Final Report Summary - TAILORED-TREATMENT (DEVELOPMENT OF TAILORED ANTIMICROBIAL TREATMENT REGIMENS AND NOVEL HOST- PATHOGEN INSIGHTS FOR RESPIRATORY TRACT INFECTIONS AND SEPSIS)
Executive Summary:
In the past 70 years, antibiotics have been one of the most important weapons against infectious diseases. Unfortunately, they are now one of the most misused drugs in the world. Importantly, this misuse has led to the development of a wide range of antibiotic resistances, representing one of the major threats to global health. A major factor in helping to prevent the development and spread of antibiotic-resistant bacteria is appropriate antibiotic treatment that is tailored to the pathogen (if any) that is actually causing the disease. However, one of the major problems facing clinicians is deciding which (if any) antibiotic therapy should be prescribed in the 12–48-hour period before the causative agent of the infection is identified. Furthermore, patients may die if they are prescribed incorrect antibiotic therapy or if no antibiotic therapy is given. On the other hand, the indiscriminate use of antibiotics, for example in treating viral and fungal infections that do not respond to antibiotics, may lead to the development of antimicrobial resistance, as well as causing unnecessary side-effects. A (rapid) method to help clinicians to tailor antimicrobial prescribing to individual patients would help reduce the development of antimicrobial resistance and unwanted side-effects associated with unnecessary treatment. The European Union-funded, 4-year TAILORED-Treatment (TTT) project established a broad-based strategy (not limited to a particular infection) that can be used to increase the effectiveness of antibiotic treatment, reduce potential side-effects of therapy, and help to limit the emergence of antimicrobial resistance in hospitalised children and adults. At the heart of the TTT project is a clinical study that involved hospitalised patients with respiratory tract and/or bloodstream infections, including both children and adults. State-of-the-art diagnostic techniques have been utilised to generate transcriptomic, proteomic, genomic, and microbiome data, which has been assembled into a single database. This database (HOPOIT) is being used to identify novel interactions that characterise both patients and their infections, in order to discover new biological markers of infection and to develop new computer tools that will enable clinicians to tailor antibiotic therapy in an appropriate and effective way to individual patients.
Project Context and Objectives:
The treatment of infectious disease using antibiotics has been one of the most important advances in modern healthcare, saving millions of lives since their discovery and widespread use. Despite their immense contribution to global healthcare, the CDC recently reported that ‘up to 50% of all the antibiotics prescribed for people are not needed or are not optimally effective as prescribed’. Antibiotic overuse typically stems from prescribing these drugs to treat nonbacterial diseases (mostly viral infections) for which they are ineffective. For example, in the USA alone, over 60 million annual cases of viral influenza are prescribed unnecessary antibiotic therapy. Antibiotic misuse has severe health and economic outcomes. Overprescription of antibiotics may cause preventable adverse events such as allergic reactions, intestinal yeast infection and antibiotic-associated diarrhea. These preventable adverse events may impact patient care and result in lengthy hospitalization. On the other hand, delayed or no antibiotic treatment in cases of bacterial disease is also common (24–40% of all bacterial infections) . While this may reduce the risk of antibiotic-related adverse events, such practices can lead to disease-related complications resulting in increased rates of morbidity and mortality.
One of the most alarming consequences of antibiotic overuse is the emergence and spread of multidrug-resistant bacteria. Resistance of microbial pathogens to antibiotics is increasing worldwide at an accelerating rate, with a concomitant increase in morbidity and mortality associated with infections caused by antibiotic-resistant pathogens. At least 2 million people are infected with antibiotic-resistant bacteria each year in the USA alone, and at least 23,000 people die as a direct result of these infections. In the EU, an estimated 400,000 patients present with resistant bacterial strains each year, of which 25,000 patients die. Consequently, the WHO has warned that therapeutic coverage will be insufficient within 10 years, putting the world at risk of entering a ‘post-antibiotic era’, in which antibiotics will no longer be effective against infectious diseases. The CDC considers this phenomenon ‘one of the world's most pressing health problems in the 21st century’.
Antibiotic overuse in hospitals and outpatient settings contributes significantly to the rising prevalence of antibiotic resistance. At the heart of this problem is the challenge of accurately distinguishing between bacterial infections (which warrant antibiotic therapy) and viral infections (for which antibiotic treatment is generally not required). This diagnostic gap is driven by the inability of current diagnostic tools to provide rapid and accurate information regarding the etiological basis of an infection.
Conventional diagnostic approaches depend on the cultivation of infectious agents and subsequent testing for antibiotic sensitivity and resistance. This process requires lengthy cultivation periods (days) and is not applicable for certain bacterial infections, or for most viral infections. Nucleic acid amplification-based tests (NAAT) for direct pathogen detection are showing considerable promise. Their advantages include high sensitivity and the simultaneous detection of multiple pathogens. Consequently, these tests are increasingly used in hospital and laboratory settings. However, NAAT protocols exhibit varying degrees of sensitivity and specificity when identifying specific pathogens and antibiotic resistance traits. In addition, NAAT diagnostic technologies usually require direct sampling of the pathogen. Such sampling is often not easily feasible if the infection site is not easily accessible (e.g. sinusitis, middle-ear infection and bronchitis) or the site of infection is unknown (e.g. fever of unknown origin). The described limitations of current diagnostic procedures leads physicians to either overprescribe (‘just-in-case’) or underprescribe (‘watchful waiting’) antibiotics, both of which can adversely impact patient care and health economics. Therefore, there is a clear need for novel solutions that will empower physicians to make early, evidence-based antibiotic treatment decisions, in order to improve patient care, reduce adverse events and limit the spread of antimicrobial resistance.
Meeting the challenges of antibiotic resistance requires adopting an integrative approach that expands the antimicrobial arsenal on the one hand and reduces antibiotic consumption on the other. New antibiotic drugs are desired, but will likely provide only a temporary solution in light of the inherent ability of microbes to develop new resistances by a variety of physiological mechanisms, including the exchange of transferable genetic elements, uptake of DNA by transformation and transduction and DNA mutation. Moreover, only a handful of new antibiotics are likely to reach the market in the near future due to the ‘broken pipeline’ associated with their development. Complementary to the development of new drugs is the need for a significant reduction in antibiotic consumption. This goal may be achieved by increasing public awareness of antibiotic misuse and its consequences, as well as by adopting new diagnostic approaches to bridge the current diagnostic gap.
New technologies are constantly emerging, providing the scientific and medical communities with powerful tools to improve the diagnosis of infectious diseases. Advances in the analysis of pathogens and the host immune response to infection in a broad and sensitive manner have led to a deeper understanding of complex host–pathogen–treatment interactions. Technologies such as next-generation sequencing provide snapshots of the patient's entire transcriptome in response to an infection or treatment, as well as enable the genetic analysis of the patient's natural microbiota (microbiome). These technologies also facilitate the search for novel genetic variation markers (single nucleotide polymorphisms [SNPs]) that may predispose individuals to specific infections and disease progression. Most importantly, advances in bioinformatics algorithms and ‘big-data’ analysis enable the integration of clinically relevant information such as host genetics, microbiota, response to treatment and data on the disease-causing agent. Mining this integrated data to generate treatment algorithms, coupled with the development of intuitive web-based interfaces, will be a major step forward in the management of infectious disease patients. Not least in the ongoing battle against antimicrobial resistance.
The EU recently funded the 4-year ‘TAILORED-Treatment’ research program, which focussed on the rapid and correct diagnosis of infectious diseases for guiding antibiotic treatment in patients presenting with respiratory tract infections and/or sepsis. The project combined state-of-the-art omics-based techniques and data with multivariate analysis methods and newly developed bioinformatics software with the goal of translating novel host-pathogen insights into treatment decision support algorithms. At the heart of the project was a large multicenter prospective clinical study that enrolled 1,222 adult and pediatric patients. Subsets of patient samples were investigated for host transcriptomic and proteomic response, nasal microbiota, SNP analysis and pathogen diagnostic proteomics. The data was assembled into a dedicated TAILORED-Treatment database (HOPOIT) , which is being investigated for significant associations between multiple biological and clinical parameters in order to determine the best combination of factors that determine and can distinguish between 'Bacterial-Infection', 'Viral-infection, 'Both Bacterial and Viral Infection' and 'No Infection' classifiers. Current analysis has identified factors that can be used to distinguish between the above classifiers with prediction accuracy of between 95% and 98%, dependent on the number of parameters included in the analysis.
Project Results:
A total of 1, 222 patients were recruited for the TAILORED-Treatment project, from the Netherlands and Israel with age distributions ranging from <1 year of age (16%) to >60 years of age (24%). An expert panel of clinicians were recruited to examine electronic Case Report Forms (eCRF) in order to determine a 'gold standard' diagnosis on which the TAILORED-Treatment classifiers of 'Bacterial-Infection', 'Viral-infection, 'Both Bacterial and Viral Infection' and 'No Infection' could be based. Using these criteria and only the eCRF data, the expert panel concluded that 75% of 522 children and 36% of 352 adults (who could be 'accurately' diagnosed by the expert panel) suffered from a viral infection.Virology studies showed that flu, RSV and rhinovirus were the most frequently detected viruses with higher proportions of viral infection in children than in adults. As expected, for 10% of the children no diagnosis was made. The inappropriate use of antibiotics was found in 41% of viral infections. SNP analysis was performed on 341 patients with 283 samples generating good quality GWAS data for statistical analysis.
The majority of SNPs in the TTT cohort were associated with patients of European descent. Analysis of SNPs associated with common infections (using a Candidate SNP Approach) identified a SNP with a frequency in the TAILORED-Treatment population, which deviated more than 5% from other European populations. Further SNPs located on a particular chromosome were also found to be interesting for further study.
MeMed applied biochemical and advanced high-throughput technologies to explore the host proteome and host transcriptome in search of novel biomarkers that distinguish between viral, bacterial, and fungal infections, both in children and adults. The clinical informative value of single biomarkers can be sensitive to inter-patient variability, including gender, age, time from symptom onset, clinical syndrome, pathogen species and co-morbidities. Therefore, protein- and RNA-based host response signatures were developed by computationally combining multiple biomarkers, which can identify different types of infections while overcoming the challenge of patient heterogeneity. The research initiated with a screening process that included bioinformatics analysis of publically available databases using our newly developed bioinformatics platform, augmented by manual curation of literature covering up to 20,000 potential proteins. This process resulted in a list of several dozens of protein candidates, some of which with an established role in the host immune response to infection, and others with no known direct link to the immune system. First, che clinical informative value of 70 potential protein biomarkers was evaluated using 80 clinical samples obtained from well characterized patients. Next, the levels of the most informative proteins were further measured in increasing number of blood samples (ranging between 500-1200 patients). The best performing single protein biomarker, identified in the screening phase, was TNF-related apoptosis-inducing ligand (TRAIL). TRAIL was induced in response to a wide range of viral infections and surprisingly, also exhibited significantly reduced levels in bacterial infected patients. Finally, statistical classification algorithms were used to identify protein signatures that can increase diagnostic performance using iterative cycles of training and validation . The signature with the highest precision included both viral- and bacterial-induced proteins: TRAIL, Interferon gamma-induced protein-10, and CRP. MeMed also developed RNA-Based immune host-response signatures for diagnosing the source of infection applying several stages: (i) RNA purification from clinical samples; (ii) quality assurance tests to verify integrity levels of purified RNA; (iii) high-throughput measurements of RNA expression levels to identify differentially expressed transcripts as potential new biomarkers; (iv) computational analysis of high-throughput measurements; (v) constructing multi-parametric immune signatures by implementing feature selection algorithms and supervised learning strategies. Comprehensive transcriptome analysis using Affymetrix HTA 2.0 arrays was performed on samples from 130 patients with well characterized infections. The process resulted in highly accurate RNA-Based immune host-response signatures that included both traditional and novel RNA transcripts.
More than 1,200 patient nose swab samples were processed and sequenced for microbiota profiling using Illumina sequencing technology. Five hundred-sixteen samples contained sufficient microbial biomass (> 1,000 16S rRNA gene copies/ul) and sequence read coverage (> 1,000 sequence reads/sample) to ensure that the 16S rRNA gene sequencing results obtained were valid and no bias was obtained from the laboratory environment or reagents used (which may be potential sources of contaminating DNA in the clinical setting when using low biomass samples such as nose swabs). At the same time, a novel microbiota platform (MYcrobiota) that facilitates the implementation of microbiota diagnostics into the clinical diagnostic laboratory by providing quantifiable bacterial identification for clinical samples was developed and evaluated. Hierarchical clustering methods identified that 9 different bacterial operational taxonomic units (OTUs) dominated the nasal samples. Preliminary investigations were also performed regarding the utility of Oxford Nanopore sequencing technology for microbiota sequencing using a subset of Illumina sequenced TAILORED-Treatment nose swab samples.
UGOT has identified mass spectrometric protein markers for: 1) species and strain identification and 2) antimicrobial resistance; in relevant bacterial respiratory pathogens (i.e. S. pneumoniae, H. influenzae, M. catarrhalis, S. aureus, etc.). The MS-proteomics methodology used was optimised for: 1) rapid pathogen identification; and 2) detection and analysis of expression of antimicrobial resistance factors (e.g. CTX-M/OXA/SHV/TEM, ampC, carbapenemases and porins). A database containing the proteomic markers of microbial pathogens of respiratory infections has been developed, based on an established database with more than 2,500 antibiotic resistance genes, representing all major resistance mechanisms. LC-MS/MS and LPI proteomics-based rapid pathogen detection has been optimized, with respect to sensitivity, specificity, etc., and applied directly to the analyses of TAILORED-Treatment clinical samples, i.e. without prior cultivation and isolation of microbial pathogens.
A dedicated TAILORED-Treatment database (HOPOIT) and graphical user interface (GUI) was developed, including a data-upload module. Multivariate analysis tools have been integrated into the HOPOIT database. . The HOPOIT system is made of 3 parts: 1) A central
relational database that stores and serves all the data produced in the project, 2) A web platform that includes data management, exploration, analysis and visualization functionalities and 3) A local application that manages all the internal processes, like file parse and dump, quality control, dataset creation or analysis threads management. Multivariate analysis tools have been added to HOPOIT, including three types of tools: 1) Variables selection tools, 2) Data mining algorithms and 3) Exploratory tools. The variable selection and data mining tools have been codified in C#, bundled in dlls (dynamic-link libraries) and tested individually. A web platform has been developed to integrate the previously developed database and the analysis tools. It includes user-friendly visualization tools to ease the use of the analysis tools and the interpretation of the results for physicians and data access control in order to protect the confidentiality of the data while enabling data sharing to the scientific and medical community. Current multivariate analysis of TAILORED-Treatment data within HOPOIT has determined that the best combination of factors can distinguish between 'Bacterial-Infection', 'Viral-infection, 'Both Bacterial and Viral Infection' and 'No Infection' classifiers with a prediction accuracy of between 95% and 98%, dependent on the number of parameters included in the analysis.
Potential Impact:
The treatment of infectious disease using antibiotics has been one of the most important advances in modern healthcare, saving millions of lives since their discovery and widespread use. Despite their immense contribution to global healthcare, the CDC recently reported that ‘up to 50% of all the antibiotics prescribed for people are not needed or are not optimally effective as prescribed’. Antibiotic overuse typically stems from prescribing these drugs to treat nonbacterial diseases (mostly viral infections) for which they are ineffective. For example, in the USA alone, over 60 million annual cases of viral influenza are prescribed unnecessary antibiotic therapy . Antibiotic misuse has severe health and negative economic outcomes, with antibiotic overprescribing being associated with preventable adverse events such as allergic reactions, intestinal yeast infection and antibiotic-associated diarrhea. These preventable adverse events may impact patient care and result in lengthy hospitalization. Importantly, one of the most alarming consequences of antibiotic overuse is the emergence and spread of multidrug-resistant bacteria. Resistance of microbial pathogens to antibiotics is increasing worldwide at an accelerating rate, with a concomitant increase in morbidity and mortality associated with infections caused by antibiotic-resistant pathogens. At least 2 million people are infected with antibiotic-resistant bacteria each year in the USA alone, and at least 23,000 people die as a direct result of these infections. In the EU, an estimated 400,000 patients present with resistant bacterial strains each year, of which 25,000 patients die. Consequently, the WHO has warned that therapeutic coverage will be insufficient within 10 years, putting the world at risk of entering a ‘post-antibiotic era’, in which antibiotics will no longer be effective against infectious diseases. The CDC considers this phenomenon ‘one of the world's most pressing health problems in the 21st century’.
The TAILORED-Treatment project has successfully obtained samples from more than 1200 patients (children and adults) suffering from respiratory infections and sepsis. Subset of these specimens have been investigated using state-of-the-art technologies in order to generate novel algorithms accurately and rapidly diagnose if a sick child or adult is suffering from a bacterial, viral, both or no viral or bacterial infection. The use of such an algorithm will allow clinicians to rapidly assess whether a patient requires antibiotics or not, thereby helping reduce antibiotic overprescribing (associated with adverse events for patients and an increasing antibiotic resistance) and consequently decreasing the 2 million people infected with antibiotic resistant bacteria each year in the USA and the 400,000 patients infected within the EU. Further, acceptance of TAILORED-Treatment algorithms will lead to reduced morbidity and mortality on a worldwide scale, as well as an overall reduction in health care budge costs.
The project has generated several peer-reviewed publications and has been the topic of poster and all presentations at a range of international scientific conferences. Further, exploitable foregrounds has been identified, what patents has been submitted and another possible patent submission is currently under review.
List of Websites:
www.tailored-treatment.eu
In the past 70 years, antibiotics have been one of the most important weapons against infectious diseases. Unfortunately, they are now one of the most misused drugs in the world. Importantly, this misuse has led to the development of a wide range of antibiotic resistances, representing one of the major threats to global health. A major factor in helping to prevent the development and spread of antibiotic-resistant bacteria is appropriate antibiotic treatment that is tailored to the pathogen (if any) that is actually causing the disease. However, one of the major problems facing clinicians is deciding which (if any) antibiotic therapy should be prescribed in the 12–48-hour period before the causative agent of the infection is identified. Furthermore, patients may die if they are prescribed incorrect antibiotic therapy or if no antibiotic therapy is given. On the other hand, the indiscriminate use of antibiotics, for example in treating viral and fungal infections that do not respond to antibiotics, may lead to the development of antimicrobial resistance, as well as causing unnecessary side-effects. A (rapid) method to help clinicians to tailor antimicrobial prescribing to individual patients would help reduce the development of antimicrobial resistance and unwanted side-effects associated with unnecessary treatment. The European Union-funded, 4-year TAILORED-Treatment (TTT) project established a broad-based strategy (not limited to a particular infection) that can be used to increase the effectiveness of antibiotic treatment, reduce potential side-effects of therapy, and help to limit the emergence of antimicrobial resistance in hospitalised children and adults. At the heart of the TTT project is a clinical study that involved hospitalised patients with respiratory tract and/or bloodstream infections, including both children and adults. State-of-the-art diagnostic techniques have been utilised to generate transcriptomic, proteomic, genomic, and microbiome data, which has been assembled into a single database. This database (HOPOIT) is being used to identify novel interactions that characterise both patients and their infections, in order to discover new biological markers of infection and to develop new computer tools that will enable clinicians to tailor antibiotic therapy in an appropriate and effective way to individual patients.
Project Context and Objectives:
The treatment of infectious disease using antibiotics has been one of the most important advances in modern healthcare, saving millions of lives since their discovery and widespread use. Despite their immense contribution to global healthcare, the CDC recently reported that ‘up to 50% of all the antibiotics prescribed for people are not needed or are not optimally effective as prescribed’. Antibiotic overuse typically stems from prescribing these drugs to treat nonbacterial diseases (mostly viral infections) for which they are ineffective. For example, in the USA alone, over 60 million annual cases of viral influenza are prescribed unnecessary antibiotic therapy. Antibiotic misuse has severe health and economic outcomes. Overprescription of antibiotics may cause preventable adverse events such as allergic reactions, intestinal yeast infection and antibiotic-associated diarrhea. These preventable adverse events may impact patient care and result in lengthy hospitalization. On the other hand, delayed or no antibiotic treatment in cases of bacterial disease is also common (24–40% of all bacterial infections) . While this may reduce the risk of antibiotic-related adverse events, such practices can lead to disease-related complications resulting in increased rates of morbidity and mortality.
One of the most alarming consequences of antibiotic overuse is the emergence and spread of multidrug-resistant bacteria. Resistance of microbial pathogens to antibiotics is increasing worldwide at an accelerating rate, with a concomitant increase in morbidity and mortality associated with infections caused by antibiotic-resistant pathogens. At least 2 million people are infected with antibiotic-resistant bacteria each year in the USA alone, and at least 23,000 people die as a direct result of these infections. In the EU, an estimated 400,000 patients present with resistant bacterial strains each year, of which 25,000 patients die. Consequently, the WHO has warned that therapeutic coverage will be insufficient within 10 years, putting the world at risk of entering a ‘post-antibiotic era’, in which antibiotics will no longer be effective against infectious diseases. The CDC considers this phenomenon ‘one of the world's most pressing health problems in the 21st century’.
Antibiotic overuse in hospitals and outpatient settings contributes significantly to the rising prevalence of antibiotic resistance. At the heart of this problem is the challenge of accurately distinguishing between bacterial infections (which warrant antibiotic therapy) and viral infections (for which antibiotic treatment is generally not required). This diagnostic gap is driven by the inability of current diagnostic tools to provide rapid and accurate information regarding the etiological basis of an infection.
Conventional diagnostic approaches depend on the cultivation of infectious agents and subsequent testing for antibiotic sensitivity and resistance. This process requires lengthy cultivation periods (days) and is not applicable for certain bacterial infections, or for most viral infections. Nucleic acid amplification-based tests (NAAT) for direct pathogen detection are showing considerable promise. Their advantages include high sensitivity and the simultaneous detection of multiple pathogens. Consequently, these tests are increasingly used in hospital and laboratory settings. However, NAAT protocols exhibit varying degrees of sensitivity and specificity when identifying specific pathogens and antibiotic resistance traits. In addition, NAAT diagnostic technologies usually require direct sampling of the pathogen. Such sampling is often not easily feasible if the infection site is not easily accessible (e.g. sinusitis, middle-ear infection and bronchitis) or the site of infection is unknown (e.g. fever of unknown origin). The described limitations of current diagnostic procedures leads physicians to either overprescribe (‘just-in-case’) or underprescribe (‘watchful waiting’) antibiotics, both of which can adversely impact patient care and health economics. Therefore, there is a clear need for novel solutions that will empower physicians to make early, evidence-based antibiotic treatment decisions, in order to improve patient care, reduce adverse events and limit the spread of antimicrobial resistance.
Meeting the challenges of antibiotic resistance requires adopting an integrative approach that expands the antimicrobial arsenal on the one hand and reduces antibiotic consumption on the other. New antibiotic drugs are desired, but will likely provide only a temporary solution in light of the inherent ability of microbes to develop new resistances by a variety of physiological mechanisms, including the exchange of transferable genetic elements, uptake of DNA by transformation and transduction and DNA mutation. Moreover, only a handful of new antibiotics are likely to reach the market in the near future due to the ‘broken pipeline’ associated with their development. Complementary to the development of new drugs is the need for a significant reduction in antibiotic consumption. This goal may be achieved by increasing public awareness of antibiotic misuse and its consequences, as well as by adopting new diagnostic approaches to bridge the current diagnostic gap.
New technologies are constantly emerging, providing the scientific and medical communities with powerful tools to improve the diagnosis of infectious diseases. Advances in the analysis of pathogens and the host immune response to infection in a broad and sensitive manner have led to a deeper understanding of complex host–pathogen–treatment interactions. Technologies such as next-generation sequencing provide snapshots of the patient's entire transcriptome in response to an infection or treatment, as well as enable the genetic analysis of the patient's natural microbiota (microbiome). These technologies also facilitate the search for novel genetic variation markers (single nucleotide polymorphisms [SNPs]) that may predispose individuals to specific infections and disease progression. Most importantly, advances in bioinformatics algorithms and ‘big-data’ analysis enable the integration of clinically relevant information such as host genetics, microbiota, response to treatment and data on the disease-causing agent. Mining this integrated data to generate treatment algorithms, coupled with the development of intuitive web-based interfaces, will be a major step forward in the management of infectious disease patients. Not least in the ongoing battle against antimicrobial resistance.
The EU recently funded the 4-year ‘TAILORED-Treatment’ research program, which focussed on the rapid and correct diagnosis of infectious diseases for guiding antibiotic treatment in patients presenting with respiratory tract infections and/or sepsis. The project combined state-of-the-art omics-based techniques and data with multivariate analysis methods and newly developed bioinformatics software with the goal of translating novel host-pathogen insights into treatment decision support algorithms. At the heart of the project was a large multicenter prospective clinical study that enrolled 1,222 adult and pediatric patients. Subsets of patient samples were investigated for host transcriptomic and proteomic response, nasal microbiota, SNP analysis and pathogen diagnostic proteomics. The data was assembled into a dedicated TAILORED-Treatment database (HOPOIT) , which is being investigated for significant associations between multiple biological and clinical parameters in order to determine the best combination of factors that determine and can distinguish between 'Bacterial-Infection', 'Viral-infection, 'Both Bacterial and Viral Infection' and 'No Infection' classifiers. Current analysis has identified factors that can be used to distinguish between the above classifiers with prediction accuracy of between 95% and 98%, dependent on the number of parameters included in the analysis.
Project Results:
A total of 1, 222 patients were recruited for the TAILORED-Treatment project, from the Netherlands and Israel with age distributions ranging from <1 year of age (16%) to >60 years of age (24%). An expert panel of clinicians were recruited to examine electronic Case Report Forms (eCRF) in order to determine a 'gold standard' diagnosis on which the TAILORED-Treatment classifiers of 'Bacterial-Infection', 'Viral-infection, 'Both Bacterial and Viral Infection' and 'No Infection' could be based. Using these criteria and only the eCRF data, the expert panel concluded that 75% of 522 children and 36% of 352 adults (who could be 'accurately' diagnosed by the expert panel) suffered from a viral infection.Virology studies showed that flu, RSV and rhinovirus were the most frequently detected viruses with higher proportions of viral infection in children than in adults. As expected, for 10% of the children no diagnosis was made. The inappropriate use of antibiotics was found in 41% of viral infections. SNP analysis was performed on 341 patients with 283 samples generating good quality GWAS data for statistical analysis.
The majority of SNPs in the TTT cohort were associated with patients of European descent. Analysis of SNPs associated with common infections (using a Candidate SNP Approach) identified a SNP with a frequency in the TAILORED-Treatment population, which deviated more than 5% from other European populations. Further SNPs located on a particular chromosome were also found to be interesting for further study.
MeMed applied biochemical and advanced high-throughput technologies to explore the host proteome and host transcriptome in search of novel biomarkers that distinguish between viral, bacterial, and fungal infections, both in children and adults. The clinical informative value of single biomarkers can be sensitive to inter-patient variability, including gender, age, time from symptom onset, clinical syndrome, pathogen species and co-morbidities. Therefore, protein- and RNA-based host response signatures were developed by computationally combining multiple biomarkers, which can identify different types of infections while overcoming the challenge of patient heterogeneity. The research initiated with a screening process that included bioinformatics analysis of publically available databases using our newly developed bioinformatics platform, augmented by manual curation of literature covering up to 20,000 potential proteins. This process resulted in a list of several dozens of protein candidates, some of which with an established role in the host immune response to infection, and others with no known direct link to the immune system. First, che clinical informative value of 70 potential protein biomarkers was evaluated using 80 clinical samples obtained from well characterized patients. Next, the levels of the most informative proteins were further measured in increasing number of blood samples (ranging between 500-1200 patients). The best performing single protein biomarker, identified in the screening phase, was TNF-related apoptosis-inducing ligand (TRAIL). TRAIL was induced in response to a wide range of viral infections and surprisingly, also exhibited significantly reduced levels in bacterial infected patients. Finally, statistical classification algorithms were used to identify protein signatures that can increase diagnostic performance using iterative cycles of training and validation . The signature with the highest precision included both viral- and bacterial-induced proteins: TRAIL, Interferon gamma-induced protein-10, and CRP. MeMed also developed RNA-Based immune host-response signatures for diagnosing the source of infection applying several stages: (i) RNA purification from clinical samples; (ii) quality assurance tests to verify integrity levels of purified RNA; (iii) high-throughput measurements of RNA expression levels to identify differentially expressed transcripts as potential new biomarkers; (iv) computational analysis of high-throughput measurements; (v) constructing multi-parametric immune signatures by implementing feature selection algorithms and supervised learning strategies. Comprehensive transcriptome analysis using Affymetrix HTA 2.0 arrays was performed on samples from 130 patients with well characterized infections. The process resulted in highly accurate RNA-Based immune host-response signatures that included both traditional and novel RNA transcripts.
More than 1,200 patient nose swab samples were processed and sequenced for microbiota profiling using Illumina sequencing technology. Five hundred-sixteen samples contained sufficient microbial biomass (> 1,000 16S rRNA gene copies/ul) and sequence read coverage (> 1,000 sequence reads/sample) to ensure that the 16S rRNA gene sequencing results obtained were valid and no bias was obtained from the laboratory environment or reagents used (which may be potential sources of contaminating DNA in the clinical setting when using low biomass samples such as nose swabs). At the same time, a novel microbiota platform (MYcrobiota) that facilitates the implementation of microbiota diagnostics into the clinical diagnostic laboratory by providing quantifiable bacterial identification for clinical samples was developed and evaluated. Hierarchical clustering methods identified that 9 different bacterial operational taxonomic units (OTUs) dominated the nasal samples. Preliminary investigations were also performed regarding the utility of Oxford Nanopore sequencing technology for microbiota sequencing using a subset of Illumina sequenced TAILORED-Treatment nose swab samples.
UGOT has identified mass spectrometric protein markers for: 1) species and strain identification and 2) antimicrobial resistance; in relevant bacterial respiratory pathogens (i.e. S. pneumoniae, H. influenzae, M. catarrhalis, S. aureus, etc.). The MS-proteomics methodology used was optimised for: 1) rapid pathogen identification; and 2) detection and analysis of expression of antimicrobial resistance factors (e.g. CTX-M/OXA/SHV/TEM, ampC, carbapenemases and porins). A database containing the proteomic markers of microbial pathogens of respiratory infections has been developed, based on an established database with more than 2,500 antibiotic resistance genes, representing all major resistance mechanisms. LC-MS/MS and LPI proteomics-based rapid pathogen detection has been optimized, with respect to sensitivity, specificity, etc., and applied directly to the analyses of TAILORED-Treatment clinical samples, i.e. without prior cultivation and isolation of microbial pathogens.
A dedicated TAILORED-Treatment database (HOPOIT) and graphical user interface (GUI) was developed, including a data-upload module. Multivariate analysis tools have been integrated into the HOPOIT database. . The HOPOIT system is made of 3 parts: 1) A central
relational database that stores and serves all the data produced in the project, 2) A web platform that includes data management, exploration, analysis and visualization functionalities and 3) A local application that manages all the internal processes, like file parse and dump, quality control, dataset creation or analysis threads management. Multivariate analysis tools have been added to HOPOIT, including three types of tools: 1) Variables selection tools, 2) Data mining algorithms and 3) Exploratory tools. The variable selection and data mining tools have been codified in C#, bundled in dlls (dynamic-link libraries) and tested individually. A web platform has been developed to integrate the previously developed database and the analysis tools. It includes user-friendly visualization tools to ease the use of the analysis tools and the interpretation of the results for physicians and data access control in order to protect the confidentiality of the data while enabling data sharing to the scientific and medical community. Current multivariate analysis of TAILORED-Treatment data within HOPOIT has determined that the best combination of factors can distinguish between 'Bacterial-Infection', 'Viral-infection, 'Both Bacterial and Viral Infection' and 'No Infection' classifiers with a prediction accuracy of between 95% and 98%, dependent on the number of parameters included in the analysis.
Potential Impact:
The treatment of infectious disease using antibiotics has been one of the most important advances in modern healthcare, saving millions of lives since their discovery and widespread use. Despite their immense contribution to global healthcare, the CDC recently reported that ‘up to 50% of all the antibiotics prescribed for people are not needed or are not optimally effective as prescribed’. Antibiotic overuse typically stems from prescribing these drugs to treat nonbacterial diseases (mostly viral infections) for which they are ineffective. For example, in the USA alone, over 60 million annual cases of viral influenza are prescribed unnecessary antibiotic therapy . Antibiotic misuse has severe health and negative economic outcomes, with antibiotic overprescribing being associated with preventable adverse events such as allergic reactions, intestinal yeast infection and antibiotic-associated diarrhea. These preventable adverse events may impact patient care and result in lengthy hospitalization. Importantly, one of the most alarming consequences of antibiotic overuse is the emergence and spread of multidrug-resistant bacteria. Resistance of microbial pathogens to antibiotics is increasing worldwide at an accelerating rate, with a concomitant increase in morbidity and mortality associated with infections caused by antibiotic-resistant pathogens. At least 2 million people are infected with antibiotic-resistant bacteria each year in the USA alone, and at least 23,000 people die as a direct result of these infections. In the EU, an estimated 400,000 patients present with resistant bacterial strains each year, of which 25,000 patients die. Consequently, the WHO has warned that therapeutic coverage will be insufficient within 10 years, putting the world at risk of entering a ‘post-antibiotic era’, in which antibiotics will no longer be effective against infectious diseases. The CDC considers this phenomenon ‘one of the world's most pressing health problems in the 21st century’.
The TAILORED-Treatment project has successfully obtained samples from more than 1200 patients (children and adults) suffering from respiratory infections and sepsis. Subset of these specimens have been investigated using state-of-the-art technologies in order to generate novel algorithms accurately and rapidly diagnose if a sick child or adult is suffering from a bacterial, viral, both or no viral or bacterial infection. The use of such an algorithm will allow clinicians to rapidly assess whether a patient requires antibiotics or not, thereby helping reduce antibiotic overprescribing (associated with adverse events for patients and an increasing antibiotic resistance) and consequently decreasing the 2 million people infected with antibiotic resistant bacteria each year in the USA and the 400,000 patients infected within the EU. Further, acceptance of TAILORED-Treatment algorithms will lead to reduced morbidity and mortality on a worldwide scale, as well as an overall reduction in health care budge costs.
The project has generated several peer-reviewed publications and has been the topic of poster and all presentations at a range of international scientific conferences. Further, exploitable foregrounds has been identified, what patents has been submitted and another possible patent submission is currently under review.
List of Websites:
www.tailored-treatment.eu