Nodal staging plays a critical role in the evaluation of disease progression and design of therapeutic strategies in the treatment of most cancer patients. For example, in breast cancer, intradermal (i.d.) or subcutaneous (s.c.) administration of a nonspecific -radiocolloid is used to identify the tumor draining or sentinel lymph node for resection. Once the surgical pathologist confirms the presence of cancer cells in the resected sentinel lymph node, additional nodes around the axilla are removed to better assess the extent of lymph node involvement. The accuracy of nodal staging is critical for judicious selection of effective therapy. The inability to surgically resect inaccessible tumor-draining lymph nodes, such as the internal mammary or supraclavicular lymph nodes in breast cancer patients, or the pelvic lymph nodes that drain the prostate in prostate cancer patents, can compromise oncologists’ ability to accurately stage and select proper therapy. Noninvasive, diagnostic imaging of cancer-positive lymph nodes could positively impact the manner in which tumor-node-metastasis (TNM) staging is performed.
Molecular imaging approaches for early-stage diagnosis of metastasis in the lymphatics were first explored by Weinstein, who used radio-labeled monoclonal antibodies that were administered (i.d.) through the footpad in preclinical studies.1, 2 Since then, several investigators have used immunolymphoscintigraphy as a tool to study lymphatic delivery of agents. Steller 3 demonstrated the dose dependence of iodinated monoclonal antibody biodistribution following i.d. administration in a murine animal model, while Wahl 4 employed iodinated intact and fragmented antibodies to assess the kinetics of their clearance from the lymphatic system in response to ambulation. Additional reports have used immunolymphoscintigraphy to study lung cancer metastases to pulmonary and mediastinal lymph nodes in dogs.5 The i.d. route of imaging agent delivery may be extremely favorable for nodal staging because antibodies and small molecules drain primarily from the lymph plexus into the lymphatic system. Small molecules pass readily through one or more nodes before exiting the lymphatic circulation via the blood circulatory system, while larger agents may be retained longer within lymph nodes. Some advantages for using a direct lymphatic approach include: (1) high lymph node uptake of antibodies, which enables significantly lower doses of labeled antibodies than that required for intravenous (i.v.) administration, (2) more rapid lymph node uptake, enabling minimal time between i.d. antibody administration and imaging, and (3) increased target-to-background ratios due to reduced nonspecific binding, which would otherwise occur in systemic i.v. administration. A more comprehensive review of lymphatic uptake and transport of proteins following intralymphatic administration can be found in an article by Porter and Charman.6
In addition to animal studies, imaging-based clinical applications of intralymphatic administration have also been explored by several investigators, mostly using radio-labeled monoclonal antibodies in their intact or fragmented form to detect tumor-draining lymph nodes in a spectrum of cancer patients ranging from melanoma, T-cell lymphoma, breast, and prostate cancers. One of the first investigations of lymph node imaging was reported by Order in 1975, in which they demonstrated accumulation of -labeled antiferrin immunoglobulin in lymph nodes of patients with breast carcinoma and lymphoma.7
Following that, DeLand conjugated -labeled antibodies to carcinoembryonic antigen (CEA) and showed high sensitivity in detecting nodal metastases after interdigital administration in patients.8 In breast cancer and T-cell lymphoma, immunolymphoscinitigraphy was performed by several investigators with -, -, and -labeled monoclonal antibodies to noninvasively assess regional lymph nodes.9, 10, 11, 12 Pelvic nodal metastases in prostate cancer have been investigated13 by bipedal intralymphatic administration of -PAY 276, while antimelanoma antibodies that have been radio-labeled for lymphoscintigraphy have also been reported in the literature.14, 15, 16 Most injections were well tolerated, but some patients have reported pain at the injection site. Most studies have reported a high sensitivity (80 to 100%) using this route of administration, but nonspecific uptake by normal lymph nodes continues to be problematic.
Although lymphoscintigraphy, or the administration of nonspecific sulfur colloid, is the “gold standard” for clinically evaluating lymph mapping, some limitations suffered by this imaging technique include the need for long gamma camera integration times, poor temporal and spatial resolution, low signal-to-noise ratios (SNRs), and the finite physical half-life of the tracer . In contrast, near-infrared (NIR) fluorophores provide an attractive opportunity that outweighs some of these disadvantages inherent in nuclear techniques. Since fluorescent reporters do not have a physical half-life, they can be repeatedly activated using appropriate excitation light. Theoretically, for a fluorophore with a nanosecond fluorescent lifetime and a quantum efficiency of 0.1, photons can be generated per second per molecule. When compared to the one-photon imaging event resulting from the radioactive decay of a radionuclide, the potential advantages of NIR fluorescent optical imaging with NIR fluorophores becomes evident. The increased photon yield can considerably improve signal-to-noise and significantly reduce image acquisition time for molecular imaging, if tissue attenuation does not significantly reduce the number of detected fluorescent photons. While NIR fluorophores also have an absorption cross section, their administration in trace doses does not provide sufficient absorption contrast for in vivo. Hence, NIR contrast may be best provided from the collection of NIR fluorescent photons rather than from the attenuation of excitation photons, which may occur from endogenous chromophores as well as from the exogenous fluorophores.
Herein, we present our work on a dual-labeled conjugate — -trastuzumab- — which targets the human epidermal growth factor receptor-2 (HER2). Overexpression of HER2 is associated with poor prognosis in about 20 to 30% of breast cancer patients.17 We have previously shown specificity in vivo in subcutaneous xenograft models.18 In this report, we first review recent results from our group describing lymphatic imaging with nonspecific optical agents. Next, we review our prior results showing tumor targeting of dual-labeled -trastuzumab- and enhanced performance of optical over nuclear imaging. We additionally show trafficking of the labeled antibody into the lymphatics and its subsequent drainage into lymph nodes using NIR optical imaging.
Real-Time Lymphatic Imaging with Near-Infrared Fluorophores
Recently, Sharma illustrated the ability to quantitatively image lymph function in a swine model after administering an i.d. injection of a nonspecific NIR fluorophore, IC-Green.19 Their work was the first to noninvasively image pulsatile lymph flow velocities of , averaging approximately along lymph channels ranging in length. A similar approach was employed by Sevick-Muraca, who used i.d. microdoses of IC-Green to image lymph propulsion in the breast and axilla of breast cancer patients undergoing sentinel lymph node mapping.20 They observed mean velocities ranging between 0.08 and . The image acquisition time for optical imaging in both swine and humans ranged from . Upon further reducing the integration time to , Kwon and Sevick-Muraca have been able to rapidly image lymph trafficking in the limbs and tail of mice.21 The average lymph flow velocity ranged from .
Figure 1 depicts representative optical images showing “packets” of IC-Green solution traversing the lymphatic channels after i.d. administration in human (b), swine (c), and mouse (d).19, 20, 21 The stick-diagram with a white light image (a) illustrates the right breast view of a breast cancer patient with four covered i.d. injection sites. The corresponding optical image from this patient (b) reveals two lymph channels and a region of lymph “pooling” through which IC-packets were propelled into the axilla. The swine model shows lymph propulsion from the hind limb up to the middle iliac node, while the mouse model depicts IC-Green trafficking to the axillary lymph node after i.d. administration into the footpad of the forepaw. These studies demonstrate the feasibility of noninvasive NIR optical imaging to qualitatively and quantitatively contribute toward the standard of care in nodal staging in preclinical and clinical studies. Most importantly, the humanstudies confirm the ability to use microdose administration (i.e., between 10 and 100 micrograms) of IC-Green, suggesting that the instrumentation is capable of assessing NIR fluorescent signals for molecular imaging after trace administration of molecular imaging conjugate.
Other investigators have also conducted NIR fluorescence imaging of IC-Green. Frangioni and coworkers have intraoperatively imaged the fluorescence from IC-Green and other NIR fluorescent conjugates administered intradermally to intraoperatively identify sentinel lymph nodes in animal models.22, 23, 24 Yodh and coworkers25 also measured fluorescence in humans after intravenous administration of . In their work, as well as in the work of others described earlier, NIR fluorescence was measured using an integrating CCD camera but was coupled with time-resolved absorbance measurements made in the frequency domain in order to perform fluorescence optical tomography from the fluorescence CW measurements. To date, there have been no reports of fluorescence enhanced optical tomography using PMT-based, time-dependent measurements of fluorescence as reported in the companion review paper by Sevick-Muraca and Rasmussen.26 Fluorescence tomography using gain modulation of an intensified CCD camera has been demonstrated in phantoms by Godavarty, 27, 28 Roy, 29 and Joshi, 30 however, using mM amounts of IC-Green. Before fluorescence tomography measurements can be conducted using microdoses in clinical relevant tissue volumes, two requirements must be met: the “noise floor” of fluorescence measurements must be reduced,26 and a molecularly targeting NIR optical imaging agent specific to disease must be used. Herein, we review and present new work on a dual-labeled imaging agent for planar NIR fluorescence imaging validated with nuclear imaging techniques.
Dual-Labeled Molecular Imaging Agents
Imaging agents that are designed with both nuclear and optical tracers combine complementary advantages from each modality and enhance the potential for detection of disease. Dual-labeled agents provide us with a unique opportunity to cross-validate NIR and nuclear imaging modalities. Some peptide-based dual-labeled imaging agents developed include a cyclopentapeptide, c(KRGDf), which is known to target alpha(v)beta3 integrins,31, 32 and a cyclic peptide, c(CGRRAGGSC), which is known to target IL-11 Ref. 33. Quantitative comparison between planar scintigraphy and optical imaging in in vivo xenograft models injected with the RGD peptide have revealed comparable sensitivities between the two modalities, but NIR fluorescence shows improved signal-to-noise ratios.32 Herein, we employ the dual-labeling strategy to image antibodies in the lymphatics.
Labeled Trastuzumab for Metastasis and Nodal Staging
The human epidermal growth factor receptor (HER) family consists of transmembrane receptor tyrosine kinases that regulate complex intracellular signaling networks to control normal and pathological cellular growth, differentiation, and survival.34 The second member, HER2, is of particular importance in breast cancer because overexpression or gene amplification of HER2 is closely associated with aggressive tumor progression and poor prognosis. Since a significant percentage of breast cancer patients are affected by the aberrant HER2 signaling, this family of receptors represents an excellent target for therapeutics. Additionally, the presence of HER2 on the plasma membrane of cells has also made it a potential target for diagnostic biomarkers. Trastuzumab (Herceptin, Genentech, San Francisco) is a humanized anti-HER2 antibody35 that interferes with HER2 signaling and is approved clinically as a therapeutic for breast cancer.36, 37 Evidence also suggests that with metastasis, HER2 is conserved or elevated with respect to that expressed on the primary tumors.38, 39
Several investigators have labeled intact, derived, and fragments [(Fab, )] of trastuzumab antibody with a number of radioactive nuclear agents such as Ga-68 (Ref. 40), Y-90 (Refs. 41, 42, 43), In-111 (Refs. 41, 42, 43), and Tc-99m (Ref. 44) to image HER2 overexpressing breast cancer xenografts, as shown in Table 1 . Smith-Jones 40 showed that sequential positron emission tomography (PET) imaging using -labeled fragments of trastuzumab can be used to quantify the loss and recovery of HER2 induced by HSP90 inhibitors in animals bearing BT-474 human breast tumors, while Tang 43, 44 have developed In-111 and Tc-99m based trastuzumab Fab conjugates to detect xenografts using whole-body scintigraphy. Huh 46 and Artemov 47 have conjugated nanocrystals and iron-oxide-based nanoparticles to detect HER2 expression with magnetic resonance imaging (MRI). Optical imaging–based strategies include intact trastuzumab labeled with a red fluorescent dye to image HER2-overexpressing human breast cancer xenografts, as shown by Hilger, 49 while recently Koyama applied spectral fluorescence imaging methods with rhodamine-green conjugated trastuzumab to visualize pulmonary metastases.50 Tada have used a confocal scanner unit coupled with an EMCCD camera to track single quantum dot particles, which have been conjugated to trastuzumab, from a capillary vessel to a cancer cell in a live animal.51
Summary of molecular imaging studies using trastuzumab.
|Imaging Modality||Imaging Agent||In vitro/In vivo:Cell Lines Used||Dose||Route||Ref.|
|Position emission tomography (PET)||-DOTA-Herceptin||In vivo (mice): BT-474MCF-7, MDA-MB-468(s.c.)||i.v.||40|
|Single photon emission computed tomography/computed tomography (SPECT/CT)||-Fab4D5, -AB.Fab4D5||In vivo (mice): tumorcells derived fromMMTV/HER2 transgenicmice (s.c.)||(300 to 500μCi)||i.v.||42|
|Scintigraphy(whole body)||[99mTc]-HYNIC-TranstuzumabFab, -trastuzumab Fab, -trastuzumab,||In vivo (mice): BT/474,SK-OV-3 (s.c.) In vivo (mice, humans)||i.v.||43, 44, 45|
|Mice: ; Human: 100 to||i.v.|
|Scintillationcounter(tissue/cells)||-Herceptin, , -Herceptin, -Herceptin||In vivo (mice): BT474(s.c.)||i.v.||40|
|In vitro: MCF-7, SKBr-3||41|
|Magnetic resonance tomography (MRT)||WSIO-HerceptinBiotinylated Herceptin withStreptavidin-SPIO mocrobeads||In vitro NIC3T6.7(s.c.) In vitro: MCF-7, MDA-MB-231, AU-565 (gel phantoms)||of Fe||i.v.||46, 47|
|Optoacoustic tomography (OT)||Herceptin conjugated withgold nanoparticles (MabNP)||In vivo: SK-BR-3, L6 ratmyoblasts||48|
|Fluorescenceimaging||Cy5.5-;labeled Herceptin||In vivo (mice): SK-BR-3,PE/CA-PJ34 (s.c)||i.v.||49|
|Herceptin-RhodG||In vivo (mice):3T3/HER2+,Balb/3T3/HER2-||in PBS||i.v.||50|
|Fluorescencequantum dots||Trastuzumab-Qdots||In vivo (mice), ex vivo (tumor frommice): KPL-4, MBA-MB-231 (s.c.)||, tissuesections stained with||i.v.||51, 52|
The first use of trastuzumab as a diagnostic imaging agent was reported by Behr 53 This was followed by a study by Perik In which In-111-labeled trastuzumab was clinically translated for scintigraphy and systemic administration in patients to help identify late-stage metastases with SPECT/CT imaging that was not previously identified.54
Dual-Labeled Trastuzumab for Cancer Imaging
Recently, we have synthesized a dual-labeled antibody — -trastuzumab- — and demonstrated imaging specificity for HER2 in vitro and in vivo. The reader is referred to Ref. 18 for details of synthesis and imaging. Figure 2 represents a schematic of the imaging agent. Since DTPA dianhyride and the NIR dye, IRDye 800CW, bind at the lysine residues of the antibody, we see a range in the ratios of binding between each of these molecules with trastuzumab, with and typically ranging between 7 and 10. The agent is stable at for extended periods of time, but DTPA dianhydride shows serum-based instability with degradation occurring at the rate of 6% per day. As an alternative, we have since employed more stable forms of DTPA, such as p-SCN-Bz-DTPA, for chelation of In-111, a gamma emitter with a half life of .
The binding of -trastuzumab- to a single SKBR3 cell that overexpresses HER2 is shown in Fig. 3 . Sytox green stains the nucleus, while the fluorescent signal from IRDye 800CW indicating local expression of HER2 is present only along the outer rim of the cell, demonstrating extracellular binding. In Fig. 3 the nuclear stain is pseudo-colored green and -trastuzumab- binding is represented in red. Prior incubation with unlabeled trastuzumab obliterates -trastuzumab- cell binding, indicating the molecular specificity of the agent.18 Figure 3 also illustrates the usefulness of optical labeling in molecular imaging, as in vitro specificity can be readily visualized from microscopy, an advantage not present in nuclear imaging techniques.
Both nuclear and planar NIR fluorescence imaging was used to evaluate the in vivo specificity of -trastuzumab- . Figure 4 shows a series of whole-body fluorescence (a and b), planar scintigraphy (c), and SPECT/CT (d) images from athymic nude mice with HER2-overexpressing SKBr3 human breast cancer xenografts after i.v. injection of -trastuzumab- . In vivo fluorescence imaging was accomplished by illuminating the animal with light from a laser diode ( and for light; DL7140-201; Sanyo) that expanded to a circular area approximately in diameter. The reemitted fluorescent light was collected with an EMCCD camera in CW mode. Filter sets used in this study included a bandpass filter with an center wavelength and an optical density of greater than 3 at (ANDV8483; Andover Corp.) for collecting IRDye 800CW fluorescence and a holographic filter with an optical density of greater than 6 (HSP 785.0; Kaiser Optical Systems) for rejecting backscattered and reflected excitation light. Image acquisition was accomplished with V11 software to collect raw fluorescence images, and with these images, data processing and analysis were accomplished with ImageJ software. The integration time for white-light and fluorescence images was . SPECT and CT images were acquired consecutively with a MicroCAT II scanner (Siemens Medical Solutions). The SPECT scan was acquired for 20 projections over for a scan time of per frame. SPECT and CT tomographic images were coegistered by geometric transformation and rendered to make the fused images with Amira (version 3.1; Konrad-Zuse-Zentrum fur Informationstechnik). Planar scintigraphy was performed by removing the pinhole collimator of the SPECT scanner, adding a planar collimator, and integrating for .
The fluorescence and nuclear signal obtained at the tumor region appear significantly higher than the rest of the body. Comparison of tumor uptake between mice after administration of dye alone and dual-labeled trastuzumab revealed significant accumulation of the dual-labeled agent in the tumor, as opposed to dye alone (p-value: 0.0011). Antibody interaction with the hepatocytes (which represent routes for degradation and clearance) increased agent accumulation in the liver and is expected with i.v. administration of an antibody-based imaging agent. Liver uptake is observed in both imaging modalities, consistent with the integrity of the dual-labeled agent.
The target-to-muscle ratios (TMRs) calculated from the planar fluorescence and planar scintigraphy images according to the approach described by Houston 32 are found to be comparable at 2.11 and 2.43, respectively, and consistent with the dual labeling on a single imaging agent. The signal-to-noises ratio (SNRs) calculated from the planar fluorescence and planar scintigraphy images are found to be dramatically different at 16.82 and 4.22, respectively. The differences can be attributed to the increased photon count rates of fluorescence imaging when compared to scintigraphy. Upon dissection of major organs and evaluation of gamma and fluorescent photon counts, the TMRs were calculated for both modalities. When optical TMR is normalized by area (or pixels) and nuclear TMR by tissue weight (in grams), no significant change is observed within each modality over time, but the signal obtained from optical imaging is statistically higher compared to that from nuclear imaging. In contrast, when both TMRs are calculated based on tissue weight, the statistical significance disappears, as represented by Fig. 5 , consistent with the origin of both nuclear and fluorescent signals from a single imaging agent. In addition, the standard error observed in optical imaging is smaller than nuclear imaging, consistent with the reduced SNR also observed in the case of the cyclic RGD peptide.32
Further, we assessed the potential of antibody-based optical imaging for lymph node imaging using i.d. administration. Upon injecting (5.4 pmole) of a solution into the footpad of athymic nude mice, we were able to image trafficking of -trastuzumab- into the axillary nodes and popliteal nodes [Figs. 6a and 6b , respectively]. A closer look at the popliteal lymph nodes, revealed lymphatic channels that picked up the agent from the injection site and drained it into the lymph node as indicated by the arrows in Fig. 6c. Both images were obtained within of injection. It is of interest to note that the volume injected in our preclinical study is considerably less when compared to the injections as reported with nuclear imaging.2, 3
Figure 7b shows a sequence of optical images taken after an i.d. injection of -trastuzumab- ( , 5.4 pmole) into the footpad of an athymic mouse. The lymph node targeted was the axillary node. The corresponding sample white light image is shown in Fig. 7a. The accumulation and clearance of the antibody-based imaging agent in the lymph node is slower than that observed by IC-Green or IRDye 800CW. IRDye 800CW has a molecular weight of 1166 Da, which is significantly smaller than antibodies (MW: 185kDa), enabling it to clear from the lymph nodes within a few hours. Antibodies requires a few days (data not shown), which can be a problem for radioimmunoscintigraphy of cancer-positive lymph nodes. Figure 7c shows the rate of clearance of -trastuzumab- averaged over time in five animals. We found variability in the uptake of the imaging agent into the axillary node among the mice, which is represented by a large error bar at [Fig. 7c], but all mice attained maximum fluorescence in nodes within . The half-life of clearance was between , which is consistent with previous reports using radio-labeled antibodies. As seen in the i.v. administration of -trastuzumab- , we also observed a fluorescence signal emanating from the liver region after an i.d. injection (albeit at later times), which is consistent with antibody degradation and clearance.
This study demonstrates the potential of using NIR fluorescence imaging for impacting tumor nodal staging. By enhancing the accuracy of nodal staging with molecular-specific imaging agents, we can offer novel approaches to diagnose sites that are often harder to access for ex vivo pathology, e.g., internal mammary nodes and supraclavicular nodes in breast and pelvic lymph nodes in prostate cancer patients. By injecting the agent directly into lymphatics through i.d. or s.c. routes, we can considerably reduce the dose administered and the amount of nonspecific binding in other organs that is characteristic of an i.v. route of administration. Since antibodies are cleared through the liver and kidneys, which are major organs surrounding the lymph nodes draining the prostate, i.d. delivery is especially advantageous in assessing metastases in prostate cancer.
Despite nuclear imaging being the clinical “gold standard” of molecular imaging, optical imaging provides superior signal-to-noise ratios with low camera integration times due to high photon count rate. These advantages have recently provided investigators with an opportunity to analyze lymph function and quantify pulsatility in humans, swine, and mice.19, 20, 21 Furthermore, they have enabled the synthesis of novel optical-based targeting agents to address other lymph-based disorders, as demonstrated by HA-NIR, a molecular target of the lymph vascular endothelial receptor (LYVE-1).19 In the present study, we have shown the feasibility of performing optical imaging in detecting lymph nodes in a preclinical animal model using molecular targets such as antibodies. -trastuzumab- is a dual-labeled antibody that exhibits high molecular specificity to HER2. We have demonstrated that picomole amounts are sufficient to visualize the axillary and popliteal lymph nodes in mice (Fig. 6). We observed that the fluorescent intensity due to dual-labeled -trastuzumab- reduced to half of its highest value at after administration [Fig. 7c]. But further studies addressing questions of specificity and sensitivity will have to be conducted to prove efficacy for cancer detection. Nonetheless, the combination of NIR and nuclear imaging of lymph node metastases in cancer patients with dual-labeled molecular agents that are administered through i.d. or s.c. routes shows promise for improving the accuracy of tumor node metastasis staging.
This work was supported by the National Institutes of Health, Grant Nos. P50 CA58183, R01 EB003132, and R01 CA112679.