**4. Discussion**

The developed mass spectrometry based analytical methods were adequately selective to quantify the amount of peculiar molecules related to the illicit treatment of fishery products with hydrogen peroxide. In particular, to quantify trimethylamine, trimethylamine-N-oxide, dimethylamine, and 2-hydroxyanisole.

Many other analytical methods were proposed to measure the analytes [33–41]. The following paragraph gives a summary about was presented.

Bilgin's and Chung's research groups performed a quantitation of some amines in a large number of fish samples using HPLC coupled to a photodiode array detector [33] or other detectors such as chemiluminescent nitrogen, SPME-GC-MS, and spectrophotometric ones [34]. The first study regarded the determination of histamine, cadaverine, and tyramine after derivatization with dansyl chloride in 63 fish samples. The declared limit of quantitation was in the range between 0.010 and 0.100 μg/mL. Chung et al. determined TMAO, TMA, DMA, and FA in 266 fish samples. They were used respectively for TMAO, HPLC coupled to a chemiluminescent detector; for the TMA and DMA SPME-GC-MS method with a carboxen/divinylbenzen/polydimethylsiloxane fiber; and LC-visible analyzer for FA after derivatization with 2,4-dinitrophenylhydrazine. The stated LOQ were 25, 10, 10, 5 mg/kg for TMAO-DMA-TMA-FA, respectively [34]. Moreover, Soncin [40] and Chan et al. [35] in their studies used as analytical methodologies the headspace solid-phase extraction coupled with GC-MS. The first performed an identification and quantitation of markers of spoilage in fish and found as indicators

four molecules. They defined the technique suitable for analyzing the volatile compounds [40]. Chan's study monitored the concentration of DMA and TMA in fish and the achieved limit of detection was 100 ppb [35].

Heude in his paper published on Food Analytical Methods in 2015 used a 1H high resolution magic angle spinning NMR spectroscopy to determine the *K*-value [36] and trimethylamine nitrogen content as parameters of freshness and quality of fish products. They studied four species of fish and highlighted as grea<sup>t</sup> advantages the possibility to not process fish samples, no extraction was required for NMR analysis [36]. Feng Li et al. in their research used ion chromatography coupled to an unsuppressed conductivity detector to measure the amount of DMA, TMA, and TMAO in aquatic products. They obtained a good lower limit of detection (60, 80, 100 ng/mL for DMA-TMA-TMAO) and quantified the analytes in three real samples [38].

Finally, the remaining methods developed [37,39,41] were based on mass spectrometry techniques in order to determine the amines of interest. Le's group using a C18-PFP column and triple quadrupole quantify TMAO in human plasma samples as a potential indicator of cardiovascular health. The determined LOD and LOQ were 1 and 6 ng/mL [37]. Romero-Gonzáles et al. measured cadaverine and TMA with UHPLC-MS with a LLOQ of 25 and 60 ppb, respectively [39]. Finally, using a hydrophilic interaction liquid chromatography coupled to mass spectrometry Wu et al. quantified in fish meals five amines with satisfactory LOD and LOQ [41].

The present method is based on ultra-high liquid chromatography coupled to mass spectrometry and demonstrates to be su fficiently sensitive to quantify the four studied amines in fishery products in 5 min only. As previously indicated, the lower limits of quantification in matrices for DMA-TMA-TMAO and CAD were 45, 70, 50, and 40 ng/mL, respectively. The selectivity of the method was remarkable in spite of the low mass to the charge value of molecular protonated ions of the analytes. This region of low *m*/*z* ratio ([M+H]<sup>+</sup> 46, 60, 76, and 103 for DMA-TMA-TMAO, and CAD) is normally a ffected by a high background noise, but using tandem mass spectrometry the signal to noise ratio was reasonably good.

To highlight the illicit treatment of fishery products with hydrogen peroxide, we focused our attention on the ratio between TMAO and TMA. As shown in Table 3, subsequently to the prohibited washing with H2O2, the ratio between the measured compounds was shifted again towards a higher concentration of TMAO, one of the endogenous molecular marker of freshness. This ratio was not completely reversed because hydrogen peroxide acted only on the skin of the fishery products, besides the alkylamine is present in higher concentration in muscles. However, the treatment could make the product look younger and fresher.

As explained in the results section, the GC-MS method was an upgrade of the Tanaka et al. GC-ECD methodology for the quantitation of hydrogen peroxide in Chinese foods. The quantitation is possible thanks to the indirect measurement of 2-hydroxyanisole generated by a potassium ferricyanide catalyzed redox reaction of anisole with H2O2 [31]. In that paper, Tanaka et al. performed a deep investigation about reaction environment (pH range), catalyzer, and hydrogen peroxide concentrations in order to obtain the highest possible yield of redox reaction. For ECD quantitation, pentafluoro-benzoyl chloride was used as a derivatizing agen<sup>t</sup> prior to analysis of the product of the reaction, 2-hydroxyanisole. The LLOQ by Tanaka et al. was 0.10 μg/mL.

With the present method based on mass spectrometry we aimed to improve the literature method both for sample preparation and sensitivity. It is well known that SPME followed by gas chromatography coupled to mass spectrometry is a highly sensitive method and the electron ionization source (EI) owns a high fragmentation repeatability.

The sample preparation herein described is free from the necessity of derivatization to detect the redox reaction product 2-hydroxyanisole. The HS SPME with carboxen/polydimethylsiloxane fiber was improved by testing various pH bu ffers, catalyzer concentrations, reaction times and fiber exposure times, stirring of solution.

We found that the best conditions to obtain the highest amount of 2-hydroxyanisole were: Solution bu ffering at pH 2.5, 0.1 M catalyzer concentration, 1 h at 60 ◦C time reaction, time of 30 min of exposure of the fiber without stirring. In these conditions a LLOQ value of 0.05 μg/mL was obtained.

With this HS SPME-GC-MS method we measured the amount of H2O2 in fishery products, comparing fresh and illicit treated samples. In the latter case, the concentration should be hypothesized by extrapolating over the calibration curve range because the signal was too abundant, and it was calculated larger than 100 ppm, as listed in Table 4.
