Image-guided tumor surgery: The emerging role of nanotechnology

Abstract

Surgical resection is a mainstay treatment for solid tumors. Yet, methods to distinguish malignant from healthy tissue are primarily limited to tactile and visual cues as well as the surgeon’s experience. As a result, there is a possibility that a positive surgical margin (PSM) or the presence of residual tumor left behind after resection may occur. It is well-documented that PSMs can negatively impact treatment outcomes and survival, as well as pose an economic burden. Therefore, surgical tumor imaging techniques have emerged as a promising method to decrease PSM rates. Nanoparticles (NPs) have unique characteristics to serve as optical contrast agents during image-guided surgery (IGS). Recently, there has been tremendous growth in the volume and types of NPs used for IGS, including clinical trials. Herein, we describe the most recent contributions of nanotechnology for surgical tumor identification.

This article is categorized under:

Therapeutic Approaches and Drug Discovery > Nanomedicine for Oncologic Disease

Implantable Materials and Surgical Technologies > Nanoscale Tools and Techniques in Surgery

Diagnostic Tools > in vivo Nanodiagnostics and Imaging

1 |. INTRODUCTION

Although the adoption of molecularly targeted therapies has improved cancer survival rates, surgical resection remains the most effective treatment option for cancer and is the primary treatment for the majority of solid tumors (Miller et al., 2016; Wyld, Audisio, & Poston, 2015). It is well-established that the ultimate cause of death in cancer is not from the primary tumor, but from metastases (Chaffer & Weinberg, 2011; Mehlen & Puisieux, 2006). As such, the complete removal of the primary tumor, or a negative surgical margin (NSM), is associated with decreased rates of local recurrence and improved survival (Haque, Contreras, McNicoll, Eckberg, & Petitti, 2006; Houssami, Macaskill, Luke Marinovich, & Morrow, 2014; Meric et al., 2003; Pawlik et al., 2005). Although surgery is potentially curative, the detection of malignant versus healthy tissue remains limited to the surgeon’s experience and visual/tactile cues. As a consequence, positive surgical margins (PSMs) occur in roughly 14–36% of tumor resection surgeries, presenting negative prognostic outcomes surgeries (Orosco et al., 2018; Tringale, Pang, & Nguyen, 2018). Additionally, PSMs necessitate further treatment, subjecting patients to additional side effects and economic burden (Tringale et al., 2018). As a result, image-guided surgery (IGS) has emerged as a promising solution for margin detection (Vahrmeijer, Hutteman, van der Vorst, van de Velde, & Frangioni, 2013).

We previously reported on the potential of nanotechnology for IGS in 2016 (Hill & Mohs, 2016). Since then, the use of nanoparticles (NPs) as contrast agents for IGS has experienced tremendous growth. Advancements in nanotechnology have allowed for improvements in IGS efficacy, the entrance of new NP classes into the IGS realm, and the initiation of several first-in-human clinical studies. Due to the ability to deliver a variety of therapeutic and imaging payload to solid tumors, multifunctional NPs hold high potential to provide imaging feedback to surgeons during all stages of treatment. Also, developments in optics and engineering have facilitated the growth of imaging modalities such as photoacoustic (PA) imaging, surface-enhanced Raman spectroscopy (SERS), and near-infrared (NIR) II fluorescence (Ding, Zhan, Lu, & Sun, 2018; Maloney et al., 2018). Nevertheless, in order to attain clinical translation, several challenges specific to nanotechnology must be addressed. Herein, we cover the progress in the use of NPs for IGS and outline their potential to improve outcomes in surgical oncology.

2 |. IMAGE-GUIDED SURGERY

Imaging modalities such as X-ray, magnetic resonance imaging (MRI), positron emission tomography (PET), computed tomography (CT), ultrasound (US), and single-photon emission computed tomography (SPECT) offer valuable information for diagnosis, staging, and preoperative planning. However, the prohibitive cost and sophisticated infrastructure of these imaging modalities limit translation to the operating room. Additionally, the resolution and contrast provided by these imaging modalities do not provide sufficient sensitivity and specificity to identify surgical margins and local metastasis (Alam et al., 2018; Frangioni, 2008). Intraoperative frozen section analysis (IFSA) of resected tissues is beneficial in reducing rates of local recurrence and is the gold standard for margin assessment (T. P. Olson, Harter, Muñoz, Mahvi, & Breslin, 2007). Nevertheless, IFSA requires an average of 24–27 min to obtain results, and final margin status is not definitive until after surgery (Chiappa et al., 2013; Jorns et al., 2012; Kennedy et al., 2015). Thus, there is a clinical need for an imaging modality that provides real-time feedback on surgical margin status.

IGS is emerging as a promising alternative for tumor margin assessment. IGS is defined, here, as surgery guided with real-time optical imaging feedback. The goals of IGS are locating the primary tumor, detection of residual disease in the resection margin, identification metastatic lymph nodes, and preservation of surrounding healthy tissue (Tummers, Warram, et al., 2018). Before surgery, the patient is injected with an exogenous contrast agent which accumulates in the tumor through passive or active targeting mechanisms while rapidly clearing from healthy tissues, providing contrast to malignant tissues. With the aid of an imaging system that is specific to the type of exogenous probe, the surgeon uses real-time image guidance to resect the primary tumor and identify any remaining cancer in the surgical cavity. As a result, the likelihood of attaining complete tumor resection, or NSM, is increased (Tipirneni et al., 2017).

The surgical tumor imaging methods with the most promise are techniques utilizing exogenous probes (Maloney et al., 2018). Although there are a variety of optical imaging modalities and exogenous IGS probes under development (M. T. Olson, Ly, & Mohs, 2019; Tipirneni et al., 2017; Tringale et al., 2018; Vahrmeijer et al., 2013), they generally fulfill the same criteria: Image of tumor is displayed in real-time; specific accumulation of contrast agent in the tumor and rapid clearance from healthy tissues, creating tumor contrast; the imaging system/contrast agent should identify tumors with high sensitivity and high resolution, including the detection of low signals and small tumors; image display and imaging system should not interfere with the surgical workflow (Mondal et al., 2014). Currently, several optical imaging modalities are compatible with tumor resection surgery. These include fluorescence, PA imaging, SERS, multimodal imaging, and theranostics, which integrates therapy with imaging.

2.1 |. Optical strategies for image-guided surgery

2.1.1 |. Fluorescence

Fluorescence-guided surgery (FGS) is the most-commonly utilized optical modality for tumor margin detection (Figure 1). Fluorescence offers several advantages over traditional biomedical imaging modalities, namely, the absence of ionizing radiation and high spatial resolution (Hill & Mohs, 2016; Vahrmeijer et al., 2013). Fluorescence imaging in the NIR window (700–1,300 nm) is superior to visible wavelengths due to low scattering, negligible tissue autofluorescence, and relatively high tissue penetration. Imaging with light <600 nm is limited by scattering in biological tissues and hemoglobin absorbance, while imaging with light >1,300 nm is hindered by high water absorbance (G. Hong, Antaris, & Dai, 2017). The properties of NIR light allow for specific imaging of tumors with a high signal-to-background ratio (SBR), imparting surgeons the ability to use visual and tactile cues in combination with real-time fluorescence imaging to identify tumor margins.

FIGURE 1.

Optical imaging modalities used for IGS with examples of surgical tumor detection from the literature: fluorescence (Reprinted with permission from Zhao et al., 2017. Copyright 2016 Springer Nature), photoacoustic (Reprinted with permission from Thawani et al., 2017. Copyright 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim), Raman spectroscopy (Reprinted with permission from Mohs et al., 2010. Copyright 2010 American Chemical Society), theranostics (Reprinted with permission from Li et al., 2018. Copyright Ivy Spring International Publisher), and multimodal imaging (Reprinted with permission from Jin, Li, Yang, & Tian, 2019. Copyright Acta Materialia Inc. published by Elsevier Ltd.)

Although 5-aminolevulinic acid (5-ALA) improved complete resection and prolonged progression-free survival in a pivotal randomized phase 3 trial, leading to approval by the food and drug administration (FDA) in 2017 (Hadjipanayis & Stummer, 2019; Stummer et al., 2006), it suffered from inconsistencies in tumor contrast across the patient population, resulting in poor sensitivity and negative predictive value (NPV; Hadjipanayis, Widhalm, & Stummer, 2015; Moiyadi, Syed, & Srivastava, 2014). This is partially due to 5-ALA’s visible emission profile (ex. 405 nm; em. 635 nm). Aside from 5-ALA, there are two FDA-approved NIR fluorophores: Methylene blue (MB; ex. 665 nm; em. 686 nm) and indocyanine green (ICG; ex. 780 nm; em. 820 nm). ICG has garnered the most interest because its relatively lower tissue absorption and higher quantum yield (QY) are favorable for biomedical imaging (Matsui et al., 2010; Schaafsma et al., 2011). Although ICG is yet to be approved for FGS, it is under clinical investigation for intraoperative sentinel lymph node (SLN) mapping and tumor identification (Olson et al., 2019). Nevertheless, ICG still suffers from limitations such as poor aqueous stability, poor photostability, nonspecific binding to proteins, concentration-dependent aggregation, and reliance on the enhanced permeability and retention (EPR) effect for tumor accumulation. These characteristics result in low tumor signal and contrast (Alander et al., 2012; Schaafsma et al., 2011; H. Wang et al., 2018). Therefore, other NIR dyes such as cyanine-based and novel fluorophores are being investigated for FGS, and seek to improve tumor signal and contrast (Figure 2; Hernot, van Manen, Debie, Mieog, & Vahrmeijer, 2019; Olson et al., 2019).

FIGURE 2.

Emission profiles of commonly used dyes and NPs for surgical tumor detection and the emission wavelengths of NPs covered in this review. The majority of dyes fall into the NIR range. Fluorophores in the green text indicate FDA-approved compounds. Red text indicates the typical excitation wavelength for SERS NPs. The excitation and emission wavelength (nm) are reported in parenthesis after each dye

Finally, there are several FDA-approved and clinical-stage FGS systems that are on the market. Specific details of FGS systems are reviewed elsewhere (DSouza, Lin, Henderson, Samkoe, & Pogue, 2016). It is worth noting that these systems offer features that are complimentary with surgical workflow, maximizing the amenability of FGS with clinical practice. As novel fluorophores enter the clinical realm, the field must align on standardized system performance requirements for FGS systems (Pogue, 2018; Zhu, Rasmussen, & Sevick-Muraca, 2014).

2.1.2 |. Photoacoustic imaging

PA imaging combines the high contrast and specificity of spectroscopy with the spatial resolution capabilities of ultrasound to generate 2D and 3D images of organs and tissues (Beard, 2011). In PA imaging, laser pulses are directed toward biological tissues, where the energy is absorbed and converted to heat. The heat causes thermal expansion, which leads to the generation of ultrasonic waves that are ultimately detected by an ultrasound transducer (Xu & Wang, 2006). Endogenous molecules with high absorption coefficients, such as oxygenated and deoxygenated hemoglobin, can be visualized with high contrast using PA imaging. As a result, these properties can be leveraged to generate real-time images of blood flow and vascular networks in tissues of interest, including tumors (Figure 1; Ku, Wang, Xie, Stoica, & Wang, 2005; Lao, Xing, Yang, & Xiang, 2008; Laufer et al., 2012).

Additionally, exogenous compounds with a high molar extinction coefficient, sharp absorption peak in the NIR range, high photostability, and low QY, can be used as PA contrast agents (Weber, Beard, & Bohndiek, 2016). Interestingly, the properties of ICG make it a suitable PA contrast agent (C. Kim, Song, Gao, & Wang, 2010). However, due to the in vivo stability issues associated with ICG, others have sought to improve PA imaging of tumors with nanoformulations of ICG or the development of novel, tumor-targeted PA contrast agents (Weber et al., 2016).

To date, there are over two dozen PA imaging clinical trials that have been completed or are actively enrolling, with several investigating the use of PA for imaging of solid tumors and SLNs (NCT03931655, NCT03897270, NCT03630601, NCT03318107, and NCT03501823). Results from a study assessing excised thyroid tumors with PA imaging identified distinct differences in PA intensity (hemoglobin, lipid, and water) between malignant and healthy tissue (50 patients), thus demonstrating the potential application of PA imaging to surgical oncology (Dogra et al., 2014).

2.1.3 |. Surface-enhanced Raman spectroscopy

SERS and surface-enhanced resonance Raman scattering (SERRS) are spectroscopic techniques that are in development for tumor detection applications. SERS relies on the detection of Raman scattering, inelastically scattered light that is created upon the interaction of light generated by the Raman probe with the tissue of interest (Opilik, Schmid, & Zenobi, 2013). Each molecule within a tissue possesses a characteristic Raman scattering profile, otherwise known as a Raman fingerprint. Although label-free SERS can be used to identify tissue characteristics from endogenous molecules, it is complex and time-consuming. Alternatively, SERS NPs can be employed to provide a high-intensity signal during surgery (Cordero, 2018; Mohs et al., 2010; Zavaleta, Kircher, & Gambhir, 2011). As a result, SERS probes are utilized for the identification of biological structures, including tumors (Figure 1). Raman spectroscopy offers distinct advantages over fluorescence, namely resistance to photobleaching, the narrow peak of Raman signals (1–2 nm), multiplexing capabilities, and fM sensitivity (Lane, Qian, & Nie, 2015). Therefore, the use of SERS NPs is a highly sensitive method to simultaneously detect multiple tumor-associated biomarkers during tumor resection surgery and examination of resected tumors.

2.1.4 |. Multimodal imaging and theranostics

The previously mentioned imaging modalities offer unique benefits for tumor detection. However, each modality also suffers from a unique set of limitations as well. Due to these drawbacks, it is not possible to obtain complete information about a tumor using a single imaging modality. Multimodal imaging minimizes these limitations by imaging tumors with multiple, complementary imaging modalities (Figure 1; Ehlerding, Grodzinski, Cai, & Liu, 2018; Louie, 2010). For example, NIR fluorescence (NIRF) offers the benefit of high-contrast, real-time imaging of tumors. Relative to PET, NIRF has lower probe sensitivity and a low depth of detection. Nevertheless, PET is not well-suited for intraoperative imaging and is more useful for preoperative and postoperative settings (Frangioni, 2008; F.-M. Lu & Yuan, 2015). Therefore, the use PET and NIRF in tandem can bridge the gap between the high sensitivity and detection depth of preoperative PET imaging with the real-time, intraoperative imaging capabilities of NIRF (Christensen et al., 2017; Li, Zhang, et al., 2018). Multimodal imaging is not only limited to NIRF/PET. It can include any combination of clinically established (X-ray, MRI, PET, CT, or SPECT) and/or intraoperative (fluorescence, PA, Raman spectroscopy) imaging modalities. Due to their ability to simultaneously deliver multiple payloads, NPs are well-suited to serve as multimodal imaging probes, many of which are described in forthcoming sections.

There are several early examples of multimodal imaging. The first successful clinical demonstration was with the combination of the radiodiagnostic agent, ^99mTc, with ICG for hybrid radio- and fluorescence-guided detection of SLNs in various solid malignancies. The lymphatic distribution pattern of ^99mTc and ICG were proven to be identical, and thus, combined radio and fluorescence detection of SLNs was clinically feasible (Brouwer, Buckle, et al., 2012; Brouwer, Klop, et al., 2012; Buckle, Brouwer, Valdes Olmos, van der Poel, & van Leeuwen, 2012; van der Poel, Buckle, Brouwer, Valdés Olmos, & van Leeuwen, 2011). Other studies have followed with radio- and fluorescently-labeled multimodal imaging probes. The carcinoembryonic antigen (CEA)-targeted humanized antibody conjugate, ¹¹In-diethylenetriaminepentaacetic acid (DTPA)-labetuzumab-IRDye800CW, was developed for combined SPECT/fluorescence imaging. ¹¹In-DTPA-labetuzumab-IRDye800CW was proven to be effective at identifying invisible submillimeter lung tumors with preoperative SPECT and intraoperative fluorescence eye (Hekman et al., 2017). Another antibody-dye conjugate, epidermal growth factor receptor (EGFR)-targeted cetuximab-IRDye800CW, was utilized for FGS and ex vivo PA imaging of pancreas tumors in a phase 2 clinical study (NCT02736578; Tummers, Millers, et al., 2018). Finally, the silica-based ¹²⁴I-cRGDY-PEG-Cornell dots (C dots) containing Cy5 were administered to patients with metastatic melanoma in a first-in-human clinical trial (NCT01266096). Although capable of dual PET-optical imaging, the pilot study only assessed the safety and feasibility of using ¹²⁴I-cRGDY-PEG-C dots as a PET contrast agent. Due to favorable safety profile and evidence of preferential tumor uptake, ¹²⁴I-cRGDY-PEG-C dots hold high potential as an optical-PET multimodal-imaging agent (Phillips et al., 2014).

NPs hold potential to further enhance drug biodistribution and pharmacokinetics (PK), maximizing the amount of drug delivered to the tumor, thus improving efficacy and limiting off-target toxicities (Tran, DeGiovanni, Piel, & Rai, 2017). Capitalizing on the multifunctional capabilities of NPs, nanotheranostics combine diagnostic agents with therapeutic agents. As a result, nanotheranostic agents allow the ability to noninvasively monitor treatment in real-time (Chen, Zhang, Zhu, Xie, & Chen, 2017). Nanotheranostics are applicable to surgical imaging and can serve as adjuvant therapy postsurgery.

Nanotheranostic agents that employ phototherapy are especially relevant to FGS. Phototherapy utilizes light and exogenous photosensitizers to elicit an antitumor effect, either through the generation of reactive oxygen species (ROS; photodynamic therapy, PDT) or heat (photothermal therapy, PTT). Due to the high tissue penetration of NIR light, NIR-responsive dyes that are used as contrast agents in FGS, such as ICG, are preferred as photosensitizers (Dolmans, Fukumura, & Jain, 2003; X. Huang, El-Sayed, Qian, & El-Sayed, 2006; Shanmugam, Selvakumar, & Yeh, 2014; Zou et al., 2016). Therefore, phototherapy is complementary to the surgical workflow, requiring minimal additional effort to induce phototherapeutic effects during surgery. As a result, phototherapy can be used in tandem with FGS, immediately eliminating any residual disease that may remain in the surgical cavity.

3 |. NANOTECHNOLOGY OVERVIEW

Recently, there has been tremendous growth in the volume and types of NPs used for IGS (Wang et al., 2019). NPs can be classified into two categories, hard and soft. Hard NPs are composed of inorganic materials and include quantum dots (QDs), noble metal, metal oxides, and lanthanide-based NPs. Optical features, such as strong absorption, photoluminescence, or magnetic properties, are inherent characteristics of the hard NPs, making them useful for bioimaging applications (Sangtani, Nag, Field, Breger, & Delehanty, 2017). Nevertheless, because hard NPs are composed of inorganic materials, the potential for toxicity and colloidal stability are significant concerns. Therefore, they are often surface-modified with biocompatible materials, such as polyethylene glycol (PEG), to improve in vivo safety and performance (Jiao et al., 2018; Sperling & Parak, 2010). On the contrary, soft NPs are composed of organic materials and include micelles, liposomes, and polymeric NPs. In addition to surface modification, soft NPs can be loaded with a diverse variety of cargo, such as fluorophores or therapeutic agents. Most soft NPs have can carry hydrophilic and amphiphilic cargo, such as ICG and other NIR dyes, providing in vivo stability to otherwise poorly soluble molecules. Interestingly, soft NPs can also be loaded with inorganic NPs as cargo, effectively producing hybrid NPs (Sangtani et al., 2017).

There are two primary mechanisms of NP uptake and retention in tumors: Passive and active targeting. Passive targeting of NPs is reliant upon the EPR effect, a phenomenon that allows for the elevated accumulation of macromolecules in solid tumors (Yashuhiro & Maeda, 1986). The EPR effect is the product of rapid tumor angiogenesis, which produces underdeveloped tumor vasculature possessing a defective endothelium with wide fenestrations (Danhier, 2016; Maeda, Nakamura, & Fang, 2013). As a result, molecules in the size range of 10–1,000 nm, or molecular weights >40 kDa, extravasate across the vascular epithelium and passively accumulate in solid tumors (Maeda et al., 2013; Torchilin, 2011). Additionally, NPs can be granted an extra degree of tumor specificity through conjugation to targeting ligands. Tumor-specific markers are leveraged for NP targeting during tumor imaging. Both hard and soft NPs are capable of conjugation to targeting ligands such as antibodies/antibody fragments, small molecules, peptides, and aptamers (Wang & Thanou, 2010). Unlike the conjugation of targeting ligands to small molecules, the functionalization of NPs is especially advantageous because the fluorophore is left chemically unmodified, limiting the possibility of altering optical performance (Kamaly, Xiao, Valencia, Radovic-Moreno, & Farokhzad, 2012). In order to ensure that adequate contrast is provided during surgery, it is essential that the target of interest is highly expressed while expression is minimal in surrounding healthy tissues (R. R. Zhang et al., 2017).

3.1 |. Challenges and critical barriers

To date, there is a substantial body of work that utilizes nanotechnology for cancer therapeutics and imaging. However, clinical outcomes of nanomedicine trials do not necessarily reflect the robust results that are observed preclinically (Maeda & Khatami, 2018). There are relatively few NP formulations have surpassed clinical hurdles and reached FDA-approval: ~50 FDA-approved NPs, while superparamagnetic iron oxide NPs (SPIONs) are the only nanoparticles approved for an imaging application, MRI (Agarwal, Bajpai, & Sharma, 2018; Han, Xu, Taratula, & Farsad, 2019; Ventola, 2017). Although the use of NPs for IGS is primarily in the preclinical development stage, it is necessary to discuss the critical challenges and barriers to successful clinical translation.

3.1.1 |. The EPR effect

The high clinical attrition rates of anti-cancer nanoparticles can be partially-attributed to tumor heterogeneity and tumor microenvironment (TME) factors that impact the efficacy of NP delivery (Hare et al., 2017). In nano-imaging, the EPR effect is often cited as the primary mechanism underlying tumor uptake and retention of NPs (Hare et al., 2017). However, the heterogeneity of human tumors is a challenge to this claim. Moreover, the EPR effect has only been reported for some tumor types, and is variable across different cancers, even among different tumors within the same patient (Maeda, 2015; Tanaka et al., 2018). In addition, the EPR effect is difficult to predict. It can be influenced by a complex set of TME factors, including tumor characteristics, vasculature, stroma, macrophages, lymphatics, and interstitial fluid pressure (Hare et al., 2017). Therefore, reliance on the EPR effect alone may be a poor predictor of tumor uptake and can result in unsatisfactory clinical outcomes. Although ongoing work is attempting to identify predictive EPR effect biomarkers and enhance EPR effect mediated tumor-targeting, the most popular solution is to bypass the EPR effect altogether and target tumor antigens (Golombek et al., 2018). As discussed in forthcoming sections, there is a strong pipeline of targeted NPs for IGS, which decreases the dependence on the EPR effect.

3.1.2 |. Challenges to translating multifunctional nanoparticles

Although multifunctional NPs can extend imaging and theranostic capabilities, they present several practical constraints. First, each added functionality of a NP requires additional synthetic steps for conjugation or loading of ligand. As a result, the synthetic yield is lowered with each step and requires more time to produce the final product (Cheng, Al Zaki, Hui, Muzykantov, & Tsourkas, 2012). Also, complex synthesis protocols for multifunctional NPs carry a high risk of batch-to-batch variability. Variations in size and ligand density could impact target binding, in vivo protein binding, and imaging performance/therapeutic efficacy. In fact, complexity in NP manufacture poses as a unique challenge to regulatory agencies, making it difficult to develop consistent guidelines for nanomedicines (Klein et al., 2019). These potential issues must be addressed before scale-up to clinical-grade certified good manufacturing practices (cGMP; Landesman-Milo & Peer, 2016).

Although the conjugation of multiple targeting ligands and contrast agents to NPs can improve performance, it can significantly increase the overall cost of a NP. For example, the annual price for a therapeutic monoclonal antibody (mAb) can be upwards of $100,000 (Hernandez et al., 2018; Shaughnessy, 2012). Although this is the annual cost for a therapeutic regiment, the NP conjugation efficiency can be less than 10%, and require a substantial amount for effective targeting (Thorek, Elias, & Tsourkas, 2009). Also, imaging systems required for NP detection can be prohibitively expensive (MRI: >$300,000, CT: $100,000–300,000, PET: >$300,000, optical imaging: $100,000–$300,000; Leary & Key, 2014), and the NP may require specific optimization for effective performance with each imaging modality (Louie, 2010; Payne et al., 2017). Taken together, the costs for the preclinical development of an IGS contrast agent are $10 million (Pogue et al., 2015), and added functionalities can further increase this number.

The most significant hurdle to multifunctional NPs is clinical development. Because FGS is in its relative infancy, there are a limited number of contrast agents that have reached FDA-approval. However, often looked to as a template for the development of FGS probes are molecularly targeted nuclear medicine contrast agents. It is estimated that these agents cost an average of ~$150 million for development in a single indication (Agdeppa & Spilker, 2009). If multifunctional NPs eventually reach the clinic, a clinical assessment dedicated to each feature of the NP would be required (i.e., preoperative imaging, intraoperative imaging, therapy; Box 1). Superiority over the standard of care would be necessary in each indication, thus setting a high bar for FDA-approval and multiplying development costs (Yu, Park, & Jon, 2012).

BOX 1. COMMONLY USED TERMS.

Intraoperative Imaging:

The imaging of anatomical structures and tissues during surgical procedures.

Image-Guided Surgery (IGS):

The use of real-time intraoperative imaging to identify and guide resection of primary tumor, local metastases, and metastatic lymph nodes. IGS includes, but is not limited to nuclear (MRI, PET, etc.) and optical (fluorescence, Raman, PA, etc.) imaging modalities.

Fluorescence-Guided Surgery (FGS):

The most commonly utilized method of IGS. FGS relies on fluorescence to identify and resect malignant tissue during surgery.

3.1.3 |. Safety concerns

Finally, concerns over safety and biocompatibility remain a critical challenge to the field. NP safety must be rigorously assessed alongside efficacy studies. Although there are safety challenges specific to each type of NP, safety, and biocompatibility issues generally can be attributed to a number of NP characteristics, including composition, shape, surface area, surface charge, catalytic activity, and/or the presence of bioreactive groups on the NP surface (Sukhanova et al., 2018). It has been well-documented that NPs have the potential to cause internal cellular damage and elicit unwanted immune responses, both of which can lead to tissue damage and other toxicities (Sukhanova et al., 2018). Another concern is the long-term biodistribution of NPs and the effect of nanomaterials on human health. This is difficult to assess in preclinical stages and may not ultimately be known until human studies are completed (Sahu & Hayes, 2017). Therefore, due to the lack of human NP exposure data, safety and biocompatibility remain a significant concern.

4 |. NANOPARTICLE STRATEGIES FOR IMAGE-GUIDED SURGERY

There have been great strides in the application of NPs for surgical tumor imaging. This review covers advancements in dye nanoformulations, SPIONs, stimuli-sensitive NPs, QDs, NIR-II NPs, persistent luminescence NPs (PLNPs), aggregation-induced emission (AIE) NPs, and SERS NPs as surgical contrast agents (Figure 3). Details of each study covered by this review are outlined in Table 1.

FIGURE 3.

Types of nanoparticles used for surgical image guidance

TABLE 1.

NPs for intraoperative tumor detection

Nanoparticle	Class	Composition	Size (nm)	Ex/Em (nm)	Target (moiety)	Modality	References
TQ-BPN	AIE	TQ-BPN + Pluronic F-127	33	630/808 (SWIR)		FGS	J. Qi et al. (2018)
L897 NPs	AIE	BPST + DSPE-PEG2000	34	711/888		FGS	Wu et al. (2019)
AIE NP	AIE	α-DTPEBBTD-C4 + DSPE-PEG2000	46	635/801		FGS	Liu et al. (2017)
Cor-AIE dots	AIE	Corannulene-PEG-TPP-TPA	47	440/680		FGS PDT	Gu et al. (2018)
HLZ-BTED dots	AIE	HLZ-BTED	60	805/1,034		FGS	Lin et al. (2019)
DTE-TPECM NPs	AIE	DTE-TPECM + lipid-PEG2000 + YSA peptide	68	410/550	EphA2(YSA peptide)	FGS PA PDT	J. Qi et al. (2018)
AGL AIE Nanodots	AIE	TPE-Ph-DCM + lipid-PEG2000	95	465/700 (10 hr)		FGS	Ni et al. (2019)
JNP15	AIE	Polyphorin + Bacteriopheophorbide NP	100	750/820		FGS PA	Lovell et al. (2011); Shakiba et al. (2016)
TPE-Th-B	AIE	TPE-Th-B + DSPE-PEG2000	110	363, 405/622		FGS	H. Gao et al. (2019)
Man-PRx800	DN	PEG-α-cyclodextrin + ZW-800–1	7	772/788	Mannose receptors (PEG/α-cyclodextrin)	FGS	Wada et al. (2018)
NIA	DN	PMLA + CTX + ICG + Tri-leucin	12	730/795	Various targets (CTX)	FGS	Patil et al. (2019)
LipImageTM 815	DN	Liposomal IR786	50	793/815		FGS	Cabon et al. (2016); Jacquart et al. (2013); Sayag et al. (2016)
NanoICG	DN	HA-PBA + ICG	80–150	780/800	CD44 (HA)	FGS	Hill et al. (2015, 2016); Qi et al. (2018); Souchek et al. (2018); Wojtynek et al. (2019)
CF800	DN	Liposomal ICG	90.7	735/820		FGS CT	Patel et al. (2016); Zheng et al. (2015)
SAMINs	DN	HA-PBA-Cy7.5 and HA-PBA-DTPAGd3+	97.8	98	CD44 (HA)	FGS MRI	Payne et al. (2017)
ICG/MSNs-RGD	DN	RGD + MSN + ICG	100	770/837	αvβ3 integrin (RGD)	FGS	Zeng et al. (2016)
NanoCy7.5	DN	HA-PBA-Cy7.5	100	52	CD44 (HA)	FGS	Hill et al. (2016); Kelkar, Hill, Marini, and Mohs (2016); Souchek et al. (2018)
LipoICG	DN	Liposomal ICG	130	800/N/A		MSOT	Shi et al. (2015)
LP-iDOPE	DN	Liposomal ICG	191	780/811		FGS PDT	Suganami et al. (2015)
RVG&IRDye800-Gd203 TNs	DN	Gd203 triangular Nanoplates-RVG-IRDye800	Edge length = 37.5–42.5	775/806	nAchR (RVG peptide)	FGS MRI	Jin et al. (2019)
ZCG	PLNP	ZnGa2O4Cr0.004	10	UV LED/696 (5 hr)		FGS CT	Ai et al. (2018)
NLPLNP	PLNP	Zn1.1Ga1.8Ge0.1 O4:Cr3+,Eu3+ @Si02 + FA	50	Red LED/696 (15 hr)	Folate receptor (FA)	FGS Theranostic	Shi et al. (2015)
Luminescent UCNPs	PLNP	NaGdF4:Yb,tm,Ca@NaLuF4 + anti-HER2 mAb	137	980/804 (24 hr)	HER2 (mAb)	FGS SPECT/CT	Qiu et al. (2018)
ZGGO:Cr	PLNP	Zn1 + xGa2 − 2xGex04:Cr + Aptamer	10–80	550/680 (5 hr)	(Aptamer)	FGS	J. Wang et al. (2017)
PDFT1032 polymer NPs	NIR-II	mPEG shell + DFT1032	68	809/1,032		FGS	Shou et al. (2018)
HI NPs	NIR-II	Liposomal HI	80	808/110	αvβ3 integrin (RGD)	FGS	Sun et al. (2017)
CH-4 T/SLB-MSN-Mdot/64Cu2+	NIR-II	SLB-MSN-Mdot + CH-4 T + 64Cu2+	112	738/1,055		FGS PET	Q. Zhang et al. (2019)
DCNP-L1-FSHβ	NIR-II	DCNPs (NaGdF4) + DNA + FSHβ	~300 in vivo	808/1,060	FSHβ receptor (FSHβ peptide)	FGS	Wang, Li, et al. (2018)
UCNP@Azo + DCNP@β-CD	NIR-II	Nd@NaGdF4 DCNPs + Azo or βCD	15 (UCNP@Azo), 8 (DCNP@β-CD)	808/1,060		FGS	M. Zhao et al. (2018)
SCH1-SCH4	NIR-II	Self-assembled PEG + CH1055 NPs	170, 80, 30, 2 (SCH1–4)	739–775/990–1,050		FGS	Ding et al. (2018)
RBCp	NIR-II	UCNPs@RB@RGD@avidin-RBC@ICG@biotin	40 (UCNPs) 7 μm (RBCs)	~808/~1,000		FGS PDT	Wang, Fan, et al. (2019)
SWCNT-coated virus	NIR-II	SWNT + M13 virus (p8) + p3 + AF584	500 × 1 (SWNT)	808/1,000–1,300	SPARC (p3)	FGS	Ceppi et al. (2019)
HT@CDDP NPs	NIR-II	TQTPA + CDDP + HA	60–70	760/1,016	CD44 (HA)	FGS Theranostic	Wang, Fan, et al. (2019)
Core-Shell Nanoparticles	NIR-II	PEO-b-PCL + NaYF4 + DiR	60–90	980/1,175	Folate receptor (Folate)	FGS	Tao et al. (2017)
RENPs@Lips	NIR-II	NaYF4/Nd 7%@NaYF4 + DPPC + Chol + PEG-lipid	72	264, 545/1,064, 1,345		FGS	Li et al. (2019)
Ag2Se-cetuximab QDs	QD	Ag2Se QD-OPA-PEG-cetuximab	2.8	720/930	EGFR (cetuximab)	Theranostic	Zhu et al. (2017)
pRF-GQDs	QD	Graphene QD	4	270/440		FGS	Fan, Zhou, Garcia, Fan, and Zhou (2017)
Bio CFQD® NPs	QD	In-based QD + ZnS + PEG	12.2	405/615		FGS	Yaghini et al. (2016); Yaghini, Turner, Pilling, Naasani, and MacRobert (2018)
QD-apt	QD	ZnS QD-PEG-A32	20	330/605	EGFRvIII (A32)	FGS	Tang et al. (2017)
Recombinant protein QDs	QD	EGFP-protein G-PbS QDs-anti HER2 mAb	30	515/1,150	HER2 (mAb)	FGS	Sasaki et al. (2015)
Gd-Ag2S Nanoprobe	QD	Gd3+-DOTA-Ag2S QDs	~25	808/1,200		FGS MRI	C. Li et al. (2015)
GERTs	SERS	Au + Si + BDT	97	785, 808/N/A		SERS Theranostic	Qiu et al. (2018)
SERRS-MSOT-Nanostars	SERS	Au Nanostar + Si + PEG + IR-780	100	785/N/A		SERS MSOT	Neuschmelting et al. (2018)
SERS NPs	SERS	Au + Si + PEG + IR-780	108	785/N/A		SERS	Andreou et al. (2016)
SERRS-NPs	SERS	Au + Si + PEG + IR-780	110	785/N/A		SERS	Harmsen et al. (2019)
Multiplexed SERS NPs	SERS	Au + Si + mAb	120	785/N/A	EGFR (mAb), HER2 (mAb), CD24 (mAb), CD44 (mAb), ER (mAb)	SERS	Kang, Wang, Reder, and Liu (2016); Wang, Ma, et al. (2017)
Multiplexed SERS NPs	SERS	Au + Si + PEG12 + DyLight 650 + mAb	120	785/N/A	CA9 (mAb), CD47 (mAb), IgG4 (mAb)	SERS	Davis et al. (2018)
PEG-R-Si-au-NPs	SERS	Au + Si + PEG	120	785/N/A		SERS	Karabeber et al. (2014); Thakor et al. (2011)
αTF-SERRS NPs	SERS	Au + Si + IR-780 + ALT-836	122	785/N/A	TF (mAb)	SERS	Nayak et al. (2017)
PET-SERRS NPs	SERS	Au + Si + IR-780 + 68Ga	132	785/N/A		SERS PET	Wall et al. (2017)
RGD SERRS	SERS	Au Nanostars + Si + IR-780 + RGDyK	140	785/N/A	αvβ3 (RGD)	SERS	R. Huang et al. (2016)
CD47-targeted SERS-NPs	SERS	Au + Si + SM(PEG)12 + Dylight 650 + anti-CD47 mAb	150	785/N/A	CD47 (mAb)	SERS	Davis et al. (2018)
αFR-NPs	SERS	Au Nanostar + Si + (1) αFR-Ab + IR780 or (2) PEG5000 + IR140	164	785/N/A	Folate receptor (ab3361 anti-folate binding protein Ab [LK26])	SERS	Oseledchyk, Andreou, Wall, and Kircher (2017)
SERRS Nanostars	SERS	Au Nanostar + Si + PEG + IR-780	100–140	785/N/A		SERS	Harmsen et al. (2015); Spaliviero et al. (2016)
F-SERS dots	SERS	Ag + Si + AF610 + mAb	352–363	612, 785/628 (AF610)	EGFR (mAb), VEGF (mAb)	SERS FGS PTT	Y. Kim et al. (2017)
FRNPs	SERS	Au Nanorod + Ag + Si + (PS)A6 DNA	60–80	785/794		SERS FGS PTT	Pal et al. (2019)
ICG@MCNPs	SPION	SPIONs + carbon NPs + ICG	10	780/816		FGS MRI PTT	Song et al. (2017)
ATF-PEG-IONP	SPION	SPION + PEG + ATF-NIR-830 + Dox or Cis	25	800/830	uPAR (ATF peptide)	FGS MRI Theranostic	Gao et al. (2017)
NF-SION	SPION	SPION + oleic acid + PEG + Cy5.5-APTES	38	675/705		FGS	Lee et al. (2018)
IONPs-ICG-HA	SPION	SPION + ICG + HA	50	780/820	H202; CD44 (HA)	FGS PTT/PDT PA	Wang, Fan, et al. (2019)
ISCs	SPION	SPION + ICG	97	850/N/A		MRI PA	Thawani et al. (2017)
800ZW-sPION@dsiO2-YY146	SPION	SPION + dSi02 + 800ZW + PEG + YY146	146	778/806	CD146 (YY146 mAb)	FGS MRI	Wang et al. (2015)
NIR-830-ZHER2:342-IONP-Cisplatin	SPION	SPION + NIR-830 + HER2 Affibody	18–24	800/830	HER-2 (Affibody)	FGS MRI Theranostic	Satpathy et al. (2014, 2019)
P-GFLG-Cy5	SS	HPMA copolymer-GFLG-Cy5	7	649/666	Cysteine Cathespins	FGS	Blau et al. (2018)
PINS	SS	PEG-b-P(EPAlOO-r-ICGl)	26	780/820	pH < 6.9	FGS	Zhao et al. (2017)
PNP	SS	SiNc + PEG-PCL	40	750/780	Self-quenching/de-quenching	FGS PDT	Li, Zhang, et al. (2018)
LUM015	SS	QSY21-GGRK-Cy5-PEG	Not reported	650/675	Cysteine Cathespins	FGS	Whitley et al. (2016)

Abbreviations: AIE, Aggregation-induced emission; DN, dye nanoformulation; Em, emission; Ex, excitation; NIR-II, near-infrared-II; N/A, not applicable; PLNP, persistent luminescence nanoparticles; QD, quantum dot; SERS, surface-enhanced Raman spectroscopy; SPION, superparamagnetic iron oxide nanoparticle; SS, stimuli-sensitive.

4.1 |. Nanoparticle formulations of established fluorescent dyes

As mentioned above, several NIR dyes are well-suited for biomedical imaging (Figure 2). However, these dyes suffer from in vivo stability issues, undergo rapid clearance, lack tumor specificity, resulting in low tumor accumulation (G. Hong et al., 2017). As a result, several have developed nanoformulations of NIR dyes (liposomes, micelles, polymeric NPs, mesoporous silica NPS [MSNs], etc.) in order to improve the uptake and retention in tumors.

As the only FDA-approved NIR dye, ICG is the most commonly used fluorophore in NIR dye NPs. For example, we have sought to improve the tumor retention of ICG through nanoformulation via self-assembled hyaluronic acid (HA) NPs. As a result of conjugating HA polymers to the hydrophobic moiety, 1-pyrenebutanamide (PBA), hydrophobic pockets are formed during the NP self-assembly process. Hydrophobic and amphiphilic molecules, such as ICG, can be loaded into HA-PBA NPs and retained in the hydrophobic pockets through π–π stacking (Svechkarev, Kyrychenko, Payne, & Mohs, 2018). Compared to free ICG, NanoICG provided up to 2.9-fold higher intraoperative contrast and improved surgical outcomes in breast, pancreatic, and prostate tumors (Hill et al., 2015, 2016; Qi, Chen, et al., 2018; Souchek et al., 2018; Wojtynek et al., 2019). Others have improved the tumor uptake of ICG with active targeting ligands. ICG/MSNs-RGD are MSNs conjugated to an RGD peptide, targeting αvβ3 integrin in brain tumors. ICG/MSNs-RGD improved tumor signal 1.7-fold versus nontargeted ICG/MSNs. Although both were effective at identifying the primary tumor, ICG/MSNs-RGD demonstrated a propensity for identifying submillimeter tumors that would have gone unnoticed with other methods of tumor detection (Zeng, Shang, et al., 2016). Additionally, a NIR polymalic acid chlorotoxin nanoconjugate (NIA) provided high contrast to orthotopic brain tumors and improved the completeness of resection (Patil et al., 2019).

Although ICG is primarily used as a NIRF contrast agent, it is also effective at providing contrast in other modalities. A liposomal formulation of ICG, LipoICG, was developed as a multispectral optoacoustic tomography (MSOT) contrast agent. LipoICG provided sharp MSOT contrast to 4T1 tumors 24 hr after administration at a resolution of ~35 μm. Additionally, LipoICG provided a stronger MSOT signal than gold nanorods, a standard MSOT contrast agent, while still possessing high biocompatibility (Beziere et al., 2015). Others have developed multifunctional ICG-based NPs. CF800 is a liposomal formulation of ICG and iohexol, a combination of FDA-approved NIRF and CT contrast agents. This dual-modality probe was effective at providing preoperative contrast enhancement (>200 Hounsfield units) to lung tumors. The CT signal was consistent with the NIRF during intraoperative imaging, where CF800 provided a TBR >5 (Patel et al., 2016; Zheng et al., 2015). Another liposomal formulation of ICG, LP-iDOPE, provided strong contrast to orthotopic brain tumors. In addition to surgery, the authors also noted that LP-iDOPE could be used as a PDT photosensitizer to eliminate any remain tumor cells in the surgical cavity (Suganami et al., 2015).

Although the use of ICG in FGS is widespread, it is still limited by a relatively-low QY, lack of conjugatable functional groups, and poor in vivo stability. Other NIR dyes in preclinical and clinical development have desirable features that can be exploited to develop more effective FGS contrast agents (Figure 2; G. Hong et al., 2017). Man-PRx800 is a mannose receptor-targeted NP conjugated to ZW800–1. Man-PRx800 displayed preferential uptake in LNs and provided high contrast during surgery. Although the primary use of Man-PRx800 was not for primary tumor margin detection, it can be beneficial in the identification of metastatic LNs (Wada et al., 2018). Our group developed NanoCy7.5 through direct conjugation of Cy7.5 to HA NPs. Compared to free Cy7.5, NanoCy7.5 provided 14.8-fold higher contrast in breast tumors. This is a product of not only the high QY of Cy7.5 but also direct conjugation of the dye to the HA NP, minimizing the amount of free Cy7.5 escaping from the nanoformulation (Hill et al., 2016; Kelkar et al., 2016; Souchek et al., 2018). Also, NanoCy7.5 was combined with HA-PBA-GdDTPA in a mixed-micelle formulation, termed self-assembled multimodal imaging NPs (SAMINs), for dual-modality MRI-NIRF imaging of breast tumors. SAMINs provided greater T₁ MRI contrast and % injected dose versus Magnevist (gadopentetate dimeglumine) with a 40× lower dose of gadolinium, while also retaining the ability to provide sufficient NIRF contrast during surgery (Payne et al., 2017). Finally, Jin et al. developed neuroblastoma-targeted Gd₂O₃ triangular nanoplates (TNs), conjugated to IRDye800CW, for dual-modality MRI-NIRF imaging. The neuroblastoma-targeted with Gd₂O₃ TNs provided a 3- to 5-fold higher MRI and NIRF signal to the tumor versus healthy surrounding tissue. As a result, resection guided Gd₂O₃ TNs improved 42-day survival versus controls and nontargeted Gd₂O₃ TNs (Jin, Li, Yang & Tian, 2019).

Due to their straightforward composition with FDA-approved or biocompatible materials, dye nanoformulations hold high potential as first-generation intraoperative contrast agents. As NIR dyes other than ICG enter the clinical realm, such as IRDye800 (Rosenthal et al., 2015), we expect to see an expansion in the use of dye nanoformulations for FGS.

4.2 |. Superparamagnetic Iron oxide nanoparticles

SPIONs are composed of either γ-Fe₂O (maghemite), Fe₃O₄ (magnetite), or α-Fe₂O₃, with diameters ranging from 10 to 100 nm (Wahajuddin & Arora, 2012). Due to their superparamagnetic properties, SPIONs are useful as preoperative MRI contrast agents, capable of providing enhanced T₂-weighted contrast to malignant tissues (Rosen, Chan, Shieh, & Gu, 2012). Furthermore, SPIONs possess an excellent biocompatibility profile (Corot, Robert, Idée, & Port, 2006). Past studies have indicated that SPIONs undergo biodegradation in vivo, limiting long-term cytotoxicity (Zimmer et al., 1995). Therefore, SPIONs are a favorable platform for biomedical imaging applications and are currently undergoing clinical assessment for MRI and preclinical development as contrast agents for FGS (Rosen et al., 2012; Zhu, Zhou, Mao, & Yang, 2017). In fact, the only FDA-approved SPION, ferumoxytol (treatment of iron-deficiency anemia in patients with chronic kidney disease), is also being used off-label as an MRI contrast agent in patients with renal failure who cannot be given gadolinium and is under clinical investigation for the detection of metastatic LNs and hepatic masses (Thakor et al., 2016). Thawani et al. developed a low-cost contrast agent termed ICG-coated SPIO-nanoparticle clusters (ISC), combining FDA-approved SPION and ICG. ISC is demonstrated enhanced contrast in preoperative T₂-weighted MRI and a strong intraoperative PA signal in a flank model of GBM. When compared to microscopic surgery, PA-guided surgery improved complete tumor removal and prolonged survival (Thawani et al., 2017).

SPIONs can be further modified by the addition of targeting ligands and therapeutic agents (Rosen et al., 2012; Wahajuddin & Arora, 2012; Zhu, Zhou, et al., 2017). For example, Wang and coworkers coated SPIONs with dSiO₂, 800ZW, and a CD-146-targeted antibody (800ZW–sPION@dSiO₂–YY146). Compared to an untargeted NP, 800ZW–sPI-ON@dSiO₂–YY146 generated an increase in NIRF signal and improved T₂-weighted MRI contrast in a mouse model of gastric cancer (Wang et al., 2015). Another group developed a SPION platform for multimodal imaging and therapy of pancreatic and ovarian tumors. In both cases, PEGylated SPIONs conjugated to NIR-830 were functionalized with targeting ligands (ATF peptide for uPAR targeting in pancreatic cancer, and a HER-2-targeting affibody in ovarian cancer), enhancing T₂-weighted MRI contrast and intraoperative NIRF contrast versus nontargeted SPIONs Additionally, the SPIONs were conjugated with chemotherapeutic agents and slowed tumor growth in comparison to free drug (N. Gao et al., 2017; Satpathy et al., 2014, 2019). Finally, Wang et al. demonstrated the use of HA/ICG-coated SPIONs (IONPs-ICG-HA) for combined FGS, PA, and PTT/PDT. IONPs-ICG-HA exploited TME features such as hypoxia, low pH, and excessive H₂O₂ production to trigger OH production by IONPs-ICG-HA. Also, ICG served as a photosensitizer to improve synergistic PDT/PTT and acted as a fluorescent imaging probe. This, combined with CD44 targeting provided by HA, resulted in the efficacious treatment of colorectal tumors and improved PA and NIRF contrast (Wang, Fan, et al., 2019). Although FGS was not the primary objective of the IONPs-ICG-HA study, it demonstrated the versatility of SPIONs for multimodal imaging/theranostics of solid tumors.

Due to the safety profile and efficacy of SPIONs, they remain an excellent platform for surgical imaging. Recent findings have made it clear that modifications to SPIONs (e.g., targeting, theranostics) have the potential to improve outcomes. However, further work is needed to assess the effects of active targeting on the pharmacokinetics and biodistribution of SPIONs (Rosen et al., 2012; Zhu, Zhou, et al., 2017).

4.3 |. Stimuli-sensitive nanoparticles

During tumor development, cellular processes become increasingly dysregulated to sustain proliferation and survival. As a result, tumors acquire hallmark characteristics that distinguish neoplastic disease from healthy tissue (Hanahan & Weinberg, 2011). These hallmarks manifest as abnormalities that are observed in the TME, such as acidic pH, hypoxia, or the upregulation of proteolytic enzymes. Consequently, these TME factors can be targeted by delivery systems and exploited by stimuli-responsive imaging probes for intraoperative imaging (Du, Lane, & Nie, 2015). With tumor-specific fluorescence activation, stimuli-sensitive NPs can maximize tumor signal and while minimizing fluorescence activation in background tissues.

Low pH is a characteristic of the TME. The rapid growth and proliferation of cancer cells require a continuous source of energy. To maintain high energy requirements, cancer cells shift their metabolism from oxidative phosphorylation to aerobic glycolysis (Warburg effect), acidifying the TME (pH_Tumor = 6.5–6.9 vs. pH_{Normal Tissues} = 7.4; Vander Heiden, Cantley, & Thompson, 2009). Recently, Zhao et al. developed a transistor-like ICG-based pH nanoprobe, termed PINS. At pH ≤6.9, PINS dissociates into protonated unimers, which are highly fluorescent, while at pH ≥7.0, PINS are retained in a nonfluorescent, micellar state. Effectiveness was demonstrated across a variety of orthotopic and metastatic tumors, in which PINS provided superior contrast versus a panel of NIR imaging probes. Also, PINS reduced the number of false positives versus fluorodeoxyglucose (FDG)-PET imaging. The surgical utility of PINS was demonstrated, dramatically improving survival after FGS in models of head and neck cancer and small occult breast nodules with a 100% confirmation rate (T. Zhao et al., 2017). Based on this robust preclinical data, PINS (ONM-100) is now under phase II clinical investigation in patients with solid tumors (Figure 4). Another group has developed a SiNc-based pH-activatable NP for combinatorial FGS/phototherapy. Upon internalization into cancer cells, the OFF/ON SiNc NPs were activated by pH-triggered dequenching of fluorescence, leading to sufficient tumor contrast and cytotoxicity (X. Li et al., 2018).

FIGURE 4.

Nanoparticles currently undergoing clinical investigation

Another characteristic of tumor cells is high expression of proteases. Proteases, such as cysteine cathepsins or matrix metalloproteinases (MMPs), have essential roles in facilitating tumor growth, invasion, and metastasis, chiefly through degradation of the extracellular matrix (Koblinski, Ahram, & Sloane, 2000). Therefore, proteases have been long sought-after as targets of cancer imaging probes (Mahmood & Weissleder, 2003). Cysteine cathepsin-activatable NPs have recently been reported (Blau et al., 2018; Whitley et al., 2016). LUM015 is a PEGylated dye-quencher conjugate linked with a cathepsin-degradable peptide sequence (QSY21-GGRK-Cy5-PEG). Upon degradation of the GGRK peptide by tumoral cathepsins, fluorescence is activated through a dequenching mechanism. After demonstrating outstanding efficacy in a genetically-engineered mouse model (GEMM) of soft tissue sarcoma (STS), LUM015 was administered to 15 patients that underwent surgical resection of STS or breast tumors and imaged ex vivo. Results demonstrated that that LUM015 was well-tolerated at three doses and provided sufficient contrast to tumors. In the majority of patients, LUM015 provided a tumor to normal tissue fluorescence ratio > 1, an indicator of strong potential as a contrast agent for FGS. Since this preliminary study, LUM015 has been undergoing various stages of clinical investigation in multiple indications, including a recently established multicenter phase 3 clinical trial (NCT03686215; Figure 4).

In addition to PINS and LUM015, other peptide-based activatable probes have initiated clinical trials (R. R. Zhang et al., 2017). Therefore, there is a high potential for the use of stimuli-sensitive probes for FGS. Nevertheless, heterogeneous expression levels of enzymes and conditions targeted by stimuli-sensitive NPs can potentially limit their efficacy (Bu, Shen, & Cheng, 2014; Fisher, Pusztai, & Swanton, 2013). To address these concerns, it is essential to stratify patients by stimuli expression levels to identify the full impact of tumor heterogeneity.

4.4 |. Quantum dots

QDs are semiconducting nanocrystals consisting of 200–10,000 atoms and are on the order of 2–10 nm in core diameter, while targeted and coated QDs can be much larger. Their composition imparts unique optical properties, including broad absorbance, large Stokes shift, high QY, long fluorescence lifetime, and long photostability. The most unique characteristic of QDs is size and composition-dependent tunable emission, allowing for the precise selection of fluorescence emission wavelength (Smith, Duan, Mohs, & Nie, 2008). The structure of a QD typically consists of a fluorescent core surrounded by a semiconductor shell. QDs can be conjugated to biocompatible ligands to improve in vivo stability, and functionalized with tumor-targeting ligands (McHugh et al., 2018). Although QDs offer optical properties that are desirable for intraoperative imaging, they also suffer from significant toxicity concerns. Until recently, the composition of QD core material has mainly consisted of heavy metals (e.g., Cd, Hg, Pb, and As), which can be released from the core during circulation. Cd toxicity is the most considerable concern because it is the most commonly used core material (Smith et al., 2008). The primary clinical manifestation of Cd toxicity is renal damage but can also cause bone, cardiovascular, immune, and endocrine dysfunction (Bernhoft, 2013). A solution to limit the risk of heavy metal toxicity is to optimize the QD shell composition to minimize the possibility of Cd²⁺ release (Oh et al., 2016). However, to eliminate the risk of heavy metal toxicity altogether, the field has been inclined to develop Cd-free QDs with biocompatible materials (In, Cu, Ag, Zn, S, and Se) that have comparable performance to Cd-containing QDs (Won et al., 2019; Zeng et al., 2016).

Yaghini and coworkers developed an In-based QD platform, termed bio CFQD®, for in vivo LN imaging. Bio CFQD® was shown to display high QY and colloidal stability in aqueous environments. After subcutaneous injection into rat paws, bio CFQD® accumulated in regional LNs located in the thoracic fat pads and exhibited fluorescence retention after LN dissection (Yaghini et al., 2016). Although a long-term biodistribution study indicated that bio CFQD® was retained in the liver and spleen after 90 days, no organ or hematological toxicity was observed (Yaghini et al., 2018). In the future, bio CFQD® can serve as an effective platform for tumor and SLN imaging.

Several groups have optimized QDs through functionalization with targeting ligands. Although it was only used for a theranostic application, Ag₂Se-cetuximab nanoprobes were shown to accumulate in orthotopic oral tumors 13 hr after injection, and provide greater tumor signal than nontargeted QDs (Zhu, Chen, et al., 2017). An aptamer-based EGFRvIII-targeted QD, QD-Apt, was shown to cross the blood–brain barrier (BBB), providing sharp intraoperative contrast to orthotopic brain tumors. Interestingly, it was demonstrated that the EGFRvIII is highly expressed in 41.82% of human gliomas, and 0.00% in normal brain tissue, thus providing potential clinical relevance to QD-Apt (Tang et al., 2017).

There are several examples of QDs that have features of other NPs classes. Fan and coworkers developed pH-responsive fluorescent graphene quantum dots (pRF-GQDs), which displayed a fluorescence shift between green and blue at pH 6.8. Although fluorescence was emitted at a visible wavelength (em. = 440 nm), pRF-GQDs exploited tumor acidosis and provided contrast to pancreatic, liver, lung, and brain tumor xenograft models. Optimization of the optical properties (i.e., developing NIR-emitting QDs) could improve the intraoperative performance of these stimuli-sensitive QDs (Fan et al., 2017). NIR-II-emissive recombinant protein QDs were developed as a mAb-conjugatable QD platform based on recombinant enhanced green fluorescent protein (EGFP)-protein G. In a proof-of-concept study, the QD was conjugated to anti-HER-2 mAbs and provided distinct contrast to breast tumors. Although it is an effective tumor-targeting QD platform, the QD contained Pb, limiting its translational potential. Therefore, this targeting platform should be applied to biocompatible QDs to improve clinical relevance (Sasaki et al., 2015). Finally, Gd-Ag₂S nanoprobes combined the high penetration depth of MRI with the real-time imaging capabilities of NIR-II fluorescence. The Gd-Ag₂S nanoprobes were proven to be safe for up to 1 month, were detectable at concentrations as low as 0.025 × 10⁻³ M with T₁-weighted MRI, and 4 × 10⁻⁶ M with NIR-II imaging. MR contrast-enhancement was observed up to 48 hr after administration to mice bearing orthotopic brain tumors. Gd-Ag₂S was able to provide high intraoperative contrast (TBR ≈ 11) to tumors with minimal background interference. Histological analysis determined that Gd-Ag₂S effectively reduced the quantity of residual tumors when compared to BLS (4.5% vs. 14%; Li et al., 2015).

4.5 |. NIR-II emitting nanoparticles

There is a large body of evidence suggesting that fluorescence in the NIR-I range (700–900 nm) is favorable for biomedical imaging. Although impacted by tissue scattering to a lesser degree than visible light, NIR-I fluorescence is still not entirely immune to scattering. In biological tissues, the magnitude of scattering is inversely proportional to the wavelength of light. Therefore, as wavelength increases, scattering decreases (Welsher, Sherlock, & Dai, 2011). Biomedical imaging in the NIR-II window (900–1,300 nm) is superior to the NIR-I window due to a higher resistance to scattering, and thus, deeper tissue penetration. Also, light in the NIR-II window is less susceptible to the absorption and autofluorescence that is observed in biological tissues and endogenous molecules (G. Hong et al., 2017; Smith, Mancini, & Nie, 2009). Consequently, NIR-II imaging has garnered interest for FGS due to its potential in generating high TBRs with minimal interference.

Nanoformulation of NIR-II fluorophores can improve in vivo stability and maximize the amount of dye delivered to the tumor. These contrast agents are typically soft NPs loaded with NIR-II fluorophores. For example, PDFT1032 polymeric NPs consisted of PDFT, a novel NIR-II-emitting polymer surrounded by a DSPE-mPEG shell. PDFT1032 generated high TBRs in osteosarcoma for up to 3 days after IV injection (Shou et al., 2018). The self-assembled, PEGylated Scheme NP platform consisted of four size-tunable NPs based on the NIR-II dye, CH1055. SCH4, the smallest NP (2 nm), exhibited rapid hepatic clearance and allowed for the identification of hepatic tumors with an excellent TBR (>7; Ding, Li, et al., 2018). H1 NPs, were functionalized with targeting ligands, allowing for the precise identification of glioblastoma and SLNs (Sun et al., 2017). A multifunctional NP termed HT@CDDP was capable as a theranostic agent for oral squamous cell carcinoma. HT@CDDP provided sufficient tumor contrast up to 24 hr after injection while also suppressing tumor growth with combinatorial chemo-PTT (Wang et al., 2019).

Another popular class of NIR-II fluorophores are lanthanide-based NPs. Although lanthanide NPs provide bright NIR-II emission that is favorable for bioimaging, they are retained in the reticuloendothelial system (RES) and give rise to long-term safety concerns. Recently, Li et al. reported on the development of a liposomally formulated NIR-II lanthanide NP. Liposomal formulation dramatically reduced the blood, liver, and spleen half-life of RENPs@Lips compared to other lanthanide NPs. As a result, more than 90% of RENPs were cleared from the hepatobiliary system 72 hr after administration. Additionally, RENPs@Lips provided sufficient interoperative contrast to identify primary tumors and SLNs in mouse models of melanoma and osteosarcoma. Taken together, these results demonstrated the potential benefits of using nanoformulation to improve PK profile, possibly expanding the use of lanthanide-based NPs to safe, in vivo use (Li et al., 2019).

In addition to simple nanoformulations, there are more complex systems that are utilized for the delivery of NIR-II contrast agents to solid tumors (Cai et al., 2019). Two unique examples include an SWNT-coated M13 bacteriophage (Ceppi et al., 2019) and red blood cell-based probe (RBCp; Wang, Fan, et al., 2019). Another innovative NP delivery strategy is in vivo assembly of up/downconversion NPs (U/DCNPs). This strategy reduces RES scavenging and background interference while maximizing tumor-specific NIR-II fluorescence signal. The Zhang group developed two such NIR-II NPs. The first, DCNP-L₁-FSH_β, is a follicle-stimulating hormone receptor (FSHR)-targeted NP for the intraoperative detection of ovarian cancer. After dual injection of each component, the NP assembled at the tumor through DNA pairing. Intraoperative NIR-II imaging revealed that DCNP-L₁-FSH_β identified tumors with a tumor-to-normal tissue ratio of 12.5 and provided enough contrast to distinguish submillimeter lesions in the peritoneal cavity (Wang, Li, et al., 2018). Similarly, the UCNP@Azo/β-CD NP system was able to increase tumor accumulation fourfold with in vivo assembly. It resulted in a 1.5-fold increase in SNR compared to ex vivo assembled NPs in peritoneal ovarian cancer metastases (M. Zhao et al., 2018).

Finally, Zhang et al. synthesized a dual-modality NIR-II/PET imaging probe, termed CH-4T/SLB–MSN–Mdot/⁶⁴Cu²⁺ nanoprobe, consisting of the NIR-II fluorophore, CH-4T, and ⁶⁴Cu²⁺ as a PET contrast agent. The hybrid MSN/supported lipid bilayer (SLB)/Mdot formulation maximized CH-4T NIR-II fluorescence while also serving as a chelator-free scaffold for ⁶⁴Cu²⁺. As a result, NIR-II fluorescence was increased 4.27-fold versus CH-4T alone. These results were replicated in a model of squamous cell carcinoma (SCC), where the NPs provided high PET contrast before surgery as well as sufficient intraoperative NIR-II contrast for tumor recognition (Q. Zhang et al., 2019).

Although intraoperative imaging in the NIR-II range is an emerging field, we have observed tremendous growth due to the superior optical properties of the NIR-II window. Evidence of the first clinical study using NIR-II imaging emerged at the end of 2019. This study utilized ICG to image liver tumors in the NIR-I/II windows. Although off-peak ICG fluorescence was used, intraoperative NIR-II imaging demonstrated superior tumor detection sensitivity, tumor-to-normal liver tissue signal ratio, and tumor detection rate compared to imaging in the NIR-I window (Hu et al., 2019). Moving forward, it is likely that innovations in materials science and engineering will drive further breakthroughs. For example, InGaAs-based fluorescence detection cameras are most-commonly utilized in NIR-II imaging, however, as other NIR-II-sensitive materials, such as InSb or HgCdTe, are incorporated into cameras, the detection limit and depth will be improved, expanding the utility of NIR-II imaging (G. Hong et al., 2017). Nevertheless, because many of the NIR-II fluorophores are novel organic dyes or inorganic nanomaterials, comprehensive safety, and biocompatibility studies are necessary (Ding, Zhan, et al., 2018).

4.6 |. Luminescent nanoparticles

PLNPs are unique in that they do not require a continuous excitation source. Rather, PLNPs emit light several minutes to hours after the cessation of excitation. Because a continuous excitation source is not required during FGS, the risk of autofluorescence and background noise is minimized, allowing for high TBRs. Although there are over 200 chemical combinations reported in the literature, PLNPs for biomedical imaging applications are composed of host materials combined with rare earth (RE) and metal transition cations (dopant), emitting light in the red/NIR window (Liu et al., 2018). The underlying physics of PLNPs is complex and is still under investigation. However, persistent luminescence begins as excited electrons are generated by exposure to excitation light. Instead of falling back to relaxed energy levels, the electrons are captured by electron traps, which are characteristic of the PLNP composition. The energy stored in the trap is gradually released through a discharge process, and the end result is the continuous generation of light (Lécuyer et al., 2016).

Several groups have developed PLNPs for the purpose of FGS. Generally, PLNPs were based on either ZnGa (Ai et al., 2018; Shi et al., 2015; J. Wang, Ma, et al., 2017) or NaGd (Qiu, Zeng, et al., 2018), complexed with common nanomaterials, and functionalized with targeting ligands. In each example, PLNPs were exposed to excitation light before administration and emitted light in the NIR range, which could be observed up to 24 hr after initial excitation. Of note, Qui and coworkers developed NIR upconversion luminescent nanoprobes functionalized with anti-HER-2 mAbs. After IV administration, the PLNPs provided an excellent contrast to metastatic LNs that penetrated 7.7 mm. Interestingly, the PLNPs were tracked with SPECT/CT imaging, results from which demonstrated consistency with NIRF imaging (Qiu, Zeng, et al., 2018). Although each cited example showed enhanced tumor contrast provided by PLNPs, only one study reported the effects of PLNP administration on surgical efficacy. Ai et al. developed ZGC, which were ZnGa₂O₄Cr_0.004-based PLNPs for the NIRF imaging of liver tumors. During surgery, ZGC provided high contrast with long-lasting NIRF liver tumors, and detected lesions that were smaller than 1 mm. Also, FGS with ZGC prolonged survival versus traditional surgery (Ai et al., 2018).

Finally, it is essential to address the questionable toxicological effects of RE metals. Several studies have demonstrated that long-term exposure to RE dust may lead to toxic effects in the lung and long-term disposition in the body (Rim, Koo, & Park, 2013). More specifically, it was found that RE oxides undergo biotransformation in macrophage lysosomes, ultimately resulting in inflammation and fibrosis (R. Li et al., 2014). In order to advance the clinical translation of PLNP, further investigation into their long-term biodistribution and toxicity are necessary.

4.7 |. Aggregation-induced emission nanoparticles

A characteristic of planar fluorophores, such as ICG, is aggregation-caused quenching (AQC) at high concentrations/ close proximity. In AQC, π–π stacking interactions between aromatic rings quenches fluorescence, limiting the fluorescence imaging capabilities of these dyes at high concentrations. However, organic nonplanar fluorophores exhibit a quality known as AIE. Their fluorescence is usually quenched by extensive intramolecular rotations and other motions in low concentrations. In high concentrations/close proximity, these motions are restricted due to π-π stacking, and the molecules regain fluorescence (Y. Hong, Lam, & Tang, 2011). Additionally, NIR-emitting AIEgens possess higher QY, broader Stokes shifts, and stronger resistance to photobleaching than conventional NIR dyes, while also being more biocompatible than QDs (Mei, Leung, Kwok, Lam, & Tang, 2015).

Improvements in AIEgen nanoformulations have allowed for their use as surgical contrast agents. AIE NPs are composed of AIEgens loaded into lipid NPs, which helps encourage aggregation and improve delivery to tumors. AIE NPs developed by Liu and coworkers were utilized for NIRF detection of peritoneal metastases. The AIE NPs provided an ultra-high TBR of ~7.2 to malignant tissues. As a result, the AIE NPs were able to identify several lesions that were 100–500 μm in size (Liu et al., 2017). TPE-Th-B NPs also displayed the ability to identify submillimeter pulmonary metastases of 4 T1 breast tumors with high NIRF contrast (H. Gao et al., 2019). Two AIE NPs used for surgical imaging demonstrated multimodal imaging capabilities. JNJ15 (Polyphorin + Bacteriopheophorbide NPs) effectively provided NIRF contrast to a rabbit model of head and neck SLNs. After the excision of SLNs, JNJ15 also provided contrast with PA imaging (Lovell et al., 2011; Shakiba et al., 2016). A more elaborate combination of multimodal imaging and PDT was demonstrated with UV-sensitive, DTE-TPECM smart function-transformable NPs. The design of DTE-TPECM NPs allowed for the unique ability to undergo reversible switching between closed- and opened-ring structures upon exposure to UV light, allowing for the production of robust PA and fluorescence/ROS signals, respectively. As a result, this versatile system offered presurgical imaging (PA), intraoperative imaging (fluorescence), and PDT capabilities, ultimately improving overall survival after surgical resection of 4T1 breast tumors (J. Qi, Chen, et al., 2018). Finally, two separate AIE NPs possessed characteristics of other NP classes including NIR-II capabilities (J. Qi, Sun, et al., 2018) and persistent luminescence (Ni et al., 2018) in a variety of tumor models. Of note, the NIR-II emitting AIE NPs (TQ-BPN), showcased the benefits of AIEgens with a superior NIR-II quantum yield (2.8%) versus common NIR-II nanomaterials such as SWCNTs (QY = 0.4%). Surprisingly, although the maximum emission of TQ-BPN is 808 nm, the off-peak, short-wave infrared (SWIR) fluorescence produced by TQ-BPN was still effective at providing NIR-II contrast.

There is a strong potential for the use of AIE NPs in FGS. As this NP class grows, it is likely that the performance of nanoformulations will be improved with higher loading capacities and functionalization with targeting moieties. Also, because all AIEgens are considered new molecular entities, it is necessary that rigorous preclinical drug metabolism, pharmacokinetics (DMPK), and toxicology studies be conducted.

4.8 |. Surface-enhanced Raman spectroscopy nanoparticles

SERS NPs are unique in that they rely on the Raman effect, rather than fluorescence, for image-guidance. SERS NPs consist of a multi-layered structure composed of a Raman reporter molecule adsorbed onto the surface of the metal core (Au or Ag NPs), falling into the 100–120 nm size range. The Raman reporter component is responsible for generating a Raman spectrum unique to the NP. Also, SERS NPs can be further modified with a biocompatible coating and/or targeting ligands to improve in vivo stability and tumor specificity (Y. Wang, Kang, Doerksen, Glaser, & Liu, 2016).

There are several methods of SERS NP-guided surgery: Nontargeted SERS, targeted SERS, multiplexed SERS, and multimodal imaging. Regardless of the imaging method, an enhanced SERS signal serves as tumor indicator and guides resection. Although nontargeted SERS NPs attribute tumor accumulation to the EPR effect and macropinocytosis, they are still effective and safe contrast agents (Andreou et al., 2016; Harmsen et al., 2015, 2019; Karabeber et al., 2014; Spaliviero et al., 2016; Thakor, Luong, et al., 2011). Notable highlights include the detection of microscopic foci as small as 100 μm with 1.5 fM sensitivity (Harmsen et al., 2015), and the detection of premalignant GI lesions with high sensitivity (93.1%) and positive predictive value (89%; Harmsen et al., 2019).

Tumor-targeting ligands improve the detection of cancer by SERS NPs. αTF-SERRS NPs were conjugated to an antiALT-836 mAb and were able to detect lung metastases as small as 200 μm (Nayak et al., 2017). Also, αvβ3 integrin-targeted SERRS NPs detected orthotopic brain tumors with fM sensitivity and allowed for the recognition of tiny lesions as small as five cells, outperforming nontargeted NPs (R. Huang et al., 2016).

SERS has the unique ability to employ multiplexed imaging for the detection of multiple tumor markers in a single sample. This is beneficial in cancer imaging, due to heterogeneous expression of tumor antigens. In order to detect multiple NP flavors, a ratiometric quantification method is required, comparing the signals of targeted SERS NPs to a nontargeted reference (Garai et al., 2013). Two imaging studies using multiplexed SERS nanoparticles were performed on resected human tumor samples. The first example applied multiplexed SERS to human nonmuscle invasive bladder cancer (NMIBC), using three flavors of topically administered SERS NPs (CD47, CA-9, and nontargeted) with a Raman endoscope system. It was found that both the targeted and nontarget NPs differentiated tumor from normal tissue with high accuracy (ROC AUC = 0.93). The nontargeted NPs exhibited fivefold deeper penetration and 3.3-fold higher tissue concentration than the targeted NPs, suggesting a strong EPR effect in NMIBC (Davis, Kiss, et al., 2018). The Liu group employed a panel of SERS NPs (EGFR, HER2, CD44, and ER) to perform multiplexed imaging of 57 human breast tumor samples 10–15 min after resection. The multiplexed SERS NPs enabled discrimination between tumor and healthy tissue, and multiple subtypes of breast cancer, achieving a sensitivity of 89.3% and a specificity of 92.1% (Wang, Ma, et al., 2017). Taken together, these studies showcased the high translational potential of multiplexed SERS imaging.

Although SERS offers high sensitivity and resolution in the detection of tumor margins, it is reliant on point spectroscopy, which requires long acquisition times and can result in residual disease being overlooked. Also, relative to other imaging modalities, SERS has a limited depth of penetration (Eberhardt, Stiebing, Matthäus, Schmitt, & Popp, 2015). Therefore, SERS has been combined with other imaging modalities that offer a wider field of view (NIRF, MSOT) or higher depth of penetration (PET) to maximize the effectiveness of SERS. Both F-SERS Dots and FRNPs combined NIRF with SERS. Combined imaging allowed for the quick recognition of tumor with NIRF, while SERS detected residual tumor with high sensitivity, including microscopic lesions that were not recognizable with NIRF (Y. Kim et al., 2017; Pal et al., 2019). SERRS-MSOT-nanostars combined SERS with multispectral optoacoustic tomography (MSOT) for the detection of GBM. MSOT allowed for the generation of 3D mapping while SERS allowed for the detection of GBM with high sensitivity and resolution (Neuschmelting et al., 2018). Finally, PET-SERRS NPs were conjugated to ⁶⁸Ga to combine PET with SERS. PET was useful in the identification of tumor and metastatic LNs before surgery, while SERS detected malignant tissue during surgery (Wall et al., 2017).

Due to the high sensitivity of SERS along with multiplexing capabilities, SERS NPs hold high promise as effective contrast agents for surgical guidance. In order to break into the clinical realm, several hurdles must be surpassed. Because the primary components in SERS NPs are Au and Ag, there are concerns over potential long-term toxic effects of the NPs (Singh et al., 2018). Although previous studies have shown that Au NPs have negligible to mild toxic effects, further long-term biodistribution and toxicology studies are necessary prior to clinical investigation (Fraga et al., 2014; Thakor, Paulmurugan, et al., 2011). Additionally, advancements to SERS imaging systems such as wide-field SERS and endoscopic systems will improve the ease of use in the clinical setting (Lane et al., 2015).

4.9 |. Nanoparticles in the clinic

To date, NPs in clinical development include dye nanoformulations, stimuli-sensitive probes, and SERS NPs (Figure 4). More-sophisticated classes of NPs (QDs, luminescent, AIE, NIR-II, etc.) have not yet reached the clinic, but hold promise. Due to the commercial availability of FGS systems and compatibility with the surgical workflow, the majority of NPs undergoing clinical investigation for intraoperative tumor identification are for FGS indications (Tipirneni et al., 2017). However, it is worth noting that several of these NPs could be compatible with PA imaging, especially ICG-containing NPs. Nevertheless, PA imaging is more appropriate for applications other than surgery (i.e., applications that require high depth of detection) and is less amenable to the operating room setting (Steinberg et al., 2019). Although SERS NPs have been utilized for real-time tumor identification in resected patient samples, clinical hurdles lie within safety concerns associated with Au NPs and limited clinical applicability SERS systems (Kircher, 2017; Singh et al., 2018; Wang et al., 2016). For these reasons, there has been more clinical progress in the FGS space when compared to PA, and SERS.

The most clinically advanced NP is LUM015, which is undergoing eight clinical trials, including a pivotal phase 3 study (NCT03686215). As discussed in the stimuli-sensitive NP section, LUM015 is a pegylated cathepsin-activatable probe that is composed of Cy5 (ex. 658 nm; em. 696), a QSY21 quencher, and a GGRK peptide linker. As a consequence of protease overexpression in the TME, the probe is cleaved, separating dye and quencher, and activating fluorescence. The phase 3 trial is a multicenter, single-arm study with the primary objective of assessing the ability of LUM015 and the LUM2.6 FGS systems to detect residual tumor in 250 breast cancer patients undergoing lumpectomy. The secondary endpoint seeks to determine the rate of PSMs after FGS. This trial is expected to be completed in early 2020 and will be the first FGS NP to be tested on a large patient population. To date, the only results reported in the literature are from a dose-escalation study in 15 breast cancer patients, which reported tumor-to-normal tissue ratios of ~4.5 and an acceptable safety profile (B. L. Smith et al., 2018).

Because tumor acidosis is a relatively universal feature of tumors, ONM-100 (PINS), a pH-sensitive poly(ethyleneglycol)-b-poly(ethylpropylaminoethyl methacrylate) copolymer NP conjugated to ICG, is seeking FDA-approval as a tissue-agnostic FGS contrast agent. ONM-100 is currently undergoing two clinical trials, with the most advanced being a phase 2a study in 45 patients with breast, head and neck, colorectal, bladder, prostate, and ovarian cancers, and is estimated to be completed in early 2020 (NCT03735680). The primary endpoint of the study is to determine the mean fluorescence intensity of tumor versus nontumor in patients treated with ONM-100 while the secondary endpoint to establish the PK profile of ONM-100.

cRGDY-PEG-Cy5.5-CDots are integrin-targeted NPs that are undergoing clinical development as dual-modality PET/fluorescence contrast agents for SLN mapping (Bradbury et al., 2013). cRGDY-PEG-Cy5.5-CDots are composed of core-shell silica CDots, PEG, Cy5.5 (ex. 679 nm; em. 696 nm), and an integrin-targeting peptide, cRGDY. There are currently two active clinical trials for cRGDY-PEG-Cy5.5-CDots, with the most advanced being a phase 1/2 study in 105 patients with head and neck, breast, and colorectal tumors (NCT02106598). The investigators hope to utilize the cRGDY-PEG-Cy5.5-CDots to conduct preoperative SLN mapping with PET followed by fluorescence-guided removal of SLNs. The secondary objective of the study is to determine whether the cRGDY-PEG-Cy5.5-CDots can image breast tumors during surgery. The study is estimated to be completed by mid-2020.

The co-administration of carbon NPs (CNP) and ICG seeks to be established as a SLN mapping strategy. CNPs, 21 nm NPs consisting of activated carbon, accumulate in lymphatic vessels and LNs, and allow for their identification via visualization of black color (Y. Lu, Wei, Yao, Pan, & Yao, 2017). Addition of ICG could potentially improve SLN mapping through NIRF imaging. Currently, two large trials are assessing the feasibility of using CNP + ICG for the guided resection of SLNs in patients with cervical (NCT03778268; 1,143 patients) and endometrial (NCT03778255; 635 patients) tumors. The primary endpoint of the studies is to determine the predictive value of CNP + ICG versus CNP alone. The secondary endpoints seek to assess the SLN detection rate, sensitivity, and false-negative rates. The estimated study completion date is in mid-2020.

Three other NP candidates that have completed late-stage preclinical/translational studies. LipImage^™ 815 is a lipid NP composed of Lipidot^™ NPs (Gravier et al., 2011) and the NIR dye, IR780 (ex. 650 nm; em. 780 nm). In a prospective proof-of-concept study, LipImage^™ 815 was administered to companion canines with spontaneous STS or subcutaneous tumors and underwent surgery shortly after. No adverse effects were observed as a result of LipImage^™ 815 administration. Furthermore, LipImage^™ 815 provided sufficient contrast to identify tumors in each dog (n = 9) and resulted in PSMs in only one case. Although a small study, this demonstrates the feasibility of using LipImage^™ 815 for intraoperative tumor detection in spontaneous tumors, which have more heterogeneity than preclinical cell lines and GEMMs. Further studies on larger populations are needed to determine the sensitivity and specificity of the probe (Cabon et al., 2016). Finally, two sets of previously-discussed multiplexed SERS NPs demonstrated value as ex vivo contrast agents in identifying disease in resected NMIBC (Davis, Kiss, et al., 2018) and breast cancer (Wang, Ma, et al., 2017).

5 |. CONCLUSION AND FUTURE OUTLOOK

This review covered the current progress and advancements in the use of NPs for surgical tumor recognition. Since our prior report (Hill & Mohs, 2016), there has been a rapid development of nanotechnology for biomedical imaging with the entrance of several new technologies into the surgical tumor detection realm (i.e., NIR-II fluorescence, AIE NPs, etc.), as well the entry of first-generation NPs into the clinic. The use of NPs for surgical image guidance will likely continue to grow and play a significant role in the field.

However, several critical issues must be addressed to ensure successful clinical translation of NPs. The field must strictly adhere to rigid preclinical guidelines to maximize the translational potential of NPs and rule out under-performers. This includes testing NP candidates in clinically relevant models, such as orthotopic patient-derived xenograft tumors, which better recapitulate the heterogeneity and tissue-specific characteristics of human tumors. Additionally, strive to achieve clinical translation with efficacious, simple, and biocompatible (FDA-approved materials if possible) nanoformulations, while also minimizing NP complexity (Ioannidis, Kim, & Trounson, 2018; Park, 2019). Finally, because long-term biodistribution and safety of NPs are a significant concern, preclinical and clinical studies must assess these items with rigor (Han et al., 2019).

As more NPs enter the clinic, the IGS field must align on clinical trial design. This includes developing clinically relevant endpoints that seek to assess the effects of the contrast agent on patient benefit (i.e., survival or PSM rate) rather than merely correlating intraoperative signal with tumor location (Tummers, Miller, et al., 2018). Pogue and coworkers have established a cost-effective clinical trial framework that takes advantage of the FDA’s exploratory investigational new drug (eIND) pathway for the early clinical development of contrast agents with first-in-human phase 0 microdose trials (Pogue et al., 2015). Also, others have developed standard preclinical and clinical benchmark guidelines for the successful development of novel intraoperative contrast agents (Koller et al., 2018; Tummers, Warram, et al., 2018). In addition to contrast agents, several commercially-available imaging systems are used for surgical imaging. Each system varies in its dynamic range and sensitivity, which can lead to discrepancies in tumor recognition. Standard detection criteria must be developed to ensure consistent imaging effectiveness (DSouza et al., 2016).

Although there are several challenges to be addressed, the future outlook of the use of NPs for surgical tumor detection is bright, with the potential to create high value for cancer patients (Pogue, 2018). As advances are made at the intersection of engineering, materials science, and medicine, the in vivo performance of NPs for surgical tumor detection will continue to improve, and the benefits of NPs will be realized. For example, recent advancements in deep learning with improvements in neural networks have made substantial impacts in cancer diagnosis, proving to be a useful tool in digital pathology (Bera, Schalper, Rimm, Velcheti, & Madabhushi, 2019), radiology (Hosny, Parmar, Quackenbush, Schwartz, & Aerts, 2018), and the classification of skin lesions (Esteva et al., 2017). This technology can be applied to surgical image guidance as a tool to assist the surgeon in distinguishing between malignant and healthy tissue, increasing the likelihood of achieving NSMs. Additionally, computational methods leveraged for the development and optimization of NPs, improving their performance (Yamankurt et al., 2019). Given opportunities and the advancements highlighted by this review, we envision that NPs will become fundamental in the ultimate goal of reducing the incidence of PSMs in surgical oncology.