestingly, in some batches (batch 9 and 10), impurities
2 and 3, derived from two distinct synthetic routes, were

310
S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
Table 2
was developed in order to qualify and quantify new
impurities found in indomethacin batches. NMR inves-
tigations directly from the bulk substance without prior
separation supported the interpretation of the results. A
total of 38 batches were investigated using current com-
pendial methods and the newly developed LC–UV and
NMR method. Although all batches met the current
compendial requirements, it was found that 16 out of the
38 batches showed impurities higher than 0.2%. The
presented isocratic HPLC–UV method was reliable, and
inexpensive, and it is applicable in laboratories all over the
world, including those in developing countries where most
generics are produced. The NMR method was found to be
extremely useful for the identification of the impurities
without prior separation. Clearly, it was found that some
bulk ware available on the market contained the new
impurity 8. Other samples were obviously derived from
two distinct synthetic routes. It appeared as if indo-
methacin batches containing both the methyl ester 2 and
ethyl ester 3, were mixed in order to reduce the impurity of
each by-product. From the point of view of good manufac-
turing practices, this practice is not acceptable. Since these
findings could not be detected with earlier compendial
methods, the development of both the new HPLC–UV and
the NMR method is an important completion of the
investigation of indomethacin drug substance. They are
especially suitable for quality control in pharmaceuticals.
Representative results of selected, investigated batches
Batch no.
HPLC results
(area under the curve in
%, identified impurity)a,b
1
2
4.2 min (0.03)
4.3 min (0.07)
5.8 min: indomethacin
2.4 min (0.02)
5.8 min: indomethacin
11.8 min (,0.01)
3
4
5.8 min: indomethacin
8.8 min (0.10)
2.3 min (0.03)
5.8 min: indomethacin
8.8 min (0.13, 2)
5
6
3.9 min (0.04)
5.8 min: indomethacin
5.9 min (,0.01)
2.3 min (0.01)
3.4 min (0.09)
5.8 min: indomethacin
10.2 min (0.1)
7
8
3.9 min (,0.01)
5.8 min: indomethacin
8.7 min (0.31, 8)
4.3 min (,0.01)
5.7 min (,0.01)
5.8 min: indomethacin
9.9 min (0.20, 3)
9
2.5 min (,0.01)
4.4 min (0.01)
Acknowledgements
5.8 min: indomethacin
8.8 min (0.4, 2)
9.9 min (0.2, 3)
¨
The authors thank the Bundesinstitut fur Arzneimittel
und Medizinprodukte (Federal Institute for Drugs and
Medical Devices), Berlin, Germany, for their generous
financial support. Thanks are also due to Dr Heidemarie
Horn and Fioralba Demirai, Institute of Pharmacy, Uni-
versity of Leipzig, for technical assistance.
10
3.9 min (,0.01)
5.8 min: indomethacin
8.8 (0.4, 2)
9.9 min (0.18, 3)
a Impurities smaller than 0.1% remain unidentified.
b Assigned impurities were verified by NMR.
References
methods was important, because the impurities could not
be detected with compendial methods published so far
(Eur. Pharm. Comm., 2000a; Brit. Pharm. Comm., 2000;
USP24, 1999a).
Our results show that there is an urgent need for
constant renewing of compendial methods. Especially in
drugs used for chronic diseases, unidentified impurities
bear a risk of adverse effects.
British Pharmacopoeia Commission, British Pharmacopoeia 2000. Lon-
don, Indomethacin, 1, 851–852.
Denk, O.M., Gray, A.I., Skellern, G.G., Watson, D.G., 2000a. Isolation
and identification of three potential impurities of pholcodine bulk drug
substance. J. Pharm. Pharmacol. 52, 819–829.
Denk, O.M., Watson, D.G., Skellern, G.G., 2000b. Micellar electrokinetic
and high-performance liquid chromatographic determination of po-
tential manufacturing impurities in pholcodine. J. Chromatogr. A 895,
9–16.
¨
¨
Eger, K., Troschutz, R., Roth, H.J., 1999. Arzneistoffanalyse: Reaktivitat,
¨
Stabilitat, Analytik [Drug Analysis: Reactivity, Stability, Analysis].
4. Conclusions
Deutscher Apotheker, Stuttgart.
European Pharmacopoeia Commission, European Pharmacopoeia 1997.
Council of Europe, Strasbourg, High Performance Liquid Chromatog-
raphy, 33.
This study has shown the necessity to re-examine
compendial methods applied to drugs that are no longer
under patent protection. The presented HPLC–UV method
European Pharmacopoeia Commission, European Pharmacopoeia 1997,

S. Hess et al. / European Journal of Pharmaceutical Sciences 14 (2001) 301–311
311
Supplement 2000. Council of Europe, Strasbourg, Trimethoprim,
1453–1456.
European Pharmacopoeia Commission, European Pharmacopoeia 1997,
Supplement 2000. Council of Europe, Strasbourg, Indomethacin,
1012–1013.
Hardman, J.G., Limbird, L.E., Molinoff, P.B., Ruddon, R.W., Gilman,
A.G., 1990. In: Goodman & Gillman’s The Pharmacological Basis of
Therapeutics, 8th Edition. McGraw-Hill, New York, pp. 659–661.
Merck and Co. Inc., 1968. 5-Methoxy- and 5-dimethylamino-1-(4-chloro-
benzoyl)-2-methyl-3-indolylacetic acid. Patent FR1534327.
Sato, J., Amizuko, T., Niida, Y., Umetsu, M., Ito, K., 1997. Simple, rapid
and sensitive method for the determination of indomethacin in plasma
by high-performance liquid chromatography with ultraviolet detection.
J. Chromatogr. B 692, 241–244.
Shen, T.Y., Ellis, R.L., Windholz, T.B., Matzuk, A.R., Rosegay, A.,
Lucas, S., Witzel, J.D., Stammer, C.H., Wilson, A.N., Holly, F.W.,
Willett, J.D., Sarrett, L.H., Holtz, W.J., Risley, E.A., Nuss, G.W.,
Winter, C.A., 1963. Non-steroid anti-inflammatory agents. J. Am.
Chem. Soc. 85, 488–489.
¨
Hartke, K., Hartke, H., Mutschler, E., Rucker, G., Wichtl, M., 1997. In:
Indomethacin. Arzneibuchkommentar, Vol. I 11. Deutscher Apotheker,
Stuttgart, pp. 1–6.
Hess, S., Doelker, M., Haferburg, D., Kertscher, H.P., Matysik, F.-M.,
Ortwein, J., Teubert, U., Zimmermann, W., Eger, K., 2001a. Sepa-
ration, analyses and syntheses of trimethoprim impurities. Pharmazie
56, 1–5.
Hess, S., Akermann, M., Ropte, D., Eger, K., 2001b. Rapid and sensitive
LC separation of new trimethoprim impurities. J. Pharm. Biomed.
Anal. 25, 531–538.
Toussaint, B., Pitti, C., Streel, B., Ceccato, A., Hubert, P., Crommen, J.,
2000. Quantitative analysis of N-acetylcysteine and its pharmacopoeial
impurities in a pharmaceutical formulation by liquid chromatography–
UV detection–mass spectrometry. J. Chromatrogr. A 896, 191–199.
Tsai, Y.H., Naito, S., 1981. Simultaneous determination of indomethacin
and its metabolites in rabbit plasma by high-pressure liquid chroma-
tography. Int. J. Pharm. 8, 203–209.
Holzgrabe, U., Waver, I., Diehl, B., 1999. In: NMR Spectroscopy in Drug
Development and Analysis. Wiley–VCH, Weinheim, pp. 5–60.
Lehr, G.J., Barry, T.L., Petzinger, G., Hanna, G.M., Zito, S.W., 1999.
Isolation and identification of process impurities in trimethoprim drug
substance by high-performance liquid chromatography, atmospheric
pressure chemical ionization liquid chromatography/mass spec-
trometry and nuclear magnetic resonance spectroscopy. J. Pharm.
Biomed. Anal. 19, 373–389.
USP24, United States Pharmacopeia, 1999a. 24th review, Rockville,
Indomethacin, 874–880.
USP24, United States Pharmacopeia, 1999b. 24th review, Rockville,
NMR spectroscopy, 1959–1965.
Yamamoto, H., 1968. 1-Acylindoles. II. A new syntheses of 1-( p-
chlorobenzoyl)-5-methoxy-3-indolylacetic acid and its polymorphism.
Chem. Pharm. Bull. 16, 17–19.