Antiretroviral therapy can halt HIV disease progression; however, HIV can develop resistance to any antiretroviral compound. Doctors must continuously monitor HIV genetic changes and adjust the combination of drugs to keep HIV under control as long as possible. 

IBM Research Lab in Haifa, together with the EuResist Network GEIE, a non-profit European Economic Interest Grouping composed of Karolinska Institute, Max Planck Institute, University of Siena, Informa s.r.l., and University of Cologne, participated in EuResist, a European Union 6th Framework project. The EuResist project recently announced the availability of a European integrated system for clinical management of antiretroviral drug resistance. 

The EuResist online system can be freely accessed to obtain a prediction of response to antiretroviral treatment for any given HIV genetic variant. The EuResist engine has been trained on the EuResist database, the worlds largest database centered on HIV resistance and treatment response, now including more than 33,000 patients. The system uses innovative technologies and models enabling a smarter and more efficient way to assist antiretroviral treatment compared to the state-of-the-art methods and expert humans. EuResist helps bring higher quality of care for the millions of people infected with HIV worldwide. 

By combining some of the largest databases and creating new prediction engines, EuResist can provide a prediction of how a specific HIV variant will react in a certain person given a certain combination of drugs. The achievement of EuResist will help bring about better medicine, lower treatment-related toxicity and cost savings, which means higher quality of care for the millions of people worldwide who are infected with the virus.

EuResists researchers have created standardized biomedical information integration technology to process and correlate data  such as treatment histories, treatment response information, and the sequence of the relevant part of the HIV genome (genotype)  from three large nationwide genotype-response databases, namely the Italian ARCA database, the German AREVIR database, and data collected from Sweden at the Karolinska Infectious Diseases and Clinical Virology department in Stockholm. EuResist's researchers have recently expanded the system to include also data from Luxembourg, Belgium (Leuven), and Spain (Catalonia). 
EuResist addresses not only the genetic make-up of the virus, which can predict what happens with the virus, but, it also uses other relevant information from the patient history. The technology allows doctors to check a patients previous drug history and then analyzes how previous treatments affect the success of the intended new treatment. Information is retrieved from the EuResist database for past treatments and correlated with the success or failure of the recommended treatment.

EuResist researchers have developed new mathematical prediction models that not only take into account the patients own history, but tap into the wealth of information that EuResist researchers have amassed. The recent expansion of the EuResist database to include information from more than 33,000 patients and 98,000 therapies, and 370,000 viral load measurements makes it the worlds biggest database centered on HIV resistance and clinical response information. Having a large database is crucial to achieve good predictions, some of the prediction methods were not applicable before due to insufficient amounts of data. Notably, more than twenty drugs can be now combined into many different three- or four-drug regimens and the number of mutations in the HIV genome believed to cause or modulate some kind of drug resistance is close to 100. 

Three different, but complementary, prediction methods have been developed in the EuResist project. Each method uses the same type of mathematical model to classify a given drug combination as successful or unsuccessful, but it is fed with different data. In fact, three approaches are used to extract data from the database to account for different aspects of the disease evolution. All three methods have similar performances, but we found that a straightforward combination of the three different techniques consistently improved the prediction, making the whole greater than the sum of its parts, and increasing the success rate for patients.
 
The EuResist infrastructure is open and interoperable, supporting standard interfaces based on Web Services and HL7 v3 CDA. EuResist researchers have defined a novel CDA Template for HIV that enables HIV researchers and caregivers to exchange HIV data in a standard way. 

The confidentiality of HIV patients is paramount. All the data managed by EuResist is completely de-identified. The server that hosts the project is situated in a highly secured environment, transport of data is encrypted, and access to the system is regulated by strict authentication and authorization policies. 

Yes

Yes

Previously HIV caregivers recommended treatment to patients based on the patient's condition, and the caregiver's experience and knowledge. HIV genome was taken into account based on algorithms developed by human experts to predict the activity of the individual drugs on the specific HIV variant. Because the treatment must include multiple drugs and putting the drugs together is very challenging, the wrong HIV treatment could have a very negative effect for the patient.  For example, HIV can change its genetic makeup and become resistant often even to drugs different than the ones being used (cross-resistance). The EuResist engine has been trained on a combination of drugs rather than single drugs.  It is the only available system predicting response to any drug cocktail, meeting the primary need of clinical practice. Now caregivers across the world, experts or novice, can consult the freely available EuResist online system to get a recommendation for the best drug combination for each individual patient. 

Today, people diagnosed with HIV in Western countries have several treatment options that give them a reasonable life expectancy. In spite of this, eradication of infection is not yet achievable, and prolonged - possibly lifelong - antiretroviral therapy favors the selection of drug-resistant viral strains. A number of patients harbor a drugresistant virus, which is currently a primary cause of treatment failure. If a patient has a history of wide resistance to drugs in the three main classes, this is a death indicator.

The EuResist project developed a statistical engine that helps doctors choose the treatment that has the highest probability of halting virus replication and impairing evolution of drug resistance. The innovative EuResist approach uses viral genotype information integrated with treatment response information from clinical practice to predict the success of a treatment regimen against any given HIV genotype.

The achievements of EuResist translate into better medicine, lower treatment-related toxicity and cost savings  giving considerable hope to the 40 million people infected with the virus worldwide.

The on-line EuResist engine has recently been evaluated by the scientific community and is now in use by HIV specialists across Europe. In addition, the system has been presented to HIV researchers and caregivers in Eastern Europe and will be proposed to low to middle income countries as a tool to optimize the clinical use of available antiretroviral compounds. 

Drug companies have also shown interest in EuResist as a possibility for better success rates with their drugs. Some of the companies are supporting or negotiating support to EuResist.

EuResist is the first and only freely available data-driven computational method to predict the success of a treatment regimen against any given HIV genotype, based not only on viral genotype information, but also taking into account treatment response information from clinical practice. It is the only system providing an estimate of activity for combination therapy, rather than for individual drugs.

EuResist is also the best and most effective system of its kind. It outperforms other state-of-the-art methodologies, including the "gold standard" Stanford HIVdb, ANRS and Rega systems, as shown in Appendix 1, and even human experts, as shown in Appendix 2.  

EuResist has managed to create the largest clinically oriented antiretroviral drug resistance database in the world. The ability to analyze clinical, laboratory and demographic data accumulated over the years significantly improves prediction of the right combination of drugs that works for the maximum amount of time. These innovations are being provided as a free tool that can help extend the lives of millions of people who are the victims of this killer disease.  

The EuResist team feels both humbled and privileged by the opportunity to put good science and state-of-the-art technologies at the service of such an important and meaningful cause.       

One of the biggest challenges faced by the multidisciplinary team during the initial stages of the project was to make sure we really understood each other. This was no small task given that the team literally spoke different languages in addition to having different cultural and professional backgrounds. There are doctors, virologists, mathematicians, computer scientists, and IT experts on the team and each came to the project with different ways of thinking and different vocabularies.

Fortunately, the project was blessed with a wonderful team of outstanding researchers united by a true passion for collaborating and helping each other achieve a common goal. You can see a picture of the EuResist team in Appendix 3. 

Additionally, the team faced several technical challenges. Integrating HIV clinical and genomic data from a large set of distinct data sources, in different languages, and with different definitions for treatments, drugs, lab results, sequencing, etc., was a Herculean task. It was also imperative that the integrated data be clean and consistent.   


The definition of the record suitable for training the models (the standard datum) has been also challenging, requiring a careful and exhaustive evaluation of all the variables involved in response to treatment. The evolution of HIV infection is multifactorial, depending on specific complex features of the virus and the host, further complicated by human intervention (i. e. therapy). HIV genotype and treatments are high-dimensional variables requiring encoding techniques. Modeling has been based on several different approaches, including sophisticated methods such as support vector machines, fuzzy logic and instance based reasoning, as well as more common logistic regression techniques. 

The EU Framework Programmes (FPs) have been the main financial tools through which the European Union supports research and development activities covering almost all scientific disciplines. The FP is proposed by the European Commission and adopted by Council and the European Parliament. The FPs are highly regarded by scientists all over Europe and the competition for funding is fierce. 

EuResist presented a proposal under FP6, and was authorized for funding. This was not an easy task, given the very strong competition. Acceptance rate in FP6 was less than 20%! EuResist succeeded based on two main ingredients: a strong interdisciplinary consortium with experienced scientists from top research institutes in Europe and an excellent proposal that presented an innovative vision together with a down-to-earth roadmap on how to achieve it. 

Exceeded

The minimum goal was to integrate three different data sources and base predictions on the available data. EuResist surpassed that goal and now integrates seven different data sources, making it the worlds largest database centered on HIV resistance and treatment response. The database now includes over 33,000 patients from Sweden, Germany, Italy, Spain, Belgium, and Luxembourg. 

The systems predictions are 76 % accurate, outperforming other commonly used HIV resistance prediction tools. The system also outperformed experts in the field. The EVE (Engine versus Experts) study compared EuResist with ten international experts confronted with 25 case histories, where all the clinical and virological information was available, in an attempt to simulate real practice in HIV specialized care. EuResist's predictions outperformed nine out of ten human experts, as shown in Figure 2.

In fact, the ICT ( Information and Communication Technologies) Health unit of the European Commission recently took a detailed look at the success garnered by the project and selected EuResist as the Project of the Month for November 2008.

The system is currently available online through a web interface that is freely available to the global medical community at http://engine.euresist.org 

EuResist has been a milestone in the HIV resistance arena.  It has also fueled debate about global data collection policy. Other large networks have requested data for related research proposals and EuResist has been invited to participate on other applications where the amount of data is critical to achieve good translational research objectives. Pharmaceutical companies are obviously major actors in the field of control of infectious diseases.  Their interest in EuResist is a clear indicator of a possible trend to support similar efforts in which multidisciplinary science coupled with effective setup of large clinical data collection can contribute to better use of anti-infective compounds. 

HIV is not the only virus that can develop resistance to medications. A similar situation happens for example with Hepatitis B and Hepatitis C. We believe the success of EuResist can improve the treatment of other diseases, such as Hepatitis B and C, bringing relief and hope to millions of patients around the world. 

Reaction from the potential user community has been very positive. We are diligently disseminating EuResist via several channels: EU official web sites, printed and web press coverage, international conferences, journals and workshops and more. We have already given more than eighty talks on EuResist,  presented seventeen posters, appeared in eighteen important press articles, and published twenty five papers. A major focus in future plans is to present and offer the system to low and middle income countries that have high rates of HIV. This should allow these areas to benefit from a leading-edge tool at the time they are expanding use of antiretroviral therapy.

The web system has been launched and is still being adapted to the users requests so, we are just beginning to compile usage statistics.  The team is confident that our system can be a relevant tool for HIV researchers and caregivers worldwide. Furthermore, both the EuResist Network GEIE organization and IBM Research Lab in Haifa are deeply committed to continue supporting, expanding and disseminating EuResist. 

A sixteen pages document describing EuResist's prolific dissemination activities can be provided upon request.