Multimodal silica nanoparticles are effective cancer-targeted probes in a model of human melanoma

Nanoparticle-based materials, such as drug delivery vehicles and diagnostic probes, currently under evaluation in oncology clinical trials are largely not tumor selective. To be clinically successful, the next generation of nanoparticle agents should be tumor selective, nontoxic, and exhibit favorable targeting and clearance profiles. Developing probes meeting these criteria is challenging, requiring comprehensive in vivo evaluations. Here, we describe our full characterization of an approximately 7-nm diameter multimodal silica nanoparticle, exhibiting what we believe to be a unique combination of structural, optical, and biological properties. This ultrasmall cancer-selective silica particle was recently approved for a first-in-human clinical trial. Optimized for efficient renal clearance, it concurrently achieved specific tumor targeting. Dye-encapsulating particles, surface functionalized with cyclic arginine–glycine–aspartic acid peptide ligands and radioiodine, exhibited high-affinity/avidity binding, favorable tumor-to-blood residence time ratios, and enhanced tumor-selective accumulation in α_vβ₃ integrin–expressing melanoma xenografts in mice. Further, the sensitive, real-time detection and imaging of lymphatic drainage patterns, particle clearance rates, nodal metastases, and differential tumor burden in a large-animal model of melanoma highlighted the distinct potential advantage of this multimodal platform for staging metastatic disease in the clinical setting.

Despite recent advances in nanoparticle probe development for biomedicine, the translation of targeted diagnostic particle platforms remains challenging (1–3). Nanoparticle-based materials currently under evaluation in oncology clinical trials are largely nontargeted drug delivery vehicles or devices for thermally treating tissue and not typically surface modified for direct detection by clinical imaging tools. Tumor-selective diagnostic probes, in addition to satisfying critical safety benchmarks, need to elucidate targeted interactions with the microenvironment and their effects on biological systems. Several key factors that limit the translation of such particle probes have been identified: suboptimum pharmacokinetics, in vivo probe nanotoxicity (4), evolving regulatory considerations, reimbursement issues, and resource-intensive scale-up of manufacturing for clinical trials. To address such issues and comply with regulatory and clinical practice guidelines, rigorous quantitative imaging approaches and analysis tools, such as PET, will be essential for evaluating a variety of probes undergoing preclinical testing or transitioning into early-phase clinical trials. By attaching a radiolabel to the particle surface, the clinical potential of these molecularly targeted probes can be sensitively and comprehensively defined according to the following set of diagnostic criteria: favorable distribution and targeting kinetics (1, 5–9), efficient renal clearance (10, 11), extended circulation (blood residence) times (5, 12, 13), pronounced tissue signal amplification (5, 6, 14), and improved tumoral penetration (15).

Recently reported design criteria have been proposed for advancing targeted probes to the clinic (11). These have largely been restricted to improving renal clearance properties of inorganic particles containing inherently toxic components. The relative lack of in vivo toxicology data associated with these probes has led to persistent and valid safety concerns (16), prompting a number of particle-design modifications (9, 17–19), including size, composition, and surface chemistry, that promote more rapid clearance through kidney (renal) or liver (hepatic) pathways (20). While such approaches may minimize toxicity, the ultimate fate and impact of residual particles in the reticuloendothelial system (RES; liver, spleen, lymph nodes, bone marrow) remains unclear, particularly if toxicity is related to prolonged organ residence times (16, 17). Accelerated particle clearance may also preclude certain clinical applications as a result of restrictions imposed upon the available imaging period. Further, while rapid particle extravasation may facilitate clearance and delivery to tumors, reduced receptor binding (5) and increased nonspecific tissue dispersal (21, 22) may also occur, increasing image background and, potentially, toxicity by prolonging retention.

Only a relatively small percentage of new imaging agents undergoing comprehensive preclinical testing potentially satisfy diagnostic criteria for translation (2). Although a number of fluorescent particle platforms have been investigated (23–26), only Cornell dots (C dots) meet the aforementioned criteria and have received the first FDA-approved investigational new drug approval of their class and properties for a first-in-human clinical trial. Surface functionalized with small numbers of targeting peptides (cyclic arginine–glycine–aspartic acid [cRGDY] peptides) and long-lived radiolabels, this α_vβ₃ integrin–targeting particle tracer has a unique combination of structural, optical, and biological properties: bulk renal clearance (10), favorable targeting kinetics, lack of acute toxicity, superior photophysical features, and multimodal (PET-optical) imaging capabilities. To the best of our knowledge, these properties have not been collectively achieved or described for any single inorganic particle platform. Importantly, the translation of this cancer-targeted, PET-optical platform represents a significant clinical advance over our previous preclinical optical imaging work using approximately 30-nm in diameter [i.d.] non-PEGylated (24, 27) and approximately 3.3- and 6-nm i.d. PEGylated, nontargeting probes (10) in non-tumor-bearing animals.

We highlight the distinct advantages of using this renally excreted, multimodal agent for tumor-selective targeting and nodal mapping studies in both small- (mouse) and large-animal (miniswine) melanoma models. By combining key benefits of PET (depth penetration, quantitation) with those of deep-red/NIR multiscale optical imaging technologies (28–30) (enhanced sensitivity/contrast, multispectral capabilities), we initially monitored uptake, lymphatic drainage, and particle clearance in melanoma-bearing mice (31–34). These methods were extended to a larger-animal spontaneous melanoma miniswine model (29) to assess the feasibility of performing intraoperative image-guided metastatic disease detection, localization, and staging; to determine whether differences in nodal tumor burden could be sensitively discriminated; and to simulate more accurately the application of sentinel lymph node (SLN) biopsy procedures in humans.

Nanoparticle design and characterization. From a clinical standpoint, modifying an inherently nontoxic and biocompatible material, amorphous silica, for targeted diagnostics and/or therapeutics offers an attractive strategy for ultimately achieving a favorable toxicologic profile in vivo, while, at the same time, preserving its versatility in a variety of patient settings. To realize such a platform, Cy5 dye-encapsulating core-shell silica nanoparticles (emission maxima, >650 nm), coated with methoxy-terminated polyethylene glycol (PEG) chains (PEG ~0.5 kDa), were prepared according to previously published protocols (see Methods) (10, 27). The neutral PEG coating prevented uptake by other cells (opsonization). The use of bifunctional PEGs enabled attachment of small numbers of α_νβ₃ integrin–targeting cRGDY peptide ligands (~6–7 ligands per particle) (34) to maintain a small hydrodynamic size, facilitating efficient renal clearance.

Peptide ligands were labeled with the positron-emitting radionuclide ¹²⁴I through the use of a tyrosine linker to provide a signal that can be quantitatively imaged in 3 dimensions by PET (¹²⁴I-cRGDY-PEG-ylated dots [¹²⁴I-cRGDY-PEG-dots]; Figure 1A). An important practical advantage of relatively long-lived ¹²⁴I (physical half-life, 4.2 d) is that sufficient signal persists long enough to allow radiodetection up to at least several days after administration, when background activity has largely cleared, and tumor-to-background contrast is maximized. PEG- and cRAD-bound, PEG-coated particles containing tyrosine residues for ¹²⁴I labeling served as the control particle probes, the latter bearing a scrambled peptide (i.e., cyclo-[Arg-Ala-Asp-Tyr]). Purification of radiolabeled samples by size exclusion chromatography (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI45600DS1) resulted in radiochemical yields of more than 95%. Hydrodynamic diameters of approximately 7, 6.8, and 6.5 nm (mean ± SD, n = 15) were measured for cRGDY-PEG-dots, PEG-dots, and cRADY-PEG-dots, respectively, using fluorescence correlation spectroscopy (FCS) (Figure 1B and Table 1). The relative brightness of the cRGDY-PEG-dots was determined to be approximately 200% greater than that of the free dye (Table 1), consistent with earlier results (10, 35). Based on these physicochemical properties, we anticipated achieving a favorable balance between selective tumor uptake and retention versus clearance of the targeted particle, thus maximizing target-tissue localization while minimizing toxicity and normal-tissue radiation doses.

Figure 1

Multimodal C dot design for α_νβ₃ integrin targeting and characterization. (A) Schematic representation of the ¹²⁴I-cRGDY-PEG-ylated core-shell silica nanoparticle with surface-bearing radiolabels and peptides and core-containing reactive dye molecules (insets). (B) FCS results and single exponential fits for measurements of Cy5 dyes in solution (black) and PEG-coated (PEG-dot, red) and PEG-coated, cRGDY-labeled dots (blue, underneath red data set) showing diffusion time differences as a result of varying hydrodynamic sizes.

Table 1

Photophysical characterization using FCS^A

In vitro receptor-binding studies. To examine in vitro receptor-binding affinity and specificity of ¹²⁴I-cRGDY-PEG-dots to melanoma and vascular endothelial cell surfaces relative to those of particle controls (¹²⁴I-PEG-,¹²⁴I-cRADY-PEG-dots), α_vβ₃ integrin–overexpressing (M21, HUVEC) and –nonexpressing (M21L) lines were used. Initially, M21 cells were incubated with increasing ¹²⁴I-cRGDY-PEG-dot concentrations in the presence of excess nonradiolabeled cRGD peptide, and γ-counting was performed (Figure 2A). Scatchard analysis of the binding data yielded a dissociation equilibrium constant, K_d, of 0.51 nM (Figure 2A, inset) and a receptor concentration, B_max, of 2.5 pM. Based on the B_max value, the α_vβ₃ integrin receptor density was estimated to be 1.0 × 10⁴ integrin receptors per M21 cell, in reasonable agreement with the previously published estimate (36) of 5.6 × 10⁴ integrin receptors per cell for this cell line. Incremental increases in integrin-specific M21 cellular uptake were also observed over a temperature range of 4°C to 37°C, suggesting that cellular internalization contributed to overall uptake (data not shown).

Figure 2

Competitive integrin receptor binding studies with ¹²⁴I-cRGDY-PEG-dots, cRGDY peptide, and anti–α_νβ₃ antibody using 2 cell types. (A) High-affinity and specific binding of ¹²⁴I-cRGDY-PEG-dots to M21 cells by γ-counting. The inset shows Scatchard analysis of binding data, plotting the ratio of the concentration of receptor-bound (B) to unbound (or free [F]) radioligand or the bound-to-free ratio (B/F) versus the receptor-bound receptor concentration; the slope corresponds to the dissociation constant, K_d. (B) α_νβ₃ Integrin receptor blocking of M21 cells using flow cytometry and excess unradiolabeled cRGD or anti–α_νβ₃ antibody prior to incubation with cRGDY-PEG-dots and nonspecific binding with controls (RAD-PEG-dots, PEG-dots). (C) Specific binding of cRGDY-PEG-dots to M21 cells as against M21L cells lacking surface integrin expression using flow cytometry. Anti–α_νβ₃ integrin receptor antibody concentrations were used at 100 times (i.e., 100x) and 250 times (i.e., 250x) the particle (i.e., ¹²⁴I-cRGDY-PEG-dot) concentration. (D) Specific binding of cRGDY-PEG-dots to HUVECs by flow cytometry. Each bar represents mean ± SD of 3 replicates.

Binding specificity of cRGDY-PEG-dots was further demonstrated with flow cytometry. Competitive binding assays showed complete blocking of receptor-mediated binding (Figure 2B) using anti–α_vβ₃ integrin antibody. No significant reduction was seen in the magnitude of receptor binding (~10% of M21) with M21L cells (Figure 2C) using either excess cRGDY peptide or anti–α_vβ₃ integrin antibody. These results were confirmed by additional γ-counting studies, and a 50% binding-inhibition concentration, IC₅₀, of 1.2 nM was determined for the ¹²⁴I-cRGDY-PEG-dot. This IC₅₀ value suggests greater potency than that recently reported for engineered, high-affinity ⁶⁴Cu-DOTA-knottin peptides (IC₅₀ ~20 nM) (37). Potency of cRGDY-PEG-dots was additionally evaluated by an anti-adhesion assay (5, 38) in order to assess particle multivalent interactions with α_vβ₃ integrin receptors. A multivalent enhancement factor of greater than 2.0 was determined for cRGDY-PEG-dots relative to monomeric cRGD peptides after incubating each with M21 cells (data not shown). A final set of studies showed that, similar to M21 cells, excess antibody effectively blocked cRGDY-PEG-dot receptor binding to HUVECs (Figure 2D).

Biodistribution and clearance studies. Biodistribution and blood, renal, and hepatobiliary clearance were evaluated by i.v. administering tracer doses (~0.2 nanomoles) of ¹²⁴I-cRGDY-PEG-dots and ¹²⁴I-PEG-dots to M21 tumor xenograft models. Although the percentage of the injected dose (ID) per gram tissue (%ID/g) for the targeted probe was measured over a 196-hour postinjection (p.i.) time interval, comparison of the ¹²⁴I-cRGDY-PEG-dot (Figure 3A) and ¹²⁴I-PEG-dot tracers (Figure 3B) was restricted to a 96-hour window, as in vivo activity for the latter was not detectable at 1 week p.i. Statistically significant (P < 0.05) differences in tracer uptake (%ID/g) values were observed for blood, tumor, and major organs at 4 and 96 hours p.i. as well as at 24 hours p.i. for tumor and several other tissues (Supplemental Table 1). The targeted probe was almost entirely eliminated from the carcass at 1 week p.i. (~3% of the ID remained). Tracer clearance half times (T_1/2) and bioavailability for blood, tumor, and major organs are shown in Table 2. For blood, the relatively long T_1/2 value (mean ± SD) for the ¹²⁴I-PEG-dot tracer (i.e., 7.3 ± 1.2 hours), which is similar to that found for the nonradiolabeled particle control in normal mice by fluorescence detection (10), decreased to 5.6 ± 0.15 hours (Figure 3A, inset) for the ¹²⁴I-cRGDY-PEG-dot tracer.

Figure 3

Pharmacokinetics and excretion profiles of the targeted and nontargeted particle probes. (A) Biodistribution of ¹²⁴I-cRGDY-PEG-dots (targeted) in M21 tumor-bearing mice at various times from 4- to 168-hours p.i. The inset shows a representative plot of these data for blood to determine the T_1/2. Sm. int., small intestine; Lg. int., large intestine; conc., concentration. (B) Biodistribution of ¹²⁴I-PEG-dots (untargeted) from 4- to 96-hours p.i. (C) Clearance profile of urine samples collected up to 168-hours p.i. of unradiolabeled cRGDY-PEG-dots (n = 3 mice, mean ± SD). The inset shows a strong correlation between the percentage of the injected particle dose excreted and the corresponding measured fluorescence signal. (D) Corresponding cumulative %ID/g for feces at intervals up to 168-hours p.i. (n = 4 mice). For biodistribution studies, bars represent mean ± SD.

Table 2

Radiation dosimetry for i.v.-injected ¹²⁴I-RGDY-PEG-dots and ¹²⁴I-PEG-dots

By appropriate organ mass-adjusted translation of the foregoing biodistribution data to humans, human normal organ radiation doses were derived and found to be comparable to those of other commonly used diagnostic radiotracers (Table 2, columns 8 and 9). Along with the findings of formal toxicity testing that the targeted probe was nontoxic at i.v. IDs 100 times the proposed human dose equivalent (based on standard body mass-normalized allometric scaling) and resulted in no tissue-specific pathologic effects (Supplemental Figure 2 and Supplemental Table 2), first-in-human targeted and nontargeted molecular imaging applications with these i.v.-injected agents are planned.

Efficient renal excretion was found for the approximately 7-nm diameter targeted and nontargeted probes over a 168-hour time period by fluorimetric analyses of urine samples. Fluorescence signals were background corrected and converted to particle concentrations (%ID/μl) based on a serial dilution calibration scheme (Figure 3C, inset; Supplemental Table 3, column 2; and ref. 10). Concentration values, along with age-dependent conservative estimates of the average urine excretion rate (39), permitted the cumulative excreted fluorescence signal (%ID) to be computed (Supplemental Table 3, column 4). Nearly half of the ID was observed to be excreted over the first 24 hours p.i. and approximately 72% was excreted by 96 hours (Figure 3C), suggesting that the bulk of excretion occurred in the first day p.i. No significant particle fluorescence in urine could be detected 168 hours p.i. Fluorimetric determinations correlated strongly with corresponding γ-counter–derived %ID values (R² = 0.965, Supplemental Table 3, column 5), suggesting that no significant free ¹²⁴I-cRGD was present. In addition, as cumulative percentage dehalogenation values of urine samples by radio-TLC averaged less than 10% over a 24-hour period, RES trapping was not significantly underestimated (Supplemental Table 3, column 6). Fecal excretion profiles of the ¹²⁴I-cRGDY-PEG-dot indicated that, on average, 7% and 15% of the ID was eliminated over 24 and 96 hours, respectively (Figure 3D). To assess particle integrity, FCS analysis of urine specimens obtained prior to injection, and at 4 and 24 hours p.i. of the targeted probe, revealed that the particle was excreted intact and without release of the encapsulated dye (Supplemental Table 4); minor differences in solvent or surface chemistry likely contributed to the observed findings.

Serial whole-body PET studies. PET imaging of integrin expression in M21 and M21L subcutaneous hind leg xenograft mouse models was performed at multiple time points p.i. after i.v. injection of either ¹²⁴I-cRGDY-PEG-dots or control particles (i.e., ¹²⁴I-PEG-dots, ¹²⁴I-cRAD-PEG-dots). Representative whole-body coronal microPET images at 4-hours p.i. of M21 and M21L tumors and at 24-hours p.i. of M21 tumor are shown in Figure 4A; a corresponding representative 24-hour fluorescence image of the tumor is shown in Figure 4B. The specific targeting of the α_vβ₃ integrin–overexpressing M21 tumor is clearly visible from these images. Average tumor %ID/g and SDs are shown for the following groups (Figure 4C): M21 (n = 7) and M21L (control) tumors (n = 5) receiving the targeted ¹²⁴I-cRGDY-PEG-dots; M21 tumor mice (n = 5) receiving nontargeted ¹²⁴I-PEG-dots; and M21 mice (n = 3) receiving the ¹²⁴I-cRAD-PEG-dots (Supplemental Figure 3A). At the time of maximum tumor uptake (~4 hours p.i.), up to 3-fold higher activity concentrations (%ID/g) were seen in M21 tumors than in controls. Differences were statistically significant at all time points p.i. (P < 0.05), except at 1 hour (P = 0.27). In vivo integrin receptor blocking studies of M21 tumors (n =3) additionally confirmed particle binding specificity, with a ~6-fold decrease in tumor uptake in targeted tracer uptake before versus after administration of excess cRGD peptide (Supplemental Figure 3B).

Figure 4

Serial in vivo PET imaging of tumor-selective targeting. (A) Representative whole-body coronal microPET images at 4-hours p.i., demonstrating M21 (left, arrow) and M21L (middle, arrow) tumor uptakes of 3.6 and 0.7 %ID/g, respectively, and enhanced M21 tumor contrast at 24 hours (right). (B) Representative 24-hour fluorescence image of the tumor. (C) In vivo uptake of ¹²⁴I-cRGDY-PEG-dots in α_vβ₃ integrin–overexpressing M21 (black, n = 7 mice) and nonexpressing M21L (medium dark gray, n = 5 mice) tumors, ¹²⁴I-PEG-dots in M21 tumors (dark gray, n = 5), and ¹²⁴I-cRAD-PEG-dots in M21 tumors (light gray, n = 3). (D) M21 tumor-to-muscle ratios for ¹²⁴I-cRGDY-PEG-dots (black) and ¹²⁴I-PEG-dots (gray). (E) Correlation of in vivo and ex vivo M21 tumor uptakes of cRGDY-labeled and unlabeled probes. Each bar represents mean ± SD.

Image-derived tumor-to-muscle uptake (%ID/g) ratios for the ¹²⁴I-cRGDY-PEG-dots revealed enhanced tumor contrast at later times (~24–72 hours p.i.), while those for ¹²⁴I-PEG-dots declined progressively with time (Figure 4D). This finding suggested that ¹²⁴I-cRGDY-PEG-dots were tumor selective, which became more apparent as blood activity was cleared during the initial 24-hour period (compare Figure 4D with Figure 3A, inset). A statistically significant correlation was found between PET-derived tumor %ID/g values for both the targeted and nontargeted probes, and the corresponding ex-vivo γ-counter–derived tumor %ID/g values (correlation coefficient r = 0.94, P < 0.0016; Figure 4E), confirming the accuracy of PET for noninvasively deriving quantitative biodistribution data.

Serial in vivo Cy5 fluorescence imaging and microscopy. In vivo fluorescence imaging studies were conducted using our targeted and nontargeted particle probes for mapping local/regional nodes and lymphatic channels in small-animal models, as the size of these structures (1–2 mm) may not be adequately resolved by PET imaging. Representative lymph node mapping was initially performed across spatial scales using the targeted probe, in conjunction with live-animal whole-body optical imaging (Figure 5A) and fluorescence microscopy techniques (Figure 5B), to visualize lymphatic drainage from the peritumoral region to the inguinal and axillary nodes in surgically exposed living animals. Peritumoral administration of the targeted probe revealed drainage into and persistent visualization of adjacent inguinal and popliteal nodes as well as more remote axillary nodes over a 4-hour interval with less than 1-mm resolution. Higher-resolution (i.e., submillimeter) fluorescence images (Figure 5B, bottom row) permitted more detailed intranodal architecture to be localized, including high endothelial venules, known to facilitate passage of circulating naive lymphocytes into the node; this may have important implications for nodal staging and the ability to detect micrometastases at earlier stages of disease. Smaller, less intensely fluorescent lymphatic branches were also visualized by fluorescence microscopy in the axillary region (data not shown). Thus, the smaller size of the targeted probe permitted the first draining, or sentinel, node proximal to the tumor to be visualized and also enabled detection of more distant nodes and visualization of the pattern of lymphatic drainage.

Figure 5

Nodal mapping using multiscale near-infrared optical fluorescence imaging. (A) Whole-body fluorescence imaging of the tumor site (T) and draining inguinal (ILN) and axillary (ALN) nodes and communicating lymphatics channels (LCs; bar) 1-hour p.i. in a surgically exposed living animal. (B) Corresponding coregistered white-light and high-resolution fluorescence images (top row) and fluorescence images only (bottom row), revealing nodal infrastructure of local and distant nodes, including high endothelial venules (HEVs). (C) Whole-body fluorescence image of the tumor site 10 minutes after subdermal PEG-dot injection. (D) Delayed whole-body fluorescence image of the tumor site 1 hour after PEG-dot injection. (E) Percentage increase in the area of fluorescence (fluor) relative to that measured at 10-minutes p.i. for targeted and nontargeted probes. Scale bars: 1.0 cm (A); 500 microns (B); 3 mm (C and D).

Differential rates of particle clearance were assessed for both nontargeted (PEG-dots) and targeted probes in intact, living mice bearing hind limb M21 xenografts versus those in controls. After subdermal, 4-quadrant, peritumoral injection, relatively rapid dispersion of both particle probes in M21 mice was observed over a 1-hour time period (measured against the initial 10-minute time point), with representative PEG-dot images shown (Figure 5, C and D). Using the fluorescence area increase (%) as a measure of particle dispersion, no significant difference was seen between the targeted and nontargeted probes over the first hour p.i. (Figure 5E). Further, in control animals, no percentage increase was found for either probe (data not shown). Although no fluorescent nodes were detected in intact mice, surgical exposure revealed lymphatic drainage from the inguinal region to the axilla for both probes, similar to that seen in Figure 5A.

Multimodal approaches were extended to a larger-animal spontaneous melanoma miniswine model (32) in order to assess the feasibility of performing real-time, intraoperative image-guided metastatic disease detection and staging and to determine whether tumor burden could be sensitively discriminated with correlative histology. After i.v. injection, whole-body dynamic ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) PET-CT scanning identified an ¹⁸F-FDG-avid melanomatous lesion adjacent to the spine on the upper back as well as a hypermetabolic right-sided SLN in the posterior neck (Figure 6A, white arrow). Subdermal, 4-quadrant peritumoral injection of ¹²⁴I-RGD-PEG-dots 48 hours later confirmed the ¹⁸F-FDG imaging findings and identified 2 additional hypermetabolic nodes, 1 within the left posterior neck (Figure 6B, arrowhead) and the second immediately anterior to the SLN 5-minutes and 1-hour p.i. by dynamic PET scanning. No additional PET-avid nodes or other suspicious areas of increased uptake were seen on a 1-hour whole-body PET scan. Corresponding optical imaging of the excised right-sided SLN demonstrated fluorescence signal within the black-pigmented (melanin-containing) gross specimen (Figure 6, C and D), which measured 1.3 × 1.0 × 1.5 cm³, as compared with the smaller posterior left-sided hypermetabolic node, which measured 1.0 × 0.6 × 1.0 cm³ (Supplemental Figure 4A). Significantly increased background-corrected SLN activity, detected with a hand-held intraoperative positron probe, corroborated these optical findings (Figure 6D, annotation).

Figure 6

Imaging of metastatic disease in a spontaneous melanoma miniswine model. (A) Whole-body dynamic ¹⁸F-FDG PET-CT sagittal and axial views, demonstrating primary tumor (green arrow) and single SLN (white arrow) posteriorly within the right (Rt) neck after i.v. injection. ant, anterior. (B) High-resolution dynamic PET-CT scan 1 hour after subdermal, 4-quadrant, peritumoral injection of ¹²⁴I-RGD-PEG-dots (SLN, arrow; left-sided node, arrowhead). (C) Whole-body Cy5 fluorescence image of the excised SLN. (D) Gross image of the cut surface of the black-pigmented SLN (asterisk), which measured 1.3 × 1.0 × 1.5 cm³, and annotated γ counted activity. (E) Low-power view of H&E-stained SLN, demonstrating scattered melanomatous clusters. (F) Corresponding high-power view of H&E-stained SLN, revealing melanoma cells (yellow arrowheads) and melanophages (white arrowhead). (G) Low-power image of a melanoma-specific marker, HMB-45, in representative SLN tissue. (H) High-power image of HMB-45–stained SLN tissue. (I) Low-power image of representative normal porcine nodal tissue. (J) High-power image of representative normal porcine nodal tissue. (K) Low-power view of H&E-stained contralateral hypermetabolic lymph node, demonstrating scattered melanomatous clusters (arrowhead). Tumor burden in this smaller node (1.0 × 0.6 × 1.0 cm³) was estimated to be 10- to 20-fold less than that in the SLN by pathological analysis. Scale bars: 1 mm (E, G, I, and K); 20 μm (F, H, and J).

H&E-stained tissue sections from the SLN showed scattered, dark melanomatous clusters on low-power views (Figure 6E), which were found to comprise both melanoma cells and melanin-containing macrophages (i.e., melanophages) on high-power views (Figure 6F), similar to findings for excised primary lesions (data not shown). Immunohistochemical staining of the SLN with a known human melanoma marker, HMB45, demonstrated positive expression of this marker on low-power views (Figure 6G) and high-power views (Figure 6H). Representative normal-appearing porcine nodal tissue harvested from the neck revealed no metastatic infiltration in representative low-power views (Figure 6I) and high-power views (Figure 6J). By contrast, low-power views (Supplemental Figure 4B and Figure 6K) of H&E-stained sections from the posterior left-sided hypermetabolic node demonstrated a few smaller-sized melanomatous clusters containing melanoma cells and melanophages (Supplemental Figure 4C). Tumor burden in this smaller node, estimated to be 10- to 20-fold less than in the SLN by pathological analysis, was sensitively discriminated by the targeted particle probe.

Here, we report on an ultrasmall, high-affinity, and efficiently cleared silica nanoparticle probe, which has been recently approved for first-in-human clinical trials and which successfully overcomes a number of the limitations of other particle platforms. This multimodal platform has advanced to the point of clinical translatability. Its applications include real-time, intraoperative detection and imaging of nodal metastases, differential tumor burden, and lymphatic drainage patterns in melanoma. Although several investigators have synthesized radiolabeled fluorescent particle probes (25, 30, 40), our multimodal agent has been radiolabeled with the long-lived positron-emitter iodine-124, and, thus, we believe it can provide unique longer-term pharmacokinetic clearance and targeting information over the course of days. The complementary nature of this platform, coupled with its small size (~7-nm i.d.), may facilitate clinical assessments by enabling the seamless integration of imaging data acquired at different spatial, temporal, and sensitivity scales, potentially providing new insights into fundamental molecular processes governing tumor biology.

The results of this study underscore the clear-cut advantages offered by PET. Using this quantitative and highly sensitive imaging tool, we were able to noninvasively extract an accurate, reproducible, and comprehensive body of data for the targeted probe: (a) molecular information, including receptor expression levels, binding affinity, and specificity; (b) in vivo distribution and targeting kinetics; (c) clearance and dehalogenation profiles; (d) blood/tissue residence times and bioavailability; and (e) radiation dosimetry.

Our in vitro results show receptor-binding specificity of the ~7-nm–targeted particle probe to M21 cells and HUVECs. Similar findings have been reported with receptor-binding assays using the same cell types but with the monovalent form of the peptide (36). Importantly, the multivalency enhancement of the cRGDY-bound particle probe, along with the extended blood and tumor T_1/2 values, are key properties associated with the particle platform that are not found with the monovalent form of the peptide.

The integrin-binding peptide, cRGD, a well-established integrin-binding molecular marker, was selected to elucidate the biological and kinetic properties of the peptide-bound particle and additionally enable the biological performance of our multimodal platform to be benchmarked against other cRGD agents (i.e., peptide tracers). However, another more tumor-specific, high-affinity ligand could be investigated as the targeting moiety in future studies. The biological properties of particle-based systems are generally quite different from those of simple (i.e., small monovalent) targeting peptides, and the choice of the proper targeting agent (i.e., molecular versus particle-based probes) in the clinical setting will rest, in part, on the application of interest, the tumor type/composition, and standard-of-care considerations.

The relatively long blood T_1/2 for the ¹²⁴I-PEG-dot tracer may be a consequence of the chemically neutral PEG-coated surface, rendering the probe biologically inert and significantly less susceptible to phagocytosis by the RES. However, recognition of the ¹²⁴I-cRGDY-PEG-dot tracer by target integrins and/or more active macrophage activity may have led to reductions in the T_1/2 value. These values, however, are substantially longer than published blood T_1/2 values of existing cRGDY peptide tracers (~13 minutes) (5), potentially leading to increased probe bioavailability, facilitating tumor targeting, and yielding higher tumor uptake over longer time periods. Moreover, the tumor T_1/2 for the ¹²⁴I-cRGDY-PEG-dot was found to be 13-times greater than that for blood, compared with only a 5-fold difference for the ¹²⁴I-PEG-dot, suggesting substantially greater target-tissue localization of the former than the latter. Such mechanistic interpretations of the in vivo data can be exploited clinically to refine diagnostic, treatment-planning, and treatment-monitoring protocols.

The greater accumulation in and slower clearance from M21 tumors, relative to that of surrounding normal structures, allows discrimination of specific tumor uptake mechanisms from nonspecific mechanisms (i.e., tissue perfusion, leakage) in normal tissues. However, a small component of the M21 tumor uptake can presumably be attributed to vascular permeability alterations (i.e., enhanced permeability and retention effects) (41), largely reflected in the observed %ID/g increases for the control tracer (¹²⁴I-PEG-dots, Figure 4C) at earlier p.i. time points. At 1-hour p.i., no significant %ID/g increases were seen in the M21 tumors over the controls. This observation may represent the effects of differential perfusion in the first hour, with tumor accumulation and retention primarily seen at later p.i. times (i.e., 24 hours). Further, in comparison with those of the clinically approved peptide tracer, ¹⁸F-galacto RGD (42, 43), nearly 2-fold greater maximum uptake values were found in M21 tumors for the targeted dots at 2-hours p.i. (data not shown), while additionally offering advantages of multivalent binding and extended circulation times.

The advantages of the combined optical-PET probe highlight the versatility of the platform for in vivo applications. This is particularly true in small-animal models, in which the ability to assess anatomic structures having sizes at or well below the resolution limit of the PET scanner (i.e., the so-called partial-volume effect) may undermine detection and quantitation of activity in lesions. In these cases, assessment of metastatic disease in small local/regional nodes, important clinically for melanoma staging and treatment, may not be adequately resolved by PET imaging, given that the size of the nodes we typically observed was on the order of the PET spatial resolution (1–2 mm). By exploiting the significantly improved photophysical features of encapsulated dyes, such as Cy5, as well as the enhanced detection sensitivity and contrast achievable at longer emission wavelengths, detailed information pertaining to the localization of superficial nodes, lymphatic function, and clearance can be acquired using deep-red/NIR fluorescence imaging strategies.

Larger-animal spontaneous melanoma models may more accurately reflect human disease and enable improved simulation of surgical procedures used in humans (i.e., SLN mapping). Locally injected ¹²⁴I-cRGDY-PEG-dot tracer and dynamic PET imaging enabled superior detection sensitivity and discrimination of metastatic tumor burden within hypermetabolic neck nodes compared with the PET imaging agent, ¹⁸F-FDG. ¹⁸F-FDG is an indicator of glucose metabolism, accumulating within metabolically active tumors that use this substrate. Traditionally used to stage clinical melanoma, it failed to accurately stage nodal disease in this representative miniswine. A number of well-known limitations are associated with whole-body ¹⁸F-FDG PET (44), particularly when imaging patients with early-stage melanoma, including a low mean sensitivity of 17.3% (0%–40%) for detecting SLNs and an inability to detect micrometastases less than about 1-cm i.d. In the present study, the ¹²⁴I-cRGDY-PEG-dot tracer was able to discriminate at least order of magnitude differences in metastatic tumor burden (i.e., 10- to 20-times greater) between the bilateral neck nodes, as determined by high-power microscopy, while ¹⁸F-FDG PET could not. Although metastatic disease was detected within the SLN, tumor was missed within the smaller (1.0 × 0.6 × 1.0 cm³) contralateral node.

Thus, for metastatic disease assessment, the use of dynamic PET imaging as a depth-sensitive, volumetric imaging tool, in conjunction with ¹²⁴I-cRGDY-PEG-dots, will offer distinct advantages in large-animal models in the intraoperative setting. Superior detection sensitivity, metabolic nodal status, and enhanced penetration for the mapping of deep-seated nodal and tissue activities may complement extended real-time fluorescence imaging assessments of metastatic tumor burden in the future. The use of PET will be critical for confirming the findings of depth-insensitive optical imaging tools, particularly in anatomic regions associated with unpredictable patterns of metastatic disease spread.

View Supplemental data

We acknowledge M. Gönen for providing assistance with biostatistical analyses and R. Toledo-Crow, S. Patel, and J. Lewis for technical assistance. This work was supported by an NIH-American Recovery and Reinvestment Act/Clinical and Translational Science Center grant to M. Bradbury and U. Wiesner. The work was further supported by the Cornell Nanobiotechnology Center, a Science and Technology Center program of the National Science Foundation under agreement no. ECS-9876771, and by the In vivo Cellular and Molecular Imaging Center P50 CA86438 grant. Technical services provided by the Memorial Sloan-Kettering Cancer Center Small-Animal Imaging Core Facility, supported in part by NIH Small-Animal Imaging Research Program (SAIRP) grant no. R24 CA83084 and NIH center grant no. P30 CA08748, are gratefully acknowledged.

Address correspondence to: Michelle S. Bradbury, Department of Radiology, Sloan-Kettering Institute for Cancer Research, 1275 York Ave., New York, New York 10065, USA. Phone: 212.639.8938; Fax: 212.794.4010; E-mail: bradburm@mskcc.org.

Conflict of interest: Hooisweng Ow and Ulrich Wiesner hold shares of the start-up company Hybrid Silica Technologies.

Reference information: J Clin Invest. 2011;121(7):2768–2780. doi:10.1172/JCI45600.

Erik Herz’s present address is: Exxon-Mobil Research and Engineering, Paulsboro, New Jersey, USA.

1. Weissleder R, Pittet MJ. Imaging in the era of molecular oncology. Nature. 2008;452(7187):580–589.
   View this article via: PubMed CrossRef Google Scholar
2. Weissleder R. Molecular imaging in cancer. Science. 2006;312(5777):1168–1171.
   View this article via: PubMed CrossRef Google Scholar
3. Manning HC, Lander A, McKinley E, Mutic NJ. Accelerating the development of novel molecular imaging probes: a role for high-throughput screening. J Nucl Med. 2008;49(9):1401–1404.
   View this article via: PubMed CrossRef Google Scholar
4. Fischer HC, Chan WC. Nanotoxicity: the growing need for in vivo study. Curr Opin Biotechnol. 2007;18(6):565–571.
   View this article via: PubMed CrossRef Google Scholar
5. Montet X, Funovics M, Montet-Abou K, Weissleder R, Josephson L. Multivalent effects of RGD peptides obtained by nanoparticle display. J Med Chem. 2006;49(20):6087–6093.
   View this article via: PubMed CrossRef Google Scholar
6. Montet X, Montet-Abou K, Reynolds F, Weissleder R, Josephson L. Nanoparticle imaging of integrins on tumor cells. Neoplasia. 2006;8(3):214–222.
   View this article via: PubMed CrossRef Google Scholar
7. Kelloff GJ, et al. The progress and promise of molecular imaging probes in oncologic drug development. Clin Cancer Res. 2005;11(22):7967–7985.
   View this article via: PubMed CrossRef Google Scholar
8. Heath JR, Davis ME. Nanotechnology and cancer. Annu Rev Med. 2008;59:251–265.
   View this article via: PubMed CrossRef Google Scholar
9. Hong H, Zhang Y, Sun J, Cai W. Molecular imaging and therapy of cancer with radiolabeled nanoparticles. Nano Today. 2009;4(5):399–413.
   View this article via: PubMed CrossRef Google Scholar
10. Burns AA, et al. Fluorescent silica nanoparticles with efficient urinary excretion for nanomedicine. Nano Lett. 2009;9(1):442–448.
    View this article via: PubMed CrossRef Google Scholar
11. Choi HS, et al. Design considerations for tumour-targeted nanoparticles. Nat Nanotechnol. 2010;5(1):42–47.
    View this article via: PubMed CrossRef Google Scholar
12. Alric C, et al. Gadolinium chelate coated gold nanoparticles as contrast agents for both X-ray computed tomography and magnetic resonance imaging. J Am Chem Soc. 2008;130(18):5908–5915.
    View this article via: PubMed CrossRef Google Scholar
13. Lucignani G. Nanoparticles for concurrent multimodality imaging and therapy: the dawn of new theragnostic synergies. Eur J Nucl Med Mol Imaging. 2009;36(5):869–874.
    View this article via: PubMed CrossRef Google Scholar
14. Beer AJ, Schwaiger M. Imaging of integrin alphavbeta3 expression. Cancer Metastasis Rev. 2008;27(4):631–644.
    View this article via: PubMed Google Scholar
15. Rhyner M, Smith A, Gao X, Mao H, Yang L, Nie S. Quantum dots and targeted nanoparticles probes for in vivo tumor imaging. In: Bulte J, Modo M, eds. Nanoparticles in Biomedical Imaging . New York, New York, USA: Springer; 2008:413–426.
16. Hauck TS, Anderson RE, Fischer HC, Newbigging S, Chan WC. In vivo quantum-dot toxicity assessment. Small. 2010;6(1):138–144.
    View this article via: PubMed CrossRef Google Scholar
17. Schipper ML, et al. Particle size, surface coating, and PEGylation influence the biodistribution of quantum dots in living mice. Small. 2009;5(1):126–134.
    View this article via: PubMed CrossRef Google Scholar
18. Zimmer JP, Kim SW, Ohnishi S, Tanaka E, Frangioni JV, Bawendi MG. Size series of small indium arsenide-zinc selenide core-shell nanocrystals and their application to in vivo imaging. J Am Chem Soc. 2006;128(8):2526–2527.
    View this article via: PubMed CrossRef Google Scholar
19. Aharoni A, Mokari T, Popov I, Banin U. Synthesis of InAs/CdSe/ZnSe core/shell1/shell2 structures with bright and stable near-infrared fluorescence. J Am Chem Soc. 2006;128(1):257–264.
    View this article via: PubMed CrossRef Google Scholar
20. Choi HS, et al. Renal clearance of quantum dots. Nat Biotechnol. 2007;25(10):1165–1170.
    View this article via: PubMed CrossRef Google Scholar
21. Ohnishi S, Lomnes SJ, Laurence RG, Gogbashian A, Mariani G, Frangioni JV. Organic alternatives to quantum dots for intraoperative near-infrared fluorescent sentinel lymph node mapping. Mol Imaging. 2005;4(3):172–181.
    View this article via: PubMed Google Scholar
22. Kobayashi H, Brechbiel MW. Nano-sized MRI contrast agents with dendrimer cores. Adv Drug Deliv Rev. 2005;57(15):2271–2286.
    View this article via: PubMed CrossRef Google Scholar
23. Hilderbrand SA, Weissleder R. Near-infrared fluorescence: application to in vivo molecular imaging. Curr Opin Chem Biol. 2010;14(1):71–79.
    View this article via: PubMed CrossRef Google Scholar
24. Burns A, Ow H, Wiesner U. Fluorescent core-shell silica nanoparticles: towards “Lab on a Particle” architectures for nanobiotechnology. Chem Soc Rev. 2006;35(11):1028–1042.
    View this article via: PubMed CrossRef Google Scholar
25. Choi J, et al. Core-shell silica nanoparticles as fluorescent labels for nanomedicine. J Biomed Opt. 2007;12(6):064007.
    View this article via: PubMed CrossRef Google Scholar
26. Kumar R, et al. In vivo biodistribution and clearance studies using multimodal organically modified silica nanoparticles. ACS Nano. 2010;4(2):699–708.
    View this article via: PubMed CrossRef Google Scholar
27. Ow H, Larson DR, Srivastava M, Baird BA, Webb WW, Wiesner U. Bright and stable core-shell fluorescent silica nanoparticles. Nano Lett. 2005;5(1):113–117.
    View this article via: PubMed CrossRef Google Scholar
28. Kim S, et al. Near-infrared fluorescent type II quantum dots for sentinel lymph node mapping. Nat Biotechnol. 2004;22(1):93–97.
    View this article via: PubMed CrossRef Google Scholar
29. Tanaka E, Choi HS, Fujii H, Bawendi MG, Frangioni JV. Image-guided oncologic surgery using invisible light: completed pre-clinical development for sentinel lymph node mapping. Ann Surg Oncol. 2006;13(12):1671–1681.
    View this article via: PubMed CrossRef Google Scholar
30. Kobayashi H, et al. Multimodal nanoprobes for radionuclide and five-color near-infrared optical lymphatic imaging. ACS Nano. 2007;1(4):258–264.
    View this article via: PubMed CrossRef Google Scholar
31. Felding-Habermann B, Mueller BM, Romerdahl CA, Cheresh DA. Involvement of integrin alpha V gene expression in human melanoma tumorigenicity. J Clin Invest. 1992;89(6):2018–2022.
    View this article via: JCI PubMed CrossRef Google Scholar
32. Ballou B, et al. Sentinel lymph node imaging using quantum dots in mouse tumor models. Bioconjug Chem. 2007;18(2):389–396.
    View this article via: PubMed CrossRef Google Scholar
33. Wang W, et al. Near-infrared optical imaging of integrin alphavbeta3 in human tumor xenografts. Mol Imaging. 2004;3(4):343–351.
    View this article via: PubMed CrossRef Google Scholar
34. Haubner R, Bruchertseifer F, Bock M, Kessler H, Schwaiger M, Wester HJ. Synthesis and biological evaluation of a (99m)Tc-labelled cyclic RGD peptide for imaging the alphavbeta3 expression. Nuklearmedizin. 2004;43(1):26–32.
    View this article via: PubMed Google Scholar
35. Larson DR, Ow H, Vishwasrao HD, Heikal AA, Wiesner U, Webb WW. Silica nanoparticle architecture determines radiative properties of encapsulated fluorophores. Chem Mater. 2008;20(8) ) :2677–2684.
    View this article via: CrossRef Google Scholar
36. Cressman S, Sun Y, Maxwell E, Fang N, Chen D, Cullis P. Binding and uptake of RGD-containing ligands to cellular α_υβ₃ integrins. Int J Pept Res Ther. 2009;15(1):49–59.
    View this article via: CrossRef Google Scholar
37. Kimura RH, Cheng Z, Gambhir SS, Cochran JR. Engineered knottin peptides: a new class of agents for imaging integrin expression in living subjects. Cancer Res. 2009;69(6):2435–2442.
    View this article via: PubMed CrossRef Google Scholar
38. Li ZB, et al. (64)Cu-labeled tetrameric and octameric RGD peptides for small-animal PET of tumor alpha(v)beta(3) integrin expression. J Nucl Med. 2007;48(7):1162–1171.
    View this article via: PubMed CrossRef Google Scholar
39. Drickamer L. Rates of urine excretion by house mouse (mus domesticus): Differences by age, sex, social status, and reproductive condition. J Chem Eco. 1995;21(10):1481–1493.
40. Nahrendorf M, et al. Hybrid PET-optical imaging using targeted probes. Proc Natl Acad Sci U S A. 2010;107(17):7910–7915.
    View this article via: PubMed CrossRef Google Scholar
41. Seymour LW. Passive tumor targeting of soluble macromolecules and drug conjugates. Crit Rev Ther Drug Carrier Syst. 1992;9(2):135–187.
    View this article via: PubMed Google Scholar
42. Haubner R, et al. Noninvasive imaging of alpha(v)beta3 integrin expression using 18F-labeled RGD-containing glycopeptide and positron emission tomography. Cancer Res. 2001;61(5):1781–1785.
    View this article via: PubMed Google Scholar
43. Poethko T, et al. Two-step methodology for high-yield routine radiohalogenation of peptides: (18)F-labeled RGD and octreotide analogs. J Nucl Med. 2004;45(5):892–902.
    View this article via: PubMed Google Scholar
44. Cheng Z, et al. Small-animal PET of melanocortin 1 receptor expression using a 18F-labeled alpha-melanocyte-stimulating hormone analog. J Nucl Med. 2007;48(6):987–994.
    View this article via: PubMed CrossRef Google Scholar
45. Bogush GH, Tracy MA, Zukoski Iv CF. Preparation of monodisperse silica particles: Control of size and mass fraction. J Non-Cryst Solids. 1988;104(1):95–106.
    View this article via: CrossRef Google Scholar
46. Sadasivan S, Dubey AK, Li Y, Rasmussen DH. Alcoholic solvent effect on silica synthesis—NMR and DLS investigation. J Sol-Gel Sci Technol. 1998;12(1):5–14.
47. Herz E, Burns A, Bonner D, Wiesner U. Large stokes-shift fluorescent silica nanoparticles with enhanced emission over free dye for single excitation multiplexing. Macromol Rapid Commun. 2009;30(22):1907–1910.
    View this article via: PubMed CrossRef Google Scholar
48. Hermanson G. Bioconjugate Techniques . 2nd ed. London, United Kingdom: Academic Press; 2008.
49. Yoon TJ, et al. Specific targeting, cell sorting, and bioimaging with smart magnetic silica core-shell nanomaterials. Small. 2006;2(2):209–215.
    View this article via: PubMed CrossRef Google Scholar
50. Piatyszek MA, Jarmolowski A, Augustyniak J. Iodo-Gen-mediated radioiodination of nucleic acids. Anal Biochem. 1988;172(2):356–359.
    View this article via: PubMed CrossRef Google Scholar
51. Eckerman K, Endo A. MIRD: Radionuclide Data and Decay Schemes . 2nd ed. Reston, Virginia, USA: Society of Nuclear Medicine; 2008.
52. Cristy M, Eckerman K. Specific absorbed fractions of energy at various ages from internal photon sources (I-VII). In: Oak Ridge National Laboratory Report . Springfield, Virginia, USA: National Technical Information Service, Department of Commerce; 1987.
53. Loevinger R, Budinger T, Watson E. MIRD Primer for Absorbed Dose Calculations . Revised ed. New York, New York, USA: Society of Nuclear Medicine; 1991.
54. Stabin MG, Sparks RB, Crowe E. OLINDA/EXM: the second-generation personal computer software for internal dose assessment in nuclear medicine. J Nucl Med. 2005;46(6):1023–1027.
    View this article via: PubMed Google Scholar