3.1 Method development
A time-consuming gradient HPLC method (acetonitrile/water/acetic acid in a ratio of 57:43:1 (v/v/v) and acetonitrile/water/acetic acid in a ratio of 90:10:1 (v/v/v) are used as mobile phase A and mobile phase B, respectively.) for related substances test of CDC is recruited in USP and BP monographs. In order to save time, an UHPLC method was developed in this study. The UHPLC method was initiated with a gradient program and flow rate obtained by converting EP method with a Agilent HPLC/UHPLC translator. When optimum resolutions were not achieved on the flow rate of 0.2 mL/min obtained by the translator, 0.3 mL/min was used for method optimization. The optimal resolution, elution time, and peak shapes were achieved using the following gradient program 0 min (100%A–0%B) → 0.6 min (100%A–0%B) → 6.6 min (0%A–100%B) → 10.0 min (0%A–100%B) → 10.2 min (100%A–0%B) → 13.0 min (100%A–0%B). In an effort to promote the response of compounds in mass spectrometry, 1% acetic acid involved in the HPLC method was replaced by 0.1% formic acid. ACQUITY UPLC BEH C18 (2.1 × 150 mm, 1.7 µm) and ZORBAX Eclipse XDB-C18 column (2.1 mm× 100 mm, 1.8 µm) were the columns initially selected for RS study. It was found that the interference of gradient to RS9 was considerable and the resolution between RS4 and RS5 (R༜1.5)was poor when using the former one. Thus ZORBAX Eclipse XDB-C18 column (2.1 mm× 100 mm, 1.8 µm) was employed to achieve the resolution of individual related substances from CDC. The optimized chromatographic conditions are shown in Section 2.2.
3.2 Chromatographic characteristics
Figure 2 shows the chromatograms of CDC test solution and all the degradation samples. Among them, ten forced degradation products (RS1-6, 8–11) and one process-related impurity (RS7) were detected. Although the amount of RS9 did not increase significantly under forced degradation conditions in this study, it was still classified as degradation product according to the study by Mehta et al.[8] Degradation of CDC was observed under acid hydrolysis (RS2 and 9), alkaline hydrolysis (RS1), oxidative (RS1-6, 8 and 10), photolytic (RS10) degradation conditions and accelerated stability study (60 ℃ for ten days) (RS2 and 11) as described in Section 2.3. None of degradation product was observed under accelerated stability studies (90% RH and 4500lx for ten days for ten days) (Fig. S1).
3.3 Mass fragmentation pattern of CDC
UHPLC-Q-TOF-MS spectrum of CDC was recorded to outline mass fragmentation pattern for assisting the characterization of related substances (Fig. 3). The protonated molecule of CDC produced a base peak at m/z 611.2613, corresponding to the formula of C33H34N6O6. The MS/MS spectrum of CDC showed main product ions at m/z 423, 395, 380, 367, 349, 338, 263, 235 and 207. The major fragment ion of m/z 423 was formed by the cleavage of the ester linkage from the parent ion, and it underwent further fragmentation to product ions at m/z 395 and 380 by the loss of N2 and N3H, respectively. The formation of fragment ion of m/z 367 can be attributed the loss of ethylene group from m/z 395, and further fragmentation gave product ion of m/z 349 and 338 by the loss of H2O molecule and N2H radical, respectively. The product ion at m/z 263 was formed due to the cleavage of 14-position C–N bond and ethyl group transfer from benzimidazole, which on further loss of N2 formed the fragment ion of m/z 235. The formation of fragment ion of m/z 207 can be attributed to the cleavage of 14-position C–N bond from m/z 395. The possible mass spectral fragmentation pathway of CDC is shown in Fig. 4, and and the chromatographic and mass spectrometric data are summarized in Table 1.
Table 1
UHPLC-Q-TOF-MS data of CDC and its related substances
Code | Retention time(min) | [M + H]+(m/z) | Formula | Dif(ppm) |
CDC | 4.793 | 611.2608 | C33H34N6O6 | 0.75 |
RS1 | 1.102 | 441.1679 | C24H20N6O3 | -2.12 |
RS2 | 3.039 | 583.2308 | C31H30N6O6 | -1.44 |
RS3 | 3.619 | 627.2567 | C33H34N6O7 | -0.84 |
RS4 | 3.727 | 627.2562 | C33H34N6O7 | − .044 |
RS5 | 3.941 | 627.2573 | C33H34N6O7 | -1.80 |
RS6 | 4.088 | 611.2616 | C33H34N6O6 | -0.56 |
RS7 | 4.189 | 597.2450 | C32H32N6O6 | 1.02 |
RS8 | 5.152 | 611.2630 | C33H34N6O6 | -2.85 |
RS9 | 6.227 | 639.2947 | C35H38N6O6 | -3.35 |
RS10 | 6.384 | 609.2456 | C33H32N6O6 | 0.02 |
RS11 | 7.318 | 639.2943 | C35H38N6O6 | -2.73 |
3.4. UHPLC-Q-TOF-MS studies on known RSs
The RS 1, 2, 6, 8, 9 and 11 were identified as EP impurity G, B, C, D, E and F, respectively, by comparing their retention times and fragment ions with those of reference substances used in the analysis of related substances of CDC. The UHPLC chromatogram of system suitability solution (containing EP impurity B, C, D, E, F, G, I and CDC references) is shown in Fig. S1. MS/MS spectrum and possible mass spectral fragmentation pathway of RS1, 2, 6, 8, 9, 11 are shown in Fig. S2 and Fig. S3, respectively.
3.5. UHPLC-Q-TOF-MS studies on unknown RSs
3.5.1 RS3 and RS5
RS3 and RS5 were the major degradation products under oxidative degradation condition with decreased retention compared to the drug (Fig. 2e). The observed accurate m/z value of RS3 and RS5 were 627.2567 and 627.2573, respectively. Based on the exact mass data, the elemental compositions of RS3 and RS5 are both C33H34N6O7. By comparing with the elemental composition of the drug, it was found that RS3 and RS5 had an additional oxygen atom due to oxidation. As shown in Fig. 3, RS3 and RS5 had the same fragmentation pattern, indicating that they were isomeric to each other. The product ions at m/z 439, 411, 354, 279 and 223 were also 16 Da more than each of the corresponding fragment ions of CDC at m/z 423, 395, 338, 263 and 207, suggesting the structural difference was attributed to biphenyl-tetrazole structure. The characteristic fragment peak at m/z 180, which was assigned as methylene-carbazole ion, was both found in MS/MS spectra of RS3 and RS5, indicating the extra oxygen atom was in tetrazole structure. Considering the weaker retention in comparison with that of CDC, it was reasonable to speculate that RS3 and RS5 were N-hydroxyl derivatives of the drug. Hydroxyl addition could be possible at both N1 and N2 positions, and it was difficult to distinguish between them merely on the basis of mass studies. The calculated cLogP values (Chemdraw software 14.0) of EP impurity E (N1-ethyl CDC) and EP impurity F (N2-ethyl CDC) were 7.86889 and 7.65889, respectively. The calculated cLogP values (Chemdraw software 14.0) of N1-hydroxyl CDC and N2-hydroxyl CDC were 6.399 and 6.031, respectively, which indicated the relative retention time of RS3 and RS5 was similar as that of EP impurities E and F. Moreover, as observed in the stability sample (60 ℃ for ten days), EP impurity F was formed in higher amount than EP impurity E, The rationale for this might be due to the steric hindrances present in EP impurity E.[7] RS3 and RS5 also displayed a similar kind of the relative extent of formation under oxidative stress. Therefore, RS3 and RS5 were proposed to be N1-hydroxyl CDC and N2-hydroxyl CDC, respectively. The formation of similar N1-hydroxyl product on the oxidative process has been reported in the case of valsartan.[11] The possible mass spectral fragmentation pathways of RS3 and RS5 are shown in Fig. 4.
3.5.2 RS4
RS4 was one of the major degradation products under oxidative degradation condition with reduced retention. Similar to RS3 and RS5, the observed accurate m/z value of RS4 was 627.2562. Elemental composition calculator suggested the same molecular formula for it as RS3 and RS5, indicating an additional of hydroxyl group compared to the drug. The MS/MS spectrum of RS4 showed product ions at m/z 439, 377, 289 and 207, which was quite different from RS3 or RS5. The fragment ion at m/z 439 indicated that the additional hydroxyl group was in candesartan structure. The characteristic fragment peak at m/z 377 was explained by the cleavage of 14-position C–N bond from the parent ion. As m/z 377 was not found in the MS/MS spectrum of CDC, it is reasonable to speculate that the additional hydroxyl group in RS4 made 14-position C–N bond easier to break. The fragment ion at m/z 289 was explained by the cleavage of ester linkage from m/z 377 followed with cyclohexyloxy transfer. According to exact mass data, the fragment of m/z 207.0767 was assigned as protonated 2-ethoxy-1H-benzimidazole-4-carboxylic acid, which was different from equivalent fragment of drug with 207.0923. Based on the above information, RS4 was identified as 14- hydroxyl CDC. Similar oxidation mechanism in benzyl position through a free radical mediated mechanism in the case of morphine has been discussed in the literature.[12–14] The possible mass spectral fragmentation pathway of RS4 is shown in Fig. 4.
3.5.3 RS7
RS7 was a process-related impurity with decreased retention compared to the drug in the test solution a (Fig. 2a). The observed accurate m/z value of RS7 was 597.2456. Based on the exact mass data, the elemental composition of RS7 is C32H32N6O6, which is 14 Da less than the mass of the drug, indicating the absence of a CH2 group in RS7. The product ions at m/z 409, 381, 363 and 221 was also 14 Da less than that of the corresponding fragments of the drug at m/z 423, 395, 349 and 235. Moreover, the product peaks at m/z 338 and 207 were also present in the MS/MS spectrum of the drug. The characteristic fragments of m/z 221 and 207 suggested the migration of a methyl group, indicating that RS7 was structurally different from the drug where the ethoxy group attached to the imidazole ring was replaced with a methoxy group. By comparing the retention time with that of a reference substance (Fig. S4), RS7 was confirmed to be CDC methoxy analogue. It is not reported as a related substance in CDC substances or preparations in the literature. According to the synthesis routine of CDC, the detritylation step was carried out in aqueous methanol containing hydrochloric acid. RS7 is possible to be the overreacting product during this step in methanol/HCl system.[15] The mass possible spectral fragmentation pathway of RS7 is shown in Fig. 4.
3.5.4 RS10
RS10 was the major degradation product under photolytic condition. The observed accurate m/z value of RS10 was 609.2456. Based on the exact mass data, the elemental composition of RS10 was C33H32N6O6, which was 2 Da less than that of CDC, which suggested the loss of two hydrogen atoms. The MS/MS spectrum of RS10 showed product ions at m/z 421, 393, 233 and 205, these ions were also 2 Da less than each of the corresponding CDC fragment ions at m/z 423, 395, 235 and 207, indicating the structural difference was attributed to biphenyl-tetrazole structure. The fragment ion of m/z 439 was formed by the loss of cilexetil moiety from the parent ion, and it underwent further fragmentation to product ion at m/z 411 by the loss of N2. A likely candidate for RS10 was cyclized analogue of CDC which was reported by Mehta et al. as a photolytic degradation product with the same MS properties.[8] The possible mass spectral fragmentation pathway of RS10 is shown in Fig. 4.