Background
Early gastric cancer is defined as gastric cancer that is confined to the mucosa or submucosa, regardless of the presence or absence of regional lymph node metastasis [
1,
2]. The prognosis for early gastric cancer is universally excellent, and strategies to improve prognosis is strictly dependent on earlier detection and accurate diagnosis since early gastric cancer can potentially be cured by endoscopic therapy such as endoscopic mucosal resection or endoscopic submucosal dissection [
3,
4]. It was reported that patients with early gastric cancer after treatment had a 1 year survival in excess of 90% [
5]. However, there is little information on the symptoms of early gastric cancer and it is often difficult to detect gastric superficial lesions using conventional endoscopy with white-light imaging [
6]. At present, the diagnosis of these lesions is always based on the pathologic assessment of endoscopic biopsy specimens. As such, superficially taken biopsies and sampling error for histopathology are common events, making it difficult to draw firm conclusions [
7]. Hence, development of a new diagnostic technique will be of important clinical meaning.
Multiphoton imaging technique such as two-photon excitation fluorescence (TPEF) and second-harmonic generation (SHG) has been widely used for biological tissue imaging [
8,
9]. TPEF exploits the auto-fluorescence of biological samples and therefore obviates the need for exogenous contrast agents, and SHG makes use of the non-centrosymmetric properties to image structural proteins and will not suffer from phototoxicity effects or photobleaching because it does not involve excitation of molecules [
10‐
12]. There are a variety of intracellular molecules including NADH, FAD and porphyrins, as well as certain extracellular components such as elastin and collagen within gastric tissues, which can generate intrinsic multiphoton signals. Therefore, in this study, we investigated the potential of using label-free, multimodal multiphoton microscopy that incorporates two-photon excited fluorescence and second-harmonic generation techniques for distinguishing early gastric cancer from normal tissues.
Methods
Imaging instrumentation
As has been described previously, a mode-locked femtosecond Ti:sapphire laser (Mira 900-F, Coherent, Inc., USA) was used as the multiphoton excitation source, and the excitation light was delivered to and the emitted light was also collected from the sample through an inverted microscope (LSM 510 META, Zeiss, Germany) [
13,
14]. A 63× Zeiss Plan-Apochromat oil immersion objective lens (Numerical Aperture (NA) =1.4) was chosen for capturing multiphoton microscopic images. The backscattered intrinsic TPEF and SHG signals were received respectively by a META detector composed of a reflective grating and an optimized 32-channel PMT array detector using 810 nm excitation wavelength. TPEF signal was detected in the wavelength range 430–716 nm and SHG signal was detected in the 389 to 419 nm wavelength range. In order to provide a visual contrast of multiphoton images, all the TPEF images adopted pseudo-colored red and SHG images were marked with pseudo-colored green.
Sample preparation
In this work, 12 early gastric cancer samples including 1 well-differentiated adenocarcinoma, 6 moderately-differentiated adenocarcinomas, 1 moderately-poorly differentiated adenocarcinoma and 4 poorly-differentiated adenocarcinomas were collected. All samples were immediately sent to the pathology laboratory for frozen section through a cryostat microtome once they were obtained from the surgeons. Three consecutive sections with 10 μm thickness were used in this study, in which two slices were used to carry out the multiphoton microscopic imaging and the middle section was diagnosed through hematoxylin and eosin staining (H&E) in order to further determine the experimental results. In addition, we also collected 12 normal gastric tissue samples for the sake of comparison. One thing to be aware of was that in order to avoid specimen shrinkage or dehydration, we added a small amount of phosphate-buffered saline (PBS) to the tissue sections during the experiment.
Histologic analysis
Firstly, every H&E-stained slice was checked by two certified pathologists in this study to avoid the inter-individual variability, and then a digital image of the H&E-stained slide was acquired by an optical microscope (Eclipse Ci-L, Nikon Instruments Inc., Japan) with a CCD (DS-Fi2, Nikon). Secondly, two investigators who were blinded to the diagnostic results confirmed all the multiphoton imaging results by comparing with the H&E-stained digital images.
Statistical analysis
Furthermore, in the development of early gastric cancer, we measured the circumference of cell nucleus and SHG average intensity per pixel so that the changes of cell size and collagen content in mucosa could be quantitatively estimated. For every multiphoton image, SHG average intensity per pixel was obtained via dividing the sum of all intensities by the total number of pixels [
15]. The results were presented as a mean followed with its standard deviation (mean ± SD). In this study, we made use of the ImageJ software for analyzing SHG images, and based on the IBM SPSS Statistics 21, we also employed the student’s t-test to carry out statistical analysis, and
P value less than 0.05 was regarded as statistical significance.
Discussion
Most gastric cancer is diagnosed at an advanced stage with a poor survival rate, despite treatment including major surgery and adjuvant therapy. The outcome could be improved by early detection and treatment [
18,
19]. Because early gastric cancer is associated with a much better prognosis than advanced disease, its diagnosis is important. Endoscopic screening is widely accepted as an available approach for early gastric cancer because it is minimally invasive, safe and convenient. However, conventional endoscopy examination has limited ability to precisely identify this lesion because of the insufficient resolution [
20]. The disadvantages existing in MRT (Magnetic Resonance Imaging), CT (Computed Tomography), US (Ultrasound), and so on should not be overlooked, such as low resolution, poor contrast sensitivity, or even potential radioactive hazard [
17]. Histological evaluation of biopsy specimens is still the golden standard for diagnosing early gastric cancer, but performing biopsy procedures has several shortcomings, including sampling error, costs, risks to the patient, and the delay in obtaining results [
21]. All these factors together might lead to underdiagnosis or overdiagnosis and therefore suboptimal treatment as well as surveillance practices.
The advantages presented by multiphoton microscopy such as not only the low photodamage and photobleaching, high-resolution, high-sensitivity as well as high-contrast, but also providing label-free detection of biological tissues in real time, may help reduce sampling errors and even eliminate the need for invasive tissue removal [
22]. TPEF from NADH as well as FAD in cells and SHG from collagen have been shown to be two major intrinsic signals for gastric mucosal tissues. MPM images show that normal gastric mucosa is mainly composed of well-aligned glands and well-organized extracellular collagen network that is essential for the maintenance of gastric glands, while abnormal mucosal tissues have disordered glands with various shapes and have chaotic extracellular collagen matrix. Stromal changes will be involved in neoplastic process for providing a tumor-promoting environment for cancer progression and metastasis [
8,
23]. The presence of collagen fibers is therefore considered to influence the development of gastric cancer. As a major component of extracellular matrix, the architectural properties of collagen fibers can be obtained by SHG imaging to reflect stromal changes in abnormal tissues. Furthermore, quantitative analysis proves that the cells will be enlarged with the progression of early gastric cancer and collagen content increases because of desmoplastic response which is characterized by dense collagen growth. A pilot study showed that the diagnosis results based on MPM have high accuracy, and the sensitivity, specificity, PPV, and NPV were 91.7%, 100%, 100%, and 92.3% respectively.
Though it has been shown that collagen changes would participate in the development of cancerization [
24,
25], however, interestingly enough, some researchers reported that malignant cells may secrete extracellular enzymes to degrade the collagen fibers at the invasion front [
8,
26,
27], while some works demonstrated that tumor invasion may activate fibroblasts and cause desmoplasia, and therefore increase deposition of extracellular collagen [
28,
29]. In this study, our data showed that collagen density increases during gastric carcinoma progression, which is consistent with the previous results [
13].
The submucosa mainly consists of loose connective tissue with blood vessels, and elastic fibers as well as collagen fibers are the major components of normal connective tissue. Elastin is a common fluorescence source, and this property makes TPEF imaging a convenient tool for illustrating the microstructures of elastin fiber and blood vessel. In normal submucosal tissues, elastin fibers have a long rope-like structure, and collagen fibers are well-organized. However, in abnormal tissues, elastin fibers have almost disappeared, and collagen fibers are sparse and disrupted because malignant cells would secrete extracellular enzymes to degrade the original tissue architecture and promote tumor progression into surrounding tissues [
30].
Additionally, because of the insufficient resolution of conventional endoscopy, endoscopic surgery may fail to remove disseminated invasive cells that lie beyond the surgical resection border, which may lead to tumor recurrence and ultimately patient death. The data obtained in this work show that MPM is capable of detecting single abnormal cell and identifying stratified structure in gastric tissues, and therefore may be helpful for determining surgical margin as well as selecting an optimal treatment when it is successfully incorporated into endoscope in the near future. Although the light penetration of MPM is still limited at present, it has been improving with the development of a variety of technologies, such as a gradient index lens-based MPM or a compact and flexible MPM probe [
31,
32]. It was reported that the penetration depth of MPM could reach millimeter-order [
33]. Maybe the combination of gradient index lens and multiphoton probe is a best method to overcome the defect of image depth.
Furthermore, the exploration of multiphoton endoscope into intravital imaging that aims to translate this technique into the clinics has been performed; especially the design of compact and flexible MPM probes for future clinical applications of gastrointestinal tract has achieved significant advancement [
21,
31,
34,
35]. With miniaturization and integration of this technology [
36,
37], MPM could be incorporated into endoscopy equipment or robotic surgical systems [
38‐
40], and thus can be used in imaging during live endoscopy to diagnose gastrointestinal diseases as well as would be helpful to determine the surgical margin accurately.