Devices that can quickly detect any bacteria, virus or fungus are currently under development. The networking of the devices will allow the health authorities to save lives through earlier identification of outbreaks of disease.
When I once wandered through an old cemetery near Philadelphia, I looked at the dates of birth and death engraved on the tombstones. I remembered that until the early 20th century, most people died before their 50th birthday. The main causes of death were infectious diseases such as smallpox, influenza and pneumonia.
Today, it is rare for infectious diseases to cause death in developed countries where improvements in sanitation, nutrition and vaccinations, as well as the use of antibiotics have virtually eliminated premature death from this type of disease. However, we may return to a period of death at a young age from infectious diseases because many microorganisms develop resistance to the available drugs, and because the pharmaceutical industry does not develop enough substitutes.
Excessive use of antibiotics is one of the most important causes of this problem, and it occurs for understandable reasons. In most cases, the diagnostic tools accepted today are unable to determine which of a wide variety of bacteria, the only organisms sensitive to antibiotics, is the cause of the disease. For the most part, old-fashioned methods that take several days to complete, which involve growing the bacteria in culture, are required to identify certain bacterial strains. The delay in treatment can be fatal, so doctors seek to encompass all possibilities by giving broad-spectrum and powerful antibiotic drugs capable of killing many types of bacteria. However, sometimes the drug destroys the sensitive bacteria and leaves the bacteria resistant to this particular drug. In this situation, these bacteria, which are resistant to antibiotics, multiply without competition from the bacteria that have been eliminated and they can spread and infect other people until they find the right conditions to allow them to make another person sick. Giving broad-spectrum antibiotics helps keep many patients healthy today, but also encourages the development of more resistant bacteria tomorrow.
There may be solutions to this paradox. Scientists are developing new molecular sensors that will allow doctors to quickly determine whether a patient is suffering from a bacterial or other infection, and which strain of bacteria is responsible for the disease. A key time-saving feature of these devices will be the immediate diagnosis of almost all pathogens at the same time, instead of examining individual microorganisms one by one. Moreover, doctors who suspect that it is a bacterial infection will not have to guess which strain is causing the disease. My research at Ibis Biosciences, now part of Abbott, provides the basis for one of these devices. Other bioengineers are developing similar products at other companies.
These devices for rapid diagnosis will reach hospitals and clinics in the coming years. However, with a little thought and planning, we can considerably increase their impact by connecting them together into a national or even global network of devices, which will provide comprehensive real-time data on the outbreak of new diseases, on food-borne diseases, on global epidemics, and possibly even on terrorist attacks biological.
It's time for an upgrade
The prevailing way to diagnose infectious diseases is based on culture methods originally developed by Louis Pasteur more than 150 years ago. Doctors take a tissue sample from the patients, for example blood, mucus or urine, and transfer it to a bottle containing a rich culture liquid or to a plate of agar, an extract of gelatin-like algae, which allow the pathogens to multiply. A day or two later, the bacteria multiply in an amount that allows identification. To test sensitivity to different drugs, the lab technicians can check if the cultures die when they are grown in the presence of different substances and how quickly this happens. But even if such an approach is faster, it is not ideal for deciding on the appropriate treatment, because it is difficult to grow many pathogens, which need a special culture liquid or a special growing environment. Sometimes it is not possible to multiply bacteria from certain patients at all because they were treated with antibiotics even before the sample was taken.
I became interested in diagnosing and monitoring infectious diseases while working at the US Defense Advanced Research Projects Agency (DARPA), researching new approaches to antibiotic discovery. Our goal was to scan thousands of chemical compounds and discover the compounds that kill many types of bacteria that have in common a certain sequence of RNA, a molecule essential to the functioning of all living things.
My colleagues and I used a device called a mass spectrometer to determine if the potential drugs bound to the bacterial RNA. A mass spectrometer actually determines the mass of molecules with great precision, and can be compared to scales. Since we knew the mass of the bacterial RNA, we could deduce the mass of each compound that bound to it, the same way we weigh a dog when we hold it in our hands while weighing it and then subtract our weight. The weight of the compound actually provided us with its identity, because each compound has a unique mass.
We soon realized that this technology could allow us to distinguish between bacteria, viruses, fungi and parasites, by weighing part of the RNA or part of the DNA of those creatures. DNA is a molecule very similar to RNA. Each strand of these molecules consists of subunits, called nucleotides, which appear a large number of times in each molecule. Nucleotides are marked with the first letter of the nitrogen-containing part of the molecule: A (adenine), C (cytosine), G (guanine) and U (uracil) in the case of RNA or T (thymine) in the case of DNA. Almost by definition, certain parts of these nucleic acids will be unique to different pathogens. Since the molecular weights of the different nucleotides (A, T, C, G and U) are very different from each other, we can determine how many nucleotides of a certain type are present in a given strand only by the result obtained in the mass spectrometer. For example, each DNA strand weighing 38,765.05 daltons (a dalton is the standard unit of measurement for atoms), as determined by a mass spectrometer, must contain 43 adenine units, 28 guanine units, 19 cytosine units, and 35 thymine units. This combination is the only one that gives exactly this result without a residue of nucleotide parts that are not found in nature. This information, in turn, indicates the strain of the microorganism being tested.
This method is similar to the method by which you can calculate the number of coins in a jar that only has US quarter coins (each weighing 5.670 grams when new) and 5 cents (nickels, each weighing 5.000 grams). If the total weight of the coins is 64.69 grams, the jar must contain 7 quarters and five nickels (64.69=5.67q+5n where q and n can only be positive integers or zero). Any other number of 25-cent coins means parts of nickels.
The process of identifying pathogens depends on the ability to distinguish, in a sample, between the RNA or DNA of the invader and that of the patient. Usually, the amount of foreign material is too small to make meaningful measurements, unless additional copies of DNA or RNA are prepared. Instead of waiting for more bacteria to grow in the culture, we use a method known as PCR (polymerase chain reaction of the enzyme that replicates nucleic acids) to make copies, or "amplify", the amount of DNA or RNA found in the sample taken from the patient. PCR has long been used to detect pathogens, but the method has so far been limited to detecting one or a few pathogens at a time. My colleagues and I decided to use PCR in combination with a mass spectrometer, in a way that would allow the identification of broad groups of organisms at once.
The key is to limit the amount of nucleic acids we need to produce to get significant results. We achieved this goal through a particularly careful selection of the DNA or RNA segments that we replicate. We choose "conserved" segments, that is, segments that contain sequences of letters that appear in all types of organisms from a certain large group, such as, for example, all organisms that are stained with what is known as Gram staining, but we make sure that these conserved segments are adjacent to regions unique to a certain strain, such as Staphylococcus aureus. Focusing on a large number of such carefully selected sequences allows us to identify precise categories and subcategories of organisms without unnecessarily prolonging the process. And so, after extracting the RNA or DNA from the microorganism, we identify the desired segments to continue the process by adding primers (primers, short segments of DNA that mimic the natural mechanism by which the cell begins the process of copying nucleotides). After the DNA replication is finished, we measure the segments in a mass spectrometer, and receive a series of numbers that can be checked against a database we prepared in advance, which contains more than 1,000 organisms that cause diseases in humans.
Together, the hardware and software serve as a universal detector of pathogens capable of identifying the type of organism causing the disease, as well as some of its unique characteristics, within a few hours.
A prototype of such a device I helped create was put to the test in 2009, when a nine-year-old girl and a ten-year-old boy with flu-like symptoms were diagnosed in two different locations in Southern California. Doctors took a mucus sample from the children's throats for a quick and standard flu test. The results revealed that the young people had the flu, but did not find out which strain it was.
The samples were sent to the US Naval Health Research Center in San Diego, which tested the prototype device. The device correctly identified that both children were infected with the same strain of virus, a strain never seen before. The device also identified that the virus originated in pigs, because the RNA letter count obtained was similar to that of swine flu strains that were stored in the database. What's more, the mass spectrometer letter count, or fingerprint, of the first two cases matched that of later samples of what was eventually called swine flu, now known as the 1 H1N2009 pandemic virus. No one can say whether the early warning saved lives, but it probably didn't, and the routine availability of technology capable of detecting new and unique strains of viruses would undoubtedly be valuable for detecting new disease outbreaks.
In view of the fact that it was important to quickly identify the new strain of influenza in 2009, universal detectors of pathogens are expected to stand out in their effectiveness, especially in diseases that doctors have no idea what causes them. The devices can also help with medication selection. For example, the same mass profile that identifies the bacterial strain also provides clues to its sensitivity to different types of antibiotics, allowing doctors to immediately prescribe the correct antibiotic and only when it is really needed. Patients will enjoy a faster recovery time, even in the case of resistant strains, because they will receive the optimal treatment sooner.
induction
Beyond testing a single person, doctors will be able to quickly determine whether several people in a given area have been infected with the same organism, for example salmonella, which is a common source of food poisoning. One might expect that once such information is in the hands of public health researchers, they will conduct a traditional investigation of the epidemic in the field, interviewing patients and tracing their origins to determine if they all have something in common, such as dining at a certain restaurant or eating a certain ingredient in a salad. Such an investigation, which has a similar outline to the one conducted by John Snow in 1854 to show that the cholera epidemic in London was caused by a particular water pump, can take weeks or months, and therefore usually only the most serious outbreaks of the disease are investigated.
However, there is a better way than this, and the key to it is in your pocket or bag right now. Most of us today carry a mobile phone that contains information about our location as part of the operating software or in one of the accompanying applications. Also, service providers collect various types of information based on which it is possible to verify our location at any given moment. If patients infected with an organism with public health implications volunteered to share their past-day location history as recorded on cell phones, epidemiologists could quickly determine whether several patients infected with the same organism visited the same location in a given time window.
In a system based on mobile phones, it will be necessary to maintain the same right to privacy as in the current epidemiological investigations with one big difference: the answers will be received much faster. Through proper coordination, the data from a well-planned network of universal pathogen detectors will enable near-instant identification of a threat to public health such as a pandemic outbreak, bioterrorism attack, or life-threatening food contamination. Also, health experts will be able to know immediately what the source of the infection is and whether it is limited to one city or if it has already spread to several cities. The results will be quickly reported to patients or health authorities as required, and doctors will be able to speed up information sharing about effective treatments.
Building such a network, which I call the "threat network", will finally transfer medical diagnosis and epidemiology from the 19th century straight into the 21st century.
What is the size of the network?
Since the spread of pollution can be viewed as a social network, we can mathematically determine how many such detectors must be connected to each other for the project to act as an effective warning system in a particular country or region. One of the easiest ways to approach the problem is to use a mathematical model known as Monte Carlo simulation, a model in which the computer runs the same scenario under different conditions to determine a range of possible outcomes. (Investment firms routinely use similar calculations to estimate the size of a client's pension fund under various potential market conditions.) When there were epidemiological data on the rate of infection, where and how sick people receive treatment, how often diagnostic tests are done, and what is the incubation time of a wide variety of pathogens Diseases, I ran the data thousands of times to determine the rate at which the network would begin to provide early warning of a national outbreak of disease.
The results were amazing. It is enough to connect to a network 200 carefully selected hospitals across the USA to encompass the entire urban population of the country. Any urban area the size of Washington DC or the city of San Diego would need about five hospitals with networked universal detectors. Under such conditions, it is enough for seven patients to seek treatment in an emergency room for immediate identification of a disease agent affecting public health, such as bird flu, anthrax, plague or food poisoning, with a probability of 95%.
This incredibly low number of machines connected to the network, or nodes, originates from what I call the "funnel effect". Most sick people stay at home to take care of themselves. Only the more severe patients will reach the hospital (the first funnel), where skilled doctors (the second funnel) will decide which of them should be diagnosed. In other words, there is no need to put biological sensors where people are, which would require more devices; Rather, it is enough for the "right" people to be channeled, like through a funnel, to the sensors.
When I made computer simulations of the most common diseases and compared the function of the "threat network" in detecting disease outbreaks to the best results of the method available today, I found that the threat network was much better. She identified the spearhead of the eruption, days to weeks before the existing system. In the real world, warning of an outbreak a few days earlier can mean the difference between life and death for thousands of people, because it allows hospitals to prepare for an influx of patients, health authorities to access drug stocks, and researchers to determine the source of the problem.
What next?
According to my calculations, establishing a network of devices in 200 hospitals will cost about 40 million dollars (assuming that the hospitals will buy the devices themselves) and then another 15 million dollars every year to maintain the network. On the other hand, according to a 2012 study that examined the 14 most common causes of serious foodborne illnesses, the direct cost of treatment and absence from work is $14 billion per year. In the US, it therefore makes sense for the American Centers for Disease Control and Prevention (CDC) to manage the network, given their roles and expertise in detecting disease outbreaks.
No one had previously developed a sophisticated epidemiological surveillance system like the threat network. Based on past experience, designing the hardware and software will probably be the easiest part. It will also be possible to solve many issues concerning regulations, laws and territories. But the biggest obstacle is that there is no single body that has the mandate, incentive or opportunity to launch such a venture, even if it is for the benefit of all humanity. It will be difficult to achieve the level of cooperation required from doctors, nurses, hospital administrators, public health experts and privacy advocates, especially in countries where most of the health care system is decentralized and privatized.
A society-wide and integrative approach to the diagnosis of infectious diseases will be considerably more effective and less expensive than the prevailing approach to public health and the medical means of identifying epidemics and biological threats. The idea of mounting the real-time monitoring of public health on the shoulders of next-generation diagnostic methods, in combination with a network and modern means of communication, holds great potential for improving patient care, saving on the use of antibiotics, and providing alerts that will allow an earlier stop of disease outbreaks or biological terrorist attacks. All we have to do is see if we are wise enough to combine efforts and create a smarter health monitoring system.
_______________________________________________________________________________________________________________________________________________________________
About the author
David J. Ecker is a scientist and inventor at Ibis Biosciences, a company belonging to Abbott.
in brief
Scientists are developing new biological sensors that can detect within a few hours with the help of a patient's sample, if the source of the disease is an infection of a virus, bacteria or fungus.
Patients will receive the appropriate treatment sooner, and doctors will be able to prescribe antibiotics only when they are really needed.
Connecting no more than 200 such biological sensors in a network would provide the US with early warning of epidemics or bioterror attacks.
The main obstacles in creating such a network are mainly political and regulatory obstacles, not technical.
More on the subject
Ibis T5000: A Universal Biosensor Approach for Microbiology. David J. Ecker et al. in Nature Reviews Microbiology, Vol. 6, pages 553-558; July 2008.
Comprehensive Biothreat Cluster Identification by PCR/Electrospray-Ionization Mass Spectrometry. Rangarajan Sampath et al. in PLOS ONE, Vol. 7, no. 6, Article No. e36528; June 29, 2012.
"Salvage Microbiology": Detection of Bacteria Directly from Clinical Specimens following Initiation of Antimicrobial Treatment. John J. Farrell et al. in PLOS ONE, Vol. 8, no. 6, Article No. e66349; June 25, 2013.
The article was published with the permission of Scientific American Israel
