Fluorescent‐guided surgery for sentinel lymph node detection in gastric cancer and carcinoembryonic antigen targeted fluorescent‐guided surgery in colorectal and pancreatic cancer

Sentinel lymph node procedures for gastric cancer resections using indocyanine green (ICG) linked to Nanocoll outperformed normal ICG but did not provide information on possible lymph node metastasis. Carcinoembryonic antigen targeted fluorescent imaging using SGM‐101 was successful in both pancreatic and colorectal cancer. A large phase III multicentre trial will soon be initiated in colorectal cancer patients.

Sentinel lymph node procedures for gastric cancer resections using indocyanine green (ICG) linked to Nanocoll outperformed normal ICG but did not provide information on possible lymph node metastasis. Carcinoembryonic antigen targeted fluorescent imaging using SGM-101 was successful in both pancreatic and colorectal cancer. A large phase III multicentre trial will soon be initiated in colorectal cancer patients. Near-infrared (NIR) photons can penetrate deeply into tissue (up to 5 to 10 mm). This is utilized by NIR fluorescence (NIRF) to provide fast and quantitative contrast images. 1,2 Fluorophores can either accumulate in tumor tissue due to either the enhanced permeability and retention (EPR) effect or by targeting a specific tumor marker.
Once activated by laser, NIRF light reflected by the fluorophore can be visualized using a dedicated imaging system. NIRF-guided surgery can help distinguish benign from malignant tumor tissue and also aid in identifying lesions or metastasis outside the standard field of resection 3 and therefore might influence clinical decision making.
Recently, research has shifted from nonspecific imaging (depending on the EPR effect) toward tumor-marker specific imaging (eg, tumor receptor targeting). NIRF imaging is currently being investigated for multiple purposes in clinical trials. This review discusses two applications of targeted and nontargeted NIRF. First, the clinical and preclinical data on NIRF-guided sentinel lymph node (SLN) identification in gastric cancer using indocyanine green (ICG) is discussed. Subsequently, the available data on carcinoembryonic antigen (CEA) targeted NIRF imaging in rectal and pancreatic cancer is reviewed.

| SLN IDENTIFICATION IN GASTRIC CANCER
Tumor resection is imperative when treating gastric cancer with curative intent. In addition to total or partial gastric resection, a standardized lymph node dissection is performed. Lymph node metastasis in gastric cancer patients is an established prognostic factor for survival. Lymph node involvement is found in 2% to 50% of patients, increasing with tumor stage. 4 Currently, a fixed number of lymph nodes is resected, regardless of the presence of lymph node metastases. However, since unnecessary lymph node dissection. During a SLN procedure only the primary draining lymph nodes (which are at the highest risk of containing tumor metastases) are identified and dissected. Whereas histological assessment on hematoxylin and eosin (H&E) stained slides can accurately identify metastases for most cancer types, it has been shown that absence of visible tumor cells on H&E slides of gastric cancer lymph nodes does not guarantee absence of micrometastases. [7][8][9][10][11] To overcome this diagnostic challenge, additional examination (eg, immunohistochemistry) is used to detect micrometastases. 12 Although several techniques are available, they are expensive and time-consuming. 13 Multiple trials have been performed to investigate methods of SLN detection and showed accuracy rates up to 99% when detecting lymph nodes with a dual tracer consisting of radiolabeled tin colloid and blue dye. 14 This dual tracer provides the surgeon with visible blue dye and audible guidance based on radioactivity measured by a handheld probe. However, the disadvantage of this technique is irreversible hampering of the operating field by the blue dye.
Intraoperative NIRF can also be used to identify draining lymph nodes in real time. In contrast to the blue dye, NIRF does not alter the operating field by dark staining and allows detection of deeper situated nodes.
The concept of using NIRF for the identification of draining lymph nodes in gastric cancer was first described by Soltesz et al (preclinical) and Kusano et al (clinical). 15,16 In the first study, fluorescent quantum dots were used in animal experiments and showed successful lymph node detection. The second study described the first inhuman trial using ICG (a cyanine dye which passively accumulates in tumor areas due the EPR effect, with its emission peak at 800 nm) and showed safe and accurate identification of gastric sentinel lymph nodes. Since the success of these studies, numerous other trials have proven the feasibility and safety of this method. However, more false-positive lymph nodes were found than anticipated due to spread of the contrast agent through lymphatic vessels to second-tier lymph nodes. [17][18][19][20][21] Hence, no additional value of the NIRF was observed, since still too many lymph nodes were unnecessarily resected. 22 To overcome this problem of migration to second-tier lymph nodes, Tummers et al suggested using ICG adsorbed to nanocolloid. 23 This principle was first described in breast cancer and skin melanoma studies. 24,25 The adsorption of nanocolloid to ICG increases its hydrodynamic diameter from <1 nm to 20 to 80 nm. Only molecules with a hydrodynamic diameter of <10 nm can migrate to second-tier lymph nodes. 26 In this way, spreading of ICG to second-tier lymph nodes can be retained. 23 The study by Tummers et al assessed the feasibility of ICG:Nanocoll for the identification of sentinel lymph nodes in gastric cancer and also investigated the prognostic utility of the detected sentinel lymph nodes. 23 ICG:Nanocoll was injected intraoperatively into four quadrants of the tumor after which fluorescent imaging was performed.
In this study included 22 gastric cancer patients, with varying tumor stages, undergoing partial or total gastrectomy were assessed. In 21 out of 22 patients, at least one fluorescent lymph node was visualized and a mean number of 3.1 (range 1 to 6) lymph nodes were detected. This is significantly lower compared to earlier reported means of 7.2 and 9.3 when using ICG alone. 17,20 The mean tumor-to- In six out of these eight patients, metastatic lymph nodes could be detected with the fluorescent signal. Nonfluorescent metastatic lymph nodes (n = 7) were found in the other two patients. All 7 nonfluorescent metastatic lymph nodes, however, were completely obliterated by tumor tissue. This suggests that lymphatic functions such as lymph flow or drainage were hampered, and the fluorescent agent was not able to reach the lymph nodes. In eight patients, the initial treatment plan was altered based on fluorescent lymph nodes found outside of the standard resection plane. In two out of these eight patients, the additional lymph nodes were tumor positive.
Tummers et al conclude that NIRF can aid in identifying additional draining lymph nodes outside of the standard plane of resection (eg, skip metastasis in extraperigastric lymph nodes, which are found in up to 11% of patients) similarly to previous studies performed in breast cancer patients. 24,27,28 Recent studies have indicated that sentinel node biopsy is most relevant in early gastric cancer cases since this subset of patients have low chances of lymph node metastasis. [29][30][31] A recent study by Kinami et al has demonstrated the feasibility of this technique in q72 early gastric cancer cases specifically, using ICG. 32 Only one false-negative case was observed, which was due to failure of frozen section diagnosis.
It still remains unclear if the use of ICG in SLN procedures has additional value, since it does not give any information on the presence of micrometastases in identified draining lymph nodes.
Possibly, tumor-marker targeted NIRF could assist in this matter. An overview of the mentioned clinical studies is depicted in Table 1.

| TARGETED FLUORESCENT-GUIDED SURGERY AND RATIONALE FOR CEA IMAGING
Since nontargeted fluorophores such as ICG, have limited applicability in specifically delineating tumor tissue, the era of receptortargeted fluorescent-guided surgery has commenced. 34 For this purpose, highly overexpressed tumor markers are required, that are not (or minimally) expressed on normal tissue. CEA is such a molecular marker. CEA is present on embryonic cells and highly expressed by many cancer types including pancreatic ductal adenocarcinoma (PDAC), colorectal cancer (CRC), and gastric cancer.
CEA is also expressed under inflammatory conditions. Overexpression of CEA on tumor tissue is seen in over 90% of all CRC and PDAC patients. [35][36][37] After production and attachment to the cell membrane, CEA is shed into the bloodstream and is therefore a measurable protein in serum. Serum CEA is the only tumor marker that has shown to efficiently monitor therapy response in CRC patients.
Consequently, monitoring of serum CEA during follow up has become standard of care. 38 | Since CEA is highly expressed by many tumor types and absent on healthy tissue, it could serve as a target for tumor-specific imaging and therapy. Previous research has already investigated CEA as a target for therapy of solid tumors including T cell-specific antibodies and radioimmunotherapy. [39][40][41] An imaging modality in which CEA could be used as a target is fluorescence-guided surgery. Recently, SGM-101, a CEA-targeted fluorescent probe (emission peak at 700 nm) was developed (SurgiMab, Montpellier, France) and is currently being tested in clinical trials. 42 The BM104 dye (a fluorescent carbocyanine dye) was conjugated to an anti-CEA chimerized monoclonal antibody and extensive preclinical research has been performed. 42 Other CEA-targeted fluorescent probes have been developed and tested preclinically although they have not been used in-human clinical trials yet. [43][44][45][46] Boonstra et al 46  The main goal of this study is to determine the optimal dose for phase II studies.

| CEA AND CRC
The cornerstone of CRC treatment is resection of the tumor with clear margins (R0 resection). Despite neoadjuvant treatment protocols for rectal cancer, up to 17.2% of rectal cancer patients and 5.3% of colon cancer patients still have positive resection margins (R+). 47,48 In rectal cancer patients with threatened or involved circumferential resection margins, R1 resections are seen in up to 25% of the patients. 49,50 Positive resection margins are correlated with higher rates of local recurrence (up to 40%) and worse diseasefree and overall survival. [51][52][53] CEA serum measurements are currently used during follow-up and can indicate local recurrence or metastatic disease when values are rising. 38 Interestingly, a recent histopathological study by  Sentinel lymph node navigation using ICG in gastric cancer Kusano et al 16 22 The detection rate, accuracy, and false-negativity rates were 90.9%, 88.9%, and 33.3% in T1 stage patients, but lower in higher tumor stages. A mean of 3.6 nodes per patients was detected.
Tajima et al 20 56 A mean of 7.2 sentinel nodes was found. A higher accuracy rate was established in T1 stage cancers compared to higher tumor stages.
Yano et al 88 130 All (100%) lymph nodes were identified and 100% sensitivity was established using ICG. All metastatic lymph nodes were fluorescent.
Kinami et al 32 72 The sensitivity, specificity, and accuracy of ICG sentinel node mapping were, respectively, 90.1%, 100%, and 98.6%. A median of six fluorescent nodes was found.
Tummers et al 23 22 Sentinel nodes outside the standard plane of resection were identified using ICG:Nanocoll. In 8 of 22 patients, the initial treatment plan was altered based on the fluorescent imaging.
CEA-targeted fluorescent-guided surgery in colorectal and pancreatic cancer Boogerd et al 33 26 Optimal dosing was set at 10 mg of SGM-101, best imaged 96 hours postinjection. Primary cohort: seven of nine colorectal tumors visualized using fluorescence (two nonfluorescent lesions were pathological complete responders).
Expansion cohort: in 6 of 17 patients the treatment strategy was altered based on fluorescence. Sensitivity 98%, specificity 62%, and accuracy 84%.
Hoogstins (unpublished data) 12 Optimal dosing was set at 10 mg of SGM-101, best imaged 96 hours postinjection. All pancreatic tumors were visible using NIR fluorescence, with a mean TBR of 1.6. Also all four metastatic lesions were visible using NIR fluorescence.   79 A study investigating staging laparoscopy using contrast-enhanced ultrasound and fluorescence (ICG) has been performed by our group. 80 Twenty-five patients were included in this study. Patients received 10 mg of ICG 1 day prior to surgery to detect possible liver metastases. Intraoperative fluorescence imaging and laparoscopic ultrasound of the liver were performed. Of every lesion suspect for metastasis a biopsy or resection was performed.

| CEA AND PANCREATIC CANCER
This study concluded that intraoperative use of ultrasound has limited value. However, laparoscopic fluorescence imaging using ICG showed the highest accuracy in detecting liver metastases compared to inspection and laparoscopic ultrasound.
The concept of fluorescent-guided surgery in pancreatic cancer has already been tested with ICG. Unfortunately, insufficient contrast between benign and malignant tissue was achieved and no additional value of optical fluorescent-guided surgery was found. 81 In search of other tumor-specific imaging targets, de Geus et al found that CEA, integrin αvβ6, epithelial growth factor receptor and urokinase plasminogen activator receptor seem suitable targets for pancreatic imaging. 82 In this study, 158 of 165 (96%) of pancreatic and periampullary adenocarcinomas could be identified using CEA immunohistochemistry. The TBR as seen in intraoperative setting was significantly lower than expected from the mouse models (TBR 3.5). A possible explanation for this can be the poor vascularization of pancreatic tumors and presence of desmoplastic stroma in pancreatic tumors, resulting in poor delivery of SGM-101. 86 However, the fact that fluorescent signal was observed in most patients demonstrates that the contrast agent was undoubtedly able to reach the tumor cells.
The limitations of this "far red" part of the NIR spectrum are more autofluorescence and lower depth penetration compared to the higher NIR wavelengths. 89 Currently, fluorescent agents targeting the endothelial growth factor receptor and the vascular endothelial growth factor (VEGF; PENGUIN trial) are also being clinically tested for pancreatic cancer imaging (Clinicaltrials.gov: NCT03384238, NCT02743975). can increase tumor specificity and enable injection at the day of surgery.

| DISCUSSION
When combining this with a fluorophore emitting light at higher wavelengths, lower background signal could be achieved.
In conclusion, this review shows that SLN procedures for gastric cancer resections using ICG:Nanocoll outperformed procedures using normal ICG. However, the clinical relevance can be argued since the fluorescent signal only indicates lymph node presence, but gives no information as for the presence of lymph node metastasis.
Tumor-specific targeting by CEA-targeted fluorescent imaging using SGM-101 was successful in both pancreatic and CRC patients.
A large phase III trial will soon be initiated in CRC patients.

CONFLICTS OF INTEREST
The authors declare that there are no conflicts of interest.

SYNOPSIS
SLN procedures for gastric cancer resections using ICG:Nanocoll outperformed normal ICG but did not provide information on possible lymph node metastasis. CEA-targeted fluorescent imaging using SGM-101 was successful in both pancreatic and CRC.