Identification and structural characterization of potential degraded impurities of ribociclib by time of flight -tandem mass spectrometry, and their toxicity prediction

Abstract

The US FDA and EMA granted approval to Ribociclib (RIBO) for the treatment of metastatic breast cancers in 2017. The formation of impurities during the storage of any pharmaceutical product can significantly impact its overall toxicity and therapeutic effectiveness, ultimately posing a safety concern. Over time, it has been observed that impurities can sometimes cause serious adverse toxic effects, potentially leading to the withdrawal of a drug from the market. Therefore, a thorough characterization of potential impurities is extremely important for identifying molecular sites susceptible to structural changes. To the best of our current understanding, the potential degradation impurities of RIBO have not yet been identified. No study documented in the available literature has reported on the structural characterization of the degradation impurities of RIBO. In this study, a comprehensive stress study, following the guidelines recommended by the International Council for Harmonisation (ICH), was conducted on RIBO under hydrolytic, oxidative, photolytic, and thermolytic conditions. The resulting degradation products were characterized using tandem mass spectrometry, primarily employing electrospray ionization with a time-of-flight mass analyzer. The atmospheric pressure chemical ionization mode was utilized for the characterization of N-oxide degradation products where Meisenheimer rearrangement occurred. One degradation product was synthesized in our laboratory and fully characterized using NMR techniques, including 1H NMR, 13C NMR, DEPT, 2D NMR, and D2O exchange experiments. The source of formylation in the generation of degradation products was investigated by employing different solvent systems. The degradation pathways were elucidated by explaining the proposed mechanisms of degradation under various conditions. The in silico toxicity of the degradation impurities was evaluated using the ProTox-II toxicity prediction platform.

Introduction

Cyclin-dependent kinase (CDK) inhibitors represent a newer generation of anticancer drugs specifically designed for breast cancer treatment. These drugs selectively block CDK 4/6 and inhibit the growth of tumors. Ribociclib (RIBO) is a selective inhibitor of CDK4/6 that received approval from the United States Food and Drug Administration (USFDA) in 2017. It is used in combination with aromatase inhibitors for the treatment of estrogen receptor-positive and human epidermal growth factor receptor 2-negative breast cancers. Beyond its efficacy in breast cancer, it has also shown potential in mitigating other resistant solid tumors. Its ability to permeate the blood-brain barrier makes it a promising candidate for the treatment of central nervous system tumors. Stress studies are an essential component of the drug discovery and development process, serving to identify and characterize the degradation products (DPs) of a drug molecule. These studies are valuable for understanding the intrinsic stability of the drug and for establishing appropriate storage conditions. Stress studies, or forced degradation studies, are conducted under more extreme conditions than accelerated stability testing to generate the DPs that are likely to form during the long-term storage of the drug. Understanding the impact of environmental factors such as temperature, oxygen, light, and pH on drug molecules assists manufacturers in selecting suitable packaging materials and specifying appropriate handling conditions. The formation of impurities during storage can negatively affect both the safety and the efficacy of a drug. An altered chemical structure can significantly contribute to the overall toxicity and therapeutic effectiveness of a drug, ultimately leading to concerns about its safety. Over time, it has been observed that impurities can cause various undesirable toxic effects, which may even necessitate the withdrawal of a drug from the market. Therefore, a comprehensive characterization of all DPs is extremely important for identifying molecular sites that are prone to structural changes. Modern sophisticated analytical techniques allow for continuous improvement in the structural characterization of DPs. Reverse-phase high-performance liquid chromatography (RP-HPLC), liquid chromatography-tandem mass spectrometry (LC-MS/MS), and nuclear magnetic resonance (NMR) spectroscopy are among the most frequently employed analytical techniques for the identification of DPs. In addition to structural characterization, the LC-MS/MS technique has proven to be a highly efficient and indispensable tool for quantifying a wide range of analytes in various biological matrices at trace levels. This makes mass spectrometry an essential technique in the pharmaceutical field. The utilization of quadrupole-time of flight (Q-TOF) assisted high-resolution mass spectrometry (HRMS) plays a crucial role in qualitative analysis for determining the structures of DPs generated during stress studies. Knowledge of the structural modifications of a parent molecule that occur during the formation of DPs is important for elucidating the degradation pathway and for defining suitable storage conditions. Various studies have reported methods for the quantification of RIBO and its metabolites in biological samples. Two separate studies by Kala et al. and Bao et al. demonstrated the application of bioanalytical methods for quantifying RIBO in the brain. Another bioanalytical method by Martínez-Chávez et al. showed the quantification of three CDK 4/6 inhibitors, including RIBO, in biological samples from humans and mice. An in vitro metabolism study conducted by Neu et al. in rat and human microsomes reported ten metabolites. Another study by Alsubi et al. discussed the bioactivation pathway of RIBO in in vitro matrices. A detailed in vitro and in vivo metabolic investigation of RIBO was reported by Sahu and Sengupta. However, to the best of our current knowledge, no forced degradation study regarding the structural characterization of the DPs of RIBO has been reported to date. In this study, a comprehensive forced degradation study was conducted in accordance with the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) guidelines (ICH Q1A(R2) and ICH Q1B) for the identification and characterization of DPs of RIBO using RP-HPLC and Q-TOF-MS/MS analysis. Furthermore, a DP was synthesized, and its structure was elucidated using NMR techniques.

Experimental

Analyte and reagents

Pure RIBO (≥99 %) was obtained from MedChem Express, USA. Laboratory reagent grade Ethyl formate was procured from SD Fine-Chem limited, Mumbai. Hydrogen peroxide (H2O2) (30 %), hydrochloric acid (HCl), acetonitrile, and methanol were of analytical reagent (AR) grade and were obtained from Fisher Scientific. AR grade formic acid, ammonium formate, and sodium hydroxide (NaOH) were purchased from Merck. Ultrapure water used for all analyses was obtained from a Milli-Q water purification system (Millipore, Milford, MA, USA).

Instrumentation

The analysis of all degradation samples was performed using an Agilent 1260 series modular HPLC system, which included a quaternary pump (DEADP1979), an autosampler (DEADA0034), and a diode array detector (DAD) (DEAAX0589). The column compartment was equipped with a temperature control system to maintain a constant temperature throughout the analysis. For the qualitative analysis of DPs, a mass spectrometer with a TOF mass analyzer (Q-TOF-LC–MS 6545 series, Agilent Technologies, USA) was employed. Ionization of the compounds was achieved using an electrospray ionization (ESI) source, operated in positive mode to generate protonated mass spectra. The operating capillary voltage was set at 3000 V, with a skimmer voltage of 60 V. The fragmentor voltage was set at 130 V, and the collision energy was 30 eV to achieve optimal fragmentation. The vaporizer temperature was maintained at 350 °C. Nitrogen (N2) gas was used for both drying (320 °C and 11 L/min) and nebulization (40 psi). The mass scan range for MS was set from 100 to 1000 m/z, while for MS/MS, it was from 50 to 1000 m/z. NMR spectra (1H and 13C) were recorded on Bruker 500 MHz and 125 MHz spectrometers, using tetramethylsilane as an internal standard. Deuterated chloroform (CDCl3) was used to dissolve the samples for NMR analysis. Both 1H and 13C NMR chemical shifts are reported in parts per million (ppm) downfield from tetramethylsilane. A Radley parallel synthesizer (Essex, United Kingdom) was used to conduct the hydrolytic (acid, base, neutral) stress study, operating at controlled temperature and rotational speed. An ultra sonicator (Antech) and a pH meter (Eutech Instruments) were utilized at various stages of the analysis. For the thermal stress study, a hot air oven (Thermo Fisher Scientific) was used to provide dry heat to the samples. A photolytic stress study was performed using a photostability chamber (Newtronics Lifecare Pvt. Ltd) equipped with UV and fluorescent light sources, which effectively controlled both temperature and humidity inside the chamber within ± 2 °C and ± 5 % RH, respectively.

Stress degradation study

The stress study was designed following the recommendations of ICH guidelines, encompassing different conditions such as hydrolysis, thermal stress, oxidation (ICH Q1A (R2)), and photolysis (ICH Q1B). Both solution and solid-state stress studies were conducted under varying conditions (milder to stronger) to achieve optimal degradation. The solution-state stress study was performed at a concentration of 500 µg/mL. Due to the limited solubility of RIBO in water, acetonitrile and methanol were used as co-solvents for solubilization. However, during the oxidative stress study, the acetonitrile and water solvent system was found unsuitable due to vigorous degradation, possibly resulting from the formation of peracids from acetonitrile. Therefore, DMSO was explored as a co-solvent. Nevertheless, DMSO was incompatible with acid hydrolysis and oxidative stress due to drug precipitation. To avoid this complication, a methanol-water mixture at a ratio of 70:30 (v/v %) was used as the co-solvent throughout the study. Hydrolytic stress experiments were carried out in 1 N HCl (acidic), 1 N NaOH (basic), and water (neutral) for durations of 48 h, 7 h, and 120 h, respectively. The acid and base hydrolytic sample aliquots were collected and neutralized with base and acid, respectively, to stop further stress reaction. The thermal stress study was performed at 80 °C for 7 days in a hot air oven by exposing a thin uniform layer (1 mm thickness) of the drug in a petri dish to dry heat. The oxidative stress study was conducted with 0.3 % H2O2 at room temperature in the absence of light to prevent photo-oxidation. Solid-state photolysis was induced by irradiating the samples with ultraviolet (UV) and fluorescent light sources, with a total exposure intensity of 1000 Wh/m2 and 6 million lux h. For solutions, the light exposure intensity was 400 Wh/m2 (UV) and 2.4 million lux h, along with control samples wrapped in aluminum foil. The internal environment of the photostability chamber was maintained at 40 °C ± 2 °C and 75 % ± 5 % RH. Subsequently, all the aliquots were diluted to a concentration of 100 µg/mL with diluents and then filtered through a 0.22 µm PTFE filter before analysis.

Validation of HPLC method

The developed RP-HPLC quantitative method was validated following the ICH Q2 (R1) guideline to assess the reliability of the analysis. The method was validated for selectivity, sensitivity (limit of detection (LOD) and limit of quantification (LOQ)), linearity, accuracy, and precision. Sensitivity, in terms of LOD and LOQ, was estimated at a signal-to-noise (S/N) ratio of 3:1 and 10:1, respectively, by analyzing multiple injections (n = 6) into the HPLC system. A calibration curve was plotted using seven calibration concentrations covering a range from 40 µg/mL to 160 µg/mL, and the linearity of the method was established. Accuracy was determined in terms of recovery. An artificial excipient mixture was prepared with microcrystalline cellulose (MCC), silicon dioxide, polyvinyl pyrrolidone, and hydroxypropyl cellulose, into which RIBO was spiked at 80 %, 100 %, and 120 % of the assay concentration (100 µg/mL). These samples were analyzed in triplicate (n = 3) by HPLC, and the recovery was calculated by comparing the peak areas of the spiked samples to that of a neat sample. Repeatability was assessed by injecting six replicates (n = 6) of the assay concentration, and the percentage relative standard deviation (% RSD) was calculated. Intra-day and inter-day precision were evaluated by injecting six replicates (n = 6) of the assay concentration on the same day and on different days, respectively. The % RSD of the repeated injections was calculated for both intraday and inter-day precision.

In silico toxicity prediction of RIBO and its DPs

Upon degradation, a drug is converted into DPs, which have structural differences from the parent compound. These structural changes may lead to toxicity, often arising from the introduction of various structural alerts. Predicting toxicity during the drug discovery and development phase plays an essential role in designing a compound. The toxicity of the DPs was predicted and compared with that of RIBO using the ProTox-II computational platform. This software deduces the toxic potential of DPs by considering different aspects of the chemical entity and develops a comparative statistical analogy with molecules of known toxicity. Different models within the platform have different training sets (containing compounds with known toxicity), which helps to establish toxicological assumptions with a confidence score. Toxicity was predicted as severe by the software when the confidence score was greater than 70 % (0.7).

Result and discussion

Optimization of HPLC parameters

The development of a quantitative HPLC method was necessary to separate RIBO from its DPs with optimal resolution and characterization. Initial method development trials were conducted using an Agilent Zorbax Eclipse C18 column (250 mm × 4.6 mm, 5 µm) with various gradient programs employing different aqueous buffers and organic solvents. Volatile mass-compatible buffers such as ammonium acetate, ammonium formate, and formic acid were used along with acetonitrile or methanol as organic modifiers. When acetonitrile was used with ammonium formate (10 mM), DP3, DP4, DP5, and DP6 were not adequately resolved from each other. Subsequently, methanol was used in combination with acetonitrile at different compositions to resolve the co-eluting DPs. This mixed organic phase and a buffer with a pH of 5 failed to resolve DP3 and DP4. Ultimately, all DPs were resolved with good peak shape when ammonium formate with an ionic strength of 10 mM (pH 4.5) was used with an acetonitrile:methanol mixture (1:1 v/v) as the organic modifier. A gradient elution program was employed, starting with 20 % organic solvent for 2 min, then increasing to 50 % at 20 min and maintained for 2 min, followed by a gradual increase to 90 % for the next 8 min, held steady for 2 min, and finally returned to the initial composition at 37 min and maintained for 3 min. A sample volume of 10 µL was injected with a mobile phase flow rate of 1 mL/min. The stress samples were analyzed at a wavelength of 272 nm to detect RIBO and its DPs. The peaks were assigned based on their retention time (RT), and secondary DPs were numbered after primary DPs. The individual UV spectra of RIBO and its degradation products are shown. The stress samples were analyzed by LC–MS/MS in positive ionization mode (ESI) with a fragmentation voltage of 130 V, collision energy of 30 eV, and skimmer voltage of 60 V. Nitrogen gas was used as both sheath gas and nebulizing gas.

Degradation behaviour of RIBO

RIBO exhibited significant instability under oxidative stress conditions when acetonitrile was used as a co-solvent. Additionally, it formed one artifact ([M+H]+; m/z 474), which might have resulted from the interaction of acetonitrile with RIBO. When DMSO was used as a co-solvent, RIBO showed low solubility under oxidative and acidic hydrolysis conditions, leading to precipitation of the drug. Therefore, methanol was used as a co-solvent for the stress study of RIBO. Upon exposure to different stress conditions, RIBO was found to be susceptible to hydrolytic (acidic and basic) and oxidative stresses. It degraded under acidic (1 N HCl for 30 h) and basic (1 N NaOH for 7 h) hydrolysis to generate DP1. Upon oxidation with 0.3 % H2O2 for 60 h, DP3, DP5, DP7, DP8, and DP9 were formed. RIBO degraded to form DP4, DP6, and DP8 under solid-state photolysis, whereas DP2 and DP8 were formed in solution-state photolysis. No degradation of RIBO was observed under thermal and neutral hydrolytic stress conditions. DP8 was formed under oxidative stress conditions due to the involvement of methanol as a solvent. DP9 is a secondary DP formed by N-dealkylation from DP8. Furthermore, DP8 was observed in solution-state photolysis where acetonitrile and DMSO were used as co-solvents, indicating that the solvents were responsible for the production of N-formylated DP8. In acidic hydrolysis, a pseudo-DP (methyl ester of DP1) was formed due to the presence of methanol in the solvent. The primary focus of this study was to identify the primary degradation products of ribociclib. The less intense peaks observed in the stress studies were considered insignificant due to their formation in minute quantities even under harsh forced degradation conditions. The stress conditions employed for the degradation study are presented in a tabular format.

Mass fragmentation behaviour of RIBO and its DPs

The MS/MS spectra of RIBO and its DPs were recorded, and the mass fragmentation pathways were established as illustrated. The mass fragmentation behavior of RIBO was elucidated, and the structure of the fragment ions was confirmed based on accurate mass measurement. Additionally, the respective fragmentation behaviors of DPs were compared to that of RIBO, enabling the identification of structural modifications during degradation. The HRMS data of structure-conclusive product ions of RIBO and its DPs have been summarized.

MS/MS fragmentation behaviour of RIBO

RIBO eluted at a retention time (RT) of 15.9 min and exhibited a protonated parent ion at m/z 435. The MS/MS spectrum of RIBO showed abundant daughter ion peaks at m/z 367 after the loss of the cyclopentene moiety (m/z 68; -C5H8 from m/z 435). This indicates the presence of a cyclopentane ring in RIBO. The loss of m/z 45 and 28 was observed in the mass spectrum from m/z 367 and 322, confirming the presence of the N,N-dimethylamine carboxamide moiety. These losses produced daughter ions at m/z 322 and 294. Other characteristic and structure-descriptive fragment ion peaks at m/z 252 (loss of C2H6N from m/z 294), 189 (loss of C5H3N3 from m/z 294), 162 (loss of HCN from m/z 189), 134 (loss of C2H4 from m/z 162), and 72 (loss of C20H26N7 from m/z 435) provide substantial evidence for the structure of RIBO. The loss of the aziridine (m/z 43) moiety from m/z 162 forms the fragment ion at m/z 119, indicating the existence of a piperazine ring in the structure. The fragmentation pattern of RIBO is illustrated, and this pattern was subsequently used for assigning the structures of its DPs.

Mass fragmentation behaviour of DP1- DP10

MS/MS fragmentation behaviour of DP1 ([M+H]+; m/z 408)

DP1 (RT 8.6 min) was formed under both acidic and basic hydrolytic conditions. The MS/MS spectrum of DP1 in positive ESI mode showed a protonated parent ion peak at m/z 408, with a possible elemental formula of C21H26N7O2+. This indicates the presence of an additional oxygen atom compared to RIBO. The even m/z value of the molecular ion confirms that the nitrogen count in RIBO was altered after degradation, and the DP contains an odd number of nitrogen atoms. Predominant fragment ion peaks at m/z 340, resulting from the loss of the cyclopentene group (-C5H8, m/z 68 from m/z 408), confirm that the cyclopentane ring remained unchanged in DP1. The appearance of the most significant daughter ion peak at m/z 322 (H2O loss from m/z 340) suggests the replacement of the terminal N,N-dimethylamine group by a hydroxyl group. The sequential loss of H2O and CO confirms the presence of a -COOH group in DP1. Moreover, the appearance of m/z 179 (at the expense of C9H11N3 from m/z 340) in the spectrum revealed the conversion of the N,N-dimethyl carboxamide residue into a carboxylic acid functionality. Further fragmentation occurred in a similar pattern to RIBO. Other structure-illustrative peaks observed at m/z 294, 252, 189, and 162 confirm that the N-(5-(piperazin-1-yl)pyridin-2-yl)-7H-pyrrolo[2,3-d]pyrimidin-2-amine residue remained unaffected and was consistent with the fragmentation of RIBO. Based on accurate mass measurement and spectral information, the structure of DP1 was proposed to be 7-cyclopentyl-2-(5-(piperazin-1-yl)pyridin-2-ylamino)-7H-pyrrolo[2,3-d]pyrimidine-6-carboxylic acid.

MS/MS fragmentation behaviour of DP2 ([M+H]+; m/z 467)

DP2 was generated in solution under photolytic stress conditions and eluted at 11.2 min. The MS/MS spectrum of DP2 under positive ESI exhibited a protonated peak at m/z 467, with the most probable elemental formula C23H31N8O3+. The additional 32 Da mass in DP2 compared to RIBO indicates the presence of two extra oxygen atoms in the DP. The product ion appeared at m/z 439 upon the loss of CO from the parent ion, suggesting that DP2 might have formed by hydroxylation on an aromatic ring. The formation of a product ion at m/z 381 due to the neutral loss of piperazine (C4H10N2) rules out the possibility of hydroxylation on the piperazine ring. Other prominent daughter ions were observed after the loss of C5H8 (m/z 467→399, 439→371, 381→313) and C2H7N (m/z 399→354). This indicates that the cyclopentane ring along with the N,N-dimethylamine moiety did not undergo hydroxylation. The occurrence of structure-specific product ions at m/z 212 from m/z 240 and m/z 184 from m/z 212 suggests that dihydroxylation could have occurred on the pyridine ring of the 1-(pyridin-3-yl)piperazine moiety, which exhibited peculiar CO losses instead of water loss. The fragment ion peak at m/z 422, formed by the loss of a CH3NO group from the parent ion, suggests the possibility that one of the hydroxylation sites might be at the ortho position to the pyridinyl nitrogen. Therefore, the sites of hydroxylation were assigned at the ortho (6-) and meta (3-) positions to the pyridinyl nitrogen atom, owing to the affinity of the hydroxyl radical (OH˙) towards the less electron-deficient meta (3-) and ortho (6-) carbons.

MS/MS fragmentation behaviour of DP3 ([M+H]+; m/z 409)

Under oxidative stress conditions (0.3 % H2O2), DP3 originated from RIBO and eluted at 12.26 min. The recorded MS/MS spectrum of DP3 under positive ESI mode showed a protonated peak at m/z 409 with a probable elemental composition of C21H29N8O+, exhibiting a 26 Da mass deficit compared to RIBO. This characteristic m/z difference was consistent throughout the fragmentation pattern of DP3. Evidence of the structural integrity of both the cyclopentane and N,N-dimethylamine residues was supported by the presence of prominent product ions such as m/z 341, 296, and 268, resulting from the sequential loss of cyclopentene, N,N-dimethylamine, and carbon monoxide from the parent ion, respectively. The loss of ammonia (NH3; m/z 17) observed from m/z 341 and 296 (producing m/z 324 and 279 peaks, respectively) suggests the presence of a free aliphatic amino residue in the structure. Furthermore, the structure-denoting product ions m/z 163 (losing C5H3N3 from m/z 268) and 136 (losing HCN from m/z 163) confirm that the piperazine ring was structurally compromised and underwent N-dealkylation. Other fragment ion peaks appeared at m/z 364 and 238, consistent with the proposed fragmentation pattern corresponding to the proposed structure. Based on accurate mass measurement and the fragmentation pattern, DP3 was structurally confirmed as an N,N-dealkylated product of RIBO.

MS/MS fragmentation behaviour of DP4 ([M+H]+; m/z 451)

The solid-state photolytic impurity DP4 eluted at 12.5 min. The MS/MS spectrum in ESI mode showed a protonated parent ion peak at m/z 451 with a suggested elemental composition of C23H31N8O2+. Both the m/z value and elemental composition indicated the presence of one additional oxygen atom in the structure. The characteristic abundant daughter ion at m/z 421, resulting from the loss of formaldehyde (HCHO; m/z 30) from the parent ion, suggests that the oxygen might be present on the terminal methyl group of the N,N-dimethylamine, as observed in other similar reported studies. The H2O loss, which produced a product ion peak at m/z 433, was less significant, pertaining to the possibility of N-oxidation. However, the possibility of N-oxidation was ruled out when atmospheric pressure chemical ionization–mass spectrometry (APCI-MS) analysis did not show the characteristic loss of 16 due to the elimination of an oxygen atom. The loss of HCHO can therefore be attributed to the hydroxylation of the methyl group of the N,N-dimethylcarboxamide moiety, as reported in the literature. Beyond m/z 353 (losing m/z 68 (-C5H8) from m/z 421), similar fragmentation was observed as in the case of RIBO, indicating that the N-(5-(piperazin-1-yl)pyridin-2-yl)-7H-pyrrolo[2,3-d]pyrimidin-2-amine moiety was not structurally compromised. Therefore, DP4 might have been formed due to the hydroxylation of the terminal N-methyl moiety.

MS/MS fragmentation behaviour of DP5 ([M+H]+; m/z 451)

The MS/MS spectrum of DP5, observed under oxidative stress conditions (RT 14.2 min), showed a protonated molecular ion at m/z 451, indicating an elemental composition of C23H31N8O2+. The 16 Da mass difference between RIBO and DP5 was due to the presence of an additional oxygen atom. The abundant product ion formation at m/z 383, 338, and 310 was a manifestation of the excess m/z 16 in DP5 compared to RIBO fragments. The emergence of product ions at m/z 383 (resulting from the loss of the -C5H8 group from m/z 451), 406, and 338 (formed by the exit of the C2H7N group from m/z 451 and 383, respectively) explains that hydroxylation had not occurred on the cyclopentane and N,N-dimethylamine residues. The characteristic neutral loss of piperazine (C4H10N2; m/z 86) from m/z 451 generated a product ion at m/z 365, clarifying that hydrolysis did not occur in the piperazine ring. The daughter ion peak observed at m/z 423, due to the characteristic neutral elimination of carbon monoxide (CO; m/z 28) from the parent ion, provides substantial evidence for the existence of an aromatic hydroxyl group, which facilitates the loss of CO instead of H2O. Furthermore, the product ion exhibited at m/z 150 was formed from m/z 178 by losing CO, which authenticates that the pyridine ring of the 1-(pyridin-3-yl)piperazine residue underwent hydroxylation. The fragmentation pattern was found to be consistent with the proposed structure. The position of the hydroxyl radical (OH)˙ was assigned at the meta (3) carbon to the pyridinyl nitrogen, as the meta-positioned carbon is the least electron-deficient and is more susceptible to the incoming OH˙.

MS/MS fragmentation behaviour of DP6 ([M+H]+; m/z 421)

DP6 was generated under solid photolysis conditions and eluted at a retention time (RT) of 14.9 min. The recorded MS/MS spectrum in positive mode showed a protonated ion at m/z 421, which is a deficit of 14 Da compared to RIBO. The cyclopentane ring was found to be unaltered, as a daughter ion at m/z 353 was observed, corresponding to the m/z 68 (-C5H8) loss from the molecular ion. The appearance of a prominent daughter ion peak at m/z 322 due to the expulsion of methylamine (CH3NH2) from m/z 353 confirms that DP6 was formed by N-demethylation of RIBO at the N,N-dimethylamine site. This was further confirmed by observing the fragmentation pattern of DP6, which reflected that the mass fragmentation pattern of DP6 and RIBO was coherent beyond m/z 322. Therefore, it was inferred that DP6 was formed due to N-demethylation of RIBO during degradation.

MS/MS fragmentation behaviour of DP7 ([M+H]+; m/z 451)

DP7 was formed during the oxidative stress experiment and eluted at an RT of 22.6 min. The mass spectrum of DP7 under positive ESI mode revealed a protonated ion peak at m/z 451 with a probable molecular configuration of C23H31N8O2+. The excess of 16 Da mass compared to RIBO corresponded to the additional oxygen atom in DP7. The appearance of peculiar product ions at m/z 434 and 433, due to the sequential elimination of a hydroxyl radical (−OH)˙ (m/z 17) and a hydrogen atom (-H) from the parent ion, respectively, suggests the possibility of N-oxide formation under oxidative stress conditions. This type of OH˙ loss was also seen in subsequent fragmentations, such as m/z 383→366 and 310→293. The product ions appeared at m/z 383, 338, and 310, indicating that neither the cyclopentane nor the N,N-dimethylamine carboxamide moieties underwent structural changes during degradation. The formation of a product ion at m/z 178 suggests that the N-oxidation might have taken place on the piperazine ring, preferentially at the tertiary nitrogen atom. This could be supported by the literature, which implies that the tertiary nitrogens of aliphatic and alicyclic systems are more favored to form N-oxides rather than aromatic tertiary nitrogen atoms. For further rationalization of N-oxide formation, a confirmatory analysis was performed in APCI-MS mode. From the APCI-MS full scan spectrum, peaks at m/z 435 and 433 were observed at the expense of oxygen and H2O, respectively, from the parent ion. The deoxygenation (loss of O) from the parent ion was due to the thermal decomposition of the N-oxide complex. Furthermore, the appearance of m/z 419 in the spectrum under APCI-MS mode corresponded to the loss of CH3OH (m/z 32) from an intermediate structure formed by Meisenheimer’s rearrangement. Other daughter ion peaks observed at m/z 392, 378, 324, 279, and 251 were consistent with the proposed structure. From the above mass spectrometry data, DP7 was confirmed as the N-oxide of RIBO.

MS/MS fragmentation behaviour of DP8 ([M+H]+; m/z 463)

DP8 was formed under oxidative and photolytic (solid and liquid) conditions (RT 24.3 min). Under positive ESI mode, the MS/MS spectrum of DP8 showed a protonated ion at m/z 463, which is 28 Da higher than RIBO. The proposed elemental formula of DP8 is C24H31N8O2+, indicating the presence of an additional carbon and oxygen atom in the structure. From the mass spectrometric data, the appearance of fragment ion peaks at m/z 395, 350, 322, 217, and 190 showed an additional mass of m/z 28 to each corresponding fragment ion of RIBO, which supported the possibility of formylation. Based on literature, the secondary nitrogen atom is the preferred site for N-formylation. The formation of m/z 190 indicates that formylation might have occurred at the secondary nitrogen atom of the piperazine residue. N-formylation reactions are usually observed in drug products due to excipients that can serve as donors of the formyl group. However, N-formylated DP was observed in the pure drug substance. The use of methanol as a co-solvent and ammonium formate as the mobile phase led to speculation that it might be an artifact in mass spectrometric analysis, as both can act as formyl donors. To confirm whether DP8 was a mass artifact or not, DP8 was synthesized in our in-house facility and analyzed by HPLC. The retention time of the DP8 and the synthesized compound was found to be exactly the same. The fragmentation pattern of both was found to be similar, which was further confirmed by nuclear magnetic resonance (NMR) analysis as described in the following section.

Synthesis and structural elucidation of DP8 by NMR

To confirm the structure of DP8 generated in photolytic and oxidative stress degradation, it was synthesized from RIBO by a simple one-step reaction with ethyl formate. The drug (1 M equivalent) was refluxed with ethyl formate (8.95 M equivalent) in a reaction tube at 60 °C overnight. The completion of the reaction was confirmed by monitoring thin-layer chromatography (dichloromethane:methanol; 95:5, v/v). After completion, ethyl formate was evaporated using a rotary evaporator, and the residue was loaded onto a silica column (230–400 mesh) with dichloromethane, with the polarity gradually increased using methanol. Fractions eluted with 5% methanol were collected, their purity was checked by TLC, and they were dried for further spectroscopic analysis. The structure of DP8 was confirmed in detail using 1H NMR, 13C NMR, and DEPT-135 experiments after the synthesis of the degradation product. A D2O exchange experiment was performed to identify exchangeable protons. The NMR spectrum of the synthesized compound was compared with that of the drug, and the characteristic changes were confirmed using heteronuclear single quantum coherence spectroscopy (HSQC) and heteronuclear multiple bond correlation spectroscopy (HMBC) experiments. All the atoms in the structures of RIBO and DP8 were assigned numerical labels. The values of 1H NMR chemical shifts and 13C NMR chemical shifts have been summarized. From HRMS data, it was observed that synthesized DP8 exhibited [M+H]+ at m/z 463 with the most probable elemental formula C24H31N8O2+, suggesting that both RIBO and DP8 contain an equal number of hydrogen atoms. The 1H NMR spectrum of DP8 showed a peak for the aldehydic proton (-CHO) (H33) at δ 8.11 (s), which was not observed in RIBO. The downfield shift of protons H29 and H31 in DP8 compared to RIBO further indicates the presence of an electronegative group at the N-30 position. The 13C NMR spectrum of DP8 confirmed the presence of an additional carbon for (N-CHO) at δ 160.77, which was absent in RIBO. The D2O exchange study of RIBO confirmed two exchangeable protons at the N20 and N30 positions, whereas in DP8, the proton at the N30 position was replaced with the (N-CHO) group, hence it showed only one exchangeable proton at the N20 position. DEPT-135 analysis of DP8 confirmed the sp2 hybridized (-CH) nature of the (N-CHO) carbon at the N-30 position. The HSQC spectrum of DP8 showed a single bond correlation between the (N-CHO) carbon and the proton (C33) corresponding to it at δ 8.11 (s), which was not observed in the HSQC spectrum of RIBO. The exact site of formylation was further confirmed by performing the HMBC experiment, which exhibited multiple bond correlations between the carbon of the (N-CHO) group at the C33 position and the protons at the C29 and C31 positions of the piperazine ring. This correlation was absent in RIBO. From the above experimental data, it was concluded that DP8 contains a formyl group in its structure. Overall, NMR data indicates the structure of DP8 as 7-cyclopentyl-2-(5-(4-formylpiperazin-1-yl)pyridin-2-ylamino)-N,N-dimethyl-7H-pyrrolo[2,3-d]pyrimidine-6-carboxamide.

MS/MS fragmentation behaviour of DP9 ([M+H]+; m/z 437)

DP9 eluted at an RT of 21.9 min and was observed in samples subjected to oxidative stress conditions. The parent ion peak was observed at m/z 437 in the ESI-MS/MS spectrum of the protonated DP9, with an elemental composition of C22H29N8O2+. It had a 26 Da mass deficit compared to DP8. Upon fragmentation, it exhibited a peculiar loss of formamide (CH3NO; exact mass m/z 45.0215), resulting in the appearance of the most abundant product ion at m/z 392. This peculiar loss was also found in further fragmentation from m/z 369 and 191 to m/z 324 and 146, respectively. The loss of H2O from m/z 437 and 324 generated product ions at m/z 419 and 306, respectively. When the cyclopentene group was lost from m/z 437, a prominent product ion appeared at m/z 369. This product ion, upon further fragmentation, lost formamide (exact mass 45.0215) and N,N-dimethylamine (exact mass 45.0578) sequentially to produce fragment ions of equal nominal mass at m/z 324, which were distinguished by careful observation of the spectrum. This proves that the cyclopentane along with the N,N-dimethyl moieties remained intact during the formation of DP9. One of the daughter ions with an exact mass at m/z 324.1567 further lost C2H7N and CO sequentially to produce product ions at m/z 279 and 251. The daughter ion with an exact mass at m/z 324.1203 lost H2O and CH3NO groups to produce daughter ions at m/z 306 and m/z 279, respectively. Apart from this, the structure-narrative fragment ion peak at m/z 191 confirmed the N-dealkylation of DP8, which forms DP9.

MS/MS fragmentation behaviour of DP10 ([M+H]+; m/z 422)

The ESI-MS/MS spectrum of DP10 showed a protonated parent ion peak at m/z 422, which is 14 m/z units higher than DP1. It was observed in the acidic hydrolyzed sample and eluted at an RT of 26.5 min. The even mass number of DP10 suggested that one nitrogen atom might be less compared to the parent compound RIBO. The product ion formed at m/z 390 after losing methanol (m/z 32) from the molecular ion (m/z 422) might be attributed to the formation of a methyl ester. The esterification of the carboxylic acid group of DP1 by methanol (co-solvent) might have taken place under acidic hydrolytic conditions (pseudo DP). The loss of cyclopentene from the parent ion generated a product ion at m/z 354. Beyond m/z 322, all fragment ions were the same as those of RIBO.

Degradation pathway of RIBO

RIBO was found to be susceptible to degradation under the tested stress conditions, resulting in the formation of DP1 to DP10. The proposed mechanisms of formation of these DPs have been illustrated. Under acidic and basic hydrolysis, RIBO degraded to form DP1 when the N,N-dimethylamine carboxamide moiety underwent heterolytic cleavage to form a carboxylic acid with the elimination of the N,N-dimethylamine moiety. In solution-state photolysis, DP2 was formed due to nucleophilic attack at the 4 and 6 positions of the pyridine ring. DP3 was formed under oxidative stress conditions due to the N-dealkylation of the piperazine ring. Under solid-state photolysis, DP4 was formed due to aliphatic hydroxylation of a methyl group at the N,N-dimethylamine moiety. Hydroxylation on the pyridine ring under oxidative stress conditions led to the formation of DP5. DP6 was formed under solid-state photolysis due to demethylation of the N,N-dimethylamine moiety. The N-oxide of RIBO (DP7) was formed due to nucleophilic attack on the tertiary nitrogen of the piperazine ring during oxidation by H2O2. DP8 was formed in solid-state and liquid-state photolysis due to N-formylation at the secondary nitrogen atom of the piperazine ring. The source of formylation was investigated by performing liquid-state photolysis in acetonitrile:water and DMSO:water solvent systems. Oxidative stress studies utilizing methanol:water and DMSO:water systems were also performed, which also generated N-formylated DP8. The formation of N-formylated DP reveals that N-formylation occurred irrespective of the solvent system. This might be due to the methyl groups in the solvent system, which were converted into formic acid upon oxidation. However, N-formylation in drug substances is rarely observed, and the source of carbonylation is difficult to justify. It was speculated that in solid-state photolytic conditions, formylation occurred due to formic acid, which might have been generated by demethylation followed by oxidation. However, in the case of oxidative stress and solution-state photolysis, the formation of DP8 might be due to the contribution of the solvent system (methanol, acetonitrile, and DMSO) as a formyl group donor to form DP8. DP9 was formed by dealkylation of DP8 under oxidative stress conditions in the presence of H2O2. DP10, a pseudo-DP, was the methyl ester of DP1, formed due to the methanol used as a co-solvent under acidic hydrolytic conditions.

Method validation

The developed HPLC method was validated following the ICH Q2(R1) guideline. Recommended validation parameters were considered to ensure the selectivity, accuracy, and precision of the method.

Selectivity

The method demonstrated the necessary capability to resolve all DPs from the analyte and from each other. The selectivity of the method was confirmed through peak purity assessment of RIBO and individual DPs using a PDA detector and LC–MS analysis. The method was found to be selective for the target analytes.

Sensitivity

LOD and LOQ were established after injecting different concentrations of RIBO in the range of 1–4 µg/mL to obtain a minimum analyte response with a signal-to-noise ratio of 3:1 and 10:1, respectively. The LOD was found to be 1.3 µg/mL (average S/N; 3.61), and the LOQ was 4 µg/mL (average S/N; 10.67).

Linearity

Linearity of the method was achieved within a dynamic range from 40 µg/mL to 160 µg/mL. The standard curve was plotted with concentration (X-axis) against response (Y-axis) for different concentrations. The linear regression equation (y = mx + c) was determined with a correlation coefficient (r2) greater than 0.999 in triplicate analysis (n = 3).

Accuracy

Accuracy was determined in terms of recovery. Samples fortified with excipients were prepared by incorporating different excipients such as MCC, silicon dioxide, polyvinyl pyrrolidone, and hydroxypropyl cellulose. RIBO was spiked into the excipient mixture at levels of 80 %, 100 %, and 120 % of the assay concentration in triplicates (n = 3). The fortified samples were centrifuged, filtered, and analyzed by HPLC. Peak areas obtained from the fortified samples and unfortified samples (without excipients) were compared to calculate the recovery. The recovery of RIBO was found to be in the range of 96.94–100.22 % with RSD values ranging from 0.34 to 0.76 %. The relative error was found to be 3.06, 0.76, and 0.22 at the 80 %, 100 %, and 120 % levels, respectively.

Precision

Repeatability was assessed by performing six injections, and the RSD was found to be 0.8 %. Intra-day and inter-day precision were evaluated by injecting six replicates (n = 6) of a 100 µg/mL concentration on the same day and on different days, respectively. The RSD was found to be 0.85 % for intra-day precision and 0.74 % for inter-day precision. The results indicate that the developed method was precise. The precision data has been summarized.

Computational toxicity assessment of RIBO and its DPs

The parent drug RIBO and its DPs were categorized into six different toxicity classes based on their acute oral toxic dose (LD50) values, in decreasing order of toxicity. Class 1 is considered the most fatal, and class 6 the least toxic. With an LD50 of 2216 mg/kg, RIBO falls under class 5. RIBO did not show any potential for toxicity as it did not generate any toxicological alerts for cytotoxicity, carcinogenicity, mutagenicity, or immunotoxicity. It lacked any binding affinity to various toxicity targets and was inert towards various toxicity pathways. Except for DP2 and DP3 (toxicity class 4), all DPs were found to be in toxicity class 5. Different toxicity endpoints were predicted for the DPs with a confidence score generated in the computational analysis. DP7 was predicted to have hepatotoxicity with a confidence score of 0.59 (59 %). Immunotoxicity was prevalent in the case of DP2, DP3, DP8, DP9, and DP10, with confidence scores of 0.56, 0.75, 0.78, 0.95, and 0.78, respectively. Carcinogenicity was predicted for DP1, DP4, and DP10 with confidence scores of 0.51, 0.59, and 0.50, respectively. Mutagenicity was observed for DP4 and DP7 with confidence scores of 0.61 and 0.57, respectively. From the in silico toxicity prediction, although RIBO did not show any major toxicity, its degradation products (DP3, DP8, DP9, and DP10) showed immunotoxicity with a confidence score above 0.70, which confers a high risk of adverse effects. Although other DPs were found to be less toxic in terms of low confidence scores (<0.7) for different toxicological endpoints, they must be thoroughly monitored through in vivo toxicity studies in preclinical species.

Conclusion

A comprehensive forced degradation study was conducted on RIBO following ICH guidelines. The selective method developed in this study can be valuable in routine quality control analysis, along with the identification of degradation products and process-related impurities. This method will also be helpful for monitoring drug-excipient compatibility studies during the development of new dosage forms. A total of 10 DPs were resolved by the RP-HPLC method and identified using LC-Q-TOF-MS/MS analysis. RIBO was found to be prone to degradation under hydrolytic, oxidative, and photolytic conditions but stable under neutral hydrolysis and thermal stress. DP8 was synthesized, and its structural elucidation was performed using 1H NMR, 13C NMR, 2D NMR, DEPT-135, and D2O exchange experiments. The toxicity potentials of RIBO and its DPs were predicted by the Protox-II web server, which indicated a high immunotoxic probability for DP3, DP8, DP9, and DP10. However, the toxicity reported in this article is only indicative, and actual toxicity should be confirmed through properly designed in vivo experimentation.