Keywords
Cardiovascular invasion - cine MRI - mediastinal masses
Introduction
Accurate assessment of the local extent of mediastinal masses is essential to determine
surgical feasibility and decide upon the future course of management. Owing to limited
space available in the mediastinum, masses appear to abut the adjacent cardiovascular
structures and sometimes it becomes difficult on contrast-enhanced CT scans to confirm
or rule out cardiovascular invasion by mediastinal masses, often over-diagnosing possible
infiltration of cardiovascular structures, thus denying patients of potentially curative
surgery.
Currently, contrast-enhanced CT scans are most commonly used for preoperative assessment
of mediastinal masses. Some criteriathat have been studied for diagnosing local mediastinal
invasion on CT include loss of intervening fat planes, broad surface of contact, and
a large angle of contact (90° or 180°) between the mass and adjacent cardiovascular
structures, indentation or distortion of vessel contour and intraluminal extension.[1], [2], [3] However, the sensitivity, specificity and accuracy of CT scan for cardiovascular
invasion vary in literature and largely depend upon the study design and patient population.
Sensitivity and specificity ranges are 40%-77.8% and 67%-99%, respectively.[4], [5]
We aimed to evaluate the additional role of steady-state free precession (SSFP) cine
MRI to visualize the differential sliding motion and presence or absence of type 2
chemical shift artefact (black boundary artefact) between the mediastinal mass and
adjacent cardiovascular structures in patients with equivocal findings of invasion
on CT scan.
Materials and Methods
A retrospective study of 44 patients was performed between 2010 and 2018 for who had
undergone cine SSFP MRI to evaluate equivocal findings of cardiovascular invasion
by mediastinal masses on CT scan. The imaging findings were correlated with intraoperative
findings. Seven patients were excluded from the study as they were deemed inoperable,
in view of extensive infiltration on imaging or detection of metastases.
CT Technique and interpretation
Contrast-enhanced CT scans of the thorax in axial and coronal planes with slice thickness
2.5-5 mm using 1 ml/kg of non-ionic iodinated contrast media were obtained using 16
slice Siemens Emotion or 64 slice GE Discovery CT750 HD CT scanners.
CT findings were considered equivocal for cardiovascular invasion when there was a
loss of intervening fat planes, broad surface of contact (>3 cm) or >90° angle of
contact between the mass and adjacent cardiovascular structures [Figure 1A] and [B] or in a multidisciplinary team meeting if the reporting radiologist and thoracic
surgeon had any suspicion of infiltration.
Figure 1 (A and B): (A) Axial CECT images illustrate the criteria used to define equivocal cardiovascular
invasion on CT. (A) shows a broad surface of contact (>3 cm) and obliteration of intervening
fat plane between mass and pericardium (arrow). (B) shows >90° angle of contact and
loss of intervening fat plane between mass and ascending aorta (arrow)
MRI Technique and interpretation
The MRI scans were performed on a 1.5-Tesla Siemens Avanto scanner. Bright blood SSFP
non-cine images were obtained at repetition time/echo time (TR/TE) of 290-360 ms/1.1-1.4
ms, with a slice thickness of 6-8 mm, slice spacing of 6.3-6.5 mm, field of view (FOV)
ranging from 340 to 350 mm, matrix ranging from 240-300 × 208-256, receiver bandwidth
of 866 Hz/Px, the phase-encoding direction was anteroposterior and frequency-encoding
direction was right to left. Cardiac-gated cine SSFP images were obtained at TR/TE
of 29-41 ms/1.1-1.3 ms with a slice thickness of 6-8 mm, FOV ranging from 340 to 350
mm, matrix of 192-256 × 208-256, receiver bandwidth of 930 Hz/Px, phase-encoding direction
was antero-posterior and frequency-encoding direction was right to left. Cine and
non-cine images were taken in axial, coronal and sagittal planes. In most patients,
oblique coronal and oblique sagittal cine images were also acquired, with the plane
of imaging perpendicular to the interface between mass and cardiovascular structure
being studied.
The presence or absence of type 2 chemical shift artifact was evaluated on both cine
and non-cine images [Figure 2A] and [B] and differential sliding motion between the mass and adjacent cardiovascular structure
was assessed on cine images [Figure 3A], [B] and Video 1]. These two parameters were used on MRI to assess invasion or adhesion
of mediastinal masses to the cardiovascular structures.
Figure 2 (A and B): (A) Non-cine sagittal image show the presence (A) of chemical shift artifact between
the mediastinal mass and ascending aorta (arrow). (B) Non-cine axial SSFP images show
the absence (B) of chemical shift artifact between mediastinal mass and right atrium
(arrow)
Figure 3 (A and B): (A) Limited images from cine axial SSFP images in end-systolic (A) shows differential
sliding motion between the main pulmonary artery (MPA) and anterior mediastinal mass
(arrow). (B) shows differential sliding motion between MPA and anterior mediastinal
mass (arrow). A sliver of pericardial fluid seen between MPA and the mass in (B)
The presence of either sliding motion or chemical shift artifact or the presence of
both was considered as indicating no invasion or adhesion. The absence of both differential
motion and chemical shift artifact was interpreted as invasion or adhesion.
Results
Demographics
Forty-four patients with mediastinal masses who had undergone additional MRI to evaluate
equivocal findings on CT were included. There were 31 males and 13 females; mean age
36.5 years; range 11-65 years. A total of 162 cardiovascular structures showed doubtful
findings on CT.
Imaging findings were correlated with surgical findings in 137 cardiovascular structures
in 37 patients. In all, 25 cardiovascular structures in 7 patients were excluded as
they did not undergo surgery owing to extensive infiltration or presence of metastases
on imaging.
Diagnoses
A wide variety of histopathological diagnoses was encountered. Thymic lesions were
the commonest including thymoma (n = 10), thymic carcinoma (n = 3) and thymic carcinoid (n = 3), others included malignant germ cell tumors (n = 5), mature teratoma (n = 3), sarcoma (n = 2), benign cysts (n = 4), neurogenic tumors (n = 5), inflammatory myofibroblastic tumor (n = 1) and lymphoma (n = 1).
MRI findings and correlation with intraoperative findings
A total of 137 cardiovascular structures with equivocal invasion on CT scan in the
37 patients were evaluated on MRI for presence of sliding motion and chemical shift
artifact. Pericardium (n = 31) was the most common structure evaluated followed by aorta (n = 19). Other structures included main pulmonary artery (n = 16), right/left pulmonary arteries (n = 11), superior vena cava (n = 14), atria (n = 18), brachiocephalic vein (n = 11), pulmonary veins (n = 8), ventricles (n = 4), inferior vena cava (n = 3), left subclavian artery (n = 1) and left common carotid artery (n = 1).
There was no invasion or adhesion in 106 cardiovascular structures on MRI while 31
structures showed invasion or adhesion on imaging. Intraoperatively, 38 cardiovascular
structures were involved and 99 structures were not invaded or adhered to.
Out of 31 structures that showed features of invasion or adhesion on MRI, 28 structures
were intraoperatively invaded (true positive for ‘invasion’) [Figures 4], [5] and Videos 2, 3] and three structures were free (false positive for ‘invasion’).
Out of 106 structures showing no invasion on MRI, 97 were intraoperatively not invaded
or adhered (true negative) [Figure 6] and Video 4] and 9 structures were involved (false negative). Details of the status
of involvement of individual cardiovascular structures on imaging and surgery is illustrated
in [Table 1].
Table 1
Illustrates the status of infiltration of individual cardiovascular structures on
imaging and surgery
|
Cardiovascular Structure
|
No Invasion/Adhesion on MRI
|
Invaded/Adhered on MRI
|
Intraoperatively Uninvolved
|
Intraoperatively Involved
|
True Negative
|
False Negative
|
True Positive
|
False Positive
|
|
Aorta
|
19
|
0
|
17
|
2
|
17
|
2
|
0
|
0
|
|
MPA
|
15
|
1
|
15
|
1
|
14
|
1
|
0
|
1
|
|
RPA/LPA
|
11
|
0
|
9
|
2
|
9
|
2
|
0
|
0
|
|
SVC
|
7
|
7
|
7
|
7
|
7
|
0
|
7
|
0
|
|
Pericardium
|
19
|
12
|
17
|
14
|
18
|
1
|
12
|
0
|
|
Atria
|
18
|
0
|
16
|
2
|
16
|
2
|
0
|
0
|
|
Ventricles
|
4
|
0
|
4
|
0
|
4
|
0
|
0
|
0
|
|
Pulmonary Veins
|
8
|
0
|
7
|
1
|
7
|
1
|
0
|
0
|
|
Inferior venacava (IVC)
|
2
|
1
|
2
|
1
|
2
|
0
|
1
|
0
|
|
Brachiocephalic vein (BCV)
|
3
|
8
|
5
|
6
|
3
|
0
|
6
|
2
|
|
Aortic arch branches (left CCA and left SCA)
|
0
|
2
|
0
|
2
|
0
|
0
|
2
|
0
|
|
Total
|
106
|
31
|
99
|
38
|
97
|
9
|
28
|
3
|
Figure 4 (A-C): (A) Axial CECT image (A) of a 39-year-old man with biopsy-proven thymic carcinoid
showing obliteration of fat plane and >90° angle of contact between anterior mediastinal
mass (M) and superior vena cava (arrow). (B) End-systolic (B) and end-diastolic (C)
limited cine axial SSFP images show the absence of chemical shift artifact and sliding
motion between the mass and SVC, indicating invasion or adherence
Figure 5 (A and B): (A) Limited cine axial MRI images of a 45-year-old lady with biopsy-proven invasive
thymoma in end-diastolic (A) illustrating the loss of chemical shift artifact and
absence of sliding motion between inferior vena cava (IVC) and the mediastinal mass,
suggesting invasion/adhesion on MRI. Intraoperatively, mass was adherent to IVC. (B)
Cine axial images of a 45-year-old lady with biopsy-proven invasive thymoma in end-systolic
phases (B) illustrating the loss of chemical shift artifact and absence of sliding
motion between inferior vena cava (IVC) and the mediastinal mass, suggesting invasion/
adhesion on MRI. Intraoperatively, the mass was adherent to IVC
Figure 6 (A-C): (A) Axial CECT image of an 11-year-old boy with biopsy-proven mature teratoma in
anterior mediastinum (arrow) showing loss of fat planes and >90° angle of contact
with superior vena cava (SVC) and ascending aorta. (B) Limited cine MRI axial images
of 11-year-old boy with biopsy-proven mature teratoma in end-systolic (B) and end-diastolic
(C) phases show the presence of chemical shift artifact and differential sliding motion
between the mass and cardiovascular structures. Sliver of pericardial fat can be appreciated
between SVC, ascending aorta and the mass in the end systolic image, that is, absent
in end-diastolic image suggesting presence of sliding motion
Using our imaging criteria for invasion, cine MRI was able to exclude or confirm invasion
of cardiovascular structures by mediastinal masses with sensitivity, specificity and
accuracy of 75.7%, 97% and 91.2%, respectively, in cases with equivocal invasion on
the initial CECT scan.
All the cardiovascular structures with the presence of chemical shift artifact showed
differential sliding motion except in one case involving pericardium, where sliding
motion was absent due to the large pericardial effusion rather than actual invasion
[Figure 7A], [B] and Video 5].
Figure 7 (A and B): (A) Limited Cine coronal images of a 32-year-old lady with biopsy-proven mature teratoma
in end-systolic (A) and end-diastolic phases (B) show a large mediastinal mass with
a broad surface of contact with the pericardium and moderate pericardial effusion.
There is a loss of sliding motion between the pericardium and the mediastinal mass.
However, chemical shift artifact is maintained (arrow) suggesting no invasion on MRI.
Intraoperatively, there was no invasion and loss of sliding motion was due to the
associated pericardial effusion
The sensitivity, specificity, and accuracy of sliding motion and chemical shift artifact
as independent parameters to exclude or confirm cardiovascular invasion is shown in
[Table 2].
Table 2
The sensitivity, specificity and accuracy of individual imaging criteria that has
been used to confirm or exclude cardiovascular invasion
|
Parameter
|
Sensitivity
|
Specificity
|
Accuracy
|
|
Sliding motion
|
75.6%
|
96%
|
90.5%
|
|
Chemical shift artifact
|
75.7%
|
97%
|
91.2%
|
Discussion
Accurate assessment of cardiovascular invasion by mediastinal masses is essential
to determine surgical feasibility and in planning the surgery. Many centers consider
the invasion of the myocardium and great vessels as contraindications for surgical
resection.[6] Computed tomography has long been the modality of choice for pre-operative evaluation
of mediastinal masses. However in some cases, the accurate assessment of cardiovascular
invasion by intrathoracic masses becomes difficult on contrast-enhanced CT, due to
the limited space in the mediastinum resulting in the mass indenting on adjacent structures
even if it does not infiltrate.
MRI has found limited use in the evaluation of thoracic pathology as it is expensive,
more time consuming than CT and lack of protons in lung tissue resulting in poor signal
return.[7] However, owing to higher soft-tissue resolution and the advent of improved and faster
imaging techniques, including the ECG-gated fast GRE cine sequences, MRI can be used
as an alternative or complimentary modality.[8], [9], [10] Cine images can be acquired using spoiled or incoherent GRE sequences (where the
residual transverse magnetization is spoiled) and balanced steady state or coherent
GRE sequences (where the transverse magnetization is refocused and added to the longitudinal
magnetization to create a steady state).[11] Balanced SSFP (bSSFP) sequences have a higher blood-tissue contrast but are more
sensitive to the field inhomogeneties, which tend to increase with increasing magnetic
strength. Hence, bSSFP sequences have become the method of choice to acquire cine
images at low magnetic field strength (1.5 T) while spoiled GRE sequences, which are
less sensitive to local field variations, are used at higher magnetic fields (3T or
higher).[12] In this study, we used bSSFP sequences to acquire the cine and non-cine images to
evaluate the presence of chemical shift artifact and differential motion between mediastinal
masses and adjacent cardiovascular structures.
Black line/black boundary artifact also called the type 2 chemical shift artifact
is seen as a black line between the apposing structures in out of phase imaging like
SSFP sequences and suggests an intact fat-water interface.[13], [14], [15] Type 2 chemical shift artifact is already being used for delineating borders of
organs surrounded by fat in other areas of the body like delineating renal borders
in abdominal imaging; identification of fat in lipomas, angiomyolipomas in the liver
or kidneys and in identifying lipid-rich adrenal adenomas. In this study, we have
used this artifact to identify the presence of fat plane between mediastinal masses
and cardiovascular structures. However, radiologists should be aware that the chemical
shift artifact can be minimized or accentuated by altering various imaging parameters
like the field of view, frequency-encoding direction and the receiver frequency bandwidth.
Choosing the frequency-encoding direction in the plane with the narrowest lipid-water
interface, decreasing the field of view and increasing the frequency bandwidth can
minimise the chemical shift artefact and vice versa.[11] These things should be kept in mind while using chemical shift artifact as a sign,
to avoid misinterpretation.
Our study shows the superiority of MR imaging over CT in identifying preserved fat
planes between abutting structures using the presence of type 2 chemical shift artefact
in cine and non-cine SSFP sequences. Out of 137 cardiovascular structures displaying
loss of fat plane on CECT, 106 structures showed the presence of a chemical shift
artifact suggesting the presence of fat-water interface and out of those 97 structures
were not invaded intraoperatively.
We found that the presence of sliding motion can be used to rule out cardiovascular
invasion with a high degree of confidence. This was particularly true while assessing
the cardiac chambers, as the differential sliding motion is more pronounced in the
normal cardiac chambers.[16] Mediastinal vessels, especially those that are away from the heart like descending
thoracic aorta, IVC, brachiocephalic vein and pulmonary veins, normally show minimum
sliding motion, hence, leading to a false interpretation of sliding motion. Out of
12, 9 (75%) cases where MRI findings did not correlate with intraoperative findings
were related to mediastinal vessels. Large intrathoracic masses can reduce the extent
of differential motion between abutting structures due to the mass effect resulting
in false negative cases.[17] We found that pericardial effusion is another factor that can affect sliding motion,
as seen in one of our cases where the differential motion between pericardium and
mass was absent due to massive pericardial effusion rather than actual invasion. To
summarize, we feel the following factors may lead to false interpretation of sliding
motion on cine MRI; mediastinal vessels that are away from the heart, large and bulky
mediastinal masses, gross pericardial effusion (limiting the motility of pericardium)
and inflammatory adhesions between the mass and adjacent cardiovascular structures.
Some previous studies evaluated the effectiveness of cine MRI in assessing cardiovascular
invasion by mediastinal masses. A study conducted by Seo et al. in 2005 evaluated the presence of sliding motion in 39 cardiovascular structures
in 20 patients with accuracy of 94.4%.[16] Ong et al. in 2016 assessed sliding motion in 48 cardiovascular structures with accuracy, sensitivity
and specificity of 88.5%, 70%, and 97.6%, respectively.[18] Both studies also assessed the presence of intervening fat planes between mediastinal
mass and cardiovascular structure using T1WI and T2WI. None of the above studies looked
at the role of type 2 chemical shift artefact in excluding invasion of cardiovascular
structures.
After a trial with few cases and realizing the potential benefits, the authors now
routinely clinically use cine MRI as a problem-solving tool for any doubt of cardiovascular
invasion by mediastinal masses on CT.
MRI-tagging sequences have been used for the diagnosis of constrictive pericarditis
by looking at adherence and immobility of the myocardial-pericardial interface.[19], [20] These sequences use a grid-like pattern of saturated areas over an image slice and
allow the study of the deformation of the grids with cardiac cycle. To our knowledge,
these sequences have not been used in the context of the invasion of cardiovascular
structures by mediastinal masses. We propose that these sequences can be used in future
studies.
The authors are aware of some of the limitations of the study. Firstly, this was a
retrospective study with its associated limitations. The accuracy of imaging findings
involving 25 cardiovascular structures in 7 patients who were not operated owing to
the advanced stage of disease or distant metastases, could not be assessed, hence,
these structures had to be excluded from the study.
Conclusion
SSFP cine MRI can be used as an effective complementary imaging modality in the evaluation
of cardiovascular invasion by mediastinal masses and help in planning the future course
of management. The presence of type 2 chemical shift artifact and differential sliding
motion between the mass and cardiovascular structure can effectively exclude invasion,
thus facilitating surgery in patients with equivocal CT findings.
