Fluorescence properties of methylene blue conjugated to normal/cancerous human tissues for spectral discrimination of breast cancer stages (I-IV)

Abstract

Laser-induced fluorescence (LIF) spectroscopy is utilized to differentiate normal/malignant breast cells/tissues and classify cancer stages, using the biocompatible methylene blue (MB) fluorophore. Cancerous cells (MCF-7) conjugated with MB exhibit a red shift of 3.96 ± 0.41 nm, corresponding to ∼2.5 times higher signal intensity versus normal cells (MCF-10A). Regarding stages, tissue-level fluorescence intensity is elevated ∼2.5 times in stage I and ∼8 times in stage IV relative to healthy ones, with red-shifts ranging from 3.73 ± 0.75 nm in stage I to 18.56 ± 2.15 nm in stage IV. Thus, the LIF-MB technique offers a rapid, non-invasive method for both the early detection and grading of malignant tumors. The novelty of this work arises from the synergic contribution of the LIF and gravimetric validation for breast cancer staging.

1. Introduction

Breast cancer is one of the most prevalent malignancies worldwide, accounting for 30% of annual cancer cases in women. It is also recognized as the second most fatal disease among women [1]. In 2023, ∼ 300,000 new cases of invasive breast cancer and ∼ 50,000 cases of ductal carcinoma in situ (DCIS) were diagnosed in the United States alone, while breast cancer–related mortality accounted for ∼14% of the female cancer deaths worldwide [2,3]. It is worth noting that half of all breast cancer cases occur in middle-aged women, with nearly half of these cases diagnosed between the ages of 40 and 65 years [4,5]. Indicators of breast cancer progression include the presence of a breast lump, changes in breast appearance, small skin depressions, nipple discharge, a sunken nipple, or red and crusted skin on the breast [6]. Disease progression is also assessed regarding the tumor size; and whether it possesses hormone receptors. The stages are classified in Table 1 [7–10].

Table 1. Breast cancer stages and corresponding tumoral distribution [7–10].

Stage	Size	Involved lymph nodes	Tumor location
I	$\le 2\text{cm}$	0	Seeded in the breast tissue.
II	$2-5\text{cm}$	1–3	Either in the breast or in the lymph nodes nearby or both.
III	$<5\text{cm}$	4–9	Spread from the breast organ to the skin, chest wall, or multiple lymph nodes in or neighboring the breast.
	$>5\text{cm}$
	Locally advanced
IV	Very large	all around	Metastatic cancer. Spread to one or more distant parts of the body, most commonly to the bones, lungs, and liver.

The death rate from breast cancer in women has remained stable due to recent improvements in detection and treatment [11]. In fact, according to the World Health Organization (WHO), age-standardized breast cancer mortality rates have declined in high-income countries by approximately 2–4% per year due to the recent improvements in early detection and treatment strategies [12]. This reduction is attributed to early detection through modern diagnostic methods, heightened awareness, and the availability of advanced diagnostic technology [9,10]. Standard diagnostic methods include X-ray mammography (XRM), ultrasonography (USG), magnetic resonance imaging (MRI), and positron emission tomography (PET), all of which are widely used to detect and characterize breast lesions, staging and to guide clinical management [13,14]. However, ionizing imaging modalities, such as X-ray mammography and computed tomography (CT) expose patients to cumulative radiation doses that may cause DNA damage, increase cumulative radiation exposure, and elevate the long-term risk of secondary malignancies, particularly with repeated examinations [15,16]. Recently, rapid and non-invasive diagnostic methods based on optical spectroscopic techniques have been developed, favoring cancer diagnosis/treatment [17,18]. These techniques are related to the morphological and biochemical characteristics of tissues, enabling the identification of subtle alterations in the optical properties; such as dispersion, absorption, and fluorescence, with lasers frequently being used as non-ionizing coherent sources to reveal the optical properties of soft tissues [18,19]. Metabolic and structural changes arise at the cellular/subcellular levels [19,20]; thus, optical spectroscopy contains valuable information for diagnosing cancerous tumors [7,8]. Laser-induced fluorescence (LIF) spectroscopy, as a real-time method, demonstrates significant features, including high sensitivity, rapidness, affordability, and portability [21,22]. In LIF spectroscopy, ultraviolet or visible light is usually used to provoke the fluorescent molecules [9]. In this study, we focus on LIF spectroscopy, based on the light absorption at shorter wavelengths and the subsequent emission at longer wavelengths. Photons are absorbed through the molecular transitions, leading to the re-emission at longer wavelengths after the occurrence of stokes shift and re-absorbance events [7].

In previous studies, the autofluorescence spectra have been reported from normal/cancerous breast cells, indicating some significant differences in the fluorescence characteristics [23]. Fluorescence imaging of single cells or cell monolayers may provide better contrast than volume-averaged fluorescence measurements of cells in suspension. Therefore, the spectral differences in the normal/malignant human breast tissues may be partially attributed to variances in the intrinsic cellular fluorescence of events [23]. Optical spectroscopy and fluorescence lifetime techniques have shown significant promise in distinguishing malignant from healthy tissues in real time. Fluorescence lifetime imaging (FLIM), in particular, provides contrast based on fluorescence decay kinetics rather than signal intensity alone, making it less sensitive to fluorophore concentration and excitation fluctuations. A recent comprehensive review has summarized the fundamental principles, methodological developments, and expanding clinical applications of fluorescence lifetime techniques in biomedical diagnostics, highlighting their growing translational relevance in oncology [24]. A time-resolved fluorescence spectroscopy (TR-FS) system was developed, capable of acquiring fluorescence lifetimes across multiple spectral bands. This system demonstrated its capability to differentiate tumor from normal tissue during surgical excision with near real-time performance [25]. Similarly, a non-destructive method combining autofluorescence imaging and Raman spectroscopy has been employed to assess sentinel lymph node status in breast cancer, achieving high specificity using histopathology as the reference standard [26]. Moreover, a UV diode laser operating at 266 nm was used to excite normal, benign, and malignant tissues to record the separate autofluorescence emissions [27]. Differences in parameters such as scatter plots, cut-off values, diagnostic indices, and receiver operating characteristic (ROC) curves have shown that the specific spectral parameters, such as the range under the peak of the fluorescence spectrum and its average intensity, can be utilized to distinguish the features of the breast tissues with high sensitivity [27]. The fluorescence quenching effects of graphene/nanodiamond nanostructures combined with MB have also been investigated using the LIF spectroscopy [28]. The addition of graphene oxide/nanodiamond (GO/ND) nanostructures to the suspensions leads to a noticeable blue-shift. It has also been confirmed that the quenching coefficient is strongly related to the molecular structure and active sites of the nanostructures [28]. Furthermore, LIF spectroscopy has been employed to measure the spectral displacement of stained breast tissue with Rhodamine 6 G (Rd6 G) to discern between normal/cancerous tissues [29]. The fluorescence spectra obtained from cancerous tissues typically show a lucid shift toward longer wavelengths (red-shift) compared to the normal ones [29]. Tissue samples have been examined according to LIF spectroscopy, either through autofluorescence or based on the conjugation with chemical and toxic fluorophores.

This study investigates the changes in the spectral patterns of fluorescence signal intensity as a diagnostic criterion to distinguish different grades of breast cancer. Specifically, the aim is to utilize LIF spectroscopy with the biocompatible MB fluorophore to not only discriminate normal/cancerous breast tissues, but also to assess the cancer grading. Methylene blue (MB) is selected as a biocompatible contrast agent due to its extant FDA approval, high quantum yield (∼0.52), and preferential accumulation in malignant tissues via its enhanced permeability and retention (EPR) mechanisms [30–33]. Compared to indocyanine green (ICG), MB provides stronger fluorescence emission at 686 nm with a superior signal-to-noise ratio and greater photostability. Conversely, ICG suffers from aggregation and self-quenching in aqueous media [33,34]. Although label-free autofluorescence avoids the use of exogenous agents, it is limited by weak signals, high background scattering, and spectral overlap with endogenous fluorophores (e.g., NADH, FAD), reducing specificity [35,36]. In contrast, MB-enhanced LIF delivers amplified, red-shifted signals in the near-infrared (NIR) window, enabling deeper tissue penetration and accurate real-time grading. Beyond its fast response, this approach is non-invasive, cost-effective, and eliminates the need for tissue excision, unlike histopathology, which is labor-intensive and prone to inter-observer variability [37]. Moreover, the excitation wavelength at 665 nm aligns with MB's absorption peak, supporting potential integration with photodynamic therapy (PDT) for theranostic applications [38]. To the best of our knowledge, there is no similar report available on the grading of the breast cancer using LIF spectroscopy, specifically examine the correlation between spectral shifts in neoplastic tissues and their corresponding grades. Accordingly, LIF spectroscopy is employed using a diode laser at 665 nm, to measure the spectral shifts in breast tissues stained with MB fluorophores. MB-assisted LIF offers a more sensitive and reliable approach for rapid and non-invasive breast cancer detection. The fluorescence emission of pristine MB solutions (without tissue) demonstrates a notable red-shift at low concentrations. This spectral red-shift also appears at high MB concentrations in both healthy and cancerous tissues. These spectral changes primarily arise from the distinct structural disorder in malignant tissues. According to these findings, normal/cancerous tissues can be effectively classified in terms of both the red-shift and the corresponding signal intensity.

2. Materials and methods

This project was conducted in accordance with the ethical code IR.IUMS.REC.1400.1005 at the Medical University of Iran. The laser spectroscopy experiments were performed in the Advanced Laser Laboratory at Amir Kabir University of Technology (AUT). Tissue samples were obtained from female patients aged over 18 years, undergoing breast biopsy or surgery at Khatam-al-Anbia Hospital, following written informed consent obtained by an oncoplastic breast surgeon. Inclusion criteria included histopathologically confirmed breast cancer (stages I–IV), while healthy tissue samples were collected from peripheral, unaffected areas of the same breast. Tissue samples were anonymized immediately after collection and labeled with coded identifiers to ensure patient confidentiality. Additionally, no confidential personal information beyond the patients’ gender, age, and cancer type was accessed.

MB (Sigma-Aldrich, analytical grade, Merck KGaA, Darmstadt, Germany; Cat. No. 457250-1GM, ≥95% purity) is used as a fluorescent probe. MB ( $C_{16}{H}_{18}Cl{N}_{3}S)$ is a formal derivative of phenothiazine, characterized as a highly hydrophilic, biocompatible, and light-sensitive substance [39]. It is a solid, odorless, dark green powder with a molecular weight of 319.86 g/mol. When dissolved in deionized water, MB forms a blue solution [40,41]. MB is used both as a fluorophore and a medication. It has been widely used as a biocompatible diagnostic agent/histological dye in clinical applications. It is FDA-approved for use in methemoglobinemia and is also widely utilized as a photosensitizer in the course of photodynamic therapy (PDT) [12–15]. Compared with other dyes such as indocyanine green (ICG), MB exhibits higher aqueous solubility, superior photostability, and a larger fluorescence quantum yield ( $\sim$ 0.52), whereas ICG tends to aggregate and self-quench in aqueous environments, leading to reduced signal stability and reproducibility [42].

2.1. Cell culture

The normal breast cell line (MCF-10A) and the cancerous one (MCF-7) are initially frozen, thawed, and cultured respectively in Dulbecco's Modified Eagle's Medium (DMEM; Sigma-Aldrich, Merck KGaA, Darmstadt, Germany; Cat. No. D7777) and DMEM/F12 supplemented with 5% horse serum (HS), 20 ng/ml epidermal growth factor (EGF), 0.5 µg/ml hydrocortisone, and 10 µg/ml insulin. Both media are further supplemented with 10% fetal bovine serum (FBS; Gibco 16000044, USA) and 1% penicillin-streptomycin (Sigma-Aldrich, Merck KGaA; Darmstadt, Germany; Cat. No. P0781-100 ML). All cell lines are maintained at 37°C in a humidified incubator containing 5% CO2. After the third passage, the cells will be ready to be used in a series of subsequent experiments.

2.2. Tissue sample preparation and dye solution

Normal and cancerous tissues from 34 female patients with various stages of breast cancer were examined. Unaffected (healthy) and cancerous samples taken from individuals were collected immediately after biopsy/surgery and subsequently stored in Falcon tubes containing 0.9% normal saline solution [43]. The samples were refrigerated at 4°C to preserve tissue structure for subsequent spectroscopic analyses. It is worth noting that all tests were performed at room temperature (22 ± 1°C) within 48 hours of tissue extraction to ensure that the samples remain normal during the experiments. All tumor samples were taken from ∼0.5-1 cm thick tumor core region as confirmed by histopathological assessment. Adjacent normal tissues were collected at least 2 cm away from the tumor margin under the supervision of surgeon and pathologist. All the samples were then cut into several pieces of ∼0.5×0.5 cm (average mass ∼ 125 mg). To quantitatively evaluate dye uptake, each tissue piece was weighed before and after staining using a Sartorius Cubis II analytical balance (precision 0.1 mg) located in a temperature and humidity-controlled room. For further experiments, each tissue pieces were then immersed for 30 minutes in a 30 µM (9.6 mg/L) solution of MB; dissolved in 100 cc of deionized water. The samples were then blotted with lint-free paper and air-dried under controlled ambient conditions to remove any residual moisture. Each dried tissue piece was then reweighed gravimetrically using the same balance, and the net mass increase (ΔW = W₁ − W₀) was recorded as the absorbed dye mass. The prepared samples were then exposed to a laser beam, and the LIF spectra were recorded consequently with excitation provided by a tunable red diode laser at 665 nm. The effective irradiance at the sample was ∼150 mW·cm⁻² which is well below the 500 mW·cm⁻² threshold commonly referenced in photo biomodulation (PBM) safety guidelines and below the typical levels associated with rapid photobleaching under continuous-wave excitation in biological tissues, with a short exposure time of 100 ms to prevent photobleaching. These exposure conditions are consistent with accepted safety limits for visible lasers, as outlined in the ANSI Z136.1 laser safety standard and related FDA guidance [44], ensuring safe measurements without causing tissue damage. The experimental setup for weighing and drying is illustrated in Fig. 1.

Fig. 1.

Schematic representation of the gravimetric procedure used to determine MB retention in breast tissue samples. Tissue pieces (average mass 125 mg) are weighed using a Sartorius Cubis II analytical balance (precision: 0.1 mg) before and after staining with a 30 µM MB solution (100 mL, 9.6 mg/L) for 30 minutes, followed by air-drying under controlled conditions.

Table 2, tabulated the measured values of MB retention, calculated from the net mass increase (ΔW = W₁ − W₀).

Table 2. Average of MB retention in healthy and malignant breast tissues of various stages, measured gravimetrically using a scale balance (precision 0.1 mg).

Tissue Stage	W0: Initial Weight (mg) ± SD	W1: Final Weight (mg) ± SD	$\mathrm{\Delta }$W: MB retention in tissue (mg) ± SD
Healthy	125 ± 0.1	125.4 ± 0.2	0.4 ± 0.2
Stage I	124.8 ± 0.1	125.4 ± 0.2	0.5 ± 0.2
Stage II	125 ± 0.1	125.6 ± 0.2	0.6 ± 0.2
Stage III	125.1 ± 0.1	125.9 ± 0.2	0.8 ± 0.2
Stage IV	124.7 ± 0.1	125.7 ± 0.2	0.9 ± 0.2

Finally, the Confocal imaging was performed using a Nikon Eclipse Ti-E inverted microscope (2011 model, Japan) in confocal mode. A 40× oil immersion objective lens (NA = 1.3) was used for high-resolution imaging. Tissue samples for microscopy were sectioned into 5–10 µm thick slices using a microtome prior to staining with 30 µM MB (9.6 mg/L) for 30 minutes. Images were acquired over a 50 × 50 µm field of view, with a pixel dwell time of 2 µs. All confocal images were processed using the NIS-Elements software (Nikon, Japan).

2.3. Experimental array

Figure 2 illustrates the experimental setup employed for laser-induced fluorescence spectroscopy of cells (Fig. 2(a)) and tissues (Fig. 2(b)) stained with MB fluorophores. A diode InGaAlP laser operating at 665 nm with an average power of 150 mW is employed to provoke the fluorophore molecules. The beam is delivered as a near-collimated spot with an effective illumination area of $\sim$ 1.0 cm², yielding a surface irradiance of $\sim$ 150 mW·cm⁻². Each LIF acquisition uses a short exposure time of 100 ms, corresponding to an energy density per acquisition of $\sim$ 15 mJ·cm⁻². A fiber optic spectrometer with a fiber diameter of 200 μm (Avantes AvaSpec 2048, NA = 0.22) is utilized for recording LIF spectra. This spectrometer is equipped with a diffraction grating of 300 lines/mm, providing a resolution of 0.4 nm. A glass slide is used as the substrate for tissue samples, while a cubic quartz cuvette (1 cm × 1 cm × 4 cm) uses as the irradiation chamber for cells. The dashed lines indicate that the experiment is performed separately on healthy/cancerous samples. The detection angle is configured to be perpendicular to the laser beam (right-angle arrangement). All graphs and figures are generated using OriginPro 2023 (OriginLab Corporation, Northampton, MA, USA).

Fig. 2.

Schematic of the LIF spectroscopy arrangement for (a) cells and (b) tissues. Irradiation is performed using a diode laser at 665 nm/ 150 mW. Dashed lines indicate separate paths for healthy and cancerous samples, preventing cross-contamination. Dual quartz cuvettes and sample holders are used. A right-angle arrangement directs the fluorescence emission toward the optical emission spectrometer via a notch filter to prevent stray laser photons from entering the spectrometer.

3. Results and discussion

Systematic experiments have been carried out in favor of both (MB + normal) and (MB + cancerous) cellular specimens, as well as on a collection of the stained normal/cancerous tissues, using LIF spectroscopy. The experimental results are divided into three categories: (I) LIF spectra obtained from the pristine MB solutions (without any cells/tissues) at different concentrations; (II) those taken from the stained cells (MB + cells); and (III) the spectra taken from the stained tissues (MB + tissues).

First, the LIF spectra of various concentrations of pristine MB solution ranging from 10–70 μM (3.2–22.4 mg/L) have been examined. Figure 3(a) displays the spectral absorbance as a function of MB concentration which increases with higher concentration. The inset shows the absorbance versus concentration at a couple of MB peaks, i.e, 610 nm and 665 nm. Figure 3(b) similarly illustrates the emission wavelength against MB concentration for the chosen excitation wavelength ( ${\lambda }_{ex}$ = 665 nm); which exhibited a nonlinear correlation between the emission wavelength and the fluorophore concentration. This attests the enhancement of the fluorescence signal alongside a noticeable red-shift at higher concentrations. Figure 3(c) illustrates fluorescence signal intensity as a function of MB concentration, which follows the Modified Beer-Lambert (MBL) formulation, given by Eq. (1) [45]:

$$I_f=\beta (1-{10}^{-\alpha Cl})e^{-kCl}$$ (1)

where $\alpha$ , $\beta$ , k and $I_f$ ascertain the excitation coefficient, proportionality factor, self-quenching coefficient, and intensity of the fluorescent emission respectively. The path length l is filled in a cuvette containing the fluorophore at a certain concentration C. Figure 3(d) depicts the spectral absorbance and emission within the same frame. Note that a maximum Stokes shift of 32 ± 0.7 nm appears for MB at 30 µM, indicating a significant overlap between spectral absorbance and spectral emission, which facilitates the re-absorption events. This obviously emphasizes that the re-absorption events frequently occur in dense solutions with respect to the dilute ones [28,46].

Fig. 3.

(a) Spectral absorbance for MB concentrations ranging 10–70 μM, showing that absorbance increases at higher MB concentrations. The inset illustrates the absorbance peaks at 610 nm and 665 nm versus MB concentrations. (b) Fluorescence emission wavelength corresponding to MB concentration, demonstrating nonlinear behavior accompanied by a notable red-shift. (c) Fluorescence signal intensity versus MB concentration based on the MBL formalism. (d) Spectral absorbance and emission are plotted in the same frame. A maximum Stokes shift of 32 nm is observed for MB at 30 µM, indicating substantial spectral overlap leading to significant reabsorption events, particularly in dense solutions against those of dilute ones. The error bars represent standard deviation (SD).

3.1. LIF of the cells

Neoplasia refers to the abnormal growth of cells of the body, wherein certain cells proliferate uncontrollably and spread to other organs during the metastatic stage [47]. Cancer cells can originate in almost any part of the body, and each tumor may contain trillions of abnormal cells [48]. Thus, the cellular experiments have been performed first, to examine MB effects on the cells of interest. The results attest a correlation between the findings and the normal/cancerous tissues characteristics in stage I, specifically in terms of signal intensity variation and the corresponding spectral shift in emission wavelength. Before performing the main LIF measurements, cell viability was assessed using a well-known and reliable method [49]. In this method, cultured cells were incubated with a 30 μM of MB solution for several minutes, then gently washed with phosphate-buffered saline (PBS) to remove excess dye. Cell counting and analysis were performed using a hemocytometer, and a microscopic image of the counted field is provided in Supplement 1 ^{(1.7MB, pdf)} Figure S1. The analysis revealed the presence of very few stained cells (indicating non-viability) in each field. The number of stained cells was usually fewer than 10 per plate. Considering the total number of cells in the fields, this represents $\sim$ 1–2% of the cell population; therefore, the overall viability was estimated to be >98%. Hence, the MB concentration and contact time under our experimental conditions did not cause significant cell death, minimizing the possibility of false-positive signals due to non-viable cells.

Next, a series of experiments were performed to examine both normal (MCF-10A) and cancerous (MCF-7) breast cells. Briefly, third-passage cells were incubated for 24 hours. After one day of incubation, the cells were released from the flask using trypsin. Then, each type of cell was transferred to a new flask with the appropriate culture medium, stained with MB solution for around 10 minutes, and finally centrifuged in microtubes to pellet the cells. The supernatant containing free MB was carefully removed, and the pellet was washed once with PBS to eliminate any residual free dye. Each group of the cell pellet was then exposed to the diode laser beam at 665 nm/150 mW for LIF spectroscopy. Figure 4 illustrates the spectra obtained from each cell colony. The fluorescence peak of MCF-7 cells is red-shifted 3.96 ± 0.41 nm relative to that of MCF-10A cells. In addition, the signal intensity is found to be ∼2.5 times larger than that of normal ones. It is worth noting that there are ∼100,000 live cells available in each flask, as counted via MTT test. Each spectroscopic assay was repeated on 5 replicates for statistical assessments. The scatter data are given in the Fig. 4 inset.

Fig. 4.

LIF spectra taken from healthy (MCF-10A) and cancerous (MCF-7) breast cells. Cancerous cells exhibit a red-shift of ∼4 nm and the signal intensity increase ∼2.5 times against those of the normal ones. Inset shows scatter data, indicating an obvious discrepancy between healthy/cancerous cells signal intensity.

Cell receptors are types of proteins located on the plasma membrane that facilitate the communication between the cell and its external environment. These receptors do function in the cell signaling by binding to the extracellular molecules [50,51]. Cancerous cells possess a unique set of surface receptors, which can provide the potential targets for tumor theranostics [52]. Overexpressed receptors are often found in the cancerous cells because those need in the interest of meeting the cell's demands for uncontrolled growth [53]. This abundance of receptors promotes greater conjugation of MB, in turn increasing the abundance of fluorophores within the cell, which results in a distinct red-shift [47], with the emission peak shifting toward longer wavelength [20], a spectral changes that can be useful for diagnostic purposes [54]. In addition, breast tissue is highly scattering media; that is, emitted photons undergo multiple scattering events within the tissue. This phenomenon enhances re-absorption events, further facilitating a significant red-shift in the fluorescence emission [36,37].

3.2. LIF of the tissues

After MB staining and tissue sample preparation, an additional gravimetric analysis was performed to quantify the actual dye mass in each specimen prior to fluorescence measurements. Each tissue piece was weighed before and after immersion in the 30 µM (9.6 mg/L) MB solution, as described in the Materials and Methods section. The net mass gain represented the amount of absorbed dye, confirming that all tissues successfully retained a measurable quantity of methylene blue. As shown in Table 2 in the Materials and Methods section, the average mass of absorbed dye gradually increases with tissue stage, ranging from $\sim$ 0.4 mg in healthy tissue to $\sim$ 0.9 mg with respect to stage IV. As further illustrated in Fig. 5, the retained MB mass exhibits a monotonic increase with cancer stage. This steady rise reflects compounding structural changes during tumor progression, such as increased porosity and interstitial permeability. Thus, the gravimetric analysis serves as an independent physical validation of uptake trends, confirming the proportionality of MB accumulation to malignancy.

Fig. 5.

Gravimetrically measured MB retention for healthy and cancerous tissues at different stages, display a monotonic increase in retained dye mass with advancing malignancy. The error bars represent standard deviation (SD).

After the weight measurement, a series of experiments have been carried out to analyze the LIF spectra taken from each normal/cancerous specimen for the purpose of neoplasia grading. As detailed in the Material and Methods section, a 150 mW InGaAlP diode laser operating at 665 nm is utilized for samples irradiation alongside prompt LIF spectroscopy with a right-angle arrangement and a notch filter to prevent the laser beam from entering the spectrometer.

Figure 6(a) displays the fluorescence emission spectra in favor of matched healthy/cancerous tissues taken from patients diagnosed with stage I breast cancer. The observations reveal a maximum red-shift of 3.73 ± 0.75 nm, corresponding to a ∼2-folds increase in fluorescence signal intensity. The inset demonstrates the scatter data from 10 different cancerous tissue samples alongside their corresponding normal peripheral tissues. It is worth noting that these peripheral tissues are regarded as the unaffected healthy samples in accordance with the ethical guidelines.

Fig. 6.

LIF spectra of stained cancerous and matched healthy breast tissues given for: (a) stage I, (b) stage II, (c) stage III, and (d) stage IV (metastasis). Cancer grading is in accordance to the observed spectral shift in the fluorescence emission. The insets illustrate the experimental scatter data of the signal intensity for both normal and cancerous human breast tissues from (a) 10, (b) 18, (c) 4, and (d) 2 patients, respectively.

Both signal intensity and the red-shift act as the characteristic parameters to discriminate healthy/cancerous tissues. When MB fluorophores are conjugated with the normal tissues, only a minimum elevation appears in the normalized signal intensity, considering as the measurement basis. The discrepancy in the signal intensity among the normal tissues arises from the differences in the fat content within the tissues [55]. The presence of fat affects the transmission of the laser beam and the subsequent detection of photons [56,57].

Similarly, regarding the stage II breast cancer, Fig. 6(b) depicts the spectra obtained from healthy/cancerous tissues. A maximum red-shift of 8.52 ± 1.48 nm and ∼3-folds elevation of the signal intensity is observed. The inset illustrates the scatter data taken from 18 patients as well as their surrounding normal tissues. Figure 6(c) displays the corresponding spectra of the stage III breast cancer. The cancerous tissues emphasize a maximum red-shift of 13.47 ± 1.69 nm, while the intensity elevates ∼6-fold against the normal ones. Additionally, the inset illustrates the scatter data of the cancerous tissues taken from 4 patients. The corresponding peripheral unaffected areas are given as the healthy ones. Eventually, Fig. 6(d) illustrates the LIF spectra in favor of metastatic stage IV breast tissue from only 2 available patients. These samples reveal a pronounced red-shift of 18.56 ± 2.15 nm, accompanied by ∼10-fold elevation in the signal intensity. The inset depicts the corresponding scatter data. It is important to note that the cancerous tissues from all stages have been examined and compared with their respective unaffected peripheral biopsied samples from the same patient's organ, ensuring the data reliability. Furthermore, the reproducibility of spectral alterations associated with cancer grading is evaluated to verify the correlation among the emission wavelength, spectral red-shift, and the corresponding signal intensity.

Regarding stage I breast cancer, it is worth mentioning that the grading primarily relies on the signal intensity rather than the red-shift. However, in advanced stages, both parameters contribute to performing the reliable grading assessments.

While variations in fat content can contribute to variability in the fluorescence signal of normal breast tissues, other factors such as tissue heterogeneity, fibrosis, and local inflammation may also affect the measured spectral features. To minimize these confounding effects, each malignant tissue sample was compared with its corresponding surrounding healthy tissue from the same patient. Despite potential variability, the observed trends in fluorescence peak shifts and intensity enhancements were consistent across samples and statistically significant, indicating that the measured LIF parameters reliably reflect differences between healthy and malignant tissues.

As part of the cancer treatment protocol, an expert submits the pathobiological report for each biopsied sample to the breast surgeon. These reports have been used as references for analyzing the LIF spectra. The typical report for each tissue is provided in the Supplement 1 ^{(1.7MB, pdf)} Table S1.

Figure 7 depicts the 2D scatter data of all true diagnosed samples within a single frame, comprising 68 matched tissue pairs from 34 individuals. It is evident that the LIF-based values have good correlation with the corresponding pathobiological report, grouping into 5 separate clusters respectively representing the healthy breast tissue and the breast cancer stages I-IV. Notable discrepancies among the tissue groups were statistically determined for categorizing/grading of the various biopsied samples.

Fig. 7.

2D scatter data illustrating the distribution of breast normal/cancerous tissues at various stages, highlighting discrete spectral discrepancies in terms of signal intensity and emission wavelength. Specifically, the scatter plot includes data from 34 healthy and 34 cancerous tissue. specimens.

Cancerous tissues contain numerous micropores caused by the damaged membranes. These cavities allow fluorophore molecules to diffuse deeper into the tissue, leading to a larger propensity for greater fluorophore deposition [58] that in turn reduces the collisional rate and subsequent interactions between suppressed molecules, resulting in a higher incidence of MB re-absorption events. Consequently, a significant red-shift occurs due to the higher concentration of fluorophores in the cancerous tissues against that of the healthy ones [59].

Fluorophore molecules demonstrate a specific chemical affinity determined by their chemical properties, allowing them to bind through conjugation process. This binding varies among different tissue types. These chemical characteristics enable the dyes to selectively interact with various components in both healthy and diseased tissues, leading to the measurable optical differences that can be utilized for identification and diagnostic imaging [29,54]. On the other hand, the fluorophore congestion results in a more pronounced red-shift. This occurs because a dense concentration gives rise to a lucid red-shift as the emission peak shifts toward the longer wavelengths [20]. Breast tissue, in particular, is taken into account as a highly scattering media, which undergo multiple scattering of emitted photons within the tissue. This phenomenon enhances the re-absorption events, leading to a significant red-shift in the fluorescence emission [36,37]. To further quantify the interaction between MB fluorophore and breast tissues, the effective retention concentration of equivalent MB in healthy and cancerous tissues is estimated based on the MBL model (Eq. (1)). This approach using the normalized fluorescence intensity data from pristine MB solutions (Fig. 3(c)), where all intensities are normalized to the maximum value at 30 μM (I_max = 1), to infer concentrations in tissues, assuming similar quenching and re-absorption behaviors.

For normalized intensities $I_{norm}=\frac{I_f}{I_{max}}with{I}_{f,max}$ corresponding to the maximum intensity at 30 μM in the pristine MB solution, the equation becomes Eq. (2):

$$I_{norm}={\beta }_{norm}(1-{10}^{-\alpha C})e^{-kC}$$ (2)

The parameters are determined by least-squares fitting to the normalized fluorescence intensity data of pristine MB solutions (10–70 μM, Fig. 3(c)). In this case, ${\beta }_{norm}=\frac{\beta }{I_{f,max}}=1.42$ , $\alpha =0.048\mu M^{-1}andk=0.032\mu M^{-1}.$

For each tissue stage, Eq. (2) is solved numerically using the Newton-Raphson method [60] in order to estimate the equivalent MB concentration, C. The estimated concentrations and the corresponding percentages of MB uptake (%) relative to the initial 30 μM staining solution are tabulated in Table 3.

Table 3. Equivalent absorbed concentrations and percentage uptake of MB in healthy and cancerous breast tissues, determined using the MBL model (Eq. (2)).

Tissue Stage	Normalized Intensity ± SD	Equivalent MB retention concentration (μM)	Equivalent MB retention concentration (%)
Healthy	0.76 ± 0.08	11.8	39.3
Stage I	0.85 ± 0.08	14.7	49
Stage II	0.88 ± 0.08	16.3	54.3
Stage III	0.93 ± 0.08	19.6	65.3
Stage IV	0.97 ± 0.08	24.1	80.3

These results reveal a progressive increase in equivalent MB concentration with increasing cancer stages, ranging from $\sim$ 12 μM (39.3% uptake) in healthy tissues with respect to $\sim$ 24 μM (80.3% uptake) in stage IV. The estimated concentrations lie within the linear region of the MBL curve (C < 30 μM), as shown above in Fig. 2(c), where the fluorescence intensity elevates nearly linearly with concentration prior to the onset of significant inner filter and quenching effects. Being within that regime ensures high measurement sensitivity and the non-linear distortion, enabling robust discrimination among the cancer stages.

A linear regression analysis was further performed using the Origin software to numerically depict the MB uptake versus cancer stages (Healthy = 0, Stage I = 1, Stage II = 2, Stage III = 3, Stage IV = 4), yielding Eq 3:

$$MBconcentration=3.14\times Stage+10.14(R^2=0.967)$$ (3)

The high $R^2$ value indicates that MB uptake serves as a reliable quantitative biomarker for cancer progression and staging. These equivalent concentrations, derived from the MBL model and calibrated spectroscopically, directly correlate with measured fluorescence intensity, making them the primary diagnostic parameter for real-time spectral discrimination. These concentrations represent the effective in-tissue fluorophore content rather than the external solution concentration, confirming that the differences in fluorescence intensity between the tissue stages are primarily caused by variations in dye uptake and local binding efficiency, rather than consequent optical data's.

Based on the fluorescence measurements and the quantitative estimates derived from the MBL model, the results were further analyzed to evaluate the relationship among fluorophore equivalent concentration, spectral shift, and corresponding emission intensity across different tissue stages. This model allows the conversion of normalized fluorescence signals into effective in-tissue MB concentrations, facilitating a direct comparison between the physical dye uptake and the associated optical response.

For better visualization, Fig. 8 summarizes the spectral results for all examined samples, illustrating the overall fluorescence behavior and its relationship to cancer progression. In particular, Fig. 8(a) displays the fluorescence spectra at each stage of cancer as a function of wavelength; Fig. 8(b) plots the MB retention derived from gravimetry, alongside the normalized signal intensity as a function of cancer stage; and Fig. 8(c) presents the emission red-shift relative to the cancer stages. Analysis of this empirical data indicated correlations among MB retention, signal intensity and the red-shift that can be used to differentiate the various stages of the breast cancer malignancy.

Fig. 8.

(a) LIF spectra of the breast cancerous tissues stained with MB in favor of stage I-VI, against that of the healthy ones. (b) MB retention and the normalized signal intensity within the tissue versus the cancer stages. (c) Red-shift in relation to the cancer stages. The error bars represent standard deviation (SD).

To provide a clear overview of the study population and associated spectroscopic data, Table 4 summarizes the clinical information of the patients, including the number of cases at each cancer stage, the age range, and the number of tissue samples investigated per patient. Additionally, it presents the associated spectroscopic data, including the mean and maximum fluorescence emission wavelengths, red-shifts with respect to both the previous stage and healthy tissue, as well as the relative increase in signal intensity within a general framework.

Table 4. Patient demographics, clinical information, and spectroscopic data.

Stage	No. of Patients	No. of Samples	Mean age	Mean peak λ (nm) ± SD	Max peak λ (nm)	Max. Red-shift (nm) against previous stage	Max. Red-shift (nm) against healthy stage	Signal Intensity Increase	Notes
Healthy	Same as cancer patients	Same as cancer samples	—	694.98 ± 0.92	696.48	—	—	—	Normal adjacent tissue used as control
Stage I	10 patients	10 cancerous + 10 normal	46.2	698.22 ± 1.12	700.21	∼ 3.73	∼ 3.73	∼ 2×	Intensity differences due to fat content
Stage II	18 patients	18 cancerous + 18 normal	51.4	700.43 ± 1.36	704.97	∼ 4.76	∼ 8.52	∼ 3×	Significant shift compared to Stage I
Stage III	5 patients	5 cancerous + 5 normal	53.7	705.10 ± 0.84	709.91	∼ 4.94	∼ 13.47	∼ 6×	More pronounced shift, smaller sample size
Stage IV	2 patients	2 cancerous + 2 normal	55.1	714.35 ± 1.75	715	∼ 5	∼ 18.56	∼ 10×	Limited data, noted in discussion

Although the number of advanced-stage breast cancer samples (stages III and IV) was limited, this constraint primarily reflects the restricted availability of freshly resected metastatic breast tissue suitable for optical measurements under controlled experimental conditions. Despite the sample limitation at these stages, the observed fluorescence trends were highly consistent and demonstrated strong correlations among cancer stage, fluorescence intensity, and spectral red-shift. Importantly, early-stage samples (stages I and II), which are most clinically relevant for early diagnosis and treatment planning, constituted the majority of the dataset. The reproducibility of spectral features within each stage, along with the clear separation between stages observed in the fluorescence spectroscopy, supports the robustness of the proposed grading approach.

Histological images of breast tissue samples are routinely analyzed to compare structural differences between healthy/cancerous tissue stages. Hematoxylin and eosin (H&E) staining reveals distinct morphological changes in cancer specimens, including elevated intercellular spaces, higher porosity, irregular nuclear borders, and irregular tissue architecture leading to the heterogeneity. In contrast, healthy tissue exhibits regular lobular structures and a uniform distribution. Confocal microscopy provides high-resolution images and precise optical sectioning, enabling detailed observation of the microscopic structures within the samples. In order to envisage the significant changes in fluorescence properties, a Nikon Eclipse Ti-E inverted laser scanning confocal microscope (LSCM; 2011 model), equipped with a 40× oil immersion objective lens (NA = 1.3), was utilized to capture the cross-sectional images of the tissues of interest. A tunable red laser set at 665 nm was employed to provoke the fluorophore molecules accordingly. Regarding LSCM preparation, the tissues were initially stained with 30 μM (9.6 mg/L) of the MB fluorophore. Subsequently, the samples were sectioned using a microtome into 5–10 µm thick slices. The images were then acquired, and the fluorescence emission is collected subsequently. A uniform distribution of the dye concentration was observed throughout the tissue depth. The MB fluorophores were evenly distributed within the tissue, allowing the laser to penetrate deeply into the tumor. The confocal microscopy images obtained from the healthy tissue and the breast cancer tissues at stages I–IV were precisely correlated with the corresponding experimental results.

Figure 9 presents a comprehensive comparison of histological, confocal, and 2D schematic analyses of healthy breast tissues across different cancer stages (I–IV). Both histology and LSCM images were obtained using a 40x objective lens. Histology images for each stage, from healthy through stage IV, are depicted in Fig. 9(a)–(e), illustrating the progressive morphological changes that occur during cancer development. Specifically, healthy tissue exhibits well-defined lobular organization and uniform epithelial layers, whereas stage I and stage II samples show mild to moderate architectural distortion, enlarged nuclei, and partial loss of tissue organization. These features become more pronounced in stages III and IV, where the tissue structure is completely disrupted, and densely packed abnormal cells dominate the field of view. LSCM images for the corresponding stages are likewise illustrated in Fig. 9(f)–(j), revealing similar trends with pattern changes that strongly correlate with the observed histological abnormalities. It is evident that the fluorophore population in the malignant tissue is significantly greater than that of the normal tissue. This observation distinctly clarifies the structural discrepancy among the normal and cancerous tissues, providing robust evidence for cancer grading.

Fig. 9.

Comparative histological, confocal, and schematic analyses of healthy and cancerous breast tissues across cancer stages. (a–e) Representative histological (H&E) images. Healthy samples exhibit organized lobular architecture and uniform nuclei, whereas cancerous tissues show progressive structural disorganization, increased nuclear size, and loss of cellular cohesion. (f–j) Corresponding confocal microscopy (LSCM) images demonstrating increased fluorescence intensity and heterogeneity with advancing cancer stages.

Together, these data facilitate a better understanding of cell proliferation in malignant tissues and provide consistent, multi-scale evidence supporting the correlation between histopathological transformations associated with breast cancer progression and the optical properties of the tissue.

In comparison to existing optical spectroscopic techniques including label-free autofluorescence, Raman spectroscopy, time-resolved fluorescence, and ICG-based imaging, the proposed MB assisted laser-induced fluorescence (MB-LIF) method offers an advantageous combination of high diagnostic performance, real-time data acquisition, and low system complexity. Label-free autofluorescence is often limited by weak intrinsic signals and spectral overlap, while Raman and time-resolved fluorescence techniques typically require longer acquisition times and more complex instrumentation. ICG-based approaches may also suffer from aggregation and photobleaching. Unlike most previously reported methods, which primarily enable binary discrimination between normal and malignant tissues, MB-LIF provides quantitative spectral features that correlate with the cancer stage. LIF offers high sensitivity, reliable and rapid signal acquisition, and an excellent signal-to-noise ratio due to the optical separation of excitation and emission. This unique grading capability, combined with its high specificity, real-time performance, and simple instrumentation, underscores the potential of MB-LIF as a practical and clinically translatable optical biopsy technique.

3.3. Statistical analysis

To evaluate the significance of fluorescence differences among normal and cancerous breast tissues across different cancer stages, the following analyses were performed using OriginPro 2023 (OriginLab Corporation, Northampton, MA, USA) and GraphPad Prism 9.

First, fluorescence peak positions were extracted from spectroscopic measurements of paired healthy and cancerous tissues obtained from the same individuals across tumor stages I–IV. For each stage, paired comparisons between healthy and cancerous tissues were performed using paired t-tests, as the normality of paired differences was confirmed using the Shapiro–Wilk test. To evaluate stage-dependent alterations in cancer-associated spectral shifts, paired wavelength differences were calculated as Δ = Healthy – Cancer for each subject, and a one-way analysis of variance (ANOVA) was performed across tumor stages, followed by Tukey's multiple comparisons test. Additionally, fluorescence peak positions of cancerous tissues were compared across different tumor stages using one-way ANOVA with Tukey's post hoc analysis. Data are presented as mean ± standard error of the mean (SEM), and statistical significance was defined as p < 0.05.

Paired analysis of fluorescence peak positions in Fig. 10(a) revealed a significant difference between healthy and cancerous tissues at tumor stages I, II, and III; however, no statistically significant difference was observed at stage IV. The lack of statistical significance at stage IV is attributed to the limited sample size and increased biological heterogeneity in advanced tumors, which can reduce statistical power despite the presence of a stage-dependent spectral trend. In addition, comparison of cancer tissues across stages demonstrated a significant stage-dependent shift in fluorescence peak positions, with higher peak values observed in advanced tumor stages.

Fig. 10.

(a) Fluorescence peak positions of healthy and cancerous tissues were compared in a paired manner within each tumor stage (Stages I–IV). Additionally, fluorescence peak positions of cancerous tissues were compared across tumor stages to assess stage-dependent spectral changes. Paired t-tests were used for healthy versus cancer comparisons within each stage, while one-way ANOVA followed by Tukey's post hoc test was applied for comparisons among cancer stages. (b) Paired wavelength differences were calculated as Δ = Healthy − Cancer for each subject at tumor stages I–IV. One-way ANOVA was used to evaluate differences in Δ values across tumor stages, followed by Tukey's multiple comparisons test. Results demonstrate a progressive increase in Δ with advancing tumor stage. Data are presented as mean ± SEM. Statistical significance is indicated as ns (not significant), p < 0.01 (**), p < 0.001 (***), and p < 0.0001 (****).

Analysis of paired wavelength differences Δ = Healthy − Cancer in Fig. 10(b) demonstrated a progressive increase in Δ values from stage I to stage IV. A one-way ANOVA confirmed a significant effect of tumor stage on Δ, and Tukey's post hoc test identified significant differences between early (stages I and II) and advanced (stages III and IV) tumor stages. These findings indicate a clear stage-dependent alteration in fluorescence spectral characteristics associated with tumor progression.

To evaluate the diagnostic performance of laser-induced fluorescence (LIF) spectroscopy in distinguishing healthy/cancerous breast tissues, receiver operating characteristic (ROC) analysis was conducted. ROC curves graphically represent a diagnostic test's ability to discriminate between two conditions (in this case, normal versus malignant tissue) by plotting the true positive rate (sensitivity) against the false positive rate (1 − specificity) across various threshold values.

In this study, fluorescence peak wavelengths obtained from LIF spectra were used as classifiers for ROC analysis. For ROC calculations, cancerous tissue samples were assigned a value of 1, while healthy peripheral tissues were assigned a value of 0. For each stage, the ROC curve was generated by comparing the peak wavelengths of cancerous tissue samples with their corresponding healthy tissues. The area under the curve (AUC) was calculated to quantify diagnostic accuracy, where an AUC of 1 indicates perfect discrimination, and an AUC of 0.5 represents random classification. Figure 11 resulting ROC curves, which demonstrate high sensitivity and specificity in differentiating healthy from malignant tissue at all stages. The curves confirm that the observed spectral shifts in peak wavelength are statistically significant and clinically relevant, supporting the use of LIF spectroscopy as a rapid, non-invasive technique for both early detection and staging of breast cancer.

Fig. 11.

ROC curve for discriminating healthy (0) versus cancerous (1) breast tissues based on fluorescence peak wavelengths obtained via LIF spectroscopy. Each curve represents the true positive rate (sensitivity) plotted against the false positive rate (1 − specificity). The area under the curve (AUC) quantifies diagnostic accuracy, with high AUC values indicating excellent discrimination between normal and malignant tissues.

These results provide independent statistical validation of the method, complementing the standard deviation and ANOVA analyses presented previously.

The calculated diagnostic metrics demonstrated a sensitivity of $\sim$ 97.47%, specificity of $\sim$ 99%, and an overall accuracy of $\sim$ 90.69%. These findings elucidate the high reliability of the MB-LIF-based approach for distinguishing cancerous from healthy breast tissue.

4. Conclusion

The scientific premise of this work arises from differential dye retention observed in healthy and malignant tissues. In principle, cancer grading can be performed by using an accurate balance to carefully measure the fluorophore residual after collection, drying, and washing. However, this would be a very sluggish method. Instead, real-time grading can be achieved by laser-mediated fluorophore excitation and consequent laser-induced fluorescence (LIF) spectrometry. This approach extends the previous optical discrimination of normal versus cancerous cells and tissues using LIF spectroscopy. Specifically, we have used spectra obtained from in-vitro specimens to develop an innovative method for the rapid diagnosis and staging of malignant breast tumors. This work addresses the cell/tissue phases of that development separately.

During the cell phase, normal (MCF-10A) and cancerous (MCF-7) cells were stained with biocompatible methylene blue (MB) following standard preparation protocols. These cells were then exposed to a 150 mW InGaAlP diode laser at 665 nm. The cancerous cells exhibited a 3.96 ± 0.41 nm red-shift in the fluorescence peak compared to healthy cells, along with a 2.5-fold increase in signal intensity than that of normal cells, primarily due to enhanced fluorophore conjugation and recombination events.

For tissue experiments, biopsied malignant tissues from breast tumors of stages I–IV, along with surrounding healthy tissues, were collected from patients at Khatam-al-Anbiya Hospital. After initial preparation, the tissues were stained with MB and exposed to laser radiation under controlled conditions. The recorded fluorescence spectra revealed a maximum red-shift of 3.73 ± 0.75 nm with a twofold increase in signal intensity for stage I compared to healthy tissue. This red-shift progressed to 8.52 ± 1.48 nm in stage II, with a threefold intensity enhancement; a 13.47 ± 1.69 nm red-shift with sixfold intensity increase in stage III, while metastatic stage IV showed the maximum red-shift of 18.56 ± 2.15 nm, accompanied by a tenfold rise in fluorescence signal. Both red-shift and signal intensity were utilized as key spectral features for differentiating and grading cancerous tissues. This method achieved a sensitivity of 97.47% and a specificity of $\sim$ 99% and an overall accuracy of $\sim$ 90.69%. comparable to or superior to values reported for other spectroscopic techniques, such as time-resolved fluorescence and autofluorescence/Raman spectroscopy. The near-perfect specificity indicates the absence of false positives, underscoring the robustness of MB-based fluorescence as a diagnostic tool. To further validate the robustness of these findings, comprehensive statistical analyses were performed. A one-way ANOVA demonstrated statistically significant differences (p < 0.05) in fluorescence peak wavelength and intensity among healthy tissues and various cancer stages. In addition, ROC analysis using fluorescence peak wavelengths as the classification parameter (healthy = 0, cancerous = 1) yielded high AUC values of 0.98 ± 0.013, confirming strong discriminatory power. These statistical results quantitatively demonstrate that the observed spectral variations are significant and not due to random variability, thereby reinforcing the diagnostic reliability of the proposed MB-LIF approach.

Corresponding to this robust discrimination, images from histology and confocal microscopy clearly attest to the structural discrepancies between healthy/cancerous tissues. Histological evaluation further confirms the progressive architectural disorganization and cellular abnormalities observed in the LSCM and LIF analyses. Confocal imaging provides high-resolution 2D visualization of tissue morphology, validating the spectral shifts observed through LIF. Thus, the combined histological and optical evidence strongly supports the correlation of the tissue structure and optical responses. Compared to histopathology, which requires fixation, sectioning, and staining, the MB-LIF technique envisage a reliable, real-time, and non-invasive assessment of tissue fluorescence within minutes, providing rapid intraoperative feedback. Additionally, MB may have utility in photodynamic therapy (PDT), highlighting its dual diagnostic and therapeutic potential. Hence, this approach holds promise for integrated diagnostic and therapeutic (theranostic) applications in breast cancer management.

Funding

Iran University of Medical Sciences https://ror.org/03w04rv71 ( 1400-2-90-21957); National Institute for Medical Research Development https://ror.org/046k81c97 ( 4001594); Amirkabir University of Technology https://ror.org/04gzbav43.

Disclosures

The authors declare no conflicts of interest.

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the corresponding authors upon reasonable request.

Supplemental document

See Supplement 1 ^{(1.7MB, pdf)} for supporting content.