Feature Article

This article charts the important design steps involved in creating a single device on which are integrated all the analytical process steps necessary for diagnosis. Two examples of innovative devices of this type are given, which are taking the microfluidics sector closer to commercialisation of this long-held goal.

FIGURE 1: Schematic diagram of the diagnostic process flow in a microfluidic device.

In the first step, the sample has to be brought onto the device through some interface. Because the type of sample can be different, for example, biopsy, swab, sputum or blood, this interface has to be adapted to the type of sample to safeguard efficient sample transfer to the device and ensure the absence of any contamination of the sample or infection risk to the operator. These “world-to-chip” interfaces are still often overlooked, but important, items during the development of micro-fluidic systems The use of existing standards from the targeted application area, for example, Luer-Lock compatible interfaces in clinical diagnostics, is increasingly being established; however, this involves disadvantages, mainly in terms of size. No interface that is dedicated to minimising the footprint has so far been successfully implemented.

FIGURE 2: Simulation of cell trajectories in cell assembly chip. (Image courtesy of BioMEMS and Sensors group, NMI, Tübingen, Germany)

The next step, that is, the various sample preparation processes such as liquefaction of the sample, the lysis of cells, extraction of deoxyribonucleic acid (DNA)/ribonucleic acid (RNA) or the sample concentration have until now typically been performed off-chip because of their complexity and the different nature of the various samples. Moving these steps onto a device represents the biggest challenge. This is mainly because usually several media (wash buffer, carrier buffer, beads and lysing agents) have to be handled sequentially as well as in parallel, which requires interfaces and plumbing in restricted device areas. Furthermore, many of these processes must be performed with high precision in terms of volume, times or sequence, and these are much easier to achieve with specialised, but often costly, laboratory equipment. Therefore, in the development of miniaturised assays that are to be performed on-chip, it is a specific requirement that the assay is as robust as possible in terms, for example, of process steps, volume control and timing. To illustrate the point, a volumetric precision of ±5% of, for example, 1 µL volume requires manufacturing precision to produce a microchannel with dimensions of 100 × 100 µm of ±2% in each dimension, a factor that directly relates to the manufacturing cost of the device. Thus, a more robust assay with reduced process precision requirements also helps to keep the fabrication costs of the devices lower. In recent years, several groups have demonstrated the integration of these sample preparation steps on-chip.^3,4 For the development of an integrated device of this type, a two-pronged approach has proven to be advisable. First, a holistic top-down strategy from the system level is necessary to ensure the inclusion of all the necessary functions as well as the definition of all interfaces, including fluidic, mechanic and optic functions. A flow diagram of all the process steps performed on the device can then be translated into individual functional modules. Second, is the development, for example, by simulation and subsequent prototyping, of the individual module such as a DNA extraction chamber or a mixing structure for the lysis buffer in which the individual functions can be validated before integration.

FIGURE 3: Module test platform with integrated membrane for DNA extraction.

There is one significant difference between microfluidics and other engineering disciplines, particularly mechanical and electrical engineering. In those fields, the development of individual modules tends to be simpler, because the mutual interactions of the individual modules are more limited and often calculable with simple restraints. This allows the assembly of module libraries that can be transferred from one development case to another. In microelectronics, an operation amplifier or a storage capacitor will behave (almost) identically regardless of the overall system layout. In microfluidics however, the performance of a single module is often to a large extent dependent on the overall system layout. A typical example would be the parameter of flow speed in a microfluidic module such as a simple T-shaped microchannel. This flow speed can be easily calculated given the dimensions of the various arms of the channel. However, because the flow speed depends (amongst other things) on the back-pressure that the different arms of the T are experiencing, the flow distribution changes depending on the back-pressures generated by pre- or succeeding modules. If this happens in a time series, the functional description of a simple module of this type can become difficult. It is, therefore, emerging as best practice in development to combine the theoretical (or modelling) approach (Figure 2 shows an example) with some experimental data from module prototypes, which are coupled together to approach the complexity of the fully integrated device. This stepwise approach also simplifies the search for, and correction of, possible errors observed in the performance of the device. Figure 3 shows this type of module test platform, in this case a membrane chip for the extraction of DNA from a sample through this membrane. Different chamber volumes, membrane areas and channel dimensions are built on this device to map out the practical perform-ance space. Noting the dimension of the attached Luer connectors, the need for a universal microfluidic interface, as mentioned above, becomes apparent.

FIGURE 4: Fully integrated test chip for the development of a cancer diagnostic system (project SmartHEALTH, EU FP6-2004-ISTNMP-2-016817).

The next process step in devices using molecular biology methods usually involves amplification of target molecules, employing methods such as conventional or isothermal polymerase chain reaction (PCR) and rolling circle amplification to increase the number of target molecules and thereby achieve better detection selectivity and sensitivity. This amplification step is then frequently followed by a separation step such as electro-phoresis, chromatography (up to now not well developed on-chip), and the use of capture probes such as DNA arrays or other filtration mechanisms to isolate the desired component spatiotemporally or remove unwanted components from the mixture.

FIGURE 5: Assay development chip using centrifugal forces for full-blood analysis (project ZentriLab, BMBF contract 16 SV2350).

An example of an integrated diagnostic device is shown in Figure 4. This was developed under the framework of the European Union funded project “SmartHEALTH,”⁵ which targets the early diagnosis of cancers such as cervical or colorectal. The image shows a development platform chip, which contains all the above mentioned process steps, including amplification and a detection area, where different types of sensors (for example, electrochemical) can be mounted on the chip and exposed to the sample through parallel microchannels.