1 r e hapt C

chapter 1. Human exposure to aflatoxins and fumonisins

Data on the prevalence of myco- clear picture of the extent to which their role in disease development, toxins in staple foods are essential mycotoxins compromise food safety and determine the efficacy of in- for all applied research into their and health, at either an individual or tervention strategies. The recent impact on health and on effective a population level. However, this is development of multitoxin analytical mitigation. Country- or region- generally not achieved in develop- methods, whether applied to food or specific knowledge enables the ing countries, primarily due to a lack to biological samples as biomark- identification of susceptible edible of country-specific data, resources, ers, has raised awareness of the crops that are responsible for toxin and analytical capacity. concurrent exposure to aflatoxin exposure in specific populations. Exposure biomarkers, such as and fumonisin as well as sometimes Prevalence data can indicate how serum aflatoxin–albumin adducts to other, unanticipated mycotoxins. effective maximum levels have been (AF–alb) or urinary fumonisin B1 in influencing food safety, while (UFB1), offer a more integrated es- Exposure to aflatoxins acknowledging that their enforce- timate of exposure from all sources ment could have food security im- for either aflatoxin or fumonisin, and Aflatoxins are mycotoxins found plications. Monitoring of prevalence offer potentially more reliable expo- in four main forms: aflatoxin B1 also provides information on how sure estimates. Measurement of ex- (AFB1), B2 (AFB2), G1 (AFG1), and various implemented strategies to posure, either by measures of food G2 (AFG2). Aflatoxins occur on a reduce contamination or exposure consumption combined with con- wide range of crops, including the levels directly affect toxin levels. tamination levels or by using bio- major staple cereals (e.g. maize), Ideally, exposure assessment, as markers of exposure, can be used edible nuts and legumes, and their one component of risk assessment, to identify the main dietary contribu- products. In general, AFB1 occurs integrates mycotoxin levels with tors to exposure, detect areas with at the highest levels and is the most food consumption patterns and thus unacceptable exposures, assess toxic. The main fungal producers provides, via risk characterization, a health impacts of mycotoxins and of aflatoxins are Aspergillus flavus,

Chapter 1. Human exposure to aflatoxins and fumonisins 1 which produces AFB1 and AFB2, This compound can be detected in income regions. However, it is im- and Aspergillus parasiticus, which the urine and milk of exposed ani- portant to note that these estimates produces all four forms. Contamina- mals, including humans. Data on are based on very limited data- tion can occur before or after harvest the carryover of AFM1 to breast milk sets, particularly in those regions at or both. are limited, but the carryover has greatest risk of high exposures. Aflatoxin contamination levels been estimated at 0.1–0.4% (Zarba can vary widely, from products that et al., 1992), and exposure of in- Exposure to fumonisins meet the strict maximum levels fants to AFM1 from human breast set by the European Commission milk has been reported in devel- Fumonisins, which are produced

(2 µg/kg for AFB1; 4 µg/kg for total oping countries (Shephard, 2004; mainly by Fusarium verticillioides aflatoxins [sum of AFB1, AFB2, Turner, 2013; Magoha et al., 2014). (Sacc.) Nirenberg and F. proliferatum

AFG1, and AFG2] for cereals and In addition, AFM1 from milk of live- (Matsush.) Nirenberg, are common nuts for direct human consumption) stock consuming AFB1-contami- contaminants of maize and maize-

(European Commission, 2010) nated feed is a further source of based products. Fumonisin B1 (FB1) to products with levels that can exposure. The 56th meeting of the is the most abundant (generally pose a risk of acute aflatoxicosis. Joint FAO/WHO Expert Committee ~70% of the total fumonisin contam- For example, determination of on Food Additives (JECFA) com- ination), and it normally co-occurs total aflatoxins in a rural market piled data on AFM1 levels found with lesser amounts of fumonisin B2 survey in four districts during an in commercial raw and processed (FB2) and B3 (FB3). Occurrence on acute outbreak in Kenya, in 2004, dairy milk (Henry et al., 2001). sorghum has also been reported showed a range of total aflatoxins of However, few data were available (Bulder et al., 2012). 1–46 400 µg/kg, with 7% of samples from Africa, and those reported are Fumonisins were evaluated by above 1000 µg/kg (Lewis et al., unlikely to reflect typical village- or JECFA in 2001 and 2012 (Bolger 2005). In 2003, data available from subsistence farm-level exposures. et al., 2001; Bulder et al., 2012). As African countries were summarized Further study is needed to better exposure is a product of both con- by Shephard (2003). More recent understand the consequences of tamination level and consumption, data, including summaries of global AFM1 ingestion from breast milk certain rural communities in de- occurrence in samples submitted and/or from the milk of livestock in veloping countries can exceed the for analysis, have been presented Africa. provisional maximum tolerable dai- by Rodrigues et al. (2011) and Global intake estimates for af- ly intake (PMTDI) of 2 μg/kg bw/day Schatzmayr and Streit (2013). latoxin (ng/kg body weight [bw]/ of fumonisin if their diet contains Recent African data have also been day) have been reported based high amounts of maize (Burger et provided by Gnonlonfin et al. (2013). on estimates of typical maize al., 2010). Examples from this literature include and nut consumption, con- Fumonisin intake estimates (µg/ groundnut cake from Nigeria (range, tamination levels, and body kg bw/day) in several regions of 20–455 μg/kg); raw groundnut weight (Liu and Wu, 2010). For Africa were recently reviewed (Wild from Kenya (non-detectable to Africa, estimates were made for and Gong, 2010), including Burki- 7525 μg/kg) and Botswana (12– the Democratic Republic of the na Faso (0–2); Bizana (1–19), Cen- 329 μg/kg); and maize from Benin Congo (range, 0–27), Ethiopia (1– tane (2–36), Transkei (4), and Kwa­ (2–2500 μg/kg), Ghana (20–355 μg/ 36), The Gambia (4–115), Kenya Zulu-Natal (0), South Africa; and kg), and Zambia (1–109 μg/kg). (4–133), Mozambique (39–180), Bomet, Kenya (< 0.1). Intakes of Other aflatoxin-contaminated food Nigeria (139–227), South Africa 0.2–26 µg/kg bw/day in Tanzanian sources reported in various African (0–17), the United Republic of Tan- children were reported (Kimanya countries include cassava, tiger zania (0–50), and Zimbabwe (18– et al., 2014). nuts, cowpeas, sorghum, okra, 43). Similarly high intakes were In Latin America, estimates of and hot peppers, although due to reported for China and countries fumonisin intake in Guatemala were consumption patterns, maize and in South-East Asia, compared with reported to be 3.5 µg/kg bw/day groundnuts dominate in terms of western Europe and North Amer- (urban) and 15.5 µg/kg bw/day (rural) level of exposure. ica at 0–1 ng/kg bw/day (Turner (Wild and Gong, 2010), and more

Aflatoxin M1 (AFM1) is a toxic et al., 2012; Schleicher et al., recently a range of 0.20–23 µg/ metabolite of AFB1 and a possible 2013). These data indicate a much kg bw/day was reported (Torres et human carcinogen (IARC, 2012). higher burden of exposure in low- al., 2014).

2 1 r e

Biomarkers for aflatoxins Health and Nutrition Examination cally significant correlation was ob- hapt C and fumonisins Survey (NHANES) were almost all served between the concentrations (99%) below the limit of detection of these biomarkers (r = 0.375, Food contamination and food intake (LOD), and the geometric mean of P < 0.001) (Shirima et al., 2013). can vary greatly within rural subsis- the positives was only 0.8 pg/mg Urinary aflatoxin and fumonisins tence farm settings and between vil- (Schleicher et al., 2013). were observed less frequently in lages and individuals. Assessments AF–alb has also been used in samples from two major cities, of both of these parameters present various studies to assess associa- Yaoundé and Bamenda, in Cam- analytical and measurement diffi- tions between aflatoxin exposure eroon (Abia et al., 2013) and from culties. In addition, there is interin- and infant and early childhood rural regions of Nigeria (Ezekiel et dividual variation in toxicokinetics growth faltering (Turner, 2013). al., 2014), although co-exposures and toxicodynamics related to toxin Typically there is greater confi- did occur. Differences in the sen- ingestion. For these reasons, con- dence in the long-term markers sitivities of the analytical methods siderable effort has been given to of aflatoxin exposure to assess between these studies limit direct developing biomarkers for aflatoxins health outcomes, as they provide comparison. A separate study from and fumonisins (Turner et al., 2012). an integrated measure over several Cameroon, looking at urinary my-

For AFB1, the peripheral blood months. Several putative biomark- cotoxin markers in young children, AF–alb biomarker has been validat- ers for fumonisin exposure have also reported aflatoxin and fumoni- ed for moderate- to long-term expo- been investigated. These include sin exposure (Njumbe Ediage et al., sure (several months), whereas the sphingoid bases in plasma and 2013). These data were comple- urinary biomarkers, aflatoxin–N7- urine and FB1 in hair, nails, serum, mented by a survey across multiple guanine and AFM1, reflect shorter urine, and faeces (Shephard et al., agro-ecological zones in Camer- exposures. The application of these 2007); however, none of these have oon, in which maize, groundnuts, biomarkers has helped establish been validated in human studies. and cassava were found to be con- the link between aflatoxin exposure UFB1 has been measured in hu- taminated with multiple mycotoxins and the development of liver cancer man samples in regions with known (fumonisins were found in 74% of (Kensler et al., 2011; IARC, 2012) high exposure to dietary fumonisins the maize samples and aflatoxins and has allowed the efficacy of in- (Gong et al., 2008a; Xu et al., 2010; in 22% of the maize, 29% of the tervention studies to be demonstrat- van der Westhuizen et al., 2011; Ri- groundnuts, and 25% of the cas- ed (Turner et al., 2005). ley et al., 2012; Torres et al., 2014). sava samples) (Ediage et al., 2014). Validated aflatoxin biomarker In general, statistically significant In a study by Probst et al. (2014), a data from sub-Saharan Africa show relationships between UFB1 and ei- total of 339 maize samples from 18 that the ranges of exposures are ther estimated or measured FB1 in- countries in Africa were assessed likely to vary greatly in many re- takes were reported; however, the for aflatoxin and fumonisin contam- gions and within and across closely data indicate that the urinary mea- ination. Aflatoxins were detected located villages and agro-ecolog- sure was only moderately reflective (LOD, 1 µg/kg) in 47% of the sam- ical zones, as well as seasonally of the level of intake. ples, with 7% exceeding 20 µg/kg and annually (Turner et al., 2012; and 6% exceeding 100 µg/kg (the Turner, 2013). The biomarker data Co-occurrence of aflatoxins maximum level was 1409 µg/kg). further highlight the early-life bur- and fumonisins Fumonisins were detected (LOD, den of exposure, including in utero 500 µg/kg) in 81% of the samples, and during early infancy. Exposures The co-occurrence of aflatoxins and with 7% exceeding 5000 µg/kg and in West African studies involve both fumonisins has been widely docu- 3% exceeding 100 000 µg/kg. Afla- maize and groundnuts as the pri- mented by both biomarker studies toxin and fumonisin co-contamina- mary sources of intake of aflatox- and food analyses. In the United tion occurred in 35% of the samples. ins. Typical biomarker levels in chil- Republic of Tanzania, AF–alb and Concentrations of co-contaminants dren younger than 5 years in Benin, UFB1 were assessed in young varied by region, but for the Coast The Gambia, and Togo range up children (Shirima et al., 2013). The Province in Kenya, for example, to 1000 pg aflatoxin–lysine/mg al- prevalence of detection of both of 50% of samples contained high bumin (Turner, 2013). By compari- the mycotoxins was high, and 82% levels of both aflatoxins (mean, son, levels of AF–alb reported from of the children were positive for 97 µg/kg) and fumonisins (mean, the recent United States National both. Also, a modest but statisti- 32 000 µg/kg) (Probst et al., 2014).

Chapter 1. Human exposure to aflatoxins and fumonisins 3 In Latin America, co-exposures Key scientific gaps mycotoxin detection in food com- to aflatoxins and fumonisins have modities is a complex task, there is also been documented. Maize from The problem of mycotoxin exposure a tool available to support countries 22 districts in Guatemala was ana- is most acute in developing coun- in this regard: the Food and Agri- lysed; 36% of 572 samples tested tries, which lack resources and ana- culture Organization of the United positive for aflatoxins (mean, 63 µg/ lytical capacity for analyses. Conse- Nations (FAO) Mycotoxin Sam- kg; range of positives, 5–2655 µg/ quently, few data are reported from pling Tool (http://www.fstools.org/ kg), and 99% of 640 samples test- developing countries and those mycotoxins/). Further, there is a ed positive for fumonisins (mean, available are usually based on only World Health Organization (WHO) 1800 µg/kg; range of positives, 10– a limited number of samples of un- programme (Global Environment 17 000 µg/kg) (Torres et al., 2015). certain quality. As a result, there is Monitoring System – Food Contam- a widening gap between the qual- ination Monitoring and Assessment Analytical limitations ity and quantity of prevalence data Programme [GEMS/Food]) that generated by laboratories in devel- collects global food contamination One limitation with urinary biomark- oped countries compared with de- data and reports food consumption er approaches is the volumes of veloping countries. There is thus a data. Average per capita food con- urine required. Even though techno- need in the developing countries to sumption data are reported based logical development of highly sen- have sampling and analytical tools on the FAO Food Balance Sheet sitive liquid chromatography-mass available that are fit for specific pur- data. It is important to note that the spectrometry (LC-MS) techniques poses, such as: database provides average con- will help support biomonitoring, the • A rapid screening method aimed sumption levels but will not capture approach itself may be limited by at the field/subsistence farm lev- the food consumption pattern at the instrumentation costs, restricting el that is inexpensive and user- subsistence farm level. Another da- analysis to specialist laboratories. friendly and has a wide dynamic tabase within GEMS/Food collects With the development of multitoxin analytical range. This could addi- occurrence data for contamina- analytical techniques based on tionally help support a rapid alert tion levels, including aflatoxins and LC-MS/MS, multibiomarker meth- system that informs responses and fumonisins in food products and ods have been developed for urinary appropriate actions for food safety. crops. It would be useful to highlight biomeasures for toxins, including • A comprehensive regional or coun- the opportunity for researchers to

FB1 and AFM1 (Solfrizzo et al., 2011; try-wide monitoring programme, add their studies to this database. Warth et al., 2012), as extensions involving the establishment of However, acquiring data on con- of multimycotoxin methods for food a reference laboratory within a sumption and contamination levels analysis. These methods have been country/region. The monitoring in subsistence farmers will remain a applied in Africa to evaluate expo- programme should be developed significant hurdle. sure (Abia et al., 2013; Shephard within existing surveillance sys- Among monitoring options, an et al., 2013; Ezekiel et al., 2014). To tems and be expanded over time. approach that might be implement- date, there have been limited efforts For example, many regions have ed is sampling at community maize to compare multimycotoxin methods national health and nutrition pro- milling facilities. For example, in from different laboratories. Thus, grammes where archived biospec- some parts of East Africa farmers currently there is greater confidence imens could be requested. Future could bring maize to a local milling in the data from single measures, national surveys of this nature may operation, where subsampling and and for increased utility these inter- be asked to collect larger volumes aflatoxin and fumonisin analyses laboratory comparison studies are of biospecimens (e.g. to support could be carried out using rapid urgently needed. An additional con- urinary xenobiotic surveillance). test kits for field application. Rela- cern is that some of the multimyco- De novo monitoring activities could tively large data collection activities toxin methods, especially for foods, include both food measures and may be possible in such settings, may be measuring contaminants of biomarkers. providing an improved surveillance, limited relevance to human health. For a successful food monitoring although this will capture only some This could result in additional costs programme, it is essential to have of the prevalence data in some re- (e.g. of measuring > 60 metabolites) effective sampling plans in place. gions and none in others. This also while potentially leading to inaccu- While it is recognized that design- may, however, provide a target site rate measurements. ing effective sampling plans for for intervention.

4 1 r e

Measures of individual exposures sure is known to be high. The lack predictive of the level of intake com- hapt C are important for epidemiological in- of reagents such as aflatoxin– pared with relationships reported for vestigations of disease causation lysine and mono-adducted AF–alb aflatoxin biomarkers. For general and for demonstration of efficacy is a major constraint and needs to be biomonitoring this is not a major is- of intervention. The development of addressed. Enzyme-linked immuno- sue; however, this is a concern when a reliable source of certified stan- sorbent assay (ELISA) approaches making assessments in relation to dards, especially for aflatoxin bio- are typically less expensive, but an putative health effects and assess- markers, would allow a substantial additional issue is a lack of commer- ing the efficacy of interventions. For increase in biomarker-directed epi- cially available kits or antibodies. the use of FB1 and AFM1, it was noted demiology research. While LC-MS provides robust data, that neither of these predicts longer- Therefore, the problem of insuf- the analytical costs are prohibitive term exposures, and while serum ficient data could also be addressed for most laboratories. Exposure of AF–alb is used for this purpose in af- by the use of individual biomark- in­fants in developing countries to latoxin biomonitoring and epidemiol- ers of exposure. Aflatoxin biomark- AFM1 also needs to be monitored as ogy, there remains a need to develop ers are well understood, but the these countries are prone to higher a long-term exposure biomarker for most useful for long-term exposure AFB1 exposures. fumonisin. An additional challenge studies, AF–alb, is currently mea- UFB1 has been measured by is the need for higher-throughput sured in only a limited number of LC-MS in several world regions, and analytical tools, which would benefit laboratories. It would be advanta- again a current concern is the cost of from a cooperative activity between geous if this analysis were more the analysis. While dose–response experts in exposure assessment and generally available, especially in relationships were reported, the uri- researchers with subject matter ex- countries where aflatoxin expo- nary measure was not as strongly pertise in mycotoxins.

Chapter 1. Human exposure to aflatoxins and fumonisins 5