Advancing metallic nanozymes for multiplexed multimodal biosensing in early disease diagnostics

Abstract

There is an ever-expanding demand for inexpensive, rapid, and reliable diagnostic sensors that simultaneously and accurately detect various biomarkers of clinical significance in human biofluids. The emergence of powerful metallic nanozymes marks the epitome of next-generation biomarker detection. Over the past seven years, researchers have fully leveraged the distinctive electrocatalytic features and versatility of metallic nanozymes within multimodal and multiplexed detection systems. Multiplexed detection using a biosensor is essential for diagnosing diseases using numerous biomarkers from a small amount of biofluids. Multimodal readouts that combine various methods offer enhanced accuracy, sensitivity, cross-validation, and real-time analysis of the targeted biomarkers. All of these components in a biosensor enable compact miniaturisation, a microfluidic platform, and the integration of sensors with wearable technologies, which will further substantiate point-of-care diagnostics. This review explores numerous designs of metallic nanozymes, probes, and signal amplification strategies applied in recent years for the ultra-selective and sensitive detection of multiple target biomarkers. Overall, these innovations are collectively paving a route in the field towards a non-invasive, robust, and efficient diagnostic platform tailored for personalised medicine and early disease detection.

---

1. Introduction

1.1. The evolving landscape of biomarker-based diagnostics

The rise in the prevalence of chronic diseases such as cardiovascular conditions, neurodegenerative diseases, cancer, and diabetes emphasises the significance of early detection of the associated biomarkers. Clinical biomarkers are crucial biomolecules present in biofluids, which indicate an individual's physiological condition.¹ The clinical biomarkers that are considered for disease diagnosis include metabolites, lipids, saccharides, proteins, and nucleic acids.² There are several current diagnostic methods that are utilised to detect biomarkers, such as enzyme-linked immunosorbent assay (ELISA), western blotting, tissue biopsy, genomic profiling, polymerase chain reaction (PCR), and immunohistochemistry.^3,4 However, these current diagnostic techniques have several limitations including being labour-intensive and slow, requiring skilled professionals, invasive procedures, unsuitable for point-of-care (POC) testing, and issues with specificity and selectivity.⁵ A biosensor platform is an alternative option to circumvent the major disadvantages of routine diagnostic methods for biomarker detection.

Biosensors are analytical tools that combine a bioreceptor (biological recognition element) and a transducer (physicochemical converter) to detect an analyte or a group of analytes.⁶ The analyte is first transported from the solution onto the surface of the biosensor. This transportation of the analyte may occur through mixing and diffusion processes.⁷ The specific bioreceptor has affinity for, and interacts with, the analyte to be detected in a highly selective but reversible manner.⁸ Bioreceptors are generally divided into two subgroups: bioaffinity and biocatalytic agents. Some examples of bioaffinity agents include lectins, nucleic acids, hormones, organelle receptors, aptamers, and antibodies. When there is an interaction between an analyte and a bioaffinity agent, a complex is formed, and this causes changes in physicochemical factors. Meanwhile, biocatalytic agents lead to the molecular alteration of the analyte, which results in the elevation or reduction of the amount of substance in the solution. Animal tissue or plant sections, microorganisms or enzymes are used as biocatalytic agents.⁹

Substrates are often added into the reaction mixture within the biosensor platform, which also react with the bioreceptor.¹⁰ This results in either the formation of selective ions such as OH⁻, NH₄⁺, H⁺, and other monovalent anions and cations, the release or consumption of gas molecules (NH₃, CO₂, O₂, etc.) or electrons being used or released. The sensor measures shifts in the chemical, biological, and physical reactions including mass, light absorption, heat, refractive index, pH, temperature, layer thickness, current, and voltage changes perceived from the bioreceptor.¹¹ The transducer then registers the changes and converts the information received into a measurable signal based on specific transducers including electrical, piezoelectric, optical, and thermal signals. The display device and the signal processors are mainly responsible for presenting the analysed qualitative or quantitative results in a user-friendly manner.¹²

Biosensors have found remarkable applications in various fields including the diagnosis and quantification of biomarkers in medicine, analysis of industrial liquids and gases, monitoring fermentation processes, biosecurity and homeland security, detection of various analytes in the food industry, chemistry and biology, veterinary medicine and agriculture, tracking environmental explosives, pollution, and other military fields.¹³ This demonstrates the versatility of biosensors and biosensing is still gaining recognition over traditional analytical techniques by virtue of its rapid-response, real-time monitoring, easy operation, high accuracy, low instrumentation cost, and good sensitivity. The application of biosensors in detecting analytes in the respective fields fulfils the required properties of accurate output and reasonable sample concentrations, reduced preparation time, high robustness, and cleanliness of the system.¹⁴ Further progress of biosensor detection in the diagnostic approach for critical diseases such as autoimmune disorders and cancer is found to be achievable. The previous literature shows that early disease monitoring through patients’ biological fluids (blood, sweat, tears, urine, saliva, etc.) is possible through biomolecular affinity, by correlating disease-related changes in RNAs, proteins, and small molecules.¹⁵

2. The role of nanozymes in biosensing

The opportunity to develop nanozymes (NZs) is seized as the activity of natural enzymes is restricted by factors such as limited shelf-life and recycling, instability at extreme pH values or temperature, and high cost. These artificial-enzyme mimics can counter the drawbacks of natural enzymes while offering additional advantages, such as being easily produced in the laboratory and providing multifunctionality.¹⁶ Unlike natural enzymes, NZs are less susceptible to degradation and denaturation as they do not require specific amino acid sequences and sensitive folding patterns to function.¹⁷ The biosensing landscape of biomarkers has been revolutionised by the discovery of the first nanozyme (Fe₃O₄ nanoparticles) over the past few decades, and the surge in NZ research continues to be innovative to this day.¹⁸ In contrast to conventional techniques, the NZ-based biosensing is easy to perform, inexpensive, durable, and rapid and has good biocompatibility.¹⁹ This aligns with the quintessential aim of improving biomarker detection by developing a cost-effective, reliable, and dynamic detection platform for diagnosis, prognosis, and monitoring of the recurrence and progression of a particular disease.

2.1. Structural engineering of nanozymes for biosensing

Nanozymes are the epicentre in improving the performance of biosensors and they are based on nanoparticles (NPs) and nanomaterials, which possess the characteristics of enzymes. Individually, NPs have excellent compatibility, a large surface area, adjustable morphology, and the ability to accelerate signal transduction due to their conductive attributes.²⁰ Efficient NZs are designed in the laboratory to enable the enzymatic processes to be regulated and improved.²¹ The mimicking activity of NZs can be tuned through adjustment of their surface composition, morphology, size, crystal defects, surface modification, and active sites’ density, which produce a unique electronic and geometric structure. External stimuli such as heat or light also further tune these properties, offering effective control of the NZ's activity.²² Additionally, numerous variants in the synthesis pathways and sources of nanomaterials can lead to variations in morphologies. Various NZs display magnetic properties, enabling easy separation and simultaneous detection of multiple biomarkers, highlighting their multifunctionality.²³ The integration of numerous NPs might boost the multifunctionality or enhance the basic properties of NZs.

2.1.1. Unlocking sensing performance through nanozyme design. The majority of nanozyme catalytic reactions occur on the surface of NPs; therefore, surface modification can alter the surface chemistry and atomic composition of NPs. The enzymatic activity also depends on the core of the NPs.²⁴ The surface chemistry fundamentally governs the NZ catalytic activity by determining substrate affinity, stability, and the reaction environment.²⁵ Functional groups such as amine, carboxyl, hydroxyl, and thiol on NZ surfaces provide binding sites and facilitate ligand attachment, enabling further modification with molecules including ligands, polymers, aptamers, small particles, and antibodies to enhance catalytic specificity and substrate recognition. In addition, structural engineering on the surface of NZs can generate a concentrated local substrate effect in the proximity of the catalytic sites and establish chiral binding interactions that mimic the active sites of natural enzymes.²⁶ As a result, substrates gain access to the active sites of NZs and facilitate their transformation. This improves catalytic performance and selectivity in complex sensing matrices.²⁷ Structural tuning via NZ surfaces can modulate catalytic activity through enhancements in stability and turnover rates, thereby providing robustness and optimising sensing outputs. Their catalytic mechanisms directly dictate the signal generation in biosensors, contributing to a low limit of detection (LOD) and high sensitivity.²⁸

Overall, the collective interplay between NZ structure, surface chemistry, and catalytic activity improves the analytical performance of NZ-based sensors (Scheme 1). Both structural design and surface engineering of NZs influence the underlying catalytic pathway, ultimately determining signal amplification, stability, specificity, and robustness in sensing applications. Effective design of NZs requires the balancing of mechanistic pathways, surface microenvironments, and structural precision to optimise their sensing behaviour.

Scheme 1 Key factors that drive effective nanozyme design and sensing performance in biosensors.

2.2. Overview of metallic nanozyme catalysis

The NZs are classified according to the NPs they are based on, namely metallic nanozymes, non-metallic-based nanozymes (such as graphene, carbon dots and quantum dots) and hybrid nanozymes (combination of metallic and non-metallic nanomaterials). The NZs can also be classified based on their enzyme-mimicking activity, which includes catalase (CAT), superoxide dismutase (SOD), peroxidase (POD) and oxidase (OXD).²⁹ In NZ-based biosensing, two main processes occur, namely distinct analyte recognition followed by a catalytic signal from the NZ. Combining novel metallic NPs to form NZs significantly provides notable amplification of biosensor output by rapidly generating detectable products, improving the signal-to-noise ratio, and lowering the detection limit. The principle of NZ detection is divided into two categories: (1) the qualitative indication of the biomarker through the indirect reaction of the NZ and the agent present and (2) the biomarker either deactivates or activates a reaction between the agent and NZ.^30,31 This review aims to provide an overview of recent developments (the past seven years) in metallic NZ-based biosensors, emphasising their applications in the early identification of biomarkers in human biofluids. The advantages of multimodal and multiplexed metallic NZ-based biosensors in enhancing diagnostic accuracy and facilitating real-time monitoring of health are also explored in this review (Scheme 2). By compiling insights from prior studies, we aim to present a comprehensive overview of recent advancements, challenges and prospects in this rapidly evolving field.

Scheme 2 Simultaneous metallic nanozyme-based detection of multiple biomarkers with the integration of various multimodal modes, and the advantages of using such biosensor systems.

2.3. Tuning redox activity and surface characteristics of metallic nanozymes

Among various NZ classes, metallic NZs are widely studied in the biomarker sensing field (Table 1), which showcases their versatility and biocompatibility in biofluids. Metallic NZs can be further subdivided into metal elements, metal oxides and metal–organic frameworks (MOFs). The valence properties and atomic structure signify the enzymatic properties of metallic NZs, which facilitate the electron-transfer process and promote the generation of reactive oxygen species (ROS). Large optical enhancement is possible for noble metals including Pt, Pd, Bi, Ag, and Au due to their distinctive plasmonic features at the nanoscale level.³² Monometallic NZs consist of only a single type of metal NP, which may possess several disadvantages in their biosensing applications, such as limited tunability, moderate catalytic efficiency, oxidation and stability issues, susceptibility to aggregation, poor selectivity, narrow temperature and pH range, and lack of synergistic effects compared to bimetallic, trimetallic, and hybrid NZs.³³ Synergistic effects refer to the boosted catalytic activity and functional performance of NZs that exceed the individual features of the combined nanoparticles.

Table 1 Summary of current metallic NZ detection of numerous biomarkers using NZ-based sensors

Group	Metallic NZs	Enzyme-like mechanism	Detection method	Biomarker detected	Range of detection	Limit of detection	Ref.
Metal-based	Au NPs	POD	Colourimetric	H2O2 and glucose	H2O2: 2 × 10^−6 to 2 × 10^−4	H2O2: 5 × 10^−7	36
					Glucose: 1.8 × 10^−5 to 1.1 × 10^−3 M
	Ag nanoclusters	OXD	Colourimetric	Hg^2+ and DNA	Hg^2+: 80 nmol L^−1 to 50 mmol L^−1	Hg^2+: 25 nmol L^−1	37
					DNA: 30 to 225 nmol L^−1	DNA: 10 nmol L^−1
	Pt–Pd	SOD	Electrochemical	HeLa cells	16 to 1536 µM	0.13 µM	38
Metal-oxide based	CeO2	OXD	Colourimetric	Dopamine and catechol	0.075 to 2 mM	Dopamine: 1.5 µM; catechol: 0.2 µM	39
	Mn3(PO4)2/MXene	SOD	Electrochemical	Superoxide from lung cancer cells (A549)	5.75 nM to 25.93 µM	1.63 nM	40
	NiCo2O4	POD and OXD	Electrochemical	Glucose	2 to 100 µM	1.6 µM	41
MOFs	Cu-NMOF@PtNPs/HRP	CAT and POD	Electrochemical	miR-155	0.50 fM to 1.0 × 10^5 PM	0.13 fM	42
	Ni/Cu MOF	OXD	Electrochemical	Glucose	1 µM to 20 mM	0.51 µM	43
	Eu/Fe-MOF	POD	Fluorescence	Cholesterol	2 mM to 8 mM	10.5 µM	44

---

2.3.1. Metal elements. The effective usage of positively charged Au NPs as natural peroxidase mimics is especially observed in the detection of glucose and hydrogen peroxide (H₂O₂) in the presence of 3,3′,5,5′-tetramethylbenzidine (TMB).³⁴ An example is a study conducted by Gao et al. on the development of an Au–Ag NZ conjugate for the colourimetric detection of glucose.³⁵ In this study, the Au NPs imitate the activity of glucose oxidase (GOx), which catalyses the oxidation of glucose and generates the byproducts gluconic acid and H₂O₂. The H₂O₂ then etches the Ag NPs, causing the initial yellow colour of Ag NPs to fade and the red colour of Au NPs to appear. The visual shift in colour change forms the basis of glucose quantification. Through a spectrophotometer, this method showed a low detection limit of 5 µM in the concentration range of 5 to 70 µM. The team discovered that the system exhibits high selectivity for glucose when L-cysteine is assembled on the surface of the Au–Ag NPs. In the absence of L-cysteine, substances such as fructose, bovine serum albumin (BSA), paracetamol (PA), uric acid (UA), and ascorbic acid (AA) caused significant interference. The L-cysteine modification on Au NPs formed a stable self-assembled monolayer structure on the surface of NZs through the thiol group. This prevents the adsorption of proteins particularly BSA on NPs, allowing only selective interaction between NZ and H₂O₂.⁴⁵ The improved selectivity for glucose detection after L-cysteine surface modification might be attributed to the zwitterionic nature of cysteine. However, the exact mechanism has not been fully elucidated at present. Overall, the recovery of glucose in human serum samples is satisfactory, as there is no marked difference between the hospital method and the developed biosensor.

In another study conducted by Arshad et al., an Au–Ag–Pt trimetallic NZ was conjugated to the anti-osteoprotegerin antibody to directly detect the biomarker osteoprotegerin through electrochemical and colourimetric modes.⁴⁶ In both modes, the signal is based on the reduction of H₂O₂ to give highly active hydroxyl radicals. In the colourimetric assay, the hydroxyl radical further reacts with TMB to display the visible blue-coloured product. The detection of osteoprotegerin electrochemically is achieved in the range of 0.1 fg mL⁻¹ to 10 ng mL⁻¹ with a LOD of 1.81 pg mL⁻¹. Whilst, the detection of osteoprotegerin through the colourimetric method is in the range of 1 fg mL⁻¹ to 100 ng mL⁻¹, and the LOD achieved is 1.87 pg mL⁻¹. High specificity for osteoprotegerin is demonstrated through the developed immunosensor with insignificant interference from other proteins. They also validated the performance of the immunosensor in human serum samples, which yields robust recovery rates and possesses possible clinical applications.

Most bare noble monometallic NZs including Pd, Pt, and other metals display some level of toxicity in biological environments based on their size, catalytic activity, and reactive surface. These NPs with various sizes may affect the normal biodistribution and cell functions within the body and potentially cause organ damage.⁴⁷ Rational NZ design through surface coatings, modifications, and hybridisation with other NPs can limit metal leakage and lower toxicity while enhancing catalytic efficiency and biocompatibility.⁴⁸ Although noble metals pose inherent toxicity concerns, advancements in surface modifications and engineering are increasingly mitigating these risks and enabling their safer use in biosensing.

2.3.2. Metal oxide NZs. The noteworthiness of metal oxide NZs is supported by their ease of modification, large-scale production, exceptional stability, and customisable functionality. Metal oxide NZs particularly exhibit outstanding optical properties due to their strong ultraviolet light absorption capacity and wide bandgap.⁴⁹ For instance, Alizadeh and co-workers fabricated a paper-based microfluidic colourimetric device using a Co₂(OH)₂CO₃–CeO₂ NZ for the immunodetection of carcinoembryonic antigen (CEA).⁵⁰ The authors functionalised primary antibodies (Ab1) on the surface of the paper through chitosan and ionic liquid mixture modification. As a signal tag, the Co₂(OH)₂CO₃–CeO₂ NZ was functionalised with secondary antibodies (Ab2), which catalysed the degradation of H₂O₂ and subsequent oxidation of TMB in the presence of CEA in a sandwich manner. This produces an observable colour change, which is subsequently analysed by an application installed on the smartphone. The detection limit of the proposed immunosensor is 0.51 pg mL⁻¹ within the CEA concentration range of 0.002 to 75 ng mL⁻¹. This immunosensor showed a promising and sensitive response to CEA in the real human serum sample, which was further cross-validated by conventional ELISA.

In addition, an immunosensor was developed by Fu et al. based on the photothermal immunosensor NZ detection of prostate-specific antigen (PSA).⁵¹ The sensing was performed through the photothermal reaction of the Fe₃O₄ NP mediated TMB-H₂O₂ colourimetric assay, and a common thermometer was utilised as the signal quantitative reader. A detection probe was developed using an Fe₃O₄ NZ-labelled antibody. They found that TMB is oxidised through the one-electron charge transfer complex mediated by Fe₃O₄ NPs, thus exhibiting colour change, and they also demonstrated a strong near-infrared (NIR) laser-driven photothermal effect via the sandwich-type immunosensor. The oxidised TMB acted as a highly sensitive photothermal probe to relay the immunoassay signal into heat via the NIR laser-driven photothermal effect. The detection range of PSA is from 0 to 0.006 mg mL⁻¹ and the lowest concentration that the platform can detect is 1 ng mL⁻¹ in human serum samples.

2.3.3. MOF NZs. MOFs are nanoscale crystalline porous hybrids consisting of organic ligands and metal-based nodes. Their suitable size, biodegradability, and good biocompatibility have attracted attention in biomedical fields, biosensing, and biocatalysis.⁵² Despite being newly introduced in biosensor applications, they are found to have fascinating enzyme-like capabilities and are considered high-potential nanomaterials. Similar to metal and metal-oxide NZs, bimetallic MOFs with organic ligands exhibit greater catalytic activity due to the synergistic effects and potential of the constituent metals, resulting in enhanced electron transfer and efficient catalytic activities.⁵³ For instance, a study conducted by Hu et al. designed a peroxidase-mimicking NZ that incorporates Au NPs into a thermally stable and porous MOF known as MIL-101.⁵⁴ The Au NPs@MIL-101 NZ was integrated to lactate oxidase (LOx) and GOx to form two integrative NZs, namely Au NPs@MIL-101@LOx and Au NPs@MIL-101@GOx. The oxidases catalyse the oxidation of target lactate or glucose to form H₂O₂. In the presence of the H₂O₂ formed, the NZ showed its ability to convert the Raman-inactive reporter leucomalachite green (LMG) into its active form malachite green (MG). MG then served as a substrate for surface-enhanced Raman scattering (SERS) and consequently amplified the signals produced. The team applied this system to living rat brain tissues to monitor the lactate and the fluctuation of lactate and glucose in the context of ischemic stroke and evaluate the therapeutic effects of astaxanthin in reducing cerebral ischemic injuries. The designed integrative NZ demonstrated the potential application for the detection of other biomarkers and the monitoring of the therapeutic plan in various diseases.

2.4. Comparative performance of metallic and non-metallic NZs

The key performance differences between metallic and non-metallic NZs are summarised in Table 2. When detecting the same biomarkers such as glucose and H₂O₂, the metallic NZ displayed its superiority compared to the non-metallic NZ. Zhang et al. reported PtS₂ nanosheets NZ that exhibited good stability, reproducibility, and intrinsic POD-mimicking activity. Through a spectrophotometer, glucose concentration in buffer and human serum was evaluated.⁵⁵ For H₂O₂ detection, the LOD was found to be 0.33 µM within the linear range of 1 to 100 µM. When PtS₂ nanosheets NZ were coupled with GOx for glucose detection, the authors recorded a LOD of 0.20 µM, within a linear dynamic range of 0.5 to 150 µM, whereas, a study combining graphene quantum dots and CuO nanostructure to form the GQDs/CuO NZ with the TMB system achieved H₂O₂ detection with a 0.20 µM LOD in a narrower linear range of 0.5 to 10 µM.⁵⁶ When coupled with GOx for serum glucose detection, a higher LOD of 0.60 µM was recorded with a 2 to 100 µM linear concentration range. Hence, a relatively low LOD and wider linear concentration range are shown by the metallic NZ compared to the non-metallic NZ in detecting both glucose and H₂O₂.

Table 2 Comparison of performance trends between metallic and non-metallic NZs

Features	Metallic NZs	Non-metallic NZs
Typical LOD	Very low LOD (may reach nM or sub-nM to low µM)	Moderate to high LOD (usually > 0.1 mM to high µM)
Linearity range	Broad linear range (around 3 to 5 orders of magnitude) when coupled with signal amplification and tuned surface chemistry	Narrow linear range (2 to 3 orders of magnitude), due to limitations in substrate affinity and NZ surface heterogeneity
Biofluid compatibility	High compatibility with blood, saliva, and serum particularly when there are surface-modifications and alloying, which reduce aggregation, prevent protein adsorption, and improve biocompatibility	Adequate biocompatibility within complex biofluids. However, they may aggregate, suffer non-specific adsorption, and matrix quenching unless functionalised
Design–function correlation	Strong correlation such that the size, shape, composition, and modification of the NZ precisely control its selectivity, catalytic activity, multifunctionality, and synergistic effects	Moderate correlation; structural modifications of NZs affect catalytic activity to some degree. However, the catalytic mechanism, selectivity and tunability are lower than those of metallic NZs, requiring additional amplification for comparable sensitivity
Response speed and catalytic function	Rapid readout (seconds to minutes) based on the high turnover via fast catalytic kinetics	Slow readout due to lower catalytic kinetics per active site, which leads to longer incubation requirements and higher catalyst loading

---

Dopamine (DA) is a crucial neurotransmitter that plays a significant role in various cardiovascular, renal, and neurological systems. Early detection of abnormal levels of DA is important in cardiovascular diseases and mental diseases, including Alzheimer's and Parkinson's. Two different studies have employed metallic and non-metallic NZs in the detection of DA. Nishan and colleagues devised a cobalt-doped hydroxyapatite (Co-HAp) NZ with POD-like activity for the colourimetric detection of DA. The designed system was rapid (2-minute response) and able to detect DA in a wide linear range of 0.9 to 35 µM with a low LOD of 0.51 µM. The applicability of the platform within physiological solutions was also evaluated.⁵⁷ In contrast, another study that used non-metallic co-doped graphene quantum dots (BS-GQDs) as a fluorescent probe demonstrated a higher LOD of 3.6 µM and a detection range of DA from 0 to 340 µM.⁵⁸ The authors did not perform DA detection in real sample analysis, likely because matrix quenching occurs due to the unfunctionalised NZ. Therefore, metallic NZ detection of dopamine is significant compared to its non-metallic counterparts due to its rapid response, low LOD, high selectivity, and good biofluid compatibility.

A PSA concentration of higher than 4 ng mL⁻¹ in human serum is a crucial indicator of prostate cancer. Metallic NZs and non-metallic NZs have been applied in the detection of PSA. A study conducted by Cao et al. reported the synergistic POD-like activity of PtNP@Co₃O₄ hollow nanopolyhedrals that were functionalised with an aptamer and magnetic beads. The sandwich structure with the PSA target was isolated from the mixture through a magnetic field, and this homogenous solution was added to two channels of detection. The sensor achieved linear ranges of 0.01 to 10 ng mL⁻¹ (electrochemical) and 0.01 to 15 ng mL⁻¹ (colourimetric), covering the clinical PSA threshold of 4 ng mL⁻¹, with ultra-low LODs of 0.0079 ng mL⁻¹ and 0.01 ng mL⁻¹, respectively.⁵⁹ On the other hand, a study conducted by Zhu et al. reported a carboxyl graphene quantum dot (CGQD) NZ combined with a PSA aptamer to form a probe, while few-layer vanadium carbide (FL-V₂CTx) nanosheets acted as a quencher for fluorescence detection of PSA. The fluorescence-based aptasensor detected PSA as low as 0.03 ng mL⁻¹ in a linear detection range of 0.1 to 20 ng mL⁻¹.⁶⁰ The metallic NZ showed versatility with dual-channel detection, achieving much lower LODs than the single-mode non-metallic NZ.

3. Rationale for multimodal and multiplexed biosensor platforms

3.1. NZ compatibility in multimodal readouts

In a NZ-based immunosensor, the highly specific recognition receptors that form a complex or interact biochemically with the biomarker initiate the analysis of signal transduction through the catalytic activities of NZs.^61,62 The main cascade of mechanisms occurring in biosensing platforms includes H₂O₂ or a target analyte that can generate hydrogen peroxide by substrate oxidation.⁶³ Besides that, the target analyte can undergo reduction and be detected by reducing the oxidised products, which consequently lowers the catalytic signals (example: glutathione),⁶⁴ and the detection of targets through their interaction with an enzyme, which simultaneously regulates the activity of the enzyme.⁶⁵ Some examples of methods used in the detection of various biomarkers with numerous substrates are SERS, mass-sensing BioCD protein array, gel electrophoresis, and ELISA.⁶⁶ NZs effectively substitute the enzyme used in ELISA. As a result, the widespread application of nanoplatforms based on NZs is continually being observed in the detection of biomolecules.

Many of the methods for biomarker detection are based on the traditional principle of target capture being recognised by reporter antibodies for signal read-out.⁶⁷ Although a single signal read-out is common and offers high sensitivity, its accuracy is often reduced due to baseline drift, intrinsic and extrinsic noise, and saturation effects, which may distort the singular results obtained.⁶⁸ Therefore, the ability of NZs to participate in reaction catalysis to provide electrochemical, photothermal, colourimetric, and fluorescence responses leads to the merging of various detection methods, including dual-mode (colourimetric/SERS, colourimetric/photothermal, colourimetric/electrochemical and colourimetric/fluorescence) or even recently multi-mode biosensing for the avoidance of single-mode detection. Multi-mode detection combines numerous signal probes leading to multiple sensing signals either under different or the same analytical conditions.⁶⁹ The advantages of a multimodal sensor include cross-validation, large information flux, small sample usage, and enhanced sensitivity, specificity, and accuracy.⁷⁰ The multimodal sensor approach aligns with the ever-growing demand for cheap, fast, reliable, and POC sensors to quantify and detect various biomarkers of clinical significance.

3.2. Concept of multiplexing in clinical diagnostics

The simultaneous detection of numerous analytes in a single sample, known as multiplexing, is advantageous for diagnosing various diseases. Multiplexing provides high accuracy and simultaneous assessment of deviation in the expression of discriminative biomarkers, which in turn allows early disease diagnosis.⁷¹ The focus on single-biomarker recognition is insufficient to determine in-depth information for disease tracking and clinical diagnosis. Therefore, the development of future-generation sensors has centred on multi-detection and multimodal approaches. In a standard biosensor, only one specific analyte in a single sample can be detected and this is not applicable for the accurate and early detection of complex diseases with a multitude of biomarkers, such as osteoarthritis (OA), rheumatoid arthritis (RA), diabetes, and cancer. Detection of various biomarkers without ambiguity can be achieved through the employment of multiplexing detection, even when there are limited sample volumes.⁷² This approach is even more attractive as many data points of quantitative and qualitative information can be collected despite the small quantities of clinical samples. Additionally, other advantages of multiplexing technologies are increased throughput, fewer errors due to fewer samples collected, reduced cost per data point, and fewer samples being collected from patients.⁷³ Hence, the favourable advantages of both multimodal and multiplexing have been implemented in the NZ-based biosensor detection of biomarkers in biofluids (Table 3).

Table 3 Diverse studies on NZ-based sensors for multiplexed and multimodal detection of biomarkers

Metallic NZs	Enzyme-like mechanism	Detection method(s)	Biomarker(s) detected	Range of detection	Limit of detection	Ref.
CeO2-CNC	POD	Colourimetric and electrochemical	HSA	Colourimetric and electrochemical: 0.1 to 1000 pg mL^−1	Colourimetric: 5.13 pg mL^−1	77
					Electrochemical: 5.85 pg mL^−1
ZIF-67	POD	Chemiluminescence (CL)	PSA	PSA: 5 pg mL^−1 to 0.2 µg mL^−1	PSA: 3.9 pg mL^−1	78
			CEA	CEA: 5 pg mL^−1 to 1 µg mL^−1	CEA: 2.1 pg mL^−1
			AFP	AFP: 1 pg mL^−1 to 100 µg mL^−1	AFP: 0.4 pg mL^−1
GNPs@MIL-53)	POD	SERS and fluorometric	Caffeine and glucose	—	Caffeine: 1.2 × 10^−11 M	79
					Glucose: 3 × 10^−8 M
Ce/Fe-Fum Bio-MOF	Laccase, POD, OXD, and catechol oxidase	Colourimetric with smartphone signalling	Acetone, glucose, uric acid, and L-tryptophan	20 to 1000 µM	—	80
Fe CDs/Mo SACs	POD	Fluorescence and colourimetric	H2O2 and UA	H2O2 : 0 to 300 µM	—	81
				UA: 0 to 200 µM
MoS2 NSs	POD	ECL, colourimetric, and photothermal	HE4	ECL: 10^−6 ng mL^−1 to 10 ng mL^−1	ECL: 3 × 10^−7 ng mL^−1	82
				Colourimetric: 10^−4 ng mL^−1 to 10 ng mL^−1	Colourimetric and photothermal: 3 × 10^−5 ng mL^−1
CuFe2O4/N, O-codoped porous carbon (MCNPC)	POD and laccase	Colourimetric, photothermal, and smartphone signalling	D-PA, CPL, neurotransmitters (dopamine and epinephrine)	D-PA: 5 to 130 µg mL^−1	D-PA: colourimetric-0.78 µg mL^−1	83
				CPL: 5 to 180 µg mL^−1	Smartphone: 2.10 µg mL^−1
				Dopamine: 5 to 60 µg mL^−1	Photothermal: 9.40 µg mL^−1
				Epinephrine: 2 to 50 µg mL^−1	CPL:
					Colourimetric: 0.70 µg mL^−1
					Smartphone: 1.90 µg mL^−1
					Photothermal: 13.87 µg mL^−1
					Dopamine:
					Colourimetric-:1.41 µM
					Epinephrine:
					Colourimetric: 0.71 µM

---

4. Cascade reaction in multiplexed strategies

Several approaches exist for multiplexed detection of biomarkers utilising labels such as magnetic microbeads, metallic NPs, and enzymes. These include the use of spatially divided regions within an electrode array or a channel network; the usage of spatially separated detection sites for each biomarker; and lastly “barcoding” where distinct labels are spatially encoded on a single transducer surface.^74,75 There are a few recent studies that display the efficiency and ability of NZ-mediated biosensors to differentiate multiple analytes in a multiplex platform.

4.1. Visual-detection of multiple analytes in serum and urine

Exosomes (EXOs) are secreted by all types of mammalian cells and consist of phospholipid-membrane-enclosed nanoscale vesicles. They are regarded as a promising biomarker in the liquid biopsy of cancers, particularly in determining the specific level of exosomal proteins on the surface of exosomes. Di and colleagues on this basis have modified EXOs with thiol-terminated 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-poly (ethylene glycol) [DSPE-PEG-SH] through hydrophobic interactions, which enabled the anchoring of Au NPs to their phospholipid membranes.⁷⁶ This created EXO@Au NZ, which exhibits POD-like activity, allowing the oxidation of TMB in the presence of H₂O₂. The corresponding antibodies (Ab) targeting the cancer-related exosomal proteins including cluster of differentiation 63 (CD63), CEA, glypican-3 (GPC-3), programmed death-ligand 1 (PD-L1), and human epidermal growth factor receptor 2 (HER2) were immobilised on 96 well-plates. As per Fig. 1A, the EXO@Au NZ was then captured by these antibodies. The POD-mimicking activity of the Au NZ amplifies the detection signal and allows the quantification of the exosomal proteins present using a microplate reader. Its promise for early cancer diagnosis is highlighted through the effective distinguishing ability of the platform between hepatocellular carcinoma (HCC) patients, hepatitis B patients, and healthy donors. Moreover, the expression patterns of these exosomal proteins enable the classification of different cancer cell lines.

Fig. 1 Metallic NZ-based biosensors to visually detect simultaneous biomarkers and analytes in serum and urine. (A) Schematic illustration depicting the visualisation of the interaction between EXO@Au NZ and Ab in the presence of the target, which triggers the colour change according to the different protein levels. Reproduced with permission from ref. 76, copyright (2020) Theranostics. (B) Schematic representation of the colourimetric and ratiometric fluorescence detection of H₂O₂ and UA detection through the Fe CD/Mo SAC NZ. Reproduced with permission from ref. 81, copyright (2024) Elsevier. (C) h-Fe₃O₄@ppy NZ-colourimetric sensing platform for GSH and H₂O₂ determination. Adapted with permission from ref. 85, copyright (2021) Elsevier.

Chen et al. reported the colourimetric and ratiometric fluorescence detection of H₂O₂ and uric acid (UA) using a noble iron-carbon dot-embedded molybdenum single-atom nanoflower (Fe CDs/Mo SAC) NZ (Fig. 1B).⁸¹ Firstly, H₂O₂ and allantoin are generated from UA through uricase catalysis. H₂O₂ was then decomposed by the Fe CD/Mo SAC nanozyme to form hydroxyl radicals, which in turn oxidised o-phenylenediamine (OPD) to 2,3-diaminophenazine (DAP). This led to the colour change in the system from colourless to yellow, indicating an increasing concentration of UA. Meanwhile, the blue fluorescence of Fe CDs/Mo SACs at 460 nm was quenched by DAP, which emitted yellow fluorescence at 590 nm. Ratiometric fluorescence is based on the analyte-induced changes in the intensity of two or more emission bands.⁸⁴ The synergistic POD-like activity of Fe CDs/Mo SACs is evident in both methods. Furthermore, the dual-mode sensor is highly selective and can accurately detect UA in human urine and serum. The recovery rate of the ratiometric fluorescence and colourimetric method was 91.3 to 103.8% (RSD: less than 3.2%) and 93.0 to 107.0% (RSD: below 2.9%), respectively. Hence, this study promotes the application and development of single-atom NZs in multimodal-multiplexed sensing processes, which have shown enhancement in the performance of NZs.

Yang et al. reported a novel ternary magnetic nanocomposite hemin-Fe₃O₄@polypyrrole (h-Fe₃O₄@ppy) with excellent POD-like properties due to the presence of Fe₃O₄ (Fig. 1C).⁸⁵ The hemin improves the catalytic activity of the Fe₃O₄ NZ by being an active cofactor for the enzyme. As h-Fe₃O₄@ppy can catalyse TMB and H₂O₂ to produce a colour change, detection of H₂O₂ was successful through the change of the absorption signal within the linear range of 0.2 to 100 µM and LOD of 0.18 µM. Simultaneously, GSH was detected in biological samples from acute coronary syndrome patients by the reduction of yellow, oxidised TMB (oxTMB) to colourless, with a range of detection from 0.5 to 80 µM and a LOD of 0.15 µM.

4.2. Detection of various biomarkers through sensor arrays and microfluidic platforms

Wang et al. constructed POD-like Ir, Ru, and Pt NZ-based cross-reactive sensor arrays.⁸⁶ Distinct colourimetric response patterns were observable and identifiable through the interaction of NZs with various analytes, including cancer cells, proteins, and biothiols. The colourimetric changes depend on the catalytic reaction by NZs to oxidise OPD in the presence of H₂O₂ (Fig. 2A). The NZ-sensor arrays demonstrated their ability to discriminate analytes such as five cancer cells (4T1, HeLa, MCF-7, Hep-G2 and A2780), nine proteins, and six biothiols. Further validation is provided by the detection of unknown samples, whereby 42 of 45 proteins and 28 of 30 biothiols were identified precisely. Proteins in human urine and biothiols in serum were also discernible, which showcases the practical applications of the sensor arrays.

Fig. 2 Metallic NZ-based sensor arrays and microfluidic platform to simultaneously detect numerous biomarkers in serum, urine, and sweat. (A) Schematic illustration of the NZ sensor arrays for the visual discrimination of analytes from small molecules to proteins, as well as cells. (a) The POD-like Ir, Ru, and Pt NZ catalysed the oxidation of OPD. (b) The consequent reaction in the presence of analytes alongside the NZ. (c) The specific pattern recognition of the analytes detected by the NZ in the cross-reactive sensor arrays. Adapted with permission from ref. 86, copyright (2018) Elsevier. (B) The mechanism involved in the colourimetric detection of glucose, acetone, L-tryptophan, and uric acid using the developed method in the oxidation of 4-AAP/2,4-DCP, OPD/CAT and TMB through the Ni:Ce/Fe-Fum Bio-MOF NZ and ABTS through the Ag-Cu:Ce/Fe-Fum Bio-MOF NZ. Adapted with permission from ref. 80, copyright (2025) Elsevier. (C) Schematic flow of the cascade signal and visual amplification induced by the Pt NZ for the ovarian cancer biomarker detection. Adapted with permission from ref. 88, copyright (2025) Elsevier. (D) Construction of a POC hydrogel filled platform loaded with an Au/CoAl-LDO NZ and various reaction substances, to enable curated reactions and generate colourimetric response towards AST, ALT, and ALP. Adapted with permission from ref. 89, copyright (2022) ACS.

In a study conducted by Beygnezhad et al., they have constructed a Ce/Fe fumarate biological metal–organic framework (Ce/Fe-Fum Bio-MOF) NZ doped with Cu, Ni, and Ag NPs to detect biomarkers (acetone, glucose, uric acid, and L-tryptophan) that are indicative of diabetes in human sweat (Fig. 2B).⁸⁰ The NZ exhibits multiple enzyme-mimicking activities that are observable through the manifestation of distinct colour changes, based on the specific chromogenic substrate. The versatile NZ can display laccase-like, POD, OXD, and catechol oxidase-like mimicking activities when tested with 4-aminoantipyrine (4-AAP)/2,4-dichlorophenol (2,4-DCP), 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), TMB and OPD/CAT, respectively. The resulting colour patterns from the sensor platform were captured and analysed via smartphone signalling, which provides a rapid and simple quantitative detection method. The detection range of the sensor array was broad, from 20 to 1000 µM. Smartphone detection is part of optical sensing as smartphones are equipped with rear and front cameras, white light-emitting diode (LED), IR sensors, proximity sensors, and ambient light sensors (ALSs).⁸⁷ The portability and user-friendly properties of this biosensor are suitable for real-time monitoring of diabetic patients.

A portable gas-driven microfluidic platform involving Pt NZ was devised for the visual amplification and ultra-sensitive detection of ovarian cancer biomarkers, as illustrated in Fig. 2C.⁸⁸ The magnetic silica NPs were functionalised with capture antibodies to capture the ovarian cancer biomarker cancer antigen-125 (CA125) and human epididymis-specific protein 4 (HE4). The Pt NZ-labelled detection antibody was added to the same magnetic silica-capture antibody conjugates, enabling the formation of sandwich-type immunocomplexes with the target biomarker. The Pt NZ has POD-mimicking activity, which decomposes H₂O₂ to form reactive species and consequently catalyses the dopamine polymerisation into polydopamine (PDA), amplifying the detection signal. The attachment of the Pt NZ to 4-mercaptophenylboronic acid (4-MPBA) further enhances the immobilisation orientation and acts as an enzymatic signal amplifier, which results in the improvement of specificity for antigen binding in the platform. In addition, the POD-catalytic activity of the Pt NZ enables catalytic cascade reactions for biomarker readout to be visually observed on the microfluidic chip. The platform displayed detection limits as low as 0.1 pg mL⁻¹ for HE4 and 0.0001 U mL⁻¹ for CA125, which significantly surpass existing methods and clinical thresholds. Rapid and reliable portable testing is available for the diagnosis of ovarian cancer through the multi-purpose NZ in the biosensor.

Liu and colleagues designed a POC platform, which integrates NZ-loaded agarose hydrogels into a portable device, as per Fig. 2D.⁸⁹ A gold-decorated cobalt–aluminium layered double oxide (Au/CoAl-LDO) NZ displays POD-like activity, oxidising colourless TMB to a blue-hue product in the presence of H₂O₂ and liver-related biomarkers including alkaline phosphatase (ALP), alanine transaminase (ALT), and aspartate transaminase (AST). The detection mechanism is based on H₂O₂ generated by ALT and AST, which subsequently initiates the Au/CoAl-LDO NZ to catalyse the chromogenic reaction of TMB. Meanwhile, ALP hydrolysed L-ascorbic acid 2-phosphate to ascorbic acid and subsequently the oxTMB products with the aid of Au/LDO became discoloured. The information on the hydrogel's colour development can be converted to hue values by coupling this platform with a smartphone. It was found that quantitative analysis of ALP, AST, and ALT showed a detection limit of 5 U L⁻¹, 10 U L⁻¹ and 15 U L⁻¹, respectively. The platform offers a low-cost and convenient method to track liver health and provide a timely warning for liver diseases.

5. Advanced signal-amplification in multimodal strategies

5.1. Electrochemical, visual, and optical manifestations of multimodal readouts

Multimodal detection for a singular or more than two biomarkers is now the new standard in the biosensor realm. Numerous prior studies indicate that reliability and accuracy could be achieved through multimodal detection in a biosensor. For instance, Miao et al. presented a robust NZ-based immunosensor capable of dual colourimetric and ratiometric fluorescence detection of human cardiac troponin I (cTnI), as shown in Fig. 3A.⁹⁰ When cTnI was introduced into the immunosensor, the CeO₂ NZ–Ab₂ conjugate oxidised OPD to DAP, displaying its POD-like properties and producing a detectable colourimetric change. At the same time, DAP effectively quenched the fluorescence of the exogenous fluorescence signal source, which is the graphite carbon nitride quantum dots (g-C₃N₄ QDs). A reliable ratiometric fluorescence readout was obtained and this method showed good self-calibration ability. The lowest detection limit of cTnI was 0.413 pg mL⁻¹ for the ratiometric fluorescence method and 0.227 pg mL⁻¹ for the colourimetric method. Hence, the NZ-based multimodal platform enables sensitive and reliable cTnI detection.

Fig. 3 Unbound and bound metallic NZ-based multimodal sensors in detecting various biomarkers in blood, urine, and serum. (A) Illustration of a CeO₂ NZ-linked immunosensor mechanism for the dual ratiometric fluorescence and colourimetric determination of cTnI. Reproduced with permission from ref. 90, copyright (2019) Elsevier. (B) Visualization of the colourimetric, photothermal (smartphone signalling), and electrochemical NZ-based immunosensing of HER2. Reproduced with permission from ref. 91 copyright (2024) ACS. (C) The glucose detection mechanism involving the bimetallic NZ through TMB substrate oxidation and colour progression on a paper microchip. Reproduced with permission from ref. 92 copyright (2024) Elsevier. (D) The colourimetric-electrochemical detection of HSA using a hybrid nanozyme. Reproduced with permission from ref. 77 copyright (2026) Elsevier.

A bimetallic single-atom NZ was developed for electrochemical-photothermal and smartphone imaging-based detection of HER2 in breast cancer, as displayed in Fig. 3B.⁹¹ Here, the iron-manganese ion N-doped carbon single-atom (FeMn-NC etch/SAC) was the peroxidase-mimetic NZ used. In the presence of HER2, GOx attached to the secondary antibody (mAb₂) generated the production of H₂O₂. The NZ then decomposed H₂O₂ to hydroxyl radicals, which in turn catalysed the conversion of TMB into oxTMB. The colourimetric signal was subsequently converted into both photothermal and electrochemical signals. The product was drop-cast on a screen-printed electrode for electrochemical signal determination through chronocurrent measurement. For photothermal detection, the same microtiter-plate was irradiated with a laser, and the temperature change of the sample was monitored via a near-infrared imaging camera on a smartphone. The lowest limit of electrochemical detection was 3.9 pg mL⁻¹ (linear range: 0.01 to 10 ng mL⁻¹) and for photothermal detection it was 7.5 pg mL⁻¹ (linear range: 0.01 to 2 ng mL⁻¹). This biosensor offers POC testing potential and enhanced accuracy for HER2 detection in early prognosis and diagnosis of breast cancer.

A study was conducted by Fareeha and colleagues for the visual detection of glucose through a microfluidic platform and colourimetric method, by utilising a NZ that consists of gold nanorods (AuNRs) and CeO₂.⁹² The NZ imitates and enhances the GOx activity in the colourimetric mode alongside the GOx added, such that the NZ catalyses the oxidation of glucose present in gluconic acid and H₂O₂. H₂O₂ then produces free hydroxyl radicals, which cause further oxidation of colourless TMB to blue-hued products (Fig. 3C). The same principle was also applied for the paper substrate devised. The biosensor demonstrated a LOD of 0.65 mg dL⁻¹ within a wide linear range of 0.5 to 500 mg dL⁻¹. Excellent recovery rates in urine (91.84% to 109.5%) and serum (93% to 98.27%) were also validated in this biosensor. Hence, the potential practical application of the paper sensor is significant for simple and portable POC for glucose monitoring, especially in low-resource settings.

Mohd Salleh et al. have engineered a hybrid nanozyme that combines CeO₂ NPs and cellulose nanocrystals (CNC), which displayed significant POD-like activity for the detection of human serum albumin (HSA).⁷⁷ The synergistic electrocatalytic behaviour of the NZ enables the generation of both colourimetric and electrochemical responses through the reduction of H₂O₂. The hydroxyl radical formed in colourimetric mode further reacts with TMB to form oxTMB (Fig. 3D). The constructed dual-detection HSA immunosensor showed a wide linear detection range from 0.1 pg mL⁻¹ to 1000 ng mL⁻¹ with a low detection limit of 5.13 pg mL⁻¹ colourimetrically and 5.85 pg mL⁻¹ electrochemically. The hybrid NZ enabled enhanced selectivity, specificity, and sensitivity and showed promising recovery rates for HSA detection in a human serum matrix. Only a single 96-well microtiter plate was used for the dual-strategy detection, leading to a reduced sample volume requirement, highlighting the cost-effectiveness of the biosensor. Furthermore, the reliability and robustness of the sensor showcased the suitability and potential for POC applications.

5.2. DNA probe-NZ based amplification modules

Aptamers (apts) are known as “chemical antibodies” and consist of short single-stranded RNA or DNA molecules with high affinity and specificity for a biomarker.⁹³ The flexibility of the aptamer structure provides significant specificity for target recognition, while the NZ is responsible for signal output.⁹⁴ In most cases, the activity of the NZ is enhanced with these modifications.

In a study conducted by Gong et al., they combined MoS₂ with Au@Pt NPs to form a NZ. They merged the NZ and the specific recognition ability of apts for the dual-detection of EXOs extracted from the human breast cancer cell line MCF-7 (Fig. 4A).⁹⁵ Probe P1 was immobilised on the MoS₂–Au@Pt NZ's surface to form MoS₂-based signal amplified nanoprobes (MNP) through Au–S bonds. The MNP could effectively catalyse the TMB oxidation in the proximity of H₂O₂, due to the POD-mimicking activity and excellent electrocatalytic capability of the MoS₂–Au@Pt NZ. There was an enhanced colourimetric and electrochemical response as a result. When target EXOs were introduced, the specific reaction between the CD63 aptamer and the EXOs led to the dissociation of the sandwich structure and the release of MNPs from the surface of the electrode, resulting in a lower electrochemical signal intensity and a reduced blue intensity of the solution. The linear range of EXOs detected through this approach was 10 to 10⁹ particles per mL and 10⁴ to 10⁹ particles per mL for electrochemical and colourimetric signals, respectively, with detection limits of 4.2 × 10³ particles per mL (colourimetric method) and 9.3 particles per mL (electrochemical method). This dual-mode sensing approach offers high recoveries and low relative standard deviations for the detection of various concentrations of MCF-7-derived exosomes in 5% human serum. Mutual verification of the assay results was guaranteed and there were minimal errors on account of assay operation and conditions, which provides a new approach for the sensitive and accurate detection of biomarkers.

Fig. 4 Coupling the NZ with DNA probes for the multi-modal detection of biomarkers. (A) The proposed MoS₂-Au@Pt NZ in the colourimetric/electrochemical dual-mode aptasensor platform for the sensitive detection of EXOs. Adapted with permission from ref. 95, copyright (2024) Elsevier. (B) Principle of the proposed aptasensor for CTC determination employing the MOF@Pt@MOF NZ and the DNA walker. Separation through magnetism and release of cDNA chains in the proximity of MCF-7 cells, followed by mapping the outline of the ITO chamber and the detection mechanism of the developed CTC sensor. Adapted with permission from ref. 96, copyright (2024) Elsevier. (C) Schematic representation of the proximity hybridization-MoS₂ NS NZ-based triple mode immunosensing of HE4. Adapted with permission from ref. 82, copyright (2020) Elsevier.

In a similar study, Zhao and colleagues developed an electrochemical and colourimetric dual-mode sensing strategy to detect and capture MCF-7 circulating tumour cells (CTCs).⁹⁶ In this study, the cellular sensing approach is based on a MOF@Pt@MOF NZ and a DNA walker, as well as a SYL3C apts with high specificity for epithelial cell adhesion molecule (EpCAM) protein to capture the CTCs (Fig. 4B). There was prehybridisation of SYL3C apts with their cDNA and they were modified on magnetic nanospheres (MNs). When MCF-7 CTCs were present, cDNA was released via the specific recognition of apts and CTCs, which were then separated by magnets and consequently triggered the DNA walking process. During the DNA walking process, the cDNA can activate the DNA replacement reaction through the MOF@Pt@MOF NZ-modified signal probe (SP). This, in turn, releases the cDNA chain to continue and participate in the cycle, resulting in the amplification of the output signal. The outstanding synergistic effect and amplification capabilities of NZ are highlighted through the good POD-like activity of NZ in the presence of TMB (colourimetric) and H₂O₂ (electrochemical) to detect CTCs. A comprehensive linear range of 5 to 5 × 10⁵ cells per mL and a detection limit as low as 5 cells per mL were obtained through the dual-mode detection. The dual-mode aptasensor was evaluated in human serum samples to detect the CTCs and showed recovery rates of 100% to 109% (RSD: 0.6% to 2.3%), proving its practicality. Overall, the recognition process in the liquid phase enhanced the nucleic acid reaction, and the direct binding of the NZ signal probe to the indium tin oxide (ITO) chamber's surface reduced steric blocking compared with binding it to the cell. This novel approach could be replicated for the detection of other biomarkers in biofluids, enriching the multimodal-multiplexed field.

Lastly, Zhang and colleagues introduced molybdenum disulfide nanosheets (MoS₂ NSs) in an immunosensing platform, which exhibited excellent photothermal effect through increasing electrode temperature and electrocatalytic activity to dissolve O₂ in electrochemiluminescence system (Fig. 4C).⁸² Here, two antibodies were conjugated to each DNA1 and DNA2, forming DNA1-Ab₁ and DNA2-Ab₂. Both were added to the surface of a glassy carbon electrode (GCE), which was immobilised with a capture DNA strand (DNA3). When the antibodies recognise the presence of HE4, both the DNA–Ab conjugates come into close proximity to each other, resulting in proximity hybridisation. Partial hybridisation with DNA3 simultaneously occurs to form a double-stranded DNA complex. The remaining single-stranded DNA segment complex then adsorbs the MoS₂ present. The MoS₂ NSs function as the NZ such that they imitate POD-mimicking activity, which oxidises ABTS to generate a colourimetric response and improve the electrochemiluminescence (ECL) performance of a luminol cathode. The ECL method depends on the light emission from luminophores, which is generated by the high-energy electron transfer reactions between an electrode and electrogenerated species.⁹⁷ Photothermal sensors rely on the temperature change that occurs due to the materials being irradiated with light of specific wavelength.⁹⁸ The oxidised ABTS, when interacting with the NZ, displays strong photothermal characteristics such that ABTS generates heat when irradiated with an NIR laser, which leads to the temperature-based quantification of HE4. The platform was able to detect HE4 as low as 3 × 10⁻⁷ ng per mL for ECL and 3 × 10⁻⁵ ng per mL for colourimetric and photothermal sensing. The proximity hybridisation-based multiple stimuli-responsive platform for photothermal sensing, colourimetric, and ECL detection of the ovarian cancer biomarker in human serum is an attractive design for a detection platform. Hence, triggering multiple signals by using a NZ to detect a range of biomarkers is now necessary in the research landscape.

6. Challenges and opportunities for the NZ-multiplexed and multimodal biosensor

6.1. Specificity, stability, and reproducibility of NZ-based biosensors

Considerable promise in the field of metallic NZ integration into multimodal and multiplexed biosensors has been elaborated in the previous sections. There remain several challenges that need to be addressed for this emerging field to reach its peak performance, which is the daily diagnostic application of biomarker detection in human biofluids. For instance, most NZs have poor activity due to the absence of a substrate-binding pocket, leading to a lack of specificity.⁹⁹ Apart from functionalising the NZ with Ab or probe, the issue could be further resolved through single-atom metallic nanozyme (SANS) design. Atomic-level tuning of SANS enables well-defined coordination and precise arrangement of mononuclear metal sites that closely simulate natural enzyme active centres. Their atomically dispersed metal atoms, anchored onto a support material, create uniform and structurally identical active sites that offer consistent catalytic performance and tunable selectivity.¹⁰⁰ This precise architecture promotes specific substrate binding, reduces non-specific interactions, and avoids atom aggregation, in contrast to natural enzymes that possess heterogeneous surface atoms with variable shape, size, and coordination. Consistent batch performance and tunable catalytic activity enhance the uniformity and selectivity of SANS,¹⁰¹ increasing the availability of active sites and thereby improving the overall activity during the detection process.¹⁰²

Achieving uniform size and composition of NZs is challenging as there might be slight variation in preparation conditions that leads to inconsistency in catalytic behaviour between different batches of NZs. Khramtsov et al. tested three individual batches of Prussian blue NPs loaded with BSA (PB@BSA5), and they showed similar yield and NZ content; yet batch-to-batch variations of ∼20 nm in size were observed despite identical synthesis conditions.¹⁰³ Similarly, another study reported a −9% to +11% deviation between target and observed Prussian blue NP sizes, with NZ stability limited to three months before aggregation occurred.¹⁰⁴ The coefficient of variation (CV) of the NZ was found to be significantly low, with 0.3 to 2.6%, signifying good batch-to-batch reproducibility. Other NZs showed a low CV of 2 to 5% for Au¹⁰⁵ and <5% for poly(D,L-lactide-co-glycolide),¹⁰⁶ whereas dextran-coated iron oxide NZs exhibited a much broader CV of 10 to 42%.¹⁰⁷ Another study conducted by Sun et al. also proved that there is low batch-to-batch variability in the different batches of Au@Pd@Pt NZ based on the sensitivity, recovery rates, and reproducibility performance of lateral flow immunoassay (LFIA) in the detection of SARS-CoV-2 nucleocapsid protein. The relative standard deviations for all the tests were less than 4% indicating stable percentages across these batches, ensuring that the NZ is reliable for practical diagnostic use.¹⁰⁸ Nevertheless, a few NZs still encounter variation in batch-to-batch morphology heterogeneity and therefore require improvement in their reproducibility.

Reproducibility issues also arise from insufficient characterisation, usually limited to TEM and hydrodynamic size, while omitting zeta potential, surface area, impurities, and metal leaching analysis, all of which can contribute to variability as well as a lack of standardisation across in-batch, intra-batch, and inter-batch measurements of NZ performance and poor experimental design.¹⁰⁹ The reproducibility issues could be refined by the creation of monodisperse NZs. Besides that, chemically entrapping the dispersed NZ on substrates or growing the NZ directly on substrates could extend the long-term storage stability of the NZ, which is usually not achievable with the colloidal state of the NZ.¹¹⁰ Additionally, lyophilisation (freeze-drying) of NZs enhances the long-term shelf life without affecting the NZ's function.^111,112

6.2. Sensitivity, standardisation, and biological matrix interaction of NZs

The intrinsic chemical structure determines NZs' catalytic performance, which requires improvement due to their low sensitivity to substrates and targets. The sensitivity and selectivity of NZs can be improved through linking NZs with molecular imprinted polymers (MIPs), which enhance the overall activity during sensing.^113,114 Not only that, further investigation of the possible changes in the activity of NZs and signal amplification after their dispersion in a biological buffer is of great importance, to ensure their applicability in the biomarker detection of human biofluids. There is also a lack of in-depth catalytic performance evaluation in terms of the connection between the in vivo biological activities of NZs at the molecular level and their catalytic characteristics, as well as the catalytic mechanisms involved.¹¹⁵ Devoting more energy to understanding each of these mechanisms will greatly benefit the regulation and trajectory of the catalytic activities of NZs.

Furthermore, ISO-compliant standardised protocols for NZ characterisation, synthesis, and clinical validation are essential for their acceptance and integration within healthcare settings. Various regulatory requirements must be fulfilled for the reliable clinical translation of NZs, such as robust matrix characterisation to assess NZ surface chemistry, aggregation state, size, and catalytic behaviour in complex biofluids, where background protein binding can affect the signal output of the biosensor. Besides this, the NZ-based assay in laboratory settings must be simple enough for non-specialists to perform reliably and smoothly, minimising user-dependent errors.¹¹⁶ Reproducibility and repeatability are critical such that the results of synthesised NZs and biosensor output must be consistent across different runs, batches, and testing sites.¹¹⁷ Batch-to-batch comparability should be demonstrated for at least three lots, ensuring manufacturing consistency, including all critical quality attributes (CQAs) and functional performance. Even during the scale-up of NZs, the NZ properties that generate biosensor signals must remain intact.¹¹⁸ Apart from this, the biosensor must maintain high selectivity to produce no detectable signal or altered signal in the presence of structurally similar biomarkers. Selectivity in clinical matrices should be benchmarked against clinical thresholds and validated using appropriate controls and calibrators.¹¹⁹ Moreover, establishing the LOD and linear range is important for defining the analytical performance, sensitivity, and alignment with clinical decision limits. NZ shelf-life and reagent stability must be validated, including temperature sensitivity and shipping requirements, to ensure the long-term reliability of the designed biosensor.¹²⁰

6.3. Limited type of NZ and machine learning integration

Furthermore, the range and type of NZ and their targets should be expanded to improve both the time-to-diagnosis and diagnostic approach of biomarker detection, especially in POC treatment.^121,122 Despite the cost-effectiveness of the methods involved in NZ production, noble metal nanomaterials (Pd, Pt, and Au) are still costly. Hence, efforts should be directed towards the green synthesis of metal nanomaterials.¹²³ Other than that, NZs should be rationally designed to enable prediction of their effect on the target and alignment of their catalytic performance with the sensing strategy.^124,125 Clear insights into the specific design of an efficient NZ are attainable through computational technology, particularly machine learning (ML). The optimal NZ properties are precisely tuned based on their specificity, stability, and catalytic efficiency.¹²⁶

A growing body of NZ literature has enabled the development of comprehensive datasets that can train machine-learning models to optimise NZ design, synthesis, and reaction conditions. These models provide data-driven recommendations and comparisons of different NZs reducing reliance or risks of trial-and-error strategies.^127,128 Pairing high-throughput computational screening with ML enables rapid exploration of NP composition, surface engineering, and doping strategies to optimise structures tailored for specific biomarkers.¹²⁹ Computational tools can also predict catalytic performance with high precision, biocompatibility in complex biofluids, aggregation behaviour, and metal leakage, while providing structure–function mechanisms.^130,131 Overall, computational modelling and data-driven design significantly accelerate NZ optimisation by improving reproducibility and scalability, shortening design cycles and enabling further reliable translation into practical biosensing applications. Although most of the ML-assisted optimisations for NZ design are still in the proof-of-concept stage and not yet fully translated to biomarker assay, the gap is actively narrowing between biosensor-oriented ML and design-oriented ML.¹³² The two cutting edges of both NZ and ML technologies provided a synergistic impact and hold a great outlook for the future of biotechnology.

6.4. Multimodal and multiplexing complexity

The significant challenge for multiplexing and multimodal biosensing is cross-interference. If the bioreceptor is unable to differentiate between the target biomarkers and other similar structural components, this will drastically reduce the specificity of immunoreactivity.¹³³ As a result, multiplexed and consequently multimodal detection might not function properly, and its application in complex clinical and biological samples is hindered. Apart from the construction of novel NZs with high catalytic and substrate selectivity, several crucial elements must be followed for the successful implementation of biosensing systems, and this includes two key components in the biosensor development, namely, recognition probes that can specifically interact with the particular biomarker and detection probes that can provide a discernible signal contrast in the final imaging process. Lastly, the readout system must differentiate the information of specific analytes through quantifying and evaluating the detection probes with the analyte accurately, without affecting the other modalities.^134,135 Regardless of the existing limitations and challenges stated, the research field of NZ detection continuously offers limitless possibilities in resolving challenges involving sensors for biomarker detection in biofluids, particularly in multimodal and multiplex modes.

7. Conclusion and future outlook

The alarming rate of chronic diseases around the globe highlights the importance of early detection of biomarkers involved in these diseases. The current diagnostic landscape carries several drawbacks including invasive procedures, slow processes, labour-intensive methods, the requirement of skilled professionals, issues with specificity and selectivity, and unsuitability for POC testing. There is still a great demand for practical and advanced medical diagnosis, despite the fact that medical treatments have advanced tremendously. Delayed treatment and diagnostics, particularly in locations such as underdeveloped countries and poor regions with inadequate medical resources, usually result in exacerbated conditions in individuals.¹³⁶ Hence, the integration of NZ-based biosensors offers new opportunities and great possibilities for the multiplexing and multimodal detection of biomarkers.

The versatility of metallic NZs is evident in a wide range of fields, including the rapidly advancing field of biosensors, in which they are capable of detecting and monitoring biomarkers at early stages of various diseases. By leveraging their enzyme-mimicking activity that is programmable and tunable, multifunctionality, biocompatibility, and ease of synthesis, NZs are ideally used in the multimodal and multiplexed sensing platform.¹³⁷ False-negative or false-positive results can effectively be circumvented by the mutual calibration of multiple signals in detecting multiple biomarkers present in limited sample volumes. Researchers have highlighted the favourable features of NZs to design reliable, cost-effective, and highly sensitive diagnostic platforms. However, challenges such as specificity, stability, non-uniform composition, lack of “design-for-purpose” of NZs, and cross-reactivity in multimodal and multiplexed POC biosensors remain, though they may be overcome. The pursuit of innovative development in NZ-based multiplexed and multimodal biosensors for biomarker detection is still ongoing, which is crucial to realise more accessible, effective, and personalised diagnostic tools in the future.

7.1. AI-driven NZ data processing

To further revolutionise and pave the way in the NZ-based diagnostic and personalised medicine field, AI-driven data processing in the detection of biomarkers could substantiate the gaps in manual studies carried out and reduce human errors.¹³⁸ Sensor systems paired with an AI algorithm have the potential to self-learn from patterns of data inputs and autonomously make decisions.¹³⁹ Xu et al. fabricated a fluorescence sensor array that employed both NZs and bioenzymes for the highly sensitive identification of Aβ42 and Aβ40 peptides (Alzheimer's biomarkers).¹⁴⁰ Two step-coupling occurred between NZs and bioenzymes, which converted the weak interaction differences between the target and sensing elements (Au NZ-DNA) into various fluorescence amplification responses via calibration of the catalytic activity. In the presence of the target, the catalytic activity either decreased or increased based on the association between Au NZ-DNA complexes and Aβ peptides. This resulted in a cascade reaction that generated fluorescence variations that were utilised to determine the fingerprints of various Aβ peptides. Through classification models of a pattern recognition ML strategy, the hybrid NZ/biocatalyst sensor array was able to identify various Aβ aggregates at a low concentration of 200 nM with 100% accuracy. Another study employed apts-modified C₃N₄ nanosheets (Apt/C₃N₄ NSs) for the ratiometric fluorescence detection of various tumour exosomes. In the presence of target exosomes, the apts binding led to the dissociation of the NZ from the complex, lowering its catalytic activity and reducing the production of 4′,6-diamidino-2-phenylindole (DAPI). ML was used to discriminate the various types of exosomal proteins and distinguish the blood of healthy controls and cancer patients. The three aptamers employed in this study produced a distinctive response pattern across the sensor units, facilitating ML-based classification. The platform achieved a low LOD of 2.5 × 10³ particles per mL, demonstrating high sensitivity, while ML further enhanced the discrimination ability and selectivity of exosomal proteins.¹⁴¹ Thus, integrating AI with NZs markedly enhances biomarker sensitivity and classification in biosensing.

7.1.1. Smartphone AI–NZ detection platform. Multimodal biomarker detection should at least include smartphone attachment for data transfer and output signal processing.^142,143 Smartphone-enabled multiplex monitoring could enhance practicality, speed, and real-time tracking of biomarkers.¹⁴⁴ Smartphone use also provides superior results and reduces any potential errors.¹⁴⁵ Besides that, smartphone-based biosensors with AI algorithms and image processing enable a rapid, robust, and low-cost system that is accessible and user-friendly for non-expert users. Personalised health monitoring is also feasible and accessible to individuals in all age groups. AI has thus far been employed in several ways, such as regression, segmentation, and classification models to assess colour changes in colourimetric assays.¹⁴⁶

Further expanding the AI–NZ scope with smartphone integration, Zhong et al. devised colourimetric detection of unsaturated fatty acids (UFAs) based on competitively inhibiting MnO₂ NZ-catalysed ABTS oxidation and producing a measurable colour change.¹⁴⁷ The “Quick viewer” smartphone application captured the colour shifts and segmented each image into a 3 × 4 grid, enabling per-well feature extraction using RGB values and deep neural network features. The model was trained for the prediction of each fatty acid's concentration and to classify individual UFAs and their mixtures based on the colourimetric patterns. It achieved R² values of 0.9969 (oleic acid (OA)), 0.9668 (linoleic acid (LA)), and 0.7393 (α-linolenic acid (ALA)), with strong accuracy for OA and LA and acceptable quantification for ALA. The trained model was then integrated into the “Intelligent analysis master” app for one-click, on-phone quantification.

In another study, a bimetallic single-atom NZ (CuZn-N) was integrated with a smartphone and deep learning (DL) algorithms to detect homocysteine, glutathione, and cysteine biomarkers.¹⁴⁸ Multiple DL frameworks were embedded in the “ThiolSense” mobile application to quantify these analytes accurately in human serum. Initially, the analyte-specific inhibition of CuZn-N activity triggered the colourimetric responses generated. Here, the mobile application system can perform automated segmentation and extraction of colourimetric input data from the smartphone-captured images, followed by robust pattern recognition. This platform enabled multiplexed and accurate identification of homocysteine, glutathione, and cysteine with detection limits between 1.17 and 1.35 nM, and within a wide linear range down to the clinical threshold. DL significantly reduced error arising from changes in ambient lighting, imaging angle, and user operation factors that frequently affect manual or conventional single-variable analyses. This was done through enabling real-time and user-independent processing of massive and high-dimensional datasets. Thus, the combination of NZ, AI, and smartphone technology offers a robust and practical tool for on-site biosensing. In this review, we have compiled and summarised previous studies that utilised metallic NZs in multiplexed and multimodal platforms. This proof of concept demonstrates the feasibility and strong potential for wearable and portable integration.

7.2. Personalised medicine and big data

The devices to measure the output signal of biomarkers should be miniaturised and portable to ensure accessibility and convenience for everyone involved. The wearable devices exploit various biological, chemical, and physical modalities to acquire physiological information from interstitial fluid, urine, blood, sweat, tears, or saliva in real time (continuously, if possible) in a non-invasive or minimally invasive manner.¹⁴⁹ Furthermore, a diverse database that compiles and includes individual-to-large-population health records is necessary to evaluate the health consensus of the countries, which is derived from access to wearable sensors or portable devices. The information is then assessed through big data analysis.¹⁵⁰ The sensors come in different shapes and forms, such as patches or bandages, textiles, wristwatches, fitness bands, contact lenses, face masks, jewellery, and glasses, as an alternative pathway to clinical diagnostics.^151–154 Again, next-generation wearable sensors and portable devices are the future trajectory in enabling multiplexed and multimodal measurement of biomarkers, whereby monitoring of physical parameters continuously and in real time is possible. Allowing time-resolved and high-resolution historical recording of the individual's health status is definitely a transformative technology required for the diagnostics field. In conclusion, bridging the knowledge gap between practical implementation and theoretical advancements is pivotal to realise the full potential and innovation of POC multimodal and multiplexed NZ-based biomarker detection in the application of routine clinical settings.

Author contributions

Batrisyia Safwah Mohd Salleh: writing – original draft, validation, methodology, data curation, conceptualisation; Minhaz Uddin Ahmed: writing – review and editing, supervision, conceptualisation, validation, resources, project administration, investigation, funding acquisition.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

No primary research results, software or code have been included, and no new data were generated or analysed as part of this review.

Acknowledgements

This work was partly supported by the Universiti Brunei Darussalam's grant UBD/RSCH/URG/RG(b)/2023/036 and the Brunei Research Council Grant No. 10. Batrisyia Safwah Mohd Salleh wishes to thank the Ministry of Education of Brunei for her postgraduate scholarship.

References

1. C. Zoccali, G. Tripepi, V. Stel, E. L. Fu, F. Mallamaci, F. Dekker and K. J. Jager, Clin. Kidney J., 2024, 17, 130.
2. A. Ahmad, M. Imran and H. Ahsan, Pharmaceutics, 2023, 15, 1630.
3. A. Passaro, M. Al Bakir, E. G. Hamilton, M. Diehn, F. André, S. Roy-Chowdhuri, G. Mountzios, I. I. Wistuba, C. Swanton and S. Peters, Cell, 2024, 187, 1617.
4. S. T. Sanjay, G. Fu, M. Dou, F. Xu, R. Liu, H. Qi and X. Li, Analyst, 2015, 140, 7062.
5. F. Arshad, D. W. M. Arrigan and M. U. Ahmed, IEEE Sens. J., 2022, 22, 15622–15634.
6. Y. Z. Lee, S. C. Low and C. L. Ngan, Biomanufacturing for Sustainable Production of Biomolecules, 2023, pp. 87–122.
7. J. D. Newman and A. P. F. Turner, Biosens. Bioelectron., 2005, 20, 2435–2453.
8. F. Karim and A. N. M. Fakhruddin, Rev. Environ. Sci. Biotechnol., 2012, 11, 261–274.
9. R. Karunakaran and M. Keskin, Analytical Techniques in Biosciences: from Basics to Applications, 2022, pp. 179–190.
10. N. Verma and A. Bhardwaj, Appl. Biochem. Biotechnol., 2015, 175, 3093–3119.
11. A. Kumar, S. Paliwal, D. Kalra, A. Varghese, S. Tripathy and S. K. Ghoshal, Mater. Sci. Semicond. Process., 2024, 174, 108225.
12. A. Cavalcanti, B. Shirinzadeh, M. Zhang and L. C. Kretly, Sensors, 2008, 8, 2932–2958.
13. P. Mehrotra, J. Oral. Biol. Craniofac. Res., 2016, 6, 153.
14. P. Damborský, J. Švitel and J. Katrlík, Essays Biochem., 2016, 60, 91–100.
15. A. C. Murphy, M. E. Wechsler and N. A. Peppas, Curr. Opin. Biomed. Eng., 2021, 19, 100318.
16. C. Padmakumari Kurup, S. Abdullah Lim and M. U. Ahmed, Bioelectrochemistry, 2022, 147, 108170.
17. C. P. Kurup and M. U. Ahmed, Biosensors, 2023, 13, 461.
18. L. Gao and X. Yan, Adv. Exp. Med. Biol., 2019, 1174, 291–312.
19. F. Arshad, M. Deliorman, P. Sukumar, M. A. Qasaimeh, J. S. Olarve, G. N. Santos, V. Bansal and M. U. Ahmed, Trends Food Sci. Technol., 2023, 136, 145–158.
20. M. Perwez, S. Y. Lau, D. Hussain, S. Anboo, M. Arshad and P. Thakur, Colloids Surf., B, 2023, 225, 113241.
21. A. M. Ashrafi, Z. Bytesnikova, J. Barek, L. Richtera and V. Adam, Biosens. Bioelectron., 2021, 192 DOI:10.1016/j.bios.2021.113494.
22. K. Xu, Y. Cui, B. Guan, L. Qin, D. Feng, A. Abuduwayiti, Y. Wu, H. Li, H. Cheng and Z. Li, Nanoscale, 2024, 16, 7786–7824.
23. S. Ahmadi, K. Rahimizadeh, A. Shafiee, N. Rabiee and S. Iravani, Process Biochem., 2023, 131, 154–174.
24. D. Manoharan, L. C. Wang, Y. C. Chen, W. P. Li and C. S. Yeh, Adv. Healthcare Mater., 2024, 13, 2400746.
25. W. Yang, X. Yang, L. Zhu, H. Chu, X. Li and W. Xu, Coord. Chem. Rev., 2021, 448, 214170.
26. S. H. Elagamy, R. H. Obaydo, A. Ashkar and H. M. Lotfy, Sens. Actuators Rep., 2025, 10, 100394.
27. Z. Zhang, Y. Li, Z. Yuan, L. Wu, J. Ma, W. Tan, Y. Sun, G. Zhang and H. Chai, Inorg. Chem. Front., 2025, 12, 400–429.
28. R. Singh, A. Umapathi, G. Patel, C. Patra, U. Malik, S. K. Bhargava and H. K. Daima, Sci. Total Environ, 2023, 854, 158771.
29. X. Ren, D. Chen, Y. Wang, H. Li, Y. Zhang, H. Chen, X. Li and M. Huo, J. Nanobiotechnol., 2022, 20, 1–18.
30. D. Liang, Y. Wang and K. Qian, Interdiscip. Med., 2023, 1, e20230020.
31. Y. Wu, D. C. Darland and J. X. Zhao, Sensors, 2021, 21, 5201.
32. P. K. Jain, X. Huang, I. H. El-Sayed and M. A. El-Sayed, Acc. Chem. Res., 2008, 41, 1578–1586.
33. Y. Y. Wu, X. Tian, Y. Jiang, H. Y. Ma, W. Wang, W. S. Zhang, J. S. Martin, Y. Yan, D. D. Qin, D. X. Han and L. Niu, TrAC, Trends Anal. Chem., 2024, 176, 117757.
34. G. Lee, C. Kim, D. Kim, C. Hong, T. Kim, M. Lee and K. Lee, ACS Omega, 2022, 7, 40973.
35. Y. Gao, Y. Wu and J. Di, Spectrochim. Acta, Part A, 2017, 173, 207–212.
36. Y. Jv, B. Li and R. Cao, Chem. Commun., 2010, 46, 8017–8019.
37. G. L. Wang, L. Y. Jin, X. M. Wu, Y. M. Dong and Z. J. Li, Anal. Chim. Acta, 2015, 871, 1–8.
38. X. Zhu, T. Liu, H. Zhao, L. Shi, X. Li and M. Lan, Biosens. Bioelectron., 2016, 79, 449–456.
39. A. Hayat, J. Cunningham, G. Bulbul and S. Andreescu, Anal. Chim. Acta, 2015, 885, 140–147.
40. S. F. Zhao, F. X. Hu, Z. Z. Shi, J. J. Fu, Y. Chen, F. Y. Dai, C. X. Guo and C. M. Li, Nano Res., 2021, 14, 879–886.
41. L. Su, W. Dong, C. Wu, Y. Gong, Y. Zhang, L. Li, G. Mao and S. Feng, Anal. Chim. Acta, 2017, 951, 124–132.
42. Z. Liang, D. Ou, D. Sun, Y. Tong, H. Luo and Z. Chen, Biosens. Bioelectron., 2019, 146, 111744.
43. B. Wang, Y. Luo, L. Gao, B. Liu and G. Duan, Biosens. Bioelectron., 2021, 171, 112736.
44. J. Yao, Z. Xie, X. Zeng, L. Wang and T. Yue, Sens. Actuators, B, 2022, 354, 130760.
45. A. Mocanu, I. Cernica, G. Tomoaia, L. D. Bobos, O. Horovitz and M. Tomoaia-Cotisel, Colloids Surf., A, 2009, 338, 93–101.
46. F. Arshad, S. Nurul Azian Zakaria and M. U. Ahmed, Microchem. J., 2024, 201, 110523.
47. N. Tagaras, H. Song, S. Sahar, W. Tong, Z. Mao and T. Buerki-Thurnherr, Adv. Sci., 2024, 11, 2407816.
48. X. Wang, C. Shu, G. Wang, P. Han, L. Zheng, L. Xu and Y. Chen, Nanoscale, 2025, 17, 10557.
49. J. Qin, N. Guo, J. Yang and J. Wei, Food Chem., 2024, 447, 139019.
50. N. Alizadeh, A. Salimi and R. Hallaj, Talanta, 2018, 189, 100–110.
51. G. Fu, S. T. Sanjay, W. Zhou, R. A. Brekken, R. A. Kirken and X. Li, Anal. Chem., 2018, 90, 5930.
52. C. He, D. Liu and W. Lin, Chem. Rev., 2015, 115, 11079–11108.
53. C. P. Raptopoulou, Materials, 2021, 14, 310.
54. Y. Hu, H. Cheng, X. Zhao, J. Wu, F. Muhammad, S. Lin, J. He, L. Zhou, C. Zhang, Y. Deng, P. Wang, Z. Zhou, S. Nie and H. Wei, ACS Nano, 2017, 11, 5558–5566.
55. W. Zhang, X. Li, T. Cui, S. Li, Y. Qian, Y. Yue, W. Zhong, B. Xu and W. Yue, Mikrochim. Acta, 2021, 188(5), 174.
56. L. Zhang, X. Hai, C. Xia, X. W. Chen and J. H. Wang, Sens. Actuators, B, 2017, 248, 374–384.
57. U. Nishan, N. Jabeen, A. Badshah, N. Muhammad, M. Shah, I. Ullah, S. Afridi, J. Iqbal, M. Asad, R. Ullah, E. A. Ali, S. Ahmed and S. C. Ojha, Front. Bioeng. Biotechnol., 2024, 12, 1364700.
58. M. Chatterjee, P. Nath, S. Kadian, A. Kumar, V. Kumar, P. Roy, G. Manik and S. Satapathi, Sci. Rep., 2022, 12, 9061.
59. X. Cao, M. Liu, M. Zhao, J. Li, J. Xia, T. Zou and Z. Wang, Anal. Bioanal. Chem., 2022, 414, 1921–1932.
60. T. Zhu, Q. Tang, Y. Zeng, S. Chen, Y. Yang, H. Wang, J. Chen, L. Guo and L. Li, Spectrochim. Acta, Part A, 2023, 293, 122474.
61. M. Broto, M. M. Kaminski, C. Adrianus, N. Kim, R. Greensmith, S. Dissanayake-Perera, A. J. Schubert, X. Tan, H. Kim, A. S. Dighe, J. J. Collins and M. M. Stevens, Nat. Nanotechnol., 2022, 17, 1120–1126.
62. S. Li, Y. Zhang, Q. Wang, A. Lin and H. Wei, Anal. Chem., 2022, 94, 312–323.
63. X. Wang, L. Qin, M. Lin, H. Xing and H. Wei, Anal. Chem., 2019, 91, 10648–10656.
64. X. Wang, G. Wei, W. Liu, Y. Zhang, C. Zhu, Q. Sun, M. Zhang and H. Wei, Anal. Chem., 2023, 95, 5937–5945.
65. L. Qin, X. Wang, Y. Liu and H. Wei, Anal. Chem., 2018, 90, 9983–9989.
66. J. Wu, X. Wang, Q. Wang, Z. Lou, S. Li, Y. Zhu, L. Qin and H. Wei, Chem. Soc. Rev., 2019, 48, 1004–1076.
67. D. Jiang, D. Ni, Z. T. Rosenkrans, P. Huang, X. Yan and W. Cai, Chem. Soc. Rev., 2019, 48, 3683.
68. G. Goumas, E. N. Vlachothanasi, E. C. Fradelos and D. S. Mouliou, Diagnostics, 2025, 15, 1037.
69. J. Xu, B. Zhang, Y. Zhang, L. Mai, W. Hu, C. J. Chen, J. T. Liu and G. Zhu, Talanta, 2023, 253, 124037.
70. N. Ullah, T. A. Bruce-Tagoe, G. A. Asamoah and M. K. Danquah, Int. J. Mol. Sci., 2024, 25, 5959.
71. F. Arshad, B. J. Yee, K. P. Ting, A. A. S. Janudin, M. N. Adzremeen bin Amir and M. U. Ahmed, Microchem. J., 2024, 207, 111826.
72. G. Jarockyte, V. Karabanovas, R. Rotomskis and A. Mobasheri, Sensors, 2020, 20, 6890.
73. T. Beduk, D. Beduk, M. R. Hasan, E. Guler Celik, J. Kosel, J. Narang, K. N. Salama and S. Timur, Biosensors, 2022, 12, 583.
74. C. Dincer, R. Bruch, A. Kling, P. S. Dittrich and G. A. Urban, Trends Biotechnol., 2017, 35, 728–742.
75. U. Zupančič, J. Rainbow, C. Flynn, J. Aidoo-Brown, P. Estrela and D. Moschou, Studies in Systems, Decision and Control, 2021, vol. 327, pp. 63–93.
76. H. Di, Z. Mi, Y. Sun, X. Liu, X. Liu, A. Li, Y. Jiang, H. Gao, P. Rong and D. Liu, Theranostics, 2020, 10, 9303–9314.
77. B. S. Mohd Salleh, F. Arshad and M. U. Ahmed, Talanta, 2026, 296, 128427.
78. F. Li, L. Guo, Z. Li, J. He and H. Cui, Anal. Chem., 2020, 92, 6827–6831.
79. X. Wang, Z. Xia, E. K. Fodjo, W. Deng and D. Li, J. Mater. Chem. B, 2022, 10, 3023–3031.
80. F. Beygnezhad, K. Dashtian, R. Zare-Dorabei and Z. Zangouei, Sens. Actuators, B, 2025, 423, 136769.
81. J. Chen, T. Lian, S. Liu, J. Zhong, R. Cheng, X. Tang, P. Xu and P. Qiu, J. Colloid Interface Sci., 2024, 667, 450–459.
82. S. Zhang, Y. Chen, Y. Huang, H. Dai and Y. Lin, Biosens. Bioelectron., 2020, 159, 112201.
83. Y. Lin, W. W. Shao, D. Wu, T. Zhang, D. N. Fei, Y. N. Kong, Y. F. Gao, S. C. Zhao and R. L. Liu, Anal. Chim. Acta, 2024, 1306, 342598.
84. M. H. Lee, J. S. Kim and J. L. Sessler, Chem. Soc. Rev., 2014, 44, 4185.
85. W. Yang, C. Weng, X. Li, H. He, J. Fei, W. Xu, X. Yan, W. Zhu, H. Zhang and X. Zhou, Sens. Actuators, B, 2021, 338, 129844.
86. X. Wang, L. Qin, M. Zhou, Z. Lou and H. Wei, Anal. Chem., 2018, 90, 11696–11702.
87. T. H. Bui, B. Thangavel, M. Sharipov, K. Chen and J. H. Shin, Chemosensors, 2023, 11, 468.
88. Y. Wang, W. Long, Y. Wang, Y. Han, F. Zeng, J. Li, B. He, Y. Song and Y. Gao, Chem. Eng. J., 2025, 503, 158289.
89. X. Liu, X. Mei, J. Yang and Y. Li, ACS Appl. Mater. Interfaces, 2022, 14, 6985–6993.
90. L. Miao, L. Jiao, Q. Tang, H. Li, L. Zhang and Q. Wei, Sens. Actuators, B, 2019, 288, 60–64.
91. Y. Wang, R. Zeng, S. Tian, S. Chen, Z. Bi, D. Tang and D. Knopp, Anal. Chem., 2024, 28, 2025.
92. F. Arshad, K. P. Ting, S. N. A. Zakaria, N. F. Mohd-Naim, Y. W. Soon and M. U. Ahmed, Appl. Mater. Today, 2024, 41, 102451.
93. C. P. Kurup, N. F. Mohd-Naim and M. U. Ahmed, Crit. Rev. Biotechnol., 2022, 42, 794–812.
94. M. A. Ansari, D. Verma, M. A. Hamizan, M. Das Mukherjee, N. F. Mohd-Naim and M. U. Ahmed, ACS Synth. Biol., 2025, 14(1), 21–40.
95. L. Gong, B. Chen, Y. Tong, Y. Luo, D. Zhu, J. Chao, L. Wang and S. Su, Sens. Actuators, B, 2024, 398, 134686.
96. J. Zhao, Y. Long, C. He, Y. Leng, Z. Huang, Y. Ma, C. Hou and D. Huo, Chem. Eng. J., 2024, 479, 147513.
97. S. M. Yoo, Y. M. Jeon and S. Y. Heo, Biosensors, 2022, 12(9), 738.
98. X. Lai, G. Zhang, L. Zeng, X. Xiao, J. Peng, P. Guo, W. Zhang and W. Lai, ACS Appl. Mater. Interfaces, 2021, 13, 1413–1423.
99. S. Ahmadi, K. Rahimizadeh, A. Shafiee, N. Rabiee and S. Iravani, Process Biochem., 2023, 131, 154–174.
100. K. Wang, Q. Hong, C. Zhu, Y. Xu, W. Li, Y. Wang, W. Chen, X. Gu, X. Chen, Y. Fang, Y. Shen, S. Liu and Y. Zhang, Nat. Commun., 2024, 15, 5705.
101. J. Shen, J. Chen, Y. Qian, X. Wang, D. Wang, H. Pan and Y. Wang, Adv. Mater., 2024, 36, 2313406.
102. J. Han, Y. Gu, C. Yang, L. Meng, R. Ding, Y. Wang, K. Shi and H. Yao, J. Mater. Chem. B, 2023, 11, 9840–9866.
103. P. Khramtsov, M. Kropaneva, M. Bochkova, V. Timganova, D. Kiselkov, S. Zamorina and M. Rayev, Int. J. Pharm., 2021, 599, 120422.
104. P. Khramtsov, M. Kropaneva, A. Minin, M. Bochkova, V. Timganova, A. Maximov, A. Puzik, S. Zamorina and M. Rayev, Nanomaterials, 2022, 12(10), 1630.
105. L. Panariello, S. Damilos, H. Du Toit, G. Wu, A. N. P. Radhakrishnan, I. P. Parkin and A. Gavriilidis, React. Chem. Eng., 2020, 5, 663–676.
106. S. Haque, B. J. Boyd, M. P. McIntosh, C. W. Pouton, L. M. Kaminskas and M. Whittaker, Curr. Nanosci., 2018, 14, 448–453.
107. A. K. Hauser, R. Mathias, K. W. Anderson and J. Zach Hilt, Mater. Chem. Phys., 2015, 160, 177–186.
108. M. Faria, M. Björnmalm, K. J. Thurecht, S. J. Kent, R. G. Parton, M. Kavallaris, A. P. R. Johnston, J. J. Gooding, S. R. Corrie, B. J. Boyd, P. Thordarson, A. K. Whittaker, M. M. Stevens, C. A. Prestidge, C. J. H. Porter, W. J. Parak, T. P. Davis, E. J. Crampin and F. Caruso, Nat. Nanotechnol., 2018, 13, 777.
109. Y. Sun, Z. Xie, F. Pei, W. Hu, S. Feng, Q. Hao, B. Liu, X. Mu, W. Lei and Z. Tong, Anal. Methods, 2022, 14, 5091–5099.
110. G. Jarockyte, V. Karabanovas, R. Rotomskis and A. Mobasheri, Sensors, 2020, 20, 6890.
111. M. S. Gatto and W. Najahi-Missaoui, Int. J. Mol. Sci., 2023, 24, 14041.
112. W. Abdelwahed, G. Degobert, S. Stainmesse and H. Fessi, Adv. Drug Delivery Rev., 2006, 58, 1688–1713.
113. L. Fan, D. Lou, H. Wu, X. Zhang, Y. Zhu, N. Gu and Y. Zhang, Adv. Mater. Interfaces, 2018, 5, 1801070.
114. K. Fan, H. Wang, J. Xi, Q. Liu, X. Meng, D. Duan, L. Gao and X. Yan, Chem. Commun., 2016, 53, 424–427.
115. X. Cheng, S. Liu and Y. Hu, Biosensors, 2024, 14, 520.
116. K. M. Tyner, N. Zheng, S. Choi, X. Xu, P. Zou, W. Jiang, C. Guo and C. N. Cruz, AAPS J., 2017, 19, 1071–1083.
117. C. N. Cruz, K. M. Tyner, L. Velazquez, K. C. Hyams, A. Jacobs, A. B. Shaw, W. Jiang, R. Lionberger, P. Hinderling, Y. Kong, P. C. Brown, T. Ghosh, C. Strasinger, S. Suarez-Sharp, D. Henry, M. Van Uitert, N. Sadrieh and E. Morefield, AAPS J., 2013, 15, 623–628.
118. US Food and Drug Administration, Drug Products, Including Biological Products, that Contain Nanomaterials: Guidance for Industry, FDA, 2022, https://www.fda.gov/regulatory-information/search-fda-guidance-documents/drug-products-including-biological-products-contain-nanomaterials-guidance-industry, (accessed: 17 November 2025).
119. S. H. Elagamy, R. H. Obaydo, A. Ashkar and H. M. Lotfy, Sens. Actuators Rep., 2025, 10, 100394.
120. A. Kausaite-Minkstimiene, A. Krikstaponyte, N. Stasyuk, G. Gayda and A. Ramanaviciene, Biosensors, 2025, 15(8), 545.
121. S. Sharma, J. Zapatero-Rodríguez, P. Estrela and R. O’Kennedy, Biosensors, 2015, 5, 577–601.
122. B. Hayes, C. Murphy, A. Crawley and R. O’Kennedy, Diagnostics, 2018, 8, 39.
123. H. Singh, M. F. Desimone, S. Pandya, S. Jasani, N. George, M. Adnan, A. Aldarhami, A. S. Bazaid and S. A. Alderhami, Int. J. Nanomed., 2023, 18, 4727.
124. T. T. S. Lew, M. H. Wong, S. Y. Kwak, R. Sinclair, V. B. Koman and M. S. Strano, Small, 2018, 14, 1802086.
125. X. Cheng, S. Liu and Y. Hu, Biosensors, 2024, 14, 520.
126. Y. Gao, Z. Zhu, Z. Chen, M. Guo, Y. Zhang, L. Wang and Z. Zhu, Biomater. Sci., 2024, 12, 2229–2243.
127. W. Gao, J. Zhong, X. Liu, D. Bai and M. Wu, Bone Res., 2025, 13, 85.
128. W. Xuan, X. Li, H. Gao, L. Zhang, J. Hu, L. Sun and H. Kan, Sci. Rep., 2025, 15, 13305.
129. Y. Yu, M. Zhang and K. Fan, Mater. Horiz., 2025, 12, 7779–7813.
130. Y. Wei, J. Wu, Y. Wu, H. Liu, F. Meng, Q. Liu, A. C. Midgley, X. Zhang, T. Qi, H. Kang, R. Chen, D. Kong, J. Zhuang, X. Yan and X. Huang, Adv. Mater., 2022, 34(27), e2201736.
131. J. Zhuang, A. C. Midgley, Y. Wei, Q. Liu, D. Kong and X. Huang, Adv. Mater., 2024, 36, 2210848.
132. Y. Gao, Z. Zhu, Z. Chen, M. Guo, Y. Zhang, L. Wang and Z. Zhu, Biomater. Sci., 2024, 12, 2229–2243.
133. G. Jarockyte, V. Karabanovas, R. Rotomskis and A. Mobasheri, Sensors, 2020, 20, 1–23.
134. M. Elsabahy, K. L. Wooley, A. Hendricksen and K. Oh, Methods, 2019, 158, 81–85.
135. Y. Fan, S. Wang and F. Zhang, Angew. Chem., Int. Ed., 2019, 58, 13208–13219.
136. S. T. Syed, B. S. Gerber and L. K. Sharp, J. Commun. Health, 2013, 38, 976.
137. H. Wei and E. Wang, Chem. Soc. Rev., 2013, 42, 6060–6093.
138. S. S. Arya, S. B. Dias, H. F. Jelinek, L. J. Hadjileontiadis and A. M. Pappa, Biosens. Bioelectron., 2023, 235, 115387.
139. Ö. Kap, V. Kılıç, J. G. Hardy and N. Horzum, Analyst, 2021, 146, 2784–2806.
140. Y. Xu, C. Qian, Y. Yu, S. Yang, F. Shi, L. Xu, X. Gao, Y. Liu, H. Huang, C. Stewart, F. Li and J. Han, Anal. Chem., 2023, 95, 4605–4611.
141. M. X. Liu, H. Zhang, X. W. Zhang, S. Chen, Y. L. Yu and J. H. Wang, Anal. Chem., 2021, 93, 9002–9010.
142. S. Madhurantakam, S. Muthukumar and S. Prasad, ACS Omega, 2022, 7, 12467–12473.
143. S. Rauf, A. A. Lahcen, A. Aljedaibi, T. Beduk, J. Ilton de Oliveira Filho and K. N. Salama, Biosens. Bioelectron., 2021, 180, 113116.
144. T. Beduk, D. Beduk, M. R. Hasan, E. Guler Celik, J. Kosel, J. Narang, K. N. Salama and S. Timur, Biosensors, 2022, 12, 583.
145. K. Bera, K. A. Schalper, D. L. Rimm, V. Velcheti and A. Madabhushi, Nat. Rev. Clin. Oncol., 2019, 16, 703–715.
146. T. Akkaş, M. Reshadsedghi, M. Şen, V. Kılıç and N. Horzum, Adv. Mater., 2025, 37, 2504796.
147. X. Zhong, Y. Qin, C. Liang, Z. Liang, Y. Nong, S. Luo, Y. Guo, Y. Yang, L. Wei, J. Li, M. Zhang, S. Tang, Y. Liang, J. Wu, Y. M. Lam and Z. Su, ACS Sens., 2024, 9, 5167–5178.
148. J. Wang, Z. Huang, X. Yu, B. Zhou, W. Xu, X. Zhou and Z. Chen, ACS Appl. Nano Mater., 2025, 8, 5481–5493.
149. S. M. A. Iqbal, I. Mahgoub, E. Du, M. A. Leavitt and W. Asghar, npj Flexible Electron., 2021, 5, 1–14.
150. A. Ahmed, R. Xi, M. Hou, S. A. Shah and S. Hameed, IEEE Access, 2023, 11, 112891–112928.
151. S. Guo, K. Wu, C. Li, H. Wang, Z. Sun, D. Xi, S. Zhang, W. Ding, M. E. Zaghloul, C. Wang, F. A. Castro, D. Yang and Y. Zhao, Matter, 2021, 4, 969–985.
152. H. Y. Y. Nyein, L. C. Tai, Q. P. Ngo, M. Chao, G. B. Zhang, W. Gao, M. Bariya, J. Bullock, H. Kim, H. M. Fahad and A. Javey, ACS Sens., 2018, 3, 944–952.
153. W. Gao, S. Emaminejad, H. Y. Y. Nyein, S. Challa, K. Chen, A. Peck, H. M. Fahad, H. Ota, H. Shiraki, D. Kiriya, D. H. Lien, G. A. Brooks, R. W. Davis and A. Javey, Nature, 2016, 529, 509–514.
154. S. Naveen Prasad, S. Mahasivam, R. Ramanathan and V. Bansal, Anal. Bioanal. Chem., 2024, 416, 6149.

---