Medical Device Link.

MOLECULAR DIAGNOSTICS

Photo by CombiMatrix Corp.

To respond effectively to recurring influenza A epidemics and a possible flu pandemic, virulent influenza A isolates must be identified quickly. Unfortunately, RNA viruses such as influenza represent a moving target for vaccines, diagnostics, and therapeutic approaches. Influenza, a single-stranded RNA virus, is subject to a high level of mutation, recombination, and genetic diversity. As a family, RNA viruses are prone to this same high level of mutation due to a lower fidelity during replication of the viral genome. Changing the genetic makeup of the virus helps it adapt to its environment and evade the host immune system.

The influenza genome is somewhat unusual in that its genetic material is divided into eight separate segments. The primary antigenic determinants for the virus reside on two of the eight segments: the hemagglutinin (H) and the neuraminidase (N) genes that code for glycoproteins that mediate the virus’s attachment and entry into the host cell. There are multiple varieties of these proteins: 16 different H subtypes and 9 different N subtypes, which have been recorded in 50 combinations (144 combinations are theoretically possible). If a host cell is infected with multiple virions, it is possible for these segments to recombine to form new hybrid strains. Because an infectious event generates a large number of viral progeny (around 1000 viral progeny per single infectious event), the rate at which mutants can be created, due to stochastic mutation and recombination, can become very high. In fact, roughly 70% of viral progeny are mutant strains of the infecting virus.

Given the large number of viruses generated per infectious event and the high level of mutation and recombination for influenza, it is no surprise that there are many strains of influenza in circulation, and new strains appearing rapidly. The large genetic diversity manifests itself as antigenic variation. This means that the ability of the host organism to successfully employ the same immunological defenses against the virus over a sustained period is limited. Because the virus remains in a state of genetic flux and can afford to test new mutant strains due to the high number of viral offspring, it can also evade immune surveillance, become vaccine resistant, and, most disturbingly, develop a resistance to drug therapy.

(click to enlarge) Table I. Comparison of selected commercial immunoassay systems for blood testing.

Type A viruses are believed to have been responsible for most known human influenza pandemics, and avian influenza viruses are therefore key contributors to the emergence of new human flu pandemics.1 The natural reservoir for most Type A influenza strains is wild and domesticated birds, especially species that live in aquatic habitats such as ducks, geese, swans, gulls, and sandpipers. The spectrum of influenza subtypes found in avian populations is remarkably broad compared with the number that have been isolated from humans. However, there are also some influenzas found in humans that do not infect birds (see Table I).

Infections of wild-bird populations present a particularly difficult challenge. The response to infection varies widely between bird species, and the geographical locations can vary significantly. To compound this problem, although avian influenza has caused large-scale death in birds, it certainly is not the only cause of mass mortality in the avian population.

Transmission of Avian Influenza

How the virus is spread among birds is not entirely understood. Transmission through avian species likely occurs through one of two routes: the migration of wild birds and the movement of farmed poultry. In the former case, the incidence of dead birds, including prominent species such as swans, has been at the forefront of public attention. These situations have sometimes been associated with outbreaks of avian influenza. However, since migratory birds are also susceptible to other lethal infections such as avian cholera and botulism, not all dead birds are the victims of H5N1 infections. In fact, between June and mid-August of 2006, more than 18 significant epidemics involving migratory birds were identified in the United States, none attributable to avian influenza.

The second mode of transmission, through farmed stocks of poultry, is more complex. Avian influenza infections may move between poultry stocks and migratory birds. Although outbreaks of H5N1 are certainly lethal in chickens, there has been evidence suggesting that the symptoms of the disease can be masked in chickens coinfected with another strain of influenza. Furthermore, immunization programs designed to vaccinate broiler stocks against other pathogenic strains of avian influenza (e.g., H9) may confer adaptive immunity onto flocks exposed to the H5 virus. This latter, disturbing possibility provides for a Typhoid Mary scenario in broiler stocks that are moved over large distances or even across international borders. Therefore, the diagnoses of mixed infections—whether real or caused by vaccination—are critical pieces of information.

Diagnostic Challenges

One challenge of creating influenza tests is effectively gauging the pathogenicity of a strain. As the mortality of sentinel species like birds may be due to a variety of causes, infections may provide only limited information. In addition, the effects of a virus in birds may be distinct from those in humans (e.g., avian influenza has been associated with encephalopathic pathology in certain bird species). Whether an influenza epidemic is associated with high mortality is critical. Although this connection can be inferred through DNA sequencing, it is difficult to determine using immunoassay methods. Indeed, lower-pathogenicity forms of H5N1 have been found in wild birds in North America on multiple occasions, in 1975, 1986, 2005, and 2006.

Using conventional diagnostics to differentiate the higher-pathogenicity form of influenza, which is not antigenically distinct, from the lower-pathogenicity form is time-consuming. During an outbreak, slow or incomplete diagnosis could lead to serious economic and public health issues.

Should the H5N1 bird flu virus become able to spread more easily through person-to-person transmission, it could develop into a pandemic human influenza virus that could undermine the public health security and economic vitality of countries around the globe.^2-4

There is already evidence that the virulence of the Asian H5N1 virus has increased in poultry, animals, and humans since it first infected humans during a 1997 outbreak in Hong Kong.^2,5,6 Significantly, mutations in the virus have caused severe disease in domestic ducks and wild waterfowl, which formerly carried the virus without developing symptoms. This has been accompanied by a collateral increase in the frequency and severity of H5N1-caused disease in nonhuman mammals.^2,6,7

Morbidity and mortality from the H5N1 virus have also increased among children and adolescents. For example, there were significantly higher rates of clinical disease and death among children in Thailand in 2004 than during the initial 1997 outbreak in Hong Kong.⁸ More than half of the people with confirmed bird flu infections have died, and at least half of all confirmed and suspected fatal cases have occurred in previously healthy children and young adults under 20 years of age.⁹ Death rates from bird flu are now higher among infants and children than among adults. In Thailand, there is a fatality rate of 89% among children younger than 15 years.¹⁰

For public health workers, the rapid change in the genetic makeup of influenza provides a challenge. First, the variation in the influenza viral strain requires a complex annual vaccination strategy that infers the most likely strain for the upcoming year. This process is necessary because it takes a considerable amount of time for the vaccine to be produced, tested, and distributed, and is typically repeated every one to two years as new vaccine-evading strains appear. Second, there is evidence that the mutation and recombination of the virus—sometimes referred to as genetic drift and genetic shift, respectively—can result in drug-resistant variants. As the virus adapts, drug therapies such as Tamiflu from Hoffman-La Roche Inc. may become ineffective. Finally, and perhaps most disturbing, current diagnostic tools for identifying influenza from the natural reservoirs of the disease (e.g., wild birds) are either slow or insensitive, or provide an inadequate level of certainty about the pathogen present in the sample.

Given the threats that H5, H7, and H9 influenza strains represent to human health and agriculture, the standard toolbox of diagnostic assays should be expanded beyond simple immunological or basic polymerase chain reaction (PCR) assays. Rapid knowledge of the exact strain, the origin of the strain, and the probable characteristics of the virus are critical for monitoring a disease outbreak and preventing its spread.

To develop timely diagnostics, especially with the level of genetic variability observed in influenza, it is important to know the genetic makeup of emerging strains. Restricting this information for geopolitical or economic reasons (e.g., technology-licensing strategies) seems at odds both with medical needs as well as with the desire of the IVD industry to develop effective tools for analyzing flu. The recent decision by the Centers for Disease Control and Prevention (Atlanta) and the Association of Public Health Laboratories (Silver Spring, MD) to make genomic sequences for new influenza viruses available to the research community through publicly accessible databases is an important step toward ensuring that outbreaks are effectively monitored and that IVD tests remain up-to-date.

Technical Approaches for Detection and Identification

A variety of technical approaches exist for identifying influenza. Cost, certainty of outcome, assay sensitivity, time to answer, test complexity, and, sometimes, instrument portability are important factors. Because conventional tests, whether close to the bedside or in the field, have significant limitations, a combination of methods is generally used in critical scenarios.

Measuring influenza reliably can be challenging for three reasons:

• Flu infections in humans often exhibit transient viremia. This means that the virus is only shed at high titers for a brief period (around three to five days after infection). Thus, limitations in assay sensitivity are a common problem.
• The most effective sampling procedures tend to be the most unpleasant (e.g., nasopharyngeal washes). As a result, sample quality can vary widely.
• Improper clinical sample shipment and storage, as well as delays in testing, often result in degradation of the sample and false-negative results.

Unless effective tests can be run promptly at the sample-collection site, significant discipline and sampling training is required. To overcome some of these limitations, samples are often stored in viral transport media, buffers designed to preserve the virus. Another option for amplifying critical sample titers is to culture the virus in the laboratory, then to test it using serological or molecular methods. This process is extremely time-consuming.

Thus, ascertaining whether a flu is of the lethal A (H5N1) strain requires that a sample be frozen at the collection site and shipped to a secure laboratory. Once there, the virus is grown in eggs, isolated, and genetically sequenced—a process that takes four or five days, not including shipping time, and runs the risk of having the samples defrost and ruined in transit. By the time a result is available, the patient may have died.

Immunological Assays. Current methods for assaying influenza can be divided into two categories: immunological assays and molecular, or nucleic acid–based, assays. Immunological assays include agglutination assays, and immunofluorescence and enzyme-linked immunoassays, as well as rapid strip-type tests.

Immunological methods are the most established of these technologies for influenza diagnostics, although molecular tests are becoming commonplace. Immunological assays are relatively quick and provide basic information (e.g., the H or N type of the virus), but little else. In addition, a number of these tests have limited sensitivity and a significant false-negative rate.¹¹ Tracking emerging diseases with serological assays is particularly challenging. Generating and testing antibodies is the most time-consuming of these procedures.

Molecular Diagnostics. Molecular methods such as real-time PCR are extremely sensitive, reasonably fast, and can be adapted to new diseases. They often alleviate the need for tedious culturing of the virus. As such, these techniques can complement serological assays. An advantage for any nucleic acid method is that the sample can be stored in denaturing buffers that help reduce false negatives.

However, since real-time PCR methods rely on one or two data points to make a measurement and are frequently subject to spurious signals, the technique often requires verification by a secondary method such as DNA sequencing. In general, the problem with additional amplification products generated by real-time PCR has been addressed by adding one to four dye-quencher probes per reaction. These provide an additional level of confidence, but also significant cost. In addition, because real-time PCR is so specific, false-negative results can appear due to pathogen mutations at the priming sites. Real-time PCR is more flexible than serological assays, but it still provides a limited amount of information (e.g., it cannot easily distinguish between closely related sequences or mutations).

There are two ways to determine pathogenicity of avian influenza in birds such as chickens. One approach is to analyze the gene sequence and to determine whether it looks like a virus with pathogenic potential. The alternative is to inoculate young chicks and observe mortality over a 10-day period. Given that in vivo testing is slow and often impractical, it is reasonable to extract as much information as possible from a sample. In fact, sequence information for influenza can yield strain information that may infer pathogenicity, susceptibility to current vaccination strategy, and drug-resistance characteristics.

Microarrays and Other Methods

A number of approaches can be used to gain extra information about the genetic composition of a virus after the nucleic acids have been isolated and amplified. These methods include microarrays, mass spectrometry, and sequencing. Each of these techniques adds more time and cost to the assay, but they increase certainty and provide significantly more information. Mass spectrometry is effective and highly sensitive, but the equipment is large and complex. Sequencing also has advantages, but the process requires a great deal of complex equipment to be dedicated for a single clinical purpose.

(click to enlarge) Figure 1. Microarray assay data from avian H9N2 influenza A sample interrogated with thousands of probes (a). Amplitude (y-axis) corresponds to electrical signal (picoamperes per electrode) associated with each influenza-strain- specific probe. Multiple loci are probed per strain type (x-axis). Strain-specific probes are grouped so that individual signals from loci can be observed individually. Summary data are also shown (b). Assay time is 4 hours.

Microarrays can provide single-base resolution, if desired, on nucleic acids and can concurrently interrogate a nucleic acid sequence with thousands of probes (see Figure 1).^12-14 This is in stark contrast to the one to four probes that can be employed in a real-time PCR run. Currently, microarrays can identify several thousand different genotypes (e.g., all known influenza strains) and provide sequence information for critical loci in the genome (e.g., known areas that confer drug resistance).

Microarrays can also analyze specific regions at disparate loci without needing to sequence in a linear fashion through whole genomes like conventional sequence analysis. This is useful as only limited genome regions tend to be informative when differentiating between related strains.

The advent of customizable microarray technology (i.e., programmable, application-specific probes) means that interrogating new or varying strains of a virus is straightforward. Because the array can be updated and manufactured in a shorter period than it takes for a new pathogen strain to appear, validation of the array content remains a rate-limiting step. The validation and testing of these systems will become more efficient as the bioinformatics and rules defining microarray probe design improve.

Technical Hurdles. Until recently, microarrays suffered from a number of technical limitations that slowed their dissemination. For example, early microarray techniques for multipathogen detection required assay times as long as 48 hours. This was due to slow amplification techniques and long hybridization times. Improvements in the speed of whole-genome amplification techniques—a reduction from 16 hours to 1 hour, the adoption of pan-specific PCR schemes—has reduced the total assay time of multipathogen detection to 4–5 hours. In addition, the combination of orthogonal PCR amplification and microarray technology permits PCR to function more broadly while improving the sensitivity and hybridization kinetics of the microarray.

Figure 2. An electrochemical microarray reader, laptop, and semiconductor microarrays by CombiMatrix Corp. (Mukilteo, WA), for rapid influenza typing.

Another challenge for multiplex technologies is that microarray cost and the complexity of instrumentation and handling have been high. However, technologies adapted from the semiconductor industry are helping to change this situation. True chip technologies, in contrast to microarrays patterned and spotted on glass, can be read directly using well-established electrochemical detection techniques already used in devices such as blood glucose monitors. These technologies are scalable, meaning that the cost per test can be adapted to the appropriate market and that improved chemical characteristics of the device can make them robust enough for multiple uses. These developments translate to order-of-magnitude drops in price and more-robust, less-expensive, and more-compact instrumentation for multiplex molecular analysis.

The integration of electrochemical detection on semiconductor microarrays enables a new generation of simple microarray instruments that ensure the seamless transit of samples through nucleic acid extraction, amplification, and hybridization onto the array. These technologies help simplify the microarray process and increase assay quality and data interpretation while decreasing the risks of sample cross-contamination. Using embedded semiconductor technologies in the field to deploy multiplex pan-pathogen technologies may aid sample handling and preservation, limiting the transport of infectious agents and assuring chain of custody.

Conclusion

In an era of rapid multinational travel, increasing populations, and global trade, emerging diseases can spread quickly and cause significant socioeconomic damage. The IVD industry plays a critical role in monitoring the emergence of new infectious agents by taking a leading role in surveillance, as well as in guiding therapeutic development. Though challenges remain, fundamental advances in how microarrays are manufactured and used demonstrate that the technology has tremendous potential as a tool for simultaneously screening for multiple pathogens.

(left to right) Andy McShea, PhD, is vice president at CombiMatrix Corp. (Mukilteo, WA). Amit Kumar, PhD, is chief executive officer at CombiMatrix. Joe Dudley, PhD, is chief scientist, biosecurity and bioinformatics, at EAI Corp. The authors can be reached at amcshea@combimatrix.com, akumar@combimatrix.com, and jdudley@eaicorp.com.