Hazard identification

Various details on tylosin, including its structure, structural identifiers, and physicochemical characteristics, are presented in Fig. 1 and Table 1.

Fig. 1

Structures of tylosin factors a, b (desmycosin), c (microcin) and d (relomycin)

Table 1 Physico-chemical characteristicsAbsorption, distribution, metabolism, elimination (ADME)

Based on its pharmacokinetic and pharmacodynamic profiles, tylosin bioavailability is likely to differ among exposure routes and species. Tylosin bioavailability following IM injection in chicken was ranged from 22.5 to 34.0%, while it was ranged from 70 to 95% in pigs, cows, and sheep after oral administration [12,13,14]. In contrast, the reported oral bioavailability in mink is poor (41%), while the IM-injected bioavailability in olive flounder is high (approximately 87%) [7, 15].

Short- and long-term toxicity

Acute toxicity studies performed using various tylosin types, such as base, tartrate, and hydrochloride, routes, and doses indicated that tylosin has low acute toxicity in multiple species [1, 3]. For instance, studies in rats showed an LD50 range from 461 to over 6200 mg/kg. Similarly, the LD50 range in mice was 321 to over 6200 mg/kg and over 800 mg/kg in dogs. Specific information can be referred to in various unpublished studies such as Anderson et al. (1961), Morten (1988), and Quarles (1983). According to the unpublished paper authored by Richards and Berkman (undated), the LD50 in cockerels following a single dose of tylosin phosphate was 3765 mg/kg after oral exposure and 501 mg/kg when injected subcutaneously. Comparative intravenous studies with tylosin A, B, and C in female rats reported respective LD50 values of 321, 193, and 189 mg/kg [3].

Several repeated dose studies have been conducted in laboratory animals (rats and dogs), poultry (chickens, turkeys, ducks, and quails), pigs, and cattle. The no-observed-adverse-effect level (NOAEL) for kidney toxicity following oral tylosin administration for two years in dogs was 100 mg/kg bw/day [3]. A similar study by Anderson et al. (1996) unpublished, in which beagles were exposed to up to 400 mg/kg bw/day oral tylosin through capsules for two years, reported no noticeable adverse effects at all concentrations except occasional diarrhea and vomiting at doses of 10–400 mg/kg bw/day. Since diarrhea and vomiting were common symptoms in untreated animals, the NOAEL was considered to be 100 mg/kg bw/day. In rats, the most critical effect was hematological changes at doses greater than 500 mg/kg bw/day; therefore, the EMA concluded that a NOAEL of 50 mg/kg bw/day was appropriate. A one-year study in Wistar rats reported that the 5000-ppm group had increased urinary pH and blood lymphocyte count, and decreased neutrophil count; therefore, the NOAEL was set at 1000 ppm (equivalent to 39 mg/kg bw/day) [9]. Based on the above studies, short- and/or long-term tylosin administration was not considered a cause of critical adverse effects.

Reproductive and developmental toxicity

Tylosin was investigated for its teratogenicity and multigenerational reproductive toxicity in rats and mice. Drug-related changes between control and treated groups and critical effects on parents or offspring, including mortality, fertility, and malformation rates, were not observed in various unpublished studies (i.e., Anderson et al., 1996; Broddle et al., 1978; EMEA, 1997; Tsubura et al., undated; Tsuchikawa and Akabori, undated). The NOAELs for teratogenicity and reproductive toxicity could not be determined from these studies. Based on these studies, exposure to tylosin was not expected to cause reproductive or developmental health concerns.

Carcinogenicity and genotoxicity

Tylosin was tested for genotoxicity in four in vitro/in vivo assays. Greis (1990) conducted an in vitro chromosome aberration assay in which Chinese hamster ovarian (CHO) cells were treated with tylosin (purity, 99.3%) at 500–1000 µg/mL in DMSO without metabolic activation. The cells were also exposed to 250–750 µg/mL tylosin in DMSO with metabolic activation. The study found no evidence of chromosomal aberrations. Another study used CHO cells for gene mutation assays and found no change in the HGPRT + locus mutation frequency when the tylosin concentration was 100–1500 µg/mL [3]. Greis (1990) also investigated potential gene mutations following treatment with tylosin in mammalian cells. Mouse L5178Y TK + / − cells treated with tylosin (purity, 99.3%) at 10–1000 µg/mL in DMSO showed that tylosin had a weak cytotoxic effect. An in vivo micronucleus assay study performed by Greis (1990) in which ICR mice were administrated up to 5,000 mg/kg bw tylosin (purity, 96%) over 48 h found no evidence of genotoxicity. Therefore, we concluded that tylosin was not a genotoxic compound. Additionally, JECFA reviewed several carcinogenicity studies in rats and found no evidence of carcinogenicity [3]. Overall, tylosin was found to have no genotoxic or carcinogenic effect.

Microbiological data

New antibiotics research has highlighted the link between gut microbiota and host health [16]. Furthermore, as the human microbiota is progressively recognized as a critical therapeutic target when using antibiotics [17], the microbiological effects of tylosin have been investigated in several studies. Tylosin residues are known to disrupt the colonization barrier of the gastrointestinal tract in humans because of their antibiotic activities against bacterial strains in the human colonic flora [3]. The most susceptible bacteria were Bifidobacterium spp. and Clostridium spp., for which MIC50 was 0.062 µg/mL. Conversely, tylosin showed little antibacterial activity against various Escherichia coli strains [3]. JECFA committees evaluated tylosin in compliance with the international cooperation on harmonization of technical requirements for registration of veterinary medicinal products (VICH) and clinical and laboratory standard institute (CLSI) guidelines and recognized the need to establish for it a mADI. Although the VICH guideline recommended assessing genera based on their MIC50, JECFA committees decided to use the MIC90 values for Bacteroides fragilis, other Bacteroides spp., Bifidobacterium spp., Clostridium spp., Enterococcus spp., Eubacterium spp., Fusobacterium spp., and Peptostreptococcus spp. to determine a calculated MIC (MICcalc) as recommended by VICH guideline. The justification for adopting the MIC90 was derived from the CLSI guideline. Based on a MIC90 of 5.44 µg/mL, the MICcalc was 1.698 µg/mL [3]. A value of 220 g was used for the colon content mass based on the colon content measured in humans. The maximum antibiotic activity available to microorganisms was 64% in human fecal inactivation studies, and the rate its metabolites reach the colon is likely 35% of tylosin in an activity based—pig data as a substitute for human data. Multiplication of these two explained factors yields 0.224, the fraction of an oral dose available to the microorganisms (FA). Therefore, JECFA concluded that an mADI of 0.0–0.03 mg/kg bw/day (rounded-up value as commonly practiced) could be established based on the MIC test and fecal binding data [3]. The formula used for the JECFA evaluation is explained below.

$$} = \frac}\left( .\mu }/}} \right) \, \times }\left( 0}} \right)}}}\left( } \right) \, \times }\left( 0}} \right)}}$$

In contrast, EMA evaluation of tylosin reported several decades ago established an mADI of 0.006 mg/kg bw/day [5]. The committee for medicinal products for veterinary use (CVPM) in European Union recommended using a specific daily fecal bolus value of 150 mL, geometric mean MIC50 for all sensitive genera, and a correction factor (CF) for microbiological risk assessment. The geometric mean MIC50 for tylosin was 0.606 µg/mL, calculated from MICs for Lactobacillus, Bifidobacterium, Clostridium, Bacteroides, Peptostreptococcus, Eubacterium, and Enterococcus. A CF of 2 was used to adjust the inoculum density. An FA of 0.5 was used to account for the nature of the fecal residues because most of the oral dose is excreted through the feces in some species, and data for humans was unavailable. Hence, the formula used by EMA to determine the microbiological risk of tylosin was as follows:

$$} = \frac}_0}} \left( 0\mu }/}} \right) \, \times }\left( \right) \, \times }\left( 0}} \right)}}}\left( } \right) \, \times }\left( 0}} \right)}}$$

However, a recent study by food safety commission of Japan (FSCJ) established an mADI of 0.005 mg/kg bw/day based on the VICH guidelines, with the only difference from the JECFA evaluation being the MICcalc. The FSCJ used a MICcalc value derived from a domestic investigation of the microbiological effects of veterinary antibiotics in 2006. They obtained a MICcalc of 0.308 µg/mL based on MIC50 in the same genera following the VICH guidelines [8]. The values for other factors (i.e., FA and colon content mass) remained unchanged. Therefore, an mADI of 0.005 mg/kg bw/day was established in Japan. The FSCJ evaluation formula was as follows:

$$} = \frac}\left( 0\mu }/}} \right) \, \times }\left( 0}} \right)}}}\left( } \right) \, \times }\left( 0}} \right)}}$$

Point of departure (POD) determination

Given the toxicological and microbiological data, the microbiological effects are likely the most sensitive endpoint. Therefore, we reevaluated the data above for mADI. The toxicological and microbiological ADIs of international organizations (i.e., JECFA, EMA, and FSCJ) are summarized in supplementary Tables 1 and 2 to help better understand our conclusions. This study recalculated an mADI following the VICH guideline [8]. The MICcalc data followed the FSCJ approach rather than the JECFA approach, as agreed by the expert committees, for the following reasons. First, in vitro MIC data obtained from Japan could be more reliable than the JECFA data because they are based on enough replicates (preferably ten isolates) and inoculum density (> 106). Second, it is possible that the normal gut microflora in Japanese individuals is comparable to that in Koreans. This opinion could be supported by recent articles on ethnicity-associated differences in gut microflora, although the mechanism remains unclear [18, 19]. However, one of the experts expressed concerns about differences in dietary intake patterns between Koreans and Japanese. Unlike the Japanese, most Koreans not only intake kimchi daily, which food fluently contains the beneficial intestinal bacteria, but also prefer salty and spicy food, and such intake habits could result in microbiome differences [20, 21]. Considering all the above, and with a realistic consideration of the Korean gut microbial communities, using the value obtained from the FSCJ seemed reasonable. These rationales are listed in Table 2. Hence, using the formula below, the most appropriate mADI would be 0.01 mg/kg bw/day.

$$} = \frac}\left( 0\mu }/}} \right) \, \times }\left( 00}} \right)}}}\left( } \right) \, \times }\left( 0}} \right)}}$$

Table 2 Korean microbiological ADI and rationalesExposure assessment and risk characterization

Chronic dietary exposure to tylosin residues was estimated using the 2010–2016 KNHANES food consumption data and the proposed MRLs. Multiplication of the two factors (MRL and consumption) yields the exposure amount. A detailed explanation of the exposure assessment model was reported by the World Health Organization [22]. A comprehensive explanation of the model was also well-documented as a manual by international organization [10]. The estimated dietary exposure was shown up to 0.2251 mg/person/day (as equivalent to 0.00375 mg/kg bw/day). The hazard index was 37.5%, indicating that the tylosin residues from use are unlikely to cause a public health concern. The exposure assessment results are presented in Table 3.

Table 3 Results of estimated chronic dietary exposure and risk (%)