Introduction

Nitrosamines are common contaminants in low amounts in many consumer goods from water to foods to tobacco smoke1. However, these products are generally exceptionally low in nitrosamine content. Based on animal studies1,2,3, exposure to N-nitroso-dimethylamine (NDMA) or N-nitroso-diethylamine (NDEA) is acutely toxic and chronically carcinogenic. Likewise, the common organic solvent N, N-Dimethylformamide (DMF) is known to be teratogenic in animals and was involved in the formation of NDMA4.

Although there is a limited number of conclusive studies in humans on the matter, NDMA, NDEA, and DMF are classified by the International Agency for Research on Cancer in group 2A (probable human carcinogen)5.

In 2018, angiotensin receptor blockers (ARB) joined the list of nitrosamine impurities sources when some valsartan6 and losartan7drug products were found contaminated with NDMA or NDEA, and recalled. Approximately 15 million patients8 had a written prescription for losartan or valsartan finished drug products at the time of the recall.

After ARB recalls, in 2019 and 2020, reports showed unacceptable nitrosamine levels in over-the-counter preparations of ranitidine9 and metformin10. While metformin was voluntarily recalled by some drug manufacturers11, the U.S. Food and Drug Administration (FDA) ordered the removal of all ranitidine preparations12. More than 20 million patients8 were taking metformin and over 1 million patients8 had a prescription for ranitidine in 2020 in the United States.

Similar measures were taken by the European Medicines Agency (EMA), i.e., suspending all ranitidine products in the European Union13 and asking manufacturers to test sartan and metformin drug products for nitrosamine impurities14 before releasing them on the market.

To assess the effectiveness of these measures within the European and American drug markets, we tested different commercially available, and post-market drug products. Investigated drugs included 13 metformin products, 17 sartan products, 1 injectable ranitidine product, 1 calcium-channel blocker, 1 anti-Parkinson agent, 1 antimicrobial agent, and 1 diuretic. The impurities we searched for were NDMA, NDEA, and DMF.

Structural similarities to ranitidine15 e. g. the dimethylamine group, and the proven reaction between DMA and nitrosating agents16 led us to test the potential of metformin finished drug products to generate NDMA in simulated gastric fluid.

Methods

Commercially available finished drug products from Europe/Romania (30) were tested for NDMA, NDEA, and DMF as single tablets. Metformin finished drug products were evaluated for NDMA production and DMF in simulated gastric fluid. We tested various available metformin drugs with several concentrations of sodium nitrite, such as 40 mM, 10 mM, 1 mM, and 0.1 mM, in stomach conditions, including a few metformin drugs from the USA market, as specified in the Results section.

Chemicals and reagents

Two nitrosamines (NDMA, NDEA), and DMF were selected as target analytes in this study. Reference standards of NDMA, NDEA, and isotope-labeled DMF standard D7-DMF were purchased from Sigma-Aldrich (St. Louis, MO). Isotopically labeled NDMA standard 13 C2-D6-NDMA was purchased from Cambridge Isotope Laboratories (Tewksbury, MA). Formic acid (LC–MS grade) was purchased from Millipore Sigma (Burlington, MA). Methanol (LC–MS grade) was purchased from Honeywell (Charlotte, NC). Milli-Q water (18.2 MΩ) was generated from a Milli-Q IQ Element system from Millipore Sigma.

Sample preparation

The single drug tablet was accurately weighed. After the methanol dilution to approximately 100 mg active pharmaceutical ingredient (API) per mL of methanol, samples were homogenized. The last tube of rinsing methanol was used as a laboratory reagent blank (LRB) tested with the drug samples. A quality control positive (QCP) sample is also prepared from a recalled lot of Valsartan. Further dilution to 25 mg API/mL is achieved with Milli-Q water, followed by vortexing, centrifugation, and filtration through a 0.2 µm nylon filter to obtain a sample extract that was spiked with an isotopic internal standard mixture containing 666 ng of 13C2-D6-NDMA and 333 ng of D7-DMF. The vial subjected to the LC-HRMS contains a sample of 500 µL with 26 ng/mL 13C2-D6-NDMA and 13 ng/mL D7-DMF. The final internal standard concentration in each sample is 40 ng/mL.

For stomach conditions testing, metformin tablets were kept for 2 h at 37 °C and pH 2.5 in 100 mL solutions with 4 different sodium nitrite concentrations: 0.1 mM, 1 mM, 10 mM, and 40 mM.

Instrumental analysis

SCIEX ExionLC AD (LC) coupled with SCIEX X500R quadrupole time-of-flight high resolution mass spectrometry (QToF HRMS) was purchased from SCIEX (Framingham, MA). The separation of the two nitrosamines and DMF is performed on an InfinityLab Poroshell 120 EC-C18, 4.6 × 100 mm, 2.7 µm analytical column (Agilent Technology, Santa Clara, CA).

Determination of target analytes is performed by SCIEX X500R QToF HRMS operated in positive atmospheric pressure ionization (APCI+) mode for the ionization of NDMA, NDEA, DMF and their isotopic labeled internal standards.

For a comprehensive description of the LC and HRMS parameters consult the supplemental data and the methods described within the reference17.

Quality assurance and quality control (QA and QC)

All calibration and QC samples are prepared in 25/75 (v/v) methanol/water, following a protocol previously described in detail here15.

Statistical analysis

Data are presented as mean and standard deviation (SD) of 3 independent experiments, unless specified otherwise. Data were collected and analyzed using Excel (Microsoft Corp).

Results

Commercial drug analysis for NDMA, NDEA, and DMF impurities

30 drugs from the European market were tested for NDMA, NDEA, and DMF impurities (Table 1). The main analyte, NDMA, was detected only in an injectable preparation of ranitidine and this was below the acceptable daily intake set by the FDA of 96 nanograms18. NDEA was not present in any of the drugs. However, DMF was present in most metformin preparations.

Table 1 Chemical impurity analysis of the first 30 finished drug products from the European Market.

Metformin drugs in simulated gastric fluid

A previous study investigated the formation of NDMA from ranitidine in gastrointestinal conditions when reacting with nitrite in a low pH medium19.

The presence of the dimethylamine group in the molecule of the biguanide, metformin, and the proven interaction between nitrites and DMA to generate NDMA led us to evaluate metformin’s potential to form NDMA in a simulated gastric environment.

8 metformin tablets were dissolved in a stomach-like fluid solution (100 mL solution with a pH of 2.5 at 37 °C and in which drugs are kept for 2 h) containing 40 mM sodium nitrite and then analyzed using the HPLC–MS method. Results are summarized in Fig. 1. All tested drugs in these conditions show NDMA levels in the thousands of nanograms per tablet, amounts which easily exceed the acceptable daily intake established by the FDA of 96 nanograms. However, a concentration of 40 mM is also hard to reach in the stomach, in vivo. We further decided to test various available metformin drugs with lower concentrations of sodium nitrite, such as 10 mM, 1 mM, and 0.1 mM, in stomach conditions. In order to assess products from both markets, we selected 2 of the metformin drugs from Europe (shown in Fig. 1), one with the highest concentration of NDMA produced with 40 mM nitrite and a random one from the remaining 7 and 4 additional commercially available metformin drugs from the USA.

Figure 1
figure 1

NDMA formation from metformin at pH 2.5 in simulated gastric fluid and 40 mM sodium nitrite concentration. A plot of the amount of NDMA at 2.5 pH in 100 mL simulated gastric fluid with 8 different metformin tablets and 40 mM sodium nitrite after 2 h incubation at 37 °C. The data represent the mean and SD of 3 independent samples, and the error bars show the SD of the measurement. The drugs are presented on the x-axis.

In Fig. 2, we present the data from the stomach conditions testing with 3 different nitrite concentrations. At 0.1 mM sodium nitrite, no NDMA was detected in any of the 6 drugs. 1 mM nitrite caused an increase above the acceptable daily intake set by the U.S. Food and Drug Administration (FDA) for some of the metformin drugs, while the 10 mM nitrite and higher concentrations led to NDMA amounts way above the acceptable daily intake set by FDA of 96 nanograms.

Figure 2
figure 2

N-Nitrosodimethylamine formation from metformin at pH 2.5 in simulated gastric fluid and in increasing nitrite concentrations 0.1, 1, 10 mM. A plot of the variable NDMA amount at 2.5 pH in 100 mL simulated gastric fluid with different metformin tablets and 10 mM, 1 mM, and 0.1 mM sodium nitrite after 2 h incubation at 37 °C. The data is presented as the mean of 3 independent samples. For the data with mean and SD of 3 independent samples see Table 2.

.

Table 2 The data represents the mean and standard deviation of 3 independent samples.

Discussion

Various reasonable and common situations exist in which concentrations of nitrite could be significantly elevated. As an example, consuming one pound of dry-cured meat that contains the regulatory limit of nitrite (625 ppm)20 would bring 800 mL of stomach volume to a nitrite concentration of over 5 mM.

All drugs tested in the presence of 40 mM of sodium nitrite in stomach conditions show NDMA levels in the thousands of nanograms per tablet, amounts which significantly exceed the acceptable daily intake established by the FDA of 96 nanograms. For example, Metformin 6 and Metformin 1 have 4255 ng/tablet and 3470 ng/tablet of NDMA, respectively. However, a concentration of 40 mM is also hard to reach in the stomach, in vivo. Thus, we repeated the experiment with 2 selected metformin drugs and 4 metformin drugs from the USA market at lower concentrations of sodium nitrite, namely 0.1 mM, 1 mM, and 10 mM. Metformin 6 showed the highest NDMA amount per tablet, and Metformin 11 the lowest.

Limitations

The in vitro model used in this study has inherent limitations and does not reflect all aspects of human physiology. However, it does provide a method for assessing the potential for physiologic nitrite reactions with drugs in the gastric fluid that lead to NDMA formation.

Conclusion

In this in vitro study, nitrite concentration had an especially important effect on NDMA quantification in metformin tablets added to simulated gastric fluid. 1 mM nitrite caused an increase above the acceptable daily intake set by the U.S. Food and Drug Administration (FDA) for some of the metformin drugs, while the 10 mM nitrite and higher concentrations led to NDMA amounts exceedingly elevated compared to the acceptable daily intake set by the FDA of 96 nanograms. These findings suggest that metformin can react with nitrite in gastric-like conditions and generate NDMA. Thus, patients taking metformin could be exposed to NDMA, when high nitrite levels are present in their stomach, therefore leading to exposure to higher-than-normal concentrations of NDMA. This metformin-nitrite interaction in the gastric medium is even more relevant when considering how the biguanide is prescribed i.e. it should be taken with food21. Further studies are needed to evaluate the interaction between nitrite content of the meal and its interaction in the gastrointestinal tract with metformin or its degradation product, DMA.