The health status of the population has clearly decreased in recent decades with especially a burst of chronic diseases, particularly obesity, type II diabetes, cardiovascular diseases, cancers, neuro-developmental disorders, neonatal malformations [1,2,3,4]. Such a dramatic increase in the prevalence of these diseases cannot be explained by genetic factors nor by infectious origins. Indeed, no Mendel law can account for this considerable rise in the number of patients over such a small number of generations. In addition, it is clear that no infection can be incriminated in most identified cases.

Understanding this phenomenon requires going back in time. In the nineteenth century, the main cause of mortality and morbidity was the spread of infectious diseases, as we had little knowledge and means to fight against bacteria and viruses, and diagnostic methods were not very effective. Large epidemics (plague, cholera, smallpox, syphilis, rabies, tuberculosis), but also infectious diseases in the specific field of perinatal care, have been responsible for significant mortality. Moreover, these infections were favored by a decrease in the immune defense for an important part of the population, due to malnutrition.

At the beginning of the twentieth century, two major revolutions took place: the introduction of bacterial hygiene thanks to Louis Pasteur and the discovery of penicillin in 1928 by Alexander Fleming, and sulfamides in 1935 by Gerhard Domagk, who led to the development of antibiotics. Alongside these major discoveries, technical progress has led to significant advances in diagnostic and examination methods, as well as in surgical treatments. This has resulted in an increase in life expectancy and a decrease in morbidity (at least in the so-called “developed countries”), which continued steadily until the 1990s.

Since the mid-1990s—early 2000s, there has been a reversal of this trend, with a slowdown in life expectancy growth in European Union and other high-income OECD countries [5], a decrease in healthy life expectancy in Europe [6] and France [7] and even a recent decrease in life expectancy in high income countries, especially United Kingdom and United States, during 2014–2015 [8]. This recent situation in the United States is keeping with projections for the twenty-first century made previously [9].

At the same time, the dominance of chronic diseases as major contributors to total global mortality has emerged [10], with especially a burden of cancers [11,12,13,14,15] and cardiovascular diseases [16]. Cancer cases increased by 33% between 2005 and 2015 [17]. In 2015, cancer caused over 8.7 million deaths globally and was the second leading cause of death in the world behind cardiovascular diseases [18]. This period was also characterized by the increase of other chronic diseases, such as type 2 diabetes and obesity which is a major contributor to the growing incidence of cancer [19], neurodegenerative disorders [20] or sexual differentiation and development anomalies [21,22,23].

Many studies highlighted the link between these diseases and environmental factors, including exposure to chemicals such as bisphenols [24, 25], lead [26], pesticides [27, 28], asbestos [29], dioxins [30], phthalates [31] and other endocrine disruptors [32,33,34,35,36,37,38,39,40,41,42].

The same established fact was first made for wildlife where many animals displayed genital tract and reproductive abnormalities related to exposure to endocrine disruptors and other xenobiotics that accumulate in ecosystems and organisms, leading to the decline of some species [43,44,45,46,47,48,49,50,51].

Although environmental causes are always multifactorial, it is clear that xenobiotics with endocrine disrupting properties, including many pesticides and other pollutants have played—and are still playing—an important role in the human chronic diseases burst and the decline of animal biodiversity.

Why are these substances so easily and massively authorized to be marketed despite their dangerousness? This is the question we propose to try to answer here, while considering concrete solutions to overcome this deficiency.


Is regulatory assessment able to protect health and the environment?

Obviously, regulatory assessment agencies both at the national level (ANSES for France) and at the European level (EFSA) do not sufficiently protect citizens’ health and the environment. Major health and environmental scandals of past years (Distilbene [diethylstilbestrol], Mediator [benfluorex], asbestos, polychlorobiphenyls [PCBs], bisphenol A [BPA], chlordecone, neonicotinoids, bovine spongiform encephalopathy, nuclear disasters) only demonstrate this established fact, and undoubtedly contribute to the huge gap that has opened between citizens and risk assessment agencies, with the latter sometimes inspiring more mistrust than trust. The causes for such a situation are multiple, and sometimes indirect and contextual. We will list and attempt to analyze the main ones through some concrete examples.

Is Regulatory toxicology suitable for most pollutants, especially endocrine disruptors?:

Many xenobiotics and other pollutants, especially endocrine disruptors (EDs), have properties that allow them to escape the screens of regulatory toxicology (RT).

  • EDs do not have the same effects according to the sex [24, 25, 31, 38], while the RT considers that a disparity of effects between male and female rats is due to the natural variability of the rats and not to the tested molecule.

  • EDs effects are not always proportional to the tested dose [52,53,54], whereas RT predicts that the effect-dose response should be monotonic.

  • EDs do not have the same effects depending on the exposure period, embryonic and fetal development being the period of greatest susceptibility [55, 56]. However, most often, RT consists of either acute toxicology tests that last 3 months in adult rats or chronic toxicology tests conducted for 2 years (whole life of rats), but rarely tests on pregnant rats with analysis of the offspring.

  • EDs may have transgenerational effects, over several generations, without the latter having been exposed to these pollutants [57, 58], but RT does not provide for transgenerational analyses.

  • Most of EDs are mixtures, such as commercial formulations of pesticides, including an active principle and co-formulants. While most of commercial formulations are more toxic than the declared active ingredient [59, 60], only the latter is assessed in the context of RT. Moreover, some surfactants proved to be more toxic as such than the active ingredient itself, as it is the case for polyethoxylated tallow amine (POEA) and alkyl-polyglucoside in Roundup formulations [61].

Are Exposure limit values sufficiently protective?:

The enacted standards of references or the provisory tolerable weekly intakes (PTWIs), as well as acceptable daily intake (ADI) are not always the most protective. First, the ADI only makes sense for molecules with a very short lifespan (less than 24 h). Indeed, for stable (and therefore bio-accumulable) molecules, such as EDs, the dose admitted on D-day is added to those absorbed on previous days, or even weeks, months or years. Second, the ADI is most often arbitrarily calculated as one hundredth of NOAEL for the declared active ingredient. This factor 100 is supposed to take into account: (1) the difference between rats (on which the tests are carried out) and humans; and (2) the difference in human sensitivity (differences between frail people—such as children, pregnant women or elderly—and healthy adults). Therefore, ADI ignores the fact that commercial formulations are more toxic than the declared active ingredient by itself. For instance, Roundup formulations proved to be 10–1000 times more toxic than glyphosate alone [62]. Consistently, the ADI of glyphosate-based herbicides should thus be reduced by a factor of 10–1000. While these exposure limit values are based on LOAELs and NOAELs and published scientific studies, serious problems are arising such as the choice of model organisms, often among the most resistant ones, the choice of exposure concentrations, or biological or biochemical parameters not sensitive enough (for example, choice of mortality or loss of mass rather than the dosage of metabolites or enzyme activities, which are much more sensitive).

Let us consider the example of the aquatic plant Hydrilla verticillata exposed to copper [63]. If we collect from this article the 7-day LOAELs for different parameters (Table 1) we notice that the difference between the most and the less sensitive LOAEL reaches a 50-fold gap according to the observed parameter: enzymatic assays are the most sensitive as well as the ionic leak test (caused by damage to the cell envelope), while biomass loss is the least sensitive parameter. However, it is often the mortality, the loss of weight, or the biomass on which the reference doses are deduced.

Table 1 LOAEL determined after 7 days of exposure of Hydrilla verticillata plants to copper in aquatic medium

Another problem is that NOAELs and LOAELs actually depend on the choices made by the experimenters in terms of concentrations. In the last example [63], the copper concentrations chosen were 0.1, 1, 5, and 25 μM. The lowest LOAEL records were for three enzymatic activities and ionic leakage. However, no concentration lower than 0.1 was tested because of the selection of concentrations decided by the authors. It is reasonable to suspect that the true experimental LOAEL is less than this approximate LOAEL. The same holds true in the case of the common lettuce L. sativa exposed to the herbicide imazethapyr (Table 2).

Table 2 LOAEL determined on various organisms exposed to the herbicide imazethapyr

The choice of the model organism is of crucial importance. If we consider now the example of the herbicide imazethapyr [64, 65], the impact on growth of the common lettuce Lactuca sativa appeared 10-times weaker than that of the green algae Pseudokirchneriella subcapitata to this chemical (Table 2). And the cyanobacteria Anabaena flos-aquae is 4.5-fold more sensitive to this herbicide than the diatom Navicula pelliculosa (Table 3). When focusing on the genotoxic outcomes, the most sensitive methodology was the nucleus test performed on Allium cepa. This test was 10-times more sensitive than the detection of chromosomal aberrations on that species, and 105-fold more sensitive than the Ames test carried out on the bacterium Salmonella typhimurium (Table 2). The Ames test is still widely used to address the mutagenicity of chemicals despite its very low sensitivity.

Table 3 Impact on growth for various organisms exposed during 4 days to imazethapyr herbicide through two commercial formulations

In official documents, NOAELs and LOAELs given for different aquatic species are not comparable with each other. They are indeed determined at different durations of exposure, for different water qualities, and without any specifications about the biological parameters tested. However, in a number of examples, these parameters vary from one species to another. For instance, the toxicological and environmental data sheet for “cadmium and its derivatives” edited by the French national agency INERIS (National Institute of Industrial Environment and Risks) in 2011 [66] highlights such an inconsistency (page 37): for the freshwater snail Physa integra the NOAEL at 28 days is 8.3 µg CdCl2/L [67], and for the planktonic crustacean Daphnia magna the LOAEL at 21 days is 0.3 µg CdCl2/L [68]. Besides the fact that the exposure time is not the same, the absence of mention of the biological parameters taken into account in these data makes any comparison between these values impossible. In fact the related articles indicates that the considered parameter for D. magna is the biomass whereas it is the survival for P. integra. It is hardly understandable why the INERIS document only indicated the NOAEL for P. integra while the authors mentioned the LOAEL for the survival of P. integra after 28 days of exposure: 27.5 µg CdCl2/L [67]. Moreover, the water hardness is not indicated in the INERIS document although it is not the same for these two data (224 CaCO3 mg/L in the case of D. magna and 45 CaCO3 mg/L in the case of P. integra) and, even worse, this parameter is known to change the bioavailability of heavy metals. Thus, it makes no sense to compare toxicological data obtained in such different conditions.

As for heavy metals present in agricultural soils and in edible crops, a study highlighted two major drawbacks in the European regulation: a/metal concentration (total or bioavailable fraction) limits in agricultural soils are lacking; b/metal concentration thresholds (existing only for cadmium and lead in crops) reported in the EU 420/2011 Regulation (April 29th 2011), are expressed on the fresh weight basis rather than on the dry weight basis. This is not suitable due to the variety of water contents within edible parts of vegetables [69].

The reference doses for food or water intended for human consumption are often displaying substantial differences among the regulatory agencies. For instance, for cadmium through the oral route, the ATSDR (Agency for Toxic Substances and Disease Registry) limit is 0.2.10–3 mg/kg/day, while that of WHO is 7.10–3 mg/kg/day, and the US Environmental Protection Agency (EPA) reference doses (RfDs) are equal to 0.5.10–3 mg/kg/day in water and 1.10–3 mg/kg/day in food [66, page 31]. Most worrying are the inconsistencies in the maximum levels displayed by legal regulations enacted by the EU. For cadmium, the European directive EU 420/2011 indicates a limit of 0.05 mg/kg fresh weight for vegetables and fruits, meanwhile excluding without apparent logic leaf vegetables, fresh herbs, leafy brassica, fungi, stem vegetables, root and tuber vegetables, for which the authorized limit is four times higher, i.e. 0.2 mg/kg fresh weight. The same holds true with lead for which limits are shifting from 0.1 to 0.3 mg/kg fresh weight. Surprisingly, it is written in the EU 420/2011 that “the default maximum levels for lead and cadmium in fruit and vegetables are not realistic for seaweed, which can naturally contain higher levels. Seaweed should therefore be exempted from the default maximum levels for lead and cadmium in fruit and vegetables”. In other words, whenever a foodstuff contains too much toxic, no limits apply to it. Are the regulations made to protect the market more than human beings? For instance, the upper limit for mercury in fishery products and muscle meat of fish is 0.5 mg/kg wet weight (European directive EU 1881/2006). However, about thirty fish species are excluded from this rule and are allowed a limit of 1 mg/kg wet weight (EU 1881/2006, page 29). The list of these species includes the most consumed fish such as tuna, emperor and swordfish. Most often, their mercury concentration is greater than 0.5 mg/kg fresh weight, which means that they could not be marketed if the lower limit applied to all species. The most toxic form of mercury found in foods is the organic form methylmercury (MeHg). MeHg intake is linked to fishery products and crustaceans. The RfD for MeHg set by the US EPA is equal to 0.7 μg/kg body weight/week [70], whereas the JECFA (committee administered jointly by the Food and Agriculture Organization and WHO) established a PTWI value of 1.6 μg/kg body weight/week [71], as though the American people were 2.2-fold more sensitive to MeHg than other people in the world. For an adult of 60 kg, according to these two standards, the intake should not exceed 42 or 96 μg/week, respectively. EFSA chose the less protective intake limit, that of JECFA. Even considering this limit, a person of 60 kg eating 200 g of a tuna meat containing 0.6 mg/kg fresh weight would ingest 120 μg of MeHg, which is 1.25-times above the EFSA PTWI and 2.8-times above the US EPA RfD. Ironically, if the upper limit in concentration of 0.5 mg Hg/kg also applied for tuna, 200 g of tuna would deliver 100 μg of mercury, about 95 μg of MeHg (the proportion of MeHg in carnivorous fish is about 95% of the total Hg), a dose that is very close to the EFSA PTWI. The choice of the less protective PTWI and of the arbitrary higher Hg upper concentration limit in food can only be explained as an allegiance measure to protect the fish market. Clearly, if the US EPA RfD and the limit of 0.5 mg Hg/kg fresh weight applied, species such as swordfish could not be marketed anymore.

Conflicts of interest in risk assessment agencies and scientific publishing

The primary goal of risk assessment relevant to public health is to provide impartial evidence on health hazards for humans. However, this goal can be altered by conflicts of interest [72], especially when they occur within evaluation agencies.

It is widely documented that regulatory bodies such as EFSA and the German Federal Institute for Risk Assessment (BfR, especially in charge of glyphosate evaluation in the EU) do not behave like the industry-independent bodies they pretend to be, mostly because of the conflicts of interests of some of their panel members. Several journalists investigating the CV of members of the scientific panels of the EFSA revealed that nearly half (46%) of the experts of this agency are in a financial conflict of interest situation with the agrobusiness and food industry [73]. In the German newspaper TAZ [74], the BfR head Andreas Hensel had defended the fact that the pesticide committee of his agency includes Bayer and BASF employees, explaining that facts about products “can only be judged by those that work with them” [75].

But conflicts of interest also concern the sphere of scientific publishing, thus possibly leading to a bias in the availability of the scientific data on which the evaluation agencies are supposed to rely. An example is given by Richard Goodman, former Monsanto employee, who was editor of the Food and Chemical Toxicology journal for a short time. The US Freedom Of Information Act allowed the release of emails revealing that the unilateral removal of a paper challenging the safety of Genetically modified Organisms [76] was driven by Monsanto through Richard Goodman, and that the latter received a retribution via a funding to the Allergen Database, a program he oversees [77].

The most famous scientific lobbying organization (to which a number of scientists sitting in evaluation agencies are linked) is ILSI (International Life Science Institute), a Washington-based international organization founded and funded by most major international companies to defend their interests [78]. Moreover, some firms resort to a serious form of scientific fraud, called "ghostwriting". It consists in obtaining, in exchange for remuneration, the approval and signature of renowned scientists for articles or reports written by employees of the company. As revealed by the "Monsanto Papers", the firm Monsanto used this scam to claim the safety of its herbicide Roundup [79].

Conflicts of interest lead organizations to apply a double standard of evaluation depending on whether the studies are from academic researchers or industry (usually the petitioner), with greater confidence being given to the latter [80]. This is all the more damaging as it is well known that the source of funding has an influence on the outcome of a study [81,82,83,84]. An example is given with BPA safety evaluations where the US Food and Drug Administration (FDA) and EFSA have neglected a large number of independently replicated government-funded studies, and given special prominence to two industry-funded studies having serious conceptual and methodological flaws [85]. Similarly, many studies on which regulatory agencies rely for ED chemicals evaluation proved to be flawed [86]. Another example is that of glyphosate license renewal application, where IARC and EFSA reports led to conflicting conclusions: while the WHO organization relied on scientific studies published in international peer-reviewed journals, the European agency relied mainly on unpublished and confidential studies from industrials. In this extreme instance, plagiarism was even evidenced] [87]: in the Renewal Assessment/Approval Report (RAR) of glyphosate by BfR (2015), the chapters on carcinogenicity, mutagenicity and reproductive toxicity contain large parts of copy-paste from a report by Monsanto Europe S.A. on behalf of the Glyphosate Task Force (2012).

Evidence of corporate companies malfeasance and undisclosed conflicts of interest with respect to issues of scientific integrity, such as ghostwriting, interference in journal publication, and undue influence of regulatory bodies, undermine the objectivity of the risk assessment agencies’ conclusions and therefore the confidence that can be placed in them.

The precautionary principle is not applied.

The precautionary principle originated in Germany (Vorsorge Prinzip) in the 1970s: it was mentioned in the works of Hans Jonas where it appeared as a corollary of the "principle of responsibility". The precautionary principle is also not new on the legislative level since it appears in the first treaties of the European Union. It is a principle of Community law on which the European Court of Justice often relies. It should therefore serve as a basis for all European and national policies by systematically prioritizing, in case of doubt, human health and the environment, and not industrial interests. We are in a diametrically opposed situation at both national and community level. Indeed, it is clear that this principle currently applies in favor of products, so as not to hinder their dissemination as long as their toxicity is not proven.

The scope of the precautionary principle is in fact very limited. It only targets hypothetical risks arising from new technologies and chemical products. It is thus distinguished from the prevention which concerns risks perfectly identified and for which a probability of occurrence can be established. Contrary to what is claimed through certain media or political invocations, the precautionary principle has absolutely nothing to do with disasters such as industrial accidents or various weather events, as well as with perfectly defined risks of everyday life. Tobacco, speed, unprotected sex and alcohol consumption kill thousands of people every year. Protecting oneself from these realities has nothing to do with the precautionary principle: this is the application of the prevention principle.

This confusion is fueled, in order to discredit the precautionary principle, by all who consider it as an obstacle to neoliberalism and technological freedom. Nonetheless, this principle is a principle of innovation, budget saving and societal progress. A report by the European Environment Agency [88] describes the cost of mistakes made for lack of precaution and how innovation and science should be linked to the precautionary principle. Indeed, the application of this principle requires a very strong research in order to better understand the risks. But this principle is also a source of innovation insofar as the costs likely to be generated by the refusal to consider the potential negative consequences of new technologies are enormous for companies as well as for the whole civil society. It is therefore necessary to research and develop alternative products and technologies that fulfill the same function or even an improved function, while overcoming the negative consequences for human health or the environment.

France is in a particularly paradoxical situation, since it is probably one of the countries in the world where the precautionary principle is the most evoked but the least applied. Contaminated blood, asbestos, pesticides, growth hormones, Mediator®: France likely holds the European record for failures in precautionary principle, even when health impacts have been fully demonstrated.

In the end, the legislation is perfectly consistent with the precautionary principle, but it is not applied because it is bypassed. Inaction is preferred whereas “the precautionary principle ought to be acting and doubting” [89]. Lobbies often obtain other regulatory provisions that counteract the texts of protection. Thus, for GMOs, the excellent Directive 2001–2018 is thwarted by the Regulation 1829/2003, which is much less protective, and on which petitioners rely to obtain marketing authorizations.

Patent protection prevents data transparency.

Studies requested by the risk assessment agencies as part of the regulatory evaluation are conducted by the petitioners themselves, and the raw data from these studies remain confidential under the pretext of intellectual property. Regulatory bodies are therefore obliged not to disclose them, which makes contradictory and transparent expertise impossible. However, transparency is a fundamental ethical imperative of scientific research, especially in the framework of assessment studies, justifying massive official regulations and policies [90]. The lack of transparency contributes to “manufactured doubt”, which even impacts independent rigorous and peer-reviewed studies. These sources of doubt have important consequences, both from a health and environmental point of view and from a social cost point of view, as it has been reported in the case of EDs assessment [86].

The raw data of industry-funded studies have only been obtained in a few particular instances at the cost of extremely heavy administrative or legal procedures. This was the case in Germany about the health risk assessment of the genetically modified maize MON863, approved for food and feed in Europe in 2005. A 90-day rat-feeding study performed under the responsibility of Monsanto Company has been subjected to questions from regulatory reviewers in Europe, and the results remained controversial. An Appeal Court action in Münster allowed public access in June 2005 to all the raw data from this study. These data were independently re-analyzed as part of a university study, which led to conflicting conclusions with, especially, signs of hepatorenal toxicity [91]. The authors concluded that longer studies (2-year, i.e., whole life of rats) were required to conclude whether corn is safe or not. These studies were not carried out, once again, in disregard of the precautionary principle.


About the risk assessment agencies

Most expertise agencies are reduced to a scientific college in charge of a strictly technical evaluation on health and environmental risks. But it is essential that the expertise is not limited to these aspects but addresses above all the questions of social utility and potential alternatives. These subjects are outside the field of expertise of scientists alone and therefore require a pluralism of expert commissions that must include representatives of civil society. This is the reason why we propose that each evaluation agency be composed, in addition to a scientific committee, of an economic, ethical and social college (EESC), like the High Council of Biotechnologies (HCB) in France. As for HCB, EESC should include representatives from trade unions, companies from various sectors of activity, consumer protection associations, environmental NGOs, patient associations, etc. But unlike the current functioning of HCB, it is important that the two colleges (EESC and the scientific committee) intervene with equal contributions (in HCB, opinions are issued from the scientific committee only, the prerogatives of EESC being limited to recommendations).

The pluralism of expert commissions as well as the multidisciplinary representation of the scientific college will have to be guaranteed by an independent structure—a High Authority of Expertise (HAE)—supervising, either at European level or at national level, all the evaluation agencies, and ensuring the transparency, the methodology and the deontology of the expertise. The HAE will be composed of members of various origins and skills, and whose appointment will be made by the constitutional assemblies and the highest jurisdictional bodies. As an example, with regard to France, we propose that the HAE be composed of: (1) deputies from different parliamentary groups; (2) persons responsible for collective expertise missions of major national research organizations; (3) a representative of each agency involved in the assessment of health and environmental risks (chosen because of its experience in ethical issues in expertise missions); (4) personalities renowned for their research work in the field of scientific expertise; (5) advisers; (6) qualified personalities in labor law, environmental law and public health law; (7) representatives of associations concerned by the ethics of the expertise; (8) a representative of each major union. Members (1) to (4) will be appointed by the Parliamentary Office for the Assessment of Scientific and Technical Choices (OPECST); members (5) will be designated by the Court of Cassation (in France this Court has jurisdiction to review the law, to certify questions of law and to determine miscarriages of justice) and by the Council of State; and members (6) to (8) will be appointed by the Economic, Social and Environmental Council. The appointment of all members of the HAE will be ratified by decree taken by the Council of State after a public call for candidates and a survey to ensure that applicants have no conflict of interest. The diversity of the members’ profiles, as well as the multiplicity of the sources of appointment will guarantee the multi-representativeness and will thus minimize the risks of collective impartiality which inevitably results from a stereotyping of the members. We propose that the term of the HAE members be 4 years, and renewable only once.

The HAE will therefore have to develop a code of deontology that will be made mandatory. This code should include an obligation of competence, of independence vis-à-vis potential petitioners (absence of conflicts of interest) and of responsibility. Thus, the HAE will have to establish a statute of the experts including the definition of their formation and competence, as well as the conditions of their remuneration (which will have to be comparable for all the agencies).

Indeed, in order to avoid sources of indirect remuneration, and thus to avoid conflicts of interest, it is important that the expert be formally compensated for the supplied service and time spent on preparation and participation in committees. The funding of the experts and of the expertise as such will have to be ensured by a specific fund administered by the HAE and fed by the petitioners.

The High Authority will also be in charge of controlling the absence of conflicts of interest by making the declarations of interests mandatory each year, and for all the experts sitting on the committees. The High Authority must be provided with the means to carry out its own investigations in order to detect any false statements and omissions. Such frauds will have to be sanctioned by the removal of the expert, and in the most serious cases, by penal sanctions.

The High Authority could be seized by all the evaluation agencies, but also the European Commission, the States, the companies concerned by a given expertise and by the approved associations, and would have the discretion to request a second opinion and to audition external experts. These external experts could be chosen by the HAE itself or suggested by the applicant for the counter-expertise.

About the assessment itself

Need for a pluralistic expertise procedure open to civil society:

The question of social utility and alternatives must be discussed before the possible health and environmental assessment. Indeed, why to engage an expensive and time-consuming evaluation if the proposed new technology or new substance does not correspond to any societal expectation? The only advantage limited to a private company applying for a marketing authorization cannot in itself constitute a contribution to the public interest. While the company reaps the commercial benefits, it is up to the whole of society to bear the social cost (health and environmental as well as financial). For instance, the contribution of EDs to diseases and disorders of human male reproduction and of neurobehavioral deficits represents annual associated costs in the EU estimated to 15 billion and 150 billions euros, respectively [92, 93].

As mentioned in the previous Sect. (3.1.), these issues are the responsibility of the whole civil society. It is of course the role of EESC which we propose the setting up in all the evaluation agencies. But we propose to supplement the opinion of EESC by that of a panel of citizens. It is often assumed that innovation cannot be assessed by the population, which is overwhelmingly incompetent. This is not false, but it neglects the widely verified possibility of enabling "ordinary" citizens to acquire a good knowledge of a difficult subject. The interest in using citizen judgment is multiple. First, it is the most relevant perspective for overcoming pressures from particular interests. Second, it is also a unique opportunity to take into account arguments that escape scientific expertise but include the fundamental relationship between humans and their fellow human beings and nature. Third, it is above all the surest way to respond to the legitimate democratic aspiration that what is good for the population can only be defined by itself. Finally, it is a way to reveal the extraordinary properties of Homo sapiens when his empowerment produces both collective intelligence and empathy towards others. In other words, citizen judgment is able to identify the solutions that are most in line with the common good, both in terms of research [94] and the evaluation of its results. However, these effects can only be achieved under specific conditions, since the will to understand and to appreciate positions of others is not shared on a daily basis by humans. It is through truly participatory procedures, such as the various forms of citizen juries, experienced by hundreds over the past half-century, that "ordinary" people can demonstrate the best of humanity. Observers of hundreds of citizens' conferences held around the world over the past 30 years are unanimous: in this procedure, there is a profound transformation of any individuals into responsible and ingenious citizens: whatever their age, their origins and levels of education, they discover that they are capable of mastering a complex subject and propose relevant solutions that are not dictated by their personal interests. But to achieve this result and to gain the credibility to transform citizens' opinions into political decisions, the procedure needs to be streamlined. It is to this end that the Citizens' Convention [95] was drawn up, inspired by the Citizens' Conferences invented in Denmark 30 years ago. With these citizens' conventions, it is possible to obtain the opinion of people without special interests (drawn by lot), sheltered from lobbies (anonymous until their opinion is given), remarkably informed (based on contradictory expertise), voluntary and without sustainable status (replaced for each expertise), and representing socio-economic diversity, all conditions that thwart the current perversions of the so-called "participatory democracy". Currently, policy makers decide on the dissemination of innovative products and technologies based on scientific expertise and industrial claims. The wise and democratic solution would be to interpose the citizen between the expert and the politician.

Modification of regulatory toxicology:

Regulatory toxicology must be adapted to the characteristics of new pollutants, their ubiquity, the multiplicity of their targets, and the fact that we are generally exposed to mixtures.

Relevance of controls:

The control diets of laboratory animals should be monitored and analyzed for possible toxic chemicals, especially those under scrutiny in a given study. For instance, it has been demonstrated that almost all diets for mice and rats contained MeHg concentrations at such important levels that the alleged control animals could not be seriously considered as control [96]. Similarly, laboratory rodent diets from 5 continents have been shown to contain toxic levels of various pollutants (pesticides, dioxines, PCBs, heavy metals) [97]. Such contamination of the “control” rat diets increases the background noise of health disorders and thus reduces the statistical significance of the observed effects in treated rats. This situation is therefore likely to question the validity of all the toxicological tests carried out on rodents. It should be checked that the strain of control animals used are really naïve in terms of sensitivity to the tested chemical. For instance, it has been demonstrated that all animals in a corporation-funded study appeared to have been exposed to an estrogenic contaminant, with the consequence that males in all groups, negative controls and BPA-exposed, had significantly enlarged prostates and lower daily sperm counts relative to negative control values from the same mouse stock examined at the University of Missouri [98]. Most probably, the “control” diet was contaminated with an estrogen-like chemical. In other words, control animals should be good and real control (i.e. healthy and fed with diets devoid of toxic chemicals). The most sensitive model species should be chosen by toxicologists, along with the most sensitive biological, biochemical or genetic parameters to define LOAELs.

Relevance of toxicological parameters:

The most protective (for human beings and all other living beings) upper limits or PTWI or tolerable intakes should be selected. Vegetables or meats from different animals should not be excluded from a regulation, nor should the upper tolerable limits in foodstuffs be increased to allow their continued marketing. The protection of human and living beings should be a priority and push the markets and industrialists to comply with the regulations and not the reverse. Especially for vegetables and leafy foodstuffs that contain variable amount of water according to the season and the place of production, the concentration limits should be given on a dry weight basis instead of the fresh or wet weight. In official documents delivered by sanitary and environmental institutions, the parameters linked to a LOAEL or NOAEL should always be indicated, along with important information such as the hardness of the aquatic medium or the carbonate content of soils, since when dealing with heavy metals, the bioavailability of these toxics can vary in a huge extent.

Bioaccumulation and combined effects of chemical mixtures:

Human populations are never exposed to single compound but to mixtures of chemicals, including some chemicals which are persistent in the environment even if they have been banned for decades. Although it is important to test chemicals one by one to determine their toxicity profiles, the evaluation of health effects should take into account effects of mixtures. This can be done by using whole-mixture approaches for cocktails of chemicals for which interactions and composition are unknown. This concerns both commercial formulations such as pesticides (see next section) and combinations of characterized chemicals, as it has been shown with simultaneous exposure to several EDs [99], several pesticides [100, 101], or various emerging pollutants [102]. This is not only valid for chemicals but also for whole-food testing such as GM crops. The potential metabolic effects of transgene insertion into a crop cannot be modeled by measuring the effects of the modified encoded-protein only. The whole GM crop should be tested by inclusion in standard rat diets. Therefore, the principle of “substantial equivalence” (i.e. close nutritional and compositional similarity between two crop-derived foods) on which is based the assessment of genetically modified crops to claim that they are as safe and nutritious as currently consumed conventional plant-derived foods [103], is not satisfactory. Indeed, secondary metabolic effects of the genetic modification, as well as the presence of pesticide (herbicide residues or/and insecticides) are ignored [104], although metabolic disorders due to pesticide residues have been observed under conditions of apparent herbicide tolerance [105,106,107]. For some chemicals with similar modes of action, additive models have been shown to be a valid approach in most cases, although it is not possible to consider that the toxicity of all chemical mixtures will follow additive models. In case of uncertainties, it is also possible to take a precautionary approach by adding safety factors in the calculation of ADI.

Evaluation of pesticides in their commercial formulation:

Pesticides are always commercialized as mixtures of chemicals. Only one of these ingredients is regulated: the active principle declared by the manufacturer. Other ingredients are considered as inert diluents although they can be more toxic than the declared active principles on non-target species [60]. When a pesticide is tested to predict its toxic effects on human health, only the toxicity of the declared active principle is measured to establish an acceptable daily intake. This is a crucial gap. All ingredients entering in the composition of a pesticide formulation should be regulated as active principles, so that their unexpected side effects would be considered. These effects encompass possible additional toxicological outcomes or the increase of the bioavailability of the alleged principal components (such in the case of the Roundup companion chemicals and glyphosate). The formulations can really introduce huge differential effects. For instance, ethoxylated adjuvants of glyphosate-based herbicides (GBH) have been shown to be active principles of human cell toxicity [62]. Also, two GBH co-formulants, polyethoxylated tallow amine (POEA) and alkylpolyglucoside (APG), have been shown to be endocrine disrupters [61]. Another example is given with the herbicide imazethapyr [65]: it is 22-fold more toxic to the algae P. subcapitata when present in the formulation 2ASU as compared to Verosil whereas the mass proportion of this chemical in Verosil is twice that in 2ASU (Table 3).

Long term (whole life) evaluation:

RT guidelines for the testing of chemical toxicity rely on the assumption that human health effects from chemical exposure can be predicted by extrapolating long-term effects from short-term exposures tested in laboratory animals. This is a valid approach to predict the toxicity of chemicals causing organ damages with linear dose–response relationships. However, this cannot be used to predict chronic effects from long-term exposures to chemicals acting as metabolic, nervous or endocrine disruptors since these effects are not always proportional to the tested doses [52,53,54, 108]. Therefore, the absence of life-long exposure tests is another major gap in the current regulatory toxicity framework. Chemicals are either tested during animal gestation, the exposure being stopped when offspring become adult, either tested when animals are adult so they are not exposed during early-life. It is now clear that chemicals can have life-long effects and influence the development of chronic diseases when the animals are exposed prenatally. While this developmental origin of health and diseases (DOHaD) paradigm is emerging [109,110,111,112], it is not addressed by current toxicity tests. A solution would be to implement new rodent toxicity tests in which the exposure would start prenatally, and the animals monitored for their whole life.

Trans-generational effects evaluation:

According to the current RT guidelines, multigenerational effects investigations are limited to studying the reproductive function after two generations in rats (when the tested chemical is supposed to be reprotoxic). Recent studies have shown that a gestational exposure to some chemical toxicants, especially endocrine disruptors, can have effects persisting until the third generation and beyond, despite the absence of exposure of these descendants to these pollutants [57, 58]. The trans-generational toxicity of these chemicals has been shown to occur via epigenetic mechanisms, although a classic hereditary transmission within bacterial communities inhabiting the human body has also been hypothesized to explain this phenomenon [113]. It is therefore essential that studies of trans-generational effects resulting from a gestational exposure be systematically conducted, over at least three generations.

Endocrine disrupting effects evaluation:

The regulation of endocrine disrupting effects of environmental pollutant has received a considerable attention in the last decade. A major step has been done in the European Union by implementing regulatory tests addressing endocrine disrupting effects on estrogen, androgen and thyroid function, as well as on steroidogenesis. Although this is a major advance, the endocrine system is not limited to the previously mentioned biological systems. Disturbances in the action of neuro-hormones (serotonin [114], dopamine [115] or endocannabinoids [116]), or hormones controlling energetic metabolism (e.g. insulin or leptin [117]), have also been shown to give rise to health effects and are not covered by regulatory toxicity tests. Moreover embryonic and fetal development being the period of greatest susceptibility for EDs [55, 56], it is essential that studies be conducted on pregnant rats, with several generations of offspring analysis (see previous section: Trans-generational effects evaluation.) Finally, the effect-dose response being generally not monotonic for EDs, the number of tested doses must be greater than two. Also, to identify, organize and utilize mechanistic data when evaluating EDs, we suggest to integrate the 10 key characteristics as described by La Merrill et al. [118].

Evaluation of the impacts on gut microbiome:

Our bodies are inhabited by communities of micro-organisms (bacteria, viruses, fungi, protozoa). Alterations in the composition and function of these ‘microbiomes’, the largest of which being the gut microbiome, can have major health impacts. The gut microbiome capability to transform environmental chemicals such as drugs or industrial chemicals has been largely underestimated [119]. This is the case for pesticides. For instance, Roundup has been shown to have a sex-dependent impact on gut microbiome [120]. Although declared active ingredients of pesticides are generally developed to act on metabolic pathways present in the target pest but not in human cells, some bacteria from the gut microbiome may have these pathways and be affected by these pesticides, such as the shikimate pathway inhibited by glyphosate [121] or the branched amino acids pathway inhibited by imidazolinone [122]. Some studies have also recently demonstrated that the lack of control of the gut microbiome composition in rodent colonies can make studies irreproducible [123]. We thus propose to include, in the battery of tests used to test the toxicity of pesticides, metagenomes analysis in order to appraise the possible impacts on the intestinal micro-organisms communities.

Abrogation of industrial protection for health and environmental issues and prevalence of the precautionary principle:

The application of the precautionary principle is, of course, not compatible with data retention relative to health and environmental issues. The role of public authorities in keeping studies confidential, alleging industrial secrecy, must be prohibited. Not only these studies and their rigor are subject to doubt, but also they cannot be reproduced and debated. This is contrary to the basic scientific principle of reproducibility of experiments and discussion of the results obtained. We therefore propose a restriction of the industrial secrecy based on Article 25 of Directive 2001/18 on GMOs, by creating an obligation to report all known impacts on health and the environment as well as raw data from previous studies conducted to better identify them. Such a provision can only be effective if it is accompanied by another one recognizing, on the part of the public authorities, an obligation of information in environmental and health matters, and therefore a pledge to divulge all the documents relating to these data.

The corollary is that evaluation agencies must rely on a maximum of studies conducted by independent academic laboratories and published in international peer-reviewed journals. A paradigmatic case of a biased opinion influenced by studies emanating from the industrial sector is that of the recognition of BPA (2,2-bis{4-hydroxyphenyl}propane) as an endocrine disruptor. In 2010 the WHO expert panel recommended no new regulations limiting or banning the use of bisphenol A (BPA), stating that "initiation of public health measures would be premature". In 2011 the EU (EU 10/2011 regulation) and in 2012 the US Food and Drug Administration did ban the use of BPA in baby bottles, however the lobbying organization Environmental Working Group called the ban "purely cosmetic". The WHO and EFSA RfD is 50 μg/kg/day BPA [124, 125]. However an estrogenic mode of action of BPA is confirmed by in vitro experiments, which describe disruption of cell function at 10–12 M or 0.23 ng/L (0.23 ppt) [126]. In human fetal leydig cells cultured in vitro, a significant decrease of testosterone secretion was recorded as soon after one single day of exposure to 10–8 M BPA or 2.3 µg/L [127]. Yet, chemical manufacturers continue to ignore these published findings because no industry-funded studies have reported significant effects of low doses of BPA, although over 90% of government-funded studies have reported significant effects. In a thorough scrutiny of the literature [98], it was shown that a total of 130 articles had been published on the BPA subject up to 2006, 119 from governments-funded teams and 11 from corporations-funded ones. A total of 109 articles (83%) reported harmful effects whereas 21 (16%) could not unravel effects. All of these 109 articles were coming from governments-funded teams, and none of the 11 articles funded by chemical corporations could pinpoint any effects. However, in expert committees and panels, there are always voices to say that the scientific data are not matching and agreeing with an absolute certainty and that in such circumstances, one cannot establish strong and protective rules or ban BPA.

The measures we advocate will therefore prevent the doubt to benefit to the commercialization of the product rather than to the precautionary principle.

General comments

Preventing risks means first determining their scope and content. It is then to set the rules of the evaluation so that the economic interest of the stakeholders does not come at the expense of safety and prevention, which implies responsibility and transparency. This is why information, expertise and responsibility are a whole, and these notions need to be linked. This link is all the more obvious because the actual costs, linked to poorly evaluated technological choices, are becoming heavier, in addition to human costs. For example, the Netherlands have estimated the cost of delaying the ban on asbestos to 19 billion euros and 34,000 premature deaths, not counting the health costs and compensation of victims [128]. An assessment based on a clear distribution of responsibilities, transparent information, pluralistic and rigorous expertise which is subjected to debate, guarantees a better quality of the choices, a process more democratic and thus a relation of confidence which no longer exists and which, nevertheless, participates in dynamism and the good health of a society. But the debate is obviously possible only if it is honest. It means that it must be contradictory and transparent, and that the public bodies responsible for giving an opinion be relieved of their conflict of interest. This is all the more important since the opinion issued is generally followed by the decision-makers.

To achieve such objectives, it is essential that the public evaluation bodies have strong and rigorous ethical control and that they have sufficient funds to benefit from internal experts and to conduct, independently, the additional studies required. This is why we propose the establishment of a high authority of expertise that will be in charge of supervising these missions.

The set of propositions developed in this manuscript constitute a coherent whole and are based on one another. They are all made in the same state of mind, that of transparency, democracy, and deontology, and for the same purpose: to put an end to health and environmental disasters, and stop the erosion of biodiversity and the explosion of chronic diseases.