BOLETIM TÉCNICO No. 10 - www.micotoxinas.com.br


ROBUST ANALYTICAL PROCESSES FOR MYCOTOXINS - THE BACKBONE OF MEASUREMENT STRATEGIES AND RISK ANALYSIS FRAMEWORK FOR FOOD SAFETY* #

Achim Boenke* #

* European Commission; 1st Thematic Programme-Quality of Life & Living Resources; Key Action 1 - Health, Food & Nutrition; 200, Rue de la Loi; 1049 Brussels; Belgium.

SUMMARY

This article shows that robust analytical processes form: (a) the fundamental basis to assess the composition and properties of food and manufactured products, and (b) are essential for risk assessment as those require representative, reliable, and valid, data to allow a correct and objective intake assessment. Robust analytical processes start with the question "What is the content of a given mycotoxin in a given food matrix?" and end with the answer to it. Consequently, sampling and robust analytical processes are needed to provide an accurate and precise answer to this question. It is important to study, evaluate, and validate analytical methods and sampling plans in order to obtain reliable and comparable results. Different types of inter-laboratory and in-house studies based on acceptance criteria, Certified Reference Materials (CRMs), and proficiency testing schemes provide useful practical tools which are explained. Accreditation of food control laboratories forms nowadays an essential requirement, helps to maintain robust analytical processes, and its link to proficiency testing is also shown. Acceptance criteria are also applied for screening and confirmation procedures. Such procedures can form integral parts of a measurement strategy aiming at a high sample throughput at relatively low costs by using a screening technique and confirmatory techniques. Consequently, monitoring, surveillance and contamination prevention programmes become more cost-and time-effective as more samples can be investigated by providing a more representative picture of toxin occurrence. When robust analytical processes are applied in such programmes, intake and risk assessment becomes more accurate and precise leading to an effective protection of the consumer.

KEYWORDS

Analytical Methods, Certified Reference Materials (CRMs), Risk Assessment, Food Safety, ELISA.


Introduction

Robust (i.e. accurate, repeatable, reproducible, and generating results comparable to results obtained by others for the same property measured) analytical processes form the fundamental basis to assess the composition and properties of food, feed, and any other materials and manufactured products. Furthermore risk assessment also requires representative and reliable, data which must be of a valid quantity. Consequently, the uncertainty factors used during intake assessment depend strongly on the use of robust analytical processes to measure the occurrence of toxins and their metabolites in various commodities, tissues, etc.. The backbone characteristic of robust analytical processes for a measurement strategy is clear. It is becomes even more clear when noting that in 1988 250 million chemical analyses were performed each day in the USA. Out of these, about 10% were poor and had to be repeated at an additional cost of US$5 billion (Hertz 1988). Robust analytical processes consist of reliable sampling plans and analytical procedures as main parts. They start with the question "What is the content of a given mycotoxin in a given food and/or feed matrix?" and end with the answer to this question. Consequently, if sampling went wrong, not even the best validated method of analysis can help in providing the correct, accurate and precise answer to the question. In addition, the risk analysis will also lead to results which may generate serious problems in the political decision making process. It is important to carefully study, evaluate, and validate analytical methods and sampling plans in order to obtain reliable and comparable results. This article will provide: (a) an overview on what is a robust analytical process; (b) information on the possible tools and their use to obtain and maintain these; (c) explanations related to the impact of robust versus non-robust analytical procedures on risk assessment; and (d) information on the research support activities in the Key Action1 on Health, Food and Nutrition within the 1st Thematic Programme (Quality of Live and Living Resources (QoL)) within the 5th Framework Programme of the European Commission related to this discussed topic.

Discussion

Robust analytical processes

The analytical process starts with a question/aim such as "What is the content of a given property in a given food matrix?" and ends with the answer to this question (Eckschlager & Danzer 1994). Each of the steps of the analytical process need to be fully under control in order to make the whole robust. They must be accurate, repeatable and reproducible in order to generate results that are comparable with others obtained by different analytical processes. Only, if this is ensured the answer to the posed question or the aim will be correct and consisting of an acceptable uncertainty. Consequently, any error in any of the individual steps may lead to a large uncertainty and will make the answer less correct, less reliable, and/or less applicable. It is also important to note that the robustness of the analytical process implies stability over several dimensions; e.g. time, chemical/biological/separative/sensing approach, and including calibration. Any decreasing or increasing trend or variability will make the answer less correct and finally less robust. On-going control steps/mechanisms/programmes often specified as quality control and quality assurance systems/schemes are established to guarantee the robustness of analytical processes in different dimensions. Such systems/schemes make use of general and sometimes very specific criteria related to: e.g. minimum sampling variability (i.e. coefficient of variation (CV) for the sampling step), representative sampling, recovery rates, minimum number of ‘suspected false negatives and positives ‘ (i.e. samples detected by the applied test as being negative or positive but which are either positive instead of negative or negative instead of positive in reality). Further examples for such criteria are: minimum cross-reactivity of immunotechnology based procedures, linearity of the calibration curve, within-and between-day variability of the injection of calibrants, as well as within-and between-laboratory repeatability and reproducibility of results, minimum degree of chromatographic peak separation, etc. (Heitzman 1994, Boenke 1995(b) & 1998(a)). These criteria are either specified in or related to regulations, by international bodies (e.g. Codex Alimentarius), in norms/written standards, in industrial product requirements, and/or consumer health needs (Council 1985, European Committee for Standardisation (CEN) 1999, Wood 1997, Verardi et al 1995, Pittet 1995). The analyst has various tools at hand which help to ensure that each individual step of the analytical process is in conformity with its specific criteria and is consequently robust.

Tools for establishing and controlling robust analytical processes

Various tools exists for establishing and controlling robust analytical processes. These include the following different types:

(1) In-house studies;

(2) Various types of inter-laboratory studies including proficiency testing schemes; and

(3) Ruggedness-tests;

(4) The use of Certified Reference Materials (CRMs).

(1) In-house studies :

In-house studies are mostly based on a detailed investigation and evaluation of one single analytical procedure by :

(1a) studying its applicability for a range of matrices or different influences on its steps by checking its compliance to various acceptance criteria (e.g. within-lab, within-day repeatability and within-lab, between-day reproducibility) for spiked and naturally contaminated materials (Luckas et al 1994);

(1b) studying its accuracy for a range of matrices or different influences on its steps by comparing it with an already validated and robust analytical procedure or a CRM.

(2) Inter-laboratory studies are either based on :

(2a) one single analytical procedure or sampling plan applied by different laboratories (Stroka et al 1998), or;

(2b) various analytical procedures or sampling plans applied by different laboratories (Pettersson 1998).

The aims of such inter-laboratory studies are in case of (2a): to validate one analytical process and derive typical performance characteristics (e.g. repeatability, reproducibility, and accuracy); and in case of (2b): to either compare different analytical processes to identify systematic errors or to certify a mycotoxin content in a given matrix in order to obtain a CRM (see also sub-section (4) on CRMs below).

In addition, both of the above types of inter-laboratory studies can be organised as the so-called 'step by step approach'. This approach is consisting of a series of inter-laboratory studies following the different steps of the analytical process. An overview of this approach in the mycotoxin area is presented in the literature (Boenke 1998(a)). It is, in particular, preferred when the aim is to certify the content of mycotoxins in different food and feed matrices to obtain a CRM (Boenke 1995(a)).

Different examples of protocols and reporting sheets for inter-laboratory studies can be found in the literature (IUPAC 1988 & 1995, & ISO 1994 for inter-laboratory studies; Reynolds et al 1997 for end-determination step studies; Nicholls et al 1994 & Reynolds et al 1998 for studies involving the clean-up and the end-determination step based on one and different analytical procedures; Thompson et al 1995 & Coker et al 1998 for sampling plan studies; Ramsey et al 1995 for collaborative trials on sampling; AOAC International 1995 for analytical procedures validation studies; ISO 1986 for method performance needs and standardisation purposes).

Proficiency testing schemes :

Proficiency testing schemes are a special form of inter-laboratory studies aiming at the comparison of a laboratory's performance against that of similar laboratories and at the evaluation of the implementation of analytical procedures by analysts in different or in the same laboratories (Maier et al 1993). These schemes are also closely linked and sometimes direct part of the formal accreditation process of analytical laboratories (Eßer 1995 & Kohl 1996). A list of different proficiency testing schemes in Europe can be obtained from the individual National Accreditation Bodies in Europe, European Accreditation for Laboratories (EAL), EURACHEM, and EUROLAB. General information on the organisation and statistical assessment of proficiency testing of analytical laboratories is published elsewhere (IUPAC 1995). Information on the feasibility of proficiency tests in sampling is explained by (Argyraki et al 1995).

(3) Ruggedness-test approach :

In-house and inter-laboratory studies can both make use of the so-called 'ruggedness-test' (Werimont 1985) approach of which a recent practical approach has been described in the literature (Courtheyn et al 1996 and Visconti et al 1999). The ruggedness-test, one of the available chemometrical approaches, offers a cost-effective manner for generation of robust analytical processes. It can reduce the full factorial design of a series of test measurements by equally maintaining the required level of information.

(4) Certified Reference Materials (CRMs) :

Certified Reference Materials (CRMs) are the cost effective tools in obtaining and maintaining robust analytical processes. The BCR (Bureau Communitaire de Référence)-, M&T (Measurements and Testing)-and the SMT (Standards, Measurements and Testing)-Programmes have produced substantially more CRMs in both the area of food as a whole as well as for mycotoxin analyses, in particular, than any other major CRM-producer (Maier et al 1997). CRMs contribute to the validation of analytical procedures and sampling plans, and also satisfy a number of other needs such as making knowledge available for the preparation of inter-laboratory and proficiency scheme test-materials, and in-house quality control materials. Furtheron they form tools for and help during the implementation of written standards/norms, EU-regulations, laboratory accreditation, laboratory Total Quality Management (TQM), etc.. They can also be employed for calibration purposes and for checking the performance of analytical procedures, sensors needed by the different end-users to show that the repeatability of the analytical process is robust and compatible with the repeatabilities of those of the certifying laboratories. CRMs are currently available for a number of mycotoxins (Boenke 1997, SMT-& IRMM (Institute for Reference Materials and Measurements) home-pages: http://www.cordis.lu/smt/home/html & http://www.irmm.jrc.be/mrm.html). CRMs provide also an important milestone in the development of robust analytical processes which are composed of screening (e.g. ELISA, & bio-sensors) and confirmatory (e.g. HPLC/DAD, GC/MS) procedures. They are a cost-attractive tool for validation and quality control of test-kits and sensors. Taking also into account the rapid growth of automation in analytical laboratories, CRMs can substantially contribute to quality control of the accurate and precise (i.e. repeatable & reproducible) work of robots used for extraction and clean-up steps. Different examples of protocols for certification exercises can also be found in the literature (European Commission, SMT-Programme 1994 for general recommendations of BCR-certification studies; Wood et al 1995; ISO 1985 for general recommendations for certification exercises; Taylor 1985 for general recommendations of NIST (National Institute of Standards and Technologies)-certifications).

Accreditation:

The so-called 'New Approach and the Global Approach Directives' in Europe were responsible for the generalised development of accreditation systems. These systems were based on the EN 45001 written standard. This written standard was derived from the existing ISO/IEC Guide 25 (ISO/IEC 1985), which was used up to this time. Laboratory accreditation can be defined as a formal recognition by an authoritative body of the technical competence of a laboratory to perform tests or calibrations (ISO/IEC 1996). This accreditation body is providing the recognition and acts a third party between the laboratory and its clients. The aim is to establish confidence between these two parties. The overall main objective of accreditation can be summarised as 'once tested, everywhere accepted' in order to avoid barriers of trade. Today, accreditation is developing towards a competitive factor as part of a commercial and survival strategy of laboratories. It has also become mandatory for official food control laboratories in Europe. It is to be kept in mind that accreditation is different from the certification of quality systems as it is related to the evaluation of quality systems (e.g. quality control and quality assurance schemes) and the technical competence (e.g. to perform reliable analyses). As explained above (see sub-section (2)) proficiency testing schemes are an integral part of accreditation and ensure that technical competence is also maintained in the accredited laboratory. It has recently been reported that the percentage of unsatisfactory results in an accredited laboratory (13%) is lower than that in a non-accredited one (41%) (Cortez 1999). In addition, a lower percentage of major faults (deviations greater than 100% from the reference value, i.e., more than double, or less than half the reference value) was obtained for results obtain by an accredited laboratory (Cortez 1999).

 

Impacts and links of robust analytical processes on monitoring and contamination prevention programmes

The consumer thinks of food as that which sustains life based on an emotional level. She/He is extremely concerned when learning that food is comprised of a number of chemicals placed there by Mother Nature herself. Consequently, she/he sometimes moves her/his attention from natural toxicity to toxicity. Natural toxins can be hazardous, in particular, when the diet is focussed on consumption of certain foods. About 25% of the world's food crops are affected by mycotoxins every year (Mannon & Johnson 1985). The Council on Agricultural Sciences and Technology (CAST) estimated that in the US about US$ 20 million is lost annually on peanuts due to aflatoxin contamination, only (CAST 1989). The prevention and early detection of mycotoxin-contaminated raw materials is essential. Consequently, monitoring and surveillance programmes (WHO 1979) have been set-up in order to:

(a) provide overall re-assurance about safety of our food raw materials;

(b) identify emerging potential problems;

(c) ascertain any long-term changes in the contents of various substances (including mycotoxins) in food;

(d) assess the health significance of any such changes in relation to major changes in dietary habits;

(e) provide background material and a basis for decisions to remedy any problems which could arise.

The execution of such programmes consists basically of:

(a) eliminating mouldy and toxic raw materials from the food chain;

(b) following the contents of various substances in different foods;

(c) performing intake calculations on the basis of data on different countries diets.

In order to reach these above aims successfully, it is clear that robust and cost effective analytical procedures allowing a high sample throughput are a pre-requisite. For example, in Denmark about 1,125 samples of different matrices were analysed from 1986 to 1992 for the mycotoxin ochratoxin A (OTA) which leads to an average of about 187 samples per year for OTA (European Commission 1998). Within the Nestlé Mycotoxin Surveillance Programme a rough estimation of between 3,000 to 4,000 mycotoxin analyses (including OTA-analyses) are performed per year by their respective factory laboratories, regional laboratories, and in the quality and safety department (Pittet 1995). It can even happen that such high number of analyses have to be further increased in a shorter time interval as shown recently by the 'Dioxin-and PCB-Crises' were about 4,000 PCB-tests could only be carried out per week in Belgium but even more than these were required to be performed (Weser Kurier 1999). A measurement strategy composed of screening and confirmatory analytical measurements could be the approach of choice. Such screening detection techniques (e.g. ELISA, Bio-sensors) allowing a high sample throughput (e.g. 100 to 200 samples per day) at relatively low costs (e.g. ~100 EURO for ~100 samples in routine) would allow that many more samples can be investigated than with conventional techniques by providing a more representative picture of toxin occurrence. Only, the positive samples would consequently be needed to be confirmed by these confirmatory techniques at higher costs (e.g. ~500 EURO for ~5-10 samples in routine by e.g. GC/MS or HPLC/MS). In order to illustrate this the following example is provided here: 3,560 samples were measured for OTA, for example, in Germany from 1987 to 1995 out of which 2,763 were reported to have an OTA content below the limit of detection and 797 contained various levels of OTA (European Commission 1998). Now, if the above prices for screening and confirmatory analytical processes would be applied, the following would be obtained:

(a) Case A, applying only confirmatory type analytical processes such as HPLC:

3,560 samples x 50 EURO = 178,000 EURO;

(b) Case B, applying screening (e.g. ELISA) prior to confirmatory type (e.g. HPLC) analytical processes:

For screening: 3,560 samples x 1 EURO = 3,560 EURO;

Out of these 3,560 samples 797 samples were found positive and their OTA content need to be confirmed by HPLC: 797 samples x 50 EURO = 39,850 EURO.

In total: 3,560 EURO + 39,850 EURO = 43,410 EURO. In comparison to Case A 134,590 EURO could have, consequently, been saved.

However, also the screening detection techniques need to be robust, i.e. they need to generate a low percentage of 'suspected false negatives' (e.g. less than 5%). This can be examined by measuring the relevant mycotoxin in a wide variety of matrices together with validated robust analytical processes. Also a low number of 'suspected false positives' need to be generated by them to avoid that many analyses need to be repeated with confirmatory chemical techniques. This can be examined by analysing the reaction kinetics and/or by the specific blocking of immunoglobulines used by the assay. Specific blocking can, however, only be applied if the assay is carried out by using monoclonal antibodies. Experiments need to be designed where monoclonal antibodies would be used for coating the antibody in the ELISA. Consequently, specific fragments of these anti-idiotypic monoclonal antibodies would then be used to block the paratopes of the coated antibodies, specifically.

Similar criteria apply to the use of models or measurement of 'precursors' or 'precursor properties' aiming at measuring the ‘precursors’ as this is easier performed when their content in a given sample is clearly related (e.g. directly proportional) to the final/parent toxin of interest.

The World Health Organisation (WHO) (WHO 1980) provides a detailed description on important acceptance/validation criteria for kits for immunoassay and other protein binding systems such as: (a) analytical validity; (b) maintenance of assay to assay reproducibility; (c) assessment of random errors; (d) assessment of systematic errors; (e) specificity; (f) stability. Specific additionally important acceptance/validation criteria can also be formulated for sensors such as: (a) robustness; (b) very little maintenance and calibration; (c) very long life times; (d) stability at different temperatures; (e) selectivity and sensitivity; (f) low answering times; (g) low signal drift in relation to its zero point (Boenke 1998(b)).

Screening detection techniques such as ELISA are currently being available, developed and/or used for the major mycotoxins such as aflatoxins, fumonisins, OTA, zearalenone, and thrichothecenes. Their detection limits range from 0.05 ng/mL for aflatoxin M1 to 0.5 ug/kg for aflatoxin B1 in food. Their detection limits for fumonisins (B1 & B2) in maize based foods and maize samples range from 3 ug/kg to 10 ug/kg and up to 200 ug/kg. The Referee as part of the Inside Laboratory Management Journal monthly published by AOAC International provides a nice overview on screening detection test kits certified by the AOAC International PERFORMANCE TESTEDSM validation process (AOAC 1996). Another validation and recognition procedure recently developed in Europe is known as MICROVAL. MICROVAL was established in 1993 as an EUREKA-project and provides recognition for various test-kits. It is officially recognised by the International and European Standardisation Bodies (ISO & CEN).

Immunoaffinity columns (IACs) form another useful integral part of robust analytical processes. They provide an improvement in matrix elimination during clean-up in confirmatory techniques. An overview on IACs was published by Scott and Trucksess in 1997 (Scott & Trucksess 1997).

Impacts and links of robust analytical processes on risk analysis

Based on the risk analysis framework proposed by FAO/WHO (FAO/WHO 1995), three major parts form this framework. These ones are: (a) risk assessment; (b) risk communication; and (c) risk management. Each one of those major parts is consisting of various steps. Robust analytical processes have an important influence on the output of the following steps of risk assessment: (a1) hazard identification, (a2) exposure assessment, (a3) risk characterisation. Regarding hazard identification, various tests such as the Ames-test, epidemiological studies and animal toxicity studies need to be carried out in order to allow qualitative structure activity relationships (QSAR) predictions, determine the "no-observed-adverse-effect-level" (NOAEL) and the "lowest-observed-adverse-effect-level" (LOAEL), and detect carcinogenic and genotoxic potentials of mycotoxins. The costs of carcinogenicity studies are about US$2 million per study (Kuiper-Goodman 1998). These provide qualitative information and are today also used for quantitative risk assessments (Kuiper-Goodman 1998). As all of these tests require analytical measurements, it is essential that robust analytical processes are applied in order to keep the overall uncertainty as low as possible.

Exposure assessment is essentially based on monitoring and surveillance programmes by including also essential information on the effect of the manufacturing and processing of foods in industry and households. It is pretty obvious that robust analytical processes are needed. In particular, non-robust sampling strategies as part of robust analytical processes may contribute to a large extent to the overall uncertainty of the results. Coker et al (Coker et al 1998) demonstrated that the uncertainty of a sampling plan for aflatoxin B1 in feedstuffs is essentially dependent on the number of incremental samples taken per composite sample. The least precise plan (CV=12.5%) consisted of a single composite sample composed of 10 incremental samples, and the analysis of one test portion from the sample. The precision of the latter could be improved (CV=9.0%) by selecting 20 incremental samples and by performing duplicate analyses per sample. The CV here consists of the CV composed of sampling, sample preparation and analysis.

The influence of robust analytical processes becomes even more important when assessing the risk of mycotoxins in organically grown crops because: (a) fungicides are either not applied or are substantially reduced/limited as more food products based on biologically grown crops are demanded by consumers; and (b) the application of research related to genetically modified crops leading to more fungus resistance. Finally, several aspects of the HACCP-concept are substantially influenced and partly depending on robust analytical processes. These different aspects are all addressed in the new mycotoxin type research projects outlined in the next section.

Within food companies monitoring programmes with robust analytical processes are needed. In addition, suitable, cost effective, as well as robust on-, in-, and at-line analytical processes are demanded by food processing industry in order to ensure that questions can be answered correctly such as 'Will processing, including expected consumer use, eliminate mycotoxins or reduce their presence to an acceptable level?' or 'Is it likely that, at the stage of food processing, mycotoxins will be freed (bound toxins) or an existing mycotoxin may increase to unacceptable levels?'.

Robust analytical processes and RTD-activities of Key Action1 under Programme1 within the EC, 5th Framework Programme

The Key Action1 on Health, Food and Nutrition within the 1st Thematic Programme is an essential part of the 5th Framework Programme (5th FP) and is based on 'problem solving'. The Key Action1 is aimed at providing scientifically validated, reliable, and cost-as well as time-effective solutions to the large area of food safety to ensure consumer health and safety, preparation and updating of sound scientifically based EU-regulations, and finally strengthening of the competitive position of European's food and food processing industry. Consequently, its research objectives and deliverables include, for example (European Commission 1999): Development of rapid and cost effective detection tests (e.g. test-kits, portable bio-& chemical sensors, analytical processes), sampling plans, and pre-and co-normative research for pathogens, hormones, and undesirable substances such as natural toxins and chemical contaminants.

These above research objectives and deliverables take actively on board that the money being spent by every European household (i.e. on average 20 % of household's disposable income (European Commission 1997) on food and drink is not leading to consumer's health damage, trade, economical and job losses (i.e. European food and drink processing industries alone employ ~2.3 Million persons and in 1996 consumption within the Community will reach ~EURO 500,000 Million (European Commission 1997)). Furtheron, the 'problem solving research approach' is also aimed at avoiding foodborne illness leading to consumer's health damages such as these ones reported in 1990, where an average of 120 cases of foodborne illness per 100,000 population from 11 European Countries occurred (Jouve et al 1998).

Already, the first call for proposals with its deadline on the 6th June 1999 generated four high quality projects related to mycotoxins, i.e. ochratoxin A, Fusarium toxins, and Alternaria toxins. These projects have started either by the 1st January or the 1st February 2000. Three of these new projects aiming at the development and validation of mycotoxin prevention strategies and the implementation of the HACCP-approach for ochratoxin A and Fusarium toxins are combined in a 'Mycotoxin Cluster'. The fourth one aiming at the development of strategies for safe and high quality organically grown food supply by anticipating Alternaria mycotoxin risk is closely associated with it. All of these new projects will also generate new robust analytical processes such as screening detection tests and confirmatory analytical processes, both of which will form essential parts of the analytical strategy described in the previous sections. Such new robust analytical processes will aim to measure and consequently control the mycotoxin formation conditions. The clustering and association of the new mycotoxin projects will ensure an active and effective exchange, dissemination, exploitation and standardisation of generated information. This generated information may also be of valid input far beyond the mycotoxin or even the natural toxins area.

Conclusions

In the previous discussion section, a description of the robust analytical process together with an overview on its broad implications and different as well as important roles was outlined. In-house studies, various types of inter-laboratory studies, CRMs, and accreditation are helpful and important tools/approaches to establish and maintain robust analytical procedures. Without those, monitoring, prevention, surveillance programmes, and risk assessment would provide less meaningful results. Consequently, risk analysis of mycotoxins in food would become less accurate and would lead either to an under-or overprotection of the consumer. Both of this can not be accepted in any society because of various reasons.

In view of these broad implications and important roles of robust analytical processes, it can be concluded that such processes form a vital pre-requisite, if not one of the essential pillars within the Community's food law and its goals which are described as follows in the European Commission's Green Paper (European Commission 1997).

The Key Action1 on Health, Food and Nutrition in its ‘problem solving’ approach is substantially contributing to the Community's food law and its goals, already. In addition, Key Action1 is also generating and supporting the development of robust analytical processes such as modern screening techniques which form an essential pillar of a cost effective measurement strategy for food safety purposes.

References

Analytical Methods Committee, 1992, Proficiency testing of analytical laboratories: Organisation and statistical assessment, Analyst, 117: 97-104.

AOAC International, 1995, Quick and Easy! - AOAC Official Methods Programme, Gaithersburg, MD, USA.

AOAC International, 1996, AOAC research institute plans co-operative validation effort in Europe, Referee, February, pp. 17-21.

Argyraki, A., Ramsey, M.H., Thompson, M., 1995, Proficiency testing in sampling: Pilot study on contaminated land, Analyst, 120.

Boenke, A., 1995(a), BCR-and M&T-activities in the area of mycotoxin analysis in food and feedstuffs, Natural Toxins, 3: 243-247.

Boenke, A., 1995(b), Standardreferenzmaterialien (CRMs) - Anforderungen und Einsatz, Qualität im Analytischen Labor - Qualitätssicherungssysteme Maßnahmen zur Qualitätssicherung - Der ganzheitliche Qualitätsgedanke, edited by S.Kromidas (VCH-Verlag, Weinheim, DE, ISBN-3-527-28683-7), pp 136-157.

Boenke, A., 1997, The Food and Feed Chains - A Possible Strategy for the Production of Certified Reference Materials (CRMs) in the Area of Mycotoxins?, Food Chemistry, 60(2): 255-262.

Boenke, A., 1998(a), Method Validation for Mycotoxin Determinations in Food and Feedstuffs, Trends in Analytical Chemistry, 17(1): 10-17.

Boenke, A., 1998(b), Pure substances and a new generation of CRMs for chemical industry, Fresenius Journal Analytical Chemistry, 360: 388-392.

CAST, 1989, Mycotoxins: Economic and health risks, Report 116, Council for Agricultural Science and Technology, Ames, IA, USA.

Coker, R., Nagler, M.J., Defize, P., Derksen, G., Buchholz, H., Putzka, H.A., Hoogland, H., Roos, A., 1998, The development of sampling plans for the determination of aflatoxin B1 in animal feedingstuffs, EUR-Report no. 18542 EN, Brussels, BE, ISBN-92-828-4438-2.

Cortez, L., 1999, The implementation of accreditation in a chemical laboratory, Trends in Analytical Chemistry, 18(9+10) : 638-643.

Council Directive 85/591/EEC concerning the introduction of Communitty methods of sampling and analysis for the monitoring of foodstuffs intended for human consumption, O.J. L372, 1985, 31 December.

Courtheyn, D., Moermans, R., Schilt, R., Boenke, A., 1996, Food Addititives and Contaminants, 13: 795-810.

Eckschlager, K., Danzer, K., 1994, Information Theory in Analytical Chemistry, John Wiley & Sons, Inc., New York, USA, ISBN-0-471-59507-1, ISSN-0069-2883.

Eßer, F., 1995, Akkreditierung - Ein Mittel zur Sicherung und Steigerung der Qualität im Laboratorium, Qualität im Analytischen Labor - Qualitätssicherungssysteme Maßnahmen zur Qualitätssicherung - Der ganzheitliche Qualitätsgedanke, edited by S.Kromidas (VCH-Verlag, Weinheim, DE, ISBN-3-527-28683-7), pp 1-33.

European Commission, SMT-Programme, 1994, Guidelines for the certification of BCR-CRMs, Doc.BCR/48/93, version 15th December 1994.

European Commission, 1997, The general principles of food law in the European Union - Commission Green Paper, O.J. COM (97) 176 final, Brussels 30/04/1997, ISBN92-78-19509-X, Catalogue number: CB-CO-97-196-EN-C.

European Commission, 1998, Reports on tasks for scientific co-operation - Report of experts participating in task 3.2.2 - Assessment of dietary intake of ochratoxin A by the population of EU Member States - Revised edition of November 1997, EUR-Report no. 17523 EN, Luxembourg, LU, ISBN-92-828-2578-7.

European Commission, 1999, Commission Decision concerning the adoption of the work programme in the framework of the specific programme for research, technological development and demonstration on "Quality of Life and Management of Living Resources" (1998-2002), O.J. n° C/99/493, 4th March 1999.

European Committee for Standardisation (CEN), 1999, Food analysis – Biotoxins - Criteria of analytical methods for mycotoxins, Report Reference no. CR 13505:1999E, February, Brussels, BE.

FAO/WHO, 1995, Application of risk analysis to food standard issues: Report of the Joint FAO/WHO Expert Consultation, WHO, Geneva, CH.

Heitzman, R.J. (editor), 1994, Veterinary Drug Residues - Residues in food producing animals and their products: Reference Materials and Methods, Second Edition, EUR-Report no.15127 EN, Blackwell Scientific Publications, Oxford, GB, ISBN-0632-03786-5, pp.5/1-5/15.

Hertz, H.S., 1988, Are quality and productivity compatible in the analytical laboratory, Analytical Chemistry, 60 (2): 75A-80A.

ISO, 1985, Certification of reference materials - General and statistical principles. ISO Guide 35-1985 E (1985), International Organisation for Standardisation, Geneva, CH.

ISO 5725-2, 1986, Accuracy (trueness and precision) of measurement methods and results - Part 2: Basic method for the determination of repeatability and reproducibility of a standard measurement method. International Organisation for Standardisation, Geneva, CH.

ISO/IEC, 1985, ISO/IEC Guide 25: General requirements for the competence of testing and calibration laboratories, Geneva.

ISO/IEC, 1996, ISO/IEC Guide 2: Standardisation and related activities - General vocabulary, Geneva.

IUPAC, 1988, Protocol for the design, conduct and interpretation of collaborative studies, Pure and Applied Chemistry, 60: 856-864.

IUPAC, 1995, Harmonised protocols for the adoption of standardised analytical methods and for the presentation of their performance characteristics, Pure and Applied Chemistry, 62(1): 149-162.

Jouve, J.L., Stringer, M.F., Baird-Parker, A.C., 1998, Food safety management tools, Report prepared under the responsibility of ILSI-EUROPE's Risk Analysis in Microbiological Task Force, April, Brussels, ISBN1-57881-034-5.

Kohl, H., 1996, Qualitätsmanagement im Labor - Praxisleitfaden für Industrie, Forschung, Handel und Gewerbe, Springer, Berlin, DE, ISBN-3-540-58100-6, pp. 146.

Kuiper-Goodman, T., 1998, Food Safety: Mycotoxins and Phycotoxins in Perspective, Mycotoxins and Phycotoxins - Developments in Chemistry, Toxicology and Food Safety, edited by M.Miraglia, H.P. van Egmond, C. Brera, J. Gilbert (Alaken, Inc.Fort Collins, Collorado, USA, ISBN-1-880293-09-9), pp 25-48.

Luckas, B., Hummert, C., Thielert, G., Kirschbaum, J., Boenke, A., 1994, Improvement of HPLC/Fluorescence and HPLC/MS procedures for the determination of DSP-and PSP-toxins in contaminated shellfish, EUR-Report no. 15339 EN, Brussels, BE, ISSN-1018-5593.

Maier, E.A., Quevauviller, Ph., Griepink, B., 1993, Inter-laboratory studies as a tool for many purposes: proficiency testing, learning exercises, quality control and certification of materials, Analytica Chimica Acta, 283: 590-599.

Maier, E.A., Boenke, A., Mériguet, P., 1997, Importance of the Certified Reference Materials Programmes for the European Union, Trends in Analytical Chemistry, 16(9): 496-503.

Mannon, J., Johnson, E., 1985, Fungi down on the farm, New Scientist, 195: 12-16.

Nicholls, S., Suett, D.L., 1994, Intercomparison studies: Insecticides (N,P-Compounds) in cereals - Second intercomparison study on cereal grain treated with insecticides, EUR-Report no. 16921 EN, Brussels, BE, ISBN-92-827-3912-0.

Pettersson, H., 1998, Intercomparison of thrichothecene analysis and feasibility to produce certified calibrants and reference materials, EUR-Report no. 18214 EN, Brussels, BE, ISBN-92-828-3224-4.

Pittet, A., 1995, Keeping the mycotoxins out: Experience gathered by an international food company, Natural Toxins, 3: 281-287.

Ramsey, M.H., Argyraki, A., Thompson, M., 1995, On the collaborative trail in sampling, Analyst, 120: 2309-2312.

Reynolds, S.L., Fussel, R., Caldow, M., James, R., Nawaz, S., Ebden, C., Pendlington, D., Stijve, T., Diserens, H., 1997, Intercomparison study of two multi-residue methods for the enforcement of EU-MRLs for pesticides in fruit, vegetables and grain - PhaseI: Intercomparison study of pesticides solutions, EUR-Report no. 17870 EN, Brussels, BE, ISBN-92-828-1666-4.

Reynolds, S.L., Fussel, R., Caldow, M., James, R., Nawaz, S., Ebden, C., Pendlington, D., Stijve, T., Diserens, H., 1997, Intercomparison study of two multi-residue methods for the enforcement of EU-MRLs for pesticides in fruit, vegetables and grain - PhaseII: Intercomparison study of spiked test materials, EUR-Report no. 18639 EN, Brussels, BE, ISBN-92-828-5084-6.

Scott, P.M., Trucksess, M., 1997, Application of Immunoaffinity Columns to Mycotoxin Analysis, Journal of A.O.A.C. International, 80(5): 941-949.

Stroka, J., Anklam, E., 1998, Validation of analytical methods to determine the content of aflatoxin, ochratoxin A, patulin in foodstuffs of vegetable origin - Immunoaffinity clean-up with liquid chromatography using post-column bromination for the determination of aflatoxins in peanut butter, pistachios, dried figs and paprika: collaborative study, EUR-Report no. 18640 EN, Brussels, BE, ISBN-92-828-5085-4.

Taylor, J.K., 1985, Standard Reference Materials: Handbook of SRM users, NBS Special Publication no. 260-100, US Department of Commerce, National Bureau of Standards, Gaithersburg, MD 20899, September, USA.

Thompson, M., Ramsey, M.H., 1995, Quality concepts and practices applied to sampling - An exploratory study, Analyst, 120: 261-1356.

Verardi, G., De Froidmont-Görtz, I., 1995, Overview on Community Legislation in the field of official control of mycotoxins in feedingstuffs, Natural Toxins, 3: 248-256.

Visconti, A., Solfrizzo, M., De Girolamo, A., Boenke, A., with participation of Gilbert, J., Brereton, P., Anklam, E., Von Holst, C., Van Egmond, H.P., Sizzo, E., 1999, Fumonisins in Maize and Maize Based Foods, EUR-Report EUR19096EN, Brussels, BE, ISBN-92-828-8018-4.

Werimont, G.T., 1985, Use of statistics to develop and evaluate analytical methods, Association of Official Analytical Chemists (AOAC), Arlington, VA, USA.

Weser Kurier, 1999, Feilschen um Dioxin-Kontrollen - Belgien streitet mit der EU wegen Waffeln und Mayonnaise, 9th August 1999, Bremen, DE.

WHO, 1980, Requirements for kits for immunoassay and other protein binding systems, Document number: WHO/BS/80.1270/HRP/80.1, Geneva, CH.

WHO, 1979, Guidelines for establishing or strengthening national food contamination monitoring programmes.

Wood, G.M., Entwisle, A.C., Farnell, P.J., Patel, S., Boenke, A., 1995, The certification of the ochratoxin A mass fraction of two wheat RMs (CRMs 471 & 472), EUR-Report no. 16825 EN, Brussels, BE, ISBN-92-827-6320-X.

Wood, R., 1997, On the Regulatory Horizon – International bodies endorse criteria-based approach to methods, Inside Laboratory Management, 23-25.