CME JOURNAL OF MAGNETIC RESONANCE IMAGING 41:1405–1412 (2015)
Original Research
In Vivo Visualization of Mesoscopic Anatomy of Healthy and Pathological Lymph Nodes Using 7T MRI: A Feasibility Study Martin T. Freitag, MD,1* Mathies Breithaupt, MSc,2 Moritz Berger, PhD,2 Reiner Umathum, PhD,2 Armin M. Nagel, PhD,2 Jessica Hassel, MD,3 Mark E. Ladd, PhD,2 Heinz-Peter Schlemmer, MD,4 Wolfhard Semmler, MD, PhD,2 and Bram Stieltjes, MD, PhD1 Purpose: To evaluate whether inguinal lymph nodes (LNs) may be visualized in vivo using 7T magnetic resonance imaging (MRI) at high spatial resolution.
Conclusion: We present a protocol with which inguinal LNs and their mesoscopic anatomy may be visualized in vivo using 7T MRI.
Materials and Methods: Twelve healthy controls and six patients with LN metastasis of melanoma were included. Examinations were performed using a 7T MRI and a transmit/receive loop coil. The protocol included a B0-map, B1-map, and T1-weighted-3D-fast low-angle shot (FLASH), T1w-Dixon-volumetric interpolated breath-hold examination (VIBE) and T2w sequences lasting 34.4 6 0.5 minutes. Signal- and contrast-to-noise of LNs, artery, muscle, and fat were quantified in controls. Metastatic features of LNs (hypervascularization, lymph vessels, fat hilus sign, tumor bulk, number of metastases, and size) were classified in patients.
Key Words: Lymph node; metastasis; neoangeogenesis; melanoma; 7T; loop coil J. Magn. Reson. Imaging 2015;41:1405–1412. C 2014 Wiley Periodicals, Inc. V
Results: Mesoscopic LN architecture such as central blood vessels and peripheral lymph vessels were observed in healthy controls with 0.5 mm3 isotropic resolution for T1w and 0.2 0.2 2 mm3 for T2w sequences. Mean signal-to-noise using 3D FLASH, Dixon VIBE and T2 TSE of healthy LN (27.2 6 7.5, 35.3 6 11.9, 31.7 6 11.1), muscle (17.6 6 4.6, 31.5 6 9.3, 7.3 6 5.4), artery (37.7 6 5.9, 42.7 6 19.7, 3.7 6 3.9), and saturated fat (3.7 6 0.9, 5.4 6 1.9, 9.3 6 5.2) and mean contrast-to-noise LN/fat (24.4 6 6.7, 39.6 6 11.1, 23.3 6 6.1) were adequate. In patients, multiple signs of metastasis could be clearly visualized.
1 Section Quantitative Imaging Based Disease Characterization, Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany. 2 Division of Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany. 3 Department of Dermatology, National Center for Tumor Diseases (NCT), University of Heidelberg, Heidelberg, Germany. 4 Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany. *Address reprint requests to: M.T.F., German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. E-mail: [email protected] Received March 14, 2014; Accepted June 11, 2014. DOI 10.1002/jmri.24686 View this article online at wileyonlinelibrary.com. C 2014 Wiley Periodicals, Inc. V
MANY COMMON CANCER TYPES such as breast cancer, skin cancer, and urologic or gynecologic tumors metastasize primarily to superficial regional lymph nodes (LNs). In tumor entities with a lymphatic metastasizing pattern, the nodal status is a major prognostic factor (1,2). In some entities, such as melanoma and breast cancer, it is even considered to be the most important factor (2–6). Despite the importance of correct LN staging in cancer, most imaging methods including ultrasound (US) (7,8), magnetic resonance imaging (MRI) (9,10), computed tomography (CT), and positron emission tomography (PET)/CT with 18fluoro-deoxy-glucose (6,11) lack sufficient sensitivity and specificity for the detection of small LN metastases. Instead of imaging, the current gold standard for initial LN staging in breast cancer and melanoma patients is an invasive sentinel node biopsy procedure. Here, technetium-99 is injected subcutaneously in the area of the primary tumor, resulting in radiolabeling of the first draining LN followed by surgical resection and histopathological analysis (12). As the majority of resected LNs do not contain tumor cells, it has been discussed that up to 96% of all sentinel nodes of melanoma patients are resected unnecessarily, resulting in overtreatment (13). Furthermore, the invasive sentinel node biopsy has false-negative numbers between 5– 10% (14,15), indicating that up to 10% of patients with negative sentinel nodes have metastasized disease, are wrongly staged, and receive their therapy too late. To overcome these limitations and to come to a more direct visualization of nodal architecture and vasculature, MRI could be advantageous at high
1405
1406
Freitag et al.
Table 1 Final Sequence Protocol Sequence B1 map B0 map T1w 3D FLASH T1w Dixon VIBE T2w TSE
Resolution (mm3) 113 113 0.5 isotropic 0.5 isotropic 0.2 0.2 2
FOV (mm x mm)
TR/TE (ms)
flip angle ( )
Averages
Acquisition time (s)
Orientation
6600/4 660/5.6 27.6/5.8 10/2.7 6584/57
90 48 15 13 120
1 1 1 1 1
81 253 [915.2 s 6 111] [425 s 6 22.4] [390 s 6 115]
Axial Axial Coronary Coronary Coronary
256 256 192 192 192
192 192 192 192 192
Acquisition time varied between individuals due to individual differences in anatomy and specific absorption rate limits. FOV: field of view; TE: echo time; TR: repetition time.
spatial resolution. MRI at ultrahigh magnetic field strengths (e.g., 7T) has already demonstrated capabilities for clinical diagnostics in a variety of studies (16). The first aim of the current study was to develop a clinically feasible protocol where LNs in the inguinal region could be visualized at such high spatial resolution that mesoscopic LN anatomy would be revealed in healthy individuals. The second aim was to test this protocol in patients with nodal metastasis of melanoma and to evaluate whether anatomical signs of metastasis could be visualized.
MATERIALS AND METHODS Volunteer and Patient Description The study was performed in accordance with the Declaration of Helsinki and was approved by the local ethics committee. Informed consent was obtained from all individuals. The imaging protocol was first established and optimized on controls and then tested on patients. Twelve healthy controls (five male, seven female; 26 6 4.9 years, range 18–34 years) and six patients (four male, two female; mean age 55.7 6 12.9 years, range 44–68 years) with a history of cutaneous malignant melanoma and suspect LN in the groin were included. All patients presented with at least one suspect LN in clinical examinations and were included to demonstrate the feasibility for clinical imaging. In seven controls, the MRI examination was established and tested including target resolutions ranging between 0.38 and 0.75 mm3 isotropic for T1w (weighted) sequences. The T2w sequence was established with 0.2 0.2 2 mm3 resolution. The final protocol (Table 1) was repeated in five controls. The LN metastases of patients were resected and confirmed by histopathology. 7T MRI The measurements were performed on a 7T wholebody scanner (Siemens Magnetom, Erlangen, Germany) equipped with a gradient system capable of 40 mT/m amplitude and 200 T/m/s slew rate. For tissue excitation and signal reception, a linearly polarized single-channel 1H surface loop coil was used (297.2 MHz, outer diameter 138 mm, inner diameter 75 mm; Rapid Biomedical, Rimpar, Germany). The participants were asked to avoid pronounced breathing to reduce movement-related artifacts. For imaging of inguinal LNs, the femoralis pulse was palpated and
the loop coil was positioned caudal to the inguinal ligament at the height of the femoralis pulse. The coil was fixed with a small sand bag and strapped to the patient table (Fig. 1a1–a3). Coil and scanner manufacturer safety standards were maintained to ensure that the specific absorption rate (SAR) limit was not exceeded. After the examination, patients were asked if the examination was well tolerated. Evaluation of the Scan Depth Inguinal LNs are located superficially in the subcutaneous fat tissue (superficial inguinal LNs) and also found numerously around the femoral vessels (deep inguinal LNs). Typically, these deep inguinal LNs are located 1–4 cm from the skin surface at the height of the large femoral vessels. To evaluate whether the loop coil could illuminate such a scan depth, a B1 map was obtained in the inguinal region to illustrate the relative flip angle as a function of spatial position using a presaturated double-contrast single-shot 2D fast low-angle shot (FLASH) sequence (resolution 1.0 1.0 2.5 mm3, field of view [FOV] 192 256, matrix 192 256, TR 5000 msec, TE 3.05 msec, four slices in axial and coronal orientation each, three averages). Four regions of interest (ROIs) with a circular area of 20 mm2 were placed in muscle tissue starting 2 cm below the skin surface up to the target depth of 5 cm in 1-cm steps. Sequences The femoral artery and vein were defined as the dorsal margin of the FOV. To determine the area of interest, a localizer with three orthogonal imaging planes followed by a coronally oriented multislice FLASH sequence (fat suppression, resolution 1.5 1.5 2 mm3, FOV 280 280, matrix size 192 192, TR 262 msec, TE 4.45 msec, 26 slices) were used. A welldefined flip angle at the ROI was achieved by using a single-voxel-spectroscopy tool and varying flip angles to adjust the reference B1 amplitude. The voxel was placed in the most representative LN in the localizer. To minimize flow artifacts of the femoral vessels, 3D flow compensation was used. The optimized sequence protocol is given in Table 1. Coronal orientation was preferred over axial orientation for the T1w and T2w sequences since LNs in the inguinal region are usually oriented parallel to the longitudinal axis of the human body.
In Vivo 7T Lymph Node MRI
1407
Figure 1. Study setup and demonstration of target depth. a1: Employed loop coil (Rapid Biomedical). a2: The loop coil is placed in the inguinal region at the height of the femoralis pulse and weighted with a sand bag to prevent artifacts from breathing and to ensure close positioning to the skin. a3: Imaging of the superficial (A) and subinguinal (B) glands by positioning the coil (red circle) tangentially and mediocaudal of the inguinal ligament (red line); adapted from (29). b1: Exemplary B1 map in axial orientation of the left inguinal region in a healthy control. b2: Corresponding 3D FLASH sequence at 0.5 mm3 isotropic resolution to demonstrate the achieved target depth. The colormap in (b1) illustrates the variation of the B1 profile in muscle tissue (asterisk) relative to the target depth. The flip angle is plotted from 25 to 100 , as indicated by the scale. The large femoral vessels were defined as the dorsal margin (red arrow: arteria femoralis) of the region of interest; hence, healthy inguinal LNs (white arrows) could be visualized. R: right; L: left.
Image Evaluation Before further evaluation, the images were screened for macroscopic artifacts. Signal-to-noise (SNR) and contrast-to-noise (CNR) measurements were performed by dividing the signal obtained from an ROI (lymph node, vessel, fat, and sartorius muscle) through the standard deviation of the noise located outside of the body in coronary orientation (Fig. 2a). Images were evaluated with regard to mesoscopic details by two radiologists in consensus (M.T.F. with 4 years and B.S. with 12 years of experience in oncologic radiology). The
Figure 2. SNR and CNR comparisons of selected anatomical structures in five healthy individuals using the employed sequences. ROIs were placed in muscle tissue, artery, lymph node, and fat as exemplarily demonstrated in the T1 gradient-echo sequence (a). SNR and CNR (b) were calculated by using the standard deviation from the background signal outside of the individual.
LNs were screened for pathological signs by the readers: peripheral hypervascularization, pronounced lymphatic vessels, number of suspicious LNs, presence of a tumor bulk, and loss of fat hilus. A score between 0 and 3 (0 ¼ not visible, 1 ¼ suggestive, 2 ¼ visible, 3 ¼ distinct) was addressed. For size criteria, an LN was regarded as metastatic if the short axis diameter was at least 1 cm or larger and a score was allocated on a three-point scale (1 ¼ 2 cm). DICOM file conversion, MPR reconstructions, ROI analysis, and data visualization was performed using Osirix
1408
Freitag et al.
Table 2 Relationship Between the Nominal and the Actual Flip Angle Parameter k Mean actual flip angles ( )
Sequence
Flip angle ( )
2 cm
3 cm
4 cm
5 cm
— T1w 3D FLASH T1w Dixon VIBE T2w TSE
— 15 13 120
1.0 6 0.2 15.0 6 3.0 13.0 6 2.6 120.0 6 24.0
0.8 6 0.2 12.0 6 3.0 10.4 6 2.6 96.0 6 24.0
0.7 6 0.2 10.5 6 3.0 9.1 6 2.6 84.0 6 24.0
0.6 6 0.1 9.0 6 1.5 7.8 6 1.3 72.0 6 12.0
Calculated from five participants with B1 maps as given in Table 1. The ROI-based analysis was performed in the axially oriented B1-map beginning 2 cm below the skin surface down to 5 cm. As a reference, muscle tissue close to the LN under investigation was chosen over fat tissue, since fat tissue is more heterogeneous (fat, LN, vessels) and the B1-mapping sequence is not optimized for fat due to the frequency-selectivity of the magnetization-preparation pulse. Kappa k ¼ Mean relative flip angle.
(17). The 3D virtual reconstruction in Fig. 6c was created using Fovia Software (Palo Alto, CA). Additional LN Work-up The rough position of LN metastasis was obtained from US-imaging employing a 14 MHz linear array (Siemens Accuson, Erlangen, Germany) prior to the MRI examination to guide later coil positioning. Histological workup was performed for all LNs including hematoxylin-eosin (tumor), D2–40 (lymphoid tissue), and CD31 (blood vessels) stainings. RESULTS Scan Depth As illustrated in the colored B1 map (Fig. 1b1), the obtained flip angles were sufficient at the height of inguinal LNs, usually located between 1–4 cm below the skin surface, up to the target depth of 5 cm (Table 2), where the femoral vessels represented the dorsal margin of the FOV.
Sequence Evaluation In the first two patients (patients #1 and #2), the T2w TSE sequence could not be performed due to incompliance (back pain). All other participants did not report any discomfort and the exam was well tolerated. Slight SNR/CNR variations across space due to B1-inhomogeneity were sometimes visible but did not influence the diagnostic interpretation up to the target depth of 5 cm. Severe motion artifacts
were not present. The use of the loop coil resulted in a characteristic cone-shaped FOV. Optimized sequence parameters are given in Table 1. In one control, fat suppression artifacts were noticed but these were not located in the area of interest. The image quality of the Dixon volumetric interpolated breath-hold examination (VIBE) water image was comparable to the 3D FLASH, with the substantially lower acquisition time favoring the VIBE sequence (Table 1). Delineation of Mesoscopic LN Architecture Muscle, artery, LNs, and fat could be visualized with excellent contrast to the adjacent tissue and sufficient SNR (Fig. 2b). In controls, mesoscopic details such as the saturated fatty hilus and LN cortex (Fig. 3a–d) could be observed. Within the FOV 6.6 6 0.5 LNs were seen, of which 87.9 6 12.6% had a fat hilus sign. LNs were always connected within an intricate and dense network of blood (Fig. 4a,b) and lymph (Fig. 4c,d) vessels. In all depicted LNs, blood vessels could be identified as inserting into the hilus region and connected to a larger vessel distal from the LN (Fig. 3a,b). In contrast, lymph vessels were identified as vessels inserting into the LN periphery, only clearly visible at 0.2 mm 0.2 mm in-plane resolution using the T2w TSE sequence (Fig. 4c,d). These vessels could also be identified in T1w sequences, but spatial resolution of the T1w sequences (0.5 mm3 isotropic) was too low compared to T2w (0.2 0.2 2 mm3) to clearly verify the peripheral connection. In general, blood vessels were T2w hypointense due to the flow-void effect (Table 1, Fig. 3d), but lymph vessels were observed to be T2w
Figure 3. 7T LN MRI. A prominent lymph node (10 mm 40 mm) of a healthy participant is depicted using the parameters for the 3D FLASH (a), T1w Dixon VIBE water image (b), T1w Dixon VIBE fat image (c), and T2w TSE (d) given in Table 1. The hilus region (red arrow), a blood vessel to the hilum (white arrow), and the sartorius muscle (white arrowhead) are marked. Note that the T1w hyperintense venous vessel (white arrow) is T2w hypointense due to the flow-void effect. R: right. L: left.
In Vivo 7T Lymph Node MRI
1409
Figure 4. Visualization of mesoscopic LN vasculature in four different healthy individuals. The LN is connected within a dense network of blood (a,b) and lymph vessels (c,d). Blood vessels (white arrows in a,b) insert into the LN hilus region, whereas lymph vessels (open arrows in c,d) are identified by insertion into the periphery of the LN and by T2w hyperintensity. Arteries and veins cannot be differentiated. a: The connection of the LNs with surrounding blood vessels and the visualization of efferent and afferent vessels (white arrows) to the LN is shown (maximum intensity projection, 12 mm). b: Visualization of very small blood vessels (white arrows) supplying the hilum region. c,d: T2w hyperintense vessels (open arrows) inserting into the lymph node periphery correspond to lymph vessels with diameters between 350 and 500 mm.
hyperintense (Fig. 4c,d), as the flow rate is much lower in the lymphatic compared to the vascular system. LN metastases were pathologically enlarged in five of six patients and considered as macro-metastases (mean size of the short axis [2.8 6 0.4] cm). In one patient, the short axis was not pathologically enlarged
(9 mm). Five of six patients showed LNs with hypervascularization in 7T MRI, of which four LNs had aberrant vessels stemming from smaller subvessels of the femoral vessels (exemplary Fig. 5a) and one LN had multiple large feeding vessels from the femoral artery (Fig. 6). Three patients showed LNs with
Figure 5. Direct visualization of neoangiogenesis, lymphangiogenesis, and tumor bulk at 7T. Histopathologically confirmed LN metastasis of a 60-year-old female patient with a history of superficial spreading melanoma of the lower extremity presenting with swelling in the left groin. Besides two macroscopic signs of malignancy (3 cm short-axis diameter and loss of the fatty hilum), the T1w Dixon VIBE maximum intensity projection (4 mm) depicts strong hilus perfusion (open arrow) and peripheral hypervascularization (a1) correlating to the histopathology CD31 staining (a2) and suggesting malignant neoangiogenesis. Red arrows denote blood vessels. T2w images show pronounced peripheral lymph vessels (b1, yellow arrows) which insert into the peripheral lymph node capsule as depicted in the D2–40 staining for lymphoid tissue (b2). Furthermore, hyperintense intranodal inhomogeneity is visible in T2w images (c1) correlating with tumor tissue in the hematoxylin-eosin staining (c2), as illustrated by the dotted line. The rectangle in a1, b1, and c1 defines the area where histopathological workup was performed.
1410
Freitag et al.
Figure 6. Visualization of neoangiogenesis. A 45-year-old female patient with superficial spreading melanoma of the left lower abdomen presenting 6 years after diagnosis with a definite LN macrometastasis in the left inguinal region. This example demonstrates the visualization of two large feeding vessels from the femoral artery (red arrow and white triangle). a: T1w 3D FLASH. b: Maximum intensity projection (7 mm) of 3D FLASH. c: 3D virtual reconstruction technique (VRT) of (a) with highlighted neoangiogenesis and vessel connection to the femoral artery (asterisk). Note the intranodal tissue inhomogeneity in the metastatic LN (white arrow). R: right; L: left.
emphasized lymph vessels inserting into the LN periphery (exemplary Fig. 5b). All patients presented with intranodal tumor bulk, most distinct in the T2w TSE sequence (compare Fig. 5a1 vs. 5b1 and 5c1). The results of pathological LN signs are shown in Table 3. The diagnosis of nodal metastasis was confirmed by histopathology in all patients.
DISCUSSION We present a series that shows that in vivo 7T MRI examination of superficial and deep inguinal LNs using high spatial resolution is feasible. Earlier ex vivo reports (18,19) about dissected LN specimens used high-resolution T1w gradient-echo sequences employing a 0.18 mm3 isotropic resolution, albeit using long, clinically unacceptable acquisition times (3–5 hours). However, their results corroborated that a clinically feasible examination protocol of LNs with high spatial resolution could have significant impact on cancer diagnostics, since detailed information on the LN micro-architecture such as vessels may be delineated and, thus, high-resolution 7T MRI might serve as a diagnostic method for the identification of small LN metastasis (18,19). Inguinal LNs are embedded in fat pads adjacent to the large femoral vessels. Therefore, fat saturation is necessary to maximize the contrast between LN and fat tissue. Fat-water separation by Dixon VIBE is achieved by acquisition of multiple images with defined fat-water phase differences, whereas 3D FLASH employs a frequency-selective saturation prepulse with a center frequency corresponding to that of lipids (20). The image quality and SNR of both T1w sequences with identical scan parameters was comparable, but the Dixon VIBE provides two compelling advantages in comparison to the 3D FLASH: the possibility to generate separate fat and water maps accompanied by a substantially lower scan time. The fat map may be used to help identify the presence of the fat hilus, a potential sign of benignity.
A healthy LN is typically only perfused by blood vessels inserting in the fatty hilus region. In contrast, peripheral vascularization is suggestive for malignancy and typical for metastases of malignant melanoma (21–23). This sign was also noticed at 7T in five of six included patients. Neoangiogenesis evolves at 2 mm metastasis size (24) and, if untreated, may lead to a tumor with multiple feeding vessels from large arteries. This hypervascularization is directly visible as T1w hyperintense vessels. From a previous study in LN metastases of breast cancer, neovascularization is a known predictor for overall and diseasefree survival (25). Thus, the direct imaging-based identification of pathological tumor vessels in LNs provides additional information about the progression of the disease and its prognosis, as multiple vessel connections are indicative of higher probability of further hematogeneous spread to central organs. Furthermore, 7T MRI was able to visualize smallest vessels with diameters between 350–500 mm inserting into the LN periphery using the T2 TSE sequence. These small vessels correspond to lymph vessels according to the normal anatomy of a LN, whereas blood vessels commonly insert via the fat hilus region (26). We have provided evidence for the depiction of this intricate network of healthy lymph vessels in vivo without the use of contrast media. It has been reported that lymphangiogenesis in the LN metastasis, similar to angiogenesis, is a prognostic factor for local and distant cancer metastasis (27). The presence of lymphangiogenesis or dilated lymph vessels due to metastatic tumor load in the LN could represent valuable diagnostic markers. In accordance, we did observe pronounced lymph vessels in three of six patients. This study has some limitations. First, the MR images were screened in consensus between two radiologists, which may potentially result in reading bias. However, in this study we did not aim to assess the diagnostic accuracy. By consensus reading we aimed to assure the optimal selection of pathological LNs and the proper description of the imaging findings.
#6 male
#5 male
#4 (female, 60y)
#2 (female, 44y) #3 (male, 73y)
The rating scale is defined as: 0 (not present), 1 (suggestive), 2 (visible), 3 (distinct). Current tomography criteria for LN pathology (enlargement, presence of fat hilus) are extended at 7T MRI by direct visualization of blood and lymph vessels and intranodal tumor inhomogeneity at high spatial resolution. *The T2w TSE sequence in patients #1 and #2 was not assessable due to reduced compliance at the end of the examination.
1/3 1/3 3/3 0/3 1
0/3
3/3 3/3 3/3 1/3 3
3/3
3/3 3/3 3/3 3/3 1
3/3
3/3 3/3 3/3 3/3 3/3 3/3 3/3 2/3 1 7
0/3* 2/3
3/3 3/3 3/3 3/3 2
0/3*
1411
Superficial spreading melanoma (pT1bN3cM0) Nodular melanoma (pT3bN1M0) Resected melanoma metastasis of the left thigh with unknown primary (TxN3M1) Superficial spreading melanoma (T3N1M0) Superficial spreading melanoma (pT4bN2M0) Acro-lentiginous melanoma (pT3bN1aM0) #1 (female, 45y)
Diagnosis (tumor stadium) Patients
Table 3 Signs of LN Malignancy in the Included Patients at 7T MRI
Number of LN metastases
Peripheral blood vessel insertion
Pronounced lymph vessels visible
LN tumor bulk (T1w, T2w)
Loss of fat hilus
Enlargement (1: < 1 cm; 2: 1–2 cm; 3: > 2 cm short-axis diameter)
In Vivo 7T Lymph Node MRI
Second, the number of included patients was small and thus some imaging features that were clearly visible in this small sample may not be as prominent in a larger group of patients. In conclusion, hallmarks of cancer such as neoangiogenesis and lymphangiogenesis may be directly visualized in vivo with the proposed protocol. This could be of critical importance considering the differentiation between inflammation and metastasis. In future studies, there are several promising MRI-based techniques that could optimize the proposed protocol for LN imaging at 7T such as contrast-enhanced dynamic sequences or time-of-flight angiography. Another very interesting approach at high resolution would be the application of contrast medium to achieve contrast enhancement within the LN body, either by shortening the T1 relaxation times (eg, gadolinium) or by exploiting the T2* effect (eg, iron nanoparticles), which could, analogous to findings of previous studies at lower resolution (28), help to identify LN metastases at an earlier timepoint. Besides its usefulness for tumor diagnostics, the use of gadolinium-based agents could further help to increase the obtained SNR and CNR and thus even further improve the image quality or target resolution, as LNs tend to take up this contrast media. However, dynamic contrast-enhanced sequences typically have lower spatial resolution compared to the standard static T1w sequences to calculate perfusion biomarkers, in favor of a higher temporal resolution. As described in the introduction, further advances in noninvasive LN imaging are urgently needed. For this purpose, MRI at high resolution, particularly considering that it is a radiation-free tomographic method, could be considered as an alternative or additional modality. Our proposed setup serves as a first step to increase the spatial resolution in superficial LNs with a transmit/receive coil in the inguinal region. To achieve broader clinical application, further coil improvement is necessary, especially to access the axilla and cervical region. High-risk patients, unclear nodal status in patients undergoing radiotherapy, or patients undergoing regular nodal staging may substantially benefit from a complementary high-resolution MRI. However, the number of patients included in the present study was limited, as feasibility was the main purpose. Therefore, a detailed systematic study using our proposed protocol is now warranted to determine whether high-resolution LN imaging at 7T really improves the sensitivity/specificity in comparison to lower field strengths or other modalities.
CONFLICT OF INTEREST The authors declare no conflict of interest.
ACKNOWLEDGMENTS We thank Siemens (Erlangen, Germany) for providing the 1H transmit/receive loop-coil. We also thank Jessica Engelhart and Vanessa Peregovich for their kind assistance.
1412
REFERENCES 1. McMahon CJ, Rofsky NM, Pedrosa I. Lymphatic metastases from pelvic tumors: anatomic classification, characterization, and staging. Radiology 2010;254:31–46. 2. de Boer M, van Deurzen CH, van Dijck JA, et al. Micrometastases or isolated tumor cells and the outcome of breast cancer. N Engl J Med 2009;361:653–663. 3. Balch CM, Soong SJ, Gershenwald JE, et al. Prognostic factors analysis of 17,600 melanoma patients: validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol 2001;19:3622–3634. 4. Gershenwald JE, Thompson W, Mansfield PF, et al. Multi-institutional melanoma lymphatic mapping experience: the prognostic value of sentinel lymph node status in 612 stage I or II melanoma patients. J Clin Oncol 1999;17:976–983. 5. Morton DL, Thompson JF, Cochran AJ, et al. Sentinel-node biopsy or nodal observation in melanoma. N Engl J Med 2006; 355:1307–1317. 6. Klode J, Dissemond J, Grabbe S, Hillen U, Poeppel T, Boeing C. Sentinel lymph node excision and PET-CT in the initial stage of malignant melanoma: a retrospective analysis of 61 patients with malignant melanoma in American Joint Committee on Cancer stages I and II. Dermatol Surg 2010;36:439–445. 7. Chai CY, Zager JS, Szabunio MM, et al. Preoperative ultrasound is not useful for identifying nodal metastasis in melanoma patients undergoing sentinel node biopsy: preoperative ultrasound in clinically node-negative melanoma. Ann Surg Oncol 2011:1100–1106. 8. Lee B, Lim AK, Krell J, et al. The efficacy of axillary ultrasound in the detection of nodal metastasis in breast cancer. AJR Am J Roentgenol 2013;200:W314–320. 9. Bellin MF, Lebleu L, Meric JB. Evaluation of retroperitoneal and pelvic lymph node metastases with MRI and MR lymphangiography. Abdom Imaging 2003;28:155–163. 10. Schipper RJ, Smidt ML, van Roozendaal LM, et al. Noninvasive nodal staging in patients with breast cancer using gadofosvesetenhanced magnetic resonance imaging: a feasibility study. Invest Radiol 2013;48:134–139. 11. Acland KM, Healy C, Calonje E, et al. Comparison of positron emission tomography scanning and sentinel node biopsy in the detection of micrometastases of primary cutaneous malignant melanoma. J Clin Oncol 2001;19:2674–2678. 12. Balch CM, Gershenwald JE, Soong SJ, et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol 2009;27:6199–6206. 13. Torjesen I. Sentinel node biopsy for melanoma: unnecessary treatment? BMJ 2013;346:e8645.
Freitag et al. 14. Estourgie SH, Nieweg OE, Valdes Olmos RA, Hoefnagel CA, Kroon BB. Review and evaluation of sentinel node procedures in 250 melanoma patients with a median follow-up of 6 years. Ann Surg Oncol 2003;10:681–688. 15. Morton DL, Cochran AJ, Thompson JF, et al. Sentinel node biopsy for early-stage melanoma: accuracy and morbidity in MSLT-I, an international multicenter trial. Ann Surg 2005;242: 302–311. 16. Kraff O, Fischer A, Nagel AM, Monninghoff C, Ladd ME. MRI at 7 Tesla and above: demonstrated and potential capabilities. J Magn Reson Imaging JMRI 2014 [Epub ahead of print]. 17. Rosset A, Spadola L, Ratib O. OsiriX: an open-source software for navigating in multidimensional DICOM images. J Digit Imaging 2004;17:205–216. 18. Korteweg MA, Zwanenburg JJ, Hoogduin JM, et al. Dissected sentinel lymph nodes of breast cancer patients: characterization with high-spatial-resolution 7-T MR imaging. Radiology 2011; 261:127–135. 19. Korteweg MA, Zwanenburg JJ, van Diest PJ, et al. Characterization of ex vivo healthy human axillary lymph nodes with high resolution 7 Tesla MRI. Eur Radiol 2011;21:310–317. 20. Delfaut EM, Beltran J, Johnson G, Rousseau J, Marchandise X, Cotten A. Fat suppression in MR imaging: techniques and pitfalls. Radiographics 1999;19:373–382. 21. Voit C, Kron M, Schafer G, et al. Ultrasound-guided fine needle aspiration cytology prior to sentinel lymph node biopsy in melanoma patients. Ann Surg Oncol 2006;13:1682–1689. 22. Thompson JF, Haydu LE, Sanki A, Uren RF. Ultrasound assessment of lymph nodes in the management of early-stage melanoma. J Surg Oncol 2011;104:354–360. 23. Torabi M, Aquino SL, Harisinghani MG. Current concepts in lymph node imaging. J Nnucl Med 2004;45:1509–1518. 24. Mihic-Probst D, Ikenberg K, Tinguely M, et al. Tumor cell plasticity and angiogenesis in human melanomas. PloS One 2012;7: e33571. 25. Guidi AJ, Berry DA, Broadwater G, et al. Association of angiogenesis in lymph node metastases with outcome of breast cancer. J Natl Cancer Inst 2000;92:486–492. 26. Clement O, Luciani A. Imaging the lymphatic system: possibilities and clinical applications. Eur Radiol 2004;14:1498–1507. 27. Rinderknecht M, Detmar M. Tumor lymphangiogenesis and melanoma metastasis. J Cell Physiol 2008;216:347–354. 28. Harisinghani MG, Barentsz J, Hahn PF, et al. Noninvasive detection of clinically occult lymph-node metastases in prostate cancer. N Engl J Med 2003;348:2491–2499. 29. Gray H, Lewis WH. Anatomy of the human body. Philadelphia and New York: Lea & Febiger: 1918.
Original Research
In Vivo Visualization of Mesoscopic Anatomy of Healthy and Pathological Lymph Nodes Using 7T MRI: A Feasibility Study Martin T. Freitag, MD,1* Mathies Breithaupt, MSc,2 Moritz Berger, PhD,2 Reiner Umathum, PhD,2 Armin M. Nagel, PhD,2 Jessica Hassel, MD,3 Mark E. Ladd, PhD,2 Heinz-Peter Schlemmer, MD,4 Wolfhard Semmler, MD, PhD,2 and Bram Stieltjes, MD, PhD1 Purpose: To evaluate whether inguinal lymph nodes (LNs) may be visualized in vivo using 7T magnetic resonance imaging (MRI) at high spatial resolution.
Conclusion: We present a protocol with which inguinal LNs and their mesoscopic anatomy may be visualized in vivo using 7T MRI.
Materials and Methods: Twelve healthy controls and six patients with LN metastasis of melanoma were included. Examinations were performed using a 7T MRI and a transmit/receive loop coil. The protocol included a B0-map, B1-map, and T1-weighted-3D-fast low-angle shot (FLASH), T1w-Dixon-volumetric interpolated breath-hold examination (VIBE) and T2w sequences lasting 34.4 6 0.5 minutes. Signal- and contrast-to-noise of LNs, artery, muscle, and fat were quantified in controls. Metastatic features of LNs (hypervascularization, lymph vessels, fat hilus sign, tumor bulk, number of metastases, and size) were classified in patients.
Key Words: Lymph node; metastasis; neoangeogenesis; melanoma; 7T; loop coil J. Magn. Reson. Imaging 2015;41:1405–1412. C 2014 Wiley Periodicals, Inc. V
Results: Mesoscopic LN architecture such as central blood vessels and peripheral lymph vessels were observed in healthy controls with 0.5 mm3 isotropic resolution for T1w and 0.2 0.2 2 mm3 for T2w sequences. Mean signal-to-noise using 3D FLASH, Dixon VIBE and T2 TSE of healthy LN (27.2 6 7.5, 35.3 6 11.9, 31.7 6 11.1), muscle (17.6 6 4.6, 31.5 6 9.3, 7.3 6 5.4), artery (37.7 6 5.9, 42.7 6 19.7, 3.7 6 3.9), and saturated fat (3.7 6 0.9, 5.4 6 1.9, 9.3 6 5.2) and mean contrast-to-noise LN/fat (24.4 6 6.7, 39.6 6 11.1, 23.3 6 6.1) were adequate. In patients, multiple signs of metastasis could be clearly visualized.
1 Section Quantitative Imaging Based Disease Characterization, Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany. 2 Division of Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany. 3 Department of Dermatology, National Center for Tumor Diseases (NCT), University of Heidelberg, Heidelberg, Germany. 4 Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany. *Address reprint requests to: M.T.F., German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. E-mail: [email protected] Received March 14, 2014; Accepted June 11, 2014. DOI 10.1002/jmri.24686 View this article online at wileyonlinelibrary.com. C 2014 Wiley Periodicals, Inc. V
MANY COMMON CANCER TYPES such as breast cancer, skin cancer, and urologic or gynecologic tumors metastasize primarily to superficial regional lymph nodes (LNs). In tumor entities with a lymphatic metastasizing pattern, the nodal status is a major prognostic factor (1,2). In some entities, such as melanoma and breast cancer, it is even considered to be the most important factor (2–6). Despite the importance of correct LN staging in cancer, most imaging methods including ultrasound (US) (7,8), magnetic resonance imaging (MRI) (9,10), computed tomography (CT), and positron emission tomography (PET)/CT with 18fluoro-deoxy-glucose (6,11) lack sufficient sensitivity and specificity for the detection of small LN metastases. Instead of imaging, the current gold standard for initial LN staging in breast cancer and melanoma patients is an invasive sentinel node biopsy procedure. Here, technetium-99 is injected subcutaneously in the area of the primary tumor, resulting in radiolabeling of the first draining LN followed by surgical resection and histopathological analysis (12). As the majority of resected LNs do not contain tumor cells, it has been discussed that up to 96% of all sentinel nodes of melanoma patients are resected unnecessarily, resulting in overtreatment (13). Furthermore, the invasive sentinel node biopsy has false-negative numbers between 5– 10% (14,15), indicating that up to 10% of patients with negative sentinel nodes have metastasized disease, are wrongly staged, and receive their therapy too late. To overcome these limitations and to come to a more direct visualization of nodal architecture and vasculature, MRI could be advantageous at high
1405
1406
Freitag et al.
Table 1 Final Sequence Protocol Sequence B1 map B0 map T1w 3D FLASH T1w Dixon VIBE T2w TSE
Resolution (mm3) 113 113 0.5 isotropic 0.5 isotropic 0.2 0.2 2
FOV (mm x mm)
TR/TE (ms)
flip angle ( )
Averages
Acquisition time (s)
Orientation
6600/4 660/5.6 27.6/5.8 10/2.7 6584/57
90 48 15 13 120
1 1 1 1 1
81 253 [915.2 s 6 111] [425 s 6 22.4] [390 s 6 115]
Axial Axial Coronary Coronary Coronary
256 256 192 192 192
192 192 192 192 192
Acquisition time varied between individuals due to individual differences in anatomy and specific absorption rate limits. FOV: field of view; TE: echo time; TR: repetition time.
spatial resolution. MRI at ultrahigh magnetic field strengths (e.g., 7T) has already demonstrated capabilities for clinical diagnostics in a variety of studies (16). The first aim of the current study was to develop a clinically feasible protocol where LNs in the inguinal region could be visualized at such high spatial resolution that mesoscopic LN anatomy would be revealed in healthy individuals. The second aim was to test this protocol in patients with nodal metastasis of melanoma and to evaluate whether anatomical signs of metastasis could be visualized.
MATERIALS AND METHODS Volunteer and Patient Description The study was performed in accordance with the Declaration of Helsinki and was approved by the local ethics committee. Informed consent was obtained from all individuals. The imaging protocol was first established and optimized on controls and then tested on patients. Twelve healthy controls (five male, seven female; 26 6 4.9 years, range 18–34 years) and six patients (four male, two female; mean age 55.7 6 12.9 years, range 44–68 years) with a history of cutaneous malignant melanoma and suspect LN in the groin were included. All patients presented with at least one suspect LN in clinical examinations and were included to demonstrate the feasibility for clinical imaging. In seven controls, the MRI examination was established and tested including target resolutions ranging between 0.38 and 0.75 mm3 isotropic for T1w (weighted) sequences. The T2w sequence was established with 0.2 0.2 2 mm3 resolution. The final protocol (Table 1) was repeated in five controls. The LN metastases of patients were resected and confirmed by histopathology. 7T MRI The measurements were performed on a 7T wholebody scanner (Siemens Magnetom, Erlangen, Germany) equipped with a gradient system capable of 40 mT/m amplitude and 200 T/m/s slew rate. For tissue excitation and signal reception, a linearly polarized single-channel 1H surface loop coil was used (297.2 MHz, outer diameter 138 mm, inner diameter 75 mm; Rapid Biomedical, Rimpar, Germany). The participants were asked to avoid pronounced breathing to reduce movement-related artifacts. For imaging of inguinal LNs, the femoralis pulse was palpated and
the loop coil was positioned caudal to the inguinal ligament at the height of the femoralis pulse. The coil was fixed with a small sand bag and strapped to the patient table (Fig. 1a1–a3). Coil and scanner manufacturer safety standards were maintained to ensure that the specific absorption rate (SAR) limit was not exceeded. After the examination, patients were asked if the examination was well tolerated. Evaluation of the Scan Depth Inguinal LNs are located superficially in the subcutaneous fat tissue (superficial inguinal LNs) and also found numerously around the femoral vessels (deep inguinal LNs). Typically, these deep inguinal LNs are located 1–4 cm from the skin surface at the height of the large femoral vessels. To evaluate whether the loop coil could illuminate such a scan depth, a B1 map was obtained in the inguinal region to illustrate the relative flip angle as a function of spatial position using a presaturated double-contrast single-shot 2D fast low-angle shot (FLASH) sequence (resolution 1.0 1.0 2.5 mm3, field of view [FOV] 192 256, matrix 192 256, TR 5000 msec, TE 3.05 msec, four slices in axial and coronal orientation each, three averages). Four regions of interest (ROIs) with a circular area of 20 mm2 were placed in muscle tissue starting 2 cm below the skin surface up to the target depth of 5 cm in 1-cm steps. Sequences The femoral artery and vein were defined as the dorsal margin of the FOV. To determine the area of interest, a localizer with three orthogonal imaging planes followed by a coronally oriented multislice FLASH sequence (fat suppression, resolution 1.5 1.5 2 mm3, FOV 280 280, matrix size 192 192, TR 262 msec, TE 4.45 msec, 26 slices) were used. A welldefined flip angle at the ROI was achieved by using a single-voxel-spectroscopy tool and varying flip angles to adjust the reference B1 amplitude. The voxel was placed in the most representative LN in the localizer. To minimize flow artifacts of the femoral vessels, 3D flow compensation was used. The optimized sequence protocol is given in Table 1. Coronal orientation was preferred over axial orientation for the T1w and T2w sequences since LNs in the inguinal region are usually oriented parallel to the longitudinal axis of the human body.
In Vivo 7T Lymph Node MRI
1407
Figure 1. Study setup and demonstration of target depth. a1: Employed loop coil (Rapid Biomedical). a2: The loop coil is placed in the inguinal region at the height of the femoralis pulse and weighted with a sand bag to prevent artifacts from breathing and to ensure close positioning to the skin. a3: Imaging of the superficial (A) and subinguinal (B) glands by positioning the coil (red circle) tangentially and mediocaudal of the inguinal ligament (red line); adapted from (29). b1: Exemplary B1 map in axial orientation of the left inguinal region in a healthy control. b2: Corresponding 3D FLASH sequence at 0.5 mm3 isotropic resolution to demonstrate the achieved target depth. The colormap in (b1) illustrates the variation of the B1 profile in muscle tissue (asterisk) relative to the target depth. The flip angle is plotted from 25 to 100 , as indicated by the scale. The large femoral vessels were defined as the dorsal margin (red arrow: arteria femoralis) of the region of interest; hence, healthy inguinal LNs (white arrows) could be visualized. R: right; L: left.
Image Evaluation Before further evaluation, the images were screened for macroscopic artifacts. Signal-to-noise (SNR) and contrast-to-noise (CNR) measurements were performed by dividing the signal obtained from an ROI (lymph node, vessel, fat, and sartorius muscle) through the standard deviation of the noise located outside of the body in coronary orientation (Fig. 2a). Images were evaluated with regard to mesoscopic details by two radiologists in consensus (M.T.F. with 4 years and B.S. with 12 years of experience in oncologic radiology). The
Figure 2. SNR and CNR comparisons of selected anatomical structures in five healthy individuals using the employed sequences. ROIs were placed in muscle tissue, artery, lymph node, and fat as exemplarily demonstrated in the T1 gradient-echo sequence (a). SNR and CNR (b) were calculated by using the standard deviation from the background signal outside of the individual.
LNs were screened for pathological signs by the readers: peripheral hypervascularization, pronounced lymphatic vessels, number of suspicious LNs, presence of a tumor bulk, and loss of fat hilus. A score between 0 and 3 (0 ¼ not visible, 1 ¼ suggestive, 2 ¼ visible, 3 ¼ distinct) was addressed. For size criteria, an LN was regarded as metastatic if the short axis diameter was at least 1 cm or larger and a score was allocated on a three-point scale (1 ¼ 2 cm). DICOM file conversion, MPR reconstructions, ROI analysis, and data visualization was performed using Osirix
1408
Freitag et al.
Table 2 Relationship Between the Nominal and the Actual Flip Angle Parameter k Mean actual flip angles ( )
Sequence
Flip angle ( )
2 cm
3 cm
4 cm
5 cm
— T1w 3D FLASH T1w Dixon VIBE T2w TSE
— 15 13 120
1.0 6 0.2 15.0 6 3.0 13.0 6 2.6 120.0 6 24.0
0.8 6 0.2 12.0 6 3.0 10.4 6 2.6 96.0 6 24.0
0.7 6 0.2 10.5 6 3.0 9.1 6 2.6 84.0 6 24.0
0.6 6 0.1 9.0 6 1.5 7.8 6 1.3 72.0 6 12.0
Calculated from five participants with B1 maps as given in Table 1. The ROI-based analysis was performed in the axially oriented B1-map beginning 2 cm below the skin surface down to 5 cm. As a reference, muscle tissue close to the LN under investigation was chosen over fat tissue, since fat tissue is more heterogeneous (fat, LN, vessels) and the B1-mapping sequence is not optimized for fat due to the frequency-selectivity of the magnetization-preparation pulse. Kappa k ¼ Mean relative flip angle.
(17). The 3D virtual reconstruction in Fig. 6c was created using Fovia Software (Palo Alto, CA). Additional LN Work-up The rough position of LN metastasis was obtained from US-imaging employing a 14 MHz linear array (Siemens Accuson, Erlangen, Germany) prior to the MRI examination to guide later coil positioning. Histological workup was performed for all LNs including hematoxylin-eosin (tumor), D2–40 (lymphoid tissue), and CD31 (blood vessels) stainings. RESULTS Scan Depth As illustrated in the colored B1 map (Fig. 1b1), the obtained flip angles were sufficient at the height of inguinal LNs, usually located between 1–4 cm below the skin surface, up to the target depth of 5 cm (Table 2), where the femoral vessels represented the dorsal margin of the FOV.
Sequence Evaluation In the first two patients (patients #1 and #2), the T2w TSE sequence could not be performed due to incompliance (back pain). All other participants did not report any discomfort and the exam was well tolerated. Slight SNR/CNR variations across space due to B1-inhomogeneity were sometimes visible but did not influence the diagnostic interpretation up to the target depth of 5 cm. Severe motion artifacts
were not present. The use of the loop coil resulted in a characteristic cone-shaped FOV. Optimized sequence parameters are given in Table 1. In one control, fat suppression artifacts were noticed but these were not located in the area of interest. The image quality of the Dixon volumetric interpolated breath-hold examination (VIBE) water image was comparable to the 3D FLASH, with the substantially lower acquisition time favoring the VIBE sequence (Table 1). Delineation of Mesoscopic LN Architecture Muscle, artery, LNs, and fat could be visualized with excellent contrast to the adjacent tissue and sufficient SNR (Fig. 2b). In controls, mesoscopic details such as the saturated fatty hilus and LN cortex (Fig. 3a–d) could be observed. Within the FOV 6.6 6 0.5 LNs were seen, of which 87.9 6 12.6% had a fat hilus sign. LNs were always connected within an intricate and dense network of blood (Fig. 4a,b) and lymph (Fig. 4c,d) vessels. In all depicted LNs, blood vessels could be identified as inserting into the hilus region and connected to a larger vessel distal from the LN (Fig. 3a,b). In contrast, lymph vessels were identified as vessels inserting into the LN periphery, only clearly visible at 0.2 mm 0.2 mm in-plane resolution using the T2w TSE sequence (Fig. 4c,d). These vessels could also be identified in T1w sequences, but spatial resolution of the T1w sequences (0.5 mm3 isotropic) was too low compared to T2w (0.2 0.2 2 mm3) to clearly verify the peripheral connection. In general, blood vessels were T2w hypointense due to the flow-void effect (Table 1, Fig. 3d), but lymph vessels were observed to be T2w
Figure 3. 7T LN MRI. A prominent lymph node (10 mm 40 mm) of a healthy participant is depicted using the parameters for the 3D FLASH (a), T1w Dixon VIBE water image (b), T1w Dixon VIBE fat image (c), and T2w TSE (d) given in Table 1. The hilus region (red arrow), a blood vessel to the hilum (white arrow), and the sartorius muscle (white arrowhead) are marked. Note that the T1w hyperintense venous vessel (white arrow) is T2w hypointense due to the flow-void effect. R: right. L: left.
In Vivo 7T Lymph Node MRI
1409
Figure 4. Visualization of mesoscopic LN vasculature in four different healthy individuals. The LN is connected within a dense network of blood (a,b) and lymph vessels (c,d). Blood vessels (white arrows in a,b) insert into the LN hilus region, whereas lymph vessels (open arrows in c,d) are identified by insertion into the periphery of the LN and by T2w hyperintensity. Arteries and veins cannot be differentiated. a: The connection of the LNs with surrounding blood vessels and the visualization of efferent and afferent vessels (white arrows) to the LN is shown (maximum intensity projection, 12 mm). b: Visualization of very small blood vessels (white arrows) supplying the hilum region. c,d: T2w hyperintense vessels (open arrows) inserting into the lymph node periphery correspond to lymph vessels with diameters between 350 and 500 mm.
hyperintense (Fig. 4c,d), as the flow rate is much lower in the lymphatic compared to the vascular system. LN metastases were pathologically enlarged in five of six patients and considered as macro-metastases (mean size of the short axis [2.8 6 0.4] cm). In one patient, the short axis was not pathologically enlarged
(9 mm). Five of six patients showed LNs with hypervascularization in 7T MRI, of which four LNs had aberrant vessels stemming from smaller subvessels of the femoral vessels (exemplary Fig. 5a) and one LN had multiple large feeding vessels from the femoral artery (Fig. 6). Three patients showed LNs with
Figure 5. Direct visualization of neoangiogenesis, lymphangiogenesis, and tumor bulk at 7T. Histopathologically confirmed LN metastasis of a 60-year-old female patient with a history of superficial spreading melanoma of the lower extremity presenting with swelling in the left groin. Besides two macroscopic signs of malignancy (3 cm short-axis diameter and loss of the fatty hilum), the T1w Dixon VIBE maximum intensity projection (4 mm) depicts strong hilus perfusion (open arrow) and peripheral hypervascularization (a1) correlating to the histopathology CD31 staining (a2) and suggesting malignant neoangiogenesis. Red arrows denote blood vessels. T2w images show pronounced peripheral lymph vessels (b1, yellow arrows) which insert into the peripheral lymph node capsule as depicted in the D2–40 staining for lymphoid tissue (b2). Furthermore, hyperintense intranodal inhomogeneity is visible in T2w images (c1) correlating with tumor tissue in the hematoxylin-eosin staining (c2), as illustrated by the dotted line. The rectangle in a1, b1, and c1 defines the area where histopathological workup was performed.
1410
Freitag et al.
Figure 6. Visualization of neoangiogenesis. A 45-year-old female patient with superficial spreading melanoma of the left lower abdomen presenting 6 years after diagnosis with a definite LN macrometastasis in the left inguinal region. This example demonstrates the visualization of two large feeding vessels from the femoral artery (red arrow and white triangle). a: T1w 3D FLASH. b: Maximum intensity projection (7 mm) of 3D FLASH. c: 3D virtual reconstruction technique (VRT) of (a) with highlighted neoangiogenesis and vessel connection to the femoral artery (asterisk). Note the intranodal tissue inhomogeneity in the metastatic LN (white arrow). R: right; L: left.
emphasized lymph vessels inserting into the LN periphery (exemplary Fig. 5b). All patients presented with intranodal tumor bulk, most distinct in the T2w TSE sequence (compare Fig. 5a1 vs. 5b1 and 5c1). The results of pathological LN signs are shown in Table 3. The diagnosis of nodal metastasis was confirmed by histopathology in all patients.
DISCUSSION We present a series that shows that in vivo 7T MRI examination of superficial and deep inguinal LNs using high spatial resolution is feasible. Earlier ex vivo reports (18,19) about dissected LN specimens used high-resolution T1w gradient-echo sequences employing a 0.18 mm3 isotropic resolution, albeit using long, clinically unacceptable acquisition times (3–5 hours). However, their results corroborated that a clinically feasible examination protocol of LNs with high spatial resolution could have significant impact on cancer diagnostics, since detailed information on the LN micro-architecture such as vessels may be delineated and, thus, high-resolution 7T MRI might serve as a diagnostic method for the identification of small LN metastasis (18,19). Inguinal LNs are embedded in fat pads adjacent to the large femoral vessels. Therefore, fat saturation is necessary to maximize the contrast between LN and fat tissue. Fat-water separation by Dixon VIBE is achieved by acquisition of multiple images with defined fat-water phase differences, whereas 3D FLASH employs a frequency-selective saturation prepulse with a center frequency corresponding to that of lipids (20). The image quality and SNR of both T1w sequences with identical scan parameters was comparable, but the Dixon VIBE provides two compelling advantages in comparison to the 3D FLASH: the possibility to generate separate fat and water maps accompanied by a substantially lower scan time. The fat map may be used to help identify the presence of the fat hilus, a potential sign of benignity.
A healthy LN is typically only perfused by blood vessels inserting in the fatty hilus region. In contrast, peripheral vascularization is suggestive for malignancy and typical for metastases of malignant melanoma (21–23). This sign was also noticed at 7T in five of six included patients. Neoangiogenesis evolves at 2 mm metastasis size (24) and, if untreated, may lead to a tumor with multiple feeding vessels from large arteries. This hypervascularization is directly visible as T1w hyperintense vessels. From a previous study in LN metastases of breast cancer, neovascularization is a known predictor for overall and diseasefree survival (25). Thus, the direct imaging-based identification of pathological tumor vessels in LNs provides additional information about the progression of the disease and its prognosis, as multiple vessel connections are indicative of higher probability of further hematogeneous spread to central organs. Furthermore, 7T MRI was able to visualize smallest vessels with diameters between 350–500 mm inserting into the LN periphery using the T2 TSE sequence. These small vessels correspond to lymph vessels according to the normal anatomy of a LN, whereas blood vessels commonly insert via the fat hilus region (26). We have provided evidence for the depiction of this intricate network of healthy lymph vessels in vivo without the use of contrast media. It has been reported that lymphangiogenesis in the LN metastasis, similar to angiogenesis, is a prognostic factor for local and distant cancer metastasis (27). The presence of lymphangiogenesis or dilated lymph vessels due to metastatic tumor load in the LN could represent valuable diagnostic markers. In accordance, we did observe pronounced lymph vessels in three of six patients. This study has some limitations. First, the MR images were screened in consensus between two radiologists, which may potentially result in reading bias. However, in this study we did not aim to assess the diagnostic accuracy. By consensus reading we aimed to assure the optimal selection of pathological LNs and the proper description of the imaging findings.
#6 male
#5 male
#4 (female, 60y)
#2 (female, 44y) #3 (male, 73y)
The rating scale is defined as: 0 (not present), 1 (suggestive), 2 (visible), 3 (distinct). Current tomography criteria for LN pathology (enlargement, presence of fat hilus) are extended at 7T MRI by direct visualization of blood and lymph vessels and intranodal tumor inhomogeneity at high spatial resolution. *The T2w TSE sequence in patients #1 and #2 was not assessable due to reduced compliance at the end of the examination.
1/3 1/3 3/3 0/3 1
0/3
3/3 3/3 3/3 1/3 3
3/3
3/3 3/3 3/3 3/3 1
3/3
3/3 3/3 3/3 3/3 3/3 3/3 3/3 2/3 1 7
0/3* 2/3
3/3 3/3 3/3 3/3 2
0/3*
1411
Superficial spreading melanoma (pT1bN3cM0) Nodular melanoma (pT3bN1M0) Resected melanoma metastasis of the left thigh with unknown primary (TxN3M1) Superficial spreading melanoma (T3N1M0) Superficial spreading melanoma (pT4bN2M0) Acro-lentiginous melanoma (pT3bN1aM0) #1 (female, 45y)
Diagnosis (tumor stadium) Patients
Table 3 Signs of LN Malignancy in the Included Patients at 7T MRI
Number of LN metastases
Peripheral blood vessel insertion
Pronounced lymph vessels visible
LN tumor bulk (T1w, T2w)
Loss of fat hilus
Enlargement (1: < 1 cm; 2: 1–2 cm; 3: > 2 cm short-axis diameter)
In Vivo 7T Lymph Node MRI
Second, the number of included patients was small and thus some imaging features that were clearly visible in this small sample may not be as prominent in a larger group of patients. In conclusion, hallmarks of cancer such as neoangiogenesis and lymphangiogenesis may be directly visualized in vivo with the proposed protocol. This could be of critical importance considering the differentiation between inflammation and metastasis. In future studies, there are several promising MRI-based techniques that could optimize the proposed protocol for LN imaging at 7T such as contrast-enhanced dynamic sequences or time-of-flight angiography. Another very interesting approach at high resolution would be the application of contrast medium to achieve contrast enhancement within the LN body, either by shortening the T1 relaxation times (eg, gadolinium) or by exploiting the T2* effect (eg, iron nanoparticles), which could, analogous to findings of previous studies at lower resolution (28), help to identify LN metastases at an earlier timepoint. Besides its usefulness for tumor diagnostics, the use of gadolinium-based agents could further help to increase the obtained SNR and CNR and thus even further improve the image quality or target resolution, as LNs tend to take up this contrast media. However, dynamic contrast-enhanced sequences typically have lower spatial resolution compared to the standard static T1w sequences to calculate perfusion biomarkers, in favor of a higher temporal resolution. As described in the introduction, further advances in noninvasive LN imaging are urgently needed. For this purpose, MRI at high resolution, particularly considering that it is a radiation-free tomographic method, could be considered as an alternative or additional modality. Our proposed setup serves as a first step to increase the spatial resolution in superficial LNs with a transmit/receive coil in the inguinal region. To achieve broader clinical application, further coil improvement is necessary, especially to access the axilla and cervical region. High-risk patients, unclear nodal status in patients undergoing radiotherapy, or patients undergoing regular nodal staging may substantially benefit from a complementary high-resolution MRI. However, the number of patients included in the present study was limited, as feasibility was the main purpose. Therefore, a detailed systematic study using our proposed protocol is now warranted to determine whether high-resolution LN imaging at 7T really improves the sensitivity/specificity in comparison to lower field strengths or other modalities.
CONFLICT OF INTEREST The authors declare no conflict of interest.
ACKNOWLEDGMENTS We thank Siemens (Erlangen, Germany) for providing the 1H transmit/receive loop-coil. We also thank Jessica Engelhart and Vanessa Peregovich for their kind assistance.
1412
REFERENCES 1. McMahon CJ, Rofsky NM, Pedrosa I. Lymphatic metastases from pelvic tumors: anatomic classification, characterization, and staging. Radiology 2010;254:31–46. 2. de Boer M, van Deurzen CH, van Dijck JA, et al. Micrometastases or isolated tumor cells and the outcome of breast cancer. N Engl J Med 2009;361:653–663. 3. Balch CM, Soong SJ, Gershenwald JE, et al. Prognostic factors analysis of 17,600 melanoma patients: validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol 2001;19:3622–3634. 4. Gershenwald JE, Thompson W, Mansfield PF, et al. Multi-institutional melanoma lymphatic mapping experience: the prognostic value of sentinel lymph node status in 612 stage I or II melanoma patients. J Clin Oncol 1999;17:976–983. 5. Morton DL, Thompson JF, Cochran AJ, et al. Sentinel-node biopsy or nodal observation in melanoma. N Engl J Med 2006; 355:1307–1317. 6. Klode J, Dissemond J, Grabbe S, Hillen U, Poeppel T, Boeing C. Sentinel lymph node excision and PET-CT in the initial stage of malignant melanoma: a retrospective analysis of 61 patients with malignant melanoma in American Joint Committee on Cancer stages I and II. Dermatol Surg 2010;36:439–445. 7. Chai CY, Zager JS, Szabunio MM, et al. Preoperative ultrasound is not useful for identifying nodal metastasis in melanoma patients undergoing sentinel node biopsy: preoperative ultrasound in clinically node-negative melanoma. Ann Surg Oncol 2011:1100–1106. 8. Lee B, Lim AK, Krell J, et al. The efficacy of axillary ultrasound in the detection of nodal metastasis in breast cancer. AJR Am J Roentgenol 2013;200:W314–320. 9. Bellin MF, Lebleu L, Meric JB. Evaluation of retroperitoneal and pelvic lymph node metastases with MRI and MR lymphangiography. Abdom Imaging 2003;28:155–163. 10. Schipper RJ, Smidt ML, van Roozendaal LM, et al. Noninvasive nodal staging in patients with breast cancer using gadofosvesetenhanced magnetic resonance imaging: a feasibility study. Invest Radiol 2013;48:134–139. 11. Acland KM, Healy C, Calonje E, et al. Comparison of positron emission tomography scanning and sentinel node biopsy in the detection of micrometastases of primary cutaneous malignant melanoma. J Clin Oncol 2001;19:2674–2678. 12. Balch CM, Gershenwald JE, Soong SJ, et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol 2009;27:6199–6206. 13. Torjesen I. Sentinel node biopsy for melanoma: unnecessary treatment? BMJ 2013;346:e8645.
Freitag et al. 14. Estourgie SH, Nieweg OE, Valdes Olmos RA, Hoefnagel CA, Kroon BB. Review and evaluation of sentinel node procedures in 250 melanoma patients with a median follow-up of 6 years. Ann Surg Oncol 2003;10:681–688. 15. Morton DL, Cochran AJ, Thompson JF, et al. Sentinel node biopsy for early-stage melanoma: accuracy and morbidity in MSLT-I, an international multicenter trial. Ann Surg 2005;242: 302–311. 16. Kraff O, Fischer A, Nagel AM, Monninghoff C, Ladd ME. MRI at 7 Tesla and above: demonstrated and potential capabilities. J Magn Reson Imaging JMRI 2014 [Epub ahead of print]. 17. Rosset A, Spadola L, Ratib O. OsiriX: an open-source software for navigating in multidimensional DICOM images. J Digit Imaging 2004;17:205–216. 18. Korteweg MA, Zwanenburg JJ, Hoogduin JM, et al. Dissected sentinel lymph nodes of breast cancer patients: characterization with high-spatial-resolution 7-T MR imaging. Radiology 2011; 261:127–135. 19. Korteweg MA, Zwanenburg JJ, van Diest PJ, et al. Characterization of ex vivo healthy human axillary lymph nodes with high resolution 7 Tesla MRI. Eur Radiol 2011;21:310–317. 20. Delfaut EM, Beltran J, Johnson G, Rousseau J, Marchandise X, Cotten A. Fat suppression in MR imaging: techniques and pitfalls. Radiographics 1999;19:373–382. 21. Voit C, Kron M, Schafer G, et al. Ultrasound-guided fine needle aspiration cytology prior to sentinel lymph node biopsy in melanoma patients. Ann Surg Oncol 2006;13:1682–1689. 22. Thompson JF, Haydu LE, Sanki A, Uren RF. Ultrasound assessment of lymph nodes in the management of early-stage melanoma. J Surg Oncol 2011;104:354–360. 23. Torabi M, Aquino SL, Harisinghani MG. Current concepts in lymph node imaging. J Nnucl Med 2004;45:1509–1518. 24. Mihic-Probst D, Ikenberg K, Tinguely M, et al. Tumor cell plasticity and angiogenesis in human melanomas. PloS One 2012;7: e33571. 25. Guidi AJ, Berry DA, Broadwater G, et al. Association of angiogenesis in lymph node metastases with outcome of breast cancer. J Natl Cancer Inst 2000;92:486–492. 26. Clement O, Luciani A. Imaging the lymphatic system: possibilities and clinical applications. Eur Radiol 2004;14:1498–1507. 27. Rinderknecht M, Detmar M. Tumor lymphangiogenesis and melanoma metastasis. J Cell Physiol 2008;216:347–354. 28. Harisinghani MG, Barentsz J, Hahn PF, et al. Noninvasive detection of clinically occult lymph-node metastases in prostate cancer. N Engl J Med 2003;348:2491–2499. 29. Gray H, Lewis WH. Anatomy of the human body. Philadelphia and New York: Lea & Febiger: 1918.
