Translating Microfluidics: Cell Separation Technologies and their Barriers to Commercialization

Abstract

Advances in microfluidic cell sorting have revolutionized the ways in which cell-containing fluids are processed, now providing performances comparable to, or exceeding, traditional systems, but in a vastly miniaturized format. These technologies exploit a wide variety of physical phenomena to manipulate cells and fluid flow, such as magnetic traps, sound waves and flow-altering micropatterns, and they can evaluate single cells by immobilizing them onto surfaces for chemotherapeutic assessment, encapsulate cells into picoliter droplets for toxicity screenings and examine the interactions between pairs of cells in response to new, experimental drugs. However, despite the massive surge of innovation in these high-performance lab-on-a-chip devices, few have undergone successful commercialization, and no device has been translated to a widely distributed clinical commodity to date. Persistent challenges such as an increasingly saturated patent landscape as well as complex user interfaces are among several factors that may contribute to their slowed progress. In this article, we identify several of the leading microfluidic technologies for sorting cells that are poised for clinical translation; we examine the principal barriers preventing their routine clinical use; finally, we provide a prospectus to elucidate the key criteria that must be met to overcome those barriers. Once established, these tools may soon transform how clinical labs study various ailments and diseases by separating cells for downstream sequencing and enabling other forms of advanced cellular or sub-cellular analysis.

Emergence of microfluidic cell sorting technologies

The number of microfluidic technologies designed to sort cells have soared in the past 15 years (Figure 1) (1). These devices, typically the size of a credit card when excluding operating equipment, are well-suited for processing cell suspensions as their microscale features approach the size of individual cells (2). Microfluidic devices are significantly smaller than their forerunners, fluorescence-assisted cell sorting (FACS) (3,4) and magnetic-assisted cell sorting (MACS) devices (5), yet they provide equal and sometimes enhanced cell processing capabilities (1). Microfluidic devices exploit a remarkable variety of means to sort cells, as we outline in a recent technical review (6), and can provide capabilities to isolate cells into individual wells (7,8), inspect cells with lasers for high-precision cytometry (9), mix liquids (e.g., for studying dose responses to drugs) (10), count cells (11), size cells (12) and lyse cells (13) (e.g., for DNA sequencing (14)) in a holistic “lab on a chip” format (15,16).

Figure 1.

Recent growth of microfluidic cell sorting technologies. (A) Number of publications versus publication year, organized by articles published in engineering, biology and medical journals and (B) total number of citations per year that resulted from the query “microfluidic cell sorting” on the Web of Science (Thomson Reuters, Co.).

Despite the major advances in microfluidic technologies for sorting cells, they have failed to supplant the predicate mainstays (i.e., FACS and MACS devices) in routine clinical testing (17). This is due in part to the many years of refinement of FACS and MACS devices stemming from increased demands for cellular analysis and concomitant technology adoption that has spurred extensive product development, enhanced automation in system operation, robustness in device design and reliability in sorting performance. FACS devices in particular can provide remarkably high speeds (i.e., up to 50,000 cells/sec) and sorting performances (i.e., purities >99.9%), especially when multiple cell-identifying parameters are used (9,18). MACS devices are a significantly cheaper alternative to FACS as they do not require precise, serial interrogation of cells and can separate cells in bulk. However, both FACS and MACS technologies suffer from several key limitations. For example, FACS devices are expensive, large, complicated to use and typically require high operating pressures that may compromise cell function or viability (19). MACS devices suffer from entrainment of non-labeled cells in target cell populations and highly non-linear magnetic forces that can result in poor capture of labeled cells positioned far from the field source (20).

Microfluidic cell sorting devices can overcome many of these limitations and, in certain instances, provide attractive alternatives to FACS and MACS. For example, microfluidic devices can offer:

• (i)Sorting performances nearly comparable to FACS and MACS devices (e.g., purities ranging, but as high as 99% (21) and throughputs up to 48,000 cells/sec (22) or ~36 mL/hr for whole blood (23));
• (ii)Excellent spatial (e.g., to a precision of 1 μm) and temporal control of cells (24);
• (iii)Reduced risk of compromising cell function by decreasing or obviating the need for high-pressure-sheath fluids (6);
• (iv)Multiplexed cell sorting capabilities in a single or multi-orifice device (e.g., by sorting across 5 outlets (25), which is not possible in conventional MACS devices (9));
• (v)Decreased unit cost (i.e., <$5 (USD) for materials per microfluidic flow cytometer (26) compared to a few tens of thousands of USD (total cost) for a midgrade benchtop flow cytometer (9));
• (vi)Simplified interfaces for increased user friendliness (27);
• (vii)Significantly decreased power consumption (e.g., ~0.1 mW (28) compared to 150 W for the BD FACSVerse™ Flow Cytometer (http://www.bdbiosciences.com/); N.B., this estimate of ~0.1 mW does not account for the power to move fluid through the device);
• (viii)Potential disposability that may allow the development of true point-of-care devices (29);
• (ix)The ability to process unmodified biological fluids (e.g., whole blood) (30);
• (x)Smaller footprints for improved portability, reduced reagent consumption and faster analytical timescales for experiments involving precise changes in temperature or controlled diffusion kinetics within the fluidic cavity (6).

Advances in diagnostic and therapeutic medicine have greatly expanded the need for cell sorting as it is now routinely performed to fractionate blood (31), debulk tissue explants (32) and isolate rare cells from biological fluids (e.g., circulating tumor cells (CTCs) (33) or hematopoietic stem cells (34)). However, there is an enormous disparity between the number of microfluidic tools developed and the number of tools undergoing commercialization.

Commercialization of microfluidic cell sorters

The first cell sorter was developed by Mack Fulwyler in 1965 at Los Alamos Scientific Laboratory (3). Less than 10 years later, Becton, Dickinson and Company (BD) in collaboration with Leonard Herzenberg of Stanford University sold the first commercial cell sorter circa 1974 (35). The first chip-based microfluidic cell sorter, however, which used a spectrophotometer to detect cells in real time for displacement by mechanical pumps, was developed in 1972 by Kamentsky and Melamed (36). Over 40 years later, only a small handful of microfluidic devices are actively progressing towards commercialization (see Table 1). Given the many technological innovations in the past two decades, these technologies are highly diverse, with some offering throughputs that equal leading FACS devices on the market (i.e., 20,000 to 50,000 cells/sec) (9). We divide these devices into two principal categories based on their separation modality. Active devices displace cells for sorting using external forces generated by magnetic, electric, acoustic or optical fields, and which usually require external powering equipment (N.B., some magnetic devices use permanent magnets and can thus sort cells without external powering equipment, as categorized separately in Table 1 (see “Active (Type I)” devices)). Passive devices, on the other hand, do not require external fields to redirect cells or fluid flow, but instead sort via internal means such as antibodies for capturing specific cells or geometric features within the device for altering cellular flow (albeit, external power is usually required to move liquid through the device, but this is generally true for all cell sorting devices).

Table 1.

Microfluidic cell sorting technologies that are commercialized, or are approaching commercialization, arranged by type of device and specifying product name, sorting mechanism, company, key metrics of sorting and application notes.

Type of Device	Product	Mechanism(s)	Company (Acquired by)	Headquarters	Key Metrics	Application Notes
Passive	ClearCell® FX System [a]	Geometric features to trap certain cells by size and stiffness (37,38)	Clearbridge Biomedics, Pte Ltd.	Singapore	8 mL/hr of whole blood, 10,000-fold enrichment	Automated and label-free isolation of CTCs
	Disposable CTC-Chip [b]	Affinity capture (39)	GPB Scientific, LLC	Richmond, Virginia, USA	10 mL/hr of whole blood	Separation of rare cells by immunospecific immobilization
Active (Type 1): External field used, but no external powering required for sorting	CTC-iChip [c]	Deterministic lateral displacement and magnetophoresis (23)	Massachusetts General Hospital	Boston, Massachusetts, USA	~36 mL/hr of whole blood (10,000 cells/msec)	Isolation of CTCs from whole blood with capabilities for positive and negative selection (in collaboration with Janssen Diagnostics, LCC)
	LiquidBiopsy® [d]	Magnetophoresis (40,41)	Cynvenio Biosystems, Inc. (Distribution: Thermo Fisher Scientific, Inc.)	Westlake Village, California, USA	5 mL/hr of whole blood, 1 cell/billion sensitivity	Automated isolation and staining of target cells from whole blood for simultaneous isolation of CTCs and cell-free DNA
Active (Type 2): External field used and external powering required for sorting	Free Flow Acoustic Chip [e]	Free flow acoustophoresis (42)	AcouSort AB	Lund, Sweden	N/A	Label-free approach to acoustically sort cells by size in a continuous flow microchip device
	GigaSort™ [f]	Pneumatic chambers to hydrodynamically displace fluids (22)	Cytonome, Inc.	Boston, Massachusetts, USA	≤ 48,000 cells/sec	Parallel microchips (24,48 or 72) for sterile, high throughput cell identification and sorting
	Jet Flow Sorter [g]	Bubble-jet flow generated from Joule heating (43)	JSR, Co.	Leuven, Belgium	~2,000 cells/sec	On-chip imaging and cell identification for multi-channel sorting (in collaboration with IMEC vzw)
	MACSQuant® Tyto [h]	Fluorescence detection and electro-mechanical switching (44)	Owl Biomedical, Inc. (Miltenyi Biotec GmBH)	Bergisch Gladbach, Germany	≤ 30,000 cells/sec	Disposable cartridge with a high frequency switch for cell sorting with 95% viability
	Sorting Chip for Cell Sorter SH800 [i]	Focuses and detects cells for post-chip sorting via FACS technology (45)	Sony Biotechnology, Inc.	San Jose, California, USA	≤ 10,000 cells/sec	Disposable cartridge that aligns automatically in a permanent workstation for cell identification
	SSAW Acoutophoresis Chip [j]	Standing surface acoustic wave (SSAW) (46)	Ascent Bio-Nano Technologies, Inc.	State College, Pennsylvania, USA	N/A	Label-free sorting of cells using “acoustic tweezers”
	Wolf™ Cell Sorter [k]	Fluorescence detection and acoustophoretic sorting (28,47)	NanoCellect Biomedical, Inc.	San Diego, California, USA	~330 cells/sec	Sorts cells while removing contaminants such as cell-free DNA

Websites:

^[a]http://www.clearbridgebiomedics.com/

^[b]http://www.gpbscientific.com/

^[c]http://www.massgeneral.org/research/resourcelab.aspx?id=51#/

^[d]http://www.cynvenio.com/technology/

^[e]http://www.acousort.com/

^[f]http://www.cytonome.com/products/gigasort/

^[g]http://www2.imec.be/be_en/research/life-sciences/high-throughput-cell-sorting.html/

^[h]http://www.miltenyibiotec.com/en/products-and-services/macs-flow-cytometry/flow-sorter.aspx/

^[i]http://sonybiotechnology.com/sh800_sorting_chip.php/

^[j]http://ascentbionano.com/

^[k]http://nanocellect.com/technology/wolf-cell-sorter/

Leading microfluidic sorting devices

Passive devices

The first class of passive sorting devices comprises those where the microchannels contain indented, protruded or curved sidewalls to direct cells. An early example was pinched flow fractionation, which uses a co-flowing stream of buffer to force cells along one wall such that the center of larger cells are slightly further from the wall than the center of smaller cells. Consequently, larger cells separate from smaller cells once the width of the microchannel broadens several-fold (48). A second example is inertial focusing, in which microchannels displaying a serpentine or Archimedean spiral pattern exert inertial forces on cells to allow them to cross streamlines at distances according to their size (49,50).

The second class of passive cell sorting devices is comprised of those that contain repeated geometric patterns within the flow cavity of the microfluidic device (51). These devices can contain a variety of features, such cylindrical or oblong columns, pores and anisotropic ridges to fractionate cells by size, stiffness or shape (e.g., ClearCell® FX System; Table 1). Deterministic lateral displacement is a classic example of this, whereby an array of posts positioned at an angle offset from the direction of flow fractionates cells by size (such that smaller cells partially circumnavigate the posts in the direction of flow and larger cells travel at the offset angle directed by the posts) (52). Other examples include microvortex-generating (53,54), size exclusion (55), cross-flow (56) and hydrodynamic (57) filtration devices, which contain various posts and pores for separating cells. Recently, Toner et al. developed the “Cluster-Chip”, which uses engineered geometric patterns to trap clusters of CTCs from unprocessed blood by virtue of their size and tight cell-cell junctions (58).

The final class of passive sorting devices is comprised of those that contain biomolecular recognition sites to promote cell attachment (59,60) or temporary adhesion (61). Several groups have developed devices that sort cells by the former mechanism (59,62). Notably, Toner et al. developed the “CTC-Chip” (Table 1), which immobilizes CTCs via geometric posts within the flow channel that are coated with antibodies (39). In contrast to such “irreversible” interactions, Karnik et al. developed a device that sorts cells by deterministic cell rolling, in which cells expressing a target ligand interact transiently with proteins on the surface to allow the cell to roll and separate from other cells (Figure 2) (63).

Figure 2.

Cell sorting by deterministic cell rolling. In this schematic, cells expressing a target ligand (purple) have specific, but transient, interactions with proteins immobilized on the floor of the channel, directing them across the channel and into the distal outlet. Cells that do not express the target ligand (blue) pass over the ridges and exit the proximal outlet. Reprinted with permission from Choi et al. (63). Copyright 2012 Royal Society of Chemistry.

Active devices

The first class of active cell sorters is comprised of those that uses electric forces to sort cells, which typically operate by first identifying cells with fluorescent labels for subsequent downstream sorting under flow (e.g., as in conventional FACS or the Cell Sorter SH800; Table 1) (64,65). Recently, several groups have developed chips that generate droplets of aqueous buffer to encapsulate cells within an oil (i.e., an emulsion) for manipulation by electric fields (66). A major advantage of this approach is that the cell-containing droplets serve as “picoliter reaction vessels”, enabling the evaluation of single cells for their sorting based on enzymatic (or other) activity instead of surface marker expression alone (67,68). One challenge with this approach has been the encapsulation of a single cell per aqueous droplet, as the number of cells per droplet is effectively dictated by Poisson statistics (69).

The second class of active sorting devices, and perhaps the most widely adopted, is comprised of those that use magnetic forces (e.g., LiquidBiopsy®; Table 1) (70). The first magnetic cell sorter, developed in 1990, was a benchtop tool that isolated cells bound to magnetic beads in a column (5). Xia et al. were among the first to miniaturize this technology into a microfluidic device (71). A large number of similar magnetic sorting devices have been developed (20). These devices are convenient due to the specificity of the magnetic field interactions on magnetic, biospecific labeling materials (which are not found in large quantities in cells) (72). Several groups are further exploiting the convenience of the magnetic labels on cells after sorting for the transportation and organization of cells along magnetographic patches for single cell analysis (7,73).

The third class of active cell sorters uses acoustic forces to sort cells via: (i) bulk acoustic standing waves (74-77), (ii) standing surface acoustic waves (SSAWs) (78-80) or (iii) bulk acoustic travelling waves (81). Petersson et al. first demonstrated the utility of bulk acoustic standing waves by separating lipids from blood cells (82). Our group recently developed elastic silicone-based particles that migrate to the pressure antinodes (i.e., instead of the nodes like most biological cells) in aqueous media (83). We have shown that cells and polystyrene beads bound to these particles can be displaced to the pressure antinodes for trapping and, potentially, sorting from unlabeled cells (84,85). Shi et al. were among the first to develop such a device that imposed SSAWs to sort cells (Figure 3) (86). A large number of SSAW devices have since been developed for bioanalytical applications (46).

Figure 3.

Cell sorting device using SSAWs. Cells flow through the device from left to right. A stream of buffer fluid (central inlet) focuses cells (outer inlets) along the walls of the device whereupon acoustic fields generated by interdigitated transducers propel cells towards the pressure node at rates proportional to their size for sorting. Reprinted from permission by Shi et al. (86). Copyright 2009 Royal Society of Chemistry.

The fourth class of active sorting devices comprises those that use optical forces. Optical tweezers have been used to sort cells in the microfluidic regime (87), but they require a significant amount of power and they typically move one cell at a time. Another type of optical device for cell sorting, optoelectronic tweezers, projects optical images onto a photosensitive substrate to form transient electrodes to direct cells for sorting (88). While this provides a low-power and more rapid alternative to optical tweezers (89), the setup is complex and cumbersome compared to other microfluidic cell sorting tools.

Integrated devices

Integrated devices are those that exploit more than one method for manipulating cells on a single chip. These integrated systems show promise for commercialization as they can provide enhanced capabilities for processing unmodified biofluids (e.g., blood) and may be also used to evaluate individual cells after sorting. A groundbreaking example is the “CTC-iChip” (Figure 4A; Table 1), which was developed by Toner et al. (90) as a successor of the “CTC-Chip” (39). The CTC-iChip combines active and passive components to isolate ultra-rare CTCs (i.e., as few as one cell in 7.5 mL of whole blood) via either positive or negative selection (23). Several other integrated devices have been developed, such as one that sorts cells by AC electric fields and then by magnetic fields (91) and another that sorts cells by hydrodynamic filtration and further fractionates those cells by magnetophoresis (Figure 4B) (92).

Figure 4.

Integrated cell sorting devices. (A) Cell sorting by the “CTC-iChip”. Whole blood enters the device (the far end of the schematic) whereupon nucleated cells (i.e., white blood cells (WBCs) and CTCs) are separated from smaller, non-nucleated cells and plasma by deterministic lateral displacement. The nucleated cells then enter a serpentine channel where they focus into a single streamline by inertial forces. Finally, CTCs are separated from WBCs via magnetophoresis. Reprinted from permission by Karabacak et al. (23). Copyright 2014 Nature Publishing Group. (B) A hydrodynamic filtration system with an integrated magnetophoretic separation component. Cells enter the device (left), are pushed toward a series of drains by a co-flowing stream of buffer fluid and are separated by size. Once separated, cells are further fractionated by magnetic forces, which pull more strongly on cells with higher expressions of a target marker (i.e., due to their higher magnetic content). Reprinted with permission from Mizuno et al. (92). Copyright 2013 American Chemical Society.

Barriers to commercialization and clinical translation

Despite these numerous advances in microfluidic cell sorting, relatively few technologies are progressing towards the market (Table 1) (93). Of the few technologies that are undergoing commercial translation, several persistent barriers have prevented these technologies from becoming a widely distributed commodity, even after successful laboratory validation and several rounds of heavy commercial investment. We believe there are a number of factors contributing to this phenomenon. We divide these factors into device-related barriers and commercialization-related barriers, as discussed below.

Device-related barriers

Device-related barriers typically arise from the focus of research labs to produce new technologies, often at the expense of practicality for the end user. For example, microfluidic devices usually contain multiple complex components (e.g., fluidic pumps, valves and sometimes a microscope station and/or a field source generator) that require control for operation and sometimes coordinated data acquisition (Figure 5). They can contain several fluidic junctions that must be manually connected to fluidic lines (e.g., pneumatic lines (94)), which increases their footprint and renders setup time-consuming. Not only so, these junctions often require custom tubes and adaptors, thus impeding their integration with standard laboratory tools (95). Further, microfluidic devices are often fabricated from polydimethylsiloxane (PDMS) via soft lithography (96,97). While convenient for laboratory-scale production, these materials are difficult to scale up to industrial levels and are known to dissolve gasses and hydrophobic compounds, thus affecting the environment of cell suspensions (98). Leveraging existing frameworks, such as the monolithic techniques from the semiconductor industry, as well as emerging techniques (e.g., 3D printing) may help to overcome these challenges (99-101); however, the ideal material for industrial-scale productions of microfluidic devices ultimately depends on the separation technique at hand. For example, microfluidic devices supporting bulk acoustic standing waves must be made from rigid materials, such as silicon, glass or polymethyl methacrylate (PMMA) (76,102), whereas optoelectronic devices are typically made, at least in part, from transparent, electrically conductive materials, such as indium tin oxide (ITO) (88).

Figure 5.

Schematic of a representative microfluidic cell sorting setup. Top (from left to right): a computer control system to regulate the operating equipment, a microscope to observe the sorting junction and a power source to generate an external field (for active devices). Bottom (from left to right): vials to collect the sorted cell populations, the microfluidic chip and an automated syringe pump (note: objects are not to scale).

Second, the lifetimes of microfluidic devices can be relatively short compared to standard benchtop tools. Due to the compact dimensions of the microchannels, air bubbles and other small obstructions can dramatically diminish their performance in cell sorting applications. As such, the lifespan of microfluidic devices is often curtailed by buildup of debris from cell suspensions, residue from native biofluids (e.g., coagulants and biofoulants) and aggregates of reagents and synthetic labeling materials. These aggregates can induce cavitation, which further diminishes performance and throughput, which necessitates longer operating times and becomes especially problematic when isolating rare cells (103).

Commercialization-related barriers

Commercialization-related barriers, which relate to intellectual property and market need (among other things), are a direct extension of the former set of barriers and are chiefly responsible for the slowed integration of microfluidic devices into the marketplace. The first major barrier of this type is the saturated landscape for new intellectual property, which was not always a major issue. In the 1960s, many of the patents awarded by Los Alamos Scientific Laboratory (e.g., the patent for the first cell sorter developed by Mack Fulwyler in 1965) were made available for licensing on a royalty-free, non-exclusive basis (3,104). This freedom to operate resulted in a number of corporations working to commercialize cell sorting technologies simultaneously, with the first commercial product appearing less than 10 years later (i.e., by BD in 1974) (35). In contrast, a flood of patent applications was filed in the 1990s to protect all potential uses of microfluidic devices. Now, in the U.S. alone, there are over 2,750 issued patents with the word “microfluidic” in their claims (from an online search on the USPTO patent full-text and image database). These patents, which contain broadly written claims, have sequestered much of the available space for the acquisition and implementation of new intellectual property. Consequently, many promising new technologies may fail to realize their commercial potential due to a lack of freedom to operate (i.e., intellectual security or threat of patent infringement), discouraging inventors from pursuing startup (or other) ventures and dissuading licensors or investors from financing these endeavors. This challenge is exacerbated when the technology overlaps with other patented technologies or techniques (e.g., cartridge-based devices), whereby licensing agreements must be forged between multiple parties, thus increasing expense and risk. N.B., some companies have successfully found ways to overcome these issues (e.g., see Dolomite Microfluidics, Inc., which has developed several modular microfluidic technologies).

The second barrier of this type is the complexity of the target market. An enormous amount of time, money and effort will be required to change the infrastructure of the cell sorting industry from manufacturing benchtop devices to microfluidic tools. Importantly, this change will never take place without a compelling advantage by a new technology fulfilling an unmet need in an efficient and practical manner (105). As such, it is essential for new technologies to possess a high market potential (i.e., a major competitive advantage in terms of improved overall performance or a significant reduction of cost, or preferably both). Incremental advances in technology are insufficient for driving the shift of the current market infrastructure to microfluidic tools for cell sorting. As shown in Figure 1, the vast majority of literature on microfluidic cell sorting appears in engineering journals, whereas very few articles have advanced into medical journals. In order to better understand the potential market for microfluidic cell sorting, significantly more biological and medical translational research must be conducted.

Thirdly, and as a direct result of the previous two barriers, limited investment opportunities post-invention have greatly hindered the forward progress of microfluidic cell sorters. The entry of these technologies into the marketplace is largely dependent on successful funding, which broadly encompasses early, midterm and late commercial investments (e.g., from bootstrapping, angel investors, corporations and venture capital firms), as well as subsidies to catalyze their intellectual property, development, optimization, standardization, manufacturability, marketability and distribution (106). Funding is dependent on a variety of factors, including the perception of commercial potential, economic factors and regulatory standards (e.g., especially for devices that involve clinical use). In the U.S., the pathways to regulatory approval are typically by 510(k) clearance (i.e., when the technology is substantially equivalent to a predicate technology) or premarket approval (i.e., a submission to the FDA to show that a new device is safe and effective), which is significantly more costly and time-intensive (105). While critical to the translation of a clinical technology, regulatory affairs are extremely time consuming and costly (e.g., obtaining premarket clearance, paying submission fees, and funding extensive pre-clinical and clinical studies), thus requiring significant investment, which can be problematic in a highly competitive market against well-established tools such as FACS machines. As such, to our knowledge, no microfluidic cell sorting technology has been approved for clinical use in the U.S. to date.

Perspective: Overcoming barriers to commercialization

The current market for cell sorting is projected to exceed $5 billion (USD) annually by 2019 (107). This rapid growth has the potential to fuel the translation of many microfluidic cell sorting technologies to the market, especially if certain strategies are implemented, as we discuss below.

Device-related solutions

We believe the most significant impediment to translation of microfluidic cell sorting devices is associated with philosophy of innovation. Microfluidic systems are usually developed first and then adapted to fit a potential market need. However, this longstanding approach must be rethought and reversed in order to enhance translation towards a significant commercial or clinical impact (108). Innovators must identify a biological problem that represents a significant unmet need, and then work to develop microfluidics tools that addresses that need.

As such, we believe there are several simple strategies can be implemented to overcome challenges related to the design of microfluidic devices. First, user interfaces must be simplified to minimize setup time and maximize ergonomic operation. Integrating devices with universal tubes and adaptors will facilitate this operation without specialized training that can easily be integrated into a laboratory or hospital setting. Devising cartridge-based apparatuses (cf., MACSQuant® Tyto; Table 1) will enable the facile loading and unloading of microfluidic devices into an external housing unit to facilitate automation and reliability as well as reduce or obviate long term issues related to clogging, fouling and the generation of bubbles. Making these cartridges disposable could eliminate the need for serial cleaning steps and the risk of cross-contamination (109). These changes will add to the commercial appeal due to perceivably higher user compliances, improved reusability and potential for sustained revenue over the life of the instrument (110).

The development of multifunctional, modular microfluidic devices will help to firmly establish their utility in the laboratory and the clinic. Microfluidic devices can independently accomplish a variety of tasks (e.g., mixing (10), trapping (8), lysing (13) and drug screening (11)); however, an easily constructible and reconfigurable device with connectable and interchangeable parts for cell sorting has not, to our knowledge, been developed. Such a design could reduce manufacturing costs and enable the integration of microfluidic cell sorting devices in the clinical workspace without changes in infrastructure, thus allowing these devices to interface with existing laboratory supplies, increasing their analytical and therapeutic value and appeal to users. We note that the C1™ Single-Cell System by Fluidigm is a closely related example of such a device by offering multiple interchangeable integrated fluidic circuits to trap, lyse and sequence single cells (https://www.fluidigm.com/products/c1-system).

Finally, given the miniaturized size of these devices and the economic trade-off of low versus high volume manufacturing, scaling up both on the device-level and the production-level is paramount. On the device-level, advances in parallelization and stacking microfluidic channels into 3D architectures may allow these devices to process larger volumes of fluid in shorter timescales (111,112). On the production-level, selecting materials suitable for high volume manufacturing, potentially by leveraging existing facilities and infrastructures, could enable the production of devices at economically viable scales (e.g., 100,000 per year for disposable devices). Manufacturing methods and materials for this level of scale may include wet etching glass, hot embossing thermoplastic materials (e.g., PMMA), lithographic processing of silicon and milling various metals such as stainless steel (113). Furthermore, innovations from emerging techniques, such as 3D printing (100), may lead to a significant reduction in overall costs (114). However, we emphasize the practice of materials selection is seldom “one size fits all”, as different devices may require different materials to function and thus must be decided on a case-by-case basis.

Commercialization-related solutions

Building a strong market strategy is critical to overcoming many of the barriers related to commercialization. It is imperative for inventors, entrepreneurs and product development experts to focus on the end market in addition to the product itself at all stages of development, from invention all the way to production. As described in a recent review, the number one reason behind the failure of most startup companies is the lack of a market (or failing to address the needs of a market) (106). Building a competent team with knowledge, expertise and vision is vital to increasing one's chances for successful commercial venture. However, even with a strong team, understanding the competitive advantage of the device and building a robust market strategy (i.e., one that addresses the needs, cultural milieu and values of the market) is essential for success (110).

Aggressively filing patent applications and pursuing strategic licensing can help to reduce risk and add value to companies commercializing microfluidic cell sorting technologies (115). Intellectual property should be developed in parallel with research, not as an afterthought or in a rush before publication. While it is not required, including strong, validating experimental data in a patent can greatly help reinforce that patent later on during times of intellectual infringement. The obligation to aggressively file patent applications and build a comprehensive patent portfolio can sometimes be perceived as a deterrent to commercialization; however, strategic licensing can mitigate this burden, whereby a technology is licensed to either a startup company or a more established entity with other, complementary patents. Such strategic licensing can allow the inventor and the licensor to forge a partnership to benefit both parties while advancing the technology as a whole. Also, field-of-use licensing enables inventors to restrict the licensee to use the technology only for a particular application, which can provide additional opportunities for licensing. In addition to licensing, forging strong partnerships between academic and industrial entities (e.g., as with the CTC-iChip) can foster an environment conducive for successful commercialization. Navigating the best path forward for securing and managing intellectual property is best decided on a case-by-case basis; however it is essential to the funding, development and forward progress of microfluidic technologies.

Concluding remarks

While the field of microfluidic cell sorting is still relatively new, its outlook is promising (27). Groundbreaking discoveries in cell sorting have revolutionized the ways in which cells are separated and analyzed, enabling the realization of high-performance microchip devices that rival the functionality and performance of standard benchtop tools, but with added features for enhanced cellular evaluation. However, persistent challenges have prevented these devices from reaching their full potential. We provide several simple strategies to improve their user friendliness and potentially expedite their translation into the clinic. As these changes take place, we believe microfluidic cell sorters will soon revolutionize the ways in which cell-containing fluids are processed in the laboratory and the hospital.