Blood sampling has become a common method in modern medical diagnosis due to its high efficiency and low cost. Many biometric tests are based on blood samples. However, the invasive sampling methods always accompanied subcutaneous acupuncture or finger pricks that usually lead to a decrease in patient compliance (Ayala et al., 2009). Meanwhile, blood sampling usually has to be performed by professionals. Although numerous efforts have been made in minimizing professional requirements during operation, the pain and discomfort still exist in the frequent finger-prick sampling with easy-to-use glucose meters. In addition, irregular operation and usage of unsterilized needles may pose a risk of blood-borne transmission of various biohazard pathogens (Sagoe-Moses et al., 2001). The disposal of sharp solid waste after blood sampling is also a problem.

Dermal interstitial fluid (ISF) is a portion of the fluid that seeps into the interstitial tissues from the arterial ends of capillaries. ISF acts as an intermediary between cancer cells and the circulatory system. Previous studies have proved that the composition and concentration of electrolytes, small molecules, and proteins in ISF are similar to those in plasma (Kool et al., 2007; Halvorsen et al., 2017; Niedzwiecki et al., 2018). Furthermore, some of the systemic and skin-derived metabolite biomarkers in ISF are unique (Muller A. C. et al., 2012; Niedzwiecki et al., 2018). Therefore, ISF is considered a promising alternative biofluid source with a lot of health-related information. The microneedle-based devices provide a noninvasive way to access this information.

The skin, the body’s largest organ, is a rich resource supply of ISF. However, the extraction of this fluid is restricted by a skin barrier. The stratum corneum, which is 15–20 μm thick, effectively prevents foreign substances from penetrating the skin (Dervisevic et al., 2020). Microneedles (MNs) are miniaturized needles with few hundreds of microns in length (Waghule et al., 2019). They can traverse the epidermis corneum and further insert into the viable epidermis while escaping from contacting any pain-sensing neurons or dermal blood vessels (Larraneta et al., 2016b). The first MN concept was proposed by Gerstel and Place in 1970 (Gerstel et al., 1976), while its blooming started until the 1990s after the development of micro- and nanofabrication techniques (Ventrelli et al., 2015). Since then, MNs have been fabricated using various materials with different physicochemical properties (Ali et al., 2019). The application of MNs initially focused on transdermal delivery of drugs (Kathuria et al., 2018) and vaccines (Creighton and Woodrow, 2019). Then, the study of MNs has been explored in many other areas including blood sampling (Xue et al., 2018), biosensing (Tasca et al., 2019), cosmetic (Yang J. et al., 2019), and cancer therapy (Moreira et al., 2019). MN-based devices have many fascinating properties, such as painless and minimal invasiveness, which show a great potential in point-of-care testing (POCT).

In general, MN-based devices can be classified into four categories according to their functional purposes (Figure 1): 1) MNs are used for transdermal sampling only with the corresponding materials analyzed by external methods 2) the MN array is designed to capture biomarkers selectively; 3) MNs are integrated in a sensor to detect or monitor the analytes within the skin; and 4) MNs are used for bio-signal recording platforms.

In the present review, we focus on the recent advances of microneedle-based devices. The four categories of MN-based devices and their applications are highlighted and discussed in detail. Finally, the future perspectives of MN-based devices are summarized.

Four types of MNs (solid MNs, hollow MNs, porous MNs, and swelling MNs) have been used for transdermal sampling (Figure 1A). The properties of MNs for transdermal sampling are summarized in Table 1.

Solid MNs and hollow MNs, made by silicon, glass, and stainless steel, are usually used for transdermal sampling (Tariq et al., 2021). Normally, these types of MNs were punctured on the skin with the ISF collected by a vacuum pump or a well-absorbent film sucking to withdraw the tissue fluid for subsequent analysis (Wang et al., 2005; Nalbant and Boyaci, 2019). For example, Kolluru et al. (2019) designed a two-component system composed of a strip of paper on the back of stainless steel MN arrays. First, the MN arrays break the skin barrier and create micropores that allow the ISF to flow to the skin surface, in which the ISF can ultimately be sucked and reserved in the paper strips due to the capillary force. The optimized paper strip–stainless steel MN array is capable of >2 µl of ISF collection within 1 min safely and conveniently on a rat skin in vivo. Nicholas et al. (2018) integrated an enzyme-based colorimetric glucose sensor to a single hollow MN device and achieved rapid determination of glucose concentration in simulated ISF with physiological relevancy. The sharp tip of hollow MNs could help the MNs to better penetrate the skin. Through the capillary effect inside the skin, the uptake of ISF within 5 s is demonstrated.

Solid MNs and hollow MNs have greatly facilitated transdermal sampling. Nonetheless, instrumental assistance and multi-step operations are usually required to maintain a negative pressure during the preparation of MNs, which limits their application. Additionally, glass- or silicon-based MNs often suffer from brittleness, which poses a potential safety risk to patients and causes environmental hazards (Sivamani et al., 2007; Larraneta et al., 2016a). One solution to the aforementioned issues is to replace glass/silicon with porous polymers in MN fabrication. Porous polymer–based MNs exhibit various distinctive advantages including desirable biocompatibility, ease of processing, large cavity, and three-dimensional connected porous structure. Liu et al. (2020) developed a phase inversion route to prepare various polymer-based MNs using cellulose acetate (CA), polysulfone (PSF), polyethersulfone (PES), polylactic acid (PLA), etc. with high porosity and interconnected pore structures. The resultant porous polymer–based MNs can extract ISF rapidly. Liu P. et al. (2021) have demonstrated a facile and effective method for interconnected structured and hydrophilic porous polymer MN modification. The coating layer of PDA provides new sites for polymers post-functionalization, which is beneficial for further PEG modification on the polymer matrix. The imported PEG improved the porous structure hydrophilicity and enhanced the molecular resistance effect. In addition, this method could be promoted to other polymer materials such as PLA and PVDF. The strategy provides a new idea for the fabrication of hydrophilic and anti-adhesion porous polymer MNs. Takeuchi et al. (2022) developed a fluidic system that directly interferes with the porous MNs, in which a capillary pump was utilized inside the microfluidic chip for continuous ISF sampling. In detail, the developed device connects a standard microelectromechanical system (MEMS)-fabricated microfluidic chip with porous flexible MNs of PDMS. It achieved continuous flow control of phosphate-buffered saline (PBS). The development of the device is beneficial for long-term health monitoring applications based on minimally invasive and continuous bio-sampling.

Another emerging good candidate for transdermal sampling is swelling MNs, which generally consist of cross-linked polymer networks and enable one-step ISF extraction without the aid of extra devices. Hydrogels are biocompatible and biodegradable. They are able to cause minimal tissue damage and little environmental threat after disposal (Herrmann et al., 2021). Chang et al. (2017) developed a methacrylated hyaluronic acid (MeHA)-based swellable MN platform, which can extract ISF efficiently for rapidly metabolic analysis. The MeHA patch could penetrate into the skin by a thumb press, extracting ≈1.4 mg ISF within 1 min, while maintaining structure integrity without leaving residues in the skin. A gelatin methacryloyl (GelMA)-based MN patch was developed in an 11 × 11 array fashion by a micromolding approach for minimally invasive ISF sampling (Figure 2) (Zhu et al., 2020). The properties of the patch could be adjusted by varying the GelMA prepolymer concentration and the crosslinking time. The swelling ratios of resulting patches are between 293 and 423%, and compressive moduli are between 3.34 and 7.23 MPa. The developed GelMA MN patch exhibits an efficient extraction of ISF. Laszlo et al. (2021) proposed two hydrogel-based MN arrays for dermal ISF proteomics sampling. The developed MN patches could restore their initial shape even after multiple mandatory washing steps to erase un-crosslinked polymers. Such a property could prevent signal-inhibiting effects and chromatographic interferences. The in vitro and in vivo biocompatibility was demonstrated. The collection of dermal ISF in vivo is also studied for subsequent proteomic applications. Zheng et al. (2020) developed a new osmolyte-composited swellable hydrogel MN for skin ISF extraction. The developed MN patch is composed of osmolytes such as maltose and hydrogels such as methacrylated hyaluronic acid. During extraction, the maltose would dissolve in the matrix and provide an osmotic pressure to enhance the diffusion rate of ISF to the hydrogel matrix. The microneedle prepared with the optimized formula could extract 7.90 µl ISF from pig ear skin and 3.82 µl ISF from mouse back skin within 3 min in vitro.

The MN array designed for biomarker capture is an emerging approach in modern medical diagnosis. As shown in Figure 1C, MN systems for biomarker capture usually contain identification elements or probes for the selection and capture of analytes.

As shown in Table 2, the recognition mechanism of MNs for biomarker capture is mainly immunoaffinity. The recognition element is usually an antibody, oligonucleotide, or other proteins that can bind to a specific molecular motif. Thus, the target analyte can be effectively separated from the complex biological matrix. Figure 3A shows a typical example about this mechanism. The nanoporous MNs were functionalized with an antibody to selectively capture estrogen (E2) (a preeclampsia biomarker), following insertion into the skin, thereby enabling PE diagnosis by measuring biomarker concentrations based on the immunoassay method (Kang et al., 2020). To provide ultra-dense binding sites for the extraction and detection of trace amounts of the target analyte in the subcutaneous ISFs, a uniform nanopore structure was created on the tissue-penetrating aluminum (Al) MN array by controlled anodic oxidation. The immuno-functionalized nanoporous MN patch’s rapid, sub-nanogram-level sensing ability was verified by in vitro E2 detection tests and in vivo skin tests of experimental animals with response signals obtained within 1 min.

Apart from immunoaffinity, Watson–Crick base pairing has also been applied in MN systems for biomarker capture. Al Sulaiman developed alginate–peptide nucleic acid-coated MNs that are capable of sequence-specific capture of target miRNAs by Watson–Crick base pairing (Figure 3B) (Al Sulaiman et al., 2019). Attaching PNA oligomers to the hydrogel’s fibers enables specific sampling, purification, and release of the only nucleic acid fragments that are complementary to the PNA sequence. By simply adapting the PNA sequence, this system can be applied in any miRNA of interest.

Yang B. et al. (2019) provide another way of biomarker capture by MNs. They employed hydrogel MN patch for fast and easy in situ capture of Epstein–Barr virus cell-free DNA (EBV Cf DNA) from ISF in 15 min with a capture efficiency as high as 93.6% (Figure 3C). Detection of EBV Cf DNA was achieved by electrochemical recombinase polymerase amplification implemented wearable flexible microfluidics, with a detection limit of 3.7 × 10² copies/μl. Animal experiments support the effectiveness of MNs for EBV Cf DNA capture. Later, they combined reverse iontophoresis and MN techniques and engineered a wearable epidermal detection system for rapid sensing of EBV Cf DNA (Yang et al., 2020). On account of the additive extraction effect of reverse iontophoresis and MNs, the engineered platform successfully separated cell-free DNA targets from ISF within 10 min, with a maximum capture efficiency of 95.4% and a threshold of 5 copies per µl. Captured cell-free DNA is directly used in the developed electrochemical microfluidic biosensor with a detection limit of 1.1 copies/µl.

Until now, MN arrays for capturing biomarkers have been reported by many groups. Corrie et al. (2010) first designed a MN-based device that captures circulating biomarkers in skin fluids in a painless manner with high specificity. They grafted a hetero-bifunctional PEG to the gold-coated microprojections, which reduced the unspecific protein binding and provided a spot for capturing attachment proteins. High specificity and high sensitivity antibody capture, extraction, and analysis have been achieved. Using the similar method, their groups realized the capture of dengue virus NS1 protein (Muller D. A. et al., 2012), recombinant P. falciparum rPf HRP2, and total IgG from the complexed matrix (Lee et al., 2014). Wang Z. et al. (2021) reported that functional biomolecules (such as antibodies) modified MNs that could penetrate the stratum corneum and periosteum, enabling the selective capture of protein biomarkers in the local ISF in a concentration-dependent manner. Due to the hydrophobic interaction between polystyrene and nonpolar residues on the protein, this interaction effectively traps the hydrophobic structures in antibodies and proteins and binds them to the polystyrene surface. Zhang et al. (2019) developed photonic crystal (PhC) barcodes integrated in encoded MNs for detecting ISF biomarkers. PhC barcode–loaded flexible MNs were elegantly fabricated by replication of dynamic ferrofluid casting micropatterns. When the prepared MNs are inserted into the skin, they could enrich certain biomarkers into their PhC barcode–modified probes. Therefore, sandwich immunocomplexes can be formed after the addition of corresponding fluorescent tags. The relative level of the biomarkers could be read out by the fluorescence signals of the barcode; at the same time, the types of biomarkers could be clearly discriminated by the reflection peaks of the PhC barcode.

It is worth noticing that aptamers can be potentially applied in MN arrays for biomarker capture. Aptamers are recognized as single-stranded oligonucleotides in 25–80 base length that exhibit specific binding affinity toward targets including amino acids, drugs, and protein biomolecules. In comparison with protein antibodies, aptamers have several advantages such as small size, strong binding affinity, high specificity, desirable biocompatibility, good stability, and lesser immunogenicity, making them widely used in the field of biomedicine (He F. et al., 2020; Ni et al., 2021). Aptamer-modified porous MN arrays are proposed to enable in situ enrichment and detection of biomarkers in ISF by Yi et al. (2021) (Figure 3D). Porous MN arrays were made by replicating negative molds containing SiO₂ microspheres and UV-curable ethoxylated trimethylolpropane triacrylate (ETPTA). Since the MN array combines advantages of aptamers and porous structures, its surface area is remarkably increased to 6.694 m²/g, so a large amount of aptamer probes (0.9459 μM) can be fixed. Furthermore, due to capillary forces, the MN array can extract ISF into its porous structure. Subsequently, biomarkers are captured and detected without sample post-processing.

MN-based sensors have been employed widely for detecting or monitoring a diverse range of analytes as listed in Table 3.

Both humans and animals produce regular electrical signals in static and active states. Monitoring these biological signals, for example, electromyogram (EMG), electrocardiogram (ECG), and electroencephalogram (EEG), can help to understand human pathology and physiology (Peng et al., 2016). Conventional wet Ag/AgCl electrode could obtain biological signals with relatively high resolution in clinical practice. However, several disadvantages limit its further application. First, Ag/AgCl electrodes always require skin preparation such as haircuts and skin scrapes (Griss et al., 2000). Second, gel electrolytes are often used to reduce the high impedance of the cuticle (Wang et al., 2017). However, the gel may cause skin irritation or allergic reactions that make the subject uncomfortable (Kato et al., 2006). In addition, the gel dries out over time, which reduces the quality of the recorded signal (Tallgren et al., 2005). Therefore, wet electrodes are suggested to fail to long-term bio-signal recording. Compared with wet electrodes, dry electrodes do not require electrolytic gel, and thus can record biological signals continuously (Baek et al., 2008; Fu et al., 2020). However, the electrode-skin interface impedance (EII) is still very high and sensitive to human movement (Yu et al., 2009; Forvi et al., 2012).

As a kind of dry electrode, the MN array electrode (MAE) was first proposed by Griss et al. (2000). MAE can penetrate the cuticle of human skin and eliminate the effect of cuticle on impedance (Ren et al., 2020b). Meanwhile, MNs do not require skin preparation for puncture and result in minimal skin trauma (Kashaninejad et al., 2021). It has been shown that MAE has a lower and more stable EII than commonly used dry electrodes. Figure 7 schematically showed the electrical equivalent circuit model of wet electrodes and MAE. Here, E_w is the half-cell potential. The interface between the wet electrode Ag/AgCl and the electrolytic gel could be simulated by the parallel connection of capacitor C_d and resistor R_d. R_g represents the electrical resistance of the gel used in the experiment. E_se represents the potential difference between the gel and the stratum corneum. The stratum corneum can also be modeled by a capacitor C_e and a resistor R_e in parallel. R_u represents the resistance of the dermis and subcutaneous tissue. The equivalent circuit of MAE is relatively simple. Since the MNs are in direct contact with the epidermis, there is only coupling between the electrodes and the conductive layers within the skin, except for the conductive resistance of the dermis and underlying tissue, described by the half-cell potential E_m and connected in parallel with the resistance R_m and the capacitance C_m.

To date, MN-based devices have been successfully used in ECG, EMG, and EEG recording (Chen et al., 2016; Pei et al., 2017; Ren et al., 2018; Sun et al., 2018; Mahony et al., 2019; Satti et al., 2021; Wang Y. et al., 2021). The properties of MNs for bio-signal recording are summarized in Table 4. Due to the relatively mature processing technology in semiconductor industries, many MN materials are based on silicon. Silicon is the first material used in MN fabrication due to its instinct excellent mechanical strength, good biocompatibility, and the matureness of micro-fabrication technologies (Li et al., 2018). Griss et al. (2000) used silicon MN array for EEG signal recording in 2000. Forvi et al. (2012) also employed silicon MNs that have good performance in recording EEG, EMG and ECG, but silicon electrodes need to be coated with metal to achieve a relatively lower impedance. In addition, the silicon material is hard and lacks flexibility, resulting in the unfitness of electrodes on the body skin, and concomitant motion generated artifact signal (Tachikawa et al., 2010). Wang et al. developed flexible parylene-based MN electrode arrays with silicon MNs, which could provide not only conformal but also robust contact (Wang et al., 2017). Compared with wet electrode, a good impedance density of 7.5 KΩ cm² @10 Hz was attained. Flexibility of P-MNEA contributes to recording stability, and P-MNEA realizes credible EEG acquisition.

Metal materials are also good candidate materials for MEA fabrication. For example, MEA with MNs with a length of 550 μm could be constructed with a magnetization-induced self-assembly approach (Wang Y. et al., 2021). The MNs could penetrate the cuticle and reach the human epidermis for high-quality signal recording. In addition, the appropriate MN penetration length will not induce harm to the human dermis; therefore, painless signal acquisition could be achieved. The improved performance of MEA in EMG and ECG recordings has been demonstrated experimentally in both able and amputee subjects. Inspired by the edge of golden margined century plant leaf, a kind of flexible dry biological electrode with MN structure was proposed (Figure 8) (Li Y. et al., 2021). The red copper sheet with 100 µm in thickness was selected as a material for the MN array due to its excellent conductivity and processability. The results showed that compared with the triangular MNs, the curved hook MNs exhibited smaller penetration resistance, larger pullout force, better damage resistance, and stronger bonding force with a flexible substrate. Compared with Ag/AgCl wet electrode, the flexible dry biological electrode with curved hook MN structure showed better time domain and frequency domain performance in EMG signal acquisition.

Apart from silicon and metal, polymers could also be considered for MN electrodes. A novel MN array–based dry electrode modified with a ring-shaped flexible polyimide substrate matrix was fabricated by Li J. et al. (2021), which was compatible with EEG signal recording on hairy scalp. The loop structure allows the hair to be passed through the MN attached to the back. The MN on the “ring” can be pressed from behind to directly touch the scalp. MNs with pointed tips, also based on polyimide, can penetrate the cuticle, the outermost layer of the skin, enabling low-impedance contact and minimal invasiveness. Satti et al. (2020) introduced a rigid parylene-coated MN electrode array for the construction of a portable ECG circuit, which was able to monitor the process of ECG reducing the motion artifacts. In comparison with the traditional Ag/AgCl electrodes, the prepared MNE showed an increased stability and durability for dynamically and long-termed monitoring ECG. The microneedles could survive the compression force until 16 N and successfully penetrate skin tissue with an insertion force as low as 5 N. The electrical properties of the MNE were characterized using impedance spectroscopy under the equivalent circuit model. The resultant wearable wireless MNE demonstrated the potentials for ECG recording with the reduction of the movement artifact noises in the process of dynamic behaviors. A recent case was reported that a flexible MAE with a Miura-ori pattern was constructed for monitoring biosignals accurately and stably (Hou et al., 2021). The mold was directly prepared using high-precision machining. The mold was consisted of micro-level needle tips and a matrix macro-level Miura-ori structure. The flexible M-MAE patches were extruded with a further PDMS mold pressing process. A series of tests of the EII and flexibility supported this minimally invasive M-MAE with outstanding biosignal sensing performance as the result of smaller skin-electrode impedance and more stable sensing signal output against bending. The M-MAE possessed benign airiness due to air-permeable channels, facilitating to get rid of sweat timely and stabilize signal output. The M-MAE sensing patch also showed to have in situ practicability and long-term stability to realize real-time biosignal recording of ECG and EMGs.