This course familiarizes students with the novel concepts being used to revamp regulatory toxicology in response to a breakthrough National Research Council Report “Toxicity Testing in the 21st Century: A Vision and a Strategy.” We present the latest developments in the field of toxicology—the shift from animal testing toward human relevant, high content, high-throughput integrative testing strategies. Active programs from EPA, NIH, and the global scientific community illustrate the dynamics of safety sciences.
Introduction to Read-Across
Loading...
Del curso dictado por Johns Hopkins University
Toxicology 21: Scientific Applications
10 calificaciones
De la lección
Module 4
The first lesson in this module will explain the concept of read-across, why it is important for regulatory risk assessment, and how it can be implemented by industry in response to REACH legislation. The organotypic cultures lesson will teach you about existing in vitro cell-based models starting from simple monolayer cultures of cell lines and ending with complex combined organ-on-a-chip and human-on-a-chip approaches.
Conoce a los instructores
Lena SmirnovaResearch Associate
Center for Alternatives to Animal Testing
Thomas HartungProfessor
Environmental Health and Engineering
[MUSIC]
Welcome, this is Thomas Hartung from Baltimore,
from the Johns Hopkins Bloomberg School of Public Health.
Welcome to another lecture in our series on Toxicity Testing in 21st Century.
And this time we're going to talk about Read-Across.
Read-Across very much emerging technology using computer
technologies to help us with some of our problems in safety assessments.
I would like to start introducing the concept of Read-Across.
And then I will show you subsequently what is happening at this moment and
why this technology is more and
more used over the last three years in the field of safety assessments.
A lot of the revolution and toxicology which we summarized under the term
of toxicity testing in the 21st century is around big data.
Because many of the new technologies,
the technology of the 21st century, are allowing us to produce
large quantities of data in a short period of time.
We have talked in this lecture series a couple of times about
the high-throughput screening programs by EPA,
the environmental protection agency and DNA HFTA EPA collaboration.
So the foot and rock administration, which you know as ToxCast and
Tox21 they're producing obviously big data.
We have also talked about Omics technologies, transcriptomics,
metabolomics, exposomics and others.
These hold some assessments of how systems are influenced by
toxic effects of substances.
They're producing sometimes many,
many gigabytes of data in one single experiment.
There's also some high content imaging as it is pursued, for example,
in the European project of EUToxRisk, but
we have not really addressed this so far in this lecture series.
Today we will focus on the last of these which is the use of legacy data.
Historical data, which are increasingly created in large databases, and
these databases allow us to mine data and
carries out something which is termed Read-Across.
Read-Across is in principle, I do not have data for
a certain substance like the one shown with the question mark here,
but I do have possibly data for neighboring substances.
Substance which share structural similarity.
So since all of these seem to be active for
a given health effect, we might interpolate that probably also
this substance in the middle is actually active with respect to this hazard.
This is the simple form of Read-Across agreeing identifying substances which
have some similarity and concluding best interpolating that the substance
of interest shares a certain property, is the data gap-filling technology.
So instead of a costly and lengthy process of producing new data,
I simply assume I can predict this property.
Very similar to this is what we call a category approach.
In this case we would assume that we don't have data for all of this five substances.
And you would choose, with some reasoning, one or two representative
molecules of them will test this and use the data for the entire group category.
Both of these approaches are quite popular because they are saving a lot of money,
they are saving a lot of time.
But they are also depending on the quality of this Read-Across very much
because it works if certain quality is ensured.
And this is exactly what we are going to talk about today.
Read-Across is increasingly being used.
This is an example for the number of scientific articles using
Read-Across between 2000 and February this year, and
you can see how publications and citations were steadily increasing
with respect to the terms Read-Across combined with toxic.
So it is something which is emerging.
And this has to do with both the availability of data, the need for
safety assessments from [INAUDIBLE] chemicals as
we have discussed in the series and also simply the capabilities
running big datasets with the respect of computing power.
The first mentioning of Read-Across as a term we could identify so
far goes back to the year 2000.
This abstract at a conference reproduced on toxicology letters said,
read-across of toxicological data and the notification of new chemicals and
I think this is the grandfather of all they've begun to present.
The reason why these technologies are being used at
the moment increasingly is especially the European REACH legislation.
We have made reference this legislation several times even recall that
Europe went ahead about ten years earlier than the most recent
reauthorization of the toxic substance control in the West.
Europe started in 2006 with the REACH legislation, which is regulating
the existing and new chemicals, industrial chemicals on the European market.
And we have seen so far two phase in deadlines 2010 and
2013 where chemicals produced or
marketed in the European Union above 1,000 tons in 2010 and
above 100 tons in 2013 had to be registered.
And quite impressively 75% of this dossiers use Read-Across.
It is one of the most common used approaches in this dossiers.
Other alternative methods, cell cultures, quantitative structure,
activity relationship and others are hardly used.
At the same time, the expertise how to do Read-Across is actually very
low in the industry with the notably exception of the big chemical
companies who have their own departments for safety assessments.
And sometimes often do have some computation
toxicologists skilled also read-across.
The majority of companies has no expertise in this at all and as you have seen,
the scientific literature on Read-Across is only emerging.
So there is, at this moment,
not really an agreement on how to do Read-Across properly.
And for the first two deadlines,
we have also seen that the European Chemical Agency,
EChA did accept of those checked so far only very few.
And this creates a situation where there's a discrepancy of the possible utility,
the possible speed up and
cost saving of this technology and the available knowledge how to perform.
Just to show you EChA's regularly reporting on the use of Read-Across.
This is one of the last reports from 2014 and they're showing in this report
which can be downloaded at the URL at the bottom of the slide.
They're showing in blue and is both two circles, how for example, for
reproductive toxicity or developmental toxicity, read across is broadly applied.
So quite substantial number of the registration dossiers are using this
technology.
This theme shows you the principles of Read-Across.
Read-Across is typically done first of all from one substance to the other.
It is based on structure activity relationships SAR, and you can see here
assuming for a given number of substances here which is a theoretical mode,
we do have some data points shown as filled circles.
And we tried to extrapolate the similar structures which have no data.
Ideally, we do this as an interpolation.
You see here that the increase in chain length of these alcohols is actually
covered by data points at the smallest and
at the longest side of the series of molecules.
Such interpolation as the favored type of Read-Across.
However, very often it is just one substance to one substance type of
comparison which is carried out.
An example of a category is given here in this published example.
You can see that for these four alcohols, some data are available.
These are all multigene city data, some prove to be negative and those which were
not performed where then substituted was a data extrapolated from the tested ones.
That's a typical application for
a relatively easy and homogeneous group.
But these things can get quite complicated.
Here is another group, these are tallied acids and esters and
you can see here, how already the physical, chemical properties,
in part, have been right across or have been estimated.
They're showing a pretty homogeneous behavior with regard to molecular weight,
water solubility and lipophilicity of these substances.
Then the question is, can we actually fill the data gaps?
You see the data available for these substances, there's relatively few.
And they also did conclude in this paper,
of which you can consult, if you're interested in more details.
Yes you can, but only if you have high quality source data.
If there is some consistency within the data for the categories or
no surprises or sudden changes in behavior.
If you are interpolating, which means, if you are having,
within a serious the edged positions covered, and
there is a good reason and justification for data gap filling.
So you need to justify why you would apply such a technology.
Read-Across has a lot of potential applications.
REACH, the European legislation and other global legislations you've heard about
REACH like programs and career in Taiwan, programs in planning and China and Turkey.
But also the authorization, they are possibly customers for
such type of approaches.
And this can be done then for the existing chemicals, the backlog of old and
tested materials, as well as for new substances which are coming to the market.
A Read-Across can be especially be used for to prioritize.
We can identify what is smelling like a problem, what is possibly deserving larger
testing, and what can very likely be put on the back burner does not rectify if
it's the same priority testing and we can use these methodologies for an assessment.
The classification and labeling, here abbreviated as C&L is giving,
as you heard earlier, labels to chemicals.
This is a sensitizer, is such a label and read across helps to assign such labels.
It is also a source of information,
usually an evade of evidence type of approach for risk assessment.
But Read-Across can also be used, as we will hear later in this lecture again,
also, earlier in the [INAUDIBLE] stewardship, already by the identification
of substances to be considered for a certain product.
New product registration can be supported especially in the US legislation by
the environmental protection agency TOXCA.
We do have premarketing notifications and
this premarketing notifications can be informed by such type of read-across.
You've heard in this lecture series about the adverse outcome phase but
also about the integrated test and strategies.
The integrated approach to testing an assessment and
they can include elements of Read-Across.
And there's perspective of using this for the neutral interest,
the nanomaterials, the pharmaceuticals and the development or
imagine almost any drug which comes to the market is not completely pure.
We cannot synthesize completely pure chemical molecules.
There will always be some reaction products.
There will always be some precursor molecules in small impurities.
We know that they're there because our chemical analytics allows us to identify
these impurities, but we cannot carry out traditional type of risk assessment.
This would require to synthesize at least 10 to 20 kg of the substance,
this is practically and on a cost side impossible.
And there's also new challenges in evaluating the safety of
illicit drugs of legal substances or abuse.
For example, we mentioned earlier the e-cigarettes which are at the moment
challenging our regulatory system.
And this is just giving you a flavor of the broad variety of
applications Read-Across could have if it's only working.
At the same time, there are some concerns because not always is structure predicting
biological effect or minimal changes in structure can have dramatic differences.
This is a very well known example, I will disclose in a second.
These are two stereoisomers of the very same drug.
The drug is Thalidomide.
Thalidomide, which was leading to a scandal in the late 50s and
early 60s because of malformation being produced.
And interestingly, as was found out later,
one of the two stereoisomers is the Teratogen and
the desired sedative effect is actually being produced by the other stereoisomer.
Unfortunately from the synthesis the drug was isomes so
it was a mixture of both of these stereoisomers.
In this lecture the scandal about 10,000 babies with malformations
of the limbs as you can see in these pictures, were identified from this
substance which was marketed as Contergan by the pharmaceutical company Grünenthal.
This was the trigger for a lot of reproductive toxicity testing in drugs and
in consequence also for chemicals.
And fortunately something like this has not happened again.
But Read-Across would not have helped us here, as you can see from this example.
So don't expect it to be the answer to all problems.
Let's take this example.
Very similar molecules, the only difference is the hydroxyl group by
chloride, both very electronegative,
but one of then is non skin sensitizer.
The other one is a very strong one, very importance skin sensitizers,
though, you would expect that the behavior of these molecules is relatively similar.
So the distinguishing in Read-Across, also,
not only the structure of analogues, you see here again,
the long chain alcohol's, as the example for structural analogues.
But there's also mechanistic analogues, which sharing, as you can see,
a certain structure in the molecule to induce certain mechanism.
But there's also mode of action analogs like these substance which are all
activators of the estrogenal receptor and are for
this reason, disruptors, as a mode of action analogs.
So it is not only a structure, it is also the shared substructures or
the shared activation of a receptor which can be used as a basis for Read-Across.
We also have to ask, what is happening to these molecules?
Very easy cases are if substances are very quickly and
completely metabolised to a common molecules.
So we can then do the risk assessment on this common metabolite.
Similarly, if spontaneously substances degrade to common products,
we might do a risk assessment for them as one category.
But we will also have to talk about other metrics for molecular similarity,
because it is not only the shared chemical substructures, it is a lot about
the descriptors of a molecule which can help us to establish similarities.
In this is case for
the chemical similarity, we need first of all data source.
Data source on chemicals associated with some type of test data.
We then would move forward and use some type of chemical similarity function,
which is a description of the molecule if an example two molecules which seem to be
rather similar, their shares some functional groups.
And now we need a way of expressing the similarity.
We call this also finger printing.
And then we can carry out on this basis our Read-Across here an example, for
structures which are inhibiting vitamin K dependent activities on blood coagulation.
The chemical's fingerprint is in the end translating our molecule into
a series of zeros and ones if there is a certain structure present or not.
And our fingerprint metric which is assessing the similarity between
the two molecules is converting these two strings of zeros and
ones into a measure of similarity.
And in the literature there's quite a few different ones.
This is a incomplete list tanimoto indices have been used most frequently
in the literature which is a comparison based on 900 different functional groups.
Whether they are present or not present in a given molecule and in this case,
the overlap between all function groups, the two molecules would be
used in order to define the Tanimoto distance or similarity.
But as you can see, there's others and I don't want to go into the details,
how mathematically the similarity is being calculated in these different approaches.
The key issues which we'll have to discuss in the rest of our lecture is
how can we support Read-Across predictions?
What can we provide as evidence that a chemical belongs to a group,
that two chemicals are similar, and what role can biological evidences play?
So going beyond structure.
And in the end for
its utility we want to know when can we replace the traditional test?
For example, an animal test.
We will discuss this question in the upcoming sections of the lecture.
[MUSIC]
Explora nuestro catálogo
Inscríbete de manera gratuita y obtén recomendaciones personalizadas, actualizaciones y ofertas.
Coursera brinda acceso universal a la mejor educación del mundo, al asociarse con las mejores universidades y organizaciones, para ofrecer cursos en línea.
© 2018 Coursera Inc. Todos los derechos reservados.
