Lighting the way for necrosis excision through indocyanine green fluorescence-guided surgery

Abstract

Background:

No objective technique exists to distinguish necrotic from viable tissue, risking over-excision in burns and loss of wound healing potential. Second Window Indocyanine Green (SWIG) is a novel fluorescence-imaging modality being studied to identify residual solid tumors during oncological surgery. SWIG has also been shown to have avidity for necrosis in animal models but translation of these findings to humans is lacking. The objective of this study was to evaluate SWIG in the identification of burn wound necrosis, and compare it to previously published indocyanine green angiography (ICGA) techniques.

Study Design:

This study utilized mouse, human skin xenograft and human patient burn models. Brightfield and SWIG near-infrared (NIR) imaging were performed on macroscopic tissue samples, which were then cryopreserved, sectioned and analyzed for microscopic fluorescence. SWIG fluorescence findings were correlated to visual assessment of the burn wound, as well as histological markers of necrosis using hematoxylin and eosin (H&E) and lactate dehydrogenase stains (LDH).

Results:

We found that SWIG identified burn necrosis in a manner dependent on the dose and timing of indocyanine green (ICG) administration and had an inverse fluorescence signal compared to ICGA. Furthermore, SWIG fluorescence identified the interface of viable and non-viable tissue.

Conclusion:

Our study confirmed that ICGA is an inconsistent and non-standardized modality to evaluate burn injuries. In contrast, SWIG imaging is a potential imaging modality to objectively prognosticate burn wound healing potential and guide intraoperative burn excision. Further studies are needed to define ratios of fluorescence intensity values to guide surgical decision-making in burn excision, and to better define how ICG is retained in necrotic tissue to enhance utility of SWIG in other disease processes.

PRÉCIS

Second window indocyanine green (SWIG) fluorescence imaging is a new technique using indocyanine green and identifies the deep border of necrotic tissue in human burn wounds. SWIG represents a potential novel fluorescence-guided modality for burn wound excision and other surgical disease processes requiring identification of necrosis.

Graphical Abstract

INTRODUCTION

Intraoperative identification of necrotic tissue by visual appearance alone is imprecise and viable tissue with regenerative capacity is often sacrificed when the current standard of visual evaluation is utilized.^1–2 Histologic analysis of biopsies, the gold standard for objective determination of tissue necrosis in burn wounds, is time-consuming and not clinically feasible. Many advanced imaging techniques are in various stages of evaluation, with most relying on perfusion or oxygenation as surrogate markers for necrosis.^3–9 Indocyanine Green (ICG) Angiography (ICGA), one such method, has been considered to delineate burn depth earlier after injury than visual evaluation¹⁰ and improve prognostication of healing potential.^11–15 However, a recent review determined that these studies lacked details on data interpretation that would enable comparison across institutions and devices, concluding that there is a need for larger human trials focused on quantifying perfusion changes in real-time.⁴ Furthermore, a consensus statement found that determining optimal dosing, concentration, and timing of ICG administration is needed before ICGA can be used for burn depth determination.¹⁶ Currently, no feasible, objective, intraoperative burn necrosis assessment technique exists.

Second Window Indocyanine Green (SWIG), which involves delayed fluorescence imaging after high dose (up to 5 mg/kg) ICG administration, has recently demonstrated utility to intraoperatively identify cancer during oncologic surgery for solid tumors.^17–19 The theorized mechanism is enhanced permeability and retention (EPR), where ICG leaks out of permeable vessels and becomes lodged in inflamed tissue surrounding the tumor.²⁰ In preclinical studies, SWIG fluorescence was identified in necrotic burn tissue with the hypothesis that ICG binds to the exposed membrane phospholipids on necrotic cells,²¹ which is a distinctly different mechanism from that of ICGA.⁴ To date, there have been no clinical studies evaluating SWIG properties in burn patients.

The objective of this study was to use a multi-model approach with mouse skin, human skin xenografts, and human burn patients to evaluate SWIG in the identification of burn wound necrosis and compare it to previously published ICGA techniques. We also sought to compare a currently available imaging system, Spy-Elite (SPY; Stryker/Novadaq, British Columbia, Canada) to a new near-infrared (NIR) imaging device, OnLume Imaging System (OIS; OnLume Surgical, Madison, WI) for fluorescence-guided surgery.

METHODS

Human Subjects Study:

Patient recruitment:

The human subjects study was approved by the University of Wisconsin (UW) Institutional Review Board. We recruited patients admitted to our center with burn wounds requiring surgery as determined by the attending burn surgeon. If eligible, written informed consent was obtained (Table 1). Seven patients were enrolled for ICGA and two for SWIG imaging.

Table 1.

Human Subject Inclusion and Exclusion Criteria

Inclusion Criteria:

• Age 18 years or older

• Patients with 2-30% total body surface area burn wounds requiring surgery

• Subject understands the study procedures and can provide informed consent to participate in the study and authorization for release of relevant protected health information to the study investigator

Exclusion Criteria:

• Contraindication to ICG injection, i.e. previous reaction to ICG (adverse event rate: 1 in 42,000) or iodine allergy

• Inability to obtain consent

• Subject with pre-existing inflammatory diseases or chronically treated before admission to the hospital with steroids, nonsteroidal anti-inflammatory drugs or biologics

• Subject with immune deficiency (HIV infection or use of corticosteroids, cytostatic drugs, tetracycline and certain bisphosphonates)

• Subject with known or suspected infections or on antibiotic therapy

• Subject known or suspected to be pregnant

Protocol:

The region of interest (ROI) of the patient’s burn and adjacent non-injured skin were identified. The day before surgery, brightfield (standard of care) and baseline NIR fluorescence images of the ROI were obtained using SPY. ICG was administered intravenously per the FDAapproved dosage for angiography studies. ICGA image capture was initiated following ingress of ICG into the burn region and then continued until the ICG began to egress. The remaining portion of the total ICG dose (up to 5 mg/kg) was then administered over the course of one hour. Approximately 24 hours after ICG administration, brightfield and SWIG images of the ROI were obtained in the operating room using SPY or OIS. A full-thickness biopsy from the ROI was obtained. Sequential tangential excisions of the ROI were performed using visual assessment by the operating surgeon (standard of care) blinded to fluorescence data. After the first and last tangential excision, the excised tissue and wound bed were imaged using SPY (patient 1 and 2) and/or OIS (patient 2).

ICGA image analysis: Due to limited dynamic range of fluorescence images captured by SPY, the gain was doubled in order to attain perceptible fluorescence contrast for publication. Perfusion calculations were performed without image adjustment. A circular ROI (5 cm in diameter) was identified within the burn (12 burns from 7 patients). Ten subsampled points dispersed evenly within each ROI were used for subsequent ICGA analysis. Two previously published methods were used to quantify ICGA fluorescence in which each subsampled point within the ROI was compared to a singular control data point identified as 100% perfusion and calculated as a percentage of this control perfusion value. First, the adjacent non-burned skin was set at 100% perfusion; burns were categorized as deep with perfusion values less than 100% and superficial with values greater than 100%.¹³ Second, the brightest region within the burn (i.e., the region of maximal perfusion) was set at 100% perfusion.^14–15 Here, a cut-off value of 33% differentiated between superficial (>33%) and deep (<33%) partial thickness burns, with the latter indicating the need for excision and grafting.^14–15

Mouse and Xenograft Study:

Mouse protocol:

All animal procedures were approved by the UW Institutional Animal Care and Use Committee and the Research Animal Resource and Compliance Office. Clinical translation of studies performed with mouse models for burn depth determination are limited by the fact that mouse skin is significantly thinner than human skin. However, as a lower cost, ethically responsible model, it is a reasonable model to test different variables initially. A customized burn device was used to create a 4 mm full thickness burn on mouse skin (100 °C x 5 seconds).²² 1 mg/kg (n=2), 2 mg/kg (n=2), or 5 mg/kg (n=3) of ICG was administered through retro-orbital injection immediately after burn injury. 5 mg/kg of ICG was injected immediately (n=3), 3 hours (n=3), or 24 hours (n=4) after thermal injury. To improve SWIG contrast for image publication, an additional mouse was injected with 10 mg/kg of ICG 24 hours after burn injury.

Xenograft protocol:

Using our established human skin xenograft mouse burn model,²³ each mouse (n=4) received two xenografts. A partial thickness burn was created on one xenograft using the customized burn device (150°C x 5 seconds)²² overlapping the interface of human xenograft and mouse skin, while the other xenograft was preserved as the non-burned control region. To mimic the clinical human standard of care, the necrotic epidermis was debrided 24 hours after burn injury, then 5 mg/kg of ICG was administered through retro-orbital injection (n=3). As above, the same procedures were conducted on one additional xenograft mouse using an ICG dose of 10 mg/kg.

Small animal ICGA and SWIG imaging: Images of the burned and non-burned control regions on both the xenograft and mouse skin were acquired immediately (ICGA) and 24 hours (SWIG) after ICG injection. Brightfield and fluorescence images of burn and control sites were captured simultaneously by the small animal imager (SAI; OnLume Surgical, Madison, WI).

Histology:

Burn and control tissues from all models were harvested for histology immediately after imaging. The tissues were bisected through the wound and cryopreserved. Sequential cryosections were stained with hematoxylin and eosin (H&E) to evaluate tissue structure, and lactate dehydrogenase (LDH) to evaluate tissue viability.²²

Microscopy:

H&E and LDH tissue sections were viewed at 40X and 200X magnification using a Ti-S inverted microscope (Nikon, Tokyo, Japan). Digital images were captured with a DS Ri2 cooled color camera (Nikon), X-Cite 120LED BOOST System lamp (Excelitas), and NIS Elements Imaging Software (Nikon). Sequential cryopreserved tissue sections were analyzed with our customized NIR microscope (OnLume, Madison, WI) at 40X and 200X magnification within 72 hours of ICG injection to image ICG fluorescence.

SWIG Image Analysis:

ROI analysis was performed with ImageJ.²⁴ For the animal images obtained using SAI, analyses quantified the fluorescence within the necrotic tissue using signal-to-background ratios (SBR), defined as the ratio of the average fluorescence intensity in the burned ROI (signal) to that of the control ROI (background). For the human images obtained with OIS, SBR analyses quantified the ratio of the average fluorescence intensity within a circular ROI 2 cm in diameter (signal; due to wound size) to that within the same-sized area of adjacent, non-burned skin >1 cm away from the burn (control).

Statistical Analysis:

We conducted all statistical analyses in Prism 8 (GraphPad 2022, San Diego, CA). Quantitative data were presented as mean ± standard deviation. Two-tailed, Student’s t-tests assuming equal variance were used to compare the mean SBR for dosing experiments in mice (unpaired), mouse versus xenograft skin (paired), and ICGA versus SWIG (paired). To evaluate effects of different ICG injection times on fluorescence intensity, we assessed differences in averaged SBR by one-way ANOVA followed by Tukey’s multiple comparison between two groups. A significance level of 0.05 was used.

RESULTS

Current ICGA methods to assess burn healing potential are inconsistent and non-standardized.

There were no discernible qualitative ICGA fluorescence patterns in 12 deep partial or full thickness burns in 7 human patients. Using SPY, we found that burn-injured tissue demonstrated large inter-patient variations in fluorescence patterns and perfusion percentages that could not be attributed to dose or timing of injection after burn injury (Figure 1). All burns required excision and grafting as determined by visual assessment and contained multiple areas with greater than 100% perfusion compared to unburned skin. Multiple burns had areas of saturated ICGA fluorescence reflecting hyperperfusion, limiting the ability to quantitatively analyze. Furthermore, multiple burn perfusion percentage values were greater than 33% when compared to the area of maximum perfusion, consistent with superficial burns based on published studies,^14–15 yet visual and histologic evaluations confirmed deep partial or full thickness burns (Figure 1, burn images B (*), I (**) and K (***)).

Figure 1: Quantifying ICGA using different published methods in burn injuries requiring surgery based on visual assessment (standard of care) demonstrates large variations in fluorescence patterns and perfusion percentages.

ICGA fluorescence was imaged in 12 burns (A-L) from 7 patients after injection of ICG. Values were obtained from 10 subsampled points within a circular ROI 5 cm in diameter (=black circles). A full thickness biopsy (=X) obtained from within the ROI at time of excision and stained with lactate dehydrogenase (blue cells = viable) illustrates depth of injury for burns corresponding to *=B (deep partial burn), **=I (deep partial burn), and ***=K (full thickness burn). Images obtained with the SPY-Elite (Novadaq-Stryker). Abbreviations: ICG, indocyanine green; ICGA, indocyanine green angiography; SD, standard deviation.

ICGA and SWIG signals are dependent on ICG dose and length of time between burn injury and ICG injection.

After administering ICG to mice at varying doses immediately after burn injury to evaluate how ICG dosage impacts fluorescence, the region of full thickness burn, confirmed by LDH viability staining (Figure 2B), lacked fluorescence with ICGA and demonstrated a bright signal with SWIG that varied in a dose dependent manner (Figures 2A, columns 1 and 2 & 2C). No differences were observed in the ICGA or SWIG signals between 1 mg/kg and 2 mg/kg dose (Figure 3). A bright halo of ICGA fluorescence surrounded the burn wound in the high dose (5 mg/kg) mice only. Given changes in perfusion related to inflammation and microthromboses that occur after a burn injury, we evaluated how the timing of ICG injection impacts ICGA and SWIG fluorescence. The ICGA and SWIG SBRs were significantly higher when ICG was injected immediately than when ICG was injected 3 or 24 hours after burn injury (Figure 2A & D). Although it did not reach statistical significance, SWIG was enhanced in burn injuries when ICG was injected 24 hours compared to 3 hours after injury (Figure 2A, columns 3 and 4).

Figure 2: ICGA and SWIG fluorescence in mouse skin demonstrates variable ICGA and SWIG fluorescence intensities based on ICG dose and time of ICG injection.

(A) ICGA and SWIG in mouse skin with corresponding brightfield images of burn created at varying lengths of time in relation to a range of ICG injection doses. (B) LDH staining of the burn region of a representative sample of mouse skin confirming that the burn injury was full thickness (lack of blue stain). (C-D) Quantitative analysis of SBR between (C) ICG dose (p=0.007 for ICGA SBR and p=0.0062 for SWIG SBR), and (D) timing of ICG injection from burn injury (ICGA SBR p values for <5 min to 3 hrs, 3 hrs to 24 hrs, and <5 min to 24 hrs: 0.009, 0.781, 0.003, respectively; SWIG SBR p values for <5 min to 3 hrs, 3 hrs to 24 hrs, and <5 min to 24 hrs: <0.001, 0.383, 0.004, respectively) at α = 0.05. * = a statistically significant difference; ns = no significant difference. ICGA SBR in mouse skin is not clinically applicable due to the contribution of fluorescence signal from underneath the thin skin. Scale bar= 1 mm. Abbreviations: ICG, indocyanine green; ICGA, indocyanine green angiography; SWIG, second window indocyanine green; LDH, lactate dehydrogenase; SBR, signal-to-background ratio.

Figure 3. No significant difference was observed for ICGA or SWIG fluorescence SBR in mouse skin using low dose ICG.

Representative brightfield, ICGA, and SWIG fluorescence images of burn injuries in mouse skin with ICG doses of 1 mg/kg and 2 mg/kg, respectively. Scale bar = 1 mm. ICGA SBR p value = 0.783; SWIG SBR p value = 0.383; a = 0.05. ns = no significant difference. Abbreviations: ICGA, indocyanine green angiography; SWIG, second window indocyanine green; SBR, signal-to-background ratio.

ICGA and SWIG imaging demonstrate complementary inverse signals.

To facilitate translation of these findings to patients, ICGA and SWIG imaging were performed on a pre-clinical in vivo human skin xenograft model using 5 mg/kg of ICG injected 24 hours after partial thickness thermal injury, as confirmed by LDH staining (Figure 4A & B).²³ In comparison to the mouse skin, the mean ICGA SBR in human skin xenograft was significantly higher (Figure 4C). No significant difference was observed between the mean SWIG SBR in mouse versus human skin xenograft. Similar to mouse skin, the fluorescence signal in the deep partial thickness burn in the xenograft had a mean SBR <1 for ICGA and a mean SBR >1 for SWIG (mean SBR >1), demonstrating a statistically significant inverse relationship (Figure 4D).

Figure 4. ICGA and SWIG fluorescence in human xenograft skin demonstrates an inverse relationship between ICGA and SWIG fluorescence signals.

(A) ICGA and SWIG in human xenograft skin with corresponding brightfield images of a burn wound in human xenograft skin. Circle = area of the burn wound (4 mm diameter). (B) LDH staining confirms a deep partial thickness burn injury in the xenograft skin. (C) Quantitative analysis of SBR for ICGA and SWIG imaging for mouse versus human xenograft skin at α = 0.05 shows a significant difference in ICGA SBR (p=0.030), but not SWIG SBR (p=0.809). (D) Quantitative SBR analysis of SBR demonstrates an inverse relationship between ICGA and SWIG in xenograft skin (p=0.004). * = a statistically significant difference; ns = no significant difference. Scale bar = 1 mm. Abbreviations: ICGA, indocyanine green angiography; SWIG, second window indocyanine green; LDH, lactate dehydrogenase; SBR, signal-to-background ratio.

Burn necrosis is associated with SWIG fluorescence in human burn patients.

SWIG imaging in human burn patients revealed retention of the fluorescence signal 24 hours after ICG injection in grossly necrotic regions requiring excision, contrasting with their lack of ICGA fluorescence (Figure 5A, ROI). These findings in a deep partial thickness burn, as verified by LDH staining (Figure 5B), confirmed the inverse relationship of ICGA and SWIG fluorescence seen in the xenograft model. SWIG fluorescence was present in areas corresponding to necrotic tissue pre-excision and was absent after excision of these regions with respect to the background (Figure 5A). Under gross observation, there was a lack of SWIG fluorescence in adjacent, non-burned tissue.

Figure 5. Depth of burn necrosis is associated with fluorescence intensity.

(A) Brightfield, ICGA, and SWIG pre- and post-excision imaging obtained with the SPY demonstrate absent fluorescence on ICGA, presence of SWIG fluorescence pre-excision, and absence of SWIG fluorescence post-excision within the region of interest (ROI) compared to the background. (B) LDH staining of the biopsy and tangential excisions (blue=viable) from the burn in (A) demonstrate a deep partial thickness burn injury. (C-D) The tangential excisions and biopsy as imaged by the SPY and Small Animal Imager (SAI) show SWIG signal decreasing with depth, as confirmed by a line scan of the fluorescence within the biopsy obtained from the SAI. (E) Brightfield and SWIG intraoperative imaging pre- and post-excision using the SPY and OnLume Imaging System (OIS) of a comparable deep partial thickness burn from a second human subject with a ROI 2 cm in diameter shows the in vivo SWIG signal qualitatively decreasing with tissue excision, and quantitatively decreasing based on the pre- and post-excision SBRs calculated using OIS data. OIS was used for real-time imaging, whereas SPY images displayed here were manipulated equally after the procedure with a higher gain (2x) during image analysis to facilitate visualization of the fluorescence. X=biopsy site. Scale bar = 1 cm. Abbreviations: BF, Brightfield; ICGA, indocyanine green angiography; SWIG, second window indocyanine green; LDH, lactate dehydrogenase; ROI, region of interest; SAI, small animal imager; OIS, OnLume Imaging System; SBR, signal-to-background ratio.

To compare the performance of SPY and OIS, we imaged the excised tissue ex vivo using both devices. Given the inability to extract raw data from SPY, we were limited to qualitative comparison of the respective images. OIS demonstrated higher qualitative sensitivity (Figure 5C). Fluorescence signal in the first tangential excision was higher compared to the last excision (Figure 5C & D). This corresponded to LDH viability staining (Figure 5B) in which most of the depth in the first tangential excision demonstrated non-viable cells, while the last tangential excision had viable cells throughout that were increasing in density through the tissue section depth (Figure 5B). The gradation of fluorescence intensity throughout the burn depth from intense to minimal signal was confirmed by imaging a full thickness biopsy (Figure 5E).

We conducted intraoperative in vivo SWIG imaging with SPY and OIS of the second patient with a deep partial thickness burn. The images obtained by SPY, without the ability to adjust the gain during image capture, demonstrated a lack of observable fluorescence sensitivity (Figure 5E; SPY images manipulated post-procedure as described). Using raw data extracted from the OIS, the SWIG SBR was greater than 2, sufficiently demarcating the burn region from surrounding tissue in vivo. After the final excision, the fluorescence signal was similar to that of the surrounding background (SBR ~ 1) (Figure 5E).

Microscopic SWIG fluorescence identifies the interface of viable and non-viable tissue.

To better understand the functional mechanisms of macroscopic SWIG, LDH staining was compared to microscopic SWIG fluorescence in mouse skin. Non-viable regions identified by the lack of LDH stain coincided with SWIG fluorescence in adjacent tissue sections (Figure 6A). Necrotic epithelial-lined structures within the skin contained the strongest ICG fluorescence signal (Figure 6A). Similarly, SWIG fluorescence in the human skin xenograft correlated to non-viable tissue (Figure 6B). Interestingly, in contrast to mouse skin, burns created in human skin had a thick, superficial eschar that lacked fluorescence signal despite being completely devoid of viable cells. This corresponded to a desiccated skin layer seen macroscopically that, when debrided, resulted in a brighter SWIG signal (Figure 6C). The findings in human skin xenograft were replicated in the full thickness biopsy obtained from our second burn patient, where a similar desiccated layer of necrotic skin seen macroscopically lacked cell viability or SWIG fluorescence microscopically (Figure 6D). In both human skin xenograft and burn patient biopsies, there was a clear transition of SWIG fluorescence identifying the interface of non-viable and viable tissue.

Figure 6. ICG is retained within non-viable tissue near the interface of non-viable and viable tissue.

(A) LDH staining and SWIG fluorescence microscopy of non-burned (*) and burned (**) mouse skin. LDH staining confirms a full thickness burn (blue = viable). Insets represent magnified regions of LDH and corresponding ICG microscopy showing minimal ICG within the non-burned tissue (*) and retained ICG signal within the burned tissue (**). (B) LDH staining and SWIG fluorescence microscopy of deep partial thickness burn in human xenograft skin. SWIG inset shows retained ICG signal within deep non-viable, burned tissue closer to the interface of non-viable and viable tissue (bracket 1), and minimal ICG within the viable tissue where there is inflammatory infiltrate (bracket 2). (C) Brightfield, ICG fluorescence imaging, and LDH staining demonstrate that the necrotic tissue resulting from thermal injury corresponds to the desiccated superficial layer lacking SWIG fluorescence seen in xenograft (B) and human (D) tissues, and masks the SWIG fluorescence signal. Debridement of this necrotic layer reveals a brighter SWIG signal. Dotted region = area of debrided epidermis. (D) Brightfield, LDH, and SWIG imaging of the full thickness biopsy from a deep partial thickness burn in a human subject (surgery on post burn day 5). Dashed region = area of necrotic tissue and inflammatory infiltrate. Punch biopsy brightfield image scale bar = 1 cm; all other images scale bar = 1 mm. Abbreviations: ICG, indocyanine green; ICGA, indocyanine green angiography; LDH, lactate dehydrogenase; SWIG, second window indocyanine green.

DISCUSSION

Burn surgeons recently highlighted the need for ongoing research on wound depth assessment (2021 American Burn Association State and Future of the Science meeting; proceedings manuscript in preparation). This study provides the first-in-human evidence that SWIG fluorescence imaging may represent a new modality to detect necrotic tissue in human burn wounds. SWIG imaging has emerged as a new application of a generic FDA-approved molecule and has benefits over ICGA for burn injury. A relatively constant fluorescence signal allows imaging to occur over an extended time frame, enabling continuous, real-time guidance of burn excision. SWIG also has direct necrosis avidity, as compared to ICGA, which uses perfusion as a surrogate marker for necrosis. Additionally, burn injury is an ideal disease process for SWIG characterization given that the superficial nature of skin tissue allows for easy fluorescence visualization and the ability to know when the injury occurred to predict the timeline of inflammation and necrosis.

Our study also illustrates the challenges associated with the generalizability of published ICGA methods for burn wound healing prognostication. Despite promising preliminary studies,^10–15 our attempts to replicate these data using SPY reinforce the lack of discrete threshold values corresponding to burn depth and changing perfusion states. They also highlight the inconsistencies between patients, users, and systems that have prevented widespread clinical adoption of this technology as a stand-alone method to delineate burn wound healing potential.

Burn injury results in a consistent pattern of injury zones (Figure 7A), where the central zone of necrosis represents irreversible injury surrounded by tissue at risk for conversion to necrosis due to inflammation and ischemia.²⁵ The bright halo of fluorescence visualized on ICGA in our mouse studies only after immediate injection corresponds to the zone of hyperemia (Figure 7A), supporting EPR as a mechanism for delivery of ICG. One explanation for the differential SWIG fluorescence signal depending on ICG injection time is that the intense immediate hyperperfusion (as evidenced by the bright halo) allows more ICG to leak into the damaged tissue. In contrast, hyperperfusion is diminished by 3 and 24 hours post-injection and the relative amount of ICG that leaks into the damaged tissue is decreased, leading to a lower SWIG signal. We propose a mechanism of SWIG in burn injury in which the inflammatory state after injury delivers the high dose ICG to the interface of necrotic and viable cells. This allows binding of the fluorescence molecule to necrotic cells (Figure 7B), thereby combining two mechanisms (EPR and necrosis-avidity) of ICG retention in burn injury.

Figure 7. Schematic illustrations of Jackson’s zones of burn injury and the mechanism of SWIG.

(A) Jackson’s burn model includes a central zone of coagulation surrounded by a zone of stasis and then a zone of hyperemia. Each zone contains differing amounts of ischemia and necrosis. The zone of coagulation has irreversible damage and contains necrotic tissue. The middle zone of stasis is an area of ischemia that can progress to necrosis, increasing the extent of injury, or heal depending on many patient and wound care factors. The outer zone is the zone of hyperemia where there is an initial increase in blood flow to the burn wound. (B) Within normal tissue, in the absence of inflammation ICG stays within the vasculature and is subsequently metabolized by the liver. In burn tissue, where there is necrosis and inflammation, ICG leaks out of permeable blood vessels and is retained in tissue by binding to damaged cell membranes, making SWIG feasible. Unbound ICG leaves the region of interest to undergo metabolism and will not contribute to SWIG fluorescence. Abbreviations: ICG, indocyanine green.

Translation of our preclinical models to patients is a priority; thus, we evaluated a 24-hour time point after injury to mimic a clinically feasible time for ICG infusion. We found that the fluorescence continued to be as localized to the burn region as it was in earlier time points. Our findings from the mouse skin model were also translated to the human skin xenograft model to determine the different local effects of thicker, clinically relevant skin tissue. By evaluating the fluorescence signal from burn injury overlapping human and mouse skin in the xenograft model, we demonstrated the effect of skin thickness on fluorescence. We also showed that ICGA in a mouse model is challenged by the fact that the organs underlying the thin skin also demonstrate fluorescence, causing signal interference. For these reasons, we confirmed our preclinical SWIG findings in an initial study of two burn patients, the first of which was imaged before OIS was available for clinical use. The comparison of these two imaging devices in our patients was critical in identifying the shortcomings of SPY, namely the inability to extract raw fluorescence data and insufficient sensitivity, which preclude in vivo imaging and data analysis across platforms. Further, the need to have bright lights turned off during imaging interrupts operative flow. In contrast, OIS has the high sensitivity required to distinguish the fluorescence without turning off operating room lights, and offers the ability to perform quantitative comparison of fluorescence across platforms at a variety of overall signal levels by capturing raw fluorescence data. Importantly, OIS the processed data is displayed to the user in real-time, resulting in better quality in vivo and ex vivo NIR imaging of human tissue.

To further develop SWIG imaging for burn injury, we must understand how ICG is retained in tissue. We demonstrated localization of ICG within burn-injured tissue that consistently identified the deep border of necrosis and viable tissue, an area that is useful to delineate intraoperatively. Further investigation is required to understand specific ICG binding and retention patterns.

While this study focused on how SWIG imaging could be used as an intraoperative tool to guide selective excision of necrotic tissue, we envision it could also be used as a prognostication tool that defines the burden of necrotic tissue early after injury to improve decisions regarding operative intervention. Improving the precision of surgical decision-making may increase the use of therapeutic options other than autologous skin grafting for wound closure. Our findings lay the groundwork for understanding how SWIG fluorescent signal and cellular necrosis overlap, and will require ongoing optimization and customization using an imaging system capable of collecting raw data. We envision that the correlation of histology and fluorescent signal, as demonstrated in this pilot study, will serve as input data to train a software algorithm in identifying the fluorescence threshold that equates to non-viable tissue with high diagnostic accuracy. In this future state, SWIG imaging is obtained at the time of excision and a threshold is set that outlines the areas of the burn that meet criteria for necrosis (Figure 8). As the surgeon performs tangential excision, the contoured region would automatically adjust to the remaining necrotic tissue until excision was complete and fluorescence was absent, thus preventing excision of tissue with regenerative capacity.

Figure 8. Schematic illustration of our proposed intraoperative SWIG fluorescence imaging technique to guide selective excision of necrotic regions within a burn injury.

This burn wound has both a deep and a superficial region. The deep (necrotic) region has a corresponding SWIG SBR > n (to be determined in future studies), indicating the need for excision of this area. The superficial region has a corresponding SWIG SBR <1, indicating that this area has regenerative capacity and does not require excision. After the necrotic area has been excised to a healthy wound bed (post-excision), there is minimal SWIG fluorescence (SBR = 1) remaining within this region, indicating that excision is complete. Abbreviations: SWIG, second window indocyanine green; SBR, signal-to-background ratio.

A major strength of this study is the use of multiple burn models to demonstrate the utility of SWIG in identifying necrotic tissues; however, we recognize the low number of patients as a limitation. Future studies will focus on enrolling patients early after injury to evaluate this modality as a tool to evaluate wound healing potential, as well as patients with burns requiring surgical excision. Finally, additional study is required to identify the influence of inflammation and necrosis on SWIG fluorescence signal at the cellular level as we define the role of SWIG imaging in burn management.

CONCLUSIONS

Using a multi-model approach with a focus on rapid translation to clinical practice, we have shown SWIG imaging as a possible future modality to detect the necrosis burden and assess healing potential within a burn injury through a unique application of an old molecule. Future work in understanding the mechanisms by which the SWIG technique identifies necrotic tissue will enhance its application across many surgical diseases.

ABBREVIATIONS

ICG

Indocyanine green

ICGA

Indocyanine green angiography

SWIG

Second window indocyanine green

EPR

Enhanced permeability and retention

NIR

Near-infrared

UW

University of Wisconsin

ROI

Region of interest

OIS

OnLume Imaging System

SAI

Small Animal Imager

H&E

Hematoxylin and eosin

LDH

Lactate dehydrogenase

SBR

Signal-to-background ratio

FDA

United States Food and Drug Administration

SD

Standard deviation

BF

Brightfield