In silico methods simply means computer-based methods.  In this context they are computer-based methods for assessing the toxicity of chemicals.  

In silico methods involve using databases of existing experimental data on the toxicity of chemicals.  That data is used to predict the toxicity of chemicals which are similar to those already tested.   The advantage of in silico methods is that they make it possible to assess the toxicity of large numbers of chemicals quickly using a database and software on a desktop computer, whereas assessments were formerly only possible from long and expensive tests in laboratories.  However, the main limitation is in the amount of existing data available; much is held by industry as confidential, and in some areas (such as long term toxicities for mammals) the amount of existing data of high quality is limited.  

In silico methods are widely used (including by regulators in the US) to screen and identify chemicals for priority testing in the laboratory. They are also used in addition to available laboratory data to generate a stronger ‘weight of evidence’.  In silico methods include QSAR models, ‘Read across’ and Virtual Screening.

The term in silico contrasts with in vivo and in vitro.  In vivo methods involve testing chemicals on living organisms, including invertebrates, fish and animals.  In vitro methods involve test-tube research, where (for example) instead of putting the chemical on the skin or in the eye of an animal, the chemical is put in a test tube with some living skin cells or eye cells taken from an animal.  Both in silico and in vitro methods are known as ‘alternative methods’.  In silico methods make use of existing data from in vivo and in vitro tests.

Using QSARs requires expertise, firstly in order to judge which model to use for a particular substance and endpoint, and secondly in order to evaluate the level of reliability in the results.  

Company managers may employ consultants or specialist employees to carry out the evaluation, or may develop the expertise to do it themselves.  We suggest all managers can benefit from a core understanding of what the models can produce, and of what factors will increase or decrease the reliability of a result, in order to be able to use the toxicity evaluation wisely as a basis for decision-making.   (They may also need this understanding if their advisors or consultants are attached to in vivo laboratories and potentially pre-disposed to prefer in vivo tests.)

Expertise and care in practice are vital for the future trajectory of QSAR models.  Now and in the future they may prove to be an increasingly valuable technology, with a potentially important function for protecting human health and the environment.  It is therefore vital to avoid the all-too-familiar trajectory of a developing technology being initially over-stated in terms of its capabilities, and then discredited when it is over-applied or used in an uninformed way.  

It is therefore in the public interest, and in the interests of all professional stakeholders, that QSAR models are explained openly, and are used appropriately and wisely in practice.  That principle underlies the ORCHESTRA project, and is why we offer critical understanding rather than promotion.

Furthermore, ORCHESTRA is devoted to make more explicit the evaluation of the results obtained from QSAR models, in particular through the VEGA platform.

ORCHESTRA is preparing a book on QSAR, which will be made freely available, describing in detail the theory and the practical use of the QSAR models.

QSARs have been used for decades in the development of pharmaceuticals, where a drug is to be developed to achieve a particular biological action.  Our interest here is in the reverse use, where the chemical is known, and the biological action is to be predicted.  That too has been the focus of research for decades.  But here we are concerned specifically with their use for evaluating toxicity within the regulatory framework of REACH, and it is early days.  

In practice, the use of in silico methods by European industry within REACH is still limited.  There are  three key practical issues that can delay their use.  They are highly inter-connected.

(i) Progress in developing models for regulatory use:  The limitations for model development are primarily the lack of good available experimental data as the basis for developing models.  The results of the many thousands of past animal experiments carried out or commissioned by industry are held as confidential by industry.   (The regulatory demand for the transparency of QSAR models inevitably requires that they not only reveal the experimental data used to develop the model, but also make available for independent review the details of how the experimental results were achieved.)  In his video interview (see documentary, part 4) Professor John Dearden made a plea to industry to release that data.  

A positive development is that it is now clearer to QSAR developers than ever before, exactly what is expected of QSAR models for regulatory use.  The regulatory framework of REACH, the OECD principles and the ECHA guidelines effectively work together to increase the demands on models in terms of rigour, reliability and transparency.    

(ii) Uptake by industry and consultants: uptake requires awareness of the models available, confidence in their reliability and confidence in regulatory acceptance if they are used.  (See the FAQ: ‘Do in silico methods require expertise, or can anyone use them?’)  The use of in silico methods for REACH also involves a simultaneous shift from simply using in vivo methods as an accepted form of evidence, to using different and complementary sources of evidence.  

(iii) Acceptance in practice by regulators (ECHA and the national competent authorities):  The ultimate responsibility of regulators is to protect human health and the environment, so there is some caution in accepting the use of alternative methods.  The experience of European regulators in using QSARs is also still limited.  

The regulation has made the shift to animal testing being ‘only as a last resort’, but process requires confidence, shared expertise and communication between these three major stakeholders.  It may be encouraged by wider interest among industry shareholders, publics, and policy makers, in realisation that there are alternatives to animal testing.  

QSAR models are available and in use for some endpoints, while being limited for others (e.g. long term mammalian toxicity).  There are several reasons, including:

1. the limited amount of high quality experimental data currently available (and released by industry) for some endpoints,
2. the higher complexity of the chemical and biological processes involved in generating the toxicity for some endpoints, and
3. the lower regulatory acceptance for certain endpoints.  

In general, for physico-chemical properties QSAR models are more reliable, because the phenomenon to be modelled is less complex, and thousands of experimental results are available in the databases.  Whereas, for models to predict chronic effects, the data are much more limited, and the phenomenon is much more complex.

The EC funded ANTARES project is evaluating existing models which could be used for REACH.  Hundreds of possible models have been identified and listed, which theoretically could be used for tens of REACH endpoints. ANTARES is currently checking the performance of the QSAR models for several endpoints.  (See: FAQs: ‘Is there an independent review of QSAR models’ and ‘What makes a good QSAR model’.)