Juvenile Toxicity: Are We Asking the Right Questions?

Abstract

The recent increase in numbers of juvenile toxicity studies over the past few years has doubtless added to our knowledge of the development of organ systems in the young rodent, but it is perhaps rather more difficult to answer the questions as to whether this body of work has served us well in terms of assessing risk in the pediatric population. If there are shortcomings, are they the result of poor study design or poor sensitivity of the model, or are the wrong questions being posed? This article provides a superficial overview of the progress made thus far and considers which aspects of study design and model choice currently fail to adequately address the major issues surrounding pediatric toxicology.

Introduction

That the finished adult is a product of previous experience has perhaps never really been a source of contention. But it is only comparatively recently that we have perhaps begun to fully appreciate the importance of low birth weight and other early childhood experiences in terms of their association with increased rates of coronary heart disease and the related disorders stroke, hypertension, and type 2 diabetes, to name but a few. These changes are thought to occur as a result of the body’s plasticity during development; it appears that structure and function of many organ systems may be changed by the intrauterine and, more importantly from the perspective of pediatric drug development, by the early postnatal environment (Barker 1992a, 1992b, 2006; Hertzmann et al. 2001; Thomas et al. 2008).

As a broad generalization, our current paradigm of juvenile studies is to designs studies that seek to investigate toxicity in the rat between the ages of ten days and early adulthood. This design seeks to examine only toxicity that occurs during the period of dosing, which is generally until a point shortly after sexual maturity. These types of study do not address the issue of the possibility of later onset changes that may be predicated by early life exposure. The majority of study designs also fail to recognize issues of early exposure and the very short window of development that the rat presents as a model. Before looking at possible ways to address some of these problems, it may be useful to examine the issues that arise in specific organ systems and consider some of the likely ramifications in man. This is by no means intended as a comprehensive review of the shortcomings of juvenile rodent models, but it hopefully takes into consideration some of the most important areas of concern.

Current Guidance

Current study designs are based mainly on the United States Food and Drug Administration Guidance for Industry: Nonclinical Safety Evaluation of Pediatric Drug Products and the European Medicines Agency’s Guideline on the Need for Non-clinical Testing in Juvenile Animals of Pharmaceuticals for Pediatric Indications. These guidelines recognize that many of the medicinal products currently in use for pediatric applications were not initially designed for this age group. As we know from embryofetal studies, organ systems are especially prone to toxicity during the period of organogenesis, and therefore organ systems such as the brain, pulmonary system, immune system, kidneys, skeletal system, gastrointestinal (GI) system, and reproductive system, which are developing in pediatric populations, may be prone to toxicity not evident in studies in adults. Juvenile studies are generally considered to be necessary when existing animal and human clinical data are insufficient to evaluate the likely safety in the target population and ideally should focus on effects not previously studied in existing clinical or preclinical evaluations. Although decisions on studies in juvenile animals are made on a case-by-case basis, studies with therapeutic agents that are known to target the organ systems noted above are considered to be particularly relevant.

Considerations of target organ toxicity will suggest appropriate dosing periods. For instance, if effects are expected on systems with a long development period, then dosing until the animal reaches adulthood will be expected, but if organ systems with shorter development time are expected targets, then shorter dosing periods will be more appropriate. Considerations of route of administration and species selection are on the same basis as for adult studies, although one species is normally considered sufficient.

Investigating Changes in the Juvenile Lung and Effects That Carry into Adulthood

Several authors have reported that reduced birth weights and lower respiratory tract infections in man in the first two years of life have been shown to be associated with impaired lung function in later life (Hoo et al. 2004, Fillipone and Barraldi 2011). Although chronic obstructive pulmonary disease (COPD) is an adult-onset lung disease most often seen in aging people with a tobacco-smoking history, only around 20% of cigarette smokers develop full-blown emphysema, clearly suggesting that there is a genetic susceptibility. The mechanisms by which these genomic alterations contribute to COPD pathogenesis are as yet barely understood, but it would appear to be clear that genomic alterations may affect lung developmental processes in early life, changes that may create the environment that is permissive for the adult onset of COPD. It has already been demonstrated in mice that mutations affecting lung alveogenesis during early postnatal life can lead to subsequent alveolar enlargement (Chen at al. 2005). A lung with such developmental abnormalities is likely to be disposed to further disruption of lung homeostasis, such as that caused by the imbalanced protease/antiprotease activities seen in COPD.

Once it is accepted that early, almost imperceptible changes in the lung may have a substantial impact on the adult, the importance of what happens to the young rat lung in the early days of life in terms of risk assessment becomes clear. One of the substantive issues surrounding the testing of lung toxicity relates to the exposure of the lung to new chemical entities (NCEs). The main surge of alveolarization from the immature saccular lung seen in the neonatal rat into the mature lung occurs between days 4 and 14; thereafter, further alveolar wall thinning occurs between days 14 and 21. Because rodent epiphyses never close, thoracic growth and alveolarization is continuous, and although modeling and remodeling of the lungs occur to an extent throughout life, it is never more rapid than during the fetal and early neonatal periods. In man, although the saccular stage is complete at birth, there is a critical period of alveolarization, and morphogenesis is from birth to two years of age, and development is complete by six to eight years of age. In terms of modeling this critical period of development in man, we are looking at a very small window of exposure in the rat, that is, from day 7, at the earliest, to around day 21. Given our standard rat protocol, this gives us a very short window of exposure. This period of exposure may be further compromised by the development of the GI tract. In the young rat, the GI tract only really becomes active just prior to weaning, which occurs on or about day 21. It seems highly probable that many small molecules will be barely absorbed prior to day 17, but as the major portion of alveolarization of the lung is complete by day 14, the actual period of exposure in the critical period of development of the lung is likely to be short at best, or in the worst case scenario, nonexistent.

There is of course one aspect at least where the rat may be of great utility. As the status of lung development in the newborn rat corresponds to the latter stages of the fetal lung in utero, it could potentially be a model for the human lung in the premature baby. As the lung in the premature baby is an area of concern that is difficult to model in other animal models, this model is worthy of consideration—we just need to work out how to intravenously dose the newborn rat.

Until these issues of exposure are addressed, it seems likely that our current study designs will not be suitable for the investigation of the consequences of early pulmonary toxicity, an area of toxicity that we know has consequences for man (Hoo et al. 2004).

The Liver: The Effect of Imprinting in Later Life

Given our current state of knowledge, we can probably surmise with a degree of confidence that acquisition of drug biotransformation capacity will be a dynamic process in early postnatal life.

Possibly the best known type of toxicity associated with reduced drug transformation is the grey baby syndrome associated with administration of chloramphenicol to infants. The UDP-glucuronyl transferase enzyme system of infants, especially premature infants, is immature, and therefore metabolizing the drug is far slower than in the adult or older child (Brunton 2006). As a result of the lack of conjugation, renal excretion is much reduced; as a consequence of these factors, the half-life increases roughly fourfold. This type of toxicity may be difficult to predict given the variation in ontogeny of enzyme development between species. However, there are aspects of liver enzyme development that are of even greater concern.

With regard to p450s, which is probably where we have the greatest insight, it is not only the drug-induced developmental switches during periods of administration that are of concern, but the longer-term consequences of early life exposure. It is not clear just how long this “longer-term” basis may be. However, there are signs that the consequences of drug treatment initiated early in childhood may not be apparent until many years later. One risk that would appear to be well established in man is the one associated with neonatal exposure to phenobarbital. In utero or perinatal exposure to phenobarbital will cause CYP isoform imprinting, even at therapeutic levels. This early imprinting seems to program the liver to overexpress constitutive CYPs over a lifetime, resulting in enhanced tumorigenesis and reduced life expectancy (Agrawal and Shapiro 2005; Gold et al. 1978). The risk assessment of this type of change is made more difficult because of the differences in the ontogeny of drug biotransformation between rodents and humans, but at the most basic level, an increase in liver weight in a juvenile study should act as an indication of a potential hazard.

As the experience with phenobarbital has shown, epigenetic flags that are set in early life can apparently alter the function of the liver for a lifetime, and there may be subtle changes unaccompanied by obvious changes that have a similar long-term effect. The potential for such changes is a worry and is clearly worthy of our attention.

The Thyroid Gland and Central Nervous System Interactions

Until comparatively recently, endocrine disruption was all about estrogenic, anti-estrogenic and anti-androgenic effects, but the potentially adverse effects that may be the result of disruption of the signaling of thyroid hormones have become an increasing cause for concern. The thyroid hormones play a major role in the development of the young in most mammalian species, so much so that it could perhaps be argued that the effects of thyroid hormones in the juvenile are the single most critical factor for normal development. In man and rat, the development of many organ systems is heavily influenced by circulating levels of thyroid hormones, but most notable are long bone development, central nervous system (CNS) maturation, and reproductive system development (onset of puberty). Of these changes, long bone development is a relatively straightforward change to measure in a rodent study, but potential changes in CNS and reproductive maturation present a considerable challenge. The difficulty of investigating thyroid-brain interactions in particular is a serious concern, especially when one considers the consequences of disruption of this axis.

Much of the available data pertaining to this type of toxicity are related to environmental chemicals. The most commonly indicted chemicals are the polychlorinated biphenyls, perchlorates, and polybrominated diphenyl ethers. Although it is difficult to demonstrate a causal link between any of these chemical and developmental effects in man, in vitro and in vivo studies have clearly identified the potential to disrupt the thyroid axis.

It is not as though the scientific community is unaware of the importance of the issues involved in disruption of the interactions of the thyroid hormones with the developing CNS. The Organisation for Economic Co-operation and Development Environment, Health and Safety Publications series on Testing and Assessment No. 57 was published in 2006. This publication is a detailed review paper on thyroid hormone disruption assays that runs to 434 pages and includes many different assays.

As the large number of assays would suggest, there is no single test that satisfies all the criteria. If we recognize that these shortcomings represent a substantial and probably unsolvable gap in our knowledge, then it is perhaps inevitable that we need to consider a more substantive examination of the brain itself. There has been plenty of recent discussion in the pages of Toxicologic Pathology with respect to the examination of the adult brain, and there would seem to be a similar groundswell of opinion within regulatory and pathological circles to extend the examination of the juvenile brain.

Although there are substantial differences in the development of the brain between rat and man, the status of the brain in the rat at ten days coincides quite nicely with the development of the brain at birth in man, a commonly selected time for commencement of juvenile toxicity studies in rats. Although most overt effects of toxicity in the brain are readily recognized, assuming the relevant area of the brain is examined, what is more difficult to evaluate is a generalized size reduction resulting from diffuse neuronal loss. Pathologists who are involved with the development of agrochemicals will doubtless be familiar with the brain morphometry technique, which involves linear measurements of predetermined areas and has become the chosen methodology for the investigation of generalized loss of neuropils in developmental neurotoxicity studies. However, there are serious doubts as to the sensitivity of such methods, and based on the observations of experienced pathologists that have performed a large number of these studies, it seems unlikely that the study design can actually detect the type of change for which it has been designed. There is every indication that the inherent variability associated with the presentation of the sections may exceed the subtle changes being sought. As well as being difficult to conduct with extreme accuracy, these methodologies are very labor intensive. However, one could perhaps make the case that what we should really be concerned with is whether CNS functionality is adversely affected by pediatric exposure.

For some time now, we have relied quite heavily on the functional observation battery. Although the methodology for these tests is now well established, there have to be some serious doubts as to their sensitivity. As an indication of this lack of sensitivity, it has been shown that functional observations in the rat are not obviously affected by thyroidectomy at day 25 of age (Eayrs 1961).

Results in other species are not obviously better. It has always seemed probable that the mini-pig, which displays a greater pharmacological similarity to humans than rodents, might provide a better animal model for assessing the efficacy and adverse effects of psychoactive agents and could lead to the identification of side effects not seen in rodent models (Van der Staay et al. 2009). A colleague of mine has investigated behavioral testing on the juvenile mini-pig using two tests, an adjusted hole-board test to assess cognitive performance and a ten-minute open field test to assess behavioral responses (Manton 2011). Three test items were selected to impair normal responses to assess the selectivity and sensitivity of each assessment. D-amphetamine sulfate was expected to elicit increases in activity and sensory responses, haloperidol was expected to elicit decreases in activity, and scopolamine hydrobromide was expected to elicit decreased learning and memory retention. This ten-minute open field test proved quite successful in detecting behavioral changes in juvenile mini-pigs treated with haloperidol or d-amphetamine, but the results of testing for learning and memory were equivocal and ultimately disappointing. It may of course be the case that the pig is relatively insensitive to the effects of scopolamine, or perhaps the test protocol itself requires some refinement. More probably, we are being overly optimistic in our expectation of what this type of testing can deliver. Although we may consider effects on memory and learning disability to be major manifestations of developmental neurotoxicity in man, and therefore obvious indicators of toxicity, it may be that they are in fact quite subtle effects in the greater scheme of things. If that is the case, then we may have to accept that current testing regimens simply do not have the sensitivity to investigate this type of change.

To place this type of change into context, let us consider the consequences of a five-point loss in IQ. At the level of the individual this may not affect the ability of that person to live a productive life, but if such a loss is experienced by an entire population, the implications for that society would potentially be profound. As an indication, a five-point decrease in average IQ in a population increases the number of individuals regarded as mentally retarded by 57% (Davidson et al. 2006). Putting this case into numbers to which we can perhaps relate, in a population such as that of the United States, say 250 million people, you would expect 6 million people with IQs above 130 and 6 million below 70. A decrease in average IQ of five points would cause the number of people scoring above 130 to decline by 3.6 million, and the number below 70 would increase by 3.4 million.

Given the serious implications of this change and the likely difficulty of assessing toxicity of this nature, it does rather suggest that the risk assessment of all chemicals that cross the blood–brain barrier or affect the thyroid hormones in the young need to be considered as potential hazards for which routine risk assessment may be very difficult.

As with the lung, there is also the question of GI tract development in juvenile rats, and hence systemic exposure. Although the issue is not as acute as for the lung, the period of development between birth and day 14 is the period of most rapid CNS development in the rat, and so the potential of lack of exposure is probably an issue for the CNS as well as the lungs.

The Reproductive System and the Onset of Puberty

As early as 1970, the connection had been made between endocrine disruption and transgenerational effects of diethylstilboestrol (DES) in adolescent girls (Williams and Schweitzer 1973). Although DES is best known for this effect on the female’s reproductive tract, it has also been shown to reduce the number of Sertoli cells when administered to neonatal rodents (Sharpe et al. 2003). In the rat, chemicals that have been classified as endocrine disruptors are generally at their most toxic around gestational day (GD) 14 to 19, but there is clear evidence that the male reproductive tract is sensitive to the effects of such chemicals for the entire period of development (Fisher 2004). As the effects of endocrine disruptors have been shown to be additive, testing new therapeutic agents in isolation may well lead to a false sense of security (Rajapakse et al. 2002).

More recently it has been observed that endocrine disruptors such as vinclozolin and methoxyclor, when administered at the critical time of gonadal sex determination, have the ability to alter the DNA methylation pattern in male germ cells. The result is transgenerational effects in the reproductive systems of the F1 generation (Anway et al. 2005).

One of the more subtle effects that may result from endocrine disruption is alterations to the time of onset of puberty. Onset of puberty is a milestone we can test reasonably easily in the rat. The vaginal opening of the female and preputial separation in the male can both provide reliable indicators of onset of puberty. With respect to vaginal opening, the experience at our laboratory is that the average time to vaginal opening is 34.3 days (± 2.69), with a range of opening from thirty-one days to forty days. A group size of ten in a study is sufficiently sensitive to detect a difference of plus or minus four days in vaginal opening. If you have group sizes of twenty, sensitivity improves and you can reasonably expect to detect a three-day difference.

Although we can measure these changes with a degree of accuracy, the sensitivity and significance to man make any risk assessment difficult to determine. One may question whether a change of this nature is necessarily a cause for concern. Indeed, in clinical practice, changes to the onset of puberty are more often considered signs of underlying pathology, but the effects of changes to the onset of puberty in itself have consequences.

The early onset of puberty has been linked to a number of emotional and behavioral problems, particularly in girls. These problems include higher rates of delinquency or conduct disorder symptoms and more frequent affiliation with older, more deviant peers. Unresolved delay of puberty may result in short spinal length and psychological problems, which if severe may be very distressing and result in deviant behavior. The deviant behavior may be so severe as to result in shoplifting and vandalism, or even suicide (Stanhope et al. 1992). Given this catalog of potential sociological ills as possible effects, it seems that this is an issue worthy of more than passing interest.

Routes of Administration

As with protocols for adult animals the regulations specify that wherever possible the expected route of clinical administration should be used. For the majority of NCEs targeted toward the older child and adolescent, the oral route is the route of choice and therefore will be the route of choice in the preclinical model. It may be a particularly obvious statement, but the young rat is a very different beast to the young human. Both man and rat are altricial animals, that is, much of the functional development takes place during the immediate postnatal period. Notwithstanding this similarity, there are fairly substantial differences in the detail of development.

In man, many intestinal enzymes are present at meaningful levels at birth, some of which are close to adult levels. For instance, in the newborn human infant, hydrochloric acid and pepsinogen secretion is around 50% of adult levels, with a pH around 4.5. By comparison, the pH of the stomach is around neutral at birth in rats, with minimal pepsinogen and acid secretion. Thereafter, in the rat, pH drops from neutral at birth to around pH 6 from days 5 to 10, then to pH 4 by day 15, and toward adult pH at weaning. These differences in GI tract maturation after birth contribute greatly toward making the rat a poor model for the infant. Ideally, to model the human infant, we would look to start dosing from day 10 or even a little before that. Certainly for small molecules dosed orally, rat toxicokinetic data obtained prior to day 15 are unlikely to be a good indicator of likely clinical bioavailability in the infant, and systemic exposure at this critical stage of development may be limited or even absent under some circumstances. In circumstances where the human infant is not the intended target, study designs with aggressive dosing from an early stage are probably not likely to generate meaningful data.

Although the oral route may be the most common route for older pediatric patients, this is not the case for the neonate or infant, for whom IV dosing or infusion becomes more important. Currently, the rat from birth to around day 14 is the only model we have for the preterm-to-neonate phase of development. Birth weights for rat pups are generally in the 6 to 7 g range, rising to around 14 to 16.5 g by day 7. Clearly the technical challenges that IV dosing or continuous infusion present are immense and not obviously surmountable. Although subcutaneous dosing is practicable from an early age and miniature osmotic mini-pumps can be used for drug delivery in newborn rats, we do not currently appear to have a very close model for the preterm/neonate in the most likely clinical scenario (Doucette at al. 2000).

Future Directions for Juvenile Testing

Modeling the GI Tract

By and large, the GI tract problems are purely and simply a question of model selection. Although the rat (and mouse), have been fine servants to toxicology and pathology over the years, they are sadly lacking with respect to being a model for the GI tract in the juvenile, more particularly the infant. As newborn and premature infants are increasingly in receipt of therapies, we need to consider an alternate. Our possible models are presumably the rabbit, dog, primate, and mini-pig. Husbandry issues make the rabbit totally unsuitable as a juvenile model, and the dog as a carnivore has a clearly different developmental regimen in the GI tract. The primate, as always, is the best model of all, but the economic and ethical issues are of substantial proportions. The mini-pig is an omnivore and has a GI tract development, which although very accelerated compared to man, is broadly comparable. The size of the newborn mini-pig and the maternal tolerance to disturbance of the litter means that mini-pigs can be dosed from birth. As such, this may well represent the best compromise, and it is worthy of consideration on this basis alone as an alternative model for juvenile toxicity studies, particularly those in which dosing in man is targeted toward a comparatively short window in the newborn and infant.

A Model for the Preterm/Neonatal Child

As previously noted, there is currently no practical real way of modeling the common clinical scenario of the preterm/neonate. Although the size of the mini-pig, nonhuman primate, and dog at birth are all sufficient to support intravenous dosing, scenarios involving dosing after preterm delivery by cesarean section require the equivalent of neonatal intensive care units, which is surely a step too far in terms of pragmatism, not to mention the ethical concerns, particularly if primates were involved. Although using one of these larger animal models could certainly be a practical solution for the neonate, there seems to be no plausible alternative to the rat as a preterm model at this stage. Without an ideal solution any choice of animal model or study design is going to have a degree of compromise, and it may be that subcutaneous dosing or neonatal implantation of mini-pumps offers the best solution.

Modeling CNS Toxicity: Could It All Be a Question of Timing?

Much of what is known to date regarding the effects of exposure of the neonate and infant to toxic agents comes from data pertaining to the extended use of antiepileptic drugs. These drugs cause cognitive side-effects, not only during use, but also long-term effects that may be detected in adults, long after medication has ceased (Sulzbacher et al. 1999). Although studies in juveniles with agents such as phenobarbital and hexachlorophene have caused CNS pathology that has been shown to be predictive of toxicity in children, these studies have by and large been at quite high doses (Fonseca et al 1976; Towfighi 1980). Clearly a major challenge is to detect potential hazards where there is no clear CNS pathology, and to assess those hazards in a risk benefit context. In this respect it would perhaps be beneficial to consider a model that we would expect to be oversensitive. In rodents at least, and almost certainly in man, the most vulnerable period for the CNS appears to be the period during which the brain undergoes its growth spurt. In man, this period begins in utero and continues for the first few years of life. In the rat, this critical period is from birth to day 14, so as long as adequate exposure can be guaranteed over this time frame, this should be a fair model. As described earlier, our current study designs not only miss a significant portion of this critical period, but they also allow time for a cleanup operation. Most pathologists would recognize the difficulty of detecting of neuronal loss after the event, compared to assessment of an acute change immediately after the damage has taken place, so surely, trying the measure neuronal loss long after the damage has occurred is counterintuitive. An initial suggestion would be that where there is cause for concern we should conduct a focussed neuro-pathology study from birth to day 14 in the rat.

As discussed earlier, dosing these tiny animals over this critical period is far from straightforward, and again, the mini-pig may be an alternative to consider if meaningful exposure is otherwise impractical. The development of the mini-pig brain is actually very similar to that of man in terms of myelination, composition, and electrical activity (Dickerson and Dobbing 1966). The CNS growth spurt begins at birth and continues for several weeks, so a similarly timed study should in theory be able to answer the same set of questions in those cases of exposure difficulty. Clearly, this is a very underresearched area and there is work to be done before this could be considered a routine, but given that both phenobarbital and valproate at therapeutic doses trigger a wave of apoptotic neurodegeneration in infant rats, this type of study may have more sensitivity than current study designs (Brodie and Dichter 1996).

There seems to be no reason why this timescale should not also be used to investigate CNS neurotoxicity mediated by endocrine disruption affecting the thyroid hormones. Thyroid hormone production, as it needs to be to ensure normal development, is at a high level at birth, and logic would suggest that disruption of signaling at any level would result in detectable pathological changes. Perhaps this type of investigation, with the incorporation of a reporter gene assay to detect interference at the level of the thyroid hormone receptor, will provide a level of sensitivity that current strategies do not possess.

Taking Advantage of the Unusual Early Immune Development of the Mini-Pig

One interesting aspect of mini-pig development is the lack of transplacental transfer of immunoglobulins from sow to piglet. The result of this is that despite its relatively advanced state of development in many organ systems at birth, the neonatal mini-pig is immunologically naïve and has essentially no leukocytes for the first day or two of life (Rothkoetter et al. 1991; Solano-Aguilar et al. 2000), including antigen-presenting cells, T-cells, and B-cells (Stokes et al. 1999). It takes up to two weeks before activated CD4+ cells appear in the lamina propria and up to 4 weeks before mature memory cells are present. In the first day or two of life for the piglet, maternal milk and colostrum are a rich source of immunoglobulins and other protein. Clearly these proteins, although potentially antigenic, do not elicit an immune response in the infant mini-pig, nor later in the adult. Could it therefore be possible to orally expose the neonatal mini-pig to a novel large molecule in the first two days of life and be left with a nonprimate model that does not raise an immune response when subsequently dosed with the large molecule?

Epigenetics: The Predictive Toolbox for the Future?

It has been highlighted on several occasions that although we do not actually attempt to determine adult risks associated with juvenile exposure, it is highly likely that such effects exist. There are likely to be several scenarios in which we are simply unable to investigate this idea with our currently employed toolkit.

The starting point is to ask ourselves just how adult toxicity is related to juvenile experiences. If you can be persuaded that some of the pathology of the adult is directly linked to epigenetic alterations to the genome in early life, then assessment of pediatric risk must clearly involve the question whether the genome retains an imprint of juvenile exposure.

On a daily basis, the principle of epigenetic changes in the juvenile genome as candidates for carrying the memory of early exposure into adult life is becoming more established. Unfortunately, this principle is simply not catered for in preclinical testing.

However, it is not all doom and gloom—there is help at hand. As is usually the case in science, with the increase in appreciation of a problem and interest comes the availability of technology that can be applied to generate solutions. In this instance, there has been an explosion in the availability of investigative tools to assess the methylation status of the genome. Polymerase chain reaction arrays are now available to quantify the methylation status of any gene of interest.

Clearly the major step we need to take now is to identify relevant panels of genes in each major organ system to give us an insight into the potential future that epigenetic aberration of the genome represents. If we can achieve this goal, we should be able to detect not only the induction of inherently disadvantageous genetic profiles, but also alterations to genes not involved in early development or that have redundant functions in the juvenile and have very subtle, difficult-to-detect adverse impacts in adulthood. These types of change are potentially wider ranging and may result in subtle degenerative changes that degrade both physiological function and anatomic structure faster than would otherwise occur in normal aging (Shi and Warburton 2010).

Looking even further out, it is not impossible to hope that analysis of the whole genome could be a routine procedure. Currently genomewide analysis of methylation status and histone modification is possible, albeit a bit unmanageable, but in the future, who knows? Clearly, if a molecular approach is being followed, then it makes sense to do a toxicity panel at the same time to support the histopathology. Hopefully, it would require only a relatively small study to generate a wealth of usable data, and more than that, may even identify significant epigenetic changes that could potentially be reversed or managed before adulthood.

Conclusions

The extent of testing NCEs in juvenile animals has increased exponentially over the past ten years, but it is questionable whether much of that testing has resulted in usable, relevant data. With this idea in mind, it is difficult to have confidence that our current approach is taking us in the right direction. There will doubtless always be a place for a screening study using the rat to identify any extreme differences in response between adult and juvenile. The rodent model may also have a part to play in neuropathology and preterm investigations, but it may be lacking in terms of investigating some of the more subtle types of change and the delayed effects that may well be of great significance in terms of making a pertinent risk assessment for man.

The guidelines are fairly clear that study designs should ideally cover the corresponding period of development in model and man. Study designs very often appear to be over conservative in terms of dosing period, which in itself is not necessarily an issue, but in these circumstances it would be desirable to see interim sacrifices timed to to critical period of development of putative target organs.

Although the rat model often represents a suitable model for aspects of toxicity in the older child and adolescent, and it is the only current solution for the preterm infant, there are some major shortcomings when it comes to modeling the important group that is the newborn and infant. That gap could perhaps be filled by selection of an alternative model. The dog is unsuitable in so many ways that it is difficult to make any case for its use in juvenile studies, and although a primate model would be ideal, pragmatism suggests that the mini-pig may represent a suitable compromise for many aspects of this particular period of pediatric development, especially in terms of small molecules, and perhaps even for large molecules. As one of the factors on which regulatory advice on study designs is based is data from adult preclinical toxicity data, the mini-pig is clearly disadvantaged compared to other species, as it is less commonly used as a second species in adult preclinical programs. However, as the regulatory agencies take a very case-by-case approach in their advice regarding juvenile studies, there would certainly be instances in which a case could made for the use of this species when the shortcomings of the other available models are considered.

Whatever the choice of model, the human evidence now emerging suggests that identification of epigenetic changes in the young animal will be an important aspect of analyzing risk. In the first instance, one might suspect the significance of such changes would be extremely difficult to contextualize, but given time, it seems possible that mapping a panel of genes in the liver, brain, kidney, testes, ovary, and lung could provide the insight we need to make meaningful predictions of potential hazards. One of the major issues we have with all animal testing, and particularly with juvenile testing, is the extrapolation of data. However, with this type of investigation it is quite conceivable that epigenetic changes seen in animal tests will also show up in man. A retrospective look at human populations should be able to identify the presence of comparable alterations, and it should be quite possible to gain some understanding of the significance of the epigenetic changes that may be induced by early life exposure to NCEs and thus form a reasonable view of the consequences of therapies vital to the survival and well-being of the neonate, infant, and child.

If we consider the range of preclinical studies we perform in adult animals, we can perhaps reflect that it was over-optimistic to expect to answer all the questions we have regarding juvenile toxicity with a single study. Clearly, the case-by-case approach is always going to be a cornerstone for planning the safety assessment of pediatric medicines, but we do perhaps need to consider that in some cases, we will need to ask ourselves a more complex set of questions than we have been used to up to now, and we should be prepared for more innovative study designs.