Evaluation of pyrrolizidine alkaloids in Korean commercial honeys and bee pollens

Abstract

Pyrrolizidine alkaloids (PAs) in 200 honeys and 63 bee pollens distributed in South Korea were investigated to evaluate their concentration and safety by simultaneous LC-MS/MS analysis. The risk of PAs' intake was also evaluated. The average PA concentration in honey and bee pollen was lower in products produced in South Korea (honey, 4.2 µg/kg; bee pollen, 306.4 µg/kg) than in ones produced outside South Korea (honey, 24.3 µg/kg; bee pollen, 327.5 µg/kg). PA concentration in bee pollen was about 30 times higher than in honey. It was thought that bee pollen had a significant effect on PA concentration in honey, as honey and bee pollen showed similar PA detection tendencies. The daily exposure to PAs from honey intake was 0.0003 µg/kg body weight (b.w.) per day for all ages, and the safe margin of exposure (MOE) calculated based on BMDL₁₀ 237 µg/kg b.w./day was 722 759, which was a safe level (MOE 10 000 or more).

Introduction

Pyrrolizidine alkaloids (PAs) are natural toxins produced by plants for self-defense. The PAs are based on a pyrrolizidine structure, and there are more than 500 different PAs in more than 6 000 species of plants. According to their structure, PAs are divided into 4 types: senecionine-, lycopsamine-, heliotrine-, and monocrotaline-type (Dübecke et al., 2011). PAs exist mainly in Asteraceae, Boraginaceae, and Fabaceae plant families with pharmacological effects, but caution is required when ingesting them due to toxicity (Schaneberg et al., 2004). Among PAs, fukinotoxin is mainly present in Petasites japonicus, senkirkine in the roots and leaves of Farfugium japonicum, syneilesine in Syneilesis palmate, and isatidine is mainly present in Senecio isatieus and Senecio retrorsus (Hartmann and Witte, 1995). When PAs contained in plants are absorbed into the body, they are metabolized by enzymes in hepatocytes, converted into toxic components, and excreted in the urine. Acute toxicity could lead to blockage of blood vessels and chronic liver damage and genotoxicity (EFSA, 2011). Unsaturated PAs with a double bond at 1,2-position are considered toxic, with the level of toxicity being highest for cyclic diesters, intermediary for non-cyclic diesters, and lowest for monoesters (IPCS, 1988; Edgar et al., 2014; Mattoks, 1986). The International Agency for Research on Cancer (IARC) classified lasiocarpine, monocrotaline, and riddelliine as Group 2B (possibly carcinogenic to humans); and isatidine, retrorsine, seneciphylline, senkirkine, Symphytum, jacobine, and 18-hydroxysenkirkine as Group 3 (unclassified human carcinogens) (IARC, 2002).

Bees make honey from nectar, honeydew, and sap harvested from flowers and sap-producing plants. Bee pollen is pollen collected from flowers by bees. Therefore, honey and bee pollen may contain PAs from various plants. In 2011, the German Federal Institute for Risk Assessment (Bundesinstitut für Risikobewertung, BfR) recognized the possibility of harm when PAs contained in honey were ingested and conducted a risk assessment of PA toxicity. As a result, it was concluded that the exposure to PAs from the consumption of honey and other foods should be kept as low as possible (BfR, 2011). In 2013, a test method to measure the PA concentration in 17 types of honey was proposed (BfR, 2013). PAs were included in the European Food Safety Authority (EFSA) report on activities related to emerging risks in the field of food hygiene in 2016, among 17 emerging risks reviewed in 2016 (Afonso et al., 2017).

There is still no standard for managing PAs in food worldwide. However, in 2005, the Dutch National Institute for Public Health and the Environment (RIVM) established a maximum non-toxic dose—no observed adverse effect level (NOAEL), of 10 µg/kg, and by applying an uncertainty factor of 100 to body weight (b.w.) per day, the tolerable daily intake (TDI) was presented as 0.1 µg/kg b.w./day (Wit et al., 2015). EFSA and BfR re-analyzed the dose-response evaluation of the riddelliine chronic toxicity test (the occurrence of hepatovascular sarcoma by dietary administration) of the 2003 National Toxicology Program (NTP) Technical Report (NIH, 2003). The toxicity reference point—benchmark dose lower confidence limit (BMDL₁₀), was set at 237 µg/kg b.w./day, and risk assessment was performed (EFSA, 2017; BfR, 2013).

In most countries, efforts were made to recognize risks posed by PAs and find ways to reduce them (CODEX, 2014). As a result, interest in the PAs' safety increased in South Koreaⁱ⁾. Hong et al. (2013) analyzed some PAs in honey, but studies on PAs in honey and bee pollen products distributed in South Korea are insufficient. Valese et al. (2021) reported the total levels of 8 PAs in bee pollen, honey, and Senecio brasiliensis collected from Santa Catarina, Brazil. Although PAs are naturally present in plants and pollen of some species, including Senecio, Crotalaria, Bacharis, Ecchium, Mimosa scabrella, and Vernonia), these compounds are transferred to the final product. Therefore, it is necessary to monitor PA contamination levels in plants, honey, and bee pollen products.

In this study, 17 PAs were analyzed in 200 honeys distributed in South Korea. In addition, the PAs were investigated by the country of origin and honey crop. The concentration of 28 PAs was also analyzed in 63 bee pollens distributed in South Korea.

Materials and Methods

Two hundred honey samples, including 8 South Korean and non-Korean miscellaneous products and 63 bee pollen samples, including 4 South Korean and non-Korean processed pollen and pollen-containing products, were purchased from off-line supermarkets and Internet stores in South Korea. Reference standards including monocrotaline, monocrotaline N-oxide, lycopsamine, lycopsamine N-oxide, retrorsine, retrorsine N-oxide, heliotrine, heliotrine N-oxide, seneciphylline, seneciphylline N-oxide, senecionine, senecionine N-oxide, echimidine, echimidine N-oxide, lasiocarpine, lasiocarpine N-oxide, intermedine, intermedine N-oxide, senecivernine, senecivernine N-oxide, europine, europine N-oxide, jacobine, jacobine N-oxide, erucifoline, and erucifoline N-oxide were purchased from PhytoLab (Vestenbergsgreuth, Germany). Trichodesmine and riddelliine were purchased from ALB Technology (Kowloon, Hong Kong). Symphytine and otosenine were purchased from Wuhan Chemfaces Biochemical Co. (Hubei, China). HPLC-grade methanol (Merck, Darmstadt, Germany) and sulfuric acid (Merck, 98%) were used as a mobile phase and sample pretreatment solvents for HPLC, respectively.

In the cartridge purification step, Oasis MCX 6 cc (500 mg) LP Extraction Cartridges (Waters, Dublin, Ireland) and ammonia (Merck, 32%) were used. For instrumental analysis, LC-MS grade ammonium formate (Sigma-Aldrich, St. Louis, MO, USA) and XBridge C18 column (100 mm × 2.1 mm, 3.5 µm, Waters, Manchester, UK) were used. For PAs' standard stock solution, a certain amount of each standard was taken and dissolved in methanol or acetonitrile to make 500–1 000 µg/mL. A mixed standard solution for simultaneous PA analysis was taken from each standard stock solution, and 5 mL of methanol was added. The final concentration of each PA was 1 000 × LOQ. The standard solution for preparing the matrix-matched calibration curve was prepared by serial dilution of a PA mixed standard solution (equivalent to 1 000 times the limit of quantitation).

Sample preparation    Honey samples were pre-treated by applying the Method Protocol of the German Federal Institute for Risk Assessment (BfR) (BfR, 2013). First, 10 g of a homogenized honey sample was placed in a 50 mL centrifuge container, and 30 mL of the extraction solution (50% methanol containing 0.05 M sulfuric acid) was added. After shaking the mixture for 5 min using a vortex mixer, it was centrifuged at 2 900 g for 10 min, and its supernatant was used as an extract. Next, the extract (30 mL) was injected into an Oasis MCX 6 cc (500 mg) cartridge activated with 5 mL of methanol and 5 mL of 0.05 M sulfuric acid in 50% methanol solution and passed at 1 drop/s. Then, 6 mL of distilled water and 6 mL of methanol were injected at the same flow rate. The solution remaining in the cartridge was removed using a vacuum pump and then collected by eluting 10 mL of 2.5% ammonia in methanol. The eluate was dried with nitrogen gas at 50–55 °C, and methanol was added to the dried product to dissolve again. After filtering with PTFE Chromacol syringe filters (0.2 µm), it was used as the final test solution.

The 40 mL of an extraction solution (0.05 M sulfuric acid in 50% methanol) was added to 0.5 g of the pulverized and homogenized sample, followed by shaking for 60 min. The mixture was centrifuged at 4 500 g for 10 minutes, and 2 mL of the supernatant was used as an extract. The extract (2 mL) was injected into an Oasis MCX 6 cc (150 mg) cartridge activated with 3 mL of methanol and 3 mL of distilled water and passed at 1 drop/sec. Then, 4 mL of distilled water flowed out at the same flow rate, and the solution remaining in the cartridge was removed using a vacuum pump. Then the solution was collected by eluting with 4 mL of 2.5% ammonia in methanol. The eluate was dried with nitrogen gas at 50–55 °C and dissolved again by adding 1 mL of 5% methanol to the dried material. After filtering with PTFE Chromacol syringe filters (0.2 µm), it was used as the final test solution.

LC-MS/MS analysis    The PA concentration in samples was analyzed using LC-MS/MS (SHIMADZU NEXERA X2/LCMS-8060, Tokyo, Japan) equipped with an electrospray ionization (ESI) source. XBridge C18 (100 mm × 2.1 mm, 3.5 µm, Waters, Manchester, UK) column was used. LC-MS/MS conditions are shown in Table 1. MS/MS conditions were optimized for the analysis of the toxins as follows: curtain gas (CUR), 25.0 psi; collision gas (CAD), 9 psi; ion spray voltage, 5.0 kV; ion source temperature, 350 °C; ion source gas (GS1), 50.0 psi; ion source gas (GS2), 50.0 psi; and source collision energy, 31–105 V (N₂). The optimal multiple reaction monitoring (MRM) conditions, including the precursor ion, product ion, qualification ion, and collision energy for the analysis of PAs in honey and bee pollen, were determined using the PA detection approach, according to Kwon et al. (2021).

Table 1. Operating conditions of LC/MS/MS for PA analysis.

Parameter	Conditions
Instrument	SHIMADZU NEXERA X2/LCMS-8060
Ionization	ESI (electrospray ionization) positive
Column	XBridge C18 3.5 µm (2.1 × 100 mm)
Time	15 min
Mobile phase	A: aqueous 0.05 mM ammonium formate and 0.1% formic acid
	B: 100% methanol with 0.05 mM ammonium formate and 0.1% formic acid
Column temperature	40 ℃
Gradient	Time	A (%)	B (%)
	Gradient 0.0	95	5
	Gradient 0.5	95	5
	Gradient 5.0	25	75
	Gradient 7.0	35	65
	Gradient 9.0	50	50
	Gradient 9.5	20	80
	Gradient 9.6	0	100
	Gradient 11.0	0	100
	Gradient 11.1	95	5
	Gradient 15.0	95	5
Flow rate	0.3 mL/min
Injection vol.	1 µL
Ionization mode	multiple reaction monitoring (MRM)

Risk assessment of PAs in honey    The daily exposure to PAs due to honey intake was calculated using the amount of honey ingested and the average body weight of the investigators who consumed honey during the 5^th and 6^th periods (2010–2015) of the National Health and Nutrition Examination Survey. PAs' MOE was calculated using the BMDL₁₀ value of 237 µg/kg body weight (b.w.) per day announced by EFSA (2017).

  

Results and Discussion

Total PA concentration in honey    Table 2 shows PA concentration in Korean and non-Korean honey distributed in South Korea. The overall PA detection rate was 75% (PAs were detected in 150 cases out of 200 analyzed). The PA detection range was 0.1–113.4 µg/kg, and the average total PA concentration was 4.9 µg/kg. Ryu et al. (2019) analyzed 7 PAs in 84 honey samples distributed in South Korea and reported that PAs were detected in 6 samples, and the PA detection range was 1.76–202.1 µg/kg.

Table 2. PA concentration in honey.

Honey type	No. of samples (No. of detected samples)			PAs (µg/kg)
	Total	Korean	Non-Korean	Range	Average	Korean	Non-Korean
Sugar-fed honey	23 (13)	23 (13)	0 (-)	0.1–20.0	1.3	2.6	-
Polyfloral honey	69 (55)	64 (54)	5 (1)	0.1–45.0	4.4	5.8	1.2
Acacia honey	62 (49)	61 (49)	1 (0)	0.3–15.2	2.7	3.5	0
Chestnut honey	11 (7)	9 (7)	2 (0)	0.5–5.0	1.8	2.8	0
Manuka honey	6 (6)	0 (-)	6 (6)	1.6–14.1	8.2	-	8.2
Eucalyptus honey	4 (4)	0 (-)	4 (4)	5.0–113.4	72.6	-	72.6
Clover honey	2 (2)	0 (-)	2 (2)	0.3–1.6	0.9	-	0.9
Others	23 (14)	11 (9)	12 (5)	0.3–69.0	5.0	2.1	18.7
	200 (150)	168 (132)	32 (18)	0.1–113.4	4.9	4.2	24.3

The PA detection rate was higher in Korean honey (79%) than in non-Korean honey (56%), but the average PA concentration was lower in Korean (4.2 µg/kg) than in non-Korean (24.3 µg/kg) honey. Dübecke et al. (2011) reported that the detection range of 10 PAs in honey produced in Central and South American countries such as Argentina and Chile was 1–1 087 µg/kg, and the average PA concentration was 67 µg/kg. On the other hand, the detection range of 10 PAs in honey produced in European countries such as Bulgaria and Hungary was 1–225 µg/kg. The average PA concentration was 26 µg/kg, which was relatively low compared to honey from Central and South American countries. Bodi et al. (2014) reported that the detection range of 17 PAs in honey distributed in Germany was 0.3–234.5 µg/kg, and the average PA concentration was 6.1–14.5 µg/kg. Valese et al. (2016) reported that PAs were detected in 99.1% of honey distributed in Brazil, and the detection range of 8 PAs was 0–423.4 µg/kg.

The results of this study are similar to those of Dübecke et al. (2011), Bodi et al. (2014), and Valese et al. (2016), and the PA concentration in honey produced in Korea was lower than that of in honey produced outside Korea. The PA detection rates by honey crops decreased in the order of Manuka honey (100%), eucalyptus honey (100%), clover honey (100%), miscellaneous honey (80%), and acacia honey (79%). Eucalyptus honey had the highest PA concentration, with the maximum PA concentration of 113.4 µg/kg and the average PA concentration of 72.6 µg/kg. Ryu et al. (2019) reported that the PA content of eucalyptus honey was the highest at 201.71 µg/kg among the types of honey distributed in South Korea. Because sugar-fed honey was harvested and matured after bees were fed sugar, it was considered that the PA detection rate (57%) and PA concentration would be relatively lower for that type of honey than for the other honey crops.

More than 500 PAs were discovered, and the strength of toxicity varied according to PAs' chemical and structural characteristics (Zhou et al., 2010). BfR (2013) selected 17 PAs known to be relatively toxic through analysis and examined their concentration in honey. The average PA concentration in honey decreased in the following order: echimidine (25.0 µg/kg), lycopsamine (16.7 µg/kg), and trichodesmine (4.1 µg/kg) (Table 3). Echimidine was detected in large amounts in eucalyptus honey. Echimidine is mainly present in Echium species such as comfrey that mainly inhabit South America and Spain (Dübecke et al., 2011). Lycopsamine is mainly found in Eupatorieae, which inhabits South American countries such as Argentina, Uruguay, and Brazil (Dübecke et al., 2011). Relatively low amounts of trichodesmine are detected, but its detection frequency is high. It is mainly detected in Trichodesma incanum, a plant of the Boraginaceae family in Central Asia. In 1950, 44 deaths were reported in Uzbekistan caused by eating bread contaminated with the seeds of this plant (Huxtable et al., 1996).

Table 3. PAs in honey.

PA (No. of detected samples)	Honey type (No. of detected samples)	Range (µg/kg)	Average (µg/kg)
Lycopsamine (9)	Polyfloral honey (1)	0.1	0.1	16.7
	Manuka honey (3)	0.6–3.6	1.9
	Eucalyptus honey (2)	31.0–42.7	36.9
	Clover honey (2)	0.2–1.6	0.9
	Others (1)	69.0	69.0
Retrorsine-N-oxide (2)	Acacia honey (2)	0.1–0.2	0.2	0.2
Trichodesmine (134)	Sugar-fed honey (12)	0.1–20.0	2.5	4.1
	Polyfloral honey (54)	0.1–45.0	5.6
	Acacia honey (49)	0.3–15.2	3.5
	Chestnut honey (7)	0.5–5.0	2.8
	Eucalyptus honey (2)	0.3–5.0	2.7
	Others (10)	0.4–4.1	1.8
Seneciphylline (2)	Manuka honey (2)	1.7–3.8	2.8	2.8
Seneciphylline-N-oxide (1)	Acacia honey (1)	0.2	0.2	0.2
Senecionine (5)	Manuka honey (4)	0.2–4.6	3.3	2.8
	Eucalyptus honey (1)	0.9	0.9
Senecionine-N-oxide (1)	Manuka honey (1)	0.3	0.3	0.3
Echimidine (10)	Polyfloral honey (1)	1.2	1.2	25.0
	Manuka honey (2)	0.2–14.1	7.2
	Eucalyptus honey (3)	67.4–72.2	70.1
	Clover honey (1)	0.1	0.1
	Others (3)	0.3–22.3	8.0
Riddelliine (2)	Manuka honey (2)	3.0–7.6	5.3	5.3

Table 4 shows the PAs by the honey's country of origin. Trichodesmine was detected frequently and in large amounts of honey from South Korea and France. Bodi et al. (2014) analyzed 17 PAs, including trichodesmine, in the honey distributed in Germany, but trichodesmine was not detected. Griffin et al. (2015a) analyzed 14 PAs, including trichodesmine, in the honey distributed in Australia. Trichodesmine was not detected, but lycopsamine (detection rate 88%, average detection amount 76 µg/kg) and echimidine (detection rate 76%, average detection amount 106 µg/kg) were detected. Valese et al. (2016) reported that senecionine and senecionine N-oxide were mainly detected in the honey distributed in Brazil because many plants belonged to the genus Senecio.

Table 4. PAs in honey by the country of origin.

Origin (No. of detected samples)	PA (No. of detected samples)	Range (µg/kg)	Average (µg/kg)
South Korea (168)	Lycopsamine (1)	0.1	0.1
	Retrorsine N-oxide (2)	0.1–0.2	0.2
	Trichodesmine (131)	0.1–45.0	4.1
	Seneciphylline N-oxide (1)	0.2	0.2
New Zealand (8)	Lycopsamine (3)	0.6–3.6	1.9
	Seneciphylline (2)	1.7–3.8	2.8
	Senecionine (3)	0.2–4.3	2.9
	Senecionine-N-oxide (1)	0.3	0.3
	Echimidine (5)	0.2–22.3	7.9
	Riddelliine (2)	3.0–7.6	5.3
USA (4)	Lycopsamine (3)	0.2–69.0	23.6
	Echimidine (1)	0.1	0.1
Australian (4)	Lycopsamine (2)	31.0–42.7	36.9
	Trichodesmine (1)	0.3	0.3
	Senecionine (2)	0.9–4.6	2.7
	Echimidine (3)	67.4–72.2	70.1
France (2)	Trichodesmine (2)	0.7–5.0	2.9
Italy (1)	Echimidine (1)	0.3	0.3

Dübecke et al. (2011) reported that echimidine and lycopsamine were mainly found in honey produced in many Latin American and European countries due to the significant influence of Echium and Eupatorium species. Relatively varied PAs were detected in honey from New Zealand, and among them, a large amount of echimidine was detected (Table 4). Riddelliine, which is relatively toxic, was detected in 2 cases of Manuka honey from New Zealand, and lycopsamine was detected in honey from the United States. Lycopsamine and echimidine were detected in large amounts of honey from Australia, and the amount was relatively higher than in honey from other countries. The results of this study are similar to the studies by Griffin et al. (2015a) and Dübecke et al. (2011) in that echimidine and lycopsamine were mainly detected in honey from America and Europe.

The PAs in different honey crops are shown in Table 5. Trichodesmine was detected in Korean honey, including sugar-fed, miscellaneous, acacia, and chestnut honey. Echimidine and lycopsamine were detected in honey imported to South Korea, such as Manuka honey, eucalyptus honey, and clover honey. Griffin et al. (2015b) reported that PAs were mainly detected in Manuka honey, forest honey, and floral honey, among the types of honey distributed in Ireland.

Table 5. PAs in honey by source.

Honey type (No. of samples)	PA (No. of detected samples)	Average (µg/kg)
Sugar-fed honey (23)	Trichodesmine (12)	1.28
Polyfloral honey (69)	Echimidine (1)	4.43
	Lycopsamine (1)
	Trichodesmine (54)
Acacia honey (62)	Retrorsine-N-oxide (2)	2.74
	Seneciphylline-N-oxide (1)
	Trichodesmine (49)
Chestnut honey (11)	Trichodesmine (7)	1.75
Manuka honey (6)	Echimidine (2)	8.24
	Lycopsamine (3)
	Riddelliine (2)
	Senecionine (4)
	Senecionine-N-oxide (1)
	Seneciphylline (2)
Eucalyptus honey (4)	Echimidine (3)	72.55
	Lycopsamine (2)
	Senecionine (1)
	Trichodesmine (2)
Clover honey (2)	Lycopsamine (2)	0.93
	Echimidine (1)
Others (23)	Echimidine (3)	4.89
	Lycopsamine (1)
	Trichodesmine (10)

PA concentration in bee pollen    Table 6 shows the PA concentration in Korean and non-Korean bee pollen distributed in South Korea. The total PA detection rate was 51% (PAs detected in 32 out of 63 cases analyzed), the total PA detection range was 0.0–2 829.1 µg/kg, and the average total PA concentration was 319.5 µg/kg. The PA detection rate was lower in Korean bee pollen (17%, detected in 4 out of 24 cases) than in non-Korean bee pollen (72%, detected in 28 out of 39 cases). The average PA concentration was lower in Korean bee pollen (306.4 µg/kg) than in non-Korean bee pollen (327.5 µg/kg). The PA detection rate for different types of pollen decreased in the following order: pollen-containing processed products (69%), processed pollen (65%), and pollen (agricultural products, 31%). The PA concentration in pollen (agricultural products) was the maximum 2 829.1 µg/kg and the highest.

Table 6. PAs in Korean and non-Korean bee pollens

Pollen type	No. of samples (No. of detected samples)			PAs (µg/kg)
Pollen type	Total	Korean	Non-Korean	Range (min–max)	Average	Korean	Non-Korean
Processed pollen	20 (13)	0 (-)	20 (13)	0.0–1 185.8	370.7	-	370.7
Pollen-containing product	13 (9)	2 (-)	11 (9)	0.0–1 199.6	311.2	0.0	367.8
Other product	1 (1)	0 (-)	1 (1)	46.0	46.0	-	46.0
Pollen (agricultural product)	29 (9)	22 (4)	7 (5)	0.0–2 829.1	297.2	334.2	181.1
Total	63 (32)	24 (4)	39 (28)	0.0–2 829.1	319.5	306.4	327.5

Table 7 shows PA concentration by the country of origin for bee pollen distributed in South Korea. The detection rate of PAs was the highest in Bulgarian pollen (67%), followed by Spanish (50%), Vietnamese (50%), and Korean (17%) pollen. The average PA concentration was the highest in Bulgarian pollen, followed by Vietnamese, Korean, Spanish, and Australian bee pollen. The maximum PA concentration was the highest in Korean bee pollen at 2 829.1 µg/kg. Among non-Korean bee pollens, it was the highest in Spanish bee pollen at 1 117.5 µg/kg. Dubecke et al. (2011) reported that the PA concentration in bee pollen originated from Southeast European countries such as Croatia and Bulgaria was low. However, the PA concentration in bee pollen from Spain was relatively high. On January 12, 2018, a large amount of PAs (1 511 µg/kg) was detected in Spanish bee pollen products, and they were returned (RASFF, 2018). The PAs were also detected in bee pollen from New Zealand, Latvia, Italy, and Colombia. Innacio et al. (2021) reported that 67% of 61 bee pollen samples in Italy contained PAs with a mean concentration of 339 µg/kg.

Table 7. PA concentration in bee pollens by the country of origin.

Origin	No. of samples	No. of detected samples	Detection rate (%)	Average (µg/kg)	Range (µg/kg)
Spain	20	10	50	290.7	0–1 117.5
South Korea	24	4	17	319.5	0–2 829.1
Vietnam	4	2	50	349.3	262.6–435.9
Australia	3	1	33	183.5	183.5
Bulgaria	3	2	67	353.3	125.1–581.4
USA	2	0	0	0.0	-
New Zealand	1	1	100	32.1	32.1
Latvia	1	1	100	179.9	179.9
Italy	1	1	100	160.7	160.7
Colombia	1	1	100	710.6	710.6
Mexico	1	0	0	0.0	-
Estonia	1	0	0	0.0	-
China	1	0	0	0.0	-
Total	63	32

The average PA concentration in bee pollen was the highest for trichodesmine (1 838.21 µg/kg), followed by echimidine N-oxide (318.39 µg/kg), and echimidine (255.37 µg/kg) (Table 8). Thus, a relatively large amount of trichodesmine was detected, but its detection frequency (6.3%, detected in 4 out of 63 cases) was low. It was mainly detected in Trichodesma incanum, a plant of the Boraginaceae family in Central Asia (Huxtable et al., 1996). On the other hand, large amounts and high detection frequency were identified for echimidine N-oxide (22.2%, detected in 14 out of 63 cases) and echimidine (25.4%, detected in 16 out of 63 cases). These are mainly contained in Echium species such as comfrey, found mainly in South America and Spain (Dubecke et al., 2011).

Table 8. PAs in bee pollen.

PA	No. of detected samples	Range (µg/kg)	Average (µg/kg)
Echimidine	16	16.1–580.9	255.37
Echimidine N-oxide	14	12.2–789.9	318.39
Lycopsamine N-oxide	6	51.3–185.7	97.03
Intermedine N-oxide	5	56.7–194.5	108.98
Lycopsamine	4	32.1–133.6	67.88
Trichodesmine	4	1 146.1–2,829.1	1 838.21
Intermedine	3	160.7–183.5	174.71
Senecionine	1	122.2	122.22
Senecionine N-oxide	1	233.8	233.84
Senecivernine	1	116.2	116.22
Senecivernine N-oxide	1	238.4	238.40

Table 9 shows PAs by the country of origin of bee pollen. A large amount of trichodesmine was detected in Korean bee pollen. In Spanish bee pollen, echimidine and echimidine N-oxide were detected in large amounts, and their detection frequency was high. Large amounts of lycopsamine and lycopsamine N-oxide were detected in bee pollen from Vietnam and Bulgaria, and the detection frequency was high. Dubecke et al. (2011) reported that Vietnamese bee pollen contained high lycopsamine and lycopsamine N-oxide levels, possibly because the flower pollen of Chromolaena odorata, mainly found in Vietnam, contains large amounts of these PAs. It was also reported that echimidine and echimidine N-oxide were mainly detected in Spanish bee pollen. Valese et al. (2021) reported that bee pollen samples collected from Santa Catarina, Brazil contained senecionine-type PAs/PANOs. In this study, senecionine-type PAs/PANOs were mainly detected in bee pollen from Colombia. Dubecke et al. (2011) investigated the PA concentration in 381 samples of honey and 119 samples of bee pollen, and the detection rate in honey (65%) was slightly higher than in bee pollen (60%). However, the average PA concentration was 1 846 µg/kg, about 30 times higher than in honey (26 µg/kg).

Table 9. PAs in bee pollen by the country of origin.

Origin (No. of detected samples)	PA (No. of detected samples)	Range (µg/kg)	Average (µg/kg)
South Korea (24)	Trichodesmine (4)	1 146.1–2 829.1	1 838.2
Spain (20)	Echimidine (14)	16.0–580.9	346.8
	Lycopsamine (1)	51.9	5.2
	Echimidine N-oxide (12)	187.5–789.9	436.9
	Lycopsamine N-oxide (4)	51.3–104.5	28.9
	Intermedine (3)	56.7–194.5	32.7
Bulgaria (3)	Lycopsamine (2)	112.9–505.5	261.1
Latvia (1)	Intermedine (1)	179.9	179.9
Vietnam (4)	Lycopsamine (2)	53.9–133.6	46.9
	Lycopsamine N-oxide (2)	107.5–185.7	73.3
	Intermedine N-oxide (2)	101.2–116.6	54.5
Colombia (1)	Senecionine (1)	122.2	122.2
	Senecionine N-oxide (1)	233.8	233.8
	Senecivernine (1)	238.4	238.4
	Senecivernine N-oxide (1)	116.2	116.2
Australia (3)	Intermedine (1)	183.5	183.5
New Zealand (1)	Lycopsamine (1)	32.1	32.1
Italy (1)	Intermedine (1)	160.7	160.7

Based on these results, it was concluded that bee pollen had a significant effect on the PA concentration of honey. In this study, high levels of trichodesmine, echimidine, lycopsamine, and senecionine were detected in honey and bee pollen. The detection rate of trichodesmine, echimidine, lycopsamine, and senecionine was also high in honey and bee pollen.

Risk assessment of honey    The MOE is a risk assessment method to prioritize risk management of carcinogens. The MOE of 10 000 was applied to assess PAs as genotoxic carcinogens (EFSA, 2017). The daily exposure to PAs due to honey intake was 0.0003 µg/kg b.w./day for all ages. The MOE calculated based on BMDL₁₀ 237 µg/kg b.w./day was 722 759 for all ages, which was a safe level (MOE 10 000 or more) (Table 10). Eucalyptus honey, with a high level of PA contamination compared to other honeys, had a relatively high daily exposure rate (0.0048 µg/kg b.w./day). However, it was a safe level based on the MOE of 10 000 or more (49 591). BfR (2011) reported that efforts should be made to keep the daily exposure to PAs from honey intake to 0.007 µg/kg b.w./day or less. EFSA (2016) concluded that the average daily exposure to PAs among infants and young children due to honey consumption was 0.003 to 0.027 µg/kg b.w./day. Thus, it is necessary to keep the total PA exposure of consumers as low as possible in the future. As a result of the risk assessment in this study, the risk level due to PAs in the honey distributed in South Korea was low for all age groups.

Conclusion

PA detection rate was slightly higher in honey than in bee pollen, and the average PA concentration in bee pollen was higher than in honey. The average PA concentration in honey and bee pollen was higher in non-Korean products than in Korean products. Trichodesmine was mainly detected in Korean products, and echimidine was detected in non-Korean products. The PA concentration in bee pollen was about 30 times higher than that of honey, and the PA detection tendencies in honey and bee pollen were similar. Therefore, it was judged that bee pollen had a significant effect on the PA concentration in honey. PA concentration in honey and bee pollen showed trends similar to previous research. Korean products are considered to be less contaminated with PAs than non-Korean ones. It is expected that the PA concentration analysis data in foods distributed in South Korea will be useful for PA safety management in the future.

Acknowledgements    This study was financially supported by the grant (No.17161MFDS013) from the Ministry of Food and Drug Safety (MFDS) of the Republic of Korea.

Conflict of interest    There are no conflicts of interest to declare.

References

•  Afonso,  A.,  García Matas,  R.,  Maggiore,  A.,  Merten,  C., and  Robinson,  T. (2017). Technical report on EFSA's Activities on Emerging Risks in 2016. EFSA Supporting publication 2017: EN-1336.
• BfR (2011). Chemical analysis and toxicity of pyrrolizidine alkaloids and assessment of the health risks posed by their occurrence in honey. BfR opinion No. 038/2011.
• BfR (2013). Determination of pyrrolizidine alkaloids (PA) in honey by SPE-LC-MS/MS, Method Protocol. BfR-PA-Honey-1.0/2013.
•  Bodi,  D.,  Ronczka,  S.,  Gottschalk,  C.,  Behr,  N.,  Skibba,  A., and  Wagner,  M. (2014). Determination of pyrrolizidine alkaloids in tea, herbal drugs and honey. Food Addit. Contam.: Part A, 31, 1886-1895.
• CODEX Alimentarius (2014). Code of practice for weed control to prevent and reduce pyrrolizidine alkaloid contamination in food and feed. CAC/RCP 74-2014.
•  Dübecke,  A.,  Beckh,  G., and  Lüllmann,  C. (2011). Pyrrolizidine alkaloids in honey and bee pollen. Food Addit. Contam.: Part A, 28, 348-358.
•  Edgar,  J.A.,  Molyneux,  R.J., and  Colegate,  S.M. (2014). Pyrrolizidine alkaloids: Potential role in the etiology of cancers, pulmonary hypertensions, congenital anomalies, and liver disease. Chem. Res. Toxicol., 28, 4-20.
• EFSA (2011). Scientific opinion on pyrrolizidine alkaloids in food and feed. EFSA J., 9, 2406.
• EFSA (2016). Dietary exposure assessment to pyrrolizidine alkaloids in the European population. EFSA J., 14, 4572.
• EFSA (2017). Risks for human health related to the presence of pyrrolizidine alkaloids in honey, tea, herbal infusions, and food supplements. EFSA J., 15, 4908.
•  Griffin,  C.T.,  Mitrovic,  S.M.,  Danaher,  M., and  Furey,  A. (2015a). Development of a fast isocratic LC-MS/MS method for the high-throughput analysis of pyrrolizidine alkaloids in Australian honey. Food Addit. Contam.: Part A, 32, 214-228.
•  Griffin,  C.T.,  O´Mahony,  J.,  Danaher,  M., and  Furey,  A. (2015b). Liquid chromatography tandem mass spectrometry detection of targeted pyrrolizidine alkaloids in honeys purchased within Ireland. Food Anal. Methods, 8, 18-31.
•  Hartmann,  T., and  Witte,  L. (1995). Chemistry, biology and chemoecology of the pyrrolizidine alkaloids. Alkaloids: Chem. Biol. Perspectives., 9: 155-233.
•  Hong,  J.G.,  Yoon,  S.H.,  Yang,  K.R.,  Lee,  S.K., and  Lee,  J.Y. (2013). Development of analytical method for the monitoring of pyrrolizidine alkaloids in foods. Korea Food and Drug Administration. No.13162MFDS012.
•  Huxtable,  R.J.,  Yan,  C.C.,  Wild,  S.,  Maxwell,  S., and  Cooper,  R. (1996). Physicochemical and metabolic basis for the differing neurotoxicity of the pyrrolizidine alkaloids, trichodesmine, and monocrotaline. Neurochem. Res., 21, 141-146.
• IARC Working Group on the Evaluation of Carcinogenic Risks to Humans (2002). Some traditional herbal medicines, some mycotoxins, naphthalene and styrene. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, 82, 1-556.
•  Inacio,  L.D.J.,  Merlanti,  R.,  Lucatello,  L.,  Bisutti,  V.,  Carraro,  L.,  Larini,  I., and  Capolongo,  F. (2021). Natural contaminants in bee pollen: DNA metabarcoding as a tool to identify floral sources of pyrrolizidine alkaloids and fungal diversity. Food Res. Int., 146, 110438.
• International Programme on Chemical Safety (IPCS) (1988). Environmental health criteria 80 pyrrolizidine alkaloids. World Health Organization (WHO), Geneva.
•  Kwon,  Y.,  Goo  Y., and  Jeong,  Y. (2021). Determination of pyrrolizidine alkaloids in teas using liquid chromatography-tandem mass spectrometry combined with rapid-easy extraction. Foods, 10(10), 2250.
•  Mattocks,  A.R. (1986). “Chemistry and Toxicology of Pyrrolizidine Alkaloids.” Academic Press, London.
• National Institutes of Health (NIH). (2003). NTP technical report on the toxicology and carcinogenesis studies of riddelliine, (CAS NO. 23246-96-0) in F344/N rats and B6C3F1 mice. NIH Publication No. 03-4442.
•  Ryu,  H.J.,  Kim,  O.H.,  Lee,  E.S.,  Kim,  M.S.,  Kim,  J.G.,  Yun,  E.S.,  Kim,  H.J., and  Kim,  M.S. (2019). Characterization and screening of pyrrolizidine alkaloids. Anal. Sci. Technol., 32, 252-261.
•  Schaneberg,  B.T.,  Molyneux,  R.J., and  Khan,  I.A. (2004). Evaporative light scattering detection of pyrrolizidine alkaloids. Phytochem. Anal., 15, 36-39.
•  Valese,  A.C.,  Molognoni,  L.,  de Sá Ploêncio,  L.A.,  de Lima,  F.G.,  Gonzaga,  L.V., and  Górniak  S.L. (2016). A fast and simple LC-ESI-MS/MS method for detecting pyrrolizidine alkaloids in honey with full validation and measurement uncertainty. Food Control, 67, 183-191.
•  Valese,  A.C.,  Daguer,  H.,  Muller,  C.M.O.,  Molognoni,  L.,  da Luz,  C.F.P.,  de Barcellos Falkenberg,  D., and  Costa,  A.C.O. (2021). Quantification of pyrrolizidine alkaloids in Senecio brasiliensis, beehive pollen, and honey by LC-MS/MS. J. Environ. Sci. Health B, 56, 685-694.
•  Wit,  L. de,  Geraets,  L.,  Bokkers,  B., and  Jeurissen,  S. (2015). Pyrrolizidine alkaloïden in kruidenpreparaten. RIVM Briefrapport 090437001/2014. Rijksinstituut voor Volksgezondheid en Milieu (RIVM), Bilthoven.
•  Zhou,  Y.,  Li,  N.,  Choi,  F.F.K.,  Qiao,  C.F.,  Song,  J.Z., and  Li,  S.L. (2010). A new approach for simultaneous screening and quantification of toxic pyrrolizidine alkaloids in some potential pyrrolizidine alkaloid-containing plants by using ultra-performance liquid chromatography-tandem quadrupole mass spectrometry. Anal. Chim. Acta., 681, 33-40.
• i) http://www.foodnews.co.kr/news/articleView.html?idxno=70288 (June 13, 2019)