1

Detection of dehalogenation impurities in organohalogenated 1

pharmaceuticals by UHPLC–DAD–HRESIMS 2

3

Erik L. Regalado*, Renee K. Dermenjian, Leo A. Joyce, Christopher J. Welch* 4

Merck Research Laboratories, Rahway, New Jersey 07065, USA 5

*Corresponding authors. Tel.: +1 732 594 5928 (C.J. Welch); +1 732 594 5452 (E.L. Regalado); 6 fax: +1 732 594 9140. 7 8 E-mail addresses: christopher_welch@merck.com (C.J. Welch), erik.regalado@merck.com (E.L. 9 Regalado) 10 11

ABSTRACT 12

The presence of dehalogenated impurities is often observed in halogen-containing 13

pharmaceuticals, and can present a difficult analytical challenge, as the chromatographic 14

behavior of the halogenated drug and the hydrogen-containing analog can be quite 15

similar. In this study we describe the chromatographic separation and unambiguous 16

identification of dehalogenation impurities or associated isomers in organohalogenated 17

pharmaceuticals using UHPLC with a pentafluorophenyl column coupled with diode-18

array and high resolution electrospray ionization mass spectrometry detection (UHPLC–19

DAD–HRESIMS). 20

21

KEYWORDS: pharmaceuticals; dehalogenation impurities; method development; 22

perfluorophenyl; UHPLC; HRMS 23

24

Important: This is an uncorrected version. Please, access to the final version through: 25

http://dx.doi.org/10.1016/j.jpba.2013.12.043 26

27

2

1. Introduction 28

Incorporation of halogen into modern pharmaceuticals has become vitally important, 29

with more than half of all recently introduced small molecules drugs containing halogen 30

atoms [1]. The incorporation of halogens, especially fluorine, serves to block metabolism 31

and enhance the bioavailability of pharmaceuticals [2-5]. Chemical transformations 32

involving low-valent transition metals (e.g. palladium coupling or catalytic 33

hydrogenation over metal catalysts) can sometimes give rise to dehalogenation impurities 34

created through hydrodehalogenation [6]. In addition, halogen-containing starting 35

materials can often be contaminated with the corresponding proteo analogs, leading to the 36

formation of impurities that can persist through further synthetic steps. Owing to close 37

structural similarities, these dehalogenation impurities can sometimes be difficult to 38

separate from the parent compound [7-9]. 39

We recently investigated a number of chromatographic method development screens 40

to identify optimum columns and conditions for resolving fluorine or other halogen-41

containing pharmaceuticals from their dehalogenated analogs [9]. The best overall 42

conditions identified in the study involved the use of perfluorophenyl (PFP) stationary 43

phases in either UHPLC or core shell HPLC modes using acetonitrile/methanol based 44

aqueous eluents containing either phosphoric or perchloric acid. In this study, we apply a 45

variation on these conditions, combined with mass spectrometric detection, to the search 46

for dehalogenation impurities in halogen-containing pharmaceuticals. 47

2. Experimental 48

2.1. Instrumentation 49

Reversed phase achiral UHPLC-PDA-MS screening experiments were performed 50

with a Waters Acquity UPLC H-Class (Waters Corp., Milford, MA, USA) system 51

equipped with a quaternary solvent delivery pump, a sampler manager – FTN 52

autosampler, a 80 Hz photodiode array detector, a Waters Acquity Single Quadrupole 53

LC/MS detector with electrospray ionization in the positive and negative mode, and 54

Waters MassLynx® software for instrument control and data processing. 55

3

Reversed phase achiral UHPLC-DAD-HRMS experiments were performed on an 56

Agilent 1290 Infinity liquid chromatography system (Agilent Technologies, Palo Alto, 57

CA, USA) equipped with a G4220A binary pump, G4212A diode array detector, 58

G4226A autosampler, and G1316C thermostated column compartment. The LC system 59

was coupled to an Agilent 6520 Q-TOF mass spectrometer equipped with electrospray 60

ionization (ESI) source in the positive mode. The system was controlled by 61

MassHunter® software. 62

2.2. Chemicals and reagents 63

Methanol and acetonitrile (HPLC Grade) were purchased from Fisher Scientific (Fair 64

Lawn, NJ, USA), formic acid (HCOOH), ammonium formate (NH4HCO2), linezolid, 65

lamotrigine, sunitinib, citalopram HBr, fluoxetine, flurbiprofen, amlodipine besylate, 66

chlorofibrate, fenofibrate, desloratadine and chlorofibric acid were all purchased from 67

Sigma–Aldrich (St. Louis, MO, USA). Aprepitant, desfluoro aprepitant, paroxetine 68

maleate, risperidone, chlorowarfarin, warfarin, ciprofloxacin HCl, ofloxacin, norfloxacin 69

and enoxacin were all obtained from the Merck Building Block Collection. Atorvastatin 70

sodium, desfluoro atorvastatin sodium, voriconazole, desfluoro voriconazole, ezetimibe, 71

desfluoro ezetimibe, desfluoro paroxetine, desfluoro risperidone, desfluoro ciprofloxacin 72

HCl and desfluoro ofloxacin were all purchased from Molcan Co. (Toronto, Ontario, 73

Canada). Ultrapure water was obtained from a Milli-Q Gradient A10 from Millipore 74

(Bedford, MA, USA). 75

2.3. Preparation of buffer solutions 76

2 mM NH4HCO2 in H2O (pH = 3.5) and 2 mM NH4HCO2 in 90% CH3CN and 10% 77

H2O (pH = 3.5) solutions: 12.6 g NH4HCO2 and 7.9 mL HCOOH were dissolved in 1 L 78

Millipore water. A 100-fold dilution of this stock solution was performed in either pure 79

water or an acetonitrile to afford the 2 mM solutions. 80

2.4. UHPLC-PDA-MS and UHPLC-DAD-HRMS conditions 81

UHPLC separations were carried out on a 2.1 mm × 50 mm, 1.9 µm Hypersil Gold 82

PFP column (Thermo Scientific, Rockford, IL, USA) by gradient elution at a flow rate of 83

4

0.6 mL/ min. The LC eluents were solvent A (water, 2 mM ammonium formate, pH 3.5) 84

and solvent B (acetonitrile, 2 mM ammonium formate, pH 3.5). The mobile phase was 85

programmed as follows: linear gradient from 5% to 95% B in 5.5 min, 95% B hold from 86

5.5 to 6.2 min and 2 min re-equilibration time. The column and samples were maintained 87

at a temperature of 40 °C and 20 °C, respectively. For UHPLC-PDA-MS screening: The 88

positive ion ESI parameters were cone voltage 45 V, desolvation gas (N2) flow rate 800 89

L/h, cone gas (N2) flow rate 20 L/h, and source temperature 150 °C. Full-scan mass 90

spectra were acquired in the mass-to-chage (m/z) range 100–1000 using the SQD 91

analyzer operating with a scan time of 0.10 s. 92

For UHPLC-DAD-HRMS experiments: The positive ion ESI parameters were 93

fragmentor 55 V, skimmer 65 V, desolvation gas (N2) temperature 350°C and flow rate 94

13 L/min, nebulizer 60 psig. Full-scan mass spectra were acquired over the range m/z 95

100–1000 at an acquisition rate of 2 spectra/s. Spectra were recorded in centroid mode. 96

G1969-85001 ES-TOF Reference Mass Solution Kit (Agilent Technologies) was used for 97

continuous auto-calibration during UHPLC-HRMS experiments to ensure precise and 98

automated mass accuracy measurements. Final solution composition: 25 μM ammonium 99

trifluoroacetate, 1 μM purine, 2.5 μM hexakis (1H, 1H, 3H-tetrafluoropropoxy) 100

phosphazine (HP-0921) in 95:5 acetonitrile:water. G1969-85000 ESI-L Low 101

Concentration Tuning Mix (Agilent Technologies) was used for tuning and calibration of 102

the Agilent 6520 Q-TOF mass spectrometer. 25 mL of the tuning mix was mixed with 103

71.25 mL acetonitrile and 3.75 ml water to give the working solution. 104

3. Results and discussion 105

The previous study to identify optimal columns and conditions to resolve 106

dehalogenated impurities from the corresponding halogen-containing pharmaceuticals 107

utilized known standards and UV detection, however real-world problem solving 108

typically involves situations in which no authentic standards for the dehalogenated 109

impurities exist. Mass spectrometry detection provides a useful tool for identifying 110

potential dehalogenation impurities based on mass. Also, incomplete chromatographic 111

resolution can sometimes be addressed by MS detection, which can allow deconvolution 112

of overlapping peaks with different masses [10, 11]. However, the phosphoric acid or 113

5

perchloric acid-based mobile phase additives used in the original study are not easily 114

compatible with MS detection. An examination of the separation of nine halogen-115

containing pharmaceuticals from their dehalogenated impurities on the PFP column using 116

a standard MS compatible mobile phase (2 mM ammonium formate / acetonitrile-water, 117

pH 3.5) showed good resolution of impurities in all cases. While the separations were not 118

quite as good as was observed with the perchlorate/phosphoric acid eluents, resolution 119

was sufficient for routine problem solving. If required, the peak shape and resolution 120

obtained using these standard gradient elution conditions (from 5 to 95% organic phase 121

over 5.5 min) could potentially be improved by adjusting elution conditions. In all 122

examples illustrated in figure 1, the halogen containing pharmaceutical is eluted later 123

than the dehalogenated analog. 124

Figure 1 125

Using these UHPLC conditions with single quadrupole MS detection we analyzed 22 126

commercial halogen-containing pharmaceuticals for the presence of dehalogenation 127

impurities. While most of these samples (footnote) showed no indication of 128

dehalogenated impurities, the three examples shown in figure 2 all showed early eluting 129

compounds with the appropriate nominal mass corresponding to the dehalogenated 130

species. The fluorine-containing antibiotic drug, linezolid (Figure 2a) shows a single 131

major component by UV detection, with the extracted ion chromatogram (EIC) of m/z = 132

338 amu (corresponding to the molecular ion [M+H]+) being easily detected. The EIC of 133

m/z 320, corresponding to a loss of 18 mass units, shows an early eluting peak which may 134

represent replacement of the fluorine substituent with a hydrogen atom ([M+2H-F]+). 135

Similarly, the antidepressant drug, paroxetine (Figure 2b) shows a strong peak in the EIC 136

of the molecular ion (m/z 330), with an early eluting impurity being observed in the EIC 137

of m/z 312, potentially corresponding to the desfluoro impurity. Importantly, while this 138

loss of 18 mass units could be attributed to the desfluoro species, it could also come from 139

Compounds analyzed: linezolid, lamotrigine, sunitinib, citalopram, fluoxetine, flurbiprofen, amlodipine

besylate, chlorofibrate, fenofibrate, desloratadine, chlorofibric acid, aprepitant, paroxetine, risperidone,

chlorowarfarin, ciprofloxacin, ofloxacin, norfloxacin, enoxacin, atorvastatin, voriconazole and ezetimibe.

6

a loss of water, or from some altogether different species. Consequently, determination of 140

the exact mass of the species in question is required to confirm the assignment. 141

Figure 2 142

The anticonvulsant drug, lamotrigine (Figure 2c) contains two chlorine atoms. 143

Interestingly, two early eluting peaks are observed in the EIC of m/z 222. These 144

presumably correspond to the two possible monochloro isomers, which would result from 145

loss of one chlorine atom each. In addition, a very early eluting impurity is observed in 146

the EIC of m/z 188, an indication of a complete deshalogenation ([M+3H-2Cl]+). Again, 147

high resolution mass analysis is required for unambiguous assignment. 148

It is important to point out that in all three cases the levels of these putative 149

dehalogenation impurities is quite low; at or below the 0.15% threshold of concern for 150

typical pharmaceutical impurities. Nevertheless, such trace impurities can sometimes be 151

important in understanding and controlling chemical processes, or as signature 152

compounds for anti-counterfeiting efforts [12, 13]. 153

Table 1 154

In order to unambiguously assign the putative dehalogenation impurities, exact mass 155

measurement using a more accurate mass spectrometer was required. We repeated the 156

analysis using a UHPLC system fitted with a diode array detector and a high resolution 157

Q-TOF mass spectrometer (Table 1 and Figure 3). High resolution mass measurements of 158

the early eluting linezolid impurity indeed confirmed it to be the expected desfluoro 159

impurity (Figure 3a, HRMS (ESI-TOF) m/z: [M+2H-F]+ calculated for C16H22N3O4 160

320.1610; found 320.1599, Δ 1.7 ppm). Similarly, the early eluting impurity from the 161

analysis of paroxetine sample is confirmed to be the desfluorinated product (figure 3b, 162

[M+2H-F]+ calcd for C19H22NO3 312.1600; found 312.1595, Δ 0.2 ppm). Finally, the 163

HRMS detection also led us to unambiguously confirm the assignments of the three 164

dehalogenated impurities detected in the lamotrigine sample (Figure 3c). Two mono-165

deschloro isomers ([M+2H-Cl]+ calcd for C9H9ClN5 222.0547; found 222.0537, Δ 1.2 166

ppm and 222.0538, Δ 0.7 ppm, respectively) and a bis-deschloro product ([M+3H-2Cl]+ 167

calcd for C9H10N5 188.0936; found 188.0932, Δ 0.7 ppm). 168

7

Figure 3 169

The approach illustrated in this study provides an example of how modern LC-MS 170

tools can be used to search for and identify suspected dehalogenation impurities in 171

halogen-containing pharmaceuticals. Detection using a lower resolution single 172

quadrupole MS detector correctly identified a number of potential dehalogenation 173

impurities that were later confirmed by exact mass measurement using TOF. However, 174

care must be taken in ascribing any loss of 18 mass units observed via low resolution MS 175

to an impurity involving the replacement of fluorine with hydrogen. For example, several 176

pharmaceuticals analyzed afforded a significant peak in the EIC of the mass 177

corresponding to the [M+H-18]+ ion. While the co-elution of this peak with the parent 178

species could easily be interpreted as a stealthy dehalogenation impurity, exact mass 179

measurement clearly shows these peaks to be dehydration products, presumably arising 180

from loss of water from the parent drug during the ionization process. 181

182

CONCLUSIONS 183

The presence of dehalogenated impurities is often observed in halogen-containing 184

pharmaceuticals, and can present a difficult analytical challenge, as the chromatographic 185

behavior of the halogenated drug and the hydrogen-containing analog can be quite 186

similar. In this study we describe the chromatographic separation and unambiguous 187

identification of dehalogenation impurities or associated isomers in organohalogenated 188

pharmaceuticals using UHPLC with a pentafluorophenyl column coupled with diode-189

array and high resolution electrospray ionization mass spectrometry detection (UHPLC–190

DAD–HRESIMS). 191

ACKNOWLEGEMENTS 192

We are grateful to MRL Postdoctoral Research Fellows Program for financial 193

support provided by a fellowship (E.L.R) and also to the MRL New Technologies 194

Review & Licensing Committee (NT-RLC) for providing funding for the instrument used 195

in this evaluation. 196

8

References 197

[1] L.M. Jarvis, New Drug Approvals Hit 16-Year High In 2012, Chem. Eng. News, 91 198

(2013) 15-17. 199

[2] J. Swinson, Fluorine - a vital element in the medicine chest, PharmaChem, 4 (2005) 200 26-30. 201

[3] R. Wilcken, M.O. Zimmermann, A. Lange, A.C. Joerger, F.M. Boeckler, Principles 202 and applications of halogen bonding in medicinal chemistry and chemical biology, J. 203

Med. Chem., 56 (2013) 1363-1388. 204

[4] M.Z. Hernandes, S.M.T. Cavalcanti, D.R.M. Moreira, d.A.W. Filgueira, Jr., A.C.L. 205 Leite, Halogen atoms in the modern medicinal chemistry: hints for the drug design, 206 Curr. Drug Targets, 11 (2010) 303-314. 207

[5] I. Ojima, Fluorine in medicinal chemistry and chemical biology, John Wiley & Sons 208 Ltd., 2009. 209

[6] K. Köhler, K. Wussow, A.S. Wirth, Palladium-Catalyzed Cross-Coupling Reactions – 210 A General Introduction, in: Palladium-Catalyzed Coupling Reactions, Wiley-VCH 211 Verlag GmbH & Co. KGaA, 2013, pp. 1-30. 212

[7] L. Turco, S. Provera, O. Curcuruto, E. Bernabè, A. Nicoletti, L. Martini, D. Castoldi, 213 Z. Cimarosti, D. Papini, C. Marchioro, R. Dams, Detection, identification and 214

quantification of a new de-fluorinated impurity in casopitant mesylate drug substance 215 during late phase development: An analytical challenge involving a multidisciplinary 216 approach, J. Pharm. Biomed. Anal., 54 (2011) 67-73. 217

[8] S. Ertuerk, A.E. Sevinc, L. Ersoy, S. Ficicioglu, An HPLC method for the 218

determination of atorvastatin and its impurities in bulk drug and tablets, J. Pharm. 219 Biomed. Anal., 33 (2003) 1017-1023. 220

[9] E.L. Regalado, P. Zhuang, Y. Chen, A.A. Makarov, N. McGachy, C.J. Welch, 221

Chromatographic resolution of closely related species in pharmaceutical chemistry: 222 dehalogenation impurities and mixtures of halogen isomers, Anal. Chem., (2013) 223

http://dx.doi.org/10.1021/ac403376h. 224

[10] C.J. Welch, B. Grau, J. Moore, D.J. Mathre, Use of chiral HPLC-MS for rapid 225 evaluation of the yeast-mediated enantioselective bioreduction of a diaryl ketone, J. 226 Org. Chem., 66 (2001) 6836-6837. 227

[11] E.L. Regalado, W. Schafer, R. McClain, C.J. Welch, Chromatographic resolution of 228

closely related species: separation of warfarin and hydroxylated isomers, J. 229

Chromatogr. A, 1314 (2013) 266-275. 230

[12] T. Almuzaini, I. Choonara, H. Sammons, Substandard and counterfeit medicines: a 231 systematic review of the literature, BMJ Open, 3 (2013) e002923. 232

[13] K. Degardin, Y. Roggo, P. Margot, Understanding and fighting the medicine 233 counterfeit market, J. Pharm. Biomed. Anal., doi:pii: S0731-7085(13)00018-6. 234 10.1016/j.jpba.2013.01.009 (2013). 235

236

Fig. 1. Reversed phase achiral UHPLC-PDA screening method for separation of halogen-containing

pharmaceuticals and their dehalogenation impurities using a standard gradient with a mass spectrometry compatible

eluent. Column: Hypersil Gold PFP (2.1 x 50 mm, 1.9 µm). Temperature: 40˚C. Detection: UV 210 nm. Sample: 0.5

µL injection of a ~0.5 mg/mL mixture of drug/dehalogenated standards. Flow rate: 0.6 mL/min. Eluents: 2mM

NH4CHO2 in H2O (pH 3.5) : 2mM NH4CHO2 in CH3CN (pH 3.5). Standard gradient: from 95:5 to 5:95 in 5.5 min,

hold at 5:95 for 0.7 min.

Fig. 2. Reversed phase achiral UHPLC-PDA-MS analysis of organohalogenated pharmaceuticals with potential

dehalogenation impurities. Standard screening gradient as described in figure 1. Temperature: 40˚C. Sample: 1 µL

injection of linezolid (MW: 337 dalton), b) paroxetine (MW: 329 dalton) and c) lamotrigine (MW: 255 dalton) at

~0.5 mg/mL. Detection: UV 210 nm and Single Quadrupole MS (fragmentor = 45 eV) showing the Extracted Ion

Chromatograms (EIC) at m/z = [M+H]+ and [(M+2H-halogen]

+.

Fig. 3. Reversed phase achiral UHPLC-DAD-HRMS analysis of organohalogenated pharmaceuticals (linezolid,

paroxetine and lamotrigine) with dehalogenation impurities. Standard screening gradient as described in figure 1.

Sample: 0.2 µL injection at ~0.5 mg/mL. Detection: UV 210 nm and HRMS (fragmentor = 55 eV) showing the

Extracted Ion Chromatograms (EIC) at m/z = [M+H]+ and [M+H-halogen]

+.

Fig. 1.

3.5 4 4.5 5 5.5

3 3.5 4 4.5 5

2 2.5 3 3.5 4

2.7 3.2 3.7 4.2 4.7

3 3.5 4 4.5 5

R = HR = F

aprepitants

atorvastatins

voriconazoles

ezetimibes

paroxetines

2.5 3 3.5 4 4.5

2.7 3.2 3.7 4.2 4.7

1.5 2 2.5 3 3.5

1.5 2 2.5 3 3.5

risperidones

warfarins

ciprofloxacins

ofloxacins

R = H R = Cl

R = H R = F

R = H R = F R = H

R = F

R = HR = F

R = H R = F

R = HR = F

R = HR = F

Fig. 2.

2.8 3.8 4.8

2.8 3.8 4.8

1.2 2.2 3.2

1.2 2.2 3.2

UV 210 nm

EIC 338 m/z

EIC 320 m/z

a) linezolid

1.2 2.2 3.2

1.9 2.9 3.9

UV 210 nm

EIC 256 m/z

EIC 222 m/z

EIC 188 m/z

c) lamotrigine

UV 210 nm

EIC 330 m/z

EIC 312 m/z

b) paroxetine

[M+H]+

desfluoro?

[M+H]+

desfluoro?

[M+H]+

Monochloro

(2 isomers)?

bis-deschloro?

2.8 3.8 4.8

1.9 2.9 3.9

1.9 2.9 3.9

1.9 2.9 3.9

min

Fig. 3.

1.2 2.2 3.2

3.2 4.2 5.2

UV 210 nm

1.2 2.2 3.2

1.2 2.2 3.2

UV 210 nm

EIC 338 m/z

EIC 320 m/z

3.2 4.2 5.2

3.2 4.2 5.2

EIC 330 m/z

EIC 312 m/z

1.1 2.1 3.1

1.1 2.1 3.1

1.1 2.1 3.1

1.1 2.1 3.1

UV 210 nm

EIC 256 m/z

EIC 222 m/z

EIC 188 m/z

7x104

0

+ESI Scan188.0932

165 185 205 220

2 x105

0

+ESI Scan1: 222.05372: 222.0538

150 200 250 300

1x106

0

+ESI Scan 256.0149

m/z180 280 380

min

min

min

a) lamotrigine

b) paroxetine

C9H7Cl2N5

[C9H8Cl2N5]+

[C9H9ClN5]+

[C9H10N5]+

C19H20FNO3

[C19H21FNO3]+

[C19H22NO3]+

[C16H21FN3O4]+

[C16H22N3O4]+

6x104+ESI Scan

312.1595

309 314 319

9x105

0

+ESI Scan330.1492

m/z300 335 370

0

7x105

0

+ESI Scan320.1599

260 320 380

1

2

1x106

0

+ESI Scan 338.1514

m/z250 350 450

a) linezolid

C16H20FN3O4

desfluoro?

desfluoro?

bis-deschloro?

Monochloro

(2 isomers)?