Stability behaviour of antiretroviral drugs and their combinations. 1: characterization of tenofovir disoproxil fumarate degradation products by mass spectrometry

Abstract

Tenofovir disoproxil fumarate (TDF), an antiretroviral drug, was evaluated for its degradation behaviour in solid and solution states. A total of twelve non-volatile degradation products were formed, which were separated on a C18 column in a gradient mode, using methanol and ammonium formate in the mobile phase. The same method was extended to liquid chromatography-high resolution mass spectrometry (LC-HRMS). First a comprehensive mass fragmentation pattern of the drug was established by direct injection and collection of HRMS and multi-stage tandem mass spectrometric (MSⁿ) data. Then LC-HRMS studies were carried on the stability samples containing the degradation products. Also headspace gas chromatography-mass spectrometry (HS-GC-MS) studies were conducted to explore the formation of volatile components. The collated information was utilized for the characterization of all non-volatile and volatile degradation products. Eventually, the degradation pathway of the drug was established under the investigated conditions.

---

Introduction

Stability, apart from safety and efficacy, is a prime requirement for any candidate, under development or developed as a medicine. For antiretroviral drugs, stability issues are more common, due to which specific formulation strategies are employed, e.g., film coating and packaging in HDPE bottle with a desiccant. Also strict conditions are recommended for storing their products, e.g., 25 °C or below.

We have recently undertaken a project in our laboratory to carry out molecular level studies to delineate intrinsic stability behaviour of antiretroviral drugs alone and when present in combination formulations. The aim is to make available this vital pre-formulation information on antiretroviral drugs to formulation scientists as part of the QbD knowledge space. The drugs selected for the studies were tenofovir disoproxil fumarate (TDF), lamivudine (3TC), emtricitabine (FTC), zidovudine (AZT), stavudine (d4T), abacavir (ABC), efavirenz (EFV) and nevirapine (NVP), along with their marketed combinations.

This first part of the series reports intrinsic degradation behaviour of TDF (9-[(R)-2-[[bis[[(isopropoxycarbonyl)oxy] methoxy]phosphinyl]-methoxy]propyl]adenine fumarate). TDF is a reverse transcriptase inhibitor, which is used for the treatment of HIV and chronic hepatitis B.^1,2 It is known to degrade via two interrelated mechanisms.^3,4 The first is dismembering of isoproxil moieties, leading to formation of formaldehyde (HCHO). The second is formation of the methylene linked dimer of TDF. A few of impurities are also named in compendial monographs, e.g., United States Pharmacopeia (USP) pending monograph mentions control of tenofovir, tenofovir isoproxil monoester and tenofovir disoproxil dimer.⁵ The latter two, tenofovir monosoproxil dimer as well as tenofovir di- and monosoproxil heterodimer are listed in monograph on the drug in the World Health Organization's International Pharmacopoeia (Ph. Int.).⁶ Recently, a study reported two alkaline degradation products of TDF, viz., methyl hydrogen ({[1-(6-amino-9H-purin-9-yl)-propan-2-yl]oxy}methyl) phosphonate and dimethyl({[1-(6-amino-9H-purin-9-yl)propan-2-yl]oxy} methyl)phosphonate.⁷ Some other studies are available on degradation behaviour of the drug, but these are limited to solution state, and structures of the degradation products were also not elucidated.^8–13

To best of our knowledge, no comprehensive report exists on solid as well as solution state degradation behaviour of TDF, which includes characterization of its all degradation products. In this paper, we provide information on mass fragmentation profile, degradation behaviour, structures of non-volatile and volatile degradation products, and degradation pathway of the drug. The study was accomplished using sophisticated instrumental techniques, viz., liquid chromatography-high resolution mass spectrometry (LC-HRMS), multi stage tandem mass spectrometry (MSⁿ), and headspace gas chromatography-mass spectrometry (HS-GC-MS).¹⁴

Experimental

Chemicals and reagents

Pure TDF was obtained as a gratis sample from M/S Aurobindo Pharma Ltd. (Hyderabad, India). Hydrochloric acid (HCl) was procured from LOBA Chemie Pvt. Ltd. (Mumbai, India) and sodium hydroxide (NaOH) was purchased from Ranbaxy Laboratories (S.A.S. Nagar, India). High performance liquid chromatography (HPLC) grade methanol (CH₃OH) and acetonitrile (CH₃CN) were procured from Aldrich (St. Louis, MO, USA) and J. T. Baker (Phillipsburg, NJ, USA), respectively. Buffer salts and all other chemicals were of analytical reagent grade. Ultra pure water (H₂O) was obtained from ELGA water purification unit (Bucks, England).

Apparatus and equipments

Accelerated stability studies were carried out in humidity (KBF720, WTC Binder, Tuttlingen, Germany) and photostability (KBWF 240, WTC Binder, Tuttlingen, Germany) chambers set at 40 °C/75% RH and ambient temperature, respectively. The photostability chamber was equipped with an illumination bank on inside top consisting of a combination of three UV (OSRAM L18 W/73) and three white fluorescent (Philips, Trulite) lamps, as recommended by ICH guideline Q1B.¹⁵ Lux meter (model ELM 201, Escorp, New Delhi, India) and near UV radiometer (model 206, PRC Krochmann GmbH, Berlin, Germany) were used to measure visible illumination and near UV energy, respectively. A Dri-Bath (Thermolyne, Dubuque, IA, USA) was used for solid state thermal stress studies. Precision water bath equipped with MV controller (Julabo, Seelbach, Germany) was used for degradation studies in the solution state.

HPLC studies were carried out on LC-2010 HT liquid chromatograph (Shimadzu, Kyoto, Japan), which was equipped with a SPD-M20A prominence diode array detector. LC column used was Pursuit XRs 5 C18 250 × 4.6 mm (Varian Inc., Lake Forest, CA, USA).

LC-HRMS studies were carried out using LC-ESI-Q-TOF-MS, in which LC part consisted of 1100 series HPLC (Agilent Technologies, Waldbronn, Germany) comprising of an on-line degasser (G1379A), binary pump (G1312A), auto injector (G1313A), column oven (G1316A) and PDA detector (G1315B). The MS system consisted of MicrOTOF-Q spectrometer (Bruker Daltonics, Bremen, Germany). System control and data acquisition were done by Hystar software (version 3.1) from the same source. The calibration solution used was 5 mM sodium formate. LTQ-XL-MS 2.5.0 (Thermo, San Jose, CA, USA) mass spectrometer was used for tandem mass experiments. The mass spectra were acquired and processed using Xcalibur software (version 2.0.7 SP1).

HS-GC-MS system consisted of HS sampler (TurboMatrix 16), GC (Clarus 600) and MS (Clarus 600C). The three were controlled using TurboMass (version 5.4.2) software (PerkinElmer Inc., Shelton, CT, USA). The GC column used for the analysis of samples was Elite-5MS, 30 m, 0.25 mm ID, 0.25 μm df.

pH/Ion analyzer (PB-11 235, sartorius, Goettingen, Germany) was used to check the pH of all solutions. Other equipments used were sonicator (3210, Branson Ultrasonics Corporation, Danbury, CT, USA), precision analytical balance (AG 135, Mettler Toledo, Schwerzenbach, Switzerland) and auto pipettes (Eppendorf, Hamburg, Germany).

Solid state stress studies

For solid state stress studies, four sets were prepared. First and second sets were further divided into three subsets, i.e., drug alone, drug + acid stressor, and drug + base stressor. Acid and base stressors were boric acid (H₃BO₃) and sodium carbonate (Na₂CO₃), respectively. The drug : stressor ratio was fixed as 1 : 2. The powders were mixed thoroughly in the vial with the help of a spatula. The first set (all three subsets) contained drug and stressors along with 50 μl H₂O in closed vials, and was stored in accelerated stability test chamber for 1 month. The second set (three subsets) was exposed to accelerated stability condition of 40 °C/75% RH for 3 months. The third set containing drug alone was kept at 60 °C and subjected to thermal exposure for 1 month. Similarly, in the fourth set, the drug was charged to the photostability chamber and exposed for sufficient duration so as to provide a total dose of 6 million lux h cool white light and 1193 W h m⁻² near-UV. After completion of the study period of all the sets, the samples were withdrawn, and were monitored for chemical changes through HPLC. Samples for HPLC analyses were prepared in water, and had concentration of 2 mg ml⁻¹. HPLC method was optimized to provide clear picture on generation of the new degradation products. The developed method involved mobile phase composed of methanol (A) and 10 mM ammonium formate (pH 3.75) (B) in a gradient mode (Tmin/B; T0/95; T5/95; T50/10; T55/95; T62/95). The column oven temperature was maintained at 30 °C during analysis, while detection wavelength, flow rate and injection volume were 254 nm, 1 ml min⁻¹ and 5 μl respectively.

Solution state stress studies

Stress studies were carried out under hydrolytic conditions according to ICH Q1A(R2) recommendations,¹⁶ following the protocols described in our previous publications.^17,18 Hydrolytic studies were carried out in water (neutral), 0.5 N HCl and 0.5 N NaOH solutions. Drug stock solutions were prepared in water and each contained 4 mg ml⁻¹ of the drug. The same was diluted with an equal volume of the stressor. The temperature of hydrolytic reactions was 80 °C. Samples were withdrawn after 3 h and neutralized with acid/alkali and/or diluted two times with water before injecting into HPLC. The HPLC method used was similar to the method employed for solid state stress studies, except injection volume, which was 10 μl in this case.

For the analysis of volatile degradation products, 2 ml of 10 mg ml⁻¹ drug solutions in water and CH₃OH : H₂O (50 : 50 v/v) were stressed in 22 ml crimp top sealed glass vials. The temperature was set at 80 °C, and monitored for 15 h. HS-GC parameters were optimized with helium as a carrier gas. The optimized column oven temperature program was: Tmin/°C; T0/40; T30/240; T40/240. All samples were injected through HS sampler with a split ratio of 5. The injector, HS oven, HS transfer line and HS needle temperature were set at 140, 80, 120, and 90 °C, respectively. Thermostat, withdraw and inject time were fixed at 15, 0.5 and 0.04 min, respectively. The mass spectrometer was operated in electron impact ionization mode (EI, 70 eV) in the mass range of m/z 25–600. GC-MS interface and ion-trap temperatures were 270 and 170 °C, respectively.

HRMS, MSⁿ and LC-HRMS studies

For establishment of fragmentation pattern of the drug, a 10 ppm solution was directly injected into the Q-TOF system. MSⁿ studies were similarly conducted by directly injecting 10 ppm solution of the drug into the LTQ-XL system at a flow rate of 10 μl min⁻¹. Critical HRMS parameters were optimized in direct analysis mode. The same were used for subsequent LC-HRMS experiments for characterization of the degradation products. Here, LC method used was similar to the method employed for solid state stress studies. To prevent condensation in the ionization source, the solvent flow into MS system was reduced from 1 ml min⁻¹ to 200 μl min⁻¹, using a diversion valve.

Results and discussion

Mass fragmentation behaviour of the drug

The HRMS line spectrum of TDF is shown in Fig. 1. The data obtained from MSⁿ studies are listed in Table 1. The best possible molecular formulae and exact masses for the fragments were calculated using an elemental composition calculator. Additionally, accurate masses for the losses, errors in mmu, and ring plus double bond (RDB) values were determined. The same are listed in Table 2. The data in Tables 1 and 2 were taken into account to establish fragmentation pathway of TDF, which is outlined in Fig. 2.

Fig. 1 HRMS line spectrum of tenofovir disoproxil fumarate (TDF) in ESI + ve mode.

Table 1 MSⁿ data of tenofovir disoproxil fumarate (TDF)

MS^n	Precursor ion	Product ion(s)
a Ions were not captured for further MS^n studies.
MS^2	520	490, 476, 448, 446, 416, 404, 386, 374, 344, 342, 330, 312, 300, 288, 282, 270, 240, 206, 176^a
MS^3	490	446, 404, 386, 374, 344, 330, 300, 288
	476	446, 330, 288
	448	404, 288
	416	386, 342, 300, 270
	312	270
	282	252
	240	176^a
	206	176^a
MS^4	446	404, 374, 342, 330, 312, 300, 288
	386	344, 300
	270	288, 252, 240, 206, 176^a
MS^5	404	374, 330, 288, 270
	344	300
	342	300
	252	136^a
MS^6	374	330
	330	312, 288
	300	288, 282, 270
MS^7	288	270, 240, 206, 176^a, 136^a

---

Table 2 HRMS data of tenofovir disoproxil fumarate (TDF)

Peak	Accurate mass	Best possible molecular formula	Exact mass	Error (mmu)	RDB	Precursor ion	Loss from precursor ion	Molecular formula for loss
[TDF + H]^+	520.1811	C19H31N5O10P^+	520.1803	0.8	7.5
a	490.1692	C18H29N5O9P^+	490.1697	−0.5	7.5	[TDF + H]^+	30.0119	CH2O
b	446.1811	C17H29N5O7P^+	446.1799	1.2	6.5	a	43.9881	CO2
c	416.1350	C15H23N5O7P^+	416.1330	2.0	7.5	[TDF + H]^+	104.0461	C4H8O3
d	404.1347	C14H23N5O7P^+	404.1330	1.7	6.5	b	42.0464	C3H6
e	386.1244	C14H21N5O6P^+	386.1224	2.0	7.5	a	104.0448	C4H803
						c	30.0106	CH2O
f	374.1241	C13H21N5O6P^+	374.1224	1.7	6.5	d	30.0106	CH2O
g	344.0774	C11H15N5O6P^+	344.0754	2.0	7.5	e	42.0470	C3H6
h	342.1340	C13H21N5O4P^+	342.1326	1.4	6.5	b	104.0471	C4H8O3
						c	74.0010	C2H2O3
i	330.1342	C12H21N5O4P^+	330.1326	1.6	5.5	f	43.9899	CO2
j	312.1235	C12H19N5O3P^+	312.1220	1.5	6.5	i	18.0107	H2O
k	300.0877	C10H15N5O4P^+	300.0856	2.1	6.5	g	43.9897	CO2
						h	42.0463	C3H6
l	288.0874	C9H15N5O4P^+	288.0856	1.8	5.5	i	42.0468	C3H6
m	282.0817	C10H13N5O3P^+	282.0751	6.6	7.5	k	18.0060	H2O
n	270.0774	C9H13N5O3P^+	270.0751	2.3	6.5	j	42.0461	C3H6
						k	30.0103	CH2O
						l	18.0100	H2O
o	252.0650	C9H11N5O2P^+	252.0645	0.5	7.5	m	30.0167	CH2O
						n	18.0124	H2O
p	240.0671	C8H11N5O2P^+	240.0645	2.6	6.5	n	30.0103	CH2O
q	206.1065	C9H12N5O^+	206.1036	2.9	6.5	n	63.9709	HO2P
r	176.0966	C8H10N5^+	176.0931	3.5	6.5	p	63.9705	HO2P
						q	30.0099	CH2O

---

Fig. 2 Mass fragmentation pathway of tenofovir disoproxil fumarate (TDF) under ESI + ve mode. Structures shown in dotted box were observed only during MSⁿ study. Structure of fragment of m/z 288, which was also identified as degradation product T1, is shown in the solid box.

The parent of m/z 520, on collision-induced dissociation (CID), fragmented via four routes involving loss of CH₂O, CO₂, C₄H₈O and C₄H₈O₃, resulting into the product ions of m/z 490, 476, 448 and 416, respectively. The precursors of m/z 490 and 476, on MS³ CID, fragmented into common ion of m/z 446 on loss of CO₂ and CH₂O, respectively. The product ions of m/z 448 and 446 further dissociated into a common fragment of m/z 404 upon neutral loss of CO₂ and C₃H₆, respectively. The precursor of m/z 490 also reduced into fragment ion of m/z 386 (loss of C₄H₈O₃), which was also formed from m/z 416 on loss of CH₂O. The fragment of m/z 416, in parallel, reduced to ion of m/z 342 on loss of C₂H₂O₃, which was also formed from m/z 446 on loss of C₄H₈O₃. The precursor of m/z 386 dissociated into ion of m/z 344 on loss of C₃H₆. The latter, and also product ion of m/z 342, reduced into a common product ion of m/z 300 on loss of CO₂ and C₃H₆, respectively. The fragment of m/z 374 (loss of CH₂O) was generated from the parent ion of m/z 404, which further formed ion of m/z 330 (loss of CO₂). The latter dissociated into fragments of m/z 312 (loss of H₂O) and 288 (loss of C₃H₆), and both were further reduced into a common product ion of m/z 270 on loss of C₃H₆ and H₂O, respectively. The precursor of m/z 300 was also reduced into ions of m/z 282 (loss of H₂O) and 270 (loss of CH₂O) that were further converted to a common fragment of m/z 252 on loss of CH₂O and H₂O, respectively. An interesting observation was formation of a higher mass product ion of m/z 288 from m/z 270 on addition of H₂O. The ion of m/z 288 was incidentally also a degradation product T1, as described later. The ion of m/z 270 further yielded fragments of m/z 240 (on loss of CH₂O) and 206 (on loss of HO₂P). The latter two reduced to a common product ion of m/z 176 on loss of HO₂P and CH₂O, respectively. Finally, the precursor of m/z 252 resulted in product ion of m/z 136 on loss of C₄H₅O₂P.

The whole fragmentation pattern showed multiple unusual losses of CH₂O, CO₂ and C₂H₂O₃. The unusual loss of CH₂O was observed from the drug and its fragments of m/z 476, 416, 404, and 270, while the unusual loss of CO₂ was showed by the drug and precursor ions of m/z 490 and 374. The fragment of m/z 416 showed unusual loss of C₂H₂O_3. Among these unusual gas phase transitions, loss of CH₂O is reported in the case of adefovir, which is a congener of TDF.¹⁹ We anticipate that the two losses of CO₂ and C₂H₂O₃ follow a similar mechanism involving attachment of isopropyl alcohol to the cleavage site generated subsequent to loss of the mentioned moiety.

Solid state degradation behaviour

A total of twelve degradation products (T1–T12) were formed under different solid state stress conditions, as shown in Fig. 3. T1 and T2 were mainly formed under basic wet condition, while T1 was also observed in wet and all accelerated conditions, though in a low concentration. T3, T4 and T5 were detected under basic accelerated condition, while T5 was also observed under wet basic condition. T6 was observed almost in each condition, except basic accelerated. T7–T12 were observed in wet neutral and wet acidic environment in variable concentrations.

Fig. 3 Chromatograms showing degradation products of tenofovir disoproxil fumarate (TDF) formed under different solid state conditions.

Solution state degradation behaviour

Of the total twelve solid state degradation products, only two were observed in solution stress studies, i.e., T1 and T6. Both T1 and T6 were observed under neutral condition, while under acidic (0.5 N HCl) and basic environment (0.5 N NaOH), the drug was completely converted into degradation product T1 in 3 h, (Fig. 4).

Fig. 4 Chromatograms showing degradation products of tenofovir disoproxil fumarate (TDF) formed under different solution state conditions.

Characterization of the non-volatile degradation products

The LC-HRMS line spectra of the all the degradation products are shown in Fig. 5, while their accurate and exact masses, best molecular formulae and errors in mmu are listed in Table 3. The characterization of all the non-volatile degradation products was attempted and the same are discussed below individually or in the groups, categorized on the basis of similarities in their fragmentation pattern and/or their mechanism of formation.

Fig. 5 LC-HRMS line spectra of degradation products T1–T12 (a–l) of tenofovir disoproxil fumarate (TDF) formed on stressing the drug in solid and solution states.

Table 3 LC-HRMS data of degradation products T1–T12 of tenofovir disoproxil fumarate (TDF)^a

[DP + H]^+	Accurate mass	Molecular formula	Exact mass	Error (mmu)	Accurate mass of fragments
a DP, degradation product.
[T1 + H]^+	288.0818	C9H15N5O4P^+	288.0856	−3.8	270.0699, 240.0615, 206.1044, 176.0960
[T2 + H]^+	300.0803	C10H15N5O4P^+	300.0856	−5.3	288.0815, 282.0677, 270.0692
[T3 + H]^+	557.1521	C18H27N10O7P2^+	557.1534	−1.3	288.0823, 270.0739, 240.0619, 206.0995
[T4 + H]^+	587.1612	C19H29N10O8P2^+	587.1640	−2.8	557.1508, 300.0849, 288.0812, 270.0723, 240.0564, 206.1061
[T5 + H]^+	587.1593	C19H29N10O8P2^+	587.1640	−4.7	300.0823, 288.0813, 282.0774, 270.746
[T6 + H]^+	404.1354	C14H23N5O7P^+	404.1330	2.4	374.1210, 330.1307, 288.0857, 270.0772, 206.1062
[T7 + H]^+	416.1333	C15H23N5O7P^+	416.1330	0.3	386.1219, 342.1294, 314.0972, 300.0827, 282.0716
[T8 + H]^+	819.2546	C29H45N10O14P2^+	819.2586	−4.0	685.1968, 569.1475, 416.1333, 404.1317, 352.1139, 343.0986, 330.1269, 315.1042, 300.0821, 294.0833, 288.0814, 285.0762, 282.0711, 270.0674
[T9 + H]^+	490.1666	C18H29N5O9P^+	490.1697	−3.1	460.1561, 430.1087, 374.1169, 356.1085, 314.0601, 288.0807, 270.0700
[T10 + H]^+	610.2103	C22H37N5O13P^+	610.2120	−1.7	592.1980, 532.1789, 494.1619, 476.1511, 378.1125, 360.1029
[T11 + H]^+	935.3023	C34H53N10O17P2^+	935.3060	−3.7	819.2608, 685.1987, 611.1977, 587.1630, 569.1533, 554.1609, 542.1609, 532.1786, 520.1797, 468.1523, 416.1324, 410.1315, 404.1328, 373.1328, 352.1065, 343.1031, 336.1251, 330.1303, 315.1053, 306.1001, 300.0815, 294.0827, 288.0830, 285.0778, 282.0725, 270.0707
[T12 + H]^+	1051.3457	C39H61N10O20P2^+	1051.3533	−7.6	935.3015, 861.3091, 819.2549, 745.2693, 685.1980, 629.2188, 611.2003, 587.1586, 569.1565, 554.1646, 542.1622, 532.1793, 526.1779, 520.1794, 468.1543, 446.1750, 431.1517, 416.1305, 410.1318, 404.1320, 401.1283, 373.1293, 352.1069, 343.1020, 336.1275, 330.1304, 315.1060, 306.1003, 300.0815, 294.0823, 288.0834, 285.0786, 282.0726, 270.0722

---

T1. Accurate mass of [T1 + H]⁺ was found to be 288.0818 Da with error and molecular formula as −3.8 mmu and C₉H₁₅N₅O₄P⁺, respectively. Its mass and hence structure was similar to a drug fragment (Fig. 2), and it also showed further fragmentation behaviour similar to that in the drug, viz., m/z 288 → 270 → 240, 206, with the latter two converting to common product ion of m/z 176. Accordingly, the structure of T1 was assigned as tenofovir (Fig. 2), which is reported as an impurity in USP pending monograph.⁵

T2. Accurate and exact masses of [T2 + H]⁺ were found to be 300.0803 and 300.0856 Da, respectively (error −5.3 mmu). Its molecular formula was worked out to be C₁₀H₁₅N₅O₄P⁺. The structure for the same was assigned as (1-(6-(methyleneamino)-9H-purin-9-yl)propan-2-yloxy)methyl phosphonic acid, justified by its fragmentation pattern (Fig. 6). Upon fragmentation, it formed product ions of m/z 288 (generated on simultaneous addition of H₂O and loss of CH₂O) and 282 (loss of H₂O). The ion of m/z 288 further dissociated into fragment of m/z 270 on loss of H₂O.

Fig. 6 Structures and fragmentation behaviour of degradation products T2, T5–T8, T11 and T12. See Fig. 2 for the structures of fragments not shown in this figure. The structures of degradation products are shown in solid boxes.

T3 and T4. These two degradation products were generated on exposure of drug in the presence of Na₂CO₃ for 3 months under accelerated stability test condition. The accurate masses of their molecular ions were found to be m/z 557.1521 and 587.1612, respectively. Elemental composition calculator suggested their formulae as C₁₈H₂₇N₁₀O₇P₂⁺ (error −1.3 mmu) and C₁₉H₂₉N₁₀O₈P₂⁺ (error −2.8 mmu), respectively.

The common fragmentation pattern of both degradation products is shown in Fig. 7. The parent of m/z 557 reduced to product ion of m/z 288 (loss of C₉H₁₂N₅O₃P) and 270 (loss of C₉H₁₄N₅O₄P). Direct formation of these two fragments indicated that the parent was formed due to addition of two tenofovir molecules after loss of water. The molecular ion of T4 fragmented into product ions of m/z 557, 300 and 288 on the neutral loss of CH₂O, C₉H₁₄N₅O₄P and C₁₀H₁₄N₅O₄P, respectively. The loss of CH₂O meant that methylene was attached with oxygen (different from T5). The fragment of m/z 288 further dissociated into fragment of m/z 270 on loss of H₂O, the same product ion as in T3. The latter in both cases underwent loss of CH₂O and HO₂P and generated fragment ions of m/z 240 and 206, respectively. These latter two on loss of HO₂P and CH₂O reduced into a common product ion of m/z 176. Based on these observations, the structures of T3 and T4 were proposed as bis((1-(6-amino-9H-purin-9-yl)propan-2-yloxy)methyl) pyrophos-phonic acid, and (1-(6-((((1-(6-amino-9H-purin-9-yl) propan-2-yloxy)methyl) (hydroxy)phosphoryloxy)methylamino)-9H-purin-9-yl)propan-2-yloxy)methylphosphonic acid, respectively.

Fig. 7 Fragmentation behaviour of degradation products T3 and T4. See Fig. 2 for missing structures here. The structures of degradation products are shown in solid boxes.

T5, T8, T11 and T12. T5, T8 and T12 were elucidated to be methylene linked dimers of tenofovir, tenofovir monosoproxil and tenofovir disoproxil, respectively. The latter two are controlled as related substances in drug monograph in Ph. Int.⁶, while T12 is also controlled in USP,⁵ respectively. T11 was characterized as tenofovir mono and disoproxil methylene linked heterodimer (controlled as an impurity in Ph. Int.⁶). Structures of the same along with their fragments are shown in Fig. 6. The parent of m/z 1051 ([T12 + H]⁺) was reduced into the product ion of m/z 935 ([T11 + H]⁺) on loss of C₅H₈O₃. [T11 + H]⁺ further reduced into fragments of m/z 861, 819 ([T8 + H]⁺), 532 and 520 on loss of C₂H₂O₃, C₅H₈O₃, C₁₄H₂₂N₅O₇P and C₁₅H₂₂N₅O₇P, respectively. The fragments of m/z 745, 629 and 611 were formed from the precursor of m/z 861 upon sequential losses of C₅H₈O₃, C₅H₈O₃ and H₂O. The precursor of m/z 819 lost C₅H₁₀O₄ moiety to form a product ion of m/z 685, which further reduced to m/z 569. Another fragment formed from the same precursor was of m/z 587 ([T5 + H]⁺) that was generated on loss of C₁₀H₁₆O₆. The same was further dissociated into product ions of m/z 300 (loss of C₉H₁₄N₅O₄P) and 288 (loss of C₁₀H₁₄N₅O₄P). The latter two on independent loss of H₂O formed fragments of m/z 282 and 270, respectively. The zoomed mass spectra of T8, T11 and T12 are shown in Fig. 8a–c. It depicted number of doubly charged fragment ions, which were confirmed via isotopic peak having difference of 0.5 Da instead of 1 (Fig. S1†). Fig. 9 shows the dissociation pathway of doubly charged fragments, where parent of m/z 526 fragmented into ion of m/z 468 upon loss of C₅H₈O₃. The latter was further reduced into the fragment ions of m/z 431 and 410 upon loss of C₂H₂O₃ and C₅H₈O₃, respectively. The precursor of m/z 431 upon the sequential losses of C₅H₈O₃, C₅H₈O₃ and H₂O generated fragments of m/z 373, 315 and 306, respectively. The ion of m/z 410 dissociated into products of m/z 343, 336 and 294 on loss of C₅H₁₀O₄, C₄H₄O₆ and C₁₀H₁₆O₆, respectively. The precursor of m/z 343 was further reduced to the ion of m/z 285 on loss of C₅H₈O₃.

Fig. 8 Zoomed mass spectra of degradation products T8 (a), T11 (b) and T12 (c).

Fig. 9 Fragmentation scheme for doubly charged fragments of degradation products T8, T11 and T12.

T6 and T7. The accurate masses of molecular ions of these two degradation products were found to be 404.1354 and 416.1333 Da, respectively (Fig. 5f and g). The exact masses (molecular formulae, error in mmu) were m/z 404.1330 (C₁₄H₂₃N₅O₇P⁺, 2.4) and 416.1330 (C₁₅H₂₃N₅O₇P⁺, 0.3), respectively. T6 had similar structure to the drug fragment of m/z 404 and hence further fragmentation was also the same (Fig. 6). T7 (m/z 416) also reduced to a few fragments that had masses similar to the drug fragments of m/z 386, 342 and 300, but structures for these fragments were somewhat different (Fig. 6). The products of m/z 386, 342 and 314 were formed from the parent, T7, on loss of CH₂O, C₂H₂O₃ and C₄H₆O₃, respectively. The latter further dissociated into fragments of m/z 300 and 282 upon sequential losses of CH₂ and H₂O, respectively. The structures for T6 and T7 were proposed as tenofovir monosoproxil (controlled in the drug monograph in USP⁵ and Ph. Int.⁶) and its methylene adduct, respectively.

T9. The degradation product T9 was proposed as carbamate product, formed due to interaction of tenofovir monosoproxil and isopropyl hydrogen carbonate on loss of H₂O. The accurate mass of its molecular ion was m/z 490.1666 and the corresponding exact mass was m/z 490.1697 (error = −3.1 mmu). Its molecular ion formula worked out to be C₁₈H₂₉N₅O₉P⁺ (Fig. 10). Initially, it fragmented to product ions of m/z 460 and 430 upon neutral losses of CH₂O and C₃H₈O, respectively. The ion of m/z 460 dissociated further to a fragment of m/z 374, which also generated two product ions of m/z 356 (loss of H₂O) and 288 (loss of C₄H₆O₂). Even the latter reduced into product ion of m/z 270 on loss of H₂O. The precursor of m/z 430 fragmented to product ion of m/z 314 on loss of C₅H₈O₃.

Fig. 10 Fragmentation behaviour of degradation product T9. For missing structures see Fig. 2. The structure of degradation product is shown in the solid box.

T10. The accurate mass of molecular ion of T10 was m/z 610.2103, with error from exact mass as −1.7 mmu. Elemental composition calculator suggested its formula as C₂₂H₃₇N₅O₁₃P⁺. The structure and fragmentation pathway are shown in Fig. 11. The parent of m/z 610 dissociated into product ions of m/z 592, 532 and 494 upon neutral loss of H₂O, C₂H₆O₃ and C₅H₈O₃, respectively. The latter ion of m/z 494 was further dissociated through the route m/z 476, 378 → 360. On the basis of all these observations, its structure was proposed as (((1-(6-(((hydroxymethoxy)methoxy)methylamino)-9H-purin-9-yl)propan-2-yloxy)methyl)phosphoryl)bis(oxy)bis(methylene)isopropyl dicarbonate.

Fig. 11 Fragmentation behaviour of degradation products T10, whose structure is shown in the solid box.

Postulated degradation pathway for non-volatile degradation products T1–T12

The postulated pathway for the formation of non-volatile degradation products T1–T12 of TDF in solid state stress conditions is outlined in Fig. 12 (Scheme A). The major degradation products of the drug, both in solid and solution states (Fig. 3 and 4, respectively), were tenofovir isoproxil monoester (T6) and tenofovir (T1), which were formed on sequential hydrolysis of two isoproxil moieties from the drug. The released isoproxil moiety was further degraded to isopropyl hydrogen carbonate, generating formaldehyde in the process (Fig. 12, Scheme B), as also reported earlier.^3,4 Isopropyl hydrogen carbonate further interacted with T6 to form a carbamate product T9. T1 formed pyrophosphonic acid (T3) by the loss of water.

Fig. 12 Postulated degradation pathway for the formation of non-volatile degradation products T1–T12 (Scheme A). Structure shown in dotted box is an intermediate. The formation of isopropyl alcohol (VT1) and formaldehyde (HCHO) from isoproxil moiety is shown in Scheme B.

The released formaldehyde formed imine via Schiff base mechanism and resulted in degradation products T2, T7 and an intermediate²⁰ (shown in dotted box in Fig. 12, Scheme A). The latter combined with oxydimethanol (formed on interaction of two molecules of formaldehyde with water^21,22) resulted in formation of the degradation product T10. The imine moiety of T2 interacted with phosphonic acid and amino groups of T1, resulting in T4 and T5, respectively. Also, the imine moiety of T7 on interaction with free amino group of T6 resulted in T8. T12 was generated from combination of imine intermediate and TDF, which on hydrolysis of one of isoproxil moiety resulted in T11. T11 could also form via two other routes, viz., (1) combination of T7 and TDF and (2) T6 and imine intermediate.

Typical behaviour observed in presence of the solvents

It was observed that when solid or solution state samples were prepared in presence of CH₃CN (CH₃CN : H₂O, 60 : 40 v/v), degradation products T1, T2, T3, T4 and T5 showed two peaks (Fig. S2†), which had similar mass and UV spectrum in each case. The presence of CH₃OH (CH₃OH : H₂O, 60 : 40 v/v), resulted in the formation of two products, which were detected during LC-HRMS studies (Fig. S3a and b†) and had mass of m/z 302.0994 and 316.1153 (Fig. S3c and d†). These two were characterized to be methyl esters of tenofovir (Fig. S3e†), formed in the presence of methanol. The same structures have erroneously been reported by Anandgaonkar et. al.⁷ as degradation products of TDF which in fact should only be considered as interaction products of drug base with the solvent.

Investigation of volatile degradation products

The chromatogram obtained on HS-GC-MS study of the drug solution in water stressed at 80 °C for 15 h resulted in the formation of one volatile degradation product, VT1 (Fig. 13a) with molecular ion of m/z 60 and base peak of m/z 45 (Fig. 14a). The same was characterized as isopropyl alcohol. Its generation was justified considering that the drug on stressing in water was converted to degradation products T6 and T1 upon hydrolysis of one and two isoproxil moieties, respectively (Fig. 4), and degradation of the latter resulted in VT1, by following the same pathway as shown in Fig. 12 (Scheme B). The same pathway was also responsible for the formation of formaldehyde, as also reported in the literature.^3,4

Fig. 13 GC chromatograms showing volatile components of tenofovir disoproxil fumarate (TDF) in H₂O (a) and CH₃OH : H₂O (50 : 50 v/v) (b).

Fig. 14 EI mass spectra of volatile components VT1 (a), VT2 (b) and VT3 (c) along with their structures and fragmentation pattern.

Because formaldehyde was responsible for formation of most of the non-volatile degradation products, a study was hence carried out to confirm its presence. For the same, CH₃OH : H₂O (50 : 50, v/v) was deliberately used as a solvent, as it is known that methanolic solution of formaldehyde has the tendency to convert into hemiformal and poly(oxymethylene) hemiformal.^22,23 The presence of these were looked into by conducting another HS-GC-MS run on the methanolic sample solution. Practically, two peaks in addition to methanol were seen in the chromatogram, labelled as VT2 and VT3 in Fig. 13b. However, their masses of 76 Da (Fig. 14b) and 106 Da (Fig. 14c) did not match hemiformal or poly(oxymethylene) hemiformal. The fragmentation behaviour of VT2 and VT3 involved the routes m/z 76 → 75 (loss of H radical) → 45 (loss of CH₂O) → 29 (loss of CH₄) and m/z 106→105 (loss of H radical) → 75 (loss of CH₂O) → 59 (loss of CH₄), 45 (loss of CH₂O) → 29 (loss of CH₄), respectively. The two were instead identified as dimethoxymethane and methoxy(methoxymethoxy)methane. Fig. 15 describes the pathway of their formation involving interaction of one and two molecules of formaldehyde with one molecule of methanol to yield initially hemiformal and di(oxymethylene) hemiformal, which further interacted with another molecule of methanol and loss of water resulting in VT2 and VT3, respectively.

Fig. 15 Scheme showing the formation of volatile components VT2 and VT3.

Conclusion

The present study provided comprehensive and comparative degradation behavior of TDF in both solid and solution states. A total of twelve non-volatile degradation products were formed under the studied conditions. The same were characterized by using LC-HRMS. HS-GC-MS studies were also conducted for the identification of volatile degradation products. In the process of characterization, an exhaustive mass fragmentation pathway of the drug was established, which might also be helpful for characterization of its impurities, interaction products, and metabolites.

Acknowledgements

The authors are thankful to Dr M. Narayanam (Biocon-BMS R&D Centre, Bangalore, India) for helpful suggestions.

References

1. G. J. Dore, D. A. Cooper, A. L. Pozniak, E. DeJesus, L. Zhong, M. D. Miller, B. Lu and A. K. Cheng, J. Infect. Dis., 2004, 189, 1185–1192.
2. S. Pol and P. Lampertico, J. Viral Hepat., 2012, 19, 377–386.
3. L.-C. Yuan, T. C. Dahl and R. Oliyai, Pharm. Res., 2001, 18, 234–237.
4. M. Fardis, in Prodrugs: Challenges and Rewards Part 1, ed. V. J. Stella, R. T. Borchardt, M. J. Hageman, R. Oliyai, H. Maag and J. W. Tilley, Springer, New York, 2007, 5.20, pp. 647–657.
5. The United States Pharmacopeia, http://www.usp.org/sites/default/files/usp_pdf/EN/USPNF/pendingStandards/m3429_authorized.pdf, accessed July 2015.
6. Tenofovir disoproxil fumarate, The International Pharmacopoeia, World Health Organization, Geneva, 2014.
7. V. Anandgaonkar, A. Gupta, S. Kona and M. V. N. K. Talluri, J. Pharm. Biomed. Anal., 2015, 107, 175–185.
8. S. Manavarthi and G. S. Chhabra, Der Pharma Chemica, 2014, 6, 401–409.
9. H. C. Bhirud and S. N. Hiremath, J. Pharma Res., 2013, 7, 157–161.
10. S. Havele and S. R. Dhaneshwar, Sci. World J., 2012, 2012, 1–6.
11. S. Sutar, S. Patil and S. Pishavikar, Degradation Study of Tenofovir Disoproxil Fumarate, https://www.lap-publishing.com/catalog/details//store/gb/book/978-3-659-57963-9/degradation-study-of-tenofovir-disoproxil-fumarate, accessed Sep 2015.
12. M. N. Trinath, B. M. Gurupadayya, S. Prasad and S. Kache, Am. J. PharmTech Res., 2013, 3, 279–293.
13. S. S. Hussen, P. Shenoy, N. Udupa and M. Krishna, Int. J. Pharm. Pharm. Sci., 2013, 5, 245–248.
14. S. Singh, T. Handa, M. Narayanam, A. Sahu, M. Junwal and R. P. Shah, J. Pharm. Biomed. Anal., 2012, 69, 148–173.
15. ICH, Q1B: Stability Testing: Photostability Testing of New Drug Substances and Products, 1996.
16. ICH, Q1A(R2): Stability Testing of New Drug Substances and Products, 2003.
17. S. Singh and M. Bakshi, Pharmaceutical Technology On-Line, 2000, 24, 1–14.
18. S. Singh, M. Junwal, G. Modhe, H. Tiwari, M. Kurmi, N. Parashar and P. Sidduri, Trends Anal. Chem., 2013, 49, 71–88.
19. X. Chen, J. Xing and D. Zhong, J. Mass Spectrom., 2004, 39, 145–152.
20. E. H. Cordes and W. P. Jencks, J. Am. Chem. Soc., 1962, 84, 832–837.
21. J. G. M. Winkelman, O. K. Voorwinde, M. Ottens, A. A. C. M. Beenackers and L. P. B. M. Janssen, Chem. Eng. Sci., 2002, 57, 4067–4076.
22. I. Hahnenstein, H. Hasse, C. G. Kreiter and G. Maurer, Ind. Eng. Chem. Res., 1994, 33, 1022–1029.
23. J. P. Guthrie, Can. J. Chem., 1975, 53, 898–906.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra17532a

---