Electroanalytical point-of-care detection of gold standard and emerging cardiac biomarkers for stratification and monitoring in intensive care medicine - a review

Abstract

Determination of specific cardiac biomarkers (CBs) during the diagnosis and management of adverse cardiovascular events such as acute myocardial infarction (AMI) has become commonplace in emergency department (ED), cardiology and many other ward settings. Cardiac troponins (cTnT and cTnI) and natriuretic peptides (BNP and NT-pro-BNP) are the preferred biomarkers in clinical practice for the diagnostic workup of AMI, acute coronary syndrome (ACS) and other types of myocardial ischaemia and heart failure (HF), while the roles and possible clinical applications of several other potential biomarkers continue to be evaluated and are the subject of several comprehensive reviews. The requirement for rapid, repeated testing of a small number of CBs in ED and cardiology patients has led to the development of point-of-care (PoC) technology to circumvent the need for remote and lengthy testing procedures in the hospital pathology laboratories. Electroanalytical sensing platforms have the potential to meet these requirements. This review aims firstly to reflect on the potential benefits of rapid CB testing in critically ill patients, a very distinct cohort of patients with deranged baseline levels of CBs. We summarise their source and clinical relevance and are the first to report the required analytical ranges for such technology to be of value in this patient cohort. Secondly, we review the current electrochemical approaches, including its sub-variants such as photoelectrochemical and electrochemiluminescence, for the determination  of important CBs highlighting the various strategies used, namely the use of micro- and nanomaterials, to maximise the sensitivities and selectivities of such approaches. Finally, we consider the challenges that must be overcome to allow for the commercialisation of this technology and transition into intensive care medicine.

Graphical abstract

Importance of rapid testing for cardiac markers in critically ill patients

Cardiovascular dysfunction is a frequent complication of critical illness. Approximately 30% of patients admitted to the intensive care unit (ICU) have underlying cardiac diseases, and approximately 50% of this group are admitted to the ICU with cardiac problems as the primary cause [1, 2]. This includes conditions such as acute myocardial infarction (AMI), heart failure (HF) and cardiogenic shock. However, cardiac complications can arise in ICU patients who have been admitted due to other critical illnesses such as sepsis [3], severe burns [4] and brain trauma [5]. ICU patients are exposed to high levels of non-cardiac stress, which in turn, increases myocardial oxygen consumption. In some patients, the myocardial oxygen supply may be reduced by hypotension, tachycardia, hypoxemia and anaemia. The heart is one of the most frequent organs to fail in critically ill patients [6, 7], and this can have several profound implications for a patient’s prognosis [8, 9]. As such, accurate assessment and monitoring of cardiac function in the ICU is vital to patient care.

Identifying cardiac dysfunction in critically ill patients, however, can be difficult. Co-morbidities and other confounding factors, along with the non-specificity of clinical symptoms, complicate diagnosis. Furthermore, the broad aetiologies behind cardiac dysfunction, are echoed by a wide range of cardiac pathophysiologies, and yet, prompt, appropriate stratification and treatment is crucial to patient outcome, since the acute nature of the dysfunction in many cases, can result in rapid patient deterioration and ultimately, death. Repeated monitoring of cardiac function over time is also vital and comprises assessment of the initial haemodynamic state, and ongoing evaluation of any change in this state that can indicate patient deterioration. Furthermore, assessment of the heart in response to administrated therapies is important in critically ill patients. Ideally, testing for cardiac dysfunction on the ICU should be performed using techniques that are rapid and can be performed repeatedly, with ease.

Cardiac biomarkers (CBs) are produced as a result of a pathological processes in the cardiovascular system. Several are now well-established and routinely used to aid diagnosis of a cardiac event, particularly within the emergency medicine setting, and to identify the progression of cardiovascular diseases [10,11,12,13,14,15]. Indeed, biomarkers of cardiac injury have been used to aid the diagnosis of AMI for over half a century, with aspartate transaminase (AST) being the first CB to be used in clinical practice [16]. However, the lack of specificity of AST for myocardial injury quickly saw it superseded by more clinically relevant CBs which could be used to identify AMI in an emergency room setting, and technologies for their rapid and accurate detection have been sought.

Although testing for certain CBs in emergency and cardiology settings is relatively commonplace, CB testing on the ICU is performed far less frequently. However, CB testing in critical illness has gained significant interest, with the hopes of providing useful information to supplement that provided by more conventional cardiac assessment methods. Indeed, diagnoses and patient stratification based on more traditional methods such as echocardiography are not always sufficient to inform the most appropriate treatment or management strategy. For example, while an enlarged right ventricle signifies pressure or volume overloading, imaging cannot aid in determining the aetiology. Similarly, ECG has been reported to have a low sensitivity for identification of AMI in the critically ill population. CBs, on the other hand aid the diagnostic process by providing information on the nature of the cardiac damage, for example suggesting myocardial stretch, inflammation or cardiomyocyte necrosis.

CBs are also important for prognostication and risk stratification in critically ill patients, with specific CBs measured early in the patient’s ICU stay repeatedly being shown to predict outcome in specific ICU subgroups. Indeed, Cardiac Troponin I (cTnI) strongly predicts mortality and/or length of hospital stay such as in the case of trauma [17], sepsis [18], pneumonia [19] and COVID-19, [20] although, cTnI should be considered more as a specific marker / “gold standard” for the diagnosis of AMI [21]. Likewise, other CBs such as Cardiac Troponin T (cTnT) [22, 23], B-type natriuretic peptide (BNP) [24], N-terminal (NT)-pro hormone BNP (NT-proBNP) [25, 26], soluble suppression of tumourigenicity 2 (sST2) [27, 28] and Heart-type Fatty Acid-Binding Protein (H-FABP) [29], have all been shown to demonstrate some degree of prognostic value in critically ill subgroups. This could be extremely useful for daily practice on the ICU since early measurement of specific CBs may help clinicians to identify early risk of deterioration and may allow for optimisation of ICU resources. Rapid, cost-effective identification of CBs on the ICU therefore holds great potential, since CB testing in this setting is under-utilised. Although point-of-care (PoC) systems for rapid analysis of Troponins and some natriuretic peptides have been adopted in some settings for ruling-out AMI in patients presenting with acute chest pain, some of these technologies are costly, and all other CBs must be analysed by clinical chemistry analysers or plate-based immunoassay in the hospital pathology department.

When considering the development of technology for the rapid detection of CBs, it is vital that the most appropriate CBs are targeted, and this is particularly relevant when considering ICU patient groups; there are a vast number of CBs encompassing enzymes, hormones, and proteins, each with their own set of key attributes and supporting literature; see Table 1. Some of these are more applicable for patient diagnosis, while others hold greater prognostic significance. It is also extremely important to consider that CBs are generally elevated in the critically ill patient population overall (see Table 1) and so require the derivation of separate reference ranges for this distinct cohort. Technologies adopted in the ICU will therefore need to cover a much broader analytical range, than those used to assess CBs in the emergency department for rule out of AMI/HF. Table 1 summarises the evidence for the clinical utility of CBs in ICU patients and provides an indication of the concentrations of each reported in this patient group; from this critically useful table, we overview the electroanalytical approaches to their determination, providing an up-to-date overview.

Table 1 Cardiac biomarkers, their source, clinical relevance, and analytical ranges

Alternative methods for the detection of cardiac biomarkers

Let us first consider the non-electroanalytical methods for the detection of cardiac biomarkers. Currently there are commercially available analysers for cTn, for example the TnI-Ultra assay (ADVIA Centaur XP immunoanalyzer, Siemens Healthcare Diagnostics) and the cTnT assay (Elecsys TnT-hs, Roche Diagnostics). The cTnI assay can achieve detection in plasma as low as 0.006 ng mL⁻¹ and spanning a range of 0.006–50 ng mL⁻¹, whereas, the cTnT assay has a limit of detection (LOD) of 0.005 ng mL⁻¹ and can detect its presence up to 50 ng mL⁻¹. These lab-based methodologies have improved significantly, with the Roche Troponin T assay able to produce results in a single hour [132]. However, there is still a huge drive for portable, reliable, and low-cost devices. Several commercial PoC benchtop devices are also available including BioMerieux Vidas, Mitsubishi Pathfast, and Radiometer AQT90, but development of more portable, hand-held, low-cost devices is still warranted.

Due to the significance of the topic, a plethora of other sensing methodologies have been reported throughout the literature for the detection of CBs. As such there are numerous reviews tackling many topics and their application to the detection of CBs which we direct the reader towards. These include general CB biosensors [133,134,135,136], lab-on-a-chip devices [137], fluorescence [138], colourimetric [139] nanomaterial-based [140, 141], acoustic-wave [142], potentiometric [143], and optical [144] to name just a few. Additionally, there have been reviews for electrochemical strategies [145,146,147], which often highlight a small number of markers or cover multiple detection methods. Herein, we focus solely on electrochemical-based strategies, giving comprehensive coverage of the published literature for the detection of a wide range of clinically proven and emerging CBs.

Current electrochemical/electroanalytical approaches to detect cardiac biomarkers

Now let us consider that the electrochemical detection of cardiac biomarkers is an area of huge interest, with a plethora of different and interesting approaches reported. As expected within the exciting field of biosensor development, there are numerous works that utilise very similar strategies to achieve their end-goal of the quantification of the target biomarkers. For example, the use of EDC (carbodiimide compounds)/NHS (N-hydroxysuccinimide) coupling to covalently attach a bio-recognition element (e.g. protein/peptide) to the electrode surface, the utilisation of Au–S bonding on gold electrodes or the electrodeposition of AuNPs are widely utilised. It is commonplace for almost identical systems to be reported, varying only the electrochemical method (cyclic voltammetry (CV), differential pulse voltammetry (DPV), electrochemical impedance spectroscopy (EIS) etc.), photoelectrochemical (PEC), electrochemiluminescence (ECL), nanomaterial used (metallic nanoparticles, graphene, carbon nanotubes etc.) or simply the target detected. As such, we have aimed to include a comprehensive survey of the literature for each biomarker discussed in the form of tables for each section, highlighting the electrodes used alongside any modifications, the recognition element used, target biomarker and electrochemical detection technique used along with the key analytical parameters and real sample matrix. We do this whilst highlighting some unique and novel advances in biosensor technology and in cases where the literature is too vast (for example cTnI, cTnT and myoglobin) we will focus predominantly on strategies presented in the last 5 years. The electrochemical biosensors utilise three key recognition elements; these are antibody, aptamer, or molecular imprinted polymers (MIPs) person-made mimics of antibodies. Antibodies are widely used in the pharmaceutical industry but can be expensive, have limited stability and require the use of animals. A new approach is to utilise aptamers, synthetic molecules that can be raised against any kind of target and can bind their target with an affinity similar to, or higher than antibodies. Aptamers are ~ tenfold smaller than antibodies and can be chemically modified at-will in a defined and precise way. They can be easily stored and delivered, an advantage over antibodies, can be reversibly heat-denatured, and have a high batch to batch reproducibility. More recently, MIPs have been developed. These are artificial, highly cross-linked polymeric receptors that are engineered towards the binding of specific target analytes. This binding interaction is facilitated by nanocavities that are disturbed throughout the synthesized polymeric network, reflecting the conformation and chemical functionalities of the imprinted molecule or species. Advantages over conventional antibodies include superior chemical and thermal stability, ability to tailor the MIP to the template, and low-cost [148]. Clearly, all three recognition elements can be used in the development of electrochemical biosensors, but the advantages of aptamers and MIPs over antibodies is clear. Despite this, their advantages are not being fully utilised.

Markers of myocardial injury

Cardiac troponin T (cTnT)

The literature for this marker, alongside cTnI, is vast and we concentrate on the last 5 years only, which are summarised in Table 2. It is clear that immunoassay, aptamer and MIP based technologies are all being explored towards cTnT detection, producing clinically relevant linear ranges and detection limits with validation in predominantly human serum. Radha Shanmugam et al. [149] reported a multi-sensor immunoassay for cTnT and cTnI based on gold electrochemical platforms decorated with zinc oxide nanorods. Figure 1A shows a schematic overview of the sensing platform. In this approach, the multi-sensor is based upon thin film fabrication technology with a few nm ZnO seed layer deposited onto the working electrode via RF-Magnetron sputtering after which acts as nucleation sites for further hexagonal shaped ZnO nanorod growth when subjected to a low temperature hydrothermal bath consisting of a zinc nitrite salt and hexamethylenetetramine dissolved in water. The resultant morphology is the vertically oriented ZnO nanostructures, with their ends functionalised with an amine reactive crosslinker molecule—(dithiobis(succinimidyl propionate)), where the NHS ester group at its terminal end provides an amino-reactive surface that forms amine linkage with primary amine groups in the antibody molecule. The authors utilised electrochemical impedance spectroscopy (EIS) and Mott-Schottky analysis on the same sensor platform to demonstrate multi-configurable modes which allowed, via a “signal off” mechanism, the simultaneous measurement of cTnT and cTnI over the range of 0.1 to 1 × 10⁵ pg mL⁻¹ with a LOD in human serum reported to correspond to 1 pg mL⁻¹ for both cTnI and CTnT [149]. The authors report that ZnO is an attractive nanostructured material due to a high isoelectric point and high catalytic efficiency with the ability to align vertically the ZnO to provide a large surface area and useful attachment sites for the antibodies. The authors extended this to measure simultaneously cTnI, cTnT and BNP, showing the successful determination in human serum over the range of 1 pg mL⁻¹–100 ng mL⁻¹ with a LOD of 1 pg mL¹ [150]. Jiang and co-workers [151] have developed an immunoassay sensor utilising electrochemiluminescence (ECL) via the fabrication of silver nanoparticles functionalized SnO₂ nanoflowers where the latter are in the range of 1–2 µm fabricated via a facile hydrothermal methodology. The SnO₂ nanoflowers were then functionalised with 3-aminopropyltrimethoxysilane (ATPES) by adding this dropwise into a solution containing the nanoflowers. These aminated nanoflowers were then dispersed into a glutaraldehyde solution to obtain aldehyde-terminated SnO₂ nanoflowers. The nanoflowers were next dispersed into an ethanol solution to form a suspension with a silver ammonia solution added to obtain silver nanoparticle modified SnO₂ nanoflowers via the traditional silver mirror reaction. To functionalise the Ag@SnO₂ nanoflowers with the cTnT antibody probe, the former were simply mixed with the latter with an incubation of 12 h. Subsequently, bovine serum albumin (BSA) was added to the same solution to block the unspecificed nonspecific binding sites. The basis of the sensor is a sandwich type immunoassay with the second antibody attached to gold nanoparticles all supported upon a GCE. In the presence of cTnT, the sensor is a “signal on” where the ECL intensity is greater when the cTnT has binded between the two antibodies. The sensing approach requires 10 mM S₂O₈²⁻ in the solution that is measuring the target cTnT. The authors believe that the silver nanoparticles serve as a co-reaction accelerator which is able to react with the co-reactant of S₂O₈²⁻ for facilitating the ECL reaction between the SnO₂ nanoflowers and S₂O₈²⁻ resulting in a stronger ECL signal compared with that of just the SnO₂ nanoflowers in the presence of S₂O₈²⁻ (no silver nanoparticles). The immunoassay exhibited a large linear range from 1 fg mL⁻¹ to 100 pg mL⁻¹ with a LOD of 0.11 fg mL⁻¹ reported. The immunoassay was shown to successfully determine cTnT in spiked human serum with recoveries in the range of 91.36 to 112.7%. Recently, Pourali and co-workers [152] reported a biosensing platform based on a sandwich immunoassay utilising CdS quantum dots (QDs). They developed a one-pot synthesis method for producing monodispersed CdS semiconducting nanocrystals (5 nm diameter), through the facile mixing of pre-cursers into a solvent system of dibenzyl ether and oleylamine. These nanocrystals were further modified with streptavidin and used for the signal enhancement. Detection was achieved through the binding of the CdS-streptavidin to biotinylated secondary antibodies followed by the use of square-wave anodic stripping voltammetry. Using this methodology, where the sensing mechanism is a “signal on”, the authors obtained a dynamic linear range from 5 to 1000 ng L⁻¹ (0.005–1 ng L⁻¹) with a detection limit of 2 ng L⁻¹. The authors tested the effect of avidin, myoglobin and CK-MB, showing that the sensor retained at least 92% of its response. They further validated their results in human serum achieving RSD values of 9.8, 7.5 and 3.6% for three different fabricated immunosensors, additionally measuring recovery values between 95.6 and 105.1%.

Table 2 A summary of the reported literature for the electrochemical detection of the markers for myocardial injury; highlighting the marker(s) targeted, electrode materials and modifications, and the electroanalytical method used alongside the measured linear range, limit of detection and real sample medium

Fig. 1

Reproduced with permission from ref [149]. Copyright Elsevier 2017. B) (i) Preparation process of the Ab₂/Ag@SnO₂ NFs signal probe; (ii) construction of the self-accelerated Ag@SnO₂ NFs-based ECL immunosensor and (iii) proposed ECL mechanism for this system. Reproduced and adapted with permission from ref [151]. Copyright Elsevier 2018

A) Schematic representation of sensing cTnI and cTnT biomarkers in a multiplexed sensor array format, utilising antibodies attached to ZnO nanorods.

Silva and co-workers developed a nano-molecularly imprinted polymer (N-MIP) assembled on reduced graphene oxide modified screen-printed graphite electrodes for sensing cTnT [164]. The biomimetic surface was obtained by first taking screen-printed graphite electrodes, which are then surface modified (via drop casting) with reduced graphene oxide; the authors attribute the use of reduced graphene oxide to improve electron transfer rates. The N-MIP was fabricated by taking the reduced graphene oxide screen-printed electrode and placing it into a solution containing cTnT, pyrrole and carboxylated pyrrole (COOH-3-Py) which is then electropolymerized via cyclic voltammetry. The authors explored a range of monomers in order to reach a maximal electron transfer; they also used organic polymers containing functional groups (carboxyl) in order to obtain more reactive biomimetic sites of the cTnT. The authors found that pyrrole and carboxylated pyrrole (COOH-3-Py) provided the best biomimetic conductive polymer where the carboxylic group in position 3 at the monomeric ring linked to the carboxylic group allowed the promotion of more interactions between reactive sites with cTnT. The authors justified their use of reduced graphene oxide in order to increase the synergy with PPy to increase electron transfer rates and promote greater numbers of biomimetic sites due to the nanostructured electrode surface area [164]. A critical parameter for N-MIPs is determining the dissociation constant, ${K}_{D}$ which can be calculated using a Langmuir isotherm model: ${I}_{CD}=\frac{{I}_{max}}{1+(\frac{{K}_{D}}{S})}$ where ${I}_{CD}$ is the current density, S is the concentration of the target (cTnT), and ${I}_{max}$ is the maximum current density. In terms of the N-MIP towards cTnT, a ${K}_{D}$ of 7.3 x 10⁻¹³ mol L⁻¹ was found compared to the control (N-NIP) of 11.6 x 10⁻¹³ mol L⁻¹ reflecting a high affinity of the biomimetic sites to low cTnT concentrations. The authors noted that the ${K}_{D}$ value of the N-MIP is comparable to that of conventional antibodies that exhibits K_D in the range of 10⁻⁷–10⁻⁹ mol L⁻¹ [164] justifying their experimental development. The N-MIP modified electrodes were found to detect cTnT over the range 0.01 to 0.1 ng mL⁻¹ with a very low LOD (0.006 ng mL⁻¹) found to be possible using DPV. The authors went further and examined the N-MIP modified electrodes in human serum comparing their response with gold-standard ECLIA assays with recoveries found over the range of 97–115%. Phonklam et al. [166] followed a similar approach for the sensing of cTnT using MIPs upon screen-printed carbon electrodes with multi-walled carbon nanotubes modified via electrodeposition with the redox probe polymethylene blue. The authors reported that the use of carbon nanotubes increased the electrode area with a three-fold increase in the peak current/signal compared to the case of a bare electrode surface. The MIP was formed via the electropolymerization of polyaniline with the sensing mechanism based upon the redox probe polymethylene blue, where the binding of the cTnT with the MIPs impedes the electron transfer of the oxidation current providing a “signal off” sensor. The sensor was found to detect cTnT over the range of 0.10–8.0 pg mL⁻¹ with a LOD of 0.040 pg mL⁻¹ using DPV. The MIP sensor exhibited an excellent binding affinity (K_D = 2.8 × 10^–13 mol L⁻¹) comparable to others formed via different fabrication strategies and found that the sensor retained more than 90% of the sensitivity after 6 weeks of storage at room temperature. The authors determined cTnT in spiked human plasma which was found to compare well with an independent electrochemiluminescence method. MIPs clearly are an active range for sensing cTnT and from inspection of Table 2 we can see a range of MIPs [164,165,166,167,168] all evaluated in real samples and providing linearly useful analytical ranges. When considering application of such technologies in the ICU setting, we must be mindful of the analytical ranges needed, since these are typically developed with AMI “rule-out” in the emergency setting in mind; typical concentrations of cTnI in ICU patients may be in the region of 1000–1500 ng L⁻¹ with many technologies being capable of translation into this setting [36].

Cardiac troponin I (cTnI)

From inspection of Table 2, a range of approaches have been reported utilising nanomaterials, such as using acetic acid functionalized graphene quantum dots (fGQDs) for an antibody free approach with a reported linear range of 0.17–3 ng mL⁻¹ and a LOD of 0.02 ng mL⁻¹ [169]. While the mechanism is attributed to hydrogen bonding interactions mediated by the carboxylic group in the fGQDs, the sensor is of limited use, if any, due to the lack of tests on real samples. As can be seen in Table 2, a large majority evaluate their sensor in real samples (human serum) which is a must for the credibility of any sensor. Ma and co-workers have reported an electrochemical immunoassay for the sensitive monitoring of cTnI using a novel controlled release system-based antigen-response [171]. Figure 1B shows a schematic overview of the electrochemical based immunosensor which is based upon Fe₃O₄-NH₂ nanospheres (mean diameter of 150 nm) produced via a one-step solvothermal methodology. The nanospheres are mixed with glutaraldehyde (GLH) to provide functionalisation sites and the cTnI antibody and incubated for 2 h. Following this, bovine serum albumin (BSA) is used to block remaining active sites. Next, the surface is modified with the cTnI antibody and cobalt phthalocyanine nanoparticles (8-10 nm diameter). Aminated polystyrene microspheres (APSM) are then used to cover the mesoporous negative charged Fe₃O₄-Ab by electrostatic adsorption. As cTnI is introduced/analysed, APSM is separated from the Fe₃O₄ nanospheres which also releases the cobalt phthalocyanine nanoparticles. These latter released nanoparticles catalyse the added hydrogen peroxide (see Fig. 1B) and provide the electroanalytical signal via a “signal on” approach. This immunoassay was able to measure cTnI from 1.0 pg mL⁻¹ to 100 ng mL⁻¹ with a LOD of 0.39 pg mL⁻¹ using amperometry. The authors went further and demonstrated their biosensor to measure cTnI in human serum with good recoveries (96.7–98.9%) and validated the proposed bioanalytical approach with ELISA indicating the biosensor to have a high accuracy and potential for clinical uptake. Mi et al. [196] have reported a ratiometric aptamer based sensing approach based upon the ECL signal of doxorubicin (Dox)-luminol or the electrochemical (EC) signal of methylene blue (MB) vs. referable EC signal of Dox. Figure 2 shows a schematic overview of the ratiometric aptamer sensor which utilises Ti₃C₂-MXene nanosheets fabricated by ultrasonic exfoliation resulting in 2 nm thick sheets indicating that they are few or single layer. The MXene nanosheets are then modified with gold nanoparticles and tetrahedral DNA (capture probe) which is combined with Au@Fe₃O₄ nanoparticles modified with Tro4-aptamer. In this approach, when cTnI binds with the aptamer, BFP (DNA sequence) is released, which hybridizes with the capture probe. A ECL luminophore (Dox-luminol complex) prepared by the cross-linking between Dox and luminol is used to amplify the ECL signal. Alternatively, the electrochemical signal of methylene blue can be used as an indicator allowing the sensor to be used as ECL _Dox-luminol/Current _Dox or Current _MB/Current _Dox (see Fig. 2). The approach provides a highly useful calibration signal (stable current signal, see Fig. 2) which increases the accuracy of detection which occurs via a “signal on” approach. The sensor is shown to be able to measure cTnI over the range 0.1 fM to 1 pM (0.00239–23.9 pg mL⁻¹) with a LOD of 0.04 fM (0.97 fg mL⁻¹). The authors demonstrated the sensor to measure cTnI in human serum and validated the measurements with ELISA which provided excellent agreement suggesting the sensor could be routinely used for the clinical measurement of cTnI.

Fig. 2

Copyright Elsevier 2019. B) An illustration of (i) the specific target recognition and BFP release, and (ii) the ratiometric biosensing mechanism for cTnI using an MXENE based sensor. Reproduced and adapted with permission from [196]. Copyright Elsevier 2021

A) Preparation procedure for the sandwich-type electrochemical cTnI immunosensor based on mesoporous Fe₃O₄. Reproduced with permission from [171].

Of note, Yang and co-workers [163] have utilised ECL for the simultaneous measurement of cTnT, cTnI and Myoglobin. Figure 3 shows the aptamer-based system and how the sensor is fabricated. The biosensor is based upon a gold macroelectrode array (2 mm diameter) which is modified with the cTnT, cTnI and Myo ssDNA aptamers and then with cysteamine (CysA). The sensor is then exposed to the analyte targets (cTnT, cTnI and Myoglobin) for 60 min which is then modified with a solution of biotinylaed antibody and the ECL probe, a ruthenium complex-labelled streptavidin (Ru1-SA). The ECL signal is based upon Ru(bpy)₃²⁺-tripropylamine (TPA) undergoing electron transfer at the electrode surface to form an excited, light emitting state; see Fig. 3. Through the use of an Electron Multiplying Charge Coupled Device (EM-CCD), the ECL intensity-potential profiles are obtained providing the analytical signal. The multi- sensor was able to measure cTnT, cTnI and Myoglobin over the following linear ranges: 0.50–4.0 ng mL⁻¹, 0.0010–0.010 ng mL⁻¹, 0.050–1.0 ng mL⁻¹ respectively with low detection limits of 0.30 ng mL⁻¹, 0.79 pg mL⁻¹ and 31 pg mL⁻¹ respectively. Despite the achievement of excellent sensitivities, the dynamic ranges would need to be extended to be of use clinically, since the 99^th centile for cTnI for both men and women is outside of this range and given that concentrations of cTnI in critically ill populations can reach tenfold higher than the upper LOD reported here. The potential applicability of the sensor was shown to be viable in human serum samples with a commercial immunoassay. Again, it is important to consider the concentrations that may be seen in ICU patients to assess whether such technologies could be translated into this setting; it is recommended that a Myoglobin assay have a dynamic range of at least 500 ng mL⁻¹ [56] and so some modifications would need to be made to achieve this. We would remind the reader, that although Myoglobin is generally considered an outdated CB for investigation of AMI/HF in emergency settings, very recent studies have shown its usefulness in sepsis and COVID-19 and have suggested it is superior to troponins in these settings [52,53,54]. Singal and co-workers [174] reported a simple yet elegant approach using a 3-dimensional graphene-multi walled carbon nanotube (G-MWCNT) hybrid prepared using a one-step chemical vapor deposition method with acetylene as a precursor source. The G-MWCNT film was transferred onto a glassy carbon electrode and modified with the cTnI antibody, attached through a molecular bi-linker, 1-pyrene butyric acid N-hydroxysuccinimide ester (PyBuNHS). The sensor exhibited a linear range from 1.0 pg mL⁻¹ to 10 ng mL⁻¹ with a LOD of 0.94 pg mL⁻¹ using EIS and was shown to be successful to determine cTnI in human serum.

Fig. 3

Copyright Elsevier 2018

A) (i) Schematic diagram of the ECL biosensor array for the detection of three targets. (ii) Diagram and (iii) photograph of gold electrode array. B) ECL intensity-potential profiles for the determination of different concentration of myoglobin (i), cTnI (ii) and cTnT (iii). In (i): (ng/mL): (a) blank, (b) 0.050, (c) 0.10, (d) 0.25, (e) 0.50, (f) 0.75, (g) 1.0; In (ii) (pg/mL): (a) blank, (b) 1.0, (c) 2.0, (d) 4.0, (e) 6.0, (f) 8.0, (g) 10.0; In (iii) (ng/mL): (a) blank, (b) 0.50, (c) 0.75, (d) 1.0, (e) 2.0, (f) 3.0, (g) 4.0. Insert, calibration curve of Myo, cTnI and cTnT. Measurement conditions: 0.1 M PBS (pH 7.4) containing 50 mM TPA at a scan rate of 50 mV/s. Reproduced and adapted with permission from ref [163].

Last, of particular note is work by Zhang and co-workers [172] who reported a complex electrode configuration but yet provided the basis of a sensor for the ultrasensitive determination of cTnI with a LOD of 33.3 fg mL⁻¹ and a linear range from 1 × 10⁻⁴ to 100 ng mL⁻¹. The sensor is a sandwich type sensor based upon the use of nanoparticles labelled with antibodies with the sensing mechanism occurring via a “signal on” approach. A glassy carbon electrode, gold nanoparticle and thionine decorated amino-functionalized microporous carbon spheres provide the sensor platform, while gold nanotubes decorated with palladium, which have an average size of 35 nm diameter, are modified with β-cyclodextrins functionalized with 3D-dimensional porous graphene. Both nanoparticle composites are modified with antibodies. The sensing mechanism is based upon the increased electrocatalytic reduction of H₂O₂ mediated by thionine, resulting in a sensitive and reliable sensor response. The sensor was shown to successfully measure cTnI in spiked human serum and was compared with ELISA. Recoveries of between 98.0% and 102.4%, RSD values ranging between 3.3% and 4.5% and the relative error (1.7% to 3.8%) between the proposed sensor and ELISA suggest that the fabricated immunosensor has potential for the clinical application for cTnI detection.

Heart-type fatty acid-binding protein (H-FABP)

The first Sandwich Enzyme-Linked Immunosorbent Assay (ELISA) was reported by Ohkaru in 1995 [234] with the first electrochemical-based assay reported in 1996 [235] and many reported over recent years. Table 2 provides a summary of electrochemical based endeavours for the detection of H-FABP using a range of different and diverse nanomaterial sensing based platforms. Feng and co-workers have developed a ratiometric immunosensor for H-FABP; Fig. 4A,B shows a schematic diagram of the fabrication steps of the sensor and how it measures H-FABP [205]. The sensor utilises gold nanodendrites (Fig. 4C) synthesised by a simple methodology with ciprofloxacin hydrochloride coupled by attaching them onto chitosan-grafted-ferrocene prepared via a Schiff-base reaction, which are immobilised upon a GCE which acts as the substrate. The label material is based upon open-pored hollow carbon spheres (OHCSs) which have an of average diameter of 115.0 nm, which are modified with polydopamine (PDA), AuPt nanoparticles and thionine; see Fig. 4A, B. The OHCSs involve taking resorcinol dissolved into water after which PMMA nanospheres prepared via an emulsion polymerisation were added into the solution with formaldehyde and polyethylene glycol 600 (acting as a pore-forming agent) under stirring. The mixture was kept at 85 ℃ in the oven for 3 days, and elevated the temperature up to 800 ℃ for 2 h under a nitrogen flow. The large specific area and high porosity of this nanocomposite provides the efficient adsorption of the thionine electrochemical probe. Figure 4C, D shows a simplified image of the electrochemical sensing platform towards H-FABP and the resulting DPV curves from increasing concentrations of H-FABP. Note that ratiometric immunosensors have two “read outs”, i.e. two analytical signals with which to monitor the output of the sensor where the thionine and the ferrocene are both electroactive. This is a common approach to utilise these electrochemical redox probes in immunosensors and can potentially allow for self-calibration with improved sensitivity and accuracy over single-signal approaches. The sensor measures H-FABP over the range 0.001 to 200.0 ng mL⁻¹ and has a very low LOD of 0.53 pg mL⁻¹. The authors demonstrated the successful determination of H-FABP in human serum with recoveries of 100.1–101.7%, indicating that the sensor holds promise in clinical application [205].

Fig. 4

Copyright Elsevier 2021

A) Schematic diagram for the construction of the ratiometric immunosensor for detecting H-FABP based on Au/Pt nanocrystals and open-pored hollow carbon nanospheres. B) TEM images of the gold nanodendrites. C) Summary of the electrochemical sensing platform and DPV signal acquired at 0.001, 0.01, 0.1, 1.0, 10 and 200 ng mL⁻¹ of H-FABP. Reproduced and adapted with permission from ref [205].

Gan and co-workers [206] reported a highly sensitive electrochemiluminescence sandwich immunosensor for H-FABP determination based on a self-enhanced luminophore coupled with ultrathin 2D nickel metal–organic framework nanosheets. The nanosheets were synthesised via a surfactant-assisted methodology with the wrapping of PEI to produce an amino group to cross-link with luminol via glutaraldehyde which was then modified with H-FABP antibodies via gentle stirring overnight followed by adding BSA to eliminate nonspecific binding sites. The underlying ELC immunosensor is based upon a glassy carbon electrode (drop cast) modified with poly(indole-5-carboxylic acid) to increase the surface area with improved conductivity. The modified electrode was then immersed into a solution of NHS/EDC/MES for 12 h to activate the carboxyl group of the poly(indole-5-carboxylic acid). After this step, H-FABP antibodies were added by dropped solutions containing the antibody onto the electrode surface followed by the addition of BSA. This ECL immunosensor mechanism is based up on a “signal on” approach and was shown to exhibit a very wide detection range from 100 fg mL⁻¹ to 100 ng mL⁻¹ with an ultra-low LOD of 44.5 fg mL⁻¹ and was shown to measure H-FABP in human serum with recover ranges from 98.7 to 102.7% with low % RSDs (4.2–8.4%). The authors ascribed the highly sensitive nature of the sensor to the following reasons: 1) the Ni-TCPP (Fe) nanosheets exhibiting good catalytic activity toward H₂O₂ decomposition but also acted as ideal nanocarrier for luminophore immobilization; 2) the use of a luminophore with a high stability, which shortens electron transport distance and reduce energy loss, effectively improving both the quantity and availability of luminol; 3) due to excellent conductivity and large surface area, poly(indole-5-carboxylic acid) (PICA) can facilitate electron transfer and significantly increase the immobilization amount of antibodies for further improvements in sensitivity [206]. The authors did comment that the conductivity of the Ni-TCPP (Fe) nanosheets is relatively low compared with noble metal nanomaterials, such that further improvement could be made through their incorporation in future sensors. Very recently, a MXene (Cd_0.5Zn_0.5S/d-Ti₃C₂T_x) composite as signal amplificator and core–shell high-crystalline graphitic carbon nitride@carbon dots as electrochemical sensor platform have been utilised as the basis of an sandwich type immunosensor which operates via a “signal on” approach [207]. The MXene was prepared by subjecting a Ti₃AlC₂ MAX phase to etching in HCL/LiF for 20 h following ultrasonic treatment and centrifugation, delaminated MXene was obtained (d-Ti₃C₂T_x). The d-Ti₃C₂T_x was added to an aqueous solution containing zinc and cadmium acetate salts, thioacetamide and subjected to a hydrothermal treatment at 180 degrees for 20 h. Following centrifugation, Cd_0.5Zn_0.5S/d-Ti₃C₂T_x is collected. This composite is then modified with H-FABP-antibody via magnetic stirring. The supporting electrochemical sensor platform was fabricated via a lengthy process which starts with carbon dots (CDs) being formed via the reaction of citric acid and ethylenediamine being heated at 250 degrees for 6 h and after cooling, impurities were removed via dialysis for 70 h. The CDs are then combined with a Ni foam template and dicyandiamide with crystallisation performed at 75 degrees for 15 h and then treated in a muffle furnace at 600 degrees for 90 min. The nickel foam is removed via treatment with 10 M acid with finally obtaining high-crystalline graphitic carbon nitride@carbon dots. These are then modified onto a glassy carbon electrode, coupled with the H-FABP antibody via drop casting, followed by the application of BSA. The immunoassay exhibited a linear range from 0.01 to 1.00 pg mL⁻¹ with a LOD of 3.30 fg mL⁻¹ [207] using DPV. While the immunosensor was shown to be highly selective in model solutions against 10 competitive proteins, with a single sensor shown to be able to be used over 50 times, no real samples were considered [207].

Last, as noted above, the detection of H-FABP is limited and generally based upon immunoassays, with very limited reports using MIPs [208]. For example, Sanati et al. [208] reported the development of a MIP based biosensor, based upon ITO modified electrochemically reduced graphene oxide (ERGO). These were modified with highly active surface area core–shelled gold nano/micro-islands (NMIs) via electrodeposition, which allows their size to be tuneable via controlling the electrodeposition process. The MIPs were fabricated via the electropolymerisation of ortho-phenylenediamine using CV in the presence of the target H-FABP. The MIP biosensor mechanism proceeds via a “signal off” and using DPV, exhibited a linear range from 1 fg mL^–1 to 100 ng mL^–1 towards H-FABP with a LOD of 2.29 fg mL^–1 which was attributed to the high surface area of the NMIs and ERGO [208]. The sensor demonstrated two key aspect of MIPs that makes them attractive as the basis of electrochemical sensors, namely, stability and selectivity. In the former, the authors demonstrated the MIP biosensor was stable after 21 days of storage with only an 8.4% decrease in the electrochemical response. In the latter, the authors explored the interference of proteins found in human serum (e.g. albumin, globulin, and fibrinogen) with no effect and also myoglobin (M_W = 17.67 kDa) and troponin T (M_W = 35 kDa) were found to have no determinantal affect upon the sensor [208].

In summary the electrochemical based sensing strategies to the determination of H-FABP are on first sight, rather limited with all based upon immunoassay technology but have the downside of having multiple components which might be hard to implement in a commercial device. That said, the majority have been shown to successfully determine H-FABP in human serum/blood samples and future work should be used to extend the number of samples measured to produce clinically relevant information for uptake as a commercially device. With the exception of [234] and [203], the above mentioned sensors already cover the desired analytical ranges needed for assessment of this specific CB in critically ill patients [43]. The use of MIPs is very limited but are simpler in terms of the number of components needed to make a sensor and provide the successful measurement of H-FABP in human serum and plasma samples. Future work should be directed to developing new MIP based sensors.

Creatine kinase-myocardial band (CK-MB)

Table 2 summarises various approaches to measuring CK-MB and on further examination, all are entirely focused upon immunoassays with none yet to utilise the potential benefits of aptamers and MIPs. The first electrochemical immunoassay was reported by Yuan and co-workers [236] using a platinum foil macroelectrode, anti-human CK-MB, NADH and ferricyanide, which provides the analytical signal. The authors demonstrated the successful determination of CK-MB in human serum and found a high correlation (0.999) with electrophoresis [236]. Moreira et al. [210] utilised gold screen-printed electrodes which are modified with a phosphorylated form of creatine (Pcrea). Figure 5A shows a schematic overview of the fabrication process where the gold SPE is modified with cysteamine and then via coupling Pcrea with N-(3-dimethylaminopropyl)-N'-ethylcarbodiimide hydrochloride (EDC) and N-Hydroxysuccinimide (NHS). As CK-MB binds to the Pcrea, it is monitored through the electrochemical response via SWV which results in a decrease (“signal off”) in the initial electroanalytical signal. The interferents cTnT, BSA, and myoglobin were studied with little effect on the electroanalytical signal and the authors found their sensor was able to measure CK-MB in synthetic urine and serum, [210] but clearly real samples are needed to progress the immunoassay. Li and co-workers [212] extended the work of Moreira et al. [210] using a GCE and demonstrated their immunoassay to work in human serum which compared favourably with an immunohistochemical staining method [212].

Fig. 5

Copyright Elsevier 2014. B) Schematic for the fabrication of the label-free ECL CK-MB immunosensor based on an CNOS/Fe₃O₄/AuNPs/CS modified SPE. Reproduced and adapted with permission from ref [213]. Copyright Royal Society of Chemistry 2019. C) Schematic illustration of the AuPdCu NWNs-based electrochemical sensor for detecting CK-MB. Reproduced and adapted with permission from ref [214]. Copyright Elsevier 2021. D) Schematic overview of the porous PdPtCoNi@Pt-skin nanopolyhedra production and their incorporation into an electrochemical immunoassay for CK-MB. Reproduced and adapted with permission from ref [215]. Copyright Elsevier 2020

A) Schematic illustration for the stepwise preparation of the biosensor for CK-MB based on creatine phosphate. Reproduced and adapted with permission from ref [210].

Adhikari and colleagues developed an ultra-sensitive label-free electrochemiluminescence (ECL) CK-MB immunosensor using a whole range of nanocomposite-modified single-walled carbon nanotube (SWCNT)-screen-printed electrodes (SPE) [213]. Figure 5B shows the schematic representation of the fabrication steps involved. A SWCNT-SPE is modified with a nanocomposite comprising carbon nano-onions (CNOs) that have been modified with Fe₃O₄ and gold nanoparticles (AuNP) and then chitosan (CS). All nanomaterials were commercially purchased with fabrication involving simple solution mixing, assembled via electrostatic interactions. The CNOs/Fe₃O₄/AuNP/CS composite was then drop casted upon the SWCNT-SPE. Next, the antibody-CK-MB is added onto the surface via drop casting, after leaving to incubate for 12 h, BSA is finally added onto the surface. The authors reported that the highly conductive behaviour of the nanoparticles with an increase in surface area due to the use of Fe₃O₄ and gold nanoparticles and the carbon nano-onions, may contribute to the enhanced ECL intensity. To monitor the binding of the CK-MB, the electrochemical probe [Ru(bpy)₃]²⁺ and tri-n-propylamine (TPrA) were selected as the luminophore and co-reactant respectively, with the electrochemical mechanism described by the following:

$$\begin{array}{*{20}l} {\left[ {{\text{Ru}}\left( {{\text{bpy}}} \right)_{{3}} } \right]^{2+} - {\text{e}}^{-} \to \, \left[ {{\text{Ru}}\left( {{\text{bpy}}} \right)_{{3}} } \right]^{3+} } \hfill \\ {{\text{TPrA }} - {\text{ e}}^{-} \to {\text{TPrA}}^{\cdot +} } \hfill \\ {{\text{TPrA}}^{ \cdot +} \to {{\text{ TPrA}}^{\cdot}} \, + {\text{H}}^{+} } \hfill \\ \begin{array}{*{20}l} \left[ {{\text{Ru}}\left( {{\text{bpy}}} \right)_{{3}} } \right]^{{{3} + }} + {{\text{TPrA}}^{\cdot}} \, \to \, \left[ {{\text{Ru}}\left( {{\text{bpy}}} \right)_{{3}} } \right]^{3+} + {\text{ TPrA fragment}} \hfill \\ \left[ {{\text{Ru}}\left( {{\text{bpy}}} \right)_{{3}} } \right]^{2+\cdot} \, \to \, \left[ {{\text{Ru}}\left( {{\text{bpy}}} \right)_{{3}} } \right]^{{{2} + }} + {\text{h}}\nu\ ( \sim {62}0{\text{ nm}}) \hfill \\ \end{array} \hfill \\ \end{array}$$

The authors demonstrated that the label free ECL based immunoassay could detect CK-MB over the range 10 ng mL⁻¹ to 50 fg mL⁻¹ with a LOD of 5 fg mL⁻¹ and the authors demonstrated the successful recovery of CK-MB from human serum with the recovery range between 98–103%. The use of screen-printed electrodes as the base electrochemical platform give potential for this to be up-taken commercially due to their low-cost, design flexibility and mass-producibility [237]. Although this sensor would be capable of detecting CK-MB values that fall outside of the normal range of 5 ng mL⁻¹ [51], further work would be needed to allow concentrations in the range of 60–100 ng mL⁻¹ in critically ill patients to be accurately reported.

Other reports have utilised nano-networks [214] such as an AuPdCu alloy fabricated by an eco-friendly one pot synthesis. The nano-networks are fabricated via a one-pot aqueous method where 4-aminopyridine along with gold, palladium and copper salts are mixed together with the reducing agent, ascorbic acid being added last, a process taking 3 h. The product was washed with water and centrifuged and dried in a vacuum. Figure 5C shows a schematic overview of the nano-network immunoassay, where the AuPdCu alloy nano-networks, which are of ~ 3 nm diameter, are drop-cast upon a GCE, which is then in turn modified with anti-CK-MB (via drop casting and 12 h incubation) and then BSA. The underlying mechanism is based upon the addition of hydrogen peroxide into the solution, which is electrochemically reduced at the alloy surface. In the presence of CK-MB which binds to the antibody, the electrochemical surface becomes inaccessible and a decrease (“signal-off”) in the electrochemical response is used as the analytical signal; Fig. 5C shows typical chronoamperograms. In using this approach, a linear range of 0.001 to 2000 ng mL^–1 was shown to be possible with a LOD of 0.88 pg mL^–1 reported and was demonstrated to be successful for the analysis of CK-MB in human serum with good recoveries (98.6–101.2%) with a RSD as low as 3.5%. Furthermore, the sensor could be stored for 28 days at 4 °C with only a 6.4% decrease in the electroanalytical signal. The authors attribute the sensor giving rise to the useful analytical performance to be due to the nano-networks providing a stable and large surface area and excellent biocompatibility for effectively capturing CK-MB [214].

Last, Wang and co-workers [215] have reported a sandwich type immunoassay for CK-MB detection which exhibited a linear range from 0.001 to 2500 ng mL^–1 with a LOD reported to be possible 0.62 pg mL^–1 using DPV. The sensor fabrication is shown within Fig. 5D where porous PdPtCoNi@Pt-skin nanopolyhedra particles (67.5–92.5 nm) are fabricated via one-pot aqueous approach and subsequent oxidation etching with nitric acid. These nanoparticles are then modified with thionine by dissolving the PdPtCoNi@Pt-skin nanopolyhedra particles into water under ultrasonication followed by the addition of thionine (Thi), stirred overnight. The Thi/PdPtCoNi@Pt-skin nanopolyhedra particles were then added into a phosphate buffer solution containing the CK-MB antibody, after which BSA was dropped into the same solution. The composite was obtained via centrifugation and washed. The electrochemical platform was produced by taking a GCE and drop casting gold nano stars, which are fabricated by dispersing thymine into water via ultrasonication until dissolved, accompanied by adding sodium hydroxide to adjust the pH to 11. Immediately after, a gold salt is added with the reducing agent ascorbic acid added dropwise with the reaction complete within 16 h. Onto this surface the CK-MB antibody was drop cast, modified with BSA and ready to use. In this sensor the thionine provides the electroanalytical signal via a “signal on” mechanism with which to indirectly measure the CK-MB. Figure 5D shows a typical DPV signal which demonstrates that the use of a chemical etching, with nitric acid gives rise to more porous PdPtCoNi@Pt-skin nanoparticles. The immunoassay was shown to determine CK-MB in human serum with good recoveries (99.2–101.0%) [215].

Myoglobin

Since the replacement of myoglobin with cTn’s as the biomarker of choice for the identification of MI, the number of reports on the development of new platforms for its detection have decreased. However, reports will continue to appear in the literature due to the far lower cost of purchasing myoglobin as a commercial analyte, making it the most accessible of the MI markers. The lower cost of myoglobin makes it especially attractive to the development of MIP based sensing platforms due to the large amount of variables needed to optimise (polymer composition, polymerisation methodology, ratios of target to monomer, template removal), in addition to the electrochemical parameters. Ribeiro and co-workers show these steps through the development of a myoglobin MIP based sensor using phenol as the MIP [223]. In this approach the MIP is made via the electrochemical polymerisation of phenol on a gold screen-printed electrode in the presence of Myo as the templating molecule using CV. The authors present the optimisation of the electrochemical parameters, template concentration, imprinting process, template extraction and analytical parameters. They performed their electrochemical oxidation of phenol at a neutral pH in order to facilitate the addition of proteins to the solution, settling on electropolymerisation of 10 mM phenol in the presence of 5 mg mL⁻¹ myoglobin forming a polyphenol MIP with a thickness of ~ 4.4 nm similar to that of the Myo protein diameter. This high quantity of myoglobin required gives an indication of why researchers are hesitant to use a similar methodology with more expensive targets. Using this sensor, the authors achieved a dynamic range of 0.001 ng mL⁻¹ to 100 µg mL⁻¹, with a LOD of 2.1 pg mL⁻¹ in buffer and 14 pg mL⁻¹ in diluted artificial serum respectively using DPV. Recently, Farahani and co-workers have reported an ultra-sensitive electrochemical sensor for myoglobin based on aptamer recognition and methylene blue loaded co-polymers for signal amplification [228]. They explored the use of two types of poly(styrene)-block-poly(acrylic acid) amphiphilic co-polymers, both synthesised through reversible addition-fragmentation chain transfer polymerisation (RAFT), investigating their self-assembly into polymeric vesicles, as well as loading and release efficiency of the electroactive probe methylene blue. It was observed that the PS₆₁-b-PAA₅₉₆ provided greater loading and release capabilities for methylene blue. The biosensing platform worked through the immobilisation of an aptamer onto a gold surface, followed by incubation with myoglobin and further incubation with a secondary aptamer. The loaded polymersomes were then attached to the ends of the secondary aptamer through EDC/NHS coupling, followed by the addition of DMF to initiate the release of the methylene blue. The presence of methylene blue was detected using DPV in conjunction with a MWCNT modified GCE to produce a linear range of 1 aM to µM and a LOD of 0.73 aM; the sensing mechanism occurs via a “signal on”. They proceeded to show no interference from the presence of haemoglobin and acceptable recoveries in human plasma. It would be advantageous for this work to explore the effect of other common interferents toward myoglobin, in addition to validation of their real sample work through the use of commercial ELISAs.

An alternative detection methodology was presented by Ma and co-workers [233], who used photoelectrochemistry for the detection of myoglobin. They cast Mn-doped CdS nanocrystal-sensitized 2D heterostructured g-C₃N₄-MoS₂ onto an ITO electrode to serve as the photoactive matrix. This heterostructured g-C₃N₄-MoS₂ effectively promoted electron transfer and resisted the recombination of electron–hole pairs, producing a high photocurrent response, with the Mn-doped CdS further increasing the obtained photocurrent. This surface was modified with myoglobin specific antibodies to form the capture part of a sandwich assay. They used this in conjunction with anti-myoglobin labelled CuO conjugates, which effectively quenched the photoelectrochemical response of the system through competition for the light-generated electrons, poor conductivity and steric hindrance. Using this methodology, they were able to detect myoglobin in the range of 1 pg mL⁻¹ to 50 ng mL⁻¹, with a limit of detection of 0.42 pg mL⁻¹. They exhibited that this system had a high specificity and sensitivity in human serum samples, achieving RSD% of 6.1% and below.

Neuroendocrine markers and indicators of myocardial stretch

Brain natriuretic peptide (BNP)

From inspection of Table 3, there are a limited amount of BNP sensors with the majority, if not all utilising immunoassay technology, and with very few using aptamer approaches. The approaches utilise a range of electrode compositions from acetylcholinesterase(AChE)-labelled anti-BNP gold nanoparticles, through to antibody labelled zinc oxide nanorods. Landim and co-workers [238] developed an immunosensor utilising screen-printed carbon electrodes (SPCE) which supported carboxylic acid functionalized multi-walled carbon nanotubes, modified with cobalt phthalocyanine (CoPc); see Fig. 6A. The electrode is then modified by drop casting with ethylenediamine (EDA), anti-BNP and glycine and left to react for 2 h. The immunosensor is based on the cobalt redox couple, which is the basis of the electroanalytical signal, with additions of BNP binding with anti-BNP, the signal decreases (“signal off”) due to the insulating nature of BNP antigen, blocking the kinetics of the interfacial electron transfer and preventing the electrochemical reduction of the CoPc, resulting in the decrease in the current [238]. The immunosensor was shown to measure BNP using LSV from 10 to 1000 ng L⁻¹ with a LOD of 3 ng L⁻¹, which is lower than conventional ELISA immunoassay for BNP quantification (14 ng L⁻¹). The authors demonstrated their sensor to measure BNP in human serum with good recoveries (96–106%). Serafín and co-workers [239] reported an immunosensor based on the immobilization of capture antibodies onto gold nanoparticles (24 nm diameter, prepared from sodium citrate and gold salt) grafted on SPCEs through aryl diazonium salt chemistry, using 4-aminothiophenol (AuNPs-S-Phe-SPCE); Fig. 6B shows an overview of how the sensor is fabricated. Initially the electrodes are modified with 4-aminothiophenol via electrochemical grafting (via CV) onto which gold nanoparticles were immobilised, after which are then modified with the antibody via drop-casting and incubation for 30 min. Last, HRP-anti-BNP is immobilised onto the electrode surface and the sensor is ready. The sensor, was shown via amperometry to detect BNP over the range 14 to 15,000 ng L⁻¹ with a LOD of 4 ng L⁻¹ and was shown to successfully detect BNP in human serum and found to be in excellent agreement with ELISAs.

Table 3 Summary of the reported literature for the electrochemical detection of the markers for myocardial stretch, neurohumoral markers and markers of extracellular matrix remodelling; highlighting the marker(s) targeted, electrode materials and modifications, and the electroanalytical method used alongside the measured linear range, limit of detection and real sample medium

Fig. 6

Copyright Wiley 2021. B) Schematic representation of the steps involved in the preparation of AuNPs-S-Phe-SPCEs (i) and HRP-anti-BNP-BNP-anti-BNP-AuNPs-S-Phe-SPCE immunosensor for the determination of BRP (ii). Reproduced and adapted with permission from ref [239]. Copyright Elsevier 2018

A) Schematic representation of the preparation and sensing of the CoPC@CNT based electrochemical immunoassay for BNP. Reproduced and adapted with permission from ref [238].

An example of a sensor platform for BNP with an even lower LOD has been recently reported by Hu and co-workers [273]. They developed a novel enhanced photoelectrochemical platform based on the successive deposition of N-doped ZnO nanopolyhedra (N-ZnO NP) and protoporphyrin IX. The N-ZnO NP provided a low band gap of 2.6 eV and was utilised as the substrate to enhance the observed photocurrents. The sensing platform was produced through casting of protoporphyrin IX followed by the N-ZnO NP. The authors then cast a DNA aptamer onto the surface of the photoelectrochemical platform to produce an ultra-sensitive, label-free “signal-off” sensor. This exhibited a wide linear range from 1 pg mL⁻¹ to 100 ng mL⁻¹, with an LOD of 0.14 pg mL⁻¹ and validation in human serum samples. In the majority of cases, the reported sensors provide very low detection levels. A thorough summary of commercial testing kits has recently been published for BNP [285] and inspection of Table 3 reveals that the detection levels are lower than commercial kits, indicating that they hold promise to be used in clinical settings. However, in some cases, the assay time can be longer than commercial kits, which is an area of future research focus.

N-terminal-pro hormone BNP (NT-proBNP)

The majority of sensors that have been developed to measure NT-proBNP are based on immunoassay with a limited amount using aptamer technology and none utilising the potential benefits of MIPs (see Table 3). Zhuo and co-workers [241] have reported an electrochemical sandwich immunosensor utilising a nano-structural gold and carbon nanotubes composite which provide immobilization sites for antibodies with gold nanochains and horseradish peroxidase (HRP) complex labelled secondary antibodies for signal amplification operating via a “signal on” mechanism; the sensor fabrication is shown in Fig. 7A. The signal amplification is based upon gold nanochains prepared by reducing a gold salt with ascorbic acid. The antibody was conjugated with the gold nanochains by simple stirring for 12 h, followed by centrifugation to remove excess reagents. HRP was then added to block the unmodified portion of the Au nanochains surface. The electrochemical platform comprises a gold macroelectrode onto which carbon nanotubes, synthesised by a chemical vapour deposition method, are immobilised. The nanotubes were acid treated to introduce carboxylic groups and then added into a solution containing a gold salt and electrochemically reduced to produce nanogold modified carbon nanotubes. Next, the addition of the antibody and BSA finishes the electrochemical platform. The immunosensor was shown to measure from 0.02 to 100 ng mL⁻¹ with a LOD of 6 pg mL⁻¹ using CV which would cover the ranges reported in the critically ill [63]. The sensor’s composition was reported by the authors to give rise to the sensitive sensing of NT-proBNP due to the carbon nanotubes promoting electron transfer and increasing the current response to hydrogen peroxide while the gold nanochains have more active sites than gold nanoparticles; thus, gold nanochains can immobilize more HRP and the current response to H₂O₂ is larger than that of the immunosensor using Au nanoparticles. No interference from cTnI, cTnT and cTnC were found, with the sensor found to be stable for up to 30 days. Esteban-Fernández de Ávila [243] have reported a novel amperometric magnetoimmunosensor using an indirect competitive format developed for the sensitive detection of NT-proBNP. Figure 7B shows a schematic representation of the sensor’s construction which involves the covalent immobilization of the antigen onto carboxyl-modified magnetic beads (HOOC-MBs) activated with EDC and sulfo-NHS, with further incubation in a solution containing variable concentrations of the antigen and a fixed concentration of an HRP-labelled detection antibody. Target NT-proBNP compete for binding with the specific HRP-labelled secondary antibody and the immunoconjugate-bearing MBs are captured by a magnet placed under the surface of a disposable gold SPE. Using amperometry, the analytical signal is measured using TMB (3,3′,5,5′-tetramethylbenzidine), an electrochemical mediator. The immunosensor measures NT-proBNP over the range 0.12 to 42.9 ng mL⁻¹ with a LOD of 0.02 ng mL⁻¹ and was shown to successfully measure NT-proBNP in human serum samples [243]. Such a dynamic range means the technology would prove useful for NT-proBNP sensing in the critically ill [63]. This study was extended to measure both NT-proBNP and CRP in human serum samples [286].

Fig. 7

Copyright Elsevier 2011. B) Schematic showing the fabrication and detection methodology for the magnetoimmunosensor for the detection of NT-proBNP. Reproduced and adapted with permission from ref [243]. Copyright Elsevier 2013. C) Schematic description for the label-free NT-proBNP immunosensor based on ABEI/GNDs/chitosan/COOK-MWCNTs. Reproduced and adapted with permission from ref [247]. Copyright American Chemical Society 2015. D) Schematic of the production, composition and read-out of the ratiometric electrochemical immunoassay for NT-proBNP based on three dimensional PtCoNi hollow multi-branches/ferrocene-grafted-ionic liquid and Co–N-C nanosheets. Reproduced and adapted with permission from ref [261]. Copyright Elsevier 2021

A) Schematic processes of the immunosensor fabrication based on Ab-HRP-AuNCs for the detection of NT-proBNP. Reproduced and adapted with permission from ref [241].

Zhang et al. [247] have developed an immunosensor based on a ITO electrode modified with carboxylic multi-walled carbon nanotubes, which have themselves been modified with chitosan, gold nanodots and N-(aminobutyl)-N-(ethylisoluminol). This is then modified with glutaraldehyde, immobilised through an amide reaction between the amino group of N-(aminobutyl)-N-(ethylisoluminol) and the aldehyde group of the glutaraldehyde. Last, the NT-proBNP antibody is connected via an amide reaction between the aldehyde group of glutaraldehyde and the amino group of the antibody. Figure 7C overviews the various steps of the sensor fabrication. Using ECL, the sensor was able to measure NT-proBNP over the range 0.01 to 100 pg mL⁻¹ with a low LOD of 3.86 fg mL⁻¹. The authors demonstrated the sensor selectivity by exploring the interferents cTnI, IgG, lysozyme, BSA which had little effect upon the sensors performance. The sensors performance was validated in human plasma samples and directly compared with ELISA giving comparable results, strongly suggesting that the developed sensor could be used for the quantitative analysis of NT-proBNP in real samples of human plasma. Such a dynamic range means the technology would prove useful for NT-proBNP sensing in the critically ill [63].

Chen and co-workers [261] have reported the development of a ratiometric immunosensor, the steps involved in its fabrication are shown in Fig. 7D. A GCE serves as the supporting electrode which is modified with PtCoNi hollow multi-branches nanostructures/ferrocene-grafted-ionic liquid. The PtCoNi hollow multi-branches nanostructures are fabricated via a hydrothermal method via mixing platinum, cobalt and nickel acetate salts with oleylamine and oleic acid under continuous ultrasonication followed but the addition of formaldehyde. The solution is then placed into an autoclave and reacted for 12 h at 190 degrees. Following this, they are washed and etched in acetic acid. The nanostructures are impressive and are composed of five or six hollow branches, with their length and width measured to be ~47 and 26 nm respectively. These PtCoNi hollow multi-branches nanostructures are drop cast onto a GCE after which the ferrocene-grafted-ionic liquid is also drop cast along with the antibody and last, BSA. The other part of the immunoassay is comprised of gold nanostars and Co–N-C nanosheets. The former is fabricated via a one-pot wet chemical reduction methodology where 5-hydroxymethyluracil is dissolved into water, adjusted to pH 10 with a gold salt added along with the reducing agent ascorbic acid, and stirred for 30 min. The final product was centrifuged and washed. The latter is obtained by a simple pyrolysis methodology. The gold nanostars, thionine and the Co–N-C nanosheets are mixed together in water, left overnight to incubate into a composite. The antibody is then added through dispersing this composite in phosphate buffer solution and leaving overnight. The immunosensor using DPV, which is a “signal on” sensor, was demonstrated to measure NT-proBNP over the range 0.001 to 10.0 ng mL^–1 with a LOD of 0.35 pg mL^–1, which again has limited use in the setting of HF diagnosis or prognostication in ICU. The selectivity of the sensor was explored towards possible interferents of cTnI, creatine kinase isoenzymes, neuron-specific enolase, and alpha fetoprotein which only caused very small changes in the peak currents towards the analytical target NT-proBNP (less than 5% RSD) and was able to successfully measure NT-proBNP in human serum. The authors attributed the sensors excellent analytical performance to be due to a combination of factors, enhanced electron transport and increased surface area from utilising 3D hollow PtCoNi multi-branches and improved biocompatibility from using the ferrocene-grafted-ionic liquid [287].

A common theme, as can be seen in Table 3, for the detection of this biomarker is the use of electrochemiluminescence (ECL). Li and co-workers [252] developed an ECL immunoassay based on the energy transfer from Luminol-Au@Fe₃O₄-Cu₃(PO₄)₂ nanomaterials (ECL donor) to Au@CuS-rGO (ECL acceptor). In this approach, the former is immobilised upon a GCE which is then modified with Ab₁ via incubation for 12 h, after which BSA is then immobilised upon rGO-Au@CuS-Ab₂ to form the sandwich type immunoreaction mechanism. While the authors do not provide an exact mechanism, it is thought that it originated from ECL resonance energy transfer (ECL-RET) where the electrode materials promote electron transfer to luminol [252, 288]. The immunosensor was able to measure from 0.5 pg mL⁻¹ to 20 ng mL⁻¹ with a LOD of 0.12 pg mL⁻¹. The authors validated their sensor in human serum and directly compared the results with ELISA which gave excellent agreement. In the above cases, and generally in biosensors, there is usually the incorporation of noble nanoparticles of various geometries and compositions, with generally the reason to increase electron transfer properties and provide binding sites. Indeed, at the nanoscale there are changes in the electronic structure and work has shown that reaction mechanism and kinetics differ at the nanoscale in comparison to the bulk [289]. The question of what type of nanoparticles provides the best electroanalytical sensor is a pertinent one. To this end, Beck et al. [259] explored a sandwich type assay for the detection of NT-proBNP, where a SPCE is modified with the capture Ab₁ label (via drop casting) and silver and gold nanoparticle Ab₂ labelled probes were explored and contrasted. The authors found that in both cases, NT-proBNP could be measured over the range 25 to 1000 ng mL⁻¹ using DPV but found that through the use of silver nanoparticles, due to their greater electrochemical activity, they provide a six-times more sensitive assay [259]. The exploration of the geometry and composition of nanoparticles used in immunoassays should be a key future research direction.

Neurohumoral markers

The detection of neurohumoral markers using electrochemical sensing platforms is sparce, with no examples found for the detection of MR-proADM or MR-proANP. This is an area that should be explored in future research. There have been some reports of electrochemical sensors for the detection of copeptin and that is where our attention turns next.

Copeptin

In terms of electrochemical based sensors, there are very few. Yang et al. [275] reported an ultrasensitive electrochemical immunoassay for copeptin determination using an ITO electrode modified with a RGO-TiO₂ nanocomposite which is then in turn modified with Ab-copeptin/BSA via coupling with EDC/NHS; the TiO₂ is in the form of nanoparticles prepared via a hydrothermal technique. The immunosensor using EIS was explored in model solutions where it was observed that as the concentration of copeptin rises, the capacitance declines, possibly because of variation in the dielectric/blocking traits of the electrolyte–electrode interface caused by the interaction between the antigen and antibody. Analysis of the EIS signal gives rise to a linear range of 10 to 500 ng mL⁻¹ (249 to 12,344 pmol L⁻¹) with a LOD of 0.15 ng mL⁻¹ (37.3 pmol L⁻¹). The interferent of ascorbic acid, uric acid, glucose, zinc, copper, mercury and cadmium salts were studied with no effect upon the electrochemical signal. The immunoassay has potential but clearly needs further work to demonstrate any potential clinical uptake. Han and co-workers [276] developed an ECL immunoassay for the measurement of copeptin based upon luminescence immune-gold nanoassemblies. The ECL immunosensor is based upon an ITO electrode onto which Cu²⁺—cysteine complexes and N-(aminobutyl)-N-(ethylisoluminol) functionalized gold nanoparticles combined with chitosan were drop cast and allowed to dry. Onto this modified ITO surface, gold nanoparticle labelled antibodies are fabricated via electrostatic interaction as well as hydrophobic interactions and weak covalent interactions [276]. This approach requires adding the copeptin antibody to a solution of prepared gold nanoparticles (12 nm diameter) followed by incubation overnight. Following this, BSA is added, and the conjugate centrifuged and cleaned with water. The antibody functionalised gold nanoparticles are simply drop cast onto the electrode surface and allowed to dry and the sensor is then ready to use. The sensor is a “signal off” where in the presence of copeptin, the ECL signal reduces due to the formation of antibody–antigen complexes. The ECL immunoassay exhibited a linear range from 0.02 to 10 pmol L⁻¹ with a LOD of 0.0005 pmol L⁻¹. The selectivity of the immunoassay was explored towards the interferents: Seven polypeptides and proteins including Y–H, 3Y–H, HGGG, MB, HAS, FABP and IgG, all at one order of magnitude higher than that of copeptin which indicated no detrimental effect upon the electrochemical signal. The authors went on to demonstrate the determination of copeptin in human serum with good recoveries (97.4–109.3%). Last, Qin et al. [277] reported the fabrication of a triethanolamine-functionalized Metallic Organic Framework (MOF) upon graphene oxide, both supported on a glassy carbon electrode which was then functionalised with anti-copeptin via modification with glutaraldehyde for 2 h after which the antibody was incubated via drop casting for 12 h. The next step was the addition of BSA and then, the sensor was ready to use via ECL Ru(bpy)₃²⁺ redox probe. The linear range was from 5 pg mL^–1 to 500 ng mL^–1 (1.24 to 12,344 pmol L⁻¹) with a LOD reported to correspond to 360 fg mL^–1 (0.09 pmol L⁻¹). The following interferents were explored upon the ECL signal GOx, human H-FABP, human cTnI, human IgG, l-cysteine, DA, and copeptin which were reported to have no effect; the authors demonstrated the successful determination of copetin in human serum with good recoveries (96–104%). This target has scope for future develop based on aptamer and MIP technology.

Markers of extracellular matrix remodelling

As with the previous section, there is very little published literature on the electrochemical detection of the markers of extracellular matrix remodelling. None were found for GDF-8 or GDF-15, and we suggest this to be a productive area of future research. There were some examples for GAL-3 and sST2 and that is where our attention turns first.

Soluble suppression of tumorigenicity 2 (sST2)

There is limited literature on the electrochemical detection of sST2, see Table 3, with some alternative detection methods explored [290]. Demirbakan and Sezgintürk reported a sST2 immunoassay using disposable graphite paper (GP) electrodes [279]. The authors took inspiration from the battery field where paper electrodes are common and have reported advantages which include: very low-cost, high electrical conductivity and practical immobilization methods. The GP utilised in this work is 0.3 mm thick with a size of 210 mm × 210 mm and was commercially purchased. The GP was modified with C₆₀ via drop-casting before incorporating carboxyl groups through the application of sulfuric acid. Following this they immobilised anti-sST2 with EDC/NHS and blocked the remaining surface with BSA. Using EIS they reported a linear range from 0.1 to 100 fg mL⁻¹, with a LOD of fg mL⁻¹ and show that the sensor platform can be stored for 10 days at 4 ºC whilst only losing 4.48% of the performance. The sensor shows no interference from cysteine, heat shock protein, protein activated kinase 2 or TNF-α and was validated through the detection in human serum samples, achieving recoveries between 100 and 113.46%.

Recently, Torrente-Rodríguez and co-workers described an electrochemical sandwich immunoassay for sST2 using SPEs [278]. They developed magnetic immunoconjugates through EDC/NHS coupling of capture antibodies onto the surface of commercially procured carboxylic acid-modified magnetic beads. Upon the binding of target sST2 and secondary antibody labelled with streptavidin and HRP was introduced to allow for a measurable signal using chronoamperometry which occurs via a “signal on” mechanism. This system required an incubation time of sST2 of 15 min, followed by an incubation time of 30 min for the secondary labelled antibodies. Using this methodology, they achieved a linear range between 76 and 2500 pg mL⁻¹, based on loading of 50 µg mL⁻¹ of capture antibody, and a LOD of 26.7 pg mL⁻¹. The sensor was used to detect the presence of sST2 in 25-times diluted human plasma samples from healthy individuals, which exhibited no matrix effects. The results from this were validated against a commercial ELISA platform, where no statistically significant differences were observed. Although the approaches to sST2 detection described in the literature have achieved excellent sensitivities, efforts now need to focus on refining the technology for optimal detection within physiologically relevant concentrations [108, 111].

Galectin-3 (Gal-3)

There have been a few recent reports on the development of electrochemical biosensors for the detection of Gal-3. The first by Tang and co-workers reported a sandwich type immunosensor that utilised various materials to enhance the sensor performance such as metal–organic frameworks (MOFs), AuNPs and nitrogen-doped graphene nanoribbons (N-GNRs) [283]. The GCE was modified with N-GNRs-Fe-MOFs@AuNPs; firstly, the N-GNRs were produced through mixing nitrogen-doped MWCNTs with H₂SO₄ and H₃PO₄ at 140 ºC followed by the addition of KMnO₄ at 65 ºC. The Fe-MOFs were produced separately [291] and decorated with AuNPs through reduction with NaBH₄. The N-GNRs and Fe-MOFs@AuNPs were then combined through sonication and stirring then drop-cast onto the surface of the GCE, before being modified with the Gal-3 specific antibody and the remaining active surface blocked with BSA. The other half of the sandwich was a AuPt-methylene blue (AuPt-MB) nanocomposite whereby, the methylene blue was allowed to form micelles in a solution of HCl and dodecyltrimethylammonium bromide (DTAB), followed by the addition of HAuCl₄ and H₂PtCl₆ to form the nanoparticles. The secondary antibody was incubated with these nanocomposites for 12 h at 4 ºC before further blocking with BSA to prevent non-specific adsorption. For the detection of Gal-3 a 6 µL sample was incubated onto the modified GCE surface for 1 h at 37 ºC, before incubation with the AuPt-MB-Ab₂ nanocomposite for a further 1 h at 37 ºC. Using DPV they reported a linear relationship for the detection of Gal-3 from 100 fg mL⁻¹ to 50 ng mL⁻¹, achieving a LOD of 33.33 fg mL⁻¹. They attribute the performance of the platform to the synergistic effect of the N-GNRs-Fe-MOFs@AuNP and AuPt-MB, with the former providing good electrical conductivity and a larger electroactive surface area and the latter providing good biocompatibility, high loading of antibodies and good signal amplification. They tested the sensing platform in human serum achieving recoveries between 97.99 and 104.84%, further validating it against a commercial ELISA platform showing the sensor provided satisfying accuracy. Although promising, as the authors note, the production time of the sensor is too long for commercial use. Additionally, a cost analysis of the sensor would be useful as, there is a large number of different materials and the use of a GCE, which could be problematic for the transition to clinical care.

Another sandwich immunoassay utilising methylene blue has been reported by Liu et al. [282]. They created their capture platform through drop-casting carboxyl graphene (CR) onto the surface of a GCE before electrochemically reducing it, conjugating the capture antibody through EDC/NHS coupling and finally, blocking the remaining surface with BSA. The secondary antibodies were attached to mesoporous silica nanoparticles along with AuNPs and MB. They utilise both the MB and AuNPs for the detection through using DPV to monitor the reduction of MB and ASV for the oxidation of the AuNPs. The detection using the Au required the use of aqua regia to give a well-defined anodic Au-stripping peak; however, it is difficult to see how this would be used in a clinical setting. Using the MB detection method, they achieved a linear range from 50 fg mL⁻¹ to 500 ng mL⁻¹ and a LOD of 2 fg mL⁻¹. They showed that the immunosensor production was reproducible over a batch of 30 electrodes, giving a RSD of 6.4%. Additionally, the sensor exhibited minimal interference from the presence of various other proteins and was shown to work in clinical serum samples, with the MB sensing method producing recoveries between 95.8 and 106%.

Piguillem and co-workers described a sandwich assay for Gal-3 based on the use of commercially procured carboxylic acid modified magnetic beads (MBs), allowing for the electrochemical measurements to be performed in buffer solution rather than the more difficult blood samples [281]. Capture antibodies were conjugated to carboxyl modified MBs through EDC/NHS coupling followed by blocking with ethanolamine. These Ab-MBs were then incubated in a solution containing Gal-3, followed by further incubation with a detection Ab and then streptavidin-HRP. After all the incubations, the solutions were washed and stored in phosphate buffer, in which the amperometric measurements could be performed. For the measurements, this solution was dropped onto an SPCE surface and inserted into a solution of 1 mM hydroquinone, followed by the addition of H₂O₂ and amperometric detection at -0.2 V (vs. Ag pseudo-reference). Through this methodology, they achieved a linear detection range between 0.028 to 5 ng mL⁻¹ and a LOD of 8.3 pg mL⁻¹. The production methodology had a RSD of 7.7% and was validated in clinical plasma samples from both health individuals and those that had experienced heart failure against commercial ELISAs. The authors continued to show the multiplex possibilities of this sensing platform, using a dual SPCE, for the detection of both Gal-3 and NT-proBNP in buffered samples.

Finally, Cerqueira et al. [284] have recently reported a MIP based biosensor for the detection of Gal-3. The MIPS were formed onto the surface of a SPCE using CV electropolymerisation of the aminophenol monomer in the presence of Gal-3 protein (5 µg mL⁻¹). The template was removed through incubation of a solution of oxalic acid (0.5 M) overnight, before thoroughly washing and storage in PBS. Detection of Gal-3 was achieved using EIS, with a dynamic range from 0.005 to 50 µg mL⁻¹. They showed that the sensor was capable of detecting Gal-3 in human serum samples, reporting an LOD 10 times lower than in buffer which is not explained. This sensor shows a glimpse of what can be achieved with MIPs for the development of these sensor platforms, and we expect further work to be published in this area.

Inflammatory markers

There are a significant number of reports of the development of electrochemical biosensors for inflammatory biomarkers due to the wide range of uses for them throughout healthcare, and not just as markers of cardiac dysfunction, see Table 4. As such we will focus on the last 5 years for each marker discussed beginning with IL-6.

Table 4 A summary of the reported literature for the electrochemical detection of the inflammatory markers linked to cardiac disease; highlighting the marker(s) targeted, electrode materials and modifications, and the electroanalytical method used alongside the measured linear range, limit of detection and real sample medium

Interleukin-6 (IL-6)

From inspection of Table 4, the majority of proposed electrochemical sensors for IL-6 are immunoassay based with some examples of aptamer and MIP based platforms. Tang and co-workers [292] reported a microfluidic immunoassay for the multiplexed detection of cancer biomarkers, including IL-6. In this work, they produced a 32-sensor array (8 electrodes per biomarker analysed) using gold electrodes modified first with a mercaptopropionic acid (MPA) self-assembled monolayer (SAM), followed by EDC/NHS coupling of the specific primary antibodies. The secondary antibodies and HRP tag were chemically attached to streptavidin-modified magnetic nanoparticles (300 nm) and drawn by a syringe into the fluidic chambers. Detection, operating via a “signal on” approach was achieved through the injection of hydroquinone and hydrogen peroxide into the microfluidic detector channels using individual syringes, with the resulting DV measurements staggered to account for the time delay in the multiplexer. Through the use of the multiplexer, they managed to connect 8 of their microfluidic devices together as one, allowing for the analysis of 256 sensors in a time of 30 min. The authors validated their results with internal controls by having 2 of every 8 sensors incorporate BSA instead of antibodies and then further validation through comparison to ELISAs. Examples such as this providing a large throughput of samples show promise for further development towards ICU settings.

An alternative multi-marker approach is reported by Wei et al. [297] for the simultaneous detection of IL-6, IL-1β and TNF-α. They achieve this through the immobilisation of specific capture antibodies for the three analytes targeted onto the GCE surface through diazonium salt electrodeposition followed by EDC/NHS coupling (Fig. 8A). When the target has bound to the surface an incubation of the secondary antibody solution is performed for 30 min. This solution contained specific antibodies for the three targets, each tagged with a different redox probe; Nile blue (NB, -0.4 V) for IL-6, Methylene blue (MB, -0.2 V) for IL-1β and Ferrocene (Fc, + 0.2 V) for TNF-α. This allowed for a DPV signal to be obtained for each individual biomarker based on the appearance of the oxidation peak corresponding to that specific redox tag (Fig. 8A (left)). Through this methodology, IL-6 was able to be detected in the range of 5 to 150 pg mL⁻¹ with a detection limit of 5 pg mL⁻¹. There was no significant interference observed from the presence of BSA, IgG PSA and CA-125 and the results were further validated in mouse serum. This system doesn’t meet the LOD requirements for identification of IL-6 in healthy individuals but could be used to indicate elevated levels in heart failure. From an ICU standpoint, the availability of such as a multi-senor would not only be useful for stratification in conditions such as HF and AMI, but would also be useful in the monitoring of other critical illnesses such as sepsis and COVID-19, provided its dynamic range was extended to accommodate those concentrations in excess of 50 ng mL⁻¹ seen in septic patients [94].

Fig. 8

Copyright American Chemical Society 2018. B) Schematic illustration of the immunosensor production and working mechanism based on porous carbon composites. Reproduced and adapted with permission from ref [311]. Copyright Elsevier 2021. C) Illustration of the synthesis procedure for NiCoO₂@CeO₂ NBs, the preparation of the electrocatalytic labels and the fabrication of the immunosensor. Reproduced and adapted with permission from reference [304]. Copyright American Chemical Society 2020. D) (i,ii) TEM images of NiCoO₂@CeO₂ NBs; (iii,iv) STEM image and elemental mapping of NiCoO₂@CeO₂ NBs. Reproduced and adapted with permission from ref [304]. Copyright American Chemical Society 2020

A) Schematic of the sensor fabrication and representative DPV for the simultaneous immunosensing of multiple cytokines in serum. Reproduced and adapted with permission from ref [297].

Liu and co-workers [296] have reported a dual-responsive sandwich immunoassay that utilises both electrochemical and ECL for the detection of IL-6. They modified a GCE surface with a composite of TiO₂ anatase mesocages (AMCs) and a carboxy-terminated ionic liquid (IL), followed by the ECL probe Ru(bpy)₃(II) and their IL-6 specific capture antibody. The AMCs and IL are utilised for immobilisation of high loadings of the other components. The mesocages were prepared by taking sodium dodecyl sulfate dissolved in hydrochloric acid solution to which titanium (IV) isopropoxide was added and kept at 80 °C for 48 h under stirring; final products were obtained by centrifugation washed thoroughly with distilled water and dried at 60 °C overnight, and then calcined at 400 °C for 30 min in air to remove the residual organics. The second part of the sandwich assay is comprised of octahedral anatase mesocrystals (OAMs), functionalised with acid phosphatase (ACP), and HRP labelled secondary antibodies. The OAMs were synthesised by taking titanate nanowires dispersed in acetic acid and then transferred into a Teflon-lined stainless steel autoclave at 200 ºC for 48 h. The resulting precipitated was obtained by centrifugation and washed with distilled water and ethanol. The final product was attained by drying the precipitate at 60 ºC for 12 h and calcined at 400 ºC for 30 min to remove the residual organics. The OAM has high crystallinity, photoelectric activity, and a nano-porous structure for immobilisation. Using the electrochemical sensing methodology, they achieved a linear range between 10 fg mL⁻¹ and 90 ng mL⁻¹, with a LOD of 0.32 fg mL⁻¹; whereas using ECL they achieved a linear range between 10 ag mL⁻¹ to 90 ng mL⁻¹ and a LOD of 3.5 ag mL⁻¹. This was tested in human serum with good recoveries. This platform showed promise, however a new ECL probe would be required for commercialisation due to possible leaking of the Ru(bpy)₃(II) from solid state biosensors.

A non-sandwich immunoassay based system with an appropriate wide linear range (0–1200 pg mL⁻¹) and low LOD (0.14 pg mL⁻¹) was reported by Liu et al. [311] (Fig. 8B). They utilised a hierarchical nanoporous mesoporous-carbon composite (NMC) decorated with AuNPs on a GCE as the base for their platform. The NMC was formed through SiO₂-nanoparticle assisted sacrificial strategy, achieving an interconnected 3D network with high surface area for deposition of the AuNPs. The anti-IL-6 was then conjugated to the AuNPs through EDC/NHS coupling before blocking the remaining active surface with BSA. DPV was used for the detection of IL-6, with the proposed platform validated in human serum against commercially available ELISA kits producing recoveries from 82.1 to 117%. Cao and co-workers [304] used nanocubes as a high surface area component for loading of their secondary antibodies and as a detection element (Fig. 8C). They used Ni²⁺ exchange, calcination and the CeO₂ growth to produce NiCoO₂@CeO₂ nanoboxes from ZIF-67 (a cobalt-based zeolitic imidazolate framework) (Fig. 8D). These nanoboxes exhibit a catalytic effect toward the oxygen evolution reaction, which changed upon binding of the target analyte. The CeO₂ nanoparticles served a dual purpose of enhancing the catalytic effect and providing sites for facile surface immobilisation of the antibodies through ester-like bridging. The GCE surface itself was modified with carboxylated graphene followed by EDC/NHS coupling of anti-IL-6 and blocking with BSA. For detection, sample was incubated onto the electrode for 50 min followed by the secondary antibody and nanocube system for 50 min. OER testing at + 1.3 V (vs. SCE) was used for detection, measuring the amperometric response, achieving a linear range from 2.5 × 10^–5 to 10 ng mL⁻¹ and a LOD of 7 fg mL⁻¹. The authors showed that this platform performed well (93.8%) for up to 30 days post fabrication and validated their results in human serum samples against a commercial ELISA kit, achieving recoveries between 95.5 and 104%. For this system to be suitable for commercial uptake the two-step incubation times would need to be reduced from the current 1 h 40 min.

Last, Tanak and co-workers have reported a multiplex system for cytokine detection, including IL-6, IL-8, IL-10, TRAIL and IP-10 in undiluted plasma samples in 5 min [309]. They fabricated their sensing platform through RF magnetron sputtering of a semi-conducting ZnO layer (200 nm) onto gold surfaces, followed by antibody immobilisation and blocking using commercial SuperBlock (blocking buffer) used to hydrolyse unbound linker sites to avoid non-specific interaction. The ZnO film was used due to its large band gap (3.367 eV) and high excitation binding energy (60 eV) which both aid in increasing sensitivity. Additionally, the ZnO is non-toxic, has high adsorption, is chemically stable and possesses good electrical conductivity. EIS was used for the detection of specific binding between the antigen and antibody, achieving detection in a wide linear range of 0.01 pg mL⁻¹ to 10 ng mL⁻¹, with an LOD of 0.1 pg mL⁻¹ for IL-6. They validated their results in pooled human blood plasma achieving a clinically accepted standard and an %RSD of ~ 10%, measured across 12 identical sensors. Further validation was obtained through the measurement of 40 patient samples (20 septic, 20 control) achieving a Pearson’s r value ≥ 0.9. This system shows the sort of validation required to provide confidence to professionals working outside of the electrochemical field.

C-reactive protein (CRP)

From inspection of Table 4, there is a wide variety of reported CRP electrochemical sensing platforms producing significantly different operational ranges due to the high concentrations of CRP present, see Table 1. As such, most reports in literature work through identifying the optimal detection ranges for their sensing platform and then diluting the samples for analysis by the appropriate factors. One such example is reported by Vilian and co-workers [323], who utilise a 100-fold dilution in the human serum samples for analysis. They report a simple immunosensor based on the formation of a 3-mercaptopropionic acid (MPA) SAM, followed by EDC/NHS coupling to the CRP antibody (Fig. 9A). The proposed sensor produced a linear range of 5 to 220 fg mL⁻¹ with a low LOD of 2.25 fg mL⁻¹, which is attributed to the gold nanowires grown on a polycarbonate surface. This gold surface was prepared through nanoimprint lithography using a customised electron beam evaporator, producing an Au film of 20 nm thickness (Fig. 9B). Detection of CRP was achieved through SWV of the [Fe(CN)₆]^3−/4− redox couple, which exhibited a reduction in peak current (“signal off”) on increasing amounts of CRP. For real sample analysis, human blood serum was diluted 100-fold in buffer solution and subsequently spiked with varying amounts of CRP, with a LOD of 4.5 fg mL⁻¹ achieved in this medium. Additionally, CRP detection was achieved in human saliva solutions using EIS, through a tenfold dilution in PBS (0.1 M), producing a LOD of 4 fg mL⁻¹.

Fig. 9

Copyright Elsevier 2019. B) Scanning electron microscopy images obtained at various magnifications of the Au/PC substrate used for the gold wire CRP sensor. Reproduced and adapted with permission from ref [323]. Copyright Elsevier 2019. C) Schematic of the preparation of rGO/Ni/PtNPs micromotors and their functionalisation with anti-CRP capture antibodies alongside SEM and EDX analysis (left), fluorescence microscopy images of the micromotors with and without streptavidin (right middle) and time-lapse images of the movement of the micromotors (right bottom). Reproduced and adapted with permission from ref [335]. Copyright American Chemical Society 2020

A) Schematic showing the fabrication of the gold wire sensor for CRP along with the detection strategy. Reproduced and adapted with permission from ref [323].

Molinero-Fernández and co-workers have described micromotor sandwich based immunoassays for the detection of CRP in preterm infant plasma [333, 335]. Micro (or nano) motors convert an external stimulus into autonomous propulsion and when used in a biosensing application this allows them to travel around a sample scavenging for the target analyte. They demonstrated the micromotors can be formed using rGO, MWCNT or carbon black (CB), with the rGO based systems producing the most efficient and reproducible functionalisation [333]. The formation of the micromachines uses a combination of rGO, Ni and PtNPs, shown in Fig. 9C. In this way, the rGO acts as functionalisation points for the CRP antibodies, the Ni layer allows for the magnetic guidance and the PtNPs provide an inner catalytic layer. Using these systems in conjunction with a microfluidic set-up and amperometric detection they were able to detect CRP between 1 and 100 µg mL⁻¹ with a LOD of 0.54 µg mL⁻¹. They used this to analyse CRP levels in preterm infant clinical samples with suspected sepsis, achieving readings using less than 10 µL sample volume in only 8 min [335]. These low sample volumes and quick turn-around time indicate the possibility of translation of this technology into clinical care.

An alternative methodology was presented by Szot-Karpińska et al. [327], who utilised bacteriophages as their recognition element. They immobilised these negatively charged bacteriophages onto a GCE surface in a layer-by-layer fashion with positively charged carbon nanofibers through electrostatic interactions. They compared systems with and without carbon nanofibers, showing that the biosensor using the CRP binding bacteriophage in conjunction with the nanofibers produced the best response, achieving a linear range of 4 to 40 µg mL⁻¹ using three layers of the modification. They propose the use of phages as an artificial alternative to the traditional use of antibodies. We note that there are not many published works for the use of MIPs as another option for the replacement of antibodies with a more stable synthetic receptor. It is suggested this could be an area of research that is explored for CRP as the detection levels do not require the same sensitivities as many other biomarkers highlighted in this review.

Lastly, Cheng and co-workers have explored utilising the synergistic effect of AuNPs and melamine signal amplification through the use of a ferrocene modified small molecular peptide (Fc-ECG) as the bio-recognition element [339]. AuNPs were deposited onto the surface for carbon based SPEs via electrochemical deposition from a aqueous solution containing a gold salt, followed by the dropwise addition of melamine and then formation of Au–S bonds between the NPs and the Fc-ECG. The sensor worked through the free thiol group on the Fc-ECG binding with CRP to form larger complexes and inhibiting electron transfer from the ferrocene tag which produces a “signal off” sensor. Using DPV a linear range of 1 to 550 µg mL⁻¹ was obtained with a LOD or 0.3 µg mL⁻¹, achieving an %RSD of 4% in serum samples with good stability over 5 days post-production. For commercial uptake into clinical settings however a sensor lifetime longer than this would be required to reduce costs and wastage of tests.

Tumor necrosis factor α (TNF-α)

As seen for the inflammatory markers above, on inspection of Table 4, there are a number of significant papers for the detection of TNF-α due to its association with a wide range of conditions. Peng and co-workers [356] have reported a versatile ECL based sensing platform utilising a GCE modified with gold nanoclusters (AuNC) and MnO₂ (Fig. 10A). The AuNC were synthesized by dissolving a gold salt into sodium hydroxide with the reducing agent N-acetyl-L-cysteine. The mixture was incubated at 37 °C for 2.5 h, obtaining a colourless solution. The solution after synthesis was subject to dialysis for more than 24 h to remove all small-molecular impurity. MnO₂ was introduced by modifying a GCE with the AuNC and electrodepositing MnO₂ from immersing the electrode into a KMnO₄ acidic solution and holding the potential at − 0.2 V for 300 s by chronoamperometry. They use an ECL-resonance-energy-transfer (RET) strategy, whereby the AuNC is the ECL donor and the MnO₂ acts as the ECL acceptor. As seen in Fig. 10A, an ELISA based protocol is performed separately using antibodies functionalised with streptavidin and alkaline phosphatase. After the capture followed by enzymatic reaction has occurred, the resultant solution was collected and the modified GCE then incubated for 4 min. This platform achieved a linear response to TNF-α of 0.06 to 31 pg mL⁻¹ with a LOD of 36 fg mL⁻¹, which corresponds to a reduction in two orders of magnitude compared to commercial ELISA kits. They attribute the excellent performance of their sensor to the independence between the ELISA and ECL parts of the system, separating the sensing interface from the complex practical samples, in addition to the dual-signal amplification provided by the ECL technology and enzyme catalysed signal amplification.

Fig. 10

Copyright American Chemical Society 2019. B) Schematic showing the development of an MBs-based immune-platform for the dual amperometric detection of RANKL and TNF at dual SPCEs. Reproduced and adapted with permission from ref [358]. Copyright Elsevier 2020

A) Schematic showing the bases of a conventional ECL-ELISA protocol and the ECL-ELISA protocol proposed by Peng and co-workers. Reproduced and adapted with permission from ref [356].

The biofouling of electrodes is a common problem in the development of biosensing platforms, an alternative method to the separation of protocols mentioned above is the incorporation of anti-biofouling layers on the electrode. Jiang et al. [343] report using mixed layers of phosphorylcholine (PPC) and phenyl butyric acid (PBA) for the development of a TNF-α sensor in whole blood samples on an ITO electrode. In this zwitterionic mixed layer the PPC is responsible for repelling the non-specific protein adsorption that plagues many electrochemical biosensors, whereas the PBA allows for the bioconjugation of antibodies to the electrode surface. The authors then use a sandwich assay (“signal on”) in which the secondary antibody is labelled with HRP for amperometric determination of the presence of the antigen. Using this protocol, they achieved a linear relationship between the current response and TNF-α concentration between 0.01 and 500 ng mL⁻¹ with a lowest detected concentration of 10 pg mL⁻¹. They validated their sensor in whole blood against a commercial ELISA kit showing variations between 2.6 and 11.7%.

Valverde and co-workers reported the dual detection of two emerging biomarkers related to breast cancer, of which TNF-α is one [358]. They utilise an SPCE with two working electrodes, dropping the appropriate solution on each electrode respectively, Fig. 10B. This system utilises a sandwich immunoassay where the capture antibodies are immobilised onto neutravidin-modified magnetic beads, and detection antibodies are labelled with HRP. The binding of the antibodies to the target is achieved separate from the electrode surface in a centrifuge tube. After incubation, the solutions were washed and resuspended in buffer solution before being placed on the working electrodes. Amperometric measurements were used to achieve a dynamic range of 9.9 to 1,000 pg mL⁻¹ for TNF-α, with a LOD of 3 pg mL⁻¹. Their results were validated against a commercial ELISA, showing favourable results. The authors also show that the sensing platform can be stored at 4 ºC for 20 days with no significant differences observed in the sensitivities and there is no significant interference from a wide range of possible competitors. Their results were further validated against commercial ELISAs in human serum samples showing excellent agreement.

Considerations for future research and progression into clinical care

Electrochemical approaches to PoC measurement of CBs in the critical care setting are extremely attractive given their speed, sensitivity, economy of production and ease of multi-panel integration. However, it is important to note that assay cut-offs have not been clearly established in this diverse patient population, and measurements during dynamic critical illness may be problematic. CB interpretation may also vary depending on individual patient characteristics and underlying illness.

From the literature on the design of the electrochemical sensing platforms, it is apparent that emphasis needs to be placed on the analysis procedure and time, production costs and storage life to provide added confidence in the technology. Most reports provide good validation for their work through measurements in real samples (human serum and plasma) and against commercial ELISAs, highlighting the promise of these technologies. The majority of work still utilises antibodies as their recognition elements and further work is expected toward using synthetic or man-made recognition elements to improve the batch-to-batch reproducibility, chemical and thermal stabilities, whilst also reducing the associated ethical concerns.

Additionally, we suggest the increased development of multiplexed sensing platforms to further enhance the possibility of commercial uptake of these technologies. As discussed, there are many markers that, although not specific for cardiac diseases, can be utilised in conjunction with the gold standard markers to provide crucial information to clinicians. In addition to increasing the confidence in the reported results, this will help to increase interest in the technology from external sources leading to increased chances of further funding or commercialisation. Future research should be directed to emerging immuno-thrombotic markers shown to be important in the diagnosis of cardiac diseases such as D-dimer and P-selectin [362,363,364], where electrochemical methods are currently limited [365,366,367,368,369,370].

Conclusions

In this review, we outline the importance of rapid testing for CBs in critically ill patients, explaining the urgent need for developments in rapid, portable and sensitive sensing platforms. We highlight current gold standards used within clinical care in addition to discussing emerging CBs and their potential use in future, data driven patient care. We provide the sources of these CBs, along with their clinical relevance and desired analytical ranges found throughout the literature as reference points for future research on these CBs. We summarise the literature reported on the development of electrochemical sensing platforms for these CBs, focussing on the last 5 years for the most popular CBs, such as cTn’s. Additionally, we explore in detail some of the interesting recent developments for each CB, highlighting how the platforms are produced, function and what key characteristics they possess. Finally, the review offers insights on where we see the field developing and what needs to happen to improve confidence in these platforms and increase the chances of commercialisation and uptake into critical care, particularly with respect to ensuring that the technology focuses on wider dynamic ranges for measurement in this unique cohort of patients.