Feasibility in the detection of sentinel lymph node-associated blood vessels using intravital microscopy in patients undergoing sentinel lymph node biopsy for melanoma

While the clinical focus on the sentinel lymph node biopsy (SLNB) is the presence of intra- or extra-nodal metastases, preclinical studies suggest that tumor-draining SLNB-associated vascular architecture and adhesion properties are altered regardless of SLNB positivity. Human intravital microscopy (HIVM) has defined blood vessel abnormalities that may impact lymphocyte adhesion and systemic drug delivery at primary melanoma sites. In this pilot study of HIVM during melanoma SLNB, we sought to determine the feasibility of obtaining HIVM observations of SLNB-associated vessels. We successfully performed HIVM in all 20 SLNB patients, and 7 were found to have nodal micrometastases by standard pathology. HIVM was capable of identifying both functional and non-functional SLNB-associated vessels based on the presence or absence of fluorescent dye uptake, respectively. Comparing vessel characteristics as a secondary exploratory objective, no statistically significant differences were noted in the diameter, flow rate, functionality, or shear stress of SLNB-associated blood vessels between positive and negative SLNBs, which may likely have been a reflection of the minimal disease burden. Nonetheless, these initial observations provide the framework to optimize future trials of HIVM in cancer patients.

Sentinel lymph node biopsy (SLNB) was originally described by Donald Morton for the staging and treatment of melanoma (Essner et al., 2002). The multicenter selective lymphadenectomy trial (MSLT1) confirmed the prognostic value of SLNB for melanoma and its function as a staging procedure, with some clinical benefit in overall survival (OS) in select patients with intermediate-thickness melanoma (Morton et al., 2014). The results of MSLT1 formed the basis of the MSLT2 trial that examined the outcomes for completion lymphadenectomy vs. observation for positive sentinel nodes. MSLT2 confirmed that immediate completion lymphadenectomy for a positive sentinel node could be omitted with no impact on OS for the majority of patients (Multicenter Selective Lymphadenectomy Trials Study Group et al., 2022). Together, these trials raised questions regarding the clinical benefit of SLNB. However, in the absence of SLNB and/or completion lymphadenectomy, full staging may be incomplete and prognostic information that can influence decisions regarding frequency of surveillance and adjuvant therapies is lacking.

By virtue of draining a primary tumor site, the sentinel lymph node represents more than just accurate staging and prognosis. On a cellular level, the sentinel node is the direct interface between the host immune system and metastatic tumor in the setting of secondary lymphoid tissue. Preclinical investigations into tumor draining lymph nodes have demonstrated how major immunologic functions such as recruitment of lymphocytes through specialized high endothelial venules (HEV) are impaired by downregulation of CCL21, a chemokine necessary for nodal entry of lymphocytes, even when actual nodal metastasis has yet to occur (Carrière et al., 2005). Soluble mediators including VEGF-D generated by primary tumors may precondition the draining node to support eventual tumor metastases by inducing changes in HEV from normally plump, cuboidal endothelium to a dilated and flat vessel architecture (Farnsworth et al., 2011). Rather than de novo generation of tumor vessels in SLNB, tumor-induced remodeling of HEVs not only sets the stage for tumor deposition, but also to become the nutrient tumor vasculature (Qian et al., 2006, 2007). Primary tumors also appear to have the ability to induce stromal remodeling in tumor draining lymph nodes that downregulates CCL21 and IL-7 to further impede lymphocyte recruitment and activation, respectively (Riedel et al., 2016). Importantly, micrometastases that have gained a foothold in the tumor draining node appear to quickly co-opt the altered vasculature as a portal for intravasation and hematologic dissemination (Qian et al., 2007; Pereira et al., 2018; Brown et al., 2018). Therefore, microscopic imaging methods that could identify potential vascular alterations or micrometastasis in a sentinel lymph node in vivo might generate opportunities to preserve the sentinel node/tumor interface and create new treatment paradigms.

Intravital microscopy (IVM) is the real-time evaluation of living tissue at the microscopic level (≥100 × magnification), which is capable of recording physiological processes such as blood flow and cell-cell interactions (Carrière et al., 2005; Gabriel et al., 2018; Nussbaum et al., 2022). In oncology, preclinical IVM has elucidated mechanisms of metastasis, development of neoangiogenesis, and immune cell-tumor interactions primarily in animal models (Beerling et al., 2011; Zomer et al., 2011; Jain et al., 2002; Jain, 2013). Independent of transillumination, IVM can be performed on exposed tissue of interest using an epifluorescent light source. IVM was only recently translated to the clinical arena in which melanoma tumors were interrogated by IVM in the operating room setting (Fisher et al., 2016). Clinical trials of human IVM (HIVM) proved feasible in the context of standard oncologic surgeries and generated real-time data of blood flow characteristics that was not previously possible (Gabriel et al., 2021, 2023; Trumbull et al., 2020). Specifically, dynamic features such as blood flow, velocity, and real-time measurements of vessel diameters in living tissue showed differences that could not have been predicted from standard histologic evaluation of the same excised tissue (Fisher et al., 2016). The success of these first in-human trials have raised the possibility of a broader application to other oncologic procedures that could also produce relevant information of in situ human tissue prior to surgical resection (Gabriel et al., 2021, 2023).

Herein, we describe a completed clinical trial of HIVM of SLNB at the time of standard melanoma surgery. The primary goal of the study was to determine the feasibility of measuring vascular parameters of human lymph nodes after exposing the lymph node just prior to surgical excision. A successful observation included the ability to identify lymph node vessels, measure vessel diameters, and visualize fluorescein and/or indocyanine green (ICG) within the observed vessels. As these were SLNBs taken for melanoma, exploratory secondary objectives included comparisons between positive and negative lymph nodes and correlation to IVM measurements. It is important to note than HIVM was performed on SLNB-associated microvessels, but not on lymphatic channels. The HIVM technology does not readily allow for identification of lymphatics, which do not uptake dye via IV administration. In addition, HIVM is limited in its ability to distinguish HEVs from other microvessels (e.g., capillaries, arterioles, venules). Thus, our observations were limited to generalized SLNB-microvessels within our fields of observation.

Patient selection

Patients scheduled to undergo SLNB for melanoma were evaluated for entry into our pilot clinical trial I-284116 “Intravital Microscopy in Identifying Tumor Vessels in Patients with Melanoma Undergoing Sentinel Lymph Node Biopsy” at the Roswell Park Comprehensive Cancer Center, Buffalo, NY (ClinicalTrials.gov identifier: NCT02857374, first posted 05/08/2016). The study was approved by the Institute Review Board of Roswell Park Comprehensive Cancer Center, and all methods were performed in accordance with the Declaration of Helsinki.

As groin lymph nodes were readily accessible to IVM observation, patients whose melanoma mapped to a non-groin sentinel lymph node were excluded. Our imaging apparatus used in this study was not readily able to be positioned in the axilla, and so we limited this study to groin SLNB. A full list of the inclusion and exclusion criteria is noted in Table 1. Eligible patients who gave informed consent were enrolled in the protocol. The following data was recorded: patient demographics (age, sex, height, and weight), tumor-specific information (tumor type, location, and stage) and intravital microscopic measurements obtained from the surgery. Clinical outcome measures, including recurrence and melanoma-specific survival, were retrospectively collected at the end of the study enrollment.

Of note, no control population was included in our pilot trial. A control population would consist of healthy volunteers without melanoma and with presumed normal lymph nodes. Because there is a 5–15% risk of lymphedema in the upper extremity and a 10–30% risk in the lower extremity with SLNB (Morton et al., 2014; Arié and Yamamoto, 2020; Moody et al., 2017), we did not include a control population due to risk and ethical considerations.

Microscope design

Sentinel lymph nodes were observed at 100 × magnification using a highly modified Olympus microscopy system as previously described by our group (Fisher et al., 2016). The microscope was attached to a cantilevered arm to extend over the patient. Given the high magnification required to observe individual vessels and flow, vibrations in the microscope were minimized using a weighted marble base of over 360 kg. Intravenously administered fluorescein was illuminated using a 467–498 nm excitation, 513–556 nm emission dichronic filter set (Spectra Services, Rochester, NY) and ICG was illuminated using a 740–800 nm excitation, 800–860 nm emission dichronic filter set (Spectra Services, Rochester, NY) with an X-Cite 120 Led light source (Lumen Dynamics, Ontario, Canada). Images were captured using a Luca EMCCD camera (Andor Technology Ltd., Belfast, Northern Ireland) controlled through the Solis acquisition and analysis program. Images were acquired with a minimum of a 0.05-s exposure at 20 frames a second.

Offline post-surgical quantification of vessel diameter and blood velocity was performed using the ImageJ software suite.

Surgical technique

All patients underwent standard lymphoscintigraphy with filtered technetium-99m sulfur colloid injected at the site of the primary melanoma. The SLNB location was confirmed by hand-held gamma probe (Neoprobe, Devicor Medical Products Inc., Cincinnati, OH). The skin overlying the sentinel lymph node was prepped with ChloraPrep (CareFusion Corp. San Diego, CA), and sterile towels and drapes were applied for exposure of the area. An incision was created with a scalpel in line with the longitudinal axis of the patient. With care to keep the lymph node capsule and perinodal adipose intact, the skin was elevated bluntly and retracted laterally with a Weitlaner retractor and the use of 3-0 Vicryl sutures (Ethicon, Raritan, NJ). Electrocautery was avoided as much as possible to prevent underlying tissue damage. In all instances, a saline soaked gauze was placed over top of the tissue to prevent desiccation before microscopic interface. After exposure of the confirmed SLNB was completed, the microscope was positioned for observation. Again, we intentionally preserved the peri-nodal tissues so as to not disturb node capsule or blood supply going to the nodal hilum. At the completion of the observation period, the microscope was withdrawn, and the lymph node was removed and sent to pathology for evaluation. Standard pathologic analyses consisted of Hematoxylin and Eosin (H&E) staining as well as melanoma-specific stains including S-100, HMB-45, and Melan A.

Intravital microscopic observations in patients

The microscope was grossly moved into position by the surgeon over the field of interest and locked into place (as depicted in Figure 1A). Fine adjustments in the x and y axis were performed by using the built-in motor controls. Once the microscope was directly overlying the exposed SLNB surface, the fluorescent light source and image acquisition software were activated (Figure 1B). Vertical control of the microscope was performed manually by the surgeon until nodal vessels came into view on the digital monitor at 100 × magnification. Lymph node vessels were clearly recognized by their well-defined architecture and branching patterns with observations focused on these areas. When present, melanoma micrometastases are typically found in the subcapsular sinus of lymph nodes (Figures 1B, C) allowing for potential detection by IVM. To facilitate stabilization of the images, respirations were temporarily held by anesthesia for a maximum of 30 s. Patient respirations can lead to suboptimal imaging, but additional anesthesia or temporary holding respirations can often restore in focus, high resolution images. Therefore, clear communication with our anesthesia colleagues was key to ensure adequate HIVM observations. Of note, holding respirations could also have impact on blood flow through changes in intrathoracic pressure. However, typically halting respirations was performed transiently for 10–15 s to complete the observation.

When a stable view of the vessels was achieved, 2.0 ml of 25% (250 mg/ml) fluorescein (Akorn Pharmaceuticals Inc., Lake Forest, Illinois) diluted in 10 ml of sterile saline was injected via a peripheral intravenous catheter. An observation was completed when fluorescein was noted to have extravasated into the nodal tissue and the vessels no longer manifested a detectable fluorescent signal. The microscope was then re-positioned with fine motor controls to observe several areas of the node. The microscopic observations were finalized when the 10 ml of diluted fluorescein was exhausted. Observations were repeated with ICG (IC-Green, Akorn Pharmaceuticals Inc., Lake Forest, Illinois) using 2.0 ml of 25% (250 mg/ml) using the same approach as with fluorescein. The entire process for HIVM observation took approximately 5–7 min. Of note, ICG observation produced less vascular detail than fluorescein and was omitted after the first three patient observations and was not included for analyses. The difference in observation detail may be related to the molecular weights of the dyes with ICG being a large molecular than fluorescein (775 g/mol vs. 332 g/mol), which could affect flow through the small microvessels, or due to variances in image capture optimization related to the different excitation and emission wavelengths of the two dyes.

Post-hoc analyses

Post-hoc analyses were performed as previously described by our group (Fisher et al., 2016). Vessels were defined morphologically, and a single vessel was described as beginning at a branch point continuing to the next branch point. To be measured, vessels had to be 100 mm in length with no interposing branch points. Vessel density was established by enumerating vessels within the field of observation. ImageJ software was used to measure vessel diameter (D) and radius (r) at the vessel's largest width. Blood flow velocity (v) was evaluated by determining the time Δ T distinct features in the fluorescent dye would take to travel a known distance Δ S and then averaging Δ S /Δ T for at least 10 points per vessel. Functionality of a given vessel was defined as a vessel that demonstrated fluorescent dye uptake/signal (Figure 2) or kinetic blood flow of red blood cells (RBC). In our experience, detection of RBCs has always been followed by dye uptake after IV administration, both of which confirm blood vessel functionality with complete concordance. Non-functional vessels neither demonstrated dye uptake or kinetic blood flow. Percent functionality was determined as mean ± s.e.m. on a per field basis. Wall shear stress was computed as τ = 32η Q /π D 3, where η is blood viscosity (assumed to be 2.2 centipoise) and Q is the blood flow rate (Q = v * π r 2 with r being the radius of the tumor vessel). Wall shear rate was determined using the formula w = v * 8/ D. The ability to discriminate between HEV and regular vessels and orders of HEV as previously described preclinically (Chen et al., 2006) was not possible due to limited fields of observation and lack of definitive anatomic references.

Statistical analyses

Demographic and clinical characteristics were summarized using the mean and standard deviation (std) for continuous variables and using frequencies for categorical variables. No statistical method was used to predetermine sample size because our study was aimed at determining feasibility and because of the exploratory nature of our secondary outcomes. Comparison of vessel characteristics between negative and positive SLNBs is presented as means with standard deviation and compared using two sample t -test. Statistical significance for all comparisons was accepted at p = 0.05. Data analyzed using R software (version 4.2.3).

Patient demographics, SLNB-specific characteristics, and treatment outcomes

Between December 2016 and August 2018, a total of 20 patients underwent SLNB with HIVM. Patient characteristics are summarized in Table 2. Of the study cohort, 7 patients (35%) had a positive SLNB(s). Characteristics of patients who had a positive SLNB are shown in Table 2. Of the 7 patients with a positive SLNB, 5 underwent completion lymph node dissection, and 2 of these patients had additional positive lymph nodes. There were no adverse events related to the administration of fluorescent dye or the HIVM procedure, similar to our previous HIVM studies. The overall disease burden within positive SLNBs was low with 6/7 having micrometastases < 2 mm or isolated tumor cells only as depicted in Figure 1C. Recurrence outcomes for each patient are also shown in Table 3. At a median follow-up of 65 months, only 1 patient with a negative SLNB was found to have a recurrence during the study period. This patient developed both lower extremity in-transit disease and nodal recurrence 15 months after surgery. Conversely, 4/7 patients (57%) with a positive SLNB developed nodal and/or distant metastases with most occurring in the first 2 years of surveillance.

HIVM SLNB vessel characteristics

All 20 patients were able to successfully undergo HIVM with observations of SLNB-associated blood vessels. Figure 2 depicts representative examples of HIVM observations using fluorescein at the time of SLNB, which align with our previous HIVM observations in other cancer histologies (Gabriel et al., 2021, 2023, 2024; Garcia et al., 2024). Vessel architecture in positive nodes were suggestive of the disorganized architecture seen in primary melanoma tumors (Fisher et al., 2016); however, unlike preclinical models of murine lymph nodes (Chen et al., 2006), HEV could not be reliably differentiated from non-HEV vessels in post-hoc analyses. Similar to our prior trials of HIVM in primary melanoma, primary sarcoma, or peritoneal carcinomatosis (Fisher et al., 2016; Gabriel et al., 2021, 2023), functional vessels were defined as those blood vessels that took up the fluorescein dye and non-functional vessels were defined as those which did not take up the dye. Thus, functional vessels (labeled with blue arrows) were portrayed by a bright white appearance, whereas non-functional vessels (labeled with yellow arrows) were portrayed by a dark black appearance, which indicated the absence of fluorescein within the blood vessel.

A comparison of vessel characteristics between negative and positive SLNBs is shown in Table 4. Functional vessels were detected in only 26% and 30% of negative and positive SLNB, respectively. There were no statistically significant differences among any of the vessel characteristics, including density of functional or non-functional vessels, mean diameter of functional or non-functional vessels, or blood flow velocity within functional vessels for either negative or positive SLNB. Within each observed field, calculations for vessel diameter and velocity were performed for all vessels observed within the field, which ranged from 3 to 12 across the patient observations. The lack of differences may have been related to the limited number of functional vessels seen across all groups as well as the minimal disease burden of micrometastatic nodal melanoma. Correlations between SLNB-associated vessel characteristics and recurrence was not permissible given the small number of recurrence events. The IVM observed areas of the lymph node were not able to be localized the with standard pathologic evaluation and direct histologic comparisons to IVM observations were therefore not possible.

To our knowledge and based on an extensive review of the current literature, this was the first study to demonstrate the feasibility of HIVM of the microvessels of tumor-draining lymph nodes. Using standard surgical techniques and our customized intravital microscope, we were able to obtain successful observations for each patient in our cohort. While our primary goal to determine the feasibility of HIVM for the SLNB was successful, our secondary exploratory goal of comparing SLNB-associated vessel characteristics and either SLNB positivity or recurrence did not yield any statistically significant differences. Yet, the findings of this clinical trial were informative to guide potential future applications of HIVM.

IVM offers certain advantages over other imaging modalities, which make it a useful tool to investigate tumor-associated vessels in this pilot study and our prior HIVM trials. IVM allows for the real-time, high resolution (< 1 μm) observation of tumor-associated vessels and potential immune cell interactions (Gabriel et al., 2018). While whole body imaging, including CT scan, PET-CT, and MRI, obtains a global visualization of large and medium sized blood vessels, these imaging modalities cannot determine vessel functionality at the capillary (microscopic) level. Lymphoscintigraphy with technetium injection can identify the draining SLNBs, but does not have a role for identifying tumor-associated or SLNB-associated blood vessels. Other high resolution imaging devices such as high frequency US, ocular coherence tomography (OCT), and photoacoustic microscopy (PAM) have been used to investigate tumor-associated microvessels (Gabriel et al., 2018). Of these other imaging modalities, IVM offers the highest resolution (< 1 μm), whereby the others are generally limited to 10–15 μm. The tissue penetration of IVM is 200 μm, which is deeper than other modalities, such as high frequency US and OCT, which are generally limited to 10–50 μm. IVM therefore provides for a deeper assessment of blood vessels within the tissue targets. In addition, the image acquisition rate for IVM (30–100 ms) is also sufficient for tracking blood flow in real time down to individual flowing red blood cells (Gabriel et al., 2020). While there is no current “gold standard” for assessing the tumor vasculature at microscopic resolution in real time in humans, for our purposes of tumor vessel imaging, IVM served as the most optimal technique.

The limited correlative findings in SLNB are in contradistinction to our previous studies of HIVM of tumor tissue. The first report of HIVM in human cancer was published by our group in 2016 for primary melanoma that was surgically resected (Fisher et al., 2016). In this first-in-human study, patient melanoma-associated vessels were characterized as disorganized and tortuous compared to normal (control) blood vessels, and up to 50% of melanoma-associated vessels did not support blood flow. In addition, human melanoma-associated vessel diameters were larger than predicted from IHC or preclinical IVM. Since our initial study, we have continued HIVM observation for peritoneal carcinomatosis derived from several different histologies (including appendix, colorectal, ovarian, and mesothelioma cancers; Gabriel et al., 2021, 2024). Most recently, we have reported our HIVM findings for retroperitoneal sarcomas and brain tumors (Gabriel et al., 2023; Garcia et al., 2024). In each of these earlier clinical trials, we demonstrated that tumor areas had on average a higher density and proportion of non-functional vessels compared to control areas. Unlike our current study in melanoma-associated SLNB, our previous trial in ovarian carcinomatosis did show statistically significant correlations between tumor-associated vessels and outcomes, including disease-specific survival and response to neoadjuvant chemotherapy (Gabriel et al., 2024).

Several factors may have contributed to differences in our current HIVM observations in SLNB vs. prior findings in tumors. All SLNB observations required surgical exposure of the tissue with the attendant potential to alter blood flow characteristics which may have accounted for the observation of only ~30% of all SLNB vessels manifesting blood flow. Thus, functional vessel assessments including flow rates were limited compared to the majority of tumor studies that did not require perturbations of the tissue prior to HIVM observation. In our prior studies of primary melanoma, brain tumors, and carcinomatosis, non-tumor control areas had notably higher proportions of functional vessels as high as 90% in areas that did not require surgical manipulation prior to observation (e.g., peritoneal lining; Gabriel et al., 2021). The inability to observe normal lymph nodes that are anatomically distant from tumor limits comparisons to a “true” normal control. Indeed, because we did not include healthy control subjects because of lymphedema risks with SLNB, we did not have a true internal control of how normal SLNB-associated vessels would appear during HIVM observations. As all SLNBs by definition are tumor-draining lymph nodes and based upon the observation that vascular changes occur within the node even prior to metastases (Carrière et al., 2005; Farnsworth et al., 2011; Qian et al., 2006, 2007), it is possible that HIVM measurements of SLNB without a normal control for comparison would be unable to detect any potential differences. Also, the structural alterations in vascular anatomy described preclinically may be minimal and below the threshold of detection for HIVM (e.g., flattening of normally cuboidal HEV; Farnsworth et al., 2011) and functional changes such as loss of adhesion molecules would not be detected (Riedel et al., 2016).

In addition, there is the possibility that pathologic analyses of the SLNB could have yielded false positive or false negatives. It is estimated that up to 10% of SLNB could be false positives (El Sharouni et al., 2021). If this were the case for any of the subjects with positive SLNB in our small cohort, then we would not expect to identify vascular differences in truly negative SLNB that were interpreted as being positive during intraoperative frozen section or final pathology (i.e., false positives). Regarding pathologic analysis, it is important to note that we did not conduct analyses to investigate associations between the HIVM vessel characteristics and pathologic vessel characteristics on permanent specimens. This was not the focus of our feasibility study, but does limit spatial correlations between the HIVM observations and pathological analyses. Lastly, by design our study focused on SLNB of the groin only. This site was chosen because our primary objective was to determine feasibility and because our imaging apparatus could not easily facilitate imaging of the axilla, which is the other main lymph node basin where we operate. Thus, there is limited generalizability of our findings to the groin SLNB, which may have anatomic vessel differences compared to other areas of the body. The SLNB-associated microenvironment may have other structural or immunologic differences from the primary tumor site as well, which could impact the differences in observations we recorded in this study compared to our prior studies at primary tumor sites.

Many studies across different tumor types have shown that primary tumor-associated vessels are structurally aberrant and behave abnormally compared to normal vessels (He and Ganss, 2021; Katsuta et al., 2019; Mangala et al., 2016). Importantly, these tumor vessel abnormalities have been shown to translate to poor responses to systemic therapies, both in preclinical animal models and in humans (Gabriel et al., 2020; Herrera et al., 2020; Mezheyeuski et al., 2019; Visioni et al., 2018). Our previous HIVM studies focused on known tumor-associated vessels compared to the surrounding grossly normal (control) tissue. However, in our current trial, it was unknown whether the SLNB would have metastatic disease or not. This study only included patients with clinically node negative melanoma (i.e., not palpable, not visible on imaging). Additionally in our cohort, the burden of disease in patients with a positive SLNBs was quite low, with the majority of positive SLNBs having only micrometastases or isolated tumor cells. Due to this low extent of disease, it is conceivable that HIVM measurement of differences between positive and negative SLNBs would be minimal even when probing for characteristically disorganized tumor vessel architecture. Because of the small cohort and exploratory nature of our secondary objectives, our study was not designed with the expectation that such minimal disease burden would produce detectable changes in SLNB vascularity. In each of our prior HIVM studies, it was readily obvious where gross tumor was present, and the tumor burden was quite high in the majority of patients (particularly those with carcinomatosis and large retroperitoneal sarcomas). Because of this limitation of low disease burden, it cannot be reliably concluded that there was no truly significant difference in the SLNB-associated blood vessels of negative vs. positive nodes.

Additionally, we recognize that our patient cohort was small, consisting of only 20 patients. The number of SLNB positive events was also small, with only 7 cases having nodal metastasis. Thus, our trial that was primarily designed for feasibility was not sufficiently powered to analyze any associations between vessels associated with either negative or positive SLNB. Similarly, our study was underpowered to analyze any associations between SLNB-associated vessels and long-term outcomes such as melanoma recurrence. It is also possible that the cellular and molecular processes that alter SLNB-associated vessels may be different from the processes occurring with the primary tumor microenvironment which may account for the lack of positive associations between SLNB-associated vessels and oncologic outcomes. Lack of sufficient power also limited our ability to detect any truly significant differences in the SLNB-associated blood vessels of negative vs. positive nodes. Both the small sample size (which limited statistical power to detect differences in vessel characteristics or melanoma-associated outcomes) and the minimal disease burden within the SLNB were significant limitations of our study to make any conclusions beyond feasibility.

Despite these limitations, our primary aim was accomplished as we demonstrated HIVM feasibility for the first time in SLNB. Troubleshooting the HIVM technique with clear communication with our anesthesia colleagues facilitated the acquisition of high resolution HIVM images. While the HIVM apparatus used in this study was patented and proprietary to the Roswell Park Comprehensive Cancer Center, we have also used a commercial HIVM device that is publicly available for purchase from Mauna Kea Technologies in some of our previous studies (Nussbaum et al., 2022; Gabriel et al., 2020, 2021, 2023; Garcia et al., 2024). We have shown that the Mauna Kea Cellvizio fluorescent confocal microscope provided similar high resolution, tumor microvasculature images in patients with peritoneal carcinomatosis, ovarian cancer, brain tumors, sarcoma, and melanoma in-transit disease. To date, the cost of performing the HIVM observations has been covered through various research grants. As we continue to explore further applications of HIVM to the multidisciplinary treatment of various cancers, we intend to perform a cost analysis to further define the feasibility of this technique.

To examine differences between positive and negative SLNB, future studies of HIVM will need to account for potential disruption of microvessels from tissue exposure either with altered surgical technique, a larger patient cohort for comparisons, or inclusion of patients with clinically positive nodal disease. As imaging technologies like HIVM continue to be refined in clinical trials, we look forward to potential novel applications of HIVM to not only characterize structural and functional differences in SLNB vessels. In addition, while our current HIVM technology does not readily allow for identification of lymphatics (which do not uptake dye via IV administration), future studies may be able to investigate SLNB-associated lymphatics. There are a few methods to introduce dye into the lymphatics, including direct subcutaneous tissue injection around the primary tumor. Similar to standard lymphoscintigraphy for SLNB identification, subcutaneous dye injection should introduce fluorescent dye into lymphatic channels, although the uptake rate would likely be quite slow compared to IV dye administration. This approach may be challenging to obtain high quality HIVM images in a timely manner that does not significantly increase operative time. Alternatively, direct injection into the lymphatics may be attempted. This approach can be performed surgically, which is often done in procedures such as lymphovenous bypass for severe lymphedema. However, direct lymphatic injection requires dilated lymphatics (i.e., as encountered in lymphedema), which would be challenging if the lymphatics are otherwise normal. These IVM applications are the subject of our group's ongoing preclinical animal studies, which still require significant optimization prior to translation to human treatment, but are nonetheless important scientific endeavors that warrant further study.