Image-guided cancer surgery using near-infrared fluorescence

Abstract

Paradigm shifts in surgery arise when surgeons are empowered to perform surgery faster, better, and/or less expensively. Optical imaging that exploits invisible near-infrared fluorescent light has the potential to improve cancer surgery outcomes while minimizing anesthesia time and lowering healthcare costs. Because of this, the last few years have witnessed an explosion of proof-of-concept clinical trials in the field. In this review, we introduce the concept of near-infrared fluorescence imaging for cancer surgery, review the clinical trial literature to date, outline the key issues pertaining to imaging system and contrast agent optimization, discuss limitations and leverage, and provide a framework for making the technology available for the routine care of cancer patients in the near future.

Introduction

Improvements in preoperative imaging techniques have made a meaningful impact on cancer patient care. However, during actual cancer surgery, the eyes and hands of the surgeon remain the dominant ”imaging modalities” used to decide which tissue needs to be resected, i.e., malignant cells, and which tissue needs to be avoided, i.e., normal cells. Palpation and visual inspection are not always sufficient for discriminating between tissue types, though, leading to irradical resections or unnecessary removal of healthy tissue. In breast cancer, for example, many of which are non-palpable, margin positivity rates range from 5 to 49%.^1,2 In some cases, ultrasound and/or x-ray fluoroscopic imaging are used during cancer surgery, but these modalities lack the possibility of using targeted contrast agents to specifically visualize certain cell types, expose patient and caregiver to ionizing radiation, and/or require direct contact with the body. Intraoperative MRI and CT scanning have played a significant role as well, especially in the field of neurosurgical image guidance.³ However, intraoperative systems are costly and complex, and are currently used mainly for neurosurgery at major medical centers.

Over the past several years, intraoperative imaging using invisible near-infrared (NIR) fluorescent light has entered the surgical theatre to fill the gap between preoperative imaging and intraoperative reality.^4,5 Whereas visible light penetrates tissue on a micron scale, NIR light (700 nm – 900 nm) can travel millimeters, up to centimeters, through tissue.⁶ Because tissue exhibits almost no autofluorescence in the NIR spectrum, the signal-to-background can be maximized using NIR fluorescent contrast agents, creating “white stars in a black sky”.⁷ In addition, it does not use ionizing radiation, making it an inherently safe technique provided that attention is paid to laser illumination levels. And, as NIR light is invisible to the human eye, it does not alter the look of the surgical field, thus minimizing the learning curve.

Specialized intraoperative imaging systems for open surgery,^8–14 laparoscopy,^15,16 thoracoscopy,^17,18 and robotic surgery^19,20 have recently become available for clinical trials.²¹ Using these systems, NIR fluorescent contrast agents can be visualized with acquisition times in the millisecond range, enabling real-time guidance during surgery (Figure 1).

Figure 1. The mechanics of NIR fluorescence imaging.

A NIR fluorescent contrast agent is administered intravenously, topically, or intraparenchymally. During surgery, the agent is visualized using a NIR fluorescence imaging system of the desired form factor, i.e., above the surgical field for open surgery or encased within a fiberscope for minimally-invasive and robotic surgery (open surgery form factor shown). All systems must have adequate NIR excitation light, collection optics and filtration, and a camera sensitive to NIR fluorescence emission light. An optimal imaging system includes simultaneous visible (i.e., white) light illumination of the surgical field, which can be merged with NIR fluorescence images. The surgeon display can be one of several forms factors including a standard computer monitor, goggles, or a wall projector (monitor form factor shown). Current imaging systems operate at a sufficient working distance that enables the surgeon to operate and illuminates a sizable surgical field.

To date, NIR fluorescent contrast agents specific for many different targets have been developed, including agents for cancer cells,^22–24 sentinel lymph nodes,^25–27, neurological diseases, ^28,29 cardiovascular diseases,^30,31 skeletal processes,³² renally cleared agents for ureter imaging,³³ and hepatically cleared agents for bile duct imaging.³⁴ Optically-active nerve cell agents^35,36 have also been described, but are yet to achieve NIR wavelengths.

Although these research results lay the foundation for a future revolution in patient care, excitement in the field needs to be tempered with the reality of the underlying physics. Like visible light, NIR light is attenuated by absorption and scatter in living tissue. In fact, total attenuation (the sum of attenuation due to absorption and scatter) is exponential as a function of depth. Thus, only 1-millionth to 1-billionth of the photons launched into a tissue as fluorescence excitation is even potentially recoverable as fluorescence emission, and only 10–25% of these photons are truly recoverable because of the finite quantum yield of most NIR fluorophores. Taken together, depth of detection for even bright targets will be constrained to ≈ 5 mm, but could be slightly more or less depending on the particular tissue’s optical properties. While this is more than adequate for tumor margin detection and many surgical applications, it requires leveraging of the technology with other imaging modalities in some clinical situations (discussed below).

Absorption and scatter also limit the ability to quantify NIR fluorescent signals. Without knowing (or measuring) absorption, scatter, and anisotropy of the tissue being imaged, only qualitative information can be inferred (reviewed in ³⁷). Correction for attenuation of excitation light³⁸ can certainly assist with target detection, although over-compensation can also cause false-positives. Full correction for optical properties can provide quantitative measurements of NIR fluorescent signal.^39,40 However, the use of quantitative metrics for intraoperative decision-making is still in its infancy. In the case of cancer margin detection, a qualitative present/absent is the relevant metric, and for perfusion studies, relative signal over time after injection of a tracer provides quantitative information without requiring the measurement of tissue optical properties.^41,42

This review focuses on the clinical applications of NIR fluorescence imaging in cancer surgery, recent developments, and the long-term potential benefit for patients.

Clinical applications of NIR fluorescence imaging

Intraoperative NIR fluorescence imaging depends on the availability of a NIR fluorescent contrast agent and an intraoperative imaging system to visualize the otherwise invisible contrast agent during surgery.⁴ Indocyanine green (ICG) is the only 800 nm NIR fluorescent contrast agent that is approved for this clinical indication by the Food and Drug Administration (FDA) and the European Medicines Agency (EMEA). Methylene blue (MB) has been applied clinically for many years as a visible (dark blue) contrast agent. As MB was introduced into clinical practice in an era when no formal approval was needed, no evaluation by the FDA, EMEA, or comparable authorities has been performed, but it remains widely used. When sufficiently diluted, MB acts as a 700 nm fluorophore and has recently been used in NIR fluorescence clinical studies.^43–45.

5-aminolevulinic acid (5-ALA) is the major substrate for protoporphyrin synthesis and has been used clinically for tumor detection (fluorescence imaging) and tumor treatment (photodynamic therapy). 5-ALA (typically administered in a topical or oral form) induces synthesis and accumulation of the fluorescent molecule protoporphyrin IX (PpIX) in epithelia and neoplastic tissues, among them malignant gliomas and meningiomas.^46–48 PpIX has two major fluorescence emission peaks with approximately half of its fluorescence emission centered ≈ 700 nm (i.e., NIR). 5-ALA has been studied extensively in the field of neurosurgery as described below.

Interestingly, because MB, 5-ALA-induced PpIX, and ICG have distinct spectral properties, they could theoretically be used simultaneously in the same patient, provided that the imaging system is capable of discriminating one from the other. Neither ICG, MB, nor 5-ALA are ligand-targeted contrast agents. The first clinical report of non-targeted (non-NIR) fluorescence surgical imaging dates back to 1948, when Moore et al used fluorescein in neurosurgical interventions for the localization of brain tumors.⁴⁹ The first clinical report of targeted (non-NIR) fluorescence surgical imaging dates back to 1992, when Folli and colleagues used carcinoembryonic antigen-targeted antibodies labeled with fluorescein to visualize colorectal carcinomas in vivo in patients.⁵⁰ In 2012, Van Dam and colleagues described the successful use of folate conjugated to fluorescein for intraoperative visualization of ovarian cancer cells for debulking surgery.⁵¹ While these landmark studies provided essential proof-of-principle, fluorescein and other visible fluorophores are not optimal for cancer surgery because high absorption and scatter result in interrogation of only the surface layer, and, importantly, high autofluorescence from surrounding tissue reduces contrast.

Nevertheless, using the existing major NIR fluorophores, ICG, MB, and 5-ALA-induced PpIX, proof of principle clinical studies in several types of surgery have been performed in recent years (Table 1).

Table 1.

Overview of Clinical Applications

Application	Cancer Type	Imaging System	Contrast Agent	Clinical Status*
SLN mapping	Breast	Open	ICG11,52–56	Clinical
	Melanoma	Open	ICG62,63, MMl69	Clinical
	Head and Neck	Open	ICG63,70, MMl27	Clinical
	Lung	VATS	ICG17,71	Clinical
	Esophagus	Open	ICG75,127	Clinical
	Stomach	Open	ICG74,128	Clinical
		Lap	ICG73,129	Clinical
	Colorectal	Open	ICG9,130, HSA800 (ex vivo)59,60	Clinical
		Lap	ICG15,58,61	Clinical
	Cervix	Open	ICG13,64,65	Clinical
	Vulvar	Open	ICG66–68	Clinical
	Endometrial	Lap	ICG72	Clinical
	Prostate	Lap/Rob	ICG131, MMI16	Clinical
	Penile	Open	MMI69	Clinical
Tumor imaging (in vivo)	Colorectal liver metastases	Open	ICG82,83,89	Clinical
	Hepatocellular carcinoma	Open	ICG10,82	Clinical
		Lap	ICG132	Preliminary
	Ovarian cancer metastases	Open	ICG93	Preliminary
	Breast cancer	Open	Bevacizumab-800CW113	Preliminary
	Insulinoma	Open	MB43	Preclinical
	Primary solitary tumor of the pancreas	Open	MB45	Preliminary
	Bladder cancer	Lap	5-ALA97	Clinical
	Brain tumors	Open	5-ALA,46–48, ICG91,92	Clinical
Tumor imaging (topical)	Bladder cancer	Cyst	5-ALA99, HAL100	Clinical
	Colon cancer	Endo	gGlu-HMRG122	Preclinical
Vital structures imaging	Bile ducts	Open	ICG133,134	Clinical
		Lap	ICG101,135,136	Clinical
	Ureters	Open	MB44	Preliminary
	Lymph Flow	Open	ICG55,102	Clinical
	Nerves	Open	Various35,118	Preclinical
Vascularization	Bowel anastomoses	Open	ICG104,105	Clinical
		Lap	ICG106	Clinical
	Reconstructive surgery	Open	ICG108–110	Clinical

^*Clinical = Multiple independent groups have reported successful application; Preliminary = Initial clinical application has been reported; Preclinical = Application has only been demonstrated in a preclinical setting.

Abbreviations: 5-ALA = 5-aminolevulinic acid; Bevacizumab-800CW = IRDye 800CW conjugated to Bevacizumab; gGlu-HMRG = g-glutamyl hydroxymethyl rhodamine green; ICG = indocyanine green; HAL = hexyl aminolevulinate; HSA800 = IRDye 800CW conjugated to human serum albumin; Lap = laparoscopic surgery; MB = methylene blue; MMI = multimodal imaging agent ICG-^99mTechnetium-nanocolloid; Rob = robotic surgery; SLN = sentinel lymph node; VATS = video-assisted thoracoscopic surgery;

Sentinel lymph node mapping

Sentinel lymph node (SLN) mapping is standard-of-care in a variety of cancers, including breast cancer and cutaneous melanoma. Currently, most centers perform SLN mapping using a radioactive tracer, a visible blue dye such as isosulfan blue or patent blue, or a combination of the two. Although in most cases acceptable results are obtained using these methods, they both have some drawbacks. Visible blue dyes stain the patient and the surgical field and cannot be visualized below the surface of tissue. Radioactive tracers expose patients and caregivers to ionizing radiation, are expensive, and imaging suffers from poor spatial and temporal resolution. For SLN mapping using NIR fluorescence imaging, contrast agents are injected at a low concentration with no staining of the surgical field, no ionizing radiation is used, and an improvement is observed over blue dyes in terms of depth sensitivity. For SLN mapping, targeted contrast agents are not necessary, and because ICG is clinically available, many SLN studies were undertaken as soon as the first intraoperative imaging systems became available (Figure 2; Table 1).⁵ To date, NIR fluorescence-guided SLN mapping has been extensively studied in breast cancer,^11,52–56 as well as in colorectal cancer, ^15,57–61 skin cancer,^62,63 cervical cancer,^13,64,65 vulvar cancer, ^66–68 head and neck, ^63,69,70 lung cancer, ^17,71 penile cancer, ⁶⁹ endometrial cancer,⁷² gastric cancer, ^73,74 and esophageal cancer. ⁷⁵

Figure 2. Examples of intraoperative NIR fluorescence imaging.

a. Example of SLN mapping using NIR fluorescence imaging in a patient with cutaneous melanoma. Displayed are the color images (left), NIR fluorescence images (middle), and pseudocolored (NIR fluorescence in lime green) merges of the two (right). The lymphatic channel (arrowhead) and SLNs (arrows) can be clearly identified percutaneously and in real-time (top panel). Identification of a SLN (bottom panel) is demonstrated using 800 nm NIR fluorescence imaging 15 min after injection of 1.6 mL of 1000 μM ICG admixed with human serum albumin around the tumor. All images were acquired in real time using the mini-FLARE¹² imaging system. NIR excitation fluence rate was approximately 8 mW/cm² and camera exposure time was 10 ms. Scale bars represent 1 cm. Reproduced with permission.⁶³

b. Example of NIR fluorescence imaging of the ureter during lower abdominal surgery in a patient with ovarian carcinoma. Intraoperative imaging of the ureter (arrow), 45 min after administration of 1 mg/kg methylene blue. Displayed are the color images (left), 700 nm NIR fluorescence images (middle), and a pseudocolored (NIR fluorescence in lime green) merge of the two (right) acquired with the mini-FLARE imaging system. NIR excitation fluence rate was approximately 1 mW/cm² and camera exposure time was 150 ms. Scale bars represent 1 cm. Reproduced with permission.¹³⁷

c. Example of fluorescence imaging in brain surgery during resection of a glioblastoma multiforme (+) using oral 5-ALA-induced PpIX. White light image (left), visible fluorescence (middle) and quantitative fluorescence images overlayed onto white light images (right) are displayed. Reproduced with permission.³⁹

These first studies demonstrate the feasibility of NIR fluorescence imaging during surgery. Comparison of NIR fluorescence imaging to blue dyes indicates that NIR fluorescence imaging may substitute blue dyes, as it outperforms the blue dyes due to increased tissue penetration depth and lack of staining of the patient and the surgical field.^53,76,77 In particular in countries were radiotracers are not allowed or available, NIR fluorescence imaging could replace blue dyes as routine SLN mapping methodology. However, as current imaging systems in combination with ICG show depth penetration less than 1 centimeter, ongoing studies are assessing the non-inferiority of NIR fluorescence compared to radioactivity, which is of particular clinical significance when lymph nodes are located deeper within the patient, for example in more obese patients.

Combination of radioactivity and NIR fluorescence

To overcome the issue of limited depth penetration, a combination of NIR fluorescence and radioactivity has been reported for SLN mapping.^16,27,69,78 This combination is complementary: the superior depth penetration of radioactivity is used to perform gross navigation to the SLN, after which the superior spatial and temporal resolution of NIR fluorescence facilitates image-guided identification and resection of the SLN. When a preoperatively injected, combined NIR fluorescent and radioactive tracer is used, no intraoperative injection of tracer is necessary and the preoperative injection by the nuclear medicine physician suffices for both preoperative lymphoscintigraphy and intraoperative SLN detection, reducing the duration of surgery and anesthesia and potentially reducing costs.^16,27,69,78

Targeted SLN tracers

An optimal tracer for SLN mapping migrates quickly from the injection site to the lymph node and remains in the first draining node without migrating to higher tier nodes. To prevent migration to higher tier nodes, radiotracers are commonly conjugated to a colloid to increase the hydrodynamic diameter or to a ligand to increase retention. In the United States, a sulfur-based colloid is most widely used, whereas in Europe, an albumin-based colloid is commonly used. Indocyanine green cannot be conjugated covalently to sulfur colloid without altering its chemical structure, although non-covalent adsorption of ICG to albumin and albumin nanocolloid appears to work well. ^25,79 A novel strategy for lymph node retention has been developed by Vera and colleagues by creating a lymphatic tracer that is specific for a receptor (mannose-binding protein) found on reticuloendothelial cells of lymph nodes.⁸⁰ While this tracer was primarily designed for radiolabeling, recent studies have been published in which this tracer is conjugated to a fluorescent label. ^26,81

Tumor imaging

Although ICG’s use in SLN mapping has paved the way for NIR fluorescence imaging in the operating room, one of the key applications in cancer surgery is intraoperative tumor visualization. In order to visualize tumors using NIR fluorescence imaging, a contrast agent should accumulate in or around a tumor. Although a variety of tumor specific NIR fluorescent contrast agents have been developed and applied preclinically, none of these has obtained full clinical approval by either the FDA or the EMEA. Using the clinically available contrast agents ICG, MB, and 5-ALA first-in-human intraoperative tumor imaging studies have been described in various cancer types.

ICG

After intravenous administration, ICG is cleared by the liver, potentially enabling intraoperative identification of liver lesions. Indeed, Ishizawa and colleagues were the first to demonstrate clear visualization of both colorectal liver metastases and hepatocellular carcinomas (HCC).⁸² In their study and subsequent studies, several hours to days after intravenous injection of ICG, a fluorescent rim could be visualized intraoperatively around the tumors, with almost no background signal in the surrounding healthy liver parenchyma.^83–85 Tumors that could be visualized using NIR fluorescence imaging were all relatively superficial (less than 8 mm below the liver surface), tumors located deeper could not be visualized by NIR fluorescence imaging.

Several clinical studies reported on NIR fluorescence imaging in patients suffering from HCCs using ICG.^10,82 Surprisingly, ICG accumulated in the HCCs itself, as opposed to the accumulation in a rim around colorectal liver metastases. Intraoperative ultrasound is routinely used before resection of liver tumors in order to identify tumors that were missed preoperatively. It is also used intraoperatively in conjunction with inspection and palpation. One of the main limitations of intraoperative ultrasound is the hampered detection of superficial and small tumors.⁸⁶ NIR fluorescence performs well in superficial and small tumors, but is unable to visualize deeper tumors. Therefore, in liver surgery, NIR fluorescence imaging can be seen as a potential adjunct to conventional imaging techniques for the preoperative and intraoperative detection of primary liver tumors and hepatic metastases. Satou and colleagues used NIR fluorescence imaging after ICG injection to identify extrahepatic metastases of HCC.⁸⁷ They report the detection of 5 metastases that were otherwise not detected in 2 out of 17 patients (11.7%), confirming the complementary value of NIR fluorescence imaging. Yokoyama and colleagues demonstrated that NIR fluorescence imaging after ICG injection identified hepatic metastases of pancreatic carcinoma that were otherwise not detected in 8 out of 49 patients (16.3%), potentially preventing those patients from unnecessary radical resections and associated morbidity.⁸⁸ A recent study by our group⁸⁹ found that in 5 of 40 patients (12.5%), ICG-based NIR fluorescence imaging detected colorectal metastases in liver that were otherwise undetectable by preoperative CT, intraoperative ultrasound, palpation, and visualization. As might be expected, all detectable metastases were less than 8 mm from the liver surface.

A novel and rapidly growing field of NIR fluorescence guided cancer surgery is related to neurosurgery. ICG has been used to intraoperatively identify high-grade gliomas, meningiomas, hemangioblastomas and pituitary tumors.^90–92 The previously mentioned tumors could be visualized by intravenously administering 5 to 25 mg ICG.

After intravenous injection, ICG binds to plasma proteins, thereby increasing its hydrodynamic diameter. This potentially enables ICG to accumulate in tumors due to enhanced permeability and retention. In 2006, Horowitz and colleagues presented the use of ICG via this mechanism for intraoperative detection of ovarian cancer metastases.⁹³

MB

MB has previously been described to stain parathyroid glands and insulinomas the color blue after high-dose intravenous and intra-arterial injection, respectively.^94,95 Winer and colleagues demonstrated preclinically that dilute MB can be used as an NIR fluorescent contrast agent for identifying insulinomas.⁴³ Although clinical use of MB for insulinomas has not yet been reported, it was successfully used to visualize a rare solitary fibrous tumor of the pancreas in a patient.⁴⁵

5-ALA

5-ALA-induced PpIX has been studied extensively in the field of neurosurgery, mostly to intraoperatively identify brain tumors such as malignant gliomas. Stummer et al. showed that visualization of PpIX fluorescence after 5-ALA administration (oral administration) led to a significant increase in the incidence of complete resection (65% compared to 36%), improved progression-free survival at 6 months (41% compared to 21%), fewer reinterventions, and delayed onset of neurological deterioration.^48,96 Furthermore, 5-ALA-induced PpIX has been used to identify the hardly visible urothelial bladder cancer during fluorescence guided cystoscopies which was reviewed by Jocham et al.⁹⁷

Topical contrast agents

Besides systemically injected contrast agents, topical application of NIR fluorescence contrast agents can be of benefit, for example in colonoscopic or cystoscopic procedures. The successful use of 5-ALA and derivatives has been described in bladder cancer.^98–100 Although visible (red) light fluorescence was used for imaging in these studies, the tumor cell-synthesized Protoporphyrin IX actually has half its emission spectrum in the NIR.

Imaging of vital structures

Iatrogenic damage to vital structures occurs frequently during cancer surgery. For example, nerve damage is relatively common and can result in postoperative pain or loss of specific functions, such as incontinence or impotence, especially in rectal cancer or prostate cancer surgery. Damage to ureters or bile ducts is relatively rare, but both are associated with severe complications, such as renal dysfunction and biliary peritonitis. Damage to these structures is usually a result of inadequate identification during surgery. For this purpose, NIR fluorescence imaging has also been described for intraoperative identification of vital structures. As ICG is cleared by the liver and excreted into the bile ducts, studies have described intraoperative NIR fluorescence cholangiographies after intravenous administration of ICG.^20,101 Using MB, which is cleared predominantly by the kidneys, ureters could be identified intraoperatively using NIR fluorescence imaging (Figure 2).⁴⁴ Due to their pelvic location, ureters can be difficult to visualize, in particular when tumor or inflamed tissue is covering them. The increased depth penetration of NIR fluorescence can therefore be of added value over visible light imaging. NIR fluorescence imaging can be of particular benefit in laparoscopic and robotic surgery, where there is diminished tactile feedback. NIR fluorescence-capable fiberscopic systems are now readily available for these applications. Finally, subcutaneously administered ICG has been used in multiple clinical studies to visualize lymphatic flow in real-time, which will likely aid in understanding the mechanism of lymphedema, a morbid complication of several cancer surgeries.^55,102

Vascularization

Vascular perfusion of anastomoses

Anastomotic leaks after intestinal surgery for cancer remain a relatively common and severe complication.¹⁰³ Occurrence of anastomotic leaks is presumably caused by insufficient perfusion of an intestinal anastomosis. NIR fluorescence can be used to visualize blood perfusion after intravenous injection of ICG. Several groups reported on its ability to assess anastomotic blood perfusion after intestinal surgery. ^104–106 Although retrospectively studied, Kudszus and colleagues described a reduction in the risk of revision due to anastomotic leak by 60% in patients whose anastomosis was examined using NIR fluorescence angiography compared to historically matched patients without imaging.¹⁰⁴

Vascular perfusion of flaps in reconstructive surgery

Vessel identification and selection is essential during cancer reconstructive surgery using free or pedicled flaps. NIR fluorescence imaging provides the opportunity to identify vessels intraoperatively and to assess perfusion in these flaps.^107,108 In breast reconstructive surgery, NIR fluorescence angiography has been used for vessel identification and selection, and to monitor venous outflow.^109,110 Newman and colleagues describe in their study that arterial and venous problems were identified intraoperatively in 4 out of 8 (50%) patients, which changed the operative plan in 3 of 4 cases, leading to a 100% flap survival outcome.¹⁰⁹

Key parameters: target, contrast agent dose, and timing

Optimal visualization is achieved by maximizing the signal-to-background ratio, thus maximizing contrast. In order to do so, the chosen target, the type of contrast agent, the route of administration (i.e. intravenous, local injection, topical spray), and the imaging goal (i.e. sentinel lymph node mapping, tumor imaging, etc.) all have to be factored in.

Perhaps the most important parameter for imaging is the target. Is it an abundant receptor on the cell surface, a sub-cellular organelle, or the substrate for a membrane transporter? Because all fluorescent techniques rely on concentration, the abundance and availability of the target are key concerns. Keep in mind, for example, that a cell surface receptor with a total number of receptors per cell (BMax) of 10⁴ can only achieve a fluorophore concentration of ≈ 17 nM, assuming a 1:1 ratio of targeted NIR fluorophore to receptor. Even the most highly abundant receptors, with a BMax of 10⁵–10⁶ can only achieve ≈ 1 μM concentration.

Within the first few seconds after intravenous injection, NIR fluorescent contrast agents highlight the arterial then venous systems (Figure 3). Over the next minutes to hours, hepatic and/or renal clearance will occur in parallel with tissue biodistribution and target binding. Some NIR fluorophores, such as MB, accumulate in certain cells by unknown mechanisms. Others are either targeted to particular receptors or membrane transporters, or accumulate passively due to structural abnormalities in the tissue (e.g., the enhanced permeability and retention or EPR effect).¹¹¹

Figure 3. Administration, Biodistribution, and Clearance of a NIR fluorescent contrast agent.

Schematic (top row) and clinical example (middle row) of the four key phases of NIR fluorescence imaging after intravenous administration of a NIR fluorophore. Top-row: Shown from left-to-right are the different phases over time: first, a NIR fluorescent probe is administered. After intravenous administration, for example, NIR fluorescence signal is visualized in the vasculature (second panel). Then, the contrast agent is distributed to all tissues in the body (third panel) but the target remains obscured. After adequate clearance time (right panel), adequate contrast between target and surrounding tissue is achieved. Bottom-row: Left: administration of NIR fluorescent contrast agent. Second panel: ICG fluorescence as observed in the vasculature of an abdominal tissue flap to be used for breast reconstructive surgery. Third panel: ICG fluorescence distribution in the liver. Right panel: ICG clearance from the liver over time, revealing a metastatic colon cancer metastasis (arrow). Shown in all panels are pseudocolored (lime green) merges of NIR fluorescence and color video acquired using the mini-FLARE imaging system. Scale bars represent 1 cm.

Bottom row: schematic of the four key phases of NIR fluorescence imaging after administration of 5-ALA. Shown from left to-right are the different phases over time: first, the non-fluorescent 5-ALA is orally administered (first panel). After uptake in the bloodstream via the gastrointestinal tract, the substance accumulates in tumor cells (second panel), where it is metabolized to the fluorescent protoporphyrin IX (third panel). After metabolism, tumor demarcation can be visualized using the inherent fluorescent properties of protoporphyrin IX (fourth panel).

As the contrast agent clears from the bloodstream and normal tissues, and hopefully retained in tumor tissue, the SBR at the target site increases. In order to determine the necessary dose and timing, dosing, biodistribution, and pharmacokinetic studies are therefore essential. For example, an imaging time too short relative to clearance will result in a low SBR due to high background. Conversely, an imaging time too long relative to target binding will result in a low SBR due to loss of signal.

In SLN mapping, the contrast agent migrates from the injection site to the sentinel lymph node. Depending on the contrast agent’s characteristics, flow to higher tier nodes might occur over time. To prevent misidentification of the SLN, dynamic imaging can be employed, starting directly after injection. Although counterintuitive, injection of a concentration of fluorophore that is too high results in a decrease of SLN signal due to quenching.⁴ In dose-finding studies for SLN mapping, an ICG injection concentration of 500 μM in a volume of 1.6 mL was found to be optimal, although these numbers may vary among tumor types, and the distance between injection site and the SLN.¹²

Current developments

Clinical studies have demonstrated the feasibility of intraoperative imaging using near-infrared fluorescence in various applications. These studies, however, are based on contrast agents that were already available clinically. In order to determine the true clinical benefit of this technique, development and clinical assessment of contrast agents tailored to specific applications is essential.

Contrast agent development and optimization

Development of novel NIR fluorescent contrast agents is dependent on the availability of clinically-compatible fluorophores. Although many fluorophores have been evaluated in a preclinical setting, to date only two fluorophores have been reported to be in the process of clinical translation: IRDye 800CW (LI-COR Biosciences, Lincoln, NE) and ZW800-1 (The FLARE Foundation, Wayland, MA). Both compounds are small molecules, characterized as non-toxic in initial toxicity studies and are conjugatable to targeting moieties. In addition to their ability to be conjugated to various targeting ligands, the purely renal clearance of ZW800-1, and the combined renal and hepatic clearance of IRDye 800CW, enable these agents to be used for imaging of ureters and bile ducts.

As recently reviewed by Scheuer and colleagues,¹¹² clinically approved targeted antibodies are available for various tumors and tumor markers, for example bevacizumab (Genentech, San Francisco, CA) against the vascular endothelial growth factor A, cetuximab against the epidermal growth factor receptor (Bristol-Myers Squibb, New York City, NY, and Eli Lilly, Indianapolis, IN), and trastuzumab against the human epidermal growth factor receptor 2 (Genentech, San Francisco, CA). Combining these already approved antibodies with a NIR fluorophore could usher in a new generation of NIR fluorescent contrast agents. Indeed, the first clinical trial using bevacizumab conjugated to IRDye 800CW has been approved and is accruing patients.¹¹³ In addition to antibodies, other targeting molecules, such as peptides, have been successfully translated into the clinical using radiolabeling and suggest potential for NIR fluorescence imaging. For example integrin αVβ3 targeting using cyclic arginine-glycine-aspartic acid (cRGD) has been demonstrated clinically in positron emission tomography studies.¹¹⁴ A fluorophore conjugated to this peptide has been described preclinically for tumor imaging (cRGD-ZW800-1).²³ To overcome disadvantages of intact antibodies, investigational contrast agents are being developed using smaller antigen binding fragments such as nanobodies.¹¹⁵ Nanobodies have a smaller hydrodynamic diameter and can therefore extravasate more easily than intact antibodies, even in areas of high interstitial pressure (i.e., solid tumors). Indeed, nanobodies against the epidermal growth factor receptor have been shown to outperform whole antibodies for tumor imaging.¹¹⁶ As heterogeneity in tumors is an important problem when selecting a targeting ligand,¹¹⁷ it is likely that a cocktail of several targeted NIR fluorophores will be employed once more tumor-specific ligands are developed and tested.

In order to visualize nerves, several research groups are developing specific contrast agents. One approach is focused on a family of fluorescent dyes that bind to the myelin of nerves.¹¹⁸ However, these contrast agents currently fluoresce in the non-optimal visible spectrum. Another approach is the use of NIR fluorescent peptides that bind to nerve sheaths.³⁵ No clinical trial has yet been reported on nerve-specific agents, although the field eagerly awaits such a development.

A distinct class of NIR fluorescent contrast agents encompasses activatable agents. These agents are non-fluorescent or low-fluorescent in the unactivated state, and become fluorescent after activation by a molecular target, such as a tumor-specific enzyme. Preclinical studies show adequate tumor detection after intravenous injection of activatable agents. ^{22,24,119–121} Other reports show rapid tumor visualization after topical application of activatable agents.¹²² Several of these agents are in the process of obtaining regulatory approval for proof-of-principle trials, however, to date no such trials are approved and accruing patients.

Imaging system development and optimization

Over the last few years, various imaging systems have been developed for intraoperative NIR fluorescence imaging by academic and industry groups.^{8–11,13–20,120} Besides imaging systems for open surgery, NIR fluorescence imaging systems are available for laparoscopic, thoracoscopic, and robotic surgery. Most of these systems mark the first production grade development iteration, with optimized successors being developed.

Key issues that inform imaging system design are the fluorescence excitation light, the optics, usability (ergonomics), and cost. Maximizing the fluence rate of the excitation light increases tissue penetration depth. However, for laser-based light sources, skin/eye exposure, irreversible photochemical bleaching of the NIR fluorophore, and tissue heating are the limiting factors. In general, fluence rates are limited to the 10–25 mW/cm² range in order to avoid these issues and the need for laser goggles. Initial imaging system prototypes were often assembled using off-the-shelf optics, which can distort and attenuate NIR light, causing suboptimal detection of NIR fluorescence.⁴ Current versions of NIR fluorescence imaging systems are manufactured with optimized optics for NIR light and do not suffer from these problems. The possibility of visualizing multiple contrast agents simultaneously, for example for tumor identification and nerve visualization, would be beneficial. For this purpose, imaging systems are in development that can simultaneously acquire multiple wavelengths. ^11,13 Furthermore, techniques are being implemented to discriminate between contrast agents that emit at a similar wavelength. ^123,124 Finally, ergonomics, both physical and the use of software and advanced algorithms to improve the efficiency of clinical workflow, should be a focus of attention during imaging system design.

Limitations and leverage

Despite improved tissue penetration when compared to visible light, an essential limitation of NIR fluorescence imaging remains the inability to visualize structures more than approximately 5 – 8 mm below the surface using current reflectance-based systems. Rather than introduce more complex optical methods, such as tomography, into surgery, the field has been migrating towards the combination of reflectance optical imaging with other modalities. Imaging modalities based on radioactive tracers, such as preoperative PET and SPECT, and intraoperative gamma probes and gamma cameras, have depth sensitivity to several cm but cannot provide real-time and precise visualization. Intraoperative ultrasound imaging also has superior depth penetration when compared to NIR fluorescence, but requires tissue contact and has problems visualizing smaller and superficial lesions. Combining NIR fluorescence imaging with these modalities leverages the key benefits, whilst overcoming the limited penetration depth of NIR light (Figure 4).¹²⁵ Another major limitation is the lack of clinically-available targeting agents. This is due to the high cost¹²⁶ and complex regulatory requirements surrounding drug development. Nevertheless, several academic and commercial groups have made progress on clinical translation and it is expected that several first-in-human clinical trials of new NIR fluorophores will commence within the next 2–3 years.

Figure 4. NIR fluorescence imaging alone or in combination with other imaging modalities.

Left panel: In this example, two superficially located targets, up to 5–8 mm deep, can be located using NIR fluorescence imaging. However, a deeper target at 25 mm would be invisible using by NIR fluorescence imaging alone.

Middle panel: Combining NIR fluorescence imaging with radioscintigraphy enables visualization of all three targets. However, spatial and temporal resolution of radioscintigraphy is poor. Once overlying tissue is removed, as guided by radioscintigraphy, NIR fluorescence can be used for more precise image guidance (see also Figure 2).

Right panel: Intraoperative ultrasound can visualize targets that are located deeper within tissue, but fails to find superficially located targets because of high acoustic reflectance. These superficial targets, though, can be visualized by NIR fluorescence thereby complementing intraoperative ultrasound. Of note, the ultrasound probe must be in direct contact with the tissue being imaged, thus precluding simultaneous imaging with NIR fluorescence.

Path to routine patient care

NIR fluorescence imaging has been proven feasible during surgery in many clinical studies. The main question, however, is when will it become part of routine patient care? In the past, the adoption of a new technology had to satisfy only two major criteria. It needed to “change patient management,” that is, make a meaningful impact on care, and it had to be “clinically realistic,” that is, it couldn’t disrupt normal workflow.⁴ Nowadays, though, as insurance companies and governments try to reign in healthcare spending, technology must also make patient care faster, better, and cheaper. Optical imaging is one of the few technologies with the potential to satisfy all three criteria. NIR fluorescence enables intraoperative imaging in real-time, without impeding the current clinical workflow. If surgeons are able to find tumors or avoid normal structures more easily then operating room time could be shortened. This in turn reduces anesthesia time and its associated risks, and means that surgeons can operate on more patients each day and hospitals can increase resource utilization. If improved tumor resection and/or normal tissue avoidance lowers recurrences and complication rates, then the healthcare system will enjoy huge cost savings. Quantification of these potential benefits, though, will require well-designed and well-executed clinical trials.

Conclusions

NIR fluorescence image-guidance during cancer surgery has the potential to improve patient management by visualizing tissue demarcation in real-time, thereby increasing the completeness of surgery and decreasing the morbidity associated with damage to normal structures. Intraoperative imaging requires a synchronous interplay between contrast agents, tumor biology, imaging systems, and algorithms. Results to date using available contrast agents and first generation imaging systems are extremely promising. Due to the relatively low depth penetration, NIR fluorescence imaging will likely be a complementary adjuvant to other imaging modalities, such as ultrasound and radioscintigraphy, in some clinical applications. Although studies focused on patient outcome and healthcare resource utilization should begin using existing imaging systems and non-targeted contrast agents, the era of targeted contrast agents for specific cancers remains on the precipice. Given the interest in the field, the next decade should clarify the role of NIR fluorescence imaging in cancer surgery and the extent to which it empowers surgeons to improve patient outcome.

Key points.

• NIR fluorescence imaging has been demonstrated to be feasible during cancer surgery using available imaging systems and contrast agents

• Clinical applicability has been described in sentinel lymph node mapping, tumor imaging, visualization of vital structures, and imaging of vascularization and perfusion

• NIR fluorescence image-guided surgery has properties that make it a good candidate for clinical acceptance: fulfills a clinical need, is relatively inexpensive, and fast

• Targeted contrast agents, necessary for full evaluation of this technique, are in advanced stages of clinical approval

• Upcoming years are essential, with novel contrast agents and optimized camera systems the technique should prove its true clinical value