Diagnosing Similar Diseases in Low Resource Settings

Diagnostics in Low-Resource Settings

A lot of excitement surrounding microfluidics has been about its promising use in diagnosis in low-resource settings. Many infectious diseases present in developing countries are manageable or treatable with available medications, but still account for 1/3 of deaths. In these areas, multiple diseases present similar symptoms, leading to misdiagnosis and thus incorrect treatment. Hundreds of blood-based microfluidic immunoassays are available for diagnostic purposes, but they’re not all created equally. They require varying levels of sample processing or analysis that prohibit their deployment in low-resource settings. Further, while some diseases may have similar symptoms, they might require different detection techniques, with varying sample volumes, reagents and processing time, making it difficult to detect multiple diseases within the same system. This is the focus of recent work from Paul Yager of University of Washington. In his Lab on a Chip paper, “Progress toward multiplexed sample-to-result detection in low resource settings using microfluidic immunoassay cards,” he and his colleagues develop a system to detect both Typhoid fever and malaria.

Malaria and Typhoid Dual Detection

The proposed card is compact and intended to integrate with the DxBox. All sample processing occurs on-card including IgG filtration. This also depicts the porous nitrocellulose membrane of the FMIA which provides high assay surface area.

The proposed card is compact and intended to integrate with the DxBox. All sample processing occurs on-card including IgG filtration. This also depicts the porous nitrocellulose membrane of the FMIA which provides high assay surface area.

The developed system is intended to integrate with the DxBox, an ongoing project focused on a point-of-care diagnostic device. As I mentioned before, different diseases might require different means of detection. In this case, the researchers decided to detect antigens generated by malaria parasites and IgM antibodies generated by the host in response to the bacteria responsible for typhoid (Salmonella Typhi). The microfluidic card is based on a flow-through membrane immunoassay (FMIA) composed primarily of nitrocellulose, instead of traditional microfluidic channels. Nitrocellulose is essentially paper and provides a lot of surface area, creating shorter assay times. Enzyme-linked immunosorbent assays (ELISA) are standard lab assays and will be replicated using FMIA. However, ELISA can be slow (more than 3 hours) due to the diffusion between the bulk fluid and the capture service, while the FMIA can perform the same task in half an hour due to its high surface area.

Immunoassay Process

The system processes blood from the same sample in different, parallel steps to test for malaria and typhoid.

The system processes blood from the same sample in different, parallel steps to test for malaria and typhoid.

The detections of both analytes are run in parallel and start with the same unfiltered blood. The card extracts the plasma with a filter, eliminating whole blood cells, and this is where the assays for malaria and typhoid diverge. The typhoid assay must filter out any IgG antibodies (which would cause false positives when testing for IgM) and dilute the sample further. This results in a four-fold increase in the sample volume used in the malaria segment. Each analyte is then captured by immobilized reagents and labeled with gold nanoparticles conjugated to antibodies. The entire process is driven by pneumatic pressure and valves. Pneumatics is cheaper than alternatives, plus it doesn’t dilute the sample with an additional liquid, but it comes at the cost of introduced air bubbles. Air vents were incorporated to eliminate bubbles, but they were not totally eradicated and still obstructed the image analysis sometimes. Within the DxBox, analysis is intended to be carried out by a webcam. However, the current design of the system created nonuniform lighting (which can be rectified), and a flatbed scanner was used instead.

Results

This microfluidic card was tested on blood samples with Typhoid or malaria. Unfortunately the researchers did not test on a large enough sample to evaluate clinical utility or determine a limit of detection for the card. Currently lab-based ELISA has a limit of detection near 4 ng/mL, which is clinically relevant. The researchers also ran each sample on ELISA and a bench-top FMIA in addition to the on-card FMIA. Comparing the quantified signal of the on-card FMIA to ELISA resulted in an R2 value of 0.73, and on-card FMIA vs bench-top FMIA had an R2 value of 0.92. These are fine results that demonstrate how closely the on-card FMIA follows the bench-top methods, but it would mean a whole lot more given a limit of detection.

Discussion

The results of this card design seem promising but will mean a lot more with more testing. The pneumatic actuation was a major hindrance to project success. While they could operate at different pressures, the actuators were unable to actually control the liquid velocity. Also, the pneumatics introduced bubbles into the card, which not only affected the assay process but the final image to be analyzed as well. While only two diseases were showcased here, the authors have indicated that there is already work to create a more complete fever symptom panel. They also acknowledged that this format could be applied to other panels aimed at diarrheal diseases and sexually transmitted diseases as well. This format could really be adapted for a variety of diseases, with the disease diagnosis as the limiting factor for card design. 

Reference:

ResearchBlogging.org

Lafleur, L., Stevens, D., McKenzie, K., Ramachandran, S., Spicar-Mihalic, P., Singhal, M., Arjyal, A., Osborn, J., Kauffman, P., Yager, P., & Lutz, B. (2012). Progress toward multiplexed sample-to-result detection in low resource settings using microfluidic immunoassay cards Lab on a Chip, 12 (6) DOI: 10.1039/C2LC20751F

Bloodhound Beads Sniffing out Heart Attacks

How would you detect a heart attack? There are some symptoms that might tell you that you are very likely having a heart attack. Although you might feel pain in the chest, shortness of breath or other known physical symptoms, that doesn’t mean you in are actually having one. Conversely, you may not experience these symptoms but an attack is well on its way. In addition to painful symptoms, an electrocardiogram can be used to further indicate if you’re having a heart attack, but it also isn’t always accurate. But what if you could detect a heart attack by monitoring cardiac specific biomarkers in the blood or saliva? Those attempts are well underway.

Detecting anything in bodily fluids is always a good time for our friend microfluidics to enter the scene. Even after the correct biomarker is established, it must be quantified in a quick and demonstrative way. While biomarker levels vary in statistically significant ways during disease, they may still occur at low levels, making detection very hard. John T McDevitt et al. fulfill this need in “A disposable bio-nano-chip using agarose beads for high performance immunoassays.” I actually worked in Dr. McDevitt’s lab while I was at Rice University, so I’m pretty familiar with his work, although I didn’t work on this project.

Agarose Beads

In this paper, Dr. McDevitt develops a system of agarose beads for immunoassays that can be implemented in a microfluidic system and tests it on C-reactive protein (CRP). CRP is a biomarker that is elevated during inflammation or heart attacks, which could make it useful when detecting or confirming a heart attack when used in conjunction with other biomarkers. This process uses a sandwich immunoassay, which we’ve seen before when detecting ovarian cancer. However, in this case, 280 µm agarose beads are used to house the immunoassay process. These beads are very porous, providing a large surface area to volume ratio, which affords many antibody-capturing locations to be stationed on and within the bead. Other beads made of polystyrene would only allow capture antibodies on the surface, and would not be able to detect as much of the biomarker in the same volume.

Protein-sensing Microarray

Microarray fabrication following PDMS mold creation from silicon master

Microarray fabrication following PDMS mold creation from silicon master

While the porosity of the bead allows the detection to be enhanced, the shape of the bead is also important in this immunoassay. This system is comprised of an array of microscale PDMS cups, each containing an agarose bead. The cups are shaped like inverted pyramids and feature a small hole at the bottom that allows for fluid to drain. This setup drives fluid through the beads by convection, as it is the only exit point for the fluid. In its simplest form, the system contains three stacked layers: the middle layer contains the microarray of inverted pyramids, which is surrounded by two PDMS layers containing the microfluidic injection and drainage channels. The middle layer is made of an epoxy, and actually originates from a lithography-produced silicon mold. The silicon mold looks exactly the same as the epoxy, but due to the cost and process to produce the silicon, it doesn’t make sense to use it in each system. Instead, it is used as a mold for PDMS, which itself is used to mold epoxy, giving us a copy of the original silicon. The resulting epoxy layer can be incorporated by irreversible covalent bonding to the PDMS layers.

CRP Detection with Agarose Beads

A  CRP capture and flagging  B  CRP detection penetrating bead  C  CRP/fluorescence dose response curve

A CRP capture and flagging B CRP detection penetrating bead C CRP/fluorescence dose response curve

The agarose beads are first incubated with CRP specific capture antibodies, capturing the CRP as it is forced through the agarose beads. Detection antibodies are fluorescently tagged and are pumped through beads, specifically attaching to the anchored CRP. The fluorescence of the beads will correspond to CRP concentration in the sample. In order to calibrate the system properly, control beads are also included in the array which contain antibodies for another antigen. This system is able to produce a precise dose-response curve. Its limit of detection is about 1 ng/ml, which is far below the physiological range of CRP, but may be used while processing saliva, which must be diluted several times since it’s highly viscous. You can also see that the fluorescent green signal infiltrates the bead and is not limited to the surface. This demonstrates the porous agarose bead’s ability to capture more CRP and deliver a stronger signal.

Modelling the Microarray with Computational Fluid Dynamics

This system was also simulated using the computational fluid dynamics program COMSOL, and the results were similar to those found in the real world. Interestingly, there are some special considerations to think about when measuring the intensity of the beads. Not all flow through the array is equal. First, there is a bit of a pressure gradient across the beads such that the beads closest to the source experience the greatest pressure and those further away experience slightly lower pressure. The increased pressure causes more CRP and tagged antibodies through the beads resulting in a higher signal. Second, the size of the drain in each pyramid may vary, which would again affect the pressure experienced by each bead and its resulting intensity. Finally, any deviations in the bead shape will alter how it sits in the inverted pyramid. Once again, smaller beads sitting lower in the holes will experience higher pressures and signals.

Computational Fluid Dynamics simulates fluid flow and antigen capture

Computational Fluid Dynamics simulates fluid flow and antigen capture

Final Thoughts

This paper presents the final component in an immunoassay intended for microfluidic chips. While there are many other processes which must be carried out before the biomarker level can be measured, this method can also be used for a range of biomarkers and is only limited by the nature of the fluid and the concentration of the biomarker.

Reference:

ResearchBlogging.org

Du, N., Chou, J., Kulla, E., Floriano, P., Christodoulides, N., & McDevitt, J. (2011). A disposable bio-nano-chip using agarose beads for high performance immunoassays Biosensors and Bioelectronics, 28 (1), 251-256 DOI: 10.1016/j.bios.2011.07.027

Detecting Ovarian Cancer with a Cell Phone and a Microfluidic Chip

This post was chosen as an Editor's Selection for ResearchBlogging.org

Author's note: This post was chosen as an Editor's Selection at ResearchBlogging.org. Thanks for the support!

Ovarian Cancer

Ovarian cancer is the fifth leading cause of cancer related mortality among women. Like many diseases, there is a stark difference in survival rates depending on detection times. When ovarian cancer is detected at stage I, there is a 90% 5 year survival rate. Compare that with the 33% 5 year survival rate when the ovarian cancer is detected in stage III and IV. This disease is unfortunately asymptomatic at early stages, drastically eliminating the odds of discovery with enough time to make a difference.

While using traditional diagnostics like imaging, biopsy, and genetic analysis is impractical for regular screening, there are alternative methods used for women who are high-risk for ovarian cancer or who have family history. Transvaginal sonography can be used annually although it has been shown to have limited efficacy. Blood serum can also be tested to indicate ovarian cancer, but this method only has a sensitivity of 72% at specificity of 95%. Sensitivity and specificity are used to measure how well a system can detect something. To calculate specificity in our case, imagine 100 women without ovarian cancer are tested, and only 5 women are incorrectly told that they have ovarian cancer. This would undoubtedly be corrected in a follow up test. But to calculate sensitivity, imagine 100 women with ovarian cancer and 28 women are incorrectly told that they do not have it.

Not only are these tests inconclusive, they are extremely invasive. In the case of transvaginal sonography, an instrument is inserted in the vagina to check the ovaries. With blood serum testing, blood obviously must be drawn. Biochips currently exist to detect ovarian cancer based on protein biomarkers or DNA sequences, but these rely on fluorescence or chemiluminescence and are designed to be used in laboratory settings. None of the previous methods lend themselves to be used in point-of-care (POC) settings. An ideal POC device would not require expensive parts, be usable by limited trained personnel or be too complex. This would allow it to be used in resource-rich and resource-limited settings, especially if it does not need a continuous power source.

Detecting Ovarian Cancer with Urine

Researchers from Harvard Medical School have developed a cell phone system to detect ovarian cancer that should address the lacking areas of diagnosis so far. “Integration of cell phone imaging with microchip ELISA to detect ovarian cancer HE4 biomarker in urine at the point-of-care” was featured in the 2011 issue 11 of Lab on a Chip. Utkan Demirci et al. demonstrate a method to non-invasively detect ovarian cancer efficiently with urine and a cell phone. At the heart of this system is an enzyme-linked immunosorbent assay (ELISA). ELISA is a very common technique used in protein detection. In this case, a sandwich ELISA is used to detect the ovarian cancer biomarker Human epididymis protein 4 (HE4). Antibodies targeted to HE4 are conjugated to horseradish peroxidase which catalyzes a substrate and causes blue color to develop. We should then be able to ascertain the amount of HE4 originally in solution by quantifying the resulting color. This process takes place in three different microfluidic channels on a microchip the size of a stamp. These three channels allow a sample to be treated in triplicate or for many samples to be tested at once.

After the urine is loaded in the microfluidic chanel, ELISA is performed resulting in a colorimetric change

After the urine is loaded in the microfluidic chanel, ELISA is performed resulting in a colorimetric change

Cell Phone and CCD Imaging

Two methods were used to detect the change in color. The first method utilized a cell phone (more specifically Sony-Ericsson i790). This took advantage of the built in camera and processing power, allowing all processing steps to be carried out on the single device. The second method uses a lensless charge-coupled device (CCD). CCDs are found in digital cameras and have completely changed the way we capture images. In fact, the cell phone used has its own CCD inside. The CCD is used directly with a computer which analyzes the image with MATLAB. Both methods take a picture of the three microfluidic channels on the chip and compare the colors of the channels to previously measured standards.

Cell phone takes an image of ELISA results and compares the color to calibrated curves

Cell phone takes an image of ELISA results and compares the color to calibrated curves

Calibration and Testing

Before this system can be tested on actual samples, it has to be calibrated with known samples. HE4 was evaluated from 1,250 to 19.5 ng/mL, which was its detection limit. I’m unsure how much urine is actually needed. Each sample was diluted twenty times, and each channel can only handle 96.75 µL including the ELISA solutions. In order to make sure that ELISA was occurring correctly on the microchip, the colored solution was transferred to a 96-well microplate and the optical density was measured with a spectrophotometer. This was validated and a strong correlation between HE4 concentration and color was found for the CCD and cell phone with high R2 values above 0.90. After this calibration, the system was used to differentiate between the urine samples of 19 women with ovarian cancer and 20 women without ovarian cancer. The standard microplate technique and the cell phone and CCD methods were able to distinguish between the normal and cancer samples with statistical significance. When operating at a specificity of 90%, the cell phone and CCD tests achieved 89.5% and 84.2% sensitivity respectively. These results indicate that the new methods can efficiently and effectively detect ovarian cancer in urine.

Strengths

  • Both the CCD and cell phone methods demonstrated their ability to distinguish the difference between healthy and ovarian cancer urine.
  • These methods are extremely portable and can be used in a POC setting.
  • No complex machines or techniques are needed, which makes it cost-effective and allows operation by minimally trained personnel.
  • The low price of these tests makes them more accessible to be used to annually screen high-risk women or to check the efficacy of treatment.
  • Urine is an attractive diagnostic fluid because it is non-invasive and not intimidating.
  • It is unclear how well this could work in early detection at stage I of ovarian cancer because the samples used had later stage cancer. It is possible that the current configuration may not be able to differentiate between normal and early stage if HE4 levels vary between stages.
  • This test could be applied to other diseases with established biomarkers and sandwich ELISAs.

Further Development

  • It is unclear how well this could work in early detection at stage I of ovarian cancer because the samples used had later stage cancer. It is possible that the current configuration may not be able to differentiate between normal and early stage if HE4 levels vary between stages.
  • This test could be applied to other diseases with established biomarkers and sandwich ELISAs.

Reference:

ResearchBlogging.org

Wang, S., Zhao, X., Khimji, I., Akbas, R., Qiu, W., Edwards, D., Cramer, D., Ye, B., & Demirci, U. (2011). Integration of cell phone imaging with microchip ELISA to detect ovarian cancer HE4 biomarker in urine at the point-of-care Lab on a Chip, 11 (20) DOI: 10.1039/C1LC20479C

SIMBAS, Everything the Blood Touches Is Our Kingdom

SIMBAS_Lab_on_a_Chip_Cover.jpg

Hey, how’s your biotin? What? No it’s not an organic metal, maybe you call it B7? You’re probably fine, but have you been depressed, lethargic or losing your hair lately? Biotin is pretty important; it’s necessary for metabolism within our cells, so I make sure I never leave home without it. It’s rare for someone to have a biotin deficiency, but if you want to know your levels, give me a drop of your blood, and I’ll have an answer from you in 10 minutes. How? Oh just my self-powered integrated microfluidic blood analysis system (but I like to call it SIMBAS for short).

The SIMBAS emerged from Berkeley and was featured on the inside cover of the 2011 Issue 5 of Lab on a Chip. Luke Lee et al. describe a device capable of picomolar detection in “Stand-alone self-powered integrated microfluidic blood analysis system (SIMBAS).” The device can analyze whole blood without a lot of bells and whistles to get in the way. I’m not saying that this doesn’t have a clever design; it just doesn’t have a battery, moving parts or specialized readers. All you need is some blood and a microscope. The device has five independent analysis streams. There are three parts to a stream, the filter, detector and the suction chamber. All of these components are featured in a slab of PDMS sandwiched between two normal glass slides. Before I describe how the features physically work together, it’s important to note the device’s driving force: low pressure. Before use, the SIMBAS is placed in a low-pressure vacuum, degassing it through the single points of entry of each stream, creating a vacuum. When blood is placed at the inlet the vacuum slowly draws it through the system.

Filtration

The degassed chip draws in blood towards the dead-end suction chambers. The wells filter the blood before passing the detection strips.

The degassed chip draws in blood towards the dead-end suction chambers. The wells filter the blood before passing the detection strips.

After the vacuum sucks in the blood, the red and white blood cells must be filtered out. One major strength of this device is that it receives unadulterated blood. However, in order to prevent the cells from muddling the detection region of the device, they are filtered out by sedimentation. The floor of the 80 µm high, 50 µm wide sample channel opens up to a 2 mm wide, 2 mm deep circular well, which acts as a filter. The large, heavy cells sediment out and take up permanent residence at the bottom depending on the flow rate of the sample. The biomarkers of interest are much lighter and will continue through without joining the blood cells. This produces plasma that is 100% blood cell free, ready for detection. Platelets sediment at lower rates, so they will still remain in the plasma, but this doesn’t seem to be an issue.

Detection

Now that the sample has been fully prepared, it is ready for detection. Remember that the PDMS was placed between two glass slides? The authors immobilized streptavidin bars on the underside of the glass roof. This leaves the streptavidin attached to the ceiling, letting it bind with the biotin that flows by. Streptavidin and biotin have one of the strongest known protein-ligand bonds, making it a perfect choice for this device. When tested, blood was spiked with fluorescently-labeled biotin at 1.5 pM. Many other detectors could be immobilized along the same stream or the other four streams. The five parallel streams allow a way to eliminate errors, or test for biomarkers that are not compatible with each other.

Volume Control

However, we’re still missing the last important feature, the suction trough, an empty region after the detectors. The volume of the trough determines how much sample is drawn through the system. Once the trough is full, flow stops, regardless of how much blood is ready to enter. Modifying the trough volume provides a method to increase the sensitivity of the device. A larger sample volume gives the SIMBAS a better chance to detect a biomarker at low concentrations.

Final Thoughts

Detection of biotin at 1.5 pM, 150 pM, 15 nM & 1.5 µM

Detection of biotin at 1.5 pM, 150 pM, 15 nM & 1.5 µM

The current setup only needs 5 µL of whole blood for each stream. To put that into perspective, modern glucometers need 1 µL at most to glucose levels. It is (relatively) a bit more blood, but you wouldn’t be doing this every day. After 10 minutes, biotin at 1.5 pM can be detected by removing the top slide and looking at it under a microscope. That’s a pretty low detection in my book. Crazy low. I’m sitting here trying to think of an accurate needle in a haystack analogy, but it’s not coming. Overall, this is a pretty innovative, yet simple device, and I’ll tell you what I think of its merits and things I’d like to see developed.

Merits:

  • To start, this is has a great design for a point-of-care device, especially in a resource-poor setting. It has no external or moving parts, and requires no power. Sometimes in microfluidics, things can get very complicated with the number of channels, pumps, reagents etc., but this has a very clean and trouble-resistant design. It can be pre-packaged under low pressure so that a user only has to open it to activate the vacuum and use it within a couple of minutes.
  • The fact that it can receive whole blood also makes it great for point-of-care. Some lab-on-a-chips actually depend on many sample preparation steps or external machines. But all the steps that are needed for it to do its job in isolation should be included, just like SIMBAS. The time between filtration and detection is pretty quick, which is important because the proteins found in the plasma change after longer separations.
  • This doesn’t require much sample, and is still able to detect biotin at the low concentration of 1.5 pM. I’m not sure how clinically relevant that number is for biotin, but if you can detect that, you can play with the configuration to bring that down (or raise it much more easily).

What I’d like to see:

  • The authors state that the biotin-streptavidin detection could be replaced by many other couples. I don’t doubt this, but I would like to see what detection levels they could achieve, since biotin-streptavidin has one of the strongest protein-ligand bonds.
  • In the paper, the authors mention that the detection strips coated on the ceiling were applied on the same day the assay was run. I’d like to know what the shelf life of a prepared card would be, which could really impact their usefulness and value.
  • As I mentioned previously, the authors detected fluorescent-tagged biotin in the blood, and examined this under a microscope. You can see the captured biotin fluorescing in their figure. But biotin isn’t naturally tagged with a fluorophore, which makes me wonder how they would normally detect a wild biomarker. Perhaps there is a noticeable difference under the microscope; otherwise they will need to introduce some tag in the blood that will act as a secondary binding agent.
  • The last thing I’d like to see is the microscope. Well, I’d like to see it removed. This microfluidics system would be even greater if it was completely stand-alone. Daniel Fletcher (also from UC Berkeley) has developed a cell phone-powered field microscope capable of fluorescence microscopy, so it isn’t impossible in a resource-poor setting, but it’s just my wishful thinking.

Reference:

ResearchBlogging.org

Dimov, I., Basabe-Desmonts, L., Garcia-Cordero, J., Ross, B., Ricco, A., & Lee, L. (2011). Stand-alone self-powered integrated microfluidic blood analysis system (SIMBAS) Lab on a Chip, 11 (5) DOI: 10.1039/C0LC00403K