1 Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was first reported in Wuhan, China, in November 2019, rapidly spreading worldwide and causing a global pandemic of coronavirus disease 2019 (COVID-19). This led to more than 500 million confirmed cases and more than 6.2 million deaths in more than 215 countries, as of April 28, 2022 [1]. The COVID-19 outbreak has had a significant economic and social impact on billions of people worldwide. Strong government sanctions, such as social distancing, quarantine, and lockdowns, have paralyzed businesses, severely weakening the global economy. This has caused many people to lose their jobs or has severely affected their household economy. Moreover, as the COVID-19 outbreak continues for an extended period, the medical community and staff, as well as the general public, are facing significant stressful situations (e.g., anxiety and depression) due to the intractable virus. Although effective vaccines and therapeutics for COVID-19 have been already developed and widely disseminated, transmission is still rapid and sustained owing to the emergence of the highly contagious variants of SARS-CoV-2 caused by mutations [2, 3].

Since the beginning of the COVID-19 pandemic, SARS-CoV-2 has accumulated mutations, and several types of SARS-CoV-2 variants identified by viral genome sequencing have been discovered in various global regions. Recent studies have shown that SARS-CoV-2 variants are more transmissible than the wild-type virus [4,5,6,7]. Additionally, the SARS-CoV-2 variants can bypass immune protection against exposure to the same virus conferred by previous vaccinations and infections [8, 9]. Fortunately, variant viruses do not continuously evolve to increase the risk of disease severity. The Omicron variant, which is currently dominant worldwide, is less severe than previous strains, especially compared to the Delta variant [10]. However, there is no guarantee that the subsequent dominant variants will not cause severe disease symptoms. In the future, devastating SARS-CoV-2 variants may emerge with higher viral transmissibility, disease severity, and vaccine-bypassing efficacy through the continued accumulation of mutations in the virus genome. To date, only a limited number of therapeutics, such as remdesivir and dexamethasone, have shown some effectiveness in reducing the mortality or severity of COVID-19. However, universally applicable therapeutics for asymptomatic and mild to moderate COVID-19 patients have not yet been developed. [11]. In the absence of universal antiviral therapeutics and reduced vaccine efficacy due to the immune escape of the variants, accurate, fast, and efficient diagnostic tools to identify patients with COVID-19 during the early stages of infection are vital for the control and further prevention of this disease.

Timely diagnosis, effective treatment, and future prevention are the most critical factors for successfully managing COVID-19 [12]. Among these, timely diagnosis plays an essential role in preventing and slowing the spread of the disease as the first line of defense. Early diagnosis of infected individuals enables immediate isolation of patients with highly contagious viruses (e.g., SARS-CoV-2), effectively controlling the spread of the disease. Moreover, early diagnosis allows physicians to provide immediate treatment, increasing the chances of cure and survival [13]. From the first outbreak of COVID-19 until now, the gold standard for confirming COVID-19 infection has been quantitative reverse transcription PCR (RT-qPCR) which amplifies small amounts of viral RNA in samples collected from an infected individual. However, this standard technique requires the analysis to be performed by trained experts and in a fully equipped laboratory, which inevitably increases the cost and time of testing. In a public health emergency where COVID-19 cases are increasing and COVID-19 testing capacity needs to be expanded, the introduction of more inexpensive, faster, easier-to-use diagnostic testing that analyzes samples at the point of use and screens for COVID-19 in a larger population is essential [14]. To compensate for the shortcomings of the current RT-qPCR assays, paper-based rapid diagnostic tests, especially lateral flow assays (LFA), have recently received widespread attention as an alternative to suppress the rapid spread and reinfection of COVID-19 [15]. Numerous LFA-based rapid diagnostic devices for diagnosing COVID-19 have been developed with great success, especially in resource-constrained environments and rapidly growing numbers of patients [16, 17]. So far, LFA-based rapid diagnostic technology has been used as an adjunct to RT-qPCR for COVID-19 confirmation due to its limited performance in terms of assay sensitivity and reproducibility [15]. However, extensive and intensive efforts are being made in academia and industry to improve the performance of LFA-based rapid inspection.

This review introduces the SARS-CoV-2 viral structure and diagnostic biomarkers derived from viral components. Then, a comprehensive introduction of paper-based diagnostic platforms, including device components, detection mechanisms for different target markers (e.g., nucleic acid, antigen, and antibody makers), and COVID-19 commercial kits, is presented. The intrinsic limitations related to the poor performance of the currently developed paper-based devices are also discussed. Furthermore, novel paper-based diagnostic platforms integrated with advanced technologies such as nanotechnology, aptamers, surface-enhanced Raman spectroscopy (SERS), and clustered regularly interspaced short palindromic repeats (CRISPR)-Cas are presented. These novel detection methods are promising for improving the diagnostic performance of previously reported paper-based tests. Finally, we discuss the prospects for the development of highly sensitive, accurate, cost-effective, and easy-to-use point-of-care COVID-19 diagnostic methods.

2 SARS-CoV-2 Viral Structure and Diagnostic Targets

The SARS-CoV-2 belongs to the genus β-coronavirus and comprises a single positive-strand RNA with a genome of ~ 30 kb. The SARS-CoV-2 genome codes ten genes that produce 26 proteins [18]. The genes are arranged in the sequence 5′ cap structure-replicase (open reading frame1/ab, ORF1/ab)-structural proteins with a [spike (S)-envelope (E)-membrane (M)-nucleocapsid (N)]-3′ poly (A) tail [19]. Among these genes, unique and specific nucleotide sequences representing SARS-CoV-2, such as genomic fingerprints, are targets for COVID-19 diagnosis. The genome sequence of SARS-CoV-2 was shared through the Global Initiative on Sharing All Influenza Data (GISAID) platform, on January 12, 2020 [20]. Subsequently, various primer and probe sets have been developed to amplify specific viral RNA sequences, and the WHO has posted these primer–probe sets [20,21,22,23], enabling the rapid development of COVID-19 nucleic acid amplification tests (NAATs). More than 100 RT-qPCR kits have been designed and prototyped and are the United States food and drug administration (FDA) emergency use authorization (EUA)-approved for COVID-19 diagnosis. These kits aim to amplify specific regions of viral genes, such as structural protein genes (N, E, S, and M) and confirmation genes (ORF1ab and RNA-dependent RNA polymerase (RdRp)) [20, 23, 24] (Fig. 1a). RT-qPCR offers high accuracy and very low analytical sensitivity; thus, it has been used as the gold standard for confirming COVID-19 infection.

Fig. 1
figure 1

SARS-CoV-2 genome, structure comprising major proteins, and variation levels of biomarkers across the duration of the infection. a The SARS-CoV-2 genome codes ten genes, and the genes are arranged in the sequence 5′ cap structure-ORF1/ab-spike (S)-envelope (E)-membrane (M)-nucleocapsid (N)-3′ poly (A) tail. Primer–probe sets for SARS-CoV-2 RNA amplification developed by research groups around the world [US Centers for Disease Control and Prevention (US-CDC) – target sequences: 28,287 ~ 28,358, 28,681 ~ 28,752, 29,164 ~ 29,230; China CDC – target sequences: 13,342 ~ 13,460, 28,881 ~ 28,979; Charité–Universitätsmedizin Berlin in Germany (Charité) – target sequences: 15,431 ~ 15,530, 26,269 ~ 26,381; National Institute of Infectious Disease in Japan (Japan NIID) – target sequence: 29,125 ~ 29,282; and University of Hong Kong (HKU) – target sequences: 18,778 ~ 18,909, 29,145 ~ 29,254]. b SARS-CoV-2 is mainly composed of four major proteins: spike (S) (red), membrane (M) (orange), envelope (E) (green), and nucleocapsid (N) (purple) proteins. c Temporal dynamics of the viral load and antigen and antibody levels. Since the types and amounts of biomarkers present in a patient's body fluid differ depending on the stage of infection, it is critical to select an appropriate biomarker and a method that can effectively detect it for an accurate diagnosis of COVID-19

Full size image

The SARS-CoV-2 genome encodes four major structural and functional proteins: the spike (S), membrane (M), envelope (E), and nucleocapsid (N) proteins (Fig. 1b) [11, 25, 26]. The S protein comprises two functional subunits, S1 and S2; S1 acts as a key protein for selective binding with the host cell receptor, angiotensin-converting enzyme 2 (ACE2), while S2 mediates membrane fusion [11, 27, 28]. The N protein is the most abundant protein in the virus and plays a crucial role in packaging and protecting viral genomic RNA. The N protein is highly immunogenic and rarely changes with disease progression [29,30,31,32]. The M protein is the most abundant structural protein that defines the shape of a virus, while the E protein is the smallest major structural protein involved in virus assembly and pathogenesis [28]. The S and N proteins are considered the most valuable antigenic biomarkers for diagnosing COVID-19 in the various detection methods for SARS-CoV-2 [33, 34].

In contrast, specific antibodies produced by the immune response to protect the body from SARS-CoV-2 infection could be another option for diagnosing COVID-19. Immunoglobulin M (IgM) is the first line of defense during viral infections appearing in the blood after SARS-CoV-2 infection, and increases rapidly, which is an indicator of early stage infection. IgG antibodies subsequently produced by the IgM antibodies are responsible for immunological memory and long-term immunity, which serves as the body's immune defense system to avoid reinfection with the same pathogen [35, 36]. IgA antibodies limit the entry of microorganisms and antigens into the susceptible mucosal barrier through respiratory mucosal epithelial protection and homeostasis regulation [37,38,39]. IgA antibody responses appear early and are characterized by intense and sustained maintenance [40]. Therefore, the detection of IgA antibodies is advantageous in the early infection stages.

Assay sensitivity is affected by the temporal profile of the viral load (or concentration of the biomarkers) across the duration of the infection. This is a critical factor in determining the diagnostic accuracy. Several studies have suggested the temporal dynamics of the viral load, antigen and antibody levels after SARS-CoV-2 infection [41,42,43,44]. In general, the viral load increased rapidly from the time of infection, peaked, and then decreased rapidly (within a few days) (Fig. 1c). After 10 days of infection, the viral load is reduced by a factor of 100 or more [41, 45]. Considering the virus structure, the quantities of viral RNA and antigens were consistent with the trend of the viral load. Therefore, the optimal time for detecting viral RNA and antigens is approximately 7 days, immediately after symptom onset. Conversely, viral RNA cannot be detected in the early or late phase of infection. Another study analyzed the serological response of COVID-19 patients to viral infections [42]. Approximately 10 days after symptom onset, the IgM response to SARS-CoV-2 is predominant over other immunoglobulins. After approximately 15 days, the IgM response decreased and disappeared. IgG antibodies are produced later than IgM production, 10–14 days after symptom onset; however, IgG antibodies persist and are detectable for a long time [43, 44]. IgM antibodies can mainly be used for the early detection of SARS-CoV-2 infection, whereas IgG antibodies may be more appropriately used to identify past infections.

The type and amount of biomarkers present in the patient's body fluid differed depending on the infection stage of the COVID-19 patient (Fig. 1c). To analyze these biomarkers, an appropriate sample preparation process according to the specimen and analysis method is also essential. In general, the upper respiratory tract sample is used in the case of RT-qPCR and antigen tests, and if possible, the lower respiratory tract sample is inspected simultaneously. Generally, the nasopharyngeal swab method is more sensitive than the nasal swab method for specimen collection in COVID-19 [46]. However, the nasopharyngeal swab method should be performed by a trained healthcare provider. The Centers for Disease Control and Prevention (CDC) noted that if both nasopharyngeal and oropharyngeal specimens can be collected, the sensitivity of the test can be maximized by combining the two specimens in one tube. On the other hand, the nasal swab method is more comfortable than the nasopharyngeal swab method and can be used easily by the general public; therefore, many diagnostic kits have been developed for nasal swab samples. Before the RT-qPCR test, RNA extraction from the sample is a crucial step to getting accurate amplification results [47, 48]. It should be carried out using proven equipment and reagents, and contamination should be prevented throughout the sampling and analysis process. In the case of the antibody tests, a small amount of blood should be collected through venipuncture or the fingerstick method. To reduce interference or increase sensitivity, any clots or erythrocyte sediments in the samples may be removed after the appropriate clotting or centrifuge method. It is essential for serum or plasma analysis kits, but these processes may be omitted for several whole blood or fingerstick kits [49, 50]. Therefore, for accurate diagnosis of COVID-19, it is crucial to select an appropriate biomarker and a method that can effectively detect these biomarkers according to the stage of infection.

3 Paper-Based Diagnostic Platform for COVID-19 Diagnosis

While the economically developed world has access to many advanced medical tools, trained personnel, and resources to perform diagnostic tests for maintaining health, these similar resources are unfortunately not as accessible in develo** countries. Rapid and accurate diagnosis is the first step toward improving healthcare conditions in develo** countries. In dire situations, such as the COVID-19 pandemic, the damage is inevitably more severe for countries marginalized from good medical services. In these countries, the supply of the COVID-19 vaccine is delayed and accurate diagnostic techniques for early diagnosis are extremely limited. Many disease-related deaths, including COVID-19, would have been preventable if the disease had been diagnosed earlier and followed by appropriate treatment. Moreover, large-scale diagnostic testing is critical to contain the COVID-19 epidemic, even in developed countries with well-established healthcare services [51]. Standard diagnostic techniques (i.e., RT-qPCR) are limited in the rapid screening of confirmed cases from a rapidly elevated number of suspected cases. For these outbreaks, it is crucial to have medical diagnostic platforms that can analyze samples on-site and provide immediate results [13].

Paper-based analytical devices show great potential in delivering POC diagnostic systems to the develo** world because of their remarkable properties such as biocompatibility, porosity, ease of modification, flexibility, chemical inertness, eco-friendliness, and ease of storage and transportation [52, 53]. Moreover, various sample types can be applied to paper-based analysis equipment, and sample transfer is possible without requiring additional power owing to capillary force. Over the last three decades, paper-based POC tests have been developed for various biomedical applications and launched as both ‘over-the-counter’ products, such as glucose monitoring and pregnancy testing, and ‘professional market’ products that can diagnose infectious disease, cardiac markers, diabetes, lipidoses, hemopathies, and several cancers.

Paper-based POC tests have also played a crucial role in the current COVID-19 pandemic. Many researchers in academia and industry have made intensive efforts to develop a simple, convenient, fast, sensitive, and accurate technology that can detect SARS-CoV-2. The technology that can best meet the strong demand for practical POC diagnostic tests in COVID-19 control is paper-based POC testing. Paper-based POC diagnostic platforms range from simple one-dimensional platforms, such as dipstick and LFA, to complex three-dimensional platforms, such as microfluidic paper-based assay devices (μPAD) and electrochemical paper-based assay devices (ePAD). Among these, LFA is a highly mature paper-based diagnostic technology that researchers and manufacturers have invested the most effort and cost in develo** COVID-19 diagnostic kits. A typical LFA comprises a sample pad, a conjugate pad, a nitrocellulose membrane, and an absorbent pad. The sample flow begins on the sample pad and meets the signal molecules that have dried on the conjugate pad. All biomarkers, antigens, antibodies, and SARS-CoV-2 RNA used to detect SARS-CoV-2 can be applied to LFA.

3.1 LFA-Based Diagnostic Platform for Detecting Viral Antigens

Unlike RT-qPCR, antigen-based diagnostics directly detect the presence of SARS-CoV-2 and related proteins in a sample taken from a nasopharyngeal swab or nasal passage without sample pretreatment and amplification. Therefore, it can provide a diagnosis of COVID-19 with faster and easier results at lower cost than RT-qPCR. Antigen-based diagnosis is based on immunoassay reactions that involve antigens and antibodies. The configuration of the general LFA diagnostic platforms for detecting SARS-CoV-2-specific antigens is shown in Fig. 2a. All proteins constituting SARS-CoV-2 can be targeted to diagnose COVID-19, but antigen tests for COVID-19 have been developed mainly targeting the S and N proteins [54,55,56]. To detect the SARS-CoV-2 antigens, a specific antibody pair that recognizes different regions of the target antigen is required. A capture antibody is immobilized on a nitrocellulose membrane to form a test line (first line), and another antibody is labeled with a signal molecule, mainly gold nanoparticles, and serves as the detection antibody. Additionally, the control line (second line) also serves to check whether the sample flowed through the nitrocellulose membrane, and additional Ig-types of antibodies that could not affect the test are used. When a nasopharyngeal swab sample of a patient with COVID-19 is loaded into the LFA device, the sample containing the target antigens flows along with the LFA strip by capillary force and first encounters the detection antibody. Target antigens are captured by both detection and capture antibodies to form a sandwich complex. After 15–20 min of sample loading, the appearance of color in the test and control lines is confirmed visually or by a portable analyzer.

Fig. 2
figure 2

Paper-based diagnostic platforms including device components, detection mechanisms for different target markers, a antigens, b antibodies, and c RNA. a To detect the SARS-CoV-2 antigens, a specific antibody pair is required. These capture and detection antibodies detect SARS-CoV-2-specific antigens (S and N proteins) while forming a sandwich complex. After 15–20 min of sample loading, the appearance of color in the test and control lines is confirmed visually or by a portable analyzer. b In serological tests (detecting IgM and IgG antibodies), the N (or S) proteins of SARS-CoV-2 are conjugated with gold nanoparticles and used as signal molecules to detect IgM and IgG antibodies. Anti-human IgM (or IgG) antibodies are immobilized on a nitrocellulose membrane to form test lines. When the sample contains the SARS-CoV-2-specific IgM or IgG antibodies, the antibodies are bound to the N (or S) protein-conjugated gold nanoparticles and finally bound to the test line, resulting in vivid color. c Isothermal amplification techniques combined with an LFA contribute to achieving POC tests for SARS-CoV-2 RNA detection. First, an isothermal amplification process is performed for target RNA amplification, and then an LFA reaction is performed so that the results can be easily checked

Full size image

Although various paper-based antigen diagnostic tests have been developed, the sensitivity of the rapid antigen test is unclear and is lower than that of RT-qPCR. The limit of detection (LOD) of antigen tests is approximately 105 copies/mL, while that of RT-qPCR is as low as 102 copies/mL [57, 58]. False-negative results may occur when the concentration of the target antigen in the clinical specimen is below the analytical sensitivity of antigen tests. Several studies have been conducted to overcome these limitations. Liu et al. presented a novel nanozyme-based chemiluminescence paper assay for detecting SARS-CoV-2 S antigen. In this case, nanozyme (Co-Fe@hemin-peroxidase) and chemiluminescent immunoassays were integrated with LFA to achieve sensitivity (360 TCID50/mL) comparable to that of an ELISA method [59]. To improve the performance of the previous signal molecule, gold nanoparticles [60], latex beads [61], cellulose nanobeads [62], and fluorescent microparticles [30] have been introduced into LFA. These signaling molecules have higher signal intensities, resulting in an approximately tenfold improvement in sensitivity compared with previous gold nanoparticle-based LFAs.

Our group also proposed a novel rapid detection method for the SARS-CoV-2 S antigen. Using the cellular receptor for SARS-CoV-2, angiotensin-converting enzyme 2 (ACE2), the SARS-CoV-2 S1 antigen was successfully detected in clinical specimens of COVID-19 patients (Fig. 3a) [63, 64]. Furthermore, we developed SARS-CoV-2 N antigen-specific single-chain variable fragment crystallizable fragment (scFv-Fc) fusion antibodies using phage display technology and applied them to the LFA platform (Fig. 3b) [62]. This scFv-Fc-based rapid diagnostic test showed high specificity that could distinguish even the N protein of SARS-CoV. Baker et al. developed a glyconanoparticle consisting of multivalent gold nanoparticles bearing sialic acid derivatives [65]. They discovered that the N-acetyl neuraminic acid is bound to the S protein and developed LFAs that exploit this interaction as a detection mechanism. This glycoprotein-based LFA showed high selectivity for the SARS-CoV S protein.

Fig. 3
figure 3

a Cellular receptor (ACE2)-based LFA for detecting SARS-CoV-2 S1 antigen, reproduced with permission from [63], copyright 2021 Elsevier. b Development of scFv-Fc-based LFA for detection of the SARS-CoV-2 N protein. Highly sensitive and specific scFv-Fc fusion proteins are rapidly screened by phage display technology, reproduced with permission from [62], copyright 2021 Elsevier. c Configuration of detecting system to quantify LFA results with the photon-counting approach and representative results for IgG antibody detection (concentrations range: from 1000 to 0.1 ng/mL), reproduced with permission from [74], copyright 2020 AIP. d LFA strip to detect anti-SARS-CoV-2 IgA antibody and the simple and universal smartphone reader to detect the optical signal from LFA, reproduced with permission from [75], copyright 2021 Elsevier. e Lateral Flow Strip Membranes (LFSM)-based on highly specific and sensitive detection of SARS-CoV-2. The LFSM assay allows simultaneous detection of the multiple regions of SARS-CoV-2 RNA in a sing test, reproduced with permission from [77, 76], copyright 2020 ACS. f Principle of reverse transcription-enzymatic recombinase amplification (RT-ERA). The RT-ERA has the capability of ultrasensitive, field-deployable, and simultaneous dual-gene detection of SARS-CoV-2 RNA, reproduced with permission from [92]. Typical LFA tests use gold nanoparticles as signal reporters; however, their sensitivities are not very high. Extensive efforts have been made to increase diagnostic sensitivity by improving the performance of signaling molecules in various studies. Liu et al. reported an advanced LFA based on gold nanoparticles for enhanced specific binding and thermal contrast amplification (TCA) for signal amplification. With TCA, the gold nanoparticles captured in the test line were excited by laser irradiation and exhibited a substantial photothermal effect, enabling the detection of subvisual positives. They successfully detected SARS-CoV-2 receptor-binding domain (RBD) antigen as low as 28.6 aM in a human nasopharyngeal swab [93]. Selenium nanoparticles exhibit favorable biocompatibility and are readily conjugated with biological molecules without losing their activity [94]. A POC selenium nanoparticle-based LFA was developed to detect SARS-CoV-2 IgM and IgG [72]. In this study, the authors made a new selenium nanoparticle-based LFA kit and visually detected anti-SARS-CoV-2 IgG and IgM antibodies in human serum within 10 min. Furthermore, the authors performed a clinical evaluation using a sample of 90 patients with COVID-19 and 263 uninfected negative controls, demonstrating a sensitivity and specificity of 93.33% and 97.34%, respectively. This selenium nanoparticle-based LFA showed superior detection limits compared to the gold nanoparticle-based LFA in IgM antibody detection and did not show cross-reactivity with influenza A, influenza B, anti-nuclear antibodies, and rheumatoid factor. On the other hand, in recent years, quantum dots (QDs) have been widely used as fluorescent signal reporters in LFA because of their excellent optical properties, such as quantifiable fluorescence intensity, broad excitation, and high light stability [95,96,13]. However, there is still an aperture that needs to be addressed to develop an ideal and universal POC diagnostic platform. Smartphones, which offer imaging, filtering, and image/data processing, can be powerful tools to compensate for current POC diagnostic platforms. Because many people always carry their smartphones, they can effectively serve as handheld readers that rapidly and accurately check for infections and contribute significantly to disease control and surveillance. In addition, it is expected that diagnosis performance will be improved through the introduction of artificial intelligence (AI). Recently, the potential application of AI-based COVID-19 diagnosis has been extensively explored in the field of lung detection imaging, such as computed tomography (CT) imaging [132, 133], chest radiographs (X-ray) imaging [134, 135]. By learning from tremendous amounts of diagnostic results through deep learning techniques and using them to present accurate current or future results, more effective infectious disease management is possible.