Progress and challenges in biomarker enrichment for cancer early detection

Cancer is a disease based on the uncontrolled growth of tumor cells unleashed by mutations that incite abnormal tissue growth and disrupt bodily functions [1]. In the United States, more than 1.9 million new cancer cases are expected to be diagnosed in 2021, and about 608 570 deaths per year are predicted due to cancer, corresponding to 1671 deaths per day [2]. Detection at the earlier stages of the disease is more likely to lead to effective treatment and a greater probability of longer survival for certain cancer types [3]. Therefore, it is of great importance to establish high statistical confidence in diagnostic accuracy and improve screening protocols for detecting cancer at early stages [4].

Cancer cells produce small quantities of biomarkers [5, 6] (markers that can be used to measure and evaluate normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention), such as extracellular vesicles (EVs) [7], circulating cancer-related nucleic acids [8], and proteins [9]. Further, these cancer cells can detach from the primary site, enter circulation, and metastasize [10] (i.e. circulating tumor cells—CTCs). We should note that EVs and CTCs can be standalone biomarkers; yet, their cargo, such as proteins and genetic materials, can themselves serve as useful biomarkers. The collection and analysis of such biomarkers can be employed to detect early signs of disease initiation and its progression thereafter [11]. However, a major obstacle in biomarker collection and enrichment is the low concentration of biomarkers that affects subsequent downstream analyses [12]. In this direction, novel methodologies have been developed to enrich and isolate biomarkers from biological matrices using a variety of technological approaches.

Solid tissue biopsy is considered a gold standard to verify disease diagnosis. However, it requires invasive surgical procedures. Meanwhile, liquid biopsy comprises the collection of diverse bodily fluids, such as blood, saliva, and urine. This clinical sample collection approach is minimally invasive and has been implemented into clinical workflows potentially for early cancer detection purposes [13–15]. The heterogeneous composition and low abundance of these biomarkers in biological samples can lead to potential errors in analyses; therefore, enrichment protocols could increase the sensing capabilities of analytical techniques.

Herein, we provide a perspective highlighting the challenges and opportunities in various enrichment methods. We thoroughly discuss each strategy to enrich biomarkers and their mechanisms of action, categorizing them as biochemical and biophysical methods (figure 1). Enrichment steps could reduce the probability of false-negative results by complementing sensitive detection methods and improving the sensitivity of the overall analysis. Building on the previous reviews, which have critically reviewed cancer early detection based on cancer type [16], biomarker type [17], and specific detection readout [18], this perspective provides a multidisciplinary approach emphasizing the isolation and enrichment of biomarkers reaching beyond the topic of CTCs [19]. Moreover, we outline and evaluate the challenges and opportunities in using isolation and enrichment methods for cancer early detection.

Zoom In Zoom Out Reset image size

Biochemical enrichment methods rely on biomarker moieties (such as antibodies, oligonucleotides, single-stranded RNA/DNA, or proteins) with specific affinity to the target molecules in order to isolate and enrich these targets for further downstream quantification [23–25]. The bioassays measure the output signals produced by a recognition event (e.g. the binding/hybridization) between a biomarker and a bioreceptor or a synthetic probe [26, 27]. Thus, biochemical enrichment protocols can use similar complementary capture bioreceptors to the bioassay sensing platforms so that they can be rapidly integrated with optical assays [28, 29], electrochemical immunoassays [30, 31], and label-based sensing such as flow cytometry [32, 33]. For example, colorimetric assays, such as enzyme-linked immunosorbent assay (ELISA), can be used as a confirmatory measure of biomarker enrichment [34]. ELISA uses bioreceptors to capture the target molecule onto the surface of the recognition area, enabling the removal of non-specific unbound molecules and a detectable signal upon the addition of an enzymatic substrate [35, 36]. This method offers ease of use but requires multiple washing steps (i.e. processing a large volume of liquid) [37]. A similar approach using microfluidics consisting of a functionalized surface with complementary antibodies can capture specific target biomarkers on the surface, while non-targeted cells and debris are washed away by passing a constant fluid flow (figure 2(a)) [20, 38–40]. In general, microfluidic-based methods enable a higher processing volume than the use of well plates, although the fluid flow could affect the binding of an analyte with a receptor due to strong shear and fluid forces that might reduce binding events.

Zoom In Zoom Out Reset image size

Recent advances have opened new avenues for on-chip ELISA systems, which can significantly lower the required assay time and sample volume using antibody-functionalized magnetic beads (figure 2(b)) [21, 41, 42]. These particles are mixed with biofluid and then bind to their specific targets; thus, enabling the recovery and collection of biomarkers using a permanent magnet [43, 44]. Advances in micro-robotics consisting of micro/nanostructures capable of converting local or external energy into mechanical energy and locomotion [45–47] have been used as 'on-the-fly' capture and isolation platforms of various biomarkers, including nucleic acid, proteins, and cancer cells from complex samples [48, 49]. The surface of the microrobot can be functionalized with a multitude of bio-affinity receptors, in which the microrobot's motion can help to increase the recognition events and enhance the fluid mixing owing to the constant movement of the functionalized microrobot compared to a static probe with the same surface functionalization [50–52]. After capture, the biomarkers can be recovered or indirectly quantified by measuring the changes in acceleration [53, 54] or optical properties of the microrobots (figure 2(c)) [22, 55].

In contrast, physical enrichment methods use the biophysical properties of specific biomarkers, such as size, density, and electric charge, to isolate these targets from biofluids [60]. These label-free approaches include the use of filtration, inertial forces, and externally applied force fields (e.g. ultrasound and electricity). Such methods enable the development of inexpensive and high-throughput isolation methodologies and minimize laborious sample preparation steps. The use of filtration for biomarker enrichment usually separates and isolates targeted biomarkers based on size and deformability [61]. For instance, some subgroups of CTCs are much larger than other blood cells, such as red blood cells or leukocytes [62]. Filtration methods use porous membranes and micropatterns integrated into microfluidic channels, which enable the separation of different components of the biofluids whereby smaller cells are passed through the filter. In contrast, the larger components of the samples, such as CTCs in this case, are retained by the filtration substrate [63–66]. Moreover, filtration methods have also been used to isolate EVs and/or exosomes from conditioned culture media and different bodily fluids [67, 68]. For example, the Exosome Total Isolation Chip (ExoTIC) reported a high isolation yield for exosomes from lung cancer cell lines and patient plasma, urine, and lavage samples. Some of the biomarker expressions were enriched when using ExoTIC with different pore-size filters compared to those using the common ultracentrifugation method (figure 3(a)) [56, 69, 70].

Zoom In Zoom Out Reset image size

Filtration-based enrichment methods can suffer from pore clogging; thus, optimizing the filter design and active area can improve the efficiency of these methods. In this direction, hydrodynamic optimization approaches can overcome clogging using shear flow gradients and centrifugal forces to sort biomarkers into specific outlet channels in a microfluidic chip (figure 3(b)) [57, 71–74].

Force fields generated by external sources (e.g. magnetic, acoustic, or electrical fields) have also been used to enrich biomarkers from biofluids enabling contactless and label-free isolation based on the biophysical properties of the biomarkers [75]. For example, magnetic levitation has been employed for isolating biomarkers, such as CTCs, without the need for labeling [76]. This approach relies on putting fluids with biomarkers in a low-magnetic-susceptibility paramagnetic fluid and a magnetic field created by permanent magnets [77], whereby the separation of healthy cells from cancer cells has been demonstrated as one of the applications [78]. The acoustic-based isolation relies on the movement of the targets toward the pressure nodes derived from the density difference between the target and the encompassing liquid medium [79]. Acoustic energy can be used to generate different types of acoustic separation methods [80, 81]. For instance, surface acoustic waves (SAWs) propagate traveling waves through a substrate that generates pressure nodes that push other biomarkers into different outlets within a microfluidic chamber (figure 3(c)) [58, 82, 83]. Moreover, the use of SAW can induce the formation of microstreaming traps using micro-posts capable of isolating CTCs [84, 85]. Focused laser beams have been used to create optical tweezers that can isolate individual cells from microfluidic chips [86–90]. The optical tweezer can sort cells into a predetermined location, relying on the difference in the refractive index between the biomarker (e.g. cells) and the surrounding fluid [91]. Electric fields have also been shown to induce positive or negative migration of biomarkers into collection electrodes, ranging from oligonucleotides to living cells (figure 3(d)) [59, 92, 93]. The isolation mechanism is commonly based on diffusiophoresis [94], which polarizes the biomarker molecules driving their migration toward the electrode with an opposite charge [95]. Although each method has intrinsic benefits, the combinatorial use of both biochemical and physical methods has improved the enrichment capability [96–98]. For instance, microfluidic pillars have been functionalized with biochemical labels to provide both selectivity and better flow-through [99–101]. Another combinatorial use of both physical and biochemical methods is the combination of filtration and antibody-functionalized magnetic beads [102], in which disease-related EVs can still be further enriched specifically following the first enrichment step using a filtration-based isolation method. Moreover, a paper-based electrophoretic bioassay has combined the widely used lateral flow technology with electrophoresis by attaching copper tapes on both sides of the strip. This setup has allowed specific detection of antibodies in whole blood using a biochemical label, whereby enrichment of the immunocomplexes can be achieved through a controlled electrophoretic flow [103]. Overall, the combination of biochemical and physical methods in biomarker enrichment has shown the potential to be translated into point-of-care diagnostic technologies.

In summary, we covered the use of enrichment methods for enhancing biomarker detection, with key differential features between biochemical and biophysical methods (table 1). Current technical developments in isolation and enrichment of biomarkers, such as CTCs, circulating tumor microemboli (CTM—clusters of CTCs), EVs, circulating tumor DNA (ctDNA), proteins, and antibodies, are critical for improving the reliability of early cancer diagnostics when direct sample collection does not provide sufficient biomarker quantity or quality to produce a readable signal using common measurement techniques due to their diluted nature in the biofluids.

Biochemical enrichment includes the use of high-affinity bioreceptors, such as antibodies, oligonucleotides, or proteins to capture target molecules; hence, isolation and enrichment of the target biomarkers are possible for a selective downstream quantification. These enrichment techniques include immunoaffinity-based platforms, microfluidics, magnetic bead separation, and self-propelled microrobots. In general, biochemical enrichment methods enable high specificity and selectivity; despite higher cost and lower throughput, they can still be used to isolate well-established and specific biomarker targets.

Physical enrichment uses the biophysical properties of target biomarkers, such as size, density, magnetic susceptibility [104], and electric charge. These methods include the use of filtration, inertial forces, and external fields. Such isolation methods can rapidly release the trapped biomarkers by removing the applied field or eluting the trapped biomarker target. Nevertheless, these methods are not as selective as biochemical label-based isolation since the disease-related components in the sample can have similar physical properties as the healthy ones; for example, this is evident in the overlapping size distribution of diseased versus healthy exosomes. The use of externally generated force fields for biomarker enrichment might require specialized instrumentation and trained users in a laboratory environment, although their miniaturization and commercialization can potentially increase their adoption as point-of-care devices.

Furthermore, most blood-based tests in the clinic use protein markers to detect cancer, such as CA125 (cancer antigen 125) for ovarian cancer, CA19-9 (cancer antigen 19–9) for pancreatic cancer, carcinoembryonic antigen for colon cancer, and prostate-specific antigen (PSA) for prostate cancer [105]. However, their diagnostic utility for cancer is limited due to their lack of specificity (e.g. elevated CA125 level is also observed with regards to pregnancy, cardiovascular, and liver diseases; similarly, PSA is prostate specific but not cancer specific) [106, 107]. Therefore, proteomic profiling can be beneficial for novel biomarker discovery, especially if the expression of such proteomes can be correlated to the genomic pathways [108]. For doing so, the collection of transcriptomes has also been of interest to study their diagnostic potential. A commonly used method for isolating nucleic acids is based on precipitation under different solvents or adsorption to solid-phase microcolumns [109]. Nevertheless, due to the nature of low concentrations of nucleic acids in the bodily fluid samples, in many cases, amplification methods are required to produce measurable signals from optical and electrochemical sensing platforms. Some common amplification methods include the use of loop-mediated isothermal amplification (LAMP) [110] and reverse transcription-polymerase chain reaction [111] to amplify a specific sequence, enabling further analytical readout by electrical, optical, or spectrophotometer methods. There are a variety of detection methods used to quantify and analyze biomarkers, but the enrichment steps can be used interchangeably; once the biomarker is isolated, it can be detected by various analytical readouts (e.g. electrochemical, optical) [27]. Hence, the enrichment of these genetic and proteomic biomarkers (and the combination thereof) could potentially detect early signs of disease, whereby the low abundance of biomarkers would have been missed and led to false negatives.

The use of biophysical and biochemical enrichment methods could potentially enable the detection of biomarkers that are not possible to detect without sample processing. For instance, (a) CTCs are considerably rare, and even the most powerful and sensitive analytical techniques could fail to detect CTCs without pre-concentration steps, (b) biofluids are complex heterogeneous mixtures with many biological components that would interfere with the detection signal producing a high background noise; hence, the ability to properly wash away the non-specific targets arguably enables the reduction of false positives or negatives, and (c) combinatorial enrichment methods can be used to help identify novel biomarkers that would be difficult to detect otherwise. However, despite the great utility of enrichment steps in cancer early detection, many challenges remain unsolved. Some of these challenges include:

5.1. Inability to detect potentially relevant low-abundance biomarkers

5.2. Variation in reporting protocols and methods can induce divergence in interpreting results

5.3. Scaling up and translation of technologies

5.4. Bio-banked samples and validating disease models

This work was supported by the Precision Health and Integrated Diagnostics (PHIND) Center at Stanford Pilot Seed Funding Program and the Canary Center at Stanford for Cancer Early Detection Seed Award. P D S was supported by the James D Plummer Graduate Fellowship, EDGE Doctoral Fellowship Program, and the Dean's Office of the Stanford School of Engineering. F S was supported by the Stanford RISE COVID-19 Crisis Response Faculty Seed Grant Program and the Schmidt Science Fellows in partnership with the Rhodes Trust. P D S and F S contributed equally to this work. TOC schematics were created with biorender.com.

No new data were created or analysed in this study.

Professor Utkan Demirci (U D) is a founder of and has an equity interest in: (i) DxNow Inc. a company that is developing microfluidic IVF tools and imaging technologies, (ii) Koek Biotech, a company that is developing microfluidic technologies for clinical solutions, (iii) Levitas Inc. a company focusing on developing microfluidic sorters using magnetic levitation, (iv) Hillel Inc. a company bringing microfluidic cell phone tools to home settings, and (v) Mercury Biosciences, a company focusing on microfluidic extracellular vesicle isolation technologies. U D's interests were viewed and managed in accordance with the conflict-of-interest policies.