Scientific opinion on the risks for animal and human health related to the presence of quinolizidine alkaloids in feed and food, in particular in lupins and lupin‐derived products

Background

Quinolizidine alkaloids (QAs) mainly occur in lupin species and other plants of the Genisteae tribe. They are secondary metabolites for defense against pathogens and other predators. QAs are biosynthesized from lysine in green tissues of the plant and stored in all organs of the plant, including seeds (Boschin and Resta, 2013).

The level of QAs depends on genotype, presence of pathogens, environmental effects and soil characteristics. More than 170 QAs have been identified in different Lupinus species and the alkaloid pattern is highly variable among different species (Boschin and Resta, 2013).

Lupins and lupin‐derived products are listed in the Catalogue of feed materials.1

Lupin‐based products, such as lupin proteins, are gaining attention to replace animal proteins and other plants ingredients in several foods such as bakery products, imitation dairy and meat products, and beverages.

Terms of Reference

In accordance with Art. 29 (1) of Regulation (EC) No 178/2002, the European Commission asks the European Food Safety Authority for a scientific opinion on the risks for animal and human health related to the presence of QAs in feed and food, in particular in lupins and lupin derived products.

Extraction and purification

QAs can be extracted from ground lupin seeds with acidic solutions (0.1–1 N HCl or trichloroacetic acid) (Wink et al., 1995; Boschin et al., 2008; Kamel et al., 2016), optionally preceded by a hexane defatting step (Boschin et al., 2008). Liquid–liquid extraction of alkaline or alkalinised extracts with chloroform or dichoromethane using Extrelut columns yields an enriched extract containing the QAs (Wink et al., 1995; Boschin et al., 2008). Although less frequently, dilute sodium hydroxide solutions have also been used to extract QAs from the seeds (Hatzold et al., 1983). Direct extraction of ground seeds by maceration in the presence of organic solvents such as chloroform and diethyl ether has also been described (Pothier et al., 1998).

Titration and TLC methods

The total alkaloid content in seeds can be estimated by titration of the QAs in their salt free form, with p‐toluenesulfonic acid in combination with a pH colour indicator, e.g. tetrabromophenolphthalein ethyl ester (Ruiz, 1977) or bromocresol green (Ruiz et al., 1977). The limit of quantification (LOQ) of these methods is around 100 mg/kg seeds (Ruiz, 1977). Chloroform extracts obtained by the extraction procedure described above, upon titration gave a reasonable indication of the total alkaloid content (Ruiz, 1978). Nevertheless, a drawback of the method is that it will determine QAs together with other co‐extracted alkaloids present in lupin seeds.

Thin‐layer chromatography (TLC) can be used to obtain qualitative information on the QA composition of lupin seeds. Separation on silica gel plates and visualisation can be accomplished with Dragendorff's reagent and ultraviolet (UV) light (Muzquiz et al., 1994). A quantitative 2D‐TLC method has been described by Karlsson and Peter (1978).

GC‐based methods

Analytical methods based on capillary high resolution gas chromatography (GC) have been in widespread use since the early 1980s. Notably the research group of Wink has published a large number of papers on the analysis of lupin seeds and other plant parts using GC methods, culminating in their overview paper of 1995, in which the spectral data and QA compositions for 56 different lupin species are presented (Wink et al., 1995). In essence, the methods in use today are still very similar to the methods developed in the 1980s. Typically GC systems are equipped with either an aspecific flame ionisation detector (FID), a nitrogen specific detector (nitrogen‐phosphorus detector, NPD) or they are coupled to a mass spectrometer (MS), typically generating electron impact (EI) spectra (Boschin and Resta, 2013). By using specific non‐polar stationary phases (typically HP‐1 (DB‐1) or HP‐5 (DB‐5)) in combination with Kovats retention indexes, databases containing spectral information on a large number of QAs have been made available (Wink et al., 1995; National Institute of Standards and Technology (NIST) database).2 Detection limits (LOD) are generally not reported, as they are considered sufficient for the analysis of lupin seeds and other plant parts. However, the LOQ of the method becomes relevant for analysis of food products containing lupin seeds as an ingredient or for processed foods. Resta et al. (2008) and Boschin et al. (2008) reported for lupanine an LOQ of 2 mg/kg (LOD = 1 mg/kg) in flour. Reinhard et al. (2006) reported for lupanine, angustifoline and 13α‐OH‐lupanine, LODs of 1, 2 and 6 mg/kg, respectively, in flour.

Derivatisation of QAs is not necessary for analysis of lupin seeds, but it is sometimes used to obtain lower LODs, particularly in complex matrices, such as food (Reinhard et al., 2006). After derivatisation, the LOD for angustifoline improved to 0.3 mg/kg and for 13α‐OH‐lupanine to 0.5 mg/kg.

Due to the lack of available reference standards, quantification is often performed by comparison of relative peak areas to that of an available standard, e.g. sparteine or lupanine (Boschin and Resta, 2013). This may introduce considerable uncertainty in the reported concentrations (Resta et al., 2008).

LC‐based methods

Liquid chromatography (LC)‐based methods are not often used for the determination of QAs. Wysocka and Chrzanowska (2004) used an high‐performance liquid chromatography (HPLC) method with a chiral analytical column and UV detection at 220 nm to study the enantiomeric purity of lupanine in L. albus and L. angustifolius cultivars. The (+)‐ and (‐)‐enantiomers of lupanine could be separated on a Chiralcel OD column (Wysocka and Chrzanowska, 2004; Przybył and Kubicki, 2011). Green et al. (2015a) used a liquid chromatography coupled mass spectrometry (LC–MS) method to quantify the major QAs of Lupinus leucophyllus (anagyrine, lupanine, 5,6‐dehydrolupanine and an unidentified QA with mass 248) in serum of Holstein cows that had been orally dosed with dried L. leucophyllus plant material. The estimated LOQs were around 10 ng/mL in serum. Sparteine, lupinine and lupanine are sometimes included in multi analyte methods covering a wider range of plant toxins (Mol et al., 2011; Carlier et al., 2015). A conference abstract reports on the development of a liquid chromatography tandem mass spectrometry method (LC–MS/MS) for the major QAs in lupin flour (Mulder and de Nijs, 2015). QAs were extracted from the flour with an acidic aqueous solvent. The extracts were analysed without further clean‐up except membrane filtration. The LOQs were < 1 mg/kg, but no further details were disclosed. Vaněrková et al. (2014) also report in a conference abstract on the development of an LC–MS/MS method for lupanine, lupinine and sparteine in lupin seeds with LOQs of 2.6, 3.3 and 0.14 mg/kg, respectively.

Analytical standards, certified reference materials and proficiency testing schemes

The availability of commercial QA reference standards is limited. Lupanine, lupinine and sparteine are available from a number of different suppliers, but the other QAs are either available from a single source only or not available at all. Isotopically labelled analogues are also not available. No certified materials are commercially available and no proficiency testing schemes for testing of QAs in lupin seeds or lupin products have been identified.

3.1.1.1. Laboratory animals and humans

Except for sparteine, there is little or no information on the toxicokinetics and metabolism of QAs that may occur in feed and food. Sparteine, as sparteine sulfate, was used as a drug to treat cardiac arrhythmias and induce uterine contractions in labour, and as a phenotypic marker of cytochrome P450 (CYP) 2D6 drug metabolism activity (Thies, 1986).

3.1.1.2. Farm animals, horses and companion animals

There is little information in farm animals, horses and companion animals on the absorption, distribution, metabolism, and excretion (ADME) of the QAs considered in 1`this Scientific Opinion. Most studies deal with wild lupins from non‐European countries (i.e. L. caudatus, L. leucophyllus and L. argenteus) containing teratogenic alkaloids (e.g. anagyrine) which are of no interest for the current opinion. Such studies will be taken into consideration in the following section only in the case they provide information on the fate of QAs considered in this Scientific Opinion.

3.1.1.3. Transfer

In general, due to their weak basic nature, plant alkaloids, including QAs, are expected to be excreted in dairy milk. Although indirect evidence has been provided of the excretion of teratogenic QAs both in cow (Panter and James, 1990) and goat milk (reviewed in Molyneux and Panter, 2009), no information could be retrieved about the mammary excretion of lupanine, 5,6‐dehydrolupanine or other non‐teratogenic QAs. It is not known whether feeding of lactating farm animals with lupins from L. albus or L. angustifolius results in occurrence of lupin alkaloids or their metabolites in cow's, sheep's or goat's milk.

In the study by Vogt et al. (1987; see Section 3.1.4), there was no measurable transfer of alkaloids into meat or eggs of laying hens fed for 168 days a diet containing 16% of three different debittered lupin grists, namely L. albus 24 P., L. albus 24 G. or L. mutabilis, with a QA concentration ranging from 8 to 60 mg total alkaloids/kg grists, corresponding to 1.1–9.6 mg total alkaloids/kg diet. However, the Panel noted that the animal QA exposure via feed in this study is low and does not allow to assess possible transfer rates.

3.1.1.4. Main biotransformation pathways of quinolizidine alkaloids

Figure 4 summarises information on the biotransformation of sparteine and lupanine (mainly oxidative) in humans, pigs, rats and rabbits. There is no information on whether conjugated metabolites might be formed from sparteine and lupanine and their metabolites. For other QAs, a general picture of their biotransformation is lacking.

3.1.1.5. Summary remarks on toxicokinetics

3.1.2.1. Acute toxicity

An overview of the acute toxicity studies with QAs is presented in Appendix H.1.

d‐lupanine, isolated from L. angustifolius seeds and purified by recrystallisation (identification by refractive index; final purity not reported), was injected i.p. in mice, rats and guinea pigs at concentrations of 15 and 100 mg/mL. Minimal lethal dose (MLD) values were 75–85, 180–192 and 210–225 mg/kg bw in mice, rats and guinea pigs, respectively. The dose killing 50% of the animals (LD₅₀) for mice was 80 mg/kg bw. Fatally dosed animals suffered from paralysis, with intermittent convulsive movements, then complete exhaustion. Necropsy revealed that death resulted from asphyxiation (no apparent liver damage or other pathological lesions were noticed) and all animals seemed to die from respiratory failure (Gordon and Henderson, 1951).

Lupanine isolated from L. albus seeds and purified by column chromatography and recrystallisation (identification by 2D‐TLC and IR; final purity not reported) and commercial sparteine were administered orally or i.p. to male Swiss mice. Oral LD₅₀s were 410 and 220 mg/kg bw for lupanine and sparteine, respectively, while i.p. administration led to LD₅₀s of 175 and 36 mg/kg bw, respectively. Identical symptoms (trembling, tonic‐clonic seizures) were observed for both alkaloids and death was caused by breathing arrest. The lower toxicity of lupanine vs sparteine was confirmed in guinea pigs (intravenous (i.v.) administration), in which a MLD of 78 and 27 mg/kg bw was identified for lupanine and sparteine, respectively. In view of the similarity of the pharmacological properties of sparteine and lupanine, the authors conclude that the lower toxicity of lupanine is probably due to differences in bioavailability and/or pharmacokinetics (Yovo et al., 1984).

Oral LD₅₀ of alkaloid extracts from L. angustifolius (cv. Fest) seeds (alkaloid profile: 49% lupanine, 39% 13α‐OH‐lupanine, 10% angustifoline, 0.7% α‐isolupanine) was 2,279 mg/kg bw for fed rats and 2,401 mg/kg bw for rats fasted for 16–18 h before dosing. Lupanine and 13α‐OH‐lupanine were further purified from the alkaloid extracts by recrystallisation and column chromatography (purity by GC: > 99.5%). For lupanine, the LD₅₀ by oral administration to fasted male Wistar rats was 1,664 mg/kg bw. Administration of lupanine and 13α‐OH‐lupanine i.p. to rats identified LD₅₀s of 177 mg/kg and 199 mg/kg, respectively. After administration of a lethal oral dose of alkaloids, rats displayed nervous signs within 2–16 min and died within 2–23 min. Signs included tremor, followed by convulsions, cyanosis, collapse and death irrespective of the route of administration. Post‐mortem analyses identified congestion of the liver and lungs in some animals. Surviving rats from all treatments showed no signs of clinical toxicity (Petterson et al., 1987).

Stobiecki et al. (1993) extracted seeds of L. angustifolius and L. albus and these extracts (containing 10% alkaloids in the dry extract) were administered by gavage to male and female mice (inbred strain 129/AoBoy/Iiw, unspecified number of animals). No deaths were recorded and an LD₅₀ of > 4,000 mg lupin extract/kg bw (corresponding to > 400 mg alkaloids/kg bw) was established. The CONTAM Panel noted that it is not clear from the study how alkaloids were measured in these extracts.

A comparative study was conducted of the effects of extracts from L. mutabilis Sweet seeds (alkaloid profile: 20% sparteine, 59% lupanine and 21% 13α‐OH‐lupanine), lupanine isolated from L. mutabilis seeds and purified by column chromatography (purity checked by TLC and GC–MS) and sparteine sulfate pentahydrate. Five doses (50, 100, 125, 250 or 500 mg/kg) were i.p. injected in male Swiss mice and acute toxicity and maximum non‐lethal dose (LD₀) were determined 8 days later (Pothier et al., 1998). LD₀s for i.p. administration were 25 mg/kg for sparteine and 64 mg/kg for lupanine and lupin extracts. Lethal i.p. doses (LD₁₀₀) were 100 mg/kg for sparteine and 250 mg/kg for lupanine and lupin extracts. The clinical signs of poisoning (shaking, excitation and convulsions) were the same for the three compounds.

Summary remarks on acute toxicity

Mice appear to be more sensitive than rats to acute toxicity (by i.p. and oral administration) of lupanine. A comparison of the acute toxicity of lupanine and 13α‐OH‐lupanine in rats (by i.p. administration) showed similar LD₅₀ values. In contrast, when comparing results from the same route of administration, several reports on mice and guinea pigs indicate that sparteine is approximately two‐ to three‐fold more acutely toxic than lupanine, regardless of the way the acute toxicity is expressed. Finally, acute symptoms following oral administration of these QAs are similar in rats, mice and guinea pigs, with death resulting from respiratory failure and toxic effects affecting the central nervous system (CNS).

3.1.2.2. Repeated dose toxicity studies

The CONTAM Panel identified five repeated dose toxicity studies in which rats received a lupin‐based diet. No studies in other experimental animals were identified. These studies are described below in detail and are summarised in Appendix H.2. No repeated dose toxicity studies on individual QAs were found. In addition to the papers described below, several other papers were identified which were considered of low relevance for the risk assessment due to a lack of information on the alkaloid concentration in the feed, poor reporting or unsuitable experimental design which focused on studying nutritional benefits of lupin seeds or mechanistic insights in experimental models of human disease. A description of these studies is provided in Appendix H.3.

In a 112‐day feeding study, three groups of Charles River rats (21–23 day old) were given a diet containing casein (control group), ground seeds (58%) of L. albus (510 mg lupin alkaloids/kg, equivalent to 45 mg/kg bw per day) and ground seeds (51%) of L. luteus (910 mg lupin alkaloids/kg, equivalent to 81 mg/kg bw per day, no specification of QAs; total lupin alkaloid content measured by titration) (Ballester et al., 1980). Food and water were available ad libitum and dietary intake and body weights were recorded weekly. Animals fed either lupin seeds at 10% protein diets without dl‐methionine performed poorly (low growth and protein efficiency ratio). In contrast when 20% protein diets supplemented with 0.3% dl‐methionine were compared L. luteus‐fed animals gained weight at rates similar to those fed with casein and a slight, but not statistically significant decrease was observed for L. albus‐fed animals. At the end of the 112‐day feeding period, no differences among the three groups were found in organ‐to‐body weight ratios of liver, kidneys, spleen, heart and adrenals. Histopathological examinations of liver, kidneys and lungs did not reveal any significant differences with the control group. The CONTAM Panel noted that no effects were observed at the only tested dose levels of 45 and 81 mg lupin alkaloids/kg bw per day, for L. albus and L. luteus, respectively.

A diet based on flour (33.6%) prepared from water‐debittered grains of L. mutabilis and containing trace amounts of lupin alkaloids (160 mg/kg, equivalent to 14.4 mg/kg bw per day as determined by GC‐NPD, no specification of QAs) was tested in Sprague–Dawley rats (21 days old) for 12 weeks (Schoeneberger et al., 1987). The protein content was 15% and the diet was supplemented with 0.2% dl‐methionine. The control group was fed a 15% protein casein diet. Rats fed on lupin diet did not show any statistically different growth performances (body weight, food consumption) in comparison to the control group. No consistent indications of significant physiological differences (haematological and biochemical parameters) were recorded. Similarly, absolute and relative weights and histopathological examinations of some organs (liver, kidney, heart, spleen and brain) were unaffected by the alkaloids. The CONTAM Panel noted that no effects were observed at the only tested dose level of 14.4 mg lupin alkaloids/kg bw per day for L. mutabilis.

Diets based on L. angustifolius (cv. Fest) flour (554 g/kg diet) which had been spiked with extracts from L. angustifolius seeds to provide dietary concentrations of 250, 1,050 and 5,050 mg lupin alkaloids/kg diet (equivalent to 25, 105 and 505 mg/kg bw per day) was fed to three groups of Sprague–Dawley rats (3 weeks old) for 90–98 days (Butler et al., 1996). No control group was included but a reference group receiving the same lupin flour‐based diet containing background levels of alkaloids (50 mg lupin alkaloid/kg diet, corresponding to 4.5 mg/kg bw per day) derived from the background level of alkaloids in lupin flour. The alkaloid profile of the flour was lupanine (44.2–44.6%), 13α‐OH‐lupanine (41.9–42.4%), angustifoline (9.9–10.9%) and α‐isolupanine (1.0%) (analytical method not reported). The diets were supplemented with 0.6% methionine. No mortality was recorded during the study. A slight transient reduction in body weight was recorded in the male group treated with 250 mg/kg diet. Food intakes were generally equivalent to those of the reference group. Haematological investigations gave little evidence of consistent effects (changes were small, often of transient nature and characterised by lack of dose response). Clinical chemistry measurements revealed no differences between lupin‐fed rats and the reference group. Relative liver weights were increased in female rats at all doses (4% increase and no dose response). Altered foci of liver parenchymal cells were seen in 1/20 females and 1/20 males in the lowest dose group, 1/20 males in the mid‐dose group, 5/20 females in the highest dose group and none in the reference group. Cells comprising these foci were basophilic or large pale‐staining cells, characterised by nuclei with prominent nucleoli and occasional mitotic figures (usually called altered hepatic foci). However, the Panel noted that these effects were not observed in a follow‐up study described below.

Diets mixed with extracts from L. angustifolius (cv. Fest) containing lupin alkaloids at dose levels of 0, 100, 330, 1,000 or 5,000 mg/kg (equivalent to 0, 10, 33, 100 and 500 mg/kg bw per day) were fed to groups of Sprague–Dawley rats (4 weeks old) for 90 days (Robbins et al., 1996). The alkaloid profile consisted of lupanine (49.5–50.7%), 13α‐OH‐lupanine (38.4–39.2%), angustifoline (8.8–10.7%) and α‐isolupanine (1.3–1.4%) (analytical method not reported). The diets were supplemented with 0.3% methionine. The control basal diets were commercial preparations with and without 4.5% maltodextrin. A statistically significant decrease in body weights was observed in both male and female rats fed a diet containing 500 mg/kg bw per day lupin alkaloids in comparison to the control diet groups. The body weights of both male and female rats fed 100 mg/kg bw per day were also lower, although the difference was statically significant only on a few time points. Feed intakes of both sexes given 100 or 500 mg/kg bw per day were consistently lower than the control groups. Some changes (with no dose response) were observed in haematological parameters and in clinical chemistry findings. The authors noted that these fell within the normal range according to age and sex of the animals. Relative liver weights at 500 mg/kg bw per day in both males and females were significantly higher (8–9%) than in control groups. Kidney weights showed some inconsistent and dose‐unrelated differences between sexes. The histopathological examination of liver, heart and bone marrow was unremarkable and none of the altered foci of liver cells reported by Butler et al. (1996) were reproduced in this study. The CONTAM Panel noted that the reduced body weight gain already recorded at 100 mg/kg bw per day might be due to reduced palatability and feed intake. Based on the increased relative liver weight observed at 500 mg lupin alkaloids/kg bw per day, the CONTAM Panel identified 100 mg lupin alkaloids/kg bw per day as a dose of lupin alkaloids present in the studied L. angustifolius cultivar that is not associated with effects is rats.

Diets containing seeds of L. angustifolius were administered to male Wistar rats (3 weeks old, 81 g initial body weight, 8 animals/group) for 28 days. Diet materials were prepared from three L. angustifolius cultivars (25.8–27.1%), i.e. Baron, Zeus and Wersal, with different lupin alkaloid concentrations (360, 410 and 560 mg/kg, respectively) (Stanek et al., 2015). The composition of QAs (measured by GC–MS) was lupanine (62.4–70.2%), 13α‐OH‐lupanine (9.5–14.4%), angustifoline (8.7–10.6%), α‐isolupanine (6.6‐8.1%), tetrahydrorhombifoline (2.5–4.1%) and isoangustifoline (1.0–1.9%). The composition of the experimental diet was 10% protein and an alkaloid concentration of 90, 110 and 150 mg/kg feed (equivalent to 8.1, 9.9 and 13.5 mg/kg bw per day for Baron, Zeus and Wersal feed), respectively. The CONTAM Panel noted that previous experiments with 10% protein without dl‐methionine supplementation showed that animals performed poorly (see Ballester et al., 1980). The control group received soybean meal. Diets containing the seeds of all L. angustifolius cultivars reduced the feed intake by 50% and the growth rate of experimental rats in comparison to the control group. Baron‐, Zeus‐ and Wersal‐based diets increased serum triacylglycerol (TG) levels, while decreased alanine transaminase (ALT) activity was observed only in Zeus‐ and Wersal‐based ones. The liver weight of rats fed diets with L. angustifolius seeds decreased with increasing dietary alkaloid concentrations. Liver histopathology revealed parenchymatous degeneration of hepatocytes (more pronounced in rats fed Zeus and Wersal diets), and congestion of portal vessels and central veins (more severe in rats fed Baron and Zeus diets). The CONTAM Panel disregarded this study because of the nutritional imbalance and the large effects on feed intake and growth rate induced by relatively low levels of alkaloids from L. angustifolius.

Summary remarks on repeated dose toxicity studies

All the repeated dose studies with laboratory animals on QAs were carried out in rats. The CONTAM Panel identified two key 90‐ to 98‐day studies performed in the same laboratory using L. angustifolius‐based flour. The major QAs of the diet were lupanine and 13α‐OH‐lupanine (about 40% each). Both studies reported an increase in relative liver weights. However, the induction of altered liver foci observed in the first study was not confirmed in the second one. Conflicting results were also obtained in the two studies on weight gains. Based on increased relative liver weight a NOEL of 100 mg alkaloids/kg bw per day was proposed in the second study.

No significant toxicological effects were reported in sub‐chronic studies on L. albus‐, L. luteus‐and L. mutabilis‐based diets at doses of 45, 81 and 14.4 mg alkaloids/kg bw per day, respectively, which were the only doses tested. It should be noted that lupanine is the major alkaloid contained in L. albus and L. angustifolius, while the more toxic sparteine is a major alkaloid present in L. luteus and L. mutabilis (see Table 1).

The PANEL noted that the only available studies on QAs repeated dose toxicity are in rats, the less sensitive species among experimental animals (see acute toxicity section). In addition, it is difficult to compare these studies because of differences in the sources of QAs (seed extracts vs flour), cultivars, presence of alkaloids other than QAs relevant for this opinion, and the control diets (commercial, lupin‐based) used for comparison. Because of these reasons and some conflicting results reported in different publications from the same laboratory, the CONTAM Panel considered that these studies have a limited value for the evaluation of QA toxicity.

3.1.2.3. Neurotoxicity

In an acute toxicity study in mice by Pothier et al. (1998), the effects of i.p. administration of lupanine and sparteine (64 and 25 mg/kg, respectively, representing the maximum non‐lethal i.p. doses) on behaviour (i.e. psychomotor activity and exploration test), were investigated. Both alkaloids elicited a weak decrease in spontaneous activity and reduced locomotor activity indicating a slight sedative action on the CNS. Also, the interaction with drugs acting on the CNS and analgesic effects were studied. No effects were seen on the interactions with drugs such as stimulating (amphetamine, pentetrazol) or depressant (pentobarbital, chlorpromazine) drugs. Similarly, these alkaloids did not show any analgesic activity (on abdominal constrictions induced by acetic acid).

The effects of the intracerebroventricular (ICV) administration of QAs extracted from L. exaltatus and L. montanus seeds were studied in adult Swiss‐Wistar rats (10 animals/group) receiving a daily ICV injection (10 μL) of sesame oil (control group) or QA extract on five consecutive days (Banuelos Pineda et al., 2005). Principal QAs present in L. montanus extracts were sparteine (35,000 mg/kg), lupanine (17,700 mg/kg), 3β‐OH‐lupanine (3,800 mg/kg) and 13α‐OH‐lupanine (3,200 mg/kg). Extracts from L. exaltatus contained lupanine (5,800 mg/kg), 3β‐OH‐lupanine (1,500 mg/kg) and only a small amount of sparteine (284 mg/kg). Clinical symptoms immediately after QA administration included tachycardia, tachypnoea, piloerection, tail erection, muscular contractions, loss of equilibrium, excitation, and unsteady walk. The contralateral unlesioned left hemisphere of the brain was analysed at the end of the treatment. A high number of neurons with degenerative alterations (30–40% vs ~ 10% in the control group) were observed in the brains of QA–treated animals, thalamus, hypothalamus, and hippocampus being the most susceptible brain areas. These structures contain cholinergic neurons, structures related to cholinergic paths and structures of the basal ganglion that possess cholinergic receptors (muscarinic or nicotinic type). These neurotoxic findings support the mode of action of QAs in the CNS proposed by Kinghorn and Balandrin (1984) and Schmeller et al. (1994). The authors conclude that neuronal damage induced by QAs from L. montanus was more severe than that of L. exaltatus (no quantification reported) in all the examined areas. These results are in agreement with the high content of sparteine present in QA extracts from L. montanus.

Following sparteine administration either ICV or i.p. in Wistar rats (8 animals/group) for 5 days, Meraz Medina et al. (2017) found that brain structures of the CNS showing toxic effects were those mainly related to cholinergic pathways (frontal, frontoparietal and striate cortex, olfactory and amygdaloid area, ventromedial hypothalamic nucleus, cerebellar Purkinje cells and the Cornus Ammonis area 1 and 3 dentate gyrus regions of the hippocampus).

Sparteine‐induced brain damage was investigated in neonatal Wistar rats (8 animals/group) receiving a single dose of 25 mg/kg bw by subcutaneous injections at postnatal day 1 and 3. Sparteine‐induced damage to neurons in the cerebral motor cortex (analysed at 7, 14, 21 and 60 days) was visible at 7 days and persisted up to 21 days. Neuronal death was due to necrosis and was significantly reduced by pretreatment with atropine suggesting that cytotoxicity of sparteine is mediated by muscarinic acetylcholine receptors (mAChRs). Transient decreases in the expression of m1–m4 subunits mAChR both at RNA and protein levels were indeed demonstrated (Flores‐Soto et al., 2006).

Yovo et al. (1984) investigated comparatively the toxicodynamics of sparteine and lupanine. Cats (n = 4) and dogs (n = 4) were injected i.v. with sparteine (cats: 0.5, 1.0, 1.5, 2.0 and 4.0 mg/kg bw; dogs: 5.0 and 7.5 mg/kg bw) and lupanine (cats: 1.0, 2.0, 3.0, 4.0 and 5.0 mg/kg bw; dogs: 5.0 and 7.5 mg/kg bw). Sparteine is more efficient than lupanine in antagonising secondary reflex hypertension caused by carotid occlusion and hypotension induced by electrical stimulation of pneumogastric nerve both in the cat and in the dog. Lupanine and sparteine show identical inhibitory action on nicotinic type hypertension induced by injection of acetylcholine (ACh) in atropine‐treated dogs.

The CONTAM Panel noted that the method of application in the above described studies (in particular ICV administration) might be of limited relevance for assessment of the toxicity of QAs via the diet. However, they give an indication of the neurotoxic effects that QAs may cause and the related mode of action.

3.1.2.4. Genotoxicity

No binding or intercalation with DNA (as measured by DNA methylgreen release from DNA and inhibition of DNA polymerase I) was identified for several QAs (anagyrine, angustifoline, cytisine, 3β‐OH‐lupanine, 13α‐OH‐lupanine, lupanine, lupinine, 17‐oxosparteine, sparteine, 13α‐tigloyloxylupanine) (Wink et al., 1998).

At variance with alkaloids belonging to other chemical classes (e.g. indole and quinoline alkaloids) which bind/intercalate in DNA, cytisine, lupanine and sparteine were unable to induce apoptosis as identified by DNA fragmentation, in HL60 cells (Rosenkranz and Wink, 2007).

Lupanine did not induce gene mutations in Salmonella Typhimurium strains TA97, TA98, TA100, TA102, TA1535, TA1538 in the presence or absence of exogenous metabolic activation (S9) (Santiago Quiles et al., 2010).

In the same study, negative results were also reported in these S. Typhimurium strains with ethanol extracts of L. termis seeds (the authors state that lupanine is the main alkaloid present in L. termis). Ethanol extracts of L. termis were also negative for mutation induction in L5178Y tk+/− mouse lymphoma cells and for micronuclei formation in the bone marrow of BalbC mice administered a single i.p. dose of L. termis extracts (635 mg/kg bw corresponding to 50% of the LD₅₀) (Santiago Quiles et al., 2010).

Sparteine ability to induce single‐strand breaks (SSBs) was measured by Comet assays in a non‐standard assay based on cultures of Tradescantia staminal nuclei. A small increase in breaks was observed at a single dose. In the same study, ethanol extracts from L. mexicanus and L. montanus increased SSBs in the same cells at all tested concentrations, but with no dose dependency (Silva et al., 2014). The CONTAM Panel noted the limitations of the study, i.e. lack of dose response, no information on toxic effects and the use of a non‐standard assay.

Unpublished data (BIBRA, 1986) showing negative mutagenic effects in S. Typhimurium of alkaloid preparations from L. angustifolius have been mentioned in reviews (Petterson, 1998; Pilegaard and Gry, 2008) (the original report of BIBRA was however not available and thus the information could not be verified by the CONTAM Panel).

3.1.2.5. Developmental and teratogenicity studies

Ballester et al. (1984) studied the effects of lupin feeding for 9 months in a multigeneration study in Wistar rats. The lupin diet containing 51.8 g flour from L. albus per 100 g diet (supplemented with 0.2% dl‐methionine) was compared to a control diet containing the same protein levels (20%) obtained by defatted soybean flour, fishmeal and dried skimmed milk. The total alkaloid level of the test diet was 250 mg/kg (corresponding to 12.5 mg/kg bw per day). The results for the F0 generations have been reported in a previous study (Ballester et al., 1982; see Appendix H.3). F1 and F2 generations (descendants of the parent F0 generation mated at 12 weeks of age) were given the same diet as their parents. The growth rate of males fed the lupin diet both at the F1 and F2 generations was higher than that of the control group. No differences were observed in females. No significant changes in the relative weights of heart, spleen, kidneys, brain and gonads were observed with the exception of relative liver weights of both lupin‐fed male and female rats being significantly lower than control animals. The histology of the liver, as well as that of the other organs, was, however, normal. There were no treatment‐related haematological changes or changes in ALT and aspartate transaminase (AST) at any stage in either generation. Reproductive performance was also unaffected by a lupin‐containing diet. The authors proposed that the difference in relative liver weights between lupin‐fed male rats and control groups is due to the unusually high liver weights of the control group (not observed in previous studies from the same group using a casein diet as control group).

Kennelly et al. (1999) tested QAs extracted from rhizomes of blue cohosh (Caulophyllum thalictroides) in in vitro rat embryo cultures. Following methanol extraction and purification of the extracts, QAs were analysed by GC–MS. The assay is an assessment of the overall growth and morphogenesis of organogenesis‐staged rodent embryos. In this assay, rat embryos are dissected from the dams at gestation day 9.5 and cultured in medium together with test compounds for 45 h. α‐Isolupanine and N‐methylcytisine isolated from the extract were tested (purity not specified). α‐Isolupanine was negative at all concentrations tested (0, 5, 20 and 80 μg/mL). In contrast, N‐methylcytisine (tested at 5, 20, 50, 80 and 250 μg/mL) caused major malformations (open anterior neural tube, disturbed eye development and twisted tail) from 20 μg/mL onwards. Anagyrine (tested at 5, 20, 80 and 250 μg/mL) inhibited overall morphogenesis at the highest concentration tested but did not cause teratogenic effects (it should be noted however that skeletal development is not scored in this assay).

In conclusion, a two‐generation study in rats indicate that a diet based on L. albus seeds did not affect the reproductive performance. In addition, an in vitro study in rat embryos assessing overall growth and teratogenic potential of purified α‐isolupanine did not cause major malformations. No information is available on the effects on development and the teratogenic potential of other QAs.

3.1.2.6. Carcinogenicity

The CONTAM Panel did not identify carcinogenicity studies.

3.1.3.1. Pharmacodynamics

Among the QAs, sparteine is the only one for which human data from previous medicinal uses as an antiarrhythmic or oxytocic drug are available (Thies, 1986). However, existing data are limited.

Sparteine19

Knowledge from use as an antiarrhythmic agent

Sparteine has been shown to diminish irritability and conductivity of the heart muscle as a sodium channel blocker in a comparable manner to quinidine and has been used therapeutically since 1873 for the treatment of cardiac arrhythmias (Thies, 1986; Blaschek et al., 2006). Thies (1986) describes that single doses of 15 to 20 mg sparteine (equivalent to 0.21 to 0.29 mg sparteine/kg bw for a person with a body weight of 70 kg) were introduced in the 1920s in the treatment of cardiac arrhythmias in France. However, Thies (1986) describes this dosage as below the threshold of pharmacological activity. According to Blaschek et al. (2006), antiarrhythmic properties have been reported for sparteine sulfate in an oral dose of as low as 20 mg (equivalent to 0.29 mg sparteine sulfate/kg bw for a person with a body weight of 70 kg)25 without clinical studies being available. In later years, (‐)‐sparteine sulfate was used in the treatment of tachycardia in single doses of 100 mg/tablet or 200 mg/tablet as a Class I antiarrhythmic (Rote Liste, 1983; Thies, 1986; Blaschek et al., 2006). The maximum single therapeutic dose was indicated to be at 4 mg/kg bw (Thies, 1986; Blaschek et al., 2006). According to clinical experiences, a single dose of 100 mg (equivalent to 1.4 mg/kg bw for a person with a body weight of 70 kg)26 still could not ensure a reliable antiarrhythmic effect. However, this could be achieved with a single oral dose of 150–200 mg (equivalent to 2.1–2.9 mg/kg bw for a person with a body weight of 70 kg). The recommended daily dosage for (‐)‐sparteine sulfate ranged from 800 to 1,000 mg per day for short‐term use and from 400 to 500 mg per day for long‐term therapy, the daily doses always being divided into 4–5 single doses (Rote Liste, 1983). Individual variations in sensitivity to sparteine sulfate are explained by the genetic polymorphism, which the isoenzyme CYP2D6 underlies. This isoenzyme is responsible for the metabolic oxidation of sparteine to 2,3‐dehydro‐ and 5,6‐dehydrosparteine (Eichelbaum et al., 1979a; Thies, 1986; see also Section 3.1.1.1). Due to the existing genetic polymorphism, these two metabolites cannot be formed by 5% of the population who thus have higher sparteine plasma levels than the normal population and who excrete more than 95% of the perorally administered dose as unchanged sparteine with the urine. These PMs were considered as a group at higher risk to develop adverse effects, which include gastrointestinal disturbances, central nervous disorders, dizziness, head pressure, accommodation disturbances in the eye, bradycardia, negative inotropy, hypertonia, changes in haemogram and liver enzymes. Use during pregnancy is considered as contraindication (Rote Liste, 1983; Thies, 1986; Blaschek et al., 2006; Aktories et al., 2009).

From the interindividual differences in sensitivity between PMs and EMs, the Panel concludes that the pharmacological activity is expected to be associated with the parent compound (‐)‐sparteine and lost upon biotransformation.

Knowledge from use as an oxytocic agent

Sparteine exhibits oxytocic properties, leading to a muscular stimulation in the virginal, puerperal and gravid uterus. During the expulsive phase of birth, sparteine induces rhythmic contractions of the uterus (Thies, 1986).

Dipont investigated the effect of orally administered sparteine sulfate on uterine contractility and if the treatment reduces the duration of labour during childbirth (Dipont, 1971). Sparteine sulfate was given to 427 women in the first stage of labour in the form of a 20% solution in drops in single doses of 100–150 mg by repeating dosage every hour if needed, up to a total dose of 600 mg. Compared to the control group, which comprised 98 labouring women, a statistically significant shortening of the duration of the first stage of labour was achieved by the treatment. The shortening amounted on average to 2 h and 8 min and was due to pronounced oxytocic activity. The average dose of sparteine sulfate administered was 402.5 mg. In this study the use of sparteine sulfate did lead neither to an increase of operative deliveries, nor to an increase in the number of perinatal complications, nor did it have a negative effect on the fetus. An adverse effect reported for the pregnant women was vomiting, which was associated by the author with the bitter taste of the sparteine sulfate (Dipont, 1971).

Sparteine sulfate was administered by intramuscular injection in repeated doses of 100–150 mg (equivalent to 1.4–2.1 mg/kg for a body weight of 70 kg) to intensify labour (Gessner and Orzechowski, 1974). However, due to the unpredictable occurrence of ‘tetanic’ uterine contractions, the use of sparteine sulfate as an oxytocic was banned by the FDA (1979).

Intoxication

The symptoms of intoxication with sparteine, for which PMs are at higher risks, are dizziness, drowsiness, headache, sweating, mydriasis and myasthenia. Via the paralysis of the phrenic nerve endings, sparteine produces in oral doses from 40 mg/kg bw onwards peripherally a curare‐like effect which may lead to respiratory arrest and death. In cases of intoxication, where peripheral respiratory paralysis has been overcome by artificial ventilation, higher oral doses from 90 mg sparteine/kg bw onwards, induce bradycardia with the risks of fatal outcome due to cardiac arrest (Thies, 1986; Blaschek et al., 2006; Aktories et al., 2009).

A child of the age of 2 years and 4 months died 3 h after accidental ingestion of 413 mg of sparteine (according to the author equivalent to approximately 30 mg of sparteine/kg bw) in the form of a medicinal product. After 2 h, the child became conspicuously tired. Respiratory paralysis was observed followed by intense pulse acceleration. Tonic seizures occurred repeatedly before death set in (Schmidt, 1961).

A 4‐year‐old boy with a body weight of 15 kg was intoxicated after accidental ingestion of 100 mg sparteine sulfate/kg bw27 in the form of tablets. A 4‐year‐old boy with a body weight of 15 kg was intoxicated after accidental ingestion of 100 mg sparteine sulfate/kg bw (28) in the form of tablets. Three hours later and after admission to the hospital, extrasystoles, a heart rate of 95 per min and gasping were observed, followed by vomiting and cardiorespiratory arrest. The boy survived after cardiopulmonary resuscitation was delivered (Landes and Robl, 1967).

3.1.3.2. Human studies

Lupanine, 13α‐OH‐lupanine

In a randomised kinetic study with a cross‐over design described in detail in Section 3.1.1.1, oral doses of 10 mg of lupanine or 10 mg 13α‐OH‐lupanine (each equivalent to 0.143 mg/kg bw for a person with a body weight of 70 kg) were administered with a two‐week wash‐out phase to eleven volunteers. With respect to the expression of CYP2D6, seven of them were classed as EMs and four as PMs. No adverse effects were observed in any of the test subjects, blood pressure and heart rate remaining unaffected (Petterson et al., 1994).

3.1.3.3. Case reports on intoxication with lupin seeds

From the standard literature, it is known that symptoms of lupin seed intoxication set in 20 min after consumption with maximum effects being observed after 4–5 h. Typical symptoms are of neurological nature and may additionally manifest in the digestive and/or cardiovascular systems. Moderate cases of intoxication represent themselves by mydriasis, dizziness, nausea, dryness of the mouth, stomach aches, vomiting, diarrhoea and/or cardiac complaints. For severe cases of intoxication prominent tiredness accompanied by curare‐like paralysis and convulsions 2–3 h after intake have been described. Death can occur by respiratory paralysis or cardiac arrest and also by choking (Schmidlin‐Mészáros, 1973; Blaschek et al., 2006; Forrester, 2006).

In general, the alkaloid pattern of the lupin seeds, which were reported to induce the intoxication, is unknown and often the botanical origin of the seeds (e.g. lupin species and variety) is not indicated.

According to the available literature, including review articles and risk assessment monographs published by Food Authorities, the number of known cases of intoxication associated with the intake of lupin seed‐based food is limited (e.g. Schmidlin‐Mészáros, 1973; ANZFA, 2001; Pilegaard and Gry, 2008; BfR, 2017a).

In the following, focus is put on case reports on poisonings with relevance for the risk assessment of the intake of QAs with lupin seed‐based food, e.g. giving information on dose effect relationships.

More than 10 published reports on poisoning cases in humans with seeds of the genus Lupinus or preparations thereof have been identified during the literature search (see Section 2.1). However, the CONTAM Panel is aware that this list may not be complete. The few cases which allow for estimation of alkaloid doses, to which subjects have been exposed, are summarised in Table 2. Other cases for which a quantification of alkaloid exposure was not possible (Smith, 1987; Marquez et al., 1991; Lowen et al., 1995; Tsiodras et al., 1999; Di Grande et al., 2004; Sarikaya and Tettenborn, 2006; Litkey and Dailey, 2007; Kurzbaum et al., 2008; Pingault et al., 2009; Awada et al., 2011; Jamali, 2011; Daverio et al., 2014) are not dealt with here in detail. In some of the case reports, authors report on inadequate debittering of lupin seeds before consumption (Smith, 1987; Lowen et al., 1995; Litkey and Dailey, 2007; Kurzbaum et al., 2008; Awada et al., 2011; Jamali, 2011; Daverio et al., 2014). One case of intoxication was even associated with the drinking of 0.5 L water from debittering of lupin seeds (Marquez et al., 1991). Schmidlin‐Mészáros (1973) and Schmidt (1961) summarise three cases from older literature of lethal intoxication of children after the intake of lupin seeds (see Table 2). Furthermore, three cases of poisoning with higher alkaloid dose estimates per kg bw are described in adults, who survived. This suggests that children may be more susceptible to intoxication with lupin alkaloids than adults and that doses from 10 mg/kg bw onwards have to be considered to be lethal in children (Schmidt, 1961; Schmidlin‐Mészáros, 1973; Malmgren et al., 2016).

The available data do not allow drawing conclusions on the botanical origin of the lupin seeds involved in intoxication. However, Schmidlin‐Mészáros (1973) and Awada et al. (2011) name L. albus or L. luteus, respectively, as species whose seeds have led to cases of poisoning.

Overall, it can be concluded that intoxication with lupin seeds do not occur frequently and rarely lead to a fatal outcome. In poisoning cases, the estimated total alkaloid intakes range from about 10–50 mg/kg bw. For children, lethal cases of poisoning have been described from 10 mg/kg bw onwards. In several publications, intoxication have been associated with inadequate debittering of lupin seeds by consumers (Smith, 1987; Lowen et al., 1995; Litkey and Dailey, 2007; Kurzbaum et al., 2008; Awada et al., 2011; Jamali, 2011; Daverio et al., 2014; Malmgren et al., 2016). No case of poisoning has been identified which was attributed to consumption of industrially produced lupin seed‐based food. The botanical origin of the lupin seeds involved in intoxication remains unclear in most case reports. However, Schmidlin‐Mészáros (1973) and Awada et al. (2011) indicate that seeds of L. albus and L. luteus, respectively, have led to cases of poisoning.

3.1.4.1. Ruminants

Cattle

A study (Johnson et al., 1986) was conducted on 12 Jersey heifers (average weight 107 kg), which were randomly allocated to two groups receiving either a diet containing soybean (control) or a diet containing Tifwhite‐78 lupin seeds (19.3%), a low alkaloid cultivar of L. albus, for 70 days. The seeds contained 680 mg lupin alkaloids/kg (dry matter (DM)) (measured by a GC method) and the resulting dietary concentration was 131 mg lupin alkaloids/kg feed (DM). Based on the reported dietary intake and weight, the CONTAM Panel calculated a corresponding dose of 4.5 mg/kg bw per day. Blood sampling was performed after 35 and 70 days of feeding. Feed consumption and body weight were measured daily. No statistically significant differences were observed in dietary intake, weight gain and feed efficiency. In addition, the lupin diet did not induce significant changes in serum biochemistry (Ca, P, Mg, total protein, albumin, BUN, total bilirubin, AP, AST, GGT, LDH). The CONTAM Panel concluded that heifers can tolerate lupin alkaloids present in the studied L. albus cultivar at a dietary concentration of 131 mg/kg, corresponding to a dose of 4.5 mg/kg bw per day. The CONTAM Panel noted that this was the only dose tested.

According to a randomised complete‐block design, 25 Friesian dairy cows were offered one of the following diets with seeds from L. albus for 70 days: control, intact crushed lupin seeds (ICL) at 150 (ICL‐15) or 300 (ICL‐30) g/kg diet, and heat detoxified crushed lupin seeds (DCL) at 150 (DCL‐15) or 300 (DCL‐30) g/kg diet. Total alkaloids were extracted,28 but only the sum of the lupanine and 13α‐OH‐lupanine was measured (GC–MS); in ICL‐15 and ICL‐30 it was 3,800 and 8,000 mg/kg diet DM, respectively, while that of DCL‐15 and DCL‐30 was 2,100 and 5,200 mg/kg diet DM, respectively. Based on the reported dietary intake and assuming a body weight of 650 kg, the CONTAM Panel calculated corresponding doses of 72, 123, 52 and 111 mg/kg bw per day. Irrespective of the nature of the seeds (intact vs heat inactivated), cows exposed to the higher lupanine and 13α‐OH‐lupanine dosages (72, 111, and 123 mg QA/kg bw/day) had a lower feed intake (12.3, 13.9 and 10.0 kg/day), respectively. In addition, feed intake and milk yield were significantly depressed in cows on intact lupins as compared to those on detoxified lupins (Mukisira et al., 1995). The authors concluded that the reduction in feed intake could be attributed to lupanine and 13α‐OH‐lupanine. Based on this endpoint, it may be concluded that dairy cows tolerate a diet containing lupin seeds from L. albus at a concentration of 2,100 mg lupanine and 13α‐OH‐lupanine/kg, corresponding to a dose of 52 mg/kg bw per day.

Sheep

In the in vitro experiments cited above (Lanca et al., 1994), rumen microbial activity was not affected by the incubation for 24 h of either sparteine up to 5.4 mM or of finely ground L. luteus seeds (cv. Brava (bitter) or Topaz (sweet)) at three levels of incorporation (5, 10 or 15 g DM basis).

No relevant studies could be identified concerning the exposure to cultivated lupins.

Goats

No relevant studies could be identified concerning the exposure to cultivated lupins.

3.1.4.2. Pigs

Based on a review of the scientific literature, Cheeke and Kelly, 1989 concluded in that pigs are reported to be more susceptible than other species to QAs contained in SLs.

Case reports of lupin toxicosis have been described in pigs. Feed refusal, unthriftiness, lethargy, sporadic deaths, reduced litter size, stillbirth, increased postnatal death losses, poor performances and intestinal stasis were reported by three different pork producers feeding growing piglets or sows with dehulled hexane‐extracted lupin seeds (most likely L. albus cv. Ultra, alkaloid content not reported) at dietary concentrations from 10% to 30% (Casper et al., 1991).

More recently (2016), a field case of lupin toxicosis has been described in northern Italy, affecting about 2,170 pigs of different categories (pregnant or lactating sows, gilts, fattening pigs) and involving several breeding farms. Poisoning was due to accidental contamination of grain legumes (peas) by bitter lupins (BLs) (L. albus). Total QA content of feed was determined by the University of Milan with a GC–MS method. There was a clear relationship between the QA concentration in the feed and the severity of clinical signs. A decrease in feed consumption was already noted after the exposure to dietary QA concentrations of 194 mg/kg; a more marked decline in feed intake was observed in animals offered diets containing 373 or 450 mg QA/kg. Some of the latter (fattening pigs, about 5%) also showed depression, recumbency, hypersalivation and vomiting along with a low mortality (about 1%) that was characterised at necropsy by torsion of the stomach and gastroenteric bloat. A remarkable reduction in feed intake, a higher incidence of clinically affected individuals (up to 50%) and a more severe clinical picture was instead reported in animals exposed to dietary QA levels of 866, 946 and 1,180 mg QA/kg feed, respectively. Mortality rose up to 20% in animals (lactating sows) fed with the highest QA levels (Boschin and Tesio, 2019, personal communication).

The tolerance of growing pigs to lupin alkaloids was tested in two different trials (Godfrey et al., 1985) in terms of changes in growth rate, feed intake, and feed conversion ratio. Total alkaloid concentration was measured by GC (method according to Priddis, 1983). In experiment 1, 9‐week‐old crossbred (Large White x Landrace) piglets (average weight 23 kg) were allocated to six groups receiving diets containing increasing amounts of lupin seeds (a mixture of L. angustifolius sweet cultivars) and consequently increasing total alkaloid concentrations (i.e. 120, 200, 280, 360, 440, 520 mg total alkaloids/kg feed, respectively) for 9 weeks. The corresponding doses calculated by the CONTAM Panel are 5, 8, 11, 14, 16 and 19 mg/kg bw per day, respectively. In experiment 2, piglets of the same breed (average weight 31 kg) were allocated to 3 groups receiving diets containing increasing amounts of BL seeds (L. angustifolius bitter cultivars) and consequently increasing total alkaloid concentrations (i.e. 50, 200, 350 mg total alkaloids/kg feed, respectively) for 7 weeks. The corresponding doses calculated by the CONTAM Panel are 2, 9 and 14 mg/kg bw per day, respectively. At the end of the treatment, 10 animals/group were killed and liver and kidneys were removed and subjected to gross and microscopic examination. No deaths or clinical signs of illness were detected in both trials irrespective of the alkaloid dietary intake. In experiment 1, a dose‐related linear decrease in body weight gain and feed intake was detected, starting from 14 mg/kg bw per day. In experiment 2, growth rate and feed intake were depressed only at the highest dose tested (14 mg/kg bw per day); neither macroscopic nor microscopic changes were detected in liver and kidney from all treated piglets. Although no control group was included, the authors concluded that growing pigs can tolerate lupin alkaloids present in the studied L. angustifolius bitter cultivars at a dietary concentration up to 200 mg/kg. This concentration corresponds to a dose of 9 mg/kg bw per day.

A further study was conducted in Polish Large White × Pietrain barrows (average weight 53 kg) (Rotkiewicz et al., 2007). Animals were fed 62 days with diets in which the soybean meal (control diet) was replaced by lupin seeds of three L. angustifolius varieties, namely Baron, Zeus and Wersal, with slightly different total alkaloid concentrations (360, 410 and 560 mg/kg DM, respectively; analytical method not reported). The corresponding concentrations in the diet were 97, 119, 156 mg/kg diet (DM). Only in barrows offered the diet containing the lupin variety Wersal with the highest alkaloid concentration, daily weight gain was statistically significantly lower than controls (680 g vs. 740 g). In lupin‐fed pigs, histopathological examinations revealed some degree of atrophy and deformation of intestinal villi and an increase in eosinophils and lymphocytes in the enteric wall. In addition, ‘proliferation of glandular and hepatic cells’ was recorded. Due to the lack of a statistical evaluation of the histopathological lesions, these findings cannot be used for hazard characterisation. Based on the reduced weight gain observed for the diet containing the highest alkaloid concentration, the CONTAM Panel concluded that pigs can tolerate lupin alkaloids present in the studied L. angustifolius varieties at a dietary concentration of 119 mg/kg. In the absence of reported feed intakes, the corresponding dose could not be calculated.

A study was designed to examine the effects of feeding two different levels of raw or germinated seeds of L. luteus (yellow lupin) or L. angustifolius (blue lupin) on zootechnical performances in growing pigs (Kasprowicz‐Potocka et al., 2013). Total alkaloid content of the Raw Yellow Lupin (RYL) seeds was 1,000 mg/kg, with sparteine and lupinine amounting to approximately 58% and 27% and low levels of lupanine (3%) and 13α‐OH‐lupanine (6%). Germinated Yellow Lupin (GYL) seeds had a lower total alkaloid content (700 mg/kg) with limited variation (about 10% less) in the percentages of the mentioned QAs. Total alkaloid content of the Raw Blue Lupin (RBL) seeds (1,500 mg/kg) was higher than that measured in RYL. As expected, RBL also showed a different QA composition, in that lupanine (69%) and 13α‐OH‐lupanine (24%) accounted for more than 90% of total QAs. Germinated Blue Lupin (GBL) seeds had a lower total alkaloid content (1,000 mg/kg) with lupanine (52%) and 13α‐OH‐lupanine (18%) accounting for only 70% of total QAs. Two experiments were performed, both lasting 33 days. In Experiment I, 50 crossbred piglets (initial bw 10 kg) were randomly allocated to one of the following treatment groups (N = 5 per sex and per group), namely control (no lupins), RYL, GYL, RBL and GBL. In experimental diets, 50% of the soybean meal protein content of the control group was replaced by the protein of RYL, GYL, RBL or GBL lupins. The same protocol was adopted for Experiment II, but the protein replacement rose to 75%. The resulting total alkaloid content (mg/kg diet) in Experiment I and Experiment II, was the following: RYL 17 and 24, GYL 10 and 15, RBL 31 and 47, GBL 18 and 27. Zootechnical performances (final body weight, feed intake, average daily gain and feed conversion ratio) and health status were monitored in both experiments. Blood samples were collected from 6 animals/group at the end of Experiment II only. Zootechnical performances were not affected in animals from Experiment 1 (50% soy protein replacement) irrespective of the lupin species (L. luteus or angustifolius) or status (raw or germinated). In Experiment II (75% soya protein replacement), feed intake was significantly depressed in RBL‐ or GBL fed piglets and a significant rise in serum alkaline phosphatase (~ 50%) of uncertain biological relevance was observed in RYL‐fed animals. However, the CONTAM Panel noted that no significant changes were observed in ALT or AST in Experiment II for all diets tested and blood enzymes were not tested in Experiment I. Therefore, the CONTAM Panel took only the zootechnical performance into account for a conclusion regarding the tolerated concentration of QAs in feed. It is concluded that piglets tolerate lupin alkaloids present in the studied L. luteus cultivar up to a dietary concentration of 24 mg/kg (1.5 mg/kg bw per day) and lupin alkaloids present in the studied L. angustifolius cultivar up to 18 mg/kg (1.0 mg/kg bw per day).

3.1.4.3. Poultry

Laying hens and broiler chicks

A study was designed to determine the lethality of total and selected QAs from a bitter variety of L. albus (cv. not mentioned) in two breeds of 10‐day‐old laying hens (ISAbrown and SHAVERcross) (Cubillos et al., 1999). Total alkaloids were extracted from seeds to obtain aqueous solutions (analytical technique not mentioned). Lupanine (71%) and 13α‐OH‐lupanine (13%) were by far the most abundant QAs in the extract, which contained less than 1% sparteine. Lupanine was purified from the total alkaloid extract and sparteine sulfate (192 mg/mL) was commercially obtained. Animals were gavaged and mortality recorded over 48 h. In ISAbrown poults, calculated LD₅₀ (mean and 95% confidence interval) were 958 mg/kg bw (854–1,070) for total alkaloids, 1,131 mg/kg bw (929–1,378) for lupanine, and 655 mg/kg bw (509–856) for sparteine base. In SHAVERcross poults, similar LD₅₀ values were found for total alkaloids and lupanine, being, respectively, 961 mg/kg bw (890–1,035) and 1,271 mg/kg bw (1,027–1,573), while much lower values were calculated for sparteine, i.e. 425 mg/kg bw (303–544), pointing to the occurrence of breed‐related differences in the sensitivity to certain QAs. Results from this study also confirm the higher toxicity of sparteine compared to lupanine (two‐ to three‐fold) observed in mammals.

Vogt et al. (1987) fed laying hens (age 24 weeks) with diets containing no lupins or diets containing 16% debittered lupin grist (from either L. albus 24 P, or L. albus ²⁴ G or L. mutabilis) that contained 1.1, 6.2, 9.6 or 1.2 mg total alkaloids/kg diet, respectively, for 168 days. Total alkaloid concentration was determined with GC (no details given); according to the authors, almost 90% of the measured total alkaloid was lupanine. Using a starting weight of 1.505 kg for the animals and a mean feed intake of 116, 120, 121 and 109 g per day, respectively, a corresponding intake of 0.08, 0.49, 0.77 and 0.09 mg alkaloids/kg bw per day could be calculated. Feeding of lupin grist did not adversely affect feed intake/efficiency, laying rate, egg weight, and egg fertility/hatchability. However, the Panel noted that the applied doses are low and not expected to cause any effects.

The adverse effects of dietary lupins were assessed in 1‐day‐old broiler chicks (unspecified breed) offered a diet including soybean meal (control) or raw (40%, group A), dehulled (40%, group B), or autoclaved (35% group C) L. angustifolius (cv. Troll) seed meals, respectively, for 21 days. The total alkaloid concentration, measured by a titration method according to Ruiz (1977), was less than 100 mg/kg in the lupin seeds, which approximately corresponded to less than 40, 40 and 35 mg alkaloid/kg feed for group A, B, and C, respectively. During the first week, two chicks from group A died and further two were euthanatised in the remaining weeks showing bradycardia, twisting of the neck, leg weakness and coma. Post‐mortem examination did not reveal significant gross changes. Skeletal deformities (limb deformity, crooked sternum and scoliosis) affected a total of seven individuals (3 from group A, 3 from group B, 1 from group C). In all treated groups, body weight gain, feed consumption, and feed to gain ratio were remarkably depressed, while the liver CYP level doubled. The authors concluded that in broilers the repeated dietary exposure to certain varieties of lupins may entail severe systemic effects (including skeletal deformities) which were not counteracted by dehulling or autoclaving (Olkowski et al., 2001). The CONTAM Panel noted the observed effects; however, due to the uncertainty regarding the concentration of total alkaloids in the feed, this study cannot be used for dose‐response assessment.

A study was designed to investigate the effects on zootechnical performance of lupin seeds (Olver and Jonker, 1997). A SL (L. albus cv. Hanti), a BL (L. angustifolius) and soaked micronised bitter lupin (SMBL) seeds were analysed for alkaloids by titration (method according to Ruiz, 1977) and found to contain < 100 mg alkaloids/kg for the SL, < 200 mg alkaloids/kg for the SMBL and > 900 mg/kg for the BL. As a result, the soaking procedures reduced the alkaloid concentration to less than 200 mg/kg lupin seeds. The trial was conducted on male Ross 1‐day‐old broilers chicks which were offered for 6 weeks a control (no lupin), a SL, a BL or a SMBL diet, containing 0, 50, 100, 200, 300 or 400 g/kg of the lupin under test. In chicks fed SLs, body weight gain, feed consumption, and feed to gain ratio were almost superimposable to control values irrespective of the added lupin amount. This was also the case for carcass characteristics (moisture, fat, protein, and ash content). By contrast, all BL‐fed chicks (both BL and SMBL at all dosages) showed a depression in body weight gain, feed consumption and feed to gain ratio. An increased moisture and carcass fat content were detected at the two highest lupin concentrations for the BL group and at the highest lupin concentration for the SMBL group, respectively. The CONTAM Panel noted the observed effects; however, due to the uncertainty regarding the concentration of total alkaloids in the feed, this study cannot be used for dose‐response assessment.

Feeding experiments with L. albus (cv. Ultra) seeds at several levels and different processing methods were performed in Single Comb White Leghorn laying hens (Watkins and Mirosh, 1987). The lupin alkaloid concentration (GC–MS) of the seeds was 80 mg/kg and only lupanine was quantified. Raw, autoclaved (121°C for 30 min) or extruded (80°C for 20 sec) lupins29 were included in the diets at different percentages (10, 15, 20, 25 or 30%) and administered for up to 32 weeks. A group fed with a basal diet (no lupins) was always included. The tested parameters were mortality, feed consumption, egg production and egg weight. No differences were observed between hens offered the basal diet or diets containing autoclaved lupins (up to 20% for 12 weeks) or extruded lupins (up to 25% for 16 weeks). Only hens administered with 25 or 30% raw lupins and for the longest duration (32 weeks) showed significant reduction in egg weight and/or egg production. The authors concluded that the L. albus (cv. Ultra) may be fed to laying hens as a protein source at levels of up to 20% (corresponding to a lupanine concentration of 16 mg/kg feed and a daily dose of 0.9 mg/kg bw based on a body weight of 2 kg and a daily feed intake of 0.113 kg/hen).

Turkeys

The tolerance to the dietary inclusion of different levels of seeds of white lupins (L. albus cv. Ultra) was investigated in growing male and female Large White (Nicholas) turkeys (Halvorson et al., 1988). The control diet consisted of corn and soybean; in lupin‐treated animals, the control diets were partially replaced by ground whole (WL) or ground dehulled (DWL) seeds. Total alkaloid content (analytical method: GC‐FID) amounted to 960 mg/kg (including 740 mg/kg ‘OH‐lupanine’30 and 150 mg/kg lupanine) in WL seeds and to 430 mg/kg (including 190 mg/kg ‘OH‐lupanine’ and 130 mg/kg lupanine) in DWL seeds used in Experiment 1, and 430 mg/kg (including 180 mg/kg ‘OH‐lupanine’ and 140 mg/kg lupanine) for WL seeds used in Experiment 2. In Experiment 1, one‐day old males were offered one of the following diets for 21 days: control, WL at 20% or 40% inclusion or DWL at 20% or 40% inclusion. The CONTAM Panel calculated the corresponding doses of 29, 61, 13 and 26 mg lupin alkaloids/kg bw. Tested parameters were body weight, average daily gain, average feed intake, and feed efficiency. In Experiment 2, one‐day old females were assigned to one of the following groups: control diet, control diet + WL (2:1, ~ 33% lupins), control diet + WL (1:1, 50% lupins), control diet + WL (0.6:1, ~ 43% lupins). At the end of the study (17 weeks), animals were killed and organs (liver, heart, pancreas, kidney, gizzard, intestine) were removed and weighed.

In Experiment 1 (3‐week feeding), there was a very similar daily feed intake in controls and all treated animals; body weight and daily weight gain were depressed only in male turkeys administered with the 40% WL diet, corresponding to a total alkaloid dose of 61 mg/kg bw per day. In Experiment 2 (17‐week feeding), mortality was not affected by the treatment and the macroscopic examination of the organs did not reveal any alteration; a very slight (up to 5%) but statistically significant reduction in body weight and average daily gain was detected in turkey hens from all treated groups along with an increase in gizzard and pancreas weight. The treated groups were exposed to concentrations ranging from 61 to 215 mg/kg feed and the corresponding doses ranged from 2.7 to 9 mg/kg bw per day. According to the authors it is doubtful that the observed effects could be related to the alkaloid contents. Based on the obtained results, the CONTAM Panel could not identify a tolerated dose from this study since effects were observed at all doses tested in the experiment with the longest and lowest exposure.

Ducks

A study was designed in Peking ducklings (Olver and Jonker, 1998) using the same experimental design of a study performed in broilers, as described above (Olver and Jonker, 1997). The tested lupin seeds included SL (L. albus cv. Hanti), a BL (L. angustifolius) and SMBLs, with an alkaloid concentration (measured by titration) of < 100 mg alkaloids/kg for the SL, > 900 mg/kg for the BL and < 200 mg/kg for the SMBL. Two hundred and forty ducklings were allocated to 16 groups and offered for 6 weeks one of the following diets containing 0, 50, 100, 200, 300, 400 g/kg of the lupins under test. Three animals per group were killed at the end of the treatment for carcass traits analysis (moisture, fat, protein, and ash content). Productive parameters (body weight gain, feed consumption, and feed to gain ratio) were recorded after 3 and 6 weeks. Animals fed either with the SL variety or with the SMBL, did not show statistically significant differences with controls in any of the tested parameters irrespective of the amount added to the diet. By contrast, BL did not affect productive traits only at inclusion levels up to 100 g/kg diet and carcass characteristics up to 300 g/kg. According to the authors, the apparent higher sensitivity of ducklings to BLs compared to broilers might be due to a more developed sense of taste in ducks, as assessed by a higher number of taste buds (200) with respect to broilers (20) (Kare and Ficken, 1963 as cited in Olver and Jonker, 1998). The CONTAM Panel noted the observed effects; however, due to the uncertainty regarding the concentration of total alkaloids in the feed, this study cannot be used for dose‐response assessment.

Japanese quails

One‐day‐old quail chicks (average weight 7 g, breed not reported) were randomly divided into four groups, namely a control group I and three further groups fed 15% L. albus seeds as follows: raw I, processed (debittered) according to a traditional Turkish method (TTM) or subjected to extrusion I. The lupin alkaloid concentration was determined by GC‐FID. The trial lasted 6 weeks and feed consumption, live weight, and feed conversion efficiency were monitored weekly. Lupanine (28,000 mg/kg) accounted for more than 80% of the lupin alkaloid content in raw lupin seeds (29,000 mg/kg) and low levels of sparteine (210 mg/kg), lupinine (15 mg/kg), cytisine (185 mg/kg) and anagyrine (115 mg/kg) were also reported. While TTM proved extremely efficacious in reducing the alkaloid concentration (lupanine concentration around 400 mg/kg, all other alkaloids not detectable), the extrusion method brought about only a negligible reduction (10%) of alkaloids. While live weight and feed conversion efficiency were lower in both R and E groups up to the fourth week of treatment, at the end of the trial there were no statistically significant differences in any of the tested parameters. It was concluded that growing performances of Japanese quails are not affected by the dietary inclusion of 15% lupin seeds from L. albus containing up to 29,000 mg alkaloids/kg seed (corresponding to 4,771 mg lupin alkaloids/kg feed DM or 826 mg/kg bw per day) (Arslan and Seker, 2002).

3.1.4.4. Rabbits

New Zealand White rabbits (initial average weight ~ 1 kg) were allocated to three experimental groups (Battaglini et al., 1991). Each group was offered a diet containing three different levels (0, 8 or 16%) of three different cultivars of L. albus seeds, namely Italian (I), Australian (A) or Australian extruded (AE) cultivars, for 80 days. Total alkaloid content was measured with an ‘official EC method’ and amounted to 5.8 (I), 5.4 (A), and 1.6 (AE) g/kg DM; the three inclusion levels resulted in the following total alkaloid content in the feed (mg/kg DM): 0, 460, 930 (I), 0,430, 860 (A) and 0, 130, 260 (AE). The corresponding doses for group I were 0, 30 and 60 mg/kg bw per day, for group A 0, 30 and 55 and for group AE 0, 10 and 19. The trial lasted 80 days, but zootechnical performances were recorded from day 45 until day 80 only. As compared to controls, in the I group, only the highest inclusion level (16%) negatively affected feed consumption, final weight and daily weight gain. In rabbits of the A group exposed to the same level (16%, same total alkaloid content) a reduced feed consumption was observed. Zootechnical parameters were not affected in rabbits receiving extruded lupins (AE). From this study, it is concluded that rabbits tolerate lupin alkaloids present in the studied L. albus cultivars at a dose of 30 mg/kg bw per day.

3.1.4.5. Fish

Rainbow trout (Oncorhynchus mykiss)

A feeding study (Serrano et al., 2011) was carried out in rainbow trout to investigate the effects of lupinine, the main QA in European L. luteus cultivars (Wink et al., 1995), accounting for almost 60% to the total alkaloid concentration in the plant. Rainbow trout (average weight 330 ± 20 g) were allocated to 8 groups receiving a diet containing 0, 50, 75, 100, 250, 500, 1,000 or 5,000 mg lupinine/kg, respectively, for 60 days. From the data provided by the authors regarding feed intake and body weight, the CONTAM Panel calculated the corresponding doses (0, 0.5, 0.8, 1.1, 2.5, 4.5 and 7.2 mg/kg bw). A 20% and 33% reduction in feed intake was detected at 500 and 1,000 mg/kg inclusion levels, respectively; the same dosages resulted in a 30% and 50% fall in weight gain, respectively. After 40 days, the treatment with the highest concentration (5,000 mg lupinine/kg) was stopped for animal welfare reasons (extremely low feed intake and weight gain). At the end of the experimental period, 9 individuals/group were sacrificed. Body, liver and spleen were weighed. Treatment effects on productive parameters, whole body composition analysis and tissue histopathology were also investigated. There was a dose‐related quadratic depression in both feed intake and growth rate, negative effects being observed at doses greater than 1.1 mg/kg bw. Feed efficiency ratio was not affected. No relevant histopathological changes were detected, except for a depletion of glycogen and lipid stores in the 7.2 mg/kg bw dose group. Splenosomatic index31 was not affected, whereas decrease in the hepatosomatic index31 was noticed at lupinine doses higher than 1.1 mg/kg bw. It is concluded that a dose of 1.1 mg lupinine/kg bw is tolerated by rainbow trout and can be considered as a NOAEL for lupinine in rainbow trout, while 2.5 mg/kg bw can be considered as a lowest‐observed‐adverse‐effect level (LOAEL).

A further study was conducted in rainbow trout to investigate the effects of sparteine (Serrano et al., 2012). Juvenile rainbow trout (average weight 61 ± 7 g) were allocated to 8 groups receiving a diet containing 0, 50, 100, 250, 500, 1,000, 2,500 or 5,000 mg sparteine/kg, respectively, for 62 days. From the data provided by the authors regarding feed intake and body weight, the CONTAM Panel calculated the corresponding doses (0, 0.7, 1.4, 3.5, 6.6, 11.9, 14.5 and 15.6 mg/kg bw). No mortality occurred. Final body weight, weight gain, feed consumption and hepatosomatic index were decreased in fish exposed to > 1.4 mg/kg bw. By contrast, treated fish did not display relevant histopathological changes in liver, kidney, spleen, and intestine, suggesting that the adverse effects of sparteine in rainbow trout are likely to reflect the decrease in feed intake. In conclusion, doses of sparteine up to 1.4 mg/kg bw per day are safe in trout and this dose can be considered as a NOAEL for sparteine in rainbow trout. The LOAEL identified from this study was 3.5 mg/kg bw.

3.1.4.6. Horses

No information about QA toxicity in horses was identified.

3.1.4.7. Companion animals

One study in cats and dogs was identified, which is described in Section 3.1.2.3.

3.1.4.8. Summary remarks

Limited data are available regarding the adverse effects caused by QAs in farm animals, horses and companion animals. Most studies used lupin seeds as source of QAs in the diet and do not report the QAs present in the diet or their relative contribution to the total QA concentration. Such studies do not allow identification of NOAEL/LOAELs. These studies, as well as studies that allowed the identification of NOAEL/LOAELs are summarised in Appendix H.6

In cattle, only studies on zootechnical performance were identified and from these, the CONTAM Panel identified 50 mg/kg bw per day as a dose of lupanine and 13α‐OH‐lupanine present in seeds from L. albus that can be tolerated. No relevant studies were identified for sheep and goats.

Case reports on lupin toxicosis are only reported for pigs, indicating that pigs are sensitive to QAs. Based on zootechnical performance, the CONTAM Panel identified 1‐10 mg/kg bw per day as a dose range of lupin alkaloids present in seeds from L. angustifolius and 1.5 mg/kg bw per day as a dose of lupin alkaloids present in seeds from L. luteus that can be tolerated by pigs. In addition to an effect on zootechnical performance, gastro‐intestinal lesions have been observed in pigs receiving a lupin seed‐based diet. However, due to the lack of a statistical evaluation, these findings could not be taken into account for identifying tolerated doses.

An acute toxicity study in laying hens, confirms the higher toxicity of sparteine compared to lupanine observed in mammals. Based on zootechnical performance, 1 mg/kg bw per day was identified as a dose of lupin alkaloids present in seeds from L. albus that can be tolerated by laying hens. In a study with one‐day‐old broiler chicks (for which no dose calculation was possible), effects on zootechnical performance, as well as severe systemic effects, were observed, however at concentrations in the feed that were higher compared to the studies in laying hens. No tolerated dose could be identified for turkeys since effects were observed on zootechnical performance at the lowest dose tested of 2.7 mg/kg bw per day. In one‐day‐old quail chicks, 826 mg/kg bw was identified as a dose of lupin alkaloids present in seeds from L. albus that can be tolerated (only dose tested).

For rabbits, the CONTAM Panel identified 30 mg/kg bw per day as a dose of lupin alkaloids present in seeds from L. albus that can be tolerated.

Pure lupinine and sparteine were tested in feeding studies with rainbow trout, and NOAELs of 1.1 and 1.4 mg/kg bw per day were identified for lupinine and sparteine, respectively, for the effect on zootechnical performance and hepatosomatic index.

3.1.6.1. Acute toxicity

Experimental animal data

The acute toxicity of lupanine was two‐ to fourfold higher in mice compared with rats when delivered either by i.p. (LD₅₀: 80 vs. 177 mg/kg bw, respectively) or orally (LD₅₀: 410 vs. 1,664 mg/kg bw, respectively). Studies in rats indicated that acute toxicity of lupanine and 13α‐OH‐lupanine is similar (LD₅₀: 177 and 199 mg/kg bw by i.p., respectively). Sparteine is two times more toxic than lupanine in mice (oral LD₅₀: 220 vs. 410 mg/kg bw, respectively) as well as in guinea pigs. Following oral intakes of lupanine, 13α‐OH‐lupanine and sparteine, the clinical symptoms of poisoning were similar in all species, with the cause of death being identified in respiratory failure and anticholinergic symptoms. The similarity of the symptoms of intoxication supports the conclusion on a similar mode of action of these QAs.

Human data

Livestock and companion animal data

In the only relevant study, oral LD₅₀ of total alkaloids from a bitter variety of L. albus, lupanine and sparteine was determined in 10‐day‐old laying hens of two different breeds (ISAbrown and SHAVERcross). In ISAbrown poults, calculated LD₅₀ (mean and 95%) were 958 mg/kg bw (854–1,070) for total alkaloids, 1,131 mg/kg bw (929–1,378) for lupanine, and 655 mg/kg bw (509–856) for sparteine. In SHAVERcross LD₅₀ for total alkaloids and lupanine were 961 mg/kg bw (890–1,035) and 1,271 mg/kg bw (1,027–1,573), respectively, for lupanine, while for sparteine an LD₅₀ of 425 mg/kg bw (303–544) was calculated. Results from this study confirm the higher toxicity of sparteine compared to lupanine observed in mice and guinea pigs and indicate possible breed‐related differences in the sensitivity to certain QAs.

Overall summary for acute toxicity

Comparison of doses reported from human intoxication with those resulting in acute toxicity in experimental animals shows that humans are more sensitive to QAs than laboratory rodents. The observed difference in acute toxicity might be linked to the differences in toxicokinetics between humans and rats and particularly the biotransformation of sparteine to lupanine in the rat, which is not reported to occur in humans. Therefore, the CONTAM Panel used the human data as a basis for hazard characterisation in the assessment of human risks due to exposure with QAs via food.

Supported by human data on sparteine, the CONTAM Panel concluded that anticholinergic and antiarrhythmic effects of QAs are the critical acute effects for the hazard characterisation. No human studies are available describing the dose‐response relationship for these effects of QAs.

Knowledge on dose ‐responses refers only to oral therapeutic doses of sparteine sulfate used in the treatment of cardiac arrhythmia, which is mainly due to the effect on the voltage dependant sodium channels. Adverse effects associated with this treatment belong to the spectrum of the anticholinergic syndrome. A dose of 20 mg of sparteine sulfate (equivalent to 0.16 mg sparteine/kg bw assuming a body weight of 70 kg) is described as the lowest single oral dose with antiarrhythmic properties, which is one‐fifth of the lowest single oral dose of sparteine sulfate associated with oxytocic activity.

Only one study in poultry was identified that supports the conclusion from experimental animals that sparteine is more toxic than lupanine. No other data in livestock or companion animals were identified.

3.1.6.2. Repeated‐dose toxicity

Experimental animal data

All the repeated dose studies with laboratory animals on QAs were carried out in rats. Single dose, sub‐chronic studies on diets containing L. albus‐, L. luteus‐, and L. mutabilis seeds did not identify significant toxicological effects. Two studies investigated a diet containing increasing concentrations of L. angustifolius flour. The major QAs in the flour were lupanine and 13α‐OH‐lupanine. Both studies reported a slight increase in relative liver weights. However, some conflicting results were reported in different publications from the same laboratory.

Due to several reasons including differences in sources of QAs in feeds, control diets and types of cultivars, the CONTAM Panel concluded that the repeated‐dose studies in experimental animals have a limited value for the evaluation of QA toxicity.

Human data

In the long‐term therapy of cardiac arrhythmias the recommended daily dosage for (‐)‐sparteine sulfate was half of that for short‐term treatment and ranges from 400 to 500 mg (equivalent to 5.7 to 7.1 mg/kg bw for a person with a body weight of 70 kg), divided into 4–5 single doses of 100 mg (equivalent to 1.4 mg/kg bw for a person with a body weight of 70 kg). Known adverse effects and contraindications were the same as for short‐term use indicated above (Rote Liste, 1983).

No other human data on adverse effects associated with long‐term exposure to lupin seeds or QAs occurring in lupin seeds are available. The CONTAM Panel concluded that the available data are too limited to identify a reference point for the human risk assessment following repeated exposure.

Livestock and companion animal data

The large majority of relevant repeated‐dose studies in livestock species (except fish, see below) were performed with the sole aim of identifying the lupin dietary inclusion levels not affecting zootechnical performances (feed intake, weight gain, feed efficiency, milk yield, egg production). The CONTAM Panel noted the considerable differences in both total and specific QA content among lupin seeds from different species and even cultivars as well as the lack of specific toxicity studies. Therefore, based on the available data, NOAEL/LOAEL for most food producing species could not be identified and were replaced by suggested tolerated concentrations/doses specifically referring to seeds from a single lupin species/cultivar. The only relevant studies performed with pure QAs (i.e. lupinine and sparteine) were conducted in rainbow trout and allowed to identify similar NOAEL values for lupinine and sparteine (1 mg/kg bw per day). In the absence of data on other QAs, the CONTAM Panel used this NOAEL for a group approach to characterise the hazard of the QAs relevant for this opinion and applied to all farmed salmonids. No relevant studies have been identified for companion animals.

3.2.1.1. Occurrence data on food

A total of 1,035 analytical results on 99 food samples were retained after data cleaning. All samples referred to lupin‐based foods. Details of the data cleaning applied for food samples are presented in Annex A (Table A.2).

Analytical results were submitted by the Netherlands (540 analytical results from 54 samples on 10 QAs) and by the University of Milan (495 analytical results from 45 samples on 11 QAs); detailed information on the number of samples versus analysed substance, sampling year and reporting country/institution can be retrieved in Table A.3 of Annex A. The sampling year spanned from 2005 until 2018, with most of the samples collected in 2016 (Annex A, Table A.3).

Table 4 provides an overview of the number of samples per food category according to different levels of the FoodEx2 classification system.

The vast majority of occurrence data were reported for samples (n = 65) classified as ‘Lupins (dry) and similar’. This type of samples represents lupin seeds that are sold as dry seeds and that can be consumed upon rehydration, debittering, boiling and brining or as flour that can be used for home‐made dishes (such as pancakes/crepes or pasta). Information on the lupin species was reported in some cases: 29 samples were classified as L. albus and 27 as L. angustifolius. Twenty‐eight samples of ‘Lupins (dry) and similar‐’ were reported to be processed (i.e. ‘grinding/milling/crushing’) and the results therefore refer to lupin seeds in the form of flour or grits. Samples reported in the category ‘canned or jarred legumes’ referred to lupin seeds in water and salt or pickled (those are normally consumed as such). In the category ‘canned or jarred legumes’, all samples were left‐censored for lupinine, tetrahydrorhombifoline and albine, while for lupanine, multiflorine and 13α‐OH‐lupanine only 13% of the samples were left‐censored. In the category ‘Lupins (dry) and similar‐’, left‐censored results were as follows: lupanine 0%, 13α‐OH‐lupanine 13%, angustifoline 11% and α‐isolupanine 18%. Details on the percentage of left‐censored data in food samples can be found in Table A.4 (Annex A).

Under the category ‘meat imitates’ different types of lupin morsels, steaks or cutlets were reported, in few cases the % of lupin flour was reported as 40%. For lupanine and 13α‐angeloyloxylupanine, no left‐censored samples were present in this category.

In the category ‘Pasta‐like products’, pasta to which isolated protein was added as ingredient was reported, while in the category ‘Pasta, plain (not stuffed), uncooked’ lupin flour was added as ingredient.

Fifty‐four samples were analysed with LC–MS/MS and an LOQ of 0.1 mg/kg was reported for all tested substances (i.e. 13α‐OH‐lupanine, 13α‐angeloyloxylupanine, 13α‐tigloyloxylupanine, angustifoline, α‐isolupanine, lupanine, lupinine, multiflorine, sparteine, tetrahydrorhombifoline). The other samples were analysed with GC‐MS and an LOQ of 1.5 mg/kg was reported for all tested substances (i.e. the substances analysed by LC‐MS/MS plus albine).

In the category ‘canned or jarred legumes’, all samples were left‐censored for lupinine, tetrahydrorhombifoline and albine, while for lupanine, multiflorine and 13α‐OH‐lupanine only 13% of the samples were left‐censored. In the category ‘Lupins (dry) and similar’, left‐censored results were as follows: lupanine 0%, 13α‐OH‐lupanine 13%, angustifoline 11% and α‐isolupanine 18%. Full details on the percentage of left‐censored data in food samples can be found in Table A.4 (Annex A).

Descriptive statistics by food category for lupanine and TotQA are presented in Table 5 while the full descriptive statistics for all the QAs can be found in Table A.5 (Annex A). TotQA is calculated as the sum of the 6 most abundant QAs in the data set available, namely lupanine, 13α‐OH‐lupanine, angustifoline, multiflorine, 13α‐tigloyloxylupanine and α‐isolupanine.

Very high levels of lupanine (up to 5,395 mg/kg) and TotQAs were found in certain samples belonging to the category ‘Lupins (dry) and similar‐’. For this reason, a clarification request was sent to the data provider. It was confirmed that a part of the reported samples (n = 22) were not present on the EU market but rather cultivated for research purposes. These research samples showed a higher mean concentration of lupanine compared to the other samples of lupin seeds (768 vs 192 mg/kg LB = UB), multiflorine (60 vs 11 mg/kg LB) and 13α‐angeloyloxylupanine (27 vs 2.6 mg/kg LB), but showed a lower mean concentration of 13α‐OH‐lupanine, 13α‐tigloyloxylupanine and angustifoline. Based on this information, the CONTAM Panel decided to exclude the research samples from the data set used for dietary exposure assessment.

3.2.1.2. Occurrence data on feed

A total of 390 analytical results on QAs were reported on 37 feed samples. Details of data cleaning are reported in Table B.1 (Annex B).

Analytical data on feed were submitted by the Netherlands (170 analytical results from 17 samples on 10 QAs) and by the University of Milan (220 analytical results from 20 samples on 11 QAs); the sampling year spanned from 2011 until 2018, with most of the analytical results from samples collected in 2017. Detailed information can be retrieved in Table B.2 (Annex B).

The University of Milan submitted 5 samples coded as ‘Legume seeds and products derived thereof (feed)’ and 15 as ‘Lupin middlings (feed)’. It should be noted that some of the samples reported in the category ‘Legume seeds and products derived thereof (feed)’ were specified by the data provider as ‘mixtures of dry legume seeds containing also dry lupin seeds (from L. albus) although with % lupin and composition unknown’. The category ‘Lupin middlings (feed)’ includes samples specified as ‘lupin flour for pig feed with percentage of lupin and composition unknown’. The Netherlands submitted 17 samples of ‘sweet lupins (feed)’, which were all L. angustifolius seeds.

Both data providers reported that several samples were suspect.37 All 20 samples of ‘Lupin middlings (feed)’ and ‘Legume seeds and products derived thereof (feed)’ reported by the University of Milan were collected from pig farms where intoxication had been reported. In the category named as ‘sweet lupins (feed)’, 6 suspect samples were reported as well. A description of the number of feed samples and percentage of left‐censored data per substance can be found in Table B.3 of Annex B.

The 17 samples reported by the Netherlands were analysed using LC–MS/MS for 13α‐OH‐lupanine, 13α‐angeloyloxylupanine, 13α‐tigloyloxylupanine, angustifoline, α‐isolupanine, lupanine, lupinine, multiflorine, sparteine, tetrahydrorhombifoline. The LOQ was 0.1 mg/kg for all QAs reported. The 20 samples reported by the University of Milan were analysed by GC–MS for the 10 substances mentioned above plus albine; all were reported with an LOQ of 1.5 mg/kg.

Descriptive statistics of the occurrence data for feed samples reported in this opinion can be found in Table 6. As for the food, the CONTAM Panel decided to focus on lupanine and TotQA as the sum of lupanine, 13α‐OH‐lupanine, angustifoline, multiflorine, 13α‐tigloyloxylupanine and α‐isolupanine. More detailed descriptive statistics for all the alkaloids can be found in Table B.4 (Annex B).

Due to the large differences in mean concentrations between suspect and non‐suspect samples, the CONTAM Panel decided not to use feed samples reported as suspect (Table 6) for the exposure assessment for farm animals, horses and companion animals. Consequently, a very limited number of samples (n = 11) were available for the exposure assessment. The CONTAM Panel therefore decided to complement the data set with data on occurrence of QAs in ‘Lupin (dry) and similar‐’ reported as food samples for which the species was reported (13 of L. albus and 21 of L. angustifolius), plus 9 samples for which the lupin species was not specified. Chemical occurrence data reported for L. angustifolius as food and feed were merged, while for L. albus only food samples were available. Table 7 shows the descriptive statistics of the final data set.

3.2.1.3. Relative variations of main QAs in lupin seeds

Figure 5 shows the relative percentage of the 6 main QAs normalised by the sum for L. albus (n = 13) and L. angustifolius (n = 21) seeds that were reported as food samples as well as L. angustifolius seeds (n = 11) reported as feed samples. All suspect samples as well as the research samples were not included.

Considering the small number of samples per category, results should be considered with caution. Lupanine was the most abundant QA in both species, followed by 13α‐OH‐lupanine (especially abundant in L. angustifolius) and angustifoline. Multiflorine and 13α‐tigloyloxylupanine were more abundant in L. albus than in L. angustifolius. These observations, although arising from a relatively small number of samples, are in agreement with the scientific literature (Table 1).

3.2.1.4. Occurrence data received during the public consultation

During the public consultation (EFSA, 2019), EFSA received analytical results from 121 samples of L. albus and L. angustifolius from a company analysed by GC. The concentration of total alkaloids ranged from 10 to 1.149 mg/kg with an average concentration of 184 mg/kg and a P90 of 368 mg/kg. The CONTAM Panel noted the lower concentration of total alkaloids in the data received during the public consultation compared to the data submitted to EFSA during the call for data. It was noted that for some of the samples only lupanine and 13‐OH‐lupanine were analysed and based on the assumption that these substances contribute for 85% to the total QA content, the total QA content was calculated.

Food

Two papers have been identified that describe the QA content in lupin‐based foods.

Reinhard et al. (2006) studied six QAs as well as ammodendrine and gramine in food on the Swiss market using GC‐FID and GC–MS (Table 8). Except from lupin coffee, the concentrations of alkaloids are very low compared to the concentration in seeds/flours analysed in the same study (see Appendix D, Tables D.1–D.4). However, lupin coffee is also the only food that is made solely from lupin seeds. Primarily lupanine, 13α‐OH‐lupanine, and angustifoline were found in the foods. The three QAs – α‐isolupanine, 13α‐angeloyloxylupanine and 13α‐tigloyloxylupanine – were only present in trace amounts in two of the lupin coffee samples, in the curry burger, the lupin tofu products and one of the breads.

Resta et al. (2008) analysed eight samples of lupin‐based commercial foods by GC‐MS (see Table 9) The results of the total QA concentration is the sum of the found QAs and is denoted ‘estimated total alkaloids’.

Feed

No information regarding the occurrence of QAs in whole plants (which would include stems, leaves and seeds) fed to livestock has been identified. In a few papers results from leaves are included and these results are summarised in Appendix D, Table D.6 to give an indication of the concentrations and distribution in plant parts other than seeds for the 11 relevant QAs. The CONTAM Panel noted that in general the same QAs are found in seeds and leaves while the distribution of the individual QAs can differ.

3.2.3.1. Debittering of lupin seeds

BL seeds need to be debittered before consumption. The traditional house‐hold methods include soaking in cold water, cooking and washing and this process lasts several days (Smith, 1987; Carvajal‐Larenas et al., 2014). In Ecuador, the National Agricultural Research Centre has proposed a commercial debittering process which consists of soaking the seeds of L. mutabilis for 14–16 h, boiling for 40 min and then washing at 40°C (Carvajal‐Larenas et al., 2013).

Carvajal‐Larenas et al. (2016) publised an overview paper about debittering of L. mutabilis seeds. In a study at commercial scale, 93% of the alkaloid content was removed after treating lupin seeds with warm water (40°C) for 90 h. A process where the seeds were soaked for 14–20 h (15°C), boiled for 1/2–2 h and then washed for 96–120 h, removed 97% of the alkaloid content while cooking for 1/2 h and extraction with cold water for 72 h removed 95% of the alkaloid content. Also, debittering with other solutions than water has been described e.g. treatment with alkaline solutions (Carvajal‐Larenas et al., 2016). According to the authors, water treatment is the only known process that is applied at commercial scale and the CONTAM Panel also considers that water treatment is the only process of interest for debittering at home.

Jimenez‐Martinez et al. have published several papers (2001, 2004, 2009, 2010) about debittering of seeds of a wild variety of L. campestris. The seeds were boiled for 6 h with water or in an acid or alkaline solution. The liquid was changed every 20 min. Water treatment including boiling removed between 89% and 99% of the original QA content.

Biological processes can also be used to debitter lupin seeds and this includes bacterial fermentation, fungal fermentation and germination. According to Carvajal‐Larenas et al. (2016), bacterial fermentation reduced the concentration of alkaloids by < 50%, fungal fermentations reduced the concentration with 50–91% while germination caused a reduction of 78%.

Gefrom et al. (2013) have studied the effect of ensiling on the concentration of QAs in crushed lupin seeds. A sweet (Bora) and a bitter (Azuro) cultivar of L. angustifolius were harvested at 65% dry matter content and four different kinds of model silages were prepared. The samples were stored at 20°C for 90 days and analysis was performed by GC–MS. The authors concluded that ensiling had no significant effect on the QA content.

Chilomer et al. (2010) investigated the influence of germination on the total amount of QAs as well as on single QAs. Seeds of L. angustifolius (cv. Graf) and L. luteus (cv. Lord) were soaked and washed before germination that was conducted in darkness for 48, 72 and 96 h at 15°C or 24°C. The samples were analysed by GC‐NPD. For both lupin varieties, the concentration decreased with time and temperature (up to 30%), so the lowest concentration was obtained after 4 days at 24°C. Also, the concentration of the different QAs differed with time and temperature.

De Cortes Sánchez et al. (2005) studied the change in QA concentration during the germination of seeds from a wild variety of L. campestris from Mexico as well as bitter seeds from L. albus L. (LO‐3,923) and L. angustifolius (1,413) from Spain. Samples were taken each day until day 9 of the germination. Determination of QAs was performed by GC‐NPD and the single substances were identified by GC–MS. For L. angustifolius, the total alkaloid concentration was the same at day 0 and day 9, while in seeds from L. albus and L. campestris the final concentrations were higher at day 9 than on day 0. Also, here the concentrations of the different QAs differ with time and the results for the individual QAs from day 0 are shown in Appendix D, Tables D.1 and D.2 for L. albus and L. angustifolius. The results from this study are not in accordance with the results from the paper by Chilomer et al. (2010).

A completely different process was developed by Haddad et al. (2006). Seeds of L. albus and L. mutabilis, with alkaloid concentrations of, respectively, 250 mg/kg DM and 55,000 mg/kg DM, were cracked, dehulled, moistened and treated by instantaneous controlled pressure drop technology (DIC), soaked for 2 h in water, and dried to a water content of 5%. The total alkaloid concentration was determined by GC‐NPD. The total alkaloid concentration decreased by 60–70%.

3.2.3.2. Changes during production of processed food

Erbas (2010) debittered seeds from a bitter variety of L. albus grown in Turkey to produce a lupin snack. The process consisted of cleaning, boiling, debittering and salting. After boiling the seeds were debittered using three different processes, but only one result (mean of the three processes) was given. The alkaloids were determined by GC‐FID with sparteine as internal standard. The total amount of alkaloids decreased from 14,400 mg/kg DM in the seeds to 6,200 mg/kg DM after boiling, to 10 mg/kg DM after debittering and below the LOQ in the snack (LOQ: ≤ 7 mg/kg). In the raw seeds besides lupanine (87%) also multiflorine (11%) and small amounts of sparteine, albine and α‐isolupanine were identified (see Appendix D, Tables D.1–D.4), but during the debittering process the QAs were removed gradually starting with sparteine after 12 h and ending with α‐isolupanine after 120 h.

Resta et al. (2008) investigated the changes in QA concentration due to food preparation. In total 16 samples of lupin protein isolates (LPI); food ingredient) and model foods were prepared (see Table 10) and analysed by GC–MS. In the protein isolates, only lupanine was identified. For the other processed foods, made using lupin flour, the authors report that the same QAs as in the seeds were found (see Appendix D, Tables D.1–D.4) but no amounts were given for the individual QAs. As can be seen from Table 10, the process from seeds to protein isolate is efficient in reducing QAs.

Based on the available information, the CONTAM Panel concluded that between 88 and 97% of the QAs present in seeds are removed by water treatment and boiling. Studies of the influence of germination on the QA concentrations show no clear effect.

3.3.1.1. Occurrence data and scenarios definition

Occurrence levels used to calculate acute dietary exposure in the different scenarios are summarised in Table 11.

Scenario 1 assumes that a consumer purchased lupin seeds and prepared them at home, upon no application of a debittering procedure or of a completely ineffective debittering procedure. In this scenario, it is therefore assumed that no reduction of the QA content is obtained. Occurrence results on samples reported in the category ‘Lupins (dry) and similar‐’ (n = 34) were used. However, based on the information provided, it was not possible to discriminate between sweet and bitter varieties. The CONTAM Panel decided to calculate dietary exposure using both the mean occurrence value of TotQA (429 mg/kg) and the highest reliable percentile (P90) to represent exposure from lupin seeds with average and high QA content.

In scenario 2, it is assumed that the lupin seeds are debittered by soaking and boiling in water and a reduction factor of 89% to the concentration of TotQA is applied (see Section 3.2.3). Occurrence results reported for ‘Lupins (dry) and similar‐’ (n = 34) were used but upon application of an 89% reduction factor. As for scenario 1, exposure was calculated both for mean and P90 to represent exposure from lupin seeds with average and high QA content.

In scenario 3, levels of TotQA in ‘Lupin seeds as canned or jarred legumes’ were used. It is assumed that the consumer buys pre‐prepared lupin seeds such as the one in brine. In this case, the number of samples (n = 8) was insufficient to calculate a reliable percentile, therefore only the mean TotQA at the LB and UB was used for exposure calculations.

Finally, scenario 4 was elaborated to simulate the exposure to QAs via the consumption of meat imitates. The mean TotQA concentration was used to calculated dietary exposure at the LB and UB. Only 8 samples were available and therefore also it was not possible to calculate a high percentile.

The CONTAM Panel considered the possibility to develop scenarios reflecting the exposure to QAs from lupin flour as an ingredient in bread and pasta and from coffee imitates. However, the number of samples on lupin‐based bread, pasta and coffee imitates was too limited to calculate dietary exposure. In addition, the uncertainty regarding he inclusion percentage of lupin flour in bread and pasta was too high to develop a scenario.

3.3.1.2. Consumption

The CONTAM Panel estimated acute dietary exposure to TotQA as defined in Section 2.2.2 from lupin and lupin‐based products.

The latest version of the EFSA Comprehensive Database (Annex A.1) was browsed for all the FoodEx2 categories containing lupins as a main constituent or as an ingredient. The consumption of ‘Lupins seeds without pods’ was reported only in a limited number of consumption days (63 for adults and less than 10 subjects in all the other population groups) in the Portuguese population. Consumption statistics in grams per day and grams per body weight per day are reported in Annex C.1. Mean and the high percentile (as the 95th) of acute consumption (in one day) were calculated for adults only as 0.8 and 1.8 g/kg bw per day, corresponding to 57 and 142 g/day of lupin seeds, respectively. It is noticeable that the portion size used by the BfR (2017a) to carry out the exposure assessment on the German population was 100 g and therefore can be considered in line with data extracted from the EFSA Consumption Database. Mean and high consumption values in adults for ‘Lupin seeds without pods’ were used for scenarios 1, 2 and 3.

The CONTAM Panel used the consumption of ‘Meat imitates’ retrieved in the EFSA Comprehensive Database as a proxy for meat imitates based on lupin seeds. The consumption across different population groups in 17 Member States is reported in Annex C.2. The largest consumers at the 95th percentile (based on more than 60 days of consumption) were the Swedish adults with a consumption of 6.9 g/kg bw per day corresponding to 330 g/day. Individual consumption data of ‘Meat imitates’ retrieved in EFSA comprehensive database were used for Scenario 4.

3.3.1.3. Results

Results of the current dietary exposure assessment for humans are reported in Annex C, and P95 dietary exposures are summarised in Table 12.

Under scenario 1, the 95th percentile of acute exposure was 761 μg/kg bw based on lupin seeds with average content of TotQA and 1,700 μg/kg bw for lupin seeds with high content of TotQA. Upon application of a debittering procedure (scenario 2) leading to an 89% reduction of exposure, the P95 acute exposure was 84 and 187 μg/kg bw for lupin seeds with average and high content of TotQA, respectively. Under scenario 3, where canned lupin seeds were considered as source of acute exposure to QAs, the mean exposure was 20–21 μg/kg bw and the P95 44‐46 μg/kg bw (LB–UB). It should be noted that these scenarios reflect the exposure of adults, but also other age groups may consume lupin seeds. However, due to limitations in the consumption data, no scenarios for other age groups, specifically children, were calculated.

In addition, according to scenario 4, consumption of lupin‐based meat imitates (Annex C.2), the P95 highest exposure was assessed to be 194–228 μg/kg bw (LB–UB) for adults from Sweden.

Studies in experimental animals

The acute toxicity of purified QAs has been investigated in experimental animals. Acute toxicity studies indicate that rats are less susceptible than mice or guinea pigs. All repeated‐dose toxicity data are confined to rats. Repeated‐dose studies have been performed using extracts from different Lupinus species, often containing different combinations of various QAs. Additional uncertainties in the interpretation of these studies arise from differences in the sources of QAs (flour versus seed extracts), cultivars, presence of lupin alkaloids other than QAs relevant for this opinion and the control diets used for the comparison.

Genotoxicity assays were performed with purified lupanine and sparteine as well as with L. termis extracts. While negative results were obtained for lupanine in a battery of assays, the data on sparteine are too limited to conclude regarding genotoxicity.

Observations in humans and hazard characterisation related to human health

Information on acute toxicity in humans is uncertain, as only sparse data were available from reports of poisoning cases with lupin seeds. The outcomes vary from fatal to less severe symptoms. Available reports do not allow to conclude on differences in toxicity of individual QAs or specific QA mixtures involved and on differences in the toxicity of seeds from different lupin species.

The reference point of 0.16 mg QA/kg bw identified by the Panel is derived by considering 20 mg of sparteine sulfate as the lowest single oral dose associated with pharmacological activity with respect to the endpoint of antiarrhythmic action. Uncertainty in the value of this reference point includes that information is missing (i) no information regarding the basis for this value, which was taken from old pharmacological reference literature, is available (ii) to which extent adverse effects may occur below the lowest‐observed‐effect‐level (LOEL), (iii) where to allocate a NOEL which is covering all known pharmacological activities including the effects on the electrical conductivity in the heart and also the oxytocic (uterotonic) action on the pregnant uterus. Furthermore, differences in the individual sensitivity to sparteine due to CYP2D6 genetic polymorphism are difficult to be quantified and it is not known whether this polymorphism may affect the sensitivity to QAs other than sparteine.

In the hazard characterisation of the QAs, the principle of dose addition was used and a similar potency was assigned to all QAs relevant for this opinion which is a conservative approach considering that sparteine is the most potent QA in terms of toxicity. As the database on relative potencies is limited and there is no information on combined effects of exposure to multiple QAs vs exposure to single QAs, there is uncertainty in using dose‐additivity with similar potencies in the hazard characterisation. Whereas it is acknowledged that interaction among QAs might occur both with regard to toxicokinetics and ‐dynamics, the CONTAM Panel considers it likely that grouping the QAs would result in an overestimation rather than an underestimation of the combined QA hazard.

Studies in farm animals, horses and companion animals and hazard characterisation related to animal health

Only sparse data are available on the kinetics of lupin QAs in farm animals, horses and companion animals. The information is limited to ruminants and pigs and no relevant studies have been performed with pure alkaloids.

Except for poultry and pigs, there is no information concerning the acute toxicity of QAs in livestock species. For repeated‐dose exposure, no relevant toxicity studies dealing with pure QAs could be identified, except for salmonids. In the large majority of toxicity studies, only endpoints related to zootechnical performances have been taken into consideration. Most studies used lupin seeds as source of QAs in the diet and do not report the QAs present in the diet or their relative contribution to the total QA concentration. Such studies do not allow identification of NOAEL/LOAELs and therefore hazard characterisation was not possible for horses, companion animals and farm animals, with the exception of salmonids. Instead, as an alternative approach, the CONTAM Panel identified doses of QAs present in seeds from L. luteus, L. albus and L. angustifolius that can be tolerated with respect to zootechnical performance. Because of the wide species‐ and cultivar‐related variation in total and specific QA content of lupin seeds, the calculated tolerated doses could not be extrapolated to all lupin species and cultivars.

For salmonids, similar NOAEL values were identified for lupinine and sparteine. In the absence of data on other QAs, the CONTAM Panel used this NOAEL for a group approach to characterise the hazard of the QAs relevant for this opinion and applied to all farmed salmonids. As for humans, the CONTAM Panel considers it likely that this grouping approach would result in an overestimation rather than an underestimation of the combined QA hazard.

G.1.2. Non‐ruminant animals