Comparative review of glycated albumin detection from conventional colorimetric assays to Raman spectroscopy, surface-enhanced Raman scattering, and surface plasmon resonance

Abstract

Albumin is the most abundant protein in plasma, and has a high susceptibility to non-enzymatic glycation. This results in the production of glycated albumin (GA). It indicates short-term glycemic status (2–3 weeks). Unlike hemoglobin A1c (HbA1c), GA is not affected by red blood cell lifespan, making it a reliable glycemia marker in conditions such as hemoglobinopathies, pregnancy, and renal diseases. Traditional colorimetric assays, such as nitro blue tetrazolium (NBT), have low specificity due to interference from other reducing agents. This review compares conventional detection with three emerging optical techniques. Raman spectroscopy (RS) and surface-enhanced Raman spectroscopy (SERS) provide reagent-free detection of GA. RS detects chemical and structural changes in albumin from changes in Raman scattering of molecular vibrations, particularly shifts in amide I and III vibrational bands. SERS magnifies these signals from localized electromagnetic fields at metallic nanoparticle spots. Also, surface plasmon resonance (SPR) detects changes in refractive index in GA after binding to a sensor surface using boronic acid derivatives and kinetic analysis. This comparative review explains how glycation of albumin, both qualitatively and quantitatively, can be measured by conventional versus RS, SERS, and SPR as better diagnostic tools.

This review evaluates the shift from colorimetric NBT to RS, SERS, and SPR for glycated albumin detection. It demonstrates how reagent-free optical techniques offer enhanced specificity and real time monitoring, essential for clinical diagnostics.

1. Introduction

Albumin glycation, the non-enzymatic attachment of reducing sugars (mainly glucose) to the amino groups of amino acid residues in albumin, is an important biochemical process that happens in hyperglycemic conditions, such as diabetes. Detecting GA is reliable in diabetes because it is an intermediate-term glycemic marker, reflecting blood glucose levels over a 2–3-week period, especially in patients who have high red blood cell turnover, where HbA1c may be unreliable.¹ Existing methods for GA detection, such as immunoassays, enzymatic assays, liquid chromatography, and electrophoresis, have their own drawbacks, including high cost, lengthy procedures, reliance on specialized reagents, and interference from similar biological variations.² These limitations highlight the need for rapid, high-resolution alternatives. Although the initial investment for optical instrumentation can be high, RS, SERS, and SPR enable direct detection of molecular signatures/fingerprints without requiring extensive chemicals or costly reagent kits.

RS, SERS, and SPR have emerged as promising alternatives that are cost-effective in per-assay cost. By detecting structural changes in albumin, such as glycation-induced alterations in Amide I and Amide III bands, RS and SERS provide high sensitivity and specificity, minimal sample preparation, and rapid analysis, making them suitable for point-of-care applications.³

2. Human serum albumin

Human Serum Albumin (HSA) is the most abundant and one of the longest known proteins of human plasma. The normal concentration of albumin in human plasma is 35–50 g L⁻¹, and it has a wide variety of physiological functions. HSA represents 50% of the normal individual's plasma protein ⁴.

Human serum albumin (HSA) is a single polypeptide of 585 amino acids with a molecular weight of about 66.5–67 kDa and carries an overall negative charge at physiological pH due to its low isoelectric point (pI 4.7).⁵ Its structure is predominantly α-helical (approx. 67%), organized into three homologous domains (I, II, III), each with subdomains A and B, forming a flexible heart-shaped globular protein stabilized by 17 disulfide bridges between cysteine residues (Fig. 1).⁶ This architecture of albumin makes multiple ligand-binding pockets, enabling the transport of metal ions, fatty acids, bilirubin, and many drugs. Its high molecular weight helps maintain plasma oncotic pressure. In physiological conditions, HSA is mainly monomeric, and 70–80% is non-glycated, but glycation at lysine and arginine residues introduces new glucose-derived compounds, forming advanced glycation end-products (AGEs), which can alter folding, stability, charge, and ligand binding.⁷ Implications for diabetes complications. Glycation-induced structural changes can also impair the function of other lysine/arginine-dependent proteins through steric hindrance and conformational disruption (Fig. 1).⁸

Fig. 1. The crystal structure of human serum albumin (PDB ID: 1AO6) was visualized using PyMOL. The HSA molecule is represented as a cartoon model with the three major structural domains displayed in different colors: Domain I (residues 1–195) is shown in light blue, Domain II (residues 196–383) is shown in green, and Domain III (residues 384–585) is shown in yellow. Major glycation-associated lysine residues, including Lys199, Lys281, Lys439, and Lys525, are highlighted in red as stick and sphere representations. These lysine residues have been identified as prominent sites of glycation adduct formation on human serum albumin through mass spectrometric characterization studies.⁹ The structural representation illustrates the spatial distribution of key lysine residues involved in HSA glycation and provides a molecular view of potential sites contributing to glycated albumin formation.

3. Protein glycation

Protein glycation is a non-enzymatic chemical process in which a sugar molecule, such as glucose, binds to a protein's amino group, typically at lysine or arginine residues, due to their nucleophilic properties. Although d-glucose predominantly exists in its cyclic pyranose forms in plasma, a small fraction remains in the linear open-chain form, and it is this reactive aldehyde group that initiates glycation.¹⁰ This reaction happens in a series of steps, starting from the formation of a reversible Schiff base (an unstable intermediate with a reversible imine – C N bond), which then undergoes rearrangement to a stable Amadori product (ketoamine structure). Over time, these products undergo chemical transformations, forming advanced glycation end-products (AGEs).¹¹

3.1. Glycated albumin formation

GA in healthy people ranges from 1% to 10%, whereas in diabetics, it may increase two- to three-fold. Glycation mainly happens in albumin due to its abundance in plasma and the more active sites in specific amino acid residues.²

The main sites for albumin glycation include lysine residues like Lys-525, which is highly reactive because its positively charged side chains have free amino groups that react with reducing sugars like glucose, forming an early glycation product. It contains a significant portion of the overall glycation. The presence of histidine residues near lysine can further facilitate this process due to their role in the Amadori rearrangement, which stabilizes the early glycation products.¹⁰ The accessibility of these free amino groups for glycation is also maintained by albumin's tertiary structure and the nearness of disulfide bonds.¹⁰

3.2. Molecular structure of glycated albumin

Glycation of HSA paves the way for many physicochemical changes, including alterations in protein conformation, an increase in molecular weight, and small changes in isoelectric point (pI). When sugar molecules bind to lysine and arginine, which have a positively charged amine group in them, they neutralize the charge and make the negatively charged albumin molecule more negative at physiological pH, decreasing pI. These changes are characterized by a transformation from a helical structure to β-sheet and development of nanofibrillar amyloids (Fig. 2).¹²

Fig. 2. The structure represents albumin undergoing non-enzymatic glycation. Initially, glucose reversibly binds to free nucleophilic amino groups of albumin (Lysine), forming a Schiff base (within hours to days). This intermediate undergoes an Amadori rearrangement, stabilizing into a ketoamine (fructoselysine/ketoamine) over 1–2 weeks. As albumin circulates, these modified concentrations increase, reflecting average blood glucose levels over 2–3 weeks. Advanced glycation can further progress over longer periods.¹³.

3.3. Chemical structure of glycated albumin

AGEs form through additional oxidation, dehydration, and cross-linking reactions that make reactive dicarbonyl compounds that affect proteins.

In oxidation, the Amadori product, a stable ketoamine formed from the initial Schiff base, undergoes oxidation. This step generates reactive oxygen species (ROS) that can further modify the protein structure. During dehydration, water molecules are removed from the Amadori product, leading to the formation of reactive dicarbonyl compounds.¹³ These compounds are highly reactive and prone to further chemical interactions. The reactive dicarbonyl compounds can then cross-link with other protein molecules or within the same protein molecule. This cross-linking contributes to the formation of AGEs, which are often structurally complex and can create large, insoluble aggregates. The glycation pattern of HSA depends on the total amount of glycation, with varying degrees of modification observed.¹⁴

3.4. Why is the detection of albumin glycation important?

Diabetes in both younger and adult populations has now become a major global health issue. According to the 10th edition of the IDF Diabetes Atlas, more than 1 in 10 people worldwide are estimated to have diabetes mellitus. The overall crude prevalence of diabetes mellitus in Sri Lanka in 2019, among individuals over the age of 18, was 23%. Hence, the early detection of diabetes mellitus is essential to improve the quality of life by preventing diabetic complications.¹⁵ Glycation of albumin evaluates an individual's glycemic control over two to three weeks. It's beneficial in monitoring glycemic control in conditions such as anemia, hemoglobinopathies, chronic kidney diseases, and pregnancy, in which HbA1c is not applicable, as those conditions affect hemoglobin levels.¹⁶ In addition to that, GA is a more reliable marker for conditions such as end-stage renal disease and patients who undergo dialysis, as albumin turnaround time is less influenced. Furthermore, high levels of GA are associated with diabetic complications such as nephropathy, retinopathy, and cardiovascular diseases. Therefore, detecting elevated glycation helps identify individuals at risk of diabetic complications. The efficacy of antidiabetic drugs can be evaluated within a short period using GA compared to HbA1c, as HbA1c takes around three months to make changes in values.¹⁷

4. Glycated Albumin detection using conventional methods

GA can be measured by enzymatic methods using proteinase and ketamine oxidase, colorimetric methods using thiobarbituric acid, immunoassay, radioimmunoassay, enzyme-linked immunosorbent assay, and ion-exchange high-performance liquid chromatography (HPLC).¹⁸ Nevertheless, these methods are not currently available in the routine laboratory setting.

4.1. NBT method for fructosamine

Fructosamine is a stable ketoamine formed when glucose non-enzymatically glycates plasma proteins, and since albumin is the most abundant and highly nucleophilic protein in plasma, fructosamine primarily reflects glycated albumin levels. A redox reaction between NBT and fructosamine is used to detect fructosamine (Fig. 3). Here, NBT serves as the oxidizing agent and fructosamine as the reducing agent. NBT is a yellow tetrazolium salt. Its chemical structure includes a tetrazole ring, which is essential to the redox process. When NBT is reduced by fructosamine, it forms formazan, a compound with a purple-blue color. This reduction process involves the transfer of electrons from fructosamine to NBT. An alkaline buffer (usually pH = 10) is used for the NBT reaction with fructosamine, which promotes the redox process.

Fig. 3. NBT reduction reaction to formazan. Glycated albumin contains ketoamine groups formed by non-enzymatic glycation of albumin. In an alkaline medium, these ketoamines undergo enolization, producing reactive intermediates that act as electron donors. These intermediates reduce Nitro blue Tetrazolium (NBT), a pale-yellow tetrazolium salt, into formazan, an insoluble purple-blue compound. The intensity of the blue color (formazan), measured spectrophotometrically at ∼530–550 nm, is proportional to the amount of fructosamine present.²⁰.

Step 1: fructosamine undergoes enolization under alkaline conditions, forming reducing intermediates that donate electrons to nitroblue tetrazolium (NBT).

Step 2: NBT accepts electrons and undergoes reduction to form the purple-blue coloured formazan product.

Step 3: fructosamine undergoes oxidation to sugar-derived products as a result of electron donation during the redox reaction. The rate of formazan production is directly related to the fructosamine concentration and is assessed with the spectrophotometric approach at a wavelength of between 530 and 550 nm. NBT-based fructosamine assay is not completely specific to fructosamine, as nitroblue tetrazolium can be reduced by other reducing agents present in biological samples. The assay is performed under alkaline conditions, which promote the enolization of fructosamine and enhance its electron-donating ability, thereby favouring fructosamine-mediated NBT reduction. However, endogenous reductants such as ascorbic acid, uric acid, and other reducing molecules may also contribute to NBT reduction under these conditions, leading to potential interference in fructosamine quantification. NBT assay quantifies the sum of all glycated serum proteins rather than glycated albumin specifically, even though HSA contributes the majority of the reducing ketoamine groups.¹⁹

Glucose also has reducing properties, but at a neutral or physiological pH, glucose could reduce NBT to some extent. However, at higher pH levels, the reactivity of glucose decreases, and the reaction with NBT is less favored. The difference in chemical structure between fructosamine and glucose further increases specificity. Fructosamine contains an amino group (NH₂), which increases its reactivity under alkaline conditions (pH = 10) and accelerates the reduction of NBT. At the same time, glucose, in its open-chain form, reacts more slowly and less efficiently in alkaline conditions.¹⁹ This minimizes glucose's contribution to the final measurement. Also, it's a time-dependent assay, measuring the rate of NBT reduction; since fructosamine reacts faster with NBT, it produces a higher rate of formazan formation compared to glucose. By controlling reaction time, the assay further reduces interference from glucose, which reacts more slowly. However, despite the optimization by alkaline pH and controlled reaction time, the NBT assay still susceptible to interferences such as endogenous reducing substances in serum, like bilirubin, uric acid, and lipids, which non-specifically reduce NBT.²⁰

4.2. Glycated Albumin detection using advanced optical techniques

4.2.1. Raman and SERS spectroscopic characterization of native and glycated albumin

Raman spectroscopy detects inelastic Raman light scattering from molecular vibrations, much weaker than elastic Rayleigh scattering by a factor of a million. High-intensity monochromatic lasers like 532 nm green (visible range) or 785 nm near-infrared diodes illuminate the analyte, with spectrometers capturing Raman shifts, which are the energy difference between incident and scattered light plotted as wavenumber (cm⁻¹) versus intensity to reveal vibrational fingerprints.³

Raman shift is an energy difference between the incident and scattered photons that corresponds to the vibrational energy levels of the molecule. The intensity of Raman peaks depends on the change in molecular polarizability during the vibration.²¹ Stronger bonds with higher force constants, such as C C triple bonds, exhibit higher vibrational frequencies and therefore appear at higher Raman wavenumbers. In contrast, bonds involving heavier atoms generally show lower vibrational frequencies because increased atomic mass increases the reduced mass of the system, resulting in lower wavenumbers. Therefore, the position of Raman peaks is determined by the combined effects of bond strength and atomic masses rather than mass alone.²²

HSA is predominantly α-helical (∼67%), has a sharp Amide I band centered at 1650–1655 cm⁻¹ (C O stretching coupled to N–H bending) and a distinct Amide III feature at 1270 cm⁻¹. Additional diagnostic peaks include the sharp phenylalanine symmetric ring-breathing mode at 1002–1003 cm⁻¹ and the tyrosine Fermi resonance doublet at 828/850 cm⁻¹.³

Non-enzymatic glycation of lysine and arginine residues triggers a protein conformational transition from α-helical structures toward β-sheet and disordered random-coil topologies. The Amide I peak broadens and shifts from 1656 cm⁻¹ (α-helix) toward 1659 cm⁻¹, reflecting the loss of α-helical hydrogen bonding and the formation of intra- and intermolecular β-sheet aggregates.²³ Exogenous vibrational modes from the attached pyranose glucose ring appear as C–O–C and C–C stretching bands at 1040 cm⁻¹ and 1125 cm⁻¹.²⁴

The exact wavenumber position and band resolution depend on the physical state of the sample. In aqueous solution (wet serum/buffer/water), hydrogen bonding with water molecules slightly downshifts the Amide I position compared to dried drop-coating deposition (DCDM), where evaporative coffee-ring concentration alters the local microenvironment but enhances the local signal-to-noise ratio.²⁵

In SERS, when the analyte (glycated albumin molecules) is in proximity to metallic nanoparticle surfaces, localized SPR of gold/silver nanoparticles produces electromagnetic hotspots which can amplify weak Raman signal by factors up to 10⁸, making it detectable even in trace concentrations.²²

In clinical serum samples, human serum albumin is the predominant protein constituent (∼35–50 g L⁻¹, accounting for 50–60% of total plasma protein). However, whole serum also contains globulins, immunoglobulins, lipids, and trace biomolecules that generate overlapping vibrational modes. Glycated albumin is mainly tracked through changes in these protein-related regions. The Amide I and Amide III bands are the main peaks of the spectra of albumin structure; glycation can shift their position and shape because it changes the peptide backbone and local hydrogen bonding.²⁶

Several overlapping bands, such as the CH₂ scissoring band at 1450 cm⁻¹ from lipids and protein side chains, as well as CH₂/CH₃ bending of albumin, overlap with glucose CH bending, making it difficult to isolate glycation-related intensity changes.²⁷ Additionally, lipid C C stretching near 1655 cm⁻¹ overlaps the Amide I band, making it harder to detect small glycation-induced Amide I shifts.²⁸ In addition, strong fluorescence and Raman bands from β-carotene in the 1150–1520 cm⁻¹ region can mask weaker protein or glucose peaks when visible lasers (e.g., 532 nm) are used, so near-IR excitation (e.g., 785 nm) is preferred to reduce background fluorescence.²⁹ The main Raman peaks related to glycated albumin are mentioned in Table 1 and Fig. 4.

Table 1. Assignment of the principal Raman vibrational bands observed in purified human serum albumin (HSA), based on established spectral characterization studies. The assigned bands correspond to the major Raman features illustrated in Fig. 4 (ref. 30).

No.	Vibrational assignment	Band position (cm−1)	Structural origin
1	Phenylalanine ring breathing	1002	Symmetric ring-breathing mode of phenylalanine residues
2	Tyrosine doublet	828–850	Fermi resonance doublet of tyrosine ring, sensitive to local hydrogen bonding and residue exposure
3	CH2 deformation (scissoring)	1447	Bending vibration of aliphatic CH2 side chains
4	Amide I	1655	C O stretching vibration of the polypeptide backbone; diagnostic of α-helical secondary structure

Fig. 4. Normalized and vertically offset Raman spectra of native human serum albumin (HSA) and glycated human serum albumin (GA) acquired from drop-coated samples derived from aqueous solutions. Asterisks indicate principal vibrational peaks, including the tyrosine doublet (828 and 850 cm⁻¹), phenylalanine ring breathing (1002 cm⁻¹), CH₂ deformation (1447 cm⁻¹), and Amide I band (1655 cm⁻¹). (reproduced from Dingari et al. (2012) under the terms of the Creative Commons Attribution (CC-BY) License).³.

Because non-enzymatic glycation occurs at specific amino acid residues such as Lys-525 and arginine sites,¹² glucose attachment does not cause dominant protein vibrational band changes. Instead, glycation induces subtle conformational alterations, such as minor peak broadening, wavenumber shifts in the Amide I, and intensity changes (Fig. 4). So, visual inspection is not enough for accurate differentiation and quantification. It requires multivariate classification models. (e.g., Principal Component Analysis [PCA], Partial Least Squares [PLS]).³

4.2.2. Sample preparation for Raman spectroscopy (RS)

Sample preparation methods for RS include drop coating and direct liquid/wet measurement. The drop-coated method involves depositing a small volume of a sample solution onto a suitable substrate, such as a glass slide, silicon wafer, or metal-coated surface. The drop-coating deposition method (DCDM) is commonly used to prepare biological samples for Raman spectroscopy, as it allows concentration of analytes through the “coffee-ring” effect.³ When a droplet of serum dries on a substrate, capillary flows drive suspended molecules from the center toward the edges of the droplet. This allows deposition of proteins and lipids at the outer ring and deposition of smaller metabolites or loosely bonded molecules in the central area of the ring.³¹

Direct liquid/wet measurement using a 96-micro well plate, cuvette, or cover slip without drying enables to do the measurements directly on wet serum samples without the need for drying, preserving the native biochemical environment of the proteins, lipids, and analytes. In this method, small volumes of serum are loaded into holders, and Raman spectra are collected through the transparent bottom of a holder made of quartz, calcium fluoride (CaF₂), or sapphire (low Raman-active substrates).³²

4.2.3. Surface-enhanced Raman spectroscopy

In SERS, the analyte is placed in proximity to metallic nanoparticle surfaces, and a localized surface plasmon resonance of gold/silver nanoparticles produces electromagnetic hotspots which can amplify weak Raman signal by factors up to 10⁸, making it detectable even in trace concentrations.²² The preparation of SERS substrates involves creating nanostructured surfaces with metal nanoparticles. Effective SERS also requires optimizing the interaction between the molecules and the metal surface to maximize the Raman signal through an enhanced local electromagnetic field.³³

SERS-based glycated albumin detection has advanced in two main substrate strategies: label-free physical adsorption substrates and chemically functionalized capture substrates. In the label-free substrate, thin silver films formed by vacuum e-beam evaporation onto glass slides have been used as SERS substrates, with glycated HSA and HSA solutions simply drop-deposited and air-dried to form a coffee-ring residue from which spectra are measured directly, without chemical modification of the surface or the analyte.³⁴

A second is chemically anchoring the target analyte to the SERS substrate to improve reproducibility and selectivity. This is done by functionalizing the metal surface with 4-mercaptophenylboronic acid, which forms a self-assembled monolayer on silver nanoparticles and covalently binds the cis-diol groups present on the glucose moieties of glycated albumin, selectively capturing GHSA from HSA–GHSA mixtures while excluding unglycated protein.³⁵ This boronic acid–based capture has also been combined with an electric field to further concentrate analyte within the SERS hotspot region, improving signal reproducibility.³⁶ The boronate ester bond formed between the phenylboronic acid and the diol group is reversible under acidic conditions.

4.2.4. Quantitative analysis

Quantitative analysis of glycated albumin (GA) using Raman spectroscopy and SERS has different analytical challenges compared to routine assays. Because non-enzymatic glycation modifies specific amino acid side chains without creating a separate GA peak, simple calibration is not enough for serum analysis.³ Raman intensity can change due to laser power, sample focus, and drying variations during drop-coating deposition.

GA-sensitive features can be normalized against the phenylalanine ring-breathing mode at 1002 cm⁻¹, whose intensity depends on total protein concentration and is largely unaffected by local microenvironment changes.³⁷

Acquire Raman spectra for both glycated and non-GA samples, focusing on the fingerprint region where key peaks related to protein structure are observed.³⁸ Measure the intensity of these peaks and construct a calibration curve by plotting peak intensity against known GA concentrations. Use this curve to quantify GA in unknown samples based on their Raman spectra. Multivariate statistical techniques, such as principal component regression (PCR) or partial least squares regression (PLSR), can use to predict GA levels in unknown samples.³

Beyond simple calibration, multivariate chemometric approaches can get information from multiple spectral features simultaneously, making them useful for quantitative analysis of complex biological samples such as glycated human serum albumin (GHSA).³⁶

Partial least squares (PLS) regression has been used make calibration models for direct Raman measurement of GHSA, while Nechaeva et al. (2020) use PLS to SERS spectra obtained using a boronic-acid-modified substrate for selective detection of GHSA in HSA–GHSA mixtures.^3,34

Slipchenko et al. (2022) combined Lasso for feature selection with Ridge regression for quantification of GHSA in HSA–GHSA mixtures, achieving a root-mean-square error of approximately 1.9% in regression.³⁶

4.3. Glycated Albumin detection using surface plasmon resonance (SPR)

4.3.1. SPR principle

Surface Plasmon Resonance (SPR) is a phenomenon that occurs by resonance of a Surface Plasmon Wave (SPW) which oscillates at a metal–dielectric interface. SPW is simply generated when photons are incident on a metal thin film. When a polarized laser is incident on a metal thin film through a high-index medium, at total internal reflection, an evanescent wave will be generated, causing the free electrons in the metal film to oscillate. The oscillation will generate an SPW along the metal surface.³⁹ At a particular angle, the SPW will oscillate at a maximum amplitude, resulting in SPR (Fig. 5). Generally, silver and gold are used as the metal thin film due to sustainability and higher sensitivity.⁴⁰ However, alternative metals such as copper (Cu), and aluminum (Al) have been used in the field of science due to cost-effectiveness and abundance. The choice of metal significantly affects the plasmonic efficiency and sensitivity of SPR sensors because these characteristics are governed by the dielectric properties of the metal. Gold and silver possess dielectric constants that support strong surface plasmon excitation with relatively low optical losses, resulting in sharper resonance curves and enhanced sensitivity to refractive index changes at the sensor surface. Silver generally exhibits higher plasmonic efficiency due to lower damping losses, whereas gold is often preferred because of its excellent chemical stability and biocompatibility. In contrast, copper and aluminum exhibit greater optical losses and plasmon damping, which can broaden the resonance peak and reduce sensor sensitivity.⁴¹ Therefore, the selection of the metal film represents a trade-off between sensitivity, stability, and cost in SPR-based biosensing applications. The resonance angle directly depends on the refractive index value (dielectric value) of the dielectric layer.⁴² Therefore, simply binding the analyte to the dielectric layer will change the overall dielectric value, leading to a change in the resonance angle. During recent years, the development of SPR-based biomolecular sensors has been highly researched due to its non-destructive and in situ sensing ability. Glycated albumin (GA) is a vital biomarker for intermediate-term glycemic control in diabetes management; GA can be detected using SPR biosensors.³⁹

Fig. 5. Schematic of a surface plasmon resonance (SPR) sensor. A polarized laser is incident through a high-refractive-index prism, which generates an evanescent wave at the metal thin-film interface, exciting free electrons and producing a surface plasmon wave (SPW). At the resonance angle of refracted light, the SPW oscillates at maximum amplitude. Binding of analyte (glycated albumin) via biomolecule (boronic acid derivatives/aptamers) alters the local refractive index, shifting the resonance angle proportionally to analyte concentration.⁴³.

4.3.2. Surface functionalization and recognition strategies

The immobilization of recognition elements onto the sensor surface is a significant factor for the effectiveness of the SPR biosensor. A common method is to employ derivatives of boronic acid, like 4-aminophenylboronic acid (APBA), which will selectively and reversibly interact with the cis-diol groups of the glucose moieties on GA to form cyclic esters with five or six members. This interaction occurs through reversible formation of cyclic boronate esters between the boronic acid group and the cis-diol groups of glucose residues on GA. Under alkaline pH conditions (pH 8.5–9.0),³⁵ boronic acid forms a boronate complex with these diol groups, providing selective recognition of glycated proteins compared with non-glycated albumin, which lacks these glucose-derived structures.³⁹ Using alkanethiols such as 11-mercaptoundecanoic acid (MUA), this chemical modification typically starts with the formation of a self-assembled monolayer (SAM), followed by the activation of carboxyl groups using a coupling chemistry between 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) and N-hydroxysuccinimide (NHS).⁴⁴ As a substitute, it can be observed that aptamer-based biosensors have become prominent because they are less expensive to produce than conventional antibodies and have a high binding affinity and thermal stability.⁴⁵ By adopting particular secondary conformations that “capture” the GA molecule through non-covalent interactions, the single-stranded nucleotide sequences offer exceptional selectivity against non-glycated human serum albumin (HSA). Aptamers enhance selectivity by forming specific three-dimensional structures that recognize GA, reducing interference from abundant serum proteins such as non-glycated HSA and other plasma components. This selective binding capability helps minimize matrix effects, which remain a major challenge in SPR-based detection of GA.⁴⁶

4.3.3. Experimental procedure and signal detection

A functionalized sensor chip, a microfluidic system, and an optical detection system generally constitute an SPR experimental setup.⁴⁷ To create a stable baseline, a running buffer of Phosphate Buffered Saline (PBS) at pH 7.4 is initially applied to the sensor surface. The binding event causes the resonance units (RU) to increase when serum samples or GA standards are injected. In the range of 0% to 25% of Glycated Human Serum Albumin (GHSA) content, which is frequently tested for quantitative analysis to plot a calibration curve, since the above-mentioned range encompasses both typical physiological and pathological values.⁴⁸ While general SPR biosensor protocols rely on standard interference testing against non-target proteins, GA-specific SPR platforms must specifically demonstrate discrimination between glycated albumin and the excess of non-glycated HSA (which lacks cis-diol groups) as well as serum globulins to confirm true recognition selectivity.⁴⁸

4.3.4. Kinetic analysis and surface regeneration

A key benefit of SPR is its ability to conduct kinetic analysis, determining the association rate constant (k_a) and dissociation rate constant (k_d).⁴⁹ The binding interaction is generally modeled using the Langmuir 1 : 1 isotherm to calculate the equilibrium dissociation constant (K_D= k_d/k_a). In the specific context of GA biosensing, these kinetic parameters quantify the affinity of phenylboronic acid monolayers or aptamer recognition elements for glycated HSA relative to unglycated HSA, providing thermodynamic metrics for surface optimization.⁵⁰ After the measurement, the sensor chip requires regeneration to remove the bound GA and restore the baseline for the next sample. For boronic acid–based sensors, the industry utilizes an acidic buffer (e.g., 10 mM glycine-HCl, pH 2.0) or a high concentration of a competing diol like sorbitol to break the boronate–diol bond. These regeneration conditions disrupt the reversible boronate ester interaction and allow the sensor surface to return to its original baseline signal. However, repeated exposure to acidic or highly concentrated regeneration solutions may affect the stability of the immobilized recognition layer, leading to gradual ligand degradation, reduced binding efficiency, and decreased sensor reproducibility over multiple cycle.⁴⁸

5. Comparative analysis

A comparative analysis of conventional and emerging analytical techniques for the detection of glycated albumin is listed in Table 2.

Table 2. Comparison of conventional and emerging analytical techniques for glycated albumin (GA) detection. NBT = nitroblue tetrazolium; RS = Raman spectroscopy; SERS = surface-enhanced Raman spectroscopy; SPR = surface plasmon resonance. Sensitivity and specificity are summarized based on reported literature.

Parameter	NBT (colorimetric) assay	Raman spectroscopy (RS)	Surface-enhanced Raman spectroscopy (SERS)	Surface plasmon resonance (SPR)
Sensor/platform preparation	None (standard optical cuvette/96-well microplate)51	None (standard quartz slide or micro-well plate)3	SERS substrate fabrication (plasmonic Au/Ag nanoparticle colloidal synthesis or vacuum deposition)52	Sensor chip functionalization (gold film coating; SAM formation via EDC/NHS coupling; boronic acid/aptamer)53
Serum sample pre-treatment	Minimal (blood centrifugation to yield serum dilution in alkaline buffer)19	Minimal (blood centrifugation to yield serum; direct drop-coating or dilution in PBS)3	Low-moderate (blood centrifugation and dilution in buffer to reduce matrix interference/fouling)54	Moderate (blood centrifugation, microfiltration and buffer dilution to avoid microfluidic clogging)55
Detection principle	Reduction of NBT to formazan by fructosamine	Inelastic light scattering of molecular vibrations	Plasmon-enhanced Raman scattering	Refractive index changes at a metal–dielectric interface
Specificity for GA	Measures total fructosamine rather than GA specifically	Based on GA-associated spectral changes	Based on enhanced signal than RS	Affinity-based recognition
LOD/sensitivity	20 µmol per L fructosamine51	14.7 µM GA calculated LOD for glycated albumin3	500 nM detection of glycated albumin with the ability to visually detect spectral features52	Reported detection limit of 0.1 µM, within the physiological concentration range of glycated albumin46
Analytical range	0–6 mmol L−1 (linear range)19	7–250 µM GA, physiological range of about 15–30% glycation3	5–25% glycation ratio (total protein 1 g L−1); LOD 500 nM in the silver-coated silicon nanowire study52	Linear response only at the lower end, up to about 1 µM. It saturates at higher concentrations. Human GA is 20–240 µM, so dilution would be needed46
Cost	Widely available, can be automated, and relatively inexpensive51	Completely reagent-free3	Costly SERS substrate and substrate synthesis36	Functionalize sensor chips, microfluidics, and readers46
Clinical validation	Validated in nondiabetic/diabetic patients	Proof of concept only	Not validated	Not validated
Key limitations	Susceptible to interference from other reducing substances	Fluorescence interference	Reproducibility issues due to substrate variability; high fabrication costs	Matrix effects, non-specific adsorption, and sensitivity to temperature fluctuations

6. Limitations of Raman, SERS and SPR

Raman spectroscopy has several limitations. Its weak signal often requires high laser power, which can heat the sample and damage it.⁴⁶ Also, the fluorescence from other highly Raman-active biomolecules can mask Raman peaks and overlap signals in complex biological matrices like serum. This makes detection challenging for low-concentration analytes.⁵⁶

In bioconjugation SERS applications, target molecules such as proteins/antigens are attached to the SERS substrate using antibodies or chemical linkers such as thiol-PEG/biotin–streptavidin. In this process, it's important to ensure these molecules bind at the right angle and position. If the binding site of the antibody is not blocked and the target molecule can be far from the surface (reducing signal).⁵⁷ Gold (Au) substrate is widely used in SERS because it is stable, but creating nanostructures such as nano stars, nanorods, or nanopillars with consistent shapes and the right spacing in between (“hot spots”) is challenging. If the surface isn't uniform, the SERS signal can vary from one spot to another, which makes unreliable quantitative results. Also, fabrication methods such as e-beam lithography and nanoimprinting are often expensive and time-consuming due to their complex instrumentation requirements. Although colloidal synthesis offers a more economical and scalable alternative, controlling nanoparticle uniformity and reproducibility remains challenging.³³

SPR sensors have practical limitations due to the high cost of optical instrumentation and the complexity of specialized sensor chips.⁵⁸ One of the major technical issues is that the non-specific adsorption of abundant proteins like non-glycated albumin and globulins causes signal interference, leading to surface fouling, which is known as the “matrix effect” of human serum.⁵⁹ Additionally, boronic acid–based sensors often require alkaline pH levels to function, which can stray from physiological norms and impact protein stability. Furthermore, the extreme sensitivity to temperature fluctuations and “bulk effects” (refractive index changes in the buffer). Finally, repeatability is limited by the degradation of the immobilized ligand layer over time due to harsh regeneration buffers.⁶⁰

7. Results and discussion

7.1. Literature search methodology

To ensure transparency and provide a comprehensive evaluation of glycated albumin (GA) detection platforms, a structured literature search was conducted using databases including PubMed, Google Scholar, Scopus, and Web of Science. The search strategy included combinations of keywords such as Glycated Albumin, “Raman Spectroscopy”, “Surface-Enhanced Raman Scattering (SERS)”, “Surface Plasmon Resonance (SPR)”, “Nitroblue Tetrazolium (NBT) assay”, and “Diabetes Mellitus”. Articles published over 39 years, ranging from 1987 to 2026, were considered to capture both established analytical methods and recent advancements in optical biosensing technologies. Studies were included if they were primary research articles or critical reviews focusing on the development, analytical performance, and validation of GA detection methods, including parameters such as specificity, sensitivity, limit of detection (LOD), linear range, sample preparation, and clinical applicability. Articles were excluded if they were not published in English, lacked sufficient methodological details, or focused solely on clinical outcomes without describing the analytical detection approach. A total of 60 articles were selected based on their relevance to the technical principles, analytical characteristics, and translational potential of four major GA detection platforms: NBT-based colorimetric assays, Raman spectroscopy (RS), surface-enhanced Raman spectroscopy (SERS), and surface plasmon resonance (SPR). The selected literature was critically analyzed to compare the advantages, limitations, and remaining challenges associated with each platform, with emphasis on their potential for future clinical translation.

7.2. Critical synthesis of analytical trade-offs

NBT assay is a low-cost, simple, and widely accessible method for routine fructosamine screening. Under alkaline conditions (pH = 10), fructosamine undergoes enolization and reduces NBT to a purple-blue formazan.¹⁹ However, it has low specificity for glycated albumin because it measures total glycated plasma proteins rather than GA alone.¹⁸

RS provides reagent-free detection of GA by inelastic light scattering corresponding to molecular vibrations.³ Glycation-induced conformational changes in human serum albumin (HSA) alter secondary structure features, producing diagnostic spectral shifts in the Amide I and Amide III bands.²⁶ However, RS in a complex serum matrix produces background fluorescence that masks the weak Raman signal.²⁸

In SERS, by placing glycated albumin in proximity to metallic nanostructures, localized surface plasmon resonance generates electromagnetic “hot spots” which amplify weak Raman scattering.²² Quantitative SERS has limitations, including poor signal reproducibility resulting from substrate non-uniformity, nanoparticle aggregation, spatial hot-spot variance, and high nanofabrication costs.³³

SPR sensors measure local refractive index changes at a metal–dielectric interface as target analytes bind to immobilized recognition elements, enabling real-time detection and kinetic monitoring.⁵⁰ Apart from high sensitivity, SPR biosensing in serum faces technical barriers due to non-specific matrix adsorption (“fouling”) from abundant non-glycated HSA and globulins, bulk refractive index fluctuations, and degradation of immobilized recognition layers over repeated acidic regeneration cycles.⁵⁹

GA testing exhibits lower clinical adoption globally than HbA1c. This is primarily due to the higher per-test consumable costs of commercial enzymatic GA assays and lack of international standardization cutoffs (unlike the established ADA/CDC clinical diagnostic cut-offs for HbA1c). Emerging RS, SERS, and SPR platforms offer an alternative by providing reagent-free, label-free molecular signatures that eliminate ongoing consumable reagent costs once initial optical instrumentation capital is invested.³ As a future direction, spectral datasets generated from well-characterized clinical cohorts could be integrated with machine-learning approaches to identify spectral features associated with clinically relevant glycemic states and develop predictive models.

8. Conclusion

RS offers many advantages over the NBT assay for detecting GA. In comparison to conventional NBT colorimetric assays, which measure total fructosamine through alkaline redox reactions and are affected by non-specific serum matrix interferences, emerging optical platforms offer high analytical specificity for GAevaluation. However, their operational mechanisms and surface requirements differ from each other.

Moreover, RS enables the simultaneous assessment of glucose and glycated albumin, offering quantitative data. Current limitations include fluorescence background, weak signal intensity, chemometrics, substrate reproducibility, and fabrication costs in SERS, and non-specific serum protein fouling in SPR currently restrict their clinical translation. These optical approaches remove ongoing reagent costs associated with enzymatic assays, but they involve investment in instrumentation. Their potential for routine point-of-care diagnostics will require standardized optical protocols and validation on large clinical cohorts. Future directions should focus on standardization of RS/SPR protocols, integration with machine learning for spectral interpretation, and clinical validation to establish RS as a complementary diagnostic tool alongside HbA1c and GA enzymatic assays.

Author contributions

Conceptualization: L. A. S. P. J., S. P. N. N. S., D. U. K., C. B. R. literature search: L. A. S. P. J., P. R. T. D. P., S. P. N. N. S. methodology and analysis: L. A. S. P. J., L. S. writing – original draft preparation: L. A. S. P. J., L. S., S. P. N. N. S. writing – review and editing: S. P. N. N. S., C. B. R., D. U. K. supervision: C. B. R., D. U. K. All authors have read and approved the final manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

This article is a review of previously published literature. No new experimental data were generated or analyzed in this study. All data discussed in this review are available in the original publications cited within the manuscript.