Introduction
Breast cancer (BC) is the leading cause of cancer in women, affecting approximately 2.3 million women annually and accounting for the highest number of cancer-related deaths [
1]. The primary treatment modalities for BC include surgery, hormonotherapy, and/or chemotherapy, depending on tumor characteristics [
2]. Successful BC surgery involves achieving complete resection of palpable tumors and the resection of small non-palpable (infra-clinical) tumors with microscopically negative margins, while preserving as much normal breast tissue as possible for optimal aesthetic breast reconstruction [
3]. Despite significant advancements in preoperative and intraoperative imaging techniques for improved tumor detection, the rate of positive microscopic margins after conservative BC surgery remains high, ranging from 14.9% to 26% in the literature [
3‐
5].
Precise localization of the breast tumor lesion and the ability to distinguish between malignant and benign tissue during surgery are critical for the successful surgical treatment of BC patients. Various pathological and imaging techniques have been reported for this purpose [
5].
Recently, near-infrared (NIR) fluorescence imaging (FI) has emerged as a promising nonionizing imaging technique for detection of cancerous tissue in different clinical conditions, including BC [
4,
6,
7]. NIR-FI utilizes light properties in the NIR spectrum (700–900 nm) to image tissue. NIR fluorescence offers advantages such as high tissue penetration (millimeters to centimeters in depth) and low autofluorescence emitted by natural fluorophores in the human body (e.g., porphyrins), enabling good discrimination between tissues containing fluorophores and those that do not, resulting in a high signal-to-noise ratio (SNR) [
8,
9].
Several fluorophores that emit in the NIR spectrum have been investigated for BC detection in preclinical and clinical studies, including specific and nonspecific fluorescent probes [
4,
8,
10‐
26]. Specific fluorophores include pegulicianine [
16], bevacizumab–IRDye800CW [
17], indocyanine green (ICG) in combination with different particles to enhance tumor avidity and specificity, such as human serum albumin [
18],
Pseudomonas aeruginosa azurin peptide p28 [
19], low molecular weight heparin (LMWH) ICG-loaded liposomes (LMWH-ICG-Lip) [
20], ICG-loaded H-Ferritin (HFn) nanoparticles [
21], and functionalized erbium-based rare-earth nanoparticles [
22]. However, specific fluorescent probes are currently not approved for clinical use and will not be discussed in this review.
Non-specific fluorophores include methylene blue (MB) [
24,
25], 5-aminolevulinic acid (5-ALA) [
26], and ICG. Among these, ICG is the most popular and widely used fluorophore. ICG is a water-soluble amphiphilic tricarbocyanine with a molecular weight of 775 Da and a hydrodynamic diameter of 1.2 nm, making it an excellent vascular and lymphatic contrast agent when injected intravenously (IV) or into the lymphatic system via subcutaneous injection. Initially developed for photography during World War II, ICG was later utilized for determining cardiac output, hepatic function, and ophthalmic perfusion. Its rapid Food and Drug Administration (FDA) registration was attributed to favorable characteristics, including confinement to the vascular compartment through binding to plasma proteins, fast and almost exclusive excretion into the bile, and low toxicity. In the 1970s, it was discovered that protein-bound ICG emitted fluorescence under illumination with NIR light (750–810 nm), peaking at around 840 nm [
27,
28]. ICG is approved by the FDA and the European Medicines Agency (EMA) for clinical applications as a vascular contrast agent. Due to its safety, affordability, and availability, ICG has become the foundation of NIR-FI for tumor detection [
29]. Intraoperative indocyanine green fluorescence imaging (ICG-FI) navigation has emerged as a promising technique for detecting cancerous tissue, including liver, colon, ovarian, head and neck, lung, and breast tumors, enabling surgeons to customize surgery based on real-time intraoperative imaging findings.
Since the first successful detection of BC tumors using ICG-FI reported by Ntziachristos and colleagues in the early 2000s, numerous preclinical and a few clinical pilot studies have demonstrated the detectability of BC tumors using ICG-FI after ICG IV injection [
4,
11,
13‐
15,
29]. The exact physiological mechanism underlying the preferential uptake of ICG in tumor tissues after intravenous injection is not fully understood. The most plausible hypothesis is the ‘enhanced permeability and retention’ (EPR) effect observed in tumoral tissue due to neoangiogenesis [
4,
6,
8,
10,
30]. Following intravenous injection, ICG acts as a macromolecule due to its high binding to plasma proteins. In healthy tissues, macromolecule-bound ICG serves as an excellent contrast agent, remaining in the intravascular compartment. In contrast, according to the EPR effect, these macromolecules are thought to extravasate from abnormal tumor vessels into the malignant tumor’s extracellular space. As the half-life of ICG in blood circulation is 3–5 min, ICG rapidly washes out from the intravascular space. Consequently, under NIR illumination, ICG that has accumulated in tumoral tissue emits a fluorescence signal that can be visualized through 5–10 mm of connective tissue thickness, resulting in the observed hyperfluorescence of tumoral tissue in contrast to surrounding normal tissue [
30‐
34].
The objectives of this study were to conduct a systematic literature review on ICG-FI for real-time detection of BC tumors in preclinical and clinical studies of perioperative imaging technologies and provide a summary of evidence-based data on the effectiveness of ICG-FI in BC.
Methods
This systematic literature review was conducted following the recommendations established by the Preferred Reporting in Systematic Review and Meta-Analysis (PRISMA).
Inclusion and Exclusion Criteria: The focus of our search was on studies that reported real-time perioperative (ex-vivo and in-vivo) ICG-FI in primary breast malignant tumors. This included the following aspects: (1) identification of primary BC tumors using ICG-FI, (2) evaluation of tumor margins after BC surgery using ICG-FI, (3) assessment of fluorescence intensity in BC tumors, and (4) accuracy of ICG-FI in detecting primary BC. In cases where two papers reported on the same population, only the first published study was included. We included either French or English language papers.
The following topics were excluded from the review: Conventional breast imaging for detection of BC, angiographic characterizations of ICG-FI, such as mastectomy flap and breast reconstructive flap vascularization and visualization, sentinel lymph node (SLN) detection, and infraclinical BC tumor marking. Additionally, editorials, reviews, commentaries, letters, and book chapters were excluded.
Sources and Literature Search: A comprehensive search was conducted in the PubMed and Scopus databases with the assistance of a professional medical librarian. The search encompassed articles published before December 2022. Furthermore, the reference lists of the retained articles were analyzed for additional relevant studies that met the inclusion criteria.
The following MeSH terms were used: “Optical Imaging”, “Indocyanine Green”, “Breast Neoplasms”, “breast neoplasms/surgery”, “Mastectomy” and “mammary neoplasms/animal”. Free search terms included: “breast cancer”, “breast neoplasia”, “breast-conserving surgery”, “mastectomy”, “breast surgery”, “fluorescence imaging”, “ICG”, “residual tumor”, “margins” and “animal”. These terms were used in various combinations.
Screening of Titles, Abstracts, and Full Texts: The titles, abstracts, and full texts of relevant studies were screened against the inclusion and exclusion criteria.
Data Extraction and Categorization: Full-text versions of studies that met the inclusion criteria were obtained for comprehensive assessment. The following data were extracted: year of publication, authors, study design, number of subjects, histological cancer type, technical details of ICG-FI (timing, volume/dose of ICG injection, type of FI system, FI intensity analysis program, depth of detection), and, when reported, the accuracy and/or detection rate.
These data were then analyzed and categorized into two groups based on the stage of the experiment: Preclinical experiences (including orthotopic tumor models and animal studies related to BC) or human clinical applications.
Discussion
FI has the potential to be a highly beneficial technique for real-time tumor identification and assessment of tumor boundaries during surgical procedures, particularly in BC. However, its clinical applications for tumor resection are currently limited, with only a few studies conducted in BC [
4,
10,
26,
33,
34,
42,
43].
In this systematic review of the literature, 23 studies were included that evaluated the efficacy of ICG-FI for discriminating between benign breast tissue and neoplastic BC. Among these studies, only 7 utilized ICG-FI in clinical settings, and all of them were in the proof-of-concept or feasibility phase [
4,
15,
16,
31‐
33]. The results from these studies show promise for the use of ICG-FI in BC surgery. The detection rate of orthotopic BC tumors and BC tumors in dogs using ICG-FI ranged from 78% to 100% and 80% to 100%, respectively [
13,
14,
40,
41]. In clinical experiences, ICG-FI was able to detect tumoral disease in approximately 8 out of 10 women (with a sensitivity between 80% and 100%) when ICG was injected shortly before surgery (within 2 h) [
4,
15,
16,
31‐
33]. The mean TBR reported for BC tumor identification in these studies varied from 1.1 to 8.5, and in human clinical studies, it ranged from 2.1 to 3.7. These TBR values are higher than the threshold detection value (1.3–1.5) by the human eye to define the tissue as hyperfluorescent regardless of cancer type, evaluation type (in vivo or ex vivo), and the FI camera system used [
31,
44].
One major challenge in BC surgery is the intraoperative assessment of breast surgical specimens during breast-conserving surgery to rapidly detect residual tumoral disease [
3‐
5]. Early evidence suggests that ICG-FI can be used for the intraoperative evaluation of surgical margin resection after breast conserving surgery, potentially improving surgical treatment outcomes for BC patients [
4,
10,
31‐
34]. With a reported high negative predictive value (100%), ICG-FI examination of the surgical bed may be able to exclude a positive resection margin with certainty, focusing intraoperative pathological evaluation only on cases where residual fluorescence is observed [
4,
31].
However, it’s important to interpret these results with caution due to the variability between studies, especially in terms of research stage (preclinical and clinical phase I settings), ICG dose, timing of ICG-FI, FI camera systems, and fluorescence intensity quantifications. More precision is needed regarding the pathophysiological mechanism of action, dosing, and timing of ICG-FI [
42‐
45].
Mechanism of action of ICG in breast tumors
The mechanism of preferential uptake of ICG in tumor tissues is not fully understood, but likely involves the EPR effect observed in tumor tissue due to abnormal neoangiogenesis. ICG molecules injected intravenously bind to serum lipoproteins and accumulate in the extravascular space of tumor tissue, emitting fluorescence under NIR illumination. The fluorescence signals can be visualized through connective tissue up to 5–10 mm thick. The rapid clearance of ICG from the intravascular space results in the observed hyperfluorescence of tumoral tissue compared to surrounding normal tissue [
7‐
9,
28,
29,
45,
46].
Another lesser-known mechanism of action of ICG in cancer cells, including BC, is its vascular contrast agent properties. During or shortly (3–10 min) after intravenous administration of an ICG bolus, the fluorescence of tumor cells is enhanced due to the binding of ICG to plasma lipoproteins. This mechanism improves the contrast between tumor and normal breast tissue, surpassing pure absorption contrast [
7,
28,
29]. Recent studies have shown higher vessel density and increased branch points of the vasculature in breast tumors, which may contribute to the angiographic effect of ICG in breast tumors [
47].
A recent study conducted by the group at Imperial College London provides further insight into the mechanism of action of ICG-FI for intraoperative detection of BC tumors [
32,
33]. Their findings suggest that the diagnostic accuracy of ICG-FI is improved when the imaging is performed during the angiography phase (within 5 min) compared to longer intervals (over 25 min) after ICG administration. The ex-vivo TBR in the angiography cohort was 3.18 (SD 1.74) compared to 2.10 (SD 0.92) in the later ICG-FI cohort, indicating better tumor detection in the angiography phase [
33]. However, it is important to note that larger and adequately powered clinical trials are necessary to confirm these findings.
In addition to the previously discussed pathophysiologic mechanisms of tumor hyperfluorescence, a new pathway for ICG accumulation within tumor cells has been described [
48‐
51]. This mechanism involves passive tumor cell targeting of ICG through increased uptake via clathrin-mediated endocytosis (CME), facilitated by the high endocytic activity of tumor cells and disruption of tight junctions. This phenomenon was initially observed in a mouse model of colorectal cancer [
48,
49] and subsequently confirmed in studies involving sarcoma and BC cell lines [
50,
51]. It appears that the affinity of ICG for phospholipid components of the cell membrane, which are altered and enriched in tumor cells, contributes to its ability to bind to and pass intracellularly via CME [
52]. Furthermore, tumor cells retain the dye for an extended period (at least 24 h) compared to normal tissues, indicating increased cellular uptake and retention as the primary mechanism of tumor fluorescence, rather than solely relying on the EPR effect. These contrasting findings highlight the complexity of the intra-tumor accumulation of ICG, suggesting that multiple mechanisms, including dysregulation of cancer cell pathways, tumor microvasculature, and the EPR effect, likely contribute to its enhanced uptake in tumors. The specific factors at play may vary depending on the tumor type. Further research is needed to better understand the mechanisms of action of ICG at the cellular level within human tumor tissue, which can provide valuable insights for the clinical use of ICG in fluorescence-guided surgery and potentially other diagnostic and treatment applications. This deeper understanding can help optimize aspects such as dosage and timing of ICG administration, which are still not fully elucidated.
ICG dose and timing
The optimal timing and dose of ICG administration for visualization and delineation of BC tumors through ICG-FI are crucial for accurate diagnosis during surgery. However, reports in the literature vary considerably in terms of protocols and inconsistent findings regarding the effectiveness of different timing and dosing strategies.
Studies have indicated that a low dose of ICG (less than 0.5 mg/kg) administered 24 h before ICG-FI is not effective for BC tumor detection [
15,
18,
19]. Even a dose of 1 mg/kg administered the day before surgery does not appear to be sufficient for satisfactory tumor visualization by FI [
12,
39,
40]. These findings suggest that preoperative injection, 24 h before surgery, is not the optimal timing for ICG administration.
It is worth noting that most clinical studies, except one, utilized intraoperative ICG injection timing (within ≤120 min) [
10]. In contrast, only 4 out of 16 preclinical studies used intraoperative timing. Additionally, studies have explored various injection times ranging from a few minutes to over 24 h, with inconsistent results. However, the interpretation of these findings is challenging due to the variations in ICG dose used across different studies.
Clinical applications of ICG-FI for intraoperative BC detection or discrimination between benign and malignant tissue have mostly been conducted with short delays between ICG injection and FI (ranging from <5 to 143 min) and lower ICG doses (0.25–0.5 mg/kg). These studies reported relatively higher efficacy (sensitivity) ranging from 72% to 100% for BC tumor detection [
4,
16,
31‐
33]. The intraoperative injection timing and lower ICG dose used in these studies make this ICG-FI strategy more easily integrated into current clinical workflows with minimal inconvenience for patients.
It is important to note that, although the literature on the application of ICG-FI in human BC is limited, the existing studies demonstrate significant heterogeneity and variation in reporting the efficacy of ICG-FI. Some studies focus solely on TBR, while others report sensitivity, specificity, negative predictive value, and false-positive rate. To ensure accurate evaluation and comparison of different timing, dosing, and FI systems/strategies, future studies should adhere to reporting complete test accuracy data.
In summary, there is a need for standardized protocols and comprehensive reporting of test accuracy data in future studies on ICG-FI for BC. This will facilitate better comparisons and understanding of the optimal timing and dose of ICG administration, as well as the effectiveness of different FI systems and strategies.
Imaging systems and FI quantification
We highlight an important limitation in the field of ICG-FI for BC detection, which is the wide variety of FI systems used in both preclinical and clinical settings. In clinical applications alone, seven different FI systems were utilized across seven different studies, with one study even testing three different FI systems in just 12 patients [
10]. This variability in FI systems makes it challenging to compare and interpret the results of these early experiences of ICG-FI in BC.
Furthermore, the handheld camera models used for intraoperative imaging may not be well-suited, especially for evaluating the breast surgical cavity after breast-conserving surgery. Although optical imaging systems may have similar characteristics, there is a lack of direct comparison between these systems. Standardization of functionality and results, along with a checklist of performance criteria, should be required and provided by manufacturers to enable meaningful comparisons between different FI cameras. Perhaps certain FI systems need to be adapted for specific uses such as angiographic assessment, sentinel node detection, or evaluation of different types of tumoral tissue. However, currently, there is no recommended device specifically tailored for ICG-FI in BC tumors, and a comparative evaluation of the existing FI systems is necessary to determine the optimal imaging approach [
53].
Another challenge in ICG-FI is the quantification of fluorescence intensity, which further complicates result interpretation and comparison. This issue is not limited to BC tumor detection but is also relevant for assessing tissue viability through vascular assessment [
54‐
56]. In the reviewed studies, fluorescence signal quantification and TBR calculations were performed using three different programs across six out of the seven clinical studies that included quantification [
4,
10,
15,
16,
31,
33]. Multiple programs with varying algorithms are being evaluated, but efforts should be made by manufacturing companies to develop quantitative imaging systems that are user-friendly and can facilitate the clinical implementation of intraoperative FI in various indications [
54].
Perspectives and limitations
Despite the heterogeneity in ICG dose, timing, and fluorescence systems used in preclinical and clinical evaluations of BC tumors with ICG-FI, the results of the few clinical studies available appear promising [
4,
16,
31,
33]. However, future prospective controlled studies are still needed to better define the optimal timing and ICG doses for ICG-FI and to strengthen the current evidence supporting its use in guiding BC surgery in clinical practice.
ICG-FI for BC tumor-guided surgery offers several advantages, including the relatively low cost of the fluorescent dye and its safety for patients and the medical team as a non-invasive agent [
57].
It is important to address some limitations of the present review. First, the level of evidence for the results obtained and presented in this review is low due to the current literature, which primarily consists of case series with a small number of patients and considerable heterogeneity. Second, the review could not provide clear-cut results regarding the optimal dose and timing of ICG injection for ICG-FI in BC tumor evaluation. One reason for this is the lack of comparison or control groups in clinical studies. The implementation of ICG-FI in BC tumor detection does not seem to follow a reliable translational approach, as most preclinical studies use a high dose of ICG and a preoperative timing of 24 h, while most clinical trials employ low doses and short intervals until FI. This difference may be attributed to the authors’ tendency to use shorter intervals between ICG injection and surgery to better align with clinical settings. Additionally, the discrepant results between preclinical and clinical studies may be explained by differences in the accumulation, distribution, and persistence of fluorescence signals after ICG IV injections between animal orthotopic tumors and true human tumors.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.