Absolute Quantitation of N-Nitrosamines by Coulometric Mass Spectrometry without Using Standards

Abstract

Carcinogenic N-nitrosamines were recently found in the sartan family of drugs and caused many drug recalls. Both of their detection and quantification are therefore important. Methods reported for N-nitrosamine quantitation rely on the use of standards and are just applicable to simple N-nitrosamines. There is an urgent need to quantify N-nitrosamines derived from drugs with a complicated structure that lack standards. To tackle the issue, this study describes a novel absolute quantitation strategy for N-nitrosamines using coulometric mass spectrometry (CMS) without standards. In our approach, N-nitrosamine is first converted into electrochemically active hydrazine via zinc reduction under acidic condition and the resulting hydrazine can then be easily quantified using CMS. To validate our method, six simple N-nitrosamines, N-nitrosodiethylamine (NDEA), N-nitroso-4-phenylpiperidine (NPhPIP), N-nitrosodiphenylamine (NDPhA), N-nitrosodibutylamine (NDBA), N-nitrosodipropylamine (NDPA), and N-nitrosopiperidine (NPIP), were chosen as test samples, and they all were quantified with excellent measurement accuracy (quantitation error ≤ 1.1%). Taking one step further, as a demonstration of the method utility, a drug-like N-nitrosamine, (R)-N-(2-(6-chloro-5-methyl-1’-nitroso-2,3-dihydrospiro[indene-1,4’-piperidin]-3-yl)propan-2-yl)acetamide (VII), was also synthesized and successfully quantified using our method at 15 ppb level in a complex formulation matrix, following solvent extraction, N-nitrosamine isolation, and reductive conversion. Due to the feature of requiring no standards, CMS provides a simple and powerful approach for N-nitrosamine absolute quantitation and has great potential for analysis of other drug impurities or metabolites.

Graphic abstract:

Introduction

N-nitrosamines (R₂N−N=O), containing a nitroso group bonded to an amine, are formed when amines react with nitrosating agents such as nitrous acid.^1–2 Under cytochrome P450 enzymatic catalysis, N-nitrosamine can be converted into diazonium, a DNA alkylating agent.³ Due to the DNA damaging potential, most N-nitrosamines are classified as Group 2A or Group 2B possible carcinogens to humans by International Agency for Research on Cancer (IARC). N-nitrosamines are considered to be part of “cohort of concern” by both European Medicine Agency (EMA) and U.S. Food and Drug Administration (FDA).^4–5 N-nitrosamines are frequently detected during food processing and water treatment.⁶ Recently unexpected finding of N-nitrosamine impurities in drugs such as angiotensin II receptor blockers, ranitidine, nizatidine, and metformin^7–8 at unacceptable levels led to costly drug recalls.⁹ Health authorities responded to these findings by putting guidance for tight control of N-nitrosamines. FDA recommends manufacturers to take appropriate measures to control N-nitrosamine impurities to acceptable levels; for example, the acceptable intake (AI) limits of N-nitrosodimethylamine (NDMA) and N-nitroso-N-methyl-4-aminobutanoic acid (NMBA) are 96 ng/day, and the AI limits for N-nitrosodiethylamine (NDEA), N-nitrosomethylphenylamine (NMPA), N-nitrosoisopropylethylamine (NIPEA), and N-nitrosodiisopropylamine (NDIPA) should not be more than 26.5 ng/day.⁷ Therefore, identification and quantitation of N-nitrosamine impurities have become an urgent task in the pharmaceutical industry.

Currently, most regulations and analytical methods focus on the common and simple N-nitrosamines and rely on the use of standard nitrosamines for quantitation. For instance, gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS)^10–13 are two main approaches for quantifying N-nitrosamines typically generated during the manufacturing process. Hyphenation of GC with other detectors was also performed for the N-nitrosamine determination in water, such as using GC-FID (flame ionization detector) and GC-NPD (nitrogen-phosphorous detector).^14–18 Determination of N-nitrosamines in food products was reported using LC-DAD (diode array detector)¹⁹ and fluorimetric detector.²⁰ Starting materials, recovered solvents, catalysts, and reagents were recognized as the root sources of N-nitrosamines. Those traditional methods can satisfy the need for quantitation, as they target N-nitrosamine whose standards are commercially available or affordable. However, analysis becomes a considerable challenge due to the lack of standards for novel N-nitrosamines, which can potentially be generated from active pharmaceutical ingredients (API) or API impurities possessing secondary or tertiary amine functional groups. Synthesis of API-derived N-nitrosamine reference standards or stable isotope-labeled internal standards could be costly and time-consuming, for which additional safety measurement needs to be in place during synthesis/purification given the potential mutagenicity nature of N-nitrosamines. For instance, the synthesis of 1 g of drug-simulant N-nitrosamine VII used in this paper costed over $15,000 and 1-month of lead time. Without standards, conventional methods are inapplicable for absolute N-nitrosamine quantitation. Although some non-chromatographic methods were developed, including electrochemistry-based sensor^21–23 and UV-photolysis coupling chemiluminescence detection to address this issue,²⁴ those methods are generally less selective and used to measure the total amount of N-nitrosamines in a sample. Therefore, a selective and sensitive method capable of absolute quantitation of N-nitrosamine without using standards would be ideal for tackling such a challenge.

In this study, to eliminate the need for using standards or calibration curves for quantitation, we present a new strategy of using electrochemistry-assisted mass spectrometry, namely coulometric mass spectrometry (CMS). We recently have applied our CMS method^25–30for direct absolute quantitation of organic compounds, peptides, and proteins. As illustrated in Scheme 1, our quantitation strategy involves the chemical reduction of N-nitrosamine to hydrazine. Then the electrochemically active hydrazine is subject to LC separation and oxidation in an electrochemical flow cell and followed with online MS detection. The oxidation results in an electrical current that can be integrated over time to obtain the total electric charge Q in the electrochemical oxidation reaction. According to Faraday’s Law, Q is proportional to the quantity of the oxidized hydrazine: Q=nzF, where n denotes the moles of the oxidized hydrazine, z is the number of electrons transferred per molecule during the oxidation reaction, and F is the Faraday constant (9.65×10⁴ C/mol). Thus, n=Q/(zF). On the other hand, from the acquired MS spectra before and after EC oxidation, the oxidation conversion yields Δi can be determined by measuring the relative change of the hydrazine peak area in the extracted ion chromatogram (EIC) upon oxidation. Thus, the total amount of the hydrazine can be calculated as n/Δi= Q/(zFΔi)). In this case, hydrazine undergoes electrochemical oxidation reaction at a low oxidation potential (0.3 V vs. Ag/AgCl), presumably by losing one electron (z=1) to form hydrazine radical cation, which subsequently loses one hydrogen to form iminium cation (Scheme 1).³¹ Since the reduction yield can also be determined by comparison the MS ion signal of N-nitrosamine before and after zinc reduction; therefore nitrosamine quantity can also be figured out based on the moles of hydrazine. Herein, to validate this CMS approach for N-nitrosamine absolute quantitation, several simple N-nitrosamines were successfully quantified with very good measurement accuracy (≤1.1%). This new quantitation method is also highly sensitive. For instance, we have obtained accurate quantitation for N,N-diphenylhydrazine at the level of ca. 0.9 pmol (0.16 ng) using our CMS method. Furthermore, quantitation of drug-simulant N-nitrosamine VII (structure shown in Scheme 2) in the presence of drug matrix at low level (150 ng VII in 10 g drug matrix; i.e., 15 ppb) using our CMS method was also successfully demonstrated.

Scheme 1.

Workflow for absolute quantitation of N-nitrosamines by CMS

Scheme 2.

Synthesis route for nitrosamine VII

Experimental section

Materials:

N-nitrosodiethylamine (NDEA), N-nitrosodibutylamine (NDBA), N-nitrosodipropylamine (NDPA), N-nitrosopiperidine (NPIP), N,N-diphenylhydrazine hydrochloride, ammonium acetate and 0.2 μm PTFE membrane Millex-LG filter were obtained from Sigma-Aldrich (St. Louis, MO, USA). N-nitroso-4-phenylpiperidine (NPhPIP) and N-nitrosodiphenylamine (NDPhA) were bought from Toronto Research Chemicals (Toronto, ON, Canada). (R)-N-(2-(6-chloro-5-methyl-1’-nitroso-2,3-dihydrospiro[indene-1,4’-piperidin]-3-yl)propan-2-yl)acetamide (VII as shown in Scheme 2) was synthesized via deprotection and nitrosation of precursor VII”. The precursor VII” was made as previously reported.³² Lithium triflate, HPLC-grade acetic acid, acetonitrile, methanol, and isopropanol were from Fisher Scientific (Waltham, MA, USA). Zinc powder was purchased from Spectrum Chemicals (New Brunswick, NJ, USA). A Millipore Direct-Q5 purification system (Burlington, MA, USA) was used to obtain purified water.

N-nitrosamine reduction by Zn:

Following a modified and reported procedure,³³ zinc powder and N-nitrosamine were mixed in methanol/acetic acid mixture (v/v 8:1), and stirred for 2 h at room temperature for reduction, under nitrogen protection. After reaction, the zinc powder was removed by filtration or centrifugation.

Extraction and isolation of N-nitrosamine VII in drug matrix:

To further validate our method for quantitation of N-nitrosamine in drug, 150 ng of N-nitrosamine VII was mixed with 10 g of a drug matrix to mimic a drug sample. The drug matrix contained corn starch, D&C red #27 aluminum lake, dicalcium phosphate, magnesium stearate microcrystalline cellulose, polyethylene glycol, polyvinyl alcohol, silicon dioxide, stearic acid, talc, and titanium dioxide. A 50 mL of isopropanol (IPA) was added to the N-nitrosamine VII sample with the drug matrix for extraction. The mixture was stirred for 15 min at room temperature to homogenize the solution, and then stirring was stopped for 5 min to allow insoluble excipients to settle down. The top liquid layer was transferred onto a small silicone pad for filtration with additional 50 mL IPA to rinse the pad to reduce sample loss, and the eluent was collected. The process was repeated 2 more times, then all eluents were pooled together in a glass flask. The merged eluent solution was vacuum dried by rotovap and reconstituted in 160 μL of MeCN/H₂O/FA (5:95:0.1 v/v/v) for LC isolation of N-nitrosamine VII. Isolation of VII from the extract was conducted using HPLC (Shimadzu, Columbia, MD): Waters XBridge Peptide BEH C18 Column, injection volume 80 μL, flow rate 1 mL/min, mobile phase A: 0.1% formic acid in H₂O (v/v), mobile phase B: 0.1% formic acid in MeCN (v/v), gradient program: 0–15 min, isocratic flow at 5% mobile phase B, then increasing mobile phase B to 95% in 30 min, then changed back to 5% in 5 min. The extracted and purified VII was vacuum dried and re-constituted in 180 μL of methanol/acetic acid (v/v 8:1) for Zinc reduction and subsequent CMS quantitation.

CMS Apparatus:

As shown in Scheme S1 (Supporting Information), LC/EC/MS setup was used for CMS quantitation. In our experiment, a UPLC (Waters, Milford, MA) was used to inject the reduced N-nitrosamines (i.e., hydrazines) into a thin-layer electrochemical flow cell consisting of a dual gold disc working electrode (WE, 3 mm I.D. each), an Ag/AgCl (3 M NaCl) reference electrode, and a stainless steel block as the counter electrode (BASi, West Lafayette, IN, USA; cell dead volume: ca. 1 μL) for oxidation, followed with online MS detection using a high-resolution Orbitrap Q Exactive mass spectrometer equipped with heated electrospray ionization (HESI) source (Thermo Scientific, San Jose, CA, USA). Ether a C18 column (Waters BEH C18, 2.1 mm × 100 mm, 1.7 μm) or a HILIC column (Waters BEH HILIC, 2.1 mm × 50 mm, 1.7 μm) was used for the separation, depending on polarities of hydrazines. WE was polished with alumina slurry before and after use. A potential of +0.30 V (vs. Ag/AgCl) was applied to the WE to induce the oxidation of hydrazines. During electrochemical oxidation, an electric current was generated and recorded by a potentiostat. Origin 2017 was used to calculate the total electric charge of Q by integrating the electric current peak with time. Before and after oxidation, ion intensities of the hydrazines were recorded with MS. The flow rate of sheath gas and the applied ionization voltage were 10 L/h and + 4 kV, respectively. The ion transfer inlet capillary temperature was kept at 250°C. Mass spectra were acquired and processed by Xcalibur (Thermo Fisher 3.0.63).

For UPLC separation using a C18 column, mobile phase A (deionized water with 0.1% formic acid) and mobile phase B (acetonitrile with 0.1% formic acid) were used with a flow rate of 200 μL/min. In a gradient elution mode, the mobile phase B was increased from 5% to 32% in 6 min and to 95% in 2 min, kept at 95% for 2 min before returning to 5%. The injection volume was 3 μL or 5 μL per injection. The isocratic mode was used for HILIC column separation, in which 5% mobile phase A (100 mM ammonium acetate in deionized water, pH 5.0), and 95% mobile phase B (acetonitrile) were pumped at a flow rate of 200 μL/min for 15 min with a sample injection volume of 3 μL.

Results and Discussion

The requirement of our CMS method is that the analyte needs to be electrochemically active. Our preliminary CV scan (Figure S1, Supporting Information) showed an oxidation peak for N-nitrosamine at 2.2 V (vs. Ag/AgCl, Figure S1b), while no N-nitrosamine reduction peak was observed 1 mM NDEA in anhydrous MeCN containing 0.1 M LiOTf using the boron-doped magic diamond (BDD) electrode. Such a high oxidation potential would only be possible in anhydrous MeCN solvent, not compatible with our LC/EC/MS runs where aqueous solvent was used. This is in line with the literature report that direct electrochemical oxidation of N-nitrosamines requires a relatively high oxidation potentials^34–35 or a modified electrode.³⁶ Thus, we employed a chemical reduction strategy to convert N-nitrosamine into hydrazine first, a common practice in the organic chemistry field, then applied CMS to quantify the resulting hydrazine, as hydrazine can be easily oxidized electrochemically in aqueous solvent. The hydrazine measured by CMS would reflect the amount of its precursor N-nitrosamine.

In our experiment, some reductants were screened, such as sodium dithionite (Na₂S₂O₄), dithiothreitol (DTT), tris(2-carboxyethyl)phosphine (TCEP), hydroxylamine, ascorbic acid, and zinc powder. Reduction product, hydrazine, was afforded only under acid condition using zinc dust as a reductant (Table S1) with nearly quantitative yield. Based on previous hydrazine electrochemical characterization,³¹ hydrazine was shown to have a high electrochemical activity (losing one electron to form hydrazine radical cation at oxidation potential 0.3 V vs. Ag/AgCl). So, using CMS to quantify the hydrazine product from N-nitrosamine reduction can be a rational strategy to determine the N-nitrosamine amount.

In our CMS experiment, N-nitrosodiethylamine I, was first examined. 3 μL of I (0.0279 mmol, 2.84 mg) and 36.6 mg zinc dust (0.560 mmol) were mixed in 450 μL methanol/acetic acid mixture (v/v 8:1), stirred for 2 h at room temperature under nitrogen protection. After reaction, the solution was filtered through 0.2 μm filter, vacuum dried, and reconstituted in 1 mL MeCN, followed by a 1000X dilution for CMS analysis (final concentration: 27.9 μM). Nanoelectrospray ionization mass spectrometry (nanoESI-MS) was used to evaluate the reduction yield by measuring N-nitrosodiethylamine ion signal change after reduction (Figure 1a). The chemical reduction yield was 99.9% by MS measurement, based on comparing the hydrazine product ion [1+H]⁺ intensity (1.33E9, measured value m/z 89.10752, theoretical value m/z 89.10732, mass error 2.24 ppm) to the total ion intensities of [1+H]⁺ and unreduced N-nitrosamine ion [I+H]⁺ (1.99E6, negligible compared to [1+H]⁺ intensity; measured value m/z 103.08688, theoretical value m/z 103.08659, mass error 2.81 ppm). The reduction yield was further confirmed by NMR analysis which showed a yield of 99.8% (Figure 1b) using dibromomethane as the internal standard. MS yield was thus adopted due to its simplicity in ion intensity measurements.

Figure 1.

(a) NanoESI-MS spectrum and (b) NMR spectrum of N-nitrosodiethylamine after chemical reduction

The resulting hydrazine 1 from reduction of N-nitrosodiethylamine (3 μL, 27.9 μM, theoretical amount: 83.7 pmol) was subjected for CMS analysis, which involved LC separation, electrochemical oxidation, and online MS detection sequentially. Before electrolysis, the protonated 1 was detected at m/z 89 (Figure 2a), and a small oxidation peak at m/z 87 was observed, probably due to the in-source oxidation of 1.^37–38. Previous literature³¹ reported that hydrazine loses one electron to form radical cation at a low oxidation potential (+0.3 V vs. Ag/AgCl) and can further loses one more electron at higher oxidation potential to form a dication (+0.75 V vs. Ag/AgCl) using gold electrode. Our CMS experiment adopted a low oxidation potential of +0.3 V vs. Ag/AgCl since it would avoid the side oxidation reactions such as water hydrolysis, which would contribute to a better current baseline. As shown in Figure 2b, the peak intensity at m/z 87 increased and the ion intensity of m/z 89 decreased after electro-oxidation. Figure 2c and 2d show the EIC of m/z 89 with the applied potential of 0 V and +0.3 V (vs. Ag/AgCl), respectively. The integrated area for the peak shown in Figure 2d was smaller by 45.8% compared to that of the peak shown in Figure 2c, indicating that the oxidation yield for 1 was 45.8% (Table S2, SI). On the other hand, the 1 oxidation current peak was detected, as shown in Figure 2f (Figure 2e shows the background current diagram for the blank solvent sample under the same + 0.3 V potential as a contrast). Based on the integration of the current peak area, the amount of the oxidized 1 was calculated to be 38.1 pmol. Therefore, our measured amount of 1 was 38.1 pmol/45.8%=83.0 pmol. Considering 99.9% reduction yield, the N-nitrosamine amount was determined as 83.0 pmol. In a triplicate measurement, the averaged amount of the N-nitrosamine was 82.9 pmol (see detailed data in Table S2, SI), which has a small quantity discrepancy of –0.9% compared with theoretical value of 83.7 pmol (the injection amount as mentioned above).

Figure 2.

ESI-MS spectra of hydrazine 1 (from N-nitrosodiethylamine reduction) when the applied potential was (a) 0 V and (b) +0.3 V (vs Ag/AgCl). EIC of [1+H]⁺ at m/z 89 when the applied potential was (c) 0 V and (d) +0.3 V (vs Ag/AgCl). Electric current responses were shown due to the oxidation of (e) a blank solvent and (f) 1.

Another simple model compound N-nitrosamine II (N-nitroso-4-phenylpiperidine) was also examined. After zinc reduction, hydrazine 2 was produced. As shown in Figure S2, the intensity of the hydrazine product ion [2+H]⁺ was 1.31E8 and the unreduced [II+H]⁺ was 3.04E5. This nanoESI-MS measurement showed that the yield of chemical reduction was about 99.8%, again, nearly quantitative. After reduction, the solution was filtered, dried, and reconstituted in 1 mL MeCN, followed by 1000X dilution for CMS analysis (final concentration: 28.0 μM, injection volume of 3 μL, theoretical injection amount: 84.0 pmol). Before electrolysis, the protonated hydrazine 2 was detected at m/z 178 (Figure 3a). A small oxidation peak at m/z 176 was observed, probably due to the in-source oxidation of 2. After electrolysis in the electrochemical cell (Figure 3b), the peak intensity at m/z 176 increased significantly. Figure 3c and 3d showed the EIC of m/z 178 with the applied potential of 0 V and + 0.3 V, respectively. The integrated area for the peak shown in Figure 3d was smaller by 32.2% compared to that of the peak shown in Figure 3c, indicating that the oxidation yield was 32.2% (Table S3). On the other hand, the oxidation current peak was detected, as shown in Figure 3f (Figure 3e shows the background current diagram for the blank solvent sample under the same + 0.3 V potential as a contrast). Based on the integration of the current peak area, the amount of the oxidized hydrazine 2 on average was calculated to be 27.0 pmol. Therefore, our measured amount of hydrazine was 27.0 pmol/32.2%=83.9 pmol. Considering the 99.8% reduction yield, the measured N-nitroso-4-phenylpiperidine (N-nitrosamine II) was 83.9 pmol. Average result from a triplicate measurement was 83.8 pmol, which is close to the theoretical injection amount of 84.0 pmol (measurement error: −0.3%, Table S3).

Figure 3.

ESI-MS spectra of hydrazine 2 (from reduction of N-nitroso-4-phenylpiperidine) when the applied potential was (a) 0 V and (b) +0.3 V. EIC of hydrazine ion at m/z 178 was recorded when the applied potential was (c) 0 V and (d) +0.3 V (vs Ag/AgCl). Electric current responses were shown due to the oxidation of (e) a blank solvent and (f) 2.

The validity of the CMS N-nitrosamines absolute quantitation approach was further conducted with 4 more N-nitrosamines. As summarized in Table 1, various N-nitrosamines at pmol range were successfully quantified using this new CMS quantitation method, with quantitation error no more than 1.1% (Figures S3–10, Tables S4–7).

Table 1.

List of N-nitrosamines quantified by CMS

In addition, we examined the quantitation sensitivity of our CMS method and 3 μL of 0.291 μM N,N-diphenylhydrazine 3 (0.873 pmol) was injected for CMS analysis. As shown in Figure 4a and 4b, hydrazine electrochemical oxidation product at m/z 183 peak intensity increased after turning on the electrochemical cell. In contrast, EIC peak area of m/z 185 was decreased by 17.8% upon electrooxidation (Figure 4c and 4d). After integrating the EC current area, we were able to quantify hydrazine as low as 0.873 pmol (~0.16 ng, quantitation error: 1.3%, Table S8). In addition, CMS can provide mass information for N-nitrosamine for their identification with high mass accuracy, as exemplified in Figures 1–3 above. Interesting, in Figure 4a, hydrazine radical cation at m/z 184 (theoretical m/z 184.09950, measured m/z 184.09966, mass error: 0.87 ppm) was also detected, as a result of inherent electrochemical oxidation during electrospray ionization, which is in agreement with the previously reported one electron transfer mechanism for hydrazine oxidation.³¹ CID MS/MS analysis showed that fragment ion of m/z 168 was formed by loss of amine radical NH₂ from m/z 184, consistent with the structure assignment of m/z 184 (Figure S11).

Figure 4.

ESI-MS spectra of 3 μL of 0.291 μM N,N-diphenylhydrazine 3 when the applied potential was (a) 0 V and (b) +0.3 V. EIC of hydrazine ion at m/z 185 was recorded when the applied potential was (c) 0 V and (d) +0.3 V (vs Ag/AgCl). Electric current responses were shown due to the oxidation of (e) a blank solvent and (f) 0.291 μM 3.

After the success of quantifying simple N-nitrosamines based on Zn reduction strategy followed with CMS quantitation of the resulting hydrazines, we explored the possibility of quantifying a N-nitrosamine stemming from drug-simulant substrate (N-nitrosamine VII, Scheme 2) that has a more complicated structure than those of simple nitrosamines examined above. Till now, quantitative analysis of complicated N-nitrosamine remains unexplored, mainly because of lack of availability of N-nitrosamine standards. However, it has become an urgent issue in the pharmaceutical field. This issue emphasizes the advantage of our approach, where no standard is needed for quantitation.

As of proof-of-concept test, N-nitrosamine VII was synthesized as shown in Scheme 2, following a reported protocol for deprotection and nitrosation of a precursor compound VII”.³² We estimated a cost of about $15,000 and 1-month time for custom-synthesis, purification and full characterization of 1 g of VII. The experimental condition for reducing N-nitrosamine VII was first optimized to prevent the formation of over-reduction amine side product. In a trail, VII (0.91 mg, 0.0025 mmol) was added with 20 equiv. Zn (3.27 mg, 0.05 mmol) in MeOH/AcOH (8:1 v/v) and stirred at room temperature. As shown in Figure S12a, no corresponding hydrazine product 7 (structure shown in Scheme 3) was observed in the MS spectrum before chemical reduction. Although nearly all of VII (0.91 mg, 0.0025 mmol) was consumed, a reduction side product, the denitrosylated amine, was also detected at m/z 335 ([C₁₉H₂₇ClN₂O+H]⁺, Figure S12b). The two reduction product structures were confirmed using MS/MS (Figure S13). After adjusting the reduction condition by reducing Zn amount from 20 equiv. to 10 equiv. and reaction time from 2h to 0.5h, no reduction side product was detected (Figure S12c) and VII was exclusively converted into hydrazine 7.

Scheme 3.

Proposed mechanism for electrochemical oxidation of hydrazine 7

Based on LC/MS analysis of the N-nitrosamine VII sample before and after reduction, the reduction yield was determined to be 47.2% (Figure S14, Supporting Information). CMS was then conducted and 3 μL of reduced N-nitrosamine solution containing hydrazine 7 (original amount of N-nitrosamine VII: 75 pmol) was injected for LC separation; the result was shown in Figure S15 (Supporting Information). The ion intensity of 7 oxidation product (the proposed oxidation mechanism shown in Scheme 3 shows the proposed oxidation mechanism for hydrazine 7 involving one electron transfer and one hydrogen loss to produce iminium ion at m/z 348. The oxidation yield for 7 was measured to be 6.5% based on signal change of the protonated hydrazine 7 upon oxidation (Figure 15c and d). Furthermore, according to the measured oxidation current (Figure S15f), the amount of the oxidized 7 was calculated to be 34.4 pmol (averaged value from a duplicate measurement). Considering the reduction yield (47.2%), the mole of the N-nitrosamine was determined to be 72.9 pmol. The quantitation error was –2.9%, compared with the theoretical injection amount (75.0 pmol, Table S9).

Our effort was then focused on quantifying trace amounts of VII. In our experiment, 150 ng of VII was used and first reduced using 0.25 mg zinc under acid condition (MeOH/AcOH=8:1 v/v) at room temperature for 0.5 h. In this case, zinc was in much excess (zinc: N-nitrosamine > 9000:1), as it was difficult to accurately weigh less amount of zinc powder due to limitation of our balance. As a result, a small amount of side reduction product, the denitrosylated amine, was formed and detected (4.1% peak area compared with the ion of the hydrazine product 7, Figure S16). The majority (95.6%) of the reduction product was hydrazine 7. The reduction yield was 41.6%, as measured by EIC signal change of N-nitrosamine VII (m/z 364) upon reduction (Figure S17). Although the retention time of side product and hydrazine 7 were similar, no EC oxidation current was arising from the amine side product under 0.3 V oxidation potential (vs. Ag/AgCl), as shown in a separate electrochemical oxidation experiment (Figure S18). It showed that hydrazine 7 can be selectively oxidized in the electrochemical cell in the presence of the amine side product. The reduced solution was filtered, vacuumed dried, and reconstituted in 160 μL solvent and 5 μL of the resulting solution was injected into the CMS system. As shown in Figure S19 and Table S10, the hydrazine product 7 was quantified by CMS to be 5.1 pmol. For the amine side product, considering the similarity of proton affinities between hydrazine and the corresponding amine (e.g., dimethylhydrazine and dimethylamine have proton affinities (PA) of 927.1 kJ/mol and 929.5 kJ/mol, respectively³⁹) and the minor amount of hydrazine product, it would be reasonable to estimate the amount of the amine side product based on its ion intensity relative to the 7 ion intensity. Such an estimation would not lead to a significant quantitation error because of their similar PA values and the minor amount of amine product (4.1% peak area compared with the reduction product 7, Figure S15). In consideration of the amine side product ratio of 4.1% and reduction yield of 41.6%, therefore, the measured amount of N-nitrosamine VII would be 5.1 pmol/(1–4.1%)/41.6%=12.8 pmol. Further, considering the dilution factor of 32, the measured N-nitrosamine would be 409 pmol which is close to the theoretical amount of 412 pmol (150 ng N-nitrosamine VII), with a measurement error of −0.7% (Table S10). Note that, in this case, CMS only used 5.1 pmol (about 1.8 ng) of hydrazine for quantitation in each run.

Finally, CMS was applied for quantitative analysis of N-nitrosamine VII in presence of drug matrix containing various excipients. In our experiment, to mimic N-nitrosamine in drug product, 150 ng of N-nitrosamine VII was doped with 10 g drug matrix (15 ppb) containing corn starch, D&C red #27 aluminum lake, dicalcium phosphate, magnesium stearate microcrystalline cellulose, polyethylene glycol, polyvinyl alcohol, silicon dioxide, stearic acid, talc, and titanium dioxide. Preliminary test showed that much lower quantitation result was obtained by directly reduction of the N-nitrosamine in the presence of excipients matrix, probably because of occurrence of condensation reactions between the reduced N-nitrosamine (i.e., hydrazine) and excipient ingredients (e.g., carbonyl groups of sugars in matrix). To overcome this challenge, N-nitrosamine VII was first extracted from the drug matrix via solvent extraction (using 2-propanol) and isolated by LC using the reverse-phase chromatography (workflow is shown in Figure S20), followed by zinc reduction to afford hydrazine 7. The extraction yield, reduction yield and amine side product percentage were determined using LC/MS analysis based on EIC peak areas (Figures S21, S22, and S23). Finally, as mentioned before, CMS was employed for the hydrazine product quantitation. As shown in Figure 5a and 5b, the oxidation product ion at m/z 348 had increased intensity when +0.3 V potential (vs. Ag/AgCl) was applied to WE. The protonated hydrazine 7 ion of m/z 350 shown in Figure 5d was smaller by 16.9% than that of the peak shown in Figure 5c, suggesting the oxidation yield to be 16.9% (Table S11). On the other hand, the 7 oxidation current peak was detected, as shown in Figure 5f (Figure 5e shows the background current diagram for the blank solvent sample under the same + 0.3 V potential as a contrast). Based on the integration of the current peak area, the amount of the oxidized 7 was calculated to be 10.0 pmol (ca. 3.5 ng). Considering the extraction and isolation yield (43.9%), the side product ratio (4.4%), the reduction yield (58.1%), and the dilution factor 10, the CMS measured amount of VII was 10 pmol/43.9%/(1–4.4%)/58.1%*10= 410 pmol on the average from a triplicate measurement. The measurement error was –1.1% compared with the theoretical amount of 412 pmol (i.e., 150 ng N-nitrosamine VII, Table S11).

Figure 5.

ESI-MS spectra of hydrazine 7 (from reduction of N-nitrosamine VII in the test sample with drug matrix) when the applied potential was (a) 0 V and (b) +0.3 V. EIC of 7 at m/z 350 was recorded when the applied potential was (c) 0 V and (d) +0.3 V (vs Ag/AgCl). Electric current responses were shown due to the oxidation of (e) a blank solvent and (f) 7.

The current zinc-based chemical reduction strategy facilitates subsequent CMS analysis. The weakness is that zinc to N-nitrosamine ratio is hard to control. When the N-nitrosamine is at trace level, it is challenging to weigh a trace amount of zinc solid. As a result, side reduction amine product can be produced due to the over-excess of Zn. However, the side product can be controlled to an insignificant amount (< 5%). Benefiting from the similar proton affinity between side product and hydrazine, the chemical reduction conversion ratio from N-nitrosamine to hydrazine can be estimated. In other words, N-nitrosamine quantitation based on zinc reduction for a trace amount of N-nitrosamine is feasible.

In our CMS experiment, the oxidation yield of hydrazine resulting from N-nitrosamine reduction varied in the cases shown above. Based on our observation, it was concentration dependent, as the surface area of gold electrode (i.d., 3 mm) used for hydrazine oxidation was limited. For example, when the injected amount of hydrazine was high, the oxidation yield was low, and vice versa. Other factors affecting the oxidation yield include the exact structure of hydrazine substrate, the mobile phase flow rate and the activity of the electrode after each polishing.

In this CMS method, benefiting from the low oxidation potential of hydrazine, we can apply a relatively low oxidation potential (+0.3 V vs. Ag/AgCl) to trigger the hydrazine oxidation. Such a low oxidation potential would trigger selective hydrazine oxidation without causing oxidation of other functional groups such as tetrazole, one common moiety existing in the sartan family of drug molecules. For instance, no oxidation was seen for tetrazole-containing compound when the applied oxidation potential was lower than 1.0 V (vs. Ag/AgCl).⁴⁰

In summary, our results show the feasibility of using CMS to quantify N-nitrosamines. The major strength of this approach is that no standards or calibration curves are needed. Therefore, it overcomes a grand challenge for synthesizing standards. This method could significantly impact pharmaceutical quality control, food safety analysis, and water treatment applications in the future. Although zinc-based chemical reduction shows some limitations, especially for controlling the N-nitrosamine and zinc reaction ratio, other reductants with better selectivity could overcome this issue. Besides, modified electrodes could be used for directly reducing or oxidizing N-nitrosamines, potentially simplifying the quantitation process. These future studies are underway.