The lateral flow immunoassay (LFIA) was emerged as a low-cost, rapid and simple detection method at the end of 1960s, and it has been used for analyzing small biomolecules at point-of-care-testing (POCT). The basic principle of it is to attach different labels to biomarkers (antigens or antibodies) to achieve macroscopic visualization or detection of microscopic immune responses by means of signal amplification. In 1976, the first radioimmunoassay based test strip was developed for the detection of human chorionic gonadotropin (hCG) in urine, and was introduced into the market in the United States. However, it wasn’t affordable for most of the average families at that time. In 1990s, in place of radioactive labels, Osikowicz et al. [1] used colored colloidal selenium for one-step chromatographic immunoassay determination of hCG in urine, which to a large extent reduce the fabrication cost. Since then, varieties of novel test strips have flooded into medical market for rapid immunoassay. Compared with traditional biochemical detection methods, LFIAs require no professional operation, much less detection time and flexible using conditions. However, it conventionally suffered from poor sensitivity and reproducibility, because the intensity of the produced signal is relatively low. Besides, since the colorimetric signals cannot be quantitatively determined by naked eyes, it was hard for LFIAs to achieve quantitative detection [2]. Hence, a growing concern has focused on the novel materials for labelling to realize signal amplification and quantification in LFIAs.

Nanomaterials, as an emerging type of material featured with size ranging from 1 to 100 nm, are considered to be ideal labels in bioanalysis and biosensing because of their unique physical and chemical properties. On the one hand, due to their much larger specific surface area than normal macroscopic particles, nanoparticles (NPs) can provide a large amount of space for surface modification, [3] and various molecular moieties can be attached to control their physical and chemical properties. On the other hand, NPs have unique advantages in optical, magnetic, electrical or thermal properties, which enable them to act as signal producers when they are combined with different analytes. The introduction of NP labels simplifies the inspection process of biomolecules and makes LFIAs more sensitive and selective. Currently, most commonly used nanomaterial in LFIAs is colloidal gold. Qualitative analysis can be achieved by observing the strip color with naked eyes. Although the method is simple and fast, it is difficult in accurate quantification [4,5]. In recent years, much effort has been made to develop new kinds of nanoparticle labels in order to make LFIAs more sensitive and accurate, and to be more adaptable to the today’s POCT. We have summarized the parameters of each nanoparticles mentioned in this paper in terms of their practical use in LFIAs, which is shown in Table 1. This review paper will briefly give a description of the structures of the LFIA test strips and illustrate the principles of two main detecting formats. And then, we will comprehensively discuss the state-of-the-art nanomaterial labels used in LFIA based on different detecting signals. At last, the conclusion and outlook will be highlighted. Compared with the previous reviews in this field [[73], [74], [75]], this paper generated a more comprehensive review of the current advanced nanomaterial labeling technology in LFIAs. Furthermore, we emphasized the importance and huge application prospects of multiplex detection in future’s LFIAs.

LFIA is a rapid diagnostic technology developed from enzyme 1inked immunosorbent assay (ELISA). It has the advantages of simple operation, low cost and short detecting time. The test strip of LFIA is combined with the sample pad, conjugate pad, nitrocellulose (NC) membrane and absorption pad (shown in Fig. 1). These are the places where immune reactions occur. We summarize the functions of each component as follows (shown in Table 2).

The general principle of LFIA is based on antibody-antigen specific interaction [5]. Specific materials are initially labelled on the antibodies. The labelled antibodies are then dispensed on the conjugate pad and will flow forward through capillary action when the sample liquid is applied on the sample pad. The liquid will pass through the test areas of the strip and finally reach to the adsorption pad at the end. Test line (TL) and control line (CL) are drawn over the NC membrane where the test result is displayed. We observe the test line to analyze the results and use control line to decide if the test is valid. In LFIAs, two formats are most commonly used for detecting: i) sandwich format, ii) competitive format.

In sandwich format LFIAs, three different antibodies are utilized. The primary antibody (Ab₁) and secondary antibody (Ab₂) are immobilized on the test line and the control line, respectively. When the sample liquid contains the target antigens (Ag), the Ag specifically bind to the labeled antibodies (Ab) on the conjugate pad to form an Ab-Ag complex. The Ab-Ag complex flow forwards on the NC membrane and will be captured by the Ab₁ on the TL to form a sandwich format Ab-Ag-Ab₁. As the Ab is labeled, we can examine the signals appearing on the TL to analyze the test result. As the liquid continues to move forward, the Ab₂ immobilized on the CL which is known as a species-specific anti-immunoglobulin antibody, can catch the Ab to form the Ab-Ab₂ complex. So, if the test is valid, signals always appear on the CL whether or not there are antigens contained in sample.

It is easy to understand that the sandwich format of LFIA requires at least two antigenic sites for the antigen, and thus, it is only suitable for large analytes.

Competitive format is used for testing small analytes which have only one antigenic site. Generally, this format utilizes two different antibodies and the format has two arrangements. For the first arrangement, the labeled antibody (Ab) is immobilized on the conjugate pad and the species-specific anti-immunoglobulin antibody (Ab₂) is immobilized on the CL. Unlike the sandwich format, the TL in the competitive format is coated with the target antigen. When there are antigens in the sample liquid, the labeled Ab specifically bind to the antigens and flows through the NC membrane. Due to competitive inhibition, the antigens on the TL will not catch the flowing Ab-Ag complex, and thus signals on the TL decreases as the antibody concentration increases. In the case of no antigens contained in the sample, the labeled Ab is captured by the Ag on the TL, where strong signals will be produced. For the second arrangement, Ab is immobilized on the TL to capture the labeled Ag on the conjugated pad. When the sample liquid applied contains the unlabeled Ag, signals on the TL will greatly reduce to show a positive result because the unlabeled Ag bind onto the TL. Similar to the sandwich format, the CL is used for the validation of the test because of the formation of Ab-Ab₂.

For NP labeled LFIAs, it is critical to combine NPs and biomolecules in a stable manner. In the past few years, many studies have paved the way for the effective bonding process. Molina-Bolivar et al. [6] compared two approaches for immobilizing IgG antibodies on latex particles: physical adsorption and covalent binding. The results indicated that the covalent immunolatex had better immunoreactivity and stabilization. Gersten et al. [7] achieved the directly coupling of antibodies by the mediation of protein A, and MasanoriYoshioka et al. [8] introduced the antibody coupling mediated by glutaraldehyde served as a bifunctional coupling agent. However, these methods are easy to cause agglomeration of particles during the preparation process, and thus affecting test results. Currently, the most common and mature coupling method is EDC/NHS activation method, which is proposed by Steve Howell et al [9]. They used N-hydroxysuccinimide (NHS) and N'-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) to activate the carboxyl group of the particles, which made it possible to link the amino group of the protein without introducing excess carbon chain. A large number of experiments have proved that this method has good stability and is easy to operate.

Based on the above achievements, various kinds of nanomaterials have been used in LFIAs for sensitive detecting, such as colloidal gold, fluorescent NPs, Raman probes, magnetic NPs and carbon NPs.

Colloidal gold refers to a gold nanoparticle suspension with a particle diameter between 1 to 150 nm. The gold nanoparticles (AuNPs) are commonly prepared by reducing the chloroauric acid (HAuCl₄) solution. Different reduction methods determine the particle size, which leads to different color exhibition. AuNPs are easy to conjugate with immunoglobulins, toxins, antibiotics, hormones and other biological macromolecules through electrostatic interaction without affecting their biological activity.

Colloidal gold LFIAs (CG-LFIAs) emerged in the early 1980s, and the detection method is mainly based on color observation. Owing to the advantages of easy synthesis, rapid detection and low cost, CG-LFIAs have been widely used in POCT such as medical diagnosis [[10], [11], [12], [13], [14]], toxic detection [[15], [16], [17], [18]], food safety [[19], [20], [21], [22], [23], [24], [25]] and other protein examinations [[26], [27]].

Because AuNPs are colorimetric labels, the analytical parameters of CG-LFIAs depend significantly from size of AuNPs. Sahoo et al. [12] discovered that the size of AuNPs mainly depended upon the concentration of sodium citrate. Besides, gold salt concentration, optimum pH and temperature in preparation also influenced the final particle size. Kim et al. [13] further demonstrated the best size of AuNPs for CG-LFIAs in detecting hepatitis B surface antigens. They used a seeded growth method in AuNPs preparation to control their size. As a result, within a particle size ranging from 34.2 nm to 137.8 nm, AuNPs of 42.7 ± 0.8 nm exhibited superior performance for the detection of CG-LFIAs.

In order to improve the sensitivity of CG-LFIAs, some studies have used enzymes as carriers to amplify the detection signal. Parolo et al. [28] proposed to use enzyme-modified AuNPs as labels. They demonstrated that the ‘enhanced label’ produced a more intensive color than that of the traditional AuNPs, and the sensitivity increased by an order of magnitude. Additionally, Panferov et al. [29] realized an ultrasensitive detection of potato virus X by dropping alkaline phosphatase (ALP) on the test strip to enhance the color intensity. They improved the detection limit by nearly 30 times, and the whole preparation process was easy to operate.

Another approach of enhancing sensitivity is to use surface-modified core-shell NPs as labels. Lu et al. [30] developed AuNPs coated with SiO₂ with high stability. The as-prepared NPs could label antibodies in a wide range of concentration, and the experiment used alpha-fetoprotein (AFP) and vanillin as samples to demonstrate the applicability. The result showed that the detection limit of AFP and vanillin decreased down to 300 pg/mL and 100 ng/g in LFIAs, respectively. Furthermore, Lin et al. [31] utilized polyethylene glycol (PEG) to modify the surface of AuNPs for the detection of Bisphenol A (BPA) (shown in Fig. 2). Because PEG has amphiphilic nature, the modification process on AuNP surfaces mainly depended on primary bonds which proved to be more reliable. Besides, the PEG-coated nanoparticles were capable of maintain good stability in a changing environment. As a result, the system could reach a naked-eye detection limit of 0.8 ng/mL, about 13 times better than that of regular AuNPs based LFIAs.

It has always been challenging for CG-LFIAs to realize quantitative detection because it is hard to accurately read out the color intensities. Liu et al. [32] for the first time applied the LFIAs in quantifying chromium ions (Cr) in water and serum samples. They examined the reflected light signals from AuNP conjugates to obtain a linear detection range from 5-80 ng/mL. Although this range was relatively small and the gradient is not accurate enough, this work opened the door for biomonitoring of Cr exposure. For the purpose of improving the sensitivity and detecting range, Oh [33] et al. developed a trap LFIA where the deletion and detection zones replaced the traditional test and control zones. When the sample was applied, the detection zones could capture the labeled analytes while the deletion zones trapped unbound ones. By measuring the ratio of color signal intensities from the two zones, the as-prepared assay achieved to detect cortisol in human saliva with a wide linear range from 0.01-100 ng/mL. By introducing an electron microscope and a scattering analyzer into the LFIAs, Yu et al. [34] realized the quantitative detection of ciprooxacin (CIP) in animal muscle. The highlight of this work went to the simple operation and good stability in examining varieties of meat samples, which showed great application prospects in POCT for food safety.

To make CG-LFIAs more portable, several researchers have developed smartphone-based system for on-site detection. Yu et al. [35] detected alkaline phosphatase (ALP) in milk samples based on a smartphone camera. Hou et al. [36] developed a dual-modality imaging system for the quantification of hCG and carcinoembryonic antigen (CEA). The system could not only detect colorimetric signals of AuNPs, but also read florescent signals of quantum dots. As a result, a good detection linearity was obtained in both labels, and the assay proved to be ideal in POCT diagnoses due to its high sensitivity and portability. Furthermore, Xiao et al. [37] reported a smartphone’s ambient light sensor (SPALS) in CG-LFIAs for detecting cadmium ions (Cd²⁺), clenbuterol and porcine epidemic diarrhea virus (PEDV). In their work, the SPALS recorded the transmitted light signals from the test strips and calculated the intensities which were finally displayed on the smartphone (shown in Fig. 3). The test result was ideal, and this approach was outstanding in easy operation and low in cost.

CG-LFIAs can also be used for multiple analytes detection [[38], [39]]. Xu et al. [40] reported a LFIA with two test lines to simultaneously detect fumonisin B1 (FB1) and deoxynivalenol (DON). In their work, they utilized silver staining on AuNPs to further enhance the signal intensity. As a result, the detecting sensitivity was demonstrated to be at least 2 times better than those traditional AuNPs-based assays. Additionally, Sun et al. [41] proposed a test strip with three test lines for detection of aminoglycoside residues in milk. Three specific antibodies conjugated with AuNPs were dispensed on the conjugate pad. The whole measuring process was proved to be rapid and sensitive without any cross-reaction.

Currently, although CG-LFIAs are the most widely used in the medical market and have been developed to detect a large variety of biomolecules, most of them can only meet qualitative or semi-quantitative detection in real clinic, which largely restricts their commercial prospects at quantitative POCT (such as detection of some tumor markers or cardiovascular markers). Thus, how to promote the quantitative CG-LFIAs to the medical market with easy operation and low cost has been an attractive research trend.

Compared with colorimetric markers, LFIAs based on fluorescent signals (FLFIAs) has the advantages of high sensitivity, good stability and quantitative ability. For the improvement of the FLFIAs, different luminescent NPs have been studied and developed.

3.2.1. Quantum dots

Quantum dots (QDs) are a class of semiconductor nanoparticles composed of elements of group III-V and group II-VI. Because the electrons and holes in QDs are quantum confinement, the continuous band structure becomes a discrete energy level structure with molecular characteristics and can emit fluorescence after being excited. Due to their high specific surface area, QDs have quantum size effects where the size strictly controls the light absorption and emission characteristics. Compared with organic dye fluorescent labelling materials, QDs have adjustable emission wavelength, wide absorption cross section, strong fluorescence emission intensity, high quantum yields, and minimal photobleaching and photodegradation [42,43].

Since the year of 2010, QDs-based LFIA has been developed for qualitative or quantitative POCT. Many virus [[44], [45]], proteins [[46], [47], [48]], nucleic acid [49], and medicine materials [[50], [51], [52]] can be detected with high sensitivity and specificity. Moreover, studies have proved that compared with colloidal gold labels, LFIA with QD labels has a lower detection limit [44,53].

In order to enhance the sensitivity of FLFIAs, Shen et al. proposed the synthesis of multishell nanoparticles to form core/shell structure of the QDs [54]. They wrapped the ZnSe/3CdSe core with a CdS/Cd_xZn_{1 - x}S/ZnS multishell to suppress exciton leakage. The experiment showed that the fluorescence quantum yields rose from 28% to 75% (Fig. 4), and the system remained stable under various physical conditions and a wide range of pH. As a result, they achieved to detect human hepatitis B surface antigen (HBsAg) by using this core/shell QDs with a sensitivity of 0.05 ng/mL.

Based on the core/shell structure, CdSe/ZnS QDs have also been developed as labels in FLFIAs [[55], [56], [57], [58]]. Anfossi et al. [58] proposed a fluorescence-quenching LFIA (FQLFIA) to detect fumonisin mycotoxins in maize flour. In the experiment, a ‘turn on’ strategy was used where they utilized metal nanoparticles as quencher to suppress the fluorescence when adding a negative sample. Once the target analytes are contained in the sample, the quencher will be displaced and the fluorescence intensity of the QD labels will recover proportionally to the amount of the target analytes. This FQLFIA with positive readout mode was proved to be highly sensitive and can reach a visual detection limit at the ng/mL level.

Another core/shell structure is based on cadmium-free synthesis [[59], [60], [61]], which proves to be environmentally friendly. Shen et al. [59] successfully synthesized QDs with photoluminescent CuInZn_xS_{2 + x}/ZnS. The relative quantum yields reached to above 40% when the core/shell QDs were transferred into water by an organic-aqueous phase transfer method. Furthermore, Wu et al. [61] carried out a two-time separate shell growth process by using CuInZn_xS_2+x as core materials to form CIZS/ZnS//ZnS QDs. They demonstrated that the double-shell QDs had excellent quantum yields and the system could realize quantitative detection of CRP with a limit of detection of 5.8 ng/mL.

Huang et al. [62] proposed a novel pitaya-type silica porous structure loaded with high-density QDs for ultrasensitive and robust FLFIA (shown in Fig. 5). Unlike the traditional sandwich-type structure, they embedded different colored QDs into silica spheres and used dense silica shell for protection. As a result, the system could quantify the CRP concentrations with a linear range from 0.125 to 300 ng/mL. Chen et al. [63] combined MnO₂ as the nanosheets with Biotin-QDs to quantitatively detect glutathione (GSH). The result indicated that the Biotin-QD-MnO₂ nanocomposite is highly sensitive and cost-effective.

For the purpose of realizing simultaneous detection, Foubert et al. [64] studied the difference between colloidal gold and QDs as labels in the multiplex detection of mycotoxins. The result demonstrated that QDs-based LFIAs had higher sensitivity and less consumption. Taranova et al. [65] designed a FLFIA in a ‘traffic light’ format where three lines of different colors were immobilized on a test strip. The water-soluble QDs were selected as the label, and the test achieved to determine the levels of three antibodies in milk according to the fluorescence intensity on the three TLs. Beloglazova et al. [60] for the first time proposed non-cadmium QDs to simultaneously detect two mycotoxins in maize and wheat. The as-prepared QDs were based on InP/ZnS core/shell nanostructure and were highlighted as environmentally-friendly. Two TLs were used to realize qualitative detection of the anlaytes with a cutoff value of 50 and 500 μg/kg for zearalenone (ZEN) and deoxynivalenol (DON), respectively. In order to achieve accurate quantitative detection, Chen et al. [66] simultaneously measured fluorescence peak intensities on the TLs and the CL to quantify CYFRA 21-1 and CEA in human serum. Qi et al. [67] utilized a dual-QDs-labelled FLFIA to realize rapid and quantitative detection of procalcitonin (PCT) and CRP.

Furthermore, Shao et al. [68] presented a novel quantum dot nonobead (QB)-based multiplexed immunochromatographic assay (QB-ICA) for the detection of aflatoxin B1 (AFB1) and ZEN (shown in Fig. 6). A streptavidin (SA)-biotin system was used in their work to display fluorescent signals. By obtaining the ratio of signal intensities on the two TLs and the CL, the QB-ICA system could achieve accurate quantitative analysis of AFB₁ and ZEN with a detection limit as low as 1.65 pg/mL and 59.15 pg/mL at 10% competitive inhibition concentration, respectively.

However, there are still some limitations in the application of QDs. First, the modified QDs have a relatively large particle size and are likely to clog during the lateral flow process. Second, QDs are prone to agglomeration when coupled with biomolecules, thereby reducing their stability and sensitivity. Additionally, the heavy metal elements contained in QDs make them potentially toxic, and a green method for synthesis needs to be concerned in future research.

3.2.2. Up-conversion nanoparticles

The up-conversion nanoparticles (UCPs) are a kind of luminescent nanomaterials doped with rare earth elements which can convert near-infrared (NIR) excitation light into visible or ultraviolet-emitting light [69]. Among them, NaYF₄ double-doped with Yb and Er is the most commonly used UCPs in LFIAs. Yb³⁺ absorbs infrared photon energy and transmits it to the Er³⁺ in a non-radiative form to emit visible fluorescence. Since UCPs have poor water solubility and dispersibility, chemical modification on their surface is necessary. Compared with QDs, UCPs have the advantages of low toxicity, anti-Stokes shifts, narrow emission spectrum, and anti-interference from biological autofluorescence, which are currently one of the most ideal fluorescent labels in FLFIAs [70].

Niedbala et al. [71] and Hampl et al. [72] for the first time utilized UCPs as labels in LFIAs for the detection of Escherichia coli O157:H7 and hCG, respectively. In their works, 10³ org/mL E. coli O157:H7 organisms were detectable in a culture medium, and 10 pg hCG was detected in a 100 μl sample. In order to enhance the sensitivity of the test, researchers have found that NaYF₄ was the most efficient matrix of UCPs among other up-converting materials because of low phonon energy of lattices [[73], [74], [75], [76], [77], [78], [79]]. Zhao et al. [77] reported an ultrasensitive UCP-based LFIA by using NaYF₄:Yb,Er nanoparticles as fluorescent labels. They chose V. anguillarum as the target analytes to demonstrate the performance of the system. The results showed that the system reached a low detection limit with a detection time of only 15 minutes. Furthermore, Liang et al. [78] prepared β-NaYF₄:Yb³⁺,Er³⁺ UCPs codoped with Li⁺ and K⁺ as the high-intensity luminescent labels in FLFIAs. The experiment indicated that the strongest green and red emissions of the codoped UCPs are at least 7 times higher than that of undoped UCPs. Additionally, Kim et al. [79] proposed another type of ion doping UCPs. In their study, Ca²⁺ was used as a heterogeneous dopant ion to form Tm@NaYF4:Ca for the enhancement of the up-conversion photoluminescence. As a result, the NaYF4:Yb,Tm@NaYF4:Ca core/shell UCPs emitted strong near-infrared light at 800 nm when applying an excitation at 980 nm (shown in Fig. 7). The as-prepared LFIA platform achieved to detect avian influenza virus (AIV) with a detection limit down to 10³ EID₅₀/mL even in dark brown-colored samples.

Like the QDs-based LFIA, quantitative detection can also be achieved in UCPs-based assays by measuring the fluorescent intensities on the TLs and CLs. Yan et al. [80] collected the emitted visible light and converted it into voltage signals. They demonstrated that the voltage ratio VT/VC of the test line and the control line was directly proportional to the amount of Yersinia pestis in the sample. As a result, a good linearity was obtained after mathematical processing. Similarly, Qu et al. [81] utilized the value of T/C ratio to realize rapid and quantitative detection of Brucella by UCP-based FLFIAs. They reached an ideal detection limit in different spiked samples, and the assay was proved to be high specificity, reproducibility and stability. Moreover, Hu et al. [82] proposed the use of UCPs for the detection of drugs. In their work, morphine (Mop) and methamphetamine (Met) from simulated-saliva samples were selected as target analytes, and they realized accurate quantification by calculating T/C ratios on the strip.

Yang et al. [83] for the first time proposed a novel FLFIA based on an electrochemical luminescence assay for quantitative detection of NT-proBNP in blood. The result showed a good linear range of NT-proBNP from 50 ng/L to 35000 ng/L with high sensitivity. This FLFIA system has a promising application prospect in POCT for clinical diagnosis of acute heart failure.

Furthermore, UCP-labels have great potential in multiplex detection [[84], [85], [86], [87], [88]]. Zhang et al. [86] realized rapid quantification of three pathogens in one test (Bacillus anthracis, Brucella spp., and Y. pestis). The assay showed high specificity and excellent stability in a wide range of pH conditions, which was ideal for POCT in the application of first-level emergency response. Additionally, Zhao et al. [87] developed a ten-channel UCP-LFIA to simultaneously detect foodborne pathogens (shown in Fig. 8). They integrated ten strips into a disc to make the system more portable and less consuming. As a result, 10 kinds of foodborne pathogens were quantified in a wide range of concentrations within 20 minutes. Recently, Zou et al. [88] studied a sensitive LFIA system by labelling broad-specific monoclonal antibodies with UCPs. The assay could achieve multiplex detection of three organophosphorus (OP) pesticides with a linear range from 0.98 to 250 ng/mL, which is expected to be a great screening tool in POCT for food samples.

However, the biggest problem existed in UCPs-based LFIAs is that the preparation of UCPs is extremely harsh and complicated. The whole process should be in vacuum or under the protection of inert gas, thus inducing a high cost. Furthermore, it is still important for future research to find more suitable doping materials, and it is also fundamental to study their up-conversion characteristics under laser excitations of different wavelengths.

3.2.3. Time-resolved fluorescence nanoparticles

The time-resolved fluorescence nanoparticles (TRFNPs) are prepared by doping lanthanides into the nanoparticles and modifying them with chelates with bifunctional groups. Lanthanide chelates are extremely suitable for labelling time-resolved fluoroimmunoassay (TRFIA) due to the unique fluorescent properties. Firstly, the fluorescence decay time of lanthanide chelates is extremely long, around 10³ -10⁶ times that of conventional fluorescence, which makes time resolution feasible. Secondly, lanthanide chelate fluorescence has a large Stokes shift and a broad excitation light spectrum, greatly reducing background interference and improving sensitivity of the detection. Hu et al. [89] for the first time systematically compared the performance of different types of lateral flow assays based on colloid gold, quantum dots and time-resolved fluorescent nanobeads in the quantification of ractopamine (RAC). They demonstrated that the TRFNPs based assay achieved the best detection limit and the widest linear range with the least reagent consumption and detection time.

Currently, the most commonly used chelate label in TRFIA is based on Eu. Huang et al. [90] measured the fluorescence counts of Eu-chelate labels in a TRFIA to quantify the concentration of the anti-PLA2R-IgG in sample. As a result, the system achieved a linear detection range of 0.03-340 mg/L and could be used in the diagnosis of idiopathic membranous nephropathy. In order to reduce interferences, Näreoja et al. [91] utilized a step of protein corona stabilization and spot-coated configuration to decrease the nanoparticles nonspecific binding. The as-prepared assay can realize the detection of thyrotropin (TSH) with a wide linear range and a limit of detection of 60 nU/L. By introducing the Eu (III)-doped polystyrene nanoparticles as labels, Hu et al. [92] proposed another rapid and ultra-sensitive TRFIA for detection of sulfamethazine in milk. In their experiment, SM₂-GA-BSA was sprayed on the TL as the detective antigen because it had a better affinity constant to monoclonal antibodies (KD = 3.214 × 10^-5 M). As a result, the system could provide a limit of detection down to 0.0045 ng/mL with a linear range from 0.05 to 10 ng/mL.

Dual-labelling technology of lanthanide ions helps TRFIAs to achieve multiplex detection, where Sm ions often serve as the second NP label. Because of the cofluorescence enhancement principle, dual-labelled TRFIAs showed better sensitivity. Wu et al. [93] presented a quantitative detection of TSH and T4 in human serum based on the dual-labelling technology (Eu³⁺ and Sm³⁺). The use of biotin further improved the sensitivity of the test. Results showed that a linearity of 0.21-80.00 mIU/L for TSH and 20-300 nmol/L for T4 were achieved with a good correlation. Additionally, haptoglobin (Hp) and C-reactive protein (CRP) were simultaneously measured by Gutiérrez et al. [94] with high sensitivity based on the dual-labelling TRFIAs.

Hou et al. [95] proposed a novel dual-label based TRFIA by using magnetic particles as an immobilization matrix and a means of separation (shown in Fig. 9). The antibodies were covalently coupled to the surface of magnetic nanoparticles rather than physical adsorption. This induced symmetrical binding sites for antigen and minor spatial hindrance for mass transfer. By this way, antibody-antigen binding equilibrium can be achieved rapidly. This assay realized the simultaneous detection of α-fetoprotein (AFP) and free β-hCG in human serum with a wide dynamic range (0.1-750 ng/mL for AFP and 0.16-450 ng/mL for hCG) and a low detection limit. Furthermore, owing to the advantages of magnetic particles, it is convenient to separate the immunocomplex and unbound components by applying a magnetic field.

Recent years, scientists have put more emphasis on applying TRFIAs to clinical POCT. The simultaneous detections of β2‐MG and ferritin [96] /β2‐MG and cystatin-C [97] in human serum based on the sandwich format (for detecting β2‐MG) and the competitive format (for detecting ferritin/cystatin-C) have been proposed. Both experiment results showed good consistency with the commercial assays. Furthermore, Zhang et al. [98] presented a simultaneous quantification of anti-mycoplasma pneumoniae (MP) IgM and IgG based on dual labels. The assay reached a linear range from 2-5500 BU/mL and 1.5-1500 BU/mL for IgM and IgG, respectively, and could be used in the early diagnosis of MP infection.

Although TRFIAs show the best sensitivity in fluorescence detection [89], it has high demand for the performance of the testing device because a quick and precise capture of the resolvable fluorescence is necessary before the end of its lifetime. Currently, most of the devices are expensive and cumbersome to operate, making it difficult for market promotion.

The surface-enhanced Raman scattering (SERS) has been emerged as a powerful tool for molecular analysis with high sensitivity and specificity. The Raman scattering spectrum carries the fingerprint information of the molecular structure and its intensity can be enhanced by using nano-structured metal surfaces as substrates. The nano-gap on the surface of the metal makes the Raman signal intensity 10¹⁰ to 10¹⁴ times higher than traditional Raman scattering due to the “hot spot” effect which is induced by the localized surface plasmon resonance (LSPR) [[99], [100], [101]]. By introducing SERS-active nanoparticles into LFIAs, SERS-LFIAs have become extremely suitable for strong specific and ultra-sensitive detection of various biomolecules [[102], [103], [104]].

In order to enhance the sensitivity of the SERS-LFIAs, different shapes of the SERS nanotags have been investigated by studies [[105], [106], [107], [108]]. Hwang et al. [106] utilized hollow gold nanospheres (HGNs) as SERS-active labels to improve the detecting sensitivity by three orders of magnitude compared with the conventional LFIA. They achieved the quantification staphylococcal enterotoxin B (SEB) with a limit of detection as low as 0.001 ng/mL. Khlebtsov et al. [107] developed a gap-enhanced Raman tags (GERTs) to further enhance the SERS intensity. The Raman molecules were embedded in a 1-nm gap between the gold nanorod core and gold shell, and the non-spherical gold nanoparticles acted as the reactants for multi-color LFIAs. With the increase concentration of applied cardiac troponin I (cTnI), the color on the test line became brighter. As a result, the embedded Raman particles showed an order of magnitude stronger SERS response than traditional SERS tags in LFIAs. The detection limit of cTnI could reach down to 0.1 ng/mL, indicating a great potential in early diagnosis of heart disease. Additionally, Fu et al. [108] proposed a flower-like gold nanoparticles to serve as SERS-active labels for the detection of β-adrenergic agonist brombuterol (BB) (shown in Fig. 10). The prepared nanotag displayed a strong signal enhancement due to the anisotropic distribution of the electromagnetic field near the surface, and an ideal result was obtained.

Another enhancing method is to use core/shell structures on the surface of SERS-active NPs. Blanco-Covián et al. [109] designed Au@Ag nanoparticles as SERS labels in the LFIA for the detection of penumolysin. Jia et al. [110] presented an attractive model based on core/shell structure where the Au@Ag NPs were dual-dyed with DTNB. They demonstrated that this SERS tag was of high bioconjugation flexibility, and the assay achieved to detect IgM with a limit down to 0.1 ng/mL, two orders of magnitude higher than the sensitivity of colorimetric LFIAs. Furthermore, Bai et al. [111] developed a SERS-LFIA system based on four kinds of nanoparticles synthesized by Au/Ag. By comparing the performance of each nanoparticles, they concluded that nanoparticles in Au@Ag-Au structure had the highest SERS activity as shown in Fig. 11, and the system could realize a detection limit of cTnI of 0.09 ng/mL, 50 times lower than visual results.

By measuring the Raman peak intensity, nearly all of the SERS-LFIA systems can perform accurate quantitative detection. Besides the references that have been discussed above in this section, studies on quantifying HIV-1 DNA [112], TSH [113] and CRP [114] have also been reported based on SERS nanotags. Tran et al. [115] presented a portable SERS-LFIA reader for the ultra-sensitive and quantitative detection of hCG (shown in Fig. 12). The assay achieved a detection time of only 5 seconds, several orders of magnitude shorter than conventional Raman instruments, and the sensitivity was 15 times higher than that of CG-LFIAs. Moreover, the whole system was cost-effective, contributing a lot to the affordable POCT in real world.

For simultaneous analysis, Wang et al. [116] used SERS-LFIA to achieve the simultaneous quantification of two different DNA markers (KSHA and BA) with a detection range from 0.1 pM-100 pM and 0.2 pM-100 pM, respectively. Liu et al. [117] realized sensitive quantification of Listeria monocytogenes and Salmonella enterica serotype Enteritidis based on Au@Ag NPs. Additionally, Shi et al. [118] proposed a sensitive SERS-LFIA for the detection of neomycin (NEO) and quinolones antibiotics (QNS) from milk samples with two AuNPs-based SERS probes. The results demonstrated that the system was highly sensitive and reliable.

Recently, a method of diagnosing multiplecardiac biomarkers has been proposed by Zhang et al. [119] In their work, test strips with three TLs were developed for simultaneous measurement of Myo, cTnI, and CK-MB. The test results were ultra-sensitive and the mutual interference proved to be little. Wang et al. [120] developed a SERS-LFIA to analyze two surface markers (CD19 and CD20) of B cell hematological malignancies. In the experiment, the sensitivity and reproducibility of SERS signals was evaluated firstly, and a dynamically monitoring of 13 patients for a month was implemented to verify the accuracy and long-term performance of the test. As a result, this detection method was of good linear correlations in quantitative analysis and could be used in POCT of minimal residual disease.

There is still a huge space for the development of SERS-LFIAs. Initially, severe background signal interference in SERS detecting remains an important issue to be solved. Thus, it is curtail to develop SERS-active substrates with high uniformity, stability and reproducibility. Secondly, proper modification method for SERS labels to eliminate non-specific adsorption in complex systems is significant for the improvement of the LFIA sensitivity. Besides, combining SERS immunoassay technology with biochips to establish a portable and ultra-sensitive detection platform could be a new trend for clinical POCT.

Magnetic nanoparticles (MNPs) (generally made from Fe₃O₄) are a new type of nanomaterials developed in recent years. Due to their good biocompatibility, they can bind to biomolecules such as enzymes and antibodies to serve as labels in the LFIAs (MLFIAs) [[121], [122], [123], [124], [125]]. Compared with fluorescent labels, LFIAs based on MNPs have very little background interference and do not need expensive and large-scale detection instruments. By analyzing the magnetic signals on the test strip with a Magnetic Assay Reader (MAR), rapid and quantitative detection could be performed on-site. Furthermore, since MNPs are paramagnetic, the labeled biomolecule can be directionally moved under an external magnetic field and the detection process becomes controllable [126].

For enhancing the sensitivity of the MLFIA, core/shell structures have been used in synthesizing MNPs. Tang et al. [127] proposed nano-Fe₂O₃ spherical particles wrapped with gold shell (MnGMs) for the rapid screening of aflatoxin B2 (AFB2). As a result, the detection limit of the assay was proved to be three times lower than conventional CG-LFIAs. Additionally, Liu et al. [128] developed gold magnetic nanoparticles (GMNPs) to perform the testing for single nucleotide polymorphisms (SNPs), which were traditionally detected by expensive and sophisticated instruments. The GNMPs based assay was combined with an amplification refractory mutation system (ARMS) and exhibited high sensitivity and specificity in genotyping pathogenic SNPs. On the basis of similar principles, Yan et al. [129] used GMNPs in an ARMS-LFIA system for CYP2C19 genotyping. The detection method was proved to be low cost and easy to operate, which had great potential in clinical POCT. Furthermore, Zhang et al. [130] developed a novel particle structure, where they used Fe₃O₄ as the core and SiO₂ as the shell. The as-prepared MNPs showed ultra-sensitivity in the detection of HBs antigen, and an analyte concentration down to 0.1 pg/mL could be detected by using an MAR.

Razo et al. [131] presented the joint use of MNPs and GNPs to realize double enhancement for detecting the PVX in LFIA testing. Due to the good biocompatibility of MNPs and GNPs and their heteroaggregates, the experiment achieved enhancement in the concentration of analytes and visibility of the labels. Compared with a conventional LFIA, the sensitivity of this test was increased by 32-fold, which reached a limit down to 0.25 ng/mL, and an early detection of PVX based infections was realized when the symptoms are still in an inconspicuous state (shown in Fig. 13).

The biggest problem in MLFIAs is excessive magnetic enrichment. Once the MNPs are agglomerated on the conjugated pad, the liquid flow channel is blocked, which will greatly increase the detection time and influence the antigen-antibody combination. The use of super-paramagnetic NPs as labels is a promising solution to this problem. Super-paramagnetic nanoparticles refer to a kind of magnetic ions whose hysteresis loops feature no remanence and coercive force. Compared with general MNPs, super-paramagnetic NPs have unique advantages of greater specific surface area and lack of hysteresis. The former enables the super-paramagnetic NPs to link more functional groups for biomolecule combinations, and the latter ensures the disappearance of magnetic enrichment during handling and storage. Wang et al. [132] studied the effect of the types of super-paramagnetic nanoparticles on detection signals in detail. They designed a series of nanoparticles with different sizes and magnetite contents as labels in MLFIAs. Four types of MNPs (MNP133, MNP111, MNP204 and MNP300) are used for the measurement of hCG, and the magnetic signal intensities are displayed on the MAR. The result indicated that the size of super-paramagnetic nanoparticles decided the detection time, and the magnetite content determined the signal intensities. Wang et al. [133] proposed another lateral-flow immunological system for the detection of Bacillus anthracis spores. To analyze the detection result, an MAR was used for reading the magnetic signals. To make the test more intuitive and easier to operate, a three-in-one detection method was developed by this group [134] two years later. The novel system can realize optical, magnetic and naked-eye detection of Bacillus anthracis spores in only one test. This detection is on the basis of “Road Closure” principle (shown in Fig. 14), where the conjugates of B. anthracis spores and antibody-decorated MNPs will block the holes in the strip. Instead of forming the test lines on the strip like a traditional LFIA, a retention line is displayed in this experiment near the sample pad. As a result, high sensitivity and specificity detection of B. anthracis spores is achieved with easy operation and low cost.

In certain conditions, proper magnetic enrichment can be beneficial. Li et al. [135] proposed a rapid detection of viable Listeria monocytogenes by using the magnetic enrichment approach. In their work, a biotin-exposure-based immunomagnetic separation strategy (IMS) was used. The streptavidin-functionalized MNPs were anchored onto the target cells through strong noncovalent interaction with biotin. The immunomagnetic separation exhibited an excellent capability for sample enrichment. As a result, the magnetically-enriched viable L. monocytogenes could be directly detected on a test strip by naked eyes. The most highlight of this experiment is that the magnetic enrichment method reduced the consumption of antibodies by 10 times compared with conventional MLFIAs.

For Simultaneous detection, Lu et al. [136] synthesized a carboxyl-modified magnetic nanobeads for the detection of neuron specific enolase (NSE) and CEA. Two test lines were used on a test strip and the cross-reactivity was proved to be little. Orlov et al. [137] proposed a novel quantitative MLFIA for multiplex detection of Botulinum Neurotoxins. They integrated several test strips with different specificity into a cylinder and linked it to an MAR. Once the sample liquid was added, it would flow evenly through the multi-channel assay, and three kinds of magnetic signals could be simultaneously measured. This detecting method was demonstrated to be highly sensitive which could be extended to a wide range of POCT use.

Although MLFIAs own unique advantages compared with other methods, it has been a great challenge to prepare MNPs with suitable and even particle size. Larger size of MNPs would slow down the flow speed and may cause “false positives” while smaller particle size would result in weak magnetic properties, affecting the MAR detection result. Furthermore, magnetic nanomaterials tend to agglomerate after prolonged placement. Thus, how to improve and maintain their dispersion in water solvents could be worth studying.

Carbon nanotubes (CNTs), made of the hexagonal grid graphite and having a structure of spatial topology, are an emerging type of nanomaterials. They have great chemical stability and high specific surface area, and have been considered as an excellent candidate for building chemical sensors [138]. Meanwhile, after the discovery of CNTs and graphene, carbon nanoparticles (CNPs) were studied as a new member of carbon nanomaterials. They have shown great application potential in a wide range of fields due to their excellent mechanical, electrical and chemical properties [[139], [140], [141]]. Moreover, compared with GNPs or luminescent nanoparticles, carbon nanomaterials have advantages of environmental protection, easy preparation and good stability. Promisingly, a number of scientists are working on applying them to sensitive and quantitative LFIAs.

Qiu et al. [142] proposed a lateral flow biosensor (LFB) based on the multi-walled carbon nanotubes (MWCNTs) to realize rapid detection of DNA sequence (shown in Fig. 15). The MWCNTs served as labels, and the carboxyl groups on their surface linked to the amine-modified DNA to form the sandwich reaction format. By viewing the characteristic black bands of CNTs and using specific software, the biosensor could achieve qualitative and quantitative detection with a limit of 40 pM target DNA. Huang et al. [143] described a novel LFB by using magnetized carbon nanotubes (MCNTs) as labels. MCNTs were formed by the deposition of magnetite nanoparticles and could immobilize antibody against CA 19-9 to capture CA 19-9 in blood as the target sample. Due to the characteristic brown bands produced by MCNTs, the detection result could be observed by both naked eyes and a portable strip reader. Moreover, this system could detect CA19-9 in blood in a concentration of 30 U/mL, which is below the cutoff value (37 U/mL). Compared with individual CNTs, a more sensitive labelling method was presented by Liu et al [144]. They decorated CNT with gold nanoparticles (CNT/GNPs) as labels for detecting Squamous cell carcinoma antigen (SCCA). In their work, the CNT/GNPs showed great stability in LFIAs, and the assay achieved to measure SCCA under a detection limit of 3.03 ng/mL, far more sensitive than reported CNTs-based LFIAs.

Owing to the excellent electrical conductivity, CNTs are also used as the signal booster in the electrochemical LFIA (ELFIA). Du et al. [145] developed an ELFIA sensor to quantitatively detect organophosphorus (OP) pesticides and nerve agents based on enzymatic reactions. CNTs were utilized to further enhance the electrical signal and make the detection more feasible at low potentials. Zhu et al. [146] embedded CNTs into the conductive paper in the ELFIA for testing the DNA damages. The CNTs paper served as the working electrode and was connected to an electrochemical analyzer for quantitative detection of 8-hydroxy-2′-deoxyguanosine (8-OHdG) with a limit of 8.85 ng/mL in urine.

Based on CNPs, Wiriyachaiporn et al. [147] reported a rapid and sensitive detection of influenza A virus by inducing the accumulation of carbon nanotags on the test zones. The result could be directly identified with naked eyes and the sensitivity reached down to 350 TCID₅₀/mL. Takalkar et al. [148] proposed a novel fluorescent carbon nanoparticle (FCNPs) for the detection of DNA. The most highlight of the study was that the synthetic procedure of FCNPs was proved to be green and simple owing to the use of carbon material, and the test result could be quantified by measuring the fluorescent signals. Moreover, Zhang et al. [149] for the first time presented a multiplex detection of three fusarium mycotoxins by utilizing amorphous carbon nanoparticles (ACNPs). Unlike typical nanomaterials, ACNPs have a diameter more than 100 nm, but have unique advantages in LFIAs due to their strong dark color and high stability. As a result, the detection limits of this assay were demonstrated to be several times better than original AuNPs-based or QDs-based LFIAs.

This paper summarized the state-of-the-art nanomaterial labelling technology in LFIA testing. Nanomaterials have unique physical and chemical properties that can be adjusted by changing their size, shape and chemical composition at the nanoscale. The high specific surface area provides a large amount of space for nanomaterials to have their surface modified with different molecules or functional groups, making them important in biological applications. By introducing nanoparticles into LFIAs, lots of biomolecules can be detected on a test strip rapidly and sensitively. Nanoparticles served as labels in LFIAs need to have strong physical signals for identification (such as color, light, magnetism and electricity etc.) and great surface stability for modification. Among them, colloidal gold is the most commonly used in commercial market for a fast and convenient detection. However, it is difficult to perform a quantitative and sensitive measurement. Studies have focused on the luminescent nanoparticles, SERS-active nanoparticles, MNPs and carbon nanomaterials to enhance the sensitivity and realize accurate quantitative detection. A partial comparison is made between CG-LFIAs and these alternate nanoparticles labelled LFIAs in terms of their practical performance, where wet set the CG-LFIA as “gold standard” (shown in Table 3). With the rapid development of nanomaterial-based LFIAs, more and more diseases can be diagnosed at an early stage, which is extremely significant for today’s POCT.

In the future, more efforts should be taken to improve the simultaneous detection of multiple biomolecules. Currently, most of the test strips can only examine a single type of molecules in one test, which will inevitably increase the operational complexity for patients who require a multiple detection. LFIAs with quantitative and simultaneous measurement will not only shorten the detecting time, but also reduce sample consumption to save costs. Furthermore, if LFIAs can combine with the cloud-enable smartphones or TV, it is possible for patients to record their immune indicators at home and stay up to date on critical health information from experts. This should be an important trend in future’s in-home POCT.

Additionally, the signal stability needs to be improved. Some NP-based LFIAs have a short lifetime and are vulnerable to changes in physical and chemical properties of liquid sample. For example, the slight fluctuation of the solution pH will quickly cause colloidal gold to be “dead”, which is irreversible and irreparable. The “dead” colloidal gold aggregates into a black color and no longer has the labelling effect. Thus, novel nanoparticles are required to be developed with better signal stability. Before introducing a novel nanomaterial to LFIAs, several considerations are necessary to ensure that the material is suitable for labelling, such as i) the maximum quality of labels that may be conjugated to immunoreagent. ii) the signal stability in changing solution, iii) the preparation cost and complexity, iv) obstacles to means of reading signals. For example, nanoparticles featured with thermal properties tend to be stable, and the thermal signals are kind of easy-reading. However, LFIAs based on thermal signals are rarely reported [[150], [151]]. We look forward it to be a great potential for commercial use because this detection method tend to be simple and inexpensive.

In summary, LFIA is in a booming stage, and with the combination of nanomaterials, we firmly believe that this technology will bring a promising future to POCT.

The authors thank the financial support from the National Natural Science Foundation of China (51802060), Shenzhen Science and Technology Program (Grant No.: KQTD20170809110344233), Shenzhen Bay Laboratory (SZBL2019062801005) and Natural Science Foundation of Guangdong Province (No. 2019A1515010762).