Introduction

Cancer survival rates vary significantly according to the stage of disease at the time it is first detected and diagnosed. As a result, early diagnosis of cancers is crucial for their successful treatment. This is especially important for cancers with asymptomatic nature in early stages such as ovarian and pancreatic cancers. For epithelial ovarian cancer (EOC), the most lethal gynecological cancer, the relative survival rate of the patients is about 46 % [1–4]. However, the chance of a 5-year survival for patients diagnosed at early stage is above 90 %, while it is about 27 % for those diagnosed at advanced stage [2, 4]. Unfortunately, in the absence of effective methods for detection, very few cancers are diagnosed at an early stage [2, 3]. This is mainly due to the relatively poor sensitivity and specificity of the current diagnostic methods [5], and as a result, early detection of cancers remains a major obstacle in the fight against cancers.

One of the key approaches used to diagnose and monitor cancers is to detect and identify protein biomarkers that can be quantified in biomedical samples [3, 6]. These protein biomarkers would be measured and monitored by performing immunoassay that yield specific signatures that can be used for the early detection of the disease [7]. Immunoassays are biochemical tests that rely on the inherent ability of antibodies to bind to the specific molecules of interest, called antigens, and are used to detect or quantify antigen biomolecules in a biomedical fluid with limited interferences from other substances. One of the commonly used laboratory tests to detect antigens (or antibodies) in the blood is the enzyme-linked immunosorbent assay (ELISA) [7]. Although biomarkers have been successfully employed, there is a growing evidence that to enhance efficiency and accuracy in cancer detection, it is useful and possibly necessary to monitor levels of biomarkers with greater sensitivity. Thus, recently, multiplex immunoassays in clinical bio-analysis have been proposed [3, 6]. In addition, various approaches have been explored such as multicolor quantum dots [8], barcode striped metal nanoparticles [9], sets of fluorescent microspheres [10], and metal chelate complexes loaded with different lanthanides [11, 12]. Though all of these methods provide valuable tools for detecting biomarkers with various bioassays, the development of test methods with significantly improved multiplicity, sensitivity, and specificity as well as simplicity of instrumentation and sample preparation is of paramount importance for real-world clinical applications.

To improve methods for cancer biomarker detection and quantification, we have recently began exploring the dual use of laser-induced breakdown spectroscopy (LIBS) and nano- and microparticle technology to develop a relatively simple, fast, accurate, and sensitive method to detect biomarkers in biomedical fluids and in particular in human blood [13, 14]. In this approach, labeling provides the specificity required to analyze molecules of relevance to the disease. Additionally, LIBS offers some advantages compared to other analytical techniques such as inductively coupled plasma atomic emission spectroscopy, matrix-assisted laser desorption ionization, and mass spectroscopy [15, 16]. LIBS requires much smaller volumes of samples with none or minimal sample preparation. The spectra are collected in real time and the experimental implementation of the method is relatively simple. Therefore, applications of LIBS include a wide variety of fields such as biomedical studies, forensic and materials analysis, extraterrestrial exploration, environmental monitoring, and more [16].

In this paper, we report on an element encoded immunoassay method based on the use of samples prepared by tagging micro and/or nanoparticles to antibodies against specific biomarker and analysis using LIBS [13–16]. We call this method Tag-LIBS [13]. We apply the method to detect a widely used cancer biomarker CA 125 for the diagnosis and monitoring of EOC in a very complex medium of human blood and show that Tag-LIBS provides tremendous advantages such as sensitivity, accuracy, and ease of use over other immunoassay techniques. Although the method is currently applied for a single biomarker, we are confident that its extension to the detection of a panel of biomarkers is feasible.

LIBS has gained ground in various applications. Though most applications rely on the use of nanosecond lasers to ablate the samples of interest, we find that for Tag-LIBS, the duration of the laser pulse greatly affects the analytical performance. Femtosecond LIBS (fs-LIBS) offers tremendous advantages over nanosecond LIBS (ns-LIBS) in terms of sensitivity, reproducibility, interference with the ambient environment, and spatial accuracy. This is in agreement with the work of other researchers who have reported that the fs-LIBS technique offers significantly improved sensitivity and reproducibility compared to ns-LIBS [16–21]. It is for these reasons that we conduct our measurements using femtosecond laser pulses to ablate the particulate sample.

To demonstrate the Tag-LIBS method, we have focused on the detection and identification the widely used biomarker for the detection of EOC: Cancer Antigen 125 (CA 125) in a very complex environment: human blood [22–25]. This high molecular weight glycoprotein is expected to be included in newly explored serum biomarker panels [24]. The ability of a new assay to analyze CA 125 in human fluids is therefore of utmost importance for developing future diagnostics.

Materials and methods

Element encoded immunoassay protocol development

For purpose of calibration of the element encoded immunoassay, we prepared a set of eight standard microparticle suspensions. These sets included various amounts of 1.5 μm titanium dioxide (TiO2) microparticles (corpuscular), e.g., 0.0 ng control; 8 ng, 78 ng, 392 ng, 783 ng, 1.6 μg, 3.9 μg, and 7.8 μg weight loads; and standard amount 46.8 μg of 1.5 μm iron oxide (Fe2O3) microparticles (Bangs Laboratories). For microparticle modification, we used biotinylated monoclonal anti-CA125 antibodies M86429B (Meridian Life Science, Biodesign International) for iron oxide particles and M86306B (Meridian Life Science, Biodesign International) for titanium dioxide particles (Fig. 1a, b). The suspensions of antibody-modified particles were mixed with phosphate-buffered saline (PBS) containing 1 % of bovine serum albumin (PBS-BSA) in predetermined proportion (Fig. 1c). The resulting suspension mixture was used for analysis of human blood plasma (Blood Bank of Delmarva) spiked with various amounts of CA 125 (Fig. 1c). To maintain homogeneity and equivalent conditions for all samples, the human blood plasma after thawing was filtered over centrifugal filter (Millipore) with a 5-μm pore size.

Fig. 1
figure 1

Two elements (Ti and Fe) Tag-LIBS assay schematic: attaching of the biotinylated group A antibodies to titanium dioxide (a) and biotinylated group B antibodies to iron oxide (b) particles modified with avidin; assaying of pre-mixed antibody modified titanium dioxide and iron oxide particles (c) with various concentrations of CA 125 in human blood plasma (d)

Full size image

After overnight incubation of the CA 125-spiked blood plasma with antibody-modified particles, the bound and unbound titanium dioxide particles were magnetically separated using magnets with residual flux density about 14.5-14.8 kg (K&J Magnetics). The unbound titanium dioxide particles were washed away. Immediately following, extensive speed-controlled shaking (vortexing) using Vortex-Genie mixer (Scientific Industries) was applied to the suspension to break down possible nonspecific complexes. These procedures were repeated four times to enhance the accuracy of the separation. To achieve a homogeneous distribution of microparticles on the filter surface, we first vortexed the particle suspension and then removed the buffer by applying partial vacuum negative pressure to the bottom part of a filter. We used the microscopic visual control to ensure the homogeneity of the microparticle distribution on a filter surface (Fig. 3(a)). After the magnetizing-washing cycle and removal of the buffer, the top part of the test tubes were cut off while the bottom part of the test tubes with the assayed microparticles were dried. Finally, the assayed microparticles were analyzed by fs-LIBS for the presence of Ti and Fe elements. Titanium (334.94 nm) and iron (261.19 and 404.58 nm) emission spectral lines were used for spectroscopic analysis. In a control experiment, we used the PBS-BSA buffer (no. CA 125) for spiking the human blood plasma. In the experiment, we used 200 μL of the human blood plasma spiked with various amounts of CA 125. Up to 20 μL PBS was used to adjust volumes. No further dilution was made. Therefore, this experiment was conducted in approximately 100 % human blood plasma.

LIBS experimental setup

The fs-LIBS experimental setup used for this work is shown in Fig. 2. CPA-Series Ti-Sapphire 150-fs ultra-short laser (Clark-MXR, Model 2210) operating at 775 nm with pulse energy of 1.6 mJ per pulse is used to generate a laser-induced microplasma. The laser pulse is focused on the surface of the target sample (the bottom part of the test tubes containing the assayed microparticles) via a 50-mm-focal-length fused silica Bi-Convex lens. A motorized stage with x-y translation is used to automatically translate the sample during the measurement. Optical emission from the plasma plume was collected in an enclosed chamber by a fiber collimation lens at 45° with respect to the laser beam and focused onto an optical fiber. The fiber was coupled to ME5000 Echelle spectrograph (Andor) with a DH734-18F O3 iStar Intensified Charge Coupled Device camera (Andor). Emission from the plasma was collected 50 ns after the laser pulse with an integration time of 700 μs using an onboard digital delay generator of the spectrograph. These parameters were determined by reducing the continuum background while at the same time maximizing the signal-to-noise ratios of the emission lines for the species (the major elemental lines). The selected time gating parameters are the result of this procedure. The image of LIBS plasma plume expansion by femtosecond laser pulse in air (Fig. 3(c)) was obtained using Andor’s iStar DH320T-18U E3 Intensified Charge Coupled Device (ICCD) camera with the same gate delay and gate width as that used for collection of optical emission from the plasma.

Fig. 2
figure 2

Schematic setup of femtosecond laser-induced breakdown spectroscopy and automated data acquisition procedure, including CPA-Series Ti:Sapphire ultra-short pulse laser (fs-Laser), ME5000 Mechelle Spectrograph, Intensified Charge Couple Device (ICCD), fiber optics (FO), sample stage (SS), fiber collimation lens (FCL), and laser energy sensor (Sen), data acquisition device from National Instruments (DAQ)

Full size image
Fig. 3
figure 3

a Microscopic picture of laser ablation craters (light spots) on a fragment of the filter surface covered with assayed microparticles. b LIBS spectra of assayed samples with various CA 125 concentrations: green, 1 U/mL; brown, 0.1 U/mL; blue, 0.01 U/mL; cyan, 0 U/mL. c Fast photography of LIBS plasma plume in air (provided to illustrate the shape and size dimensions of the plasma generated by the LIBS process). d LIBS spectra of calibration samples with various titanium dioxide microparticle loads: brown, 7.8 μg; green, 3.9 μg; blue, 0.39 μg; cyan, 0 μg; and constant iron oxide 46.8 μg load. Dots represent averaged experimental data; lines represent fitted curves

Full size image

Results and discussion

To achieve both sensitivity and specificity in the detection of CA-125, we combined the methods of fs-LIBS-based multi-element analysis with magnetic particle separation. In this work, for Tag-LIBS assay development, we employed non-fluorescent metal microparticles tagged to specific monoclonal antibodies. We used two types of microparticles: magnetic (iron oxide) and not magnetic (titanium dioxide). Both types were modified with relevant monoclonal antibodies and were equally involved in identification of the proteins of interest. Also, the iron oxide microparticles had an additional carrier function of extracting the CA 125-bound particulate complexes from the plasma blood sample. Herein, the element encoded microparticles (TiO2, Fe3O4) were specifically bound via monoclonal antibodies to the proteins of interest (CA 125) to be assayed and analyzed by fs-LIBS.

To achieve higher sensitivity with laser ablation technique, different parameters such as selection of materials as well as tagging procedure and assay protocol were optimized. We recently demonstrated that the intensity of the emission lines from the plasma plume of the ablated mixture of particles directly relates to its area number density [26]. In general, with the assumption of uniform distribution of a constant number of analyzed particles, the larger particles yield stronger LIBS signals [27]. In the Tag-LIBS assay, we used microparticles of about 1.5 μm in diameter that were dispersed over the analyzed surface.

After completing the incubation, washing, and drying steps of the assay, we analyzed the microparticles dispersed over the filter surface by fs-LIBS. In the LIBS analysis, we collected a series of 100–150 LIBS spectra from the non-overlapping spots of the two to three filter surfaces covered by assayed or standard microparticles (Fig. 3(a)). Each LIBS spectrum was collected from a single laser pulse ablation per single spot on a filter surface (Fig. 3(a)). The LIBS emission spectra of the assayed microparticles are shown in Fig. 3(b), and the LIBS emission spectra of the standard sets of microparticles are shown in Fig. 3(d). The simultaneous spectroscopic detection of Ti and Fe elements above background levels demonstrate the presence of CA 125 in the blood plasma (Fig. 5) at levels about 2 orders of magnitude smaller than the typical 3–5-U/mL limit of CA 125 detection that in vitro diagnostic enzyme-linked immunosorbent assay (ELISA) method can provide [28].

From processing of microparticle suspension such as mixing and depositing and based on the microscopic visual verification (Fig. 3(a)), we assumed the relative spot-to-spot homogeneity of the microparticle distribution over the analyzed surface. The ablation spot size was measured to be about 100 μm, much greater than the 1.5-μm size of particles (Fig. 3(a)). To increase the reproducibility of the assay, and to partially compensate for the possible variations in the plasma plume emission intensity due to the various external factors (i.e., fluctuations of the laser beam intensity and spatial shape or position of the plasma plume) [29], the LIBS spectra with the intensities less than the threshold (40 % of the mean intensity from every sample which is about 4000 counts) of Fe I at 404.58-nm emission line were eliminated. In addition, we have chosen the mean intensities of Fe I and Fe II lines as references because the weight loads of iron oxide microparticles were the same in all samples.

The ablation plume dynamics is also important for quantification of emission of radiation [30]. The laser ablation of a sample initiates the plasma plume formation with broadband bremsstrahlung and recombination continuum emission which interferes with atomic line detection [30]. For the purpose of background subtraction, we determined the profile of the continuum emission for every LIBS spectrum by second-order polynomial fitting [27]. The obtained continuum emission curve was subtracted from LIBS spectra, and the averaged spectrum was generated (lines in Fig. 3(b, d)). The intensities of the Ti and Fe emission lines were determined by a Lorentzian or Gaussian curve fitting of the averaged spectra.

To develop a quantitative and reproducible multiplex assay for biomarker detection, it is useful to select chemical elements with distinct, intense, and isolate spectral lines. LIBS analysis of titanium dioxide (TiO2) microparticles in the presence of iron oxide (Fe3O4) microparticles reveals that a relatively intensive emission line of Ti at 334.94 nm has minimal spectral overlap with the nearest emission line of Fe at 335.5 nm (Fig. 3(b, d)). Therefore, titanium dioxide microparticles have been chosen to pair with iron oxide microparticles for assay development.

Reproducibility of the Tag-LIBS assay was achieved by the following procedures. The influence of matrix effects on emission spectra due to variability of atomic number density of microparticles was minimized by using constant amounts of iron oxide particles [15, 16]. Also, the fs-LIBS spectra were collected in the absence of liquid media (blood plasma) inside of an enclosed chamber in order to better control the ambient environment (Fig. 2). These procedures provide a comparison of the optical signatures of the samples and the standards at nearly identical conditions. The constancy of iron oxide number density was monitored continuously with the relevant Fe emission lines at 261.19 and 404.58 nm (Figs. 4, and 5b). Figure 5b shows that the intensities of the emission lines of Fe II at 261.19 nm were within the experimental uncertainty (1.5 % relative change) and the intensities of the Fe I 404.58 nm were slightly increasing (12 % relative change) depending on concentrations of CA 125 in human blood samples. At the same time, the Ti II-associated fs-LIBS signals at 334.94 nm revealed an approximately linear (R = 0.97) dependence (more than two times of relative change) over the biomarker concentration range of 0 to 0.1 U/mL (Fig. 5b). It is important to mention that the matrix effects accompanying LIBS may play a role in the slightly changing intensities of the fs-LIBS emission lines of Fe II and Fe I depending on concentrations of CA 125 [14, 15]. By comparing the Tag-LIBS assay output (Fig. 5) to the calibration curve shown in Fig. 4, we estimate that at a 0.1-U/mL concentration of CA 125, about 43 % of total amount of titanium dioxide microparticles were bound to iron oxide microparticles. Extrapolating the linear part of the assay (Fig. 5b), we estimate that 50 % titanium dioxide microparticle binding level can be achieved at 0.17 U/mL concentration of CA 125. In agglutination assays, the condition when the number of bound antigen molecules is about equal to half of the total binding sites on particulate surface ratio results in the maximum agglutination of antigen molecules in the assay [31]. This condition also determines the upper limit of detection [32]: between 0.1 and 1 U/mL concentration of CA 125 in this experiment (Fig. 5). The binding capacitance of microparticles is another important parameter characterizing the ability of the assay to work over a range of analyte concentrations. In this experiment, the binding capacitance of titanium dioxide particles to CA 125 was estimated as 0.07 units per assay. Although in our research we were concentrated primarily on the sensitivity of the assay, a wider dynamic range of assay is important for practical applications. The dynamic range could be extended by increasing the particulate load to provide more surface area for binding proteins of interest, by serial dilutions of the sample [33], and by employing kinetic methods [34].

Fig. 4
figure 4

Calibration curve: intensities of titanium Ti II at 334.94 nm (black dots and solid curve) and iron Fe II at 261.19 nm (grey triangles and striped curve) of emission lines depending on titanium dioxide microparticle load with the error bars as standard deviation of the mean obtained by batch-to-batch statistical method. The error bars as standard deviation of the mean obtained by combined statistical method are smaller than the size of symbols (dots and triangles)

Full size image
Fig. 5
figure 5

Two elements (Ti and Fe) Tag-LIBS CA 125 Assay. a A bar chart of LIBS intensities of Ti II at 334.94 nm and Fe II at 261.19 nm of emission lines depending on concentration of CA 125 in human blood plasma by two analytical approaches: horizontal patterned bars for Ti II and dark grey bars for Fe II by combined statistical method, diagonal patterned bars for Ti II, and light grey bars for Fe II by batch-to-batch statistical method, depending on concentration of CA 125 in human blood plasma. Dot-dashed line represents level of negative control. b Linear part of the Tag-LIBS assay with a left Y-axis LIBS intensities of Ti II at 334.94 nm (dots and solid curve) and Fe II at 261.19 nm (triangles and long striped curve), and Fe I at 404.58 nm (squares and short striped curve with a right Y-axis), depending on concentration of CA 125 in human blood plasma. The error bars as standard deviation of the mean obtained by combined statistical method are smaller than the size of symbols (dots, triangles, and squares)

Full size image

Developing sensitive assays is important for early and accurate diagnosis of cancers [3, 5]. To obtain the sensitivity of the Tag-LIBS assay in biomolecular terms, we estimate the number of recorded CA 125 molecules at 0.01 U/mL concentration level. We estimated that the number of CA 125 molecules at a 0.01-U/mL concentration in 0.2-mL blood plasma approximately resembles the number of bound titanium dioxide particles. This estimation is based on published data on the molecular weight of CA 125, a comparison of the Ti emission line intensities at 334.94 nm shown in Fig. 3(b, d) and Fig. 4, and that the specific activity of CA 125 (mucin 16) is 2.5 106 U/mg, and its average molecular weight is 2353 kDa [35, 36]. These yield a ratio of measured titanium oxide particles to a number of CA 125 molecules of about 1.6 and a concentration of CA 125 molecules of about 0.01 U/mL, much lower than the 5-U/mL analytical limit of sensitivity of the typical in vitro diagnostic ELISA assay [28].

The reproducibility is a key component in cancer detection and diagnosis [36]. To assess the reproducibility of the Tag-LIBS assay, we utilized two different statistical approaches: batch-to-batch and combined uncertainty analysis. The coefficients of variation (CV) of spectral peak heights at different concentrations of CA 125 were determined as the percentage ratio of standard deviation of the mean to the mean value. In the first analytical approach, the LIBS assay spectral data obtained from three experiments were analyzed separately. In this method, the mean values of spectral peak heights from three different LIBS assay experiments (about 150 LIBS spectra each) were compared to determine the batch-to-batch reproducibility of the assay. The obtained CV values were in a range from 5 to 24 % at different concentrations of CA 125 (Fig. 5). In the second analytical approach, the LIBS assay spectral data obtained from three experiments were combined as a single set for analysis. The intensities of total 450 LIBS elemental spectral peaks at 261.19 nm for Fe and at 334.94 nm for Ti were taken into account to determine the mean values and uncertainties of spectral peak height measurements. The CV values obtained by the combined method were in a relatively narrow range from 2 to 3 % (Fig. 5). Therefore, examining a greater number of the LIBS spectra of assayed microparticles can significantly improve the accuracy of the spectral analysis in encoded Tag-LIBS assay that is essential for future clinical applications.

Conclusions

We have developed a new approach for the sensitive and accurate detection of biomarkers in human blood. This approach, based on tagging specific biomarkers and then analyzing them using LIBS, yields promising results for CA 125. We expect that the method described here can be applied to any biomarker and indeed to a multiple set of biomarkers at once. We are therefore confident that Tag-LIBS opens new ways for earlier detection and monitoring of cancer (or other diseases) biomarkers in human blood plasma with high sensitivity. Potentially Tag-LIBS provides the tool for earlier identification of the presence of smaller, early-stage tumors [37], especially when biomarkers associated exclusively with cancer conditions (so-called ideal biomarkers) to be discovered are employed [36]. The lowest measured 0.01-U/mL concentration of CA 125 is hundreds of times less than typical ELISA method provides [27] and is at least five times better than other commercially available techniques (e.g., among seven immunoassays for CA 125 evaluated in [38], the limit of detection of CA 125 ranged from 0.05 U/mL on Access 2 (Beckman Coulter, Brea, CA) to 1.45 U/mL on the AxSYM (Abbott) [38]).

The 2 106 amplification factor achieved using the fs-Tag-LIBS assay (particle weight/CA 125 molecular weight) makes possible for the detection of biomarker CA 125 with estimated near single molecule per particle efficiency (based on the published molecular weight and the specific activity of CA 125) in about 100 % human blood plasma.