REVIEW ARTICLE
Positron Emission Tomography/Magnetic Resonance Imaging Evaluation of Lung Cancer Current Status and Future Prospects Soon Ho Yoon, MD,*w Jin Mo Goo, MD, PhD,*wz Sang Min Lee, MD,*w Chang Min Park, MD, PhD,*wz Hyo Jung Seo, MD,y and Gi Jeong Cheon, MD, PhDzy
Abstract: Various designs of positron emission tomography/magnetic resonance imaging (PET/MRI) systems have been recently introduced to clinical practice, which have overcome preexisting technical challenges concerning the fusion of PET and MRI systems. Although further improvements are still necessary especially for bony lesions, quantification using current MRI-based attenuation correction techniques has been shown to be comparable to that of PET/computed tomography (CT) systems. On the basis of the results of previous whole-body MRI studies, PET/MRI is expected to show even better performance than PET/CT in M-staging especially for brain and liver metastases. Another advantage of PET/MRI over PET/CT, in addition to good soft tissue contrast, is the potential reduction in radiation dose. The next important hurdle to overcome for its clinical application is the development of time-efficient protocols for lung cancer evaluation and interpretation of discordant results from both modalities. Multiparametric imaging through PET/MRI will help radiologists better understand tumor biology and better evaluate treatment response. Key Words: positron emission tomography, magnetic resonance imaging, positron emission tomography and computed tomography, lung cancer, review
(J Thorac Imaging 2014;29:4–16)
I
n addition to improvements in imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI), advances in combined multimodality imaging technologies such as positron emission tomography (PET)/CT have improved clinical patient care especially in oncologic imaging. Since their introduction in 2000,1 PET/CT scanners have become widely implemented and have enormously contributed to the diagnosis, treatment, and prediction of prognosis in oncologic patients.2 Thereafter, a number of studies including a randomized controlled trial on the use of PET/CT for preoperative staging of non–small cell lung cancers have proven the superior efficacy of PET/CT in cancer staging over conventional staging methods.3–6 At present, PET/CT has
From the Departments of *Radiology; yNuclear Medicine; zCancer Research Institute, Seoul National University College of Medicine; and wInstitute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Korea. Supported in part by a grant from Guerbet Korea. The authors declare no conflicts of interest. Reprints: Jin Mo Goo, MD, PhD, Department of Radiology, Seoul National University College of Medicine, 101 Daehangno, Jongnogu, Seoul 110-744, Korea (e-mail: [email protected]). Copyright r 2013 by Lippincott Williams & Wilkins
4 | www.thoracicimaging.com
become a standard imaging modality in lung cancer staging, and its cost-effectiveness has been validated.7 Despite the many advantages of MRI, including excellent soft tissue contrast resolution and absence of radiation exposure, it has been an underutilized imaging modality for evaluating thoracic diseases, as considerable challenges in MRI acquisition for the lung still exist because of low proton density, inhomogeneity of the magnetic file in the lung, and cardiac and respiratory motion artifacts.8 With recent technical advances such as fast imaging and parallel imaging techniques in MRI, previous limitations have been partially overcome. In addition, new methods using either a moving-table platform in combination with the body coil or a specially designed rolling-table platform with 1 body phased-array coil have made it feasible to perform whole-body MRI for lung cancer staging.9–11 Considering the success of PET/CT in oncology practice and the innate limitations of PET/CT in terms of soft tissue contrast resolution and radiation exposure, a fusion of PET and MRI has garnered much attention as it is a conjunction of molecular information offered by PET and anatomic/ potentially functional information with high contrast and spatial resolution provided by MRI.12 We can anticipate the potential strengths and limitations of PET/MRI as this modality has components of both PET and MRI. The various kinds of radioisotopes used in PET can depict several important biological properties including glucose metabolism, tissue hypoxia, perfusion, and apoptosis; and MRI can yield images reflecting cellular density, perfusion, hypoxia, and metabolic features in tissue in addition to basic T1-weighted and T2-weighted MRI images offering high-resolution anatomic information with different contrasts according to the evolution of several MRI sequences. In addition to these potential advantages, PET/MRI also has an important advantage over PET/CT in terms of radiation dose. Although the underlying technical obstacles have not yet been fully resolved, various designs of PET/MRI systems including fully integrated hybrid scanners have been recently implemented in clinical practice. In this review, we discuss the instrumentation, technical aspects, results of studies obtained with PET/CT and whole-body MRI, recent results with integrated PET/ MRI, and potential applications related to PET/MRI in lung cancer.
HARDWARE DESIGN OF PET/MRI SYSTEMS An ideal combination of PET and MRI is a hybrid system in which PET and MRI signals can be acquired simultaneously. However, the integration of PET and MRI systems within a single gantry has innate technical J Thorac Imaging
Volume 29, Number 1, January 2014
J Thorac Imaging
Volume 29, Number 1, January 2014
difficulties, as the signal acquisition process of the 2 modalities is completely different causing interference between the 2 systems.13 Indeed, PET detectors can be interfered with by the strong B0 static magnetic field, abrupt changes in the B1 gradient magnetic field, and radiofrequency pulse during the signal acquisition of an MRI system. Conversely, MRI detectors can also be affected by electromagnetic radiation from PET electronics, magnetic field inhomogeneity derived from PET devices, and eddy currents. Unless these interferences can be resolved, the simultaneous acquisition of PET and MRI images will result in significant degradation of image quality. There are 2 major strategies in the design of a PET/ MRI system adopted by vendors (Fig. 1). One is the tandem arrangement, and the other is the concurrent arrangement of PET and MRI scanners. The tandem arrangement enables sequential acquisition of PET and MRI signals without major modification to PET and MRI scanners, which are located to avoid mutual interference either in the same
PET/MRI Evaluation of Lung Cancer
space or in 2 separate spaces. Ingenuity TF (Philips Healthcare, Best, Netherlands) operates PET and MRI scanners with back-to-back configurations in the same room. In the case of the Discovery PET/CT and MRI system (GE Healthcare, Milwaukee, WI), PET/CT and MRI scanners are combined and sequentially performed in 2 separate rooms. Compared with the concurrent type of arrangement, PET/MRI systems with tandem arrangement have demerits not only in the potential anatomic mismatch, which may occur due to patient motion, but also in temporal mismatch, which may occur due to the sequential acquisitions of PET and MRI images. Biograph mMR (Siemens Healthcare, Erlangen, Germany) adopted a fully integrated PET/MRI system using the concurrent configuration. One of the biggest technical difficulties in the integration of PET/MRI systems is related to the photosensor of the PET detector, as the photomultiplier tube traditionally used in PET or PET/CT systems is very sensitive to the magnetic field. An introduction of avalanche photodiodes, which are insensitive to magnetic fields, has enabled the development of hybrid PET/MRI scanners such as Biograph mMR, although avalanche photodiodes are inferior to photomultiplier tubes in terms of gain and temporal resolution. The implementation of silicon photodiodes (Geiger-mode avalanche photodiodes) in hybrid PET/MRI scanners is currently underway as silicon photodiodes have higher gain and faster temporal resolution enabling time-of-flight PET in addition to the advantages of avalanche photodiodes.14 With the technical evolution regarding PET, MRI scanners, and RF shielding, simultaneous acquisition of PET and MRI signals is now possible without significant interference in Biograph mMR hybrid scanners.15 Simultaneous acquisition provides better anatomic and temporal registration of MRI and PET signals allowing more robust interpretation of disease biology.16 Software-based image registration is another option to generate PET/MRI images by fusing separately acquired PET and MRI images.17 Several computation algorithms fusing PET and MRI images have been proposed with varying degrees of success in registration.18 The trimodality system, which combines PET/CT and MRI systems located in separate rooms, has shown a weakness with regard to the radiation dose, as the CT scanner of PET/CT is used for image acquisition. Comparison of features of integrated, sequential, and coregistered PET/MRI systems is detailed in Table 1.
TECHNICAL ISSUES IN PET/MRI MRI-based Attenuation Correction
FIGURE 1. Schematic diagrams of current commercial wholebody PET/MRI systems. A, Tandem arrangement of PET and MRI scanners in the same room (Ingenuity TF; Philips Healthcare). B, Tandem arrangement of PET/CT and MRI scanners in separate rooms (Discovery PET/CT and MRI system; GE Healthcare). PET and MRI signals are sequentially acquired in the tandem systems. C, Concurrent arrangement of PET and MRI scanners wherein simultaneous acquisition of PET and MRI signals are possible (Biograph mMR; Siemens Healthcare). PET detector composed of avalanche photodiodes is integrated between radiofrequency body coil and gradient, primary magnet coils. r
2013 Lippincott Williams & Wilkins
Attenuation correction is essential for the quantification of PET data as some portion of photons emitted from radiotracers is variably absorbed across different body tissues before arrival at the PET detector. To correct the attenuation of emitted photons throughout the body, an attenuation map with information regarding the distribution of attenuation coefficients is needed. Contrary to the PET/CT system in which the CT attenuation coefficient is linear to electron density, MRI signal intensity in the PET/MRI system has no direct relationship to electron density. Several approaches deriving attenuation coefficients from MRI images have been proposed,19 and routine applications of MRI-based attenuation corrections for whole-body PET/MRI are based on the acquisition of www.thoracicimaging.com |
5
J Thorac Imaging
Yoon et al
TABLE 1. Comparison of Integrated, Sequential, and Coregistered PET/MRI Systems
Integrated Sequential Image Coregistration PET/MRI PET/MRI in Separate PET and System System MRI Systems Simultaneity in PET and MRI acquisition Accuracy of PET/MRI registration PET attenuation correction Time-of-flight PET Flexibility of scheduling
High
Intermediate
Low
High
Intermediate
Low
MRI-based MRI-based
CT-based
Not Possible possible* Less Less flexible flexible
Possible More flexible
*Avalanche photodiode used in Biograph mMR does not support timeof-flight PET. Time-of-flight PET may be applicable in integrated PET/MRI systems using silicon photodiodes in the future.
dedicated MRI sequences such as the DIXON sequence with in-phase/opposed-phase images. The Dixon MRI sequence provides separate water and fat images and creates an attenuation map (m map) with 4 distinct tissue classes: background, lung, fat, and soft tissue (Fig. 2). The lungs are identified by connected-component analysis of the air in the inner part of the body. In addition, difficulty in differentiating air, lung, and bone on conventional MRI sequences makes MRI-based attenuation correction more complicated in the PET/MRI system. Importantly, as bony structures are not considered by currently applied MRIbased attenuation correction methods, these structures undergo a systematic undercorrection of PET data. Maximum errors of decreased uptake in bone are up to 13% or 17% in standardized uptake value (SUV).20–22 Another option for attenuation correction in the PET/MRI system is Water
Fat
Volume 29, Number 1, January 2014
the use of an anatomic atlas template to coregister patient data sets. Contrary to easier registration in the brain, the atlas-based method itself remained challenging for the whole body. Recently, the atlas-based method in conjunction with regional segmentation derived from the MRI sequence was reported to be technically feasible and was shown to provide better results than pure regional segmentation using the MRI sequence.23 Despite this source of error in SUV, recent studies using an integrated PET/MRI system revealed that the PET image quality was subjectively rated high and that the diagnostic value was comparable to that of standard PET/ CT systems.24,25 In the clinical application of PET wherein the diagnosis is often based on qualitative evaluation, accuracy in quantification may not be crucial as far as these errors do not affect image quality (Fig. 3).26 However, PET/ MRI-based therapy response assessment in oncologic disease requires accurate PET quantification, and therefore further improvement in MRI-based attenuation correction is necessary.
Alignment Precise alignment is necessary to characterize pathologic findings when combining different imaging modalities. However, bulk patient motion or various physiological motions of respiration, bowel movements, or urinary bladder filling can cause misalignment in hybrid imaging. Anatomic alignment is expected to improve with simultaneous acquisition of PET/MRI compared with retrospectively fused PET and MRI. A recent study dealt with this issue and confirmed that the alignment of hybrid data sets acquired in simultaneous whole-body PET/MRI was more accurate than retrospective fusion in abdominal organs.27 In the urinary bladder, the alignment of simultaneous PET/MRI was more accurate than PET/CT. When the effect of different breathing protocols of simultaneous, sequential, and MRI-gated data acquisition was compared, the alignment of thoracic PET/MRI with expiratory breath-hold or free-breathing MRI was more accurate than with inspiratory MRI.
In phase
Opposed phase
µMap
FIGURE 2. MRI-based attenuation correction using the Dixon sequence. MRI water and fat images are generated by in-phase/opposedphase images. These water and fat images are combined and segmented to produce the MRI-based attenuation correction map (m map).
6 | www.thoracicimaging.com
r
2013 Lippincott Williams & Wilkins
J Thorac Imaging
Volume 29, Number 1, January 2014
PET/MRI Evaluation of Lung Cancer
FIGURE 3. SUVmax of lung cancer at PET/MRI and PET/CT. PET/MRI and PET/CT were performed in a 58-year-old woman with adenocarcinoma. A, CT shows a 27 mm irregular nodule (arrow) in the right upper lobe. B, PET/CT was obtained 1 hour after (C) PET/MRI. The SUVmax of lung cancer at PET/MRI and PET/CT (arrows) were 3.01 and 5.75. The SUVmax at PET/MRI was 47.7% lower than that at PET/CT.
Other Sources of Artifacts and Errors In calculating SUV, other sources of attenuation, such as surface coils, and additional positioning devices must be considered. These devices can cause scattering and attenuation of the PET signal. With regard to MRI contrast agents, ingestion of an oral iron oxide–based MRI contrast agent has been shown to affect the MRI-based attenuation map even though PET quantification is not affected by intravenous injection of the MRI contrast agent.28 In addition, metallic implants can produce a signal loss resulting in underestimated uptake in the region surrounding the metallic implants, and truncation artifacts may arise from the difference in the transverse fields of view
of PET and MRI (Fig. 4). To recognize attenuation-related artifacts, it is important to evaluate the non–attenuationcorrected images. Partial volume effects are another source of potential error that can affect the quantitative accuracy of PET. SUV can be underestimated with decreasing tumor volume, and therefore it can be affected by applying different partial volume correction methods.29 Thus far, studies have shown that reconstruction-based partial volume correction, which includes 3-dimensional (3D) point spread function, outperforms image-based methods in terms of accuracy.29 To reduce motion artifacts, physiological gating such as electrocardiogram gating and MRI-based triggering of
FIGURE 4. Truncation artifact at PET/MRI. A, PET images and (B) T1-weighted volumetric interpolated breath-hold examination coronal images are used for image fusion (C). As the transverse field of view (FOV) of MRI and PET is 50 and 59 cm, respectively, there are some parts of arms beyond the MRI-FOV. This results in underestimation of the PET image information along the arms (arrows) and is called truncation artifact (C). r
2013 Lippincott Williams & Wilkins
www.thoracicimaging.com |
7
Yoon et al
PET data can be applied. One requirement in applying these techniques is to consider the associated increase in examination time. An alternative approach to reducing respiratory motion–related errors is the MRI-based respiratory motion correction of PET that can improve tumor visibility, delineation, and maximum SUV.16
Examination Time Unlike PET/CT, the design of PET/MRI protocols for whole-body coverage is more complex, and time efficiency is an especially important issue. PET/MRI can be performed in approximately 25 minutes without further morphologic MRI. However, with a number of additional diagnostic sequences and with the use of an intravenous contrast, the examination time may increase up to 90 minutes or more. With MRI sequences such as diffusion, perfusion MRI, and MRI spectroscopy, more functional and biological properties can be obtained at the cost of a longer examination time. Therefore, the numbers and types of MRI sequences required need to be specifically chosen and optimized according to the clinical tasks of PET/ MRI.30 A continuous moving-table data acquisition scheme can be applied in postcontrast studies.31
IMAGE ACQUISITION WITH AN INTEGRATED PET/MRI SYSTEM As most clinical studies on PET/MRI were carried out with the integrated PET/MRI system, this section regarding image acquisition is based on integrated PET/MRI. Most differences in designing PET/MRI scanning protocols come from the MRI protocols. Optimizing the PET/MRI protocol is an important issue, which is under active investigation. Due to time constraints, especially for wholebody imaging, priority among various MRI sequences should be considered in adding dedicated MRI for specific body parts. This issue is essentially the same for sequential PET/MRI systems except that they need more examination time to perform the same protocol. The MRI part in PET/ MRI plays a role for anatomic localization, attenuation correction of PET data, and tissue characterization based on selected MRI techniques.
Patient Preparation Before examination, in the case of administration of F-fluoro-2-deoxyglucose (18F-FDG), fasting, control of glucose level, and having the patient rest between 18F-FDG administration and imaging to minimize muscle uptake are required for patient preparation. In addition to preparations for stand-alone MRI, care should be taken so that only dedicated coils approved for PET/MRI are used, as the surface coils may cause additional attenuation of 511 keV photons. 18
PET/MRI Acquisition Protocols As the PET/MRI scan protocol for the staging of lung cancer has not been standardized as yet, the general principles for PET/MRI scans and the specific scan protocol in our center will be introduced. The predominant factors for PET/MRI scan time are the number and type of the MRI sequence. Despite the variability of whole-body MRI scan protocols across the literature,3,9–11,32,33 a whole-body MRI scan can be performed with a combination of a coronal or axial fast T2-weighted sequence (eg, half-Fourier acquisition single-shot turbo spin-echo sequence), fast T1weighted sequence (eg, volume-interpolated 3D-spoiled
8 | www.thoracicimaging.com
J Thorac Imaging
Volume 29, Number 1, January 2014
gradient-echo sequence), short-t inversion recovery (STIR) sequence, and diffusion-weighted MRI sequence covering organs in which metastasis frequently occurs. In addition, contrast-enhanced MRI scans are essential for better detection of metastases in solid organs including the brain, liver, and leptomeninges.34–36 The initial step is obtaining MRI localizers to plan the subsequent acquisition. At each bed position, while PET imaging is acquired, the MRI sequence for attenuation correction, which takes 19 seconds per bed position, is acquired first, followed by subsequent simultaneous MRI sequences for diagnostic purposes. As the axial range of a single bed position is 25.8 cm with a 6.1 cm overlap between adjacent bed positions in the current integrated PET/MRI system, 2 to 5 bed positions are required for lung cancer staging to cover a body part or the whole body. MRI for a specific region of interest can be acquired additionally with or without simultaneous PET scanning according to diagnostic purposes. With regard to PET scan, the number of bed positions and the acquisition time for each should be determined for PET imaging, which is obtained in the step-and-shoot mode. PET acquisition time is 2 to 4 minutes per bed position, but this can be increased for simultaneous MRI acquisitions. The dedicated PET/MRI scan protocol for lung cancer preliminarily used in our center was developed for the purpose of both clinical practice and academic research (Fig. 5) and is expanded from a basic whole-body PET/MR scan protocol (Fig. 6).
FUTURE PROSPECTS ON CLINICAL APPLICATION IN LUNG CANCER Currently, there are very few studies investigating pulmonary malignancies using an integrated PET/MRI machine. Herein, we briefly review the results of preliminary studies based on coregistration of PET and MRI or PET/MRI systems to better comprehend the potential role of integrated PET/MRI in lung cancer.
Detection and Assessment of Pulmonary Nodules At present, CT is the standard and best imaging modality for the detection of pulmonary nodules due to its high spatial and temporal resolution. Despite improvement of MRI techniques, current MRI sequences used in PET/MRI systems are not sufficient for visualization of small lung nodules of 5 mm or less.37–40 The detection rate of MRI with various sequences has ranged between 45.5% and 96.0%, and the detection rate of spin-echo sequences was better than that of gradient-echo sequences.8,37–40 In a prospective comparative study on wholebody MRI and PET/CT, more pulmonary metastases were found on PET/CT than on MRI (170 vs. 139 pulmonary metastases).3 In addition, in a recent study that used an integrated PET/MRI system, radial volumetric interpolated breath-hold examinations and PET acquired with PET/MRI detected 70.3% of all nodules and 95.6% of FDG-avid nodules, and 88.6% of nodules 5 mm in diameter or larger were detected compared with the PET/CT system, which detected 69 nodules including 45 FDG-avid nodules.41 This weakness of PET/MRI in detecting pulmonary nodules might not be problematic on the initial workup of lung cancer staging given that chest CT is routinely performed at initial diagnosis. However, substitution of PET/CT with PET/MRI during follow-up in patients with lung cancer might only be possible in combination with chest CT until better MRI sequences are developed for detection of metastatic nodules in the lung.11,42 r
2013 Lippincott Williams & Wilkins
J Thorac Imaging
Volume 29, Number 1, January 2014
PET/MRI Evaluation of Lung Cancer
A Time (min) 1
0
36
PET
41
BP PET (each 7min)
PET abd. (5min)
51
56
No PET
PET thorax (10min)
CM MRI
AC
Localizer
Cor T1 TSE
Ax Ax T2HASTE VIBE FS
5 bed positions Head to thigh
AC
Ax DWI
1 bed position abdomen
AC
Ax DWI
Ax DC E
Ax VIB E
Ax Ax Ax Ax VIBE VIBE VIBE VIBE
1 bed position Thorax
Head to thigh
B Time (min) 0
1
13
BP PET(each 6 min)
PET
MRI
Localizer AC
Cor STIR
Ax VIBE
2 bed positions Thorax to upper abdomen
FIGURE 5. Examples of whole-body (A) and partial-body (B) acquisition protocols of PET/MRI in lung cancer patients. AC indicates attenuation correction; Ax, axial; BP, bed position; CM, contrast medium; cor, coronal; DCE, dynamic contrast enhancement; HASTE, half-Fourier acquisition single-shot TSE; TSE, turbo spin-echo; VIBE, volumetric interpolated breath-hold examinations.
As PET/CT can also fail to detect small metastatic lung nodules found on chest CT,43 PET/MRI with low-dose CT may be the best option for postoperative surveillance in lung cancer patients. FDG-PET/CT has demonstrated efficacy in differentiating malignant from benign nodules, but this has limited value in the detection of ground-glass nodules and has yielded a considerable number of false-positive results for these nodules, mainly because of inflammation.44,45 On the basis of the concept that malignant lesions demonstrate increased cellularity, high tissue disorganization, and increased extracellular space tortuosity compared with benign lesions, malignant lesions show high signal intensity on diffusion-weighted imaging (DWI) and low apparent diffusion coefficients (ADC). However, granulomas, active inflammation, or fibrous nodules may also show high signal r
2013 Lippincott Williams & Wilkins
intensity, similar to malignant nodules.8 The use of dynamic contrast-enhanced MRI demonstrated high diagnostic capability in distinguishing malignant from benign nodules, with sensitivities ranging from 94% to 100% and specificities from 70% to 96%.8
Assessment of TNM Staging T-Staging T-staging is assessed mainly on the basis of tumor size and invasion of adjacent tissue according to the seventh edition of the TNM staging system for lung cancer.46 When compared with the CT scan, the MRI scan showed higher diagnostic performance in assessing locoregional invasion to the chest wall, mediastinum, and diaphragmatic pleura, as MRI offers clear visualization of the extrapleural fat plane, www.thoracicimaging.com |
9
Yoon et al
J Thorac Imaging
Volume 29, Number 1, January 2014
FIGURE 6. Whole-body PET/MRI protocol performed in a 73-year-old man with lung cancer. PET/MRI starts 1 hour after intravenous administration of 18F-FDG with 7 minutes per bed position. As scan range is from head to lower thigh, the number of bed positions is usually 5 or 6. After gradient-echo localization, a coronal T1-weighted turbo spin-echo sequence is performed (A). Then the T2weighted half-Fourier acquisition single-shot turbo spin-echo axial sequence (B) and the T1-weghted volumetric interpolated breathhold examination axial sequence are obtained. Diffusion-weighted images (b = 50, 400, 800) (C) and apparent diffusion coefficient map (D) are also obtained. Afterward, T1-weighted contrast-enhanced sequencing (E) is performed. Fused PET and MRI images (F, G) are interpreted for evaluation of lesions.
pleural abnormality on T1-weighted and T2-weighted images, chest rib destruction on STIR images, and adhesion between the chest wall and mass on dynamic cine MRI images.47–50 In a study by Plathow et al,51 whole-body MRI obtained with 1.5 T MRI provided correct T-staging in all patients, whereas PET/CT did not correctly stage chest wall invasion in 4 of 52 patients. In another study by Yi et al,11 when PET/CT and whole-body MRI obtained with 3 T MRI were compared, there were no significant differences in T-staging assessment. In a recent study that included 10 patients with pathologically proven or clinically suspected lung cancers, an integrated PET/MRI system excluded the infiltration of the mediastinal pleura by the tumor suspected at PET/CT in 1 patient and led to understaging due to slightly smaller measurement of the tumor size compared with PET/CT.25
10 | www.thoracicimaging.com
N-Staging Although CT is the standard modality in lung cancer staging, the major weakness of CT is its low accuracy in Nstaging because only size is considered to determine metastatic lymph nodes. In comparison, FDG-PET has shown better performance in determining metastatic lymph nodes. MRI is also superior to CT in diagnosing mediastinal lymph node metastasis, as metastatic lymph nodes are able to be differentiated from benign reactive lymph nodes by high signal intensity on the STIR sequence and diffusion restriction on DWI (Fig. 7; Table 2).32,53,54,56 Cardiactriggered and/or respiration-triggered STIR turbo spinecho imaging has been recommended for N-staging, and when compared with coregistered FDG-PET/CT, STIR sequence showed better performance.53 However, in r
2013 Lippincott Williams & Wilkins
J Thorac Imaging
Volume 29, Number 1, January 2014
PET/MRI Evaluation of Lung Cancer
FIGURE 7. Apparent diffusion coefficient (ADC) map on evaluation of lymph node metastasis. A, A 59-year-old man had a 3.1 cm mass (arrow) in the left upper lobe with a 7 mm left lower paratracheal lymph node (arrow) and a 13 mm left upper lobar lymph node (not shown). B, PET/MRI showed a mass (SUVmax: 14.65) in the left upper lobe, a 13 mm lobar node (SUVmax: 12.84), and a 7 mm left lower paratracheal lymph node (SUVmax: 4.5) (arrows). C and D, Diffusion-weighted image (b = 400) and ADC map showed diffusion restriction in the mass in the left upper lobe and 2 lymph nodes (arrows). The concordant findings of left lower paratracheal lymph node on the ADC map and at fused PET/ MRI increased confidence in reporting left lower paratracheal lymph node metastasis, which was further confirmed by pathologic results.
another study by Yi et al,11 there were no significant differences between PET/CT and whole-body MRI in Nstaging assessment. With regard to DWI, Nomori et al56 reported that DWI was significantly more accurate than FDG-PET because of less overstaging and fewer falsepositive results of DWI. When the diagnostic performance of FDG-PET/CT was compared with that of FDG-PET/ CT combined with DWI and T2-weighted MRI for preoperative regional lymph node evaluation, sensitivity increased from 46% to 69% with a slight decrease of
specificity from 96% to 93%.32 This difference is mainly caused by inflammatory lymph nodes in which FDG uptake usually increases while there is no restricted diffusion. However, the low spatial resolution and image distortion of DWI sequences are limitations in detecting some small metastatic lymph nodes.8 Although there may be some potential gains in diagnosing lymph node metastasis with a combination of PET and MRI over PET/CT, it was not evident in preliminary studies.10,25 In a recent study that evaluated 10 patients
TABLE 2. Summarized MRI Criteria for Assessing Mediastinal Lymph Node Metastasis in Lung Cancer Patients
MRI Sequence and Measurement Qualitative evaluation T1-weighted and T2-weighted spin-echo/ gradient-echo sequences Short axis diameter Morphologic characteristics STIR turbo spin-echo sequence SI of lymph node DW MRI SI of lymph node with a b value of 1000 s/mm2 Quantitative evaluation T2-weighted triple-inversion black-blood fast spin-echo sequence Lymph node to tumor ratiow STIR turbo spin-echo sequence Lymph node to saline ratioww Lymph node to muscle ratiowww DW MRI ADC value
Cutoff Value for Diagnosing Metastatic Lymph Nodes
Z10 mm32,52 Eccentric cortical thickening or obliteration of fatty hilum32,52,53 Higher than muscle and equal to or less primary lesion54,55 Higher than muscle and equal to or less primary lesion55
> 0.8452 Higher than 0.6052–54 Higher than 1.4055 < 0.24-2.510 3 s/mm232,55,56
For normalization of SI of lymph node, the average SI of lymph node is divided by that of tumor (w), or 0.9% normal saline phantom (ww), or rhomboid muscle (www). ADC indicates apparent diffusion coefficient; SI, signal intensity.
r
2013 Lippincott Williams & Wilkins
www.thoracicimaging.com |
11
Yoon et al
J Thorac Imaging
Volume 29, Number 1, January 2014
FIGURE 8. Brain and bone metastases at PET/MRI. A, CT shows a 28 mm enhancing lesion (arrow) abutting a calcified granuloma in the right upper lobe in a 68-year-old man. There was no significant lymph node enlargement in the mediastinum and both hilar areas. B, PET/MRI shows increased FDG uptake of the lesion (SUVmax: 8.1) in the right upper lobe (arrow). Further, multiple metastases in the brain (C) and bone (B) were detected at PET/MRI. C, T1-weight contrast-enhanced volumetric interpolated breath-hold examination axial images depicted a 3 mm enhancing metastatic lesion (arrow) and multiple metastases in variable sizes in the brain. The nodule in the right upper lobe was confirmed as adenocarcinoma.
with proven or suspected lung cancer using PET/CT followed by integrated PET/MRI, a negative lymph node was correctly diagnosed with PET/MRI, whereas PET/CT showed an FDG-avid lymph node in 1 patient.25 This difference may be caused by decreased FDG uptake in benign lymph nodes on delayed phase.
M-Staging MRI provides superior soft tissue contrast in frequent extrathoracic metastatic organs from lung cancer such as the brain and liver where detection of metastasis is limited on PET/CT. Owing to these clear advantages of MRI over PET/CT in these organs, one of the most beneficial effects of PET/MRI over PET/CT in lung cancer stating may be on M-staging.57 In studies that compared whole-body MRI and PET/CT, whole-body MRI was superior in detecting brain and liver metastases and PET/CT in detecting lymph node and soft tissue metastases.11,58 Due to the high physiological FDG uptake in the brain, FDG-PET detected only 61% of the cerebral metastases that were detected with MRI.59 Although MRI of the brain embedded in whole-body MRI detected fewer brain metastases than dedicated brain MRI (27 vs. 40), all of the missed metastatic lesions by whole-body MRI were
Positron Emission Tomography/Magnetic Resonance Imaging Evaluation of Lung Cancer Current Status and Future Prospects Soon Ho Yoon, MD,*w Jin Mo Goo, MD, PhD,*wz Sang Min Lee, MD,*w Chang Min Park, MD, PhD,*wz Hyo Jung Seo, MD,y and Gi Jeong Cheon, MD, PhDzy
Abstract: Various designs of positron emission tomography/magnetic resonance imaging (PET/MRI) systems have been recently introduced to clinical practice, which have overcome preexisting technical challenges concerning the fusion of PET and MRI systems. Although further improvements are still necessary especially for bony lesions, quantification using current MRI-based attenuation correction techniques has been shown to be comparable to that of PET/computed tomography (CT) systems. On the basis of the results of previous whole-body MRI studies, PET/MRI is expected to show even better performance than PET/CT in M-staging especially for brain and liver metastases. Another advantage of PET/MRI over PET/CT, in addition to good soft tissue contrast, is the potential reduction in radiation dose. The next important hurdle to overcome for its clinical application is the development of time-efficient protocols for lung cancer evaluation and interpretation of discordant results from both modalities. Multiparametric imaging through PET/MRI will help radiologists better understand tumor biology and better evaluate treatment response. Key Words: positron emission tomography, magnetic resonance imaging, positron emission tomography and computed tomography, lung cancer, review
(J Thorac Imaging 2014;29:4–16)
I
n addition to improvements in imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI), advances in combined multimodality imaging technologies such as positron emission tomography (PET)/CT have improved clinical patient care especially in oncologic imaging. Since their introduction in 2000,1 PET/CT scanners have become widely implemented and have enormously contributed to the diagnosis, treatment, and prediction of prognosis in oncologic patients.2 Thereafter, a number of studies including a randomized controlled trial on the use of PET/CT for preoperative staging of non–small cell lung cancers have proven the superior efficacy of PET/CT in cancer staging over conventional staging methods.3–6 At present, PET/CT has
From the Departments of *Radiology; yNuclear Medicine; zCancer Research Institute, Seoul National University College of Medicine; and wInstitute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Korea. Supported in part by a grant from Guerbet Korea. The authors declare no conflicts of interest. Reprints: Jin Mo Goo, MD, PhD, Department of Radiology, Seoul National University College of Medicine, 101 Daehangno, Jongnogu, Seoul 110-744, Korea (e-mail: [email protected]). Copyright r 2013 by Lippincott Williams & Wilkins
4 | www.thoracicimaging.com
become a standard imaging modality in lung cancer staging, and its cost-effectiveness has been validated.7 Despite the many advantages of MRI, including excellent soft tissue contrast resolution and absence of radiation exposure, it has been an underutilized imaging modality for evaluating thoracic diseases, as considerable challenges in MRI acquisition for the lung still exist because of low proton density, inhomogeneity of the magnetic file in the lung, and cardiac and respiratory motion artifacts.8 With recent technical advances such as fast imaging and parallel imaging techniques in MRI, previous limitations have been partially overcome. In addition, new methods using either a moving-table platform in combination with the body coil or a specially designed rolling-table platform with 1 body phased-array coil have made it feasible to perform whole-body MRI for lung cancer staging.9–11 Considering the success of PET/CT in oncology practice and the innate limitations of PET/CT in terms of soft tissue contrast resolution and radiation exposure, a fusion of PET and MRI has garnered much attention as it is a conjunction of molecular information offered by PET and anatomic/ potentially functional information with high contrast and spatial resolution provided by MRI.12 We can anticipate the potential strengths and limitations of PET/MRI as this modality has components of both PET and MRI. The various kinds of radioisotopes used in PET can depict several important biological properties including glucose metabolism, tissue hypoxia, perfusion, and apoptosis; and MRI can yield images reflecting cellular density, perfusion, hypoxia, and metabolic features in tissue in addition to basic T1-weighted and T2-weighted MRI images offering high-resolution anatomic information with different contrasts according to the evolution of several MRI sequences. In addition to these potential advantages, PET/MRI also has an important advantage over PET/CT in terms of radiation dose. Although the underlying technical obstacles have not yet been fully resolved, various designs of PET/MRI systems including fully integrated hybrid scanners have been recently implemented in clinical practice. In this review, we discuss the instrumentation, technical aspects, results of studies obtained with PET/CT and whole-body MRI, recent results with integrated PET/ MRI, and potential applications related to PET/MRI in lung cancer.
HARDWARE DESIGN OF PET/MRI SYSTEMS An ideal combination of PET and MRI is a hybrid system in which PET and MRI signals can be acquired simultaneously. However, the integration of PET and MRI systems within a single gantry has innate technical J Thorac Imaging
Volume 29, Number 1, January 2014
J Thorac Imaging
Volume 29, Number 1, January 2014
difficulties, as the signal acquisition process of the 2 modalities is completely different causing interference between the 2 systems.13 Indeed, PET detectors can be interfered with by the strong B0 static magnetic field, abrupt changes in the B1 gradient magnetic field, and radiofrequency pulse during the signal acquisition of an MRI system. Conversely, MRI detectors can also be affected by electromagnetic radiation from PET electronics, magnetic field inhomogeneity derived from PET devices, and eddy currents. Unless these interferences can be resolved, the simultaneous acquisition of PET and MRI images will result in significant degradation of image quality. There are 2 major strategies in the design of a PET/ MRI system adopted by vendors (Fig. 1). One is the tandem arrangement, and the other is the concurrent arrangement of PET and MRI scanners. The tandem arrangement enables sequential acquisition of PET and MRI signals without major modification to PET and MRI scanners, which are located to avoid mutual interference either in the same
PET/MRI Evaluation of Lung Cancer
space or in 2 separate spaces. Ingenuity TF (Philips Healthcare, Best, Netherlands) operates PET and MRI scanners with back-to-back configurations in the same room. In the case of the Discovery PET/CT and MRI system (GE Healthcare, Milwaukee, WI), PET/CT and MRI scanners are combined and sequentially performed in 2 separate rooms. Compared with the concurrent type of arrangement, PET/MRI systems with tandem arrangement have demerits not only in the potential anatomic mismatch, which may occur due to patient motion, but also in temporal mismatch, which may occur due to the sequential acquisitions of PET and MRI images. Biograph mMR (Siemens Healthcare, Erlangen, Germany) adopted a fully integrated PET/MRI system using the concurrent configuration. One of the biggest technical difficulties in the integration of PET/MRI systems is related to the photosensor of the PET detector, as the photomultiplier tube traditionally used in PET or PET/CT systems is very sensitive to the magnetic field. An introduction of avalanche photodiodes, which are insensitive to magnetic fields, has enabled the development of hybrid PET/MRI scanners such as Biograph mMR, although avalanche photodiodes are inferior to photomultiplier tubes in terms of gain and temporal resolution. The implementation of silicon photodiodes (Geiger-mode avalanche photodiodes) in hybrid PET/MRI scanners is currently underway as silicon photodiodes have higher gain and faster temporal resolution enabling time-of-flight PET in addition to the advantages of avalanche photodiodes.14 With the technical evolution regarding PET, MRI scanners, and RF shielding, simultaneous acquisition of PET and MRI signals is now possible without significant interference in Biograph mMR hybrid scanners.15 Simultaneous acquisition provides better anatomic and temporal registration of MRI and PET signals allowing more robust interpretation of disease biology.16 Software-based image registration is another option to generate PET/MRI images by fusing separately acquired PET and MRI images.17 Several computation algorithms fusing PET and MRI images have been proposed with varying degrees of success in registration.18 The trimodality system, which combines PET/CT and MRI systems located in separate rooms, has shown a weakness with regard to the radiation dose, as the CT scanner of PET/CT is used for image acquisition. Comparison of features of integrated, sequential, and coregistered PET/MRI systems is detailed in Table 1.
TECHNICAL ISSUES IN PET/MRI MRI-based Attenuation Correction
FIGURE 1. Schematic diagrams of current commercial wholebody PET/MRI systems. A, Tandem arrangement of PET and MRI scanners in the same room (Ingenuity TF; Philips Healthcare). B, Tandem arrangement of PET/CT and MRI scanners in separate rooms (Discovery PET/CT and MRI system; GE Healthcare). PET and MRI signals are sequentially acquired in the tandem systems. C, Concurrent arrangement of PET and MRI scanners wherein simultaneous acquisition of PET and MRI signals are possible (Biograph mMR; Siemens Healthcare). PET detector composed of avalanche photodiodes is integrated between radiofrequency body coil and gradient, primary magnet coils. r
2013 Lippincott Williams & Wilkins
Attenuation correction is essential for the quantification of PET data as some portion of photons emitted from radiotracers is variably absorbed across different body tissues before arrival at the PET detector. To correct the attenuation of emitted photons throughout the body, an attenuation map with information regarding the distribution of attenuation coefficients is needed. Contrary to the PET/CT system in which the CT attenuation coefficient is linear to electron density, MRI signal intensity in the PET/MRI system has no direct relationship to electron density. Several approaches deriving attenuation coefficients from MRI images have been proposed,19 and routine applications of MRI-based attenuation corrections for whole-body PET/MRI are based on the acquisition of www.thoracicimaging.com |
5
J Thorac Imaging
Yoon et al
TABLE 1. Comparison of Integrated, Sequential, and Coregistered PET/MRI Systems
Integrated Sequential Image Coregistration PET/MRI PET/MRI in Separate PET and System System MRI Systems Simultaneity in PET and MRI acquisition Accuracy of PET/MRI registration PET attenuation correction Time-of-flight PET Flexibility of scheduling
High
Intermediate
Low
High
Intermediate
Low
MRI-based MRI-based
CT-based
Not Possible possible* Less Less flexible flexible
Possible More flexible
*Avalanche photodiode used in Biograph mMR does not support timeof-flight PET. Time-of-flight PET may be applicable in integrated PET/MRI systems using silicon photodiodes in the future.
dedicated MRI sequences such as the DIXON sequence with in-phase/opposed-phase images. The Dixon MRI sequence provides separate water and fat images and creates an attenuation map (m map) with 4 distinct tissue classes: background, lung, fat, and soft tissue (Fig. 2). The lungs are identified by connected-component analysis of the air in the inner part of the body. In addition, difficulty in differentiating air, lung, and bone on conventional MRI sequences makes MRI-based attenuation correction more complicated in the PET/MRI system. Importantly, as bony structures are not considered by currently applied MRIbased attenuation correction methods, these structures undergo a systematic undercorrection of PET data. Maximum errors of decreased uptake in bone are up to 13% or 17% in standardized uptake value (SUV).20–22 Another option for attenuation correction in the PET/MRI system is Water
Fat
Volume 29, Number 1, January 2014
the use of an anatomic atlas template to coregister patient data sets. Contrary to easier registration in the brain, the atlas-based method itself remained challenging for the whole body. Recently, the atlas-based method in conjunction with regional segmentation derived from the MRI sequence was reported to be technically feasible and was shown to provide better results than pure regional segmentation using the MRI sequence.23 Despite this source of error in SUV, recent studies using an integrated PET/MRI system revealed that the PET image quality was subjectively rated high and that the diagnostic value was comparable to that of standard PET/ CT systems.24,25 In the clinical application of PET wherein the diagnosis is often based on qualitative evaluation, accuracy in quantification may not be crucial as far as these errors do not affect image quality (Fig. 3).26 However, PET/ MRI-based therapy response assessment in oncologic disease requires accurate PET quantification, and therefore further improvement in MRI-based attenuation correction is necessary.
Alignment Precise alignment is necessary to characterize pathologic findings when combining different imaging modalities. However, bulk patient motion or various physiological motions of respiration, bowel movements, or urinary bladder filling can cause misalignment in hybrid imaging. Anatomic alignment is expected to improve with simultaneous acquisition of PET/MRI compared with retrospectively fused PET and MRI. A recent study dealt with this issue and confirmed that the alignment of hybrid data sets acquired in simultaneous whole-body PET/MRI was more accurate than retrospective fusion in abdominal organs.27 In the urinary bladder, the alignment of simultaneous PET/MRI was more accurate than PET/CT. When the effect of different breathing protocols of simultaneous, sequential, and MRI-gated data acquisition was compared, the alignment of thoracic PET/MRI with expiratory breath-hold or free-breathing MRI was more accurate than with inspiratory MRI.
In phase
Opposed phase
µMap
FIGURE 2. MRI-based attenuation correction using the Dixon sequence. MRI water and fat images are generated by in-phase/opposedphase images. These water and fat images are combined and segmented to produce the MRI-based attenuation correction map (m map).
6 | www.thoracicimaging.com
r
2013 Lippincott Williams & Wilkins
J Thorac Imaging
Volume 29, Number 1, January 2014
PET/MRI Evaluation of Lung Cancer
FIGURE 3. SUVmax of lung cancer at PET/MRI and PET/CT. PET/MRI and PET/CT were performed in a 58-year-old woman with adenocarcinoma. A, CT shows a 27 mm irregular nodule (arrow) in the right upper lobe. B, PET/CT was obtained 1 hour after (C) PET/MRI. The SUVmax of lung cancer at PET/MRI and PET/CT (arrows) were 3.01 and 5.75. The SUVmax at PET/MRI was 47.7% lower than that at PET/CT.
Other Sources of Artifacts and Errors In calculating SUV, other sources of attenuation, such as surface coils, and additional positioning devices must be considered. These devices can cause scattering and attenuation of the PET signal. With regard to MRI contrast agents, ingestion of an oral iron oxide–based MRI contrast agent has been shown to affect the MRI-based attenuation map even though PET quantification is not affected by intravenous injection of the MRI contrast agent.28 In addition, metallic implants can produce a signal loss resulting in underestimated uptake in the region surrounding the metallic implants, and truncation artifacts may arise from the difference in the transverse fields of view
of PET and MRI (Fig. 4). To recognize attenuation-related artifacts, it is important to evaluate the non–attenuationcorrected images. Partial volume effects are another source of potential error that can affect the quantitative accuracy of PET. SUV can be underestimated with decreasing tumor volume, and therefore it can be affected by applying different partial volume correction methods.29 Thus far, studies have shown that reconstruction-based partial volume correction, which includes 3-dimensional (3D) point spread function, outperforms image-based methods in terms of accuracy.29 To reduce motion artifacts, physiological gating such as electrocardiogram gating and MRI-based triggering of
FIGURE 4. Truncation artifact at PET/MRI. A, PET images and (B) T1-weighted volumetric interpolated breath-hold examination coronal images are used for image fusion (C). As the transverse field of view (FOV) of MRI and PET is 50 and 59 cm, respectively, there are some parts of arms beyond the MRI-FOV. This results in underestimation of the PET image information along the arms (arrows) and is called truncation artifact (C). r
2013 Lippincott Williams & Wilkins
www.thoracicimaging.com |
7
Yoon et al
PET data can be applied. One requirement in applying these techniques is to consider the associated increase in examination time. An alternative approach to reducing respiratory motion–related errors is the MRI-based respiratory motion correction of PET that can improve tumor visibility, delineation, and maximum SUV.16
Examination Time Unlike PET/CT, the design of PET/MRI protocols for whole-body coverage is more complex, and time efficiency is an especially important issue. PET/MRI can be performed in approximately 25 minutes without further morphologic MRI. However, with a number of additional diagnostic sequences and with the use of an intravenous contrast, the examination time may increase up to 90 minutes or more. With MRI sequences such as diffusion, perfusion MRI, and MRI spectroscopy, more functional and biological properties can be obtained at the cost of a longer examination time. Therefore, the numbers and types of MRI sequences required need to be specifically chosen and optimized according to the clinical tasks of PET/ MRI.30 A continuous moving-table data acquisition scheme can be applied in postcontrast studies.31
IMAGE ACQUISITION WITH AN INTEGRATED PET/MRI SYSTEM As most clinical studies on PET/MRI were carried out with the integrated PET/MRI system, this section regarding image acquisition is based on integrated PET/MRI. Most differences in designing PET/MRI scanning protocols come from the MRI protocols. Optimizing the PET/MRI protocol is an important issue, which is under active investigation. Due to time constraints, especially for wholebody imaging, priority among various MRI sequences should be considered in adding dedicated MRI for specific body parts. This issue is essentially the same for sequential PET/MRI systems except that they need more examination time to perform the same protocol. The MRI part in PET/ MRI plays a role for anatomic localization, attenuation correction of PET data, and tissue characterization based on selected MRI techniques.
Patient Preparation Before examination, in the case of administration of F-fluoro-2-deoxyglucose (18F-FDG), fasting, control of glucose level, and having the patient rest between 18F-FDG administration and imaging to minimize muscle uptake are required for patient preparation. In addition to preparations for stand-alone MRI, care should be taken so that only dedicated coils approved for PET/MRI are used, as the surface coils may cause additional attenuation of 511 keV photons. 18
PET/MRI Acquisition Protocols As the PET/MRI scan protocol for the staging of lung cancer has not been standardized as yet, the general principles for PET/MRI scans and the specific scan protocol in our center will be introduced. The predominant factors for PET/MRI scan time are the number and type of the MRI sequence. Despite the variability of whole-body MRI scan protocols across the literature,3,9–11,32,33 a whole-body MRI scan can be performed with a combination of a coronal or axial fast T2-weighted sequence (eg, half-Fourier acquisition single-shot turbo spin-echo sequence), fast T1weighted sequence (eg, volume-interpolated 3D-spoiled
8 | www.thoracicimaging.com
J Thorac Imaging
Volume 29, Number 1, January 2014
gradient-echo sequence), short-t inversion recovery (STIR) sequence, and diffusion-weighted MRI sequence covering organs in which metastasis frequently occurs. In addition, contrast-enhanced MRI scans are essential for better detection of metastases in solid organs including the brain, liver, and leptomeninges.34–36 The initial step is obtaining MRI localizers to plan the subsequent acquisition. At each bed position, while PET imaging is acquired, the MRI sequence for attenuation correction, which takes 19 seconds per bed position, is acquired first, followed by subsequent simultaneous MRI sequences for diagnostic purposes. As the axial range of a single bed position is 25.8 cm with a 6.1 cm overlap between adjacent bed positions in the current integrated PET/MRI system, 2 to 5 bed positions are required for lung cancer staging to cover a body part or the whole body. MRI for a specific region of interest can be acquired additionally with or without simultaneous PET scanning according to diagnostic purposes. With regard to PET scan, the number of bed positions and the acquisition time for each should be determined for PET imaging, which is obtained in the step-and-shoot mode. PET acquisition time is 2 to 4 minutes per bed position, but this can be increased for simultaneous MRI acquisitions. The dedicated PET/MRI scan protocol for lung cancer preliminarily used in our center was developed for the purpose of both clinical practice and academic research (Fig. 5) and is expanded from a basic whole-body PET/MR scan protocol (Fig. 6).
FUTURE PROSPECTS ON CLINICAL APPLICATION IN LUNG CANCER Currently, there are very few studies investigating pulmonary malignancies using an integrated PET/MRI machine. Herein, we briefly review the results of preliminary studies based on coregistration of PET and MRI or PET/MRI systems to better comprehend the potential role of integrated PET/MRI in lung cancer.
Detection and Assessment of Pulmonary Nodules At present, CT is the standard and best imaging modality for the detection of pulmonary nodules due to its high spatial and temporal resolution. Despite improvement of MRI techniques, current MRI sequences used in PET/MRI systems are not sufficient for visualization of small lung nodules of 5 mm or less.37–40 The detection rate of MRI with various sequences has ranged between 45.5% and 96.0%, and the detection rate of spin-echo sequences was better than that of gradient-echo sequences.8,37–40 In a prospective comparative study on wholebody MRI and PET/CT, more pulmonary metastases were found on PET/CT than on MRI (170 vs. 139 pulmonary metastases).3 In addition, in a recent study that used an integrated PET/MRI system, radial volumetric interpolated breath-hold examinations and PET acquired with PET/MRI detected 70.3% of all nodules and 95.6% of FDG-avid nodules, and 88.6% of nodules 5 mm in diameter or larger were detected compared with the PET/CT system, which detected 69 nodules including 45 FDG-avid nodules.41 This weakness of PET/MRI in detecting pulmonary nodules might not be problematic on the initial workup of lung cancer staging given that chest CT is routinely performed at initial diagnosis. However, substitution of PET/CT with PET/MRI during follow-up in patients with lung cancer might only be possible in combination with chest CT until better MRI sequences are developed for detection of metastatic nodules in the lung.11,42 r
2013 Lippincott Williams & Wilkins
J Thorac Imaging
Volume 29, Number 1, January 2014
PET/MRI Evaluation of Lung Cancer
A Time (min) 1
0
36
PET
41
BP PET (each 7min)
PET abd. (5min)
51
56
No PET
PET thorax (10min)
CM MRI
AC
Localizer
Cor T1 TSE
Ax Ax T2HASTE VIBE FS
5 bed positions Head to thigh
AC
Ax DWI
1 bed position abdomen
AC
Ax DWI
Ax DC E
Ax VIB E
Ax Ax Ax Ax VIBE VIBE VIBE VIBE
1 bed position Thorax
Head to thigh
B Time (min) 0
1
13
BP PET(each 6 min)
PET
MRI
Localizer AC
Cor STIR
Ax VIBE
2 bed positions Thorax to upper abdomen
FIGURE 5. Examples of whole-body (A) and partial-body (B) acquisition protocols of PET/MRI in lung cancer patients. AC indicates attenuation correction; Ax, axial; BP, bed position; CM, contrast medium; cor, coronal; DCE, dynamic contrast enhancement; HASTE, half-Fourier acquisition single-shot TSE; TSE, turbo spin-echo; VIBE, volumetric interpolated breath-hold examinations.
As PET/CT can also fail to detect small metastatic lung nodules found on chest CT,43 PET/MRI with low-dose CT may be the best option for postoperative surveillance in lung cancer patients. FDG-PET/CT has demonstrated efficacy in differentiating malignant from benign nodules, but this has limited value in the detection of ground-glass nodules and has yielded a considerable number of false-positive results for these nodules, mainly because of inflammation.44,45 On the basis of the concept that malignant lesions demonstrate increased cellularity, high tissue disorganization, and increased extracellular space tortuosity compared with benign lesions, malignant lesions show high signal intensity on diffusion-weighted imaging (DWI) and low apparent diffusion coefficients (ADC). However, granulomas, active inflammation, or fibrous nodules may also show high signal r
2013 Lippincott Williams & Wilkins
intensity, similar to malignant nodules.8 The use of dynamic contrast-enhanced MRI demonstrated high diagnostic capability in distinguishing malignant from benign nodules, with sensitivities ranging from 94% to 100% and specificities from 70% to 96%.8
Assessment of TNM Staging T-Staging T-staging is assessed mainly on the basis of tumor size and invasion of adjacent tissue according to the seventh edition of the TNM staging system for lung cancer.46 When compared with the CT scan, the MRI scan showed higher diagnostic performance in assessing locoregional invasion to the chest wall, mediastinum, and diaphragmatic pleura, as MRI offers clear visualization of the extrapleural fat plane, www.thoracicimaging.com |
9
Yoon et al
J Thorac Imaging
Volume 29, Number 1, January 2014
FIGURE 6. Whole-body PET/MRI protocol performed in a 73-year-old man with lung cancer. PET/MRI starts 1 hour after intravenous administration of 18F-FDG with 7 minutes per bed position. As scan range is from head to lower thigh, the number of bed positions is usually 5 or 6. After gradient-echo localization, a coronal T1-weighted turbo spin-echo sequence is performed (A). Then the T2weighted half-Fourier acquisition single-shot turbo spin-echo axial sequence (B) and the T1-weghted volumetric interpolated breathhold examination axial sequence are obtained. Diffusion-weighted images (b = 50, 400, 800) (C) and apparent diffusion coefficient map (D) are also obtained. Afterward, T1-weighted contrast-enhanced sequencing (E) is performed. Fused PET and MRI images (F, G) are interpreted for evaluation of lesions.
pleural abnormality on T1-weighted and T2-weighted images, chest rib destruction on STIR images, and adhesion between the chest wall and mass on dynamic cine MRI images.47–50 In a study by Plathow et al,51 whole-body MRI obtained with 1.5 T MRI provided correct T-staging in all patients, whereas PET/CT did not correctly stage chest wall invasion in 4 of 52 patients. In another study by Yi et al,11 when PET/CT and whole-body MRI obtained with 3 T MRI were compared, there were no significant differences in T-staging assessment. In a recent study that included 10 patients with pathologically proven or clinically suspected lung cancers, an integrated PET/MRI system excluded the infiltration of the mediastinal pleura by the tumor suspected at PET/CT in 1 patient and led to understaging due to slightly smaller measurement of the tumor size compared with PET/CT.25
10 | www.thoracicimaging.com
N-Staging Although CT is the standard modality in lung cancer staging, the major weakness of CT is its low accuracy in Nstaging because only size is considered to determine metastatic lymph nodes. In comparison, FDG-PET has shown better performance in determining metastatic lymph nodes. MRI is also superior to CT in diagnosing mediastinal lymph node metastasis, as metastatic lymph nodes are able to be differentiated from benign reactive lymph nodes by high signal intensity on the STIR sequence and diffusion restriction on DWI (Fig. 7; Table 2).32,53,54,56 Cardiactriggered and/or respiration-triggered STIR turbo spinecho imaging has been recommended for N-staging, and when compared with coregistered FDG-PET/CT, STIR sequence showed better performance.53 However, in r
2013 Lippincott Williams & Wilkins
J Thorac Imaging
Volume 29, Number 1, January 2014
PET/MRI Evaluation of Lung Cancer
FIGURE 7. Apparent diffusion coefficient (ADC) map on evaluation of lymph node metastasis. A, A 59-year-old man had a 3.1 cm mass (arrow) in the left upper lobe with a 7 mm left lower paratracheal lymph node (arrow) and a 13 mm left upper lobar lymph node (not shown). B, PET/MRI showed a mass (SUVmax: 14.65) in the left upper lobe, a 13 mm lobar node (SUVmax: 12.84), and a 7 mm left lower paratracheal lymph node (SUVmax: 4.5) (arrows). C and D, Diffusion-weighted image (b = 400) and ADC map showed diffusion restriction in the mass in the left upper lobe and 2 lymph nodes (arrows). The concordant findings of left lower paratracheal lymph node on the ADC map and at fused PET/ MRI increased confidence in reporting left lower paratracheal lymph node metastasis, which was further confirmed by pathologic results.
another study by Yi et al,11 there were no significant differences between PET/CT and whole-body MRI in Nstaging assessment. With regard to DWI, Nomori et al56 reported that DWI was significantly more accurate than FDG-PET because of less overstaging and fewer falsepositive results of DWI. When the diagnostic performance of FDG-PET/CT was compared with that of FDG-PET/ CT combined with DWI and T2-weighted MRI for preoperative regional lymph node evaluation, sensitivity increased from 46% to 69% with a slight decrease of
specificity from 96% to 93%.32 This difference is mainly caused by inflammatory lymph nodes in which FDG uptake usually increases while there is no restricted diffusion. However, the low spatial resolution and image distortion of DWI sequences are limitations in detecting some small metastatic lymph nodes.8 Although there may be some potential gains in diagnosing lymph node metastasis with a combination of PET and MRI over PET/CT, it was not evident in preliminary studies.10,25 In a recent study that evaluated 10 patients
TABLE 2. Summarized MRI Criteria for Assessing Mediastinal Lymph Node Metastasis in Lung Cancer Patients
MRI Sequence and Measurement Qualitative evaluation T1-weighted and T2-weighted spin-echo/ gradient-echo sequences Short axis diameter Morphologic characteristics STIR turbo spin-echo sequence SI of lymph node DW MRI SI of lymph node with a b value of 1000 s/mm2 Quantitative evaluation T2-weighted triple-inversion black-blood fast spin-echo sequence Lymph node to tumor ratiow STIR turbo spin-echo sequence Lymph node to saline ratioww Lymph node to muscle ratiowww DW MRI ADC value
Cutoff Value for Diagnosing Metastatic Lymph Nodes
Z10 mm32,52 Eccentric cortical thickening or obliteration of fatty hilum32,52,53 Higher than muscle and equal to or less primary lesion54,55 Higher than muscle and equal to or less primary lesion55
> 0.8452 Higher than 0.6052–54 Higher than 1.4055 < 0.24-2.510 3 s/mm232,55,56
For normalization of SI of lymph node, the average SI of lymph node is divided by that of tumor (w), or 0.9% normal saline phantom (ww), or rhomboid muscle (www). ADC indicates apparent diffusion coefficient; SI, signal intensity.
r
2013 Lippincott Williams & Wilkins
www.thoracicimaging.com |
11
Yoon et al
J Thorac Imaging
Volume 29, Number 1, January 2014
FIGURE 8. Brain and bone metastases at PET/MRI. A, CT shows a 28 mm enhancing lesion (arrow) abutting a calcified granuloma in the right upper lobe in a 68-year-old man. There was no significant lymph node enlargement in the mediastinum and both hilar areas. B, PET/MRI shows increased FDG uptake of the lesion (SUVmax: 8.1) in the right upper lobe (arrow). Further, multiple metastases in the brain (C) and bone (B) were detected at PET/MRI. C, T1-weight contrast-enhanced volumetric interpolated breath-hold examination axial images depicted a 3 mm enhancing metastatic lesion (arrow) and multiple metastases in variable sizes in the brain. The nodule in the right upper lobe was confirmed as adenocarcinoma.
with proven or suspected lung cancer using PET/CT followed by integrated PET/MRI, a negative lymph node was correctly diagnosed with PET/MRI, whereas PET/CT showed an FDG-avid lymph node in 1 patient.25 This difference may be caused by decreased FDG uptake in benign lymph nodes on delayed phase.
M-Staging MRI provides superior soft tissue contrast in frequent extrathoracic metastatic organs from lung cancer such as the brain and liver where detection of metastasis is limited on PET/CT. Owing to these clear advantages of MRI over PET/CT in these organs, one of the most beneficial effects of PET/MRI over PET/CT in lung cancer stating may be on M-staging.57 In studies that compared whole-body MRI and PET/CT, whole-body MRI was superior in detecting brain and liver metastases and PET/CT in detecting lymph node and soft tissue metastases.11,58 Due to the high physiological FDG uptake in the brain, FDG-PET detected only 61% of the cerebral metastases that were detected with MRI.59 Although MRI of the brain embedded in whole-body MRI detected fewer brain metastases than dedicated brain MRI (27 vs. 40), all of the missed metastatic lesions by whole-body MRI were
