The changing face of pathogen discovery and surveillance

Abstract

The pace of pathogen discovery is increasing dramatically. This reflects not only factors that enable the appearance and globalization of new microbial infections but also improvements in methods for ascertainment. New molecular diagnostic platforms; investments in pathogen surveillance in wildlife, domestic animals and humans; and the advent of social media tools that mine the world wide web for clues to outbreaks of infectious disease are proving invaluable in early recognition of threats to public health. Additionally, models of microbial pathogenesis are becoming more complex, providing insights into the mechanisms by which microorganisms can contribute to chronic illnesses like cancer, peptic ulcer disease and mental illness. Here we review the contributions of each of these elements to infectious disease emergence and strategies for addressing the challenges of pathogen surveillance and discovery.

Introduction

When the H1N1 influenza virus struck in 1918, little was known about how infectious diseases emerge or their routes of transmission. Indeed, even the identity of the causative agent as a virus (i.e. capable of passing through a filter) rather than the bacterium Hemophilus influenza championed by some leading microbiologists, was in dispute until late in the course of the pandemic¹. The 1918 virus had a case-fatality rate estimated at 10-20%, spread to 6 continents, infected ~500 million people, and killed approximately 3% of the world's population ^2,3. The SARS coronavirus pandemic of 2003, the first pandemic of the 21^st century, had case-fatality rate near 10% but infected far fewer people (8,096)⁴. Transmissibility as estimated by the R0 (the number of cases generated through contact with an infected individual) is similar for influenza (2-3)⁵ and SARS (2-5)^6,7. However, the global response to SARS was facilitated by advances in epidemiology and microbiology that enabled rapid containment and identification of the causative agent as a novel coronavirus. The discovery process is faster still today. Whereas two large teams invested several days in characterizing SARS coronavirus genetic material amplified in tissue culture by using classical dideoxy sequencing techniques^8,9, high throughput sequencing platforms employed in more recent outbreaks such as LuJo in South Africa have allowed identification of novel agents in clinical materials in 48 to 72 hours¹⁰.

As a consequence of globalization of travel and trade, infectious agents are expanding in geographic range and appearing in new contexts. Thus, clinicians and public health officials must be prepared to detect and respond to the unexpected. The ongoing development of new antimicrobial drugs, therapeutic antibodies, vaccines and probiotics means that early and accurate diagnosis can have profound implications for medical management and public health. This is particularly true for viral infections where, until recently, opportunities for effective intervention were limited to HIV, hepatitis C virus and herpesvirus infections. Surveillance and discovery efforts are bearing fruit in chronic disorders and in studies of normal physiology as well as in the investigation of acute diseases like pneumonia, diarrhea, meningitis/encephalitis and hemorrhagic fevers. The links between Helicobacter pylori and peptic ulcer¹¹, human papillomavirus and cervical cancer¹² and polyomaviruses and Merkel cell carcinoma¹³ are prominent examples in which microorganisms contribute to the pathogenesis of disorders that were idiopathic only a few years ago. Insights into the role of the human microbiome in nutrition, allergies and autoimmunity have led to implementation of the same surveillance and discovery platforms employed to investigate classic infectious diseases ^14,15. Finally, although there have been no recent examples of bioterrorism, the risk has only increased with political instability and the accessibility of synthetic genomics that enable the creation or recreation of virulent pathogens.

In this essay, I will discuss factors that contribute to the emergence (and reemergence) of infectious diseases, the evolution of strategies and tools for pathogen surveillance and discovery, and future prospects for the field. To guide the reader we provide a time line of events and innovations described in the text (See Timeline)

Factors in microbial emergence

Globalization of travel and trade

Travel and trade are increasingly global. For instance, the number of international airline flights has nearly doubled over the past 15 years from just under 500,000 in 1996 to close to 850,000 in 2011 (U.S. Bureau of Transportation Statistics: www.bts.gov/). John F. Kennedy airport for example, one of two international airports in the greater New York metropolitan area, USA, hosts nonstop flights to 100 international destinations, and serves nearly 12 million international customers annually (http://www.bts.gov/press_releases/2012/bts017_12/html/bts017_12.html). Similar data apply worldwide for airports serving large urban centers. This means that an infected individual or mosquito can rapidly travel around the world in less than 24 hours, making it not surprising that air travel has been implicated in the global dissemination of HIV, SARS coronavirus, West Nile virus, chikungunya virus, influenza virus and M. tuberculosis^16-18.

Indeed it is perhaps more remarkable that so few outbreaks of infectious disease have been attributed to air travel, especially given that the transportation of plants and animals has continued to grow dramatically with the development of global agribusinesses and urbanization. Whereas the global population and food production have increased at similar rates (74% and 100%) since 1975, the international food trade has burgeoned by more than 200% (United Nations: http://esa.un.org/unpd/wpp/Excel-Data/population.htm, http://faostat.fao.org/). A hundred years ago, most fresh food was produced and consumed in a radius of a few kilometres, it is now not unusual for individuals to consume plants and animals harvested thousands of kilometres away¹⁹.

Agricultural practices

Contamination of meat has impacted international trade of livestock on a number of occasions²⁰, with examples including agents that can threaten humans such as prions, influenza viruses and Rift Valley fever virus, or agents that threaten the livestock themselves such as foot and mouth disease virus and Schmallenberg virus²¹. However, bacteria, viruses and parasites, particularly those present in faeces, can also contaminate fruits and vegetables to cause disease in humans and other animals, resulting in costly food recalls and affecting consumer demand^22,23. High-density farming of livestock, poultry and fish is frequently associated with the use of antibiotics as growth promoters that results in the emergence of antibiotic-resistant bacteria^24,25.

Centralization of food production and processing, particularly of ground meat or raw fruits and vegetables, has resulted in outbreaks of infectious diseases that may be distributed over large geographic areas. Furthermore, illegal trafficking in wildlife with pets or food, estimated to exceed US $10 Billion and US $15 Billion per annum in sales, respectively, is difficult to monitor (U.S. Department of State, Wildlife Trafficking, http://www.state.gov/e/oes/env/wlt/index.htm). Nonetheless, there is evidence that these activities are associated with the introduction of microorganisms into new environs that may pose threats to public health. Analysis of bushmeat from bats, rodents and primates that was confiscated in major ports has revealed evidence of foamy viruses, herpesviruses and pathogenic bacteria²⁶. Imported pets have been linked to outbreaks of human infection with poxviruses and salmonella, as well other pathogens²⁶.

Although plant pathogens do not infect humans or other animals, infection of food crops can have dire economic consequences. Recently, with greater appreciation of the importance of pollinators in food production and the recognition of colony collapse disorder, increased attention has been directed towards the potential for emerging infections of honeybees (Apis mellifera) and other pollinators, particularly with regards to viruses, fungi and external parasites²⁷. Mariculture (ocean aquaculture) is also at risk of emerging infectious diseases, as demonstrated by recent reports of novel viruses in farmed salmon^28-30. Attention has also increasingly focused on the role of land use dynamics in infectious disease emergence³¹. Deforestation and the expansion of agriculture and the extractive industries, particularly in tropical regions with high wildlife biodiversity, has led directly or indirectly to the emergence of HIV/AIDS, Nipah virus and filoviruses^32,33.

Climate change and mass migration

Global warming is already extending the geographic range of mosquitos and ticks that harbor and transmit Plasmodium sp. and arboviruses, resulting in outbreaks of malaria, dengue and yellow fever in new locations³⁴. Recent examples in North America include the appearance of dengue fever in Florida in 2010 ^35,36 and a surge in cases of West Nile encephalitis in Texas in 2012³⁷. Mass migration (due to war, natural disaster, poverty and desertification) can lead to increases in the population density, not only of humans, but also of disease vectors like rodents and ectoparasites that carry pathogenic viruses and bacteria. These factors, along with poor sanitation, malnutrition, lack of access to vaccines, and exposure to contaminated food and water create a perfect storm for the emergence and transmission of infectious diseases^38,39.

Laboratory analyses

Culture, once the mainstay for detection of organisms in the laboratory, is still emphasized in some public health organizations and remains vital to clinical microbiology, chiefly as a tool for testing the utility of drugs. However, genetic methods have moved to the forefront in microbial surveillance and diagnostics^40-44. The foundation for most of these methods is the polymerase chain reaction (PCR) developed in 1983 by Kary Mullis (See Timeline). PCR requires minimal equipment and operator training, can be completed in minutes rather than days, is less expensive, and has been adapted to portable instruments that can be used in the field in developing countries or near a patient's bedside. Furthermore, like other genetic methods, PCR may succeed in detecting an organism in instances where fastidious requirements confound cultivation. Most PCR assays approved for clinical applications test for the presence of a single type of bacterium or virus. Such assays, described as singleplex, are used to screen for any evidence of infection (e.g., hepatitis B virus in blood products used for transfusion) or to quantitate microorganism levels when assessing response to therapy (e.g. HIV burden in serum or plasma of subjects receiving anti-retroviral medication).

Multiplex Assays

Multiplex PCR, initially implemented for screening human genetic polymorphisms has been extended to microbiology wherein assays have been developed that allow simultaneous screening for the presence of up to 30 different microorganisms⁴⁰. Such assays are particularly important for differential diagnosis in medicine where many distinct infectious agents may be implicated in diseases like pneumonia, diarrhea or meningitis/encephalitis. Thus, although multiplex PCR assays are rare outside of research and public health laboratories, there is reason to believe that they will ultimately gain wider acceptance.

An even broader platform is DNA microarrays where millions of genetic probes are bound to glass or silicon wafers and tested for their capacity to bind complementary sequences in clinical and environmental samples. Binding is typically detected through the measurement of fluorescent molecules attached to nucleic acid amplified from sample extracts. Such microarrays have the potential to survey the entire known microbial world; however, implementation has been hampered by low sensitivity and cumbersome processing. Recent prototypes have been developed that suggest it may be possible to circumvent these obstacles through use of portable devices that employ nanofluidics and electronic nanocircuitry^45,46.

Genomics and Metagenomics

The most disruptive advances in microbial surveillance have been achieved in DNA sequencing. The emergence of high throughput sequencing over the past decade has enabled the discovery of new microorganisms, rapid resolution of the causes of outbreaks of infectious diseases, and the development of metagenomics, a field wherein investigators inventory the complex microbial communities found in humans, domesticated animals, wildlife, plants and various environments. Although initially confined to specialized laboratories due to high costs for instruments and supplies costs along with requirements for sophisticated personnel, technical improvements have increased access for a broader segment of the research community. One index to the evolution of sequencing technology is the per-base cost which decreased from $5,000 per megabase in 2001 with classical dideoxy methods, to $15 per megabase in 2008 with pyrosequencing (http://www.genome.gov/sequencingcosts), to $0.5 per megabase in 2012 with the Illumina platform⁴⁷. Another is the time required for obtaining sequence data. Whereas the SARS coronavirus genome was sequenced over the course of a week by a large team in 2003⁸ a single investigator could sequence that same genome in a few hours in 2012. Recent examples of the power of advancements in methods for genomic sequencing include reports on the evolution of influenza⁴⁸, hepatitis C⁴⁹ and human immunodeficiency viruses⁵⁰, the human origin of livestock associated methicillin-resistant Staphyloccocus aureus ⁵¹and the spread of antibiotic resistant Klebsiella pneumoniae between and in health care institutions⁵².

Metagenomic analyses^40,53,54 have revealed dynamic relationships between microorganisms and hosts that influence normal physiological processes like digestion^55,56 and immune response⁵⁷, may be factors in the pathogenesis of autoimmune diseases⁵⁷ and cancer ⁵⁴, and likely contribute to global climate regulation through effects on marine plankton⁵⁸. The challenge now is not in obtaining sequence data but in analyzing it. Millions of sequence reads must be assembled into continuous strings of genetic information, and identified as originating in a microorganism or a host by using algorithms that compare the sequences obtained for similarity to those already catalogued in existing databases. Few investigators now have the in-house processing power and expertise required for these types of analyses; however, access to large computer clusters can be achieved through cloud computing and high-throughput sequencing software that is rapidly becoming more user friendly.

Microbial surveillance and forecasting

Passive and active surveillance

Surveillance is broadly divided into the categories of passive and active. Whereas passive surveillance employs data that already exist or are collected routinely, active surveillance implies a new investment in and/or processes for microorganism collection and analysis. A classic example of passive surveillance is the concept of reportable diseases. Most regional and national public health authorities maintain lists of infectious diseases and laboratory tests indicative of infection that must be reported (CDC National Notifiable Diseases Surveillance System, http://wwwn.cdc.gov/nndss/script/downloads.aspx) These include infections characterized by human-to-human transmission, such as sexually transmissible diseases (e.g., syphilis or gonorrhea) and vaccine-preventable diseases (e.g., measles), as well as those where detection indicates the presence in the environment of an infectious agent that poses a substantial threat to public health, such as hemorrhagic fever viruses or highly pathogenic bacteria like Yersinia pestis, the agent of the bubonic plague. On the non-human side, agricultural authorities monitor infections like foot and mouth disease, which have important economic implications. Passive surveillance is informative and inexpensive, but may underestimate the true frequency of an agent or a disease. Furthermore, by definition it cannot detect risk prior to onset of symptoms.

Surveillance Using Social Media

Largely due to the influence of Joshua Lederberg⁵⁹, a pioneer in microbial genetics and the use of computers for communication as well as data analysis, internet based infectious disease surveillance is well established. ProMED-mail (Programme for Monitoring Emerging Infectious Diseases), created in 1994, provides continuous free email updates on new or evolving outbreaks and epidemics⁶⁰. Submissions from a grassroots network of readers are curated by a panel of experts who post submissions with commentary in 5 languages to a listserve comprising more than 60,000 subscribers in 185 different countries. GPHIN (Global Public Health Intelligence Network)⁶¹ scans news services across the globe in 9 languages for information concerning outbreaks. Unlike ProMED-mail, GPHIN is a fee-based, private subscription, and does not systematically validate its posts, although the WHO began providing verification services when GPHIN was added to the Global Outbreak Alert Response Network (GOARN) in 2000.

HealthMap⁶² is a hybrid of the passive and active surveillance strategies employed respectively by ProMED-mail and GPHIN. It integrates reports from news media, ProMED-mail and official documents into a user-friendly map that displays real-time updates of disease emergence. HealthMap also allows for public submission of geo-referenced observations or “crowd sourcing” of apparent disease occurrence via its website or a number of smart phone based apps such as Outbreaks Near Me. Google Flu Trends (http://www.google.org/flutrends/) is similar but aggregates search data to estimate global influenza activity.

InSTEDD (Innovative Strategies to Emergencies, Diseases and Disasters), proposed during Larry Brilliant's talk at the 2006 TED Conference “One Wish to Change the World”, develops open source tools to improve global information collection and exchange (http://www.instedd.org/). What is anticipated but not yet achieved are systems that aggregate medical service utilization data, prescription and over-the-counter drug purchases, and other chatter that could promote situational awareness and focus epidemiological investigation. Zoonoses, infections that originate in wildlife or domestic animals, account for more than 70% of emerging infectious diseases⁶³; thus, to be proactive, a substantive surveillance system⁶⁴ for humans must also include nonhuman animal surveillance.

Modeling infectious disease emergence

Quantitative analyses of emerging and re-emerging infectious diseases have enabled the identification of geographic hotspots of infectious disease emergence and of underlying drivers (primarily human activities) that facilitate the process^31,65-68. The recent development of high quality, global scale data sets of human demographics, agricultural production, land-use change, travel and trade patterns, climate, and wildlife distribution has greatly improved the resolution and specificity of predictive modeling³⁸. This has resulted in substantive advances in risk analyses from the original hot spots maps³⁸ (Fig. 1). These risk algorithms are being used to focus passive and active surveillance programmes on sites, populations, professions, and species of domestic animals and wildlife where there is an increased probability of known or novel high threat pathogen emergence.

Figure 1. Hot spots of outbreaks for recently emerging and re-emerging infectious diseases.

Green: low density of outbreaks; red: high density of outbreaks. Zoonotic infections are highlighted in red text. Ref. 112 and unpublished data from Peter Daszak (EcoHealth Alliance)[Will ask author for permissions for this data and for the use of the maps].

An example is the United States Agency for International Development's Emerging Pandemic Threats programme (http://avianflu.aed.org/eptprogram/) wherein hot spots models are used to prioritize regions and countries for investments in surveillance, laboratory diagnostics, outbreak response, and collaboration in 20 countries in Africa, South America, and Asia. Algorithms used to build hot spots models are continuously tested and modified in light of experimental data derived from human, domestic animal and wildlife sample analysis.

Proof of causation and mechanisms of pathogenesis

Finding footprints of a microorganism is only the first step in establishing a causative role in disease. In some instances the connection is immediately apparent because precedent supports plausibility, for example, finding a new type of ebolavirus in an individual with a hemorrhagic fever, or a new stain of Vibrio cholera in a diarrhea outbreak. However, in others the link is more tenuous. Host factors can have a profound impact on susceptibility to infection and consequences thereof. Morbidity and mortality can be high with otherwise innocuous agents in individuals with immunological deficits, whether due to genetic mutations, age, malnutrition, infection (e.g. HIV/AIDS) or complications of cancer treatment or transplantation.

Mechanisms of disease can vary (Box 1). Microorganisms may cause damage at the site of infection as a direct result of replication, or as an indirect effect of innate or adaptive immune responses to microbial gene products. They may also induce neoplasia through interference with cell cycle controls. Although not yet confirmed in human disease, work in animal models indicates that viruses can reduce production of hormones or neurotransmitters vital to normal physiology without apparent cell or organ damage⁶⁹. Linkage to infection is facilitated in each of the foregoing examples because the microorganism, its nucleic acid and/or protein are found at the site of pathology. More difficult to recognize are instances where expression of microbial toxins have remote effects, or infection induces immune responses to the microorganism that break tolerance to self. Clostridium bacteria, for example, can infect the skin or gastrointestinal tract and produce toxins that act on the nervous system to cause spasm (C. tetani)⁷⁰ or flaccid paralysis (C. botulinum)⁷¹. Streptococcal infection of the skin or oropharynx can result in autoimmunity, culminating in cardiac damage (rheumatic heart disease) and brain dysfunction (Sydenham chorea)⁷².

The best-established criteria for proof of causation were formulated by Loeffler and Koch in the 1880s. Popularly known as Koch's postulates,⁷³ they require that an agent be present in every case of the disease, specific for the disease and sufficient to reproduce the disease after culture and inoculation into a naïve host. Rivers⁷⁴ modified them by acknowledging the significance of neutralizing antibodies to an agent as evidence of infection. Fredericks and Relman⁷⁵ noted that pathogens may be recognized by molecular methods before they can be cultured, and allowed as evidence the presence of microbial sequences as well as of infectious microorganisms. Thus, although Koch's postulates remain the gold standard, they need not be fulfilled to implicate an agent. Indeed, a focus on Koch's postulates may impede the successful discovery and response to emerging pathogens and models for infectious disease.

Through the discovery and characterization of nearly 500 viruses, we have developed a three level scoring system for establishing levels of confidence in strength of association from possible to definitive (Box 2). Poor design or execution of pathogen discovery projects can lead to spurious links between infectious agents and diseases that result in inappropriate and potentially dangerous treatments or rejection of health promoting interventions like vaccines⁴⁰. The effort required to break these links can be greater than that invested in building them, particularly in disorders where the prognosis is grim and/or treatment options are limited. Based on experience with amyotrophic lateral sclerosis (enteroviruses), mental illness (bornaviruses), autism (MMR vaccine) and myalgic encephalomyelitis/chronic fatigue syndrome (XMRV and pMLV), we have developed a strategy for trying and acquitting microorganisms that addresses social as well as scientific considerations (Box 3).

Progress in Microbial Detection: Field Cases

Over the last three decades, innovations in genetic and information technologies have enhanced and expedited the rate of detection and solution of outbreaks of infectious disease. The following examples illustrate this progress in methods for acquiring public health intelligence.

In 1976, the Centers for Disease Controls was alerted that 11 war veterans had died from pneumonia after returning from the US Bicentennial Convention of the American Legion in Philadelphia ⁷⁶. A case definition was established^76,77, and Pennsylvania health officials were notified of a potential statewide epidemic. Public-health personnel searched hospitals, news reports and obituaries, while a telephone hotline was established to accept tips from the general public. Ultimately, 221 cases were identified. All were individuals who had visited the convention hotel lobby or walked along the adjacent street. Initial efforts to culture an infectious agent failed^76,78. Histological staining of lungs of individuals with Legionnaire's disease revealed inflammation but no microorganisms.

The breakthrough came when Joseph McDade, a rickettsia expert at the CDC, recognized liver disease in some victims and in guinea pigs inoculated with extracts from patients. He inoculated embryonated chicken eggs with livers extracts from guinea pigs, and then reproduced the disease by inoculating additional guinea pigs with the extracts from the embryonic chicken eggs. This enrichment through passage in these model systems resulted in the discovery of a novel fastidious gram-negative bacterium, Legionella pneumophila.

The 4 months that elapsed between the onset of the outbreak in Philadelphia and the identification of the causative agent would not be required today. Whereas almost 2 weeks passed before the CDC was notified in 1976, alerts are now distributed at near real-time with services like ProMED-mail and HealthMap. Additionally, access to an electronic registry of the convention's guests would obviate the need for a statewide, grassroots search for cases. Culture-independent methods of pathogen discovery would also enhance response time.

Recent improvements in surveillance and microbial forensic science were illustrated by the 2011 outbreak in Europe of a Shiga-toxin producing Escherichia coli O104:H4 illustrates ^79,80. In May, Germany's national public-health agency—the Robert Koch Institute—sent representatives to investigate a cluster of hemolytic–uremic syndrome associated with bloody diarrhea in Hamburg. By the time the outbreak resolved, more than 3,000 people were infected and 40 died. Economic losses were substantive, particularly in Spain, where early inaccurate links to cucumbers from Spain led to a import ban on produce^81,82.

The initial clues to the identity of the casual agent were obtained by using PCR^83,84. Genomic characterization was rapidly achieved using high-throughput sequencers⁸⁵. Within 3 days of receipt of a clinical sample, sequence data was released into the public domain for global, crowd-sourced bioinformatics analysis⁸⁶. Genome assembly was completed in the next 24 hours, which enabled the development of specific diagnostic tests and insights into pathogenesis and phylogenetic origin of the bacterium. Integration of laboratory findings with patient surveys ultimately led to implication of bean sprouts from a single farm in Lower Saxony⁷⁹.

Another comparison involves the discovery of Borna disease virus (BDV)⁸⁷ and the related but distinct avian bornavirus. Borna disease, named for a town in Saxony where a characteristic fatal meningoencephalitis was described in horses, has been known in the veterinary literature since the 1700s^88-90. The transmissibility of disease was first demonstrated in 1920s; however, nearly 60 years were required to establish methods for viral culture and another 10 years were needed for classification as a novel non-segmented, negative-strand RNA virus⁹¹. The capacity to culture the virus led to serological assays that provided evidence of a connection to human neuropsychiatric diseases in 1983⁹².

Intrigued by the potential importance of BDV in human disease and challenged by the failure of efforts to characterize the virus by electron microscopy or to isolate BDV nucleic acids, I initiated a subtractive cloning project that culminated in identification of the first BDV sequences in 1990⁹³. An additional 4 years were required to determine its genomic organization^94,95. With specific cloned reagents in hand, I reasoned that it would be straightforward to determine whether BDV was indeed a human pathogen; however, the issue lingered until 2011 when blinded multicenter analyses, molecular and serologic, ultimately revealed no evidence of human infection ⁹⁶.

By contrast, avian bornavirus (ABV), the causative agent of proventricular dilatation disease (PDD), a wasting syndrome in psittacine birds (parrots), was identified in only a few days.^97,98 The breakthrough was enabled by access to genome databases and culture-independent methods for pathogen discovery, including viral microarrays and high-throughput sequencing. Subsequently, ABV PCR assays and serology have allowed for the investigation and containment of outbreaks in aviaries⁹⁹.

Future prospects

Although we will continue to see instances where classical approaches to microorganism hunting like culture and pursuit of Koch's postulates will succeed, pathogen discovery has evolved from a whodunit exercise performed by solitary investigators to team efforts of microbiologists, cellular and systems biologists, geographers, mathematicians and other specialists. Models of disease have expanded from simple one-to-one relationships between organ damage and the presence of a single agent therein to consider more complex mechanisms that may enable recognition of links between microorganisms and mental illness, obesity, vascular disease, cancer and autoimmunity.

Growth in international travel and trade has led to the globalization of infectious diseases. It has also fostered a new appreciation of the relationship between land use, particularly in the developing world, and the appearance of zoonoses. Globalization of risk across national and species boundaries has promoted the development of international health regulations¹⁰⁰ that emphasize technology transfer and data sharing, as well as programmes that proactively survey, not only humans, but the entire animal kingdom for insights into potential threats to public health and economic welfare.

The integration of human and animal medicine, the advent of tools for rapid and efficient molecular characterization of microorganisms and hosts, and emphasis on the use of social media to promote early detection of risk promises the potential for development of a truly global immune system.

Box 1. Mechanisms of microbial pathogenesis.

• Direct damage at site of replication due to cell lysis, apoptosis or autophagy

• Indirect damage at site of replication due to expression of proteins that serve as targets for humoral or cell-mediated immune responses

• Elaboration of toxins or other products that have deleterious local or systemic effects

• Induction of expression of host cytokines and chemokines that have deleterious local or systemic effects

• Abrogation of tolerance for self, resulting in autoimmunity

• Immunosuppression resulting in opportunistic infection

• Neoplasia

• Disturbance of differentiated cell functions

• Disruption of embryogenesis

Box 2. Levels of Certainty in Pathogen Discovery.

Level 1. Possible causative relationship

The initial clue in pathogen discovery is evidence of exposure to a microorganism in one or more individuals with disease. This evidence may be the presence of a microorganism in blood, other body fluids, faeces or tissues that can be grown on media or in cultured cells or animals. It may also be a nucleic acid or protein component of a microorganism, a specific adaptive immune response to a microorganism, or visualization by light, immuno- or electron microscopy. Keys features of level 1 include:

• Microorganism is isolated and cultured

• Microbial nucleic acids are detected by polymerase chain reaction, DNA microarray or sequencing

• Microbial proteins are detected using immunological methods or mass spectroscopy

• Antibodies to a microorganism are detected using immunological methods

• Morphological evidence of the presence of a microorganism is found by microscopy or electron microscopy

Level 2. Probable causal relationship

More confidence in the clinical significance of the association is achieved when a causal relationship is biologically plausible. Evidence of biological plausibility may include the presence of microbial nucleic acid, protein or microorganism-specific antibody in or adjacent to cells showing signs of disease, or precedent for a similar disease due to a similar agent in either the same or a similar host. The strength of the association is increased when the concentration of the microorganism (or nucleic acid, protein or antibody) is high, the antibody response indicates recent exposure (IgM and/or a recent increase in IgG titre), and there is evidence of infection in other individuals, all of whom have disease. However, a microorganism can be implicated in disease without a robust immune response, particularly in chronic infections.

Different microorganisms can cause similar diseases. Although clusters of disease are the ideal proving ground, many opportunities for discovery involve only one or a few cases; indeed, clusters may not be appreciated as such until details of common exposure (e.g., through travel, food, water, intermediary hosts) become apparent. Keys features of level 2 include:

• Microbial burden (levels of an infectious bacterium, virus, fungus, parasite or associated nucleic acids or protein) is high

• Antibody response is consistent with recent exposure (presence of IgM or an increased IgG titre)

• Evidence of infection is found in more than one individual with disease

• Location of a microorganism, its nucleic acid or proteins, or immune response to the microorganism is correlated with disease

• Similar diseases are known to occur with exposure to similar microorganisms in the same or other hosts

Level 3. Confirmed causal relationship

Proof of causation can be achieved through fulfillment of Koch's postulates or by mitigation or prevention of disease through use of a microorganism-specific drugs, antibodies or vaccines. Although not formally required we insist on replication of results by independent investigators. Keys features of level 3 include:

• Koch's postulates are met

• Drugs mitigate or prevent disease and reduce levels of the microorganism, its nucleic acid or proteins, or the immune response to the microorganism

• Vaccines mitigate or prevent disease

Box 3. The road to pathogen de-discovery.

Step 1. Initial Finding

Report links disease or syndrome to an infectious agent, toxin or other factor.

Step 2. Failure to reproduce

Independent efforts by other investigators fail to confirm the statistically significant association described in the initial report. Members of the scientific community begin to question validity of the association but cannot exclude the possibility that the apparent difference in results may reflect variability in the samples or assays employed.

Step 3. Plausible doubt

A biologically plausible explanation is developed that accounts for findings in the initial report that may have led to misinterpretation. Mainstream scientific community rejects validity of the association. However, some investigators, patients and advocates might continue to believe the merits of the initial report, necessitating Step 4.

Step 4. Comprehensive De-discovery

A well-powered, blinded analysis of samples from well-characterized subjects—that includes the investigators responsible for the initial report and uses a strategy approved in advance by the investigators, representatives of patients and patient advocacy groups—fails to replicate the findings of the initial report.