| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Harvard Medical School and Beth Israel Deaconess
Medical Center, One Deaconess Rd., Boston, MA 02215.
2 Department of Radiology, Hospital de Basurto, Bilbao 48013, Spain.
3 Averion Inc., 4 California Ave., Framingham, MA 01701.
Received August 24, 2004;
accepted after revision December 14, 2004.
Address correspondence to I. Pedrosa.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Twenty-five MR examinations were retrospectively reviewed independently by two radiologists to rule out thrombosis in the central veins of the body. The presence of venous thrombus was assessed separately in 80 veins using true FISP and gadolinium-enhanced T1-weighted images. Diagnosis was confirmed by another imaging technique (sonography, CT, and/or conventional venography) in all positive cases. Negative examinations were confirmed using imaging, clinical follow-up, or both.
RESULTS. Venous thrombosis was present in 25 veins in 18 patients. True FISP images had a lower sensitivity (66%) and specificity (70.9%) for the diagnosis of venous thrombosis than gadolinium-enhanced MR images (p < 0.01).
CONCLUSION. True FISP images have lower sensitivity and specificity in the diagnosis of venous thrombosis than gadolinium-enhanced T1-weighted gradient-echo images. True FISP images should not be used exclusively for the diagnosis of venous thrombosis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
These attributes are favorable for depicting luminal morphology and potentially for depicting intraluminal filling defects of blood vessels. True FISP images are routinely obtained as a localizer for the gadolinium-enhanced 3D acquisition in our institution. We have observed that venous thrombus detected on gadolinium-enhanced MR images can be visualized on true FISP images. On the basis of this observation, we hypothesized that true FISP images may allow the diagnosis of deep vein thrombosis without requiring administration of IV contrast medium. Such a capacity would be attractive, particularly for patients with limited IV access. In addition, this approach has the potential to shorten the total examination time and decrease the cost of the study by eliminating the need for contrast medium.
The aim of this study was to assess the accuracy of true FISP images of the chest, abdomen, and pelvis in patients with suspected deep venous thrombosis of the central veins of the body and compare those results with results from gadolinium-enhanced MR venography.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our MR protocol for evaluation of deep veins in the chest, abdomen, and pelvis includes true FISP images, typically oriented perpendicular to the vessel or vessels where thrombosis is suspected. This approach allows identification of the vessel or vessels of interest in an expeditious manner. Three-dimensional gadolinium-enhanced MR acquisitions are subsequently prescribed using true FISP imaging to ensure appropriate coverage of the area of interest. We routinely use gadolinium-enhanced MR images for the diagnosis including source images and multiplanar and maximum-intensity-projection reconstructions. Thus, all patients were studied with both true FISP and gadolinium-enhanced 3D fat-suppressed T1-weighted gradient-echo sequences.
This study was approved by the committee on clinical investigations at our institution, and informed consent was not required for this retrospective review.
Based on the information obtained from each patient's chart, the time between the onset of symptoms and the date of the MR examination was arbitrarily tabulated with the following scale: 1, less than 1 week; 2, 1-2 weeks; 3, 2-4 weeks; or 4, more than 4 weeks.
MRI Protocol
All images were obtained on 1.5-T scanners (Vision or Symphony, Siemens
Medical Solutions) equipped with high-performance gradients (25 mT/m; slew
rate, -125 T/m sec). True FISP images were obtained in the axial, sagittal,
and/or coronal planes perpendicular to the vessel of interest, depending on
the clinical history of the patient. These images allow quick identification
of the blood vessels in the area of interest and help to prescribe the
gadolinium-enhanced 3D acquisition. We encourage our MR technologists to use
these images to ensure sufficient anatomic coverage for the 3D MR acquisition
by recognizing the veins of interest within the prescribed volume. Cardiac
gating was not used during acquisition of the true FISP images. True FISP
images were obtained using the following parameters: TR/TE, 4.8/2.3; flip
angle, 70°; slice thickness, 5 mm; 1-mm gap; matrix, 134 x 256;
bandwidth, 560-650 Hz/pixel; and a rectangular field, the latter when possible
based on body habitus. These parameters provided good image quality and
signal-to-noise ratio and were selected on the basis of the parameters
available in our MR scanners and the need to cover the area of interest in 1-2
breath-holds.
Axial true FISP images were obtained in the chest, abdomen, and/or pelvis in all patients except one in whom only sagittal images of the chest were obtained. Sagittal true FISP images of the chest were obtained in nine patients for evaluation of the subclavian or axillary veins. Coronal images were obtained in eight patients for rapid identification of the vascular anatomy. Coronal images were not used for this study because blood vessels are usually displayed in the plane of the image and partial volume effects in the coronal plane could impair the assessment of intravascular filling defects.
Gadolinium-enhanced MR venography was performed using a 3D fat-suppressed T1-weighted gradient-echo sequence with the following parameters: 4.2/1.7; flip angle, 25°; matrix, 160 x 256; and slice thickness, 3-5 mm before interpolation. The orientation of this volumetric acquisition and the slab thickness were established on the basis of the requirements of anatomic coverage and patient body habitus. Gadolinium-enhanced MR images were prescribed using the anatomic information provided by the true FISP images to ensure adequate coverage of the region of interest.
In all patients, 3D T1-weighted gradient-echo images were obtained before and after administration of a double dose (0.2 mL/kg of body weight) of gadopentetate dimeglumine (Magnevist, Berlex Laboratories). Contrast material was administered in a biphasic manner (0.1 mL/kg of body weight at 2 mL/sec followed by 0.1 mL/kg of body weight at 0.8 mL/sec). Three postcontrast breath-hold data sets were acquired, the first timed for the arterial phase. The second and third acquisitions were performed at 40 and 90 sec after the arterial phase while approaching the equilibrium phase.
The arterial phase was captured using a timing examination with administration of 1 mL of gadopentetate dimeglumine followed by 20 mL of normal saline solution at a rate of 2 mL/sec as previously described [6]. The timing examination was acquired at the level of the pulmonary arteries, mid kidney, or aortic bifurcation for the thoracic, abdominal, and pelvic MR examinations, respectively. On the basis of the timing examination result, a "vein-free" arterial phase acquisition was synchronized to the maximal arterial enhancement. All images were acquired during suspended respiration at end expiration.
The 3D data set acquired during the arterial phase was subtracted from those acquired during the venous phases at 40 and 90 sec after the arterial peak. This results in two 3D data sets that emphasize venous structures [7]. For this study, the unprocessed 3D data sets acquired at 40 sec after the arterial peak were used for the diagnosis of venous thrombosis.
MRI Interpretation
A retrospective review of the MR examinations in all patients was performed
independently by two experienced radiologists using a local PACS monitor
(Centricity, GE Healthcare). True FISP images were evaluated first for the
presence of intraluminal filling defects in the central veins of the body.
Both reviewers independently evaluated each of the main central veins of the
neck and chest, abdomen, and pelvis on true FISP images. Only those vessels
that were visualized perpendicular to the plane of the true FISP images were
evaluated. Therefore, axial images were used for evaluation of the internal
jugular, superior vena cava, inferior vena cava, iliac, and common femoral
veins. Sagittal images were used for assessment of the right and left
subclavian veins.
Diagnosis of venous thrombosis was made when a filling defect with lower signal intensity than that of the normal veins was detected within the lumen of the vein. A 5-point scale was used for the diagnosis of deep vein thrombosis on true FISP images: 1, negative; 2, probably negative; 3, equivocal; 4, probably positive; and 5, positive.
Both interpreters then reviewed the same vessels on contrast-enhanced 3D T1-weighted gradient-echo images for the presence of intraluminal filling defects. The presence or absence of venous thrombosis on these images was also tabulated using a similar 5-point scale to the one described for the true FISP images. Two cases with an equivocal result for gadolinium-enhanced MR venography were reviewed, and a final result was given by consensus of both reviewers. MRI findings were confirmed in these two patients by conventional venography and CT angiography of the chest, respectively.
In those cases in which the reviewers identified thrombus on the contrast-enhanced images, true FISP images were again reviewed for assessment of the signal intensity within the vessel at the same level where the thrombus was detected on gadolinium-enhanced images. The signal intensity within the vein on true FISP was compared with that of the adjacent veins in the same image. A 5-point scale was used for signal intensity in these cases: 1, slightly higher; 2, isointense (high signal intensity); 3, slightly lower; 4, much lower; and 5, black.
Proof of Diagnosis
The proof of diagnosis was based on the gadolinium-enhanced MR images along
with correlations to other imaging studies in those cases deemed positive for
thrombosis. The reported sensitivity and specificity of gadolinium-enhanced MR
venography for the diagnosis of venous thrombosis are 100%
[1-3].
Available confirmatory studies included sonography, conventional venography,
and CT in all patients with positive findings for thrombosis. Negative
gadolinium-enhanced MR venography examinations were confirmed with sonography,
CT, and/or clinical follow-up.
Statistical Analysis
The weighted kappa statistic was used to quantify the agreement between the
two reviewers. The weighted kappa was used to construct a statistical test for
evaluating a null hypothesis of no agreement between the reviewers. Results of
this statistical test were the basis for or against combining scores across
reviewers.
Sensitivity and specificity were calculated by collapsing the 5-point scale used for the diagnosis of venous thrombosis on true FISP and gadolinium-enhanced images to a 3-point scale. The positive and probably positive categories were combined into one and the negative and probably negative categories into another. The inconclusive categories remained intact because these results could not be classified as either positive or negative. The average of the two reviewers' estimates of sensitivity and specificity were reported. In a similar fashion, false-positive, false-negative, and inconclusive averages were reported.
The marginal homogeneity test was used to evaluate the differences between true FISP images and gadolinium-enhanced images in the diagnosis of venous thrombosis.
In patients with venous thrombosis, the Jonck-heere-Terpstra test was used to evaluate the relationship between the signal intensity of the thrombus on true FISP images and the time between the onset of symptoms and the date of MRI.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Eighteen (72%) of the 25 studies had venous thrombosis in 25 vessels shown on gadolinium-enhanced MR images. The diagnosis of venous thrombosis was confirmed in all positive cases by another imaging technique, including sonography (n = 6), conventional venography (n = 10), or CT (n = 2). One patient with thrombus in the left iliac vein on MR venography underwent CT angiography (CTA) of the chest; CTA depicted pulmonary embolism, and these findings were used as proof of diagnosis. In the six negative cases, gadolinium-enhanced MR images alone served as proof in three patients. These three patients have had no evidence of venous thrombosis on subsequent clinical follow-up at 3 months, 2 years, and 3 years, respectively. One patient had a positive CTA of the chest for pulmonary embolism and a questionable filling defect in the right subclavian vein. Both reviewers agreed that this was related to turbulent flow, and no thrombosis was noted at this level on gadolinium-enhanced MR venography. In the other two patients, the negative result was confirmed with sonography (n = 1) or CT (n = 1).
The weighted kappa (
w) test indicated that there was
significant evidence to support agreement in the diagnosis of venous
thrombosis between the two reviewers for both true FISP and
gadolinium-enhanced images (w= 0.73 and p < 0.01
vs w= 0.97 and p < 0.01, respectively). On the
basis of these statistical results, combined scores of the reviewers were
calculated for the diagnosis of venous thrombosis by averaging the reviewers'
scores per vessel for each imaging technique separately.
Reviewers' interpretations for true FISP and gadolinium-enhanced MR images are included in Table 1. Using the vessel-specific average score, the marginal homogeneity test revealed a significant difference between the two imaging techniques for the diagnosis of thrombosis (p < 0.01).
|
For the diagnosis of venous thrombus, true FISP images had a sensitivity of 66.0% and a specificity of 70.9% compared with the gold standard, gadolinium-enhanced MR images. Overall accuracy of true FISP images for the diagnosis of venous thrombosis was 69.4% (Figs. 1A, and 1B). The inconclusive results (score of 3) for true FISP images occurred in 26.0% of veins with venous thrombosis and 21.8% of veins without venous thrombosis. Both reviewers had two false-negative diagnoses of venous thrombus on true FISP images in 25 veins with thrombus (8.0%) (Figs. 2A, and 2B). Both reviewers had false-positive results on true FISP images in five and three of 55 veins without thrombus, respectively. The estimate of the false-positive rate was 7.3% (Figs. 3A, 3B, and 3C).
|
|
|
|
|
|
|
|
|
|
|
The remaining patients who were scanned between 2 and 4 weeks (n = 4) and more than 4 weeks (n = 4) from the onset of the symptoms had venous thrombus in eight and five veins, respectively. The thrombus in these patients showed much lower or black signal intensity compared with that of the adjacent vessels. Two patients had occlusion of the right subclavian vein and superior vena cava, respectively. These vessels were not visualized likely due to chronic thrombosis; therefore, the signal intensity of the thrombus was not evaluated in these two venous segments. Patients with acute thrombus (< 1 week) (n = 7) showed venous thrombus with slightly lower (n = 1), much lower (n = 4), or black (n = 2) signal intensity on true FISP images. Hyperintense thrombus was not noted in any patient with acute (< 1 week from the onset of symptoms) (n = 5) or chronic (> 4 weeks from the onset of symptoms) (n = 1) venous thrombosis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Gadolinium-enhanced MR venography offers a faster acquisition alternative to TOF imaging for the diagnosis of venous thrombosis. Gadolinium-enhanced MR venography can be performed by injecting contrast medium in any vein of the body, typically in the nonaffected side [11]. Gadolinium-enhanced MR venography has excellent sensitivity, specificity, and accuracy in the diagnosis of venous thrombosis of the central veins of the chest, abdomen, and pelvis [1, 2].
The use of gadolinium-enhanced MR venography as a gold standard deserves comment. Three prior reports have indicated that the sensitivity and specificity of gadolinium-enhanced MR venography for the detection of venous thrombosis is 100% [1-3], strong evidence for its ability to serve as a reference standard. It should be noted, however, that a recent report by Baarslag et al. [12] showed "disappointing" results for gadolinium-enhanced MR venography.
In the study performed by Baarslag et al. [12], the authors did not describe the location of the deep vein thrombosis for each patient or vascular segment. The rationale behind their use of a relatively small field of view (30 cm) for gadolinium-enhanced MR venography was not described. That small field of view excluded the visualization of veins with conventional venography-positive thrombus. Because those cases were considered as false-negative results, the sensitivity of gadolinium-enhanced MR venography was perhaps artificially reduced. Interestingly, the authors did not specifically describe cases in which gadolinium-enhanced MR venography missed venous thrombosis in the vascular segments that were included within the field of view.
To add credence to our methods, we pursued correlative imaging and clinical follow-up, the latter for thrombus-negative cases, for confirmation. In our study, on a case-by-case basis, we had correlative imaging showing that none of the patients with negative gadolinium-enhanced MR venography had evidence of thrombus on another imaging examination or subsequent follow-up. Furthermore, all segments (and hence all cases) that were positive had confirmatory correlative imaging. Thus, our results support the prior studies that affirm gadolinium-enhanced MR venography as a gold standard for venous thrombosis.
Patients with end-stage renal disease, cancer, or a hematologic disease are referral candidates for MR evaluation of the central veins of the body especially after multiple failed attempts of central venous catheterization. In these patients, peripheral venous access is not always available and in those circumstances gadolinium-enhanced MR venography may not be possible. We were anticipating that true FISP MR venography might provide a fast and reliable examination of the central veins of the body without the need for IV administration of contrast medium, as had been previously proposed [13].
True FISP imaging is a coherent steady-state technique in which the gradients are fully balanced to recycle the transverse magnetization in long T2 species [14]. Blood vessels show high signal intensity on true FISP images due to the high T2-to-T1 ratio of blood [14] and the fact that the intrinsic T2 of blood is relatively long [15].
True FISP imaging has been proposed as an alternative to gadolinium-enhanced MR angiography in patients with suspected acute aortic dissection and aneurysm [4]. Similarly, a comprehensive evaluation of the hepatic vasculature in potential liver donors has been proposed using true FISP imaging [5]. MR venography with high-resolution true FISP imaging has been recently proposed as a noninvasive, fast approach for the detection of venous thrombosis [13]. This technique allowed thrombus visualization by providing high contrast compared with the surrounding blood pool within the veins of the abdomen, pelvis, and lower extremities in less than 10 min.
Pulsation artifacts are frequent on true FISP images [14]. True FISP images were equivocal for the presence of venous thrombus in 26.0% of the veins with venous thrombus and 21.8% of the veins without thrombus in our series. We found a false-positive result of venous thrombosis in 7.3% of our cases using true FISP images. These results are likely secondary to the presence of pulsation artifacts and inhomogeneous signal within the veins, which resulted in decreased confidence in the diagnosis of both reviewers. We found inhomogeneous signal intensity within the veins on true FISP images, probably related to pulsatile flow, in a substantial number of studies. These artifacts were seen in our series as intraluminal areas of decreased signal intensity at the level of the confluence of veins or in areas where the veins change abruptly in direction and can be misinterpreted as venous thrombus. Differentiation between venous thrombus and pulsation artifact on true FISP images can be difficult.
Characterization of thrombus age can be clinically important, particularly when thrombolytic therapy is a consideration. Froehlich et al. [16] reported increased enhancement in the vessel wall in the first 2 weeks after deep venous thrombosis. Those authors found a correlation between this finding and inflammatory response in the wall of the thrombosed vessel at pathology [16]. We did not evaluate gadolinium-enhanced MR images for the presence of wall enhancement.
In our series, two patients were scanned between 1 and 2 weeks after the onset of symptoms and true FISP images failed to reveal thrombus that was detected on gadolinium-enhanced images; it is presumed that the non-visualization is due to the thrombus having T2/T1 characteristics similar to those of the blood pool. Similar findings were also noted in a third patient who underwent an MR venography examination 26 days after the onset of symptoms. Corti et al. [17] reported an increase in signal intensity on T1- and T2-weighted images of carotid thrombi induced in swine. At 1 week after induction of the thrombus, the presence of methemoglobin with short T1 relaxation time is responsible for the increased signal intensity on T1-weighted images [17]. Increased signal intensity on T2-weighted images is due to increased water content of lysed RBCs [17]. The relative increase in signal intensity was significantly higher on T2-weighted images than that on the T1-weighted images, and the strongest statistical significance occurred during the first 3 weeks. This phenomenon may account for the high-signal-intensity thrombus seen in our three patients. We believe that the presence of high-signal-intensity thrombus on true FISP images is suggestive of subacute thrombus occurring between 1 and 2 weeks before the MR examination.
This study has several limitations. An unintentional selection bias is possible resulting from the retrospective nature of the study and from the standard screening of patients referred to MRI. This screen assesses for patients who have MR-incompatible implanted or indwelling devices, have a severe concern for claustrophobia, have an allergy to gadolinium, or are pregnant. The information about those excluded at screening is not routinely tracked and therefore analysis is not possible. Also, this retrospective review cannot ensure a systematic correlation of MRI findings with other imaging techniques on a per-vessel basis for all cases.
In this study, the positive result for venous thrombosis on gadolinium-enhanced MR venography included verification with another imaging technique. However, it should be noted that some segments lacking venous thrombus were not verified. Specifically, several patients with upper extremity thrombus without thrombus within the superior vena cava on gadolinium-enhanced MR venography had sonography confirmation limited to the upper extremity thrombus. In these patients, lack of thrombus in the superior vena cava was assumed on the basis of the normal appearance of this vessel on the gadolinium-enhanced MR images alone. Similarly, several patients with unilateral thrombosis of iliac or femoral veins and no thrombus on the contralateral side had confirmation of the thrombus in the affected side with conventional venography. In those cases, conventional venography of the contralateral side was not performed. However, a sensitivity of 100% (0 false-negatives) has been reported for the detection of venous thrombosis with gadolinium-enhanced MR venography, thus supporting its function as a standard of reference [1-3].
We used the shortest TR available in our scanner (4.8 msec) for the true FISP images. A further decrease in TR may reduce some artifacts, although it is unclear whether this change would impact in the detection of venous thrombus.
The use of cardiac triggering (not used in our study) may reduce pulsation artifacts and thus improve the homogeneity of the signal within vessels. We obtained our true FISP images without cardiac triggering because our initial intent was to pursue anatomic information to prescribe the gadolinium-enhanced 3D acquisitions; furthermore, the triggered true FISP sequence was unavailable on the Vision scanner (Siemens Medical Solutions). Cardiac triggering has limitations: It may increase the total acquisition time and is unlikely to benefit patients with dysrhythmia. More studies are necessary to assess the impact of different scanning strategies, including cardiac triggering, in true FISP venography.
Recognizing that the suggested improvements to the true FISP protocol may reduce some false-positives, it would be unlikely for these to eliminate the false-negatives because the latter are related to the relaxation features of thrombus and blood rendered by this sequence. Thus, at this time, it seems prudent to avoid using true FISP images as an exclusive means for the diagnosis of venous thrombosis.
However, the high signal intensity of thrombus on true FISP images was associated with the subacute phase after the onset of symptoms, albeit in a small number of individuals. Thus, a phenomenon responsible for a pitfall for the diagnosis of venous thrombus holds potential—when used with gadolinium-enhanced MR venography—to characterize thrombus. If reliable, such a capacity may help identify candidates likely to benefit from fibrinolytic therapy [18]. More studies are necessary to confirm these results and implications.
In conclusion, the sensitivity, specificity, and accuracy of true FISP imaging for the detection of thrombus in the central veins of the body are too limited for this sequence to serve as the sole means for diagnosis. True FISP imaging holds promise for thrombus characterization. More studies are encouraged to assess the accuracy of true FISP images for age characterization of venous thrombus.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  H. Kandpal, R. Sharma, S. Gamangatti, D. N. Srivastava, and S. Vashisht Imaging the Inferior Vena Cava: A Road Less Traveled RadioGraphics, May 1, 2008; 28(3): 669 - 689. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
