Medical Device Link.

Originally Published IVD Technology May 2004

In Person

Amit Kumar, PhD, is president and chief executive officer of CombiMatrix Corp. (Mukilteo, WA). He can be contacted HERE.

What good is an IVD if its results cannot be generated or read properly? As advances in basic science translate into progress in developing diagnostic tests, the latest and greatest tools are being incorporated into IVD detection systems, thereby keeping the IVD industry on par with technologies developed in other medical device sectors. 

To learn what advances lie ahead in detection technologies for IVDs, IVD Technology editor Richard Park spoke with Amit Kumar, PhD, president and chief executive officer of CombiMatrix Corp. (Mukilteo, WA). In this interview, Kumar discusses the latest trends in and emerging advances in detection technologies for IVDs, as well as the challenges involved in developing them.

IVD Technology: What have been the biggest advances in detection technologies during the past few years?

Amit Kumar: Most of the advances have been driven by improved detectors developed by the microelectronics industry including better charge-coupled device (CCD) cameras, photodiodes, and photomultiplier sectors. In addition, improvements have been made in optical light sources in terms of cost, size, and stability. Imaging techniques have also improved. Specifically, in the area of fluorescence detection, better and newer dyes and labels have been developed. 

Another area of detection for biological applications that has emerged over the last few years is nonoptical detection techniques such as mass spectrometry. Traditionally, mass spectrometry was used for nonbiological applications, but recently with surface-enhanced laser desorption ionization techniques, mass spectrometry has been used increasingly for looking at proteins and so forth in a number of biological applications. 

An additional advance is that the specific types of samples that are to be detected have become more amenable and versatile. For example, people are thinking in terms of evaluating a number of assays in parallel as opposed to in a single modality. At CombiMatrix we build DNA microarrays that enable customers to study thousands of DNA assays all at once, as opposed to one at a time. 

Improving Sensitivity

What are the latest trends in detection technologies? 

In general, the trend has been to develop better sensitivity. We are now able to detect signals at a much lower level than we used to because the backgrounds are now lower, so signal-to-noise is better. Many systems that used to be very expensive are now much more amenable to use for broad applications. 

In addition, there have been advances in software development. Often software is used to process diagnostic imaging, and the algorithms used in these programs are becoming more sophisticated, enabling much better background correction and normalization capabilities. These software improvements along with hardware advances have made detectors smaller, more compact, and portable in many cases. They have also led to the development of systems that are much more inexpensive and versatile.

Does developing greater sensitivity for these tests lead to a challenge for specificity?

Absolutely. In general whenever you are developing a detection technology there is always a balance between sensitivity and specificity. Sensitivity is driven by not just the size of the signal that you can get, but also the size of the signal relative to the background. Today our electronics capability is so good that if your system can generate a signal, then you can amplify it and do all kinds of things to understand what that signal means. 

The most important component of properly reading the signal is reducing the background. The background might be due to various nonspecific things occurring in the biology of the IVD you are developing, but could also be due to certain levels of background signal that are inherent in the electronics or optics. 

For example, in fluorescence detection, a common problem is that regardless of how well an optical collection system, light path, and so forth are designed, a little bit of stray incident light can get into the detector channel and can set the background limit. If you can cut that incident light down significantly, then the little bit of electronic noise in your detector channel will be your background limit. 

So the important thing to address when you are looking at sensitivity is signal to noise. Sensitivity and specificity often go hand in hand. A lot of times if you increase your sensitivity you lose specificity and vice versa, but I think we are making tremendous strides in that area.

Overcoming Obstacles

What are some of the primary challenges IVD manufacturers encounter when designing and developing detection technologies?

One of the key differentiators between the manufacture of an IVD product relative to a general product is that the development of IVD products requires some very specific detection challenges. For example, when building a mass spectrometer for general use, it has to be very versatile so that it can perform many functions. If you are building a mass spectrometer for a very specific assay, which is what IVD manufacturers do, then the product becomes much more restricted and focused on a particular type of detection experiment. 

In many ways the product development becomes simpler, but in many ways it is much more complicated because now IVD manufacturers are subject to the scrutiny of regulating bodies like FDA. Another clear challenge is presented by the strict metrics that IVD manufacturers have to adhere to with respect to reproducibility and robustness because their instruments will be used for clinical purposes, and that is not always the case for general laboratory instruments.

What are some of these strict metrics that make the development of detection technologies challenging? 

The metrics include measurements like the coefficient of variation. The variability in the result of a particular assay is driven by a number of factors, including variability in the sample and sample handling, and also variability in the ability of the detection system to produce an accurate result. 

Suppose you were to look in a single sample for a protein marker that was indicative of a particular disease. If you split that sample into 10 different aliquots, ran all of them on a single detection system, and got 10 very different results, obviously that system is not going to be acceptable to a regulatory body. Certainly patients and physicians do not want a system that is going to generate results one day that are different from the results it gives you the next day. So it is very important that there are very strict reproducibility metrics. 

The system has to be very robust so that on a particular day when is very humid you get the same result you receive on another day that is not humid. In addition, an IVD manufacturer typically deals with the health of a patient and therefore patient quality of care issues. There are liability issues. So the concerns that go along with measurements used in patient care are much more critical for an IVD product than for a general laboratory instrument. 

How do IVD manufacturers overcome these challenges when developing detection technologies?

They use a couple of key approaches. IVD manufacturers typically make systems that are very specific to the assay that they are developing, so there is not a lot of excess capability on each machine. That has two advantages. One is that the user cannot change certain parameters on the system that take it completely out of calibration, for example. And another is that IVD manufacturers can focus on the key metrics and parameters that are important for the particular assay. 

One area that is of interest at CombiMatrix is the use of DNA microarrays for diagnostics. Right now the regulatory bodies don’t have a good feel for how to regulate those types of products, and one of the problems is that there is a lot of variability in results that are achieved using microarrays. In order for microarrays to become a real-world commercially viable IVD product, the detection systems and the microarrays themselves have to meet a certain minimum set of metrics in terms of reproducibility, performance, and sensitivity. The industry and the regulatory agencies are going to have to work together to define these requirements. 

What challenges are involved in taking technologies that more or less originated from other industries and adapting them for use in IVDs?

The challenges are common to both general lab instrument manufacturers as well as IVD manufacturers. One of the challenges in developing detection for IVDs has been the advancement of the chemistry and the biology in such a way that they can parallel the advancement that has occurred on the electronic side. 

One example is a big advance that has come from the electronics industry—the availability of laser diodes with wavelengths that can go all the way across the spectrum from the red to the blue. In the past, fluorescent dyes like fluorescein and its derivatives have been utilized for various types of IVD as well as general laboratory assays. Over the last few years, a whole new series of dyes have been designed specifically to work with new types of lasers. 

In order for that parallel advancement to take place, do IVD companies have to bring in experts in those fields to work with their engineers and scientists? 

In some specific cases that may be necessary, but in general, IVD manufacturers or developers of IVD systems have the biologists and the chemists that know how to run and develop the assays and the engineering team to build instruments. Those engineers typically will understand enough about the biology side and the new advances on the electronic side that they should be able to put it all together. 

There are always cases when a technology is very new, and a consultant can be brought in to help out, but in general that is not absolutely necessary. An engineer who built a system with a gas laser in the past can very easily adapt to using a diode laser. You don’t need to bring in a diode laser expert to show him how to do it. 

What are the advantages of the new dyes for laser diodes?

The main advantage of these dyes is that they can be used with laser diodes. Early laser diodes typically operated in the red, and as a result you couldn’t use a lot of the other dyes like fluorescein that had been used previously with the traditional lasers, not the diode lasers, and as a result we needed dyes that would allow you to use diode lasers. 

Another benefit of dyes is that in biological systems there are not a lot of things that autofluoresce in the red part of the spectrum. As a result you get less interference from nonspecific background effects, and in principle you can build better, higher-sensitivity detection systems. 

Dyes that work with laser diodes also allow you to build detection instruments that are much smaller, compact, and less expensive than if you had to use a big gas laser. With the added benefit of having dyes in the area where there is less background interference from biological material, systems with much higher sensitivity can be built. And then there are a bunch of hybrid dyes that rely on a resonant energy transfer, and various other things. These dyes enable you to separate the wavelength of incident light with the wavelengths of omitted light from far enough away that you can build much better optical systems resulting in much better signal-to-noise, hence much better sensitivity.

Molecular Diagnostics

How have the developments in molecular diagnostics affected development of detection technologies over the years?

Developments in molecular diagnostics are driving a lot of the most innovative detection modalities. Molecular diagnostics is a holy grail for developers of methods for DNA analysis. Whether it is PCR, multiplex PCR, or microarrays, the goal is to make those research tools into diagnostics. 

Over the last few years, we have learned many techniques for detecting the presence and the quantity of nucleotides and molecules in different types of samples. These approaches all incorporate some sort of detection, mostly optical and fluorescent in nature. 

Other factors that have influenced detection development include the availability of new dyes and labels, as well as the availability of lasers that can produce wavelengths of different colors. Detection methods have developed in parallel with the improvement and understanding of methods to measure DNA, and I think those trends will continue.

It is important to note that developments in the understanding of how DNA correlates with disease have put pressure on IVD manufacturers and designers to improve detection technologies. In most cases, the most important information obtained from a DNA measurement or molecular measurement comes from very low levels of nucleic acids and therefore greater sensitivity is incredibly important for detection systems used in molecular diagnostic applications. 

Another big advance in molecular diagnostics has been the concept of multiplexing. In clinical diagnostics, you can look at two or three or four markers and get a fair amount of information about some endocrinological state, an infectious disease, cholesterol levels, or various other things like that. In molecular diagnostics, most often you have to look at a number of genes. Therefore you cannot just look at one, two, three or four and get a lot of information. You may have to look at hundreds to thousands of genes at once to gather information on a particular phenotypic condition. So multiplexing on things like PCR assays and DNA microarrays, which are being developed by many companies, is a very important factor. As a result, detection manufacturers have to begin focusing on ways to look at multiple samples all at once as opposed to one sample at a time in a serial fashion.

If you are looking at a thousand different DNA experiments on a DNA microarray and it takes 30 seconds to look at each one, it will take 500 minutes to do the analysis. However, if you can do it all in parallel and do it in a few seconds, that is certainly much more powerful and much more applicable for IVD manufacturers and also for general laboratory research.

What are the advantages or disadvantages of chip- versus bead-based microarrays?

There are so many advantages and disadvantages to both that I could write a dissertation on that question. Chip-based approaches have been traditionally used in the IVD industry, whether that chip is on a piece of a glass or on a piece of silicon. They tend to be amenable to various types of analyses. Bead-based systems require you to keep track of the beads through some sort of labeling mechanism, and often it is difficult to muiltiplex bead-based assays. Some companies are multiplexing thousands to hundreds of thousands of assays on chips, but we will see how those go. 

We think chip-based systems are much more valuable because they allow you to build the DNA in situ. You cannot build the DNA in situ with bead-based systems. When building the DNA in situ, you actually build the DNA sequence one base at a time at a site attached to the surface, which is what you do on a chip. On a bead-based system, you typically build the DNA off-site and then you attach it to the bead after it has been built. 

If you are building the sequence in situ and you use a very simple method to design and build different DNA sequences, you can very easily customize the content on the chip to the requirements of the researcher. So a researcher can ask for chips that have any gene, any genome, any length, and all sorts of other variables that are of interest. We have researchers who have asked us to put the genome of a virus as well as the genes of the host human cell on the same chip, and we can do things like that because the technology is chip based. 

With bead-based arrays, customization is a little bit more difficult because you have to keep thousands and thousands of beads in beakers and then mix them according to the requirements of the customer. If the customer does not like a probe that has been affixed to a bead, then you have to remake a whole new bead and a whole probe. In the case of chips, nothing is made ahead of time. The whole chip is based on the customer’s requirements. 

How have CombiMatrix’s detection technologies been applied in diagnostic tests?

We are working with partners in three diagnostic areas. One is Rational Diagnostics (Seattle) and the University of Washington (Seattle) on some programs in lymphoma. We are also working with a research group at Case Western Reserve University (Cleveland) on Alzheimer’s diagnostics. Lastly, we are working with Saint Jude Medical (St. Paul, MN) on influenza. 

In each case it is really not just the detection that is important—it is the detection in concert with the capabilities of the chip, especially the ability to rapidly customize the DNA content on that chip. All three of these programs utilize optical detection methods right now, but eventually they will be moving into electrochemical (e-chem) detection methods. 

Rational Diagnostics and the University of Washington have data-banked a large number of tissues and the correlated clinical histories from patients to document what kind of lymphomas they had, the results of certain drugs, and so forth. We are looking at the genetic framework of those particular tissues to understand what genes were correlated with what outcomes and responses to particular drugs. Once we have done that analysis we hope to find a subset of genes that we can put on a chip, and whenever someone is suspected of having lymphoma we can run that chip on that individual and based on the genes that are over- or underexpressed, we can determine which drug would be most effective in treating the disease. 

We think we will get to the point where we can actually effect personalized medicine. Lymphoma, Alzheimer’s, and influenza are three diseases that we are focusing on right now, but we will advance that into a whole number of other diseases. I think the key enablement here is the ability to very rapidly customize the DNA content of the chip. 

Impact of Biodefense

How have the emergence of biodefense and bioterrorism concerns affected the development of detection technologies? 

We are actually receiving capital from the U.S. government to build systems that could be used on the battlefield by a soldier for the detection of various biological agents like anthrax and plague. I view systems for detection of biological or chemical warfare as just a subsection of IVD detection. There is one key difference, however, and that drives a lot of the requirements for IVD systems for biodefense. That key factor is that a bio detector will most likely be utilized in a nontraditional setting, not in a laboratory. 

For example, many systems might be utilized on a battlefield or in a high-risk target environment like an airport, train station, subway station, or post office. As a result, the system has to be robust and adaptable enough to work in those environments. For example, if you are building a system that is going to be operated in the desert in the Middle East, it has to be able to withstand the conditions of that particular scenario. 

In addition, characteristics like portability and size are becoming important. Obviously if a soldier is going to carry something in his backpack, it has to be much smaller than typical laboratory instruments and it has to be operated off of a battery. 

Ease of use is critical because systems that were operated in a laboratory would often require a fair amount of technical manipulations to be done by a highly trained technician. And even automated systems for clinical diagnostics that are ubiquitous today require a lot of robotics and lab information management systems. Those are all things that you cannot take on a battlefield, so ease of use is a very important characteristic. And of course the requirements of sensitivity and specificity and other traditional metrics remain. 

On a battlefield, you will need to be able to have a system that has automated operation, operates for 24 hours, 7 days a week, and generates faster results because you don’t want to be sitting there for an hour to determine if there is anthrax in the air. So all of the requirements of traditional IVDs remain in terms of sensitivity and specificity, but now you have the added requirements of portability, robustness, size, versatility that are required for operation in a number of new environments. 

How have IVD manufacturers attempted to address these issues of portability, durability, speed, and the ability to test for a broad spectrum of biothreat agents simultaneously? 

The multiplexing aspect is very critical because out on the battlefield you have no idea what your enemy may be able to expose you to and, in any high-risk environment, you don’t know what agents will show up. You can never, ever, identify everything. But the more you can identify and the more sensitive a test can be, the better the chance of protecting potential victims of an attack, so we are building systems that can multiplex a number of biological agents on a single chip. 

In principle, we should be able to look at anywhere from two to a dozen different agents. If the agent that an enemy is using can’t be detected by the chip you’re using, the chip is not going to do you any good. So you want biothreat detection systems to be as broad as possible. The U.S. government has identified some key agents of importance and we and many other companies are working on systems that can identify those. 

IVD manufacturers are dealing with the added requirements of robustness and portability that are necessary for terror and warfare applications by focusing on specific components and systems that are able to address those needs. For example, we are using a lot of microtechnology, microfluidics, and nanotechnology. 

We are trying to build a system that is about the size of two PDAs put together, something that is small enough to be carried in a pocket and is operated off of a battery. 

To develop a system that is this small, you must use chips with nanotechnology incorporated into them. You also have to utilize very compact sources of detection, and we are using an e-chem detection modality that eliminates the need for light sources and optics. So you get around the traditional types of design and development and go after solutions that are a little bit more innovative to develop systems that are appropriate for defense applications. The general life science community will also benefit from these new solutions. Once these very compact and versatile systems are developed for defense applications, they will be applicable for traditional IVD applications, including point-of-care diagnostics or point-of-site diagnostics. So I think these are trends that are tremendously beneficial for IVD manufacturers. 

Reagent-Free Detection

What sort of detection technologies have been developed at and by CombiMatrix? 

We have developed two detection techniques, the first is a standard CCD-based fluorescence imaging of microarrays. This technique has been incorporated into an instrument that we developed with Roche (Basel, Switzerland) called the Matrix Array hybridizer reader. This hybridizer reader is based on traditional fluorescence imaging with a light source and so forth. 

The second and most important detection format is our e-chem detection approach that utilizes the electrochemical capabilities of our chip with very robust enzyme labels. The enzymes are capable of amplifying the signal much more than fluorescence labels, and this approach has numerous advantages over fluorescence as well as other optical approaches including chemiluminescence, electrical chemiluminescence, etc. 

The first advantage is that the e-chem detection modality enables very small, portable, and inexpensive instruments. As I mentioned earlier, we are building systems that we think could be the size of about a couple of PDAs put together, smaller than a laptop, and that is enabled by fact that with e-chem detection, there is no longer a need for expensive and large light sources, detectors, or optics. As a result, we used this technology to build portable detection systems for biological and chemical warfare agents funded by the U.S. Department of Defense (DoD), as noted earlier. 

But for IVD manufacturers as well as for DoD applications, the e-chem detection modality produces better signal-to-noise ratios resulting in more sensitivity. This is because the background is not determined by stray light, as is the case in most fluorescent systems. Rather, it is determined by nonforadaic electrical currents. In other words, the background signal that drives the signal-to-noise ratio is much lower for e-chem detection systems. 

E-chem also allows for a number of different detection formats. For example, detection can be performed amperometrically, meaning by measuring current. You can do it potentiometrically, by measuring voltage. You could also do measurements in real time and obtain the kinetics of binding, which you cannot do with fluorescence detection.

In addition, because the chip itself is the detector and we essentially have on-chip detection, we can embed the detector in a microfluidic set and create a totally automated disposal system that can be used for a number of different applications. All the reagents are included. The whole system is self-enclosed. You run the test and then you can throw it away. 

Our vision for this technology is to build a very inexpensive reader system that costs perhaps less than $1000 and can read low- and high-density microarrays very efficiently. This type of cost metric can never be achieved with fluorescence detection approaches. We think that this technology is going to broaden the use of microarrays in the R&D markets and hence hasten the ability to bring microarrays into the IVD marketplace.

Since this is basically a fluorescent-free assay, do you believe that fluorescence may be used less extensively in the future?

That is inevitable. Eventually there will be other modalities that will be used, and I think e-chem approaches are going to be one of the most popular approaches. Today most people use fluorescence because they understand how to conjugate their samples with fluorescence dyes and samples have been processed that way for a while. But just as when radioactive immunoassays were the standard 20 years ago and they are slowly being phased out and everything is moving in the direction of fluorescence, eventually I think things will move in the direction of e-chem detection because of its many advantages.

There are a number of companies that have attempted to build e-chem detection systems. Whether a company is developing fluorescence detectors or e-chem detectors, the advances in the actual assay or microarray that it is going to be measuring will also have a tremendous impact on the type, quality, cost, and capability of the detection system. 

Right now, array readers cost anywhere from $100,000 to $400,000, so there are not that many institutions with the budget to use them. But imagine a world where an array reader costs just a few hundred dollars. In principle, every laboratory could have several of those machines running in parallel. If we bring down the cost of microarrays, which we are doing today, then the concept of microarrays for molecular diagnostics becomes a real possibility. 

Today, using microarrays for molecular diagnostic applications is very difficult. Maybe the instrument cost is manageable because you can purchase an instrument for $400,000, but you cannot run a $300, $400, or $500 chip on every patient that comes into a doctor’s office. The U.S. healthcare system just won’t be able to support that. But if you can build a very inexpensive instrument and chips that cost a few dollars, then it becomes more feasible to develop IVDs using a multiplexed microarray type of product. And that is what we would like to make a reality. 

In addition, as you reduce the cost of microarrays and instruments, more progress can be made in understanding the correlation between disease and genes, and therefore the science that underlies the ability to build IVDs will be furthered. I think that, over the next year or so, e-chem detection is going to revolutionize the industry. 

Future Trends

What future challenges will emerge when developing detection technologies?

There will be economic challenges as well as performance challenges. Performance challenges are the standard characteristics that we have talked about already like cost, size, robustness, regulatory approval, and reproducibility of results. In the molecular diagnostics area, which I think offers the greatest growth opportunity for IVD manufacturers, the key is robust systems that generate reproducible results all of the time. 

We have all heard of situations in which different array platforms give very different results on the same samples. In fact, some array platforms have even produced different results on the same samples, even when using the same type of array over and over again. These differences do not arise from the detection approach, but rather from the quality of the arrays and the method of sample preparation. The industry has to be very confident that they can run a particular sample on a particular array and know that they will get an accurate result. 

This was also the case 40 or 50 years ago when clinical chemistries were being developed. You would get a lot of different results depending on how you ran those clinical chemistries, but over a decade things became standardized. Now if you run your cholesterol test at a lab in California and a lab in Massachusetts, you will probably get the same results, plus or minus a little bit. That doesn’t exist in the molecular diagnostic arena today, but it will eventually have to. It will be driven by regulatory issues, economic issues, and by the knowledge and experience of leading manufacturers of these types of detection systems and arrays. 

What new trends can we expect to see in the future in the area of detection technologies?

We will see less-expensive, more-versatile, and more-compact instruments that are more sensitive as well. In addition, I think we will see much more robust arrays and assays on the market. As we learn a little bit more about molecular diagnostics and how genes correlate with diseases, we are going to see revolutionary new approaches to using IVDs. 

We are already seeing that IVDs can be tremendously valuable in determining which patients will respond to which drugs. One example of this type of drug/diagnostic pairing is Herceptin, a drug for metastatic breast cancer by Genentech (South San Francisco, CA) and another example is Gleevec, a drug for various types of leukemia by Novartis (Hanover, NJ). 
Over the next 10–15 years, we will see progress toward the realization of personalized medicine where IVDs, and specifically molecular diagnostics, are going to make revolutionary changes in the way physicians treat patients. The ultimate outcome is going to be tremendously better healthcare. 

From an economic standpoint, the result for IVD manufacturers is going to be tremendous commercial opportunities, especially for those that are building very versatile and robust systems that take advantage of the latest technological advances occurring in the industry. 

Copyright ©2004 IVD Technology