Risk assessment for nitrosated pharmaceuticals: A future perspective in drug development

Since June 2018, thousands of drug products from around the world had to be recalled due to the unexpected presence of nitrosamines (NAs). Starting with the pharmaceutical group of sartans, antidiabetic drugs, antihistamines, and antibiotics also became the subject of investigation. The occurrence of NAs has shown that pharmaceutical companies and regulatory agencies did not focus on these substances in the past during drug development. In this study, we incorporated a nitrosation assay procedure into high‐resolution supercritical fluid chromatography (SFC)–mass spectrometry screening to test the potential of direct nitrosation of active pharmaceutical ingredients (APIs). The forced degradation study was performed with a four‐fold molar excess of sodium nitrite, relative to the drug substance, at pH 3–4 for 4 h at 37°C. Chromatographic separation was performed on a porous graphitic carbon column by SFC. The mass analysis then focused on direct N‐nitrosation or N‐nitroso compounds (NOCs) formed after dealkylation. Substances (n = 67) from various pharmaceutical classes were evaluated and 49.3% of them formed NOCs, of which 21.2% have not yet been reported in the literature. In addition, for two APIs, which are known to form an unidentified NOC, the structure could be identified. A few substances also showed multiple NOCs and even N,N’‐dinitroso‐species. As NAs are carcinogens, they have to be eliminated or at least limited to prevent cancer in patients, who rely on these drugs. This study contributes a procedure that can be implemented in preapproval drug development and postapproval risk assessment to prevent unexpected findings in the future.

other active pharmaceutical ingredients (APIs) were reported to be contaminated with NAs. Ten months after the first report of a detected NA, about 22% of all tested API batches and 18% of all drug product batches containing valsartan, losartan, and irbesartan showed levels of NDMA, N-nitrosodiethylamine (NDEA), and/or N-nitroso-N-methyl-4-aminobutyric acid (NMBA) above the acceptable limits in the European Union ( Figure 1). [1] To date, more than 1800 drug product batches (sartans, antidiabetic drugs, antihistamines, and antibiotics) have been recalled in the United States ( Figure 2) due to NA detection. [2] The U.S. Food and Drug Administration (FDA) and the EMA additionally initiated a total withdrawal of all ranitidine [3,4] and varenicline drug products [5,6] from the market, as the occurrence of NAs could not be eliminated.
This triggered a worldwide scientific evaluation of nitrosamine contamination in all drug substances and medicinal products. [7][8][9] Marketing authorization holders have been requested to review all their supply chains and medicinal products for the possibility of NA formation and their root cause. A major part of this referral is the consensus that analytical measurements are necessary to detect and control N-nitrosamines and to mitigate their occurrence in medicinal products. Lists of nine potential known NAs have been established with interim limits for acceptable daily intake on a toxicological basis, [3,8] as NAs are mutagenic and carcinogenic substances from a "cohort of concern" as defined by the ICH guideline M7. [10] This unexpected finding of N-nitrosovarenicline shows that nitrosation of APIs or other drug product ingredients is possible and has not been addressed yet. [5,6] For the scientific investigation, the EMA published a multilevel approach with a strict time frame (Figure 3) for the implementation of a regulatory referral process. This approach includes a request for evaluation (step 1), additional confirmatory testing (step 2), and follow-up risk mitigation (step 3), if NAs are detected or likely to occur. In this EMA publication, a rather new scenario has been introduced within step 2, which has not been discussed yet: "one or F I G U R E 1 Number of out-of-specification (OOS) results for analyzed active pharmaceutical ingredients (APIs) and finished dosage forms (FDFs) as of April 15, 2019, according to the European Medicines Agency (EMA). [1] From 758 tested APIs, 165 were positive (21.8%), and 320 of 1802 tested FDFs (17.8%) were above the acceptable daily dose of the specific nitrosamine (NA). For N-nitrosodimethylamine (NDMA), 39.5% of the APIs (70 of 177) and 21.2% of the FDFs (253 of 1193) were contaminated; for N-nitrosodiethylamine (NDEA), 16.1% of the APIs (82 of 509) and 11.0% of the FDFs (67 of 609) were contaminated; for N-nitroso-N-methyl-4-aminobutyric acid (NMBA), 18.1% of the losartan APIs were contaminated (13 of 72)-note that other sartans and sartan drug products were not analyzed at the time of data acquisition [1] F I G U R E 2 Cumulative cases of recalled or withdrawn drug products after Food and Drug Administration (FDA) safety alerts since 2018 [2,3] -note that rifampicin and rifapentine were also subject to safety alerts, but not recalled, to prevent drug shortage. No specified nitrosamines were reported after January 4, 2021. HCT, hydrochlorothiazide; NDEA, N-nitrosodiethylamine; NDMA, N-nitrosodimethylamine; NMBA, N-nitroso-N-methyl-4-aminobutyric acid more new N-nitrosamines have been detected in a medicinal product which has not yet been assessed." [11] This was the first time that more than the nine [12] common nitrosamines were directly addressed, without further details being provided.
The main source for NAs previously has been the use or carryover of sodium nitrite (NaNO 2 ) within API synthesis and drug product manufacturing. Additionally, the use of recycled and/or contaminated raw and starting materials, carryover or cross-contamination of NA intermediates, degradation processes generating, for example, nitrosyl or oxime functionalities, and the use of certain packaging materials (e.g., nitrocellulose lidding foil) or any other nitrosating agents in the presence of secondary or tertiary amines were discussed as a source of NAs. [12][13][14][15] Recently, mainly organic solvents and short-chain aliphatic amines (e.g., dimethylamine, diethylamine, or N-methyl-2-pyrrolidone) have been considered to be the precursors of nitrosamines, but as the varenicline case shows, nitrosation should also be considered for APIs.
In 2003, Adachi et al. [16] reported a case of 12 liver injuries with severe health consequences (one patient required liver transplantation and another patient died) due to the ingestion of a Chinese weight-loss dietary supplement. This supplement, which was labeled as herbal medicine, contained N-nitrosofenfluramine. In addition, it was already shown in the 1970s that in vivo and in vitro nitrosation of APIs [17][18][19][20] is possible. It is, therefore, obvious that the potential for NA formation has not yet been scrutinized to the end, as nitrosation studies are not mandatory during drug development and registration or forced degradation studies.

| Method implementation
As our workgroup has already developed a universal and selective supercritical fluid chromatography-tandem mass spectrometry SFC-MS/MS method for 16 aliphatic, cyclic, and aromatic nitrosamines using Quality-by-Design principles, [21] analysis was easily extended to nitrosated APIs. Three commercially available nitrosated APIs ( Figure 4) were spiked at the 2-ppm level to their corresponding drug products to evaluate method suitability for this new group of nitrosamine species.
Selectivity and detectability were evaluated at a minimum of two drug product batches from different manufacturers by tuned selected reaction monitors. All N-nitroso derivatives were fully separated from their respective APIs.
The implementation experiments demonstrated that SFC-MS/ MS is able to separate and detect nitrosated APIs very efficiently and sensitively (resolution factor > 1.5 and signal-to-noise ratio > 1000 at the 2-ppm level), which is necessary during drug and process development. Out of the investigated model compound samples (Table 1) Of the 33 drug-nitrite interaction products, 24 (72.7%) are already commercially available or reported as known N-nitroso compounds (NOC). [22][23][24] Seven drugs (21.2%) have not yet been reported to form N-nitroso derivatives and two additional drugs (opipramol and amoxicillin) were already reported [22][23][24] to form F I G U R E 3 Time frame and expectation for nitrosamine referral according to EMA/425645/ 2020 [11,12] for active pharmaceutical ingredients (APIs) and finished products (FPs). EMA, European Medicines Agency; NA, nitrosamine; TTC, threshold of toxicological concern F I G U R E 4 Commercially available and investigated N-nitroso derivatives of active pharmaceutical ingredients (APIs) for method implementation (original APIs displayed with the red UV channel only, to prevent oversaturation of the mass detector, and three mass transitions of each N-nitroso compound species by electrospray ionization tandem mass spectrometry). T A B L E 1 Positive and negative findings by SFC-TOF-MS of the investigated APIs after nitrosation with the NAP test and corresponding N-nitroso derivatives with mass error (only findings with a mass error of ≤5 ppm are reported) and NOC yield (intensity ratio of MS signals in extracted ion chromatograms between an N-nitroso derivative and API) Terbinafine -

Torasemide -
Tramadol - Sotalol, which is already known to form N-nitrososotalol, showed two NOC species ( Figure 5) with the same mass spectrum that were chromatographically separated by SFC on the graphite column, due to its high isomeric separation performance, [25] but only partially with LC.
As the first N-nitrososotalol peak in the SFC, the chromatogram shows a more abundant [M+H] + adduct and the second peak shows a more abundant [M+Na] + adduct in the HRMS spectra; we, therefore, suggest that it was nitrosated at two different amine entities (secondary amine and methanesulfonamide). Multiple NOC species were also observed in felodipine and mirabegron ( Figure 5). For mirabegron, an N,N'-dinitroso derivate was additionally detected (most likely due to nitrosation at the secondary amine and phenylacetamide structure). Felodipine is manufactured as a racemic formulation and has a planar, symmetrical 1,4-dihydropyridine structure. By NO binding to the amine moiety, felodipine deracemizes and forms two isomers that were separated by SFC. chromatographic and spectrometric behavior would then be possible (level 1 structure confirmation). [26] Furthermore, MS/MS analysis is able to detect NAs in very low quantities, due to the targeted approach. This allows precise and sensitive root-cause analysis, as requested by health authorities.
On the contrary, the NAP test results are able to exclude theoretically proposed nitrosamines, derived from the chemical structure of the API or excipient. If the predicted structure cannot be detected after incubation, it is highly unlikely that it will form in the product.

| CONCLUSIONS
Although evidence to support major risk concerns for patients' health is missing or at least very low, [1,27] it has become apparent that the analytical focus of drug analysis was too narrow in the past. Not all nitrosamines are inevitably mutagenic or carcinogenic (e.g., N-nitrosodiphenylamine), as Elder et al. [28] have shown in their review, but without the knowledge of NA formation, it is already evident that this public health problem will persist.
Nevertheless, a lot of NAs are known for their high potential to cause cancer in almost all organs [29] and case studies have shown that not only small aliphatic or aromatic NAs but also nitrosated APIs can cause cancer. [16] Additionally, in vitro data of nitrosated APIs, which were summarized by Brambilla and Martelli, [22] suggest a high genotoxic potential of some drug-nitrite interaction products (Figure 6), some of them even higher than NDMA.
After completion of this study, the Canadian health authority published a short notice on the recall of orphenadrine tablets, used as a muscle relaxant, as certain batches showed elevated levels of

N-methyl-N-nitroso-2-[(2-methylphenyl)phenylmethoxy]ethanamine
(N-nitrosonororphenadrine), [30] another unexpected NOC species, formed from the API. Also, irbesartan tablets were recently recalled in the United States due to the potential presence of N-nitrosoirbesartan. [31] It is, therefore, apparent that the security of drug supply and possibly also the health of patients may be affected, as this group of nitrosamines has not been addressed to date. Such findings could have been avoided if appropriate nitrosation assays had been performed before registration.
In the present study, it was demonstrated how direct risk assessment of pharmaceuticals can be performed using a state-of-theart SFC-HRMS screening method to integrate nitrosated API species into drug development and forced degradation studies. Thus, the potential formation of NOCs can be detected, controlled, and limited.
If this had been performed earlier, varenicline, which was recently elevated to global "essential medicines" status by the WHO, [32] could have been prevented from having to be recalled.
The "nitrosamine crisis," with all its negative effects on patient health and the supply assurance of drug products, should be used to draw advantages for the future. All health participants will benefit from the increased awareness. The lifecycle of drug products will also

| NAP test and sample preparation
According to the 1980 IARC monograph from the WHO, [34] a standardized procedure was applied with an excess of nitrite. Before NAP testing, all drug products were ground and dispersed (for solid dosage forms) or directly dissolved (for liquid and semisolid dosage forms) in the incubation medium.
All samples were incubated at 37°C for 4 h with an API concentration of 10 mmol/l in a 40-mol/l sodium nitrite solution at pH 3.5 (with 1 mol/l hydrochloric acid). The pH was adjusted in the sodium nitrite solution, measured (pH 3.5 ± 0.5), and corrected if necessary after the addition of sample material. After incubation, samples were centrifuged and the particle-free supernatant was analyzed by SFC-HRMS.
Only peaks with a mass error of ≤5 ppm were investigated and reported.