Novel Self‐Assembled Multifunctional Nanoprobes for Second‐Near‐Infrared‐Fluorescence‐Image‐Guided Breast Cancer Surgery and Enhanced Radiotherapy Efficacy

Abstract Breast‐conserving surgery (BCS) is the predominant treatment approach for initial breast cancer. However, due to a lack of effective methods evaluating BCS margins, local recurrence caused by positive margins remains an issue. Accordingly, radiation therapy (RT) is a common modality in patients with advanced breast cancer. However, while RT also protects normal tissue and enhances tumor bed doses to improve therapeutic effects, current radiosensitizers cannot meet these urgent clinical needs. To address this, a novel self‐assembled multifunctional nanoprobe (NP) gadolinium (Gd)–diethylenetriaminepentaacetic acid–human serum albumin (HSA)@indocyanine green–Bevacizumab (NPs‐Bev) is synthesized to improve the efficacy of fluorescence‐image‐guided BCS and RT. Fluorescence image guidance of the second near infrared NP improves complete resection in tumor‐bearing mice and accurately discriminates between benign and malignant mammary tissue in transgenic mice. Moreover, targeting tumors with NPs induces more reactive oxygen species under X‐ray radiation therapy, which not only increases RT sensitivity, but also reduces tumor progression in mice. Interestingly, self‐assembled NPs‐Bev using HSA, the magnetic resonance contrast agent and Bevacizumab‐targeting vascular growth factor A, which are clinically safe reagents, are safe in vitro and in vivo. Therefore, the novel self‐assembled NPs provide a solid precision therapy platform to treat breast cancer.


Introduction
Tumor-free surgical margins are critical parameters for breast-conserving surgery (BCS) as negative margins effectively reduce local tumor recurrence and distant metastasis, thereby optimizing clinical outcomes for patients. [1] However, intraoperative tumor margin identification and sensitivity is poor as it mainly relies on palpation and visual inspection. Specifically, ≈20-40% of patients require further surgery or eventually undergo a mastectomy. [2] Resectomies increase complication risks, potentially delay systemic treatments, and increase costs and health care burdens. Also, intraoperative margin evaluation approaches remain challenging for conventional imaging methods such as X-ray radiography, magnetic resonance imaging (MRI), and computed tomography (CT), as sensitivity and specificity are often limited, and methods are difficult to apply in operating rooms. [3] Thus, a real-time, specific modality evaluating margins is urgently required in clinical settings. [4] Recently, free survival rate was significantly improved (p = 0.0085). Also, probe relaxation rates were improved, and were used for the preoperative diagnosis of breast masses and RT sensitization. Our nanoplatform has several advantages: i) highly biocompatible with excellent imaging and therapeutic capabilities, ii) enhanced T1 contrast capabilities and NIR-II fluorescence properties for precise imaging and guided surgery, and iii) good in vivo stability and easy to construct.
For the first time, we developed a highly biocompatible and feasible therapeutic reagent, which may become a promising multifunctional nanoplatform for the precision treatment of breast cancer in the future.

NPs-Bev Synthesis and Characterization
We explored the formation of self-assembled nanocomposites using different input HSA, GdCl 3 , and ICG ratios. The input 1:5:2 (HSA:Gd 3+ :ICG) molar ratio was selected to fabricate NPs. As shown (Scheme 1), DTPA groups were covalently bound to HSA amine moieties, and the resultant DTPA-HSA conjugate was characterized using matrix assited laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) ( Figure S1a, Supporting Information). An increase in molecular weight was observed, from 66.9 kDa (native HSA) to 72 kDa (DTPA-HSA). Also, Gd 3+ was chelated to DTPA-HSA. Then, ICG, as a hydrophobic drug, was linked with Gd DTPA-HSA molecules to form nanosized Gd DTPA-HSA@ICG particles via self-assembly. Using a particle size analyzer and transmission electron microscopy (TEM), NPs were 7-10 nm and spherical in shape (Figure 1a,b), which meant that HSA had successfully induced Gd DTPA-HSA self-assembly into nanoparticles. Zeta potential analyses showed that the Gd DTPA-HSA@ICG-Immunoglobulin G (NPs-IgG) potential (control tracer), which was NP conjugated to normal IgG, and NPs-Bev went from positive to negative when compared with NPs ( Figure 1c). NPs-IgG and NPs-Bev dissolved in phosphate buffered saline (PBS) or 10% fetal bovine serum (FBS) exhibited stable NIR-II fluorescence signals (Figure 1d). Subsequently, sodium dodecyl sulfatepolyacrylamide gel electrophoresis showed that NIR fluorescence signals in NPs-Bev and NPs-IgG were consistent with gel protein positions, indicating that Bev or IgG had successfully conjugated to NPs ( Figure S1b,c, Supporting Information). Furthermore, the fluorescence intensity of both probes was extremely stable after continuous observations for 96 h in PBS or 10% FBS ( Figure S5af, Supporting Information). These characteristics provided important indications for the safe use of these NPs in organisms.
To investigate NP optical properties, the fluorescence signal intensities of ICG, NPs-Bev, and NPs-IgG molecules were measured in PBS (Figure 1e) or 10% FBS (Figure 1f). The absorption peak of ICG was 786 nm, NPs-Bev was 790 nm, and NPs-IgG 788 nm, indicating that absorption waves underwent infrared peak shifts. The emission waves of ICG, NPs-Bev, and NPs-IgG were measured in PBS (Figure 1g) or 10% FBS (Figure 1h). The emission peak of ICG was 795 nm, NPs-Bev 796 nm, and NPs-IgG 795 nm. Additionally, both probes also exhibited long trailing signals in the NIR-II region (Figure 1i). To investigate ICG stability in NPs-Bev, we dialyzed NPs-Bev with PBS or PBS plus 10% FBS, www.advancedsciencenews.com www.advancedscience.com Scheme 1. a) Schematic showing Gd DTPA-HSA@ICG-Bevacizumab (NPs-Bev) selfassembly formation using HSA, Gd-DTPA, ICG, and Bevacizumab. b) NPs-Bev used for NIR-II surgical navigation and radiotherapy sensitization.
with ICG aqueous solution as a control. ICG release was relatively low in PBS ( Figure S2, Supporting Information). However, in the ICG group in 10% FBS, ≈88% ICG was released after 48 h. By contrast, the percentage ICG released by the NPs-Bev group in 10% FBS was ≈22% after 48 h, and was probably due to ICG stably embedding into NPs-Bev, with little leakage into the physiological environment.
NPs-Bev and NPs-IgG were dissolved in PBS at pH 7.4 (normal tissue) and 6.5 (tumor tissue) under visible light, with no significant changes observed after 48 h ( Figure S3a, Supporting Information). To determine NPs-Bev and NPs-IgG stability at different pH conditions, NIR-I fluorescence absorbance ( Figure S3a To further analyze the NIR fluorescence properties of ICG, we simulated tissue in 1% intralipid solution to test ICG penetra-tion depth in NIR-I and NIR-II devices. ICG penetrated up to 7 mm at the NIR-II region in 1% intralipid solution, but <5 mm at the NIR-I region ( Figure S6a, Supporting Information). The fluorescence intensity signal of lower limb blood vessels in mice under NIR-II region fluorescence imaging was ≈2.3 times when compared with the NIR-I region ( Figure S6b-e, Supporting Information). Thus, NPs-Bev showed better contrast traits in the NIR-II region and reached deeper penetration depths.

NPs-Bev Cell Uptake
To clarify vascular growth factor A (VEGF-A) expression levels in breast cancer patients, we searched the Clinical Proteomic Tumor Analysis Consortium (CPTAC) sample database (http://ualcan. path.uab.edu/index.html) and observed that VEGF-A expression levels in breast cancer patients were higher when compared with normal breast tissue (p < 0.01) (Figure 2a). Levels were the highest in triple-negative breast cancer (TNBC) MDA-MB-231 cells when compared with other subtypes (Figure 2b). These data were consistent with VEGF-A protein levels in our breast cancer cell lines (Figure 2c).

Fluorescence Targeting Tumors and Various Microtumor Models In Vivo
To determine the optimal imaging concentration and time, we performed continuous imaging observations according to ICG quantification, and observed that 2 mg kg −1 and injection at 36 h were optimal imaging concentration and time parameters ( Figure S9a,b, Supporting Information). To further verify NPs-Bev tumor-targeting specificity in vivo, MDA-MB-231 tumor-bearing mice were injected with NPs-Bev and NPs-IgG through the tail vein. NIR-II fluorescence imaging showed that the NPs-Bev group exhibited strong fluorescence in the tumor. Furthermore, the maximum TBR in the NPs-Bev group (6.77 ± 0.45) was higher when compared with the NPs-IgG group at 36 h postinjection (2.48 ± 0.52) (p = 0.0004), indicating that the NPs-Bev probe targeted MDA-MB-231 tumor-bearing mice (Figure 3a,b). To determine NP biodistribution levels in vivo, the To test probe sensitivity, a multiple microtumor model was constructed to examine if fluorescent NPs could real-time, accurately identify small tumors. We observed that real-time tumor fluorescence and bioluminescence imaging signals were highly consistent (Figure 3b,c), and the corresponding tissues,

In Vitro and In Vivo NPs-Bev MRI
Gd has five unpaired 3d electrons and may be used as a T1shortening agent for MRI. [9] Thus, we compared the clinical contrast agents Gd-DTPA, NPs-Bev, and NPs-IgG MRI performances in vitro. Our results showed that relaxation degrees were 7.59 and 7.50, respectively, according to curve slopes between 1/T1 and the Gd concentration in a 0.5 T magnetic field (Figure 4a,b and Figure S8a,b (Supporting Information)).

NIR-II-Fluorescence-Image-Guided Surgery in MDA-MB-231-Luc Tumor-Bearing Mice
NIR-II-fluorescence-image-guided tumor surgery was performed at 36 h postinjection since the highest TBR was confirmed at this time. Furthermore, to simulate intraoperative tumor surgery, we evaluated fluorescence-image-guided surgery feasibility in a MDA-MB-231-luc tumor-bearing mouse model. For surgical resection, NPs-Bev molecules were injected into mice. As shown (Figure 5), the NIR-II navigation surgical group was intra- venously injected with NPs-Bev (ICG dose = 2.0 mg kg −1 ) and tumors were resected at 36 h postinjection. We then completely removed the tumor under NIR-II surgical navigator guidance, and observed that the smallest recognizable residual tumor was ≈0.5 mm (Figure 5a). The tumor and adjacent muscle tissue were excised, formalin-fixed and paraffin-embedded (FFPE) into blocks (Figure 5b), and sectioned into 10 μm slices for fluorescence imaging and 4 μm slices for hematoxylin and eosin (H&E) staining and VEGF-A immunohistochemistry (IHC). To compare fluorescence (FL) signals between tumor and muscle tissue in FFPE and 10 μm sections, FL was captured, and the tumor's FL mean ratio was nearly 40 times when compared with muscle levels (p = 0.020) in FFPE block, and the signal was about 8 times stronger than that of muscle in 10 μm sections (Figure 5c Figure S11 (Supporting Information)), indicating that NPs-Bev showed excellent performance as a NIR-II fluorescence agent.

NIR-II Fluorescence Surgical Imaging in Spontaneous Breast Cancer Mice
To further simulate human breast cancer development, NPs-Bev molecules were explored as NIR-II fluorescence agents in fluorescence-imaging-guided surgery in spontaneous breast cancer mice. In vivo and ex vivo fluorescence imaging of Mouse Mammary Tumor Virus-Polyomavirus middle T antigen (MMTV-PyVT) mice injected with NPs-Bev through the tail vein showed highly intense fluorescence signals in the mammary gland, whereas normal breast tissue signals in wild-type mice were extremely weak (Figure 6a-d). Subsequent fluorescence analysis showed that the breast tumor exhibited a higher fluorescence intensity when compared with the normal breast, and the AUC for fluorescence discrimination between malignant and benign tissue was 0.9710 (95% CI: 0.9294-1.0; p < 0.01), suggesting that NPs-Bev-based NIR-II fluorescence imaging accurately distinguished tumors from healthy tissue during surgery (Figure 6e). The NIR fluorescence imaging system was used to scan mammary tissue slices and analyze NP distributions in mammary tissue. Conspicuously, fluorescence signals in microscopic tumors had higher intensities when compared with normal breast tissue (Figure 6f). All sections were stained with H&E and analyzed by IHC. As shown (Figure 6g), microscopic NPs-Bev fluorescence images clearly distinguished invasive carcinoma from healthy glands, consistent with H&E staining and VEGF-A IHC (Figure 6h), and clearly demonstrated that NPs accurately identified tumor properties.

Radiation Sensitization in Cells
To evaluate NPs-Bev sensitization by RT, we used 2ʹ,7ʹdichlorodihydrofluorescein diacetate (H2DCFDA) to monitor intracellular reactive oxygen species (ROS) levels. [13] Using X-ray irradiation (6 Gy), intracellular fluorescence in MDA-MB-231 cells incubated with NPs-Bev was significantly stronger when compared with PBS-incubated cells (p = 0.0004). Also, fluorescence signals from nonirradiated PBS and NPs-Bev groups were negligible (Figure 7a,b). Thus, NPs-Bev significantly increased intracellular ROS production under X-ray irradiation. Moreover, colony formation assays were performed to evaluate long-term NPs-Bev radiation sensitization (Figure 7c). X-rayirradiated MDA-MB-231 tumor cells showed a small number of viable cell colonies when incubated with nanoparticles, significantly less than the radiation group ( Figure S12, Supporting Information). The effects of NPs-Bev combined with RT on apoptosis were also analyzed. Flow cytometry showed that apoptosis in the NPs-Bev plus RT group was significantly higher when compared with the PBS plus RT group (p = 0.0011) (Figure 7d,e).
We also determined cytotoxicity levels of different NPs-Bev concentrations toward MDA-MB-231 cells. Gd 3+ concentrations up to 50 × 10 −6 m exerted no obvious cytotoxicity effects in tumor cells, indicating excellent nanoparticle biocompatibility. However, under X-ray irradiation, nanoparticle cytotoxicity (Gd 3+ concentration = 12.5 × 10 −6 m) was significantly higher when compared with radiation alone, indicating that RT mediated by nanoparticles greatly improved radiation killing abilities toward tumor cells (p < 0.0001) (Figure 7f).

In Vivo Radiosensitization Studies
Owing to Gd compton scattering under X-ray irradiation, we evaluated tumor elimination efficiencies in mice injected intravenously with NPs-Bev. Representative fluorescence images in mice before treatment with NPs-Bev and PBS, with or without Xray RT, are shown ( Figure S13, Supporting Information), which demonstrated that the novel probe could specifically target breast tumors, and we subsequently performed tumor radiotherapy. As shown (Figure 8a,b), nonirradiated NPs-Bev-treated mice showed almost no inhibited tumor growth. However, under X-ray irradiation, NPs-Bev-treated mice showed effective RT sensitization and significant tumor regression when compared with mice treated with X-ray radiation alone. Additionally, weight assessments of resected tumors at day 21 confirmed significant tumor eradication and tumor growth inhibition in RT-sensitized animals when compared with the other groups (Figure 8c). There were no differences in body weight between the four groups PBS, NPs-Bev, RT, and RT + NPs-Bev ( Figure S14, Supporting Information). Furthermore, the antitumor effectiveness in different tumor tissues was analyzed by IHC. Tumor sections stained with the Ki-67 marker indicated that in mice with NPs-Bev combined with RT, animals showed fewer hyperproliferative tumor cells when compared with the other groups (Figure 8d,e). Additionally, caspase-3 activity was significantly increased in NPs-Bev combined with Xray radiation animals when compared with the other groups (Figure 8f). Therefore, NPs-Bev + RT treatments activated caspase-3, possibly by inhibiting proliferation and inducing apoptosis in cancer cells.

NPs-Bev Biosafety In Vivo and Ex Vivo
Cell counting kit-8 assays were used to examine NPs-IgG and NPs-Bev cytotoxicity in MDA-MB-231, T-47D, and MCF-10A cells. No significant cell viability inhibition was observed with in-creasing probe concentrations (1, 10, 50, and 100 μg mL −1 ) for 72 h ( Figure S15, Supporting Information), thereby indicating that probes did not affect cell viability and were suitable for imaging in vivo.
To verify in vivo safety, mice were randomly divided into two groups after the tail vein injection of NPs-Bev and PBS. H&Estained organs showed no obvious pathological changes in both groups ( Figure S16a, Supporting Information). Thus, NPs-Bev toxicity in mice was negligible and showed that probes could be safely used in future clinical trials. Venous blood was also collected for liver and kidney functions or routine blood analysis on days 1, 3, 7, and 28. The results showed that alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatine kinase MB (CK-MB), UREA, and creatinine (CREA) levels showed no significant changes ( Figure S16b,c, Supporting Information). Additionally, no significant differences were observed in white blood cell (WBC), lymphocyte (Lymph), monocyte cell (Mon#), neutrophil granulocyte (Gran#), red blood cell (RBC), hemoglobin (HGB), and platelet (PLT) levels. We also observed no differences

Discussion
Positive margin is highly associated with poor tumor localization accuracy and inaccurate tumor removal through viewing examination and touch-based feedback. It is now recognized that obtaining a clean margin during the first surgery remains an important physical behavior in BCS. [14] Imaging-guided surgery is becoming more important in clinical settings. It helps surgeons identify small-sized diseased growths or leftover wounds which are easily missed during surgery, and guide intraoperative surgical margin in addition assessment. [15] Therefore, this approach may improve prognoses in patients undergoing cancer surgery.
Molecular NIR fluorescence is a novel imaging modality with several distinct advantages, including no ionizing radiation and visualization when compared with positron emission tomography, CT, and other imaging methods. [16] In 2013, we successfully used ICG to surgically navigate a sentinel lymph node biopsy in breast cancer. [17] Keating et al. [18] used the Artemis NIR imaging device combined with ICG to guide BCS, and observed that half of patients had residual fluorescence signals in the tumor bed after tumor removal (6/12), but final pathological examinations confirmed negative margins. Therefore, this simple dye method generates high false positive rates in evaluating BCS margin status. Intriguingly, the development of tumor-specific targeting molecular probes has accelerated optical molecular imaging for the accurate determination of surgical margins in BCS. [19] For example, monoclonal antibodies, such as Bev and cetuximab have been developed to target VEGF-A and epidermal growth factor receptor, which are characteristic tumor cell surface markers.
Indeed, using FDA-approved antibodies to target NIR dye distribution is highly favorable as pharmacokinetic and pharmacodynamic properties are already known from drug approval studies, thereby making it easier to customize tracers for clinical use. [20] In previous research, we synthesized the NIR tracer Bev-800CW which targeted VEGF-A at high doses at both macro-and microlevels, and observed an 88% increase in the intraoperative detection of tumor margins. [19] However, due to absorption and scattering effects of water molecules, organic biomolecules, other substances on the band, and autofluorescence from biological tissues, the depth and resolution of NIR-I fluorescence imaging remains limited in clinical applications. [21] Of note, NIR-II fluorescence imaging significantly reduced spontaneous fluorescence and photon absorption and scattering in biological tissue, and generated high resolution and penetration. [22] Many NIR-II fluorescent materials have been developed, including small molecular organic dyes, [23] carbon nanotubes, [24] quantum dots, [25] and rare earth materials, [26] for use in biomedicine. Unfortunately, due to poor biocompatibility and unstable optical properties, the clinical application of these materials is restricted. Recently, FDA-approved ICG was used in NIR-II imaging as its QY was higher when compared with most synthetic NIR-II emission contrast agents. [27] In our study, ICG penetrated up to 7 mm in the NIR-II region in 1% intralipid solution, but less in the NIR-I region (5 mm). Moreover, fluorescence intensity signals in mice in the fluorescence NIR-II imaging region were over 2 times greater when compared with the NIR-I region. Therefore, ICG facilitated the clinical application of NIR-II fluorescence imaging in vivo.
Suo et al. [28] synthesized a Bev-ICG NIR-II probe which targeted VEGF-A in a rat colorectal cancer model, and showed that probe injection and NIR-II fluorescence endoscopy guidance identified tumors that were previously difficult to find under white light conditions. In our study, NPs-Bev showed superior performance as a NIR-II fluorescent contrast agent, with fewer postoperative tumor recurrences (0/8, 0%) when compared with the white light group. We also constructed a multiple small tumor model and showed the probe accurately identified small tumors. We then used the spontaneous breast cancer transgenic mouse model [14] to simulate breast cancer growth patterns, and showed that NPs-Bev accurately distinguished between cancerous and healthy glandular tissue, achieving an AUC = 0.9710. NP-based NIR-II fluorescence image guidance improved complete tumor resection and relapse-free survival rates and accurately distinguished between benign and malignant breast tissue.
As Gd 3+ functions as a MRI contrast agent, commercial MRI contrast agents such as Gd 3+ chelate and Gd-DTPA were first approved by the FDA for clinical use. [29] Unfortunately, due to poor targeting and a short cycle life, Gd-DTPA is difficult to target to tumors as weak signals are generated. [30] Encapsulating discrete Gd 3+ chelates into nanoassembled capsules is a simple and effective way to prepare MRI contrast agents, and generates high relaxation imaging agents capable of carrying large loads. [31] We previously reported that nanostructured Gd had good relaxation properties and generated high tumor MRI imaging signals. [32] Consistent with these results, we confirmed that NPs-Bev enhanced tumor contrast signals in T1-weighted MR images.
Importantly, effectively increasing energy deposition in tumor areas or improving tumor targeting may improve tumor radiation therapy efficacy and reduce side effects associated with this therapy. [33] In the last decade, several nanomaterials were developed and tested as radiosensitizers, of which high Z nanoparticles (HZNPs) attracted considerable research attention. Radiosensitization effects were observed in HZNPs constructed of gold, silver, bismuth, and Gd. Of note, motexafin Gd demonstrated good clinical benefits and rendered tumor cells more sensitive to RT (ClinicalTrials.gov Identifier: NCT00003411). Another highly anticipated reagent is activation and guidance of irradiation by X-ray (AGuIX) nanoparticles, novel Gd chelate nanoparticles which have shown enhanced RT efficacy in multiple clinical trials. [34] In our previous study, gadolinium oxide rare earth particles were nanosized to reduce Gd nanoparticle toxicity and degradation in the acidic tumor microenvironment to facilitate rapid in vivo metabolism for BCS and radiotherapy sensitization. [9] While research results are promising, nanoparticles with aforementioned inorganic structures have issues such as uncontrolled metal leakage and heavy metal toxicity. [35] In recent years, nanomaterial engineering research has opened up diagnosis and treatment avenues for the treatment of different diseases. [36] In particular, self-assembled organic nanomaterials have fine structures, convenient processing, low costs, good biocompatibility, enhanced permeability and www.advancedsciencenews.com www.advancedscience.com retention effects, multifunctional properties, and outstanding application potential in the biomedical field. [37] HSA is a natural biological macromolecule with good biocompatibility and low immunogenicity; [38] it has been used as a carrier for the antitumor drug paclitaxel and is widely used in clinical settings. [39] More importantly, Gd exhibits high hydrophobicity and may be used as an "adhesive" component between adjacent albumins to induce albumin assembly to form large nanoparticles. The process is similar to FDA-approved paclitaxel-albumin nanoparticles (Abraxane), [40] and may be a viable way to reduce the potential toxicity of Gd nanoparticles, improve biocompatibility, and avoid endoplasmic-reticuloendothelial system deposition.
We developed a novel self-assembled multifunctional NP (NPs-Bev) to integrate RT sensitization and facilitate real-time efficacy. In NPs-Bev-treated mice, effective RT sensitization and significant tumor regression was generated under X-ray irradiation when compared with mice treated with X-ray radiation alone. Additionally, preliminary toxicological studies demonstrated good NP tolerability, and no significant pathological changes were observed in main mice organs. Therefore, our NP showed good biosafety performance.
In summary, we constructed a multifunctional molecular imaging system based on the self-assembly pharmaceutical adjunct HSA, Gd-DTPA, and Bev to provide a research platform for the accurate navigation of BCS and RT sensitization. Recently, Rosenthal et al. [41] designed a roadmap to regulate the development, formulation, current good manufacturing practice, and translation of fluorescent tracers. We aim to promote the clinical progress of molecular probes in China, and develop more multifunctional probes based on this roadmap. We anticipate that our work will provide accurate diagnosis and treatment outcomes for breast cancer in the future. NPs-Bev and IgG-NP Synthesis: 2 g HSA (Beyotime Biotechnology, China) was dissolved in 30 mL 0.1 m NaHCO 3 (pH 8.2). Then, 2 g diethylenetriaminepentaacetic acid dianhydride was dissolved in 10 mL dimethyl sulfoxide (DMSO) and added to the HSA solution. The pH was adjusted to 8.2 using 1 m NaOH. The solution was stirred for 2 h at room temperature and dialyzed against deionized water. Then, 1 g GdCl 3 (J&K Scientific, China) was added at a pH of 6.5 to generate Gd DTPA-HSA. Mass spectrometry was then conducted to verify the Gd linking efficiency. Then, 60 mg Gd DTPA-HSA solution was mixed with free ICG (J&K Scientific, China) (10 mg dissolved in 2 mL DMSO). The solution was stirred for 12 h at room temperature. Finally, the mixture was purified using a Sephadex G50 column (GE 121 Healthcare, USA).
Nanoparticle Characterization: The molecular weight of synthetic probe precursors was determined by mass spectrometry (Bruker Daltonics, MA, USA). Particle size was determined by transmission electron microscopy (JEOL, Japan). Gel electrophoresis (BIO-RAD, CA, USA), zeta potentiometer monitoring (Malvern, UK) the potential of the probe and multispectral laser imaging system (Azure Sapphire, USA) used to detect the conjugation of NPs and antibody was successful. Gd content in solution was determined by inductively coupled plasma optical emission spectrometry (Thermo Fisher Scientific, USA), and Gd content in mouse tissue was calculated. Absorption spectra were analyzed using a Multiskan Spectrum Microplate Spectrophotometer (Thermo Fisher Scientific, USA).
ROS Cell Levels: MDA-MB-231 cells were seeded in a 12-well slide chamber at 1 × 10 5 cells per well, cultured overnight, and incubated with/without NPs-Bev ([Gd 3+ ] = 25 × 10 −6 m) for 4 h. Plates were then irradiated (or not) with X-rays (6 Gy), and H2DCFDA concentrations which were used to assess ROS levels by using kits according to manufacturer's instructions (Thermo Fisher Scientific, USA). Fluorescence images were obtained using confocal microscopy (Zeiss, Germany) and analyzed using ImageJ software.
Flow Cytometry: MDA-MB-231 cells were seeded in a 12-well slide chamber at 1 × 10 5 cells per well, cultured overnight, and incubated with/without NPs-Bev ([Gd 3+ ] = 25 × 10 −6 m) for 4 h. Plates were then irradiated (or not) with X-rays (6 Gy), and apoptosis quantified using an Annexin V binding kit (Beyotime, China) and flow cytometry. Flow cytometry was performed as previously described. [42] Ex Vivo and In Vivo MRI: Samples were separately prepared for the MRI phantom study. NPs-Bev and NPs-IgG were prepared at concentrations of 0.025, 0.05, 0.1, 0.2, and 0.4 mm with respect to Gd 3+ ions in 1× PBS buffer. Deionized water was used as a control. Longitudinal relaxation times were measured to calculate sample relaxation rates. MDA-MB-231 cells bearing tumor mice were imaged by T1-weighted MRI to evaluate NPs-Bev as it specifically targeted tumors. Gd-DTPA and NPs-IgG were used as controls. Animals were imaged using a 9.4T MRI scanner (Bruker, Germany). MRI signal intensities in regions of interest were tested upon the intravenous injection of NPs-Bev, and at 0.5, 1, 2, 4, 12, 24 36, and 48 h postinjection. For quantitative comparisons, TBR ratios were calculated and analyzed using Radiant DICOM Veiver2020.2.
NIR-II Fluorescence Imaging and Biodistribution: Mice bearing subcutaneous MDA-MB-231-Luc tumors (volume = 200-300 mm 3 ) were randomly divided into two groups (n = 3) and injected with NPs-Bev and NPs-IgG through the tail vein at equivalent ICG doses (2.0 mg kg −1 ). Then, Bev blocking experiments were performed in vivo. MDA-MB-231-Luc tumorbearing mice were randomized into two groups (n = 3) and injected with PBS or Bev (250 mg kg −1 ) at 30 min before NPs-Bev injections (ICG = 2.0 mg kg −1 ). Mice were then anesthetized and fluorescence signals collected by the NIR-II imaging system (Suzhou Yingrui Optical Technology Co., Ltd., China) at different times. NPs-Bev ex vivo biodistribution was also calculated. MDA-MB-231-Luc tumor-bearing mice were intravenously injected with NPs-Bev and humanely euthanized at 36 h to collect and visualize tumors and major organs for NIR-II imaging analysis.
Fluorescence-Image-Guided Surgery in a Multiple-Microtumor Model: In mice bearing 30-60 mm 3 MDA-MB-231-Luc microtumors, bioluminescence imaging was used to calculate the number of microtumors. Then, mice were injected with NPs-Bev through the tail vein (ICG = 2.0 mg kg −1 ) and NIR-II fluorescence imaging performed at 36 h postinjection, and the number of microtumors calculated. Consistency of bioluminescence and fluorescence imaging in calculating tumor number and pathology was the gold standard.
Fluorescence-Image-Guided Surgery in MDA-MB-231-Luc Tumor-Bearing Mice: Mice bearing 300-400 mm 3 MDA-MB-231-Luc tumors were randomized into two groups (n = 8); a visible light navigation (VL) surgical group and a NIR-II navigation surgical group (NIR-II). Groups were intravenously injected with NPs-Bev (equivalent ICG dose = 2.0 mg kg −1 ). Tumors were resected in the NPs-Bev group at 36 h postinjection. In the NIR-II group, mice were placed under a NIR-II imaging navigation surgical instrument for tumor resection. If residual fluorescence was detected, the tumor was removed until no residual fluorescence signals remained. In vivo mean fluorescence intensity (MFI) information was collected and analyzed. In the VL group, mouse tumors were surgically removed under visible light without surgical navigation. Mice were monitored every other day for tumor recurrence. Bioluminescence was performed at 35 days postinjection and mice with tumor recurrence volumes > 1500 mm 3 and experiencing 25% body weight loss were humanely euthanized.
Fluorescence Imaging in Spontaneous Breast Cancer Transgenic Mice: 6-8 weeks old MMTV-PyVT spontaneous breast cancer mice (n = 3) and wildtype mice (n = 3) were intravenously injected with NPs-Bev (ICG = 2.0 mg kg −1 ). Mice were humanely euthanized and NIR-II fluorescence imaging performed at 36 h postinjection. Then, the 1st-5th breast tissue of both groups were resected, fluorescence imaging performed using NIR-II imaging, and in vitro tumor MFI calculated. Histopathologies were assessed by two breast cancer specialists, and receiver operator characteristic (ROC) curves were fitted using MFI data and histopathology results. Additionally, 10 μm paraffin block slices were imaged using fluorescence flatbed scanning (Odyssey CLx, USA). Finally, adjacent 4 μm sections were prepared for H&E and IHC (VEGF-A) staining to evaluate correlations between fluorescence intensity and histology.
NPs-Bev Biosafety: NPs-Bev or PBS was injected into healthy BALB/C mice through the tail vein. Mice were reared in a normal barrier environment and body weights monitored. Peripheral blood was collected at days 1, 3, 7, and 28 after probe injection, and liver and kidney function indices examined. Mice were sacrificed by cervical dislocation, and the main organs removed for H&E staining to observe if histological morphology and pathological changes such as necrosis and degeneration had occurred. Liver and kidney functions and routine blood indices such as ALT, AST, CK-MB, UREA, CREA, WBC, Lymph, Mon#, Gran#, RBC, HGB, and PLT were compared with control mice (injected with PBS).
Statistical Analysis: SPSS 19.0 (SPSS Inc., Chicago, IL, USA) was used for statistical analysis. Differences between variables were measured using Student's t-tests, Pearson's chi-squared tests, and Spearman's log-rank tests. Statistical significance was accepted at p < 0.05.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.