Open Access
Add to BibSonomy
Abstract
This study proposed label-free fluorescence lifetime imaging and phasor analysis methods to discriminate different grades of cervical intraepithelial neoplasia (CIN). The human cervical tissue lesions associated with cellular metabolic abnormalities were detected by the status changes of important coenzymes in cells and tissues, reduced nicotinamide adenine dinucleotide (phosphate) (NAD(P)H) and flavin adenine dinucleotide (FAD). Fluorescence lifetime imaging microscopy (FLIM) was used to study human cervical tissues, human cervical epithelial cells, and standard samples. Phasor analysis was applied to reveal the interrelation between the metabolic changes and cancer development, which can distinguish among different stages of cervical lesions from low risk to high risk. This approach also possessed high sensitivity, especially for healthy sites of CIN3 tissues, and indicated the dominance of the glycolytic pathway over oxidative phosphorylation in high-grade cervical lesions. This highly adaptive, sensitive, and rapid diagnostic tool exhibits a great potential for cervical precancer diagnosis.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Cervical cancer is one of the leading causes of cancer deaths in women, and it is also the fourth most common malignant tumor in women worldwide, with about 530,000 new cases and 270,000 deaths each year [1–3]. Cervical cancer usually develops from cervical intraepithelial neoplasia (CIN), which is a precancerous lesion categorized as low-grade squamous intraepithelial lesions (LSIL) and high-grade squamous intraepithelial lesions (HSIL). LSIL is sometimes called CIN1, which is of low risk and usually resolves without treatment. HSIL includes moderate dysplasia (also called CIN2) and severe dysplasia (also called CIN3), which are of high risk and may develop into cervical cancer [4–6]. The transformation from human papillomavirus (HPV) infection into cervical cancer takes about 5-10 years; but if cervical cancer is diagnosed at an early stage, or is found at the CIN stage of precancer, the chance of a cure can be greatly increased [7]. It was reported that the 3-year local control rate for patients with early-stage and advanced-stage cervical cancer is 87% to 95% and 74% to 85%, respectively [8]. Therefore, early detection and diagnosis of precancerous lesions are essential for appropriate treatment.
The routine screening test for cervical neoplasia was previously a conventional Papanicolaou smear, which was replaced by liquid-based cytology (LBC) in the past two decades. However, LBC testing requires several visits to the hospital and may take a few weeks, consuming considerable resources and time. Cervical biopsy requires staining of the pathological tissues and manual reading of the tissue characteristics, which relies heavily upon the selection of the collection sites. As for patients with small lesions, random collection of sites may potentially lead to a missed diagnosis. Due to various limitations of current detection techniques, new technologies are urgently needed to improve the speed, sensitivity, and specificity of cervical neoplasia screening.
Cervical lesions are caused by rapid division and proliferation of cells. Therefore, the metabolic requirements of cells increase, contributing to changes in their metabolic state and microenvironment [9–13]. There are coenzymes in cells and tissues—reduced nicotinamide adenine dinucleotide (phosphate) (NAD(P)H) and flavin adenine dinucleotide (FAD)—that are primarily involved in cellular metabolism and can emit fluorescence when excited [14]. There are also various endogenous fluorophores, such as elastin, collagen and protoporphyrin IX (PpIX). Elastin was demonstrated to be present in the dermal layer and connective tissues [15], collagen was reported as the major component of extracellular matrix [16], and PpIX accumulation was reported in many tumor-related diseases [17,18].
Fluorescence lifetime imaging microscopy (FLIM) can be used to monitor cellular metabolic abnormalities, tissue lesions, and the rates of glycolysis vs. oxidative phosphorylation by detecting lifetime changes in NAD(P)H and FAD in cells or tissues [19–24]. The metabolism is maintained in an abnormal state throughout the entire process from normal to low-risk lesions and then high-risk lesions. We found in a previous study that the average fluorescence lifetime of tissues can discriminate between normal and abnormal, but cannot directly distinguish among the different grades of lesions [25]. The average fluorescence lifetime, providing one-dimensional information, may show the same average lifetime information when components and corresponding proportions are different, so further exploration is needed to obtain more detailed information to provide further distinctions. In 2007, Digman and her colleagues introduced the concept of trajectories in the phasor domain for the first time [26], and in 2011, Stringari et al. [27] used a phasor approach to study the differentiation of stem cell metabolism. The phasor approach has the capability of simplifying FLIM image analysis—reducing the number of problems associated with exponential fitting—and provides a global representation of the fluorescence decay curve collected for each pixel. In recent years, the phasor approach has been used for pre-diagnosis of tumor pathology [28], a screening tool for basal cell carcinoma [29], and for the detection of breast cancer [30].
In this study, we used a phasor FLIM approach to systematically analyze cervical tissues from low- to high-risk lesions and to study the pathways of tumor metabolism using normal cervical epithelial cells.
2. Materials and methods
2.1 Participants and tissue samples
The cervical tissue samples were provided by the Department of Gynecology at the Central Hospital of Wuhan, China, and the study was approved by the Medical Ethics Committee of the Central Hospital of Wuhan. We obtained the samples from cervical biopsy, cervical conization, or total hysterectomy. We first immersed the samples in 10% formalin solution for fixation, embedded them in paraffin, and then cut them into several paraffin-embedded blocks of dimensions 2×2×0.3 cm. Then, tissue slices of 4-µm thickness were cut from each block and placed on slides. According to conventional protocol, tissue slices were deparaffinized, re-hydrated, and stained with hematoxylin and eosin (H&E) for histopathological examination. The total number of cases in our study was 24, and the sample collection time was from January 2015 to September 2018. The enrolled patients whose cervical tissues were analyzed in this study were not treated with any special medications before surgery or biopsy, and had no chronic diseases or any history of malignancy. In accordance with the histopathological diagnosis, we placed 2 patients into the normal control group, 3 patients into the CIN1 group, 2 patients into CIN2, and 17 patients into CIN3. The age range for all patients was between 27 and 65 years. All the tissue samples of the normal group were obtained from patients who underwent biopsies. When patients with a history of oncology or chronic disease were excluded, very few normal cases remained. It was reported that CIN1 were less in the CIN population [31–33]. Therefore, in this study, CIN1 cases were much less than high-grade CIN.
The remaining unstained paraffin-embedded tissue blocks were collected after diagnostic examination for FLIM study. One or 2 pieces of slices that contained intact cervical epithelium were cut from paraffin-embedded blocks for each patient. As our group reported previously, the autofluorescence lifetime could be steady for 14 days after fixation and mounting [34]. Thus, the unstained slices were studied using FLIM during the 12-14 days after biopsy or surgery. It should be noted that different fixation and mounting media may have different effect on the autofluorescence study of tissues.
2.2 Cell samples
Human cervical epithelial cells (HcerEpic, purchased from Beijing Beina Bio Co., Ltd) were cultured in Eagle's minimum essential medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS). HcerEpic cells were seeded in Petri dishes and incubated in a fully humidified incubator at 37°C in 5% CO2 in compressed air until they reached 80% confluence, and then observed. To study the changes in NAD(P)H and FAD during cellular metabolism, cells were incubated with CoCl2 (200 µM, Sigma-Aldrich) or 3-bromoacetone (300 µM, Sigma-Aldrich) for 90 min to inhibit cellular glycolysis or oxidative phosphorylation, respectively [35–38]. Untreated cells were set as the control groups. Each experiment was repeated independently 3 times.
2.3 Standard samples for phasor FLIM
The standard samples in this study included NADH (7.5 mM, pH 7.0, Sigma) in 90% phosphate buffer saline (PBS) and 10% 4-(-2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer, FAD (0.75 mM, pH 7.0, Aladdin) in 90% PBS and 10% HEPES buffer, PpIX in DMSO (0.5 mM, Macklin), elastin in powder form (≥99% pure, Sigma), and Rhodamine B in deionized water (1 mM, Sigma). The protein-bound NADH solution was obtained by mixing NADH (0.375 mM) with malate dehydrogenase (MDH, 0.375 mM, Aladdin) in HEPES buffer for 90 min at room temperature.
2.4 FLIM experimental setup
The FLIM images were acquired by a time-correlated single photon counting (TCSPC) FLIM system (SPC-150, Becker & Hickl, Germany) on a laser scanning confocal microscope (Olympus, FV300/IX 71, Japan) with an oil immersion objective lens (40×, NA=1.2) (Fig. 1(A)). The excitation laser was a 405 nm picosecond laser at a repetition rate of 50 MHz (BDL-405-SMC, Becker & Hickl, Germany), and the laser power was controlled as 10-50 µW. The emitted fluorescence light was separated from the excitation light by a 405-nm dichroic mirror. The x and y laser scanning signals were generated by a multifunctional I/O device (NI USB-6363) as 256 × 256 pixels per frame and were synchronized with TCSPC data acquisition. The optical resolution of this method is around 290 nm. For the light source of 50 MHz repetition rate, the pulse interval is 20 ns, thus complete decay curves within several ns can be obtained during the interval, and no incomplete decays need to be accounted for in this study.
Fig. 1. (A) FLIM setup. (B) Schematic of the collected tissue. (C) A typical FLIM image of a patient diagnosed as CIN2 with a 430 nm long-pass filter. Scale bar: 50 µm. (D) The phasor FLIM plot of image (C).
For the cellular experiments, the fluorescence was separated by a 508-nm dichroic mirror, then a 447 ± 30 nm bandpass filter was used for NAD(P)H signal detection, and a 532-nm long-pass filter was used for FAD signal detection. Five different regions were analyzed for each group of cells.
The fluorescence spectra of different fluorophores were reported with the emission peaks of collagens, elastin, NADH, FAD and PpIX at 390 nm, 410 nm, 460 nm, 550 nm and 620 nm, respectively [39,40]. When FLIM images of the tissues or cells were collected with a 447 ± 30 nm band-pass filter for NAD(P)H, the signals might partially have the components of elastin and collagen. However, NAD(P)H predominantly exists in the epithelium and collagen mainly exists in the stroma [41], and in this work, the samples were imaged and analyzed at cervical epithelial area, thus the main fluorophore here is NAD(P)H, while extracellular matrix components (collagens type I and III and elastin) play a less relevant role, as reported by previous cervical tissue fluorescence spectra studies [42–44].
When FLIM images were collected with a 532 nm long-pass filter for FAD, the signals of PpIX could be detected. But in our previous work, the fluorescence emission of PpIX at around 620 nm for the cervical tissue samples was very weak [45] and the effect of PpIX signals in cervical tissues is little.
For both standard and tissue samples, fluorescence signals were collected by a 430 nm long-pass filter. Each tissue sample containing intact cervical epithelium was imaged at the cervical epithelial area for 6-10 regions. When FLIM images of tissues were collected with a 430 nm long-pass filter, the phasor FLIM data are mainly the combination of NAD(P)H and FAD.
The fluorescence signal was collected by a cooled PMT (PMC-100-1, Becker & Hickl, Germany) with dark counts of 40 at 22°C. Each FLIM image was acquired within 1.5 min. The instrument response function (IRF) was measured under the same conditions as the FLIM observation by detecting scattering of excitation light with non-fluorescent TiO2 nanoparticles under the microscope.
2.5 Data analysis
Fluorescence lifetime data can be analyzed by multi-exponential fitting with the convolution of the IRF using SPCImage software (Becker & Hickl, Germany). The mean lifetime of each pixel was calculated by the equation:
where τi is the lifetime of one fluorescent component and ai is the corresponding contribution of the exponential component.In this work, triple-exponential function was used in the calculation of τm when the FLIM images were collected by a 430 nm long-pass filter, otherwise, double-exponential decay function was applied for NAD(P)H or FAD signals. The τm distribution curve of the 256 × 256 pixels of the FLIM image was then obtained with SPCImage software as previously reported [34].
Furthermore, each FLIM image was converted to a scatter-plot phasor FLIM, where every pixel of the FLIM image was converted to one pixel of the phasor plot. The coordinates of phasor pixels were defined by the following expression [46]:
In a multi-component fluorescent lifetime system, the phasor plot coordinates can be calculated using the following formulas [46]:
The phasor approach transforms the time-domain data to a normalized phase domain by calculating the discrete Fourier transformation numerically. The vector in phase domain originated in (0,0) and points towards the half circle (centrum at (0.5,0), radius = 0.5), which was named “universal circle”. If the fluorescence decay is a single-exponential function, the data of phasor is on the semicircle. Besides, short lifetimes of fluorophores are close to the point (1,0), while long lifetimes are close to the point (0,0) [27,46]. Generally, samples with multi-exponential decay, such as tissues or cells, are inside the semicircle.
In this work, the acquired FLIM data were imported into the SimFCS software (Laboratory for Fluorescence Dynamics, Irvine, CA, USA) for processing as follows, with Rhodamine B solution selected as the reference sample. The fluorescence decays of Rhodamine B solution detected by this system were simulated with the convolution of the IRF with a single-exponential decay function. The obtained fluorescence lifetime was 2.0 ns, which was consistent with previous work [47]. The Rhodamine B FLIM image was loaded in SimFCS software as a referenced image, and the fitting parameters were used as criteria for analyzing other FLIM images. Then, each FLIM image of tissue or cell samples was converted into a phasor scatter plot by the software (Fig. 1(C) and (D) is shown as an example). In this work, the intensity of background signals is < 1% of the fluorescence emitted by cells or tissues, thus the background correction need not be considered [48].
According to the density distribution of the scatter plot, the densest position was selected as the central coordinates of each phasor diagram, which were then used for statistical analysis. To obtain a density map of the central coordinates statistical distribution, Matlab software (The MathWorks, USA) was used to equally divide the phasor range of interest into 8 × 8 sub-areas. The number of central coordinates in each sub-area was used to draw a bivariate histogram plot with bilinear color interpolation. In the density graphic, white, blue, yellow and red colors represent successive increases in density.
3. Results
3.1 Phasor FLIM analysis of cervical tissues
Figure 2(A-C) shows FLIM images with a 430 nm long-pass filter of cervical epithelium from a normal person and two patients who had been respectively diagnosed with CIN1 and CIN3. Clear cervical tissue structures can be observed in all FLIM images. The cervical tissue from a normal person showed a clear epithelial cell layer structure and a distinct basal cell boundary (Fig. 2(A)). Figure 2(B) (CIN1) showed about one third of the epithelium having mild dysplasia with abnormal cells. The FLIM image of CIN3 showed that the abnormally shaped cells gradually invaded the epidermis from the basal layer boundary, and the cells were denser and more disordered (Fig. 2(C)). Figure 2(D) shows the phasor distribution of normal, CIN1 and CIN3 samples, in which CIN3 is located at the lower right portion, CIN1 at the upper left, normal sample at the upper right. The central coordinates of the highest density were quite different from each other, which demonstrated that this method could be used to distinguish between low-risk and high-risk samples in a phasor plot.
Fig. 2. FLIM images with a 430 nm long-pass filter of (A) a normal cervical tissue sample, (B) a CIN1 sample, and (C) a CIN3 sample. Scale bar: 50 µm. (D) Phasor FLIM plot of the FLIM images A, B and C. (E) The density graphic of the central coordinates (n=133) of phasor FLIM analysis obtained from 24 patients, including normal (n=14), CIN1 (n=29), and CIN2 and CIN3 (n=90). White, blue, yellow and red colors represent successive increases in density.
For further statistical analysis, we applied phasor analysis to 6 − 10 FLIM figures for each slice, and the central coordinates for each phasor were collected. The density map for the center coordinates is presented in Fig. 2(E) for a total of 24 patients. There were two patients in the normal group, three patients in the low-risk CIN1 group, and 19 in the high-risk group, including 2 CIN2 and 17 CIN3. In Fig. 2(E), the data from three groups can be clearly distinguished. The data from normal tissues were primarily distributed to the upper middle, and the central coordinates of this area were roughly (0.590, 0.508). The data corresponding to low-risk CIN1 are mainly observed at the upper left, with central coordinates of (0.490, 0.507). The data from high-risk lesions—including CIN2 and CIN 3—are principally concentrated to the lower right, and the coordinates of the central region were (0.658, 0.472). These three parts can be clearly distinguished in the phasor map, which further demonstrated that phasor FLIM has the ability to distinguish low and high grades of cervical precancerous lesions from normal tissues. Among them, the distribution of CIN1 data was on the left side of normal. This may result from the fact that CIN1 was usually caused by a combination of low-risk HPV and high-risk or only low-risk HPV infections, but CIN2 and CIN3 are primarily caused by persistent infection with high-risk HPV [49]. CIN1 has a high possibility of natural recovery, which may imply that its inherent cellular metabolic changes are different from those of CIN2 and CIN3. The data derived from high-risk lesions CIN2 and CIN 3 are at the lower right portion of Fig. 2(E). This may indicate a phasor data shifting to the bottom right for cervical lesions due to the requirement of rapid cell division.
Based on the phasor analysis results (Fig. 2(E)), a line of s=0.490 can be applied to distinguish HSIL in cervical tissues. Most of the normal and low-risk CIN1 slice data are mainly above the line s=0.490 (62.8%, 27/43), while those from high-risk CIN2 and CIN3 are principally located below the line (80.0%, 72/90); this can then be used as an effective approach for the diagnosis of high-grade precancerous lesions in cervical tissues.
3.2 High sensitivity of phasor FLIM
For a CIN2 or CIN3 patient, it is often possible that lesion areas are relatively small. As shown in Fig. 3, we obtained and examined two different sites of a CIN3 tissue sample by FLIM with a 430 nm long-pass filter, one from the lesion center and the other from a healthy site. A clear epithelial structure distribution appeared in the healthy sites of CIN3 (Fig. 3(A)), while the image from the lesion site showed obvious changes in pathologic structure (Fig. 3(B)). If only the healthy portion of CIN3 was used for histopathological examination, a misdiagnosis might result.
Fig. 3. Two FLIM images with a 430 nm long-pass filter from (A) a healthy site and (B) a lesion site in one CIN3 patient. Scale bar: 50 µm. (C) On the basis of Fig. 2(E), 34 scatter points were added. Each scatter point represents the central coordinates of phasor analysis for each healthy site slice from five CIN3 patients.
In Fig. 3(C), 34 black dots were added on the basis of Fig. 2(E). These black dots represent the phasor data obtained from 34 FLIM images of 5 CIN3 patients at healthy tissue sites. It can be seen that the black dots of the healthy sites from CIN3 patients in the phasor domain are all 100% (34/34) below the boundary line s=0.490. We suggest that for high-risk precancerous lesions, even the detected areas of tissues may appear normal in histopathological structure; and that the phasor approach can distinguish abnormal metabolism in cervical tissues.
3.3 Metabolic pathways in HcerEpic cells
To further explore the pathways of tumor metabolism, either CoCl2 or 3-bromopyruvate (3-BP) was used to inhibit specific metabolic pathways in normal cervical epithelial cells (HcerEpic). CoCl2 can inhibit oxidative phosphorylation [24,25] and 3-BP can inhibit glycolysis [26,27]. Figure 4 shows FLIM images of control cells (untreated) and cells co-incubated with CoCl2 or 3-BP, respectively. We demonstrated that the lifetime for NAD(P)H (with a 447 ± 30 nm band-pass filter) and FAD (with a 532 nm long-pass filter) in CoCl2-treated cells decreased when compared to the control group. In contrast, the lifetime of NAD(P)H and FAD increased in 3-BP-treated cells. The phasor plots (Fig. 4(D), (H) and (L)) revealed this tendency clearly. The lifetimes of autofluorescence in phasor plots all shifted toward the right compared with normal data when the cells were treated with CoCl2, which was consistent with the trend observed in normal to CIN3 tissues in Fig. 2(D). The above results suggest that when cervical tissue develops high-grade lesions, the intracellular metabolic pathway tends to switch to anaerobic glycolysis. This is also consistent with various studies [50–54] showing that cancer cells with high metabolic demand increased their rate of glycolysis under hypoxic conditions.
Fig. 4. FLIM images of normal cervical epithelial cells (HcerEpic) with different treatments. (A-C) total autofluorescence with a 430 nm long-pass filter, (E-G) NAD(P)H with a 447 ± 30 nm band-pass filter, (I-K) FAD with a 532 nm long-pass filter. Scale bar: 10 µm. (D, H, L) the corresponding phasor plots, in which green represents control, orange represents CoCl2-treated, and blue represents 3-BP-treated cells.
Figure 5(A) shows the distribution of several standard samples in the phasor domain that are known as endogenous fluorophores in cells or tissues, including NADH, FAD, elastin, and PpIX [55]. The central coordinates in the phasor FLIM diagram are (0.981, 0.130) for free NADH, (0.606, 0.487) for free FAD, (0.503, 0.412) for elastin, (0.378, 0.459) for MDH-bound NADH, and (0.087, 0.352) for PpIX. It can be observed that the phasor distribution for standard samples containing a single substance was basically on the circle, which was consistent with the general experimental results [27].
Fig. 5. (A) Phasor FLIM histogram of standard samples. (B) Phasor plot with the color scale representing the switch between bound and free NADH.
Figure 5(B) shows a phasor histogram of HcerEpic cells with different metabolic pathways, and the color scale represents the switch between bound and free NADH. This can be readily observed as the increased shift toward glycolysis leads the phasor data toward free NADH. Our standard sample results—as well as the experimental cell results— further indicated the dominance of the glycolytic pathway over oxidative phosphorylation in high-grade cervical lesions.
4. Discussion
Current primary screening for cervical cancer has been using the Papanicolaou (Pap) smear test, followed by a colposcopically-directed biopsy and subsequent histopathological examination if the smear is abnormal. The current method needs sample staining and it takes at least two visits and 2-6 weeks for a patient, which is time- and resource-consuming. The sensitivity of Pap test was reported only 60-85%, and the specificity was estimated 95-98% [3]. For HSIL samples, Pap test sensitivity is in the range of 70% to 80% [56]. On the other hand, cervical biopsies highly rely on proper biopsy-site selection. Especially, Small size of a lesion can lead to false-negative results. As reported by Cohen’s group, factors that limit sensitivity include small size of a lesion, inaccessible location of the lesion, the lesion not being sampled, or small size of the abnormal cells [7].
As a comparison in the present work, 34 areas at healthy tissue sites from 5 CIN3 patients were measured by phasor FLIM. If these samples were stained with H&E and examined by histopathological diagnosis, all of them would be determined negative since no morphological changes in tissues have occurred. In Fig. 3, all the data 100% (34/34) is below the boundary line, indicating the biochemistry of the tissue changes. Therefore, this result suggests that the phasor FLIM approach may be sensitive for early diagnosis of cervical precancer, especially for patients who have relatively small lesions. That is, if samples come from random collection sites, the phasor method has reliable sensitivity and does not strictly depend on the location and size of lesions. Furthermore, in order to avoid misdiagnosis using histopathological methods, additional slices must be examined at different sites of excised tissue for patients underwent cervical conization or total hysterectomy. The use of the phasor FLIM approach may in the future also greatly reduce the workload of the pathologist.
In Fig. 5, considering that the s coordinates of both free and bound NAD(P)H were lower than that of free FAD, the boundary line s=0.490 suggests that the NAD(P)H contribution to the HSIL group increased relative to normal or LSIL. This result relates to redox ratio as defined by Chance et al. [57], which is the intensity ratio of FAD/NAD(P)H. Several investigators demonstrated similar results of higher NAD(P)H and lower FAD fluorescence intensity in different cancerous diseases, such as breast cancer [58] and bladder cancer [12]. Recently, Alam et al. [59] and Wallrabe et al. [60] defined a fluorescence-lifetime redox ratio of the fractional amplitudes of the bound decay components of NAD(P)H and FAD. The advantage of the FLIRR is that it does not depend on the fluorescence intensity of NAD(P)H and FAD in the cells or tissues. Since the phasor FLIM was studied based on the autofluorescence of tissues with a 430 nm long-pass filters, the phasor results indicated the changes of the combination of NAD(P)H and FAD, and the overall state of the tissues.
There are several established optical technologies for cancer detection, such as confocal micro-endoscopy, reflectance spectroscopy and fluorescence spectroscopy, optical coherence tomography. The big advance of these optical technologies, as well as the present phasor FLIM, is real-time, fast and sensitive [61]. Confocal micro-endoscopy is sensitive on the changes in nucleus of the cells that reflects abnormal duplication and growth. Reflectance spectroscopy and optical coherence tomography have remarkable performance when there are abnormal blood vessels for vigorous growth [61]. As for changes of biochemistry of the cells and tissues, fluorescence spectroscopy and FLIM both show great potential for cancer detection. This work provides the basis for the development of minimally invasive optical screening, and can be additive to current methods by increasing the sensitivity. Meanwhile, highly sensitivity on abnormal metabolism in tissues might lead to false-positive on inflammation or benign tumors, which need to be further studied in the future.
5. Conclusions
We performed a phasor FLIM study on human cervical tissues that can distinguish among different stages of cervical lesions from low risk to high risk. Experiments designed to inhibit specific metabolic pathways in HcerEpic cells also confirmed that the glycolytic pathway caused the cellular phasor data to move toward the lower right in the phasor domain. Our data agreed with a trend for high-risk CIN2 and CIN3 compared to normal tissue data in phasor plots. Thus, we conclude that high-risk CIN2 and CIN3 tended to manifest a higher rate of glycolysis in overall metabolic pathways. In addition, the phasor approach possessed high sensitivity, especially for healthy sites in CIN3 tissues. We herein propose a highly adaptive, sensitive, and rapid diagnostic tool that exhibits great potential for cervical precancer diagnosis in the future.
Funding
National Natural Science Foundation of China (11574056, 61575046); Ministry of Science and Technology of the People's Republic of China (SINO-SERBIA2018002); Fudan University-CIOMP Joint Fund (FC2017-007, FC2018-001); Pioneering Project of Academy for Engineering and Technology, Fudan University (gyy2018-001, gyy2018-002).
Acknowledgements
The authors acknowledge all the patients involved in this study.
Disclosures
The authors declare no conflicts of interest.
References
1. J. Gu, C. Y. Fu, B. K. Ng, S. Gulam Razul, and S. K. Lim, “Quantitative diagnosis of cervical neoplasia using fluorescence lifetime imaging on haematoxylin and eosin stained tissue sections,” J. Biophotonics 7(7), 483–491 (2014). [CrossRef]
2. J. Gu, C. Y. Fu, B. K. Ng, and S. K. Lim, “Quantitative diagnosis of cervical precancer using fluorescence intensity and lifetime imaging from the stroma,” in Photonics Global Conf., (IEEE, 2012), 1–5.
3. I. M. Orfanoudaki, D. Kappou, and S. Sifakis, “Recent advances in optical imaging for cervical cancer detection,” Arch. Gynecol. Obstet. 284(5), 1197–1208 (2011). [CrossRef]
4. H. Kelly, H. A. Weiss, Y. Benavente, S. de Sanjose, P. Mayaud, Y.-l. Qiao, R.-M. Feng, H. DeVuyst, V. Tenet, and A. Jaquet, “Association of antiretroviral therapy with high-risk human papillomavirus, cervical intraepithelial neoplasia, and invasive cervical cancer in women living with HIV: a systematic review and meta-analysis,” Lancet HIV 5(1), e45–e58 (2018). [CrossRef]
5. K. Tainio, A. Athanasiou, K. A. Tikkinen, R. Aaltonen, J. Cárdenas, S. Glazer-Livson, M. Jakobsson, K. Joronen, M. Kiviharju, and K. Louvanto, “Clinical course of untreated cervical intraepithelial neoplasia grade 2 under active surveillance: systematic review and meta-analysis,” BMJ 360, k499 (2018). [CrossRef]
6. X. H. Tao, Y. Guan, D. Shao, W. Xue, F. S. Ye, M. Wang, and M. H. He, “Efficacy and safety of photodynamic therapy for cervical intraepithelial neoplasia: a systemic review,” Photodiagn. Photodyn. Ther. 11(2), 104–112 (2014). [CrossRef]
7. D. Saslow, C. D. Runowicz, D. Solomon, A. B. Moscicki, and C. Cohen, “American Cancer Society Guideline for the Early Detection of Cervical Neoplasia and Cancer,” J. Low. Genit. Tract. Dis. 7(2), 67–86 (2003). [CrossRef]
8. W. Small, M. A. Bacon, A. Bajaj, L. T. Chuang, B. J. Fisher, M. M. Harkenrider, A. Jhingran, H. C. Kitchener, L. R. Mileshkin, and A. N. Viswanathan, “Cervical cancer: A global health crisis,” Cancer 123(13), 2404–2412 (2017). [CrossRef]
9. R. J. DeBerardinis, J. J. Lum, G. Hatzivassiliou, and C. B. Thompson, “The biology of cancer: metabolic reprogramming fuels cell growth and proliferation,” Cell Metab. 7(1), 11–20 (2008). [CrossRef]
10. B. G. Wouters and M. Koritzinsky, “Hypoxia signalling through mTOR and the unfolded protein response in cancer,” Nat. Rev. Cancer 8(11), 851–864 (2008). [CrossRef]
11. A. Marusyk and K. Polyak, “Tumor heterogeneity: causes and consequences,” Biochim. Biophys. Acta 1805(1), 105–117 (2010). [CrossRef]
12. S. Palmer, K. Litvinova, E. U. Rafailov, and G. Nabi, “Detection of urinary bladder cancer cells using redox ratio and double excitation wavelengths autofluorescence,” Biomed. Opt. Express 6(3), 977–986 (2015). [CrossRef]
13. L. Liu, Q. Yang, M. Zhang, Z. Wu, and P. Xue, “Fluorescence lifetime imaging microscopy and its applications in skin cancer diagnosis,” J. Innovative Opt. Health Sci. 12(05), 1930004 (2019). [CrossRef]
14. J. V. Rocheleau, W. S. Head, and D. W. Piston, “Quantitative NAD (P) H/flavoprotein autofluorescence imaging reveals metabolic mechanisms of pancreatic islet pyruvate response,” J. Biol. Chem. 279(30), 31780–31787 (2004). [CrossRef]
15. R. P. Mecham, “Elastin in lung development and disease pathogenesis,” Matrix Biol. 73, 6–20 (2018). [CrossRef]
16. Y. Cui, X. Zhang, K. You, Y. Guo, and L. Geng, “Nanomechanical Characteristics of Cervical Cancer and Cervical Intraepithelial Neoplasia Revealed by Atomic Force Microscopy,” Med. Sci. Monit. 23, 4205–4213 (2017). [CrossRef]
17. K. Takahashi, N. Ikeda, N. Nonoguchi, Y. Kajimoto, S.-I. Miyatake, Y. Hagiya, S.-I. Ogura, H. Nakagawa, T. Ishikawa, and T. Kuroiwa, “Enhanced expression of coproporphyrinogen oxidase in malignant brain tumors: CPOX expression and 5-ALA–induced fluorescence,” Neuro-Oncology 13(11), 1234–1243 (2011). [CrossRef]
18. G. A. Wagnieres, W. M. Star, and B. C. Wilson, “In Vivo Fluorescence Spectroscopy and Imaging for Oncological Applications,” Photochem. Photobiol. 68(5), 603–632 (1998). [CrossRef]
19. M. C. Skala, K. M. Riching, A. Gendron-Fitzpatrick, J. Eickhoff, K. W. Eliceiri, J. G. White, and N. Ramanujam, “In vivo multiphoton microscopy of NADH and FAD redox states, fluorescence lifetimes, and cellular morphology in precancerous epithelia,” Proc. Natl. Acad. Sci. U. S. A. 104(49), 19494–19499 (2007). [CrossRef]
20. M. A. Yaseen, J. Sutin, W. Wu, B. Fu, H. Uhlirova, A. Devor, D. A. Boas, and S. Sakadžic, “Fluorescence lifetime microscopy of NADH distinguishes alterations in cerebral metabolism in vivo,” Biomed. Opt. Express 8(5), 2368–2385 (2017). [CrossRef]
21. J. L. Lagarto, B. T. Dyer, C. B. Talbot, N. S. Peters, P. M. W. French, A. R. Lyon, and C. Dunsby, “Characterization of NAD(P)H and FAD autofluorescence signatures in a Langendorff isolated-perfused rat heart model,” Biomed. Opt. Express 9(10), 4961–4978 (2018). [CrossRef]
22. R. Leben, M. Köhler, H. Radbruch, A. E. Hauser, and R. A. Niesner, “Systematic enzyme mapping of cellular metabolism by phasor-analyzed label-free NAD (P) H fluorescence lifetime imaging,” Int. J. Mol. Sci. 20(22), 5565 (2019). [CrossRef]
23. X. Liu, D. Lin, W. Becker, J. Niu, B. Yu, L. Liu, and J. Qu, “Fast fluorescence lifetime imaging techniques: A review on challenge and development,” J. Innovative Opt. Health Sci. 12(05), 1930003 (2019). [CrossRef]
24. J. L. Lagarto, B. T. Dyer, N. S. Peters, P. M. W. French, C. Dunsby, and A. R. Lyon, “In vivo label-free optical monitoring of structural and metabolic remodeling of myocardium following infarction,” Biomed. Opt. Express 10(7), 3506–3521 (2019). [CrossRef]
25. Y. Wang, C. Song, M. Wang, Y. Xie, L. Mi, and G. Wang, “Rapid, label-free, and highly sensitive detection of cervical cancer with fluorescence lifetime imaging microscopy,” IEEE J. Sel. Top. Quantum Electron. 22(3), 228–234 (2016). [CrossRef]
26. V. R. Caiolfa, M. Zamai, G. Malengo, A. Andolfo, C. D. Madsen, J. Sutin, M. A. Digman, E. Gratton, F. Blasi, and N. Sidenius, “Monomer–dimer dynamics and distribution of GPI-anchored uPAR are determined by cell surface protein assemblies,” J. Cell Biol. 179(5), 1067–1082 (2007). [CrossRef]
27. C. Stringari, A. Cinquin, O. Cinquin, M. A. Digman, P. J. Donovan, and E. Gratton, “Phasor approach to fluorescence lifetime microscopy distinguishes different metabolic states of germ cells in a live tissue,” Proc. Natl Acad. Sci. USA 108(33), 13582–13587 (2011). [CrossRef]
28. O. Golfetto, E. Hinde, and E. Gratton, “Laurdan fluorescence lifetime discriminates cholesterol content from changes in fluidity in living cell membranes,” Biophysical J. 104(6), 1238–1247 (2013). [CrossRef]
29. T. Luo, Y. Lu, S. Liu, D. Lin, and J. Qu, “Phasor–FLIM as a Screening Tool for the Differential Diagnosis of Actinic Keratosis, Bowen’s Disease, and Basal Cell Carcinoma,” Anal. Chem. 89(15), 8104–8111 (2017). [CrossRef]
30. T. Luo, Y. Lu, S. Liu, D. Lin, and J. Qu, “Enhanced Visualization of Hematoxylin and Eosin Stained Pathological Characteristics by Phasor Approach,” Anal. Chem. 89(17), 9224–9231 (2017). [CrossRef]
31. W. K. Huh, R. M. Cestero, F. A. Garcia, M. A. Gold, R. S. Guido, K. McIntyre-Seltman, D. M. Harper, L. Burke, S. T. Sum, and R. F. Flewelling, “Optical detection of high-grade cervical intraepithelial neoplasia in vivo: results of a 604-patient study,” Am. J. Obstet. Gynecol. 190(5), 1249–1257 (2004). [CrossRef]
32. M. Kim, S. Ishioka, T. Endo, T. Baba, and T. Saito, “Obstetrical prognosis of patients with cervical intraepithelial neoplasia (CIN) after “coin-shaped” conization,” Arch. Gynecol. Obstet. 293(3), 651–657 (2016). [CrossRef]
33. F. Narducci, B. Occelli, F. Boman, D. Vinatier, and J.-L. Leroy, “Positive margins after conization and risk of persistent lesion,” Gynecol. Oncol. 76(3), 311–314 (2000). [CrossRef]
34. X. Wang, Y. Xie, M. Huang, L. Yao, Y. Wang, Y. Fei, J. Ma, and L. Mi, “Effect of Fixation and Mounting on Fluorescence Lifetime of Cellular Autofluorescence,” IEEE J. Sel. Top. Quantum Electron. 25(1), 1–6 (2019). [CrossRef]
35. D. Y. Hwang and F. Ismail-Beigi, “Glucose uptake and lactate production in cells exposed to CoCl2 and in cells overexpressing the Glut-1 glucose transporter,” Arch. Biochem. Biophys. 399(2), 206–211 (2002). [CrossRef]
36. M. H. Koseoglu and F. I. Beigi, “Mechanism of stimulation of glucose transport in response to inhibition of oxidative phosphorylation: analysis with myc-tagged Glut1,” Mol. Cell. Biochem. 194(1/2), 109–116 (1999). [CrossRef]
37. S. Matsumoto, K. Saito, H. Yasui, H. D. Morris, J. P. Munasinghe, M. Lizak, H. Merkle, J. H. Ardenkjaer-Larsen, R. Choudhuri, and N. Devasahayam, “EPR oxygen imaging and hyperpolarized 13C MRI of pyruvate metabolism as noninvasive biomarkers of tumor treatment response to a glycolysis inhibitor 3-bromopyruvate,” Magn. Reson. Med. 69(5), 1443–1450 (2013). [CrossRef]
38. L. S. Ihrlund, E. Hernlund, O. Khan, and M. C. Shoshan, “3-Bromopyruvate as inhibitor of tumour cell energy metabolism and chemopotentiator of platinum drugs,” Mol. Oncol. 2(1), 94–101 (2008). [CrossRef]
39. Y. Pu, G. C. Tang, W. Wang, H. E. Savage, S. P. Schantz, and R. R. Alfano, “Native Fluorescence Spectroscopic Evaluation of Chemotherapeutic Effects on Malignant Cells using Nonnegative Matrix Factorization Analysis,” Technol. Cancer Res. Treat. 10(2), 113–120 (2011). [CrossRef]
40. Q. Fang, T. Papaioannou, J. A. Jo, R. Vaitha, K. Shastry, and L. Marcu, “Time-domain laser-induced fluorescence spectroscopy apparatus for clinical diagnostics,” Rev. Sci. Instrum. 75(1), 151–162 (2004). [CrossRef]
41. S. C. Tu and K. B. Sung, “Extracting fluorescence efficiency and emission spectra of cervical tissue,” in Frontiers in Optics / Laser Science, OSA Technical Digest (Optical Society of America, 2018), JTu3A.107.
42. R. Drezek, K. Sokolov, U. Utzinger, I. Boiko, A. Malpica, M. Follen, and R. Richards-Kortum, “Understanding the contributions of NADH and collagen to cervical tissue fluorescence spectra: Modeling, measurements, and implications,” J. Biomed. Opt. 6(4), 385–396 (2001). [CrossRef]
43. S. Devi, P. K. Panigrahi, and A. Pradhan, “Detecting cervical cancer progression through extracted intrinsic fluorescence and principal component analysis,” J. Biomed. Opt. 19(12), 127003 (2014). [CrossRef]
44. C. K. Brookner, M. Follen, I. Boiko, J. Galvan, and R. Richards-Kortum, “Autofluorescence Patterns in Short-Term Cultures of Normal Cervical Tissue,” Photochem. Photobiol. 71(6), 730–736 (2000). [CrossRef]
45. Y. Jing, Y. Wang, X. Wang, C. Song, J. Ma, Y. Xie, Y. Fei, Q. Zhang, and L. Mi, “Label-free imaging and spectroscopy for early detection of cervical cancer,” J. Biophotonics 11(5), e201700245 (2018). [CrossRef]
46. M. A. Digman, V. R. Caiolfa, M. Zamai, and E. Gratton, “The Phasor Approach to Fluorescence Lifetime Imaging Analysis,” Biophys. J. 94(2), L14–L16 (2008). [CrossRef]
47. R. Mercadé-Prieto, L. Rodriguez-Rivera, and X. D. Chen, “Fluorescence lifetime of Rhodamine B in aqueous solutions of polysaccharides and proteins as a function of viscosity and temperature,” Photochem. Photobiol. Sci. 16(11), 1727–1734 (2017). [CrossRef]
48. H. Szmacinski, V. Toshchakov, and J. R. Lakowicz, “Application of phasor plot and autofluorescence correction for study of heterogeneous cell population,” J. Biomed. Opt. 19(4), 046017 (2014). [CrossRef]
49. A. C. Vidal, J. S. Smith, F. Valea, R. Bentley, M. Gradison, K. S. Yarnall, A. Ford, F. Overcash, K. Grant, and S. K. Murphy, “HPV genotypes and cervical intraepithelial neoplasia in a multiethnic cohort in the southeastern USA,” Cancer, Causes Control 25(8), 1055–1062 (2014). [CrossRef]
50. G. Solaini, A. Baracca, G. Lenaz, and G. Sgarbi, “Hypoxia and mitochondrial oxidative metabolism,” Biochim. Biophys. Acta 1797(6-7), 1171–1177 (2010). [CrossRef]
51. H. Zhang, P. Gao, R. Fukuda, G. Kumar, B. Krishnamachary, K. I. Zeller, C. V. Dang, and G. L. Semenza, “HIF-1 inhibits mitochondrial biogenesis and cellular respiration in VHL-deficient renal cell carcinoma by repression of C-MYC activity,” Cancer Cell 11(5), 407–420 (2007). [CrossRef]
52. H. Zhang, M. Bosch-Marce, L. A. Shimoda, Y. S. Tan, J. H. Baek, J. B. Wesley, F. J. Gonzalez, and G. L. Semenza, “Mitochondrial autophagy is an HIF-1-dependent adaptive metabolic response to hypoxia,” J. Biol. Chem. 283(16), 10892–10903 (2008). [CrossRef]
53. S. R. Kantelhardt, D. Kalasauskas, K. König, E. Kim, M. Weinigel, A. Uchugonova, and A. Giese, “In vivo multiphoton tomography and fluorescence lifetime imaging of human brain tumor tissue,” J. Neuro-Oncol. 127(3), 473–482 (2016). [CrossRef]
54. V. K. Ramanujan, J.-H. Zhang, E. Biener, and B. A. Herman, “Multiphoton fluorescence lifetime contrast in deep tissue imaging: prospects in redox imaging and disease diagnosis,” J. Biomed. Opt. 10(5), 051407 (2005). [CrossRef]
55. A. Croce and G. Bottiroli, “Autofluorescence spectroscopy and imaging: a tool for biomedical research and diagnosis,” Eur. J. Histochem. 58(4), 2461 (2014). [CrossRef]
56. M. E. Sherman, M. Schiffman, R. Herrero, D. Kelly, C. Bratti, L. J. Mango, M. Alfaro, M. L. Hutchinson, F. Mena, and A. Hildesheim, “Performance of a semiautomated papanicolaou smear screening system : Results of a population-based study conducted in Guanacaste, Costa Rica,” Cancer 84(5), 273–280 (1998). [CrossRef]
57. B. Chance, B. Schoener, R. Oshino, F. Itshak, and Y. Nakase, “Oxidation-reduction ratio studies of mitochondria in freeze-trapped samples. NADH and flavoprotein fluorescence signals,” J. Biol. Chem. 254, 4764–4771 (1979).
58. A. Uppal and P. K. Gupta, “Measurement of NADH concentration in normal and malignant human tissues from breast and oral cavity,” Biotechnol. Appl. Biochem. 37(1), 45–50 (2003). [CrossRef]
59. S. R. Alam, H. Wallrabe, Z. Svindrych, A. K. Chaudhary, K. G. Christopher, D. Chandra, and A. Periasamy, “Investigation of Mitochondrial Metabolic Response to Doxorubicin in Prostate Cancer Cells: An NADH, FAD and Tryptophan FLIM Assay,” Sci. Rep. 7(1), 10451 (2017). [CrossRef]
60. H. Wallrabe, Z. Svindrych, S. R. Alam, K. H. Siller, T. Wang, D. Kashatus, S. Hu, and A. Periasamy, “Segmented cell analyses to measure redox states of autofluorescent NAD(P)H, FAD & Trp in cancer cells by FLIM,” Sci. Rep. 8(1), 1 (2018). [CrossRef]
61. B. Hill, S. Lam, P. Lane, C. MacAulay, L. Fradkin, and M. Follen, “Established and Emerging Optical Technologies for the Real-Time Detection of Cervical Neoplasia: A Review,” J. Cancer Ther. 08(13), 1241–1278 (2017). [CrossRef]
