Detection of Lymph Node Involvement in Hematologic Malignancies Using Micromagnetic Resonance Lymphangiography with a Gadolinum-Labeled Dendrimer Nanoparticle

Abstract

Animal models of lymphoma should reflect their counterparts in humans; however, it can be difficult to ascertain whether an induced disease is intralymphatic or extralymphatic based on direct visualization. Current imaging methods are insufficient for identifying lymphatic and intralymphatic involvement. To differentiate intralymphatic from extralymphatic involvement, we have developed a magnetic resonance imaging–based lymphangiography method and tested it on two animal models of lymphoma. A gadolinium (Gd)–labeled dendrimer nanoparticle (generation-6; ∼220 kDa/∼10 nm) was injected interstitially into mice bearing hematologic malignancies to perform dynamic micromagnetic resonance lymphangiography (micro-MRL). Both a standard T1-weighted 3D fast spoiled gradient echo and a T2/T1–weighted 3D fast imaging employing steady-state acquisition (3D-FIESTA-C) were compared in an imaging study to differentiate intralymphatic from extralymphatic involvement of tumors. The lymphatics and lymph nodes were visualized with both methods in all cases. In addition, 3D-FIESTA-C depicted both the lymphatic system and the extralymphatic tumor. In an animal model, 3D-FIESTA-C demonstrated that the bulk of the tumor thought to be intralymphatic was actually extralymphatic. In conclusion, micro-MRL, using Gd-labeled dendrimer nanoparticles with the combined method, can define both the normal and abnormal lymphatics and can distinguish intralymphatic from extralymphatic diseases in mouse models of malignant lymphoma.

Introduction

Malignant lymphoma is the fifth leading cause of cancer deaths in the United States [1]. Extralymphatic disease, which is reportedly an important negative prognostic indicator, may alter the regimen of chemotherapy [2,3]. Animal models are useful in evaluating new drug and radiation therapies; however, their validity depends, in part, on the fidelity with which they reproduce the characteristics of human lymphomas. It can be difficult to distinguish nodal tissue from tumor tissue on histology, and a disease in the vicinity of lymph nodes is often assumed to be intralymphatic. Conventional imaging methods cannot differentiate between intralymphatic and extralymphatic diseases because they cannot define the normal lymphatic system. No imaging method is currently available to reliably distinguish if extralymphatic lymphoma in living mice is in the lymphatic or in the perilymphatic space. This limitation arises because normal lymphatic tissue cannot be distinguished from abnormal lymphatic tissue. As a result, some animal models of lymphoma cannot be validated with regard to the presence or absence of extralymphatic diseases. This raises questions about their ability to predict tumor behavior in humans.

We have recently developed a micromagnetic resonance lymphangiography (micro-MRL) method, which employs interstitial injection of a nanoparticulate magnetic resonance imaging (MRI) paramagnetic contrast agent, to visualize the anatomy and physiology of most deep lymphatic vessels and lymph nodes [4]. This agent is passively but efficiently taken up by the lymphatics and readily opacifies draining lymph nodes. Standard T1-weighted MRI scans clearly demonstrate both the lymphatic channels and the lymph nodes, and may be useful for sentinel node imaging with interstitial injection around the primary cancer [5]. Micro-MRL could also be useful in characterizing animal models of lymphoma. In contrast, T2-weighted images, which can be used for whole-body tumor screening as well as for positron emission tomography scan [6–13], are useful in depicting the full extent of diseases, including extralymphatic involvement. Thus, it would be useful to design a single sequence that simultaneously displays the lymphatic system and extralymphatic diseases to fully evaluate the extent of lymphoma.

In the late 1990s, a pulse sequence called True FISP, which took advantage of faster gradient switching and higher receiver bandwidth to create a fast scan with signal dependence based on tissue T2/T1 ratio, was developed. As a result, high-signal structures, such as opacified lymphatics, yield signals because the shortening of T2 due to the paramagnetic effects of a gadolinium (Gd)–based contrast agent is less than the shortening of T1 due to those same agents [14–16]. The nonenhancing tumor is also visualized because the T2 of tumors is usually prolonged compared to that of normal tissues. Thus, both opacified lymphatics and tumors are visualized with the contrast-enhanced True FISP method. The first version of this 3D steady-state coherent imaging pulse sequence (3D-FIESTA) became available on clinical MRI instruments in 2001 [14–16]. However, images acquired with 3D-FIESTA suffered from strong banding artifacts, especially at the boundaries of air and water. This artifact was especially noticeable at the surface of the body and at the spaces adjacent to the gastrointestinal tract [17]. The new version, 3D fast imaging employing steady-state acquisition (3D-FIESTA-C), which was introduced in 2003, greatly reduced the banding artifacts. Because the image contrast on 3D-FIESTA-C depends on the ratio of the T2/T1 signals, this sequence has the potential to depict both tumor tissue and paramagnetic contrast agent with better sensitivity than T1-weighted sequences alone [17]. Therefore, micro-MRL using 3D-FIESTA-C is potentially able to define the 3D topologic relationship between the lymphatic system and the tumors.

In this study, we performed micro-MRL with a nanosize MRI contrast agent, generation-6 (G6), and assessed T2/T1–weighted 3D-FIESTA-C compared with standard T1-weighted 3D fast spoiled gradient echo (3D-fSPGR) MRI in normal mice. Then, to establish if this combined method might be useful in characterizing and monitoring mouse models of human lymphoma, we used two mouse models of hematologic malignancies with intralymphatic and extralymphatic lesions.

Materials and Methods

Preparation of Paramagnetic Contrast Agent

The G6 polyamidoamine (PAMAM-G6) dendrimer (Dendritech, Inc., Midland, MI) has an ethylenediamine core, 256 terminal-reactive amino groups, and a molecular mass of 58,048 Da. The PAMAM-G6 dendrimer was concentrated to ∼5 mg/ml and diafiltrated against 0.1 M phosphate buffer at pH 9. The PAMAM-G6 dendrimer was reacted with a 256-fold molar excess of 2-(p-isothiocyanatobenzyl)-6-methyl-diethylenetriaminepentaacetic acid (1B4M) [18] at 40°C and maintained at pH 9 with 1 M NaOH for 24 hours. An additional 256-fold molar excess of 1B4M was added as a solid after 24 hours. Resulting preparations were purified by diafiltration using Centricon 30 (Amicon Co., Beverly, MA). Subsequently, PAMAM-G6 dendrimer–1B4M conjugate (∼3 mg containing 4 μmol of 1B4M) was mixed with 8 μmol of Gd(III) citrate in 0.3 M citrate buffer overnight at 40°C. Excess Gd(III) was removed by diafiltration using a Centricon 30 filter (Amicon Co.) while simultaneously changing the buffer to 0.05 M PBS. The purified sample (G6) was diluted to 0.2 ml with 0.05 M PBS, and ∼5 μl was used in each mouse middle phalange. Quality control was performed as previously described [19].

Mouse Models

All mouse studies were approved by the Animal Care and Use Committee of the National Institutes of Health. Ten-week-old athymic nu/nu mice served as controls (n = 7). Animal disease models employed were athymic nu/nu mice bearing a PT-18 xenograft [20], a mast cell lymph node metastasis model (n = 8), and SCID/NOD mice (n = 4; National Cancer Institute, Frederick, MD) with Karpas 299 anaplastic large cell lymphoma.

PT-18 cells (10⁷) were injected into the left mammary pad of athymic nu/nu mice, and 8 of 15 mice developed tumor masses in the left axillary lymph nodes as well as in the left mammary gland within 3 weeks.

A lymphoma model of Karpas 299 [21] was created by a tail vein injection of 2 × 10⁶ Karpas 299 cells in SCID/NOD mice. The mice developed one to four extralymphatic soft tissue tumors by 3 to 5 weeks after the injection of Karpas 299 tumor cells.

Dynamic 3D-Micro-MRL

Mice were anesthetized with an intraperitoneal injection of 1.15 mg of sodium pentobarbital (Dainabot, Osaka, Japan) and then injected with 0.1 μmol Gd/5 μl G6 contrast agent directly into the midphalanges of all four extremities, for a total of four injections. All dynamic micro-MRL images were obtained using a 1.5-T superconductive magnet (Signa LX; General Electric Medical System, Waukesha, WI) with a 1-in. round surface coil (birdcage type) fixed to a custom-constructed coil holder. The mice were wrapped with gauze to maintain normal body temperature and were placed at the center of the coils. A 3D-fSPGR (repetition time/echo time = 14.3/7.0 milliseconds; bandwidth = 31.2 kHz; flip angle = 30°; four excitations; 36 slice-encoding steps; scan time = 4 minutes, 23 seconds) with chemical fat suppression was acquired at 10, 20, 30, and 40 minutes postinjection of the contrast agent. A 3D-FIESTA-C (Signa LX; General Electric Medical System) (repetition time/echo time = 9.1/2.0 milliseconds; bandwidth = 41.7 kHz; flip angle = 45°; two numbers of excitation; scan time = 2 minutes, 46 seconds) was acquired 15, 25, 35, and 45 minutes postinjection of the contrast agent. The coronal images were reconstructed with 0.6 mm of section thickness and 0.3 mm of overlap (two 512 matrix zips). The field of view was 8 × 4 cm. The in-plane matrix was 512 × 256 for 3D-fSPGR and was 384 × 256 for 3D-FIESTA-C. The slice data were processed into 3D images using the maximum intensity projection method (Advantage Windows; General Electric Medical System). The image resolution was 156 × 156 × 600 μm for 3D-fSPGR and was 208 × 312 × 600 μm for 3D-FIESTA-C. After imaging, the mice were sacrificed by CO₂ inhalation and then dissected to obtain histologic specimens.

To directly compare 3D-fSPGR (T1-weighted) and 3D-FIESTA-C (T2/T1–weighted) for visualization of the lymphatic drainage, serial dynamic micro-MR lymphangiograms of normal athymic nu/nu mice (n = 7) were obtained with 3D-fSPGR and 3D-FIESTA-C after injection of the contrast agent. Images of bilateral axillary and lateral thoracic lymph nodes and bilateral lymphatic vessels were independently examined and rated (0–2) by two board-certified radiologists using the following scoring system: 0 = invisible; 1 = partially visible; and 2 = completely visible. Any discrepancies between the two reviewers were resolved by consensus.

To evaluate the topologic relationship between hematologic tumors and the lymphatic system, serial dynamic micro-MR lymphangiograms of the PT-18 xenograft/lymph node metastasis model (n = 8) and the systemic Karpas lymphoma model (n = 4) were acquired with 3D-fSPGR and 3D-FIESTA-C after injection of the contrast agent, as described above.

Histologic Analysis

After finishing the micro-MRL study, tumors and lymph nodes around bilateral axillary and lateral thoracic regions were dissected. Tumors in the neck, thorax, and axilla, and axillary or lateral thoracic lymph nodes were removed, fixed in 10% formalin, and stained by hematoxylin–eosin (H–E) to correlate histology with micro-MRL findings.

Statistical Analysis

A Kruskal-Wallis test with Bonferroni-Dunn correction was used for the analysis of the visualization of lymph nodes and lymphatic vessels. All tests were two-sided, and P < .005 was considered significant after the Bonferroni-Dunn correction.

Results

3D-FIESTA-C Is Superior to 3D-fSPGR in Identifying Lymphatics

The 3D-FIESTA-C and 3D-fSPGR methods were compared in normal control mice. Four small lymph nodes, bilateral axillary and lateral thoracic, were visualized equally well with 3D-fSPGR and 3D-FIESTA-C (Figure 1). However, based on the ratings of the observers, the lymphatic vessels connecting the lymph nodes were significantly better visualized with 3D-FIESTA-C than with 3D-fSPGR at 10 and 40 minutes postinjection (P < .005) (Table 1). The thin lymphatic ducts were especially well depicted with 3D-FIESTA-C compared with 3D-fSPGR. Thus, the 3D-FIESTA-C method was more sensitive than 3D-fSPGR in visualizing the lymphatic system.

Figure 1.

Micro-MRL obtained with 3D-FIESTA-C demonstrates better visualization of the normal lymphatic flow than micro-MRL acquired with 3D-fSPGR. MRL images of the upper body of a mouse with 3D-fSPGR (a; animation 1) and 3D-FIESTA-C (b; animation 2), which were serially obtained at 10 minutes postinjection of the G6 contrast agent, are shown from a series of 3D dynamic micro-MR lymphangiograms. 3D-FIESTA-C especially showed the connection of thinner lymphatic vessels (arrows) between lymph nodes better than 3D-fSPGR. Arrowheads indicate injection sites. Broken arrows on (b) indicate the gall bladder.

Table 1.

Visualization Score of Four Lymph Nodes and Lymphatic Vessels on the Upper Body (n = 7)

	Time (min)
	10	20	30	40
3D-fSPGR lymph nodes	13	14	14	14
3D-FIESTA-C lymph nodes	14	14	14	14
3D-fSPGR lymphatic vessels	6	9	8	7
3D-FIESTA-C lymphatic vessels	8*	10	9	9*

Each score is the sum of consensus-based ratings (performed by two readers) for seven mice, with higher scores representing better visualization.

^*P < 0.005 compared with 3D-fSPGR.

3D-FIESTA-C Is Superior to 3D-fSPGR in Visualizing the Relationship between Tumors and the Lymphatic System

The 3D-FIESTA-C and 3D-fSPGR methods were compared in mice bearing Karpas 299 lymphoma (systemic; Figure 2, a and b) and PT-18 (xenograft/metastasis; Figure 2, c and d). The best imaging time, based on the greatest enhancement of normal lymph node tissues for both methods, to demonstrate metastatic nodules was at 20 or 30 minutes postinjection of the G6 agent. As previously demonstrated, the lymphatic system was well depicted on 3D-fSPGR images (Figure 2, a and c) [5]. However, contrast-enhanced 3D-FIESTA-C (Figure 2, b and d) not only demonstrated the lymphatics, but also identified the extralymphatic extension of the Karpas 299 and PT-18 tumors. In the Karpas 299 lymphoma model, contrast-enhanced 3D-FIESTA-C simultaneously identified the normal lymphatic system as high-signal structures and the extralymphatic tumors as moderate signal structures, compared to background. Therefore, the contrast-enhanced 3D-FIESTA-C method was able to demonstrate both the normal lymphatics and their relationship to extralymphatic tumors (Figure 2b). In contrast, in the PT-18 tumor model, the contrast-enhanced 3D-FIESTA-C method demonstrated that tumors were confined within the lymphatic system and appeared as filling defects within the nodes (Figure 2d). These findings were confirmed by histologic analysis, as shown below (Figure 3). In addition, the combination of these two methods enabled the evaluation of cystic dilatation of lymphatics in the lymph node due to tumor metastasis, blockage of normal lymph flow, and the absence of lymphatic flow in cystic structures (Figure 3).

Figure 2.

3D-micro-MR lymphangiograms with 3D-FIESTA-C visualized the relationship between the lymphatic system and tumors in mice. Two series of 3D dynamic micro-MR lymphangiograms of mice with Karpas 299 tumors (a and b; animation 3 and 4) and PT-18 tumors (c and d; animations 5 and 6) obtained with repeated 3D-fSPGR (a and c) and 3D-FIESTA-C scans (b and d) at 20 and 25 minutes postinjection of the G6 contrast agent. On micro-MR lymphangiograms with 3D-FIESTA-C, a normal lymphatic system is depicted and is superimposed on Karpas 299 tumors (arrows) (b). In contrast, in a mouse with PT-18, the opacified normal lymphatic tissue fills in around intranodal metastasis (d).

Figure 3.

Histologic analyses revealed the growth of PT-18 tumors in lymph nodes, which received lymphatic flow from the primary site. 3D-micro-MR lymphangiograms of axillary lymph nodes in a mouse bearing Karpas 299 tumor (a) and PT-18 metastatic tumors (b and c) are shown with histology (H–E staining, ×20). The axillary lymph nodes in a mouse bearing Karpas 299 tumor are shown in a normal lymph node without metastasis. We noticed this subtle inhomogeneity in all normal lymph nodes possibly because of the location of the lymphatic orifice, from which the contrast agent came into the lymph nodes. This inhomogeneity was clearer in earlier time points and was less clear in later time points. The lymph node tissue with metastatic tumor cells was not enhanced (b). The lymphatic vessels with multilocular cystic dilatation in the lymph node, which were depicted on micro-MRL, were demonstrated by histology (c). The walls of the cystic structures were lined by PT-18 tumor cells.

Histologic Analysis

In mice bearing Karpas 299, 11 of 13 tumors adhered to adjacent muscles, blood vessels, or bones. Bilateral axillary and lateral thoracic lymph nodes of normal sizes, which were adjacent to Karpas 299 tumors at the sites where micro-MRL was indicated, were separately identified. All lymph nodes were normal on H–E staining (Figure 3). In contrast, in mice bearing PT-18 xenografts in bilateral mammary glands, tumors were found in bilateral mammary glands in all eight mice examined. Additional tumors were found at the site of bilateral axillary lymph nodes (n = 16) and on either side of the lateral thoracic lymph node (n = 12). When PT-18 tumors involved lymph nodes, the lymph nodes were visibly changed in shape and were clearly depicted by micro-MRL. All enlarged lymph nodes showed a massive infiltration of eosinophilic PT-18 cells on specimens stained with H–E (Figure 3). In addition, 4 of 16 PT-18 tumors in the axillary lymph nodes contained single or multilocular cystic structures that were almost always lined by PT-18 cells [Figures 2 (c and d) and 3]. The existence of all four cystic lesions can be predicted by the combination of micro-MR lymphangiograms taken with 3D-fSPGR or 3D-FIESTA-C.

Discussion

Improvement in the outcome of malignant lymphoma has been a remarkable success story in hematology, but lymphoma remains a significant cause of morbidity and mortality. Moreover, the incidence of lymphoma has been increasing by 4% per year for the past four decades [1]. Accurate animal models present an opportunity to test new drug combinations, targeted therapies, and multimodal treatment combinations.

For an animal model to be predictive, it must reflect the characteristics of its human counterpart. Although most lymphomas are intralymphatic, an undefined portion is extralymphatic, and patients with extralymphatic lymphoma have a poorer response to therapy and, therefore, have poorer prognosis [3]. To emphasize its importance, lymphomas affecting tissues outside the lymph system (“extra(lymph)nodal”) have “E” added to their stage (e.g., stage IIE) [2]. No imaging method currently in widespread use can reliably determine whether a tumor is growing within or outside the lymph nodes because normal lymphatic vessels and normal lymph nodes cannot be easily visualized using conventional imaging modalities. Although a few mouse models of lymphoma are available, no method that can be used to determine if diseases are intralymphatic or extralymphatic exists. This is critical to the predictive value of animal models.

Micro-MRL enables the visualization of the lymphatic system and is able to identify intralymphatic diseases by demonstrating the absence of signal intensity within nodes. Extralymphatic diseases can be simultaneously distinguished from intralymphatic diseases and normal lymphatics using the 3D-FIESTA-C sequence. Indeed, it was initially thought that Karpas 299 tumors formed bulky lymphadenopathy until it became clear by micro-MRL that the disease was actually extralymphatic. By accurately mapping the lymphatic system with micro-MRL, Karpas 299 tumors developing in the soft tissues and muscles and outside the lymphatic system were readily diagnosed and confirmed by histologic analysis (Figure 3). Tumors and normal-sized lymph nodes were separately obtained by dissection of mice bearing Karpas 299. The lymph nodes near the tumor were totally intact, as shown in Figure 3. In contrast, the PT-18 model demonstrated intralymphatic metastases but also developed intranodal cystic masses that were easily detected by micro-MRL. Intranodal lymphoceles are an uncommon feature of human lymphomas. Therefore, micro-MRL is a useful analytic tool for characterizing animal models of lymphoma.

The 3D-FIESTA-C method enabled direct visualization of tumors with T2 prolongation of the tumor tissue while still depicting enhanced lymphatics and normal nodal tissues. Unlike 3D-FIESTA-C, contrast-enhanced T1-weighted 3D-fSPGR sequence could not directly depict tumors, unless the tumor had peripheral enhancement provided by the G6 agent through lymphatics. Both noncontrast and contrast-enhanced 3D-FIESTA-C methods depicted cystic masses as high-signal lesions [17]. Cystic masses can be misdiagnosed as dilated lymphatic vessels if only 3D-FIESTA-C is employed. In contrast, contrast-enhanced 3D-fSPGR with G6 agent can detect the lymphatic flow within cystic structures, thus indicating their identity as parts of the lymphatic system (Figure 2c).

A number of imaging methods are used for lymph node diseases, but they cannot distinguish between intralymphatic and extralymphatic metastases even in humans. Detection of lymphoma lesions using nuclear medicine techniques, especially ¹⁸fluoro-d-glucose [22,23] and lymphoscintigraphy [24,25], is successfully used clinically. However, all nuclear medicine techniques have insufficient resolution to accurately define the internal anatomy of lymph nodes or lymphatic vessels. X-ray lymphangiography using lipiodol, which is capable of visualizing the lymphatics, is no longer routinely performed because of difficulty in canulating the lymphatics in the foot, difficulties in interpreting results, and potential complications [26–29]. All of these methods are not available in mouse models because of insufficient spatial resolution. MR lymphography with ultra small particles of iron oxide might become a useful method in mouse models. To circumvent these shortcomings, we and others have recently developed MR lymphangiography, which has 10-fold greater spatial resolution (1 vs 10 mm) and 10-fold greater temporal resolution than lymphoscintigraphy [4,5,30]. This new imaging method relies on the selective uptake of Gd-labeled dendrimer nanoparticles (approximately 10 nm in diameter) within the lymphatic system [31]. In our previous studies, imaging was performed with T1-weighted pulse sequences to take advantage of the T1 shortening afforded by paramagnetic Gd to visualize the functional anatomy of lymphatic system [4,5].

Our results suggest that this new micro-MRL method may be of value not only as a new imaging method, but also as a platform technology for drug delivery to lymphomas. Intralymphatic treatments are difficult to justify because of uncertainty regarding the delivery of the agent and the systemic nature of the disease. This method clearly demonstrates that the lymphatic system allows assessment of pharmacokinetics within the lymph nodes and could potentially be used to deliver activatable therapeutic agents to targeted lymph nodes.

This method has several limitations. Because intralymphatic metastases may obstruct lymphatic flow, thus leading to the development of collateral lymphatics [31], micro-MRL might be limited in such cases. However, because micro-MRL is still able to depict the afferent and collateral lymphatic flows into the metastasized lymph node, it is unlikely that the method would fail to opacify a node due to lymphatic obstruction [31]. In reality, as previously demonstrated, in a spontaneous lymphoma model [4] and in the PT-18 model herein, complete blockade of lymphatic flow rarely occurs in lymph nodes involved in hematologic tumors. In both cases, all lymph nodes were stained well with nanosize agents (G6 or G8) despite intralymphatic diseases. Another limitation is that the high-signal intensity associated with extralymphatic tumors in FIESTA-C is nonspecific and inflammatory lymph nodes might also have the same appearance.

The micro-MRL with the G6 agent can differentiate normal from abnormal lymph nodes and also neoplastic from infectious lymph nodes [4]. However, reactive enlargement of lymph nodes by inflammation might not be easily differentiated from hematologic malignancy because of similarity in histology. In addition, because micro-MRL with the G6 agent enhances normal lymph node tissues rather than tumor nodules, a large lymphoid follicle may be misdiagnosed as a small tumor nodule. The minimum size of a metastatic tumor nodule that can be detected by the micro-MRL with the G6 agent is under investigation using several different tumor models, including cancers and hematologic malignancies.

In conclusion, we have developed a method of dynamic micro-MRL based on a combination of interstitial injection of G6 Gd-labeled dendrimer nanoparticle and dynamic micro-MRL using 3D-FIESTA-C and 3D-fSPGR. The micro-MRL using 3D-FIESTA-C is generally sufficient to analyze lymphatic tumors. However, only if there are cystic foci in the tumors will 3D-fSPGR contribute to obtaining additional information. The method is capable of displaying the functional anatomy of the lymphatic system in a 3D display. The high spatial and temporal resolutions of this method make it potentially superior to conventional methods in terms of lymphatic imaging, which is used to determine whether a lymphoma is growing within or outside the lymphatic system. This could lead to a more accurate characterization of animal models of lymphoma and potentially a more accurate localization of diseases in patients with lymphoma.

Abbreviations

micro-MRL

micromagnetic resonance lymphangiography

PAMAM

polyamidoamine

1B4M

2-(p-isothiocyanatobenzyl)-6-methyl-diethylenetriaminepentaacetic acid

G6

generation-6

SPGR

spoiled gradient echo

FIESTA

fast imaging employing steady-state acquisition