Point of Prevention | Inside Precision Medicine

Point-of-care (POC) testing and diagnostics for diseases or other medical conditions are common. They include at-home testing such as glucose monitors, pregnancy tests, and the COVID-19 rapid antigen test. In a clinical setting, doctors have at their disposal quick tests for strep throat, prothrombin time tests to monitor blood clotting for patients taking the anticoagulant warfarin, hemoglobin tests to check for anemia or other blood disorders, and cardiac marker tests that can quickly determine whether a heart attack has occurred during an emergency.

Some key characteristics of POC tests are that they provide an answer within minutes or hours, not days, are easy to use with little or no training or expertise, and are inexpensive. Quick answers, such as whether a person has a high-risk infection, allow clinicians to start treatments as soon as possible to avoid more severe symptoms and disease spread.

Although POC is ubiquitous for many conditions, available options for cancer diagnosis or screening are limited. Risk screening for prostate cancer, breast cancer, and colorectal cancer are readily available in the developed world, but are not viable in a POC setting. Mammography for diagnostic purposes and colonoscopy can provide same-day results, but require expensive equipment and trained personnel. In contrast, a prostate-specific antigen test for prostate cancer typically returns results within a few days.

Yet, the need for POC testing is pressing. It is well known that treating cancer at earlier stages leads to better outcomes. Starting treatment within 24 hours of diagnosis, as opposed to days or even weeks, would make a significant impact. Having easy-to-use, inexpensive POC testing options would be especially important in rural and low-resource settings, where availability of care is limited.

“If there was a test for something like cervical cancer in these settings with limited infrastructure that could provide a diagnosis on the spot, people could begin treatment right away,” said Casey Howard, associate director of technology development at the Center for Innovation and Translation of POC Technologies for Expanded Cancer Care Access (CITEC), a multi-country collaboration of bioengineers, oncologists, and international global health partners created to foster the development of POC cancer diagnostics. “It’s especially important because even with just a screening test, the needed follow-up for a diagnostic often doesn’t happen.”

For POC diagnostics for low-resource settings, the World Health Organization encourages focus on its ASSURED (affordable, sensitive, specific, user-friendly, rapid and robust, equipment-free, delivered) criteria to help guide diagnostic development. When viewed through this lens, it is not surprising that researchers are examining paper-based substrates for flow tests, photonics, and optical imaging to identify tagged cancer biomarkers and develop POC cancer diagnostics.

Counting on photonics

One promising technology that can be incorporated into a POC cancer diagnostic is being developed at the University of Illinois Urbana-Champaign, in the lab of Brian Cunningham, PhD, a professor of electrical and computer engineering. Called LOCA-PRAM, the biosensing technology combines photonic resonator absorption microscopy (PRAM) for imaging with the localization with context aware (LOCA) algorithm. This combination of photonics and artificial intelligence detects and counts individual molecules of relevant biomarkers like RNA, DNA, and proteins.

PRAM imaging is based on the interaction between light and a nanostructured surface called a photonic crystal (PC) biosensor. It can detect molecules that have been tagged with gold nanoparticles (AuNPs). When a patient sample derived from saliva, blood, or tissue from a cheek swab is introduced to the PC, the targeted molecules are labeled with AuNPs. When the biosensor is lit from below with a low-intensity LED, it produces a bright-field image where the tagged biomarkers appear as distinct black dots.

Earlier versions of PRAM had already shown high sensitivity for detecting a range of biomarkers like microRNAs, cytokines, and cancer-specific DNA. This method of detection sets PRAM apart from other methods and positions it for POC development. “PRAM enables highly sensitive and selective digital-resolution detection and quantification without the need for droplet partitioning or enzymatic amplification,” Cunningham noted.

The captured PRAM images are analyzed by the LOCA algorithm, a deep learning model that was trained by Han Lee, a graduate researcher working in the Cunningham lab.

One of the challenges of adapting PRAM as a potential diagnostic tool finding a method to accurately count the AuNP-tagged molecules captured on the PC surface as these counts map directly to biomarker concentration and disease burden. To derive accurate counts, Lee used images from a single point-spread function (PFS), then registered PRAM images with scanning electron microscope images to isolate a true single-particle signal and treat that as the ground-truth PSF. He then used the PSF to build training sets, with the objective of predicting both the number of AuNPs and their sub-pixel locations.

“This explicit localization step is what prevents undercounting. When two particles are close enough to look like one blur, the model still resolves two distinct peaks and credits two counts. We also include ‘hard negatives’ such as scratches, dust, and other artifacts, so the network learns to ignore non-particle features,” Lee said. “The result is robust, single event-level detection and counting across real-world noise and artifacts, which improves quantitative accuracy and could effectively lower the limit of detection.”

Translating the technology to POC

For a clinical POC application, Cunningham says a patient would provide a sample via an oral rinse, swab, or fingerstick of blood. The sample could then be injected into a disposable microfluidic cartridge loaded with the necessary reagents and the AuNPs to tag the targeted biomarkers—DNA, RNA, proteins, or antibodies. “We envision the cartridge then being inserted into a machine that actuates the cartridge and incorporates a simple PRAM detection instrument,” said Cunningham, who estimates that the entire process, from loading the cartridge to delivery of results, could be completed in less than 30 minutes.

The potential of LOCA-PRAM for POC use is further heightened by its makeup, which includes inexpensive optical components such as LEDs, image sensors, and photonic surfaces that could produce a compact PRAM device.

LOCA-PRAM is currently being tested in projects for the detection of prostate cancer, lung cancer, and liver cancer, and could have even broader applications. “We are thinking of LOCA-PRAM as being useful in several cancer-related diagnostic tests that include biomarkers to guide therapy selection, therapy effectiveness monitoring, and remission monitoring,” Cunningham said.

PRAM and its related assays have been patented with an eye toward commercialization. In the meantime, ongoing work in the Cunningham lab is focused on optimizing the microfluidic cartridge and validating the LOCA algorithm using larger clinical datasets.

A POC sandwich

Another biosensing technology with potential as a POC diagnostic is being developed at the Hong Kong Polytechnic University. The biosensor is designed to detect tumor markers like carcinoembryonic antigen (CEA), which is a glycoprotein associated with malignancies like colorectal, pancreatic, gastric, lung, breast, ovarian, and medullary thyroid cancers. The sandwich structure immunoassay combines luminescent quantum dots, a microfluidic biochip, and a machine vision algorithm to create a portable, noninvasive POC diagnostic system. Using a saliva sample, the platform demonstrated a detection limit of about 0.021 ng/mL, outperforming commercial lateral flow assay strips, according to the researchers.

Within the biochip are microchannels that concentrate and separate biomarkers from the saliva sample. Any CEA proteins present in the sample are bound and tagged by two antibodies which are identified when exposed to ultraviolet light. The signals are then captured and analyzed by a machine vision algorithm using a smart phone as the interface. This method produces quantitative real-time detection of tumor biomarkers without the need for blood sampling.

“The portable microfluidic biochip is mainly used for targeted biomarkers for improving the detection sensitivity,” said Yuan Liu, PhD, an assistant professor in the Department of Applied Physics at The Hong Kong Polytechnic University. “The simple machine vision algorithm is utilized for advancing the integrability and accessibility of the biosensing platform.”

Development of this method was led by Jianhua Hao, PhD, director of the research center for nanoscience and nanotechnology at Hong Kong Polytechnic, and was informed by the limitations of applying existing cancer diagnostics in a POC setting. While radiology and tissue biopsy remain standard tools for cancer diagnosis, they are not available in most limited resource settings. Other diagnostic methods targeting the identification of proteins, such as ELISA, provide high specificity for detection but typically take a day to deliver results. Lateral flow assays provide convenience and a low cost, but often lack sensitivity, particularly when tumor marker concentrations are low in non-blood samples like saliva. Recognizing the lack of POC options, Hao and team noted that “point-of-care tumor markers’ biodetection with good convenience and high sensitivity possesses great significance.”

To test their method and confirm the device’s sensitivity and accuracy, the team used human-sourced saliva samples spiked with known concentrations of CEA. They then mixed the samples with quantum dot–antibody conjugates, introduced the mixtures into the chip, and captured the luminescent responses. Their results demonstrated improved specificity compared with commercial lateral flow assay strips, which further indicated the system’s potential for use both in the clinic and potentially for home-based cancer screening.

To interpret the optical data, the team developed a Python-based machine vision algorithm that processes captured images of the biochip and extracts quantitative information from the luminescent patterns. Liu said the integration of hardware and software was central to the system’s development. “In my opinion, this designed biosensing platform, to a certain extent, combines the ‘hardware’ (such as the quantum dots’ luminescence, microfluidic biochip, related electronic components, and so on) and the ‘software’ (such as the simple machine vision algorithm) for achieving the intelligence and convenience of POC cancer screening and diagnostics.”

Hao noted that saliva-based testing offers an important advantage for POC use, eliminating the risks associated with blood collection. The system’s small size and integration with smartphones make it adaptable for use in community clinics, mobile screening units, and resource-limited settings. “Combining these infusive abilities, our elaborate biosensing platform is expected to exhibit potential applications for the future point-of-care tumor markers diagnostic area.”

The role of smartphones

As with the technology for CEA detection, the role of smartphones for POC diagnostics is growing. For instance, paper-based analytical devices (PADs) can provide low-cost, portable, and user-friendly alternatives to technologies that are impractical for POC, like liquid chromatography, mass spectrometry, and capillary electrophoresis, due to their bulk and associated high costs.

When combined with smartphones, PADs belie their humble paper origins by leveraging inherent on-board advanced technologies such as high-resolution cameras, wireless communication, and app-controlled processing capabilities to capture, analyze, and transmit test results in real time. Paired with smartphones, PADs can conduct rapid POC colorimetric or fluorescence-based detection of cancer biomarkers, allowing clinicians in resource-limited settings to access important diagnostic insights without the need for a central laboratory.

In some instances, smartphone-read PADs can identify cancer biomarkers like proteins or enzymes linked to tumor presence or progression, and are particularly adept at reading optical images to recognize changes in color intensity or fluorescence that can indicate the presence of disease. This capability is poised to be increasingly important for the development of future POC cancer testing methods.

“Optical technologies have become powerful for POC diagnostics for a combination of reasons. The key optical components, such as image sensors, lasers, LED, and optical fibers have become inexpensively manufacturable (and are) small, while providing high performance in terms of intensity of the sources and sensitivity of the sensors,” stated Cunningham. “Meanwhile, many of the key technologies in biology research and diagnostics utilize light-emitting and light-absorbing reporters, which include gold particles, fluorescent dyes, quantum dots, and many others, all of which interface well with target molecules that are nearly the same size. The resulting combination of capabilities results in very high detection and quantitative accuracy that is needed for in vitro diagnostic detection.”