IVD Technology Magazine
IVDT Article Index

The promise of miniaturized clinical diagnostic systems

Kurt Petersen, William McMillan, Gregory Kovacs, Allen Northrup, Lee Christel, and Farzad Pourahmadi

Miniaturization of fluidic and sensing components using both micromachining and precision injection molding is making possible fully integrated, miniaturized diagnostic systems and components. These next-generation systems have generated a great deal of excitement, and rightfully so.

Miniaturization offers the promise of rapid, on-site analysis. Real-time diagnosis of sexually transmitted diseases such as AIDS and chlamydia has tremendous value. Obtaining a test result while the patient is still in the office allows for immediate treatment or counseling decisions. Point-of-care testing of patients for blood chemistry or drugs could also speed treatment decisions, perhaps saving lives.

Pillars of a microfabricated DNA-capture chip by Cepheid (Sunnyvale, CA).

Testing candidate blood donors for the major blood viruses before they give blood would save the blood product industry time and money associated with pooling and processing blood that is later found to be contaminated. Donor testing could also improve the safety of the blood supply for transfusion medicine, especially during times of serious donor blood shortages when rapid donation campaigns are under way.

The recent incidents involving significant morbidity and mortality due to consumption of contaminated beef and fruit juice point out the need for fast, easy, on-site testing for food contaminants. With a portable system, it would be possible to test food stocks such as beef cattle and poultry flocks for the presence of pathogens, thus preventing their introduction into a food processing plant.

Field testing of air and water for bacterial, viral, or chemical contamination offers the ability to immediately identify and trace sources of infection or pollution and to provide warning of terrorist acts or the use of biological weapons.

Benefits of Scale

Besides the obvious benefit of portability, miniaturization offers additional advantages based on fundamental scaling principles. An area in which scaling effects are particularly favorable is the thermal processing of samples. Many biomolecular amplification and detection techniques require isothermal heating or thermal cycling. The high thermal isolation and low thermal mass achievable by miniaturization of the thermal components of fluidic systems can greatly reduce power requirements and thermal time constants. In addition, localized heating can be used to enhance reaction, mixing, reconstitution, and other processes.

This approach can greatly speed up thermal cycling in polymerase chain reaction (PCR).¹ In PCR, mixtures of sample plus a "PCR mastermix" are subjected to a temperature cycling regimen in which the target DNA is doubled during each temperature cycle of several seconds at 95°C and several seconds at 72°C. In today's commercial instruments, each cycle of this type takes about 2 minutes; 40 cycles take about 1.5 hours. Much of this time is consumed by the transition between the widely separated reaction temperatures. Reduction of the thermal mass of the reaction system, however, has made it possible to realize complete thermal PCR cycles in as little as 10–20 seconds; 40 cycles can now be completed in only 10–15 minutes. Researchers have amplified DNA and RNA using PCR, then fluorescently detected extremely low concentrations of DNA in a single tube in times as short as 15 minutes.^1,2 Another team also amplified DNA using PCR, then performed an electrophoretic separation, all in an integrated assembly and all within 15 minutes.³

High-speed DNA amplification has other important benefits. During the periods when the sample temperature is making a transition from high to low or low to high, extraneous, undesirable reactions occur that consume important reagents and create unwanted and interfering chemicals. Rapid transitions ensure that the sample spends a minimum of time at undesirable intermediate temperatures, so that the amplified DNA product has optimum fidelity and purity.

Fundamentally, all chemical and biological reactions occur on the molecular scale, one molecule reacting with another. To the extent that reactants can be brought together on a microscale in any type of system, reaction efficiency and speed can both be improved.

Mixed Blessings

The microfluidic channels that make possible this high thermal isolation and low thermal mass are one of the mixed blessings of miniaturization. In general, as channel cross-sectional dimensions are reduced to the 100-µm range, fluidic flows are laminar (nonturbulent) at the velocities that are practically achievable. This phenomenon has profound impact on microscale fluidics, because mixing of fluids in laminar flow is entirely a function of diffusion. The impact can be positive, in that diffusion-driven interaction of laminar-flow streams can be exploited to remove certain molecules from a fluid stream while leaving behind those with lower diffusion coefficients. Such an application offers the opportunity to concentrate specific molecules with high diffusion coefficients.^4,5

On the negative side, laminar flow eliminates the enhancements to mixing provided by turbulence in larger volumes of fluids. Fortunately, several creative solutions to the mixing problem have been put forth, including lamination of fluids in multiple thin sheets to enhance diffusion, formation of multiple plumes of fluid to increase interaction with a flowing sheet, and reciprocating mixing ("chaotic advection").^6–9

Laminar flow offers the additional benefit of helping to prevent the formation of bubbles, which in the very small sample sizes usually processed in miniaturized systems can completely invalidate quantitation. Nonetheless, bubbles are often formed during the introduction of fluids into a system. In the very narrow channels of a miniaturized system, purging these bubbles can be a daunting task. As fluid channels become narrower, the available hydrostatic pressure that can be applied to a trapped bubble decreases (assuming a fixed driving pressure). This problem has spurred development of new approaches for purging, including the use of gas-permeable membranes or the deliberate filling of fluidic systems with water-soluble gases such as carbon dioxide prior to the introduction of fluids (any CO₂ bubbles simply dissolve in the fluid).^10,11

Another mixed blessing of miniaturization is sample size. Microfluidic systems permit, and in most cases dictate, the use of extremely small sample sizes. This can be a substantial benefit in situations such as drug monitoring in neonates. Transfusions are required when 8 ml of blood is removed from a neonate weighing 1 kg, and many premature neonates weigh less than that. However, the volume of sample that must be addressed to detect a given analyte is not arbitrary but is determined by the following equation:

where h is the sensor efficiency (0 h 1), N^_A is Avogadro's number (6.02 x 10²³ mol^–1), and C is the concentration of analyte (moles/L). Thus, the sample volume required is fundamentally dictated by the concentration of the desired analyte(s).

Manz et al. examined the use of very small sample sizes in a landmark paper published in 1990.¹² They concluded that microfluidics is a valid and desirable approach for diagnostic applications and summed up their analysis in a graph of target analyte concentration versus sample volume. This graph shows that biological chemicals associated with clinical chemistry assays (between 10¹⁴ and 10²⁰ copies/ml) and immunoassays (between 10⁷ and 10¹⁸ copies/ml) might be readily assayed with very small sample volumes, in the range between picoliters and microliters. Typical chemicals suitable for this type of analysis can be seen in Table I.

Table I. Concentrations of typical diagnostic analytes in human blood or other samples.

Unfortunately, numerous chemicals (and organisms) are routinely present at much lower concentrations, from less than 100 to 10⁷ copies/ml. These low-concentration samples (also seen in Table I) clearly include most sources of DNA, which must be detected and analyzed in a growing number of new diagnostic procedures.

In order to accommodate such extremely low concentrations and to analyze this important class of target analytes, we must look beyond the boundaries of Manz's original graph, which spanned a concentration range from 10⁸ to 10²¹ copies/ml and a volume range from 10 ml to 10^–18 L. An expansion of this graph (seen in Figure 1) clearly shows that the minimum sample volume required for accurate DNA assays is 100 µl.

Figure 1. Process flow for the fabrication of high-aspect-ratio DNA capture surfaces in silicon.

Synchronous Microfluidic Systems

An important focus of our efforts to design microfluidic-based diagnostic systems is therefore sample preparation of larger input volumes. Currently, the preparation of complex biomedical samples prior to the diagnostic procedure can be quite laborious and time consuming. In fact, for most clinical diagnostics, it is the rate-limiting step. Many intensive manual operations, such as vortexing, centrifugation, mixing with reagents, phase separations, creation of dilution series, and other steps, are required.

The developers of most current diagnostic instruments, rather than studying and addressing the principles and rationale behind many of these operations, merely automate entrenched and unoptimized manual methods. While this strategy can minimize labor, it does not offer fundamental advances in time, efficiency, or performance.

The new microfluidic technologies offer the opportunity to take a completely different, although not entirely new, approach. This approach is best understood by considering the chemical processing industry. Chemical processing plants routinely and inherently rely upon continuous-flow reactors, process sites, and process conditions. Reactants enter such systems at one end, while products and waste materials exit at the other end after traversing numerous processing and measurement sites. As much as possible, the systems split the fluids into smaller streams or regions (such as in heat exchangers, nozzle arrays, and perforated plates) because fluid processing strategy must be based on the principle of assuring that every fluid molecule is subjected to the same microenvironment, whether for mixing, reacting, temperature control, catalytic activity, or other processes. An example of a microfluidic chip designed to accomplish one such function is shown in Figure 2.

Figure 2. Scanning electron micrograph of DNA-capture chip surface. The pillars are 200-µm high with a pitch of 34 µm.

This same approach can be applied to biochemical analysis, with biomedical fluids continuously flowing through channels, reaction sites, processing sites, and measurement sites. Small portions of the flowing sample can be reacted or otherwise processed at each site as the sample stream passes through the site. For example, active sites might consist of microfluidic chips with micron-sized features for molecular capture, high-efficiency mixing, reactions, and thermal transfer. Reagents can be introduced to the sample stream at appropriate points along the flow path. A crucial advantage of flow-through systems, which use what we call a "synchronous microfluidic systems" approach (as opposed to the traditional bolus approach), is that the volume of sample fluid being processed can be indefinite, ranging from microliters to many milliliters.

These systems would have broad application in areas such as pathogen detection, where several milliliters of sample must be processed and analyzed. An example of the use of this technique is the nucleic acid probe assay, a test for determining the presence and quantity of a target "fingerprint" DNA or RNA sequence. Benchtop and portable systems that are easy to use by relatively unskilled personnel yet still provide on-the-spot results in minutes instead of days are now within reach. By combining recently developed microfluidic technologies for extracting nucleic acid from complex biological samples (see sidebar) with miniaturized chemical and optical detection systems, the powerful world of nucleic acid probe analysis can be made available to the doctor's office, the emergency room, the food processing plant, and even tomorrow's battlefield.¹

An additional benefit of the synchronous microfluidic systems approach is that it permits the combination of complex fluidic scenarios and reagent formulations to provide multiplexing. In many infectious disease syndromes, more than one candidate microorganism can cause the disease, making the ability to distinguish among several pathogens highly desirable. Furthermore, it may be important not only to identify the causative agent but also to determine its antimicrobial resistance or susceptibility as an aid in treatment selection. In another application of multiplexing, the identification of the stage of disease, such as in hepatitis, can be identified by simultaneously detecting multiple blood markers.

Microelectromechanical Systems

To take full advantage of this multiplexing, a system would require multiple data outputs. This brings into play another feature of miniaturization, that of microelectromechanical structures. Microelectromechanical systems (MEMS) offer not only multiple data outputs but also electronic control of (and, therefore, improvement in) quality, performance verification, and the traceability of final results. Electronics, incorporated into the synchronous microfluidic systems approach, can be used for the following:

• To store data related to calibration, lot codes, and expiration dates.

• To measure and verify the proper fluidic and electronic operation and sequencing of the cartridge.

• To compile and store additional data concerning the patient, the test results, the date of the test, and other parameters for archival or future verification purposes.

Conclusion

New technologies incorporating advanced microfluidic techniques allow faster analysis, fully automated analysis, improved sensitivity, and enhanced precision. In addition, the new field of microfluidics offers an opportunity to discover and explore fundamental new principles for biochemical processing, with the potential to greatly improve the performance, sensitivity, and cost of diagnostic procedures.

References

1. Northrup MA, Ching MT, White RM, et al., "DNA Amplification with a Microfabricated Reaction Chamber," in Proceedings of Transducers '93, the Seventh International Conference on Solid-State Sensors and Actuators, Yokohama, Japan, Institute of Electrical and Electronics Engineers (IEEE), pp 924–926, 1993.

2. Northrup MA, Gonzalez C, Hadley D, et al., "A MEMS-Based Miniature DNA Analysis System," in Proceedings of Transducers '95, the Eighth International Conference on Solid-State Sensors and Actuators, Stockholm, Sweden, IEEE, p 764, 1995.

3. Woolley AT, Hadley D, Landre P, et al., "Functional Integration of PCR Amplification and Capillary Electrophoresis in a Microfabricated DNA Analysis Device," Anal Chem, 68(23):4081–4086, 1996.

4. Brody JP, and Yager P, "Diffusion-Based Extraction in a Microfabricated Device," Sensors and Actuators, A58(1):13–18, 1997.

5. Brody JP, Osborn TD, Forster FK, et al., "A Planar Microfabricated Fluid Filter," in Proceedings of Transducers '95, the Eighth International Conference on Solid-State Sensors and Actuators, and Eurosensors IX, Stockholm, Sweden, vol 1, pp 779–782, 1995.

6. Branebjerg J, Gravesen P, Krog JP, et al., "Fast Mixing by Lamination," in Proceedings of the Ninth Annual Workshop on Micro Electro Mechanical Systems, San Diego, pp 441–446, 1996.

7. Larsen UD, Branebjerg J, and Blankenstein G, "Fast Mixing by Parallel Multilayer Lamination," in Proceedings of the Second International Symposium on Miniaturized Total Analysis Systems, mTAS '96, Basel, Switzerland, pp 228–230, 1996.

8. Miyake R, Lammerink TSJ, Elwenspoek M, et al., "Micro Mixer with Fast Diffusion," in Proceedings of the IEEE 1993 Micro Electro Mechanical Systems Workshop (MEMS '93), Ft. Lauderdale, FL, IEEE, pp 248–253, 1993.

9. Evans J, Liepmann D, and Pisano AP, "Planar Laminar Mixer," in Proceedings of the IEEE 10th Annual Workshop of Micro Electro Mechanical Systems (MEMS '97), Nagoya, Japan, IEEE, pp 96–101, 1997.

10. Anderson RC, Bogdan GJ, Barniv Z, et al., "Microfluidic Biochemical Analysis System," in Proceedings of Transducers '97, the 1997 International Conference on Solid-State Sensors and Actuators, Chicago, vol 1, pp 477–480, 1997.

11. Zengerle R, Leitner M, Kluge S, et al., "Carbon Dioxide Priming of Micro Liquid Systems," in Proceedings of the IEEE 1995 Micro Electro Mechanical Systems Workshop (MEMS '95), Amsterdam, IEEE, pp 340–343, 1995.

12. Manz A, Graber N, and Widmer HM, "Miniaturized Total Chemical Analysis Systems: A Novel Concept for Chemical Sensing," Sensors and Actuators, B1(1–6):244–248, 1990.

Fabrication of a DNA-capture microchip

A primary goal of flow-through technologies is the elimination of off-line sample processing and multiple handling of the sample and reaction products. This can be of particular benefit in polymerase chain reaction (PCR), where contamination is a significant concern.¹ For ease of use and to reduce the possibility of sample-to-sample contamination, nucleic acid probe assay systems using PCR will likely be based on disposable cartridges. Such cartridges combine low-cost plastics technology with selected silicon micromachined components for critical functions such as nucleic acid extraction.

Extraction, purification, and concentration of nucleic acids from complex biological samples are the first steps in any nucleic acid probe assay. These steps are necessary because the nucleic acid from any living organism is carefully protected inside the cell. Once the cell is chemically broken open (or lysed), its nucleic acid is released into the solution and is accessible. However, other biological molecules may also be present in the solution. Some of these molecules, such as proteins and metal complexes (for example, hemoglobin) bind with the nucleic acid and otherwise interfere with the PCR amplification reaction. This requires that the nucleic acid must be isolated while these inhibitors are washed away. This complex series of chemical processes is the most difficult and time-consuming part of nucleic acid probe assays.

A typical current protocol for DNA purification involves a number of steps whereby various reagents are added to the sample, and the sample is centrifuged to separate precipitated components from the solution. Such a procedure is quite involved and, in many cases, is still done manually. Many nucleic acid purification kits are available commercially (Biorad Labs, Hercules, CA; Promega, Madison, WI; and Qiagen Inc., Santa Clarita, CA), each for a particular DNA or RNA and input sample. As a capture medium, these kits use a glass matrix or membrane or a silica gel because nucleic acids bind to glass or other silica-type surfaces under the proper chemical conditions.²

The authors have fabricated DNA capture surfaces of oxidized silicon using deep reactive ion etching (DRIE), a technology that is ideal for creating high surface area structures.^3,4 This process is simpler than competing technologies such as LIGA.⁵ A continuous flow of a fluid biological sample over the surfaces of these chips results in a gradual accumulation of nucleic acid on the large silicon dioxide surface area. We anticipate using these microchips in flow-through cartridges capable of reagent containment, mixing, and delivery for rapidly processing clinically realistic sample volumes (sometimes several ml) while still maintaining small internal fluid volumes. Our broader goal is the development of an automated microdiagnostic system capable of taking a raw sample input and giving a PCR result in a few minutes.

Figure 1. Process flow for the fabrication of high-aspect-ratio DNA capture surfaces in silicon.

Figure 1 shows the process used to produce our DNA-capture microchips. Fabrication starts with a 100-mm silicon wafer, double-side polished, approximately 400-µm thick. The wafer is first oxidized to produce a thin pad oxide and then coated with silicon nitride. The back side is patterned and etched in KOH to produce fluidic ports that are 200-µm deep. The front side is next patterned with the DRIE mask and etched in a plasma etcher to produce a field of up to 5000 pillars, each 200-µm high. At the end of this etch, the front-side pattern meets the back-side ports. The wafer is further oxidized to coat the internal surfaces with silicon dioxide. The silicon nitride prevents additional oxidation of the top surface. The nitride and underlying pad oxide are now removed by plasma and wet etching to produce a bare silicon upper surface. A cover of Pyrex is anodically bonded to the wafer to finish the process. The wafer is then sawed into square die, 3.85 mm on a side.

Figure 2. Scanning electron micrograph of DNA-capture chip surface. The pillars are 200- µm high with a pitch of 34 µm.

Figure 2 shows a scanning electron micrograph of one chip surface consisting of 200-µm high columns, each about 20 µm in diameter, with a pitch of 34 µm. After anodic bonding to the top cover, the chip has a total internal surface area of 36 mm² in an "active" area of 3.5 mm².

We tested these DNA-capture microchips using both low concentration (at or below 105 copies of target DNA) and high concentration (on the order of 100–1000 ng/ml) solutions. These two regimes have relevance to different clinical situations. Samples used in the diagnosis of infectious diseases often contain a very small amount of pathogenic nucleic acid of interest. On the other hand, samples for testing genomic DNA, such as might be present in a sample containing lysed white blood cells, typically have large concentrations of DNA.

Both the low- and high-concentration solutions were obtained by diluting high-concentration DNA stock solutions with a glass-binding solution from a commercially available DNA purification kit. This binding solution is a chaotropic (protein denaturing) salt solution, intended to both denature proteins that might be present in a real clinical sample and produce conditions that allow binding to glass. We took wash and elution reagents from the same commercially available kits. Wash solutions are typically ethanol-based with several other ionic components, and elution solutions are either TE (10 mm Tris-HCl and 1 mm EDTA) or water.

In the standard protocol used, the DNA solution was passed through the chip, during which time DNA present in the solution became bound to the silicon dioxide–coated microstructure. The internal volume of the chip is about 0.2 ml, leading to a residence time of 200 ms at a flow rate of 1.0 ml/sec. The chip was then flushed with wash solution to eliminate the salts and other PCR inhibitors that might be present in the sample. Finally, elution reagent was passed through the chip, releasing the DNA back into the fluid stream. The eluant was collected in small aliquots for analysis.

For high concentration studies, the quantities of DNA are large enough that fluorescence techniques can be used to detect the DNA. For low concentrations, however, PCR must be used to first amplify a target sequence, and only then can the presence of the target DNA be verified, usually by gel elecrophoresis. Quantitation of the PCR product is more difficult, and the use of calibration standards run through PCR in parallel is mandatory.

For detecting the products of elutions of high-concentration DNA, we used a Carl Zeiss fluorescence microscope in conjunction with a photon-counting photomultiplier tube and software from Photon Technologies, Inc. (Monmouth Junction, NJ). The microscope was fitted with a filter set suitable for fluorescein-labeled DNA that allows excitation of the sample at 490 nm and detection at 515 nm and beyond. We processed the high-concentration samples using a range of DNA concentrations (100–1000 ng/ml), a range of DNA flow rates (0.1–5.0 ml/sec), a range of wash flow rates (0.5–5.0 ml/sec), and a single-elution flow rate (0.5 ml/sec).

For low-concentration studies, we processed chip output through a standard PCR protocol using a Perkin Elmer 9700 thermal cycler. We then processed the PCR product through gel electrophoresis, typically with 1% agarose gels and ethidium bromide staining. Typically, 500 ml of the input test solution was passed through the chip at 0.5 ml/sec, followed by 250 ml of wash solution.

The total binding capacity of high-surface-area chips was investigated using an input of 400 ng of DNA (400 ml of a 1000-ng/ml solution) followed by 400 ml of wash and then elution. All flow rates were 0.5 ml/sec. By comparing each elution signal to the standards and summing them we determined that 11–12 ng of DNA was captured, then eluted, from the chip. The maximum binding capacity of glass has been reported to be approximately 40 ng/cm².⁶ Since the internal surface area of the chip is approximately 0.36 cm², it is evident that an amount of DNA consistent with the maximum binding capacity of glass was captured in this experiment.

We determined that DNA flow rates higher than 0.5 ml/sec led to reduced capture efficiency, but wash flow rates could be increased up to 5.0 ml/sec without loss of efficiency. We obtained extraction efficiencies of about 50% and concentration factors of about 10 using the optimal conditions. These results, which are comparable to results obtained with commercially available kits, demonstrate the viability of using such microchips as an element in a miniaturized, flow-through diagnostic system.

Back to main article

References

1. Saiki RK, Scharf S, Faloona F, et al., "Enzymatic Amplification of b-globin Genomic Sequences and Restriction Site Analysis for Diagnosis of Sickle-Cell Anemia," Sci, 230:1350–1354, 1985.

2. Boom WR, Adriaanes HMA, Kievits T, et al., 1993, Process for isolating nucleic acid, U.S. Pat. 5,234,809.

3. Bhardwaj JK, and Ashraf H, "Advanced Silicon Etching Using High-Density Plasmas," in Proceedings of the Society of Photo-Optical Instrumentation Engineers, Micromachining and Fabrication Technology, vol 2639, Bellingham, WA, pp 224–233, 1995.

4. van Driënhuizen BP, Maluf NI, Opris E, et al., "Force-Balanced Accelerometer with mG Resolution, Fabricated Using Silicon Fusion Bonding and Deep Reactive Ion Etching," in Technical Digest of the 1997 International Conference on Solid-State Sensors and Actuators, Transducers '97, Chicago, pp 1229–1230, 1997.

5. Marques C, Desta YM, Rogers J, et al., "Fabrication of High-Aspect-Ratio Microstructures on Planar and Nonplanar Surfaces Using a Modified LIGA Process," J MEMS, 6(4):329–336, 1997.

6. Vogelstein B, and Gillespie D, "Preparative and Analytical Purification of DNA from Agarose," Proc Natl Acad Sci, 76(2):615–619, 1979.

Kurt Petersen, PhD, is president and chief operating officer; William McMillan is director of biotechnology; Gregory Kovacs, MD, PhD, is head of the science advisory board; Allen Northrup, PhD, is vice president and chief technical officer; Lee Christel, PhD, is director of MEMS technology; and Farzad Pourahmadi, PhD, is director of microfluidics for Cepheid (Sunnyvale, CA).