Introduction

To date, semicarbazide (SEM) is determined in food as a marker to detect the illegal use of the banned antibiotic nitrofurazone [1–4]. However, it was recently found that SEM is also present in certain foods packed in glass jars and bottles closed with metal lids sealed with plastic gaskets that are foamed using the chemical blowing agent azodicarbonamide (ADC) [5]. The products concerned included fruit juices, jams, conserves, honey, baby food, pickles and sterilised vegetables, mayonnaise, mustard, sauces and ketchup. ADC is also used as a flour additive in some countries such as Canada [6], USA [7] and Brazil [8] owing to its dough-improving properties; however, this practice is banned in the European Union [9]. Recent studies proved that SEM can also be formed in processed food such as coated poultry products [8] and in bread [10] prepared with ADC-containing flour. Formation of SEM in processed food samples, such as carrageenan-containing food, treated with hypochlorite for disinfection and bleaching have recently been reported by Hönicke et al. [11].

When present in food samples as a metabolite of nitrofurazone, most SEM is present in the protein-bound form, which is stable for several weeks. Under mildly acidic conditions, SEM can be released from the proteins and derivatised with 2-nitrobenzaldehyde (2-NBA), producing the hydrazone derivative. In this way the total SEM content can be analysed [4]. As SEM does not absorb in the UV range of the spectrum, it is analysed using HPLC-MS/MS by which a low detection limit and high selectivity characterisation are achieved.

Few methods have been published which allow the selective determination of the protein-bound SEM fraction [11, 12]. All the other methods described in the literature determine the total SEM content of the sample including both the protein-bound fraction and the free SEM.

The main concern about SEM from the toxicological point of view relates to its potential genotoxicity and/or carcinogenicity [13]. Some studies indicate that SEM is both a weakly genotoxic [14] and carcinogenic [15] agent.

The main purpose of this work is to summarise and discuss the information which has been recently published since SEM was detected in samples in which the fraudulent use of nitrofurazone can be excluded. This survey aims to elucidate the need, or lack of one, to look for an alternative marker for nitrofurazone, as well as to identify the steps that must be taken to eliminate SEM from food, — whatever its origin. Emphasis is put on analytical methods to determine SEM in food.

Sources of SEM

Use of nitrofurazone

Nitrofurans form a group of antibiotics used for the treatment of gastrointestinal and dermatological infections including salmonellosis in cattle, swine, poultry, fish and shrimps. Nitrofurans are also used to treat bacterial diseases in bees and residues have consequently been found in honey [3, 16]. The use of four nitrofurans, namely furazolidone, furaltadone, nitrofurantoin and nitrofurazone, has been banned within the countries of the European Union (Commission Regulation 1442/95 (1995) for furazolidone and Commission Regulation 2901/93 (1993) for the other nitrofurans), Australia (1993), The Philippines (2001), Brazil (2002), Thailand (2002) and the USA (2002) as a result of their carcinogenicity and mutagenicity [17, 18]. Nevertheless, these antibiotics are still used in some countries. Residues of nitrofurans were found, for example, in fish products such as shrimp and catfish originating from southeast Asia. The same residues were found in certain meat, including poultry, swine, cattle, duck and rabbit, imported from the same region [1]. Nitrofuran-type antibiotics are characterised by a very fast metabolism such that they have “in vivo” half-lives of only a few hours [19–21]. It has also been observed that the decrease in the concentration of the parent compound is accompanied by the accumulation of some metabolites in proteins, and that these protein-bound metabolites are stable and persistent in the body and can be analysed several weeks after administration of the antibiotic. These findings were first observed for furazolidone [22]. More recently SEM was shown to be a stable side chain metabolite of nitrofurazone and therefore suitable as a marker for nitrofurazone abuse [23] (Fig. 1a). After feeding nitrofurazone to broilers, Kennedy et al. [23] detected tissue-bond SEM in concentrations proportional to the feed nitrofurazone. SEM has thus far been used as the marker residue of nitrofurazone in official residue controls. Very recently, Cooper and Kennedy reported that SEM and other nitrofuran metabolite residues have been found at parts per million concentrations in the retina of pigs fed therapeutic doses of nitrofuran antibiotics [24].

Fig. 1
figure 1

Chemical structure of SEM and compounds which may liberate SEM on degradation

Full size image

Thermal degradation of ADC

ADC used as a blowing agent in jar gaskets

Recently SEM has been found in some foods packed in jars and bottles closed with metal lids sealed with plastic gaskets. Very likely SEM was formed during heat treatment of the blowing agent azodicabonamide (Fig. 1b) and then migrated from the seals into the food. The ADC decomposes into gaseous products (34%) and solid residues (61%), the latter being comprised mainly of hydrazodicarbonamide (HDC) (34%) and urazole (27%) [25–27] (Fig. 1b). Some concern existed about the fact that SEM could be an artefact of the analytical process itself, which as discussed, implies a long hydrolysis step to release SEM from proteins and a simultaneous derivatisation with 2-NBA at 37–40°C overnight. Stadler et al. [28] tried to elucidate if SEM was formed by thermal degradation of ADC by extracting the gaskets in hot water and directly analysing SEM by LC-MS/MS without derivatisation. The SEM was determined in foamed poly(vinyl chloride) seals of metal lids, as well as in commercially available ADC. It was shown that SEM was formed from thermally heated ADC at temperatures higher than 180°C, reaching a maximum at 220°C and after heating for 30 min. This finding excludes the formation of SEM as an artefact of the extraction step, which was performed at 60°C. The SEM could also be measured in HDC and urazole prior to heat treatment. After exposure to the same pyrolysis conditions as those applied to ADC, SEM was also formed from HDC and urazole, although to a lesser extent. The contribution of SEM as an impurity is apparently minor compared to the formation via thermal degradation pathways from ADC [28].

As mentioned above, SEM was determined by LC-MS/MS without performing derivatisation with 2-NBA. The three major fragment ions at m/z 59, 44 and 31 were also observed for ADC and HDC under comparable infusion MS/MS conditions. Theoretically, it is feasible that SEM would be formed during ionisation or in the interface region of the mass spectrometer upon decomposition of ADC and HDC. This would hamper SEM being discerned from ADC or HDC on the basis of the MS/MS pattern. The problem was chromatographically resolved using a reversed-phase HPLC column, which allowed good separation of SEM from its tentative progenitors. By excluding the possibility of SEM formation as an artefact of the analytical technique in this way, it could be deduced that SEM was formed directly or indirectly on thermal degradation of the plastic gaskets.

The use of ADC as a blowing agent in gaskets will be prohibited within the Member States of the European Union from August 2005 [29].

ADC used as a flour additive

ADC is used as a flour additive in some countries such as the USA, Brazil and Canada. The thiol groups of flour proteins are readily converted to disulfide bridges by ADC, which is in turn, reduced to biurea [30]. This reaction improves the physical properties of the flour and is commonly used in the cereal industry to improve the quality of flours, particularly those poor in gluten.

Containers with poultry products coming from Brazil were analysed for residues of nitrofurans and SEM was detected in some of them. A high proportion of contaminated samples comprised chicken covered with flour salt and spices (64% of the positive cases were salted or seasoned chicken and 27% were covered chicken) [8]. Pereira et al. [8] analysed both SEM-free flour samples and commercial ADC-treated flour. Analysis of the former samples provided negative results for SEM, whereas the latter gave positive results ranging from 2.2 μg/kg to 5.2 μg/kg. These findings seemed to indicate that ADC is responsible for the SEM contamination in flour. To confirm this hypothesis, untreated flour was spiked with ADC. All the spiked samples became contaminated with SEM. The yield of the reaction was 0.1%. The authors postulated that biurea resulting from the decomposition of ADC undergoes a further transformation to SEM. Considering that in 1 kg of covered chicken there are approximately 20–120 g of flour, and that in Brazilian flour there are about 10–40 mg/kg ADC, 0.2–5.0 μg/kg SEM could be expected in the chicken [8].

In the study discussed here above, flour was not heated. However, the flour used for production baking goods and also baking sheets to avoid adhesion would probably follow the high-temperature decomposition pattern observed in the process of producing gaskets and hence SEM might be formed. Recently, Becalski et al. [10] published the results of a study in which the flour was heated to temperatures expected during the baking process and the bread made from ADC-containing flour was baked under commonly used conditions. SEM was formed in large quantities only in those flours containing ADC. In untreated flour only minor peaks with the retention time of the 2-NBA-derivatised SEM were found, probably due to some interference or to cross-contamination with ADC-fortified flours in the mills. The SEM values found in non-heated ADC-containing flour were in good agreement with the levels found by Pereira et al. [8] The results obtained by Becalski et al. [10] suggested that the degree of browning correlated with the amount of SEM present in the heated flour sample, although in dry-heated flour, when the temperature was increased beyond 200°C, the yield of SEM decreased dramatically. Such a decrease could be due to either decomposition of the formed SEM or by reaction of SEM with decomposition products of flour. The concentration of SEM was higher on the bread crust, probably because of a higher temperature compared with the centre of the bread. Also the influence of moisture was evaluated: after adding water to the flour and baking the bread at 200°C, the concentration of SEM was slightly higher than that found after dry heating. Reduction of the decomposition of SEM by the cooling effect of water, or hydrolysis of biurea formed from ADC, could explain that result.

Formation of SEM by hypochlorite treatment of nitrogen-rich food

High levels of SEM have been found in carrageenan and in egg powder and lysozyme; the last two were obtained by a process involving clean-up chromatography on a carrageenan resin column. The carrageenan was considered to be responsible for the contamination of the processed egg products [11], since the incoming material tested negative. Carrageenan is a food additive (E407), used as a thickening, gelling and suspending agent, for example in ice cream, pudding, yogurt, fruity jelly, chocolate milk and meat products. Carrageenan is a mixture of polysaccharides [31, 32] obtained from red seaweed by several processing steps, including alkaline extraction, precipitation, drying, grinding and blending. Semi-refined carrageenan or processed Euchema seaweed (PES) is a more economically processed food additive (E407a) because large-scale precipitation steps are not applied. PES has a higher content of acid-insoluble matter than carrageenan, namely cellulose. To eliminate the cellulose residues present in PES, alcohol or salt precipitation is performed. Bleaching is carried out by addition of a sodium hypochlorite solution containing 0.05–0.1% active chlorine.

Hönicke et al. [11] analysed carrageenan and PES before processing and found positive results for SEM, especially in PES with concentrations higher than 1 μg/kg, whereas typical values were <1 μg/kg in carrageenan. After processing, SEM levels of up to 400 μg/kg were found in PES [11]. Analysis of PES before and after bleaching provided values of 6.5 and 68 μg/kg, respectively. The additional precipitation step applied to produce carrageenan was shown to reduce the concentration of SEM to 0.6 μg/kg. In view of these results, Hönicke et al. [11] concluded that SEM was formed during bleaching.

To further study the possibility of forming SEM by processing food with hypochlorite, Hönicke et al. [11] treated shrimps, chicken, milk, egg white powder, soybean, red seaweed, carrageenan, locust bean gum, gelatine, starch and glucose with hypochlorite solutions containing 0.015, 0.05, 1 or 12% active chlorine. After treatment with the two lower concentrations, which are commonly used for disinfection and bleaching, respectively, no significant increase in the SEM concentration was detected. After treatment with 12% active chlorine, 2–65 μg/kg SEM was found in shrimps, chicken, soybean, flakes, red seaweed, carrageenan and starch, and up to 130 and 450 μg/kg in egg white powder and gelatine [11].

In the same work, no SEM was found after treatment of glucose with active chlorine, which confirms the hypothesis that SEM is formed from nitrogen-containing substances having an amidino or ureido residue such as arginine, histidine, citrulline, creatine or creatinine. The SEM can also be formed by reaction of urea with chloramine, which in turn can originate from the reaction of hypochlorite with ammonia. SEM was also found after hypochlorite treatment (0.015% active chlorine) of arginine, creatine, creatinine and urea (0.01–1 g of sample). Histidine and citrulline did not produce SEM under those conditions.

Hypochlorite is commonly used for disinfection during egg braking operations and it could, theoretically, induce formation of SEM in eggs. As mentioned above, the use of SEM-contaminated carrageenan columns could also introduce SEM contamination into processed egg products.

Can SEM be naturally present in food?

Hönicke et al. [11] also detected SEM in shrimps and prawns that had not been bleached or disinfected or exposed to nitrofuran or any other of the potential SEM sources described in this review: it was therefore concluded that SEM could occur naturally in some animals and plant species. This hypothesis has been corroborated by Saari and Peltonen [12] who recently found SEM in crayfish trapped in a natural environment by non-professional fishermen, which excludes nitrofurazone as source of SEM. The water used to boil the crayfish was from Helsinki where ozone is used to disinfect tap water instead of chlorination. The only question that remains open for the authors of that study is if SEM could have been formed during cooking, since all the crayfish analysed had been cooked. As demonstrated by Becalski et al. [10], an increase in SEM formation may be induced by the high temperatures used in food processing.

Analytical methods for the determination of SEM

Determination of total SEM

Methods for the determination for the metabolites of nitrofurans are based on the analysis of the 2-nitrobenzaldehyde imine-type derivatives (Schiff bases) of the metabolites with UV and mass spectrometric (MS) detection [4, 33]. In recent years, the latter detector has replaced the former owing to its superior detection limits and selectivity. In the particular case of SEM, it was not until 2001 that Leitner et al. [4] published the first method to determine SEM using liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS) as a result of the lack of absorbance of this analyte in the UV region. That method, with small variations, is practically the only one used to analyse SEM in food matrices today.

Briefly, the method consists of four steps: hydrolysis of the protein-bound SEM to get it free in solution and simultaneous derivatisation of the free SEM with 2-nitrobenzaldehyde to form the respective hydrazone derivative, clean-up by liquid–liquid or solid-phase extraction (SPE)— , a more elegant approach, and final LC-MS/MS determination.

Hydrolysis and derivatisation with 2-NBA

As mentioned before, SEM forms stable protein-bound compounds. A hydrolysis step is required to liberate free SEM in solution. The hydrolysis is performed under mildly acidic conditions using 0.12–0.2 mol/L HCl [2–4, 8, 10, 11]. The same mild acidic conditions are also required to perform the derivatisation of SEM with 2-nitrobenzaldehyde (2-NBA). Derivatisation is required because SEM posses a low mass located in the range of MS background noise; furthermore, being a very polar compound, its retention time on a reversed-phase column would be very low. The mass of the hydrazone derivative of SEM is 209 Da, and the signal is less affected by the MS background. The selectivity of the method is also increased because the hydrophobicity of the derivatising agent increases the retention time on the reversed-phase column.

To improve the quantitative aspects of the analysis, an internal standard is added, in some cases after the addition of HCl to follow derivatisation together with the SEM present in the sample, and in some other cases after the derivatisation has taken place. In the original method, Leitner et al. [4] used 4-nitrobenzaldehyde (4-NBA) as internal standard. A disadvantage of using this internal standard is that it has to be added after the derivatisation with 2-NBA to avoid side-reactions with the 2-NBA, which can occur under acidic conditions, such as exchange of the 2-nitrophenyl with 4-nitrophenyl groups. 4-NBA semicarbazone also has a much lower sensitivity than 2-NBA derivatives under the MS conditions used. In addition, the signal suppression effect observed in electrospray could depend on the retention time and produce inaccurate quantification even if an analogue product is used as internal standard [34]. This effect could even mask the detection of some confirmatory transitions, since 2-NBA and 4-NBA derivatives both follow the same m/z 209→192 fragmentation, leading to ambiguous or erroneous interpretations; however, the use of isotopically labelled internal standards eliminates these drawbacks. For this reason 13C15N2-SEM [10, 11] and d4-SEM derivatives [3] are currently used.

Internal standards have also been used to evaluate the efficiency of the hydrolysis–derivatisation step. With that purpose in mind, Leitner et al. [4] synthesised the 4-nitrobenzaldehyde semicarbazone, and Becalski et al. [10] produced the 13C15N2-semicarbazone of 2-nitrobenzaldehyde from labelled SEM. A detailed description of how to prepare stable isotope-labelled 2-nitrobenzaldehyde derivatives of four metabolites of nitrofuran antibiotics has been given by Delatour et al. [34]. By using the derivatised internal standard, Leitner et al. [4] calculated an efficiency for the derivatisation reaction of 70% at all the concentration levels studied. Another group reported recoveries for the internal standard between 30% and 60% depending on the matrix [11]. Becalski et al. [10] obtained recoveries of 30–40% when analysing SEM in flour. Such a low recovery was attributed to the high amount of reactive carbonyl compounds present in the samples, the concentration of which is expected to increase after heating the flour. The reactive carbonyls might compete with the 2-nitrobenzaldehyde, an hypothesis which is confirmed by the twofold increase in the recovery value after increasing the amount of 2-NBA added by two and a half times.

Some studies [3, 4] have revealed a degradation of the underivatised metabolites spiked during the HCl-mediated hydrolysis, an effect that was enhanced when the 2-NBA was added with a delay of some hours after the spiking. To avoid an overestimation of the SEM present in the matrix due to a different behaviour of the metabolite as such and of the internal standard, the latter can be added after the hydrolysis and derivatisation reactions. The main drawback of this approach is that it does not allow the evaluation of the extraction recovery of the bound metabolite but only the efficiency of the liquid–liquid extraction or the SPE performed thereafter. Probably for this reason some authors [10, 11] add the internal standard at the beginning of the hydrolysis and derivatisation reactions.

Hoogenboom et al. [22, 35] performed some experiments supplying 14C-labelled furazolidone and reported that only a small fraction of the analytes could be released from authentic samples. Unfortunately, such an experiment has not yet been conducted for SEM.

SEM has also been determined without a derivatisation step after hydrolysis in water at 60°C in gaskets [28]. Analysing the same gaskets by both the direct SEM determination and determination after 2-NBA derivatisation, it was observed that the latter method underestimates the level of SEM by a factor of 3. However, this method, successfully applied to the determination of SEM in the PVC gaskets of food jars, failed to detect SEM in carrageenan [11] because of insufficient sensitivity resulting from a high matrix effect which resulted in a limit of detection of >100 μg/kg.

Clean-up

Most published studies describe clean-up of extracts prior to the final LC-MS/MS determination of SEM using solid-phase extraction cartridges. Polystyrene–divinylbenzene copolymer was used as a sorbent material, which enables a strong and quite selective retention of the nitro-aromatic derivatives (owing to π–π interactions), whereas most of the matrix compounds are more weakly retained. Cartridges used in the SEM analysis are summarised in Table 1. Elution of the retained SEM is usually performed with 3 mL of ethyl acetate.

Table 1 LC-MS/MS analytical methods used for the determination of SEM in food samples
Full size table

A more traditional approach involves performing liquid–liquid extraction [8, 10, 12] with ethyl acetate. It is generally claimed that clean-up with SPE cartridges leads to a decrease in time of analysis and in the amount of organic solvents used. The latter does not apply in the particular case of SEM analysis. The SPE clean-up implicates the use of 3 mL ethyl acetate plus 3 mL methanol for the conditioning of the cartridge and another 3 mL ethyl acetate to elute the retained SEM, whereas liquid–liquid extraction is commonly performed by extracting twice with only 4 mL ethyl acetate.

In some studies some extra clean-up steps are introduced. For instance Becalski et al. [10], when analysing SEM in flour, applied a de-fatting step by extracting the sample twice with 5 mL n-pentane after the addition of HCl and before adding 2-NBA. This de-fatting procedure is not likely to remove semicarbazones present in the food, because losses of spiked derivatised SEM, 2-nitrobenzaldehydesemicarbazone, were less than 1%.

Edder et al. [2] also applied a de-fatting step when analysing fat meats. Samples were extracted with n-hexane, after the pH of the extracts was adjusted to 7.4, for analysis of SEM in honey samples; a liquid–liquid extraction of the derivatised SEM in 15 mL of ethyl acetate immediately before the SPE clean-up increased the recovery of SEM by 5%. The increase went up to 100 and 75% for other nitrofuran metabolites [3].

LC-MS/MS determination

No special considerations need to be kept in mind when performing the liquid chromatographic analysis of SEM after derivatisation with 2-NBA, and a wide variety of C18 reversed-phase columns have been successfully applied as summarised in Table 1. Only in direct determination of SEM in PVC gaskets was a column with a particular high resolution used in order to separate SEM from other likely precursor, as discussed previously [28].

Most work on SEM determination uses triple quadrupole mass spectrometers. Only one article has been published in which an ion trap spectrometer is used [12]. It is well known that ion trap instruments have problems for quantitation purposes, as they do not maintain good repeatability. In the work by Saari and Peltonen [12], day-to-day variations of 17% were obtained, which is acceptable according to the authors because in SEM analysis the within-day and between-day variations do not play a major role, as only a positive identification is required. Ion-trap instruments feature more in studies dealing with compound identification. The daughter ion spectrum is easy to achieve and some indicative results can even be obtained for a second-generation daughter ion spectrum (MS/MS/MS).

When analysing SEM as a marker of nitrofurazone, the analysis should be performed according to the EU criteria for the analysis of veterinary drug residues in living animals and animal products [36], which describes a system of identification points used to define the number of ions and their corresponding ratios that should be measured when using confirmatory MS techniques. Nitrofurans belong to group A according to the European regulation and therefore there is not a maximum residue limit. For this class, four identification points are required at least. To determine SEM the transition reactions 209→166 and 209→192 would represent 2.0 identification points each.

Not all the published LC-MS/MS methods satisfy the requirements of the Commission Decision 2003/181/EC [37], which sets the minimum required performance limit (MRPL) for the determination of the residues of the banned nitrofurans in food at 1 μg/kg (Table 1). The performance criteria of the analytical method are not reported at all in some papers mainly dealing with mechanisms of formation of SEM [8, 28].

Analytical method for the selective determination of protein-bound SEM

One method has been published for the specific determination of tissue-bound SEM [11]. The method consists of a sequence of extraction of the free SEM with a mixture of water and ice cold methanol, then ice cold methanol and finally ice cold diethylether. To the remaining residue, the same method of hydrolysis with HCl and derivatisation with 2-NBA is applied as for the determination of total SEM. In fact, the tissue-bound fraction of SEM is calculated indirectly by subtracting the extracted amount of SEM from the total SEM content, measured by applying the common hydrolysis–derivatisation approach (Table 1). When the method was applied to determine SEM in carrageenan, the results showed that 70 μg/kg of SEM was tissue-bound and only 9 μg/kg was present in the extractable fraction. It is known that SEM reacts rapidly with substances containing a carbonyl moiety. These compounds are present in food and thus SEM formed, for instance after hypochlorite treatment of the sample, might therefore be rapidly bound.

Conclusions

Recent findings showing that SEM can originate from the degradation of ADC (by reaction of nitrogen-containing substances with hypochlorite) or can even be naturally present in some living organisms put a question mark on the feasibility of continuing the use of SEM as a marker for the detection of the banned antibiotic nitrofurazone. Initially, it was thought that the tissue-bound SEM would serve as an indicator of nitrofurazone because SEM deriving from other sources would be free in solution. However, it has been proved that tissue-bound SEM may occur naturally at low levels in some aquatic animals [12], and may also be rapidly formed in carbonyl-containing food if treated with hypochlorite [11]. Differentiation between SEM coming from nitrofurazone or from other alternative sources might not be possible according to some authors. Some extensive studies are currently being carried out on this topic, which will hopefully allow some final conclusions to be drawn on the matter.

Besides the use of SEM as marker of the fraudulent use of nitrofurazone, the presence of SEM in baby food packed in glass jars with metal lids sealed with plastic gaskets is of particular concern because this type of food represents the basis of the diet for a lot of babies between 6 and 12 months old and because the ratio of SEM intake to body weight is the highest for this group of population.

The only existing recommendation on maximum limits for SEM in food is that indicated by the Commission Decision 2003/181/EC, which sets the minimum required performance limit at 1 μg/kg for the determination of residues of the banned nitrofurans in food. Although proficiency test exercises on contaminated samples are still to be conducted, it has to be stressed that satisfactory results in terms of performance criteria have been obtained when applying most of the LC-MS/MS published methods, as summarised in Table 1.