All-fiber label-free optical fiber biosensors: from modern technologies to current applications [Invited]

Biosensors are established as promising analytical tools for detecting various analytes important in biomedicine and environmental monitoring. Using fiber optic technology as a sensing element in biosensors offers low cost, high sensitivity, chemical inertness, and immunity to electromagnetic interference. Optical fiber sensors can be used in in vivo applications and multiplexed to detect several targets simultaneously. Certain configurations of optical fiber technology allow the detection of analytes in a label-free manner. This review aims to discuss recent advances in label-free optical fiber biosensors from a technological and application standpoint. First, modern technologies used to build label-free optical fiber-based sensors will be discussed. Then, current applications where these technologies are applied are elucidated. Namely, examples of detecting soluble cancer biomarkers, hormones, viruses, bacteria, and cells are presented.


Label-free optical fiber technologies
Label-free optical fiber biosensors (OFBs) are refractometers that target specific molecules through a biofunctionalization process [1].The building block for OFBs is a device, either inline within the fiber or on the fiber tip, that probes small changes in the refractive index (RI) in the environment surrounding the fiber [2].The RI sensitivity is then estimated by tracking the spectral changes of one or multiple spectral features and can be performed either measuring spectral shifts or intensity changes.
The biofunctionalization process then enables a selective detection of a target molecule, by immobilizing a corresponding bioreceptor over the surface of the biosensor.As the concentration of the target molecules increments, particles with RI higher or lower with respect to their surrounding medium are increasingly bound over the sensitive part of the biosensor, within the skin depth of the incoming light.As a result, the sensor response to the target biomolecules shows a similar trend to the RI changes, but with a selective behavior [3].Several configurations for biosensing can be arranged, as sketched in Fig. 1.The main differentiation, besides geometrical and fabrication aspects, is the modality of light propagation within the sensing elements.

Multimode fiber biosensors
Biosensors made with multimode fibers (MMFs) typically employ fibers having large core and thin cladding, which facilitate the interaction of light with the surrounding medium [4].The availability of weakly guided, higher-order modes promotes the interaction between the light propagating in the fiber core with the outer environment, and this can be easily implemented using glass or plastic fibers in which the cladding is only few micrometers thin.

Multimode fiber biosensors
Biosensors made with multimode fibers (MMFs) typically employ fibers having large core and thin cladding, which facilitate the interaction of light with the surrounding medium [4].The availability of weakly guided, higher-order modes promotes the interaction between the light propagating in the fiber core with the outer environment, and this can be easily implemented using glass or plastic fibers in which the cladding is only few micrometers thin.
The most common approach for biosensing using large-core fibers is based on the surface plasmon resonance (SPR) effect, that occurs at the interface between the fiber medium and a metallic thin film (typically 30-100 nm thick) [5].In the most common configuration, the metallic film is sputtered around a section of the multimode fiber, having the dual purpose of hosting the evanescent plasmonic mode and providing a surface where bioreceptors are immobilized [6].; this configuration resembles the Kretschmann free-space architecture, in which the wavelength shift of the resonant wave occurs as a result of the light excitation with a plurality of modes.The most common approach for biosensing using large-core fibers is based on the surface plasmon resonance (SPR) effect, that occurs at the interface between the fiber medium and a metallic thin film (typically 30-100 nm thick) [5].In the most common configuration, the metallic film is sputtered around a section of the multimode fiber, having the dual purpose of hosting the evanescent plasmonic mode and providing a surface where bioreceptors are immobilized [6].; this configuration resembles the Kretschmann free-space architecture, in which the wavelength shift of the resonant wave occurs as a result of the light excitation with a plurality of modes.
Recent works regarding fiber-optic SPR biosensors have investigated several components of the plasmonic sensors.The effect of different coatings was evaluated by Vitoria et al. [7], who reported three different plasmonic effects (lossy mode resonance, surface plasmon resonance, and surface exciton polariton resonance) depending on the dielectric permittivity of the metallic film.Sharma et al. reported SPR effects for various types of multimode fibers [4], while Cennamo et al. reported the use of larger plastic fibers for SPR sensing [8].Localized SPR (LSPR) biosensors make use of noble metal nanoparticles [9] as a thin surface, localizing the plasmonic effects into single or clustered particles [10].Plasmonic effects can also be exploited with tapered MMFs, in order to enhance the sensitivity, using a flexible fiber format in which the diameter varies along the fiber section; such sensors, recently labeled as "Waveflex", have been recently reported using simple [11] or cascaded tapers [12].Other configurations for plasmonic biosensors include U-bent fibers, in which a large-core fiber is folded with a tight bending angle in order to increase the amount of evanescent light; this structure has been exploited using SPR [13] and LSPR [14] configurations using tight fiber bending radii.
Plasmonic sensors provide high versatility, as they have been demonstrated in detection of particles of different size, and allow the use of low-cost setup based on a broadband light source and a low-cost spectrometer as detector [15]; they also allow in-line sensing for multiple targets [16].For this reason, they are mainly used as benchtop or portable instrumentation, alternative to lab-on-chip sample collection and processing methods.

Single-mode fibers exciting multi-mode propagation
Single-mode fibers (SMFs) provide a better light propagation property for light, having a larger mode confinement and an almost ideal tolerance to fiber bending.Due to high overlap integral in SMFs, the portion of the fundamental mode available for sensing the surrounding RI is negligible.Thus, several biosensors employ SMF fibers but exciting higher order modal propagation, either in the cladding or combining SMFs with multimode spans.
The tilted fiber Bragg grating (TFBG) is a device similar to a uniform fiber grating, but capable of exciting counterpropagating cladding modes [17]; such modes are confined between the cladding and the outer medium, and since the cladding is a multimode structure, they appear as a comb of evanescent modes in the transmission spectrum [18].The TFBG sensitivity to RI and to biomolecules is mainly encoded in the lowest order cladding modes, particularly at the occurrence of the cut-off condition in which the mode converts between cladding-propagating and evanescent [17].In order to expand the TFBG sensitivity, a plasmonic TFBG structure has been reported [19].In this configuration the TFBG is coated with a gold-sputtered thin film hosting the bioreceptors, while the grating excites the cladding modes incident onto the film and exciting the SPR resonance; the highest sensitivity is recorded for intermediate modes in which the RI sensitivity of the grating overlaps to the SPR mode [20].Due to their high stability, plasmonic TFBGs have also been employed for in-situ biological sensing [21].
The other in-fiber device that excites surface-probing cladding modes is the long-period grating (LPG); this is a grating with a long pitch (hundreds of micrometers) which generates forwardpropagating cladding modes interfering with the core mode [22].Due to the co-propagating interaction, the wavelength spacing between neighboring modes is large, in the order of hundreds of nanometers, and therefore LPG biosensors typically detect one resonant mode [23].The LPG has a long form factor, up to few centimeters, and requires the periodic inscription of defects into the core [24]; however, the arc-induced fabrication methods substantially simplify the fabrication, requiring only a telecom splicer [25].Currently, LPGs are one of the most sensitive types of refractometer, having RI sensitivity in excess of 100,000 nm/RIU when operating at the dispersion turning point [26]; however, such extreme performance figures can be observed only for RI values closer to the fiber effective index.
Another inline sensing configuration that uses a single-mode excitation into a multi-modal structure is the modal interferometer.In the most common configuration (SMS, single-multisingle mode device), a multi-mode fiber cavity is spliced at both ends of a single-mode fiber, forming a multi-path interferometer with multiple fringes at infrared wavelengths [27].Other variations of this structure enhance the RI sensitivity by using no-core fiber cavity [28], or by etching the cavity to enhance evanescent light [29], or by using only single-mode fibers using tilted splices to excite the second-order propagation mode [30].

Single-mode fiber biosensors
Biosensors based on all-SMF structure guarantee the optimal structure in terms of light propagation, as the entire power is carried only by the fundamental mode having the best tolerance to fiber bending and allowing the use of infrared interrogators that have the highest performance ratings in terms of speed and wavelength resolution.SMF biosensors also can be easily multiplexed for the parallel detection of a plurality of analytes [31].All-SMF biosensors in general do not achieve the highest RI sensitivity, but they have excellent biological performance in terms of detection limits due to their inherent stability.
A common approach in SMF biosensors is the use of wet-etching in order to remove almost completely the fiber cladding; in this structure, light is confined at the interface between the core and the outer medium, obtaining a change of group index and intensity that varies when the surrounding RI increases.This effect is at the base of etched Fiber Bragg Grating (eFBG) biosensors [32], in which a Bragg grating is etched by removing the fiber cladding by rapid etching with hydrofluoric acid, controlling the etching rate [33].The changes of the biological analytes in an eFBG result in both a shift of the Bragg wavelength, and a change of the intensity of the reflection spectrum.In alternative, the etched-tilted Bragg grating applies the etching process to a TFBG, obtaining a multi-wavelength structure in which each cladding mode has a RI dependence [34].
A recent approach to SMF sensing makes use of the so-called reflector-less biosensing approach, in which the detection is performed in reflection facilitating in-situ detection, but without the inscription of a grating or an interferometer; rather, the system interrogates the reflected waves occurring due to Rayleigh scattering.The building blocks of reflector-less biosensors are optical backscatter reflectometer (OBR) interrogators, commonly used in distributed sensing, and high-scattering SMF fibers, with a scattering increment over 10,000 times, having the core doped with a high density of scattering centers [35].Reflector-less biosensors have been demonstrated for protein detection, having a facile and scalable fabrication and sensitivity similar to eFBGs.Several works have reported this approach, enabling the RI sensitivity by means of fiber etching [36], fiber tapering [37], a combination of etching and tapering [38].
Another all-SMF structure with versatile use in biosensing is the fiber-optic ball resonator (BR) [39]; it consists of a spherical tip having 400-600 µm diameter, fabricated by means of a CO 2 laser splicer using a method similar to diffractive lens fabrication, but using a SMF fiber input.Unlike whispering gallery modes sensors, the BR is a self-contained sensor that interrogates the weak quasi-random reflection pattern at the interface between the spherical tip and the outer medium.The surface of the sphere can be functionalized using a silanization process [40], a metallic thin film [41] or metallic nanoparticles [42] in order to host a plurality of bioreceptors targeting multiple protein domains [43].

Interferometric biosensors
Optical fiber interferometers implement within an optical fiber light path the most common interferometric structures, such as Mach-Zehnder, Fabry-Perot, or Sagnac loop devices [44].In general, the interferometer is a device in which the RI sensitivity can be encoded directly into the structure, either by modulating the optical path length or by affecting the reflectivity of fiber mirrors.
The Mach-Zehnder interferometer is based on the interference or co-propagating light paths.The configuration proposed by Ran et al. is based on harmonic microfiber grating [45], in which the interference fringe is formed by exciting multiple harmonic modes in a microfiber that are recoupled at the opposite end.The resulting structure forms a multi-path interferometer, in which the wavelength shift of the main mode is detected with a high sensitivity.Other Mach-Zehnder configurations have been reported using a photonic crystal fiber [46] and with a microfiber inline cavity [47].
The use of a Sagnac loop for in-fiber biosensing has been reported by Gao et al. [48].In this configuration, the interfering beams are formed within a fiber ring, using a splitter in order to launch the light clockwise and counterclockwise.The interfering beams are modulated by the refractive index in one portion of the sensor, where a microfiber is located and biofunctionalized.A similar configuration was reported by Li et al. using a microfiber loop [49].Finally, the Fabry-Perot interferometer (FPI) is also commonly used for in-fiber cavities in biomedical sensing configurations.The FPI is formed within an optical fiber by cascading two or more inline mirrors within the fiber propagation path, formed either using an air gap [50] or fibers having different group index.The recent work presented by Rakhimbekova et al. [51] proposes an FPI sensor having the simplest fabrication, simply based on splicing and cleaving of SMF fibers.In this device, by splicing a SMF to a high-scattering fiber a weak mirror is formed, while a second mirror is formed by cleaving the fiber tip; due to the high number of scattering centers, an additional set of randomly distributed reflections overlap to these contributions, forming a semi-distributed interferometer (SDI).By functionalizing the outer surface of the SDI, it is possible to detect proteins at low concentrations.The SDI properties and RI sensitivity can be varied by optimizing the high-scattering fiber, achieving the highest sensitivity when using fibers with elongated defects drawn at lower temperatures [52].While in SDI sensors the refractive index varies the reflectivity of the mirror forming the interferometer, in core-exposed FPI sensors the fiber core is directly exposed to the analyte under investigation, which causes the change of the optical path length along the cavity [53].

Applications of label-free optical fiber technologies in biosensors
This section of the paper will mainly focus on soluble cancer biomarker proteins with high clinical relevance to which three or more label-free OFBs have been developed such as human epidermal growth factor 2 (HER2), CD44, apolipoprotein E, and carcinoembryogenic antigen (CEA).Tables 1 and 2 show the examples of some of the label-free OFBs for the detection of different analytes.Figure 2 shows variety of analytes detected by all-fiber label-free OFBs ranged according to their size.

HER2 protein
HER2 is one of the most popular cancer biomarkers used as a target in label-free OFBs systems.HER2 is overexpressed in 20-30% of breast cancer in humans [54].Extracellular domain of HER2 can be cleaved, and soluble HER can be found in serum [63].A tapered fiber fabricated on a commercial fusion splicer was integrated with an FBG.The FBG region, being insensitive to RI changes, serves as a thermometer to compensate to the sensitivity of tapered region of the sensor to ambient RI change.The sensor was embedded into microfluidic channel in order to measure analyte of interest down to 2 ng/mL [54].Another biosensor for HER2 detection also had a temperature compensation ability; they integrated microfiber Bragg grating with a regular FBG region to detect HER2 protein with an LoD (limit of detection) of 50 ng/mL [64].Taking into account the importance of HER2 in breast cancer diagnosis, other label-free systems were fabricated including silica microfiber interferometer sensor [65].Binding of HER2 to the biosensor surface resulted in a wavelength shift in the interferometric fringe.The performance of the biosensor was tested in serum.An ultra-low levels of the protein were detected when a ball resonator sensor was combined with a TFBG region: 151.5 ag/mL in buffer and 3.7 pg/mL in diluted serum [66].Other biosensors used an aptamer to capture HER2 protein as was done with a plasmonic TFBG-based biosensor developed by Lobry et al [67].Demodulation technique of tracking lower envelope of the cladding mode resulted in a wavelength shift which  Gold-coated supermode interferometer Antibodies Insulin 100 ng/mL -Robustness -Ease-of-fabrication -Reproducibility TRL4 [58] was much higher than other reported methods that track individual modes.Sensor response was also amplified by HER2-binding antibodies.

CD44
CD44 is one of the most well-known biomarkers for cancer stem cells, which are a subset of cells responsible for cancer recurrence, resistance to therapy and metastasis [68].CD44 is also found in a soluble form in blood after being cleaved from its membrane-bound form.CD44 protein is another cancer biomarker which was used as an analyte of interest in several label-free OFBs systems.A first OFB-based biosensor was developed using a ball resonator sensor which was built using a telecommunication-grade fiber in a fast and a robust manner [41].The biosensor was further improved by using another surface functionalization to achieve ultra-low LoD and a wider concentration range of CD44 protein [40].When green-synthesized nanoparticles were used both a substrate for antibody attachment and increasing reflectivity of the sensor, a dramatic increase in the sensitivity was observed [42].A ball resonator-based biosensor coated with ZnO also demonstrated a sensitive binding of CD44 protein on its surface with an additional possibility of regeneration without damaging the functionalized surface [69].

CEA
CEA is a diagnostic biomarker widely used for the detection of gastric, colon, breast and lung cancer [70].Fiber laser-based lasso-shaped biosensor having ultra-high optical resolution was built for detection of CEACAM5 protein, a CEA -related cell adhesion molecules 5 in Serum [55].Polystyrene@gold nanospheres were employed on the surface of microfiber sensors due to its synergettic effect: it increases the surface area of the sensor and enhances the evanescent field [71].The biosensor's LoD towards CEA was 3.54 × 10 −17 M and 5.27 × 10 −16 M in phosphate-buffered saline solution and serum respectively.A label-free biosensor based on optical microfiber coupler was built for CEA detection [72].The sensor had an antifouling surface to reduce non-specific binding and used a Sigma human serum as a wavelength shift reference to avoid the differences in human serum samples.

Apolipoprotein E
Apolipoprotein E is known to regulate lipid metabolism and restrict innate immune suppression.Its level is elevated in several types of cancer including non-small cell lung cancer patients (in serum), ovarian cancer (in cell lines and tissues), non-invasive bladder cancer (in urine), pancreatic cancer (in serum and tissue).It is believed that detection of this protein could improve diagnosis of ovarian cancer in terms of sensitivity and specificity [73].Shorter survival (overall and progression free) of colorectal cancer patients being treated with chemo-and antibody-therapy was correlated with the increased level of this protein [74].An SPR platform based on unstructured fiber with high surface roughness that enable re-scattering of light from the surface plasmons was fabricated for the detection of apolipoprotein E [75].The impact of two sensor functionalization methods on lowering the LoD was studied with biotin-neutravidin with (orientated antibody immobilization) being superior to cross-linking method (linking antibody covalently to the polyelectrolyte layer).The system was further multiplexed to detect an additional protein (clusterin) on a single sensor [76].Another OFB platform based on a cleaved tip of OF coated with metallic (gold and silver) nanospheres for localized SPR was also built for the same two proteins [77].Some portion of transmitted light is scattered after interacting with the nanoparticles and is recoupled to the fiber.Fewer particles on the surface were shown to result in higher sensitivity, and the biosensor was able to detect clinically relevant levels of two proteins with a limited cross-reactivity.

Other cancer biomarkers
Some of the works on label-free OFB focused on measuring cancer biomarkers with ultrahigh sensitivity.Thus, one study aimed at developing a biosensor against a renal cancer biomarker (carbonic anhydrase IX) with an ultra-low LoD.They were able to detect the protein of interest down to 13.8 zM in phosphate-buffered saline and 0.19 aM in diluted serum based on a microfiber fabricated using fully automatic hydrogen-oxygen tapering system [78].The biosensor also successfully detected renal carcinoma cells expressing this protein with an LoD of 180 cells/ml.Ucci et al [79] detected a promising serological protein -cripto1 protein -present in breast cancer patients.The work focused on obtaining a uniform antibody coverage of the sensor (phase-gradient plasmonic optical fiber meta-tip) as well as their oriented immobilization in order to lower LoD.
Other studies on label-free OFB systems demonstrating the detection of cancer biomarkers include a POF-based SPR biosensor for measurement of: 1) vascular endothelial growth factor by aptamers using a simplified experimental setup [80]; 2) neuron-specific enolase by TFBG with a high tilt angle and a nanosheet of black phosphorus [81]; 3) a gastric cancer biomarker (CA125) using SPR biosensors embedded into microfluidic chip [82]; 4) interferon-gamma protein using an in-line dual-optofluidic waveguide biosensor with a temperature compensation [83].

Thyroglobulin
Thyroglobulin is a dimeric protein and is used within the thyroid gland for the biosynthesis of the thyroid hormones thyroxine and triiodothyronine.Elevated levels of serum thyroglobulin have been observed in individuals with goiter and several other clinical disorders.Currently, the measurement of thyroglobulin remains the primary method for post-surgical monitoring of differentiated thyroid carcinoma.The detection of differentiated thyroid cancer biomarker thyroglobulin was investigated using reflection-type LPGs as fiber optic refractive index transducers.This would the real-time identification of biomarkers in the needle washout of fine-needle aspiration biopsies, assisting in the diagnosis of papillary and medullary thyroid carcinomas.
The biosensor was functionalized with an anti-thyroglobulin monoclonal antibody, resulting in a highly effective and sensitive detection system.This allowed the successful identification of human-purified thyroglobulin at concentrations as low as sub ng/ml in laboratory settings.Moreover, validation experiments conducted on clinical samples demonstrated the potential of this biosensor for future clinical applications [56].
Another biosensor based on localized SPR for the detection of thyroglobulin was developed.The effectiveness of a localized SPR sensor chip based on fiber optics is enhanced through the integration of gold nanoparticles (AuNPs) onto an optical fiber and the incorporation of a microfluidic channel.This design prevented the evaporation of biomolecule solutions and the loss of AuNPs, thereby improving the reproducibility of detecting different concentrations of thyroglobulin [84].
Another biosensor has been developed for the purpose of detecting thyroglobulin.This biosensor utilizes a high-performance fiber optic SPR sensor, which incorporates a dome array and nanogaps (DANG).This biosensor exhibits improved sensitivity for the label-free and realtime detection of biomolecules.The enhanced sensitivity is achieved through the implementation of a unique polymer bead method, which generates nanogaps that significantly improve the sensor's ability to detect changes in refractive index.Consequently, this biosensor achieves LoD at 38 fg/mL, and demonstrates excellent selectivity in detecting thyroglobulin antibody-antigen interactions.Also, when tested with serum samples, the biosensor exhibits a recovery rate of 103% [85].

Estradiol
Estrogens have a crucial role in human physiology and have attracted growing interest because of their presence in the environment and the human food chain, with a special focus on their occurrence in aquaculture methods.The utilization of estrogenic compounds like estradiol in aquaculture, with the objective of promoting the development of mono-sex populations to enhance growth efficiency, presents significant health and environmental hazards as a result of hormone pollution.A plasmonic biosensor incorporating a spoon-shaped waveguide was designed and evaluated for its ability to detect estradiol concentrations in various water samples.The results exhibited the sensor's flexibility and selectivity in distinguishing estradiol from dexamethasone and progesterone [57].
A biosensor incorporating a gold film-coated TFBG was created for the purpose of detecting environmental estrogens with high sensitivity.This biosensor operates by transforming the interaction between nuclear estrogen receptors and environmental estrogens into detectable signals.The biosensor has notable sensitivity and wide-ranging identification abilities, enabling the accurate quantification of environmental estrogens in water samples at concentrations as low as nanograms per liter.This presents an opportunity for continuing monitoring of worldwide environmental endocrine disruptors [86].

Insulin
Insulin is a hormone that regulates the level of blood glucose and the storage of energy, it is synthesized by the pancreas.An increased concentration of glucose in the body can lead to the development of diabetes and may initiate myocardial infarctions and renal problems in situations when there is impaired insulin synthesis or use.The incidence of diabetes mellitus is steadily increasing, leading to increased rates of mortality and morbidity.A wide range of biosensors have been created to enable the continuous monitoring of diabetes management and prevention of complications associated with this condition.For the detection of insulin, a gold-coated optical fiber supermode interferometer was developed.The deposition of a thin layer of gold on the end facet of a multicore fiber resulted in the formation of an interferometer that showed sensitivity to RI changes and was dependent on polarization.The highest sensitivity was seen in a sample coated with 10 nm of gold.This finding resulted in the development of a biosensor with the ability to detect insulin at a concentration of 100 ng/mL [58].

Cortisol
Cortisol is an important glucocorticoid hormone and has an impact on the metabolism of proteins and carbohydrates, with its primary physiological function being the regulation of glucose consumption in peripheral tissues.Additionally, cortisol plays a crucial part in controlling blood pressure, immunological function, and anti-inflammatory mechanisms.Elevated cortisol levels have been identified to be linked to several health conditions, including osteoporosis, hypertension, diabetes mellitus, increased vulnerability to infections, and depression.In recent years, there has been a growing interest in fiber optic biosensors as a significant technological development providing the highly sensitive detection of cortisol.The levels of cortisol show circadian shifts, reaching their highest point during the early morning hours and declining during the night increased risk of certain clinical conditions of significant socio-economic value, including age-related cognitive decline, obesity, and mental health conditions, occurs when cortisol values persistently deviate from the physiological range.SPR-based Plastic Optical Fiber (POF) sensor, integrated with a gold and palladium alloy, was engineered for the detection of cortisol, showing biosensing capabilities.The significant contribution was the development of a highly sensitive, low LoD (1 pg/mL) SPR and POF technology-based biosensor for cortisol monitoring, biofunctionalized with anti-cortisol antibodies [87].Another immunosensor with a D-shaped fiber optic configuration coated with a gold film was developed for the purpose of detecting cortisol using SPR technology.The significance of employing a D-shaped configuration fiber optic immunosensor for cortisol detection is enhanced using a gold coating instead of an AuPd alloy.This modification improves repeatability and reduces sensitivity to moisture [88].

Viruses and bacteria
This section will focus on viruses and bacteria detectable by OFBs, including COVID-19, influenza, E. coli, and Staphylococcus aureus.

COVID-19
The ongoing COVID-19 pandemic, caused by SARS-CoV-2, highlights the rapid spread of emerging viruses [89].Highly contagious viral infections, particularly those transmitted via droplets, like SARS-CoV-2 and its variants, possess the capacity for rapid dissemination, underscoring the profound societal implications of their transmission [90].The study of Janik et al. ( 2023) introduced a novel microcavity-based fiber optic sensor using a microcavity in-line Mach-Zehnder interferometer (µIMZI) for the real-time detection of SARS-CoV-2 virus-like particles (VLPs).The sensor demonstrated exceptional sensitivity, detecting SARS-CoV-2 VLPs with a limit of detection as low as single nanograms per milliliter, offering a promising tool for rapid, label-free, and highly sensitive virus detection [59].

Influenza
Influenza viruses belong to the orthomyxoviridae family and are characterized by enveloped structures and segmented genomes of negative-sense RNA [91].Their classification is determined by the presence of hemagglutinin and neuraminidase surface glycoproteins [92].SPR sensor based on a side-polished fiber optic platform was developed for the rapid detection of avian influenza virus subtype H6.The sensor, featuring a thin gold film and a self-assembled monolayer of isopropanol on the detection surface, exhibited a detection limit of 5.14 × 10 5 EID50/0.1 mL, with an average response time of 10 minutes.This compact and cost-effective fiber optic sensor holds promise for online and remote sensing applications, making it suitable for epidemiological surveillance and rapid diagnosis of avian influenza subtype H6 [61].

E. coli
Infections from pathogenic agents pose significant risks to global health and result in substantial socioeconomic challenges [93].µIMZI based on an optical fiber showed a highly sensitive and efficient label-free detection of pathogenic E. coli O157:H7 bacteria.Utilizing peptide aptamers as bioreceptors, the optimized sensor surface modification enables the detection of concentrations as low as 10 CFU/mL, with a minimal sample volume in the order of hundreds of picoliters [60].Another fiber-optic platform namely plasmonic fiber optic sensor using bacteriophage T4 as a bio-recognition element exhibited sensitivity and specificity for selective detection in mixed samples, even in the presence of significantly higher concentrations of non-target bacteria.The sensor demonstrated the capability to detect E. coliB40 at concentrations as low as 1000 CFU/mL [94].A stable long-period fiber gratings (LPFGs) was also used for the detection of E. coli with covalently immobilized bacteriophage T4.Unlike SPR sensors, LPFG eliminates the need for moving parts or metal deposition, ensuring high accuracy, compactness, and cost-effectiveness.The sensor reliably detected E. coli concentrations as low as 10 3 CFU/mL [95].The study of Celebanska et al. showed an E. coli detection in water utilizing LPFG along with α-mannoside receptor.The receptor, mimicking natural bacteria adhesion, is covalently immobilized on the LPFG platform, ensuring stable and reproducible operation.The sensor demonstrated a limit of detection of 10 3 CFU/mL [96].Another LPG coated with tantalum oxide demonstrated exceptional refractive index sensitivity exceeding 11,500 nm/RIU detecting bacteria E. coli with high sensitivity of 10.21 nm/log(ng/mL) [97].

Staphylococcus aureus
Staphylococci, prevalent purulent cocci among diverse microorganisms, represent a significant contributor to cross-infections within hospital settings, making rapid and real-time detection crucial for its early prevention.LPFG sensor, utilizing egg yolk antibody (IgY), was developed for the label-free detection of S. aureus.The sensor was fabricated with laser writing technology and demonstrated selectivity, fast detection time (around 20 min), and suitability for on-site screening of trace pathogenic bacteria with a detection limit of approximately 33 CFU/mL.[62].Another study introduced a highly sensitive OFB for detection of inactivated S. aureus.The biosensor, based on a tapered singlemode-no core-singlemode fiber coupler (SNSFC) structure with pig immunoglobulin G (IgG) antibody, demonstrated robustness and stability.Notably, it achieved a remarkable limit of detection of 3.1 CFU/mL, with a response time of less than 30 minutes [98].Moreover, a sensitive detection of methicillin-resistant Staphylococcus aureus (MRSA) was conducted using LPG along with ionic self-assembled multilayer (ISAM) film.The biosensor, functionalized with monoclonal antibodies targeting penicillin-binding-protein 2a of MR staphylococci, exhibited high sensitivity of 10 2 CFU/mL.The biosensor also demonstrated specificity in differentiating MR and non-MR bacteria and holds promise for detecting MR staphylococci in clinical samples [99].

Normal and cancer cells
In recent years, several attempts have been made to develop label-free OFBs and other methods for the detection of different types of cancer cells [100,101].This part aims to provide an overview of cancer and non-cancer cell detection by the application of label-free OFB.There is a growing body of literature that recognizes the importance of cancer cell detection to diagnose cancer at an early stage and improve the diagnostic measurement to prevent cancer development and progression or metastasis.Tumorigenesis is composed of several specific molecular traits, which are closely interlinked with each other.The seminal paper by Hanahan and Weinberg [102] has defined certain hallmarks of cancer, which serves as a profound basis for cancer biology.One of the crucial hallmarks is the ability for metastasis, therefore it is important to develop new biosensing platforms to detect cancer cells at an early stage.

Non-cancer cells
Shevchenko et al. [103] demonstrated the label-free and real-time monitoring of cellular behaviour of the NIH-3T3 fibroblast cell line and its response to various chemical stimuli by using the SPR-TFBG based on the single-mode fiber.The detection of cell detachment, absorption, and the inhibition of cellular metabolic activity were reported in this study.An experimental setup was composed of the inserted fiber inside the cell culture equipment and incubated for up to 30 minutes.As an adhesion protein, fibronectin was applied to attach fibroblast cells to the surface of sensors in combination with gold coating.Importantly, they demonstrated that the sensor can be regenerated using trypsin and ethanol, implying the possibility of recycling the same sensor for further experiments.

Breast cancer cells
Breast cancer is considered to be a leading cause of mortality after lung cancer worldwide, affecting women patients and leading to death due to metastasis [104,105].There are several works dedicated to the label-free detection of circulating breast cancer cells by the application of OFBs.The detection of circulating breast cancer cells by the label-free method using the tilted fiber grating has been reported by Loyez et al. [106].This work introduced the circulating tumor cells (CTC) detection based on aptasensors at low cells concentration range.They reported the rapid detection of metastatic breast cancer cells by the application of the all-fibre plasmonic aptasensor with a LoD of 49 cells/mL within 5 minutes.MAMA2 receptor against mammaglobin-A protein was used as an aptamer, which is expressed on the surface of the breast cancer cells, MDA-MB-415.In addition, they applied gold nanoparticles during the functionalization step to enhance the sensor's response and achieved two-fold amplification.Furthermore, they conducted specificity studies by detecting normal cell lines, including human embryonic kidney cells (HEK-293) and non-tumorigenic mammary epithelial cells (MCF10A), since they do not express breast cancer biomarkers (MAM proteins) as a control.Also, it has been reported that the detection of 10 cells/mL was possible by this method.Taken together, this study demonstrates the possibility of circulating cancer cells detection for diagnostic purposes by OFBs.
Another work on circulating breast cancer cells detection by the label-free method using a plasmonic TFBG comes from Chen et al. [107] The biosensor was designed by implementing a 18°TFBG and surface coating with the gold film at 50 nm thickness.The target cell of this work was human breast cancer cells (BT549), and the sensor was functionalized with membrane receptor GPR30 antibody which has a high affinity for the BT549 cell line and is commonly expressed in other breast cancer types.It was possible to detect the cell within 20 minutes and LoD was 5 cells/mL, implying the feasibility of this study to detect CTC in label-free and real-time conditions.Also, the real-time detection of CTC breast cancer cells in the bloodstream by the application of plasmonic fiber sensors was reported [108].This study involves the detection of Michigan Cancer Foundation-7 (MCF-7) breast cancer cells by functionalizing the sensor surface with the epithelial cell adhesion molecule (EpCAM) antibody-based receptor.The LoD was reported to be approximately 1.4 cells/µL and the needle-like sensor probe was characterized with a sensitivity of 1933.4 nm/RIU.Moreover, the circulating blood system was further designed using the mouse's whole blood in flowing conditions to resemble the real conditions.The successful detection of blood flow up to 10 cells per microliter during the 15 minutes was reported.A proposed fiber was composed of multimode fiber (MMF) for the light transmission purpose and SMF with gold film for the CTC sensing role.
Another work by Zu et al. [109] also presented a novel detection method of CTC, including human (MCF-7, MDA-MB-231) and mouse breast cancer cells (4T1) by using the combination of TFBG probe with the nanotube-functionalization and microfluidic device.Specifically, they implemented the halloysite nanotubes (HNTs) coating for the sensor surface to form slit-like patterns to enhance the biosensor sensitivity.A broadband source (BBS) was applied to excite the TFBG at a wavelength range between 1450-1560 nm.This study was characterized by a good linear response and LoD of 10 cells/mL and with the microfluidic controlling at 0.03 cells/mL which corresponds to the minimum detectable cell flow.This implies that such an approach can be integrated into the patient's vessel.Furthermore, the higher adhesion of cancer cells to the HNT-coated surface in comparison with normal cells has been reported.This study also presented the detection of human gastric carcinoma cells (MGC-803) and hematocytes as a control to compare with cancer cells.Microfluidic chips served as a fixating agent and enabled the dynamic flow of normal and tumour cells during the detection.The in-situ detection of several cancer cell types and hematocytes at a low volume in a fast and accurate way was achieved by combining the microfluidic device with the multi-resonance optical fiber.
An alternative approach using the etched multicore fiber characterized by compactness, labelfree, portable device, and ultra-sensitive properties for cancer cells detection has been proposed by Singh et al. [110].The sensing platform was composed of multicore fiber (MCF) which is arranged in a hexagonal shape and was combined with SMF.Several cancer cell lines including the hepatoblastoma cell line (HepG2), murine hepatoma cell line (Hepa 1-6), human breast cancer cells (MCF-7), adenocarcinomic human alveolar basal epithelial cells (A549), and normal cell lines, composed of neonatal connective fibroblast (NCF) and human fetal hepatocyte cell line (LO2) were detected in this study.The authors compared the sensing performance of the etched sensors using HF solution (hydrofluoric acid) in a controlled way and non-etched sensors during the cell detection and concluded that etched fibers were more sensitive.Several nanoparticles were applied in this work to further enhance the sensitivity of the biosensor after etching.If the gold nanoparticles were used for increasing the sensitivity, the graphene oxide (GO), and copper oxide nanoflowers (CuO-NFs) were applied in order to support the biocompatibility of the sensor.The combination of the abovementioned nanomaterials provides the better sensing performance.Importantly, the surface of the proposed sensor was incubated with 2-deoxy-D-glucose (2-DG) after the coating with NP because cancer cells have increased consumption and dependence on glucose compared to normal cells [111].Furthermore, the incubation of tumour cells with the glucose transporter (GLUT) receptors was performed to analyse the treated and non-treated cells by etched fiber in this study.It was reported that the strong signal corresponded to the increased number of cells during the measurements, which might be associated with the change of RI of the surrounding media, where the sensor was exposed.Finally, the LoD of this study was reported to be 3, 2, 2, 2, 4, 10 cells/mL, which correspond to the HepG2, Hepa1 6, A549, MCF-7, LO2, and NCF cells respectively.
Kaur et al. [112] also proposed another approach for the detection of several cancer types, including human breast, cervical, blood, and adrenal glands by using the 2-D (two-dimensional) materials-based optical fiber SPR biosensor at 1550 nm.The proposed sensor was composed of the core, metal, and 2-D material layers.This work was based on the detection of the difference in the RI of normal and cancerous cells with power loss and angular shift.The transfer matrix method was used to calculate the angular power loss in this work.HeLa, PC12, MDA-MB-231, MCF7, and normal cell line, Jurkat cells are used as target cell lines in this study.The range of RI of the targeted tumorous cells varied between 1.39 and 1.401.Importantly, the LoD was calculated to be 0.43 × 10 −5 .

Blood cancer
The detection of human acute leukaemia cells (K562) by using 12°TFBG probe with the combination of microfluidic chip, has been reported in the paper by Guo et al. [113].This study is based on the difference between the RI of cells (1.3342 to 1.3344) and intracellular density alteration of normal and cancer cells, also known as the "density alteration in non-physiological cells or DANCE" phenomenon.The possible fluctuations of temperature and power level were prevented due to the differential spectrum approach in this work.The detection principle of this work is focused on targeting the difference between the intracellular cell density of human acute leukaemia cells.Also, the different stages of leukaemia cells were reported.The K562 cell line was separated through discontinuous sucrose gradient centrifugation (DSGC), and the bio-samples were divided into three groups, consisting of S40, S50 and S60.The experimental setup was composed of polydimethylsiloxane (PDMS)-based microfluidic channels for the fixation of the sensing probe, and an electronic-controlled pump was applied for the injection of bio-samples to prevent possible contamination.This study was characterized with LoD 2 × 10 −5 RIU cells/mL and a wavelength shift sensitivity of 180 nm/RIU.

Prostate cancer
The label-free detection of prostate cancer cells was introduced in the work by Ding et al. [114].They focused on the detection of the noncancerous prostate (BPH-1) cells and prostate cancer (PC-3) cells, which was based on the RI differences of the normal and cancer cells.The proposed sensor applied a photonic-crystal biosensor in a total-internal-reflection (PC-TIR) configuration, which was able to distinguish the RI difference between normal and cancer cells.Moreover, the authors observed the epithelial-mesenchymal transition (EMT) of PC-3 cells by adding the soluble TGF-01 factor.

Skin cancer
Malachovska et al. [115] presented the cell-specific detection via membrane-binding proteins by the application of the SPR based on the fiber optic immunosensor.The sensor structure is based on the integrated gold-coated TFBG in the core by using a laser.This type of sensor functions at 1550 nm in the infrared spectrum, making it beneficial for the investigations of cellular interactions, and suitable and compatible with telecommunication-grade measurement equipment.The functionalization step included the incubation of the sensor surface with the monoclonal antibody against the extracellular domain of the epidermal growth factor receptor (EGFR) which is mostly expressed by several cancer types and considered to be one of the common cancer biomarkers.The detection of the human epidermoid carcinoma (A431) cell line, which is highly expressing the EGFR, at the concentration of 2 to 5 million cells/mL has been reported.It was possible to detect and monitor cells in a short period of time.The human cervical carcinoma (OMC1) cell line was used as a negative control for the specificity studies.This study reported the LoD varying between the range of 1.1 × 10 6 cells/mL and 2.0 × 10 6 cells/mL.Also, it has been shown that by the application of the controlled thermal annealing, the adhesion of gold to the silica surface was optimized.

Discussion
When analyzing the current trends for optical fiber biosensors based on a label-free approach, we can decouple the working principle of the technology from the specific biomolecule being detected.Plasmonic biosensors have received the widest attention in terms of research study, and to date the technology has reached a substantial maturity in terms of hardware, software, and coating fabrication [116,117].The recent studies are diversifying the typology of resonances under investigation [7], and their implementation using coatings having permittivity with different sign; from a system point of view, however, plasmonic biosensors guarantee the best affordability since they can operate with low-cost spectrometers and visible light sources [118], although their biological sensitivity is often limited.
The main innovations are however appearing in the field of single-mode fiber biosensors; this follows the overall trend of biomedical optical fiber sensors that are consistently exploiting the capabilities offered by custom-made optical fibers, that can customize properties such as mode field propagation [119], bioresorbing properties [120], and scattering content [121].The new generation of sensors employing reflector-less architectures, for example, removes the need for in-fiber grating fabrication, thus substantially simplifying the fabrication processes and potentially scaling up the possibility to access mass markets.An important effort has to be placed on drawing the standards of these new fibers in order to achieve a successful commercial deployment; that would also allow users to draw specific splicing recipes for these fibers.The future efforts will also need to shift from intensity-sensitive to wavelength-shifting sensors, in order to improve robustness in practical applications.
Looking at the biological applications, most of the sensors have been validated in the scenario of protein detection, in liquid analytes such as serum [40], urine [122], or saliva [123] ; to date, this remains the main application for OFBs, and the successful implementation in cardiovascular and cancer biomarker detection can unlock a substantial potential for commercial applications, as well as a tool for the real-time biomarker detection suitable for pharmacologic studies [124].An important point of analysis for future label-free OFBs is the impact of the analyte over the measurement: since each sensor detects the differential response of the sensor exposed to the liquid analyte containing various concentrations of the target proteins, it would be important to disambiguate the effect of the analyte itself from the sensitivity to the specific protein.Artificial intelligence (AI) platforms can aid this task, as they can process large datasets extracting the features related to each effect.
On the other side, fewer works with less maturity in terms of bioreceptors and working principles have been reported for targets that have size approaching or comparable to the the wavelength (such as hormones or viruses), or larger than the wavelength (such as cells).For these systems, the working principles vary and sensors having a larger geometrical size and length of few millimeters have been preferred so far.However, we expect to witness significant research in the incoming years regarding these analytes, since the performance limits and the multiplexing capacities offered by OFBs are unmatched by other optical biosensors based on fluorescence or other labeled technologies, or by other electrochemical diagnostic platforms [125].In attention, the attention to biosafety and real-time diagnostic that has been received during the SARS-CoV-2 pandemic is acting as a strong incentive towards the deployment of real-time OFBs as well, and it is likely that this will translate into a wider spectrum of diagnostic capabilities and with an increased consideration for portability, wearability, and connectivity [126].

Conclusion
In conclusion, this work reviewed modern approaches for fiber-optic label-free biosensors, and their main biomedical applications.From a technological standpoint, plasmonic sensors are becoming the most popular architecture for low-cost systems, with the capability of providing an optical alternative to electrochemical biosensors, owning the capability of a rapid and accurate detection.On the other side, the use of telecom-grade single mode fibers can lead to systems having a higher biological accuracy (usually characterized by an ultralow detection limit); systems based on single/multi-mode propagation, all-single mode propagation, and interferometry have been discussed, with the capability of either achieving ultrahigh sensitivity, or a low-cost probe with high repeatability and reliable in situ sensing.
OFBs find their main applications in protein sensing, for a wide detection of cancer, cardiovascular, or other biomarkers.The capability to work in liquid media makes them a good candidate for in-situ sensing.The versatility of the technology, and the numerous possibilities for its biofunctionalization, have brought recent research to tend towards the detection of larger molecules, such as viruses, bacteria, hormones and cells.This work reviews the most recent findings in these domains, providing insights on the surface chemistry and bioreceptors used in detection systems.
Future research in fiber optic biosensors, owing to their high biological performance, should mainly address the applications.The possibility to detect in real time and with a label-free, specific, and low-limit technology using a microscopic probe should open important avenues in the accurate detection of biological targets.Therefore, it is envisioned that the main research trend would improve the maturity of the technology, working directly with clinical samples and in diagnostic conditions in order to provide a significant impact in each application.

Figure 1 .
Figure 1.Illustrations of the main architectures of fiber optic biosensors, arranged by light propagation configuration.

Fig. 1 .
Fig. 1.Illustrations of the main architectures of fiber optic biosensors, arranged by light propagation configuration.

Fig. 2 .
Fig. 2. Schematic showing the variety of targets detected by all-fiber label-free optical fiber biosensors ranged according to their size.Popular targets are shown in bold.For biosensors detecting cells, the used ligands are shown in brackets.

Funding.
Science Committee of the Ministry of Science and Higher Education of the Republic of Kazakhstan (AP14869161); Nazarbayev University (20122022FD4134).

Table 1 . Examples of all-fiber label-free biosensors for the detection of hormones and cancer biomarkers
. (TRL = technology readiness level).